Economics of Employment and Unemployment [1 ed.] 9781608765485, 9781604567410

145 98 7MB

English Pages 377 Year 2009

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Economics of Employment and Unemployment [1 ed.]
 9781608765485, 9781604567410

Citation preview

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ECONOMICS OF EMPLOYMENT AND UNEMPLOYMENT

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ECONOMICS OF EMPLOYMENT AND UNEMPLOYMENT

MANON C. FOURNIER AND

CHLOE S. MERCIER

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

EDITORS

Nova Science Publishers, Inc. New York

Copyright © 2009 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Economics of employment and unemployment / [edited by] Manon C. Fournier and Chloe S. Mercier. p. cm. Includes index. ISBN 978-1-60876-548-5 (E-Book) 1. Unemployment--Developing countries. 2. Full employment policies--Developing countries. 3. Manpower policy--Developing countries. I. Fournier, Manon C. II. Mercier, Chloe S. HD5852.E38 2009 331.109172'4--dc22

2008052453

Published by Nova Science Publishers, Inc.    New York

CONTENTS Preface Chapter 1

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor Tomáš Sirovátka and Petr Mareš

1

Chapter 2

Evidence on the Relationship Between Unemployment and Health Anton Nivorozhkin and Laura Romeu Gordo

33

Chapter 3

Spanish Unemployment and the Ladder Effect Fabrice Collard, Raquel Fonseca and Rafael Munoz

61

Chapter 4

Juvenile Delinquency and Its Forensic Considerations Dr. B. R. Sharma, M.B.B.S., M.D. Professor, Aparajita Sharma, B.A.

105

Chapter 5

Wage and Employment in a Finance-Led Economy Célia Firmin

125

Chapter 6

Employment, Occupational Disparities, and Wages in India: A Decomposition Analysis Rajarshi Majumder and Dipa Mukherjee

151

Short- and Long-Run Effects of Productivity Shocks in an Intertemporal Model Pu Chen, Gang Gong, Armon Rezai and Willi Semmler

181

Chapter 7 Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

vii

Chapter 8

Unemployment Insurance: Factors Associated with Benefit Receipt United States Government Accountability Office

Chapter 9

Inflation, Unemployment, and Labor Force Change in European Countries Ivan O. Kitov

263

On the Linkages Between Monetary Policy, Inflation, Output and Unemployment in the EURO Zone Priti Verma and Rahul Verma

309

Chapter 10

199

vi Chapter 11

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Index

Contents Inflation, Deflation and Employment: A Macrodynamic Approach Toichiro Asada

325 349

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

PREFACE Employment and unemployment disparities affect the livelihood of various developing countries. Such issues that will be discussed in this book include the job availability, type, sector of employment and job earnings of these societies. This book provides a comprehensive study of the improvement of poverty, inequality and achievement of employment for socio-economical groups all across the globe. Chapter 1 - The purpose of the study is to explore the relationship between the forms of labour market marginalization — understood here in terms of labour market status and job quality — on the one hand, and income disadvantage, material deprivation and social exclusion on the other. Public policies that aim to improve the labour market status and levels of income of those disadvantaged on the labour market are also discussed. The authors use data gathered in a survey on social exclusion in which 2,500 respondents were interviewed, they were either welfare benefit recipients or considered their situation similar to their. The authors demonstrate that marginalization on the labour market is evident not only in relation to unemployment (often repeated and long-term) but at the same time in temporary, low paid and poor quality jobs. The incomes of those employed in the lowest segment of the labour market and of the unemployed are very similar while deprivation of the unemployed is greater in many respects, e.g. in opportunities to influence the course of their lives and the life opportunities of their own as well as of their families in particular. The authors identify under-use of welfare benefits and measures that might improve the standard of living and human capital of those who are disadvantaged. A portion of the disadvantaged remain active on the labour market and identify employment incentives, yet the authors also identify poverty traps which emerge in the case of those who become discouraged and welfare dependent. Chapter 2 - In this chapter the authors review recent studies on the impact of unemployment on individual health and well-being. The authors start the discussion with an overview of the theoretical background of the relationship between unemployment and wellbeing. Next, the authors summarize recent empirical studies with the objective to provide a basis for scholars who want to contribute to the literature in this field. Chapter 3 – This chapter aims to examine to what extent a "ladder" effect may contribute to explain changes in unemployment in Spain. The "ladder" effect arises when highly-skilled workers who do not find a job that matches their skills, accept jobs that were previously occupied by less qualified staff. The authors develop a dynamic general equilibrium model. The model is then calibrated for the Spanish economy. The authors’ results replicate the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

viii

Manon C. Fournier and Chloe S. Mercier

observed decline in the ratio of high- to low–skilled vacancies, and explain how firms substitute high for low–skilled employment. These results also suggest that in the Spanish case, ladder effect can be better explained by increases in training costs interpreted as a biased-shock against low-skilled workers. Chapter 4 - Juvenile Delinquency and the problems related to it have been faced by all societies, all over the world; however, in the developing world the problems are all the more formidable. The process of development has brought in its wake a socio-cultural upheaval affecting the age-old traditional ways of life in the congenial rural milieu. In fact, various scientific advances and concomitant industrialization and urbanization have ushered in a new era, which is characterized by catastrophic changes and mounting problems. Cities have sprung up with heterogeneity of population, cultural variations, occupational differentiations and overcrowded conditions. As a result, social disorganization and maladjustment have taken place following a perennial influx of people from their rural habitat to the urban squalid slums. Juveniles are adversely affected by these changing conditions. At the same time, the traditional social control system that served as a preventive check against any antisocial activity is gradually giving way. Consequently, the problem of juvenile deviance and antisocial propensities is rearing its ugly head. This paper examines the causes of delinquency and attempts to analyze the prevalence and prevention of deviance among adolescents and young adults. Chapter 5 - Since the beginning of the 1990’s, some OECD countries have followed a financialization trend. This trend results in an increase in distributed dividends and a decrease in the accumulation rate and in the wage share. The growth slowdown increases the unemployment rate. This is the case for example in France and Germany. The object of this chapter is to analyse wage and employment determination in a financeled economy. For this, the authors will use a post Keynesian “stock-flow” macroeconomic model and numeric simulations. The introduction of financial variables modifies investment and consumption equations’ form. Financialization results in a new macroeconomic dynamic which affects employment and wage determination. Employment, and more precisely unemployment, plays a central role in the income distribution dynamics. In this institutional framework, the distributive conflict modifies the income distribution. This conflict depends on the unemployment rate and on the shareholder negotiation power. It results in wage share fluctuations. These fluctuations are central in the macroeconomic dynamics. Indeed, the consumption evolutions modify the firms’ capacity utilisation rate and so investment. More, financialization represents the shareholders’ power increase in management decision. In this framework, profitability norms set by shareholder in the short term may reduce investment. The authors see that, in a Keynesian perspective, the introduction of financial variables increases the effective demand constraint which firms face. The growth slowdown is linked to two determinants. The first is the new income distribution dynamics, with a decrease in the wage share. The second is the new investment behaviour, with an increase in the investment selectivity due to high required profitability norms. So, financialization raises unemployment due to growth slowdown. Then the unemployment rate explains a part of the decrease in the wage share. The authors will analyse income distribution effects on investment and consumption functions in a finance-led economy. As the authors see, in this context, wage and employment represent the adjustment variables.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Preface

ix

Chapter 6 - Disparities in livelihood among socio-economic groups are a major problem in developing countries, and India is no exception. Such differentials are caused mainly by disparities among various groups in terms of job availability, type and sector of employment, and earnings there from. However, these differences transcend the boundaries of current generation and through their impact on capability formation and asset creation tends to perpetuate, and even accentuate, disparities in living standards of future generations, endangering social sustainability altogether. Policy framing for inclusive development as envisaged by the Millennium Development Goals would therefore require a thorough study of employment and earning differentials and causes thereof. The present paper explores the issue of Employment and Wage Differentials in the Indian labour market over the last decade across social classes, regions, gender, and job types. Using both parametric and nonparametric techniques, differences in employment rates, its nature, occupational distribution, wage rates, and total earnings have been explored. The roles played by Discrimination in labour market, both during entry and during wage setting, and that by Endowment in explaining the occupational and earning disparities have been examined through modern Decomposition techniques. The dynamics of these issues have also been explored against the prevalent theoretical wisdom that in a globally integrated world economy, vertical disparities may increase while horizontal disparities within groups would come down. The results and inferences drawn from this paper are expected to become a comprehensive study of Indian labour market in recent times and have serious policy implications for removal of poverty and inequality and achievement of Millennium Development Goals. Chapter 7 - In this chapter the authors study the relationship betweeen unemployment and productivity growth with respect to the short and long run. In order to do so, the authors propose a two-stage dynamic general equilibrium model. The predictions of which the authors then take to US unemployment and productivity data. Using MLE and SVAR, the model’s predictions appear to be correct: Productivity growth has an abiguous effect on unemployment — reducing it in the long run, while temporarily increasing it in the short run. Chapter 8 - Unemployment Insurance (UI), established in 1935, is a complex system of 53 state programs that in fiscal year 2004 provided $41.3 billion in temporary cash benefits to 8.8 million eligible workers who had become unemployed through no fault of their own. Given the size of the UI program, its importance in helping workers meet their needs when they are unemployed, and the little information available on what factors lead eligible workers to receive benefits over time, GAO was asked to determine (1) the extent to which an individual worker’s characteristics, including past UI benefit receipt, are associated with the likelihood of UI benefit receipt or unemployment duration, and (2) whether an unemployed worker’s industry is associated with the likelihood of UI benefit receipt and unemployment duration. Using data from a nationally representative sample of workers born between 1957 and 1964 and spanning the years 1979 through 2002, and information on state UI eligibility rules, GAO used multivariate statistical techniques to identify the key factors associated with UI benefit receipt and unemployment duration. In its comments, the Department of Labor stated that while there are certain qualifications of the authors’ findings, the agency applauds the authors’ efforts and said that this report adds to the authors’ current knowledge of the UI program. Chapter 9 - Linear relationships among inflation, unemployment, and labor force are obtained for two European countries — Austria and France. The best fit models of inflation as a linear and lagged function of labor force change rate and unemployment explain more than

x

Manon C. Fournier and Chloe S. Mercier

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

90% of observed variation (R2>0.9). Labor force projections for Austria provide a forecast of decreasing inflation for the next ten years. In France, inflation lags by four years behind labor force change and unemployment allowing for an exact prediction at a four-year horizon. Standard error of such a prediction is lower than 1%. The results confirm those obtained for the USA and Japan and provide strong evidence in favor of the concept of labor force growth as the only driving force behind unemployment and inflation. Chapter 10 - This chapter examines whether monetary policy shocks have any varying degrees of effect on the inflation, output and unemployment indicators of the European Union member countries.The authors hypothesize that a positive monetary policy shock increases the output and the price levels causing a decrease in unemployment. The results show that a one standard deviation positive shock in monetary policy positively affects inflation in France and Italy. Also there is negative response of unemployment to the monetary base in the case of Germany, France and Italy. The authors do not find any such significant relationship between output and monetary policy. Overall, monetary policy shocks have asymmetric effect on inflation and unemployment. Chapter 11 - In this chapter, the authors formulate a series of simple Keynesian macrodynamic models which can provide a systematic interpretation of the problems of inflation, deflation and (un)employment. Although the structures of the authors’ models are quite simple and transparent, there are many endogenous variables so that the dimensions of their systems are high. In fact, their models consist of four-dimensional and five-dimensional systems of nonlinear differential equations. The authors investigate the solutions of the systems analytically, and then the authors provide their economic interpretations. The models in this paper are designed to contribute to a theoretical interpretation of the macroeconomic performance of the Japanese economy during the period 1980s – 1990s. In the latter part of the paper, the authors compare the solution of one of their models and the basic macroeconomic data of the Japanese economy during this period, and the authors show that the solution of that model is consistent with the performance of the Japanese economy. The final section is devoted to the concluding remarks.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 1

POVERTY, DEPRIVATION AND SOCIAL EXCLUSION: THE UNEMPLOYED AND THE WORKING POOR* Tomáš Sirovátka1 and Petr Mareš2 Faculty of Social Studies, Masaryk University Joštova 10, 602 00 Brno, Czech Republic

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT The purpose of the study is to explore the relationship between the forms of labour market marginalization — understood here in terms of labour market status and job quality — on the one hand, and income disadvantage, material deprivation and social exclusion on the other. Public policies that aim to improve the labour market status and levels of income of those disadvantaged on the labour market are also discussed. We use data gathered in a survey on social exclusion in which 2,500 respondents were interviewed, they were either welfare benefit recipients or considered their situation similar to their. We demonstrate that marginalization on the labour market is evident not only in relation to unemployment (often repeated and long-term) but at the same time in temporary, low paid and poor quality jobs. The incomes of those employed in the lowest segment of the labour market and of the unemployed are very similar while deprivation of the unemployed is greater in many respects, e.g. in opportunities to influence the course of their lives and the life opportunities of their own as well as of their families in particular. We identify under-use of welfare benefits and measures that might improve the standard of living and human capital of those who are disadvantaged. A portion of the disadvantaged remain active on the labour market and identify employment incentives, yet we also identify poverty traps which emerge in the case of those who become discouraged and welfare dependent. 1 e-mail: [email protected] ,Phone + 42 (0)549496559 Fax + 42 (0)549491920 2 [email protected] ,Phone + 42 (0)549496559 Fax + 42 (0)549491920

2

Tomáš Sirovátka and Petr Mareš

INTRODUCTION

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

After the fall of the communist regime in 1989 the former Czechoslovak labour market went through a significant transformation. The labour market has existed before yet it was malformed due to strong regulatory interventions based on political decisions, these involved above all subsidies for fields characterized by a high share of unskilled labour, the infrastructure of such fields was also financed from the state budget involving a form of redistribution. This led to constant over-employment and the survival of ineffective industries as well as individual companies whose production was not exposed to competition. The privileging of the principle of hierarchy as opposed to meritocratic principles, levelling of wages and wage regulation exercised by the state combined with strong unions explains why the price of labour was never a market one. State monopoly in terms of the only major employer of all employees at the same time protected employees (it provided a social security 3 net) as well as exploited them, however, it mainly effectively controlled them. After 1989 in Czechoslovakia and after its break-up in the Czech Republic a real labour market was gradually formed and unemployment also grew in the process. Until 1995 the growth was gradual and unemployment remained under 5% due to economic as well as noneconomic reasons. On the one hand, the fact that the first unemployment wave was not primarily the result of dismissals but rather the most active individuals employed in state 4 companies left those in order to create a new private sector [Možný, 1994] played a role. On the other hand the breakdown of social consensus, destabilization of the society and a halt in the transformation of the society as a result of mass dismissals represented a nightmare that 5 the new elites inherited from the old ones. The conviction that low unemployment rates and thus social consensus were conditions of a successful transformation belonged among the most significant myths that accompanied the social transformation in the Czech Republic (as 6 well as in Slovakia) shortly after 1989. This myth was to consolidate the continuity and discontinuity of reality in the course of changes and for some time it fulfilled this role successfully. However, in the long run this myth was impossible to maintain and it was gradually questioned and abandoned. Although after 1996 the unemployment rate in the Czech Republic did not reach the same level as in neighbouring post-communist countries (15-20 %), it nonetheless gradually grew to the current 8-10 % (and in some regions even significantly higher). 3

The fact that unemployment was not a real threat for the population was compensated by wage levelling and extensive income redistribution. It was very rare to lose a job (those who lost it voluntarily were immediately criminalized as an undesirable). However, anyone could have lost their jobs (which they liked and were qualified to do) if they were disliked by those in power. Thus under the communist regime the labour market belonged among the major tools of social control. Because the state had all jobs under control there was no escape for anybody. 4 For more detail see Mareš, Rabušic [1994]. 5 We should not forget that at the end of the 1980s the continuity of full employment represented one of the last sources of legitimacy of the communist regime. Although many knew - and the majority suspected - that the price paid for full employment was the declining effectiveness and competitiveness of the Czech economy and the decline and failing of industrial plants, housing, transport, telecommunications etc. A number of authors deal with this topic in detail, see e.g. Mareš, Musil [1995] and Možný [1994]. 6 For more on the role of myths and counter-myths in the transformation process in the Czech Republic after 1989 see Kabele [1998].

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

3

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The establishment of a functioning labour market in the Czech Republic after 1989 thus represents a significant interference in the life of the whole society, individual social classes and strata, actual households and families as well as individuals. In terms of the topic addressed in this study the following points are most important: 1. The space for internal markets that dominated the economy under the previous 7 regime shrank. After five decades of existence on the primary labour market where they were elevated – with the help of wage levelling and improved status of unskilled labour – and protected from poverty under the previous regime numerous social categories returned to the secondary labour market. This did not involve only individuals and categories of professionals (in many cases it involved a significant drop in income and living standards as well as a drop in status) but also many fields and regions which previously benefited from the concentration of privileged workforce and huge state investments in transport, housing and social 8 infrastructures. 2. Together with the enlargement of the secondary labour market a relatively strong shadow economy came into being as a domain of self-employed people, small entrepreneurs and companies that attempted to evade taxes. It involves many very unstable jobs and due to problems with balance of payments in a portion of companies also social risks (there is a danger that employees’ health and social insurance contributions will not be paid by the employer, delayed wages, contract termination due to enforced agreement etc. ). 3. A portion of the workforce has been marginalized, this applies mainly to unskilled labourers and some ethnic groups, in particular the Roma (these tend to be unskilled as well). The marginalization is reinforced by discrimination. This applies not only in the case of the Roma but also the elderly, women – in particular with young children and disabled people [Sirovátka, 1997a, 1997b]. In this respect foreign workers without work permits represent a separate case. The number of illegal employees from other countries (the most numerous group being Ukrainians) on the Czech labour market is estimated to be twice the number of foreigner workers with a work permit. 4. Marginalized territories have come into being where a portion of the population is stuck following the destruction of original industrial fields, e.g. following the decrease in coal mining and steel manufacture in some regions. We often encounter the “marginalization of the marginalized” as these are regions where social classes 7

These worked on the principle of hierarchy and loyalty, a part of the state’s social policy was also realized in these and as already mentioned they also served purposes of social control over the population exercised by the political power. These internal labour markets of state-owned companies shaped the hierarchy of the employees according the principles of seniority and loyalty. The companies performed with respect to their employees part of state social policy measures as well as the social control of the population by the totalitarian political regime. The state was de facto the only employer and thus it was capable to enforce people to follow its will, threatening them that they might loose the (good) job. 8 It is not surprising that in an environment (defined socially as well as territorially) thus affected there is a continued sense of anomy and nostalgia related to the certainties of the past [Rabušic, Mareš, 1996] combined with support for left-wing parties, including support for the untransformed communist party which in the Czceh Republic regularly gets 10-16 % of the vote (in affected regions it has even 20-30% support).

4

Tomáš Sirovátka and Petr Mareš with low or significantly specialized human capital dominate. While in this respect prior to 1996 the most affected areas were the little industrialized districts characterized by re-structuralization of ownership in agriculture (decline of big state and co-operative farms) after this year what prevails is the marginalization of districts characterized by mining and industries typical of the 19th century which are in addition also characterized by long-term social problems such as high degrees of migration, high divorce and crime rates etc.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

LABOUR MARKET AND NEW POVERTY Although in the last five years the general unemployment rate in the Czech Republic has not exceeded the European Union average, the risk of unemployment is unequally distributed in the society. Above all education significantly influences opportunities on the labour market. The high rate of (long-term) unemployment persists in the group of people with primary education while among individuals with university education the unemployment rate is very low. Inequality in employment opportunities has been constantly increasing since 1990 [Vyhlídal, Mareš 2006]. We should not disregard the influence of ethnicity either although in the Czech Republic we can only predict the relationship between membership of the Roma ethnic group and (long-term) unemployment. Labour Offices do not keep track of the ethnicity of the registered unemployed. The Czech labour market is also significantly structured in terms of gender, age and region. Depending on these characteristics the workforce has different appeal to the employers, for example, women have a different significance on the labour market from men, young people are perceived in a different way than middle-aged people or the elderly, education has significantly different weight in terms of employment opportunities and the regions strongly influence the shape of the (local) labour market [Katrňák and Mareš 2007]. The high portion of long-term unemployed in the total number of unemployed indicates that for a portion of the unemployed life without paid employment has become a well-known experience. Research on the long-term unemployed [Mareš, Sirovátka and Vyhlídal 2003] found that – if we exclude individuals who have not yet set out on a career path – two thirds of respondents experienced more than one actual loss of a workplace. In fact one third of them have experienced at least the third instance of unemployment in the course of their 9 careers. Only 12% of the respondents worked in their last job for longer than two years and half of the respondents lost their previous jobs within 12 weeks of signing a contract [Mareš, Sirovátka and Vyhlídal 2003]. Thus we can identify the contours of a group of marginalized people whose career is fragmented and the total period of unemployment in their careers is not negligible (at the time of the current instance of unemployment 35% of the respondents experienced a total of more than 1 year of unemployment in the course of their careers,

9

We have to take into account that it concerns a period of roughly 10 years when for these individuals unemployment is theoretically possible and ca 5 years when the likelihood of losing their jobs is the same as in the highly developed countries of the European Union.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

5

actually 15% of respondents experienced more than 2 years of unemployment in total).10 We identify a higher share of unemployment in their extended families as well. Scheme: Typology of households affected by long-term unemployment [Mareš, Sirovátka and Vyhlídal 2003]. BOTH PARTNERS LONG-TERM UNEMPLOYED

ONE OF THE

PARTNERS LONG-TERM

IN TIME OF SURVEY

UNEMPLOYED IN TIME OF SURVEY

The respondent has lost a job already more than once and at the same time his/her partner lacks a permanent job

The respondent has lost a job already more than once and at the same time his/her partner has a job

11% The respondent has already at least once lost a job and at the same time his/her partner lacks a permanent job

30% The respondent has already at least once lost a job and at the same time his/her partner has a permanent job

11% The respondent lost a job for the first time and his/her partner lacks a permanent job

17% The respondent lost a job for the first time and his/her partner has a permanent job

6%

25%

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: The shares are calculated from households for which the question is relevant (the respondents have a partner and the lost job was not their first one).

Almost 30% of the households of those respondents for whom the question had relevance were affected by double unemployment. Half of the cases actually involve individuals experiencing their third or further instance of unemployment. These data demonstrate that in the Czech society “new poverty” has emerged which is linked to the labour market and is characterized by a number of attributes of social exclusion: long-term and repeatedly unemployed individuals and their households and households in which unemployment cumulates – a number of their members are unemployed. In the Czech case the relationship between poverty and unemployment is particularly significant as the occurrence of income poverty among those living in unemployed households is significantly higher than in other EU member states [OECD 2004]. Poverty accompanies in particular long-term or periodically repeated unemployment, in the majority of cases these overlap in certain types of households 11 [Sirovátka, Mareš 2005]. Each of them and in particular their cumulation are conditioned by the same socio-demographic characteristics or structural factors. Long-term unemployment is influenced above all by low levels of education or the complete lack of qualification, insufficient work experience, health conditions that limit active participation in the labour 10

Gallie [1994] points out that the long-term unemployed people include mostly those stuck in the secondary labour market who are deprived and demoralized by repeated job losses and prolonged periods of unemployment. 11 At the moment in the Czech Republic more than half of the unemployed are long-term unemployed. Their unemployment has lasted for more than a year.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

6

Tomáš Sirovátka and Petr Mareš

market, low age or ethnic origin (very often a combination of these factors is involved). Periodically repeated unemployment is marked mainly by the nature of local labour markets which do not offer sufficient employment opportunities while individuals affected by unemployment are linked to these labour markets in the majority of cases. In general “new poverty” and “social exclusion” are mainly linked to the undergoing qualitative changes on labour markets, these are considered to be consequence of processes linked to the transformation to knowledge society, de-industrialization and the global extension of economic competition [cf. in particular Bauman 1998; Beck 2000; Schmidt, Gazier 2002 etc.]. Those who cannot adapt to these changes sufficiently are marginalized on the labour market and in a society of “producers” and “consumers” lose their economic function as well as their social status. Arguments about the roots of marginalization on the labour market tend to develop along two lines, we can understand them as complementary although at the ideological-political level they are contradictory. One theoretical line is based on the individualization thesis and understands marginalization on the labour market as a failure and consequence of individual deficit of human capital whether in the sense of skills or motivation. In contrast another theoretical line works with the presumption that objective disadvantages cumulate in individual cases and explains marginalization on the labour market within the context of the cycle of deprivation and social exclusion [Whelan, Layte 2003]. Some theories, however, also deal with the failure of mechanisms of the labour market – i.e. discrimination, segmentation and the duality of the labour market [Berger, Piore 1980; Offe 1985; Lindbeck, Snower 1989 etc.]. In relation to the above outlined two theoretical stances on the roots of marginalization on the labour market two “remedial” strategies have been identified: the first one is mainly linked to nominal and wage flexibility. It attempts to solve the problem of marginalization on the labour market and to increase employment rates with the help of highly flexible wages and increased adaptation of the workforce to market indications. The other one is a strategy of functional flexibility which attempts to identify a longer term solution in the continual improvement of the quality of human capital and skills and also in adapting work regimes and the organization of work – cf. in particular Standing 1999; Lødemel, Trickey 2001; Schmidt, Gazier 2002; van Berkel, Møller 2002; Serrano Pascual 2004. The European Employment Strategy [European Commission 2003] works with 12 both approaches under the presumption that the roots and forms of marginalization on the labour market are relatively complex. That is why it stresses the internal relationship among three goals: employment (and employability), quality of work (and productivity) and social inclusion. Readiness, motivation and the ability to gain and maintain meaningful (thus also high-quality) employment is a key and a pre-condition to achieving the three outlined goals, similarly to the real achievability of full employment. Apart from the reasons behind the processes of marginalization also the consequences of marginalization on the labour market are analyzed, in particular the consequences of longterm unemployment in relation to poverty, material deprivation and social exclusion [cf. Paugam 1995; Gallie 1999; Standing 1999; Heery, Salmon 2000; Gallie, Paugam, Jacobs 2002]. Similarly, attention has centred on the link between the consequences of 12

The strategy pays attention to a wide spectrum of policies – active measures on the labour market, the education system including lifelong education, benefit and tax systems, anti-discrimination measures, social counselling and services etc.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

7

marginalization on the labour market and social institutions that protect the marginalized workforce against deprivation, i.e. the institutions of family and the social state [see particularly Gallie, Paugam 2000; Saraceno 2002]. Attention is paid not only to benefit systems as well as social services as tools protecting against poverty and deprivation but also 13 to strategies of activation, support for employability and inclusion on the labour market. Some authors [Esping-Andersen 1990, 1999; Paugam, Gallie 2000; Esping-Andersen et al. 2002; Schmidt, Gazier 2002] argue that the analysis of these institutional measures can lead to the identification of specific “labour market regimes”. The liberal labour market regime individualizes the unemployment problem, it prefers strategies of wage flexibility, deregulation of the labour market and low welfare benefits. In contrast the universalist regime is based on the hypothesis of cumulating disadvantages and prefers the strategy of so-called functional flexibility of the workforce, a typical form of providing re-qualification opportunities while maintaining a satisfactory level of welfare benefits. The corporate – conservative – regime polarizes the risk of unemployment (it protects mainly and effectively the workplaces or at least the wages of regular workforce – typically male breadwinners) and thus it deepens the gap between “insiders” and “outsiders”. When regulating the labour market it tends to rely on decreasing employment offers (in the form of early retirement, long breaks from work while caring for children) rather than on supporting the functional flexibility of the labour market. In general the liberal regime is expected to result in significant deprivation of a large segment of the workforce – working as well as non-working poor – and that in terms of wages, job quality as well as other areas of their lives. Although the income status of the poor should be better than that of the unemployed (welfare benefits are low), as a consequence of the stress on benefit targeting using means testing there might be poverty traps emerging in the case of low-income groups, also a portion of them do not take up benefits. The universalist regime is expected to yield generally low levels of income deprivation as well as access to employment, relatively decent job quality (in terms of the level and security of income, job security, promotion etc.) and small differences in all areas of deprivation as a consequence of decent income support as well as a high intensity of measures supporting functional flexibility. In a corporate regime we can expect significant differences in the degree of deprivation, those affected by disadvantages in wages (and welfare benefits) and in job quality, access to employment or participation in measures supporting functional flexibility will be mainly so-called “outsiders” on the labour market (such as less qualified, younger workforce, women with children). The impact of marginalization on the labour market on material deprivation and social exclusion has been under-researched in the Czech Republic. The relatively high share of longterm unemployment and the relatively high specific unemployment rates of some groups such as the unskilled, young people, women-mothers, self-employed women with children etc. [Mareš, Sirovátka 2005] are widely recognized yet the Czech general unemployment rate – when compared with other transforming post-communist countries – is among the lowest and in a pan-European comparison it is among the average ones. Similarly, until now analyses embedded in a variety of data sources [Sirovátka, Trbola 2003; Večerník 2004; Sirovátka et al. 2005] show that in the Czech Republic the extent of income poverty is not too high, it 13

Here activation strategy is understood as a dynamic relationship among the complex of various policies and employment [Lødemel, Trickey 2001; van Berkel, Møller 2002; Serrano Pascual 2004 and others].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

8

Tomáš Sirovátka and Petr Mareš

affects at most 7–8 % of the population (which is the lowest among all European countries), similarly about 6–7 % of households are on welfare benefits in order to guarantee the subsistence minimum. The majority (about three thirds) of benefit recipients are unemployed people mainly as a result of the relatively high share of long-term unemployed among them. The social system appears to be relatively effective in terms of eliminating the risk of poverty mainly thanks to the highly targeted nature of welfare benefits towards groups of people with low incomes, however, this does not by far apply to all groups, e.g. it does not apply to the unemployed and single-parent families with children [Sainsbury, Morissens 2002; Sirovátka et al. 2005]. It is also acknowledged that despite different risks of poverty for the unemployed (exceptionally high) and the employed (exceptionally low), among the poor – measured according to Eurostat methodology – the share of individuals (including children) living in households with an adult wage-earner and the share of those living in households with no adult wage-earners is approximately equal [Sirovátka et al. 2005]. Differences in the extent of deprivation of the poor have been researched only partially, cf. Večerník 2004. Some analyses have already pointed out the strong deprivation of a significant share of the unemployed (e.g. breadwinners) while other groups of the unemployed did not suffer from deprivation [Mareš, Sirovátka 2005]. At the same time other analyses also demonstrate that the social system does not create appropriate employment incentives at least for part of the workforce [Sirovátka, Žižlavský 2003; Večerník 2004] and also that active measures aimed at the inclusion of marginalized workforce on the labour market are not implemented sufficiently in the case of those who are more disadvantaged than others [Sirovátka et al. 2004]. Analyses have also pointed out that thus far the Czech Republic has not experienced the problem of large numbers of working poor. This, however, does not have to be valid when income and material deprivation is taken into account as its extent in the Czech Republic is considerably higher, various indicators suggest that its degree is more than double compared to income poverty [Večerník 2004; Sirovátka, Mareš 2005]. The relationships among unemployment, material deprivation and social exclusion as well as the influence of public policies on these relationships thus in some respects appear to be contradictory and they deserve further exploration. This study aims to clarify the relationship between the marginalized status on the labour market on the one hand and income disadvantage, material deprivation and social exclusion on the other. We thus understand and analyze marginalization on the labour market in the sense of repeated and long-term unemployment as well as restricted access to employment, respectively sole access to jobs that are characterized by high instability and “low quality”. We explore material deprivation and social exclusion as multidimensional phenomena. We also explore policies aimed at improving the labour market status, income and material conditions of disadvantaged people. We attempt to analyze inter-relations among marginalization on the labour market and/or material deprivation and social exclusion. One of our concerns is to identify coping strategies of those who are at risk of marginalization on the labour market and/or material deprivation and social exclusion.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

9

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

DATA The study works with data gathered in the survey “Social exclusion and social policy” (“Sociální exkluze a sociální politika”) conducted under the auspices of the Faculty of Social Studies, Masaryk University (empirical data collection was conducted by the Focus agency at the end of 2004 and beginning of 2005). The sample consisted of individuals demonstrating signs of income disadvantage, characteristics of their households were also taken into account. The sample consisted of 2,500 individuals who either I) stated that due to insufficient income they were entitled to welfare benefits in the course of 2004 or II) stated that in 2004 they considered applying for welfare benefits as subjectively they saw their situation as comparable (i.e. equally difficult) to that of benefit recipients. The latter category of respondents made up about a third of the sample. The reason why we included a subsample of respondents who do not demonstrate “objective” signs of income disadvantage (i.e. repeated receipt of income support) but rather a “subjective” indication of income disadvantage (they consider their situation comparable to that of benefit recipients) is that there are signs that also in the Czech Republic (apart from over-use) also the under-use of benefits occurs – the non-take-up of welfare benefits to which potential recipients are entitled and the extent of this non-take-up cannot be neglected [Mareš 2001]. Thus restricting ourselves only to benefit recipients would involve a significant oversight. During the survey the presumption that there is a category of the poor who are entitled to welfare benefits but do not take them up was confirmed, for example, the average per capita income (using the socalled equivalence scale at the calculation) in the category of benefit recipients was essentially comparable to the average income in the category described as “subjectively” income disadvantaged (4,700 CZK as opposed to 4,830 CZK). Our selection was informed by the national statistics of the Ministry of Labour and Social Affairs on the structure of benefit recipients: quotas were set up according to region, gender, age and type of household in which the recipients live. Quota selection was conducted in two phases: first according to region so that the respondents’ share of regions would reflect the structure of the population (this was also of practical value in terms of access to respondents). In the second phase the quotas were filled according to gender, age and the type of household while we approached and subsequently interviewed income disadvantaged individuals both according to the “objective” and the “subjective” indicators (in this case following a similar selection procedure, presuming that with increased numbers of benefit recipients the number of those who only consider applying for benefits also grows). Thus it involves a intentional quota selection which aimed to sufficiently represent the main types of respondents according to gender, age and type of household in which they live and it reflects the structure of the population only approximately. The quotas were based on an analysis of the structure of social benefit recipients [Sirovátka et al. 2005] in order to capture the main “types” of income disadvantaged people as they were identified in the analysis, i.e. ca 30 % of people aged under 25, ca 50 % of people aged 25 to 45 and ca 20 % aged over 45, approximately the same ratio (50%:50%) of men and women and further a similar ratio of respondents living in childless households, nuclear family households and single parent households (about a third of each type of household). In comparison with the population the set quotas in essence respect the age structure of the income disadvantaged but in order to fill the numbers of respondents in the compared types of households the share of respondents in nuclear family

10

Tomáš Sirovátka and Petr Mareš

households (actually among the income deprived they only make up less than 20% of the group) is over-represented and in contrast the share of single adult households is overrepresented (as a matter of fact they make up more than half of benefit recipients). Students and recipients of old age and disability pensions made up 10% of the sample, we excluded them from the analysis presented in this paper as in this case we are concerned above all with the impact of forms of marginalization on the labour market on material deprivation and social exclusion. Thus we analyzed a sample of 2,225 respondents who are more or less accessible as stable workforce for the labour market. It follows from the above that we do not consider the sample statistically representative of benefit recipients, nor do we consider it representative of income deprived people or the poor in the Czech Republic. A statistically representative sample of such a specific subpopulation which would be numerous enough is difficult to construct because we do not 14 know the extent of the non-take-up of welfare benefits. The sample that we had at our disposal, however, includes basic “types” represented among income disadvantaged people in terms of the socio-demographic structure. Moreover, the sub-sample of respondents selected according to “subjective” indicators of income disadvantage allows us to take into account the highly likely occurrence of non-take-up of benefits entitled individuals, thus we could avoid the sampling error of type I to which Halleröd [Halleröd 1995] alerts, i.e. the error of relying on the purely “objective” criteria of poverty. A wider spectrum of types of income disadvantaged individuals with a sufficiently big sample thus allows us to identify and compare various demonstrations of marginalization on the labour market, material deprivation and social exclusion and the exploration of relationships among these phenomena. We devote particular attention to the comparison of the status of employed people with permanent jobs, short-term unemployed ones and long-term or repeatedly unemployed people and the impact of the cumulation of unemployment in respondents’ households on their material deprivation and social exclusion.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

A NOTE ON METHODOLOGY Apart from usually used indicators of marginalization on the labour market such as repeated and long-term unemployment (lasting for more than 12 months) we also work with the concept of “job quality” (characterized by evaluation of its stability, wage, promotion opportunities, training opportunities, physical demands, necessity to commute long distances). When exploring material deprivation and social exclusion we work with the neutral term “income disadvantage” which (as we already mentioned) uses objective as well as subjective indicators of income disadvantage. We then deal with the relationship among income and material deprivation and social exclusion. When analyzing income we deal with declared income in the respondents’ households which is then calculated per capita using the Eurostat equivalence scale [2000]: i.e. a respondent of weight 1.0, another adult weight 0.5, 15 children weight 0.3. We analyze material deprivation as a multidimensional phenomenon 14

Perhaps the only way of acquiring a bigger representative (sub)sample of people with lowest incomes are widespread and expensive representative surveys of the Microsensus type. 15 In our sample these are dependent children aged under 18.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

11

while we explore its various components: income deprivation, deprivation of basic needs (food, clothing, holidays), deprivation in household equipment and housing deprivation. Social exclusion also involves – apart from material deprivation – deprivation in health and social contacts. These indicators are often used when exploring the extent and structure of poverty, the nature of material deprivation and social exclusion and experts deem them relevant for international comparison [Eurostat 2000]. As a result they were used to some extent also in the survey conducted by the Czech Statistical Office entitled “Social conditions in households” in 2001 and they continue to be used in surveys coordinated by Eurostat (“Statistics on Income and Living Conditions” respectively “EU Survey on Income and Living Conditions”). In addition to basic indicators of deprivation which are introduced by Eurostat to explore poverty and social exclusion we also worked with other indicators of deprivation which we understand as “supplementary” and that in the sense that they exceed basic needs and material deprivation. Yet they are important in that they indicate access to life opportunities, capabilities [Sen 1992], that is possibilities to function in a certain social structure. In concrete terms in this case it involves the possibility to change one’s living conditions, to influence one’s personal development, one’s future as well as the future of one’s children. It is exactly these circumstances that correspond to the widely used definition of social exclusion as deprivation in access to institutions that influence life opportunities, possibilities to share in majority living standards and participate in various areas of social life [Room 1995; Atkinson 1998; Atkinson et al. 2002]. In concrete terms a chance to acquire a mortgage, to pay health insurance and accident cover, supplementary pension insurance, the ability to provide for children’s education, to satisfy one’s cultural interests etc. We assessed the cumulative extent of material deprivation and social exclusion using aggregate indicators: we used cumulative index I that includes 23 items of material deprivation we worked with, index II with only 13 items that were introduced by Eurostat as internationally comparable [2000] and index III, which again has 23 items, however, these are first compiled into a sub-index for each area of deprivation and then the sub-indices create the overall index of deprivation (thus individual areas weigh approximately the same while in the case of index I the value is influenced by - apart from others - the number of items included for individual areas of deprivation). And finally index IV was created for the narrower circle of 12 selected items that had the best results in the reliability test. In fact it was this index that has included also “supplementary” items (yet in terms of the concept of social exclusion these are significant indicators) and also included items in the area of material deprivation that are traditionally considered of key importance. The value of cumulative indices expresses the percentage of items in which the respondent is significantly (very) deprived from the total number of items in the index (thus the range of values is from 0 to 100). For an overview of items and reliability test see appendix.

12

Tomáš Sirovátka and Petr Mareš

MARGINALIZATION ON THE LABOUR MARKET, DEPRIVATION AND SOCIAL EXCLUSION Forms of Marginalization on the Labour Market Income disadvantaged people belong mainly to the marginal (disadvantaged) workforce in more respects: they have lower qualifications, are more frequently unemployed, are more frequently temporarily employed. For example, in this group the share of those with primary education when compared to their share in the workforce is double. In contrast, the share of Table 1. Structure of income disadvantaged people by labour market status (in %)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Permanent job

Temporary Unemployed: job short-term

Total (N) Of which: Man Woman

38.6 (859)

15.9 (353)

8.4 (187)

Unempl.: repeatedly or longterm 27.7 (617)

50.9 49.1

46.6 53.4

31.4 68.6

50.2 49.8

8.3 91.7

44.4 55.6

– 29 years 30 – 44 years 45 and above

28.4 47.7 23.9

38.2 40.2 21.5

45.7 31.4 22.9

41.6 38.6 21.8

59.8 34.0 6.2

38.0 40.8 21.2

Primary education Secondary vocational education Secondary education University education

7.5 52.4

13.0 59.2

21.3 45.7

24.5 50.5

29.7 46.4

16.3 51.8

31.8 8.4

22.1 5.7

28.2 4.8

21.8 3.2

22.5 1.4

26.3 5.6

Nuclear family Single-parent family Childless couple Single adult

43.3 30.3 15.9 10.6

32.3 33.7 18.1 15.9

27.8 42.8 18.7 10.7

24.8 27.4 28.2 19.6

46.4 48.3 2.9 2.4

35.4 32.7 18.7 13.2

Health problems restrict employment possibilities

6.6

8.0

27.6

16.8

9.2

11.6

Declared Roma origin

4.7

4.0

7.4

10.5

18.7

7.7

Inactive

Total

9.4 (209)

100 (2.225)

university educated people is less than half while the share of people with secondary vocational education is slightly higher in contrast with the slightly lower share of people with 16 secondary education. The share of individuals with permanent jobs among the income 16

Secondary vocational education refers to secondary education without maturita (a type of school leaving exam) and secondary education refers to secondary education with maturita.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

13

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

disadvantaged when compared to the general workforce is approximately one half, the share of those with temporary jobs more than double and the share of the unemployed is ca. fivefold. One of the key characteristics of the income disadvantaged people is some form of their marginalization on the labour market. In our sample among the income disadvantaged there is 17 a slight dominance of currently employed individuals over unemployed and inactive ones. However, the hypothesis about the marginal status of the income disadvantaged on the labour market is also supported by other indicators: [a] The type of current or last job: only in 56% of the cases did it involve a permanent job, in 7% self-employment, however, in 31% of the cases the job was a fixed-term one, in 1% of the cases public work and in 5% of the cases casual work without a contract. [b] Repeated occurrence of unemployment: not only that more than a third of the individuals are unemployed, the majority of the currently unemployed were longterm or repeatedly unemployed. Repeated unemployment is a more frequent characteristic than long-term unemployment: only a third of the currently unemployed are long-term unemployed (i.e. for more than a year) while half of the sample are repeatedly unemployed. However, the main point is that the share of the long-term and/or repeatedly unemployed in the sample of income disadvantaged people is more than threefold the share of short-term and first-time unemployed. [c] Hidden unemployment: in the given cases the unemployed are prevented from access to benefits and active programmes of employment policy: although only 3 % of the unemployed are not or have never been registered at a Labour Office we identify more than 8 % who were removed from such registration by the Office and thus there are about 11 % of unregistered unemployed. Apart from that we identify also 12 % of unemployed who can be labelled discouraged (they would accept a job offer but they are not actively seeking one) and that mostly because they do not believe that they would find a job or due to personal or family circumstances. And finally only 2 % of the unemployed state that they would rather not want paid work. [d] Occurrence of cumulated unemployment in the households of income disadvantaged people. If we consider the status of households with two partners on the labour market we find that a quarter of them have a current or past experience of parallel unemployment. Almost a half (48 %) of income disadvantaged people live in a household in which currently one member is employed, however, 38 % live in households where both partners are employed; only in about 14 % of the cases are there partners who are both currently unemployed (in the large majority of cases as a result of unemployment and sometimes also due to economic inactivity). The 17

However, in this respect the sample (as we already noted) cannot be understood as representative. For example, according to in this respect the sample (as we already noted) cannot be understood as representative. For example, according to the Microsensus of 2002 among the people with incomes below the poverty line (according to the Eurostat definition) 32 % are employed, 47 % unemployed and 21 % inactive, however, this does not include pensioners and students (according to a special analysis conducted by the Czech Statistical Office for this project). In our sample thus there are fewer unemployed and inactive people and in contrast more employed, nonetheless we can compare the groups of employed and unemployed well and this is indeed our analytical aim.

14

Tomáš Sirovátka and Petr Mareš majority of income disadvantaged women – whether they are employed or not – however find support in an employed husband (ca 75 %) but only about 50 % of unemployed men have wives in employment (contingency coefficient = .390, sign.= .000).

If we follow the OECD typology on the status of members of a household on the labour market then the majority is made up of respondents whose households are either “fully employed” (both partners are wage-earners or it is a household with one adult who is employed) or at least “partly employed” (one of the partners is a wage-earner) as compared to “non-employed” ones (none of the members of the household is in employment). This is particularly true of respondents who live in a nuclear family household (in this case over 30 % of households are fully employed and 50 % partly employed, contingency coefficient = 0.475, sign.= .000). Table 2. Respondents by the status of their households on the labour market and by type of household (in %)

Nuclear family Single-parent family Childless couple Single adult Total

Fully employed hshds Partly employed hshds Non-employed hshds Total 32.7 49.3 18.0 100 51.6 – 48.1 100 31.3 50.5 40.9

25.0 – 22.2

43.8 49.5 36.8

100 100 100

Notes: classification of households according to the OECD – employed households are those in which one (in the case of single adult households) or two (in other cases) members are wage-earners, partly employed households are those in which at least one of more adults acts as a wage-earner and in non-employed households none of their adult members act as wage-earners.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Job Quality Since among the income disadvantaged people we find a relatively high share of currently employed individuals it is relevant to deal – apart from the stability of a job – also with job quality. Job quality, as we have already pointed out, is exactly one of the characteristics linked to the stability of employment and in more general to possibilities of inclusion on the labour market and in the society at large. That is why “job quality” has been since 2003 together with full employment and social inclusion one of the key goals of the European Union’s employment strategy. Indicators of the job quality confirm that whether currently employed or not income 18 disadvantaged individuals move mainly on the secondary labour market. This then 18

It is characterized – as opposed to the primary labour market – by employment offers that typically involve low demands on qualifications, employment uncertainty, non-standard forms of employment such as temporary contracts or the absence of contracts, forced part-time work, bad work conditions and low wages (also because this segment of the labour market is – in contrast to the primary labour market where trade union negotiations

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

15

Table 3. The share of (very and rather) negative evaluations of selected characteristics of the quality of a job (in %) by the respondent’s labour market status Permanent employment

Temporary employment

Short-term unemployed

Total

coef. Eta (sign.=.000)

49.5 45.4

Repeated and longterm unemployed 66.4 49.7

Lacking stability Lacking a decent wage Lacking promotion opportunities Lacking training opportunities Too tiring Long commute

30.7 41.2

53.1 66.8

45.1 48.8

.308 .162

63.0

77.3

45.4

49.7

67.0

.186

56.2

75.2

59.6

68.4

64.1

.235

53.9 38.9

47.5 35.9

45.2 38.0

55.2 34.0

51.2 37.0

.106 .079

Overall job quality index (scale 1–5)

3.36

3.67

3.49

3.64

3.50

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: The question was “Does your job provide you with opportunities?”. The unemployed evaluated their last job. The table includes the responses of (definitely) yes. Coef. ETA – relationship of the given variable to the variable “status on the labour market”.

means that for the majority of them their jobs do not provide promotion and training opportunities (this applies to ca two thirds of them); for about a half of them the job is physically too tiring, almost a half of them do not get a decent wage, it does not provide stability and in roughly 40% of the cases it also demands sacrifices in the form of long commutes (this, however, involves additional expenses and unpaid time). The greater stability of the current/past employment is more often mentioned by people who have a current stable job and in contrast least often in the case of currently unemployed and that repeatedly or longterm (Eta = 308, sign. = .000). It is surprising that in the case of people with temporary jobs we find the worst indicators of job quality and that as far as wages are concerned as well as promotion and training opportunities – and that in comparison with the currently unemployed, even in comparison with repeatedly and long-term unemployed (in the case of currently unemployed their previous job is concerned). In addition the evaluation of income and promotion and training opportunities made by those in stable employment does not significantly differ from the evaluation made by the currently unemployed. There is no difference either in the evaluation of demands and commuting and the evaluation of physical exhaustion connected with a job. Thus it appears that the stability of a job is the only characteristic that distinguishes the quality of the (last) job according to the current status on the labour market. The certainty which is the highest in the case of those with a “permanent” job (with an indefinite contract) on the other hand tends to be connected with the low quality of the job in other respects: low wages, limited promotion and training opportunities, exhaustion connected with work, commuting. Such characteristics of employment evaluated by those in employment are comparable with qualities of work as understood by unemployed individuals, even by the play a key role – fully competitive). Employees are not significantly protected in collective agreements or employment legislation, fluctuation is high, promotion and training opportunities are insignificant, for more detail see [Sirovátka 1997a: 19–27].

16

Tomáš Sirovátka and Petr Mareš

repeatedly and long-term unemployed ones. Individuals in temporary employment surprisingly provide the worst characteristics of their jobs. In the context of these findings we can deduct various strategies on the part of individuals belonging to the marginalized and income disadvantaged workforce. As we see employment that is positively defined in terms of qualities set by us is for a great part of them very likely out of reach and the marginal income disadvantaged workforce under these circumstances splits into further categories: one, smaller, part maintains lower quality but more stable jobs and they probably mainly value the stability and certainty that such jobs provide. Another, dominant, part alternates between unemployment and temporary or less stable jobs which are again of lower quality also in other characteristics – they can be driven to these partly due to job offers (an urgent need for income from employment) as well as the hope that in due time they will acquire more stable and eventually better paid jobs. Finally, one part of the workforce remains unemployed more permanently partly because they sense a slight chance of gaining a more stable and higher quality job and partly because they evaluate the low incomes from employment that is currently on offer (this mainly concerns low wages).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Income Disadvantage A large part of the income disadvantaged individuals and households involve employed people. In general it could be expected that in this category of income disadvantaged the motivation to search for and accept a job will be high and that employment allows them to get at least somewhat higher wages than unemployed or inactive people. Thus if we start with the wages declared by respondents they significantly drop – in comparison with permanently employed individuals (the average monthly income per person in the household that they live in when using an equivalence scale is 6,200 CZK) – in particular in the case of economically inactive individuals (the average monthly income per person based on the equivalence scale is in their households 5,050 CZK) thus the income of the inactive ones is about a fifth lower than that of the stably employed. It is, however, somewhat surprising that the income of the unemployed (average monthly income per person in their households is according to the equivalence scale 5,950 CZK) is not very different from that of the stably employed and that the situation of the temporarily employed individuals is even worse than the situation of the unemployed (average monthly income per person in their households is according to the equivalence scale 5,600 CZK, i.e. their incomes are 10 % lower than those of stably employed while in the case of the unemployed the difference is negligible). When exploring the per capita income in respondents’ households in relation to the status of all members of their households on the labour market (OECD typology) we can point out similarly that the income situation of those respondents who live in non-employed households is actually better than the situation of respondents living in partly employed households. Per capita incomes in non-employed households are less than 10% lower than those in the fully employed income deprived households.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

17

Table 4. Average monthly income per member of the household in which the respondent lives Employment in household Fully employed Partly employed Non-employed

Average Stand. deviation Average Stand. deviation Average Stand. deviation F (sign. .000) Eta

Per person in CZK per month 5060 2456 4320 1484 4673 2722 12,645 .134

Per person

(Eurostat) 6123 2698 5693 1977 5754 3053

N 603 402 385

4,032 .076

Note: 1,390 respondents of the 2,290 declared their incomes. Per person calculation according to the equivalence scale: respondent weight 1.0, another adult person weight 0.5, children weight 0.3.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Status on the labour market and unemployment, even cumulated unemployment, thus do not play such a significant role in determining the income levels of income deprived respondents and their households. We can actually say that partial employment or temporary jobs do not pay off for income deprived respondents and their households. In comparison the respondents’ income situation differs significantly according to the type of family in which they live. Those who live on their own or as childless couples (with the per capita income of 5,150 CZK) and in particular one-member households (per capita income of 7,100 – 8,400 CZK) have in this comparison higher incomes by a third or more – calculated per person in the first case with the same weight for all members or with the use of the equivalence scale in the second case – than those who live in nuclear family households (4,450 – 5,700 CZK) and single-parent families (4,350 CZK – 5,250 CZK), coef. ETA = .426 (F = 102,663, sign = .000). If we are guided by the poverty line established in the Eurostat convention (60 % of median per capita income according to the equivalence scale) and based on data from the Microsensus of 2002 of 6,156 CZK per capita (according to the equivalence scale) we can see the average incomes of our respondents and their households – whether with children or without them – they are still about 10 % under this line, however, the incomes of respondents’ households without children are significantly higher than the poverty line.

Material and Social Deprivation Because incomes are only an indirect indication of poverty and material deprivation and moreover they do not always necessarily indicate social exclusion in terms of access to social 19 contacts, institutions and life opportunities , a number of authors prefer the measurement of subjective material deprivation [Nolan, Wheelan; Nolan 1996; Barnes 2002; Wheelan, Layte, Bertrand 2002]. Similarly Eurostat also introduced indicators of poverty measurement and 19

In fact these are the common defining elements of the concept of social exclusion cf. Room [1995]; Atkinson et al. [2002].

18

Tomáš Sirovátka and Petr Mareš

social exclusion that are based on subjective assessment of deprivation in various areas of life [Eurostat 2000]. In comparison with the findings on the whole Czech population in the survey “Social conditions in households” (further SCH) of 2001 (see columns in Table 5) which took into account the Eurostat methodology, according to our data there is relatively significant deprivation of income disadvantaged individuals in a number of areas. A quarter of them have smaller or bigger problems in coping financially and similarly a half of them cannot pay their bills and rent on time (compared to 17 %, resp. 13 % in the whole population according to SCH). It is thus not surprising that in the area of basic needs – our indicators involve food, clothing, holidays – we find a significant degree of deprivation when almost half of the respondents in our sample cannot afford to eat meat, fish or poultry every other day, about 60 Table 5. Selected indicators of material deprivation and social exclusion by the status of members of the respondent’s household on the labour market (in %)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Income ŠCH 2001 Fully Partly Non-empl. CC (individuals) deprived empl. empl. household (sig.000) total househld househld Financial problems Very difficult to cope Problems with paying bills* Deprivation in basic needs Eat meat/fish/poultry every other day Buy new clothes Have at least a week’s holiday away from home Deprivation in household equipment Telephone Colour TV set Car Housing deprivation No bathroom or shower Lack of space Dampness (ceiling, floor, walls...) Health problems (very) poor health/experiences health problems Social deprivation Meets friends less than once a month

16.6 12.8

24.9 50.2

15.4 43.9

20.8 47.3

37.9 58.8

.229 .134

41.1 47.7 49.7

47.7 59.8 73.0

47.0 50.2 62.9

50.1 56.4 70.7

59.4 72.4 85.5

.111 .199 .221

5.8 1.1 13.8

12.1 3.8 47.5

7.7 1.6 43.5

6.3 1.0 38.8

19.4 7.2 59.7

.180 .149 .173

1.4 22.0 12.6

1.7 34.3 22.1

0.5 29.9 17.6

1.4 37.4 23.6

3.0 37.4 26.2

.086 .078 .094

12.5

35.8

35.2

32.5

38.5

nv

24.9

17.5

19.4

18.0

15.1

nv

Note: * in the last 12 months (rent and energy); CC – association between the given item and the respondent’s labour market status. Source: Czech Statistical Office – Survey on Social Conditions in Households, data 2001 (n = 27,000). Survey Social Exclusion, data November 2004, n = 2,225.

percent of them cannot afford to buy new clothes and three quarters cannot afford to go on a holiday. In contrast with the general population of the Czech Republic their households are significantly worse equipped with long-life products: for example, approximately a half of

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

19

them cannot afford even an old, used car (compared to 14% in the whole population). In comparison with the population of the Czech Republic in the case of income deprived people there is also a relatively higher degree of deprivation in housing conditions: a third of them suffers from lack of space in their flats and almost a quarter suffer from dampness in their flats (compared to 22 %, resp. 12 % in the whole population). And they also more often refer to health problems (36 % compared to 12 % in the population). In contrast, income disadvantaged people do not often suffer from deprivation in social contacts – actually the unemployed miss contact with friends less frequently than others. Gallie, Paugam and Jacobs [2002] arrived at a similar finding in the analysis of deprivation of the unemployed. This, however, does not mean that the social contacts that they maintain would provide the needed social capital – as they mainly lack wider but significant social networks [Granovetter 1974] and these cannot be mobilized to gain quality employment. In some of the analyzed items the degree of deprivation (subjective income deprivation – coping financially, further the ability to buy new clothes, to afford a holiday) is related medium strongly or rather weakly (in other items) to the status of the individuals on the labour market. According to Whelan, Layte [2003] permanent poverty is considered a decisive indication of deprivation and a source of social exclusion. In the sample of the income deprived individuals we find about 35 % of respondents who have felt poor for more than two years and it is interesting that this share is approximately the same as in the case of fully employed households and also in partly employed households as well as in non-employed ones. The difference among the households according to their status on the labour market is restricted only to the share of those who “do not feel poor at all”, in fully employed households there are 45 % of them, in partly employed ones 35 % and in non-employed households 23 %. The share of those who feel poor in the short term is, in contrast, 21 %, resp. 30 % and 40 % (contingency coef. = .223, sign. = .000). Deprivation in those areas that indicate the possibility of conducting “a mainstream lifestyle” and to a certain extent also the possibility to influence their own fates or their children’s fates is relatively strong in the case of the income deprived people and at the same time further differentiated according to status on the labour market than it is in the case of indicators of material deprivation.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 6. Supplementary indicators of material deprivation and social exclusion – by the status of the members of the respondent’s household on the labour market (in %) Total Respondent (or someone in his/her household): Has a mortgage Has supplementary pension insurance Has health insurance/accident cover Can support children’s study (if s/he has any) Can go out at least once a month (concert, theatre etc.)

25.6 17.5 37.9 52.5 40.1

Fully Partly NonCC empl. empl. empl. (sign. .000) household househld househld 32.6 31.5 14.5 .194 24.6 17.4 9.8 .174 47.3 45.0 23.0 .229 61.3 61.0 33.7 .249 48.3 40.0 31.0 .154

Note: The question was: “Have you or someone in your family …?”

20

Tomáš Sirovátka and Petr Mareš

In particular in the case of respondents living in non-employed households in comparison with those who live in fully and partly employed households (the latter two types do not differ much in this respect), it is evident that there is a restricted possibility to develop one’s cultural interests or to enable children to study (only about a third of the households declare it) and there is a similarly weak participation in supplementary pension insurance and use of mortgages (10 to 15 %). In this respect it is evident that employment and income from paid work provide the income deprived and their households with not an exactly higher living standard (as we saw in the indicators of income and material deprivation) yet they provide a certain stability and the possibility to dispose of income and decide with greater certainty about its use, to plan financially. And exactly this can be considered a component of participation in the mainstream lifestyle similarly to the active participation in the labour market. When assessing the overall degree of deprivation and social exclusion with the help of the used deprivation indices I, II and III respondents are on average strongly deprived – approximately 28–31 %, at the same time the status on the labour market has at most medium influence on their deprivation: while there is not much of a difference among the unemployed, the inactive and the temporarily employed (indices of “strong deprivation” reach 31 to 36 %) deprivation of the income deprived with permanent jobs does not exceed 25 %, actually it tends to be lower. The values of index IV which we consider the most appropriate measure of material deprivation (it involves a limited number of items, while it includes supplementary indicators of social exclusion, best reliability test) on average reach about 40 %. However, the values of the index are significantly lower in the case of persons with permanent jobs (significant deprivation is at 31%), compared to persons with temporary jobs (42 %) and in particular unemployed and inactive ones (46 %), coef. ETA = .290 (F = 67,868, sign = .000). Table 7. Values of cumulated indices of “strong deprivation” (values 0 to 100) by respondent’s status on the labour market index IV (12 selected items)

22.0 29.5 32.1 32.0

index III (6 compiled items) 24.0 31.6 32.9 33.4

34.7

31.6

31.0

46.0

Total

31.2

27.7

29.2

39.6

F (sign. .000) Eta

48.410

48.342

40,457

50.882

.283

.283

.228

.290

index I (13 items)

index II (23 items)

Permanent employment Temporary employment Short-term unemploymnt Long-term/repeated unemployment Economically inactive

25.0 33.3 35.6 36.1

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

respondent:

31.3 42.1 45.9 45.7

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

21

The values of the indices differ similarly depending on the status of the whole household in which the income disadvantaged live on the labour market. In concrete terms there is a difference between respondents living in fully or partly employed households on the one hand and those who live in unemployed households on the other: strong deprivation is on average at 24–29 % (respondents in employed households) as opposed to 34–37 % (respondents in non-employed households) in indices I to III and 34 % to 37 % (respondents in fully and partly employed households) as opposed to 48 % (respondents in non-employed households) in index IV ( coef. ETA = .279, F = 94,200, sign. = .000). In contrast to the influence of the status of a household on the labour market on the overall deprivation of the income disadvantaged individuals, the influence of the size and status of the respondent’s family on the overall material deprivation is significantly weaker than its influence in the case of income deprivation. Thus we find the lowest deprivation in the case of respondents who live in childless couples’ households (index IV = 34 %) and lower deprivation in the case of respondents living in nuclear families or living on their own (index IV = 38–39 %) as compared to respondents who live in single-parent households (index IV = 45 %), coef. ETA = .177, (F = 23,938, sign. = .05). The positive influence of the labour market status (employment) on decreased levels of material deprivation can be explained in particular with the opportunity to regularly dispose of income as the level of income – as we have demonstrated already – does not differ significantly. This hypothesis is also supported by the medium strong correlation of the cumulative deprivation index with the job quality index which, apart from income, also represents stability and perspectives linked to employment (the value of Pearson’s correlation coefficient between the job quality index and the index of material deprivation IV is 0.291, the value of correlation with the index of material deprivation I is .298 – sign. = .000).

MEASURES OF SOCIAL POLICY

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Take-Up and Evaluation of Welfare Benefits We have demonstrated that while the correlation between the labour market status and the level of average income in income deprived households is negligible, the association with material deprivation is medium strong – and that because it is above all permanent employment that provides the stability and the opportunity to dispose of income and to influence the future of one’s own family. What then is the relationship between labour market status and dependence on welfare benefits? What is the role of welfare benefits when eliminating material deprivation? Currently (at the time of interviewing) about half of the income deprived respondents were recipients of welfare benefits and about a quarter of them were on non-employment benefits while the difference between respondents living in employed and non-employed households or employed and unemployed individuals was significant: only 30 % of individuals living in fully employed households and similarly 30 % of individuals in permanent employment received welfare benefits while in the case of unemployed and inactive respondents it was about 70 % (contingency coef. CC = .364, sign. .000). If we were to trace whether they at all or during a year at least for some period of time and repeatedly

22

Tomáš Sirovátka and Petr Mareš

took up welfare benefits the difference between employed and unemployed respondents would be similar: 50 % (employed) as opposed to 90 % (unemployed). When taking up unemployment benefits, understandably, the difference according to the respondent’s labour market status is significant – nonetheless among the unemployed benefits are not taken up by more than a half: either they are no longer entitled to those or they are long-term unemployed as these benefits are provided only for six months or they were removed from the Labour Office register and thus they lost their entitlement to these. When comparing various factors (see Table 8) it becomes clear that the status of the respondents’ households on the labour market is a key variable determining the (non)dependence on welfare benefits (these involve income support to reach subsistence minimum) among the income disadvantaged individuals while other characteristics such as education, health, family status, do not play such a significant role. Thus we find that (non)dependence on welfare benefits as well as subjective income deprivation depend above all on the labour market status of the members of a household, however in the case of (non)dependence this is significantly stronger. Apart from this the family type plays a certain Table 8. Logistic regression – share of the likelihood that the respondent: a) does not take up welfare benefits, b) does not experience difficulties coping financially. Currently not on income support Exp(B) .13 .46 ref.

sign. .000 .000

Nuclear family Single parent family Single adult household Childless couple

.58 .33 .84 ref.

.000 .000 .317

.62 .47 .44 ref.

.004 .019 .017

Primary education Vocational training Secondary education University

.75 .91 1.25 ref.

.230 .668 .299

.34 .67 1.09 ref.

.000 .119 .774

Health problems No health problems

.97 ref.

.729

.64 ref.

.000

Non-employed household Partly employed household Fully employed household

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Not experiencing (significant) financial difficulties Exp(B) sign. .33 .000 .56 .001 ref.

Chi square

Model Summary 455,990 (9) sig. .000

Model Summary 232,032 (9) sig. .000

-2 Log likelihood Nagelkerke R Square

2553,492 .253

2267,004 .147

role in income deprivation (single-parent and nuclear families are more income deprived and also more dependent on welfare benefits than households without children). Finally some other individual characteristics also play an important role, above all education (lower

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

23

education is linked to higher deprivation) and health (worse health is connected with higher deprivation); characteristics such as gender, age and ethnicity do not play a statistically significant role (and they were not included in the model) – in the case of ethnicity we need to point out that statistical significance was also influenced by the low number of thus declared cases (40). Thus we identified the strong influence of the labour market status on dependence on welfare benefits while the influence of the labour market status on the average income per person in the household was essentially nil. Actually the households of income deprived respondents in particular those with children are close to the subsistence minimum – whether they are employed or not. Nonetheless, income from paid work eliminates the take-up of welfare benefits in that it slightly increases the income of the household and in addition it results in the obligation to declare income. Should they no longer be entitled to welfare benefits and a subsequent decrease in their incomes occurs they have to apply for benefits again which can result at least in a delay in acquiring finances. Thus as a consequence of employment (in particular temporary) income disadvantage is not necessarily so effectively eliminated as dependence on welfare benefits. In addition more constant personal characteristics such as education and health have a certain impact on subjective income deprivation – as illustrated in Table 8 – although, on the other hand, they do not have a strong influence on current dependence on benefits. We can expect that these characteristics in the long term influence, apart from others, also the character of accessible jobs and thus also the income level, the ability to accumulate finances and as a consequence the degree of income deprivation. Income disadvantaged individuals are generally rather critical of the level/generosity of welfare benefits: 73 % consider them too low to actually help them in their situation. Moreover, 46 % of respondents consider the benefit application process too complicated and 46 % consider the likelihood of application rejection high. As a consequence of the take-up of welfare benefits a third (34 %) of income deprived feel stigmatized and say they would keep this fact secret from their friends. And thus we can expect that the non-take-up of welfare benefits will not be rare. The factors that influence the non-take-up of benefits have been already discussed by Mareš [2001]. His expectation on the non-take-up of benefits corresponds with our finding that a part of the representative sample (although in 2004) did not apply for welfare benefits although they consider their situation as difficult as that of welfare benefit recipients and they also stated incomes at the same level as those of welfare benefit recipients. Precisely the difficulties encountered in the application process and the high likelihood of not being awarded benefits play a significant role. Non-take-up of benefits which they expect to be entitled to is more frequent in the case of respondents living in families with children with one employed parent. There are likely to be numerous reasons – on the one hand with these households there is certain income from paid work which is supplemented with welfare benefits so it gets close to the subsistence minimum or slightly over it and thus welfare benefits are not so often applied for due to their smaller contribution to the household, uncertainty of the claim, complicated application process and/or stigma linked with it. In opinions on welfare benefits we did not identify any significant differences due to the respondents’ status on the labour market. Can we conclude from the responses that for the majority of actual or potential welfare benefit recipients the system of welfare benefits does not inhibit the motivation to accept a job offer? We can see on the one hand that the level of

24

Tomáš Sirovátka and Petr Mareš

benefits is judged by the respondents mostly as low, however, we also see that a significant part of the income disadvantaged individuals are employed although they assess the quality of a job as low and incomes as not much higher than those of the unemployed, moreover, about a third of those in employment have only temporary jobs and their incomes are actually lower than those of the unemployed. Nonetheless employment of any kind clearly provides them with some sense of stability and an opportunity to dispose of an income and it also provides in a variety of ways a certain protection from three forms of deprivation that we link to “social exclusion”. And in fact a large part of the income disadvantaged people are well aware of these advantages. Yet it remains a question whether income from paid employment when compared to welfare benefits is a sufficient incentive exactly for those who are long-term dependent on welfare benefits, some of whom lose the hope of finding (quality) employment if they actually aspire to such. We have indeed identified a certain albeit small share of the so-called discouraged. In their case in terms of employment offers a negative role can be played by the fact that welfare benefits often enable an income actually comparable to the incomes of the employed in low income households which changes quite regularly albeit depending on the type of household, between 70 and 90 % of the replacement rate [Sirovátka, Žižlavský 2003; 20 Večerník 2004; OECD 2004]. Moreover, as soon as uncertainties linked with acquiring welfare benefits enter the scene when deciding about work activities the unemployed have to calculate these into the equation in the form of additional costs linked to possible future unemployment. That is why Atkinson [1999] alerts to the fact that the calculation of costs connected with future unemployment is reflected in the reservation wage for which the unemployed are willing to enter employment and the higher the uncertainty the higher the reservation wage. Jordan [1992] uses research conducted in low income households in England to argue that long-term unemployed often (and rather rationally) prefer the security of welfare benefits to insecure employment, risk of losing entitlement to benefits and the insecurity of repeated take-up of benefits. In our case the analysis of the database of welfare benefit recipients shows that the average period of uninterrupted take-up of benefits is one and a half years (!) and almost thirty percent of recipients are on benefits for more than two years [Sirovátka, Trbola 2003] thus a similar strategy can be expected on their part.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Chances of Terminating the Marginal Status on the Labour Market and Ending Deprivation There are a number of factors at stake that influence the possibility of terminating poverty and deprivation, such as the overall situation on the labour market, the composition of the social structure and mobility channels, influence of the institution of family and also policy measures: these involve institutional arrangements of access to education including lifelong education, social services, protection against discrimination on the labour market etc. For the 20

We should keep in mind that while in the Czech Republic minimum wage in relation to average wage is still lower than in EU-15 countries, the relation of the guaranteed subsistence minimum to minimum wage is on the contrary higher than in the majority of the EU-15 countries. It is mainly the consequence of lower work productivity and thus lower purchasing power.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

25

category of people who have worse chances on the labour market and are marginalized on it policy measures aimed at their inclusion actually often play a key role. Our income disadvantaged respondents assessed their chances of freeing themselves from poverty mainly as small while the differences among them according to their status on the labour market are not very large (contingency coef. CC = .110, sign. = .05). The assessment of chances of breaking off from poverty is medium strongly associated with enduring subjective poverty (contingency coef. CC = .259, Sign. = .000) when people who have felt subjectively poor for longer than two years are utterly sceptical regarding chances of breaking off from this situation (about 45 % of them see no chance and others – apart from exceptions – see only a small chance of breaking off from poverty). In the case of those whose income deprivation is related to labour market status (unemployment, instability and low job quality) – which applies to most of our respondents – the opportunity to change this situation in the longer term can lie in requalification/vocational training programmes thus in improving their employability, human capital and consequently opportunities for employment. That is why some authors [EspingAndersen et al. 2002; Kvist, Jaeger 2002] consider the stress on increasing the human capital of unskilled workforce affected by structural change a key aspect when distinguishing regimes of the welfare state in the context of “new social risks”. In our case regardless of their status on the labour market about 5% of the sample of income deprived respondents actually participated in re-qualification courses, in the case of the unemployed this was somewhat higher – almost 7 % which is, by the way, precisely in line with the statistics provided by the Ministry of Labour and Social Affairs for 2004. Apart form that, more than a quarter of the income deprived individuals were interested in requalification/vocational training yet they did not pursue it. Among the unemployed the interest in re-qualification was even higher (31 %) than among others. We cannot say that interest in re-qualification would decrease or increase with the duration of subjective poverty. However, unfortunately, actual participation in re-qualification courses decreases with the duration of deprivation. This means, for example, that those people who are exposed to income deprivation for longer than two years participate in re-qualification training only in 3 % of the cases. In this respect our finding again rather precisely reflects other findings on the relatively small scope of active employment measures in the Czech Republic in the case of most disadvantaged groups of the unemployed according to data from the Ministry of Labour and Social Affairs [Sirovátka et al. 2004]. Forms of re-qualification and training in which income deprived individuals participate also differ rather significantly in terms of their labour market status (contingency coef. CC = .412, sign. = .000) because employed individuals make use of more beneficial forms of improving human capital than the unemployed. Most frequently (in 40 % of cases) participants in re-qualification courses take part in programmes organized by the Labour Office, these involve more than half of the unemployed participants of such courses. About 20% of participants took part in courses aimed at increased qualifications or vocational training and in the remaining cases (almost 30 %) these involve studying at secondary schools or universities. These higher forms of distance studies are attended by half of those participants in re-qualification courses who are employed and from the unemployed it is only 15% of all participants in re-qualification courses.

26

Tomáš Sirovátka and Petr Mareš

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

CONCLUSIONS Labour market status is a determining factor in income disadvantage as the risk of low income (below subsistence minimum) as well as the risk of subjective income deprivation in the case of the unemployed is a number of times higher than in the case of employed people. Despite that income disadvantaged people and their households constitute a heterogeneous group. Employed individuals represent a significant part of them. On the other hand the share of the unemployed is also relatively high and moreover the most dominant group is that of the repeatedly and long-term unemployed. Among the households of income disadvantaged people more than one third comprises households with no employed members. However, a large part of these is made up of households of unemployed individuals. In contrast in the case of nuclear family households we most frequently identify parallels in partners’ employment and unemployment (in half of the cases), in fact even here we find a significant share of households with both partners in employment. Marginal status on the labour market is also represented in other ways: more than a third of income disadvantaged individuals have a temporary job or one without a contract. The disadvantage is also rather strongly demonstrated in the low “quality” of jobs as a larger part of the income disadvantaged individuals is not satisfied with income in the current or previous job, about half of them are dissatisfied with promotion and training opportunities, the physically demanding nature of the job and commuting – with no difference among currently employed or unemployed people. The income level of employed and unemployed individuals differs only slightly, there is more of a difference according to the composition of the household. Childless households are significantly above the poverty line or subsistence minimum while households with children are on average close to the subsistence minimum and below the poverty line. In contrast, income and material deprivation which are relatively strong in a number of dimensions (household’s financial status and debt, basic needs, equipment) are more differentiated according to the status of the respondents and their households on the labour market (depends to a lesser degree on the presence of children in the household). The most significant differentiation according to labour market status occurs in the case of income disadvantaged individuals in relation to deprivation indicators that take into account “mainstream” lifestyles and the opportunities to influence one’s own way of life as well as that of one’s children. Labour market status of the income disadvantaged individuals even more strongly (than subjective material deprivation) influences dependence on welfare benefits while other individual characteristics have less significance. Nonetheless in income deprivation we can identify the influence of such characteristics e.g. education and health that in the long term determine not only access to employment but also job quality. Income disadvantaged individuals usually consider welfare benefits too low, apart from that almost a half assess the application process as too complicated and the likelihood of being denied benefits as too high. A third of them feel stigmatized by benefit take-up. These circumstances might be related to the fact that only half of the income deprived people currently take up benefits and many do not apply for them at all although they consider the option. Almost 90% of income deprived people state that they have no or very small chance of breaking off from income and material deprivation regardless of their actual labour market status. Only 5 % of them currently participate in re-qualification courses or training

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

27

programmes which could actually most improve their chances on the labour market, another quarter of them express interest in training yet they do not pursue it. We can characterize income deprived people generally as a marginal workforce comprising a number of segments, a portion of them has rather lower quality but more stable jobs. The stability of employment and income for them, regardless of the value of the income, represent a chance to influence their own lives and to maintain a lifestyle and chances that are accessible to the “mainstream” society. Yet this stability often works to the detriment of other job qualities: above all the level of income, it seems. For example, these employees are sufficiently wage flexible depending on the opportunities that they identify on the labour market. A different segment of the income disadvantaged people combines temporary or less stable jobs with short-term unemployment in an effort to gain a more stable and eventually more beneficial job and thus opportunities to acquire more stability and ways of influencing one’s life. It is significant that the quality of the job that they accept (including wages) is in their opinion often low, actually it is assessed worse than the previous job of the currently unemployed. The incentive to accept such a job appear to be aspirations for a better job as well as insecurity and stigmatization linked to (repeated) take-up of welfare benefits. And finally there is that part of the income disadvantaged individuals who remain unemployed longest or repeat unemployment often. This is apparently partly because they realistically assess their low chances of acquiring a more stable and higher quality job but also because they do not want to risk losing entitlement to welfare benefits and the uncertainty connected with re-applying for them. It is in particular this segment of the income disadvantaged group that gets into a state of longer term dependence on welfare benefits: in fact the benefits are on average higher compared to income from paid employment. It is interesting that although there are significant differences in dependence on welfare benefits among these segments of low income households differences in the quality of the current or accessible jobs are negligible, similarly to differences in income. Yet differences in material deprivation are somewhat bigger and they are manifest in particular in areas that are connected to the mainstream lifestyle and the opportunity to influence one’s own life. In the context of the so-called “regimes” of the labour market our findings indicate that income disadvantaged and marginalized groups on the labour market are relatively strongly bound to the secondary segment of the labour market which is based on the principles of wage and nominal flexibility. It enables them to acquire jobs mostly of low quality thus jobs that are relatively little paid and linked to less security. Support for improving human capital and chances of terminating material deprivation are perceived as low in the case of these jobs. The situation in this segment can be understood as close to the “liberal labour market regime” when apart from little access to “quality” workplaces income and material deprivation of a relatively significant part of the employed and the unemployed is remarkable (although it is still differentiated). For a segment of the income deprived individuals this situation is an incentive and they accept the demand for wage flexibility and the advantage of income, for others – with respect to low quality and insecurity of a possible job acceptance (and consequently the uncertainty connected with the renewed take-up of social benefits in case of the loss of such an insecure job) on the one hand and on the other with respect to the relatively acceptable level of benefits in comparison with income from paid employment – dependence on welfare benefits is certainly not a preferred alternative but under the given circumstances definitely an acceptable one.

28

Tomáš Sirovátka and Petr Mareš

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Atkinson, A. B. (1999). The Economic Consequences of Rolling Back the Welfare State. Cambridge, MA: MIT Press. Atkinson, A. B. (1998). Poverty in Europe. Oxford: Blackwell Publishing. Atkinson, A. B., B. Cantillon, E. Marlier and B. Nolan. (2002). Social Indicators: The EU and Social Inclusion. Oxford: Oxford University Press. Barnes, M., C. Heady, S. Middleton, J. Millar, F. Papadopoulos, G. Room and P. Tsakloglou. (2002). Poverty and Social Exclusion in Europe. Cheltenham: Edward Elgar. Baumann, Z. (1998). Work, Consumerism and the New Poor. Oxford: Oxford University Press. Beck, U. (2000). The New Brave World of Work. Cambridge: Polity Press. Berger, S. and M. J. Piore. (1980). Dualism and Discontinuity in Modern Societies. Cambridge: Cambridge University Press. Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism. Oxford: Oxford University Press. Esping-Andersen, G. (1999). Social Foundations of Postindustrial Economies. Oxford: Oxford University Press. Esping-Andersen, G. and D. Gallie, A. Hemerijck, J. Myles (2002). Why We Need a New Welfare State ? Oxford: Oxford University Press. Eurostat (2000). European Social Statistics: Income, Poverty and Social Exclusion. Luxembourg: Eurostat. Gallie, D. (1999). „Unemployment and Social Exclusion in the European Union. European Societies 1 (2): 139–167. Gallie, D. and S. Paugam (eds.). (2000). Welfare Regimes and the Experience of Unemployment in Europe. Oxford: Oxford University Press. Gallie, D., S. Paugam and S. Jacobs. (2002). Unemployment, Poverty and Social Isolation. European Societies 5 (1): 1–32. Gazier, B. (1999). Employability. Theory and Practice. Geneva: ILO. Giddens, A. (1998). The Third Way: The Renewal of Social Democracy. Cambridge: Polity Press. Granovetter, M. S. (1974). Getting a Job: A Study of Comtacts and Carreers. Cambridge, MA: Harvard University Press. Halleröd, B. (1995). The Truly Poor. Direct and Indirect Consensual Measurement of Poverty in Sweden. European Journal of Social Policy 5 (2): 111–129. Heery, E. and J. Salmon (2000). The Insecure Workforce. London: Routledge. Jordan, B. 1992. Trapped in Poverty. Labour Market Decisions in Low Income Households. London: Routledge. Kabele, J. (1998). Přerody. Principy sociálního konstruování. [Transformations. Principles of social construction.] Praha: Karolinum. Katrňák T. and P. Mareš. (2007). “Segmenty zaměstnaných a nezaměstnaných v České republice v létech 1998 až 2004” Sociologický časopis, 43(2):281-303. Kvist, J. and M. M. Jaeger. (2002). Up for Challenge? Western European Welfare States under Pressure. Copenhagen: The Danish National Institute for Social Research.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

29

Layte, R. and Ch. T. Whelan. (2003). Cumulative disadvantage or individualisation? A comparative analysis of poverty risk and incidence. European Societies 4 (2): 209–233. Lindbeck, A. and D. J. Snower. (1989). The Insider-Outsider Theory of Employment and Unemployment. Cambridge, Massachussetts: MIT Press. Lødemel, I. and H. Trickey (eds.). (2001). An Offer You Can’t Refuse.Workfare in International Perspective. Bristol: The Policy Press. Mareš, P. (2001). Nevyužívání sociálních dávek. Praha: VÚPSV ČR. Výzkumná zpráva. Mareš, P. and L. Rabušic. (1994). Nezaměstnanost v České republice na počátku 90. let v regionálním pohledu. [Unemployment in the Czech Republic at the beginning of the 1990s from a regional perspective.] Sociologický časopis 30(4): 475-498. Mareš, P. and L. Musil (1994). The Legitimacy of Privatization: Two Case Studies of Privatized Enterprises. Czech Sociological Review 30(2):187-197. Mareš P., T. Sirovátka and J. Vyhlídal (2003). Dlouhodobě nezaměstnaní – životní situace a strategie. [Long-term unemployed – life opportunities and strategies.] Sociologický časopis, 39(1):37-54. Mareš, P. and T. Sirovátka (2005). Unemployment, Labour Marginalisation and Deprivation. Czech Journal of Economics and Finance 55 (1–2): 54–67. Možný, I. (1994). Pokus o mimoekonomické vysvětlení současné plné zaměstnanosti v České republice. [An attempt at a non-economic explanation of the current full employment in the Czech Republic.] Sociologický časopis, 30(4): 463-474. Nolan, B. and Ch. T. Whelan (1996). Resources, Deprivation and Poverty. Oxford: Oxford University Press. OECD (2004). Benefits and Wages. OECD Indicators. Paris: OECD. Offe, C. (1985). Disorganised Capitalism. Cambridge: Polity Press. Paugam, S. (1995). The Spiral of Precariousness: A Multidimensional Approach to the Process of Social Disqualification in France. pp. 49–79 in G. Room. Poverty and Social Exclusion in Europe. Bristol: The Policy Press. Rabušic, L. and P. Mareš (1996). Je česká společnost anomická? [Is the Czech society anomic?] Sociologický časopis, 32(2):175-188. Room, G. (1995). Beyond the Threshold: The Measurement and Analysis of Social Exclusion. Cambridge: Polity Press. Sainsbury, D. and A. Morissens (2002). Poverty in Europe in the mid-1990s: the effectiveness of means-tested benefits. Journal of European Social Policy 12 (4): 307– 327. Saraceno, Ch. (ed.) (2002). Social Assistance Dynamics in Europe. National and Local Poverty Regimes. Bristol: The Policy Press. Sen, A. (1992). Inequality Re-examined. Oxford, New York: Oxford University Press. Serrano Pascual, A. (ed.) (2004). Are Activation Policies Converging in Europe? The European Employment Strategy for Young People. Brussels: ETUI. Schmid, G. and B. Gazier (eds.) (2002). The Dynamics of Full Employment. Social Integration Through Transitional Labour Markets. Cheltenham, UK, Northhampton, MA, USA: Edward Elgar Publishing. Sirovátka, T. (1997a). Marginalizace na pracovním trhu. [Marginalization on the labour market.] Brno, Masarykova univerzita.

30

Tomáš Sirovátka and Petr Mareš

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Sirovátka, T. (1997b). Sociální a ekonomické faktory marginalizace na pracovním trhu. [Social and economic factors of marginalization on the labour market.] Sociologický časopis, 33(2):169-188. Sirovátka, T., M. Horáková, V. Kulhavý and M. Rákoczyová (2004). Efektivnost opatření aktivní politiky zaměstnanosti v roce 2003. [Effectiveness of measures of active employment policy in 2003.] Praha, Brno: Výzkumný ústav práce a sociálních věcí (www.vupsv.cz). Výzkumná zpráva. Sirovátka, T. and M. Žižlavský (2003). Nezaměstnanost a pracovní pobídky. [Unemployment and job incentives] Politická ekonomie 51 (3): 339–406. Sirovátka, T. P. Kofroň, M. Rákoczyová, O. Hora and R. Trbola (2005). Chudoba a sociální vyloučení v České republice ve srovnání se zeměmi Evropské unie. [Poverty and social exclusion in the Czech Republic compared to the countries of the European Union.] Praha, Brno: Výzkumný ústav práce a sociálních věcí (www.vupsv.cz). Výzkumná zpráva. Sirovátka, T. and P. Mareš (2005). „Poverty in the Czech Republic and the policies to combat it.“ Pp. 252–277 in S. Golinowska, E. Tarkowska, I. Topińska. Ubóstwo i wykluczenie spoleczne: badania, metody, wyniki. Warszawa: IPISS. Sirovátka, T. and R. Trbola (2003). Sociální dávky a jejich příjemci.[Welfare benefits and their recipients.] Praha, Brno: Výzkumný ústav práce a sociálních věcí (www.vupsv.cz). Výzkumná zpráva. Standing, G. (1999). Global Labour Flexibility. Seeking Distributive Justice. Houndmills, Basingstoke, London: Palgrave Macmillan. New York: St. Martin´s Press. Van Berkel, R. and I. H. Møller (eds.) (2002). Active Social Policies in the EU. Inclusion through Participation? Bristol: The Policy Press. Večerník, J. (2004). Who Is Poor in the Czech Republic ? The Changing Structure and Faces of Poverty after 1989. Czech Sociological Review 40 (6): 807–834. Vyhlídal, J. and P. Mareš (2006). Měnící se rizika a šance na trhu práce. Analýza postavení a šancí vybraných rizikových skupin na trhu práce [Changing risks and opportunities on the labour market. An analysis of selected at-risk groups on the labour market]. Praha: Výzkumný ústav práce a sociálních věcí. Whelan, Ch. T., R. Layte and M. Bertrand (2002). Multiple Deprivation and Persistent Poverty in the European Union. Journal of Social Policy 12 (2): 91–105.

APPENDIX Overview of items in deprivation indices I to IV index I (13 items according to Eurostat, 2000) Reliability: alpha = .7345 F = 109,4514 (prob. .0000)

Appendix (Continued).

index II (23 items)

index III (23 items, compiled into six sub-sets)

index IV (12 selected items)

alpha = .6743 F =724, 2615 (prob. .0000)

alpha = .5814 F =2050,066 (prob. .0000)

alpha = .7423 F =505,8052 (prob. .0000)

Poverty, Deprivation and Social Exclusion: The Unemployed and the Working Poor

Financial deprivation

Financial deprivation

- has great difficulties to make ends meat

- has great difficulties coping financially

- has problems paying rent and energy bills

- has problems paying rent and energy bills

Deprivation in basic needs and lifestyle

Deprivation in basic needs and lifestyle

- does not eat meat,fish, poultry every other day

- does not eat meat,fish, poultry every other day

- cannot buy new clothes

- cannot buy new clothes

- cannot afford one-week holiday away from home

- cannot afford one-week holiday away from home

Financial deprivation

Financial deprivation - has great difficulties to make ends meat - has problems paying rent and energy bills

Deprivation in basic needs and lifestyle

Deprivation in basic needs and lifestyle - does not eat meat,fish, poultry every other day - cannot buy new clothes - cannot afford oneweek holiday away from home - cannot afford to heat flat properly - cannot provide for children’s education - cannot afford to go out once a week to a concert, a play, a meal

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

- cannot afford to heat flat properly - cannot provide for children’s education - cannot afford to go out once a week to a concert, a play, a meal

Deprivation in housing

Deprivation in housing

- lack of space in flat - damp flat - no bathroom/shower in the flat

- lack of space in flat - damp flat - no bathroom/shower in the flat - shares flat with another family - lack of light in flat - insufficient heating - no flush toilet - no kitchen

Deprivation in long-life equipment - no colour TV set - no telephone - no car

Deprivation in long-life equipment - no colour TV set - no telephone - no car - no washing machine - no PC

Appendix (Continued).

31

Deprivation in housing

Deprivation in housing - lack of space in flat - damp flat

Deprivation in Deprivation in longlong-life equipment life equipment - no telephone - no car

32

Health problems - experiences significant health problems

Tomáš Sirovátka and Petr Mareš

Health problems - experiences significant health problems

Social life Social life - meets neighbours, friends, - meets neighbours, friends, family less than once a month family less than once a month

Health problems - experiences significant health problems Social life - meets neighbours, friends, family less than once a month

TOMÁŠ SIROVÁTKA works as a professor at the Department of Social Policy and Social Work, Faculty of Social Studies, Masaryk University in Brno, Czech Republic where he lectures on social policy and the labour market. His main research interests include unemployment and social policy. PETR MAREŠ works as a professor at the Department of Sociology, Faculty of Social Studies, Masaryk University in Brno, Czech Republic. His research interests concentrate on social problems, inequality, poverty and unemployment. Apart from lecturing on these topics he also convenes courses on the methodology of social sciences, qualitative research and data analysis.

ACKNOWLEDGEMENTS

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

This study was written with the support by the Ministry of Education of the Czech Republic (Reserach schneme MSM 0021622408 "Social Reproduction and Social Integration").

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 2

EVIDENCE ON THE RELATIONSHIP BETWEEN UNEMPLOYMENT AND HEALTH Anton Nivorozhkin1a and Laura Romeu Gordob a

Institute for Employment Research (IAB), Nürnberg, Germany b German Centre of Gerontology (DZA), Berlin, Germany

ABSTRACT In this chapter we review recent studies on the impact of unemployment on individual health and well-being. We start the discussion with an overview of the theoretical background of the relationship between unemployment and well-being. Next, we summarize recent empirical studies with the objective to provide a basis for scholars who want to contribute to the literature in this field.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION Reduction of unemployment is one of the main keystones of economics policy. Politicians and scholars devote much effort in analysing which are the factors determining unemployment (especially long term unemployment) and in designing policies which would increase the reemployment rates of the long term unemployed. An important aspect which needs to be considered for the design of the right activation policies is the situation of the long-term unemployed. Most of recent labour market reforms (like the German labour market reform of 2005) contemplate the reduction of the entitlement period of unemployment subsidies. The idea behind this kind of reforms is that unemployment benefits may work as a disincentive for job search. In this context, it is important to investigate whether this is the case for all long term unemployed or whether there are groups of individuals for whom unemployment benefits do not have any effect on 1

Corresponding address: Anton Nivorozhkin. Regensburgerstr. 104, 90478 Nürnberg (Germany) [email protected] [email protected]

Anton Nivorozhkin and Laura Romeu Gordo

34

their incentives. Concretely, it is important to investigate the following questions: Are long term unemployed really individuals who don’t want to work? Or are they individuals who were in the past active in the labour market and they see now how their chances to find a job again decrease as the time goes by? It is for sure not possible to draw a clear picture. Probably for some long-term unemployed, unemployment subsidies really act as a disincentive in finding a new job. However, for other unemployed who can not influence their situation because employers are not willing to hire them, unemployment subsidy does not play a role in the determination of the duration of the unemployment. For individuals who remain involuntarily unemployed, inactivity may have an effect on their well-being. First, they face financial difficulties; and furthermore, they are confronted with the loss of self-esteem derived from the lack of perspectives in a society in which work is one of the main key stones. In the literature we find evidence of the negative impact of unemployment on mental and physical health. However, solid empirical evidence is rare. Most of the empirical studies carry out cross-sectional analysis. This structure does not allow distinguishing whether the correlation between unemployment and ill-health is caused by the negative effect of unemployment on health or whether individuals with poor health are more likely to become unemployed. In the present chapter, we review the main studies which deal with the effect of unemployment on well being and health. The main objective is to provide a basis for scholars who want to contribute with solid evidence to the literature in this field.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2. THEORETICAL BACKGROUND The objective of the present study is to analyse the relationship between unemployment and health. The starting point for our analysis is the review of the theories developed by Jahoda and Warr, since they offer a comprehensive framework for analysing how unemployment affects health. Furthermore, we look at other theories (Freud, Erikson) which explain the reasons for work being such an important determinant of health. Finally, we describe the stages model which explains the phases that unemployed go through after job loss. Jahoda (1982) suggests that besides the manifest function of employment (financial remuneration) there are other benefits from working (latent functions). The author argues that employment: • • • • •

Imposes a time structure on the day, Implies regularly shared experiences and contact with others, Links an individual to goals and purposes which transcend his/her own, Defines aspects of personal status and identity, Enforces activity.

According to the author, an individual who is unemployed suffers from the absence of these latent functions. His mental health is affected because there are some needs which are not covered.

Evidence on the Relationship Between Unemployment and Health

35

For Jahoda, it is not the economic strain that provokes an impoverishment in the individual’s mental health. Rather, it is the absence of employment by which the individual covers his basic psychological needs (Ezzy, 1993). Jahoda also suggests that leisure cannot be an alternative to employment since the latent functions are not accomplished by leisure (Ezzy, 1993). Jahoda’s functional theory does not account for the possibility of an improvement in mental health when quitting a dissatisfying job. That is, for Jahoda unemployment is always worse than employment whatever the characteristics of the job, and reemployment is always better than remaining unemployed. Jahoda’s theory, however, does not offer an explanation of why there is a variation from individual to individual in the effects of unemployment on mental health. An alternative view of the impact of unemployment on mental health is presented by Warr. According to Warr (1987), mental health is assumed to be influenced by the environment in a manner analogous to the effect of vitamins on physical health. The availability of vitamins is important for physical health up to but not beyond a certain level. At low levels of intake, vitamin deficiency gives rise to physiological impairment and illhealth, but after attainment of specified levels there is no benefit derived from additional quantities. Warr suggests that some employment features are important to mental health in a similar manner: their absence leads to impairment in mental health, but their presence beyond a required level does not yield further benefit. These determinants of mental health are: •



Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.



• •

• •

Opportunity for control. The environment of the unemployed offers fewer opportunities to control activities and events. Therefore, they have a smaller scope for making decisions. Opportunity for skill use. The opportunity for skill use at work is a source of satisfaction which unemployed cannot use. Externally generated goals. The unemployed have a lack of goals determined by the environment. They do not have the same obligations and targets as when they were employed. Warr suggests that “role-generated requirements give rise to organised sequences of actions, in which specific targets and their overall structure provide ‘traction’ which draws people along”. Variety. Unemployed have more restricted behaviour and environment. Environmental clarity. Unemployed have a lower predictability in relation to the consequences of their actions. They have also a lower predictability when it comes to other people in the environment, and therefore they cannot foresee the reaction to some of their actions. Furthermore, they are not aware of the normative expectations concerning their behaviour. Availability of money. Most unemployed suffer from financial worries and this can give rise to many processes that are likely to impair mental health. Physical security. This issue is related to the previous one: availability of money. When shortage of money is extreme the individuals can be physically threatened. This happens, for example, when the financial resources of the individual are not enough to cover the family housing requirements.

Anton Nivorozhkin and Laura Romeu Gordo

36 •

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.



Opportunity for interpersonal contact. Unemployed suffer a reduction in the number and, as a rule, in the quality of interpersonal contacts. Valued social position. Unemployed suffer a decline in social position.

Warr’s model explains the negative impact of unemployment by arguing that there is an impoverishment in the environment of the unemployed. In the framework of this model, differences in mental health among different groups of unemployed can be explained by the differences in the environments of these groups. That is, unemployment does not affect all individuals in the same way, since there may be differences between groups in all the environment features presented above. The vitamin model also accounts for the possibility of having a positive effect on losing a job when this job is dissatisfying. If the environment of a dissatisfied worker is having a negative effect on his mental health, some benefit can be gained from quitting the job. For the same reason, employment in a dissatisfying job can produce a negative effect (e.g. Ezzy, 1993). It can also happen that after reemployment the environment is not better for the individual, and therefore the individual does not obtain any benefit from reemployment. In an empirical analysis, the Warr and Jahoda theories can be tested by the introduction of interaction effects between unemployment and other variables. An illustrative example is the interaction effect between ‘being unemployed’ and ‘social support’. Social support is expected to reduce the impact of unemployment through the coverage of some of the latent functions of employment introduced by Jahoda, and through the improvement of the environment of the individual (Warr’s theory). Through the introduction of interaction effects we can also identify the groups that may be more affected by unemployment. Warr in his theory accounts for the possibility of differences in the effect of unemployment between groups, whereas Jahoda does not. In an empirical analysis, this can be tested by introducing interaction effects between unemployment and some characteristics of the individual, such as age or education. In this way we can test whether there are age or educational groups that are especially affected by unemployment. Warr’s model also accounts for the possibility of a positive effect from losing a dissatisfying job and for the possibility of a negative effect from being reemployed in a dissatisfying job. The idea is that the individual who was not at all satisfied with his job may not be affected by unemployment in the same way as a person who was completely satisfied with his job. In the same way, individuals who are reemployed in a dissatisfying job may not recover in the same way as people who are reemployed in a satisfying job.

2.1. Other Theories There are also other theories which explain how health may be affected by labour status. Freud claimed that work ties us to reality. The routine imposed by employment helps individuals to avoid creating ‘new problems’ because they have too much free time: “If we are not obliged to get up in the morning and apply ourselves to a job then we are in danger of being overwhelmed by fantasy or emotion. The unemployed broken-hearted adolescent has

Evidence on the Relationship Between Unemployment and Health

37

time to dwell on her problems, while the girl working in the post office has to concentrate on her work” (see Smith, 1985). Erikson is the founder of the life span developmental theory (see Goldsmith et. al, 1996). According to this theory, an individual’s healthy psychological ego development depends on successful completion of each stage of human development. Specifically, a healthy transition from adolescence to adulthood is contingent upon the attainment of a desirable occupational identity. Therefore, Erikson would expect lack of success in the labour market to diminish an individual’s sense of worth. The final theory reviewed in this section is the stages model. This model explains the phases which unemployed go through when quitting their job: “First there is shock, which is followed by an active hunt for a job, during which the individual is still optimistic and unresigned; he still maintains an unbroken attitude. Second, when all efforts fail, the individual becomes pessimistic, anxious, and suffers active distress; this is the most crucial state of all. And third, the individual becomes fatalistic and adapts himself to his new state but with a narrower scope” (see Eisenberg et. al, 1993).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3. REVIEW STUDIES In this section, we review recent studies which summarize the existing literature dealing with the relationship between unemployment and health. Most studies classify the literature according to the outcome variable like morbidity, mortality or mental health. Some studies present classification according to methodological design (macroeconomic and microeconomic studies and plant closure studies). Wilson and Walker (1993) offer a review of studies that analyse the effect of unemployment on different health measures. The authors suggest that the main explanation for the relationship between unemployment and health is that unemployment is often inextricably combined with other social disadvantages. Unemployment is a very important factor in the development of social deprivation. The authors conclude that it is demonstrated in the literature that unemployed and their spouses have a higher mortality rate. Their children are also exposed to higher perinatal and infant mortality rates. Another important conclusion concerns the effects of unemployment on family life: children of unemployed parents face a higher risk of abuse. Furthermore, unemployed families are reported as having a high incidence of wife battering and domestic violence. Suicide and attempted suicide seems to be more common among the unemployed. Depression and neurosis too appear to be higher for unemployed and their families. In terms of physical morbidity, the authors conclude that there is a significant increase in general practitioner consultations for unemployed and their families. Finally, Wilson and Walker (1993) conclude that there appears to be an anticipatory effect of job loss on individual health. Shortt (1996), in his study reviews the literature on the effects of unemployment on mortality, morbidity, and mental health of women and young individuals. In line with Wilson and Walker (1993) Shortt concludes that unemployment accounts for at least some increase in mortality. The author also concludes that there is an adverse effect of unemployment on physical and mental health. He also concludes that there is a pathological impact of unemployment on children and spouses of the unemployed.

Anton Nivorozhkin and Laura Romeu Gordo

38

The author points out that most of the studies are focused exclusively on males. However, the few studies which focus on women and young individuals identify similar negative effects of unemployment on mental and physical health. Shortt reviews selected macro studies (e.g. Brenner, 1977; 1979; 1983; 1987) and concludes that these studies often do not properly address causality between unemployment and health. Therefore, the author recommends that future studies should concentrate on micro-level analysis. Jin et al. (1995) analyse epidemiological studies which investigate the association between unemployment and ill-health. The authors assess the findings according to the epidemiological criteria for causation. These criteria take into account the following aspects: temporal direction, strength of association, dose-response relation, consistency of findings, experimental evidence, specificity, analogy and biological plausibility. The results on the influence of unemployment on morbidity, mortality, suicide, alcohol consumption and use of mental and general health care services are in line with the studies reviewed by Wilson and Walker (1993) and Shortt (1996). In addition, the authors review the link between unemployment and deaths due to motor vehicle accidents. They conclude that the literature does not identify an empirical association between unemployment and deaths due to motor-vehicle accidents. Goldney (1997) reviews ten studies examining the influence of unemployment on cardiovascular diseases. The author argues that since none of these studies present a longitudinal structure, it is not possible to confirm the direction of causality. Murphy and Athanasou (1999) present a review of recent studies that investigate the effect of unemployment on mental health. The sixteen studies included meet the following criteria:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. The use of standardised psychological tests as measures of the dependent variable. 2. Conducted with a longitudinal design. 3. Published in the last 10 years in English-language scientific journals. Based on the results of the studies, the authors conclude that individuals who have lost their job have worse mental health than the employed individuals. The authors also discuss the magnitude of the effect of unemployment on mental health. They conclude that moving from unemployment to employment not only has a positive significant effect on mental health. Schwefel (1986) reviews the existing literature in German-speaking countries. He presents evidence on illness and the entry into unemployment, illness and the duration of unemployment, and illness and the reintegration of the unemployed. The author reviews the literature which identifies specific ‘problem groups’ like social and mental problem groups, unemployed youth, unemployed women, unemployed whitecollar workers, unemployed elderly, children of the unemployed and short-time workers. Schwefel summarises the main results from the reviewed literature as follows: • •

Unemployment and overwork can induce similar psychosomatic impairments. The unemployment effects in the 1930s are different from the unemployment effects in the 1970s. The effects of unemployment in the 30s had to do more with impoverishment. Unemployed individuals in the 1930s suffered physical deprivation.

Evidence on the Relationship Between Unemployment and Health

39

In the 1970s and today, unemployment imposes more psychological than physical burdens. Smith (1985) published a series of articles in the British Medical Journal in order to make doctors aware of the powerful effect that unemployment has on ill-health. The articles were published in the 1980s, when there was high unemployment in England. In this period, the interest on the consequences of unemployment grew and led to the publication of such a series. Kasl and Jones (1998) present a comprehensive overview of the impact of unemployment and retirement on several health indicators. In their study, the authors discuss conceptual and methodological frameworks adopted in the literature. They make a comprehensive review of the studies which deal with the effect of unemployment on mortality and physical morbidity, biological and behavioural risk factors, and on mental health and well-being. The authors also review the literature on the impact of job insecurity and threatened job loss, and the impact of retirement. The authors summarise the conclusions of the reviewed studies in the following points: • • •





Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.



Unemployment appears to be associated with a higher total mortality of about 20%30% in most studies. The impact of unemployment on physical morbidity is also evident, but the results are more variable and more difficult to interpret. Biological indicators of stress reactivity and disease risk provide rather good evidence of their acute sensitivity to some aspects of the unemployment experience (including anticipation) but chronic elevations in relation to enduring unemployment are infrequently documented. Behavioural and life style risk factors, such as smoking or exercise, show sporadic evidence of impact, as well as considerable complexity of findings: some of these variables seem implicated in selection rather than causation. Unemployment clearly increases psychological distress, particularly symptoms of depression, but overt diagnosable disorders are probably not elevated. The increases in distress seem reversible upon re-employment. A variety of indicators of physical and psychological morbidity and cardiovascular risk are likely to show adverse effects under conditions of heightened job insecurity. High community levels of unemployment have a negative impact on depressive symptoms of employed individuals (urban setting), an effect which can be interpreted as being due to threatened job loss.

On the methodological side, Kasl and Jones (1998) conclude that the main issue of the literature is the distinction between causation and selection: “Does the observation of poorer physical and mental health reflect the impact of unemployment or does it, instead, denote the influence of prior characteristics of the individual who later become unemployed?” They suggest that the evidence supports both the causation and the selection interpretations, and that the two interpretations are not incompatible. This is the issue that has mainly determined the design of the studies that analyse the impact of unemployment on health and well-being.

Anton Nivorozhkin and Laura Romeu Gordo

40

The first distinction made by Kasl and Jones among the different study designs is whether the study is based on aggregated or individual data. Next, they distinguish between controlled randomised programs and observational (non-experimental) studies. The usual design is the second one, i.e. both longitudinal and cross-sectional observational studies. They classify the longitudinal studies in three types of design: • •



Natural experiments. A typical example is a plant or factory closure study. Longitudinal comparisons of the employed and unemployed. It is, however, difficult to obtain the right data for such longitudinal comparisons. The authors suggest that the design is rather weak when no baseline data are available on health and social characteristics of the two cohorts. Follow-up of the unemployed to detect benefits of re-employment.

Kasl and Jones suggest that cross-sectional designs normally cannot distinguish between unemployment causing poor health (causation hypothesis) and poor health causing unemployment (selection hypothesis). There may be self-selection processes which cannot be controlled for the use of a cross-sectional design. Goldsmith et al. (1996) review the literature which relates unemployment and selfesteem. They conclude that there is no consensus relating to the impact of unemployment on self-esteem. They suggest that this is due to methodological problems in the considered studies that had not been solved up to the time of the review. The three statistical problems that Goldsmith et al. suggest are: omitted variables, unobserved heterogeneity and data selection. In their paper they present a new estimation of the relationship between unemployment and self-esteem and try to overcome the methodological problems found in the reviewed literature.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4. CROSS-SECTIONAL AND LONGITUDINAL STUDIES Most of the review articles analysed in the last section classify the studies using the dependent variables as criteria. The studies are classified according to the variable on which unemployment has an effect: on morbidity, on mortality, on suicide, on visits to the General Practitioner, on alcohol and cigarette consumption, on traffic accidents, etc. In this section, the studies are classified according to their structure in cross-sectional and longitudinal studies. This classification is important if we take into account the main methodological problem in the analysis of the relationship between unemployment and health: the endogeneity between these two variables. The endogenous relationship between unemployment and health derives from two sources (Gallo et al. 2000): •



There is a reverse causal relationship between these two variables: unemployment may lead to a reduction in health and, at the same time, a decline in health may lead to involuntary job loss. Individual unobserved factors may be associated with both the likelihood of involuntary job loss and health status.

Evidence on the Relationship Between Unemployment and Health

41

If these aspects are not taken into account, biased results may be obtained. In order to overcome the endogeneity problem, panel data structure is more adequate than cross-sectional structure. The use of panel data provides two major benefits for estimation (Carrasco 2002): •



Elimination of estimation bias due to unmeasured heterogeneity. In a panel data design, it is controlled for the non-observable factors by introducing into the equation an individual effect (constant through time) and by using the right estimation method to overcome the problem of correlation between these individual effects and other explanatory variables. Introduction of a dynamic structure in the model. The dynamic structure is important in order to overcome the problem of reverse causality. In longitudinal models, the direction of causality can be identified by defining the right temporal sequence.

Because of these aspects, the study structure is very important. Therefore, in the present review the studies are classified in cross-sectional and longitudinal studies. First, the problem of reverse causality is analysed in greater depth.

4.1. Causation and Selection Hypothesis: Are They Mutually Exclusive? The proportion of persons with health problems among the unemployed is likely to be higher than among the general population (Arrow, 1994). There are two general hypotheses to explain this phenomenon:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• •

Causation hypothesis: unemployment causes health problems. Selection hypothesis: individuals with health problems are more likely to be fired from their work.

It is very likely that both hypotheses are correct. This is, the relation between unemployment and health is not trivial, and probably both directions of the causal relationship play an important role. However, most of the studies are interested in testing just one of the directions of this relationship. Specifically, most of the literature has tested the causation hypothesis. When testing this hypothesis, and in order to avoid having biased estimates, it is important to consider that ill-health also has a certain effect on the probability of being unemployed. The literature gives a wide variety of solutions to this problem. The most practical point of view is the one from Smith (1985). Smith argues the following: ‘...but we should not, I think, become too obsessed with trying to work out whether poor health or unemployment comes first because either way it adds up to a great many people in poor health not having jobs’. Although he is right in the sense that the result is the same whatever the explanation, it is important to know what the causes are in order to overcome the problem with the design of the right policy tools. Therefore, some effort is needed in order to analyse deeply the relationship between these two variables. Authors like Winkelmann and Winkelmann (1998) and Elkeles and Seifert (1993) analysed the problem of double directionality of the causality in the relationship between

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

42

Anton Nivorozhkin and Laura Romeu Gordo

unemployment and health. One of the main lessons of these studies is that the use of panel data helps to understand better the endogenous relationship between these variables. When using cross-sectional data for testing such a causal relation, it is difficult to infer the direction of causation. In a cross-sectional study that tests the causation hypothesis, it is not possible to determine whether an individual’s health contributes to job loss or whether it is job loss that affects health. Furthermore, the presence of unobserved common determinants of health and unemployment may lead to a spurious correlation or omitted variable bias (Winkelmann and Winkelmann (1998)). Panel data is more suitable to overcome the problem of reverse causality due to the dynamic structure and, furthermore, with repeated observations for the same individual over time, it is possible to control for unobserved but time-invariant individual specific effects that are correlated with unemployment (Winkelmann and Winkelmann (1998)). Elkeles and Seifert (1993) test both hypotheses (causation and selection) with longitudinal data. They select individuals who were unemployed for at least six months prior to the survey dates and who were employed the preceding year, and they compare several health indicators at the time of unemployment with the same health indicators from the preceding year when the individuals were employed. The authors suggest that if the causation hypothesis is true, one would expect the health of the unemployed to deteriorate with time. If the selection hypothesis is true, an improvement or at least stability would be expected. Their conclusion is that irrespective of the level of health satisfaction before loss of employment, no overall deterioration occurred after loss of employment. Elkeles and Seifert (1993) also analyse the transition from unemployment to employment. They suggest that although health satisfaction of the re-employed did not improve, it was at quite high level. This supports the idea that healthier persons are more likely to be re-employed than individuals with poor health. They conclude therefore that the poorer health of the unemployed can be explained as a consequence of selection processes. Arrow (1994, 1996) criticises the study of Elkeles and Seifert (1993). The main points of his critic are that the sampling design for testing selection hypothesis is inappropriate, and that the authors base their conclusions on descriptive statistics only. The author uses Cox partial-likelihood regressions and concludes that the hypothesis that bad health constitutes a risk to employment is not true. However, for groups who are vulnerable in the labour market (foreigners and female workers) chronic illness or a long absence from work for health reasons is positively associated with the risk of unemployment. Stewart (2001) analyses the impact of health status on the duration of the unemployment spells. She first analyses the impact of impaired health on the duration of unemployment spells using a duration analysis framework. The results show that impaired health significantly increases the length of unemployment spells. In conclusion, the author points out that this selection bias effect must be taken into account when measuring the impact of unemployment on health status.

4.2. Cross-Sectional Studies As pointed out in earlier sections, cross-sectional designs do not satisfactorily solve the problem of endogeneity between unemployment and health. However, there are several

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Evidence on the Relationship Between Unemployment and Health

43

studies which use this design and try to solve the question of the possible reverse causality with extra tools. Theodossiou (1998) uses data from the BHPS (British Household Panel Study) for his cross-sectional study which analyses the effect of low-pay and unemployment on psychological well-being. He uses logistic regressions which relate the categorical dependent variables (six different measures of mental distress) to the individual’s employment status and other personal and work experience characteristics. The way in which Theodossiou addresses the problem of the double direction of causality is by choosing the questions concerning recent changes in individuals’ psychological state. The author argues that in this way the individuals who are included in the study suffer changes in psychological status due to their employment status rather than the contrary. The results of the study show that unemployment has negative psychological consequences on the individual. Clark and Oswald (1994) also use data from the BHPS for the analysis of the relationship between unemployment and happiness. The question which the authors address is whether or not individuals are choosing to be unemployed. The authors estimate ordered probit models. They regress an individual’s well-being on a set of personal characteristics. Although the authors contemplate the problem of reverse causality, they do not solve it and refer to the work of Warr, Jackson and Banks (1988) which sets the direction of unemploymentunhappiness. The main conclusion is that being jobless is significant and negatively correlated with well-being, and consequently individuals do not choose to be unemployed. Another important conclusion of this study is that long term unemployed show less distress than those who recently lost their jobs. This implies a certain adaptation of the individual to his situation. Hamilton, Merrigan and Dufresne (1997) analyse the relationship between mental health and unemployment by using maximum likelihood, a simultaneous equation of generalised probit. They conclude that there is an endogenous relationship between employment and mental health: higher values of employment are associated with improved mental health, and improved mental health is also associated with a higher index of employability. In his study, Rodríguez (1999) analyses the relationship between marginal employment and self-rated health using German and English data. He uses a logistic regression in which the outcome variable is divided into two groups, one of which includes reports of good or excellent health and the other fair, bad or very bad health. In order to control for a possible reverse causation effect, the model includes adjustments for previous health status. In addition, in order to control for previous experience with job instability, the model includes previous unemployment.

4.3. Longitudinal Studies In this section we review several studies which use a longitudinal design in order to overcome the problem of endogeneity between labour status and health. As an introduction to this review, we present an overview of the main advantages of working with longitudinal data.

Advantages of Using Longitudinal Designs Apart from making it possible to account for unobserved individual heterogeneity and for

Anton Nivorozhkin and Laura Romeu Gordo

44

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

dynamics as we saw above, there are other advantages of working with panel data over the cross-sectional studies (Mátyás and Sevestre (1992)). First, the number of observations is much larger, and this is likely to produce more reliable parameter estimates. At the same time, longitudinal data allows researchers to specify and test more sophisticated models. It also alleviates the problem of multicollinearity, when explanatory variables vary in two dimensions they are less likely to be highly correlated. To summarise, apart from controlling being better for the two sources of endogeneity (reverse causality and non-observed effects), longitudinal data has other advantages which make this structure the most desirable for the analysis of unemployment and health.

Review of Longitudinal Studies Twisk (1997) summarises the main longitudinal models used in the estimation of epidemiological relations, and proposes an estimation method for these models. The author presents four different longitudinal models which are limited to fixed-effects models (see Table 1). Twisk (1997) suggests that in most cases a combination of different models can be the best way of answering a particular epidemiological question. For the estimation of these models, the author recommends the use of GEE (Generalised Estimating Equations). With GEE, the relations between variables of the model at different time points are tested simultaneously. A pooled analysis of cross-sectional (between subjects) and longitudinal (within subjects) relationships is carried out. Therefore, the standardised regression coefficient combines the between-subject and the within-subject effect into one coefficient. Another important advantage of GEE compared to the maximum likelihood approach is that GEE is suitable for the analysis of both continuous and discrete variables. The disadvantage of GEE is that the method does not provide any information on how well the model fits the data. Winkelmann and Winkelmann (1995, 1998) use data from the first six waves (1984-90) of the GSOEP in order to test whether unemployed individuals are satisfied or dissatisfied with their lives in relation to individuals who are out of the labour force and employed individuals. The authors attempt to establish the size of non-pecuniary costs of unemployment relative to the pecuniary costs. Winkelmann and Winkelmann conclude from their descriptive analysis that there is a decrease in satisfaction for individuals who were employed and are now unemployed, and there is also an increase in satisfaction for people who were unemployed and are now employed. The authors suggest that these symmetric effects support the causation hypothesis. Another interesting conclusion is that for individuals who went out of the labour force the effect on life satisfaction was lower than for the individuals who went into unemployment. There is little variation in the life satisfaction indicator for the individuals who remained unemployed. This means that individuals did not get used to their status. For the econometric analysis of the relationship between employment status and life satisfaction, the authors use a limited dependent panel model after reducing the dimension of the explained variable into satisfied/dissatisfied dichotomy. They obtain logit estimates for five different models: • •

The first model is a standard logit regression for pooled data. The second model is the fixed effects logit model.

Evidence on the Relationship Between Unemployment and Health • •

45

The third and the fourth models explore the robustness of the fixed effects logit results under modified specifications. The last model introduces age-specific effects of unemployment and leaving the labour market. Table 1. Longitudinal Models Proposed by Twisk.

Model 1: Actual Values Yit=β0+∑Jβ1jXijt+∑Kβ2kZikt+

Model 2: Timelag Yit=β0+∑Jβ1jXijt

Model 3: Changes (Yit-Yit-

Model 4: Autoregression Yit=β0+∑Jβ1jXijt+β

β3t+∑Mβ4mGim+εit

-1+...

J 1)=β0+∑ β1j(Xijt-Xijt-1)

2Yit-1 +...

+... Xijt = independent variable j of Xijt-1 = Yit-1 = observation of Yit-1 = observation subject i at time t. independent subject i at time t-1. of subject i at time tZikt = time-dependent covariate variable j of 1. subject i at time tβ2 = autoregression k of subject i at time t. 1. β3 = regression coefficient of coefficient. time. Gim = time-independent covariate m of subject i. Model 1: Actual Values The coefficients β1j show the magnitude of the relation between the longitudinal development of Yit and the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

development of predictor variables.

different

Model 2: Timelag In some situations, it can be difficult to distinguish between cause and effect. To tackle part of this problem a model can be used in which the temporal sequence of cause and effect is built in.

Model 3: Changes In the combined analysis the longitudinal withinsubjects relationships will be more or less overruled by the cross-sectional between-subjects relations when the variation in actual values betweensubjects exceeds the changes over time withinsubjects. Because of this limitation, a model can be used in which the crosssectional part is removed from the analysis. The cross-sectional part is removed from the analysis. Not for modelling changes dichotomous outcomes.

Model 4: Autoregression Some times in longitudinal analysis, the results can be influenced by the relative stability of the related variables. To correct for the relative stability of the related variables an autoregressive model can be used.

Pool together longitudinal and Pool together cross-sectional relationships into longitudinal and one regression coefficient. cross-sectional relationships into one regression coefficient. Also for dichotomous outcome Also for Not for modelling changes Also for dichotomous variables. dichotomous dichotomous outcomes. outcome variables. outcome variables.

Anton Nivorozhkin and Laura Romeu Gordo

46

When comparing the pooled logit model with the fixed effects model the authors conclude: • The fixed effect model is the better model. • The detrimental effect of unemployment on satisfaction persists after fixed effects are taken into account. They also conclude that the non-pecuniary costs of unemployment by far exceed the pecuniary costs. Goldsmith et al. (1996) analyse the relationship between unemployment and self-esteem. They first review cross-sectional and longitudinal studies which link self-esteem and unemployment. The conclusion of this review is that the relationship between these two variables is unclear. The authors identify three primary problems: •

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• •

Omitted variable bias. By stratifying the data (usually by gender), the wide variety makes it difficult to control individual specific personal characteristics that influence self-esteem. This may cause biased estimators. Unobserved heterogeneity. Data selection (treatment of labour force dropouts). Most of the studies eliminate from the sample individuals who are out of the labour force. However, this group of individuals must also be taken into account since they suffer a different effect on their self-esteem than the unemployed.

The authors offer a new approach which yields to unbiased estimates of joblessness on self-esteem. They used data from the National Longitudinal Survey of Youth to control for the employment history of the individuals between the years 1978-1980. The self-esteem data is from 1980. To solve the problem of unobserved heterogeneity, factors that alter the likelihood of becoming jobless and that are also expected to contribute to self-esteem were included in the regression. In order to analyse the relationship between self-esteem and the explanatory variables, the authors use an Ordered Probit. The main conclusion is that individuals exposed to a completed spell of joblessness in the most recent subperiod have a significantly lower level of self-esteem than comparable individuals who were employed throughout the total sample period. Another interesting conclusion is that there are no significant differences between the effects of being unemployed and the effects of being out of the labour force. Goldsmith et al. (1996) discompose the impact of joblessness on selfesteem in anxiety, alienation and depression. They conclude that joblessness damages selfesteem by generating feelings of depression. Winefield et al. (1991) use cross-sectional and longitudinal comparisons in order to test the psychological impact of unemployment and unsatisfactory employment. They compare four groups of young individuals who were surveyed annually over a period of seven years and who initially did not differ on any of the psychological measures of well-being. These four groups are: satisfied employed, dissatisfied employed, unemployed and full-time tertiary students. The main finding of this study is that dissatisfied employed were generally as worse off as the unemployed.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Evidence on the Relationship Between Unemployment and Health

47

Beale and Nethercott (1985; 1988) use general practice records of patients and their relatives who suffered a factory closure and the general practice records of a control group to analyse the relationship between unemployment and health. The main conclusion is that men who suffered a factory closure visit general practitioners more often for episodes of illness requiring four or more consultations than those who were in secure employment. One important conclusion of this study is that the increase in morbidity began two years before the factory closure. The later finding points to a certain anticipatory effect. Kessler et al. (1989) combine cross-sectional and longitudinal analyses to obtain conclusions about the effects of employment and reemployment on distress. They carry out a cross-sectional analysis (using logistic regression equations) and the coefficients show that all distress measures were significantly elevated among the currently unemployed compared to the stably employed. The results show that distress is not negatively associated with reemployment as expected. One possible explanation for this result is that people who are highly distressed by job loss are willing to accept any job whatever. Another explanation could be that extreme distress is associated with a more intensive job search. Kessler et al. (1989) also estimate the effect of reemployment on subsequent change in emotional functioning. They suggest that reemployment reduces the average symptom level, and also the risk of experiencing symptoms potentially severe enough to warrant professional intervention. The authors also suggest that the reemployed may not have returned completely to their emotional state prior to job loss; complete recovery comes after a year of being reemployed. They also analyse whether the recovery is different depending on the kind of job that the individual gets. The results suggest that reemployed are, in the short term, happy to have any job. Gerlach and Stephan (1996) use German data (GSOEP data for the waves from 1984 to 1993) in order to analyse the effect of unemployment on unhappiness. They use a fixed effects model with individual and time effects for men and women of three age classes. They conclude that, from the groups analysed, men 30 to 49 years old suffer most from unemployment, and women in the age group 50 years and older suffer least from unemployment. Bardasi and Francesconi (2000) analyse the effect of non-standard employment on mental health using UK data. The authors argue that many of the endogeneity problems arise from the presence of individual-specific fixed endowments which are correlated with each other and with unobserved endogenous inputs. They carry out three kinds of estimations: an OLS model (level analysis), a first-differenced fixed-effects model and a two-period lagged first-difference model (since the authors argue that a first-differenced model does not satisfactorily solve the endogeneity problem). The authors conclude that a two-period lagged first-differences model yields consistent estimates under some orthogonality conditions. The results show that there is evidence of only a limited effect of all types of flexible employment on the mental health scores. Graetz (1993) uses Australian data to analyse the health consequences of employment and unemployment. He first carries out a cross-sectional analysis, using one-way analysis of variance (ANOVA) to test for statistically significant group differences. The conclusion is that employment status has an important effect on psychological health. For those who are employed, quality of employment has also an important impact. In the longitudinal analysis, pairwise comparisons are used to assess changes of the variables over time, while betweengroup comparisons are used to test for predisposing differences. From this analysis the author

Anton Nivorozhkin and Laura Romeu Gordo

48

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

concludes that the health consequences of employment depend mainly on the quality of work, and not on prior health differences that may predispose some people to find their jobs satisfying or dissatisfying. The general conclusion from this study is that unemployment is not always worse that employment; it depends on the quality of work. Wadsworth, Montgomery and Bartley (1999) in their study test the relationship between unemployment and socio-economic and health capital acquired by age 33 years, and the association of pre-labour market factors with health and socio-economic capital at 33 years. Logistic regressions were used for the analysis, and it was controlled for the effect of the childhood and adolescent factors which are known to influence acquisition of capital. The authors conclude that even six years after their last experience of unemployment, the individuals were more likely than others to be found in socially and materially less favourable conditions than others of the same background and educational attainment. In addition, men who had previous long periods of unemployment had adopted less favourable health behaviour. Gallo et al. (2000a) use data from the first two waves of the American Health and Retirement Survey (HRS) to investigate the health effects of involuntary job loss among older workers. The dependent variables used in the model are physical functioning and mental health. And the independent variables are a dummy variable for involuntary job loss, measures of baseline health and several socio-economic variables. In order to estimate this model they used OLS. The authors suggest that endogeneity can derive from two sources: bad health may lead to involuntary job loss, and unobserved factors may be associated with both the likelihood of involuntary job loss and with follow-up health status. The results of the estimations indicate that longer unemployment spells lead to poorer physical functioning and mental health at follow-up. Furthermore, reemployment was positively associated with physical functioning and mental health. Gallo et al. (2000b) use data from the GSOEP in order to analyse the relationship between job displacement and self-assessed health. The study compares displaced workers and continuously employed. They use a residualized change model, in which they control for the baseline self-assessed health and for standard demographic characteristics. The authors first estimate the model using OLS, and later use GEE. The estimation results suggest that the association between displacement and follow-up self-assessed health is not significant. Clark et al. (2001) use the first eleven waves of the GSOEP in order to analyse the psychological impact of past unemployment. To address the problem of simultaneity, the authors use a fixed-effects model. The main conclusions of this analysis are: ƒ ƒ ƒ ƒ

Current unemployment is associated with lower levels of subjective well-being. Past unemployment reduces the well-being of those who are currently employed. Well-being effect of current unemployment is attenuated for those who have experienced more unemployment in the past. Another interesting conclusion of their study is that the larger the drop in life satisfaction from employment to unemployment, the smaller is the probability that the individual remains unemployed one year later.

Evidence on the Relationship Between Unemployment and Health

49

5. ADAPTATION TO UNEMPLOYMENT In the previous sections we reviewed the literature on the effect of losing a job. However, whether or not the individual adapts himself to his new role has also been an important topic in the literature. Some authors argue that unemployed get used to their situation, and after a few months their health and well-being is stabilised at a low level and does not suffer significant variations. This view is explained by the stages model (reviewed in section 1.1). This model tells us that, after suffering a shock, the individual is still optimistic and looks actively for a job. Then, when all the efforts fail he becomes pessimistic and suffers distress. Finally, the individual becomes fatalistic and adapts to his new role, but with lower wellbeing. However, other studies conclude that unemployed do not adapt themselves to their situation and that the drop in health and well-being increases as the length of the unemployment spell increases. Next, we review briefly some studies that defend one and the other position.

5.1. Adaptation Evidence Warr and Jackson (1987) examine changes in mental health associated with continuing unemployment. They analyse which factors may influence this process of adaptation. In order to measure adaptation they used changes in reported health. The factors that were introduced in the analyses in order to test whether they were affecting adaptation or not were: commitment to having a job, availability of money, age, social relationships and health condition. The results show that there is a certain adaptation process. The authors classify the situation of long-term unemployed in the following categories: • •

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.



Constructive adaptation: a number of individuals develop interests and activities outside the labour market. Resigned adaptation: the individuals make some improvements. However, these small improvements are accompanied by negative changes in other aspects (reduced aspiration, autonomy and competence). Despair: some individuals suffer from low levels in all aspects: reduced well-being and also low levels of aspiration, autonomy and competence.

The results show that stronger commitment to having a job and being middle aged are associated with less adaptation. Clark and Oswald (1994) in their analysis of the relationship between unemployment and happiness conclude that long time unemployed show less distress than those who recently lost their jobs. Therefore they conclude that there is a certain adaptation process. Clark et al. (2001) in their analysis of the psychological impact of past unemployment, conclude that there is a certain habituation of the individual to the unemployment experience. The authors conclude that the effect of current unemployment on well-being is attenuated for those who had unemployment experiences in the past.

50

Anton Nivorozhkin and Laura Romeu Gordo

Eales (1988) analyses the relationship between unemployment and depression and anxiety. The author concludes that most of the disorders arising after job loss are developed within three months. This result like the previous ones supports the step theory.

5.2. Non-Adaptation Evidence

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Frese and Mohr (1987) in their study of the effect of prolonged unemployment on depression argue that financial problems and depression increase and hope decreases with prolonged or repeated unemployment. The authors use analyses of covariances to show that the long-term exposure to the daily hassles of unemployment (financial problems and disappointments) is what increases depression. Summarising, unemployed do not feel better when the period of unemployment increases. On the contrary, the problems are accentuated. Winkelmann and Winkelmann (1998) in their descriptive analysis of the relationship between unemployment and happiness conclude that unemployed do not get used to their situation. Wadsworth et al. (1999) in their study do not answer directly the question of whether or not unemployed get used to their situation. Instead, they measure the irreversible effects of persisting unemployment on the individual. They conclude that there is a depreciation of socio-economic capital for the individuals who have been unemployed early in their careers. Furthermore, they are more likely to adopt less favourable health behaviour. Some studies also analyse how the probability of reemployment is affected by distress suffered by unemployed. Kessler et al. (1989) use panel data in order to estimate the effects of distress on the probability of reemployment. The results show that distress is not negatively associated with reemployment. The authors argue that this may be due to the fact that individuals who are highly distressed by job loss are willing to accept the next job that comes along. Another possible explanation for this result is that extreme distress is associated with more intense job search. They also analyse the impact of reemployment on subsequent change in emotional functioning. The evidence showed that reemployment reduced the average symptom level. Furthermore, the authors conclude that the recovery was not significantly different depending on the quality of the job.

5.3. Anticipation Evidence Studies which focus on plant closure identify an anticipation effect on health when announcing the closure. Beale and Nethercott (1985) analyse the consequences on health when a plant closes. They conclude that there is a significant increase in the number of times that the employees and their spouses consult their doctors. They also conclude that this increase in morbidity began two years before the plant closure, at a time when it was clear to the families that the plant was going to be closed.

Evidence on the Relationship Between Unemployment and Health

51

6. EXPLANATORY VARIABLES The objective of this section is to review the main covariates used in the models which analyse the effects of unemployment on health and well-being. The reviewed studies include not only employment status variables but also other kind of control variables which we review and classify in this section and which are presented in Table 2. The dependent variables can be classified in mental health indicators, physical health indicators, and life satisfaction. There are some control variables which are included almost in every model. These control variables are mainly: age, sex, race, marital status, education and number of children (or members) in the household. There are some studies (Theodossoiou (1998), Rodríguez (1999) and Graetz (1993)) which also include housing conditions. In the analysis of Graetz (1993) the birthplace is also included. Clark and Oswald (1994) control for whether or not the region where the individual lives is a region of high unemployment. Gallo et al. (2000) introduce in their analysis a dummy variable indicating whether the individual is from East or from West Germany. These control variables are assumed to be important determinants of mental and physical health and of life satisfaction, and therefore they are included in almost every model. Table 2. Explanatory Variables of the Reviewed Studies Study • Gore (1978)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• Warr and Jackson (1985)

Dependent Variables • • •

Depression Illness symptoms Cholesterol

Explanatory Variables

(Social support) • Individual perception of wife, friends and relatives as supportive or unsupportive. • Frequency of activity outside the home. • Respondent’s perceived opportunity for engaging in social activities which are satisfying and which allow him to talk about his problems. (Perceived economic deprivation) • Income comparisons with friends and neighbours. • Difficulty in ‘getting by’ financially. • Psychological ill- (Employment situation) health. • Employment commitment. • Reported health. • Job seeking. • Reported health (Financial situation) change. • Income change. • Financial strain. • Number of dependants. • Money problems. • Financial support. (Personal situation) • Non-money problems. • Emotional support. • Social contact. • Institution membership.

Anton Nivorozhkin and Laura Romeu Gordo

52

Table 2. (Continued). • Dooley et al. (1987)

• PERI symptoms (a 25- (Control variables) item psychological • Age. symptom checklist from the • Sex. Psychiatric Epidemiology • Ethnicity. Research Instrument). • Socio-economic status. • CESD (20-items on the (Life event variables) Centre for Epidemiological • Desirable and undesirable job events. Studies Depression Scale) • Desirable and undesirable other (non-job events). (Social support and help utilisation) • Variable for the job sphere. • Variable for the non-job sphere. (Job security) • Perceived job security. • Objective job security.

• Frese and Mohr (1987)

• Depression (German translation of Zung’s scale)

• Eales (1988)

• Kessler et al. (1989)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• Kessler et al. (1989)

• Financial difficulties. • Hope for control. • Internal/External control. • General activity level. • Psychiatric disorder (Personal characteristics) (present state examination) • Age. • Marital status. • Life events and difficulties schedule. (Employment situation) • Length of unemployment. • Occupational status. (Controls) • Distress at time 2. • Age. • Sex. • Education. • Race. • Marital Status. (others) • Distress at time 1. • Reemployment. • Probability of (Controls) becoming reemployed. • Age. • Sex. • Education. • Race. • Marital Status. (others) Distress at time 1.

Evidence on the Relationship Between Unemployment and Health

53

Table 2. (Continued).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• Graetz (1993)

• Psychiatric disorder (Demographic attributes) (GHQ). • Sex. (for employed, studying and • Age. unemployed respondents) • Birthplace. • Marital status. (Living arrangements) • Lives with parents. • Nature of occupancy. (Socio-economic status) • Income. (Labour force experiences) • Job spells. • Looking for work spells. • Job satisfaction. • Duration of unemployment. • Arrow • Employment duration (Personal characteristics) (1994) • Age. • Gender. • Nationality. • Chronic illness. • Sick-leave > 42 days. (Employment situation) • Past unemployment. • Employment sector. • Firm size. (Personal characteristics) • Clark and • Mental well-being Oswald • Sex. (1994) • Age. • Age*Age. • Education. • Race. • Marital status. • Number of children. • Health. • Region (region with high unemployment). (Employment status) • Dummy variables indicating the employment status of the individual. (Personal characteristics) • Winkelm • Life satisfaction ann and • Age. Winkelmann • Age*Age. (1995) • Marital status. • Good health. (Income) • Log of the family income. (Employment status) • Variables indicating whether the individual is unemployed, out of the labour force, selfemployed, part-time employed). (Employment status of the partner) • Variables indicating whether the partner is unemployed, out of the labour force, selfemployed, part-time employed).

Anton Nivorozhkin and Laura Romeu Gordo

54

Table 2. (Continued) • Gerlach and Stephan (1996)

• Goldsmit h et al. (1996)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• Winkelm ann and Winkelmann (1997)

• Theodoss iou (1998)

• Life satisfaction (per (Personal characteristics) sex and age groups) • Marital status. • Degree of disability. • Satisfaction with health. (Income) • Household income per capita. (Employment status) • Variables indicating whether the individual is full-time, part-time employed or unemployed). (Personal characteristics) • Level of self-esteem. • Perceived personal locus of control. • Position in the life cycle. • Marital status. • Sex. • Race. • Presence of young dependants. (Income) • Wage on most recent job. • Accumulated financial assets. (Employment spell) • Being currently unemployed or out of the labour force. • Duration of current unemployment or time spent out of the labour force. • Past time spent in spells of unemployment, out of the labour force, or time in both states of joblessness. • Skills acquired. (Personal characteristics) • Life satisfaction • Age. • Age*Age. • Marital status. • Good health condition. (Employment situation) • Current labour market status (unemployed / out of the labour force). • Duration of the unemployment spell. • Duration*Duration. (Interaction effects between employment status and age groups) • Psychological well- (Control variables) being • Age. • Sex. • Ethnicity. • Marital status. • Number of children. (Personal characteristics) • Education. • Housing characteristics. (Employment status) • Dummy variables indicating whether the individual is unemployed, low paid, high paid, not in the labour force.

Evidence on the Relationship Between Unemployment and Health

55

Table 2. (Continued) • Rodrigue z (1999)

• Bardasi and Francesconi (2000)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• Gallo al. (2000)

et

• Perceived health status (Fixed personal characteristics) 1993. • Age. • Sex. (Other individual and household characteristics) • Education. • Marital status. • Household income. • Home ownership. • Number of family members. (Background risk factors) • Previous health status. • Previous unemployment. • Time spent on unpaid housekeeping work. (Employment status 1992) • Full-time employed (with permanent, with fix-term or without contract). • Working 20 to 30 hours per week (with permanent, with fix-term or without contract). • Working less than 20 hours per week (with permanent, with fix-term or without contract). • Unemployed. • Housewives/husbands. • Students. • Retired. • Other. • Mental health (GHQ) • Number of cigarettes smocked. • Education. • Work experience. • Types of non-standard employment (e.g. being on a fixed term contract working long hours, being on rotating shifts). • Follow-up self- (Personal characteristics) assessed health (SAH) • Age. • Male gender. • Marital status. • Education. • East German. • Baseline SAH. (Employment status) • Job displacement. • Blue-collar occupation. • Hourly wage. • Services (if baseline employment in services industry).

Anton Nivorozhkin and Laura Romeu Gordo

56

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 2. (Continued) • Gallo al. (2000)

et



Physical functioning.

• Gallo al. (2000)

et



Mental health.

• al.

et



Life satisfaction

Clark

(Personal characteristics) • Age. • Education. • Male. • White. • Marital status. (Income) • Labour income. • Non-housing net worth. (Health) • Hypertension. • Cancer. • Heart disease. • Heavy smoker. • Heavy drinker. • Obese. • Baseline physical functioning. (Employment status) 3 different regressions using for each one, one of the following explanatory variables: involuntary job loss, employment duration and reemployment. (Personal characteristics) • Age. • Education. • Male. • White. • Marital status. (Income) • Labour income. • Non-housing net worth. (Employment status) • 3 different regressions using for each one, one of the following explanatory variables: involuntary job loss, employment duration and reemployment. (Employment status) • Unemployed at time t. • Past unemployment. (Individual and other characteristics for males and females) • Age. • Age Square. • Income. • Years of education. • Disability. • Out of labour force/ Part-time/ Selfemployed. • Number of children. • Married/ Separated/ Divorced. • House owner. (Interaction terms) Unemployment*Past unemployment. Current and past unemployment with: age, education and children.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Evidence on the Relationship Between Unemployment and Health

57

Other important group of explanatory variables are the ones that control for the financial situation and for the socio-economic status of the individual. The main variables included are: individual income, household income, financial strain, wage on most recent job, and income change (after a change in the labour status). An interesting variable included in the study of Gallo et al. (2000) is the non-housing net worth, since the financial strain for individuals who have lost their jobs may be very different depending on their net worth. Gore (1978) introduces the perceived economic deprivation. The variables that he uses are comparative income of friends and neighbours and difficulty in ‘getting by’ financially. These control variables are interesting since they indicate how the individuals perceive own financial situation in comparison to their relevant ones. These indicators of the relative financial situation of the individual, when available, may be more relevant that absolute financial indicators when explaining mental health. Health indicators are also included in the models which analyse follow-up health. In these models, the baseline health indicators are introduced in order to control for the previous health condition of the individual. Kessler et al. (1989) use distress at time 1 as a control variable in order to explain distress at time 2. Gallo et al. (2000) use the baseline self-assessed health to explain the follow-up self-assessed health. Some studies include social support. Gore (1978) in his analysis introduces the individual perception of wife, friends and relatives as supportive and unsupportive, frequency of activity outside the home and respondent’s perceived opportunity for engaging in social activities which are satisfying. Warr and Jackson (1985) introduce as control variables emotional support, social contact and institution membership. For the employment related information different indicators are used. These variables can be classified in the following categories: being unemployed or not, characteristics of the job (if the individual is employed), past spells of unemployment and duration of the unemployment spells. Graetz (1993) introduces job satisfaction as a explanatory variable. Dooley et al. (1987) introduce perceived and objective job security as explanatory variable. Winkelmann and Winkelmann (1995) also introduce the employment status of the partner as explanatory variable. Another interesting variable is employment commitment which was introduced by Warr and Jackson (1985) in their analysis. Only few studies introduce interaction effects between market labour transitions or status and other variables. Winkelmann and Winkelmann (1998) introduce in their analysis interaction effects between employment status and age groups. Dooley et al. (1987) introduce in their model interaction effects between undesirable job and non-job events and other variables. By introducing interaction effects it can be tested which are the factors which moderate or stress the effect of unemployment. In the typical case of the interaction effect between unemployment and age, we can test which age groups are more affected by unemployment. It is also interesting to interact unemployment with the variables which indicate social support. In this way it can be tested whether or not social support plays an important role in moderating the effect of unemployment. Another interesting interaction is the one between unemployment and satisfaction with the last job. The idea is that the dissatisfaction of the individual with his last job can be a moderator for the effect of unemployment. The unsatisfied individuals may be less affected by unemployment.

58

Anton Nivorozhkin and Laura Romeu Gordo

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Arrow J.O. (1994) The influence of health on unemployment in Germany: a duration model. DIW-Vierteljahresbericht. Hefte für Wirtschaftsforschung, 1333-138. Berlin. Arrow J.O. (1996) Estimating the influence of health as a risk factor on unemployment: a survival analysis of employment durations for workers surveyed in the GSOEP. Soc. Sci. Med., vol.42(12), 1651-1659. Bardasi E. and Francesconi M. (2000) The effect of non-standard employment on mental health in Britain. ISER Working Papers. Paper 2000-37. Colchester: University of Essex. Beale N. and Nethercott S. (1985) Job-loss and family morbidity: a study of a factory closure. Journal of the Royal College of General Practitioners, vol.35, 510-514. Beale N. and Nethercott S. (1988) The nature of unemployment morbidity. Journal of the Royal College of General Practitioners, vol.38, 200-202. Brenner M.H. (1977) Health costs and benefits of economic policy. International Journal of Health Services, vol. 7, 581-623. Brenner M.H. (1979) Mortality and the economy: A review, and the experience of England and Wales, 1936-1976. Lancet, 568-573. Brenner M.H. and Mooney A. (1983) Unemployment and health in the context of economic change. Soc. Sci. Med., vol.17, 1125-1138. Brenner M.H. (1987) Economic change, alcohol consumption and heart disease mortality in nine industrialized countries. Soc. Sci. Med., vol.25, 119-132. Brenner M.H. (1987) Relation of economic change to Swedish health and social well-being, 1950-1980. Soc. Sci. Med., vol.25, 183-195. Carrasco R. (2002) Estimation of dynamic discrete choice models for panel data. PACO Training Session. Luxemburg June 2002. Clark A.E. and Oswald A.J. (1994) Unhappiness and unemployment. The Economic Journal, vol.104, 648-659. Clark A.E., Georgellis Y. and Sanfey P. (2001) Scarring: the psychological impact of past unemployment. Economica, vol.68, 221-241. Dooley D., Rook K. and Catalano R. (1987) Job and non-job stressors and their moderators. Journal of Occupational Psychology, vol.60, 115-132. Eales M.J. (1988) Depression and anxiety in unemployed men. Psychological medicine, vol.18, 933-945. Elkeles T. and Seifert W. (1993) Unemployment and health impairments. Longitudinal analyses for the Federal Republic of Germany. European Journal of Public Health, vol.3(1), 28-37. Ezzy D.(1993) Unemployment and mental health: a critical review. Soc. Sci. Med., vol.37(1), 41-52. Frese M. and Mohr G (1987) Prolonged unemployment and depression in older workers: a longitudinal study of intervining variables. Soc. Sci. Med., vol.25(2), 173-178. Gallo W.T., Bradley E.H., Siegel M. and Kasl S.V. (2000a) Health effects of involuntary job loss among older workers: findings from the health and retirement survey. Journal of Gerontology, SOCIAL SCIENCES, vol.55(3), 131-140. Gallo W.T., Bradley E..H. and Kasl S.V. (2000b) The effect of job displacement on subsequent health. Paper presented in the GSOEP‘ Users Conference 2000. Berlin.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Evidence on the Relationship Between Unemployment and Health

59

Gerlach K. and Stephan G. (1996) A paper on unhappiness and unemployment in Germany. Economic Letters, vol.52, 325-330. Goldney R.D. (1997) Unemployment and health: a re-appraisal. Int. Arch. Occup. Environ. Health 70, 145-147. Goldsmith A.H., Veum J.R. and Darity W.Jr. (1996) The impact of labour force history on self-esteem and its component parts, anxiety, alienation, and depression. Journal of Economic Psychology, vol.17, 183-220. Gore S. (1978) The effect of social support in moderating the health consequences of unemployment. Journal of Health and Social Behaviour, vol.19, 157-165. Graetz B. (1993) Health consequences of employment and unemployment: longitudinal evidence for young men and women. Soc. Sci. Med., vol.36(6), 715-724. Hamilton V.H., Merrigan P. and Dufresne E. (1997) Down and out: estimating the relationship between mental and unemployment. Health Economics, vol.6, 397-406. Jahoda M. (1982) Employment and unemployment. Cambridge University Press. Cambridge. Jin R.L., Shah C.P. and Svoboda T.J. (1995) The impact of unemployment on health: a review of the evidence. Canadian Medical Association Journal, vol.153(5). Kasl S. and Jones B.A. (1998) The impact of job loss and retirement on health. In Social Epidemiology. Berkman L.F. and Kawachi I. eds. Oxford University Press. Kessler R.C., Turner J.B. and House J.S. (1989) Unemployment, reemployment, and emotional functioning in a community sample. American Sociological Review, vol.54, 648-657. Mátyás L. and Sevestre P. (eds.) (1992) The Econometrics of Panel Data. Kluwer Academic Publishers: The Netherlands. Murphy G.C. and Athanasou (1999) The effect of unemployment on mental health. Journal of Occupational and Organizational Psychology, vol.72, 83-99. Rodríguez E. (1999) Marginal employment and health in Germany and the UK: does unstable employment predict health? Wissenschaft Zentrum Berlin DP FS I 99-203. Schwefel D. (1986) Unemployment, health and health services in German-speaking countries. Social Science and Medicine, vol.22(4), 409-430. Shortt S.E.D. (1996) Is Unemployment Pathogenic? A review of current concepts with lessons for policy planners. International Journal of Health Services, vol.26(3), 569-589. Smith R. (1985) “Bitterness, shame, emptiness, waste”: an introduction to unemployment and health. British Medical Journal, vol.291, 1024-1027. Smith R. (1985) Gissa job: the experience of unemployment. British Medical Journal, vol.291, 1263-1266. Smith R. (1985) He never got over losing his job: death on the dole. British Medical Journal, vol.291, 1492-1495. Smith R. (1985) I couldn’t stand it any more: suicide and unemployment. British Medical Journal, vol.291, 1563-1566. Smith R. (1985) I’m just not right: the physical health of the unemployed. British Medical Journal, vol.291, 1626-1629. Smith R. (1985) I feel really ashamed: how does unemployment lead to poorer mental health. British Medical Journal, vol.291, 1409-1412. Smith R. (1985) We get on each other’s nerves: unemployment and the family. British Medical Journal, vol.291, 1707-1710.

60

Anton Nivorozhkin and Laura Romeu Gordo

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Smith, R. (1985) What’s the point. I’m no use to anybody: the psychological consequences of unemployment. British Medical Journal, vol.291. Stewart J.M. (2001) The impact of health status on the duration of unemployment spells and the implications for studies of the impact of unemployment on health status. Journal of Health Economics, vol.20, 781-796. Theodossiou I. (1998) The effects of low-pay and unemployment on psychological wellbeing: a logistic regression approach. Journal of Health Economics, vol.17, 85-104. Twisk J.W.R. (1997) Different statistical models to analyze epidemiological observational longitudinal data: an example from the Amsterdam growth and health study. Int. J. Sports Med., vol.18, 216-224. Wadsworth M.E.J., Montgomery S.M. and Bartley M.J. (1999) The persisting effect of unemployment on health and social well-being in men early in working life. Soc. Sci. Med., vol.48, 1491-1499. Warr P. and Jackson P (1985) Factors influencing the psychological impact of prolonged unemployment and re-employment. Psychological Medicine, vol.15, 795-807. Warr, P. (1987) Work, Unemployment and Mental Health. Oxford University Press: New York. Warr P. and Jackson P. (1987) Adapting to the unemployed role: a longitudinal investigation. Soc. Sci. Med., vol. 25(11), 1219-1224. Warr P.B., Jackson P.R. and Banks M. (1988) Unemployment and mental health: some British studies. Journal of social issues, vol.44, 47-68. Wilson S.H. and Walker G-M. (1993) Unemployment and Health a Review. Public Health, vol.107, 153-162. Winefield A.H., Tiggemann M, and Winefield H.R. (1991) The psychological impact of unemployment and unsatisfactory employment in young men and women. British Journal of Psychology, vol.82, 473-486. Winkelmann L. and Winkelmann R. (1995) Unemployment: where does it hurt? Centre for Economic Policy Research DP. 1093. Winkelmann L. and Winkelmann R. (1998) Why are the unemployed so unhappy? Economica, vol.65, 1-15.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 3

SPANISH UNEMPLOYMENT AND THE LADDER EFFECT Fabrice Collard1, Raquel Fonseca2 and Rafael Munoz31 1

University of Toulousse, Toulousse, France 2 RAND Corporation, France 3 World Bank, Spain

ABSTRACT

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

This paper aims to examine to what extent a "ladder" effect may contribute to explain changes in unemployment in Spain. The "ladder" effect arises when highlyskilled workers who do not find a job that matches their skills, accept jobs that were previously occupied by less qualified staff. We develop a dynamic general equilibrium model. The model is then calibrated for the Spanish economy. Our results replicate the observed decline in the ratio of high- to low-skilled vacancies, and explain how firms substitute high for low-skilled employment. These results also suggest that in the Spanish case, ladder effect can be better explained by increases in training costs interpreted as a biased-shock against low-skilled workers.

JEL classification: E24, J64 Keywords: Matching models, low-skilled unemployment, labour supply and labour demand mismatch

1

Fabrice Collard, GREMAQ and University of Toulousse. Raquel Fonseca, RAND Corporation –Correspond author: [email protected] and Rafael Muñoz, World Bank.

62

Fabrice Collard, Raquel Fonseca and Rafael Munoz

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION Since the late seventies, there has been a reduction in the demand for low-skilled relative to highly skilled workers in many industrialized countries. This phenomenon was usually accompanied by a deterioration of economic conditions of low–skilled workers. In Europe, this problem has manifested itself in a large increase of the unemployment rate, mainly due to a decrease in the demand of unskilled workers (OECD (1994), Berman et al. (1998)). In the U.S. and U.K., this phenomenon was associated with greater wage inequality across groups of different skills (see Krugman (1994), Acemoglu (1999, 2002)). In Europe, part of this is due to the increase in the average education level of workers (Wasmer et al. (2007)). The large increase in the unemployment rate of low-skilled European workers may be explained by both supply and demand on the labour market. A plausible explanation is job competition between highly and low-skilled workers. By applying to low-skilled jobs, highly skilled workers increase their probability to get a job, hence displacing low-skilled ones. This form of skill upgrading has been defined as the "ladder effect". But a ladder effect may also arise from the demand side and not only from supply-side induced crowding-out of lowskilled workers. Indeed, it may well be that firms wish to hire highly skilled workers for lowskilled jobs to avoid large training cost (see Thurow (1975)), or because they have a higher productivity (see Gautier (2002)). Yet, another demand mechanism arises from "skill mismatch" resulting from the presence of relative wage rigidities as the economy is hit by biased technological shocks. In the end, evidence that highly skilled workers may occupy low-skilled positions has been documented in several countries (see Van Ours and Ridder (1995), and Muysken and Ter Weel (1998), for the Netherlands, and Green et al. (1999), for the UK). One of the most problematic issues in the Spanish labour market is the high persistence of the unemployment rates for the low-skilled workers. From the beginning of the 80’s and especially from the middle of the 80’s, the average level of education in the Spanish labour force has grown considerably (see Blanco (1997)). At the same time, the unemployment rate of the low–skilled workers rose too. For instance, illiterate workers and those with primary and low-secondary schooling have experienced a marked increase in their respective unemployment rate (from 19% in 1988 up to 23% in 1996 and 21% in 2004). Over the same period, the proportion of low-skilled jobs filled by highly skilled workers, a measure of the ladder effect rose from 9.8% in 1988, 15.1% in 1996 to 14.1% in 2004.2 There is evidence that supply-side causes to the ladder effect may be important in Spain. For instance, Alba-Ramirez (1993) finds that Spanish highly skilled workers occupying lowskilled jobs are mainly young skilled workers, without experience, who need a first job in order to obtain on-the-job training. After some time, they become the main job turnover group, which implies they move into jobs that require higher educational levels. Garcia2

We use our own computations with the Linked Labour Population Survey (Encuesta de Población Activa enlazada, EPA hereafter). Our ladder effect indicator is computed with this formula

N lh Lh / ( N lh Lh + N l Ll )

where N is employment and L is labour force. The superscript denotes the education: high-skilled (h) and low-skilled (l). The subscript indicates the occupation job. Education: h=superior and upper secondary and l=low secondary, primary and without studies. Occupation h=range 1-3 in EPA classification and l=range 0 and range 4 to 9 from EPA classification.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Spanish Unemployment and the Ladder Effect

63

Serrano and Malo (1996) conclude that in Spain, education is a substitute for on-the-job training. Hence, Spanish firms may hire highly skilled workers to reduce investment in jobspecific human capital formation. The key point of this paper is that the incentive to do so will be high when the economy is hit by biased technological shocks and this can lead in turn to over-education and the ladder-effect. It turns out that Spain is a country with a low level of on-the-job training (OECD, 2004) suggesting that firms may be shifting training costs to the general education system. But the literature on the ladder effect has focused mainly on supply-side mechanisms, while adding demand-side mechanisms in a partial equilibrium setting. For instance, Dolado et al. (2000) and Dolado et al. (2002) investigate the potential impact of the increased proportion of highly educated workers and of job competition on lowskilled employment opportunities in Spain. They build an Albrecht and Vroman (2002)3 model where they allow for on the job search by "overeducated" workers. Their results are based on the assumption that highly educated workers and low educated workers are equally productive in unskilled jobs. Their results focus on the youth over-education in Spain and are generally supportive of a supply-induced ladder effect. In contrast to these studies, our assumptions will allow differential value functions and wages for the two categories when they work in unskilled jobs. In addition, we allow for different preferences for firms when faced with both categories of workers. Recent studies for European countries and an extensive empirical literature support the wage differential between these two categories of workers when they fill up a low skilled job (see Hartog (2000), and in particular for Spain, Wasmer et al. (2007)). We also consider dynamic general equilibrium effects4 and therefore we consider the interaction between demand and supply mechanisms generating the ladder effect. Moreover, focusing on younger workers makes the ladder effect a transitory problem while this problem is of a more permanent nature. It is also found for women and older workers in the economy, particularly in the presence of biased technological change. The aim of this paper is to examine to what extent a ladder effect may contribute to explain changes in the unemployment rate and unemployment differences across skill groups in Spain. To this aim, we develop an inter-temporal general equilibrium model with two types of workers (highly and low-skilled)) and two types of jobs. Following general inter-temporal general equilibrium models of Merz (1995) and Andolfatto (1996), we build a model with endogenous capital, interest rate and wage determination. The production technology is such that highly skilled jobs can be filled only by educated workers, while low–skilled jobs may be filled by both types of workers. We assume that trade in the labour market is represented by a matching process with a Nash wage bargaining mechanism (Pissarides (2000)). The interaction between demand and supply and wages is important in terms of the posting of vacancy by firms and the evolution of the endogenous probabilities of filling such vacancies. We also distinguish between two kinds of shocks: (i) a supply change (ii) and a demand change. We consider as a demand change, a general training change for low-skilled workers 3

There is a large literature that incorporates two side heterogeneity for workers and jobs into partial equilibrium frameworks, i.e. Acemoglu (1999), Marimon and Zilibotti (1999), Muysken and Ter Weel (1999), Delacroix (2003), Davis (2001), Gautier (2002), Albrecht and Vroman (2002), Shi (2002). 4 In Collard et al. (1999), we have incorporated two sided heterogeneity into a dynamic general equilibrium framework. Other papers following our dynamic general equilibrium approach with different contributions are Pierrard and Sneessens (2002), Moreno-Galbis and Sneessens (2004) and Khalifa (2006).

Fabrice Collard, Raquel Fonseca and Rafael Munoz

64

that can be interpreted as a skill-biased technological shock against low-skilled workers. Finally, we consider an increase in the relative size of the highly skilled labour force as a supply change. We calibrate the model using Spanish micro and macro data. In our results, general equilibrium effects are found to be important for both supply and demand shocks. First, it allows to generate evidence for a ladder effect with changes in both supply a demand side and to determine their mechanism. We can also identify and quantify these changes. Second, we find that a demand shock appears to explain a larger share of the ladder effect than a comparable supply shock. This is even more evident for the persistence of unemployment for low-skilled workers. Since the demand change is through general training for low-skilled workers, these results would suggest that more research should be devoted to understanding how training of workers can help their adaptation to new technologies. Third, the model can closely reproduce the decline in the relative number of vacancies (highly skilled vs. lowskilled) in Spain over that period channeled through a demand change. The second section presents the model, showing the specific circumstances that generate a ladder effect. The third section describes the data and our calibration procedure. The forth section analyses the response of some key variables to the introduction of both demand and supply changes. We evaluate the implications of these shocks on the model steady state, thereby enabling us to perform comparative static exercises. The last section offers some concluding remarks.

2. A MODEL OF THE LADDER EFFECT 2.1. Trade in the Labour Market We consider an economy with two types of workers: highly skilled and low–skilled. Highly skilled workers can perform low–skilled jobs, whereas low–skilled workers are h

unproductive in highly skilled jobs. In each period there are L highly skilled workers and

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Ll low–skilled workers. Lh and Ll are exogenous. Because firms observe the workers’ skill level, low–skilled workers can only apply to low–skilled jobs. Highly skilled workers can apply to both types of jobs. This creates an asymmetry among workers and a competition exits between highly and low–skilled workers on the low–skilled job market. This is the source of the ladder effect. Although there are no differences in terms of productivity between highly and low-skilled workers when working on a low–skilled job, firms can incur into a general training cost when they hire low–skilled workers. All variables V j ,t , U j ,t , N j ,t and H j ,t , where j,τ ∈ {h, l} , are ratios divided by their τ

τ

τ

τ

h

corresponding labour force. Let N h ,t denote the proportion of highly skilled workers working h

as highly skilled. N l ,t denotes the proportion of highly skilled workers working as lowh

skilled, and U t denotes the highly skilled unemployment rate. We have

N hh,t + N lh,t + U th = 1.

Spanish Unemployment and the Ladder Effect l

65

l

Likewise, denote N t and U t the proportion of low–skilled working and unemployed. We have

Ntl + U tl = 1. Following Pissarides (2000), we assume that trade in the labour market is an uncoordinated and costly activity. Whenever a firm posts vacancies only a fraction of each of h

l

them will be filled V j ,t and V j ,t vacancies rate highly skilled and low-skilled jobs respectively. We take a constant returns to scale Cobb-Douglas matching function. This function relates the number of matches to the number of vacancies and the number of job seekers. We assume that high-skilled workers. This group can fill highly skilled jobs or lowskilled jobs. And low-skill educated workers can fill only the rest of low-jobs not filled by highly skilled workers. We distinguish two matching functions to fill low-skilled jobs: workers with more education take first low-skilled vacancies.5 Let us now be more precise on the timing of events, and first consider highly skilled workers. A given highly skilled individual first looks for a highly skilled job, such that

H hh,t ≡ H hh ⎛⎜⎝Vt h , U th + N lh,t ⎞⎟⎠ are formed in period t . Implicit in this formulation is the fact that job seekers are composed of highly skilled workers who do not work in highly skilled jobs in period t . When the highly skilled job seeker does not match with a firm in period t , she goes on the low–skilled labour market and attempts to get a match as a low–skilled worker.6 Then,

H lh,t ≡ H lh ⎛⎜⎝Vt l Ll / Lh ,1 − H hh,t − (1 − s ) N hh,t ⎞⎟⎠ are formed on this market. The level of low– skilled vacancies posted by firms has to be adjusted for the relative size of the two populations, in order to preserve the absence of size effects in the matching process. Finally, low–skilled workers can be employed on a low–skilled job — not already occupied by highly skilled workers — such that the level of hiring for the low–skilled is

H tl ≡ H l ⎛⎜⎝Vt l − H lh,t Lh / Ll , U tl ⎞⎟⎠ , where only those vacancies not filled by the highly skilled workers remain available to the low–skilled workers. Like the previous case, the ratio L / L adjusts for the different sizes of each group. It is worth noting that, as in Pissarides (2000), each matching function only depends on aggregate quantities, thus reflecting the fact that firms and job seekers have no control on the matching process. This assumption reflects the existence of the traditional positive trade externalities and congestion effects, associated with the matching process. The evolution of the level of each type of employment is therefore given by:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

h

5

l

We assume that the efficiency matching factors of highly skilled workers will be larger than the efficiency matching factor of low-skilled workers. 6 Notice that highly skilled job seekers who do not find a skilled job immediately search for an unskilled job.

Fabrice Collard, Raquel Fonseca and Rafael Munoz

66

N hh,t +1 = H hh,t + (1 − s ) N hh,t

(1)

N lh,t +1 = H lh,t

(2)

N tl+1 = H tl + (1 − μ ) N tl

(3)

where s, μ ∈ (0,1) denote the constant exogenous separation rates for each type of employment. The second law of motion (equation (2)) represents the fact that highly skilled workers do not occupy a low–skilled job for more than one period, but rather go back on the search. We can interpret it as a high turnover rate job for highly skilled workers, which is consistent with data.7

phh,t is the probability that a highly skilled unemployed worker will be employed in a h

highly skilled job in the next period; pl ,t is the probability that a highly skilled unemployed l

worker will be employed in a low-skilled job; and pt is the probability that a low-skilled unemployed worker will be employed in a low-skilled job in the next period. Thus:

p = h h ,t

H hh,t U th + N lh,t

,p = h l ,t

H lh,t 1 − H hh,t − (1 − s ) N hh,t

Highly skilled Unemployed Workers (Uh) phh

s

Nhh

and ptl =

H tl . U tl

(4)

Low skilled Unemployed Workers (Ul)

phl

1

pl

Nhl

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Nl High vacancy jobs

Low vacancy jobs

Figure 1. Flows in and out of employment

In Figure (1) we can observe flows in and out of employment. Below we express the unemployment dynamic to explain the evolution in both labour markets: 7

Skilled workers experiment more turnover than unskilled workers. They are easier also to migrate their jobs. Our assumption is that skilled workers take the unskilled job as a Temporary position. However they still looking for improving their jobs. Unskilled workers (even if temporary), they continue to look for in an unskilled job.

Spanish Unemployment and the Ladder Effect

67

U th+1 = 1 − H hh,t − H lh,t − (1 − s ) N hh,t Utl+1 =1− Htl −(1− μ ) Ntl .

2.2. Firms We assume a continuum of firms with measure one. In each period, the jth firm has access to constant returns to scale technology represented by the following production function: 1−α −θ

Y j ,t = AK αj ,t ( Lh N hh, j ,t )θ ⎛⎜⎝ Lh N lh, j ,t + Ll N lj ,t ⎞⎟⎠

(5)

h

where K j ,t is the level of physical capital, N h, j ,t is the highly skilled employment rate in h

l

high-skilled jobs, N l , j ,t is the low–skilled employment rate in low skilled jobs and N j ,t is the employment rate of low-skilled workers in low-skilled jobs. We assume there is perfect substitution in terms of productivity between low-skilled workers and highly skilled workers in order to fill a low-skilled job. The value of A is a positive constant that represents the level of total factor productivity. Finally, α , θ ∈ (0,1) respectively denote the elasticity of output with regards to physical capital and highly skilled employment. Each period, the firm invests a level I j ,t to form capital that accumulates as:

K j ,t +1 = I j ,t + (1 − δ ) K j ,t where

(6)

δ ∈ (0,1) is the constant depreciation rate. It also posts vacancies. V jh,t and V jl,t are

vacancies rates for respectively highly and low–skilled jobs, and the firm incurs a linear cost ϖ h and ϖ l per posted vacancy. These vacancies determine the employment that will be used

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

in the following period. The laws of motion of each type of employment are given by:

N hh, j ,t +1 = qhh,tV jh,t + (1 − s ) N hh, j ,t

(7)

Ll Lh

(8)

N lj ,t +1 = qtl ⎛⎜⎝1 − qlh,t ⎞⎟⎠ V jl,t + (1 − μ ) N lj ,t

(9)

N lh, j ,t +1 = qlh,tV jl,t

where qh,t , is the probability of filling highly skilled vacancies, ql ,t , is the probability of h

h

l

filling a low–skilled vacancy with a highly skilled individual and qt , is the probability of

Fabrice Collard, Raquel Fonseca and Rafael Munoz

68

filling a low–skilled vacancy with a low–skilled worker.8 These probabilities are thus given by

qhh,t =

H hh,t Vt h

, qlh,t =

H lh,t Lh H tl l q and = . t Vt l Ll Vt l − H lh,t Lh / Ll

(10)

It is worth noting that these probabilities are determined by aggregate quantities and thus reflect the trade externalities implied by the search process. wh,t , wl ,t and wt are the h

h

l

bargained wages. And finally, when hiring a low–skilled worker, the firm has to train her and therefore incurs a proportional cost per hiring. We suppose that a low-skilled worker needs training when she is hired in a new firm, (i.e. it can lead to a change in technology). We also suppose that a highly-skilled worker has an exogenous general skill from her education, which avoids this training cost. The period t instantaneous profit can be expressed as9

Π j ,t = Y j ,t − whh,t N hh, j ,t Lh − wlh,t N lh, j ,t Lh − wtl N lj ,t Ll

(11)

− I j ,t − ωhV jh,t Lh − ωlV jl,t Ll − H tl Ll Each firm j determines its factor demand and investment plan— both on the good and the labour market — maximising its market value:10 ∞

ϒ( S Fj,0 ) = ∑ Rt Π j ,t t =0

subject to (5)–(11) where ϒ ( S j ,0 ) denotes the value of firm Rt = F

the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

S

F j ,t

real

interest

⎧ ⎨ ⎩

= K j ,t , N

h h , j ,t

,N

rate. h l , j ,t

,N

Finally, l ⎫ j ,t ⎬⎭ .

the

set

Hereafter X

k j ,t

of , X

each h h , j ,t

, X

∏τ t

firm’s h l , j ,t

=0

(1 + rτ ) −1 and rt is

state

and X

l j ,t

variables

is

will denote the

Lagrange multipliers associated to the capital and employment laws of motion respectively. The first order conditions associated with the control variables investment, I j ,t and vacancies, V j ,t and V j ,t , are given by h

l

X kj,t = 1

(12)

8

This probability needs to be adjusted for the relative size of the two populations.

9

After rearranging,

H tl Ll = qtlVt l Ll − qtl H lh,t Lh = qtl (1 − qlh,t )Vt l Ll .

Spanish Unemployment and the Ladder Effect

69

qhh,t X hh, j ,t = ωh

(13)

qlh,t X lh, j ,t + qtl (1 − qlh,t ) X lj ,t = ωl + qtl (1 − qlh,t ).

(14)

Equation (12) represents the optimal level of investment, the marginal cost of capital goods for the firm that is one. Equation (13) represents the optimal level of highly–skilled vacancies posted by a firm. It states that the firm will post highly–skilled vacancies up to the point h

h

where the expected marginal value of filling an additional highly–skilled job ( qh ,t X h , j ,t ), is just compensated by the marginal cost to post a highly–skilled vacancy ( ωh ). The first order condition (14), the marginal value for the firm to fill a low-skilled job receives a similar interpretation, up to the point the marginal cost of posting a vacancy is complemented by an additional cost of training to hire a low-skilled worker. If we suppose no training cost then

qlh,t X lh, j ,t + qtl (1 − qlh,t ) X lj ,t = ωl . A part of the left side is the expected marginal value corresponding to fill the vacancy with a highly-skilled worker and the other part the expected marginal value corresponding to fill the vacancy with a low-skilled worker. Both depending on their different probabilities to fill this vacancy. The parameters will be chosen in such a way that the firm prefers highly–skilled workers to low–skilled ones to fill a low-skilled vacancy11:

X lh, j ,t ≥ X lj ,t .

(15)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.3. Households We now present the behaviour of each type of household — highly–skilled and low– skilled, indexed by i . Following Andolfatto (1996), households of the same type are assumed to be identical ex ante. The random matching functions and separations in the labour market induce different states in the labour market which can lead to ex post heterogeneous wealth positions, which would then make the problem intractable as we would have to keep track of each individual story. For the sake of simplicity, we assume that there exists a perfect insurance market, which allows risk averse households to fully insure against the different income fluctuations and labour market transitions.12 We implicitly suppose that social protection through family is very important in Spain. We assume that the labour force is randomly assigned across jobs at the beginning of each period. Thus, the representative household assumption can be made and the probability of 10

The interested reader is lead to Appendix A.1 for the optimality conditions associated to the firm’s problem. We have checked that assumption 15 is validated ex-post. 12 As frequent in this literature (Chéron and Langot (2004) among others), we assume complete income insurance market. Given perfect insurance markets, optimal households’ behaviour is derived using a dynamic program where ex-post heterogeneity on the labor market does not matter: risk-averse households insure themselves fully against heterogeneous wealth positions (for further details see Andolfatto (1996))). We implicitly suppose that social protection, i.e. via family support is very important in Spain. 11

Fabrice Collard, Raquel Fonseca and Rafael Munoz

70

employment status in any period is given by the different proportions of the employment status. A detailed description of the household problem with full insurance is provided in Appendix A.2.

2.3.1. Low–Skilled Households In each period, a low–skilled household i can be in two alternative states in the labour l

market. We assume that it is employed with probability N t and unemployed with probability

U tl = 1 − N tl . Depending on the state, the instantaneous utility function of the low–skilled household i is given by: =

uil,t lƒ i ,t

log(Cil,t − Γl )

if employed

l∗

lƒ i ,t

= log(C − Γ )

u

if unemployed

l∗

l

where Ci ,t and Ci ,t respectively denote the level of consumption of an employed and l∗

unemployed low–skilled household. Γ and Γ represent a utility cost — expressed in terms of physical goods — associated with the state in the labour market. This cost is assumed to be constant over the business cycle. l

l

The household enters the period with a level of assets Bi ,t carried over from the previous period, from which she gets interest revenues. When employed, the household receives the l

real wage, wijt , bargained with the firm. When unemployed, she receives an unemployment insurance associated with the insurance contract signed with an insurance company.13 These l



revenues are then used to consume Ci ,t or Ci ,t , and to buy new assets. Households take fully insurance against the different income fluctuations and labour market transitions. A perfect insurance system avoids the loss of wealth associated unemployment.14 They Therefore, the consolidated budget constraint after the insurance contract faced by the low–skilled household15 is ƒ

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

N tl Cil,t + (1 − N tl )Cil,t + Bil,t +1 ≤ N tl wijtl + (1 + rt ) Bil,t .

(16)

The problem of the representative household i is therefore to maximise the expectation of the discounted sum of its instantaneous utility with respect to the consumption and the assets she holds: 13

Appendix A.2 proves that households will choose to be fully insured against the risk of unemployment.

14

Where the optimal insurance after optimatisation of the problem is:



Šitl = wijl ,t + Γ l − Γl . and

after the

insurance the marginal utility functions of individuals are the same but the consumption values are different, such as 15





Citl = Citl + Γl − Γl .

Implicit in this formulation is that the insurance problem has already been solved.

Spanish Unemployment and the Ladder Effect ∞

∑β

t

t =0

⎧ ⎪ ⎨ ⎪ ⎩

71

ƒ ⎫

N tl uil,t + (1 − N tl )uil,t ⎪⎬⎪ ⎭

subject to equation (1).

2.3.2. Highly–Skilled Households Like low–skilled households, a highly–skilled household faces different states in the h

labour market. She can be either employed as a highly–skilled worker with probability N h ,t , h

employed as a low–skilled worker with probability N l ,t , or unemployed with probability

U th = 1 − N hh,t − N lh,t . For each state, the instantaneous utility function of the highly–skilled household i is given by:

= log(Chh,i ,t − Γ h ) if employed as high-skilled = log(Clh,i ,t − Γlh ) if employed as low-skilled

uhh,i ,t ulh,i ,t ƒ



ƒ

= log(Cih,t − Γ h )

uih,t

if unemployed

h∗

where Ch ,i ,t , Cl ,i ,t and Ci ,t respectively denote the employed household’s consumption h

h

when she works as a highly–skilled or low–skilled worker and her consumption when unemployed. Γ represents a cost, in terms of physical goods, associated with the highly– h

skilled working activity. Γ l and Γ h

h∗

are the same costs previously defined for low-skilled

households. These costs are assumed to be constant over the business cycle. The consolidated budget constraint after the insurance contract faced by the highly– skilled worker is similar to that faced by the low–skilled. Now however, the household may be in three alternative states ƒ

N hh,t Chh,i ,t + N lh,t Clh,i ,t + U thCih,t + Bih,t +1 ≤

(17)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

≤ N hh,t whh,ijt + N lh,t wlh,ijt + (1 + rt ) Bih,t . The highly–skilled worker solves the same problem as the low–skilled worker. She maximises the discounted sum of its instantaneous utility with respect to consumption and the assets she holds: ∞

∑β t =0

t

⎧ ⎪ ⎨ ⎪ ⎩

ƒ ⎫ ⎪ ⎬ ⎪ ⎭

N hh,t uhh,i ,t + N lh,t ulh,i ,t + U th uih,t

subject to equation (17).

Fabrice Collard, Raquel Fonseca and Rafael Munoz

72

2.4. Wage Determination Following Pissarides (2000), we assume that wages are determined by a Nash bargaining process between the firm and the household. The rent is shared according to the Nash solution of the bargaining problem. Because of the coexistence of different types of workers, there are different wage bargaining processes which give rise to different levels of wages. At the beginning of every period , there is a re-negotiation between firms and workers. Let

ξ hh , ξ hl and ξ l denote the exogenous parameters, which measure the bargaining

power of highly–skilled households applying for highly–skilled and low–skilled jobs and low–skilled workers, the level of wages in a symmetric equilibrium can be shown to be:16. ⎛ ⎜

whth = ξ hh ⎜⎜θ ⎜ ⎝

⎞ Yt h h ⎟ ⎟ p X + ht ht ⎟ + ⎟ Lh N hth ⎠

(18)

⎛ ⎞ N lth ( wlth − Γ l ) ⎟ + (1 − ξ hh ) ⎜ Γ h + h 1 − N ht ⎝ ⎠

⎛ Y jt wlth = ξlh ⎜ (1 − α − θ ) h h ⎜ L N l , jt + Ll N ljt ⎝ ⎛ ⎜

Y jt

⎜ ⎝

Lh N lh, jt + Ll N ljt

wtl = ξ l ⎜⎜ (1 − α − θ )

⎞ h l ⎟⎟ + (1 − ξ h )Γ ⎠

(19)

⎞ ⎟

+ ptl X tl ⎟⎟ + (1 − ξ l )Γ l

(20)

⎟ ⎠

which just amounts to the standard total rent sharing rule between the participant of the bargaining process. The firm acquires the gain in marginal labour productivity and the expected marginal value of a newly created job, X h,t and X t . When a highly-skilled worker

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

h

l

is in a low-skilled vacancy, the separation rate is 1 , and the gain for the firm is only the marginal labour productivity.17 The share accrued by the worker is given by the differential in the disutility of work. It is worth noting that for the highly-skilled worker, who fills a highlyskilled vacancy takes into account the disutililty of work and of working in a low-skilled vacancy.18

16

See appendix A.3 for a detailed exposition of the bargaining process We have checked that the total destruction rate is consistent with our calibration of separation rate of highly skilled workers in a low-skilled job. For more details see Section 3.2.

17

18

Notice as well that





Γl = Γ h = 0

and

Γl = Γlh .

Spanish Unemployment and the Ladder Effect

73

2.5. Equilibrium ∞

Perfect foresight equilibrium of this economy is a sequence of prices {Pt }t =0 =

{whh,t , wlh,t , wtl , rt }t∞=0

and

a



sequence



H ∞

F ∞

quantities {Qt }t = 0 = {{Qt }t = 0 , {Qt }t =0 } .

of



{QtH }t∞=0 = {Chh,t , Clh,t , Cth , Citl , Citl , Bth , Btl }t∞=0

and

{QtF }t∞=0 = {Yt , I t , K t , N hh,t , N lh,t ,

N tl ,Vt h ,Vt l }t∞=0 such that: ∞

H ∞

2. given a sequence of prices {Pt }t = 0 , {Qt }t = 0 is a solution to the representative household’s problem; ∞

F ∞

3. given a sequence of prices {Pt }t = 0 , {Qt }t = 0 is a solution to the representative firm’s problem; ∞



4. given a sequence of quantities {Qt }t =0 , {Pt }t = 0 clears the goods markets in the sense 5.

Yt = Ct + I t + ϖ hVt h Lh + ϖ lVt l Ll + H tlVt l

6. and the capital markets.19 7. wages are set according to the rent sharing mechanism. 8. labour market flows are determined by hiring functions, H h,t , H l ,t and H t . h

h

l

3. DATA AND CALIBRATION We are now to analyse the response of some key variables following two alternative types of shocks: either a relative labour demand shock (changes in training costs) or a relative labour supply shock (changes in proportion of highly skilled workers). Since the model has no any analytical solution, we rely on numerical simulations of the model. The model is calibrated for Spanish data. Simulation exercises are solved with the 20 DYNARE software developed by Juillard (1996).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3.1. The Data The model is calibrated for Spanish quarterly data. Macroeconomic time series are borrowed from Puch and Licandro (1997), who elaborated on the National Accounts of the Spanish Economy (Contabilidad Nacional de España). Aggregate consumption is given by the sum of non-durable consumption and government expenditures, and investment is the sum of durable consumption and fixed investment. For our estimations and stylised facts we have used the definitions of education and occupation in the labour force survey. In order to obtain data consistent with the measure of 19

Where

Ct

is equal to the sum of different households consumption taking into account their different labour

market probabilities. 20 More about DYNARE and the underlying relaxation algorithm can be found in Boucekkine (1995), Juillard (1996, 2001).

Fabrice Collard, Raquel Fonseca and Rafael Munoz

74

the "ladder effect", we rely on the Linked Labour Population Survey (Encuesta de Población Activa enlazada, EPA hereafter).This quarterly survey collects panel data of individuals during six consecutive quarters’ periods. The sample runs from 1987:1 to 1996:4.21 It covers a large number of individuals and characteristics, such as formal education attainment, occupation, employment status, age and gender. It defines 5 groups of education and 10 groups of occupation (range 0–9). Highly skilled workers are defined, for our purpose, as those with a level of education greater or equal to upper-secondary education. Therefore, low–skilled workers essentially consist of illiterate and uneducated workers, primary or low– secondary educated workers. Highly skilled occupations are taken to be managers, professionals and technicians and support professionals (range 1–3 in EPA classification), the rest (range 0, 4–9 in EPA are armed forces, clerks, service, skilled agricultural/fishing, CraftTransport, Plant-Manufacture and unskilled group) is taken to define low–skilled occupations. Ratios are reported in Table 1. Table 1. Aggregated Ratios and Probabilities (average)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Aggregated Ratios and Probabilities capital/output ratio

k/y

9.8458

investment/output ratio

i/ y

0.2877

consumption/output ratio

c/ y

0.6923

high–skilled workers in high–skilled occupations

N hh

0.4027

high–skilled workers in low–skilled occupations

N lh

0.4102

low–skilled workers in low–skilled occupations

Nl

0.7958

prob. for a high–skilled to find a high–skilled job

phh

0.0526

prob. for a high–skilled to find a low–skilled job

plh

0.0915

prob. for a low–skilled to find a low–skilled job

pl

0.1384

ratio high over low–skilled population

Lh / Ll

0.2847

not filled high–skilled vacancies over highly skilled labour force

Vhƒ

0.0093

not filled low–skilled vacancies over low–skilled labour force

Vl ƒ

0.0082

21

The employment steady state values are almost the same for whole of the period 1988-2004, however for a more accuracy calibration for the period 1988-1996.

Spanish Unemployment and the Ladder Effect

75

The "ladder effect" is measured as the proportion of low-skilled jobs filled by highly skilled workers. This proportion rose from 9.8% in 1988 to 15.1% in 1996.22. Then, the h

percentage of highly educated workers who occupy a job with low–skilled requisites ( N l in our model) is the key variable of this proportion. Finally, those highly educated workers in a highly skilled occupation and those low educated workers in a low–skilled occupation are h

l

h

represented as N h and N l , respectively. The highly skilled (low–skilled) population, L l

( L ), is given by the sum of highly (low) educated employed and unemployed workers.23 Table 1 reports the probability of finding a job for each type of individual. For our estimations and stylised facts we have used the definitions of education and occupation in the h

l

linked EPA survey.24 For instance, ph (or ph ) — the probability that a highly skilled unemployed individual finds a highly skilled (or a low–skilled job) — is measured as the proportion of highly educated unemployed workers who find a highly skilled (low–skilled) job from one period to the other. Likewise, the probability that a low–skilled unemployed l

finds a (low–skilled) job, p is measured as the proportion of low–skilled unemployed who are employed in a (low–skilled) occupation from one period to other, corrected by total population. Vacancies are measured by the number of vacancies not filled at the end of period as reported by the National Employment Office (INEM). These data also range from 1987:1 to 1996:4 and are categorised in terms of occupations following the same definition as EPA data for employment.25 They are taken in terms of proportion of the high-skilled or low-skilled total population.

3.2. Calibration

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 2 reports the calibrated value of behavioural parameters. These values are obtained from the model to their empirical counterpart. For instance, the different elasticities of output are set in steady state to be constant returns to scale Cobb-Douglas. The elasticity of output with respect to the capital α is 0.3471 and the elasticity of output with respect to the highly skilled labor θ is 0.1143. The discount rate, β , is set such that, using the Euler equation associated to the capital accumulation decision, the model matches a capital/output ratio of 9.8 — obtained from Spanish Quarterly data. Likewise, the depreciation rate, δ , is 0.0292.

22

We have analysed these proportions for the whole of the economy. Although the ladder effect is relevant in all groups of workers, it seems to have a stronger importance in women and young workers than in among older men. 23 Notice that we have calibrate the employment variables exactly as we have defined them in the model. 24

The proportions and probabilities data are that we have used to solve the model (except to

steady state, see calibration section). 25 National definitions according with R.D. 2240/79 from 14 of August. Using by INEM and EPA.

plh,t

value set in

Fabrice Collard, Raquel Fonseca and Rafael Munoz

76

ωl and ωh are set in steady state.26 We suppose that the cost of being

Cost of vacancies h∗

l∗

unemployment is Γ = Γ = 0. We also suppose that the cost of working in a low-skilled job is the same for both high and low skilled workers. Those together with the cost of working as highly-skilled worked in a highly-skilled job Γ supposed that Γ > Γ = Γ > Γ h

h l

l

h∗

h

are set in steady state. It is

l∗

=Γ .

Training costs are set to zero in our benchmark case, but a sensitivity analysis to changes in this parameter will also be considered. The total factor productivity is set such that, in steady state, output is equal to 1. Table 2. Behavioural parameters Behavioural parameters elasticity of output wrt capital

α

0.3471

elasticity of output wrt high–skilled labour

θ

0.1143

depreciation rate

δ

0.0292

discount factor

β

0.9940

cost of posting low-vacancies

ωl

0.8626

cost of posting high-vacancies

ωh

0.0652

cost of working in a high-job

Γh

0.7314

cost of working in a low-job

Γlh = Γl

0.5363

cost of not working

Γ h = Γl





Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

training costs

0.0000 0.0000

total factor productivity

A

0.6083

elasticity of matching w.r.t. vacancies

γ

0.5000

{ξi j }i , j∈{h,l }

0.5000

bargaining power

γ , the elasticity of the matching process with regards to the number of vacancies is assumed to be the same in each function and is set to γ = 0.5 .27 This lies within the range of estimated values for the Spanish economy. The bargaining power of households 26

The cost of posting a low vacancy is larger than the cost of posting a high vacancy. It can be interpreted as firms receive a signal from highly skilled workers through their education.

We suppose that γ = 0, 5 , we aim to collate in a simple way in a baseline model are the heterogeneity of agents with skill disparities.

27

Spanish Unemployment and the Ladder Effect

77

( {ξi }i , j∈{h,l } ) is also set to 0.5 , such that it is equal to the elasticity of matching with respect j

to vacancies. As shown in Hosios (1990), this implies that the Nash bargaining process yields a Pareto optimal allocation of resources. First of all, the three matching functions are specified as

H hh,t = H hh(Vhh,t )γ (1 − N hh,t )1−γ γ

⎛ Ll ⎞ Hlh,t = H lh⎜ Vhh,t h ⎟ (1− Nhh,t +1 )1−γ L ⎠ ⎝ ⎛ l⎜ ⎜ ⎜⎜ ⎝

γ

⎞ Lh H = H Vt − l H lh,t ⎟⎟ (1− Ntl )1−γ ⎟⎟ L ⎠ l t

l

In the steady state, we set H l ,t = N l ,t . Given the probabilities of finding a job, and the level h

h

of the employment rates, we are able to compute the average number of hiring as

H hh = phh (1 − N hh ), and and H l = p l (1 − N l ). h

In order to solve the model, we get the value of pl at steady state using the probability of finding a low-skilled job by a highly skilled workers pl ,t = 1− H h h

H lh,t

h h ,t − (1− s ) N h ,t

h

h

. These numbers can

l

then be used to calibrate H h , H l and H in steady state respectively. As we have remarked in the description of the model, we assume that firms prefer to hire a highly skilled h

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

worker in a low-skilled job that is the efficiency factor of the matching function H l is higher than the other parameters and its influence can be seen in that their wages become more competitive. The exogenous quit rates are calibrated using equations (7)-(10) evaluated in steady state. Therefore, we obtain

s=

H hh Hl and μ = . N hh Nl

The separation rate for highly skilled workers in a highly skilled job s , is larger than the separation rate for low-skilled workers, μ . This is consistent as a high turnover rate in the skilled workers. We set the separation rate for highly skilled workers in a low-skilled job to the value of one. In our economy, the total destruction is about 11%. This is consistent with the average of destruction rate for Spanish firms in Díaz-Moreno and Galdón (2000).28

28

They estimate quarterly job flows for Spanish economy for the period 93:II to 95:I.

78

Fabrice Collard, Raquel Fonseca and Rafael Munoz h

l

Skilled and unskilled vacancies, V and V , are defined as the sum of highly and low– skilled hiring functions and unfilled highly and low–skilled vacancies. This permits obtaining h

the probability that a firm fills a highly skilled vacancy with a highly skilled worker, qh and h

the probability of filling a low–skilled vacancy with a highly or low–skilled worker, ql and

q l , as defined in equation (10). Wage variables are set in steady state. Notice that highly skilled wages are highest. And slightly low-skilled wages larger that highly skilled wages when the worker is highly educated in a low-skilled job.29 These numbers are reported in Table 3.

4. ANALYSIS OF RESULTS This section proposes an analysis of the response of some key variables characteristic of the labour market to exogenous permanent shocks to (i) the training cost of low-skilled workers and (ii) the relative size of the highly skilled labour force. We first perform a static exercise assessing the steady state implications of such changes in the model. This comparative analysis enables us to offer some insights on the role played by the ladder effect in Spanish unemployment dynamics.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

4.1. Demand Side: Training Cost We model the relative demand shock as a change in the cost of training new low-skilled employed.30 This change in the training cost may be seen as a consequence of a technological change. We assume that highly skilled workers do not require training: Given their education level, we assume that they have exogenous general skills. In our model the increase of this training cost can be interpreted as a skilled-biased technological shock against low-skilled workers. Highly skilled workers are willing to accept low-skilled vacancies temporarily. Despite their high separation rates, firms hire highly skilled workers to avoid general training costs inherent to the hiring of low-skilled workers. This training cost affects employment asymmetrically. An increase in this training cost decreases total employment but more largely low-skilled employment. Thus, we now analyse the effects of a permanent increase in the training cost paid by the firm when hiring a low–skilled worker, χ . Figures 2–3 report the effects of a permanent 29

Groeneveld, J. Hartog (2004) find heterogeneity in wages and promotions for workers with different education related to be under or over educated in their job. Buchel and Mertens (2000) find that overeducated workers are less likely and undereducated workers are more likely to experience above average wage growth. In particular for Spain, Wasmer et al. (2007) find evidence of wage penalty for over-qualified workers.

30

We have also considered a permanent non-biased technological shock. As in literature the increase in the productivity of labour increases employment and decreases unemployment. However the effect in the equilibrium unemployment rates will be neutral. Interesting further research will be to keep in temporary technological shocks in order to experience if there exists some relationship between technological shock and “ladder effect”.

Spanish Unemployment and the Ladder Effect

79

shock in χ ranging from values 0.0 to 0.2, on the steady state of some key variables characterising the labour market. Table 3. Calibrated parameters (Labour Market) Labour market variables and probabilities hirings of high–sk. workers for high–skilled jobs

H hh

0.0314

hirings of high–sk. workers for low–skilled jobs

H lh

0.0547

hirings of low–sk. workers for low–skilled jobs

Hl

0.0282

h h

0.2014

h

0.7234

effi. factor of high–sk. workers for high–sk. jobs effi. factor sk. workers for low–sk. jobs

Hl

effi. factor sk. workers for low–sk. jobs

H

l

0.3275

prob. of finding a low-sk. jobs by a high-sk. work.

plh

0.6867

wages of high-sk. workers for high-skilled jobs

whh

0.9027

wages of high-sk.workers for low-sk. jobs

wlh

0.5632

wages of low-skilled workers

0.5801

high–skilled separation rate

wl s

low–skilled separation rate

μ

0.0355

prob. firm fills h-skilled vac. with h-skilled workers

qhh

0.7715

prob. firm fills l-skilled vac. with h-skilled workers

qlh

0.7620

ql

0.7751

high–skilled vacancies

Vh

0.0407

low–skilled vacancies

Vl

0.1532

prob. firm fills l-skilled vac. with l–skilled workers

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

H

0.0780

An increase in χ induces firms to reduce low-skilled employment. It therefore increases the marginal value of low-skilled jobs, and decreases the marginal values of highly skilled ones (see Figures 2 and 3). As the wage bargaining process implies that wages are strongly correlated with the marginal productivity of employment, the real wage paid to highly skilled workers when employed on highly skilled jobs, reduces more than that received by low– skilled workers on low–skilled occupation (see Figure 2), reducing the wage gap among them. The increase in the value of the shock has two opposite effects for the highly skilled workers. On the one hand, the cost of low–skilled employment rises and firms will reduce the hiring of both low and highly skilled workers. On the other hand, firms will partially substitute highly skilled workers for low–skilled ones to fill low–skilled vacancies as the former do not require any training. Overall, the low-skilled employment decreases most (see

Fabrice Collard, Raquel Fonseca and Rafael Munoz

80

Figure 2). This substitution effect explain the huge drop in N . Besides, noteworthy is the l

l

h

h

fact that N drops dramatically more than both N l and N h (see upper–left panel of Figure 1), illustrating the substitution effect between highly and low–skilled workers in low–skilled occupations due to the increase in the training cost. The reduction in the marginal value and the increase in the training cost therefore discourages posting vacancies, whatever their type, as illustrated by Figure 1. For values of χ larger than 0.1, the reduction in the number of highly skilled vacancies is larger than the reduction in low–skilled vacancies. This reduction in both types of vacancies and the increase in unemployment reverse the congestion effect.31 Therefore, the probability that a firm fills a vacancy raises whatever the type of posted vacancy. But, as the increase in unemployment is higher for low–skilled workers than for highly skilled ones, the rise in this probability is l

much higher in the case of low–skilled workers, q (by an order of about 5 when χ takes value 0.2) . Likewise, the implied negative trade externality explains the larger decrease in the probability to find a job for low–skilled workers. The investment behaviour of firms leads to employment reduction for these types of occupation. Our results are consistent with Van Ours and Ridder (1995), which give evidence that low- and highly skilled unemployment are strongly correlated and that low-skilled unemployment fluctuates more strongly. For instance, in face of a value of χ = 0.1 , N h and h

N lh decrease by 5%, while N l drops by an amount of 10%. This shows up in the measure of the ladder effect reported in Figure 3. Employment

0

60 % Deviation

−5 % Deviation

Unemployment

80

−10 h N hh

N

−15

40 Uh Ul Total

20

l

Nl −20

0

0.05

0.1 χ

0.15

Vacancies

0

V Vl

% Deviation

% Deviation

−15 −20

0.05

0.1 χ

0.15

0.1 χ

0.15

0.2

0.15

0.2

Wages

−1 −2

Wh hh W l

−3

−25 0

0.05

0

−10

−30

0

1

h

−5

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0

0.2

0.2

−4

Wl 0

0.05

0.1 χ

Figure 2. Steady state implication of training cost shock 31

Under the presence of coordination failures on the labour market, an increase of competition among firms creates a congestion effect that lowers the probability of filling up a vacant job.

Spanish Unemployment and the Ladder Effect

81

The ladder effect, measured in terms of stocks,

Lh N lh × 100, Lh N lh + Ll N l illustrates the previous analysis. As the ladder effect increases by 3% when χ = 0.1 to 11% in face of χ = 0.2 , once again reflecting the expected substitution effect between highly and low–skilled workers that is at work in the face of a training cost applied to low–skilled workers. Prob. to fill up vacancies

100

qh h qh

−10

80

ql

−20

l

60 40 20 0

0

0.05

0.1 χ

0.15

% Deviation

% Deviation

−10 Hh h h H l

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0

0.05

p h ph

−50

pl

l

0

0.05

0.1 χ

0.15

0.2

0.15

0.2

Marginal values

60

Xh h Xh

40

X

l l

20 0

Hl −20

h

−40

80

−5

−15

−30

−60

0.2

Hirings

0

Prob. to find a job

0

% Deviation

% Deviation

120

0.1 χ

0.15

0.2

−20

0

0.05

0.1 χ

Figure 3. Steady state implication of training cost shock

This effect is far from proportional, and the higher the training cost, the larger the ladder effect becomes. It could be interpreted that in a society with rapid technological progress, those less qualified will need more training to be able to work even in less qualified vacancies. Stated otherwise, the low–skilled vacancies will require larger skills and the cost of qualifying low–skilled workers will rise more than proportionally. Then, highly skilled workers take priority over low-skilled workers to fill these vacancies. This leads to an increase in low-skilled unemployment. The effect is an increase in the “ladder" effect indicator.

Fabrice Collard, Raquel Fonseca and Rafael Munoz

82

4.2. Supply Side: Skilled Labour Force There is a large amount of literature about the increase of skills in the labour force (see Green et al. (1999)). This section analyses the effects of a permanent increase in the relative size of the highly skilled labour force — a positive shift in L / L — which might be interpreted as an increase in the aggregate education attainments of workers. Figures 5–6 report the effects of a shock ranging from 1 to 19% on the steady state of some key variables characterising the dynamics in the labour market. h

l

Ladder effect (flow)

12

10

% Deviation

8

6

4

2

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0

0

0.02

0.04

0.06

0.08

0.1 χ

0.12

0.14

0.16

0.18

0.2

Figure 4. The ladder effect when χ changes

The first direct implication of a permanent increase in the relative availability of highly skilled workers is to shift downward the probability for this type of worker to find a job. Indeed, the higher relative availability of this type of labour creates a congestion effect on the supply side which, via the matching process, makes it much harder for highly skilled workers h

h

to find a job. Thus, as can be seen from Figure 5, both ph and pl drop. For instance, the probability for a highly skilled worker to find a highly skilled job diminishes by 4% in face of a 5% permanent increase in L / L . On the contrary, it exerts a positive trade externality that h

l

Spanish Unemployment and the Ladder Effect

83

benefits the firms and makes the probability of filling a vacancy higher. Noteworthy is that h

h

this effect does exist both for highly skilled jobs ( qh ) and low–skilled jobs ( ql ), as skilled workers may apply to such jobs. Nevertheless, the effect is more pronounced on highly skilled jobs. Conversely, the probability to fill a low–skilled job with a low–skilled worker l

( q ) is lowered, as low–skilled workers are proportionally scarcer, thus increasing the competition among firms and creating a relative congestion effect. Employment

4

Unemployment

20

2

Uh l U Total

15

% Deviation

% Deviation

0 −2 −4

h

−6

Nh h N

−8

Nl

l

−10

1

1.05

10 5 0

1.1 1.15 Δ(L /Ll)/(Lh/Ll)

−5

1.2

1

1.05

h

Vacancies

20

Vh Vl

% Deviation

% Deviation

0

0 −10 −20 −30

1.2

Wages

1

10

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

−1 −2

Wh h Wh l

−3

1

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

1.2

−4

Wl 1

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

1.2

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 5. Steady state implication of relative labour force shock

The marginal values of employment do not all increase. Notice they depend on the labour market tightness through the wage setting mechanism. Indeed, only the marginal value of low–skilled jobs increase while that of highly—skilled occupations decrease. This implies that firms post a higher number of low–skilled vacancies whereas the number of highly skilled decreases, as Figure 4 shows. This together with the evolution of the probability to fill vacancies implies that employment of low–skilled workers and highly skilled workers employed as low–skilled increase, whereas that of highly skilled decreases. Therefore low– skilled unemployment diminishes while highly skilled unemployment increases, the overall effect on total unemployment being slightly positive. The overall effect on wages is easily understood in the light of previous results as increases in the marginal productivity of all types of employment exerts an upward pressure

Fabrice Collard, Raquel Fonseca and Rafael Munoz

84

on wages, which is countered by the decrease in the tightness of the highly skilled labour market. Thus, while the real wages paid in compensation to low–skilled job increase, those paid to highly skilled job decreases. Beside these aggregate effects, the measure of the ladder effect — as reported in Figure 76 — reflects the earlier story, as increases in the relative size of the highly skilled labour force lead to an increase in the ladder, implying that more extensive use of low–skilled workers in low–skilled jobs is at work. This is explained as the increase in the probability to fill a low–skilled vacancy with a highly skilled worker makes it worth increasing the use of this type of labour. In others an increase in the availability of highly skilled workers implies that firms substitute low–skilled workers for highly skilled ones in low–skilled occupations. However, this is not enough to compensate for the rise in the highly skilled labour supply, implying a larger level of highly skilled unemployment. It seems to provide a pretty good explanation to the positive correlation between the rise in the number of educated workers and in the highly skilled unemployment rate in the Spanish economy during the eighties. Prob. to fill up vacancies

20

qh h qh

15

0

% Deviation

% Deviation

l

ql

10 5

1

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

−15

1.2

Hirings

1

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

1.2

Marginal values

10

% Deviation

0

% Deviation

l

15

2

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ph h ph pl

4

−2 −4 −6

Hh h h

−8

Hl

5 0 −5

Xh h Xh

−10

Xl

H

l

−10

−5

−10

0 −5

Prob. to find a job

5

1

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

1.2

−15

l

1

Figure 6. Steady state implication of relative labour force shock

1.05

1.1 1.15 Δ(Lh/Ll)/(Lh/Ll)

1.2

Spanish Unemployment and the Ladder Effect

85

4.3 Comparative Analysis This analysis deserves additional comments that may shed light on the recent Spanish experience. We have introduced both demand and supply changes. Both shocks are significant to explain evolution of labour market variables, but the importance of these changes is different.

Ladder Effect and Relative Wages From the last two sections, we observe that both training cost and relative labour force changes are important to explain the "ladder effect" and reproduce the same relative wages evolution. They have similar effects such a ladder effect indicator increases. Both explain that highly skilled workers take low-skilled jobs leading to low-skilled unemployment persistence. When we introduce a change in the training cost, both highly skilled employment and lowskilled employment proportions decrease (see Figure 1). Since the former drops more than the latter, the ladder effect indicator increases very significantly. Nevertheless, when a change in the relative highly skilled labour force occurs the ladder effect is due to both an increase in highly skilled employment working in low-skilled jobs and the fact that low-skilled employment increases at a slower rate (see Figure 4). In both shocks real wages paid to low– skilled job increase, while paid to highly skilled job decreases. Ladder effect (stock)

1.6

1.4

1.2

% Deviation

1

0.8

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.6

0.4

0.2

0

1

1.02

1.04

Figure 7. The ladder effect when

1.06

Lh / Ll

1.08

1.1 1.12 Δ(Lh/Ll)/(Lh/Ll)

changes

1.14

1.16

1.18

1.2

Fabrice Collard, Raquel Fonseca and Rafael Munoz

86

Employment and Vacancies Their effects on unemployment and vacancies rates are totally different. Both changes reproduce the increase in highly skilled unemployment observed the studied period. However, the training cost change only reproduces the employment. Low-skilled unemployment only increases with the introduction of a training cost. In the other shock, the low-skilled unemployment rate decreases. Hence, the training cost can better explain the evolution of the "ladder effect".32 Vh/Vl 0.45

0.4

0.35

0.3

0.25

0.2

0.15

0.1

0.05

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 8. Vacancy ratio

96 3

96 1

95 3

95 1

94 3

94 1

93 3

93 1

92 3

92 1

91 3

91 1

90 3

90 1

89 3

89 1

88 3

88 1

87 3

87 1

0

V h /V l

Figure 8 reports the ratio of vacancies for the Spanish economy for our benchmark sample. The ratio of highly to low–skilled vacancies diminishes within this period. Indeed, as predicted by the model under any shock the ratio of highly skilled vacancies to low-skilled vacancies reduces. In the face of a labour supply shock, firms post a lower number of highly skilled vacancies. This reduction is even larger in the case of a greater training cost. Decreases in the number of highly skilled vacancies would rather be explained by both labour force and the training cost shocks. When there was a change in the number of highly skilled workers, low–skilled vacancies increased in the model. There is a decrease in the low–skilled vacancies posted when the training cost shock. However their decrease is less than the highly skilled vacancies. Then both shocks explain the decrease of the highly to low skilled vacancies.

Spanish Unemployment and the Ladder Effect

87

These results are understood in our framework of high turnover for highly skilled employees who work in low-skilled jobs. The model is built with a exogenous separation rate of 1 for this group. It leads to highly skilled employees to accept low-skilled vacancies temporarily, given that a highly skilled worker really wants to fill a highly skilled job. It is consistent with the Spanish context where fixed-term employment contracts are more flexible from 1984. The number of temporary employment has growth from 15.6% in 1985 to 33.6% in 1996 (see Dolado et al., 2002) and a high turnover rate (Diaz-Moreno and Galdon (2000)). From our results, firms prefer to take a highly skilled worker for a low-skilled job to avoid general training costs of low-skilled workers that can be interpreted as a technological change. This leads to an increase in low-skilled unemployment. It is also consistent with the idea that general skills of highly skilled workers allow more flexibility to change the job. The change in the labour supply can also reproduce the increase in the indicator of ladder effect in Spain. These results indicate further research should be done to study interrelations between demand and supply.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

5. CONCLUSION This paper attempts to shed light on alternative explanations for the "ladder effect" phenomenon, which is a significant source of Spanish unemployment persistence. We have used a calibrated version of the model to assess the implications of labour demand and supply shocks. The labour demand shocks are related to the increase of the training cost in low-skilled vacancies and are seen as a biased technological progress against low–skilled workers. Labour supply shocks, often associated with increases in the number of highly-skilled workers, are introduced as increases in the relative availability of a highly– skilled labour force. Our results indicate that the ladder effect generated by the model may account for the recent Spanish experience. The model can replicate the observed decline in the ratio of high to low–skilled vacancies, and shows how firms substitute high for low– skilled employment. We argue that the Spanish ladder effect reflected by an increase in the training cost as a result of a biased shock against low–skilled workers is better reproduced than the increase in the number of highly-skilled workers. The positive change in the training cost of the low-skilled workers better reproduce the evolution on employment and in particular the decrease of low-skilled employment, as well as the evolution of the vacancy data. Since the demand change is through general training for low-skilled workers, these results would suggest that more research should be devoted to understanding how training of workers can help their adaptation to new technologies. What remains to be investigated is the extent to which the ladder-effect merely reflects the role of education as an entry screening device by firms, which in turn would imply that official job descriptions may have little relation to job-content.

REFERENCES Acemoglu, D., (1998). Changes in unemployment and wage inequality: an alternative theory and some evidence, American Economic Review, 89 (5), 1259-1278.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

88

Fabrice Collard, Raquel Fonseca and Rafael Munoz

Acemoglu, D., (2002). Direct Technical Change, Review of Economic Studies, 67, 781-809. Alba-Ramirez, A., (1993). Mismatch in the Spanish labour Market, The Journal of Human Resources, 28, 259-278. Andolfatto, D., (1996). Business cycles and labour–market search, American Economic Review 86, 112–132. Albrecht, J.&Vroman S., (2002). A Matching Model with Endogenous Skill Requirements. International Economic Review 43, 283-305. Berman E., Bound J., & Machin S., (1998). Implications of skill-biased technological change: international evidence, Quaterly Journal of Economics, 113, 1245-1279. Blanco, J.M., (1997). Comentarios acerca del desajuste educativo en España, El Mercado de Trabajo en Perspectiva Europea in Papeles de la Economia Española, 72. Boucekkine, R. (1995). An alternative methodology for solving nonlinear forward-looking models, Journal of Economic Dynamics and Control, 194: 711-734. Buchel, F., & Mertens, A., (2000). Overeducation, Undereducation, and the Theory of Career Mobility. Max Planck Institute for Human Development, Berlin paper presented at EALE/SOLE Conference. Cheron A., & Langot, F. (2004). Labor market search and real business cycles: reconciling Nash bargaining with the real wage dynamics. Review of Economic Studies, 69(3), 533563. Collard F., Fonseca R., & Munoz R., (1999) The Ladder Effect: A Tale for Spanish Unemployment Persistence IZA Summer School mimeo April, 1999. Davis, S., (2001). The Quality Distribution of Jobs and the Structure of Wages in Search Equilibrium. National Bureau of Economic Research Working Paper 8434. Delacroix, A., (2003). Heterogeneous Matching with Transferable Utility: Two Labor Market Applications. International Economic Review, 44(1), 313-330. Díaz-Moreno, C., & Galdón, J.E. (2000). Job creation, job destruction and the dynamics of Spanish firms, Investigaciones Económicas, vol. 243, 545–562. Dolado, J. J., Felgueroso, & Jimeno, J. F. (2000). Youth labour markets in Spain: Education, training, and crowding-out. European Economic Review 444-6 pp. 943-956. Dolado, J.J., Jansen, M., & Jimeno, J.F., (2002). A matching model of crowding-out and onthe-job search (with and application to Spain). IZA DP No. 886, October 2003. Dolado, J.J., García-Serrano, C., & Jimeno, J.F. 2002. Drawing Lessons from the Boom of Temporary Jobs in Spain. The Economic Journal Vol. 112 , 270-95 Garcia-Serrano, C. & Malo, M.A., (1996). Desajuste Educativo y Movilidad Laboral en España, Revista de Economia Aplicada. Número 11 Vol. IV , pp. 105-131. Gautier, P.A., (2002). Unemployment and Search Externalities in a model with Heterogeneous Jobs and Heterogeneous Workers, Economica, 69:21-40. Green, F., Mcintosh, S. & Vignoles, A., 1999. Overeducation and Skills-Clarifying the o

Concepts, CEP Discussion paper n 435. Hansen, G. D. 1985. Indivisible labour and the business cycle. Journal of Monetary Economics, 16, pp. 309–327. Hartog, J., (2000). Over-education and earnings: where are we, where should we go? Economics of Education Review 19, 131-147.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Spanish Unemployment and the Ladder Effect

89

Juillard, M. 1996. DYNARE: a program for the resolution and simulation of dynamic models with forward variables through the use of a relaxation algorithm, CEPREMAP, Working Paper 9602. Juillard, M. 2001. DYNARE: A program for the simulation of rational expectation models, Computing in Economics and Finance 213, Society for Computational Economics. Khalifa S., (2006). Heterogeneous Workers and Occupations: Unemployment, Inequality and Crowding Out. Thesis manuscript, Johns Hopkins University, 2006. Krugman, P., (1994). Past and Prospective Causes of High Unemployment, Reducing Unemployment: Current Issues and Policy Options in a Symposium Sponsored by the Federal Reserve Bank of Kansas City. Marimon, R., & Zilibotti, F., (1999). Unemployment Versus Mismatch of Talents: Reconsidering Unemployment Benefits. The Economic Journal, , 109(455): 266-291. Merz, M., (1995). Heterogeneous job-matches and the cyclical behavior of labor turnover. Journal of Monetary Economics 43, 91-124. Moreno-Galbis, E., & Sneessens, H. (2004). Unemployment, capital-skill complementarity and embodied technological progress. Recherches économiques de Louvain Vol.73, 241272 Muysken, J., & Ter Weel, B.J. (1998). Overeducation and Crowding Out of low-skilled workers, in I. Borghans and A. de Grip eds. The overeducated worked? The economic of skill utilization. Edward Elgar: Chentengam. pp. 109-132 March. OECD 1994. OECD Jobs Study, OECD, Paris. OECD 2004. Education at a Glance 2004. Groeneveld, S. & Hartog, J., (2004). Overeducation, wages and promotions within the firm. Labour Economics 11 , 701–714 Pissarides, C.A., (2000). Equilibrium Unemployment Theory, Massachusetts Institute of Technology, second edition. Pierrard, O. & Sneessens, H. R., (2003). Low-Skilled Unemployment, Biased Technological Shocks and Job Competition, IZA Discussion Papers 784, Institute for the Study of Labor (IZA). Puch, L. & Licandro, O., (1997). Are there any special features in the Spanish business cycle?, Investigaciones Económicas, 212, 361-394. Shi, S., (2002). A Directed Search Model of Inequality with Heterogeneous Skills and SkillBased Technology. Review of Economic Studies, vol. 69(2), 467-91, April. Thurow, L. C., (1975). Generating Inequality: Mechanisms of Distribution in the U.S. Economy. New York: Basic Books. Van Ours, J.C., & Ridder, G., (1995). Job matching and job competition: Are lower educated workers at the back of job queues? European Economic Review 39 1717-1731. Wasmer, E., Fredriksson, P., Lamo, A., Messina, J. & Peri, G. (2007). The Macroeconomics of Education in Europe, in Brunello, G., Garibaldi, P. & Wasmer E. Education and Training in Europe, Oxford University Press

Fabrice Collard, Raquel Fonseca and Rafael Munoz

90

APPENDIX A Appendix A.1 : Decisions Rules of Firm J . Recall that production function of the firm j is given by: 1−α −θ

Y j ,t = AK αj ,t ( Lh N hh, j ,t )θ ⎛⎜⎝ Lh N lh, j ,t + Ll N lj ,t ⎞⎟⎠

.

(A.1.1)

Accumulation of capital is,

K j ,t +1 = I j ,t + (1 − δ ) K j ,t

(A.1.2)

The law of motion of each type of employment are given by:

N hh, j ,t +1 = qhh,tV jh,t + (1 − s ) N hh, j ,t

N

h l , j ,t +1

(A.1.3)

Ll =q V h L h l ,t

l j ,t

(A.1.4)

N lj ,t +1 = qtl ⎛⎜⎝1 − qlh,t ⎞⎟⎠ V jl,t + (1 − μ ) N lj ,t .

(A.1.5)

At period t instantaneous profit can be expressed as

Π j ,t = Y j ,t − wth N hh, j ,t Lh − wtl N lh, j ,t Lh − wtl N lj ,t Ll − I j ,t − ωhV jh,t Lh

(A.1.6)

−ωlV jl,t Ll − H tl Ll .

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Then the firm solves the recursive problem

max ϒ( S Fj ,t ) = Π j ,t +

1 ϒ( S Fj ,t +1 ) 1 + rt +1

subject to (A.1.1)–(A.1.6). We form the Lagrangian for this problem. Let us denote the Lagrange multipliers associated to K j ,t , N h , j ,t , N l , j ,t and N j ,t respectively by X j ,t , h

h

l

k

X hh, j ,t , X lh, j ,t and X lj ,t . The first order conditions associated to the control variables investment, I j ,t and vacancies, V j ,t and V j ,t , are given by h

l

Spanish Unemployment and the Ladder Effect

X kj,t = 1

X hh, j ,t =

91 (A.1.7)

ωh

(A.1.8)

qhh,t

qlh,t X lh, j ,t + qtl (1 − qlh,t ) X lj ,t = ωl + qtl (1 − qlh,t ).

(A.1.9)

Equation (A.1.7) represents optimal level of investment, the marginal cost of capital goods for the firm that is one. Equation (A.1.8) represents the optimal level of highly–skilled vacancies posted by a firm. The first order condition (A.1.9) is the marginal value for the firm to fill a low-skilled job complemented by an additional cost of training to hire a low-skilled worker. The marginal values of the capital and the different types of employment for the firm are given by the envelope theorem as:

Ω = k jt

Ω

h h , jt

∂K j ,t

=

Ωlh, jt = Ωljt =

∂ϒ j ,t ( S Fj ,t )



∂ϒ j ,t ( S Fj ,t ) ∂N

h h , j ,t

∂ϒ j ,t ( S Fj ,t ) ∂N

h l , j ,t

∂ϒ j ,t ( S Fj ,t ) ∂N

l j ,t

Y j ,t +1

+1− δ

K j ,t +1



Y j ,t h

LN

=(1−α −θ )

=(1−α −θ )

h h , j ,t

− whj ,t + (1 − s ) X hh, j ,t Y j ,t

h

(L N

h l , j ,t

+LN ) l

l j ,t

Y j ,t h

(L N

h l , j ,t

+LN ) l

l j ,t

− wlh, j ,t

− wlj ,t +(1−μ ) X lj ,t .

Combining both envelope conditions and first order conditions, we get the Euler equations h

h

l

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

related to K j ,t , N h, j ,t , N l , j ,t and N j ,t ,

⎛ Y j ,t +1 ⎞ + 1 − δ ⎟ = 1 + rt +1 ⎜⎜ α ⎟ ⎝ K j ,t +1 ⎠

X hh, j ,t =

⎛ ⎜ ⎜ ⎜ ⎜ ⎜ t +1 ⎜ ⎜ ⎝

1 1+ r

θ

Y j ,t +1 Lh N hh, j ,t +1

(A.1.10)

⎞ ⎟

− whj,t +1 + ⎟⎟

+(1 − s) X

h h , j ,t +1

⎟ ⎟ ⎟ ⎟ ⎠

(A.1.11)

Fabrice Collard, Raquel Fonseca and Rafael Munoz

92

X lh, j ,t =

X lj ,t =

⎛ ⎜ ⎜ ⎜ ⎜ ⎜ t +1 ⎜ ⎜ ⎝

1 1+ r

⎛ ⎜ ⎜ ⎜ ⎜ ⎜ t +1 ⎜ ⎜ ⎝

1 1+ r

(1 − α − θ )

⎞ ⎟

Y j ,t +1 ( Lh N lh, j ,t +1 + Ll N lj ,t +1 )

− ⎟⎟ ⎟ ⎟ ⎟ ⎟ ⎠

−w

h l , j ,t +1

(1 − α − θ ) −w

⎞ ⎟

Y j ,t +1 ( Lh N lh, j ,t +1 + Ll N lj ,t +1 )

l j ,t +1

+ (1 − μ ) X

(A.1.12)

l j ,t +1

− ⎟⎟ ⎟ ⎟ ⎟ ⎟ ⎠

.

(A.1.13)

Furthermore, parameters will be chosen in such a way that the firm will prefer to hire a highly skilled worker instead of a low skilled worker for an unskilled vacancy,

X lh, j ,t ≥ X lj ,t . This condition will be satisfied at the steady state.

Appendix A.2 : The households We follow Andolfatto (1996), in order to solve the problem of households. Workers flows are determined according to the matching process we described in section 2.3. Therefore, workers are randomly selected playing a game of “musical chairs". At the beginning of each period, the whole labour force is randomly shuffled across a given set of jobs.

A.2.1 The Low-Skilled Consumer Problem This section presents the derivation of the optimal behaviour of the low–skilled consumer, insisting on insurance issues. At the beginning of each period low–skilled households face different probabilities of being employed or unemployed as this is contingent on its status in the labour market in the Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

N

previous period. This therefore implies that 2 different possible stories in the labour market are to be considered after N periods, each of which corresponds to a particular individual employment path and therefore a different story of accumulation. This leads to heterogeneity, which makes the resolution of the model extremely complicated. For the sake of simplicity, we follow Hansen (1985), and assume that there exists a perfect insurance system which may eliminate ex–post heterogeneity. The instantaneous utility functions are given by

uil,t = log(Cil,t − Γl ) (employed) ∗





uil,t = log(Cil,t − Γl ) (unemployed)

(A.2.1) (A.2.2)

Spanish Unemployment and the Ladder Effect

93

l∗

l

where ui ,t and ui ,t are the respective instantaneous utility functions for employed and l

l∗

unemployed households. As in the main body of the text, we let Cit and Cit denote the respective low-skilled household’s consumption when employed and unemployed. Γ and l



Γl can be interpreted as a utility cost, expressed in terms of goods, associated with the l∗

situation of the household in the labour market. We assume Γ > Γ , which ensures that the consumption of an employee is greater than that of an unemployed household. At the very beginning of each period — before the matching process has taken place — the low-skilled household does not know what the situation, either employed or unemployed, will be in that period. As a consequence, the household seeks to maximise her expected value. l

l

Since there N t denotes the percentage of low–skilled households that are employed. And

β ∈ (0,1) is the discount factor of the household. The low-skilled household maximises the problem ∗

V L ( Sil,t ) = N il,t uil,t + (1 − N il,t )uil,t + β V L ( Sil,t +1 ),

(A.2.3)

l∗

where state variables are Si ,t = Si ,t {B j .t , Bi ,t } depending whether the household is employed l

l

l

l

or unemployed, are to be considered. We made use of the fact that by definition of pt and the l

law of motion of N t +1

N tl+1 = ptl (1 − N tl ) + (1 − μ ) N tl

(A.2.4)

The household faces the two budget constraints

Citl + τ tlŠitl + Bitl +1 ≤ (1 + rt ) Bitl + witl , if employed ƒ

ƒ

(A.2.5)

ƒ

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Citl + τ tlŠitl + Bitl +1 ≤ (1 + rt ) Bitl + Šitl , if unemployed l

(A.2.6)

l∗

where Bi ,t and Bi ,t denote bond holdings carried over from the previous period. At the l

beginning of each period. The household receives the real wage, wit when employed and the l

insurance payment, Šit , when unemployed. Its expenditures, either employed or unemployed are consumption, insurance contracts purchased at price τ t and bonds. l

The problem of the household is therefore to solve the Bellman equation — and therefore l∗

maximise the intertemporal utility function — subject to (A.2.5) and (A.2.6). Λ i ,t and Λ i ,t , l

denote the Lagrange multipliers associated with budget constraint of the representative low-

Fabrice Collard, Raquel Fonseca and Rafael Munoz

94

skilled household when employed and unemployed respectively. The problem may be stated as a Lagrangian in the following way:33

L = N tl uil,t + (1 − N tl )uil,∗t + βV L ( Sil,t +1 ) + N tl Λ li ,t ⎛⎜⎝ (1 + rt ) Bitl + witl − Citl − τ tlŠitl − Bitl +1 ⎞⎟⎠ ⎛

ƒ

ƒ

ƒ



+ (1− Ntl )Λli ,∗t ⎜⎜⎜ (1 + rt ) Bitl + Šitl − Citl − τ tlŠitl − Bitl +1 ⎟⎟⎟ ⎜ ⎝

⎟ ⎠

The first order necessary conditions associated to the problem are therefore First order lƒ

conditions with respect to Cit , Cit , Šit are: l

l

(Citl − Γl ) −1 = Λ lit ƒ

(A.2.7) ƒ

(Citl − Γl ƒ ) −1 = Λ itl

(A.2.8) ƒ

−α il,tτ tl Λ itl + (1 − α il,t )(1 − τ tl )Λ itl = 0

(A.2.9) lƒ

Equations (A.2.7) and (A.2.8) state that the Lagrange multipliers Λ it and Λ it are equal to the l

marginal value of consumption, considering the alternative states of being employed or unemployed respectively. The first order condition (A.2.9) represents the marginal value to be fully insured against unemployment. lƒ

l

The first order conditions related to Bit +1 and Bit +1 are:

Λ =β l it

l∗ it

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Λ =β

∂V L ( Sil,t +1 )

(A.2.10)

∂Bitl +1 ∂V L ( Sil,t +1 )

(A.2.11)



∂Bitl +1

Finally the envelope therefore implies

∂V L ( Sil,t ) ∂B

l it

33

= (1 + rt )Λ lit

Note that the Lagrange multipliers are just normalized by

(A.2.12)

N tl

and

1 − N tl

for convenience.

Spanish Unemployment and the Ladder Effect

∂V L ( Sil,t ) ∗

∂Bitl

ƒ

= (1 + rt )Λ itl

95

(A.2.13)

Envelope conditions together with first order conditions related to optimal portfolio composition yields the following Euler conditions:

Λ lit = β (1 + rt +1 )Λ itl +1 ƒ

(A.2.14)

ƒ

Λ lit = β (1 + rt +1 )Λ lit +1 .

(A.2.15)

Insurance Company for low-skilled households: The expected profit of the insurance company is given by the difference from the gain of the prime of insurance times the insurances and the payment of insurance to people who were unemployed. We assume free entry on the insurance market. Profits are driven to zero such that

Π t = τ tlŠitl − (1 − N tl )Šitl = 0 We assume the company insures the current unemployment risk. The company can not differentiate from the state of low-skilled households. People insurance against the probability of being unemployed and the optimal prime for the company is

τ tl = 1 − N tl .

(A.2.16)

Using these results in the optimal choice for insurance (A.2.16) in equation (A.2.9), we end up with ∗

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Λ li ,t = Λ li ,t for low-skilled households, whether they were employed or unemployed in the previous period. This condition actually states that the marginal utility of wealth is independent from the state of the household in the labour market, which therefore eliminates heterogeneity in saving behaviour. Using this condition together with equations (A.2.7) and (A.2.8), we obtain ∗



Citl = Citl + Γl − Γl . Low-skilled households have different consumption levels when l∗

employed or unemployed, Ci ,t and Ci ,t , but they prefer to be completely insured in terms of l

utility. It also implies that equations (A.2.14) and (A.2.15) are equivalent and households accumulate the same quantity of bonds whether they are employed or unemployed, ∗

Bitl +1 = Bitl +1 = Bit +1 . As a matter of fact, households choose to be completely insured, and have the same wealth in any state. This implies that saving decisions are independent of the

Fabrice Collard, Raquel Fonseca and Rafael Munoz

96

employment history of the household. The optimal insurance is found using constraints (A.2.5) and (A.2.6): ∗

Šitl = wijl ,t + Γ l − Γl . As soon as households choose to be fully insured, and provided the employed labour force is selected randomly across jobs at the beginning of each period, we are back with the standard representative consumer, and the notation used to distinguish employment status may be eliminated. The representative low-skilled household i maximises the expectation of the discounted sum of its instantaneous utility with respect to the consumption and the assets she holds: ∞

∑β t =0

t

⎧ ⎪ ⎨ ⎪ ⎩

ƒ ⎫

N tl uil,t + (1 − N tl )uil,t ⎪⎬⎪ ⎭

subject to (A.2.17). Equation (A.2.17) is the consolidated budget constraint after we introduce perfect assurance: ƒ

N tl Cil,t + (1 − N tl )Cil,t + Bil,t +1 ≤ N tl wijtl + (1 + rt ) Bil,t .

(A.2.17)

A.2.2.The Highly-Skilled Consumer Problem Like the problem of the low–skilled worker, the highly–skilled worker faces different states on the labour market. ƒ

Chh,it , Clh,it , and Cith denote the respective highly-skilled household’s consumption and ∗

Γ h , Γl and Γ h can be interpreted as a utility cost, expressed in terms of goods, associated h l with the situation of the household in the labour market. We suppose that Γl = Γ , given that h∗

they work in a low-skilled job. We further assume Γ > Γ > Γ , which ensures that consumption of an employee in a highly-skilled vacancy is greater than that in a low-skilled vacancy and both greater than that of an unemployed household. h

l

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ƒ

Bhh,it , Blh,it and Bith denote contingent claims purchased by the highly-skilled household in the previous period. At the beginning of each period, the household receives the value of bonds purchased in the previous period, highly-employed, low-employed or unemployed. She also receives the wage when employed as a highly-skilled, the wage when employed as a lowskilled plus the insurance payment for this unsatisfactory situation and the insurance payment when unemployed. Its expenditures either employed or unemployed are consumption, insurance and purchase of bonds. The instantaneous utility contingent to be employed in a highly-skilled or low-skilled job and unemployed are,

uhh,it = log(Chh,it − Γ h ) (employed in a highly − skilled job)

(A.2.18)

Spanish Unemployment and the Ladder Effect

97

ulh,it = log(Clh,it − Γl ) (employed in a low − skilled job) ∗

(A.2.19)



uith∗ = log(Cil,t − Γl ) (unemployed)

(A.2.20)

As in the problem of a low–skilled worker, the highly-skilled household maximises the recursively problem

V H ( Sith ) = N hh,t uhh,i ,t + N lh,t ulh,i ,t + (1 − N hh,t − N lh,t )uih,t∗ + βV H ( Sith+1 ),

(A.2.21)

where state variables have three alternative states depending whether the household is employed as highly-skilled or low-skilled or unemployed. β ∈ (0,1) is the discount factor. h

h

h

We made use of the fact that by definition of ph ,t , pl ,t and the law of motion of N h ,t +1 and

N lh,t +1 . N hh,t +1 = phh,t (1 − N hh,t ) + (1 − s ) N hh,t

(A.2.22)

N lh,t +1 = plh,t (1 − N hh,t +1 ).

(A.2.23)

Highly skilled worker maximises her household’s problem taking into account her discounted expected value — solve the Bellman equation subject to the budget constraints ∗











Chh,it + τ thŠith + τ th Šith + Bhh,it +1 ≤ (1 + rt ) Bhh,it + whh,it

(A.2.24)

Clh,it + τ thŠith + τ th Šith + Blh,it +1 ≤ (1 + rt ) Blh,it + wlh,it + Šith ƒ

ƒ

ƒ

Cith + τ thŠith + τ th Šith + Bith+1 ≤ (1 + rt ) Bith + Šith h∗

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where Šit



and Šit , are the insurance contracts with respective prices h

(A.2.25) (A.2.26) ∗

τ th and τ th . The

household chooses among both types of contracts to be insured against the probability of being unemployed or employed in a low-skilled job. When she works as a low–skilled h

employee, she receives Šit to compensate the wage differential with respect to a highly– h∗

skilled job. When unemployed, she receives Šit as unemployment benefit. Λ h,it , Λ l ,it and h

h

ƒ

Λ ith denote the Lagrange multipliers associated to budget constraint of the representative highly-skilled household when highly-employed, low-employed and respectively. As in the problem of the low–skilled worker, we first form the Lagrangian

unemployed,

Fabrice Collard, Raquel Fonseca and Rafael Munoz

98

L = N hh,t uhh,i ,t + N lh,t ulh,i ,t + (1 − N hh,t − N lh,t )uih,t∗ + β V H ( Sith+1 ) +N Λ h h ,t

+N Λ h l ,t

⎛ ⎜ h ⎜ h , i ,t ⎜ ⎜ ⎝

⎛ ⎜ h ⎜ l ,i ,t ⎜ ⎜ ⎜ ⎝

(1 + rt ) Bhh,it + whh,it − Chh,it − ⎞⎟ h∗ h∗ t it

−τ Š − τ Š − B h h t it

h h ,it +1

⎟ ⎟ ⎟ ⎠

(1 + rt ) Blh,it + wlh,it + Šith − Clh,it − ⎞⎟ h ∗ h∗ t it

⎟ ⎟ ⎟ ⎟ ⎠

−τ Š − τ Š − B h h t it ⎛ ⎜ ⎜

ƒ

h l ,it +1



ƒ



(1 + rt ) Bith + Šith − Cith − ⎟⎟⎟ ⎟. (1− N − N )Λ ⎟ h h h∗ h∗ hƒ ⎜ ⎟ ⎜⎜ −τ t Šit − τ t Šit − Bit +1 ⎟⎟ h h ,t

h l ,t

h∗ ⎜ i ,t ⎜⎜ ⎝





h∗

The first order conditions with respect to Ch,it , Cl ,it , Cit , Šit and Šit are: h

h

h

(Chh,it − Γ h ) −1 = Λ hh,it A.2.27

(A.2.27)

(Clh,it − Γl ) −1 = Λ lh,it A.2.28

(A.2.28)



ƒ

ƒ

(Cith − Γ h ) −1 = Λ ith A.2.29

(A.2.29) ƒ

− N hh,tτ th Λ hh,it + N lh,t (1 − τ th )Λ lh,it − (1 − N hh,t − N lh,t )τ th Λ ith = 0 ∗





(A.2.30)

ƒ

− N hh,tτ th Λ hh,it − N lh,tτ th Λ lh,it + (1 − N hh,t − N lh,t )(1 − τ th )Λ ith = 0.

(A.2.31)

Equations (A.2.27), (A.2.28) and (A.2.29) state that the Lagrange multipliers Λ h,it , Λ l ,it and h

h

ƒ

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Λ ith are equal to the marginal value of consumption in the corresponding state on the labour market. The first order conditions (A.2.30) and (A.2.31) represent the marginal values to be fully assured taking into account the probability of being employed in a low-skilled job or unemployed respectively. h

h



The first order conditions related to Bh,it +1 , Bl ,it +1 and Bit +1 are:

Λ

h h ,it

∂V H ( Sith+1 ) =β ∂Bhh,it +1

(A.2.32)

∂V H ( Sith+1 ) ∂Blh,it +1

(A.2.33)

Λ lh,it = β

Spanish Unemployment and the Ladder Effect ∗

Λ ith = β

∂V H ( Sith+1 ) ƒ

∂Bith+1

.

99

(A.2.34)

h

h



Envelope theorem related to Bh,it , Bl ,it and Bit yields

∂V H ( Sith ) = (1 + rt )Λ hh,it ∂Bhh,it

(A.2.35)

∂V H ( Sith ) = (1 + rt )Λ lh,it ∂Blh,it

(A.2.36)

∂V H ( Sith ) h∗ it

∂B



= (1 + rt )Λ ith .

(A.2.37)

Envelope conditions together with first order conditions related to optimal portfolio composition yields the following Euler conditions:

Λ hh,it = β (1 + rt +1 )Λ hh,it +1

(A.2.38)

Λ lh,it = β (1 + rt +1 )Λ lh,it +1

(A.2.39)

ƒ

ƒ

Λ ith = β (1 + rt +1 )Λ ith +1 .

(A.2.40)

Similarly to the low-skilled problem, we can express the problem of an insurance company for highly-skilled households as ∗





Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Π = τ thŠith + τ th Šith − N lh,tŠith − (1 − N hh,t − N lh,t )Šith = 0, h∗

from which we get τ t = N l ,t and τ t = (1 − N h,t − N l ,t ) . h

h

h

h

This result together with household’s first order conditions implies:

Λ hh,it = Λ lh,it = Λ ith

ƒ

Bhh,it =Blh,it =Bithƒ which, together with the Frischian demand for consumption, yields

Fabrice Collard, Raquel Fonseca and Rafael Munoz

100

Chh,it − Γ h = Clh,it − Γl = Cith ƒ − Γ ƒ . Therefore, the levels of insurance are given by Šit = wh,it + Γ − Γ − wl ,it h



h

l

h

h

and



Šith = whh,it + Γ h − Γ h . To summarise, as in the low–skilled worker problem, as soon as households choose to be fully insured, and provided the employed labour force is selected randomly across jobs at the beginning of each period, we are back with the standard representative consumer, and the notation used to distinguish employment status may be eliminated. A representative highlyskilled household i therefore maximises the expectation of the discounted sum of her instantaneous utility with respect to the consumption and the assets she holds: ∞

∑β t =0

t

⎧ ⎪ ⎨ ⎪ ⎩

ƒ ⎫ ⎪ ⎬ ⎪ ⎭

N hh,t uhh,i ,t + N lh,t ulh,i ,t + (1 − N hh,t − N lh,t )uih,t

subject to (A.2.41). Equation (A.2.42) is the consolidated budget constraint when we introduce perfect assurance: ƒ

N hh,t Chh,i ,t + N lh,t Clh,i ,t + U thCih,t + Bih,t +1 ≤ N hh,t whh,ijt + N lh,t wlh,ijt + (1 + rt ) Bih,t .

(A.2.41)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Appendix A.3 : Wage determination This section is devoted to the exposition of the wage bargaining process, which is determined by a Nash bargaining criterion, therefore yielding a surplus sharing rule. At the beginning of every period a re-negotiation simultaneously occurs between the firm and workers of each group. Otherwise there will be many wages as workers and the macroeconomic dynamic would be done with respect to a wages distribution. Therefore, workers negotiate their wages with the firm and they account for separation and hiring probabilities. The wage setting behaviour is obtained maximising the following Nash criterion with respect to the wages which maximise the weighted product of the workers’ and the firm’s net return from the different job match. τ

The gains of the firm correspond to marginal values. Thus, let Ω l , jt the surplus of the firm associated to each group of employment. The gains for workers’ correspond to the sum of the utilities when they are employed minus the sum of utilities when the negotiation fails and they become unemployed. Let

Ψτlit Λ it

the different surplus of workers associated to each

group of workers in terms of goods. Where (τ , l) = {(h, h), (h, l ), (l ,.)} and

ξlτ are the

exogenous parameters of bargaining powers of each group of workers. Thus, the Nash bargaining criterion problem to solve is

Spanish Unemployment and the Ladder Effect τ τ ⎞1−ξl ⎟ l , jt ⎠

⎛ max ⎜Ω τ ⎝

wlijt

101

ξlτ

⎛ Ψτlit ⎞ ⎜ ⎟ . ⎝ Λ it ⎠

We define the surplus of the firm and the workers as following:

A.3.1. The Firm j Surplus Let us first characterise the surplus which accrues to a firm j when it employs a highly– skilled worker on a highly–skilled position. This is essentially given by the marginal value of a highly–skilled employment, Ω h, jt . From the optimal condition associated with a highly– h

skilled employment, we get

Ω hh, jt = θ

Y jt h

LN

− whh,ijt + (1 − s ) X hh, jt

h h , jt

(A.3.1)

Likewise in the case of a highly–skilled worker employed on a low–skilled position, the marginal value of employment is given by

Ωlh, jt = (1 − α − θ )

Y jt h

LN

h l , jt

+LN l

l jt

− wlh,ijt .

(A.3.2)

Finally, when it hires an additional low–skilled worker, this gain is given by

Ωljt = (1 − α − θ ) h

Y jt h

LN

h l , jt

+LN l

l jt

− wijtl + (1 − μ ) X ljt

(A.3.3)

l

where X h, jt and X jt are the Lagrange multipliers associated to employment laws of

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

motion.

A.3.2.The Household i Surplus Let us now present the determination of the surplus which accrues to each type of worker. The low–skilled worker: A low–skilled worker, when employed in period t , instantaneously derives utility associated to her extra gains in the labour market ( Λ it times the wage revenues net of disutility of labour in terms of goods). Furthermore, in the next period, she may still be employed with probability (1 − μ ) or laid off with probability μ . Therefore, the utility gain of an employed low–skilled worker in the labour market, ϒ l ,it , is given by e

Fabrice Collard, Raquel Fonseca and Rafael Munoz

102

ϒle,it = Λ it ( wijtl − Γl ) + β ⎛⎜⎝ (1 − μ ) ϒle,it +1 + μϒul ,it +1 ⎞⎟⎠ where Λ it 34 in the marginal utility of consumption. Likewise, when unemployed, the lowskilled household gets, ϒ l ,it u

*

ϒ ul ,it = −Λ it Γl + β ⎛⎜⎝ pl ,t ϒle,it +1 + (1 − pl ,t ) ϒ ul ,it +1 ⎞⎟⎠ Then, the net surplus of a low-skilled worker in the labour market is given by ƒ

Ψ lit ≡ ϒ le,it − ϒ lh,it = Λ it ( wijtl + Γ l − Γ l ) + β (1 − μ − pl ,t )Ψ lit +1

(A.3.4)

The highly–skilled worker: A highly–skilled worker may be employed either in a highly–skilled or a low–skilled vacancy, or may be unemployed. When employed as a highly–skilled worker, she instantaneously derives utility Λ it (Wt − Γ ) , and in the following period she may be h

h

unemployed or employed on either a highly or a low–skilled position. Therefore, the utility gain in the labour market for a highly–skilled worker employed on a highly–skilled position is given by ⎛ ⎜

ϒ hh,it = Λ it ( whh,ijt − Γ h ) + β ⎜⎜ ⎜ ⎝

(1 − s ) ϒ hh,it +1 + splh,t ϒlh,it +1 + ⎞⎟ ⎟ ⎟ ⎟ ⎠

+ s (1 − plh,t ) ϒ uh,it +1

h where ϒ h,it , ϒl ,it and ϒ uit respectively denote the utility gain when a highly–skilled worker

h

h

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

employed in a highly–skilled job, in a low–skilled job or unemployed worker. Following the same procedure, the value of a highly-skilled worker employed in a low– skilled position is

ϒ

h l ,it

= Λ it (W

h l ,ijt

−Γ )+ β l

⎛ ⎜ ⎜ ⎜ ⎜ ⎝

phh,t ϒ hh,it +1 + (1 − phh,t ) plh,t ϒ lh,it +1 + ⎞⎟ + (1 − plh,t )(1 − phh,t ) ϒ uh,it +1

and similarly the value when unemployed is

ϒ

34

h u ,it



= −Λ it Γ + β

⎛ ⎜ ⎜ ⎜ ⎜ ⎝

phh,t ϒ hh,it +1 + (1 − phh,t ) plh,t ϒ lh,it +1 + ⎞⎟ + (1 − plh,t )(1 − phh,t ) ϒ uh ,it +1 l∗

From Appendix A.1, we can see that Λ i ,t = Λ i ,t = Λ i ,t l

⎟ ⎟ ⎟ ⎠

.

⎟ ⎟ ⎟ ⎠

Spanish Unemployment and the Ladder Effect

103

The surplus of a highly–skilled worker, when employed on a highly–skilled position, is actually given by the gain from being employed as a highly-skilled worker minus the gain from being employed as a low–skilled worker times the probability that this event occurs, minus the gain from being unemployed, times the probability of being unemployed. Such that the overall surplus is given by (using the law of large number) ƒ

Ψ hh ,it ≡ ϒ hh,it − plh,t −1ϒ lh,it − (1 − plh,t −1 ) ϒ uh ,it = Λ it ( wlh,ijt + Γ h − Γ l )

Ψ hh,it = Λ it ( whh,ijt − Γ h − ⎛ N lh,t + ⎜1 − ⎜ 1− N h h ,t ⎝

Nlh,t 1− N

h h ,t

( wlh,ijt − Γl )

⎞ h∗ h h ⎟⎟ Γ ) + (1 − s − ph,t )Ψ h,it +1 . ⎠

(A.3.5)

When bargaining on a low–skilled job, the only opportunity left is unemployment such that the surplus is now given by the difference between the utility gains from being employed on a low–skilled position minus the utility gains from being unemployed ƒ

Ψ lh,it ≡ ϒ lh,it − ϒ uh,it = Λ it (Wl ,hijt + Γ h − Γ l ).

(A.3.6)

We now examine the Nash bargaining process that determines the real wage in each case. The Nash–Bargaining process The wage setting behaviour is obtained maximizing the following Nash criterion

⎛ ⎜ ⎝

max Ω τ Wlijt

τ ⎞1−ξl l , jt ⎟⎠

τ

⎛ Ψτlit ⎞ ⎜ ⎟ ⎝ Λ it ⎠

ξlτ

where (τ , l ) = {( h, h ), ( h, l ), (l ,.)} . The first order condition associated to this program – Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

making use of the definitions of the corresponding surpluses — yields τ

Ψ l ,it ξ lτ τ Ω = . l , jt Λ it 1 − ξ lτ The marginal values of employment of the household and the firm together with the above equation implies

Fabrice Collard, Raquel Fonseca and Rafael Munoz

104

⎛ h ⎜ ⎜ h ⎜ ⎜ ⎝

whh,t = ξ θ

Yt + phh,t X L N hh,t h

⎞ h ⎟ ⎟ h ,t ⎟ ⎟ ⎠

⎛ ⎞ Γh + ⎜ ⎟ + (1 − ξ hh ) ⎜ N lh,t h l ⎟ ⎜ + 1 − N h ( wl ,t − Γ ) ⎟ h ,t ⎝ ⎠

⎛ Y wlh,t = ξlh ⎜ (1 − α − θ ) h h jt l l ⎜ L N l , jt + L N jt ⎝ ⎛ l ⎜⎜ ⎜ ⎜ ⎝

w = ξ (1 − α − θ ) l t

Y jt h

LN

h l , jt

+LN l

l jt

(A.3.7)

⎞ h l ⎟⎟ + (1 − ξ h )Γ ⎠

(A.3.8)

⎞ l ⎟⎟ t ⎟ ⎟ ⎠

(A.3.9)

+ płt X + (1 − ξ l )Γ l

assuming a symmetric equilibrium, and using the fact that, in our calibration, We have l∗

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

assumed Γ = Γ

h∗

= 0.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 4

JUVENILE DELINQUENCY AND ITS FORENSIC CONSIDERATIONS B. R. Sharm1 and Aparajita Sharma2 Dept. of Forensic Medicine and Toxicology, Govt. Medical College & Hospital, Chandigarh, 160030, India Govt. College for Girls, Sector 11, Chandigarh, India

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Juvenile Delinquency and the problems related to it have been faced by all societies, all over the world; however, in the developing world the problems are all the more formidable. The process of development has brought in its wake a socio-cultural upheaval affecting the age-old traditional ways of life in the congenial rural milieu. In fact, various scientific advances and concomitant industrialization and urbanization have ushered in a new era, which is characterized by catastrophic changes and mounting problems. Cities have sprung up with heterogeneity of population, cultural variations, occupational differentiations and overcrowded conditions. As a result, social disorganization and maladjustment have taken place following a perennial influx of people from their rural habitat to the urban squalid slums. Juveniles are adversely affected by these changing conditions. At the same time, the traditional social control system that served as a preventive check against any antisocial activity is gradually giving way. Consequently, the problem of juvenile deviance and antisocial propensities is rearing its ugly head. This paper examines the causes of delinquency and attempts to analyze the prevalence and prevention of deviance among adolescents and young adults.

Keywords: delinquency; juveniles; juvenile delinquent; juvenile justice system; conduct disorder 1 2

E-mail: [email protected] # 1156 – B, Sector – 32 B, Chandigarh – 160030. India

106

B. R. Sharma and Aparajita Sharma

INTRODUCTION The term juvenile delinquency applies to violation of criminal code and certain patterns of behavior that are not approved for children and young adolescents. It may be grouped as individual delinquency (in which only one individual is involved and the cause of delinquent act is traced to individual delinquent), group supported delinquency (committed in companionship and the cause is attributed not to the personality of the individual but to the culture of the individual’s home and neighborhood), organized delinquency and situational delinquency [1]. A delinquent young person is disobedient and wayward, runs away from home and school, cannot be controlled by the parents and teachers, is not amenable to any kind of discipline, is self-willed and habitually acts in a manner injurious to the welfare and happiness of others and himself. Different scholars have classified juvenile delinquents on different basis. Hirsch classified them in six groups on the basis of kinds of offences committed: Incorrigibility (for example, disobedience and keeping late hours), Truancy (staying away from school), Larceny (ranging from petty thefts to armed robbery), Destruction of property (both public and private), Violence against individual or community and sexual offences ranging from homosexuality to rape. Eaton and Pole, classified delinquents into five groups according to the offence: Minor violations (disorderly conduct and minor traffic violations), Major violations including thefts, Property violations, Addiction, and bodily harm including homicide and rape. Trojanawicz classified them as accidental, ill socialized, aggressive, occasional, professional and gang-organized. Psychologists have classified juvenile delinquents on the basis of their individual traits or the psychological dynamics of their personality into five groups: mentally defective, psychotic, neurotic, situational and cultural [2]. Workers from different fields, proposed different theories of delinquency from time to time.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

SOCIOLOGICAL THEORIES Sociological theories of juvenile delinquency put emphasis on the environment, social structures and the learning process. Merton’s Anomie Theory [3] states that when there is a discrepancy between the institutionalized means that are available within the environment and the goals that individuals have learnt to aspire for in their environment, strain or frustration is produced, norms break and deviant behavior may result. Shaw and McKay’s Cultural Transmission Theory [4] described that delinquency is transmitted through personal and group contacts, and lack of effective social control agencies contributes to the high incidence. George Herbert Mead’s ‘Role Theory’ and ‘Theory of Self’ stated that becoming a delinquent and assuming a criminal identity involves more than merely associating with law violators. The associations have to be meaningful to the individual and supportive of a role and selfconcept that he wants to be committed to [2]. Cloward and Ohiin’s Success and Opportunity Structure Theory [5] states that faced with limitations on legitimate avenues of access to their goals and unable to revise their aspirations downwards, the lower socioeconomic class youths experience intense frustrations that result in their exploring non-conformist and illegitimate alternatives.

Juvenile Delinquency and Its Forensic Considerations

107

Cohen’s theory of the delinquent sub-culture [6] seeks to explain working class delinquency and the logic of the theory has implications for culturally supported deviance in general. It assumes that delinquency is in some way related to a discrepancy between culture goals and the availability of legitimate opportunities for achieving them. It assumes that in a society, children of all social levels are to a great degree judged by the same set of standards, and their self-respect and sense of adequacy are largely determined by their performance in terms of these standards, especially when they move out of the home situation and find themselves in competition with other youth in school and occupational settings. Cohen called these standards ‘the middle-class criteria of status’ because they are most consistently applied by those segments of the population that control the gateway to success, such as teachers, business and professional people, ministers and civic leaders. They include such things as a high level of ambition, an ability to defer gratification of immediate needs in the interest of achieving long term goals, self discipline, the possession of skills of potential academic, occupational, and economic value, and so on. Growing up in working-class homes is less likely to produce young people with the ability to perform well in terms of these criteria and being disadvantaged in the competitive pursuit of status, they are likely to find themselves ‘at the bottom of the heap’ with their self-respect damaged. Despite the sub-cultural theory having met with a number of criticisms, there is evidence that the differences among delinquent groups and the delinquent and non-delinquent groups are in fact related to differences in their opportunity structure.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

PSYCHOBIOLOGICAL THEORIES Psychobiological theories assume that independent variables in determination of crime and delinquency are some observable or hypothetical aspect of the biological structure of the person that manifests in his/her behavior. The prototype of such theories is Cesare Lombroso’s theory of the born criminal. Lombroso’s conception of born criminal, though sharply criticized at the earlier stage attracted many scholars to work on his general idea of a ‘criminal type’. The American anthropologist E. A. Hooton claimed, on the basis of his comparison of criminals and non-criminals that criminals showed various traits of biological inferiority and degeneration and that criminality was the behavioral manifestation of such biological inadequacy [7]. Sheldon [8] reported that the ‘mesomorphic physique’ characterized by a sturdy and muscular athletic frame work associated with a distinctive type of temperament called ‘somatotonia’ characterized by such traits as assertiveness of posture and movement, love of physical adventure, abounding and restless energy, need for and enjoyment of exercise, produce aggressive, energetic, daring types of people that include generals, athletes, and politicians as well as the delinquents. Sheldon’s research has been subjected to devastating criticism but his ideas were given a new lease on life by a later study [9]. However, it continues to be debated as to whether the temperament of the mesomorph, rooted in an inherited biological constitution has an inherent propensity for danger, violence, and predatory behavior, or because the strenuous and active life of the lower classes, streetdwelling delinquent tends to harden the body into a mesomorphic contour, or whether the boys with the strength and energy of the mesomorph find it easier than their scrawnier or

108

B. R. Sharma and Aparajita Sharma

more corpulent peers to gain status through the kind of accomplishments that are rewarded in the delinquent gang.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

PSYCHIATRIC THEORIES Psychiatric perspectives on crime and delinquency rest heavily on conceptions of motivation and personality derived from psycho-analytic theory. According to this theory, behavior is motivated by impulses and drive energy of a fundamentally biological nature (the Id), and modified by socialization experiences, which provide the individual with the capacity for thought and rational assessment (the Ego), and internal restraints in the form of conscience (the Superego). In the classic statement of this position, delinquent behavior results when the restraining forces are too weak to curb inherent aggressive and destructive tendencies. Such tendencies are universal to the species. While holding to the insight that ‘every man has within him the capacity to commit the most objectionable antisocial acts, no matter how civilized or sophisticated his social training has made him, modern psychiatry has moved considerably beyond the simple determinism implied in the earlier model. Halleck [10] observed that ‘while some aggressive and some sexual activity is often correlated with a weakening of control mechanisms - - - the act of law violation is often a deliberate, planned and complicated operation which may require a great deal of ego strength. Halleck and many other psychiatrists see behavior as adaptive, and problem solving. Specifically crime is seen as an adaptation to stress - - - best understood in terms of the manner in which the individual experiences the biological, psychological and socially determined situations of his existence. Just as the causal factors of delinquency are diverse and numerous, so are the definitions. Sociologists define deviance as any behavior that members of a social group define as violating their norms. This concept applies both to criminal acts of deviance as well as to noncriminal acts that members of a group view as unethical, immoral, peculiar, sick, or otherwise outside the bounds of respectability [11]. According to another definition, delinquency is a condition arising in the matrix of socio-personal disorganization in the sequence of experience and influences that shape behavior problems. It is the product of dynamic social process, involving numerous variables and the failure of personal and social controls. It is a symptom of deep socioeconomic and social ailments [12]. According to the legal concept, a delinquent juvenile is one who commits an act defined by law as illegal or delinquent, and who is adjudicated ‘delinquent’ by an appropriate court.

PUBLIC HEALTH, MEDICINE AND PSYCHOLOGY Researchers in public health, medicine, and more recently, psychology have come to appreciate the value of studying poverty in its own right. Initially this meant descriptive analyses demonstrating physical and psychological sequelae of poverty or low socioeconomic status but psychologists have begun to move beyond a social address perspective, turning their attention to underlying explanations for poverty’s harmful impacts on children and their families. Many studies in the recent past, reported that poor children confront widespread

Juvenile Delinquency and Its Forensic Considerations

109

environmental inequities. Compared with their economically advantaged counterparts, they are exposed to more family turmoil, violence, separation from their families, instability, and chaotic households. Poor children experience less social support, and their parents are less responsive and more authoritarian. Low-income parents are less involved in their children’s school activities. The air and water poor children consume are more polluted. Their homes are more crowded, noisier, and of lower quality. Low-income neighborhoods are more dangerous, offer poorer municipal services, and suffer greater physical deterioration. Predominantly low-income schools and day care are inferior. The accumulation of multiple environmental risks rather than singular risk exposure may be especially pathogenic aspect of childhood delinquency [13-48].

DELINQUENCY AND CONDUCT DISORDER

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Delinquency is sometimes equated with conduct disorder however, although known to overlap, they may not be the same. Many delinquents do not have conduct disorder (or any other psychological disorder) while many of those with conduct disorders do not offend. Conduct disorders are characterized by severe and persistent antisocial behavior, which is more serious than ordinary childhood mischief. In the preschool period, the disorder usually manifests as defiant and aggressive behavior in the home, often with over activity, disobedience, temper tantrums, physical aggression to siblings or adults, and destructiveness, while in later childhood, the disturbance often becomes evident outside especially at school as truanting, delinquency, vandalism, and reckless behavior, or alcohol or substance abuse. To constitute conduct disorders, these behaviors have to be persistent and both ICD-10 [49] and DSM-IV [50] require the presence of at least three symptoms from a list of 15 for a duration of at least 6 months. The disorder is commonly found in children from unstable, insecure, rejecting families living in deprived areas, broken homes, or from homes in which family relations are poor. While genetic factor may play a role, children with brain damage and epilepsy are prone to conduct disorders. On the other hand, delinquency is related to low social class, poverty, poor housing, and poor education, and the family factor seems to play an important role. A study reported that half of boys with criminal fathers are convicted, compared with a fifth of those with fathers who are not criminals [51].

Neglect, Exploitation and Hostile Home and Societal Environment Neglect, exploitation and hostile home and societal environment would nurture deviant attitude among juveniles. Juveniles engaged in prostitution and children of sex-workers; children abused sexually or physically; underfed children; abandoned children; slum children; street children; refugee children; and juvenile servitude constitute the pool of neglected and destitute juveniles [52]. Within the family, children may be forcibly engaged in domestic servitude and used as first choice to assist parents on the field by the small farm families. Outside the family, they are engaged as indented servitude in workshops, hotels, small industries, footpath vending, fire-works, carpet weaving etc. For the employers, child workers

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

110

B. R. Sharma and Aparajita Sharma

are advantageous as they cannot form unions, could be exploited for longer hours for meager wages and could be even used in hazardous and unhygienic work environment [53]. Despite the Universal Human Rights Declaration aimed at safeguarding the rights of children - leisure, learning and play – the social reality continues to be grim. Child labor is not a new phenomenon, what is, however, new is its perception as a social problem. In the recent past, there has been a distinct change in the value orientation and attitudinal ethos of the legitimizing groups of society vis-à-vis child labor [54]. In the pre-industrial agricultural society in developing countries, children worked as helpers and learners in hereditarily determined family occupations under the benign supervision of adult family members. The work place was an extension of the home and the work was characterized by personal and informal relationship. The tasks of technology that the work involved, were simple and nonhazardous, which the child could learn smoothly, almost unconsciously over the years through imitation and association. With the advent of industrialization and urbanization, the social scenario changed. The family members no longer work as a team and in castesanctioned occupations. The child has to work as an individual person, either under an employer or independently, without enjoying the benevolent protection of his guardian. His work exposes him to various kinds of health hazards emanating from the excessive use of chemicals and poisonous substances in industries and the pollutants discharged by them [55]. Child abuse or maltreatment is commonly divided into five categories; physical abuse, emotional abuse, sexual abuse, neglect and exploitation. Although any of these forms may be found separately, they often occur together [56]. Studies reveal that abuse and neglect of children occurs in families from all walks of life, and across all socioeconomic, religious, and ethnic groups. There is no single, identifiable cause of child maltreatment; rather, it occurs as a result of an interaction of multiple forces impacting the family. For example, lack of preparation or knowledge of critical issues surrounding parenting, financial or other environmental stressors, difficulty in relationships, stress of single parenting, and depression or other mental health problems can all lead to abusive or neglectful behavior [57,58]. The ways in which children are raised vary considerably between and within cultures. Subsequent research has confirmed that certain parenting styles tend to correlate with certain behavior in the children, although the outcomes are by no means absolute. The ‘authoritarian’ style, characterized by strict, inflexible rules, can lead to low self-esteem, unhappiness, and social withdrawal. The ‘indulgent-permissive’ style, which includes little or no limit setting, coupled with unpredictable parental harshness, can lead to low self-esteem, poor impulse control, and aggression. The ‘indulgent-neglectful’ style, one of uninvolvement in the child’s life and rearing, puts the child at risk of low self-esteem, impaired self-control and increased aggression. The ‘authoritative-reciprocal’ style marked by firm rules and shared decision making in a warm, loving environment, is believed to be the style most likely to result in selfreliance, self-esteem, and a sense of social responsibility [59]. However, it is generally agreed that a number of factors that play an important part in a youngster’s delinquent behavior can be divided into two groups, individual factors and situational factors. The individual factors include personality traits like submissiveness, defiance, hostility, impulsiveness, feeling of insecurity, fear, lack of self-control and emotional conflicts while situational factors may be attributed to family, companions, movies, school environment, work environment etc.

Juvenile Delinquency and Its Forensic Considerations

111

NEED TO UNDERSTAND ADOLESCENCE

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Adolescence is characterized by profound biological, psychological and social developmental changes. The biological onset of adolescence is signaled by rapid acceleration of skeletal growth and the beginnings of physical sexual development. The psychological onset is characterized by acceleration of cognitive development and consolidation of personality formation. Socially, adolescence is a period of intensified preparation for the coming role of young adulthood. At the beginning of adolescence, thinking usually becomes abstract, conceptual, and future oriented. Many adolescents show remarkable creativity, which they express in writing, music, art, and poetry. Creativity is also expressed in sports and in adolescent’s interests in the world of ideas – humanitarian issues, morals, ethics, and religion [60]. The peer group factor plays an all-important role at this stage. The school experience accelerates and intensifies separation from the family. More and more, adolescents live in a world unfamiliar to parents. Home is a base; the real world is school, and the most important relationships, besides the adolescent’s family, are the persons of similar ages and interests. Adolescents attempt to establish a personal identity separate from their parents but close enough to the family structure. Although adolescents tend to rely on peers for day-to-day support, the social support provided by parents has a stress-buffering effect in emergency situations [61]. Adolescents often view themselves through the eyes of their peers, and any deviation in appearance, dress code, or behavior can result in diminished self-esteem. Parents need to be aware of the sudden, frequent changes in friendships, personal appearance, and interests but must abrogate their authority. Risk-taking behavior in adolescence can involve alcohol, tobacco, and other substance use; promiscuous sexual activity; and accident-prone behavior, such as fast driving, skydiving, and hang gliding. Most mortality statistics for teenagers cite accidents as the leading cause of death, with vehicular accidents accounting for about 40% of all teenage deaths [62, 63]. The reasons for risk-taking behavior vary and may relate to counterphobic dynamics, the fear of inadequacy, the need to affirm a sexual identity, and group dynamics such as peer pressure. The behavior may also reflect some adolescent’s omnipotent fantasies, in which they view themselves as invulnerable to harm and injury.

UNDERSTANDING CRIME Crime is a universal problem and exists in every society in one form or the other. It is related to human behavior, but not all behaviors fall in the category of crime. The nature of crime is so complex that criminal behavior cannot have a unanimous definition though it has been defined both in social and legal terms. The social explanation of crime emphasizes the non-legal aspects and in this sense, ‘crime is a behavior or an activity that disregards the social code of a particular community. It looks for the source of crime in the vary nature of society rather than the biological or psychological nature of the individual. According to a legal definition, ‘crime is the intentional act or omission in violation of criminal law committed without defence or justification.’ In other words, ‘crime’ refers to all acts which

112

B. R. Sharma and Aparajita Sharma

are forbidden by law and which have an adverse effect on social interests, and for which punishment is prescribed by law. Earliest reference to the word "crime" dates back to 14th century when it conveyed to the mind something reprehensible, wicked or base. Any conduct which a sufficiently powerful section of any given community feels to be destructive of its own interest, as endangering its safety, stability or comfort is usually regarded as heinous and it is sought to be repressed with severity and the sovereign power is utilized to prevent the mischief or to punish anyone who is guilty of it. A crime presents three characteristics: (1) it is a harm, brought about by human conduct which the sovereign power in the State desires to prevent; (2) among the measures of prevention selected is the threat of punishment; and (3) legal proceedings of a special kind are employed to decide whether the person accused did in fact cause the harm, and is, according to law, to be held legally punishable for doing so. Protection of society is the basic reason of treating some acts as crime. Indeed it is one of the aims of punishment [64].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Why Is a Law Enacted? What Object(s) It Seeks to Serve? It is from the time of the Renaissance and the Reformation when men, as a result of these great revolutionary movements broke away from rule of custom and tradition, that legislation began its career as an instrument of social and political, and even religious, change. However, the laws made must respect the right to liberty and property; and laws must be made for the good of the people. The laws and legislation should be in conformity with the spirit of the people, its traditions, its philosophy of life, even the physical surroundings of the people, including the climate. Macaulay believed in the efficacy of law in improving people and their character. He wrote: "When a good system of law and police is established, when justice is administered affordably and firmly, when idle technicalities and unreasonable rules of evidence no longer obstruct the search for truth, a great change for the better may be expected which shall produce a great effect on the national character" [65] According to Ihering, law is a means to an end. He laid down the following general principles of legislation: “Laws should be known to be obeyed; should answer expectations; should be consistent with one another; should serve the principle of utility; should be methodical; must be certain to be obeyed, must not become a dead letter; are necessary to ward off the danger of the operations of egoism or self-interest, the ordinary motives of human action. Law and legislation must aim at justice, which is that which suits all. Laws are interconnected and like human beings lean on one another" [65] The fundamental principles on which the political life of the people is based are individuality, equality and justice. After securing the life and liberty of the State and of the individual, laws and legislations take on the task of serving and promoting the good life of the State and the people. For good life, morality is necessary and to maintain morality legislation is a must. Legislation, therefore, is the framework, which is required to be made for good life.

Juvenile Delinquency and Its Forensic Considerations

113

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

EVOLUTION OF JUVENILE JUSTICE SYSTEM In most countries of the world, the concept of delinquent behavior is confined to the violation of the ordinary Penal Laws of Country carried out by boys or girls up to the age of eighteen years. State laws prohibit two types of behavior for juveniles: the first includes behavior, which is criminal for adults, as for example, murder, rape, fraud, burglary, robbery, etc. and the second includes status offenses like running away from home, unruly or ungovernable truancy, etc. Juvenile justice is commonly understood as a notion of fairness and justice and an alternative system of dealing with children through laws. Here the emphasis is protective, restorative, and reintegrative with care and rehabilitation. Thus the Juvenile Justice aims at evolving effective mechanisms and creating the necessary environment for care, protection, development and rehabilitation of juveniles in conflict with law. Jane Addams and other turn-of-the-century social reformers brought the juvenile court into being in order to rescue misbehaving children from the harshness and rigidity of the adult criminal justice system [66]. In contrast to the criminal justice system, designed to apprehend, prosecute, and punish offenders for the protection of society, the goals of the child saving movement were the identification, evaluation, and treatment of maladjusted youths for their own benefit, and ultimately the society’s. Beginning in Chicago in 1899 and eventually spreading throughout the world, the juvenile court was envisioned as a humane, informal, treatment-oriented social agency that would save delinquent children from what the reformers saw as the failure of many immigrant and other impoverished parents to cope with the adverse effects on their children of an increasingly urban and industrial environment. Initially, the jurisdiction of the juvenile court was broad, encompassing both criminal and non-criminal misbehavior. Children who had committed no crimes but who were alleged to be truant, disobedient, engaging in undesirable behavior, associating with unacceptable companions, or otherwise beyond the control of their parents (ungovernable) were brought before the juvenile court, together with those who were charged with criminal offences, in order that they might receive the treatment, discipline, and care they required. These so-called ‘status offenders’ were treated the same as youths who were charged with criminal offences, though in many jurisdictions they were given a different label, such as ‘persons in need of supervision’. However, there were theoretical flaws and practical problems in the child saver’s program for curing juvenile delinquency. From the outset, juvenile courts lacked the resources even to attempt to carry out their program. There were not enough qualified mental health professionals available to the court to perform the evaluation, referral, and treatment services that were required. The coercive nature of involuntary court-imposed therapy, however benevolently intended, must have seemed punitive from the child’s perspective. In addition, the court’s treatment-oriented approach was thought to provide insufficient protection for the community from certain juveniles who committed serious crimes or were repeat offenders. Thus as early as 1903, only four years after its establishment, the juvenile court of Chicago transferred 14 children to the adult criminal system [67]. On the other hand, critics pointed to the unbridled discretion, rehabilitation pretension, and punitive realities of the juvenile court. The rising tide of criticism of the juvenile court culminated in the 1960s and thereafter in legislative reforms and judicial decisions that mandated procedural protection for juveniles and narrowed the court’s jurisdiction [68]. The

114

B. R. Sharma and Aparajita Sharma

New York Family Court Act of 1962 introduced due process including its provision for ‘law guardians’ by appointing independent lawyers to represent juveniles who came before the court. In 1966, the United States Supreme Court decided the landmark case of Kent v. United States, holding that the transfer, or ‘waiver’ of a juvenile to an adult criminal court was “an invitation to procedural arbitrariness and a decision of such tremendous consequences that it ought not be made without a hearing that provided the essentials of due process and fair treatment” [69]. In 1967, the Supreme Court expanded on its decision on Kent and ordered the provisions of due process protection to all juveniles in delinquency proceedings, characterizing the existing informal juvenile court procedures as “a kangaroo court” [70]. Finally, in 1977 the United States Congress enacted legislation requiring states seeking federal financial assistance for juvenile delinquency program to cease confining status offenders (children charged with non-criminal misbehavior) in secure detention or correctional facilities intended for juvenile delinquents [71]. However, the combined effect of the introduction of due process protections into juvenile delinquency proceedings, due process restrictions on the transfer of serious offenders to adult courts, and the effective removal of status offenders from the court’s delinquency jurisdiction, together with an apparent increase in the incidence of serious juvenile crime, have left the juvenile courts with a clientele that is more violent, more disturbed, and less amenable to superficial therapeutic interventions than that envisioned by Jane Addams and her fellow child savers.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

AN OVERVIEW OF THE INCIDENCE OF JUVENILE DELINQUENCY Despite the fact that psychology and psychiatry have succeeded in entering our schools, our courts, our legislatures, our workplaces, and, bit by bit, our homes, the newspaper headlines continue to report about an increase in youth violence, robberies, theft and deceit, a decay of morality, child abuse, alcoholism, a growing drug industry and usage, the disintegration of the family, an increase in teenage pregnancies and teen suicide, a lack of motivation among the students. At one time, sending children to school meant a guarantee of a structured, nurturing, and effective education; today parents are concerned about the failure of modern education declining moral standards and escalating drug abuse, crime, and violence in schools. Comparing the data on juvenile crime rates continues to be a challenge because most of the countries use different ways to define “juvenile,” define “violent crimes,” classify crimes, and measure crime rates. Therefore, the researchers generally restrict themselves to comparing trends based on police figures of arrests and convictions. However, in countries where alternative data are available (e.g. victim surveys, self- report studies, health care and vital statistics), these often present a different picture. The Office of Juvenile Justice and Delinquency Prevention and the Federal Bureau of Investigation providing current statistical data on juvenile arrests reported continuing decline in almost every major category of crime. However, despite the continuing decline in arrest rates, juvenile offenders in residential placement (n= 108,931 on 10/27/99) increased by 3% from 1997 to 1999. The total includes a 12% increase in placement of juveniles in a residential facility for a technical violation. Between 1980 and 2000, the arrest rate for all

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Juvenile Delinquency and Its Forensic Considerations

115

offenses reflected a 35% increase for juvenile females and a decline of 11% for juvenile males [72]. The United States Department of Health and Human Services, Administration on Children, Youth and Families, through the National Child Abuse and Neglect Data System, released April 2002, reported that 879,000 children were found to be victims of child maltreatment. Maltreatment categories typically include neglect, medical neglect, physical abuse, sexual abuse, and psychological maltreatment. Almost two-thirds of child victims (63%) suffered neglect (including medical neglect); 19% were physically abused; 10% were sexually abused; and 8% were psychologically maltreated [73]. According to a report, in 1981, juveniles accounted for over one out of three arrests for robbery, one out of every three arrests for crime against property, one out of six arrests for rape, and one out of eleven arrests for murder. In 1981, about one teenager out of every fifteen in the nation was arrested. Between 1983 and 1991, crimes committed by juveniles under eighteen showed another staggering increase: robberies increased five times, murders tripled, and rapes doubled. More than five hundred children arrested for rape in 1991 were twelve. During the 1996-1997 academic school years, 6,093 students were expelled for bringing firearms or explosives to school. Some 14000 young people were attacked on school property every day, and 160 thousand children missed school every day because of the fear of violence. A large percentage of these at-risk, antisocial youth are vulnerable to gang recruitment and membership [74]. From 1980 through 2002, the proportion of murders with a juvenile offender that also involved multiple offenders gradually increased. In the first half of the 1980s, about one-third of all murders with juvenile offenders involved more than one offender; in 2002, this proportion was nearly half (48%). Similarly, the proportion of murders with a juvenile offender that also involved an adult gradually increased, from less than 25% in the first half of the 1980s to 39% in 2002 [75]. According to a report [76], the rate of juvenile violence rose sharply in the mid-1980s or early 1990s. In some countries, the official figures increased between 50 and 100 percent. In England and Wales (counted together) in 1986, for example, approximately 360 of every 100,000 youths ages 14–16 were “convicted or cautioned by the police” for violent crimes while in 1994, that figure had climbed to approximately 580 per 100,000. In Germany the growth rate of juvenile delinquency was even higher. In 1984, the number of 14- to 18-yearolds suspected of violent crime in the former West Germany was approximately 300 per 100,000; by 1995, that figure had more than doubled, to approximately 760 per 100,000. Rates in the former East Germany were between 60 and 80 percent higher [75]. In Britain, the total rate of violent crimes against individuals had increased a frightening 1,200% during the period 1960 to 1993. The number of robberies has increased by 2,700% in the same period. 93% of Britain's crime was against property. Although the total number of reported crimes in France has fallen slightly in recent years, juvenile delinquency has continued to rise sharply by 81% over the past ten years. One in five of those charged was under eighteen [74]. Other Western countries' crime rates parallel the trend of the United States, Canada, Britain and France. In Australia, the number of serious assaults, for example, has risen 391% between 1973 and 1992, while the robbery rate increased 190%. In New Zealand, the total number of violent offences increased 615% between 1960 and 1990, from 2,937 to 20,987. The crime rate in Greece also increased 1,268% between 1980 and 1990. In Germany the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

116

B. R. Sharma and Aparajita Sharma

assault and theft rose 71%. During 1993 - 1997, the number of crimes committed by German children up to age fourteen surged 10.1% [76]. A report published by Juvenile division of the National Policy Agency of Japan reported the year-wise incidence as 141775 in 2002, 144404 in 2003, 134847 in 2004, 123715 in 2005 and 112817 in the year 2006 reflecting a decrease in the incidence of crimes committed by young children and adolescents [77]. According to a report, juvenile courts, all over China tried 393,543 defendants aged less than 18 years during the period 1999 to 2005 convicting 393,115 and acquitting 428 [78]. According to the survey, China’s approximately 1.5 million prison inmates include 19,000 juveniles. Official statistics show that 317,925 juveniles were arrested from 1998 to 2003, making up 7.3 percent of the criminal suspects arrested during that period. Chinese authorities arrested 69,780 juveniles in 2003, accounting for 9.1 percent of all criminal suspects arrested that year, and an increase of 12.7 percent over the number of juveniles arrested in 2002. 75.3 percent of the juveniles were held on allegations of encroaching on the property of others. 17.4 percent were charged with assault and infringing on the rights of others [79]. Another survey by a juvenile delinquency prevention office found that most young criminal suspects were aged between 15 and 16. In 1996 this age group made up 95.4 percent of Beijing’s juvenile offenders, that dropping to 82 percent in 2002. However, an increasing number of offenders aged under 14 have been arrested in recent years [79]. A study of data published by the National Crime Records Bureau, Ministry of Home Affairs, Government of India [80] on incidence and rate of juvenile delinquency in India reported that the incidents of juvenile crimes have declined from 12588 in 1991 to 9267 in 2000. However, during the same period offences like burglary, arson, hurt, molestation etc. by young persons have increased. Despite the declining incidence of juvenile delinquency both at the absolute, and the relative levels, it is too often reported that rural and urban India have pervasive practices of child labor, juvenile servitude, domestic juvenile servitude and trafficking of juvenile girls [81]. Such reports demand for examination of the problems confronting juveniles. In terms of number of juvenile offenders in criminal cases in Taiwan for the past ten years (1990 to 1999), the lowest number incurred in 1990. There were 17,286 juvenile in offenders that year. Due to the fact that Department of Health (of Executive Yuan) included Amphetamine as one of drugs under administration, the number of juvenile offenders increased to be 25,472 persons in 1991. In 1992, there were 30,719 offenders. In 1993, there were 30,780 offenders and the number was the highest in the 10-year period. In 1994, the number decreased to be 28,378 persons. In 1995, the number increased to be 29,287 persons. In 1996, the number slightly increased to be 29,680 persons. The number started to decrease continuously from 1997. In 1997, there were 24,766 juvenile offenders. In 1998, there were 23,094 persons. In 1999, there were 21,224 persons. It was a decrease of 8,456 persons or a decrease by 28.49% as compared to 1996 [82]. According to a report from Pakistan, the ratio of juvenile offenders repeating crimes is more than two times that in any other country and the tendency of repeating crimes is growing among child offenders and may increase still further. The total number of detention centers in Pakistan is about 90,000, but none fulfills the legal requirement of appointing a psychologist and the detention was not reformative [83]. In general, the victims of violent crimes committed by juveniles were other juveniles, as evidenced by victimization trends in the European Union countries surveyed. For example, in

Juvenile Delinquency and Its Forensic Considerations

117

the Netherlands in 1995, young people ages 15–17 were four times more likely than adults (25 and older) to be the victims of assault. Juveniles in Germany were also more likely to be the victims of violent crime than were members of other age groups. The victimization rates from 1984 to 1995 for young children and adults were relatively stable however, the victimization rates for teenagers (ages 14–18) and young adults (ages 18–21) climbed precipitously, from approximately 300 per 100,000 in each age group in 1984 to approximately 750 per 100,000 in 1995 [84]. An overview of the state of the crime levels and penal systems in the Scandinavian countries, as portrayed by available statistical sources, indicates that the crime level in Scandinavia (as regards traditional offences) is similar to or lower than that of other western European countries. The extent of drug abuse in the Scandinavian countries also appears to be on a par with or lower than it is in the rest of Europe. Increases in crime rates during the postwar period have been very substantial in the Scandinavian countries just as they have been elsewhere in Europe – indicating that the recorded increases in traditional crime in Europe may have common structural roots. The 1990s have witnessed a stabilization in theft rates, albeit at a high level. Increasing equality between the sexes might have contributed to an increase in the reporting of violent and sexual offences against women (and children), making these offences more visible. The system of formal control in the Scandinavian countries is characterized by relatively low police density, a clear- up rate that has declined, the imposition of fines in a high proportion of criminal cases and relatively low, but recently increasing prison populations [85 - 88].

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

PREVENTION The prevention of juvenile delinquency is an essential part of crime prevention in society. By engaging in lawful, socially useful activities and adopting a humanistic orientation towards society and outlook on life, adolescents and young children can develop noncriminogenic attitudes. The successful prevention of juvenile delinquency requires efforts on the part of the entire society to ensure the harmonious development of adolescents, with respect for and promotion of their personality from early childhood. The need for and importance of progressive delinquency prevention policies and the systematic study and the elaboration of measures should be recognized. These should avoid criminalizing and penalizing a child for behavior that does not cause serious damage to the development of the child or harm to others. Such policies and measures should involve: [a] The provision of opportunities, in particular educational opportunities, to meet the varying needs of young children and to serve as a supportive framework for safeguarding the personal development of all young persons, particularly those who are demonstrably endangered or at social risk and are in need of special care and protection; [b] Specialized philosophies and approaches for delinquency prevention, on the basis of laws, processes, institutions, facilities and a service delivery network aimed at reducing the motivation, need and opportunity for, or conditions giving rise to, the commission of infractions;

118

B. R. Sharma and Aparajita Sharma [c] Official intervention to be pursued primarily in the overall interest of the young children and guided by fairness and equity; [d] Safeguarding the well-being, development, rights and interests of all young persons; [e] Consideration that youthful behavior or conduct that does not conform to overall social norms and values is often part of the maturation and growth process and tends to disappear spontaneously in most individuals with the transition to adulthood; [f] Awareness that, in the predominant opinion of experts, labeling a young person as "deviant'', "delinquent" or "pre-delinquent" often contributes to the development of a consistent pattern of undesirable behavior by young persons.

Implementation of the preventive programs in accordance with national legal systems, should be focused at the well-being of young persons from their early childhood. While the young persons should have an active role and partnership within society and should not be considered as mere objects of socialization or control, Technical and scientific co-operation on practical and policy-related matters, particularly in training, pilot and demonstration projects, and on specific issues concerning the prevention of youth crime and juvenile delinquency should be strongly supported by all Governments, the United Nations system and other concerned organizations. Childhood experiences are important in the development of criminality, however, not all criminals reveal their criminality early in life. While the origins of criminal behavior in childhood are a complex matter, delinquency is reasonably predictable early in some children’s lives. Similarly, antisocial behavior in the form of juvenile delinquency is predictive of adulthood crime. It seems evident, though, that early problem behavior should not be neglected for two reasons – it is predictive of later, more serious, problems and, if it is acted on, then even simple interventions may be effective at reducing future delinquency. For the forensic psychologist, it is useful to know that different risk factors are associated with different patterns of offending, however, it is unrealistic, nevertheless, to expect rapid progress in understanding the development of criminal careers and appropriate caution must be exercised when drawing conclusions.

REFERENCES Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[1] [2] [3] [4]

[5]

Becker HS. (1996) Social problems: A modern approach, John Wiley & Sons, Inc., New York, p 226 – 238. Pradheepa P, Murlidharan K. (2005) Juvenile delinquency – a socioeconomic and family perspective, Social Welfare 51 (10): 9 - 12 Merton RK. (1957) Social theory and social structure, The Free Press, Glencoe, Illinois, Shaw C, McKay HD. (1931) Social factors in juvenile delinquency, US Government Printing Office, Washington Cloward R, Ohiin LE. Delinquency and Opportunity: a theory of delinquent gangs, The Free Press, Glencoe, Illinois 1960 Cohen AK. Delinquent Boys New York Free Press 1955

Juvenile Delinquency and Its Forensic Considerations [6]

[7] [8] [9] [10] [11] [12]

[13]

[14] [15] [16] [17] [18]

[19] [20]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[21]

[22]

[23]

[24] [25]

119

Srivastava VK. Society and crime - an introduction to criminology and the anthropology of crime; Proceedings of National Conference on Advances in Forensic Science 2001; Department of Anthropology, University of Delhi; p.15 - 32. Sheldon WH. Varieties of delinquent youth 1949; New York: Harper. Sheldon EA, Glueck E. Physique and delinquency 1956; New York: Harper. Halleck S. Psychiatry and the Dilemmas of crime 1967; New York: Harper, pp 60 Calhoun C, Light D, Keller S. Sociology, 5th Ed. Alfred A knopt, New York 1989 Tomovic VA. Definitions in sociology: convergence, conflict and alternative vocabularies, Diliton Publications Inc. St. Catherine Ontario 1979 Ackerman BP, Izard CE, Schoff K, Youngstrom E, Kogos J. Contextual risk, caregiver emotionality, and the problem behaviors of six- and seven-year-old children from economically disadvantaged families, Child Development, 1999; 70: 1415–1427. Ackerman BP, Kogos J, Youngstrom E, Schoff K, Izard CE. Family instability and the problem behaviors of children from economically disadvantaged families, Developmental Psychology, 1999; 35: 258–268. Alaimo K, Olson CM, Frongillo EA. Food insufficiency, family income, and health in U.S. preschool and school-aged children, Am J Public Health 2001; 91: 781–786 Aneshensal C, Sucoff C. The neighborhood context of adolescent mental health, J Health and Social Behavior, 1996; 37: 293–310. Bartlett S, Hart R, Satterthwaite D, de la Barra X, Missair A. Cities for children: Children’s rights, poverty and urban managment. London Earthscan 1999 Becker HJ. Who’s wired and who’s not: Children’s access to and use of computer technology. The Future of Children, 2000; 10: 44–75. Bradley RH. The home environment. In S. L. Friedman & T. D. Wachs (Eds.), Measuring environment across the lifespan, Washington, DC: American Psychological Association 1999 Bradley RH, Corwyn RF. Socioeconomic status and child development, Annual Review of Psychology, 2002; 53: 371–399 Bradley RH, Corwyn RF, McAdoo H, Coll C. The home environments of children in the United States: Variations by age, ethnicity, and poverty status. Child Development, 2001; 72: 1844–1867. Brody GH, Ge X, Conger RD, Gibbons F, Murry V, Gerrard M, Simons R. The influence of neighborhood disadvantage, collective socialization, and parenting on African American children’s affiliation with deviant peers Child Development 2001; 72: 1231–1246. Bronfenbrenner U, Evans GW. Developmental science in the 21st century: Emerging theoretical models, research designs, and empirical findings Social Development, 2000; 9: 115–125. Caspi A, Taylor A, Moffitt T, Plomin R. Neighborhood deprivation affects children’s mental health: Environmental risks identified in a genetic design. Psychological Science, 2000; 11: 338–342. Chen E, Matthews KA, Boyce T. Socioeconomic status differences in health: What are the implications for children? Psychological Bulletin, 2002; 128: 295–329. Dearing E., McCartney K, Taylor B. Change in family income-to-needs matters more for children with less, Child Development, 2001; 72: 1779–1793.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

120

B. R. Sharma and Aparajita Sharma

[26] Duyme M, Dumaret A, Tomkiewicz S. How can we boost IQs of dull children? A late adoption study. Proceedings of the National Academy of Sciences, 1999; 96: 8790– 8794. [27] Evans GW. Environmental stress and health, In A. Baum, T. Revenson, & J. E. Singer (Eds.), Handbook of health psychology 2001, Mahwah, NJ: Erlbaum pp. 365–385. [28] Evans GW. A multimethodological analysis of cumulative risk and allostatic load among rural children. Developmental Psychology 2003; 39: 924–933. [29] Evans GW, English K. The environment of poverty: Multiple stressor exposure, psycho-physiological stress and socio-emotional adjustment, Child Development 2002; 73: 1238–1248. [30] Evans GW, Wells NM, Chan E, Saltzman H. Housing and mental health, J Consulting and Clinical Psychology 2000; 68: 526–530. [31] Evans GW, Wells NM, Moch A. Housing and mental health: A review of the evidence and a methodological and conceptual critique. J Social Issues 2003; 59: 475–500. [32] Gennetian LA, Miller C. Children and welfare reform: A view from an experimental welfare program in Minnesota, Child Development 2002; 73: 601–620. [33] Grant KE, Compas BE, Stuhlmacher A, Thurm A, McMahon S, Halpert J. Stressors and child and adolescent psychopathology: Moving from markers to mechanisms of risk. Psychological Bulletin 2003; 129: 447–466. [34] Haines MM, Stansfeld SA, Head J, Job RFS. Multilevel modelling of aircraft noise on performance tests in schools around Heathrow Airport London. International J Epidemiology and Community Health 2002; 56: 139–144. [35] Hoff E. The specificity of environmental influence: Socioeconomic status affects early vocabulary development via maternal speech, Child Development 2003; 74: 1368– 1378. [36] Hoff E, Laursen B, Tardiff T. Socioeconomic status and parenting. In M. H. Bornstein (Ed.), Handbook of parenting, 2nd ed., 2002; Mahwah, NJ: Erlbaum, pp. 231–252 [37] Johnson MP, Ladd H, Ludwig J. The benefits and costs of residential mobility programmes for the poor, Housing Studies 2002; 17: 125–138. [38] Lengua LJ. The contribution of emotionality and self-regulation to the understanding of children’s response to multiple risk. Child Development 2002; 73:144–161. [39] Leventhal T, Brooks-Gunn J. The neighborhoods they live in: The effects of neighborhood residence on child and adolescent outcomes, Psychological Bulletin 2000; 126: 309–337. [40] Leventhal T, Brooks-Gunn J. Moving to opportunity: An experimental study of neighborhood effects on mental health. Am J Public Health 2003; 93: 1576–1582. [41] Linver MR, Brooks-Gunn J, Kohen D. Family processes in pathways from income to young children’s development. Developmental Psychology 2002; 38: 719–734. [42] Magnusson KA, Duncan GJ. Parents in poverty. In M. H. Bornstein (Ed.), Handbook of parenting 2nd ed. 2002; Mahwah, NJ: Erlbaum pp. 95–121. [43] Matte T, Jacobs D. Housing and health: Current issues and implications for research and progress. J Urban Health Bulletin of the New York Academy of Medicine 2000; 77: 7–25. [44] McEwen B S. The neurobiology of stress: From serendipity to clinical relevance. Brain Research, 2000; 886: 172–189.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Juvenile Delinquency and Its Forensic Considerations

121

[45] McNeely CA, Nonnemaker J, Blum R. W. Promoting school connectedness: Evidence from the National Longitudinal Study of Adolescent Health, J School Health 2002; 72: 138–146. [46] National Center for Education Statistics, Condition of America’s public school facilities: 1999 (NCES 2000–032). Washington, DC: U.S. Department of Education 2000. [47] Evans GW. The environment of childhood poverty, American Psychologist 2004; 59 (2): 77 – 92 [48] American Psychiatric Association; Diagnostic and statistical manual of mental disorders: DSM-IV, 4th ed. Washington DC, 1994. [49] World Health Organization; The ICD-10 Classification of Mental and Behavioral disorders; Geneva, 1992. [50] Wizner S. The mental health professional in the Juvenile Justice System In Golder M, Mayou R, Cowen P eds Oxford textbook of psychiatry 4th Ed 2002, New York, Oxford University Press, p 836 – 838 [51] Mukherjee S. Childhood robbed Deccan herald (Sunday) Dec. 14 2003 [52] Sharma BR, Harish D, Bangar S, Gupta M. Violation of children’s rights: can forensic medicine help? J Indian Academy of Forensic Medicine. 2005; 27 (3): 139 - 144 [53] Sundar I. Trends in juvenile delinquency in India, Social Welfare 2005; 51 (10) 26 – 32 [54] Sharma BR, Bangar S. Rights of Children vis-à-vis Reality Today. Social Welfare 2005; 2 (8) 26 – 29 [55] Sharma BR. Medicolegal aspects of child abuse, Physician’s Digest 2005; 14 (1): 41 – 48 [56] Sharma BR, Gupta M. Child rape – save the innocents. J Social Welfare 2004; 51 (9): 3 –8 [57] Sharma BR, Gupta M. Innocence abused: an overview of the nature, causes and prevention of child abuse. J Indian Academy of Forensic Medicine 2003; 25 (3): 87 – 92 [58] Manassis K Child-parent relations: attachment and anxiety disorders In Silverman WK eds Cambridge child and adolescent Psychiatry, New York, Cambridge University Press, 2001 [59] Cotton NS. Normal adolescence, In Sadock BJ, Sadock VA. Eds Kaplan and Sadock’s comprehensive textbook of psychiatry. 7th ed Baltimore: Lippincott Williams and Wilkins, 2000, p 2550 [60] 61). Jarvinen DW, Nicholls JG. Adolescent’s social goals, belief about the cause of social success, and satisfaction in peer relationships. Dev Psychol 1996; 32: 432. [61] Sharma BR, Causes, Impact, Prevention and Medicolegal aspects of Child abuse and Neglect. In Stanley M Sturt edited Child Abuse: New Research, 1st edn. 2006, Hauppauge New York, Nova Science Publishers, Inc. p 147 – 173. [62] Sharma BR, Singh VP, Sharma R, Bangar S. Unnatural deaths in India – a profile. J Indian Academy of Forensic Medicine 2004; 26 (4): 140 – 146. [63] Sharma BR. Is attempted suicide an offence? Aggression and Violent Behavior (in press)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

122

B. R. Sharma and Aparajita Sharma

[64] 6 Sharma BR, Sharma AK, Harish D. Abolition & restoration of Section 309 IPC – an overview. A. A. Internet J Forensic Medicine & Toxicology 2006; 7 (1): Available at http://www.geradts.com/anil/ij/vol_007_no_001/papers/paper003.html [65] Addams J. The Child, the Clinic and the Court, New York, New Republic, Inc. 1925 [66] Cook County Charity Service Report (1920) cited in Twentieth Century Fund Task Force Policy towards young Offenders, Confronting Youth crime 55 (1978) [67] President’s Commission on law enforcement and administration of justice, Task Force Report: Juvenile Delinquency and Youth Crime1967. [68] 383 US 541, 554 – 555, 562 (1966) [69] 387 US 1 (1967) [70] 42 US Code 5633 (12) 1977 [71] Office of Juvenile Justice and Delinquency Prevention (OJJDP) and the Federal Bureau of Investigation (FBI) report on current statistical data on juvenile arrests 2002. http://www.cwla.org/programs/juvenilejustice/htm [72] The U S Department of Health and Human Services, Administration on Children, Youth and Families, through the National Child Abuse and Neglect Data System (NCANDS), The Summary of Key Findings from calendar year 2000 Annual Report (released April 2002) http://www.cwla.org/programs/juvenilejustice/htm [73] Statistics: youth violence, juvenile delinquency, drug abuse. http:// Christian-parentinglearninginfo.org/chap06.htm [74] Snyder HN, Sickmund, M. Juvenile Offenders and Victims: 2006 National Report. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. [75] U.S. Department of Justice, Office of Justice Programs, National Institute of Justice Trends in Juvenile Violence in European Countries 1998. [76] National Policy Agency of Japan, Juvenile Division, 2007 [77] Official Data, available at: www.china.org.cn/english/government/157592.htm [78] Federal Interagency Forum on Child and Family Statistics. 2005. America’s Children: Key Indicators of Well-Being, 2005. Washington, DC: U.S. [79] Crime in India; National Crime Records Bureau, Ministry of Home Affairs, Government of India, 1998, 1999 and 2000 [80] Narayana KS. Dimensions of juvenile problems: institutional and non-institutional, Social Welfare, 2005; 51 (10) 13 – 25 [81] DeVoe J, Peter K, Ruddy S, Miller A, Planty M, Snyder T, Rand M. Indicators of School Crime and Safety: 2003. Washington, DC: National Center for Education Statistics and Bureau of Justice Statistics. [82] Daily Times December 12, 2006. [83] Walter R, McDonald. Child Maltreatment: Reports from the States to the National Child Abuse and Neglect Data System. Washington, DC: U.S. 2005, Department of Health and Human Services, Children’s Bureau.] [84] Annual report of the Finnish Prison and Probation Services, Helsinki: Criminal Sanctions Agency, 2002 [85] Aromaa K. (2000) Trends in Criminality, In: Crime and Criminal Justice in Europe, 2000; Strasbourg, pp. 13-34. [86] Barclay G, Tavares C. International comparisons of criminal justice statistics 2000, Home Office Statistical Bulletin, 12 July 2002. London

Juvenile Delinquency and Its Forensic Considerations

123

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[87] Barclay G, Tavares C. Wilby E. International comparisons of criminal justice statistics 2001, Home Office Statistical Bulletin, 24 October 2003. London.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 5

WAGE AND EMPLOYMENT IN A FINANCE-LED ECONOMY Célia Firmin Centre d’Economie de la Sorbonne (MATISSE) – Université Paris I, Panthéon-Sorbonne – CNRS, France

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Since the beginning of the 1990’s, some OECD countries follow a financialization trend. This trend results in an increase in distributed dividends and a decrease in the accumulation rate and in the wage share. The growth slowdown increases the unemployment rate. This is the case for example in France and Germany. The object of this chapter is to analyse wage and employment determination in a finance-led economy. For this, we will use a post Keynesian “stock-flow” macroeconomic model and numeric simulations. The introduction of financial variables modifies investment and consumption equations’ form. Financialization results in a new macroeconomic dynamic which affects employment and wage determination. Employment, and more precisely unemployment, plays a central role in the income distribution dynamics. In this institutional framework, the distributive conflict modifies the income distribution. This conflict depends on the unemployment rate and on the shareholder negotiation power. It results in wage share fluctuations. These fluctuations are central in the macroeconomic dynamics. Indeed, the consumption evolutions modify the firms’ capacity utilisation rate and so investment. More, financialization represents the shareholders’ power increase in management decision. In this framework, profitability norms set by shareholder in the short term may reduce investment. We will see that, in a Keynesian perspective, the introduction of financial variables increases the effective demand constraint which firms face. The growth slowdown is linked to two determinants. The first is the new income distribution dynamics, with a decrease in the wage share. The second is the new investment behaviour, with an increase in the investment selectivity due to high required profitability norms. So, financialization raises unemployment due to growth slowdown. Then the unemployment rate explains a part of the decrease in the wage share.

126

Célia Firmin We will analyse income distribution effects on investment and consumption functions in a finance-led economy. As we will see, in this context, wage and employment represent the adjustment variables.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

INTRODUCTION Since the beginning of the 1990’s, some OECD countries have followed a financialization trend. This trend results in an increase in distributed dividends and a decrease in the accumulation rate and in the wage share. The growth slowdown increases the unemployment rate. This is the case for example in France and Germany. The interaction between income distribution and macroeconomic dynamics is a key question in post-Keynesian analysis. It can be tackled from different points of view. While some authors study the evolution of factors determining income distribution and notably analyse the impact of growth on distribution (for example, Kaldor, 1956), others take the exact opposite approach, by analysing income distribution effects on growth (Kalecki, 1965). Nevertheless, most post-Keynesian papers are based on the assumption of a managerial economy, in which the influence of financial actors is not fully taken into account. That said, a trend to financialization has been developed by heterodox economists in many countries, from various points of view: from an institutional perspective (Dore, 2000 and 2002), a Marxian one (Duménil and Lévy, 2001, 2005), a regulationist one (Boyer, 2000 and Plihon, 2003), or even from a post-Keynesian angle (for example, Cordonnier, 2006 and Stockhammer 2004). Some authors give a broad definition of financialization. For Dore (2002), the concept of financialization specifies changes which occur in national models of capitalism, at institutional and macroeconomic levels. The definition used here is close to Dore’s and also follows the French regulation school, so that financialization is defined as the transition to an accumulation regime in which financial institutions occupy the top position in the institutional hierarchy (Aglietta and Rébérioux 2005, Boyer 2000, Plihon 2002 and 2003). The operation of the accumulation regime is then widely determined by the configuration of financial institutions and its evolution. More generally, financial actors’ decisions occupy a central place in macroeconomic dynamics (Duménil and Lévy, 2001, 2005). This idea is similar to Stockhammer’s (2005-6): financialization is a regime in which shareholders influence firms’ objectives and behaviour and, in this way, macroeconomic dynamics. Changes in monetary and financial institutions have led to a macroeconomic dynamics which is partly linked to changes in the forms of financing investment (Plihon 2003 and Boyer 2000). This has resulted in new investment and consumption behaviour. It may therefore be asked how changes in financing investment have induced knock-on effects on the economy. The consequences of financialization on income distribution will be dealt here from a post-Keynesian perspective. This issue leads to questions about the relationship between financialization, income distribution and growth. How does financialization influence income distribution? How do changes in income distribution act on economic growth when behaviour has changed along with financialization? More precisely, it is pertinent to understand the relationships between the decreasing wage-share and the accumulation rate since the mid-1980s, in a context of increasing equity issues and higher distributed dividends. We will see that, in a Keynesian perspective, the introduction of financial variables increases the effective demand constraint which firms face. The growth

Wage and Employment in a Finance-Led Economy

127

slowdown is linked to two determinants. The first is the new income distribution dynamics, with a decrease in the wage share. The second is the new investment behaviour, with an increase in the investment selectivity due to high required profitability norms. So, financialization raises unemployment due to growth slowdown. Then the unemployment rate explains a part of the decrease in the wage share. We will see that, in this context, wage and employment represent the adjustment variables. To analyse the links between financialization, unemployment and income distribution, we will see in a first section the Post-Keynesian model of income distribution and financialization. Then, we will propose a “stock-flow consistent” model within a Kaleckian framework and we will solve it with numerical simulations.

POST-KEYNESIAN MODELS OF INCOME DISTRIBUTION AND FINANCIALIZATION Post-Keynesian Models of Income Distribution The post-Keynesian approach to income distribution rests mainly on two schools of thought: the Cambridge school and the Kaleckian tradition. A third framework can be added, based on “wage-led – profit-led” models. The Cambridge models are built on a full-employment assumption. Theses models analyse the growth effects on income distribution and not the contrary, as Kaleckian models do. Investment doesn’t determine the production level but the income distribution. In these models, the evolution of distribution ensures a balance between saving and investment (Kaldor 1956, Pasinetti 1962). The variation of the total amount of profits is the mechanism which provides this stability. An exogenous rise in investment induces an increase in the profit-share of value-added which in turn increases saving. Indeed, in theses models, the propensity to save is constant and the profit share is linked to the investment on income ratio. The adjustments and the return to stability rest on the differences between the propensities to save of households and firms and on prices flexibility. One of the main results of these models is that the profit rate is set only by a macroeconomic relationship. It depends on the accumulation rate and the firms’ propensity to save. So, the profit rate is independent from Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

technological factors:

FT g with g the accumulation rate and s p the capitalists’ = K sp

propensity to save. By this way, wage-earners behaviours don’t act upon the distribution between wages and profits and upon the profit rate (Pasinetti 1962, Kaldor 1966). Kaldor (1966) introduces financial markets and firms owned by shareholders. In this model, the adjustment mechanism is different from other Cambridge models (for example Kaldor, 1956 or Pasinetti, 1962). Indeed, in Kaldor (1966), it is not the income distribution but fluctuations of the valuation ratio which leads to goods-market equilibrium. Those fluctuations depend on the evolutions of the equities rate of return if the profit rate and the accumulation rate are determined. These evolutions are explained by households saving behaviours and by equities behaviours by firms.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

128

Célia Firmin

In the Cambridge models (Kaldor 1956 and 1966; Pasinetti, 1962; Robinson, 1962), the profit share is thus growing with the investment. In other words, there is an inverse relationship between the real wage rate and the profit rate. This is not the case in Kaleckian models. The kaleckian approach analyses the determinants of income distribution and the links between distribution and growth. This approach is set in a context of the underutilization of productive capacity (Kalecki, 1990). Kalecki (1990) points out two factors of income distribution. Firstly, the degree of monopoly on the goods market, contributes to set up the markup rate fixed by firms on unit cost composed in a large way by wages. Then, the distributive conflict constitutes the second factor of income distribution. The distribution fluctuations have consequences on growth regimes. The Kaleckian models establish a positive relationship both between the real wage rate and the accumulation rate, as well as between the profit rate and the real wage rate (Kalecki 1965, Stockhammer 1999). In theses models in fact, a fall in wages has no reason to induce a fall in prices if the goods market structure is unchanged. By this way, the real wage diminishes and induces by consequences a decline in wage-earners consumption and so in the economic activity and employment. It induces a decrease in the amount of profits. Profits are determined by past decisions in investment and by capitalist’s propensity to save. The distribution effects on economic activity are linked to differences in the propensity to consume between social classes: between wage-earners and capitalists. In the majority of Kaleckian models, the investment function is determined by the capacity utilization rate and by the current profit rate (Stockhammer, 1999 and Lavoie, 19921). The main result of these models is the recessionary effects on the utilisation rate and the accumulation rate, due to an increase in the mark-up rate and hence the profit-share. Cambridgian and Kaleckian models thus emphasize different relationships between the wage-share and the accumulation rate. The main divergence concerns the assumption about the degree of productive capacity utilization. Lastly, the “wage-led – profit-led” models analyse the links between income distribution and growth. The analysis rests on the idea that profit-share is one of the factors of accumulation (Bhaduri and Marglin, 1990). The model is also based on the assumption of capacity under-utilization. Wage-earners’ propensity to consume is equal to 1 and profits are paid only to capitalists. Based on these assumptions, the redistribution of profits to wages induces a rise in consumption and a fall in saving. Total income fluctuations rest on the investment sensibility to the demand recovery and to the profitability decrease. In these models, the accumulation rate is determined both by the utilization rate and profitability. The effects of a rise in real wages, and so of a decrease in the markup rate and the profit share, depends on the investment sensibility to these two factors. Economic activity can thus be led by an increase in real wages or in profits, according to the extent of the capacity effect or the profit effect. The reactions of the investment function thus determine the dynamics of the economic system. The assumptions and the objects to be analysed in the different kinds of post-Keynesian models thus lead to several conclusions about the role of distribution and its effects on macroeconomic dynamics. While an increase in the wage-share is always a factor of recovery in Kaleckian models, the improvement of investment is instead accompanied by a rise in the 1

These elements are used, between others, in “stock-flow” models of Godley and Lavoie (2001-02) or Dos Santos and Zezza (2004), even if they used others variables to determine the accumulation rate.

Wage and Employment in a Finance-Led Economy

129

profit-share in Cambridgian models but, in this case it is a consequence and not a cause. In “wage-led – profit-led” models, distribution is also a factor of growth but the effects of a change in the wage-share are linked to the economic situation. These three kinds of models have been extended to the present institutional context, in order to analyse the effects of financialization.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The Links Between Financialization and Income Distribution in Post Keynesian Models Few Cambridge analyses exist of the links between financialization and income distribution. Commendatore (2003) develops a model that extends Kaldor (1966), by trying to specify the profit rate determinants in an economy in which shareholders influence firms’ management decisions. Nevertheless, in contrast to Kaldor’s model, the valuation ratio is exogenous, while the investment share financed by equity issue is endogenous. A valuation ratio value superior to unity leads to a greater profit-share in national income. Social classes’ introduction in the model leads to a generalized version of Cambridge equation. The profit rate becomes determined by capital accumulation, by capitalist propensity to save out on profits and on capital gains, and by financial market imperfection degree. As for managerial economy, wage-earners propensity to save has no impact on income distribution. In a shareholder economy, the profit rate and income distribution are thus determined by capital market structure, through the valuation ratio; by firms’ financing strategies with the share of investment financed by internal funds along with the desired investment; and by capitalist saving decisions. Some of the analyses on the links between financialization and income distribution are developed from a Kaleckian standpoint. The models which deal with this issue analyse how a change in exogenous variables induces a change in investment and profits. Stockhammer (2005-06) considers an economy in which shareholders have important power on the orientation of management decisions. In this context, as with Kaleckian models which study managerial economies, an increase in the profit-share or in the propensity to save induces a slow down in growth. A rise in shareholder power leads firms to trade off between growth and profits, which tends to decrease the desired level of investment. However, the improvement in shareholder power may induce a rise in equity prices, which leads to an increase in household wealth. This wealth effect may enlarge consumption and thus balance the direct negative impact on investment. Nevertheless, Stockhammer, as well as Poterba (2000), underline the weakness of this effect. An increase in shareholder power also leads to a decrease in the amount of investment per unit of profit. In other words, for a given amount of profit, investment is lower than in managerial economies. This result raises the issue of determining the origin of other sources of profits within the institutional framework. Thus Cordonnier (2006) develops the role of the distribution of dividends and their consumption as the explanation of the increase in profits, as investment stagnates. Dividend payments do not actually represent a production cost for firms but only a form of spending which diminishes firms’ liquidity, but not their profits. Dividends consumption can thus create differences between capital profitability and the accumulation rate, and is comparable to a profit multiplier. In other words, capital holders tend to substitute consumption to investment to make profits. Van Treeck’s analysis (2007) converges on this idea in proving that an increase

130

Célia Firmin

in dividends on the capital ratio induces a decrease in the saving rate, and then a recovery in growth. Kaleckian analyses are thus centred on the question of the links between the decline in investment and the increase or stagnation of profits, in a context of slowing growth (Stockhammer 2004). For Lazonich and O’Sullivan (2000), the new institutional context is characterized by “downsize and distribute” strategies, which formalize a new income distribution composed of a strong concentration of equity holdings, which are unfavourable to wage-earners. Kaleckian analyses focus is thus to explain the maintenance of profits in an investment slowing down context and to see how a change in the propensity to save and in the profit share exert an influence on growth. Finally, the last analytical tradition of the links between financialization and income distribution is in line with “wage-led – profit-led” models. Boyer (2000) developed a model analysing the institutional possibility of a regime led by household consumption of wealth. This regime would then be “finance-led” and its dynamics would rest on income distribution fluctuations. Indeed, the increase of financial profitability norms tends to reduce real wages, but also to raise financial property incomes. By this way, financialization modifies households’ income composition and level. Wealth effect linked to financial assets holding then tends to modify consumption behaviours. On the other hand, financial profitability norms change the investment function equally, inducing a negative effect on accumulation. These norms thus influence firms and households behaviours and the income distribution. It creates a new demand regime. Indeed, if households have fully integrated financial asset income into their behaviour, then a rise in the real wage bill may reduce consumption. The effects of a wage cut are thus unspecified and are determined by the regime configuration. Four demand regimes may exist, depending on the intensity of the wealth effect and the accelerator effect. If the wealth effect is important and if the development of financial markets leads to a generalization of investment behaviour determined by profitability, then an increase in the profit-share may induce a positive effect on demand. A virtuous circle of financial growth can exist in this configuration. However, if the economy is still dominated by wage relationships, so that wages remain the first determinant of consumption, then an increase in financial profitability norms creates a slow down in economic activity. In this way, the development of financial markets extends the wealth-based regime, but raises economic instability.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

FINANCIALIZATION AND EMPLOYMENT: A MODEL Some Theoretical Links that Have To Be Specified From the analyses presented above, it would be interesting to examine more precisely three main theoretical links. From a Kaleckian perspective, it is interesting to question the ways in which financialization affects income distribution. For this, it is essential to take into account the distributive conflict that is exogenous in Kaleckian models (Lavoie and Godley 2001-2; Stockhammer, 2005-6). Boyer (2000) represents a model formalising the endogenous determination of the wage-share, in a finance-led accumulation regime and taking into account distribution conflict. Indeed, in Kaldorian models, the income distribution depends from macroeconomics and behavioural variables. In Kaleckian analyses on contrary, the

Wage and Employment in a Finance-Led Economy

131

markup rate and so the income distribution is usually presented as an exogenous variable (Lavoie and Godley, 2001-2; Stockhammer, 2005-6). Yet, many studies show the decrease trend of the wage share in value added, in a financialization context. In the framework of a Kaleckian analysis, it is meaningful to introduce shareholders’ influence in distribution conflicts. This allows analysis of the dynamics of how income distribution is determined in a finance-led economy. For this, social classes have to be introduced into the model, rather than the distinction between households and firms, as in Commendatore (2003). The determination of the mark-up rate thus becomes endogenous. Besides, few macroeconomic models integrate simultaneously income distribution dynamics, their impact on growth, and the behavioural function changes linked to financialization. In particular, it is interesting to analyse how investment has evolved with the financialization process and how distribution changes influence investment within this institutional framework. In the Boyer model (2000), an increase in the profit-share may raise investment. This idea has to be qualified in a Kaleckian point of view. In contrast, it has been shown that, in the Stockhammer model (2005-6), a rise in the profit-share is followed by a decline in investment and economic activity. In other words, the analysis focuses on how a recovery in the profit-share is followed by a rise in the profit rate or not. With the possible wealth effects linked to the development of equity holdings by households, it is essential to see how the Kaleckian result of a recessionary effect of a decrease in the wage-share is maintained in a finance-led economy. Financialization analysis relates to the study of the investment financing modalities impact on its own specification and, more, on most of behavioural function modifications and thus the macroeconomic dynamics. We will point out the links that should exist between the investment financing modalities and income distribution determining. Finally, from a Kaleckian perspective, it is useful to examine more deeply the links between dividend consumption and its effects on growth and the profit rate, as developed among others by Cordonnier (2006) and Van Treeck (2007). In particular, it is worth analysing how these types of behaviour may influence income distribution, by setting up a specific macroeconomic dynamics. In other words, we will analyse how their influence on growth and profits can create wider effects, mainly on distribution and employment.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Hypothesis and Methodology The model presented in this section is a “stock-flow consistent” model, which draws on the methodology developed by Godley and Lavoie (2001-2 and 2007), among others. The accounting structure of the model depends on the working-out of transactions and balancesheet matrices (see annexes). These matrices constitute the model skeleton (Taylor, 2004) and its first construction stage. As pointed out by Dos Santos and Zezza (2004), with this methodology, the accounting framework is rigorously specified and integrates financial flows and stocks, with complete balance-sheets. Each financial asset holds a counterpart on liability: “Every financial asset has a counterpart liability, and budget constraints for each sector describe how the balance between flow of expenditure, factor income and transfers generate

132

Célia Firmin

counterpart changes in stocks of assets and liabilities2”. This framework allows a complete description of the model and also a check of theoretical hypotheses: each flow has to come from somewhere and to go somewhere and saving as loans add up to the wealth or debt stocks. By this way, columns and lines are equal to zero. More, this framework allows identifying precisely the relationships between transactions of different sectors, in the same time period as between time periods. Agents’ balance-sheets are thus interdependent. Once these matrixes are developed, each variable is defined by a behavioural function or an accounting identity. The model is then resolved by numerical simulations. When it converges on a steady-state, the model’s properties are analysed by introducing a shock on one of the exogenous variables. The economy described here is a closed economy without a public sector (neither State nor central bank), so that:

Y =C + I

(1)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Banks do not hold net wealth because the interest rates on loans and deposits are equal. Firms just issue equities and do not hold monetary liquidities, as in the Godley and Lavoie model (2001-2). Loans are just grants to firms and there is no inflation. Money is endogenous and follows firms’ demand for financing (Keynes 1937). Loans supply can not be greater than demand and there is no credit rationing in the model. Firms can hold excess productive capacities but there is no inventory stock. Firms’ sector use two production factors, capital and work, but it is vertically integrated, intermediate consumptions are thus not taken into account. The distinction between social classes is not made according to income level but by type of income. Two “pure” household types are distinguished: wage-earner households and shareholder households. Their propensities to consume differ, as in Kaleckian models. The economy is made up of four institutional sectors: wage-earner households, shareholder households, firms and banks. It is also made up by three markets: goods and services market, equities and credit market. Exponents on variables represent the demand (d) or the supply (s) and indexes represent the institutional sectors (a for shareholder households, w for wage-earner households, f for firms and b for banks). The index (-1) represents the reference period, for example for interests paid on loans contracted in the year n-1.

A Kaleckian Model on Financialization and Wage and Employment Wage-Earner and Shareholder Households Wage-earner households receive only their wages as income and are considered here as a “pure” case, with a propensity to consume equal to one, as in Taylor (1991) or Kalecki (1965). The wage-earners’ consumption function may thus be written as follows, with α = 1 being the propensity to consume and W wages:

2

Godley and Lavoie (2001-2), p.101-102

Wage and Employment in a Finance-Led Economy

133

C w = α ⋅W

(2)

The development of financial activities is followed by a rise in distributed dividends. This results in the development of new consumption behaviour, which rests on property income. Wage-earners are not the sole type of consumers, another type has to be added, namely shareholder households:

C = Cw + Ca

(3)

Shareholder households have to take two decisions. First, they have to decide the amount of their income which will be consumed, then they have to decide on the allocation of their savings between deposits and equities. The effects on consumption of capital gains, linked to equity price fluctuations, are taken into account here. The shareholders’ consumption function takes the following form, where β 1 is the propensity to consume for shareholders’ income,

Ra , β 2 the propensity to consume on capital gains and CG a capital gains: C a ( +1) = β ⋅1 Ra + β 2 ⋅ CGa

(4)

Kaldor’s hypothesis (1955-56) of different propensities to consume or saving between work and capital incomes is used here. The propensity to save profits is superior to that of wages, so that: α > β 1 . Moreover, some studies argue in favour of a weak wealth effect on consumption (Stockhammer 2005 and Poterba 2000). Thus, a different propensity to consume on dividends and interest and on capital gains, is used:

1 > β1 > β 2 > 0 Shareholders’ income is constituted of dividends, Da and interest on deposits, rm ⋅ M a (−1) : d

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

R a = D ad + rm ⋅ M a (−1)

d

(5)

The interest rate is exogenous, rm = rm and M a (−1) represents the deposits made in the s

previous period. Dividends are set by firms, D f , and only shareholders hold equities so that:

D ad = D sf

(6)

Capital gains depend on equity prices, pe variations related to the equities stock of the previous period:

CG = Δp e ⋅ e (−d 1)

(7)

Célia Firmin

134

The second shareholders’ decision is about their wealth distribution between equities and deposits. Equity purchases depend on their liquidity preference, γ 1 , as well as anticipated a

rates of return on equities ( re ) and deposits, rm . The introduction of rates of return into portfolio behaviour comes from Tobin (1969). In other words, shareholders desire to hold a proportion of their wealth, Va in equities, but this proportion is modulated by the relative rates of return on equities and deposits:

[

(

)]

ead ⋅ pe = γ 1 + γ 2 rea − rm ⋅ Va

(8)

Shareholders’ wealth is made up of deposits and equities, and depends on their saving, S a , plus capital gains:

Δ(Va ) = S a + CG

(9)

With their saving:

S a = Ra − C a

(10)

The rate of return on equities depends on the rate of dividend payments and on capital gains accruing on the equities’ stock in value:

re ( +1) =

D sf ( +1) + CG ( +1) e sf ⋅ p e

(11)

The anticipated rate of return on equities depends on the anticipation of dividend payments and on equity price variations:

r = Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

a e

D sf ( a ) e sf

Δp ea + pe

(12)

As anticipations are based on the past, shareholder households refer to past periods, taking into account the dividend growth rate and the equity price as indicators of evolution:

D sf ( a ) e sf

D sf ( −1) ⎛ ΔD sf ( −1) ⎞ ⎜1 + ⎟ = s e f ( −2) ⎜⎝ D sf ( −2 ) ⎟⎠

And: Δp e = Δp e (−1) a

(13)

Wage and Employment in a Finance-Led Economy

135

The equity price clears the equities market, as e f = e a : s

* e

p

(e ⋅ p ) − p = * e

* e

d

⋅ Δe sf

(14)

e sf ( −1)

According to their budget constraint, the variation in deposits is:

Δ (M a ) = Δ(V a ) − p e⋅ Δ(e ad ) − e ad( −1) ⋅ Δ( p e )

(15)

The Evolution of Profitability Norms and Investment From a post-Keynesian point of view, firms investment can be constrained by the level of economic activity, represented here by the rate of capacity utilization, u. We use here Kaldor and Kalecki causality between investment and saving. An increase in the investment leads to an increase in profits and so in saving. Investment is also determined by the comparison between anticipated returns on investment, represented by the profit rate profitability norms,

FT , and financial K

ρ . With uncertainty, anticipated returns are set by past returns (Kalecki,

1966; Keynes, 1936). Indeed, firms invest only if anticipated returns are equal to a certain norm. With the development of financial activities, this norm is now set in financial markets and no longer in bonds market, as suggested by Keynes (1936). Boyer (2000) thus introduces financial profitability norms into the investment function. This exogenous variable represents an indicator of shareholders’ growing influence. Increasing financial profitability norms create rising investment selectivity. We use this assumption in our model with a comparison between the anticipated returns on investment and the financial profitability norms. This norm is closed to the Return on Equity and the 15% set by agreement. The investment function can thus be written as:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

g ( +1) = With

I (+1) K

= i0 + a ⋅ (

FT − ρ) + b ⋅u K

(16)

δ the depreciation rate and μ the productivity:

K t = K ( −1) (1 − δ ) + I u=

Y μ⋅K

(17)

(18)

In this model, firms can finance their investment by internal saving, by equity issues, and by contracting loans. Self-financing depends on total profits, FT = Y − W , less distributed

Célia Firmin

136

dividends, D f , and interest on loans, rl ⋅ L(−1) , set by the previous loan stock and the s

d

interest rate set by banks:

S f = FT − D sf − rl ⋅ L(−1)

(19)

Self-financing is thus a residue, obtained after dividends and interest payments. The status given to self-financing is a consequence of financial actors’ dominant position.3 Distributed dividends depend on considering firms’ policy, in the face of shareholder demand constraints, represented by χ , and on net profits on interest:

D sf = χ (FT − rl ⋅ L(−1) )

(20)

This equation allows simulations to be conducted for the impact of the growing trend of distributed dividends, shown in Section 2. The second source of funds is the equity issue which takes care of a share x, that is exogenous, of investment not financed by internal saving. This behaviour draws on Godley and Lavoie (2001-2) and Kaldor (1966). Given uncertainty, firms make their decision according to the financial situation of the last period, with x as an exogenous variable:

( )

Δ e sf ⋅ p e = x ⋅ (I − S f (−1) )

(21)

Finally, firms finance the last part of investment by loans:

ΔL f = (1 − x) ⋅ ( I − S f (−1) )

(22)

Endogenous Money and Balanced Accounts Banks grant loans and collect the deposits from shareholder households. So:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ΔM s = ΔM ad

(23)

Banks are assumed to grant all loans requested by firms, so there is no credit rationing:

ΔLs = ΔLdf

(24)

Interest rates on loans are considered as exogenous and take the same value as interest on deposits ( rm = rl ). Banks thus do not make profits, and loans are equal to deposits. This

3

Indeed, it is possible to establish another rule, by determining the share held for self-financing and in introducing distributed dividends as remainder.

Wage and Employment in a Finance-Led Economy

137

equality allows the accounting structure of the model to be verified.4 In this model, money supply is thus endogenous and follows the credit demand of firms. The demand of money is also endogenous and follows households’ liquidity preference.

The Determination of Wages and Employment in a Finance-Led Economy In a finance-led economy, the “corporate governance” gives a first influence to shareholders in firms’ management decisions. In this way, financial profitability norms become an important factor of the income distribution. Shareholders’ influence thus has to be taken into account in the distributive conflict. This conflict depends on the unemployment rate and on the shareholder negotiation power. The kaleckian mark-up rate principle is useful to integrate the distributive conflict as income distribution determinant. Nevertheless, we do not consider here the degree of monopoly on goods market as determinant of the mark-up rate, but the distributive conflict between social classes, as Goodwin (1967). So, in this model, the wage-share is endogenous and depends on the unemployment rate, ur . Employment, and more precisely unemployment, plays a central role in the income distribution dynamics. This rate represents an indicator of workers’ negotiating power. An increase in the unemployment rate leads to a decrease in the wage-share, because of the weaker negotiating power of wage-earners. Nevertheless, in a finance-led economy, shareholders exert an influence on income distribution. In this model, shareholders’ influence is determined by financial profitability norms, ρ . These norms are positively related to firms’ mark-up rates. As this mark-up rate rises, the profit-share increases and so do distributed profits. Besides, the larger the gap between the rate of return on assets, re and financial profitability norms, the more the mark-up rates tends to increase. Firms are thus in an intermediate situation, arbitrating between wages-earners’ and shareholders’ interests. This position is similar to Stockhammer’s (2004), according to which firms’ preferences depend on the economy’s institutional situation, because managers are in an ambiguous class position. Their income is made up of wages and profits. Financialization corresponds in a large part to the convergence of managers’ preferences with those of shareholders, i.e. the search for an increase in profits. The mark-up rate concept is thus used here to indicate distributive conflict. With p as a parameter establishing managers’ sensitivity to shareholders’ and wage-earners’ influences, it is possible to write:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

π ( +1) = p ⋅ ( ρ − re ) ⋅ ur + π

(25)

The multiplicative form permits taking shareholder pressure for higher a mark-up rate and wage-earner pressure for a lower one into account. For example, if the unemployment rate is low, shareholders will not obtain a rise in the mark-up rate, even if required financial norms are high, due to trade union power. In contrast, a low unemployment rate will not necessarily be conducive to an important cut in the mark-up rate, if required financial norms remain high.

4

The missing equation is thus L error in the model.

=M

. This equation must always be verified to check that there is no accounting

Célia Firmin

138

When the unemployment rate or the norms of financial profitability rise, firms increase their mark-up rate and thus reduce the wage-share. Wages do indeed depend negatively on the mark-up rate:5

W =

Y 1+ π

(26)

Unemployment, UN , is determined by the comparison between the employment level, N , and employment corresponding to full capacity utilization:

UN N fe

(27)

UN = N fe − N

(28)

ur =

Employment depends on the production level and on productivity, assumed to be constant:

N=

Y

(29)

μ

Full employment depends on full capacity production, Y fc , and productivity:

N fe =

Y fc

(30)

μ

Full capacity production depends on the previous period’s capital stock and on full capacity capital ratio, ϑ , assumed to constant:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Y fc = K ( −1) ⋅ ϑ

(31)

Changes to firms’ financial structure thus induce the development of new modalities for distributing income which affect consumption and investment behavior simultaneously. This leads to change in the evolution of growth determinants and so in turn to changes in the income distribution, in particular via unemployment and rate of return on equities. The income distribution between profits and wages constitutes the first stage of the distributive scheme. The gross profits are used to pay the interests on loans. The net profits are thus getting and are used to distribute dividends to shareholders. Finally, the last operation of the

5

And the profit-share is thus positively linked to the mark-up rate:

FT 1 = Y 1+ π

Wage and Employment in a Finance-Led Economy

139

distributive scheme is the firms saving determination. Firms’ cash flow is thus a residue set at the end of the period (figure 1). The unemployment rate is thus a central variable to explain the income distribution dynamics, even in a finance-led economy. More, we will see with numerical simulations that income distribution evolutions explain the growth dynamics and then the unemployment rate. Financial profitability norms

Income distribution between wages and profits due to mark-up rate

Wages

Gross profits

Interests on loans

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Net profits

Dividends

Firms saving

Figure 1. The income distribution chronology.

Célia Firmin

140

Solving the Model with Numerical Simulations To carry out simulations, values are assigned to parameters according to the model’s main stylized facts. Then, the model is solved numerically by an approximation process, allowing a steady state to be found. From this state, simulations are executed by introducing exogenous shocks on parameters or exogenous variables, with one change at a time. For each simulation, the equality between the stock of loans and the stock of deposits is checked. We know with this methodology that there is no accounting error if this identity is gets in each simulation. This methodology is developed in particular by Godley and Lavoie (2001-2; 2007). There is no specific, long term steady state i.e. the situation reached after the shock is different from the initial state. The long term is thus a succession of short terms. The steady state concept is used as an analytical one. Such a theoretical construct is in fact never reached in practice, due to constant changes in exogenous parameters and variables. For this reason, while simulations are carried out, it is important to make a distinction between the initial effects linked to the shock (in the first periods of the shock’s dynamics) and the final effects (at steady state). The graphs represent the results after the shock divided by the base steady state. Financialization induces a change in firms’ financial modalities. They have especially recourse to self-financing, but during the 1980s they also increased their issue of equity. This investment financing by equities corresponds to a rise of x in the model, the share of investment not financed by internal saving which is financed by equity issue. A rise in distributed dividends and in required financial profitability norms, which represent some important features of financialization, will also be analyzed.

An Increase in the Investment Share Financed by Equity Issues As we have said, financialization induces a change in firms’ financing modalities, with an increase in the issue of equities. In our model, an increase in the investment share financed through equities generates a slowdown in the accumulation rate, in the utilization rate, in the profit rate and so in the growth rate (graph 1).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Graphs 1. Impact of a rise in the issue of equities on income distribution, the growth, accumulation, utilization and profit rates 1,0002

1,005

1

1

0,9998

0,995

0,9996

0,99

0,9994

0,985

0,9992

0,98 1

0,999 1

4

7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61

4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 Mark-up rate

W age share

Growth rate

Accumulation rate

Utilization rate

Profit rate

Wage and Employment in a Finance-Led Economy

141

Two factors explain the growth slowdown. First, the increase in the equity stock tends to decrease their rate of return (graphs 2). Dividends distributed per share diminish as equity prices rise, because of the increase in supply. So, the difference between required financial profitability norms and the real rate of return on equities increases and leads firms to raise their mark-up rate, with the aim of increasing dividends. Graphs 2. Impact on the equities rate of return and the unemployment rate 1,05 1,04 1,03 1,02 1,01 1 0,99 0,98 0,97 0,96 0,95 1

4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 Equities rate of return

Unemployment rate

Distributed dividends by equity 1,02 1 0,98 0,96 0,94 0,92 0,9 0,88

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1

8

15

22

29

36

43

50

57

64

71

78

85

92

99

The increase in the mark-up rate induces a fall in the wage-share. It creates a decline in the consumption and hence in the utilization rate. The second effect is the fall in the shareholders’ consumption, which is also affected by the shock. The fall of equity prices reduces capital gains and so shareholders’ consumption. Moreover, the decrease in the profit rate leads to a decline in dividends distributed by firms. We find here the same results as Godley and Lavoie (2001-2) and Skott and Ryoo (2007). In fact, the fall in the utilization rate creates a negative impact on investment (Equation 16). The slowdown of economic activity reduces the amount of profit and also the profit rate. This recalls Kalecki’s (1965) and Kaleckian authors’ conclusions (Godley and Lavoie 200102 and Stockhammer 2005, among others). A rise in the profit-share may be followed by a decrease in the profit rate, because of its recessionary effects on economic activity. The fall of the profit rate reduces investment, which in turn diminishes the profit rate. The model here indicates the double relationship between profits and investment, found by Robinson (1962).

Célia Firmin

142

For firms, the only positive effect of an increase in investment financed by equities is the decrease of their debt. Interest payments fall, but not sufficiently to compensate for the decline in profits, and so to maintain a stable level of distributed dividends. In the long run, the economy converges on a growth path lower than that of the first steady state. The growth slowdown raises the unemployment rate, which accentuates the mark-up increase. The investment decline reduces the equity issued by firms and in doing so stabilizes the rate of return on equities. A change in the financing structure of investment generates an impact on the level of investment by these effects, on the rate of return and then on the distribution of income. Managers thus adopt shareholders’ interests in a context of financialization, by increasing the mark-up rate to preserve the rate of return on equities, thus penalizing wage-earners. A slowdown of economic activity and employment results, via a distribution mechanism. This in turn induces a decrease in profits, and hence in the rate of return on equities.

An Increase in the Rate of Return on Distributed Dividends The increase in distributed dividends is one of the main features of financialization. A rise in the profit-share distributed to shareholders, χ creates a positive effect on growth. Indeed, it improves the rate of return on equities, and thus reduces the gap between required financial profitability norms and real returns on equities (graph 3), which is one of the determinants of the mark-up rate. As the shareholders’ situation improves, firms can thus give back a share of value-added to wage earners. Graph 3: Impact of a rise in distributed dividends rate on markup rate determinants 1,2 1 0,8 0,6 0,4

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0,2 0 1

4

7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 Equities rate of return

Unemployment rate

The increase in distributed dividends is thus followed by a decrease in the mark-up rate, and so the wage-share goes up (graph 4).

Wage and Employment in a Finance-Led Economy

143

Graph 4. Impact of a rise in distributed dividends rate on the mark-up rate and the wage share 1,001 1 0,999 0,998 0,997 0,996 0,995 0,994 0,993 0,992 1

8

15

22

29

36

43

50

Mark-up rate

57

64

71

78

85

92

99

Wage share

This induces a rise in wage earners’ consumption, and thus in the productive capacity utilization rate, which leads firms to invest. The upward trend in the utilization rate is also accentuated by the consumption of profits by shareholders. So, there is a second positive effect of a rise in distributed dividends. Profit consumption by shareholders tends to increase the profit rate and, in return, the accumulation rate (graph 5). The recovery of growth decreases the unemployment rate, which is the second determinant of the mark-up rate, and so reinforces the downward trend of the mark-up rate. The decrease in the saving rate is thus the key factor of the recovery. Graph 5. Impact of a rise in distributed dividends rate on the growth, accumulation, utilization and profit rates 1,15 1,1 1,05 1 0,95

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0,9 1

4

7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 Growth rate

Accumulation rate

Utilization rate

Profit rate

These results confirm those of Cordonnier (2006) and Van Treeck (2007) about the positive consequences of dividend consumption on profits and growth. The consumption of profits constitutes a profit source for firms and may in a certain way compensate for the decrease in investment which follows financialization, as is shown in the next simulation.

Célia Firmin

144

An Increase in Financial Profitability Norms The growing influence of shareholders is the main characteristic of financialization and is represented in the model by an increase in ρ . In our perspective, a finance-led economy represents a regime where shareholder influence is a key variable in firms’ management decisions. Whatever the parameters, an increase in the norms of required financial profitability creates a slowdown in growth, as well as in accumulation and the rate of profit (graph 6). The first effect is directly negative on the investment function. The increase in shareholder influence induces growing investment selectivity, as in Boyer (2000). The slowdown in the accumulation rate induces a decline in the growth rate, and an increase in the unemployment rate. Two factors tend to augment the mark-up rate and so to decrease the wage share (graph 7). First, the rise in financial profitability norms creates a direct negative impact on the wage share. Such an increase modifies the distributive conflict in favour of shareholders. Graph 6. Impact of a rise in financial profitability norms on growth, accumulation, utilization and profit rates 1,05

1

0,95

0,9

0,85

0,8

0,75 1

4

7

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Growth rate

10 13 16 19 22 25 28 31 34 37 40 43 46 49 Utilization rate

Accumulation rate

Profit rate

The slowdown of investment and growth diminishes profits, self-financing and also distributed dividends. It results in an increase in firms’ debt and in a decline in the rate of return on equities and so in an increase in the mark-up rate (graph 7 and 8). The decrease in distributed dividends is reinforced by the increase in interests paid by firms. The profits slowdown diminishes the firms’ financing capacities and by this way increases their debt. Besides, the demand for equities decreases with the slowdown of the equities rate of return. It creates a decrease in capital gains and by this way in the equities rate of return.

Wage and Employment in a Finance-Led Economy

145

Graph 7. Impact of a rise in financial profitability norms on income distribution and equities rate of return 1,01

1,005

1

0,995

0,99 1

8

15

22

29

36

43

50

57

Mark-up rate

64

71

78

85

92

99

Wage share

1,4 1,2 1 0,8 0,6 0,4 0,2 0 1

5

9

13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 Equities rate of return

Unemployment rate

More, the growth slowdown induces a rise in the unemployment rate and so a decrease in workers influence in the distributive conflict. It results in an upward trend in the mark-up rate. Graph 8. Impact of a rise in financial profitability norms on firms’ debt ratio Debt rate 1,04

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1,03 1,02 1,01 1 0,99 0,98 1

8

15

22

29

36

43

50

57

64

71

78

85

92

99

The decline of shareholders’ consumption with the decrease in distributed dividends and in capital gains reduces the utilization rate but also firms’ profits. Indeed, we have seen that profits consumption is a factor of the profit rate (Cordonnier, 2006; Van Treeck, 2007). The decline in the profit rate explains at its turn the slowdown of the accumulation rate. We find

146

Célia Firmin

here the double relationship between investment and profits demonstrated by Robinson (1962). An increase in the norms of required financial profitability thus does not lead to a rise in the return on equities, but on the contrary to a decrease of this rate. These norms influence the productive system at the same time, by the decrease in investment and in the wage share, due to upward pressures on the mark-up rate. If the increase of these norms is too high, the model becomes unstable. As in Boyer (2000), there is a limit to shareholder influence which, if it is not respected, leads the economy towards instability. Three consequences of financialization have been identified here with this model and the presentation of stylised facts: i) the increase in investment financed by equities, ii) the increase in distributed dividends, and iii) the reinforcement of shareholders’ influence. The simulations demonstrate that the increase in investment financed by equities and in shareholders’ influence induce a slowdown in economic growth and employment, and a decrease in the wage-share. Nevertheless, the rise in distributed dividends creates opposing effects on income distribution and growth and employment. These contradictory effects on income distribution, growth and employment, raise the question of the stability of this growth regime.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

CONCLUSION Post-Keynesian models on managerial economies provide a coherent analytical framework to examine financial variables. These models take into account the links between monetary, financial and real variables. The models developed by Godley and Lavoie (2001, 2007) provide a coherent stock-flow framework which is especially suited to analyzing different aspects of financialization. In this “stock-flow” perspective, Kaleckian analysis provides a coherent explanation of the financialization, in which a decrease in the wage-share and a slowdown in growth occur simultaneously. This kind of model explains factors changing the income distribution and the impact of these changes on macroeconomic dynamics. The use of numerical simulations allows these different aspects of financialization to be taken into account, which is relevant to “traditional” Kaleckian analysis. Even in a financialization process, with the model presented here, the economic dynamics remains dependant on the wage bill. With financialization, wages and employment seem to become adjustment variables used by firms to reach financial profitability norms. The increase in shareholders’ influence decreases the wage share and the accumulation rate. It results in a growth slowdown and by this way in a raise in the unemployment rate. This last reinforces the downward trend of the wage share. This conclusion supports Boyer (2000), for whom financialization occurs in an economy which is always dominated by the wage relationship, and in which wages are the main factor of consumption, so that the increase in financialization induces negative effects simultaneously for wage-earners and the economy as a whole.

Wage and Employment in a Finance-Led Economy

147

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES AGLIETTA, M. and REBÉRIOUX, A. 2005. Corporate governance adrift: a critique of shareholder value, Edward Elgar, Cheltenham BHADURI, A. and MARGLIN, S. 1990. “Unemployment and the real wage: the economic basis for contesting political ideologies”, Cambridge journal of Economics, vol.14, n°4, pp.375-393, décembre BOYER, R. 2000. “Is a finance-led growth regime a viable alternative to Fordism? A preliminary analysis”, Economy and Society, n°1, Volume 29, p.111-145 CLÉVENOT, M. and MAZIER, J. 2005. “Investment and rate of profit in a financial context: the French case”, Working Paper, n°06-2005, December, CEPN, http://www.univparis13.fr/CEPN/wp2005_06.pdf COMMENDATORE, P. 2003. “On the Post Keynesian Theory of growth and institutional distribution”, Review of political economy, vol 15, n°2, p.193-209 CORDONNIER, L. 2006. « Le profit sans l’accumulation : la recette du capitalisme gouverné par la finance », Innovation : Cahiers d’économie de l’innovation, vol.23, n°1, p.51-72 DORE, R. 2000. Stock market capitalism: welfare capitalism, Oxford University Press DORE, R. 2002. “Stock market capitalism vs. welfare capitalism”, New Political Economy, vol.7, n°1, p.115-127 DOS SANTOS, C.H. and ZEZZA, G. 2004. “A Post-Keynesian consistent macroeconomic growth model: preliminary results”, The Levy Economics Institute of Bard College, Working paper n°402, New York, février DUMÉNIL, G. and LÉVY, D. 2001. “Costs and benefits of neoliberalism. A class analysis”, Review of international political economy, vol.8, n°4, p.578-607 DUMÉNIL, G. and LÉVY, D. 2005. “From prosperity to neoliberalism. Europe before and after Working paper, the structural crisis of the 1970s”, http://www.jourdan.ens.fr/~levy/dle2002d.htm GODLEY, W. and LAVOIE, M. 2001-02. “Kaleckian models of growth in a coherent stock-flow monetary framework: a Kaldorian view”, Journal of Post Keynesian Economics, Vol.24, n°2, p.101-135 GODLEY, W. and LAVOIE, M. 2007. Monetary economics. An integrated approach to Credit, Money, Income, Production and Wealth, MacMillan, London KALDOR, N. 1955-56. “Alternative theories of distribution”, The review of economic studies, vol.23, n°2, p.83-100 KALDOR, N. 1966. “Marginal productivity and the macro-economic theories of growth and distribution”, Review of Economic Studies, Octobre, n°33, p.309-319 KALECKI, M. 1965. Theory of economic dynamics: an essay on cyclical and long-run changes in capitalist economy, Augustus M.Kelley, New York KALECKI, M. 1990. Collected works of Michal Kalecki, volume 1, J.Osiatynski (ed.), Clarendon Press, Oxford KEYNES, J.M. 1936. The general theory of employment, interest and money, Macmillan, London KEYNES, J.M. 1937. “Alternative theories of the rate of interest”, The Economic Journal, vol.47, n°186, juin, pp.241-252

Célia Firmin

148

LAVOIE, M. 1992. Foundations of Post Keynesian-Economic Analysis, Edward Elgar, Cheltenham LAZONICK, W. and SULLIVAN, M.O’. 2000. “Maximizing shareholder value: a new ideology for corporate governance”, Economy and Society, n°1, volume 29, p.13-35 PASINETTI, L. 1962. “Rate of profit and income distribution in relation to the rate of economic growth”, The review of economic studies, vol. 29, n°4, p.267-279 PLIHON, D. (dir.) 2002. Rentabilité et risque dans le nouveau régime de croissance, Commissariat général du plan, La Documentation française, Paris POTERBA, J.M. 2000. « Stock market wealth and consumption », Journal of Economics Perspectives, vol.14, n°2, p.99-118 ROBINSON, J. 1962. Essays in the Theory of Economic Growth, Macmillan, London SKOTT, P. and RYOO, S. 2007. “Macroeconomic implications of financialization”, communication to 11th Research Network Macroeconomic Policies: Finance-led capitalism? Macroeconomic Effects of Changes in the Financial Sector, Berlin, 26-27 October STOCKHAMMER, E. 1999. “Robinsonian and kaleckien growth. An update on post-Keynesian growth theories”, Working paper, n°67, October, Vienna University STOCKHAMMER, E. 2004. “Financialization and the slowdown of accumulation”, Cambridge journal of economics, vol. 28, n°5, p.719-741 STOCKHAMMER, E. 2005-6. “Shareholder value orientation and the investment-profit puzzle”, Journal of Post Keynesian economics, vol 28, n°2, p.193-215, hiver TAYLOR, L. 1991. Income distribution, inflation and growth, MIT Press, Cambridge TAYLOR, L. 2004. Reconstructing macroeconomics: Structuralist Proposals and Critiques of the Mainstream, Harvard University Press VAN TREECK, T. 2007. “Reconsidering the Investment-Profit Nexus in Finance-Led Economies: an ARDL-Based Approach”, IMK Working Paper, Hans Böckler Stiftung, n°01/2007, http://www.boeckler.de/pdf/p_imk_wp_01_2007.pdf

ANNEXES Balance Sheets

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Workers

Firms

Monetary Deposits

-M

Equities

-e

Capital Loans

+K

∑ (net wealth)

Banks

s f d f

0

s b

⋅ pe

Shareholders +

M

+

e ad ⋅ p e

+L

Vf

0

∑ 0 0 +K 0

s b

-L

d a

+ Va

+K

Wage and Employment in a Finance-Led Economy

149

Transactions Matrix

Consumption

Workers

Firms Current

− Cw

+ Cs

+W

+Is −W -Sf

Investment Wages Nondistributed profit Interest on loans Interest on deposits

Banks Capital

0

0

0

+ rl ⋅ Ls(−1)



+ rm

− ΔLs s + ΔM

e sf ⋅ pe

− D sf 0

− Ca

+Sf

+ ΔLd

0



0

rm ⋅ M (−s 1)

Dividends

Shareholders

0

-

∆ Monetary deposits Equities

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Banks Current

−Id

− rl ⋅ Ld(−1)

∆ Loans



Firms Capital

0

0

0

⋅ M ad(−1)

0

0 - ΔM a

0



0

d

ead ⋅ pe

+ Dad

0

0

0

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S.Mercer

ISBN:978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 6

EMPLOYMENT, OCCUPATIONAL DISPARITIES, AND WAGES IN INDIA: A DECOMPOSITION ANALYSIS Rajarshi Majumder11 and Dipa Mukherjee22 1

Reader, Department of Economics, University of Burdwan, Burdwan, West Bengal , India 2 Senior Lecturer, Department of Economics, Narasinha Dutt College, Howrah, West Bengal, India

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Disparities in livelihood among socio-economic groups are a major problem in developing countries, and India is no exception. Such differentials are caused mainly by disparities among various groups in terms of job availability, type and sector of employment, and earnings there from. However, these differences transcend the boundaries of current generation and through their impact on capability formation and asset creation tends to perpetuate, and even accentuate, disparities in living standards of future generations, endangering social sustainability altogether. Policy framing for inclusive development as envisaged by the Millennium Development Goals would therefore require a thorough study of employment and earning differentials and causes thereof. The present paper explores the issue of Employment and Wage Differentials in the Indian labour market over the last decade across social classes, regions, gender, and job types. Using both parametric and non-parametric techniques, differences in employment rates, its nature, occupational distribution, wage rates, and total earnings have been explored. The roles played by Discrimination in labour market, both during entry & during wage setting, and that by Endowment in explaining the occupational & earning disparities have been examined through modern Decomposition techniques. The dynamics of these issues have also been explored against the prevalent theoretical 1

Corresponding Author. Reader, Department of Economics, University of Burdwan, Burdwan, West Bengal – 713104, INDIA. Tel: 91 342 2656566 Ext 438. email: [email protected] 2 Tel: 91 33 26438049. email: [email protected]

152

Rajarshi Majumder and Dipa Mukherjee wisdom that in a globally integrated world economy, vertical disparities may increase while horizontal disparities within groups would come down. The results and inferences drawn from this paper are expected to become a comprehensive study of Indian labour market in recent times and have serious policy implications for removal of poverty & inequality and achievement of Millennium Development Goals.

Keywords: employment, wages, gender differences, social xxclusion, occupational choice, regional disparity, job function, wage decomposition, discrimination; endowment gap, parametric methods, censored regression, qualitative variable, India JEL Codes: E24; J21; J23; J24; J71; J31; R23; C14; C24; C25.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

INTRODUCTION Inequality in Wages and Earnings are responsible for much of the disparity that exists in assets, consumption, health care, educational attainment, and other accepted indicators of well-being, especially in developing countries. This causation also tends to perpetuate, or even accentuate, inequalities through their impact on human capital formation. As a result, intervention to reduce wage differentials across various sections of the society becomes a major policy objective for developing countries like India. To do so, one must carefully explore the levels, trends, and probable causes of wage and earning differentials in the country. In economic terms, wage differentials in labour markets may operate along a number of dimensions – gender, caste, region, sector, or regularity of job – which effectively reduces the opportunity for some groups to gain access to various socio-economic services, prevents their capability formation, and therefore limits their participation in the labour market. This tends to perpetuate the ‘exclusion’ of the disadvantaged groups – most frequently the rural, female, scheduled caste & tribe, casual workers being the affected ones. Since wage differential would certainly exist across different occupations because of the differing levels of skill endowment required for such jobs, differences among two groups in Occupational Distribution of workers would lead to dissimilarity in mean wages earned by the groups. Such differences in occupational pattern may be because of variations in Endowment or skill patterns of the two groups of workers, and also because of Entry Barrier in Job Market where some groups are discriminated against even if they have adequate skill. Apart from such hierarchical structure, evidences of spatial or inter-personal wage differences even within occupations are also quite substantial in India. A part of such wage differential may be again because of endowment gaps – differences in educational attainment and training – among the different competing groups. On the other hand, pure discrimination in terms of lower wage rates paid to certain groups is also common. In this paper we examine inequalities in occupation and earnings between various subgroups in the Indian wage labour market, and then econometrically decompose those gaps into separate components, explainable through differences in Occupational pattern and Wage Rates, subdivided into a part due to Endowment factors such as Education & Training, and the other representing Discrimination in both employment and wages.

Employment, Occupational Disparities, and Wages in India

153

While wage differential would exist due to skill differences and discrimination, the pattern or magnitude of it and its components are also expected to change with changing economic order. In fact, the last decade of Liberalisation-Privatisation-Globalisation (LPG) in India have initiated new dynamics in the labour market. On one hand, it is expected that with increased investment, trade, & output, more and better employment opportunities would emerge and labour mobility would increase, leading to narrowing down of horizontal wage differences among workers with similar skills and in same type of jobs. On the other, as demand for new skills and occupations increase, vertical wage differences are expected to rise. The overall impact would depend on how broad-based the labour market in the economy is, whether movement across skill-barriers is relatively easy, and whether institutions for retraining and re-deployment are in place. It has however been generally argued that LPG in India has lead to deterioration of the position of the workers in the labour market both in terms of job-availability and wages offered. In this context the present paper also attempts to explore the post-reform trends in wage differentials and the relative importance of the Endowment and Discrimination Effects.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

WAGE INEQUALITY – A BRIEF REVIEW Rising wage inequality among workers in developed countries, especially in USA and UK, have been well documented. Studies on developing countries, particularly Latin American countries since the 1980s have also been in the public domain [see Wood (1997), and Katz and Autor (1999), for a review of those studies]. While studies on wage differentials in India, particularly the post-reform dynamics of it, have been sparse, handful of those that exist (Kingdon, 1998; Kingdon & Unni, 2001; Duraiswamy, 2002; Galbraith, 2004; Dutta, 2005; Madheswaran and Attewell, 2007) either focus on specific industries, or specific types of workers, or ignore the possibility of segmented labour markets & hence segmented wage functions for different groups. The present paper therefore adds value to existing literature on four counts. First, it takes into consideration all wage workers in the economy – casual & regular, male & female, and in all non-farm occupations and regions.1 Second, it uses segmented wage functions for different groups to estimate the impact of endowments (proxied by educational attainment) on wages. Third, it augments the existing literature by applying modern wage decomposition techniques to India and examining separately the extent of wage differences attributable to Occupational Disparity, Wage Disparity, Skill-gap and Labour Market Discrimination. Fourth, it indicates what part of the aggregate disparity is due to inter-group differences and what part is due to intra-group differences, as also the postreform trends in their relative importance.

DATABASE AND METHODOLOGY Database We use NSSO data on Employment and Unemployment from the 50th, 55th and 61st Round surveys of NSSO pertaining to the years 1993-94, 1999-2000, and 2004-05. Wages are

Rajarshi Majumder and Dipa Mukherjee

154

converted to real wags at constant 1999-00 prices and Wage Earnings (per Manday) are used for our study. We compare and estimate wage differentials across the following sub-groups: [a] [b] [c] [d]

Sectors – Rural vs. Urban; Regions – Groups of Low and High Per Capita Income States; Gender – Females vs. Males; Social Groups – Backward castes (Scheduled Castes & Tribes) vs. Advanced castes; and, [e] Job Types – Casual vs. Regular workers.

Since the wages and earnings of the Farming class are determined in a completely different manner than the rest of the workers, we leave out the Farmers-Fishermen-Hunters class of workers from our study and concentrate on the Non-farm workers – both rural and urban, male & female, casual & regular, from backward and advanced social classes, and from all the states and Union Territories of India. Farmers' earnings may be explored as a separate issue altogether in a future study.

METHODOLOGY Calculating Earning Disparity Disparities at the aggregate level are studied using CVs and Theil indices in Earnings. Discrepancies among sub-groups are explored using Inter-group ratios and Absolute differences between the mean earnings of the sub-groups. To examine relative importance and trends in intra-group & inter-group earning inequality the decomposition version of Theil Index (TI) has been used. The TI is a member of the General Entropy Index class and is given by: TI = (1/n)Σ (yi/y).ln (yi/y);

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where yi-s are individual values, y is mean value, and n is number of observations. The bounded version of it, called the Relative Theil Index (RTI) is given by: RTI = (TI / ln n), which ranges between 0 and 1. The decomposition version of TI is as follows: TE = Σθk.TEk + Σθk. ln (yk/y); where θk is share of k-th subgroup in total income; TEk is Theil Index for k-th subgroup; and yk is mean income of k-th subgroup, the sum taken over the sub-groups. The first term in the RHS measures total intra-group disparity while the second term measures the total inter-group disparity. The general practice is to calculate TE, the intra-

Employment, Occupational Disparities, and Wages in India

155

group TEs, total intra-group disparity, and then compute total inter-group disparity as the residual. This helps us to explore relative contribution of Within-group and Between-group disparities in Earnings.

Occupation Disparity The disparity between two groups regarding Occupational Distribution is generally measured by the Segregation Index:

D=

1 ∑ pi1 − pi 2 ; 2 i

where pi1 and pi2 are proportion of Group-1 and Group-2 workers respectively in the ith occupation, assuming that the second group is notionally the disadvantaged group. If there are k occupational groups, then D will vary between 0, indicating perfect similarity between the two groups, to (k/2) for perfect mismatch between them. The occupational disparity between the two groups may be caused by – (a) Endowment Gap between the two groups; and (b) Entry Barrier or Job market discrimination wherein group-2 members are prevented entry to certain occupations even with similar endowments leading to their being underrepresented in those jobs.

Occupation Choice Under such situation, the Occupational Choice may be thought to be a Multinomial Logit Model (MLM) where Probability of an individual being in the ith occupation depends on her endowment. In that case, assuming kth occupational group to be the reference group, we have

ln

P(Occ = i ) Pi = = α i + ∑ β ij X j + ui = Z i ; for i = 1, 2, 3, . . k-1; where Xj are P(Occ = k ) Pk j

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Endowments. On estimation of the MLM, one gets the estimates of the coefficients α and β. From these, individual probabilities can be estimated as

Pˆi =

ˆ

e Zi k −1

1+ ∑ e

; for all i = 1, 2, 3, . . k-1; Zˆ i

i

while for the reference group k, we have

(1)

Rajarshi Majumder and Dipa Mukherjee

156

Pˆi =

1 k −1

1+ ∑ e

.

(1a)

Zˆ i

i

Expected Value or Mean of these probabilities over the population would give us Expected Proportion of the population in each Occupational Group. When calculated sub-group wise, it would give us Expected Proportion of sub-group workers respectively in each of the k occupational groups, i.e.

⎛ Nˆ ⎞ 1 N1 ˆ pˆ i1 = E ⎜⎜ i1 ⎟⎟ = ∑ Pi1 ⎝ N1 ⎠ N1 n =1

(2)

where Pi1 is Estimated Probability of a Group-1 worker being in the ith occupational group; Ni1 is Expected Number of Group-1 worker in the ith occupational group; and N1 is Total Number of Group-1 workers. These can be obtained for each of subgroups and each of the k occupational classes. These estimated proportions are therefore dependent on expected probabilities, which are in turn dependent on estimated parameters of the MLM (group and occupation specific α and β) and the group average endowment levels (Xs). While estimating Pi1 and Pi2 we have used group specific coefficients and endowments. If now, Pi2 is estimated using X2 as usual but the Group-1 coefficients instead of its own we would get Expected Probabilities for Group-2 workers with own endowment but selection function structure of the advantaged group. These would be called Non-Discriminatory Probabilities and given by ˆ

Pi′2 =

eαˆi1 + β i1 X 2 k −1

; for all i = 1, 2, 3, . . k-1;

1+ ∑ e

(3)

αˆ i 1 + βˆi 1 X 2

i

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

while for the reference group k, we have

Pk′2 =

1 k −1

(3a)

1+ ∑ e

αˆ ii 1 + βˆii 1 X 2

i

Since Pi′2 s are Non-Discriminatory Probabilities, Non-Discriminatory Proportions therefore can be obtained by taking their Expectations or Average over the sub-group population, i.e.

⎛ Nˆ ′ ⎞ 1 pˆ i′2 = E ⎜⎜ i 2 ⎟⎟ = ⎝ N2 ⎠ N2

N2

∑ Pˆ ′ . n =1

i2

(4)

Employment, Occupational Disparities, and Wages in India

157

Decomposition of Occupational Disparity Once Pi′2 s are obtained, one can decompose actual occupational disparity between group-1 and 2 into the two components – Endowment Gap, and Entry Discrimination as: Actual Disparity = pi1 − pi 2 = ( pi1 − pi′2 ) + ( pi′2 − pi 2 ) .

(5)

The first term on the RHS is the difference between Actual group-1 proportions and the Non-discriminatory group-2 proportions. The coefficient structures of both of them are identical and the difference between them arises because of endowment gaps between the two groups. Therefore this term can be called Occupational Disparity because of Endowment Gap. The second term is the difference between the Non-discriminatory group-2 proportions and Actual group-2 proportions. The endowment structures of both of them are identical and the difference is due to difference between coefficient structures of the two groups. Therefore this term can be called Occupational Disparity because of Entry Discrimination. Thus we can decompose Actual Occupational Disparity into that due to Endowment and Discrimination. The Segregation Index (SI) can also therefore be calculated as Actual SI, SI due to endowment gap, and SI due to Discrimination.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Wage Function Wage function determination has followed the basic works of Mincer (1973) where Wage is dependent on endowment of individual workers. A log linear model with (log of) Wage per manday as the dependent variable is used: yij = ln (Yij) = θij.Xij ; where Yij is Wage per manday and Xij is Endowment vector of jth group worker engaged in ith occupation, and θij is the coefficient vector, which can also be interpreted as Returns to Endowment levels. Therefore, Yi1 = exp(θi1.Xi1), and Yi2 = exp(θi2.Xi2). These wage functions are determined separately for each subgroup (Males, Females, Rural, Urban, etc.) and each occupation. One major issue of concern is non-participation of a large number of people in the workforce though they have substantial endowment levels leading to the problem of ‘Selection Bias’. The problem has been widely discussed in theoretical and empirical literature and it has been argued that ‘Selection Bias’ is a problem “only when the (non)selection is non-randomly distributed across the sub-groups” [see Leung and Yu (1996), Vella (1998), Puhani (2000), Peracchi (2007), and Adamou (2006) for a detailed discussion on this and the points below]. In these situations, commonly used method is the 2-Stage Heckman Filter method where a Selection Function is first estimated as a Logit (or Probit) model with participation in the wage labour market as the dependent variable and several causal variables as the explanatory ones. This is used to compute the Inverse Mills Ratio (IMR) and use that as an additional explanatory variable in the estimation of the Wage function. However, this requires the causal variables in the Selection function and the Wage function to be significantly different to avoid multicollinearity in the Wage function estimation. It is argued in literature that it is quite difficult to logically obtain such different

Rajarshi Majumder and Dipa Mukherjee

158

variable vectors while using workforce data and if the Non-selection is somewhat random then not much is lost by not using the Heckman Filter. In our study we find that non-workers are quite randomly distributed across sectors, regions, and social classes. Only for gender groups, non-workers are more among females compared to males. However, here too, the differences are not statistically significant. Hence the problem of Selection Bias is negligible and we do not use Heckman Filter.

Decomposition of Gross Earnings Once Wage functions are estimated for each of the subgroups, we have to decompose actual earning differences into its components. Decomposition has been done in literature using Blinder-Oaxaca (Blinder, 1973; Oaxaca, 1973) formulation where a part of the difference is accounted for by Endowment Gap and the other Unexplained part by Discrimination or Remuneration Gap. This approach was later extended to overcome the Index Number Problem – problem of arriving at a Nondiscriminatory Wage Rate for both the advantaged and disadvantaged groups (Greenhalgh, 1980; Cotton, 1988; Neumark, 1988; Oaxaca and Ransom, 1994). Further extension was done to incorporate disparity in occupational distribution into the earning gap (Brown et al, 1980; Banerjee and Knight, 1985). We have combined the Blinder-Oaxaca, Cotton-Neumark, and Brown-Banerjee models to decompose Gross Earning Gaps between groups of workers as explained hereafter [similar methodology have been used by Madheswaran and Attewell (2007), but for urban workers only]. Gross Earning per manday received by a particular group (Y1) is dependent on – [a] What proportion of them are in different occupations; and [b] What wages they earn within those occupations.

In other words, Y1 =

∑Y . p i1

i1

(6)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

i

pi1 being proportions of group-1 workers in ith occupation and Yi1 being wages received by them. To be more explicit, a disadvantaged group have lower earnings because – i.

most of its members are in low paying occupations compared to the other group – Occupational Disparity (A); ii. even with similar occupational pattern, most of their members have low endowments and hence earn lower wages – Endowment Gap in Wages (B); iii. even with similar occupational pattern and endowment levels, they are offered lower wages because of Wage discrimination – Remuneration Gap (C). B and C together is the Total Wage Gap. Gross Earning Gap = Y1 – Y2; can be decomposed as follows:

Employment, Occupational Disparities, and Wages in India Y1 – Y2 =

∑Y .p − ∑Y i1

i1

i2

i

. pi 2

i

= ∑ Yi1 .( pi1 − pi 2 ) + ∑ pi 2 .(Yi1 − Yi 2 ) i

159

i

= A + (B+C)

(7)

The second component, or Wage Gap, can be decomposed further as follows:

∑p

i2

i

~

~ ~ ~ ~ .(Yi1 − Yi 2 ) = ∑ pi 2 .[(Yi1 − Yi1 ) + (Yi1 − Yi 2 ) + (Yi 2 − Yi 2 )]

(8)

i

~

where Yi1 and Yi 2 are Non-discriminatory earnings of the two groups. Yi1 = exp(θi1.Xi1)

(9)

Yi2 = exp(θi2.Xi2)

(10)

Non-discriminatory earnings are obtained by using group-specific endowment vector along with a non-discriminatory wage structure or coefficient vector, i.e.

~ ~ Yi1 = exp(θ i . X i1 )

(11)

~ ~ Yi 2 = exp(θ i . X i 2 )

(12)

The non-discriminatory coefficient vector is obtained as weighted average of group specific coefficients, i.e.

~

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

θi =

N i1 N .θ i1 + i 2 .θ i 2 ; Ni Ni

(13)

where Ni1 and Ni2 are number of group-1 and group-2 workers respectively in the ith occupational group, and Ni = Ni1 + Ni2. Thus the expanded Wage Gap decomposition is of the form:

Rajarshi Majumder and Dipa Mukherjee

160

∑p

i2

i

~ ~ ~ ~ .(Yi1 − Yi 2 ) = ∑ pi 2 .[(Yi1 − Yi1 ) + (Yi1 − Yi 2 ) + (Yi 2 − Yi 2 )] i

~ ~ ~ ~ = ∑ pi 2 .[(Yi1 − Yi 2 ) + (Yi1 − Yi1 ) + (Yi 2 − Yi 2 )] i

~ ~ ~ = ∑ pi 2 .[exp{θ i ( X i1 − X i 2 )} + exp{(θ i1 − θ i ) X i1} + exp{(θ i − θ i 2 ) X i 2 }] i

~ or ∑ pi 2 .(Yi1 − Yi 2 ) = ∑ pi 2 .[exp{θ i ( X i1 − X i 2 )}] i

i

~ ~ + ∑ pi 2 .[exp{(θ i1 − θ i ) X i1}] + ∑ [exp{(θ i − θ i 2 ) X i 2 }] ............(14) i

i

= Endowment Effect in Wage + [Overpayment to group-1 + Underpayment to group-2] = Endowment Effect + Remuneration Effect (Wage Discrimination) =B+C Total Earning Gap therefore becomes: Y1 – Y2 = Occupation Effect + Endowment Effect in Wages + Remuneration Effect = A + (B + C). This gap can be further decomposed and re-arranged as follows: Y1 – Y2

= ∑ Yi1. pi1 − ∑ Yi 2 . pi 2 i

i

= ∑ Yi1.( pi1 − pi 2 ) + ∑ pi 2 .(Yi1 − Yi 2 ) i

i

~ or Y1 − Y2 = ∑ Yi1.( pi1 − pi 2 ) + ∑ pi 2 .[exp{θ i ( X i1 − X i 2 )}] i

i

~ ~ + ∑ pi 2 .[exp{(θ i1 − θ i ) X i1}] + ∑ [exp{(θ i − θ i 2 ) X i 2 }]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

i

i

~ = ∑ Yi1.( pi1 − pˆ i′2 ) + ∑ Yi1.( pˆ i′2 − pi 2 ) + ∑ pi 2 .[exp{θ i ( X i1 − X i 2 )}] i

i

i

~ ~ + ∑ pi 2 .[exp{(θ i1 − θ i ) X i1}] + ∑ [exp{(θ i − θ i 2 ) X i 2 }] i

i

~

or Y1 − Y2 = ∑ Yi1.( pi1 − pˆ i′2 ) + ∑ pi 2 .[exp{θ i ( X i1 − X i 2 )}] + ∑ Yi1.( pˆ i′2 − pi 2 ) i

i

~

~

i

+ ∑ pi 2 .[exp{(θ i1 − θ i ) X i1}] + ∑ [exp{(θ i − θ i 2 ) X i 2 }] i

............. (15)

i

= (A1 + B) + A2 + (C1 + C2); where A1 = Earning Difference due to Occupational Disparity because of Endowment Gap;

(16)

Employment, Occupational Disparities, and Wages in India

161

B = Earning Difference due to Wage Gap because of Endowment gap; A2 = Earning Difference due to Occupational Disparity because of Entry Barrier; C1 = Earning Difference due to Wage Gap because of Overpayment; and, C2 = Earning Difference due to Wage Gap because of Underpayment. Therefore, Gross Earning Difference = E + (A2 + C) =E+D = Gross Endowment Effect + Gross Discrimination Effect. This expanded version of the Decomposition formulation is used to examine the relative contribution of various components in explaining Earning Differentials across Sector, Region, Gender, Social Class, and Job-type in India. The above explains the methodology in detail.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

WAGE DIFFERENTIALS: MAGNITUDES - PATTERNS – TRENDS Before we embark on the issue of wage differentiation, let us take a brief look at the levels and trends in wages. Trends in wages can be explored using both Wage per Manday or Wage Rate, and Wages per Worker per Week or Average Weekly Wage. While the former is purely a measure of wage rates, the latter reflects availability of job per week also. Here we concentrate on Wage Rates or Wage Earnings per manday only. Also, we have left out the Farmers-Fishermen-Hunter Occupation Group and are analysing only the non-farm workers. Wage rates, or average wage per man-day, which was 88 rupees in 1993, increased to Rs. 117 in 1999, and to Rs. 118 in 2004 (Table 1). While Wage Rate increased at 7.2 percent pa during 1993-99, it increased by only 0.2 percent pa during 1999-2004 period. At the Aggregate level Wage disparity is substantially high as indicated by the Coefficient of Variation (CV) and Theil indices (Table 2). Wage Disparity across Sub-groups It is quite natural that different occupations would have different wage levels depending on skill requirement and actual (labour) demands. It is observed that the hierarchy has remained stable over time with the Administrative, Professionals & Technical workers (Grade-1 workers) at the top of the wage ladder, Clerical, Sales, and Service occupations (Grade-2 workers) at the middle, and the Production, Transport and other unclassified labourers (Grade-3 workers) at the bottom. Disparity across occupations had increased substantially in the first quinquenna after 1993, but has dropped thereafter, coming back to the initial level (Table 2).

Rajarshi Majumder and Dipa Mukherjee

162

Table 1. Average Wages - 1993-2004 - in constant 1999-00 prices Wage per Man-day (Rs) 1993 1999 2004 119 178 249 140 196 193 239 362 438 125 174 175 58 76 75 62 90 78 61 85 77 80 104 100 43 55 51

Groups Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

68 103

91 138

89 140

LIG States HIG States

84 91

106 128

107 138

Female Male

59 93

89 122

83 126

Backward Advanced

64 95

90 124

84 131

44 112 1993 53

57 151 1999 81

54 150 2004 84

Rural Urban

Casual Regular Aggregate

Source: Authors’ calculations based on NSSO (1995), NSSO (2001), and NSSO (2006).

Table 2. Key Disparity Measures – 1993-2004 Indicators Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

CV - across Occupations (%)

Wage per Week 1993 1999 2004 60.2 64.5 61.6

Wage per Man-day 1993 1999 2004 59.3 63.8 59.1

Urban-Rural Ratio HIG-LIG Ratio

2.6 1.2

2.6 1.3

2.7 1.5

1.5 1.1

1.5 1.2

1.6 1.3

Male-Female Ratio Advanced-Backward Ratio Regular-Casual Ratio

2.1 1.5 3.7

2.0 1.4 3.9

2.1 1.6 4.2

1.6 1.5 2.5

1.4 1.4 2.6

1.5 1.6 2.8

151.3 0.47

281.2 0.49

272.3 0.52

116.2 0.29

125.3 0.39

125.8 0.39

Overall CV (%) Overall Theil Index Source: Same as Table 1.

Employment, Occupational Disparities, and Wages in India

163

Spatial disparity has been analysed from two standpoints – that between Rural and Urban areas, and that between Low Income Group (LIG) and High Income Group (HIG) states. Rural Urban disparities in wage rates have increased throughout the period of study. Within the sectors, intra-group disparities in wage rates are on the rise in both rural and urban areas (Table 2 & 3). If we consider the LIG and HIG states, it is natural to expect that wages in the latter would be higher than that in the former. However, what is disconcerting is that the disparity between them is increasing in the post-reform period, especially in the second quinquenna. It is thus quite evident that benefits of the recent economic boom in India have accrued mainly to the HIG states, which is not surprising given the concentration of FDI and Domestic Private Investment in selected states. Within group disparity also have increased consistently in both the groups of states (Table 2 & 3). What is evident therefore is that spatial disparity in wages and earnings have increased in the last decade both across rural-urban setting and across the lagging and advanced states. Table 3. Intra-Group and Inter-Group Disparity in Wages per Man-day 1993 0.24 0.29 1.5

1999 0.36 0.38 1.5

2004 0.33 0.40 1.6

Theil Index within LIG States Theil Index within HIG States HIG-LIG Ratio

0.29 0.29 1.1

0.37 0.39 1.2

0.38 0.39 1.3

Theil Index within Female Theil Index within Male Male-Female Ratio

0.40 0.27 1.6

0.56 0.36 1.4

0.49 0.37 1.5

Theil Index within Backward Theil Index within Advanced Advanced-Backward Ratio

0.24 0.29 1.5

0.34 0.39 1.4

0.32 0.39 1.6

Theil Index within Regular Theil Index within Casual Regular-Casual Ratio

0.10 0.25 2.5

0.22 0.35 2.6

0.17 0.39 2.8

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Groups Theil Index within Rural Theil Index within Urban Urban-Rural Ratio

Source: Same as Table 1.

Gender wage disparity declined during the first quinquenna of our study period indicating convergence in male and female wages. Thereafter, differences increased but have so far remained lower than the initial period. Intra-group disparities have increased continuously during this period for the males, while for the females intra-group disparity first increased and then decreased (Table 2 & 3). Wage inequality among the Backward classes and the Advanced classes have shown signs of first coming down but has increased substantially thereafter, indicating divergence among social classes in terms of wage earnings in the labour market. Disparities within the classes have also increased significantly over the decade (Table 2 & 3).

Rajarshi Majumder and Dipa Mukherjee

164

While it is quite obvious that earnings of Regular workers will be higher than that of Casual workers, what is alarming is that the Regular workers earn a daily wage rate of about three times that of the Casual workers. In addition, this disparity is further widening during our study period. Within the groups, disparities among both casual and regular workers are rising, though in case of regular workers there are signs of convergence during 1999-04 period (Table 2 & 3).

Disparity Within and Between Groups – Decomposition Version of Theil Index So far we have discussed the disparities in Wage Rates both between the different subgroups and within them. Using the decomposable version of Theil index one can also examine the relative magnitudes of Within groups and Between groups differences. It is observed that disparities within all the subgroups have increased in the last decade, the increase being most striking in the first quinquenna of the study period (Table 4). For Occupational Groups, Between Group disparity accounts for one-fourth to one-third of the aggregate wage differences. For all the others, most of the wage inequality is due to Within Group differences, though the relative contribution of Between Group differences are increasing. Only for differences across Job Types, the relative contribution of Between Group differences is decreasing. Table 4. Decomposition of Disparity – Theil Index – Wages per Man-day Groups

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Occupational Groups

Total Within Group Disparity 1993 1999 2004 0.18 0.29 0.27 (62.6) (74.3) (69.5)

Between Group Disparity 1993 1999 2004 0.11 0.10 0.12 (37.4) (25.7) (30.5)

Urban-Rural

0.27 (95.5)

0.37 (95.2)

0.38 (97.1)

0.02 (4.5)

0.02 (4.8)

0.01 (2.9)

HIG-LIG

0.28 (97.8)

0.38 (97.6)

0.38 (97.1)

0.01 (2.2)

0.01 (2.3)

0.01 (2.9)

Male-Female

0.28 (97.8)

0.38 (99.5)

0.38 (97.1)

0.01 (2.2)

0.01 (2.3)

0.01 (2.9)

Advanced-Backward

0.28 (97.8)

0.38 (97.7)

0.38 (97.1)

0.01 (2.2)

0.01 (2.3)

0.01 (2.9)

Regular-Casual

0.22 (78.4)

0.33 (83.5)

0.35 (91.3)

0.07 (21.6)

0.06 (16.5)

0.03 (8.7)

Source: Same as Table 1. Notes: Figures in parenthesis are percentages to total disparity

Employment, Occupational Disparities, and Wages in India

165

Summary It can thus be commented that inter-group or vertical differences in wage rates in India are rising in the last decade except the convergence between males and females. Added to this is the growing horizontal differential within groups. The new economic regime with its focus on private initiative is therefore consolidating both vertical and horizontal segmentation in the labour market in terms of wage rates. Given the ground rules of economic growth at present – rising regional concentration of economic activities, increasing skill-biased growth, sudden expansion of specific sectors & allied occupations – the results are not surprising. Policy makers therefore need to intervene in spreading the labour demand across much wider physical space and also across much broader skill/education base. This will solve the problems of disparities in wage rates and also that in job availability so that differentials in total annual income, the cynosure of human advancement, narrow down, both across and within groups. Given that relative contribution of inter-group disparity is highest among Occupational groups, it is pertinent to infer that aggregate wage inequality, and also that between the other sub-groups are caused by occupational disparities up to a substantial extent. It is this that provokes us to explore occupational distributions among groups of workers and decompose wage gaps among them into that explained by Endowment Gap and that due to pure Discrimination Effect, which we explore next.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

OCCUPATIONAL DISPARITY As has already been commented, researchers have pointed at differences in occupational disparity as one of the major reasons behind wage differences between groups of workers. Therefore we must first discuss the occupational pattern of the various sub-groups. It is observed that there are substantial differences between the sub-groups we have formed regarding the occupational distribution of workers (Table 5). While the share of bluecoloured workers or Grade-3 workers (those engaged in production, transport and other miscellaneous occupations) are the highest for all the subgroups, relative share of the other two types - Grade-1 workers (white-collared or those in Professional, Technicians and Administrative occupations) and Grade-2 workers (pink-collared or those in Clerical, Sales and Service related occupations) are different for the subgroups. Also, we would explore what part of the occupational disparity is due to differences in endowment and what part is due to entry barrier. Let us discuss them in detail. Table 5. Wage Employment by Occupation and Sub-groups - 1993-2004 (in Millions excluding Farmers) Year 1993

Groups Rural Urban LIG States

Tec 1.97 1.84 1.06

Pr Adm Clr Sls Srv Prd Trp nec 2.73 0.24 2.90 0.91 2.36 11.34 1.24 4.60 2.73 0.99 5.75 2.04 4.34 11.58 1.62 2.89 1.64 0.12 1.01 0.15 2.32 3.80 0.02 1.22

Total 28.28 33.78 11.34

Rajarshi Majumder and Dipa Mukherjee

166

Table 5. (Continued)

1999

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2004

HIG States Female Male Backward Advanced Casual Regular Aggregate Rural Urban LIG States HIG States Female Male Backward Advanced Casual Regular Aggregate Rural Urban LIG States HIG States Female Male Backward Advanced Casual Regular Aggregate

2.75 0.95 2.86 1.42 2.39 2.26 1.56 3.81 2.64 2.73 1.27 4.10 1.12 4.25 1.95 3.42 2.99 2.38 5.36 0.74 1.89 0.64 1.99 0.42 2.22 0.08 2.55 1.64 0.99 2.63

3.81 0.64 4.82 0.14 5.32 2.90 2.56 5.45 3.09 3.08 2.16 4.01 0.80 5.37 0.10 6.06 3.39 2.78 6.17 4.14 4.00 3.05 5.10 1.34 6.80 0.14 8.01 5.59 2.56 8.14

1.10 0.09 1.13 0.14 1.08 0.49 0.74 1.22 0.21 1.23 0.10 1.35 0.12 1.32 0.05 1.40 0.57 0.87 1.45 0.37 1.44 0.18 1.63 0.15 1.67 0.04 1.78 0.91 0.91 1.82

7.65 1.25 7.41 0.19 8.47 4.01 4.65 8.65 3.39 6.37 1.21 8.56 1.34 8.42 0.31 9.45 4.44 5.33 9.76 3.46 6.49 1.44 8.51 1.83 8.12 0.23 9.72 6.13 3.82 9.95

2.80 0.37 2.58 0.83 2.11 1.28 1.66 2.95 1.04 2.59 0.18 3.45 0.33 3.30 0.69 2.94 1.58 2.05 3.63 1.78 3.49 0.26 5.01 0.84 4.44 0.68 4.59 3.28 1.99 5.27

4.38 2.03 4.68 1.70 5.00 3.21 3.49 6.70 2.65 4.99 2.52 5.12 1.90 5.75 1.83 5.82 3.45 4.19 7.65 3.36 6.30 3.92 5.74 3.53 6.14 1.70 7.96 5.86 3.81 9.66

19.12 5.96 16.96 13.07 9.85 10.38 12.55 22.92 11.97 13.05 2.96 22.07 5.13 19.90 13.31 11.71 10.10 14.92 25.02 18.14 15.45 4.35 29.24 10.66 22.93 19.61 13.97 21.73 11.86 33.59

2.84 0.55 2.31 0.84 2.03 1.46 1.40 2.86 1.95 2.05 0.03 3.97 0.57 3.44 1.08 2.93 1.79 2.22 4.00 2.55 2.24 0.03 4.75 1.10 3.68 1.01 3.77 3.18 1.60 4.78

6.26 3.20 4.29 6.11 1.38 4.65 2.84 7.49 5.89 3.17 1.25 7.80 3.03 6.03 7.85 1.21 6.36 2.70 9.05 7.43 2.40 1.49 8.34 4.82 5.01 8.87 0.96 8.03 1.80 9.83

50.71 15.03 47.03 24.45 37.62 30.63 31.44 62.07 32.84 39.27 11.69 60.42 14.33 57.78 27.17 44.94 34.67 37.44 72.10 41.98 43.71 15.37 70.32 24.69 61.00 32.37 53.32 56.35 29.34 85.69

Source: Same as Table 1. Notes: The occupational groups are: Tec – Technical; Pr – Professionals; Adm – Administrative; Clr – Clerical; Sls – Sales; Srv – Service; Prd - Production etc; Trp – Transport; nec – Unclassified Labourers

Sectoral Differences It is observed (Table 6) that about 60 percent of rural workers are in Grade-3 occupations and only about 15 percent are in Grade-2 occupations, the corresponding figures for the urban workers being 45 percent and 18 percent respectively. Thus the rural workers are over represented in Grade-3 jobs, and under represented in the top and middle grade jobs. It is also observed that the gaps are widening over time as indicated by the Segregation Index.

Employment, Occupational Disparities, and Wages in India

167

Table 6. Occupational Distribution of Workers (%in various occupations) – Rural/Urban Occupation Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

1993 7.0 9.6 0.8 10.3 3.2 8.3 40.1 4.4 16.3

Rural 1999 8.0 9.4 0.7 10.3 3.2 8.1 36.5 5.9 17.9

2004 1.8 9.9 0.9 8.3 4.2 8.0 43.2 6.1 17.7

1993 5.5 8.1 2.9 17.0 6.0 12.9 34.3 4.8 8.6

Urban 1999 6.9 7.9 3.1 16.2 6.6 12.7 33.2 5.2 8.1

2004 4.3 9.2 3.3 14.8 8.0 14.4 35.4 5.1 5.5

Urban-Rural Gap 1993 1999 2004 -1.5 -1.1 2.6 -1.6 -1.5 -0.7 2.1 2.5 2.4 6.8 5.9 6.6 2.8 3.4 3.8 4.5 4.6 6.4 -5.8 -3.2 -7.9 0.4 -0.7 -0.9 -7.7 -9.9 -12.2

Source: Same as Table 1.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

As we have noted earlier, a part of such differences can be explained by differences in endowment level of the two groups of workers. If occupational engagement depends on skill, education and training it is quite natural that occupational patterns would be different for differently endowed people. Using Multinomial Logit Model and education and training as explanatory variables, estimated non-discriminatory occupational distribution for each of the groups can also be determined. Gaps in these proportions would be due to endowment gaps between the groups. The remaining would be pure discrimination or entry barrier whereby some groups are underrepresented in some occupations even with similar endowment pattern. In terms of the sectoral differences it is observed that Endowment Differences is about half of the Entry Barrier effect, indicating that discrimination of rural workers in terms of occupational pattern is substantial (Table 11). This is mainly because of absence of adequately diversified non-farm sector in the rural areas due to which non-farm rural workers are engaged mainly in construction and allied jobs. Hence their occupational pattern is relatively skewed. It is also observed that over time the sectoral segregation index have increased. A possible reason may be lack of educational facilities in the rural areas and the emergence of new jobs in the post-reform period mainly in the urban areas. As we will see later, this has serious repercussion on the sectoral earning gaps.

Regional Differences Another dimension of spatial disparity is difference between various regions. Instead of geographical grouping we have divided the Indian states into two groups – Low Income Group (LIG) states with per capita income less than the national average, and High Income Group (HIG) states with PCI more than the national average. It is observed that occupational patterns in the two groups of states are almost similar (Table 7). The HIG states are marginally over represented in Grade-2 jobs while the LIG states are marginally over represented in Grade-1 jobs. The segregation index in this case therefore is quite low.

Rajarshi Majumder and Dipa Mukherjee

168

Table 7. Occupational Distribution of Workers (%in various occupations) – LIG/HIG States Occupation Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

LIG States 1993 1999 7.4 8.6 9.5 9.8 1.6 1.7 13.1 12.8 4.2 4.6 10.5 10.0 33.9 29.1 4.8 5.2 15.2 18.3

2004 2.9 9.9 1.6 10.9 5.8 10.4 38.6 5.6 14.3

HIG States 1993 1999 4.9 6.3 8.1 7.4 2.3 2.3 14.8 14.2 5.3 5.5 11.1 11.2 39.9 39.9 4.5 5.9 9.0 7.2

2004 3.4 8.7 3.1 13.0 6.8 13.0 40.4 5.4 6.1

HIG-LIG Gap 1993 1999 2004 -2.4 -2.3 0.5 -1.3 -2.3 -1.2 0.7 0.7 1.5 1.7 1.4 2.2 1.1 0.9 1.0 0.6 1.2 2.6 6.0 10.7 1.9 -0.3 0.8 -0.2 -6.1 -11.1 -8.1

Source: Same as Table 1.

However, here too entry barrier is larger compared to endowment differences, indicating that disparate job availability and job diversification are more important factors in explaining occupational disparity among regions, rather than the skill factor. Also, the difference increased sharply in the immediate post-reform period and though decreased thereafter, the levels in 2004 was almost same as that in 1993. Differences due to entry barrier followed similar pattern, but endowment gaps have continuously increased in the post-reform period. This may have been caused by continuous migration of skilled labour force from the backward states to advance states in search of greener pastures. Availability of new and high paying jobs, better educational opportunities, and better living in the better-off states are taking away the best of workers from the LIG states leading to such rise in endowment differences (Maharatna, 2003, 2005)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Gender Differences A major aspect of interpersonal disparities is the issue of gender gaps and gender discrimination. While historically occupational segregation along lines of gender have been existent, it is expected that with development such boundaries will become less rigid and the two groups will be more or less similarly distributed across the occupations. Also, since we have left out the farm sector and are exploring only the wage-workers, the general observation that a large proportion of women are engaged in domestic duties and therefore under-represented in the other ‘occupations’ does not hold here. In spite of this, we find that the disparity in occupational pattern between males and females was substantially high in 1993, with a segregation index of over 20. Females were over represented in Grade-1 jobs and grossly underrepresented in Grade-3 jobs (Table 8). They were also over represented in Service related jobs, as expected. During the next decade, the segregation index increased as this structural pattern became reinforced with women shifting further from production related jobs to Services, Technical, and Professional occupations. The exact nature of such changes will be clearer when we examine the wage and earning dynamics at a later section.

Employment, Occupational Disparities, and Wages in India

169

Table 8. Occupational Distribution of Workers (%in various occupations) – Female/Male Occupation Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

1993 9.3 14.5 1.1 8.9 1.3 20.5 33.5 0.2 10.8

Female 1999 10.8 18.5 0.9 10.3 1.6 21.6 25.3 0.3 10.7

2004 4.2 19.8 1.2 9.4 1.7 25.5 28.3 0.2 9.7

1993 5.4 7.5 2.2 15.1 5.5 8.6 37.7 5.6 12.3

Male 1999 6.8 6.6 2.2 14.2 5.7 8.5 36.5 6.6 12.9

2004 2.8 7.2 2.3 12.1 7.1 8.2 41.6 6.8 11.9

Male-Female Gap 1993 1999 2004 -3.9 -4.1 -1.4 -6.9 -11.9 -12.6 1.1 1.4 1.1 6.2 3.8 2.7 4.2 4.1 5.4 -11.8 -13.1 -17.4 4.2 11.2 13.3 5.4 6.3 6.6 1.5 2.2 2.2

Source: Same as Table 1.

Coming to the components of occupational disparity, we find that the Entry Barrier was more than thrice of the Endowment differences in 1993, indicating strong bias in job selection along gender lines. In 1999, both the endowment difference and the entry barrier increased marginally. Thus, during the initial years, job discrimination was the major factor influencing occupational disparity among males and females. During 1999-2004, while the entry barrier remained unchanged, endowment differences increased marginally, indicating that the occupational divide among the genders is reinforced in recent times. At present therefore, gender disparity in occupational pattern is due to both differences in endowment and due to entry barriers.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Social Group Disparity Another dimension of inequality that is quite significant in the Indian context is that between the social groups – Scheduled Castes & Tribes (Backward Classes) on the one hand, and the rest (Advanced Classes) on the other. It has been observed by several researchers that substantial socio-economic disparity does exist between the Advanced and the Backward classes in India [see Nayak (1995) for a brief review]. Similar differences are observed in their Occupational pattern also. More than 60 per cent of the backward class workers are engaged in Grade-3 occupations and only about 8-14 per cent in Grade-1 occupations (Table 9). Contrary to this, just about 50 per cent of the advanced class workers are in Grade-3 jobs and more than 18 per cent are in Grade-1 jobs. This segregation marginally decreased during 1993-99 period but has again increased thereafter. Significant, however, is the fact that only here the differences are more due to endowment gaps rather than due to entry barrier. Thus, educational backwardness and lack of technical skill are preventing the backward classes from diversifying into Grade-1 and 2 jobs and about two-third of them are continuing in Grade-3 jobs. Recent efforts to extend affirmative actions to the job market therefore have to be preceded by efforts to provide basic education and training to the backward classes. Enabling endowment formation would be more suitable a policy action rather than simply creating more quotas in professional education and private sector jobs.

Rajarshi Majumder and Dipa Mukherjee

170

Table 9. Occupational Distribution of Workers (%in various occupations) – Backward/Advanced Occupation Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

Backward 1993 1999 6.3 7.8 4.3 5.6 0.6 0.9 8.3 9.3 2.4 2.3 13.5 13.2 39.7 35.8 3.7 4.0 21.3 21.1

2004 1.7 5.4 0.6 7.4 3.4 14.3 43.2 4.5 19.5

Advanced 1993 1999 6.1 7.4 10.2 9.3 2.4 2.3 15.7 14.6 5.5 5.7 9.9 10.0 36.1 34.4 4.9 6.0 9.1 10.4

2004 3.6 11.2 2.7 13.3 7.3 10.1 37.6 6.0 8.2

Adv-Back Gap 1993 1999 2004 -0.2 -0.4 1.9 6.0 3.7 5.7 1.8 1.4 2.1 7.4 5.2 5.9 3.0 3.4 3.9 -3.5 -3.3 -4.2 -3.6 -1.3 -5.6 1.2 2.0 1.6 -12.2 -10.7 -11.3

Source: Same as Table 1.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Disparity among Job Types The most notorious of disparities in the wage labour market is that between the regular workers and the casual workers. While only about 7 per cent of workers in India are in the organised sector and the regularity of the rest are themselves questionable, there are also a large mass of casual workers who do not even have regular work available to them. A major difference exists between these two groups of workers in terms of their occupational pattern. More than 80 per cent of the Casual Workers are in Grade-3 jobs and a few are in Administrative, Sales, and Service occupations (Table 10). Their representations in the other occupations are almost negligible. Contrary to this, the regular workers are more evenly spread out – only about 35 per cent of them are in Grade-3 jobs, about 40 per cent in Grade-2 jobs, and the remaining 25 per cent in Grade-1 jobs. Moreover, the proportion of casual workers with no specific occupation is increasing marginally over the study period. Their condition is thus similar to the females in the sense that in absence of regular job they are also taking up sundry jobs to make a living. It must also be remembered that in addition to being engaged predominantly in the Grade-3 jobs, they are only irregularly employed and hence the low wages that are paid are also very infrequent, making their economic condition quite insufferable. The disparity between these two groups is rising over time as indicated by the segregation index – which is also highest among all the inter-group comparisons we have made. If we now examine the components of such segregation, we find that while endowment gaps do exist, entry barrier is the dominant factor behind such striking difference between the Regular and the Casual workers. Also, while the difference in educational and training levels are narrowing down over time, the discrimination factor is going up. This indicates that regular jobs are becoming more and more restricted and people with adequate endowments are now unable to find regular jobs and are forced to take up casual jobs. This again points out the worsening of labour market conditions in the last decade.

Employment, Occupational Disparities, and Wages in India

171

Table 10. Occupational Distribution of Workers (%in various occupations) – Casual/Regular Occupation

1993 5.8 0.6 0.6 0.8 3.4 7.0 53.5 3.4 25.0

Technical Professionals Administrative Clerical Sales Service Production etc. Transport Labourers nec

Casual 1999 7.2 0.4 0.2 1.1 2.5 6.7 49.0 4.0 28.9

2004 0.2 0.4 0.1 0.7 2.1 5.3 60.6 3.1 27.4

1993 6.4 14.1 2.9 22.5 5.6 13.3 26.2 5.4 3.7

Regular 1999 7.6 13.5 3.1 21.0 6.6 12.9 26.1 6.5 2.7

Regular-Casual Gap 2004 1993 1999 2004 4.8 0.5 0.4 4.5 15.0 13.6 13.1 14.6 3.3 2.3 2.9 3.2 18.2 21.7 19.9 17.5 8.6 2.2 4.0 6.5 14.9 6.3 6.2 9.7 26.2 -27.3 -22.9 -34.4 7.1 2.0 2.5 3.9 1.8 -21.3 -26.2 -25.6

Source: Same as Table 1.

Summary The occupational disparity among population sub-groups in India is therefore rising in the recent times, indicated by the increasing Segregation Indices (Table 11). Major part of this disparity is explained by entry barrier that in turn is due to differences in availability of jobs and diversity of economic structure. Also, at the face of increased mechanisation, skill is becoming less important than those factors. It therefore seems that the occupational disparity in India is mainly a demand side problem.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

EARNING DIFFERENCE DUE TO OCCUPATIONAL DISPARITY It is thus evident that Occupational pattern of different sub-groups of workers are quite dissimilar. A major reason behind earning differential between groups is this dissimilar occupational pattern of them, even if wages were identical. In this section we estimate and analyse earning differential due to occupational disparity assuming uniform nondiscriminatory wage structure.2 It is observed that substantial gaps do exist between the groups in this regard. Let us explore them. Table 11. Occupational Disparity and Segregation Index Occupation Rural-Urban Female-Male Backward-Advanced Casual-Regular LIG-HIG States

Source: Same as Table 1.

Actual Gaps 1993 1999 16.6 16.4 22.7 29.0 19.5 15.7 48.6 49.1 10.2 15.7

2004 21.7 31.3 21.2 60.0 9.5

Endowment Diff 1993 1999 2004 31.2 9.3 10.9 6.4 7.0 9.0 14.8 10.0 13.4 27.9 25.2 24.6 4.9 7.0 7.1

Entry Barrier 1993 1999 2004 34.1 15.2 20.6 22.3 26.5 26.3 8.9 7.8 9.4 40.2 43.3 55.2 9.1 13.7 8.5

Rajarshi Majumder and Dipa Mukherjee

172

Earning gap between Urban and Rural areas due to occupational pattern has quadrupled during 1993-2004 period (Table 12). Most of this difference is due to endowment gap among the workers, which is increasing over time. Differences between HIG and LIG states have also increased sharply during this period, and here too endowment difference is the dominant factor, though entry barrier has emerged to be sizeable in recent times. Table 12. Gross Earning Gaps due to Occupational Disparity – Total and Components Occupation Urban-Rural Gap Male-Female Gap Advncd-Bckwd Gap Regular-Casual Gap HIG-LIG Gap

Earning Gaps 1993 1999 2004 8.2 13.9 31.1 -5.0 -18.5 -6.7

Endowment Diff 1993 1999 2004 10.7 14.6 22.1 -12.5 0.5 1.7

Entry Barrier 1993 1999 2004 -2.5 -0.7 9.0 7.5 -19.0 -8.4

17.0

16.4

34.1

13.7

13.5

20.4

3.2

2.9

13.7

27.0 0.9

41.3 0.0

59.6 15.0

21.0 1.2

27.7 2.2

34.7 7.2

5.9 -0.4

13.6 -2.2

24.8 7.8

Source: Same as Table 1.

Considering inter-personal earning gaps due to occupational disparity, gender gap emerges to be negative, indicating that even with existing occupational pattern women would have earned more than the men if the wage structure were uniform and not discriminatory against female workers. Earning gaps due to occupational pattern among social classes and job-types are quite high and have more than doubled during the study period, most of which are due to endowment differences. This indicates that lower earning by backward classes and casual workers in recent years is accentuated by the fact that they are in low wage occupations. The significant rise in earning gap due to entry barrier to occupations in recent years indicates worsening of labour market conditions where endowments no longer ensure reasonable earnings.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

WAGE GAPS We have been looking at Occupational Disparity and its role in explaining Earning Differentials. As already mentioned in the section on Methodology, the second component of such differential is the Wage Gap between two groups of workers even within the same occupation. We may recall that a part of this Wage gap is due to differences in endowment of the two groups of workers – known as Endowment Effect in Wage setting. Another part may be pure discrimination in wage setting where workers with same endowment and in same occupation are paid lower wages – known as Remuneration Gap in Wage Setting. A log-linear wage function with (log of) daily wages being dependent on endowment factors, proxied by Educational & Technical training levels is used. The explanatory variables are expressed as Dummy Variables. The gaps in wages are then decomposed into Endowment Effect and Remuneration gap using steps as discussed in the Methodology section. We now explore the levels and trends in wage gaps (Table 13).

Employment, Occupational Disparities, and Wages in India

173

Table 13. Gross Earning Gaps due to Wage Disparity – Total and Components Occupation Urban-Rural Gap Male-Female Gap Advncd-Bckwd Gap Regular-Casual Gap HIG-LIG Gap

Earning Gaps 1993 1999 2004 27.1 32.9 25.9 37.0 44.2 39.9

Endowment Diff Remuneration Gap 1993 1999 2004 1993 1999 2004 6.3 9.8 7.0 20.8 23.0 18.9 6.6 8.1 2.7 30.3 36.0 37.1

13.9

16.1

14.1

7.2

9.2

6.9

6.7

6.9

7.2

42.1 7.7

53.9 21.2

36.2 21.9

6.0 3.9

5.8 6.0

10.2 6.4

36.0 3.8

48.1 15.2

26.0 15.5

Source: Same as Table 1.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Spatial Wage Gap The non-farm rural and urban wage gap diverged in the immediate post-reform period. Thereafter, it decreased and came below its initial level. Similar pattern is observed for the two components – endowment effect and remuneration effect. What is significant however, is the fact that only about 25 per cent of the total wage gap is explained by endowment differences, whereas more than 75 per cent is due to pure discrimination in wage setting. This indicates a depressed wage labour market in the rural areas vis-à-vis the urban areas, caused mainly by concentration of economic activities in the urban areas after the reforms. This confirms our earlier inference regarding occupational disparity. More alarming is the spatial wage gap between the HIG and the LIG states, especially the rise in this gap in the post-reform period from only 7 Rupees per manday in 1993 to 22 Rupees per man-day in 2004. While in 1993 about two-third of this gap was due to endowment effect, the relative magnitudes reversed and in 2004 about 70 per cent of the wage differential was due to remuneration gap. This again points to economic activities being concentrated in the HIG states in recent times, creating substantial demand for labour, and in the process raising wage rates sizeably in those regions relative to the LIG states. Spatial differences in wages are therefore more along the lines of divergence in labour demand and wage rates rather than skill component of workers therein. This perhaps is because of the nature of economic activities nowadays, especially in the production and the services sector, where introduction of advanced equipment has made skill component less important than before.

Gender Wage Gap Differences between Male and Female wage rates are substantially high and have been increasing further during the study period. A major part of this difference is due to discrimination in wage setting even for similarly endowed females, whose impact is increasing over time. While in 1993 about 15 per cent of the wage gap was due to endowment differences, in 2004 it accounted for less than 10 per cent of the total, indicating that contrary to desired objectives, discrimination against women in the workplace is rising. Availability of

174

Rajarshi Majumder and Dipa Mukherjee

women workers at substantially lower wages than men, and deteriorating labour market conditions forcing women to take up whatsoever jobs are available are possible reasons behind such dynamics.

Wage Gap among Social Classes Though the least among the wage gaps in recent years, differences between the backward classes (SC-STs) and the advanced classes are no less significant. The gap has also increased over time along with a fall in the endowment effect and a rise in the remuneration effect. Remarkably, it is only here that the effect of endowment was higher than that due to discrimination in the initial years, although in recent period the remuneration effect has marginally outweighed the endowment effect here too. It is thus necessary to not only safeguard the backward classes against low wage setting, but to improve their endowment levels through imparting of quality education and hands-on skill formation. Also, bringing them to technical and vocational training institutes would be much more effective than the headlines-hogging policies of reserving seats in blue chip Management Institutes. Targeting regions where these groups are spatially concentrated is also expected to bring better results rather than spreading the effort across the nation.

Wage Gap among Job Types

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Wage disparity among the Regular and Casual workers is the most glaring of the differences with the former earning more than three times per manday compared to the latter. This difference increased significantly in the immediate post-reform period but decreased sharply thereafter to come below its initial magnitude. Here too, remuneration gap or discrimination is the main reason, explaining about 80 per cent of the wage gap in 1993. Though its relative contribution decreased in later years, it still explains about 70 per cent of the wage gap between job types. Thus, lower wages of the casual workers are not because they have lesser educational levels or skill, but mainly because they are unable to enter the already saturated regular job market and are forced to take up ill-paid unclassified jobs.

Summary It is thus observed that the wage rates are segmented along lines of location, gender, social class, and job types. Wage differences are increasing in the post-reform period and these differences are mainly because of discrimination in wage setting, except among social classes where it is mostly due to endowment gap.

Employment, Occupational Disparities, and Wages in India

175

EARNING DIFFERENCES: TRENDS AND COMPONENTS We have so far analysed earning gap due to occupational disparity and that due to wage gap, including their components. Let us now combine them and explore the Gross Earning Differentials (Table 14). Table 14. Gross Earning Gaps - Total and Components (combined Occupation and Wage Disparity) Occupation Urban-Rural Gap Male-Female Gap Advncd-Bckwd Gap Regular-Casual Gap HIG-LIG Gap

Earning Gaps 1993 1999 2004 32.4 47.8 52.7 33.2 31.9 42.0

Endowment Diff 1993 1999 2004 17.0 24.4 29.1 -5.9 8.6 4.4

Discrimination 1993 1999 2004 15.4 23.4 23.6 39.1 23.3 37.6

28.9

32.8

46.4

20.9

22.6

27.3

8.0

10.2

19.1

64.2 5.4

88.3 21.4

96.4 32.4

27.1 5.1

33.6 8.3

44.9 13.6

37.1 0.3

54.7 13.1

51.5 18.8

Source: Same as Table 1.

Gross Earning Differences It is observed that the Gross Earning Differences between the different sub-groups are quite substantial – the disparity being the highest between the Regular and Casual workers (Rupees 96 per manday in 2004), and lowest between the HIG and LIG states (Rupees 32 per manday in 2004). The earning differences have increased by about 50 per cent during the study period for all the sub-groups except the differential between HIG and LIG states where it has become 6 times its initial value! Thus, the last decade has seen rising inequality in wage earnings per manday in India both at spatial and inter-personal level, but most significantly among the regions.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Occupational Disparity vs. Wage Gap Most of the Earning Differences is due to Wage Gap, i.e. due to unequal wages even within similar occupations for the gender and regional groups. For the sectoral, social class, and job-type disparities also, wage gaps were predominant in the initial period, but occupational effects have surpassed wage effects in recent times. This indicates that the lower earnings of rural workers, backward classes, and casual workers are mainly because of their concentration in less paying occupations in recent times.

176

Rajarshi Majumder and Dipa Mukherjee

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

EARNING DIFFERENCES: ENDOWMENT VS. DISCRIMINATION We have so far explored Gross Earning Gap due to Occupational disparity and that due to Wage Gap. Both of them contain components that are due to differences in endowment – education and training in this case – between the sub-groups. The other parts are due to Discrimination – Entry Barrier in Occupational Choice and Remuneration Gap in Wage setting. We can therefore combine the Endowment components on one hand and the Discrimination components on the other and examine the relative importance of them in explaining Gross Earning Differentials. It is observed that Endowment gaps and Discrimination are equally important in explaining Urban-Rural gross earning differences, with the former being marginally more so. This implies that there are substantial differences in educational attainment and skill formation between the urban and rural workers. In addition, wage labour market in rural areas is depressed and wage rates are considerably lower there compared to urban areas. The earning gap between HIG and LIG states have increased tremendously during the last decade mainly because of a significant rise in the discrimination effect. This indicates that the wage difference is not that much due to difference in education and skill in the two types of regions but because economic activities are polarised leading to lower wages in the LIG states even for equally endowed workers. The gender earning difference is predominantly due to Discrimination, which explains about 90 per cent of the total difference between male and female workers. This suggests that women have lower bargaining power in the wage labour market irrespective of their endowment level. However, the rising magnitude of endowment effect in recent years indicating increasing disparity between males and females in terms of education and training is a matter of grave concern. Earning difference between the social classes on the other hand is mainly because of endowment differences, though the relative importance of discrimination is found to be rising in recent years. This indicates that capability formation among SC\STs is still far behind the general caste and concerted effort must be made for improving the education and training levels among these groups. As we have noted earlier, earning disparity between the regular and casual workers are most glaring. Here too endowment differences and discrimination are equally important, though, the latter is marginally more so. It therefore appears that relatively less endowed workers are unable to find regular job and as they enter casual wage labour market they earn substantially less wages than what their qualifications deserve.

CONCLUSION We have explored Earning Differences among various spatial and socio-economic groups of workers in India over the last decade. The following major conclusions are arrived at. At the spatial level, disparities in earnings are increasing in the last decade, especially among the state groups. This differential is the aggregate result of rising occupational disparity among regions and differences in wage rates. It is also observed that the role of Discrimination and Entry Barrier is more important than that of endowment in explaining the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Employment, Occupational Disparities, and Wages in India

177

disparities. Concentration of economic activities in urban areas and HIG states is pulling up wages in those areas relative to LIG states and rural areas. In addition, diversification of employment structure is also restricted mainly to the advance regions. It appears that a spatial polarisation in operation with a virtuous cycle operating in the advance regions and a vicious cycle in the lagging areas, further widening the gap between them. Absolute level of endowment gap between HIG and LIG states is also increasing. This perhaps is the result of internal migration and brain drain from lagging regions to advance regions of the country. However the relative importance of endowment in wage setting is itself declining leading to lower skill premium in wages. At the inter-personal level, earning differences among males and females are also increasing, due to divergence of both occupational pattern and wages. The predominant factor behind wage disparities among genders is discrimination, and its importance vis-à-vis endowment difference is increasing over time. This highlights the lower bargaining power of women in the wage labour market. Thus the rising share of women in the workforce has to be viewed with concern. The absolute magnitude of endowment gaps between men and women is also increasing over the decade. A major thrust of ‘women empowerment’ must therefore be educational expansion and skill formation among the fairer sex. Earning differences between social classes are mostly due to skewed occupational distribution of the backward classes. This is mainly due to their lower endowment levels. Thus human capital formation and basic educational progress among the scheduled castes and tribes should be the main thrust of policy action. This would achieve the twin objectives of enabling them to access white collared occupations and also raise their wage levels. Consequently, earning differences between social groups would come down. The most striking of earning differences is that between the regular and the casual workers. Here too, the main reason is concentration of casual workers in relatively less paying occupations. Unlike the backward classes, their endowment levels are not much different compared to regular workers. It appears that in the new economic regime employment opportunities are deteriorating and substantial casualisation of workforce is taking place. In absence of regular job availability, even skilled and qualified workers are forced to take up casual jobs. Most of these are the grade-3 jobs and hence the casual workers are stuck in a low earning trap. It thus emerges that earning differentials are increasing in India in recent times due to a variety of reasons. While economic polarisation is the main factor for some groups, lack of skill and exploitation are main reasons elsewhere. Reduction of earning disparity therefore requires a multi-pronged and targeted approach. Spatial spread of economic activities, diversification of rural economy, preventing exploitation of women in wage setting, expansion of education and skill among backward classes, and ensuring fair wages for casual jobs, are some of the suggested action areas. This calls for improving conditions of work, creating more jobs in lagging regions, and also empowering people to access those jobs. Only a concerted approach can reduce disparities in earnings and usher in a more egalitarian society.

Notes 1

The One-digit National Occupational Classifications in India are as follows: 0 – Technical and related workers; 1 – Professional workers; 2 – Administrative, executive and managerial workers; 3 – Clerical and related workers; 4 – Sales workers; 5 – Service workers;

178

Rajarshi Majumder and Dipa Mukherjee

6 – Farmers, fishermen, hunters, loggers and related workers; 7-8-9 – Production and related workers, transport equipment operators, and Labourers not elsewhere classified. 2 The Non-discriminatory Wage structure is the Wages received by the advantaged group – typically the Urban, Male, Advanced Class, Regular worker, and those in the HIG states.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Adamou, A. (2006) – ‘Sample Selection Problem Analysis with Probit Selection Equation and only Exogenous Explanatory Variables: A General Discussion and the Heckit Critique’, Dept of Economics, University of Cyprus. (from http://www.econ.ucy.ac.cy/~adamos) Banerjee, B and J B Knight (1985) – ‘Caste Discrimination in the Indian Labour Market’, Journal of Development Economics, 17: pp 277-307. Blinder, A. (1973) – ‘Wage Discrimination: Reduced Form and Structural Estimates’, Journal of Human Resources, Vol. 8 (4) Brown, R S, M Moon and B S Zoloth (1980) – ‘Incorporating Occupational Attainment in Studies of Male/Female Earnings Differentials’, Journal of Human Resources, 15, pp 328. Cotton, C J (1988) – ‘On the Decomposition of Wage Differentials’, Review of Economics and Statistics, 70, pp 236-43. Duraiswamy, P. (2002) – ‘Changes in the Returns to Education in India, 1983-94: By Gender, Age-Cohort and Location’, Economics of Education Review, Vol. 21 (6) Dutta, Puja Vasudeva (2005) – ‘Accounting for Wage Inequality in India’, PRUS Working Papers 29, Poverty Research Unit at Sussex, University of Sussex Galbraith, JK, D. Roy Chowdhury and S Shrivastava (2004) – ‘Pay Inequality in the Indian Manufacturing Sector, 1979-1998’, University of Texas Inequality Project Working Paper 28. Greenhalgh, C (1980) –‘Male-Female Differentials in Great Britain: Is Marriage an Equal opportunity?’ Economic Journal, 90, pp 751-75. Katz, L. and D. Autor (1999) – ‘Changes in Wage Structure and Earnings Inequality’, in O. Ashenfelter & D. Card (eds) Handbook of Labor Economics, North Holland Kingdon, G. G. (1998) – ‘Does the Labour Market Explain Lower Female Schooling in India?’ – Journal of Development Studies, Vol. 35 (1). Kingdon, G. G. and J. Unni (2001) – ‘Education and Women’s Labour Market Outcomes in India’, Education Economics, Vol. 9 (2) Leung S.F., and S. Yu (1996) – On the Choice between Sample Selection and Two-Part Models’, Journal of Econometrics, 72: 197—229. Madheswaran, S. and Paul Attewell, (2007) – ‘Caste Discrimination in the Indian Urban Labour Market: Evidence from the National Sample Survey’, Economic and Political Weekly, Vol , No. , October, 2007. Maharatna, A. (2003) – ‘Migration, Mediocrity and ‘Misery’’, Economic and Political Weekly, Vol. 38, No. 08 February, 2003 Maharatna, A. (2005) – ‘West Bengal, Relatively Speaking: Population, Economy and Society Since 1970s’, Paper presented at the Annual Conference of The Indian

Employment, Occupational Disparities, and Wages in India

179

Econometric Society held at Department of Economics, Jadavpur University, Kolkata during January 2005; Mincer, J. (1974) – ‘Schooling, Experience and Earnings’, Columbia University Press, New York. Nayak, P. (1995) - “Economic Development and Social Exclusion in India”, in Social exclusion and South Asia. Geneva, ILO. Neumark, D (1988) – ‘Employers Discriminatory Behaviour and the Esti mation of Wage Discrimination’, Journal of Human Resources, 23. Oaxaca, R L and M R Ransom (1994) – ‘On Discrimination and the Decomposition of Wage Differentials’, Journal of Econometrics, 61, Oaxaca, R. (1973) – ‘Male-Female Differentials in Urban Labour Markets’, International Economic Review, Vol. 14 Peracchi, G.D.L.F. (2007) – ‘A Sample Selection Model for Unit and Item Nonresponse in Cross-sectional Surveys’, CEIS Tor Vergata Research Paper Series, Working Paper No. 99, March 2007, (from http://papers.ssrn.com/paper.taf?abstract_id=967391) Puhani P. A. (2000) – ‘The Heckman Correction for Sample Selection and its Critique’, Journal of Economic Surveys, 14: 53—68. Vella F. (1998) – ‘Estimating Models with Sample Selection Bias: A Survey, Journal of Human Resources, 33: 127—169. Wood, A. (1997) – ‘Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom’, World Bank Economic Review, Vol. 11 (1).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

DATABASE NSSO (2006) – Unit level Records on Seventh Quinquennial Survey on Employment and Unemployment in India 2004-05, NSS 61st Round - July 2004-June 2005, National Sample Survey Organisation, Government Of India NSSO (2000) – Unit Level Records on Sixth Quinquennial Survey on Employment and Unemployment in India 1999-2000, NSS 55th Round - July 1999-June 2000, National Sample Survey Organisation, Government Of India NSSO (1996) – Unit Level Records on Sixth Quinquennial Survey on Employment and Unemployment in India 1993-1994, NSS Fiftieth Round - July 1993-June 1994, National Sample Survey Organisation, Government Of India. Websites: www.mospi.nic.in, www.planningcommission.nic.in

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S.Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 7

SHORT- AND LONG-RUN EFFECTS OF PRODUCTIVITY SHOCKS IN AN INTERTEMPORAL MODEL1 Pu Chen2, Gang Gong3, Armon Rezai4 and Willi Semmler5 2

Department of Economics, University of Bielefeld, Germany Department of Economics,School of Economics and Management, Tsinghua University, Beijing, P.R. China 4 Schwartz Center of Economic Policy Analysis, New School, New York, USA 5 Schwartz Center of Economic Policy Analysis, New School, New York, USA, and Center for Empirical Macroeconomics, Bielefeld, Germany 3

ABSTRACT

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In this paper we study the relationship betweeen unemployment and productivity growth with respect to the short and long run. In order to do so, we propose a two stage dynamic general equilibrium model. The preditions of which we then take to US unemployment and productivity data. Using MLE and SVAR, the model’s preditions appear to be correct: Productivity growth has an abiguous effect on unemployment; reducing it in the long run, while temporarily increasing it in the short run.

Referee: Tarron Khemarj, New College of Florida JEL Classification: E24, E30, O40 Keywords: RBC models, unemployment, productivity growth

1

We gratefully acknowledge the valuable contributions of Tarron Khemraj in preparation of this paper. Department of Economics, University of Bielefeld 3 Department of Economics,School of Economics and Management, Tsinghua University, Beijing 4 Schwartz Center of Economic Policy Analysis, New School, New York 5 Schwartz Center of Economic Policy Analysis, New School, New York, and Center for Empirical Macroeconomics, Bielefeld 2

182

Pu Chen, Gang Gong, Armon Rezai et al.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION The relationship between productivity growth and unemployment has been debated ever since the classical economists. Already Ricardo asked whether technical progress is a virtue or vice. Most economists maintain that long run technical progress and growth have led to a rising standard of living in advanced countries. Others claim that technical progress and productivity growth may have contributed to unemployment. This is often stated with respect to European economies with its high rate of unemployment since the end of the 1980’s. The nexus of productivity and employment is also important for the study of Okun's Law. If employment is correlated with output, but does not reveal a one-to-one relationship as Okun (1962) states it, the relationship may change over time, due to changing growth rates of productivity. Thus, the study of the impact of productivity on employment becomes a relevant issue. After Okun's study was published in 1962, many authors have been involved in this discussion of the relationship of productivity and employment either from the short run or long run perspective. Particularly relevant authors are Tobin (1993), Kaldor (1985), Solow (1997) and Rowthorn (1999). We will discuss their contribution, in section 2 of the paper. We also want to mention that ever since the Real Business Cycle (RBC) theorists have postulated technology shocks as driving force of business cycles, an extensive controversy over the relationship of employment and productivity growth has started. In RBC models technology shocks, output and employment (measured as hours worked) are predicted to be positively correlated. This claim has been made the focus of numerous econometric studies. Employing the Blanchard and Quah (1989) research agenda by using VAR estimates, studies by Gali (1999), Gali and Rabanal (2005), Francis and Ramey (2004) and Basu et al. (2006) find a negative correlation of employment and productivity growth, once account is taken of both demand and supply shocks affecting output. While most of the econometric work has studied the effects of productivity growth on employment (hours worked), using a VAR methodology, we want to shift the emphasis to the nexus of productivity growth and unemployment in our paper. Although unemployment rates may be impacted by population growth, demographic shifts, changing labor market participation rates of certain parts of the population and so on, one might presume that the demand side of labor, the offered employment by firms, is the most essential factor for driving the unemployment rate. In this paper we frame the discussion of the relationship between productivity and unemployment into a dynamic general equilibrium model, in which the representative firm and representative household optimize their decisions. The short run and the long run relations between the productivity and the unemployment are the result of the specific dynamic property of the model due to demand shocks and technology shocks. To quantify these short run and long run relations we apply two econometric methods. First, we employ a Maximum Likelihood (ML) method to estimate the short run and the long run effect simultaneously. Second, we use a structural VAR with long run restrictions similar to the work starting with Blanchard and Quah (1989). Here we presume that in the long run non-technology shocks cannot exert a permanent effect on productivity.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

183

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2. STYLIZED FACTS Using the data set by Francis and Ramey (2004), which among other data contains time series for productivity growth and employment from 1889 to 2002, one can observe the following stylized facts. Over the last 100 years, total employment grew tremendously and was in 2002 6.5 times higher than in 1889. This corresponds to an annual growth rate of 1.6% in employment. At the same time, however, labor productivity increased by 2.4% p.a. and was in 2002 13.5 times higher than it was in 1889. Real output expanded in this period with an annual growth rate of 3.4% and was in 2002 67 times its 1889 value. This shows that the US economy has undergone significant changes in the last 100 and something years. However, the picture is even more complicated as the dynamics are not as smooth and constant as one might hope. Not only has there been tremendous structural change (change of employment from agriculture to manufacturing and subsequently to the service sector), but there are also shifts in time trends. As a result the economy's characteristics have clearly changed after 1950. Especially when looking at productivity growth, one can observe that changes have become less volatile and more persistent. The correlation between various parameters varies also with the time span considered. Nonetheless, Uhlig (2006) points out that all of these coefficients are positive and that, therefore, technical progress and growth in GDP are certainly not harming employment and over most periods net employment is created.6 Uhlig (2006) used the data set developed by Francis and Ramey (2004). Using data supplied by the BLS, one can expand the set by the unemployment series for this period. This enables one to show that this assessment of Uhlig and others has to be treated with caution. Separating the long and short run effects by taking 10 years averages and the actual deviation thereof, one can relate short and long run productivity growth and unemployment.7 A closer examination of the data, then, reveals clear differences for the long and short run and, as mentioned above, for the periods before and after World War II (WWII) (ie 1890 - 1930 and 1945 - 2002). Uhlig's conclusions that productivity growth does not harm employment and that the structural change after WWII is not significant for this assessment can be proven to be incorrect. In the short run productivity and unemployment can be positively correlated. Yet, for the different periods, the long and short run relationship between productivity growth and unemployment can take on slightly different slopes. Specifically, the short run productivity growth and unemployment are positively correlated after the Second World War, while the result is ambiguous in the period 1890 to 1930. Due to those structural changes in the relationship between unemployment and productivity growth, we limit our investigation to the post-WWII period. For this time span data availability and data quality is also much better and this allows us to use quarterly data. This data is, again, taken from the BLS. Specifically, it includes the unemployment rate and productivity in non-farming business from 1959Q1 to 2005Q4. Figures 1 display the 6

Of course, there are distributional aspects of the effects of productivity on employment. Productivity and output growth might increase inequality. Economic growth might not arrive at low levels of income. Uhlig (2006) also hints at those distributional aspects of the productivity and employment nexus.

Pu Chen, Gang Gong, Armon Rezai et al.

184

relationship between these two series. As for the previous data, one can observe that unemployment and short and long run productivity growth are correlated positively and negatively, respectively. Note that the averaging period that has been used here is 12 years.

Figure 1. Post-WWII: Short and Long Run Effects (Quarterly)

These trends can also be observed in table Table 1 which depicts the correlations between the unemployment rate and short and long run productivity growth. While short run productivity growth and unemployment are weakly positively correlated, long run productivity growth is strongly negatively correlated with unemployment. One can also see that the two productivity series are virtually not correlated at all. This is important for section 4 where they are assumed to be independent of each other.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 1. Correlations of Unemployment, Short- and Long-Run Prod. Growth

Unempl. Short Run Long Run

Unempl. 1 0.209 -0.757

Short Run 0.209 1 -0.0398

Long Run -0.757 -0.0398 1

One relevant factor that may influence the relation between the technology growth and the unemployment is demand. To eliminate this influence we use in this study the purified technology growth data given in Basu et al. (2006) and study the relation between the technology growth and unemployment. A scatter plot of the unemployment rate on the purified technology growth shows a slightly positive correlation among them.

7

Note that we consider moving averages as preferable over HP-filtering since it leaves the time series relatively unaltered.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

185

Figure 2. Technology Growth and Unemployment in Short Run

A sample regression of unemplt on DTBFK t leads to the following results. Table 2. Estimation Output for Linear Model Variable Constant DTBFK

Coeff. 0.0575 0.0648

Std Error 0.0023 0.1520

T-Stat 24.76 0.42

Signif. 0.00 0.67

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The positive coefficient before DTBFK is insignificant, implying that there is no linear relation between DTBFK t and unemplt . To explore the possible long-run relation among the technology growth and the unemployment we generate scatter plots of their moving averages at various bandwidths. It can be clearly seen that while the short-run correlation is slightly positive, the long-run correlation at a period length of 10 years become negative. Given these results one is tempted to conclude that productivity shocks (here measured in growth rates of productivity) are likely to increase unemployment in the short term and to reduce unemployment in the long term, as some of the classical literature and also some of the critics of the RBC literature focus on.

186

Pu Chen, Gang Gong, Armon Rezai et al.

Figure 3. Technology Growth and Unemployment in Short run and Long run

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

3. A MODEL OF THE SHORT- AND LONG-RUN EFFECTS OF PRODUCTIVITY SHOCKS In order to explain the above-mentioned short- and long-run effects of technology shocks, we suggest a macroeconomic model that follows closely established standards in the literature. It allows for both productivity shocks as well as demand shocks. This will help us to explain the puzzle that in the short run a technology shock may have a different effect on unemployment than in the long run. Our model is based on an intertemporal decision model where the household determines its consumption and leisure pattern with respect to a budget constraint and an accumulated capital stock. The economy is subject to a continuous stream of technology shocks. In our variant there is a search and matching in the labor market which will allow labor market transactions to take place out of equilibrium. If supply and demand for labor do not match, there are constraints for households and households are subsequently allowed to re-optimize. As money is not included in this model, price dynamic is neglected and wages representing real wage developments. To take a shortcut, we can presume, as the dynamic general equilibrium (DGE) model does, that the economy is characterized by a representative household and a representative firm. Agents enter market exchanges in three markets: the product, the labor and the capital

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

187

market. The household owns all factors of production and sells factor services to the firm and buys its products for consumption or accumulation of the capital stock. The product market is assumed to be imperfectly competitive, with the firm facing a perceived demand curve and a sticky price. Unlike the standard DGE model with competitive markets, the market in this model will be re-opened at the beginning of each period t , necessary to ensure adjustment in response to a non-cleared labor market after the first round of a matching process. The non-clearing of the market in the matching process is caused by wage stickiness as the sequence of wages

{wt }t∞= 0 is contracted and preset at t = 0 and will not be allowed to change even if the market does not clear. The decision process, therefore, has two stages: in a first step, households determine their consumption and labor supply pattern, in a second step, in case labor demand does match labor supply, households re-optimize their consumption plans following the realized transactions on the factor market. In the first step, at period t = 0 , the household expects a series of technology shocks

{Et At +i }i∞= 0 and real wages and interest rates {Et wt +i , Et rt +i }i∞= 0 . The decision problem of the household is then to choose a sequence of planned consumption and labor effort

{ctd+i , nts+i }i∞= 0 such that ⎡∞ i ⎤ d s max Et ⎢∑β U (ct +i , nt +i )⎥ {ctd+ i ,nts+ i }i∞= 0 ⎣ i=0 ⎦

(1)

subject to

ctd+i + itd+ i

= rt +i kts+i + wt +i nts+i + π t +i

kts+ i +1

1 (1 − δ )kts+ i + f (kts+ i , nts+ i , At +i ) − ctd+i 1+ γ

=

[

]

β designates the intertemporal preference rate, δ the depreciation rate, π firms' profits and γ stands for the Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where superscripts d and s stand for "demand" and "supply",

stationarity parameter. Using standard dynamic programming techniques, this optimal d

s



planning problem can be solved to yield the solution sequence {ct + i , nt +i }i = 0 ; however, from d

s

each sequence only the first tupel (ct , nt ) is actually carried out. d

d

In period t = 0 , the firm decides upon its inputs (kt , nt ) given expected demand for its products Eyt related to its perceived demand curve. Standard (one-period) profit maximization yields the factor demand functions:

ktd

=

f k (rt , wt , At , Eyt )

ntd

=

f n (rt , wt , At , Eyt )

(2)

Pu Chen, Gang Gong, Armon Rezai et al.

188

As the capital market is supposed to be perfectly competitive, the rental rate of capital,

rt , adjusts in each period such as to clear the market: kt = kts = ktd . On the labor market, however, the fixed wage contract does usually not allow to clear the market.8 We presume nominal wage rigidity in the first period, such that actual employment does not correspond to labor supply for that period. In order to determine the matching process and actual transactions on the labor market, a matching function employing a rule for employment has to be defined.9 For the short side rule there is indeed a long tradition of macroeconomic modeling with specification of the non-clearing labor markets, see, for instance, Benassy (1995, 2002) and Malinvaud (1994). We want to allow labor transactions off the labor demand schedule. So the matching rule we want to use here can be described as:

nt = ωntd + (1 − ω )nts ,

(3)

where ω measures the degree to which employment is determined by labor demand and

(1 − ω ) by labor supply. In this context then, nt is the actual employment.10 Once the factor inputs have been determined through this matching process (3), defining the employment nt , the firm proceeds with deciding its output level. Note that the firm is constrained not only by a potentially non-cleared labor market but also by the prospects of a non-cleared product market demand, Eyt (recall that prices are fixed). Hence the firm will select the optimal capital stock11 to optimize the following program:

max ktd

yt − rt ktd − wt nt

s.t. yt = f ( At ktd , nt )′ yt ≤ yˆ t where yˆ , is the realization of Eyt in period t , yielding the output supply function

yts = f (kt , nt , At ) . Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

8

This may nevertheless happen if either the representative firm has perfect foresight on the sequence of technology shocks or the wage contract is done in the form of a contingency plan. Both will be excluded here; see Gong and Semmler (2007) on a discussion on this latter point. In an extension of our model we presume that the

wage is partially adjusted to some optimal wage, ω , but it is still very sticky, see Gong and Semmler (2007). 9 In disequilibrium literature the short side of the market is supposed to determine the outcome, formalized by the minimization rule: *

nt = min(ntd , nts ). However, such an assumption may be too restrictive, as employment may need time to adjust from one period to the other. 10 Note that we presume that there may be differential impacts of the labor supply or labor demand on the actual employment. So the matching outcome is not necessarily determined by a Nash bargaining process. 11 Notice that capital markets clear instantaneously and capital can be adjusted at no cost following an unfavorable realization of the demand shock.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

189

Yet, the main issue is, once employment and output have been determined, the household needs to re-optimize triggered by the difference between actual and planned employment levels, resulting from the above matching process (3). Given the realized factor, transactions (kt , nt ) the new optimal planning program is: ∞ ⎡ ⎤ d + E U ( c , n ) β iU (ctd+i , nts+i )⎥ max ⎢ ∑ t t ctd i =1 ⎣ ⎦

(4)

subject to

kts+ i kts+ i +1

=

1 [(1 − δ )kts + f (kt , nt , Ai ) − ctd ] 1+ γ 1 = [(1 − δ )kts+i + f (kts+ i , nts+ i , At +i ) − ctd+i ], i = 1,2.... 1+ γ

which can be used to derive the consumption demand based on realized transactions in the factor markets and the realization of the technology shock in period t . Next, we need to add certain specifications regarding the preference function, the technology shock and the stationarity of the time series data the model may generate. The household's instantaneous utility function over consumption, c , and leisure, l = 1 − n is:

U (c, n) = ln(c) + θln(1 − n) with θ the elasticity between consumption and leisure to be estimated with the data. Moreover, technological shocks are supposed to follow and AR(1) process:

At +1 = α 0 + α1 At + ε t whereε t : N (0, σ ε2 )

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The stationarity parameter,

γ , can be recovered by calculating the trend growth rate of

output. Finally, employment, nt , is based on (normalized) hours worked (sample mean N ). Further the specification of the model can be summarized as follows12 - The evolution of the -- stationarized -- capital stock

kt +1 =

[

1 (1 − δ )kt + At kt1−α (nt , N /0.3)α − ct 1+ γ

The technological evolution

]

Pu Chen, Gang Gong, Armon Rezai et al.

190

At +1 = α 0 + α1 At + ε t The production function

yt = At kt1−α (nt N /0.3)α Labor supply

n s = G11 At + G12 kt + g1 Labor demand

⎧ (0.3/ N )(eyt /At )1/α kt(α −1)/α if Eyt < (αAt Z t /wt )1/(1−α ) kt At n =⎨ 1/1−α ) kt (0.3/ N ) if Eyt ≥ (αAt Z t /wt )1/(1−α ) kt At ⎩(αAt Z t /wt ) d t

Actual employment

nt = ωntd + (1 − ω )nts Consumption decision

ct = G21 At + G22 kt + G23 nt + g 2 Expected production

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Eyt = yt −1 The essentials of our model are as follows. Households' demand for goods may be constrained by the firms' actual demand for labor. This way households also constrain the product market in buying less consumption goods than firms would like to be bought. The non-cleared labor market is derived from a multiple stage decision process of households facing constraints in the labor market, but firms are likely to be also constrained, namely constrained in the product market. Here is where the role of demand comes relevant. This additional component of our model gives rise to a further interaction of the labor market and the product market constraints, allowing for non-cleared markets. For further details see Gong and Semmler (2006, 2007). Overall, however, if firms face constraints on the product market due to insufficient demand, this may explain the technology puzzle, namely that positive technology shocks may have, only a weak effect on employment in the short run -- a phenomenon inconsistent with 12

For further details, see Gong and Semmler (2006).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

191

DGE models, where technology shocks and employment are predicted to be positively correlated. One might predict that such a model matches better time series data of advanced economies such as the US and the Euro-area (see Ernst et al. (2006)). If the economy works as above sketched, there is also an important role for aggregate demand. In the standard DGE model there are only technology shocks. They are the driving force of business cycles. Technology shocks are measured by the Solow residual. The Solow residual is computed on the basis of observed output, capital and employment, and it is presumed that all factors are fully utilized. There are several reasons to distrust the standard Solow residual as a measure of technology shock. First, Mankiw (1989) and Summers (1986) have argued that such a measure often leads to excessive volatility in productivity and even the possibility of technological regress, both of which seem to be empirically implausible. Second, it has been shown that the Solow residual can be expressed by some exogenous variables, for example demand shocks arising from military spending (Hall (1988)) and changed monetary aggregates (Evans (1992)), which are unlikely to be related to factor productivity. Third, the standard Solow residual can be contaminated if the cyclical variations in factor utilization are significant. Considering that the Solow residual cannot be used as a measure of technology shock, since it may embody non-technology shocks researchers have now developed different methods to measures technology shocks correctly. There are basically three strategies. The first strategy is to use an observed indicator to proxy for unobserved utilization. A typical example is to employ electricity use as a proxy for capacity utilization (see Burnside, Eichenbaum and Rebelo (1996)). Another strategy is to construct an economic model so that one could compute the factor utilization from the observed variables (see Basu and Kimball (1997) and Basu et al. (2006)). A third strategy uses an appropriate restriction in a VAR estimate to identify a technology shock (see Gali (1999) and Francis and Ramey (2004, 2005)). In the standard DGE model the above mentioned authors find that the technology shock in fact is negatively correlated with employment if one measures technology shocks by the corrected Solow residual.13 Though the standard DGE model predicts a significantly high positive correlation between technology and employment, most of the recent empirical research demonstrates, at least at business cycle frequency, a negative correlation. As our preliminary empirical evidence of section 2 shows while this seems to hold true in the short run, the nexus between productivity and employment may be different in the long run. One should, then, distinguish the short and long run effects for both productivity and demand shocks. Traditionally, only technology shocks have been seen to have persistent effects. In terms of the effects on output and employment in this view, demand shocks have only a short run effect output and employment but not affecting output and employment in the long run. On the other hand, productivity increases appear to have long run effect on output. A set up like this is also presumed in recently used VAR tests with supply and demand shocks. Blanchard and Quah (1987) for example, presume that supply shocks (productivity

13

See also Gong and Semmler (2006, chs. 5 and 9), and Basu et al. (2006).

192

Pu Chen, Gang Gong, Armon Rezai et al.

shocks) have permanent effects on output, but not employment. Demand shocks have, due to nominal rigidities, only a temporary effect on both output and employment.14 Yet a model as the one introduced above will predict that in the short run technology shocks may indeed have a negative effect on employment (positive effect on unemployment). This is predicted to occur when demand is constrained through the above described two stage decision making process. The productivity shock, however, may lead to a increasing employment (reduction of unemployment) in the long run and thus there may be a persistent positive productivity effect on employment in the long run.15

4. THE ECONOMETRIC METHODS AND RESULTS Next, we will employ two econometric methods to estimate the short and long-run effects of productivity growth on unemployment. First, we construct a model with short-run and long-run components to present the shot-run and long-run effect explicitly. Using maximum likelihood method we can estimate both the short-run and long run-effect as well as the frequency of the long run. Second, we employ the generally used technique of a structural vector autoregression (SVAR) with the long run restriction that non-technology shocks can not permanently increase productivity growth. The data used for these estimations are the purified technology growth and the unemployment rate from 1947 to 1996.

4.1. A Short-Run/Long-Run Components Model In order to gage the short and long run effects of productivity growth on unemployment, we assume that both series consist of a short run and a long run component:

z t = z tL + z tS , y t = y tL + y tS ,

(5)

where zt is the time series of the unemployment rates i.e. zt = unemplt and yt is the time

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

series of the purified technology growth i.e. yt = DTBFK t . Further we assume that the long-run component can be calculated as the moving average of the time series.

14

This position is also replicated in the study of monetary policy analysis, where it is usually assumed that monetary policy shocks only have temporary effects. However, following Blanchard (2005), Semmler, Zhang and Greiner (2005, ch. 6) show that monetary policy affects persistently, both, the real interest rate as well as the real activity and employment. 15 We want to note, however, that in this context then demand may or may not have effects on unemployment in the long run, but it may have a persistent effect on productivity in the long run. This may be pursued further in a framework suggest by Tobin (1993). We do not pursue this line of research, here, due to our emphasis on productivity and unemployment.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

z tL =

1 B 1 B L z , y = yt −s . ∑ t −s t B ∑ B s =1 s =1

193

(6)

Assuming that there exist the following long run and short run relations:

z tL = αy tL + u tL , z tS = βy tS + u tS ,

(7)

where ut : N (0, σ L ) and ut : N (0, σ s ) are independent normal distributed disturbances. 2

L

S

2

Inserting (7) into (5) we obtain

zt = αytL + βytS + utS + utL .

(8)

The density function for the model ((8)) is:

f (α , β , σ S2 , σ L2 ; zt , yt | B) ⎛ 1 exp⎜⎜ − zt − αytL − βytS 2 2 2 2 + 2( ) σ σ 2π (σ s + σ l ) S L ⎝

=

⎛ 1 exp⎜⎜ − zt − αytL − β ( yt − ytL ) 2 2 2 2 2π (σ s + σ l ) ⎝ 2(σ S + σ L )

=

⎛ 1 exp⎜⎜ − zt − βyt − (α − β ) ytL 2 2 2π (σ + σ ) ⎝ 2(σ S + σ L )

=

2 ⎛ 1 1 B ⎛ ⎞ ⎞⎟ ⎜ exp − ⎜ z − βyt − (α − β ) ∑ yt − s ⎟ ⎜ 2(σ S2 + σ L2 ) ⎝ t B s =1 ⎠ ⎟⎠ 2π (σ s2 + σ l2 ) ⎝

(

1



(

1

2 s

2

(

1

2 l

) ⎞⎟⎟ 2



(9)

) ⎞⎟⎟ 2



1

The economic hypotheses under investigation are: Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

) ⎞⎟⎟

=

α < 0 and β > 0 .

Remarks: •

• The long run short run model differs from a simple regression model zt = β yt in L

that it adds an "additional" explanatory variable yt into the regression equation. Hence the goodness of fit will better. It is easy to see that the existence of the L

estimate, unless yt is colinear with yt . •

• If

α = β , the short run and the long run relations are identical. The short run long

run model reduces to the simple regression model.

Pu Chen, Gang Gong, Armon Rezai et al.

194 •

L

• If B is given, yt can be calculated from the data and the short run long run model L

is equivalent to a model with two regressors: yt and yt . •

• If B is unknown, one can calculate the MLE for every possible B and choose B such that the likelihood function is maximized. Since for B = 1 we have the simple regression model and for B = T we have the simple regression model with a constant (if the constant has not been in the specification), there fore we will have an optimal estimate for B .



• The assumption of iid ut and ut are in general too restrictive. Relaxing this

S

L

assumption is necessary in order to be applicable for more general cases. A direct generalization of this assumption is that the disturbance may follow an AR(1) process: ut + ut = S

L

ρ (utS−1 + utL−1 ) + ε t

Applying the maximum estimation method as described above to model (8) extended by Remarks 4 and 5, we obtain Bˆ = 15 and the estimates summarized in the following table. Table 3. Estimation Output for Linear Model Variable Constant

ΔyL ΔyS RHO

Coeff. 0.0658 -2.3084

Std Error 0.0047 0.8384

T-Stat 13.75 -2.75

Signif. 0.00 0.01

0.3891

0.0870

4.46

0.00

0.7192

0.1304

5.51

0.00

It shows that all of the included variables are highly significant. An increase in long run

1% productivity growth over a period of 15 years appears to translate in a twice as high

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

reduction of unemployment rate. The effect of short run productivity growth appears to have a slightly positive effect on unemployment. These results confirm the previous preliminary examination in section 2. In the next section these results are analyzed in the SVAR framework in order to gain deeper insight in the relationship between productivity and unemployment.

4.2. VAR with Long-Run Restrictions Another way to gage the effects of productivity growth on unemployment is to use the VAR technique. Following the methodology by Gali (1999), we assume that there are technology and non-technology shocks. The assumption that non-technology shocks do not affect productivity growth is, then, used as a long run restriction.16 The model is similar to Gali's benchmark model, except we replace his measure of employment with the 16Generally, we could, however, presume that non-technology shocks, along the line of Kaldor (1957), have some effect on productivity but this will be captured by the productivity component.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

195

unemployment rate in order to be able to study effects of technology shocks - taken to mean random technology changes - on unemployment. Our approach is different from that of Blanchard and Quah (1989) who assumed supply disturbances do not affect unemployment in the long run. To study the effect of technology shocks, we use a structural VAR with long run restrictions. We presume that in the long run non-technology shocks cannot exert a permanent effect on productivity growth. If we go by Okun's law, then this long run effect on output should also translate into a long run effect on unemployment (see Khemraj et al. (2006)), although the Okun coefficient might be diminishing over time. However, we do not impose this restriction. The long run restriction a typical VAR is written in vector moving average form as given in the following equations: ∞



k =0

k =0





k =0

k =0

DTBFK t = ∑c11 (k )ε tT− k + ∑c12 (k )ε tNT −k unemplt = ∑c21 (k )ε tT− k + ∑c22 (k )ε tNT −k

ε tT and ε tNT are the technology and the non-technology shocks, respectively. If productivity (growth) is unaffected by non-technology shocks in the long run, it must be that the cumulative effect of such shocks must be equal to zero. That is





c (k ) = 0 . Using this

k = 0 12

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

restriction we are able to identify the structural VAR and study the effects of technology and non-technology shocks on unemployment. The likelihood ratio test suggests a lag length of 2 for the reduced form VAR, also the information criteria AIC, BSC and HQ confirm this lag length. Running a VAR with the above described restrictions, one can compute the impulse response functions (IRFs) using structural decomposition. Figures 4 depict these functions, the effects of the technology shocks on the unemployment. Unemployment is negatively affected by the technology shocks in the long run. Comparing this with the results in Basu et al. (2006) and Gali (1991), we confirm largely that the technology shock has a positive employment effect in the long run.

Figure 4. IRF of Unemployment to One S.D. of Technology Shock

196

Pu Chen, Gang Gong, Armon Rezai et al.

5. CONCLUSION To study the effects of productivity growth on unemployment we formulate a dynamic general equilibrium model in which the agents optimize their decision in two stages. The dynamic general equilibrium model predicts a positive effect of the productivity on the unemployment in the short run and a negative unemployment effect in the long. This reversed effect of the productivity growth on the unemployment can be identified in a diverse set of empirical data. To quantify the short run and long run effect of the productivity growth on unemployment we apply two econometric methods. In order to concentrate on the effect of technology growth on unemployment we use the purified technology growth data given by Basu et al. (2006). We develop a short run long run model to identify the short run and long relation between the technology growth and unemployment. The maximum likelihood estimation shows that the positive short run effect and the negative long run effect are highly significant. The ML estimation also helped us to draw the line between the two time horizons in an optimal way. We also apply the SVAR with long run restrictions that the nontechnology shocks do not have permanent effect on the productivity growth. Our result confirm largely the result in the literature: Basu et al. (2006) uses the methods of SVAR with long run restrictions and came to the result that the hours worked reacted to the technology shock negatively in the short run but positively in the long run. Gali (1999) used SVAR with long restrictions came to the result that the productivity growth had negative influence on the employment/hours worked. The initial negative effect is fully reversed over time. We, here, concur as our results show that the technology growth affects unemployment positively in the short run but negatively in the long run.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Basu, S., J. G. Fernald, and M. S. Kimball (2006). Are technology improvement contractionary? American Economic Review 96, 1418–1448. Basu, S. and M. S. Kimball (1997). Cyclical productivity with unobserved input variation. NBER Working Papers 5915, National Bureau of Economic Research. Benassy, J.-P. (1995). Money and wage contract in an optimizing model of the business cycle. Journal of Monetary Economics 35, 303–315. Benassy, J.-P. (2002). The Macroeconomics of Imperfect Competition and Nonclearing Markets. Cambridge, MA: MIT Press. Blanchard, O. (2005). The role of shocks and institutions in the rise of European unemployment: The aggregate evidence. In W. Semmler (Ed.), Monetary policy and Unemployment. The US, Euro-area, and Japan. Volume in Honor of James Tobin. Routledge. Blanchard, O. and D. Quah (1989). The dynamic effects of aggregate demand and supply disturbances. American Economic Review 79, 655–673. Burnside, C., M. Eichenbaum, and S. T. Rebelo (1996). Sectoral solow residual. European Economic Review 40, 861–869.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Short and Long Run Effects of Productivity Shocks in an Intertemporal Model

197

Ernst, E., G. Gong, W. Semmler, and L. Bukeviciute (2006, August). Quantifying the impact of structural reforms. Working Paper Series 666, European Central Bank. Evans, C. (1992). Productivity shock and real business cycles. Journal of Monetary Economics 29, 191–208. Francis, N. and V. A. Ramey (2004). The source of historical economic fluctuations: An analysis using long-run restrictions. NBER Working Papers 10631, National Bureau of Economic Research. Francis, N. and V. A. Ramey (2005). Is the technology-driven real business cycle hypothesis dead? Shocks and aggregate fluctuations revisited. Journal of Monetary Economics 52, 1379–1399. Gali, J. (1999). Technology, employment, and the business cycle: Do technology shocks explainaggregate fluctuations? American Economic Review 89, 249–271. Gali, J. and P. Rabanal (2005). Technology shocks and aggregate fluctuations: How well does the rbc model fit postwar u.s. data? IMF Working Papers 04/234, International Monetary Fund. Gong, G. and W. Semmler (2006). Stoachstic Dynamic Macroeconomics: Theory and Empirical Evidence. Oxford University Press. Gong, G. and W. Semmler (2007). Business cycles, wage stickiness and nonclearing labor market. CEM Working paper. Hall, R. E. (1988). The relation between price and marginal cost in U.S. industry. Journal of Political Economy 96, 921–47. Kaldor, N. (1957). A model of economic growth. Economic Journal 67, 591–624. Kaldor, N. (1985). Economics Without Equilibrium (The Arthur M. Okun memorial lectures). M.E. Sharpe. Khemraj, T., J. Madrick, and W. Semmler (2006). Okun’s law and jobless growth. Policy Note 3, Schwartz Center for Economic Policy Analysis. Malinvaud, E. (1994). Diagnosing Unemployment. Cambridge: Cambridge University Press. Mankiw, N. G. (1989). Real business cycles: A new Keynesian perspective. Journal of Economic Perspectives 3, 79–90. Okun, A. (1962). Potential GNP: Its measurement and significance. Proceedings of the Business and Economic Statistics Section, American Statistical Association. Rowthorn, R. (1999). Unemployment, wage bargaining and capital-labour substitution. Cambridge Journal of Economics 23 (4), 413–25. Semmler, W., W. Zhang, and A. Greiner (2005). Monetary and Fiscal Policy in the Euroarea, Macro Modelling, Learning and Empirics. Amsterdam: Elsevier. Solow, R. (1997). Learning from ”Learning by Doing”: Lessons for Economic Growth. Stanford Univerisity Press. Summers, L. H. (1986). Some skeptical observations on real business cycles theory. Federal Reserve Bank of Minneapolis Quarterly Review 10, 23–27. Tobin, J. (1993). Price flexibility and output stability: An old Keynesian view. Journal of Economic Perspectives 7 (1), 45–65. Uhlig, H. (2006, May). Discussion of ”the source of historical economic fluctuations: An analysis using long-run restrictions” by Neville Francis and Valerie A Ramey. SFB 649 Discussion Paper No. 2006-042.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 8

UNEMPLOYMENT INSURANCE: FACTORS ASSOCIATED WITH BENEFIT RECEIPT* United States Government Accountability Office

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

WHY GAO DID THIS STUDY Unemployment Insurance (UI), established in 1935, is a complex system of 53 state programs that in fiscal year 2004 provided $41.3 billion in temporary cash benefits to 8.8 million eligible workers who had become unemployed through no fault of their own. Given the size of the UI program, its importance in helping workers meet their needs when they are unemployed, and the little information available on what factors lead eligible workers to receive benefits over time, GAO was asked to determine (1) the extent to which an individual worker’s characteristics, including past UI benefit receipt, are associated with the likelihood of UI benefit receipt or unemployment duration, and (2) whether an unemployed worker’s industry is associated with the likelihood of UI benefit receipt and unemployment duration. Using data from a nationally representative sample of workers born between 1957 and 1964 and spanning the years 1979 through 2002, and information on state UI eligibility rules, GAO used multivariate statistical techniques to identify the key factors associated with UI benefit receipt and unemployment duration. In its comments, the Department of Labor stated that while there are certain qualifications of our findings, the agency applauds our efforts and said that this report adds to our current knowledge of the UI program.

WHAT GAO FOUND Certain characteristics are associated with the likelihood of receiving UI benefits and unemployment duration. UI-eligible workers that GAO studied are more likely to receive UI *

Excerpted from CRS Report GAO-06-341, dated March 2006.

United States Government Accountability Office

200

benefits if they have higher earnings prior to becoming unemployed, are younger, have more years of education, or if they have a history of past UI benefit receipt when compared with otherwise similar workers. GAO found that past experience with the UI program has a particularly strong effect on the future likelihood of receiving UI benefits. However, some characteristics, such as receiving a higher maximum weekly UI benefit amount, are not associated with a greater likelihood of receiving UI benefits. UI-eligible workers who receive UI benefits have longer unemployment duration than workers with similar characteristics. Also, UI-eligible workers are more likely to experience longer unemployment duration if they have lower earnings before becoming unemployed or have fewer years of education. Other characteristics associated with longer unemployment duration include being AfricanAmerican, female, or not belonging to a union. GAO found no relationship between past UI benefit receipt and subsequent unemployment duration. UI-eligible workers from certain industries are more likely than similar workers in other industries to receive UI benefits and experience shorter unemployment duration. Specifically, GAO’s simulations show that the likelihood of receiving UI benefits during a first period of unemployment is highest among workers from the mining and manufacturing industries. Furthermore, the likelihood of receiving UI benefits when unemployed increases with each previous period of UI receipt across all industries, and the most notable increase occurs in public administration. First-time unemployed workers from construction and manufacturing experience significantly shorter unemployment duration than workers from other industries.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Source: GAO simulations based on GAO analysis of NLSY79 data. Figure. Simulated UI Benefit Receipt Rates for UI-Eligible Workers during Successive Periods of Unemployment, by Past UI Receipt Status.

ABBREVIATIONS BLS BPE CPI-U CPS HQE NLSY79

Bureau of Labor Statistics base period earnings Consumer Price Index for All Urban Consumers Current Population Survey high quarter earnings National Longitudinal Survey of Youth 1979

Unemployment Insurance: Factors Associated with Benefit Receipt

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

OLF SIC SMSA SOC UI WBA

201

out of the labor force Standard Industrial Classification Standard Metropolitan Statistical Area Standard Occupational Classification Unemployment Insurance weekly benefit amount

Unemployment Insurance (UI), established in 1935, is a complex system of 53 programs that provide temporary cash benefits to eligible workers who become unemployed through no fault of their own.[1] Eligibility for UI benefits, benefit amounts, and the length of time benefits are available are determined by state law, within broad federal guidelines. Benefits are financed through federal and state employer payroll taxes. In fiscal year 2004, employers paid about $39.3 billion in UI taxes, and 8.8 million workers received UI benefits totaling $41.3 billion. Decades of program experience and administrative data have resulted in a firm understanding of the composition of UI caseloads and the overall cost of the program. However, this understanding of the UI program has been based on snapshots of the UI beneficiary population at any given time. Additional research has provided limited information on the types of workers who are likely to receive UI benefits and on how UI requirements and benefits affect individuals’ movement into and out of the workforce, including how UI receipt affects the duration of unemployment. However, because of the difficulty of tracking the same workers over time, the circumstances that give rise to individual workers’ use or nonuse of the UI program and how this may, in turn, affect individuals’ patterns of unemployment over the course of their entire working careers are still not well understood. In 2005, we reported on the results of our analysis of a unique database that tracked a single group of individuals over time.[2] Examining this database, we found that 85 percent of a nationally representative sample of late baby boom workers (workers born between 1957 and 1964) had experienced unemployment at least once between 1979 and 2002. Workers who experienced unemployment were unemployed an average of five times over this 23-year period. Moreover, we found that of those who were eligible for UI benefits at least once, only 38 percent at some point received UI benefits. About half of the workers receiving UI benefits received them more than once. Finally, we reported that the rate at which unemployed workers received UI benefits varied across industries. As Congress reviews the ability of labor programs to meet the needs of the workforce in the new century, it will be important to understand why fewer than half of workers eligible for UI benefits receive them and the other half do not, as well as what factors cause workers in some industries to seek benefits multiple times over the course of their careers. In this context, you asked us to determine (1) the extent to which characteristics of individual workers, including a history of past UI benefit receipt, are associated with the likelihood of UI benefit receipt and unemployment duration, and (2) whether an unemployed worker’s industry is associated with UI benefit receipt or unemployment duration. To answer these questions, we analyzed data from the National Longitudinal Survey of Youth 1979 (NLSY79). This survey provides information that is not typically available from other data sources. The dataset contains information from ongoing periodic interviews with a nationally representative sample of individuals who were born between 1957 and 1964. At the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

202

United States Government Accountability Office

time of our analysis, the database contained information from interviews conducted between 1979 and 2002. There were 12,686 individuals in the sample in 1979. The survey provides a wide range of detailed information about these individuals, including their work histories, income, family composition, and education. Using the dataset, we analyzed a single birth cohort over time; therefore, our findings do not represent the experience of workers of all ages during this time period. Using this survey information and information on states’ UI program eligibility rules for each year from 1978 through 2002,[3] we estimated whether individuals from the sample were eligible for UI benefits following a job separation. We identified 5,631 workers who met the conditions for UI eligibility—a group that we refer to as “UI-eligible workers”—who collectively experienced 15,506 separate periods of unemployment during the study period (1979-2002). We used a multivariate statistical model to identify the key factors associated with UI benefit receipt and unemployment duration for our subsample of UI-eligible workers. The model allowed us to isolate the effect of a particular characteristic by statistically controlling for a number of other characteristics. In this report, we refer to the results for individual characteristics in comparison with “otherwise similar workers.” By this phrasing, we intend to show that we have controlled for all other characteristics that may be related to the characteristic being studied. For example, the test of the effect of age on benefit receipt was conducted while controlling, for example, for earnings and education—two characteristics that are correlated with age. In addition, we modeled UI benefit receipt and unemployment duration together to control for the likely correlation that exists between these two outcomes. To illustrate how changes in different characteristics affect the likelihood of UI receipt and unemployment durations, we used the results of our multivariate statistical model to simulate how changes in observable characteristics affect the likelihood of UI receipt and unemployment duration. The simulated results are calculated from our statistical model estimates, holding selected characteristics constant, as noted throughout the report. For example, to understand how changes in workers’ education affect their likelihood of receiving UI benefits, we set the number of years of education at the same value for all workers in our sample and then used the model estimates to simulate the likelihood of UI receipt for each worker. We then calculated the average likelihood of receiving UI benefits. We repeated this process for different years of education. Unless otherwise noted, simulated likelihoods of UI receipt and simulated unemployment duration are for workers experiencing unemployment for the first time. See appendix I for a more complete discussion of our methodology, including limitations of our analysis. We assessed the reliability of the NLSY79 dataset and found it to be sufficient for our analysis. Our work was conducted from May 2005 through February 2006 in accordance with generally accepted government auditing standards.

RESULTS IN BRIEF Certain characteristics are associated with the likelihood of receiving UI benefits and unemployment duration. Based on our analysis of workers during the first half of their working lives, UI-eligible workers are more likely than other workers to receive UI benefits if

Unemployment Insurance: Factors Associated with Benefit Receipt

203

they have higher earnings, are younger or have more years of education, or, most notably, if they received UI benefits in the past. In particular, UI-eligible workers who received UI benefits before are more likely than other workers to receive UI benefits again and this likelihood increases each time they are unemployed and receive UI. Other factors, including a high local unemployment rate, increase the likelihood of receiving UI. UI-eligible workers who receive UI benefits have longer periods of unemployment than workers who do not receive benefits. Similarly, workers who have fewer years of education, lower earnings, or no union membership experience longer unemployment than workers who do not have these characteristics. Workers who received UI benefits in the past, however, were unemployed about as long as similar workers who had not received UI in the past. UI-eligible workers from certain industries are more likely than other workers to receive UI benefits and experience shorter unemployment duration, although no clear industry trend emerged. Specifically, our simulations show that •



The likelihood of receiving UI benefits during a first period of unemployment is highest among workers from mining and manufacturing. Furthermore, the likelihood of receiving UI benefits when unemployed increases with each previous period of UI receipt across all industries, and the most notable increase occurs for workers from the public administration sector. The unemployment duration for first-time unemployed workers from construction and manufacturing is significantly shorter than the unemployment duration experienced by workers from other industries. While unemployment duration varies across all industries, this variation is not affected by whether workers were unemployed in the past, or whether they received UI in the past.

In its comments, the Department of Labor stated that, while there are certain qualifications of our findings, Labor applauds our efforts and said that this report adds to our current knowledge of the UI program. Labor also provided technical comments, which we incorporated where appropriate.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

BACKGROUND The UI program was established in 1935 and serves two primary objectives: (1) to temporarily replace a portion of earnings for workers who become unemployed through no fault of their own and (2) to help stabilize the economy during recessions by providing an infusion of consumer dollars into the economy. UI is made up of 53 state-administered programs that are subject to broad federal guidelines and oversight. In fiscal year 2004, these programs covered about 129 million wage and salary workers and paid benefits totaling $41.3 billion to about 8.8 million workers. Federal law provides minimum guidelines for state programs and authorizes grants to states for program administration. States design their own programs, within the guidelines of federal law, and determine key elements of these programs, including who is eligible to receive state UI benefits, how much they receive, and the amount of taxes that employers must pay to help provide these benefits. State unemployment tax revenues are held in trust by

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

204

United States Government Accountability Office

the Department of Labor (Labor) and are used by the states to pay for regular weekly UI benefits, which typically can be received for up to 26 weeks. During periods of high unemployment, the Extended Benefits program, funded jointly by states through their UI trust funds and by the federal government through the Unemployment Trust Fund, provides up to 13 additional weeks of benefits for those who qualify under state program rules. Additional benefits, funded by the federal government, may be available to eligible workers affected by a declared major disaster or during other times authorized by Congress. To receive UI benefits, an unemployed worker must file a claim and satisfy the eligibility requirements of the state in which the worker’s wages were paid. Although states’ UI eligibility requirements vary, generally they can be classified as monetary and nonmonetary. Monetary eligibility requirements include having a minimum amount of wages and employment over a defined base period, typically, about a year before becoming unemployed, and not having already exhausted the maximum amount of benefits or benefit weeks to which they would be entitled because of other recent unemployment. In addition to meeting states’ monetary eligibility requirements, workers must satisfy their states’ nonmonetary eligibility requirements. Nonmonetary eligibility requirements include being able to work, being available for work, and becoming unemployed for reasons other than quitting a job or being fired for work-related misconduct. In all states, claimants who are determined to be ineligible for benefits are entitled to an explanation for the denial of benefits and an opportunity to appeal the determination. Since UI was introduced, researchers and those responsible for overseeing the program have monitored the size, cost, and structure of the program and its effects on individuals’ movement into and out of the workforce, including which types of workers receive UI benefits. Much of what is known about the dynamics of the UI program has been based on snapshots of the UI beneficiary population at any given time. Labor regularly gathers UI program data from the states, including each state’s eligibility requirements, employers’ UI tax rates, program revenues and costs, and numbers of claims received and approved. An extensive amount of research has been devoted to the effect of UI benefit receipt on unemployment duration. Specifically, researchers have found that receiving UI benefits increases unemployment duration. Much of this research is focused on measuring how changes in the amount of UI benefits increase the amount of time that an unemployed worker takes to find a new job.[4] Although much is known about UI caseloads and about the relationship between UI benefits and unemployment duration, less is known about the patterns of UI receipt among individual workers over their entire working careers. What is known about the patterns of UI benefit receipt over an extended period for individual workers comes from a few studies that are fairly limited in scope. In one study, researchers analyzed 1980-1982 survey data and found that among unemployed workers who were eligible for UI, younger or female workers were less likely to receive UI benefits, while union workers, workers from large families, or those with more hours of work from their previous jobs were more likely to receive UI.[5] In another study, using UI administrative data from five states, researchers found that between 36 and 44 percent of UI claims from 1979 to 1984 were from workers who had received UI benefits more than once and that middle-aged workers and workers with higher earnings were more likely to be repeat UI recipients.[6] Another study, based on survey data from the NLSY79, found that 16 percent of young adults had received UI benefits more than once between 1978 and 1991 and that as many as 46 percent of those who received UI were repeat recipients.[7] This study also found

Unemployment Insurance: Factors Associated with Benefit Receipt

205

that workers who were women or Hispanic or whose fathers had more years of education were less likely to become repeat recipients than workers who were men or non-Hispanic or whose fathers had fewer years of education. In 2005, we analyzed the NLSY79 to determine the extent to which individual workers received UI benefits during their early working lives.[8] We found that 38 percent of workers born between 1957 and 1964 received UI at least once between 1979 and 2002, with almost half of these individuals receiving UI benefits more than once. (See figure 1.) We also found that the rate at which unemployed workers received UI benefits varied across industries, but we did not control for any of the other factors that may have helped to explain this variation.

Source: GAO analysis of NLSY79 data. Note: Sampling errors were within plus or minus 5 percentage points at the 95 percent confidence level. Figure 1. Incidence of UI benefit receipt from 1979 through 2002, for workers born between 1957 and 1964.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

CERTAIN CHARACTERISTICS ARE ASSOCIATED WITH UI BENEFIT RECEIPT AND UNEMPLOYMENT DURATION Earnings, age, education, and most notably past UI benefit receipt are all associated with the likelihood of receiving UI benefits for UI-eligible workers. Education, earnings, and union membership, and current UI benefit receipt, are associated with unemployment duration.

Unemployed Workers with Higher Earnings, Younger Workers, Workers with More Education, or Those Who Received UI in the Past Are More Likely to Receive UI Benefits Unemployed workers are more likely to receive UI benefits if they have higher earnings prior to becoming unemployed, are younger or have more years of education, or have a history of past UI benefit receipt, when compared to workers with similar characteristics.[9] We found that past experience with the UI program has a particularly strong effect on the future likelihood of receiving UI benefits. In addition, UI-eligible workers are more likely to receive UI when the local unemployment rate is high. However, some characteristics, such as

206

United States Government Accountability Office

the weekly UI benefit amount that a worker is eligible to receive, are not associated with a greater likelihood of receiving UI benefits.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployed Workers Who Have Higher Earnings or Are Younger or Have More Years of Education Are More Likely to Receive UI Unemployed workers who have higher earnings or are younger or who are more educated are more likely to receive UI benefits than otherwise similar workers. With respect to earnings,[10] our simulations show that the likelihood of receiving UI tends to increase as the amount earned in the year prior to becoming unemployed increases (see figure 2). For example, a UI-eligible worker with earnings between $10,000 and $11,999 in the year before becoming unemployed has a 36 percent likelihood of receiving UI, whereas a worker who earned roughly twice as much (between $20,000 and $24,999) has a 45 percent likelihood of receiving UI.[11] The likelihood of receiving UI is lowest among workers with the lowest earnings (i.e., less than $10,000 in the year before becoming unemployed). There is generally little difference in the likelihood of receiving UI among workers earning $18,000 or more. Concerning age, our simulations show that the likelihood of receiving UI peaks at about age 25 and decreases thereafter (see figure 3). More specifically, a 25-year-old unemployed worker who is eligible for UI is more than twice as likely to receive UI as an otherwise similar 40-year-old unemployed worker.

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are for the average likelihood of receiving UI during first-time unemployment at different levels of earnings. The overall average likelihood of receiving UI during first-time unemployment is 33 percent. See appendix I for methodology and estimation results. Figure 2. Simulated likelihood of receiving UI benefits for UI-eligible workers, by prior-year earnings.

This result confirms our 2000 finding that low-wage workers are less likely to receive UI benefits than workers with higher earnings even when they have worked for the same amount of time.[12] Our current result also controls for other worker differences, such as which

Unemployment Insurance: Factors Associated with Benefit Receipt

207

industries the workers were employed in or whether they were ineligible for benefits, which we had not previously been able to rule out as explanations for the variation in likelihood of receiving UI. The relationship between higher earnings and a higher likelihood of receiving UI benefits is also consistent with economic theory that predicts that workers with higher earnings prior to becoming unemployed will be more reluctant to accept lower reemployment wages and are therefore more likely to take advantage of UI benefits as a way to subsidize their job search efforts.[13]

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the average likelihood of receiving UI during first-time unemployment at different ages. The overall average likelihood of receiving UI during first-time unemployment is 33 percent. See appendix I for methodology and estimation results.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 3. Simulated likelihood of receiving UI renefits for UI-eligible workers, by age.

Previous studies have found that younger workers are less likely to receive UI benefits than older workers.[14] However, these previous studies did not include as much information about workers’ past unemployment and UI benefit receipt histories as our current analysis. Therefore, because older workers have more of this experience than younger workers, it is possible that our analysis has controlled for the effect of this past experience more completely than these previous studies, resulting in a more precise estimate of the effect of age. We are unable to explain why younger workers are more likely to receive UI benefits than otherwise similar older workers. However, it is possible that older workers, who have had more time to accumulate financial assets, may have more private resources available to help them cope with unemployment than younger workers.[15] Alternatively, younger workers may be less optimistic about how long it will take for them to become reemployed. Unemployed workers with more years of education are more likely to receive UI benefits than otherwise similar workers with fewer years of education. Specifically, our simulations show that the likelihood of receiving UI increases for each additional year of schooling that a UI-eligible worker has completed before becoming unemployed (see figure 4). For example, a UI-eligible worker with a college education (one who has completed 16 years of schooling)

United States Government Accountability Office

208

when he or she becomes unemployed is almost one-fifth more likely to receive UI than a UIeligible worker with a high school education (12 years of schooling).[16]

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the average likelihood of receiving UI during first-time unemployment at different education levels. The overall average likelihood of receiving UI during first-time unemployment is 33 percent. See appendix I for methodology and estimation results.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 4. Simulated likelihood of receiving UI benefits for UI-eligible workers, by education level.

Although the impact of education on the likelihood of receiving UI benefits has been analyzed in other research, this research found no significant education effect [17]. However, to the extent that workers with more years of education are better able to access and understand UI program rules, they may also be more likely to know when they are entitled to benefits and to have the information that they need to file successful benefit claims. Other factors, including gender, marital status, job tenure, and the local unemployment rate are also associated with UI benefit receipt. Controlling for all other characteristics among this UI-eligible group, • • •



a woman is 29 percent more likely to receive UI benefits than a man, a married worker is 13 percent more likely to receive UI than an unmarried worker, a longer tenured worker is more likely to receive UI—for example, a worker with 4 years of tenure at his or her most recent job is 12 percent more likely to receive UI than a worker with 1 year of job tenure, and being in an area with high unemployment raises the likelihood that an unemployed worker will receive UI—for example, a worker living in an area with an unemployment rate of 9 percent is 10 percent more likely to receive UI than a worker living in an area with an unemployment rate of 5 percent.

Unemployment Insurance: Factors Associated with Benefit Receipt

209

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Our finding that women are more likely to receive UI benefits than otherwise similar men differs from the results of previous research, which generally found no statistically significant differences. Nevertheless, our analysis controls for more worker characteristics than these previous studies, and it is likely that we have more carefully isolated the effect of gender from that of other characteristics related to gender, such as workers’ occupations or industries. It is not immediately clear why women are more likely to receive UI benefits, however. We are likewise unable to explain why married workers are more likely to receive UI benefits than otherwise similar unmarried workers.[18] Our findings on job tenure are consistent with previous research. However, the higher likelihood of UI benefit receipt associated with more years of job tenure is likewise difficult to explain. It might be that workers with longer job tenures have acquired more skills that are not as easy to transfer to another employer, relative to workers with less job tenure, and anticipate longer job searches. The higher likelihood of receiving UI benefits among workers living in areas with higher unemployment is likely due to the higher number of unemployed workers relative to available jobs, which may make workers more willing to apply for UI benefits as they engage in what are likely to be longer job searches. In contrast to our findings above, a key UI program element, the weekly UI benefit amount that UI-eligible workers are entitled to, is not associated with a greater likelihood of receiving UI benefits. Using our model estimates, we simulated increases in weekly UI benefit amounts of 10 percent and 25 percent and a decrease of 10 percent and found that these changes had no effect on the likelihood of UI benefit receipt. This finding is consistent with the work of others, who have found that increases in the weekly benefit amount have mixed, but generally small effects on UI benefit receipt.[19] Collectively, these results suggest that UI benefit levels have modest effects on individuals’ decisions about whether or not to receive UI benefits, after controlling for other factors.

Unemployed Workers Who Received UI in the Past Are More Likely to Receive UI During Subsequent Unemployment Unemployed workers who have received UI benefits during a prior period of unemployment are more likely to receive UI benefits during a current period of unemployment than otherwise similar workers who never received UI benefits (see figure 5). For example, when workers experience their first UI-eligible period of unemployment, their likelihood of receiving UI is 33 percent. During a second UI-eligible period of unemployment, the likelihood of receiving UI is 48 percent for workers who received UI during the first unemployment period but only 30 percent for workers who did not receive UI. Furthermore, the likelihood that these UI-eligible workers will receive UI benefits during successive periods of unemployment increases each time that they receive UI benefits and decreases each time that they do not.[20] This finding suggests that a worker’s first unemployment experience has a lasting and self-reinforcing effect. To the extent that workers know about the UI program and whether or not they are eligible, receiving or not receiving UI benefits may be a personal choice based on unobserved worker characteristics or preferences. Alternatively, if workers do not have good information about UI, those who receive UI benefits may know more about the UI program than those who do not receive UI, and their knowledge about the program could make it easier to apply for and receive benefits during a subsequent period of unemployment.

210

United States Government Accountability Office

Source: GAO simulations based on GAO analysis of NLSY79 data. Note: Simulations are the average likelihood of receiving UI during a current unemployment period for two extreme cases: (1) workers who always received UI benefits during previous unemployment and (2) workers who never received UI during previous unemployment. The average likelihood of receiving UI during first-time unemployment for all UI-eligible workers is 33 percent. See appendix I for methodology and estimation results. Figure 5. Simulated likelihood of receiving UI benefits for UI-eligible workers during successive periods of unemployment, by past UI receipt status.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Receiving UI Benefits, Along with Other Factors, Is Associated with Unemployment Duration Overall, unemployed workers who receive UI benefits have longer unemployment duration than otherwise similar workers who do not receive UI benefits.[21] Several other characteristics are also associated with unemployment duration. Specifically, UI-eligible workers are more likely to experience longer unemployment duration if they have lower earnings before becoming unemployed or have fewer years of education. Other characteristics associated with longer unemployment duration, after controlling for other factors, include being African-American or female or not belonging to a union. We found no relationship between past UI benefit receipt and subsequent unemployment duration.

Receiving UI Benefits Is Associated with Longer Unemployment Duration Whether or not an unemployed worker receives UI during a specific period of unemployment has the strongest effect on how long that period of unemployment is likely to last. Overall, UI-eligible workers who receive UI benefits during a period of unemployment remain unemployed for about 21 weeks on average, whereas otherwise similar workers who do not receive UI remain unemployed for about 8 weeks. This result is consistent with economic theory that predicts that receiving UI benefits reduces the costs associated with unemployment and allows workers to engage in longer job searches.[22] That is, an

Unemployment Insurance: Factors Associated with Benefit Receipt

211

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

unemployed worker who receives UI benefits faces less pressure to accept the first job offer they receive and can search longer for a more desirable job than an unemployed worker who does not receive UI. Another possible explanation for the strong association between UI receipt and longer unemployment duration may be that workers who expect to experience longer unemployment may be more likely to apply for UI than those who expect to return to work quickly.

Unemployed Workers with Lower Earnings and Less Education Tend to Have Longer Unemployment Duration Unemployed workers with lower earnings tend to have longer unemployment duration than otherwise similar workers with higher earnings. This finding holds for workers who are receiving UI benefits, and for workers who are not receiving UI benefits. Specifically, our simulations show that UI-eligible workers who receive UI benefits and have relatively high earnings ($30,000 and higher) in the year prior to becoming unemployed have unemployment duration that is as much as 9 weeks shorter than workers with earnings that are below $16,000.[23] The results are similar for UI-eligible workers who do not receive UI benefits (see figure 6). Our result is consistent with other research that has found that higher previous earnings tend to reduce unemployment duration.[24] Researchers have suggested that the association between higher earnings and shorter unemployment duration may be due, in part, to the higher cost of unemployment for workers with higher earnings, relative to the cost for workers with lower earnings.[25] Specifically, the cost of unemployment in terms of lost wages is greater for workers with higher earnings, because they forego a higher amount of potential earnings in exchange for the time they spend on unpaid activities, such as job search, home improvement, or recreation. Our model estimates also indicate that unemployed workers who have more education tend to have shorter unemployment duration than otherwise similar workers with less education. For example, simulations show that on average, UI-eligible workers with a 4-year college education (16 years of schooling) who receive UI benefits remain unemployed about 2 weeks less than workers with a high school education (12 years of schooling).[26] (See figure 7.) The results are similar for UI-eligible workers who do not receive UI benefits. This finding is consistent with past research indicating that less education is associated with longer unemployment duration, because workers with less education have fewer work-related skills.[27] Unemployed workers’ race or ethnicity, gender, union membership status, and length of most recent job tenure are also associated with unemployment duration. Specifically, simulations show that UI-eligible workers who are African-American or women, who do not belong to labor unions, or who have less years of job tenure before becoming unemployed tend to have longer unemployment duration than otherwise similar workers. As seen in table 1, these associations exist whether or not workers receive UI benefits. Our findings are generally consistent with prior research. In particular, longer unemployment durations have been found to be associated with being African-American, female, or not belonging to a union.[28] Two possible explanations for the differences in employment outcomes for African-American workers include labor market discrimination, and limited access to social networks that may enable these workers to find jobs more quickly.[29] Likewise, longer unemployment duration among female workers may be due to

212

United States Government Accountability Office

labor market discrimination, or to differences in how they value paid work versus nonemployment activities, relative to men.[30] Likewise, the associations between shorter unemployment duration and union membership or longer job tenure may reflect the greater access of these workers to reemployment opportunities than otherwise similar workers or because of a greater likelihood of being recalled to their previous jobs.[31]

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UIeligible workers not receiving UI benefits. See appendix I for methodology and estimation results.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 6. Simulated unemployment duration for UI-eligible workers, by prior-year earnings and UI receipt status.

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UI-eligible workers not receiving UI benefits. See appendix I for methodology and estimation results.

Figure 7. Simulated unemployment duration for UI eligible workers, by education level and UI receipt status.

Unemployment Insurance: Factors Associated with Benefit Receipt

213

Table 1. Simulated Unemployment Duration for UI-Eligible Workers by Current UI Receipt Status and Other Characteristics Worker characteristics

Race or ethnicity White Hispanic African-American Gender Male Female Union membership status Union member Not a union member Tenure at most recent joba 10 years 1 year Overall average duration

Unemployment duration (median weeks) Receiving UI benefits Not receiving UI benefits 19 21 25

8 8 11

20 22

8 9

19 21

8 9

20 21 21

8 8 8

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. See appendix I for methodology and estimation results. a Simulated decreases in median weeks of unemployment are less than 1 week per additional year of tenure at most recent job, regardless of whether workers received UI or not.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployment Duration Is Not Associated with Past UI Receipt Past UI receipt has no significant effect on subsequent unemployment duration. Although receiving UI during a current period of unemployment is associated with longer unemployment duration, past UI receipt does not affect current unemployment duration. Specifically, simulations show that unemployment duration tends to decrease by about the same amount (typically, 1 week or less) from one unemployment period to the next, regardless of whether a worker received UI benefits in the past or not, and regardless of whether or not the worker receives UI benefits in the current period.

CERTAIN INDUSTRIES ARE ASSOCIATED WITH UI BENEFIT RECEIPT AND UNEMPLOYMENT DURATION Unemployed workers in certain industries are more likely to receive UI benefits and experience shorter unemployment duration than otherwise similar workers from other industries. Simulations show that first-time unemployed workers from mining and manufacturing are more likely to receive UI than workers from other industries. Moreover, the strength of the association between past and current UI benefit receipt varies across industries. The increase in the likelihood of receiving UI from one unemployment period to the next is highest for public administration and is lowest for agriculture and construction. Furthermore, simulations indicate that UI-eligible workers from industries with higher proportions of unemployment periods that result in UI receipt are no more likely to become

214

United States Government Accountability Office

repeat UI recipients than workers from other industries. With respect to unemployment duration, UI-eligible workers from construction and manufacturing have shorter unemployment duration than workers from other industries. .

Unemployed Workers from Mining and Manufacturing Are More Likely to Receive UI Benefits Unemployed workers from mining and manufacturing are more likely to receive UI than otherwise similar workers from other industries. For example, first-time unemployed workers from the manufacturing industry are about two-thirds more likely to receive UI benefits than workers from the professional and related services industry (see table 2). Although UIeligible workers from mining are more likely to receive UI benefits than workers from other industries, just 2 percent of the unemployment periods that result in UI benefit receipt come from the mining industry. (See figure 8.)[32]

The Relationship Between Past and Current UI Receipt Is Strongest for Public Administration Unemployed workers who have received UI benefits in the past are more likely to receive UI benefits during a current period of unemployment than otherwise similar workers who never received UI benefits, across each industry (see table 3). However, the increase in the

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 2. Simulated Likelihood of Receiving UI Benefits for UI-Eligible Workers from Different Industries Industry Mining Manufacturing Public administration Wholesale and retail trade Agriculture, forestry, and fishing Business services Construction Finance, insurance, and real estate Industry Transportation and public utilities Entertainment and recreation services Professional and related services Personal services All industries

Simulated likelihood of receiving UI benefits (percent) 46 40 37 35 34 31 31 31 Simulated likelihood of receiving UI benefits (percent) 29 26 24 23 33

Note: Simulations are the average likelihood of receiving UI during first-time unemployment for workers from different industries. The parameter estimates for the mining, manufacturing, public administration, wholesale and retail trade, agriculture, forestry, and fishing, business services, and construction industries are statistically significant relative to the professional and related services industry at the 95 percent confidence level. See appendix I for methodology and estimation results. Source: Simulations based upon GAO analysis of NLSY79 data.

Unemployment Insurance: Factors Associated with Benefit Receipt

215

likelihood of receiving UI benefits associated with past UI benefit receipt is not the same across all industries. Specifically, this effect is strongest for workers from public administration and is weakest for workers from agriculture and construction.[33]

Source: GAO Analysis of NLSY79 data. Note: Total does not equal 100 percent due to rounding.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 8. Distribution of all periods of UI benefit receipt across industries.

These results show that although UI-eligible workers in some industries are more likely to receive UI benefits when they experience unemployment for the first time, their likelihood of receiving UI benefits again when they become unemployed a second or third time is not necessarily higher than it is for workers from other industries. For example, the likelihood of receiving UI benefits for workers from the manufacturing industry who are unemployed for the first time is relatively high—about 40 percent. This likelihood increases to 52 percent during a second period of unemployment for workers who have already received UI benefits, and to 65 percent during a third period of unemployment for workers who received UI each time they were unemployed. By comparison, the increase in the likelihood of receiving UI between the first and third periods of unemployment is higher for most other industries, especially public administration. Specifically, the likelihood of receiving UI benefits for public administration workers who are unemployed for the first time is 37 percent. This likelihood increases to 69 percent during a second period of unemployment for workers who have already received UI, and to 92 percent during a third period of unemployment for workers who received UI each time they were unemployed. (See figure 9.) Administrative unemployment insurance data have shown that repeat UI recipients tend to be from industries that are more seasonal, such as manufacturing and construction. Our results, however, suggest that this is not because workers with past UI receipt from these industries are more likely to receive UI benefits when they become unemployed than otherwise similar workers from other industries. Rather, it may be that workers from such seasonal industries are unemployed more often on average than workers from other industries, or that a larger proportion of unemployed workers from such industries have collected UI previously.

216

United States Government Accountability Office

Table 3. Simulated Likelihood of Receiving UI Benefits During Different Periods of UI-Eligible Unemployment for Workers with Past UI Receipt, by Industry Industry

Mining Manufacturing Public administration Wholesale and retail trade Agriculture, forestry, and fishing Business services Construction Finance, insurance, real estate Transportation and public utilities Entertainment and recreation services Professional and related services Personal services All industries

Simulated likelihood of receiving UI benefits during current UI-eligible unemployment period, given past UI receipt (percent) First Second Third Unemployment Unemployment Unemployment Perioda Period Perioda 46 57 69 40 52 65 37 68 91 35 52 70 34 42 50 31 48 66 31 40 51 31 64 91 29 46 66 26 45 67 24 39 58 23 38 56 33 48 64

Note: Simulations are the average likelihood of receiving UI during a first unemployment period, a second unemployment period with UI receipt during the prior unemployment period, and a third unemployment period with UI receipt during both prior unemployment periods. The positive effect that each prior UI receipt period has on the likelihood of current UI receipt is statistically significantly larger for the public administration industry relative to the professional and related services industry at the 95 percent confidence level, and smaller for the agriculture and construction industries. The simulations also incorporate the industry effects and the industry interactions with the number of prior periods of unemployment. See appendix I for methodology and estimation results. a Workers experiencing their first period of unemployment did not have past UI receipt. Source: Simulations based upon GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployed Workers from Construction and Manufacturing Have Fewer Weeks of Unemployment Unemployed workers from construction and manufacturing have shorter unemployment duration than otherwise similar workers from other industries. (See table 4.) Furthermore, simulations based on our model estimates show that differences in unemployment duration across industries exist whether or not UI benefits are received. Specifically, UI-eligible workers from construction who receive UI benefits have the fewest weeks of unemployment on average (17 weeks), when compared with workers from other industries. Likewise, UIeligible workers from construction who do not receive UI benefits also have the fewest weeks of unemployment, on average (6 weeks).

Unemployment Insurance: Factors Associated with Benefit Receipt

217

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the average likelihood of receiving UI during a first unemployment period, second unemployment period with UI receipt during the prior unemployment period, and a third unemployment period with UI receipt during both prior unemployment periods. The positive effect that each prior UI receipt period has on the likelihood of current UI receipt is statistically significantly larger for the public administration industry relative to the professional and related services industry at the 95 percent confidence level, and smaller for the agriculture and construction industries. The simulations also incorporate the industry effects and the industry interactions with the number of prior periods of unemployment. See appendix I for methodology and estimation results.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 9. Simulated effect of past UI benefit receipt on the likelihood of receiving UI in subsequent periods of unemployment, for selected industries.

Certain Occupations Are Associated with UI Benefit Receipt and Longer Unemployment Duration The likelihood of receiving UI benefits varies across occupations, but generally not as much as it does across industries. Specifically, UI-eligible managers are about one-fifth more likely to receive UI than otherwise similar transportation equipment operators, and one-half more likely to receive UI than professional and technical workers (see table 5). UI-eligible workers who have received UI benefits in the past are more likely to receive UI benefits during a current period of unemployment than UI-eligible workers who never received UI benefits, across each occupation. Specifically, this effect is strongest for sales and

United States Government Accountability Office

218

service workers and weakest for transportation equipment operators and craftsmen (see table 6).[34] Table 4. Simulated Unemployment Duration for UI-Eligible Workers, by Industry and UI Receipt Status

Industry Construction Mining Business services Manufacturing Finance, insurance, and real estate Wholesale and retail trade Public administration Professional and related services Entertainment and related services Personal services Agriculture, forestry, and fishing Transportation and public utilities Overall average duration

Simulated unemployment duration (median weeks) Receiving UI Not receiving UI benefits benefits 17 6 17 6 18 7 19 7 21 8 22 9 23 9 24 10 24 10 24 10 26 11 27 12 21 8

Note: Simulations are the median duration of unemployment during first-time unemployment. The parameter estimates for the construction and manufacturing industries are statistically significant relative to the professional and related services industry at the 95 percent confidence level. See appendix I for methodology and estimation results. Source: Simulations based upon GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 5. Simulated Likelihood of Receiving UI Benefits for UI-Eligible Workers from Different Occupations Occupation Managers and administrators Farmers, farm laborers, and foremen Machine operators (nontransportation) Craftsmen Laborers (nonfarm) Transportation equipment operators Clerical and unskilled workers Service workers (excluding private household) Sales workers Professional and technical workers Overall average

Simulated likelihood of receiving UI benefits (percent) 39 38 38 35 34 33 33 28 28 25 33

Note: Simulations are the average likelihood of receiving UI during first-time unemployment for workers from different occupations. The parameter estimates for managers and administrators, farmers, farm laborers, and foremen, machine operators, craftsmen, laborers, transportation equipment operators, and clerical and unskilled workers are statistically significant relative to professional and technical workers at the 95 percent confidence level. See appendix I for methodology and estimation results. Source: Simulations based upon GAO analysis of NLSY79 data.

Unemployment Insurance: Factors Associated with Benefit Receipt

219

Unemployment duration also varies across occupations. UI-eligible professional and technical workers have longer unemployment duration than otherwise similar workers from other occupations. Specifically, professional and technical workers have unemployment duration that is 5 weeks longer than average for workers receiving UI and 3 weeks longer than average for workers not receiving UI (see table 7).[35] Past experience with UI benefit receipt has no significant effect on unemployment duration, regardless of a worker’s occupation. Table 6. Simulated Likelihood of Receiving UI Benefits during Different Periods of UI-Eligible Unemployment for Workers with Past UI Receipt, by Occupation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Occupation Managers and administrators Farmers, farm laborers, and foremen Machine operators (nontransportation) Craftsmen Laborers (nonfarm) Transportation equipment operators Clerical and unskilled workers Service workers (excluding private household) Sales workers Professional and technical workers Overall average

Simulated likelihood of receiving UI benefits during current UI-eligible unemployment period, given past UI receipt (percent) Second unemployment Third unemployment First unemployment period period perioda 39 38 38

52 54 50

65 70 62

35 34 33 33 28

46 45 42 53 50

56 58 51 73 74

28 25 33

66 39 48

94 56 64

Note: Simulations are the average likelihood of receiving UI during a first unemployment period, a second unemployment period with UI receipt during the prior unemployment period, and a third unemployment period with UI receipt during both prior unemployment periods. The positive effect that each prior UI receipt period has on the likelihood of current UI receipt is statistically significantly larger for sales workers and service workers relative to professional and technical workers at the 95 percent confidence level, and smaller for transportation equipment operators and craftsmen. The simulations also incorporate the occupation effects and the occupation interactions with the number of prior periods of unemployment. See appendix I for methodology and estimation results. a Workers experiencing their first period of unemployment did not have past UI receipt. Source: Simulations based upon GAO analysis of NLSY79 data.

CONCLUDING OBSERVATIONS Although the UI program has existed for over 70 years and serves millions of workers each year, little is known about workers who receive UI benefits on a recurring basis or about workers who are eligible for UI benefits but never receive them. We found that UI-eligible workers during the first half of their working lives with certain demographic characteristics and from certain industries have a greater likelihood of receiving UI benefits multiple times and experiencing longer unemployment durations than otherwise similar workers. Although our results are generally consistent with past research, our analysis includes additional

United States Government Accountability Office

220

information about workers’ past experiences that provides new insight into the factors that distinguish workers who receive UI benefits from those who do not. In fact, the single most important factor associated with eligible workers receiving benefits is whether or not they received benefits during previous unemployment, suggesting that a worker’s perception of UI when they are faced with unemployment is key to whether that worker will ever use the program. Moreover, it does not appear that previous UI recipients from industries where UI benefit receipt is more likely, such as construction and manufacturing, are any more likely to receive benefits if unemployed again than similar workers from other industries. Rather, it appears that workers from these industries are simply more likely to face the choice of whether or not to file for UI benefits more often than their counterparts in other industries. In addition, while the patterns for UI receipt and unemployment duration we identified for this group during the first half of their working lives may not change significantly as they enter the second half of their working lives, it remains to be seen whether the issues they face in the years leading up to their retirement will reshape their use of the UI program. Table 7. Simulated Unemployment Duration for UI-Eligible Workers, by Occupation and UI Receipt Status

Occupation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Craftsmen Sales workers Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Service workers (excluding private household) Managers and administrators Clerical and unskilled workers Farmers, farm laborers, and foremen Professional and technical workers Overall average duration

Simulated unemployment duration (median weeks) Receiving UI Not receiving UI benefits benefits 16 6 18 7 19 7 20 8 20 8 23 9 23 23 26 26 21

9 10 11 11 8

Source: Simulations based upon GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment for workers from different occupations. The parameter estimates for craftsmen and machine operators are statistically significant relative to professional and technical workers at the 95 percent confidence level. See appendix I for methodology and estimation results.

AGENCY COMMENTS We provided a draft of this report to Labor officials for their review and comment. Labor applauded GAO’s efforts to determine the extent to which an individual worker’s characteristics are associated with the likelihood of UI benefit receipt and with unemployment duration and noted that the study adds to current knowledge of the UI program, particularly with regard to the impact of past UI benefit receipt on current UI receipt. However, Labor

Unemployment Insurance: Factors Associated with Benefit Receipt

221

also noted that there are several issues related to our methodology that may limit the utility of our findings for policymaking. While we agree that there are limitations inherent in our methodology, we believe that these limitations have been noted throughout the report, and that they do not compromise the overall validity of our results. Nevertheless, we have provided additional clarifications, as appropriate, to address Labor’s technical comments. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution of it until 30 days from its date. At that time, we will send copies of this report to relevant congressional committees, the Secretary of Labor, or other interested parties. We will also make copies available to others upon request. The report will be available at no charge on GAO’s Web site at http://www.gao.gov. If you or members of your staff have any questions about this report, please contact me at (202) 512-7215. Other major contributors are listed in appendix III. Sincerely yours, Sigurd R. Nilsen, Director Education, Workforce and Income Security Issues

APPENDIX I: ANALYSIS OF UI BENEFIT RECEIPT AND UNEMPLOYMENT DURATION

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Overview We analyzed the factors affecting unemployment insurance (UI) benefit receipt by statistically modeling the determinants of UI benefit receipt and unemployment durations simultaneously. We model UI benefit receipt in conjunction with unemployment durations to allow for correlations that may exist between the two outcomes for a given individual. For example, an unemployed person anticipating a lengthy unemployment period might be more likely to receive UI benefits than a person expecting a short unemployment period. Alternatively, the receipt of UI benefits may lengthen an unemployment period by allowing an individual to spend more time looking for new employment. In addition, our model controls for a number of observable factors about each unemployed worker’s situation, including recent employment experience, prior unemployment and UI benefit receipt experience, information about UI program factors, including benefit levels, and demographic characteristics. The model was developed and estimated by Dr. Brian McCall, Professor of Human Resources and Industrial Relations, University of Minnesota, under contract to GAO. This appendix describes (1) the data used in the analysis, including how the data were prepared, (2) the econometric model that was estimated, (3) the results from two specifications of the econometric model, and (4) the limitations inherent in the analysis.

Data Used We used the Bureau of Labor Statistics’ (BLS) National Longitudinal Survey of Youth 1979 (NLSY79) for our analysis. The NLSY79 is an ongoing longitudinal survey of individuals who were between the ages of 14 and 22 in 1979, the first year of the survey.[1] A primary focus of the NLSY79 is on individuals’ labor force patterns, and the data are

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

222

United States Government Accountability Office

collected at a very detailed level. This detail allows us to track the weekly employment, unemployment, and earnings histories of the individuals in the sample. The NLSY79 also contains less detailed information about individuals’ UI receipt during unemployment.[2] The NLSY79 does not contain direct information about an individual’s UI eligibility status, which is a function of previous employment and earnings, among other things, and varies by state of employment.[3] We estimate an unemployed individual’s UI eligibility status using data that are available in the NLSY79. There are three main reasons why the NLSY79 database provides the most suitable data for our analysis. First, the longitudinal nature and level of detail of the data allow us to control for an individual’s history of unemployment and UI receipt, which is a major contribution of this work. Second, respondents were first surveyed at a young age, which reduces the likelihood that we do not observe periods of unemployment and UI receipt early in a person’s working career. Third, the detailed data allow us to estimate an individual’s UI eligibility status, allowing us to focus our analysis on unemployed individuals whom we estimated to be eligible for UI benefits while also reasonably controlling for differences in UI program rules across states. A few limitations to the NLSY79 database should be mentioned. First, the sample began with 12,686 individuals in 1979, but has decreased in size over time due to attrition.[4] Second, the data are self-reported and thus subject to recall error. We assessed the reliability of the NLSY79 data by interviewing relevant BLS officials, reviewing extensive NLSY79 documentation, and performing electronic tests of the NLSY79 data for missing or corrupt information that might negatively affect our analysis. On the basis of these reviews and tests, we determined that the data were sufficiently reliable to be used in our analysis. We considered using administrative state UI data as an alternative to the NLSY79. Although such administrative data could provide information about all UI recipients in a state, these data could not provide information about UI-eligible unemployed workers who did not receive benefits. Also, because these data are not designed for research purposes, there is limited information available about individuals that can be used to control for differences, such as demographic characteristics. Finally, there is also no nationally representative data source for administrative UI data. For each individual in the NLSY79 database, we created a detailed weekly history of employment and unemployment, including whether UI benefits were received during unemployment. Our definition of unemployment is not the strict definition used in the BLS’s Current Population Survey (CPS). We define unemployment to include both the weeks in which an out-of-work person is looking for work (the standard CPS unemployment definition) and the weeks during which the individual reports being out of the labor force (OLF). We did require that an individual spend at least 1 week actively looking for work after a job loss to reduce the likelihood that the person had permanently left the labor force. Other research has addressed the effect that the UI program plays on the percentage of weeks of nonemployment that a person reports that he or she was looking for work.[5] For each unemployment period experienced by an individual, we estimate the person’s UI eligibility status. Although states determine UI eligibility using a number of criteria, we focus on the following three: (1) the unemployment must be the result of a job loss that was not caused by the individual, (2) the individual must have earned a specified amount of money during the time preceding the unemployment, and (3) the individual must be actively looking for new employment. The NLSY79 provides the information necessary to estimate

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployment Insurance: Factors Associated with Benefit Receipt

223

whether these criteria are met by an unemployed individual. For criterion 1, the NLSY79 provides information about the reason that a job was lost. Only those unemployed individuals who lost a job through no fault of their own were deemed to be UI-eligible.[6] For the monetary eligibility criterion 2, we compiled a detailed set of UI eligibility and benefit criteria for each of the 50 states and the District of Columbia over the period 1978 to 2002.[7] When these criteria were combined with the NLSY79’s detailed employment and earnings histories, we were able to determine monetary eligibility for UI with reasonable accuracy, as well as the weekly benefit amount and the number of weeks of benefits a person was eligible to receive.[8] For criterion 3, we considered as UI-eligible only those unemployed individuals who reported actively looking for work during at least 1 week of their unemployment. We erred on the side of overestimating the eligibility based on criterion 3, because individuals who self-report information about nonemployment may not fully realize the impact that “looking for work” versus “being out of the labor force” has on UI eligibility, especially if they did not receive UI benefits. Although this estimation method is not perfect, we believe that it captures some of the most important features of UI eligibility. It is similar to the methods used by other researchers.[9] In addition to estimating the UI eligibility status of individuals at the time of each of their unemployment periods, we also created the other variables used in our analysis. The empirical model outlined in the following subsection focuses on UI benefit receipt and unemployment duration. UI benefit receipt during unemployment was determined using the monthly measure provided in the NLSY79.[10] The duration of unemployment, as defined above, is measured in weeks from the week after a job was lost to the week a new job was begun. We censor duration to be no longer than 100 weeks. To isolate the impact that a variable has on the likelihood of UI benefit receipt and unemployment duration, our model controls for a great number of other factors that were observable at the start of, and throughout, the person’s unemployment. One set of variables relates to the employment experience of the individual immediately preceding unemployment, including industry and occupation of the lost job (measured at the one-digit Standard Industrial Classification [SIC] and Standard Occupational Classification [SOC] level), union status and tenure at the job lost, earnings (base period earnings [BPE] and high quarter earnings [HQE]), whether the job was lost because of a plant closing, and the calendar year and month the unemployment began.[11] We group both earnings measures into brackets to allow for nonlinear effects. All dollar values are adjusted for inflation to 2002 dollars using the BLS’s Consumer Price Index for All Urban Consumers (CPI-U). We also control for the state unemployment rate during the month that unemployment began, and, in the duration equation, for the time-varying state monthly unemployment rate over the period of unemployment. A second set of variables summarizes UI program factors, such as the weekly benefit amount (WBA) a person is eligible to receive, the number of weeks of benefits a person is eligible to receive, whether the state has a waiting period before benefits can be received, and whether permanent or temporary extended benefits are in effect.[12] We also control for the percentage of new UI claims that are denied by a state (in the receipt equation) and the percentage of continuing UI claims that are denied by a state (in the duration equation). In the unemployment duration equation, we also allow the parameter estimates for WBA, remaining weeks of benefits, and extended benefits to vary over the period of unemployment. This is done by interacting these variables with a cubic function of the number of weeks

224

United States Government Accountability Office

unemployed. Again, all dollar values are adjusted for inflation to 2002 dollars using the BLS’s CPI-U. A third group of variables relates to a person’s history of unemployment and UI benefit receipt as measured at the start of an unemployment period. This group of variables includes the number of times the person had been unemployed and the number of times a person had received UI benefits previously (in the receipt equation) and whether or not the person had been unemployed and whether or not the person had received UI benefits previously (in the duration equation). We also interact these variables with industry and occupation dummy variables to investigate whether previous unemployment and UI receipt affect the likelihood of current UI receipt and unemployment durations differently across industries. These interactions with industry and occupation are done in separate specifications of the model. A fourth group of variables relates to a person’s demographic characteristics at the time of unemployment. These include age, race, gender, marital status, number of years of schooling, health limitations, whether a spouse has used UI previously, family size, number of children, number of children between the ages of 0 and 2, whether the person lives with his or her parents, state of residence, and whether the person lives in a Standard Metropolitan Statistical Area (SMSA) as opposed to a rural area. We limit our analysis to the nonmilitary sample of NLSY79 respondents.[13] In addition, we drop individuals with insufficient information to estimate UI eligibility with reasonable accuracy. Data for an individual were included up to their first missed interview.[14] Individuals without any unemployment, and those without unemployment that was estimated to be UI-eligible, were not used in the analysis. Also, individuals who were missing data required by our econometric model were not used in the analysis. This yielded a sample of 5,631 individuals who had been unemployed and eligible for UI benefits at least once, resulting in a total of 15,506 separate periods of UI-eligible unemployment.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Econometric Model To investigate the key factors associated with UI benefit receipt, including the role of prior UI benefit receipt (repeat UI recipiency), we used a dynamic econometric model that jointly determines UI benefit receipt and unemployment duration. As mentioned above, the reason for modeling these outcomes jointly is to allow for the likely correlations that exist between them.[15] In addition to modeling UI receipt and unemployment duration jointly, our model allows prior unemployment and prior UI receipt to influence current UI receipt and unemployment duration to allow for the correlations that possibly exist over time for an individual. We used a complementary log-log specification to model the probability of UI receipt during an individual’s kth unemployment period, k=1, …, K, as:

where xu (k) is a vector of exogenous variables measured at the start of the kth unemployment period, all of which are assumed to be independent of the unobserved random variable ξu,

Unemployment Insurance: Factors Associated with Benefit Receipt

225

which helps control for unobserved heterogeneity. Variables in xu (k) include demographic characteristics, characteristics about the lost job, and UI program information. The vector zu (k) is a vector of endogenous variables pertaining to past unemployment and past UI benefit receipt, which are measured at the start of an individual’s kth unemployment period and may be correlated to ξu. We modeled unemployment durations using a discrete-time hazard function, which gives the probability of an event occurring during a discrete time period, conditional upon not having experienced the event prior to that time. This can be thought of as the escape rate from unemployment during a specific time period. We assume that the conditional probability that an individual’s kth period of unemployment ends in the interval (m-1,m], given that it exceeds m-1, where m indexes the number of weeks, follows a complementary log-log specification:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

for m=1, …, M, and where xd (k) is a vector of exogenous variables measured at the start of the kth unemployment period, all of which are assumed to be independent of the unobserved random variable ξd, which helps control for unobserved heterogeneity. Variables in xd (k) include demographic characteristics, characteristics about the lost job, and UI program information. The vector zd (k) consists of endogenous variables pertaining to current UI benefit receipt, past unemployment, and past UI benefit receipt, which are measured at the start of an individual’s kth unemployment period and may be related to ξd. The parameter vector αd =(αd ,αd ,...,αd ) is the baseline hazard function. Letting i, i = 1, …, I, index individuals and k index an individual’s unemployment periods, we define qi (k) to be equal to 1 when individual i has a kth unemployment period, and 0 otherwise. Also, we define ci (k) to be equal to 1 when individual i’s kth unemployment period is complete, and 0 otherwise. Using this notation, individual i’s contribution to the likelihood function can be written:

where the vector of parameters θ is to be estimated and contains βj, γj, ξj and αd, where j = d,u. We assume that the distribution of the unobserved random variables (ξd ,ξu ) is such that there are 3 different types of individuals in the population, with the fraction of each type equal to ph, where

226

United States Government Accountability Office

Combining these possibilities, we write an individual’s likelihood contribution as:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

where p =( p1 , p2 , p3 ). Taking logarithms and summing over all individuals yields the full log likelihood function for the sample:

This likelihood is computed in FORTRAN and maximized using the BHHH algorithm.[16] A number of features outlined above are simplifications of a more general version of this model, and were introduced to help reduce the number of parameters to be estimated by the model. First, the baseline hazard function, αd was assumed to be independent of the unemployment period number, k. Second, the parameters associated with the exogenous (βj,j = d,u) and endogenous (γj,j=d,u) variables were assumed to be independent of the unemployment period number, k. Third, the unobserved random variables (ξj,j= d,u) were assumed to be independent of the unemployment period number, k. Although this assumption is not as general as allowing each individual to have different unobserved components over time, it does help control for unobserved differences between individuals that may influence UI receipt and unemployment durations. Because of the complexity of the empirical model outlined above, interpreting the parameter estimates is difficult.[17] As a result, we use the output from the model to simulate the effect that changes in certain variables have on the likelihood of UI receipt and the duration of unemployment for the average unemployed person in our sample. For example, to understand differences in UI receipt and unemployment durations by industry, we simulate the likelihood of UI benefit receipt and unemployment duration for the average person in our sample for each of the possible industries, and then compare the results. To do this, we use the model’s output to calculate every person’s likelihood of UI receipt and escape rate from unemployment—conditional upon receiving and not receiving UI—assuming all were in the first industry grouping when they lost their job. Averaging over all individuals yields the average probability of UI receipt and the averaged (week by week) survivor function.[18] The averaged survivor function can be used to compute the expected median duration of unemployment.[19] We then repeat this process, successively, assuming that all individuals were in another industry grouping when they lost their jobs, until all industry groups have been simulated.

Unemployment Insurance: Factors Associated with Benefit Receipt

227

The simulated average likelihood of UI benefit receipt and median unemployment duration can then be compared across industries to estimate the differences by industry. Using all individuals for each simulation, and reporting results for the average unemployed person, helps insure that differences in the simulation results (e.g., industry 1 versus industry 2) reflect only the variables (industry 1, industry 2) being simulated.[20] To describe results that are not related to past experience with unemployment and UI benefit receipt, we present simulations that are specific to first-time unemployment—a simple and clearly defined scenario (the observable trends also hold for unemployed individuals with prior unemployment and UI receipt experience).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

RESULTS We report parameter estimates from two specifications of our model. The first specification includes interaction terms between industry and our measures of past UI benefit receipt and past unemployment. These results are presented in tables 8 and 9 for the UI benefit receipt equation and the unemployment duration equation respectively. The second specification includes interaction terms between occupation and our measures of past UI benefit receipt and past unemployment. These results are presented in tables 10 and 11 for the UI benefit receipt equation and the unemployment duration equation respectively. We included the industry and occupation interactions in separate specifications to avoid the issues brought about by multicollinearity.[21] Because the results for the noninteraction terms are similar between the two specifications, we focus on those from the industry-interaction specification (tables 8 and 9). After discussing these results, we discuss the results for the occupation-interaction specification (tables 10 and 11). Tables 8, 9, 10, and 11 are structured as follows. The first column in each table lists the variable names; the second column, the parameter estimates; the third column, the estimated standard errors; and the fourth column, the t-statistics. The last column contains asterisks that signify statistical significance. One asterisk (*) signifies statistical significance at the 90 percent confidence level (t-statistics greater than or equal to 1.65 in absolute value); two asterisks signify statistical significance at the 95 percent confidence level (t-statistics greater than or equal to 1.96 in absolute value) and three asterisks (***) signify statistical significance at the 99 percent confidence level. Parameter estimates discussed below are statistically significantly different from zero at the 95 percent confidence level unless stated otherwise.[22] To conserve space, the tables do not present the parameter estimates for the unobserved heterogeneity (ξd and ξu), state, year, and month effects. Table 8. Parameter Estimates for UI Receipt Equation from Industry-Interaction Specification

Past unemployment and UI receipt Number of previous UI receipt spells Number of previous unemployment spells Industry Agriculture, forestry, and fishing

Parameter estimate

Standard error

t-statistic

0.714 -0.072

0.086 0.017

8.26 -4.27

*** ***

0.438

0.211

2.07

**

228

United States Government Accountability Office

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 8. (Continued)

Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) Industry * number previous UI receipt spells Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing Industry * number of previous unemployment spells Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing UI program variables Weekly benefit amount (WBA) Potential UI benefit duration Waiting week for UI benefits Denial rate for new UI claims Extended UI benefits in effect

Parameter estimate 0.868 0.294 0.672 0.221 0.475 0.292 0.310 -0.077 0.104 0.560 -0.030 0.614 0.122 0.296 0.261 0.187 0.244 0.075 0.134 0.081

Standard error 0.242 0.135 0.108 0.162 0.109 0.174 0.142 0.198 0.226 0.188 0.095 0.100 0.138 0.088 0.092 0.090 0.115 0.101 0.191 0.092

t-statistic 3.59 2.17 6.20 1.36 4.36 1.68 2.19 -0.39 0.46 2.98 -0.31 6.15 0.88 3.39 2.85 2.07 2.12 0.75 0.70 0.88

*** ** ***

-0.301 -0.115 -0.229 -0.160 0.033 0.010 0.565 -0.005 -0.011 0.160 0.487 0.158

0.101 0.201 0.091 0.093 0.120 0.109 0.344 0.122 0.161 0.227 0.239 0.098

-2.97 -0.57 -2.53 -1.72 0.27 0.10 1.64 -0.04 -0.07 0.70 2.04 1.61

***

-0.043 -0.129 -0.040 -0.057 -0.001 -0.047 -0.013 -0.012 0.023 -0.052 -0.061 -0.061

0.035 0.056 0.022 0.019 0.028 0.020 0.032 0.025 0.035 0.045 0.040 0.021

-1.23 -2.29 -1.83 -2.97 -0.02 -2.30 -0.41 -0.48 0.65 -1.15 -1.54 -2.90

0.064 -0.623 0.053 -1.448 0.133

0.058 0.592 0.124 0.836 0.097

1.09 -1.05 0.43 -1.73 1.36

*** * **

*** *** *** *** ** **

** *

**

** * *** **

***

*

Unemployment Insurance: Factors Associated with Benefit Receipt

229

Table 8. (Continued)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Personal characteristics Years of education Armed Forces Qualifying Test score African-American Hispanic Hispanic * male Married Age Age-squared Male Lives in SMSA (urban) Health limitations Spouse used UI in past Spouse used UI in past * male Live with parents Family size Live with parents * family size Children under age 2 Number of children Recent employment experience Union member Tenure Tenure-squared State unemployment rate Base period earnings brackets Under $2,000 $2,000-$3,999 $4,000-$5,999 $6,000-$7,999 $8,000-$9,999

$10,000-$11,999 $12,000-$13,999 $14,000-$15,999 $16,000-$17,999 $18,000-$19,999 $20,000-$24,999 $25,000-$29,999 High quarter earnings $0-$999 $1,000-$1,999 $2,000-$2,999 $3,000-$3,999 $4,000-$4,999 $5,000-$5,999 $6,000-$6,999 $7,000-$7,999 $8,000-$8,999 Year effects Month effects State effects

Parameter estimate

Standard error

t-statistic

0.569 -0.295 0.005 -0.084 0.225 0.167 20.541 -41.787 -0.357 -0.111 0.041 0.270 -0.688 -0.160 -0.478 0.572 0.091 -0.017

0.128 0.097 0.059 0.086 0.098 0.050 4.510 8.035 0.051 0.050 0.110 0.100 0.173 0.097 0.177 0.231 0.060 0.025

4.44 -3.06 0.08 -0.98 2.30 3.35 4.55 -5.20 -7.01 -2.24 0.38 2.70 -3.97 -1.65 -2.71 2.47 1.51 -0.68

-0.003 0.140 -0.015 0.314

0.048 0.029 0.004 0.150

-0.05 4.85 -4.20 2.09

*** *** **

-1.450 -1.383 -1.177 -0.799 -0.780

0.278 0.196 0.168 0.152 0.140

-5.21 -7.05 -7.02 -5.25 -5.56

*** *** *** *** ***

-0.528 -0.565 -0.381 -0.279 -0.282 -0.143 0.015

0.127 0.126 0.116 0.109 0.108 0.089 0.086

-4.16 -4.50 -3.28 -2.56 -2.62 -1.61 0.17

*** *** *** ** ***

-0.180 0.076 0.333 0.443 0.410 0.270 0.096 0.170 0.030 Included Included Included

0.260 0.196 0.153 0.129 0.113 0.103 0.092 0.090 0.092

-0.69 0.39 2.18 3.44 3.65 2.63 1.05 1.89 0.33

*** ***

** *** *** *** *** ** *** *** * *** **

** *** *** *** *

230

United States Government Accountability Office Table 8. (Continued) Parameter estimate

Unobserved heterogeneity effects

Standard error

t-statistic

Included

Note: In the final column an asterisk signifies statistical significance at the 90 percent confidence level, two asterisks signify statistical significance at the 95 percent confidence level, and three asterisks signify statistical significance at the 99 percent confidence level. The notation X * Y signifies an interaction between the variables X and Y. The omitted category for industry is professional and related services and for occupation is professional and technical workers. The omitted category for BPE is $30,000 and above and for HQE it is $9,000 and above. Sample includes 5,631 individuals with a total of 15,506 unemployment spells. The maximized log likelihood value is -63,438.514.

Source: GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 9. Parameter Estimates for Duration Equation from Industry-Interaction Specification

Past unemployment and UI receipt Previous UI receipt Previous unemployment Industry Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing Occupation Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) Industry * previous UI receipt Agriculture, forestry, and fishing

Parameter estimate

Standard error

t-statistic

0.155 0.101

0.090 0.093

1.73 1.09

-0.088 0.301 0.314 0.213 -0.135 0.069 0.121 0.268 -0.031 -0.024

0.198 0.273 0.156 0.107 0.233 0.104 0.202 0.153 0.191 0.299

-0.45 1.10 2.01 1.99 -0.58 0.67 0.60 1.76 -0.16 -0.08

0.029 0.005

0.236 0.084

0.12 0.06

-0.046 -0.024 -0.106 0.030 -0.025

0.062 0.070 0.044 0.050 0.048

-0.74 -0.33 -2.43 0.61 -0.51

0.005 -0.030 -0.055 -0.112

0.061 0.052 0.105 0.044

0.08 -0.59 -0.52 -2.57

-0.004

0.162

-0.02

*

** **

*

**

**

Unemployment Insurance: Factors Associated with Benefit Receipt

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 9. (Continued)

Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing Industry * previous unemployment Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing UI program variables Receiving UI Weekly benefit amount (WBA) WBA * receiving UI Remaining UI benefit duration Waiting week for UI benefits Denial rate for continuing UI claims Extended UI benefits in effect Personal characteristics Years of education Armed Forces Qualifying Test score African-American Hispanic Male Married Married * male Age Age-squared Age * male Age-squared * male Lives in SMSA (urban)

Parameter estimate 0.006 -0.123 -0.136 -0.007 -0.128 -0.165 -0.104 0.101 0.088

Standard error 0.287 0.111 0.099 0.144 0.110 0.213 0.138 0.188 0.221

t-statistic 0.02 -1.10 -1.37 -0.05 -1.16 -0.78 -0.75 0.54 0.40

0.303 -0.080

0.221 0.098

1.37 -0.82

0.053 -0.455 -0.253 -0.187 0.139 -0.126 -0.175 -0.276 -0.199 0.012

0.196 0.295 0.159 0.111 0.239 0.109 0.214 0.163 0.199 0.305

0.27 -1.54 -1.60 -1.69 0.58 -1.16 -0.82 -1.69 -1.00 0.04

-0.239 0.224

0.247 0.090

-0.97 2.48

-1.256 0.031 0.067 -0.014 0.030 0.488 0.042

0.195 0.035 0.059 0.009 0.064 0.215 0.054

-6.45 0.90 1.14 -1.64 0.47 2.27 0.78

***

0.235 0.251 -0.254 -0.078 -1.022 -0.137 0.294 -5.371 9.472 7.759 -13.614 -0.040

0.069 0.056 0.033 0.037 0.397 0.037 0.049 2.796 5.010 2.905 5.150 0.027

3.40 4.49 -7.74 -2.13 -2.57 -3.73 6.05 -1.92 1.89 2.67 -2.64 -1.48

*** *** *** ** ** *** *** * * *** ***

*

*

**

**

231

232

United States Government Accountability Office Table 9 (Continued).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Parameter estimate -0.095 0.136 -0.045 -0.118 0.114 0.024

Health limitations Spouse used UI in past Live with parents Family size Live with parents * family size Live with parents * family size * male Children under age 2 -0.098 Number of children -0.004 Recent employment experience Union member 0.084 Tenure 0.050 Tenure-squared -0.008 Lost job due to plant closing -0.179 State unemployment rate (time -0.030 varying) Base period earnings brackets Under $2,000 -0.389 $2,000-$3,999 -0.367 $4,000-$5,999 -0.360 $6,000-$7,999 -0.286 $8,000-$9,999 -0.239 $10,000-$11,999 -0.165 $12,000-$13,999 -0.186 $14,000-$15,999 -0.137 $16,000-$17,999 -0.125 $18,000-$19,999 -0.101 $20,000-$24,999 -0.039 $25,000-$29,999 -0.135 High quarter earnings Under $1,000 -0.051 $1,000-$1,999 0.103 $2,000-$2,999 0.060 $3,000-$3,999 0.088 $4,000-$4,999 0.031 $5,000-$5,999 0.021 $6,000-$6,999 0.009 $7,000-$7,999 -0.034 $8,000-$8,999 -0.029 Time interactions (t = number of weeks unemployed) UI receipt * Extended Benefits * (t-1.536 1) UI receipt * Extended Benefits * (t3.351 1)-squared UI receipt * Extended Benefits * (t-0.179 1)-cubed

Standard error 0.055 0.051 0.051 0.089 0.136 0.094

t-statistic -1.73 2.68 -0.87 -1.33 0.84 0.26

* ***

0.033 0.014

-2.93 -0.30

***

0.029 0.018 0.002 0.046 0.007

2.86 2.75 -3.07 -3.87 -4.04

*** *** *** *** ***

0.103 0.088 0.080 0.079 0.074 0.072 0.069 0.066 0.064 0.066 0.056 0.053

-3.77 -4.18 -4.49 -3.64 -3.22 -2.29 -2.71 -2.07 -1.95 -1.53 -0.71 -2.54

*** *** *** *** *** ** *** ** *

0.115 0.096 0.083 0.074 0.066 0.062 0.057 0.057 0.056

-0.44 1.07 0.72 1.20 0.46 0.33 0.17 -0.60 -0.51

0.740

-2.08

2.822

1.19

0.246

-0.73

**

**

Unemployment Insurance: Factors Associated with Benefit Receipt

233

Table 9 (Continued).

Remaining UI benefit duration * (t1) Remaining UI benefit duration * (t1)-squared Remaining UI benefit duration * (t1)-cubed UI receipt * (t-1) UI receipt * (t-1)-squared UI receipt * (t-1)-cubed UI receipt * WBA * (t-1) UI receipt * WBA * (t-1)-squared UI receipt * WBA * (t-1)-cubed Year effects Month effects State effects Unobserved heterogeneity effects

Parameter estimate 0.569

Standard error 0.201

t-statistic 2.83

-3.172

2.624

-1.21

-0.256

0.956

-0.27

10.713 -24.169 1.564 -1.052 2.425 -0.141 Included Included Included Included

1.987 5.707 0.441 0.683 2.065 0.163

5.39 -4.24 3.55 -1.54 1.17 -0.87

***

*** *** ***

Note: In the final column an asterisk signifies statistical significance at the 90 percent confidence level, two asterisks signify statistical significance at the 95 percent confidence level, and three asterisks signify statistical significance at the 99 percent confidence level. The notation X * Y signifies an interaction between the variables X and Y. The omitted category for industry is professional and related services and for occupation is professional and technical workers. The omitted category for BPE is $30,000 and above and for HQE it is $9,000 and above. Sample includes 5,631 individuals with a total of 15,506 unemployment spells. The maximized log likelihood value is -63,438.514. Source: GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 10. Parameter Estimates for UI Receipt Equation from Occupation-Interaction Specification

Past unemployment and UI receipt Number of previous UI receipt spells Number of previous unemployment spells Industry Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration

Parameter estimate

Standard error

t-statistic

0.678 -0.094

0.059 0.016

11.58 -5.84

-0.044 0.302 0.004 0.346 0.256 0.285 0.307 0.253 0.045 -0.090 0.426

0.143 0.171 0.095 0.079 0.106 0.077 0.121 0.092 0.127 0.157 0.122

-0.30 1.77 0.04 4.38 2.41 3.73 2.54 2.75 0.36 -0.57 3.48

*** ***

* *** ** *** ** ***

***

234

United States Government Accountability Office

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 10. (Continued).

Valid missing Occupation Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) Occupation * number previous UI receipt spells Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) Occupation * number of previous unemployment spells Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) UI Program variables Weekly benefit amount (WBA) Potential UI benefit duration Waiting week for UI benefits Denial rate for new UI claims Extended UI benefits in effect Personal characteristics

Parameter estimate -0.174

Standard error 0.070

t-statistic -2.48

**

0.573 0.101 0.351 0.437 0.534

0.154 0.202 0.122 0.126 0.121

3.72 0.50 2.88 3.46 4.42

***

0.353 0.364 0.549 0.108

0.167 0.135 0.239 0.128

2.12 2.70 2.29 0.85

** *** **

-0.155 0.830 0.141 -0.204 -0.100

0.082 0.292 0.097 0.071 0.069

-1.89 2.84 1.45 -2.89 -1.45

* ***

-0.276 -0.108 0.035 0.283

0.073 0.082 0.182 0.091

-3.79 -1.31 0.19 3.13

***

0.025 -0.013 -0.009 -0.007 -0.060

0.025 0.036 0.021 0.020 0.021

0.98 -0.35 -0.44 -0.36 -2.91

0.020 -0.049 -0.073 -0.023

0.027 0.024 0.052 0.022

0.73 -2.03 -1.39 -1.02

0.068 -0.529 0.050 -1.482 0.126

0.058 0.589 0.121 0.828 0.095

1.16 -0.90 0.41 -1.79 1.33

*** *** ***

***

***

***

**

*

Unemployment Insurance: Factors Associated with Benefit Receipt

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 10. (Continued).

Years of education Armed Forces Qualifying Test score African-American Hispanic Hispanic * male Married Age Age-squared Male Lives in SMSA (urban) Health limitations Spouse used UI in past Spouse used UI in past * male Live with parents Family size Live with parents * family size Children under age 2 Number of children Recent employment experience Union member Tenure Tenure-squared Lost job due to plant closing State unemployment rate Base period earnings brackets Under $2,000 $2,000-$3,999 $4,000-$5,999 $6,000-$7,999 $8,000-$9,999 $10,000-$11,999 $12,000-$13,999 $14,000-$15,999 $16,000-$17,999 $18,000-$19,999 $20,000-$24,999 $25,000-$29,999 High quarter earnings $0-$999 $1,000-$1,999 $2,000-$2,999 $3,000-$3,999 $4,000-$4,999 $5,000-$5,999 $6,000-$6,999 $7,000-$7,999 $8,000-$8,999

Parameter estimate 0.537 -0.265 0.008 -0.084 0.239 0.164 18.880 -38.467 -0.362 -0.131 0.071 0.271 -0.696 -0.170 -0.485 0.577 0.109 -0.028

Standard error 0.128 0.096 0.058 0.086 0.097 0.050 4.432 7.863 0.051 0.050 0.110 0.100 0.170 0.097 0.177 0.232 0.060 0.024

t-statistic

0.000 0.130 -0.014 -0.250 0.354

0.048 0.029 0.004 0.085 0.148

0.01 4.50 -3.71 -2.94 2.40

-1.423 -1.344 -1.135 -0.808 -0.778 -0.492 -0.564 -0.365 -0.292 -0.272 -0.136 0.019

0.278 0.196 0.168 0.152 0.140 0.127 0.126 0.117 0.109 0.108 0.090 0.086

-5.11 -6.85 -6.77 -5.32 -5.55 -3.87 -4.48 -3.12 -2.68 -2.51 -1.50 0.22

-0.209 0.080 0.339 0.441 0.416 0.275 0.115 0.169 0.054

0.257 0.195 0.152 0.129 0.113 0.104 0.093 0.090 0.091

-0.81 0.41 2.23 3.43 3.69 2.65 1.24 1.88 0.59

4.19 -2.75 0.14 -0.98 2.46 3.30 4.26 -4.89 -7.13 -2.64 0.64 2.71 -4.10 -1.76 -2.74 2.49 1.84 -1.17

*** ***

** *** *** *** *** *** *** *** * *** ** *

*** *** *** ** *** *** *** *** *** *** *** *** *** **

** *** *** *** *

235

236

United States Government Accountability Office Table 10. (Continued).

Year effects Month effects State effects Unobserved heterogeneity effects

Parameter estimate Included Included Included Included

Standard error

t-statistic

Note: In the final column an asterisk signifies statistical significance at the 90 percent confidence level, two asterisks signify statistical significance at the 95 percent confidence level, and three asterisks signify statistical significance at the 99 percent confidence level. The notation X * Y signifies an interaction between the variables X and Y. The omitted category for industry is professional and related services and for occupation is professional and technical workers. The omitted category for BPE is $30,000 and above and for HQE it is $9,000 and above. Sample includes 5,631 individuals with a total of 15,506 unemployment spells. The maximized log likelihood value is -63,453.973. Source: GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 11. Parameter Estimates for Duration Equation from Occupation-Interaction Specification

Past unemployment and UI receipt Previous UI receipt Previous unemployment Industry Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation and public utilities Wholesale and retail trade Finance, insurance, and real estate Business services Personal services Entertainment and recreation services Public administration Valid missing Occupation Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen

Parameter estimate

Standard error

t-statistic

0.111 0.270

0.076 0.095

1.47 2.83

-0.031 -0.105 0.052 0.013 -0.003 -0.069 -0.062 -0.007 -0.205 -0.005

0.081 0.111 0.051 0.042 0.062 0.039 0.068 0.050 0.058 0.076

-0.38 -0.95 1.03 0.32 -0.05 -1.75 -0.92 -0.14 -3.51 -0.07

-0.153 0.190

0.072 0.033

-2.14 5.71

0.117 0.340 0.087 0.444 0.287

0.251 0.188 0.123 0.146 0.118

0.47 1.81 0.71 3.03 2.42

0.240 0.231 -0.002

0.206 0.136 0.234

1.16 1.70 -0.01

***

*

***

** ***

* *** **

*

Unemployment Insurance: Factors Associated with Benefit Receipt

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 11. (Continued).

Service workers (excluding private household) Occupation * previous UI receipt Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) Occupation * previous unemployment Managers and administrators Sales workers Clerical and unskilled workers Craftsmen Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Farmers, farm laborers, and foremen Service workers (excluding private household) UI program variables Receiving UI Weekly benefit amount (WBA) WBA * receiving UI Remaining UI benefit duration Waiting week for UI benefits Denial rate for continuing UI claims Extended UI benefits in effect Personal characteristics Years of education Armed Forces Qualifying Test score African-American Hispanic Male Married Married * male Age Age-squared

Parameter estimate 0.123

Standard error 0.122

t-statistic 1.01

0.056 -0.074 -0.063 -0.117 -0.053

0.147 0.241 0.101 0.099 0.092

0.38 -0.31 -0.63 -1.18 -0.58

-0.141 0.079 0.017

0.118 0.108 0.202

-1.20 0.73 0.09

-0.065

0.110

-0.59

-0.199 -0.387 -0.202 -0.422 -0.338

0.258 0.197 0.125 0.149 0.120

-0.77 -1.96 -1.62 -2.84 -2.80

-0.222 -0.308 -0.085

0.213 0.138 0.238

-1.04 -2.24 -0.36

**

-0.249

0.125

-2.00

**

-1.247 0.028 0.066 -0.014 0.030 0.474 0.042

0.195 0.035 0.059 0.008 0.064 0.214 0.054

-6.41 0.79 1.12 -1.68 0.47 2.21 0.77

***

0.240 0.242 -0.255 -0.080 -0.994 -0.133 0.289 -5.011 8.897

0.070 0.056 0.033 0.037 0.399 0.037 0.049 2.825 5.061

3.43 4.31 -7.74 -2.16 -2.49 -3.59 5.91 -1.77 1.76

*** *** *** ** ** *** *** * *

** *** ***

* **

237

238

United States Government Accountability Office

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Table 11. (Continued).

Age * male Age-squared * male Lives in SMSA (urban) Health limitations Spouse used UI in past Live with parents Family size Live with parents * family size Live with parents * family size * Male Children under age 2 Number of children Recent employment experience Union member Tenure Tenure-squared Lost job due to plant closing State unemployment rate (time varying) Base period earnings brackets Under $2,000 $2,000-$3,999 $4,000-$5,999 $6,000-$7,999 $8,000-$9,999 $10,000-$11,999 $12,000-$13,999 $14,000-$15,999 $16,000-$17,999 $18,000-$19,999 $20,000-$24,999 $25,000-$29,999 High quarter earnings brackets $0-$999 $1,000-$1,999 $2,000-$2,999 $3,000-$3,999 $4,000-$4,999 $5,000-$5,999 $6,000-$6,999 $7,000-$7,999 $8,000-$8,999 Time interactions (t = number of weeks unemployed) UI Receipt * Extended Benefits * (t-1)

Parameter estimate 7.571 -13.275 -0.041 -0.096 0.133 -0.042 -0.120 0.112 0.023

Standard error 2.917 5.164 0.027 0.055 0.051 0.051 0.089 0.135 0.094

t-statistic 2.60 -2.57 -1.51 -1.75 2.62 -0.83 -1.35 0.83 0.24

* **

-0.095 -0.004

0.033 0.014

-2.83 -0.30

***

0.087 0.050 -0.007 -0.175 -0.029

0.029 0.018 0.002 0.046 0.007

2.94 2.75 -3.03 -3.80 -4.02

*** *** *** *** ***

-0.387 -0.365 -0.359 -0.285 -0.236 -0.161 -0.187 -0.133 -0.134 -0.097 -0.048 -0.134

0.103 0.088 0.080 0.078 0.074 0.072 0.068 0.066 0.064 0.066 0.055 0.053

-3.76 -4.16 -4.49 -3.65 -3.18 -2.24 -2.73 -2.02 -2.10 -1.48 -0.86 -2.52

*** *** *** *** *** ** *** ** **

-0.063 0.096 0.053 0.084 0.027 0.022 0.011 -0.033 -0.032

0.115 0.096 0.083 0.074 0.066 0.062 0.057 0.057 0.056

-0.55 0.99 0.63 1.14 0.41 0.36 0.19 -0.58 -0.57

-1.582

0.734

-2.16

* ***

**

**

Unemployment Insurance: Factors Associated with Benefit Receipt

239

Table 11. (Continued).

UI Receipt * Extended Benefits * (t-1)-squared UI Receipt * Extended Benefits * (t-1)-cubed

Remaining UI benefit duration * (t1) Remaining UI benefit duration * (t1)-squared Remaining UI benefit duration * (t1)-cubed UI receipt * (t-1) UI receipt * (t-1)-squared UI receipt * (t-1)-cubed

UI receipt * WBA * (t-1) UI receipt * WBA * (t-1)-squared

UI receipt * WBA * (t-1)-cubed Year effects Month effects State effects Unobserved heterogeneity effects

Parameter estimate 3.496

Standard error 2.805

t-statistic 1.25

-0.191

0.244

-0.78

Parameter estimate 0.572

Standard error 0.201

t-statistic 2.85

-3.185

2.626

-1.21

-0.257

0.955

-0.27

10.722 -24.286 1.577 Parameter estimate -1.049 2.455 Parameter estimate -0.145 Included Included Included Included

1.976 5.667 0.438 Standard error 0.680 2.051 Standard error 0.161

5.43 -4.29 3.60 t-statistic

***

*** *** ***

-1.54 1.20 t-statistic -0.90

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: In the final column an asterisk signifies statistical significance at the 90 percent confidence level, two asterisks signify statistical significance at the 95 percent confidence level, and three asterisks signify statistical significance at the 99 percent confidence level. The notation X * Y signifies an interaction between the variables X and Y. The omitted category for industry is professional and related services and for occupation is professional and technical workers. The omitted category for BPE is $30,000 and above and for HQE it is $9,000 and above. Sample includes 5,631 individuals with a total of 15,506 unemployment spells. The maximized log likelihood value is -63,453.973. Source: GAO analysis of NLSY79 data.

Industry-Interaction Specification UI Receipt Equation Table 8 summarizes the parameter estimates for the UI receipt equation of the industryinteraction specification. A positive parameter estimate for a variable implies that an increase in the variable increases the likelihood of UI benefit receipt. A negative parameter estimate implies that an increase in the variable decreases the likelihood of UI benefit receipt. For example, the parameter estimate for years of education is 0.569, meaning that unemployed individuals with more years of education have a higher likelihood of receiving UI benefits than otherwise similar individuals with fewer years of education. The single asterisk signifies

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

240

United States Government Accountability Office

that the parameter estimate for years of education is statistically significant at the 95 percent confidence level. The results in table 8 show that the number of prior unemployment periods and the number of prior UI benefit receipt periods are strong predictors of an unemployed individual’s likelihood of receiving UI benefits. The parameter estimate for the number of prior unemployment periods is -0.072, which indicates that each additional prior unemployment period experienced by an individual reduces the likelihood of UI benefit receipt during current unemployment. Alternatively, the parameter estimate for the number of prior UI receipt periods is 0.714, which indicates that each additional prior UI receipt period experienced by an individual increases the likelihood of UI benefit receipt during current unemployment. The fact that the parameter estimate for the number of previous UI receipt periods is larger in absolute value suggests that this is the stronger of the two effects. To illustrate the magnitude of the effects, table 12 presents simulations of the likelihood of UI receipt by past unemployment and past UI receipt experience. According to the table, the average simulated likelihood of UI receipt for unemployed individuals with one previous unemployment period is 48 percent if UI was received in the previous unemployment period, but only 30 percent if UI was not received in the previous unemployment period.[23] Thus, for individuals with one previous unemployment period, the average likelihood of UI receipt is 60 percent higher (18 percentage points) for those who received UI benefits in their previous unemployment period. The remainder of the table shows that UI receipt exhibits significant occurrence dependence. Specifically, an individual who does not receive UI benefits during unemployment becomes less likely to receive them during future unemployment, while an individual who does receive UI benefits during unemployment becomes more likely to receive them during future unemployment.[24] Our model and data do not allow us to determine the underlying reasons for these associations. There are several possible reasons for the strong relationship between past UI receipt and current UI receipt, however. If unemployed people do not know they are eligible for benefits, or think that UI benefits are not worth the effort to apply, or are overoptimistic about finding employment, then there may be a “learning effect” that results from having received UI benefits which increases the likelihood of future use. Alternatively, if people do not apply for benefits because of a misperception of UI as a welfare program, then having received benefits once may soften such an outlook and increase the likelihood of future use. The results in table 8 also show that the likelihood of UI benefit receipt varies by the industry of the job lost by unemployed individuals. The industry variable is categorical in nature, so the parameter estimate for a particular category is an estimate of the effect of being in that category relative to an omitted category. The omitted category for industry is professional and related services. Table 8 shows that unemployed individuals from the mining, manufacturing, public administration, wholesale and retail trade, agriculture, forestry and fishing, business services, and construction industries are more likely to receive UI benefits than similar individuals from the professional services industry, because their parameter estimates are positive and statistically significant. To illustrate the magnitudes of these differences, table 13 presents the average simulated likelihood of UI receipt by industry under the specific assumption of first-time unemployment. The average simulated likelihood of UI receipt during first-time unemployment is 45.6 percent for unemployed miners, but only 24.3 percent for unemployed professional service workers. Table 13 clearly demonstrates that

Unemployment Insurance: Factors Associated with Benefit Receipt

241

there are significant differences across industries in unemployed individuals’ likelihoods of UI benefit receipt during first-time unemployment. Table 12. Simulated Likelihood of Receiving UI Benefits for UI-Eligible Workers during Successive Periods of Unemployment, by Past UI Receipt Status Unemployment period First Second Third Fourth Fifth Sixth

Simulated likelihood of receiving UI benefits (percent) Always received UI benefits Never received UI previously benefits previously — 33 48 30 64 28 78 25 88 23 94 21

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: Simulations are the average likelihood of receiving UI by unemployment period for two extreme cases: (1) individuals always received UI benefits during previous unemployment, and (2) individuals never received UI during previous unemployment. N/A denotes that there is no applicable value. See accompanying text for details. Source: GAO simulations based on GAO analysis of NLSY79 data.

To test whether or not the effects of previous experience with unemployment and UI receipt differ by industry, we also included the industry categories interacted with both the number of previous unemployment periods and the number of previous UI receipt periods. As was the case above, the parameter estimates are calculated relative to the omitted professional and related services industry. The parameter estimates for the industry interactions with the number of prior unemployment periods indicate that unemployed individuals from the mining, manufacturing, and wholesale and retail trade industries exhibit stronger occurrence dependence than unemployed individuals from the professional services industry.[25] That is, each additional previous unemployment period has a stronger negative effect on the likelihood of receiving UI benefits for unemployed individuals from these three industries relative to similar individuals from the professional services industry.[26] The parameter estimates for the industry interactions with the number of previous UI receipt periods show that unemployed individuals from the agriculture and construction industries exhibit weaker occurrence dependence than individuals from the professional and related services industry.[27] That is, each additional previous UI receipt period has a weaker positive effect on the likelihood of receiving UI benefits for unemployed individuals from these three industries relative to similar individuals from the professional services industry. Unemployed individuals from the manufacturing industry also have weaker occurrence dependence, but the result is only statistically significant at the 90 percent confidence level. Unemployed individuals from the public administration industry exhibit stronger occurrence dependence than individuals from the professional services industry. A similar result occurs for unemployed workers from the finance, insurance, and real estate industry, but the result is only statistically significant at the 90 percent confidence level. The other industries showed no statistically significant effects compared to those from the professional services industry.[28]

242

United States Government Accountability Office Table 13. Simulated Likelihood of Receiving UI Benefits for UI-Eligible Workers from Different Industries

Industry Mining Manufacturing Public administration Wholesale and retail trade Agriculture, forestry, and fishing Business services Construction Finance, insurance, and real estate Transportation and public utilities Entertainment and recreation services Professional and related services Personal services All industries

Simulated likelihood of receiving UI benefits (percent) 46 40 37 35 34 31 31 31 29 26 24 23 33

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: Simulations are the average likelihood of receiving UI during first-time unemployment. The parameter estimates for the mining, manufacturing, public administration, wholesale and retail trade, agriculture, forestry, and fishing, business services, and construction industries are statistically significant relative to the professional and related services industry at the 95 percent confidence level. See accompanying text for details. Source: GAO simulations based on GAO analysis of NLSY79 data.

To illustrate the magnitudes of these differences, table 14 presents the average simulated likelihood of UI receipt by industry and by the number of previous unemployment and UI receipt periods. Column 1 presents the simulations for first-time unemployment (see table 13). Column 2 presents the simulations assuming one prior unemployment period with UI receipt. Column 3 presents the simulations assuming two prior unemployment periods, both with UI receipt. Table 14 shows that, although unemployed individuals from the mining and manufacturing industries have the highest average simulated likelihoods of UI receipt for first-time unemployment, this is not the case if individuals have received UI benefits previously. For unemployed individuals with two prior UI receipt periods, those from the public administration, wholesale and retail trade, entertainment services, transportation, and business services industries are about as likely or are more likely to receive UI benefits again than similar individuals from the mining and manufacturing industries. Administrative unemployment insurance data have shown that repeat UI recipients tend to be from industries that are more seasonal, such as manufacturing and construction. Our results, however, suggest that this is not because workers from these industries who have received UI before are more likely to receive UI benefits when they become unemployed than similar workers from other industries. Rather, it may be that workers from such seasonal industries are unemployed more often on average than workers from other industries, or that a larger fraction of unemployed workers from such industries have collected UI previously. Our model also controls for UI program factors, but the results in table 8 show that after controlling for other observable characteristics, these factors had no statistically significant impact on an unemployed individual’s likelihood of UI receipt. These program factors include the estimated amount of weekly benefits an unemployed individual was eligible to

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployment Insurance: Factors Associated with Benefit Receipt

243

receive, the estimated duration of those benefits, and the state-specific denial rate for new UI claims.[29] Weekly benefits and the potential duration of benefits are functions of earnings, which we controlled for (and are discussed below). The parameter estimates in table 8 also show that a number of personal characteristics are associated with an unemployed individual’s likelihood of UI benefit receipt, including education, age, and gender. For instance, the parameter estimate on years of education is 0.569, which indicates that each year of education increases an unemployed individual’s likelihood of receiving UI benefits. The direction of the age effect on the likelihood of UI benefit receipt is difficult to interpret from the parameter estimates in table 8, because it is included as a polynomial to allow for nonlinear effects. Figure 10 presents a graph of the average simulated likelihood of UI receipt by age for the specific case of first-time unemployment. The graph shows that the likelihood of UI receipt increases until about the age of 25 and then decreases thereafter. For example, the average simulated likelihood of UI receipt during first-time unemployment for 25-year-olds is 10 percentage points (39 percent) higher than for 35-year-olds. While other research has found that older individuals are more likely to receive UI benefits, other researchers generally do not control for individuals’ past unemployment and UI receipt experience as completely as we did.[30] Because age and experience with both unemployment and UI receipt are correlated, age may act as a proxy for these experience measures when they are not controlled for. Table 8 also shows that several measures relating to the recent employment experience of unemployed individuals (excluding industry and occupation, which are discussed elsewhere) affect an unemployed individual’s likelihood of UI benefit receipt. For instance, table 8 shows that an unemployed individual’s likelihood of receiving UI benefits increases with earnings. We include two earnings measures: base period earnings and high quarter earnings. Each measure is grouped in earnings brackets and entered into the equation as a categorical variable to reflect nonlinear effects. As was the case with industry, each estimated effect is relative to an omitted category. For BPE the omitted earnings bracket is $30,000 and above and for HQE the omitted bracket is $9,000 and above. The pattern of parameter estimates for BPE shows that an unemployed individual is more likely to receive UI benefits, the higher his BPE (at least up to $20,000). The pattern of parameter estimates for HQE shows that an unemployed individual is more likely to receive UI benefits if his HQE are between $2,000 and $6,000. Figure 11 presents a graph of the average simulated likelihood of receiving UI benefits by base period earnings for the specific case of first-time unemployment. The level of HQE is varied to maintain a ratio of HQE to BPE of about 25 percent to approximate steady employment during the base period. The figure shows that unemployed individuals who earned more than $14,000 in their base period had a likelihood of UI receipt of over 40 percent, while individuals who earned less than $6,000 had a likelihood of UI receipt of less than 20 percent. Table 8 shows that employment experience measures other than earnings also affect the likelihood of UI receipt. For instance, an individual’s likelihood of UI receipt increases with tenure up to 9 years, after which it decreases. Also, an individual’s likelihood of UI receipt increases as the state unemployment rate increases. Interestingly, the parameter estimate on the plant closing variable is a statistically significant -0.263, indicating that unemployed individuals are less likely to receive UI benefits if they lost their jobs because of a plant closing. Union status does not have a statistically significant effect on an unemployed individual’s likelihood of UI receipt.

United States Government Accountability Office

244

Table 14. Simulated Likelihood of Receiving UI Benefits During Different Periods of UIEligible Unemployment for Workers with Past UI Receipt, by Industry

Industry

Mining Manufacturing Public administration Wholesale and retail trade Agriculture, forestry, and fishing Business services Construction Finance, insurance, and real estate Transportation and public utilities Entertainment and recreation services Professional and related services Personal services All industries

Simulated likelihood of receiving UI benefits during a current UIeligible unemployment period, given past UI receipt (percent) Second First unemployment Third unemployment perioda unemployment period period 46 57 69 40 52 65 37 68 91 35 52 70 34

42

50

31 31 31

48 40 64

66 51 91

29

46

66

26

45

67

24

39

58

23 33

38 48

56 64

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: Simulations are the average likelihood of receiving UI during a first unemployment period, a second unemployment period with UI receipt during the prior unemployment period, and a third unemployment period with UI receipt during both prior unemployment periods. The positive effect that each prior UI receipt period has on the likelihood of current UI receipt is statistically significantly larger for the public administration industry relative to the professional and related services industry at the 95 percent confidence level, and smaller for the agriculture and construction industries. The simulations also incorporate the industry effects and the industry interactions with the number of prior periods of unemployment. See accompanying text for details. a Workers experiencing their first period of unemployment did not have past UI receipt. Source: GAO simulations based on GAO analysis of NLSY79 data.

Unemployment Duration Equation Table 9 summarizes the parameter estimates for the unemployment duration equation of the industry interaction specification. A positive parameter estimate implies that an increase in a variable increases the escape rate from unemployment, thereby decreasing the duration of unemployment. A negative parameter estimate implies that an increase in a variable decreases the escape rate from unemployment, thereby increasing the duration of unemployment. For example, the parameter estimate for years of education is a statistically significant 0.235, which implies that unemployed individuals with more years of education have higher escape rates from unemployment than otherwise similar individuals with fewer years of education. As a result, unemployed individuals with more years of education will tend to have shorter unemployment durations than those with fewer years of education.

Unemployment Insurance: Factors Associated with Benefit Receipt

245

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the average likelihood of receiving UI during first-time unemployment at different ages. The overall average likelihood of receiving UI during first-time unemployment is 33 percent. See accompanying text for details.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 10. Simulated likelihood of receiving UI benefits for UI-eligible workers, by age.

We found that after controlling for other observable characteristics, the single most important predictor of unemployment duration is whether or not an individual receives UI benefits during the current unemployment period. The parameter estimate on the dummy variable for UI receipt status is -1.256, which implies that receiving UI benefits while unemployed reduces an individual’s escape rate from unemployment, thereby increasing unemployment duration. Simulations show that the median duration of unemployment is 8 weeks for individuals who do not receive UI benefits, but 21 weeks when they do receive UI benefits. We also allowed the effect of UI receipt to vary with the number of weeks of unemployment. These results indicate that a UI recipient’s escape rate from unemployment increases until about the 33rd week of unemployment. After 33 weeks, the escape rate decreases again until about the 72nd week, and then increases until 100 weeks.[31] The parameter estimates in table 9 show that having experienced prior unemployment or prior UI receipt has no statistically significant effect on unemployment duration. This result, however, is conditional upon whether or not an individual currently receives UI benefits. The unconditional effect of having previously received UI benefits is to increase unemployment duration. As stated earlier, we found that unemployed individuals who have previously received UI benefits are significantly more likely to receive UI benefits during current unemployment. Because those individuals who receive UI benefits during unemployment have longer unemployment duration, the unconditional effect of having previously received UI benefits is to increase unemployment duration.

246

United States Government Accountability Office

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are for the average likelihood of receiving UI during first-time unemployment at different levels of earnings. The overall average likelihood of receiving UI during first-time unemployment is 33 percent. See accompanying text for details.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 11. Simulated likelihood of receiving UI benefits for UI-eligible workers, by prior-year earnings.

Table 9 also shows that there is an association between the industry from which an individual lost a job and the duration of unemployment. As in the UI receipt equation, the omitted category for industry is professional and related services. Table 9 shows that unemployed individuals from the construction and manufacturing industries have higher escape rates from unemployment than otherwise similar individuals from the professional services industry, because their parameter estimates are positive and statistically significant. The parameter estimate for business services is also positive, but is only statistically significant at the 90 percent confidence level. The effects for the other industries are not statistically significant relative to the professional services industry. To illustrate the magnitudes of these differences, table 15 presents the median simulated duration of unemployment by industry for the specific case of first-time unemployment. The median duration is about 17 and 19 weeks, respectively, for unemployed individuals from the construction and manufacturing industries who receive UI benefits, but is about 24 weeks for those from the professional services industry. To test whether or not the effects of previous experience with unemployment and UI receipt on the duration of unemployment differ by industry, we also included the industry categories interacted with the indicators for both previous unemployment and previous UI receipt. As stated above, the effects are relative to the omitted category of professional and related services. The parameter estimates in table 9 indicate that there are no statistically significant differences across industry types by previous experience with unemployment or previous UI receipt, conditional upon current UI receipt status.[32]

Unemployment Insurance: Factors Associated with Benefit Receipt

247

Table 15. Simulated Unemployment Duration for UI-Eligible Workers, by Industry and UI Receipt Status

Industry Construction Mining Business services Manufacturing Finance, insurance, and real estate Wholesale and retail trade Public administration Professional and related services Entertainment and related services Personal services Agriculture, forestry, and fishing Transportation and public utilities Overall average duration

Simulated unemployment duration (median weeks) Receiving UI benefits Not receiving UI benefits 17 6 17 6 18 7 19 7 21 8 22 9 23 9 24 10 24 10 24 10 26 11 27 12 21 8

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: Simulations are the median duration of unemployment during first-time unemployment. The parameter estimates for the construction and manufacturing industries are statistically significant relative to the professional and related services industry at the 95 percent confidence level. See accompanying text for details. Source: GAO simulations based on GAO analysis of NLSY79 data.

Table 9 also shows that only one UI program factor (other than current UI receipt) has a statistically significant impact on an individual’s unemployment duration. Specifically, individuals who are unemployed in states with higher denial rates for continuing UI claims have higher escape rates from unemployment. That is, these individuals tend to become reemployed more quickly than those in states with lower denial rates. The parameter estimates in table 9 show that a number of personal characteristics affect an individual’s unemployment duration, including education, race, gender, and marital status. For example, the parameter estimate on years of education is 0.235, which indicates that each year of education increases an individual’s escape rate from unemployment. The simulations reported in table 16 show that unemployed individuals with 16 years of education (roughly a college education) have median unemployment duration that is about 1.9 weeks shorter than unemployed individuals with 12 years of education when UI benefits are received, and 1.1 weeks when UI benefits are not received. The parameter estimates for race show that AfricanAmericans have significantly lower escape rate from unemployment than Hispanics, who in turn have slightly lower escape rates than whites. Table 17 displays simulations of median unemployment duration by race for the specific case of first-time unemployment. Simulations showed that the age effect, although statistically significant, did not have much of an impact on the median duration of unemployment. In table 9, the parameter estimates for gender are difficult to interpret because gender is interacted with other variables in our specification, including age. Simulations show that unemployed men have median unemployment durations that are about 2 weeks shorter than for unemployed women when UI benefits are received; and about 1 week shorter when UI benefits are not received. The parameter estimates for marital status show that married women tend to have longer unemployment durations than do unmarried women and married men tend to have shorter unemployment durations than do

248

United States Government Accountability Office

unmarried men.[33] Although the age effects in table 9 are statistically significant, simulations showed that age had minimal effect on the median duration of unemployment. Table 16. Simulated Unemployment Duration for UI-Eligible Workers, by Education Level and UI Receipt Status

Years of education when unemployment began 9 10 11 12 13 14 15 16 17 18 19 20 and higher

Simulated unemployment duration (median weeks) Receiving UI benefits Not receiving UI benefits 22 22 21 21 20 20 19 19 18 18 18 17

9 9 9 8 8 8 8 7 7 7 6 6

Source: GAO simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UIeligible workers not receiving UI benefits. See accompanying text for details.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The last set of parameter estimates in table 9 relates to the recent employment experience of unemployed individuals (excluding industry and occupation, which are discussed elsewhere). Most of the parameter estimates in this grouping are statistically significant at the 95 percent level. Specifically, unemployed individuals who belonged to a union at the job that was lost had a higher escape rate from unemployment than otherwise similar individuals who were not union members. The simulations in table 18 show that union members had median unemployment durations that were 2 weeks shorter than nonunion members when UI benefits were received and 1 week shorter when UI benefits were not received. Simulations also show that an individual’s unemployment duration decreases modestly with job tenure until 7 years, after which it increases slightly. Table 17. Simulated Unemployment Duration for UI-Eligible Workers, by Race/Ethnicity and UI Receipt Status Simulated unemployment duration (median weeks) Race or ethnicity Receiving UI benefits White 19 Hispanic 21 African-American 25

Not receiving UI benefits 8 8 11

Source: GAO simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UIeligible workers not receiving UI benefits. See accompanying text for details.

Unemployment Insurance: Factors Associated with Benefit Receipt

249

Table 18. Simulated Unemployment Duration for UI-Eligible Workers, by Union Status and UI Receipt Status Simulated unemployment duration (median weeks) Union memberships status when unemployment began

Receiving UI benefits

Not receiving UI benefits

Union member

19

8

Not a union member

21

9

Source: GAO simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UIeligible workers not receiving UI benefits. See accompanying text for details.

Of our two measures of an individual’s earnings, only the base period earnings proved to have a statistically significant effect on the duration of unemployment.[34] The pattern of parameter estimates for BPE shows that unemployed individuals with low BPE have lower escape rates from unemployment than otherwise similar individuals with higher BPE. That is, lower-earning individuals tend to have longer unemployment periods. Figure 12 graphs simulations of median unemployment duration by BPE for the specific case of first-time unemployment.[35] Individuals with BPE below $6,000 tend to have longer unemployment duration than unemployed individuals with higher BPE.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Occupation-Interaction Specification We also estimated a specification of our model with interaction effects between the occupation categories (as opposed to industry) and our measures of past unemployment and past UI receipt experience. These results are presented in tables 10 and 11. A comparison of these results with those from tables 8 and 9 shows that the overall results of the two specifications are very similar. Therefore, only the occupation estimates will be discussed here. Because occupation is included as a categorical variable, the parameter estimates are relative to an omitted group, which is professional and technical workers. The estimates in table 10 show that unemployed managers, machine operators, craftsmen, laborers, transportation workers, and clerical workers are more likely to receive UI benefits than similar professional and technical workers. Table 19 presents the average simulated likelihood of receiving UI benefits by occupation for the specific case of first-time unemployment. Although the range is not as wide as for industry (see table 13), the table shows that there are differences in the likelihood of UI receipt by occupation.

250

United States Government Accountability Office

Source: Simulations based on GAO analysis of NLSY79 data. Note: Simulations are the median duration of unemployment during first-time unemployment. Overall average duration is 21 weeks for UI-eligible workers receiving UI benefits and 8 weeks for UI-eligible workers not receiving UI benefits. See accompanying text for details.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 12. Simulated unemployment duration for UI-eligible workers, by prior-year earnings and UI receipt status.

The interactions between occupation and the number of previous unemployment periods in table 10 indicate that unemployed machine operators and laborers exhibit stronger occurrence dependence than otherwise similar professional and technical workers.[36] That is, each additional previous unemployment period has a stronger negative effect on the likelihood of receiving UI benefits for unemployed individuals from these two occupations relative to similar professional and technical workers.[37] The parameter estimates for occupation interacted with the number of previous UI receipt periods show that unemployed transportation operators and craftsmen exhibit weaker occurrence dependence than otherwise similar professional and technical workers.[38] That is, each additional previous UI receipt period has a weaker positive effect on the likelihood of receiving UI benefits for unemployed individuals from these two occupations relative to otherwise similar individuals from professional and technical occupations. Managers also showed weaker occurrence dependence, but this estimate is only statistically significant at the 90 percent confidence level. Unemployed sales workers and service workers exhibit stronger occurrence dependence than otherwise similar professional and technical workers. The other occupations showed no statistically significant effects compared with professional and technical workers.[39] To illustrate the magnitudes of these differences, table 20 presents the average simulated likelihood of UI receipt by occupation and by the number of previous UI receipt periods. Column 1 presents the simulations for first-time unemployment (as in table 19). Column 2 presents the simulations assuming one prior unemployment period with UI receipt. Column 3 presents the simulations assuming two prior unemployment periods, both with UI receipt. Table 20 shows that although unemployed managers and machine operators have among the highest average simulated likelihoods of UI receipt for first-time unemployment, this is not

Unemployment Insurance: Factors Associated with Benefit Receipt

251

the case if individuals have received UI benefits previously. In the case of unemployed individuals with two prior UI receipt periods, sales workers, service workers, clerical workers, and farmers are about as likely, or are more likely, to receive UI benefits than otherwise similar managers and machine operators. Table 21 shows that there is also an association between the occupation from which an individual lost a job and the duration of unemployment. Specifically, unemployed craftsmen and machine operators have higher escape rates from unemployment than similar professional and technical workers, because the estimates are positive and statistically significant. The effects for the other occupations were not statistically significant relative to professional and technical workers. To illustrate the magnitudes of these differences, table 21 presents the median simulated duration of unemployment by occupation for the specific case of first-time unemployment. The median duration is under 20 weeks for unemployed craftsmen and machine operators who receive UI, but is almost 26 weeks for professional and technical workers. To test whether or not the effects of previous experience with unemployment and UI receipt on the duration of unemployment differ by occupation, we also included the occupation categories interacted with the indicators for both previous unemployment and previous UI receipt. As stated earlier, the effects are relative to the omitted category of professional and technical workers. The parameter estimates in table 11 indicate that the interactions for prior unemployment are negative and statistically significant for craftsmen, sales workers, machine operators, laborers, and service workers. This suggests that unemployed workers from these occupations have lower escape rates from unemployment relative to professional and technical workers as the number of past unemployment periods increases.[40] The parameter estimates for the interactions between occupation and past UI receipt showed no individual statistical significance.[41]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

LIMITATIONS OF THE ANALYSIS Although our analysis was performed using the most appropriate dataset and methodology available, there are a number of limitations to the analysis that could not be avoided and should be highlighted. Although the NLSY79 is the best available dataset for our purposes, it lacks some information that could have improved our analysis. It does not provide information about whether an unemployed individual attempted to collect UI benefits or not, only whether the individual did collect benefits. It also does not provide information about whether an individual was aware of his or her eligibility for benefits. As a result, we had to estimate each unemployed individual’s UI-eligibility status. An unemployed worker’s awareness of the UI program and knowledge of its basic rules could have a large impact on his or her decision to apply for benefits. This awareness may also be correlated with other observable characteristics (education and earnings, for example). Not controlling for awareness may affect the estimates of such variables.

252

United States Government Accountability Office Table 19. Simulated Likelihood of Receiving UI Benefits for UI-Eligible Workers from Different Occupations

Occupation Managers and administrators Farmers, farm laborers and foremen Machine operators (nontransportation) Craftsmen Laborers (nonfarm) Transportation equipment operators Clerical and unskilled workers Service workers (excluding private household) Sales workers Professional and technical workers Overall average

Simulated likelihood of receiving UI benefits (percent) 39 38 38 35 34 33 33 28 28 25 33

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Note: Simulations are the average likelihood of receiving UI during first-time unemployment for workers from different occupations. The parameter estimates for managers and administrators, farmers, farm laborers, and foremen, machine operators, craftsmen, laborers, transportation equipment operators, and clerical and unskilled workers are statistically significant relative to professional and technical workers at the 95 percent confidence level. See accompanying text for details. Source: GAO simulations based on GAO analysis of NLSY79 data.

The NLSY79 also lacks information about an unemployed worker’s former employer that could help estimate UI receipt and unemployment duration. Although our results control for industry, firms within an industry have different labor turnover patterns that result in different UI tax rates through experience rating.[42] The lack of perfect experience rating may even encourage firms to use temporary layoffs and recalls as a way of managing its labor force during demand fluctuations.[43] An individual who works for a firm with high labor turnover or with a high UI tax rate may be more aware of the UI program and, thus, more likely to receive benefits. Another limitation of the NLSY79 is that it includes only information about the specific group of individuals who were between the ages of 14 and 22 in 1979. Thus, any findings based on the NLSY79 are specific to this group and do not represent the experiences of workers of all ages during the 1979-2002 period. A methodological limitation is that we assume that the time between unemployment spells is fixed. One might expect individuals who have been unemployed and received UI benefits to change their subsequent work behavior, either to increase or decrease their chances of using the program in the future. For example, a person who received UI benefits while unemployed may search for more stable employment in order to reduce the likelihood of experiencing a layoff in the future. We do not incorporate such possibilities into our model because this would require a third equation to model employment duration, which would be a more complex and time-consuming analysis.

Unemployment Insurance: Factors Associated with Benefit Receipt

253

Table 20. Simulated Likelihood of Receiving UI Benefits during Different Periods of UIEligible Unemployment for Workers with Past UI Receipt, by Occupation

Occupation

Managers and administrators Farmers, farm laborers, and foremen Machine operators(nontransportation) Craftsmen Laborers (nonfarm) Transportation equipment operators Clerical and unskilled workers Service workers (excluding private household) Sales workers Professional and technical workers Overall average

Simulated likelihood of receiving UI benefits during a current UI-eligible unemployment period, given past UI receipt (percent) Second Third First unemployment unemployment unemployment period period perioda 39 52 65 38 54 70 38 50 62 35 46 56 34 45 58 33 42 51 33 53 73 28 50 74 28 25 33

66 39 48

94 56 64

Note: Simulations are the average likelihood of receiving UI during a first unemployment period, a second unemployment period with UI receipt during the prior unemployment period, and a third unemployment period with UI receipt during both prior unemployment periods. The positive effect that each prior UI receipt period has on the likelihood of current UI receipt is statistically significantly larger for sales workers and service workers relative to professional and technical workers at the 95 percent confidence level, and smaller for transportation equipment operators and craftsmen. The simulations also incorporate the occupation effects and the occupation interactions with the number of prior periods of unemployment. See accompanying text for details. a Workers experiencing their first period of unemployment did not have past UI receipt. Source: GAO simulations based on GAO analysis of NLSY79 data.

Table 21. Simulated Unemployment Duration for UI-Eligible Workers, by Occupation and UI Receipt Status

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Simulated unemployment duration (median weeks) Occupation Craftsmen Sales workers Machine operators (nontransportation) Transportation equipment operators Laborers (nonfarm) Service workers (excluding private household) Managers and administrators Clerical and unskilled workers Farmers, farm laborers, and foremen Professional and technical workers Overall average duration

Receiving UI benefits 16 18 19 20 20 23

Not receiving UI benefits 6 7 7 8 8 9

23 23 26 26 21

9 10 11 11 8

254

United States Government Accountability Office

Note: Simulations are the median duration of unemployment during first-time unemployment for workers from different occupations. The parameter estimates for craftsmen and machine operators are statistically significant relative to professional and technical workers at the 95 percent confidence level. See accompanying text for details. Source: GAO simulations based on GAO analysis of NLSY79 data.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

BIBLIOGRAPHY Anderson, Patricia M., and Bruce D. Meyer. “The Effect of Unemployment Insurance Taxes and Benefits on Layoffs Using Firm and Individual Data,” NBER Working Paper No. 4960. Cambridge, Massachusetts: National Bureau of Economic Research, 1994. Berndt, E. K., B. H. Hall, R. E. Hall, and J. A. Hausman. “Estimation and Inference in Nonlinear Structural Models,” Annals of Economic and Social Measurement, Vol. 3, No. 4 (1974): 653-665. Blank, Rebecca M., and David E. Card. “Recent Trends in Insured and Uninsured Unemployment: Is There an Explanation?” The Quarterly Journal of Economics, Vol. 106, No. 4 (1991): 1157-1189. Calvó-Armengol, Antoni and Matthew O. Jackson. “The Effects of Social Networks on Employment and Inequality.” The American Economic Review, Vol. 94, No. 3 (2004): pp. 426-454. Card, David E., and Phillip B. Levine. “Unemployment Insurance Taxes and the Cyclical and Seasonal Properties of Unemployment,” Journal of Public Economics, Vol. 53, No. 1 (1994): 1–29. Feldstein,Martin. “Temporary Layoffs in the Theory of Unemployment,” The Journal of Political Economy, Vol. 84, No. 5 (1976): 937-958. Gritz, R. Mark, and Thomas MaCurdy. “Measuring the Influence of Unemployment Insurance on Unemployment Experiences,” Journal of Business and Economic Statistics, Vol. 15, No. 2 (1997): 130-152. Gruber, Jonathan. “The Wealth of the Unemployed,” Industrial and Labor Relations Review, Vol. 55, No. 1 (2001): 79-94. Krueger, Alan B., and Bruce D. Meyer. “Labor Supply Effects of Social Insurance,” NBER Working Paper 9014. Cambridge, Massachusetts: National Bureau of Economic Research, 2002. McCall, Brian P. “Repeat Use of Unemployment Insurance,” in Laurie J. Bassi and Stephen A. Woodbury, editors, Long-Term Unemployment and Reemployment Policies. Stamford, Connecticut: JAI Press, Inc., 2000. Meyer, Bruce D. “Unemployment Insurance and Unemployment Spells,” Econometrica, Vol. 58, No. 4 (1990): 757-782. Meyer, Bruce D., and Dan T. Rosenbaum. “Repeat Use of Unemployment Insurance,” NBER Working Paper 5423. Cambridge, Massachusetts: National Bureau of Economic Research, 1996. Mortensen, Dale T. “Unemployment Insurance and Job Search Decisions,” Industrial and Labor Relations Review, Vol. 30, No. 4 (1977): 505-517. Needels, Karen E., and Walter Nicholson. An Analysis of Unemployment Insurance Durations since the 1990-1992 Recession. Prepared for the Department of Labor. 1999.

Unemployment Insurance: Factors Associated with Benefit Receipt

255

O’Leary, Christopher J., and Stephen A. Wandner, editors. Unemployment Insurance in the United States: Analysis of Policy Issues. Kalamazoo, Michigan: W. E. Upjohn Institute for Employment Research, 1997. Topel, Robert H. “On Layoffs and Unemployment Insurance,” The American Economic Review, Vol. 73, No. 4 (1983): 541-559.

Related GAO Products Unemployment Insurance: Better Data Needed to Assess Reemployment Services to Claimants. GAO-05-413. Washington, D.C.: June 24, 2005. Unemployment Insurance: Information on Benefit Receipt. GAO-05-291. Washington, D.C.: March 17, 2005. Women’s Earnings: Work Patterns Partially Explain Difference between Men’s and Women’s Earnings. GAO-04-35. Washington, D.C.: October 31, 2003. Unemployment Insurance: States’ Use of the 2002 Reed Act Distribution. GAO-03-496. Washington, D.C.: March 6, 2003. Unemployment Insurance: Enhanced Focus on Program Integrity Could Reduce Overpayments. GAO-02-820T. Washington, D.C.: June 11, 2002. Unemployment Insurance: Increased Focus on Program Integrity Could Reduce Billions in Overpayments. GAO-02-697. Washington, D.C.: July 12, 2002. Unemployment Insurance: Role as Safety Net for Low-Wage Workers Is Limited. GAO-01181. Washington, D.C.: December 29, 2000.

REFERENCES [1] [2]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[3]

[4]

[5]

UI programs are administered by the 50 states, the District of Columbia, Puerto Rico, and the Virgin Islands. GAO, Unemployment Insurance: Information on Benefit Receipt, GAO-05-291 (Washington, D.C.: Mar. 17, 2005). We considered an individual to be UI-eligible if that individual experienced an involuntary job loss, reported receiving a minimum amount of wages over a minimum period of time as defined by the state where the individual lived, and was actively looking for new employment. Our method of estimating eligibility tends to overestimate the number of UI-eligible individuals. For a more complete discussion of our methodology, see appendix I. Alan B. Krueger and Bruce D. Meyer, “Labor Supply Effects of Social Insurance,” NBER Working Paper 9014 (Cambridge, Massachusetts: National Bureau of Economic Research, 2002). Rebecca M. Blank and David E. Card, “Recent Trends in Insured and Uninsured Unemployment: Is There an Explanation?” The Quarterly Journal of Economics, vol. 106, no. 4 (1991).

256 [6]

[7]

[8] [9]

[10]

[11] [12] [13]

[14] [15] [16] [17] [18]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[19]

[20]

[21]

[22]

United States Government Accountability Office Bruce D. Meyer and Dan T. Rosenbaum, “Repeat Use of Unemployment Insurance,” NBER Working Paper 5423 (Cambridge, Massachusetts: National Bureau of Economic Research, 1996), p. 20. See Brian P. McCall, “Repeat Use of Unemployment Insurance,” in Laurie J. Bassi and Stephen A. Woodbury, editors, Long-Term Unemployment and Reemployment Policies (Stamford, Connecticut: JAI Press, Inc., 2000). GAO-05-291. The results described in this report are statistically significant at the 95 percent confidence level, unless otherwise noted. For a complete list of findings from our multivariate statistical model of the key factors associated with UI benefit receipt, see table 8 in appendix I. Earnings refers to base period earnings, which we define as the amount of earnings received during the first four of the last five full calendar quarters before a worker becomes unemployed. This definition is consistent with the time frame states generally use to determine eligibility. The average and maximum earnings for the unemployed workers in our sample are $15,524 and $597,950, respectively. GAO, Unemployment Insurance: Role as Safety Net for Low-Wage Workers Is Limited, GAO-01-181 (Washington, D.C.: Dec. 29, 2000). For economic theory concerning the relationship between job search and unemployment insurance, see Dale T. Mortensen, “Unemployment Insurance and Job Search Decisions,” Industrial and Labor Relations Review, vol. 30, no. 4 (1977). See Blank and Card, and McCall. See Jonathan Gruber, “The Wealth of the Unemployed,” October 2001, Industrial and Labor Relations Review, vol. 55, no. 1. The average number of years of schooling completed by UI-eligible workers, at the time when they became unemployed, is 12 years. See Blank and Card, p. 1185. We specifically tested for the effect of spousal income on the likelihood of receiving UI to determine whether marital status was masking some underlying effect of additional family income, and found this not to be the case. See David E. Card and Phillip B. Levine, “Unemployment Insurance Taxes and the Cyclical and Seasonal Properties of Unemployment,” Journal of Public Economics, vol. 53, no. 1 (1994); Patricia M. Anderson and Bruce D. Meyer, “The Effect of Unemployment Insurance Taxes and Benefits on Layoffs Using Firm and Individual Data,” NBER Working Paper No. 4960, December 1994; and Robert H. Topel, “On Layoffs and Unemployment Insurance,” American Economic Review, vol. 73, no. 4 (1983). As noted above, relatively few UI-eligible workers who receive UI benefits receive them multiple times. See GAO-05-291 for a more complete discussion of the incidence of repeat UI benefit receipt. For the parameter estimates of these and other variables included in our multivariate statistical model of the key factors associated with unemployment duration, see table 9 in appendix I. The variables reported here are those that were statistically significant at the 95 percent confidence level. See Mortensen.

Unemployment Insurance: Factors Associated with Benefit Receipt

257

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[23] The average prior-year earnings amount for this sample is $15,524. [24] See Karen E. Needels and Walter Nicholson, An Analysis of Unemployment Durations Since the 1990-1992 Recession, UI Occasional Paper 99-6, prepared for the Department of Labor, 1999, p. 94. [25] See Bruce D. Meyer, “Unemployment Insurance and Unemployment Spells,” Econometrica, vol. 58, no. 4 (1990), p. 771. [26] The average number of years of schooling completed by UI-eligible workers, at the time when they became unemployed, is 12 years. [27] Needels and Nicholson, p. 6. [28] See Needels and Nicholson. [29] See Antoni Calvó-Armengol, and Matthew O. Jackson, “The Effects of Social Networks on Employment and Inequality,” The American Economic Review, Vol. 94, No. 3, (2004) for a discussion of the effects of individuals’ social networks on employment outcomes. [30] See Needels and Nicholson, and GAO, Women’s Earnings: Work Patterns Partially Explain Differences between Men’s and Women’s Earnings, GAO-04-35 (Washington, D.C.: Oct. 31, 2003). [31] See Needels and Nicholson. We did not control for the likely effect of an expected job recall. [32] The percentages in table 2 and figure 8 are not comparable. The percentages in table 2 represent an individual worker’s likelihood of receiving UI when UI-eligible unemployment occurs, whereas the percentages in figure 8 compare the relative proportions of unemployment spells with UI benefit receipt coming from different industries. [33] Although the association between past UI receipt and current UI receipt is statistically significant for all industries combined, differences in this association among industries were statistically significant only for public administration, agriculture, and construction. [34] Although the association between past UI receipt and current UI receipt is statistically significant for all occupations combined, differences in this association among occupations were statistically significant only for sales and service workers, and for transportation equipment operators and craftsmen. [35] The largest differences between industries in median weeks of unemployment are 10 weeks for workers receiving UI and 5 weeks for workers not receiving UI.

APPENDIX I [1]

[2]

NLSY79 data begin in 1978. Interviews for the NLSY79 were conducted annually until 1994, and biennially beginning in 1996. We used data through 2002, which were the most recent NLSY79 data available. UI receipt information is provided on a monthly basis in the NLSY79. Because this information is only given on a monthly basis, it cannot be used to accurately measure the number of weeks of UI receipt during unemployment.

258 [3]

[4]

[5]

[6]

[7] [8]

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[9] [10]

[11]

[12]

[13]

[14]

United States Government Accountability Office State UI programs determine eligibility using a number of criteria, including the following conditions: (1) the unemployment must be the result of a job loss that was not caused by the individual, (2) the individual must have earned a specified amount of money during the time preceding the unemployment, and (3) the individual must be actively looking for new employment. See Center for Human Resource Research, Ohio State University, The National Longitudinal Surveys NLSY79 User’s Guide, prepared for the Department of Labor, 2002. R. Mark Gritz and Thomas MaCurdy, “Measuring the Influence of Unemployment Insurance on Unemployment Experiences,” Journal of Business and Economic Statistics, vol. 15, no. 2, (1997), examined the role that UI rules have on an individual’s choice to report himself or herself as unemployed (CPS definition) as opposed to out of the labor force. They found that, in addition to having longer nonemployment periods, UI recipients report being unemployed in the CPS sense for a greater proportion of their nonemployment period. It appeared from the NLSY79 data that a number of respondents did not differentiate between being laid off and being discharged or fired. As a result, we include those who report being either laid off or discharged or fired as satisfying the first UI eligibility rule. The NLSY79 reports a number of other reasons for leaving a job, including having found better work, low pay, pregnancy, illness, change of job by spouse or parents, other family reasons, job’s interference with school, the end of a program, bad working conditions, and entrance into the armed forces. See U.S. Department of Labor. Employment and Training Administration, Significant Provisions of State Unemployment Insurance Laws (Washington, D.C., 1979-2002). Although UI eligibility is based upon the rules in the state where an individual is employed, we used state of residence for our estimates because state of employment was not available in the NLSY79. Thus, people who work in one state but live in another may not be classified correctly. However, we believe that only a small percentage of such data are classified incorrectly and, thereby, our results should be only minimally affected. See Gritz and MaCurdy, 1997, and McCall 2000 for examples. We consider only an individual’s first period of unemployment with UI receipt during a person’s “benefit year.” A benefit year is the 52-week period during which UI benefits can be claimed. The base period is the period of time during which earnings are counted toward UI eligibility. It generally covers a year. We define the base period as the first four of the last five completed calendar quarters. High quarter earnings refers to the quarter of highest earnings during the base period. Permanent extended benefits are triggered by high unemployment rates in a state, and provide for up to 13 additional weeks of benefits to UI-eligible individuals. Temporary extended benefits are available periodically, as authorized by Congress. The NLSY79 began with 12,686 individuals in 1979, 1,280 of whom were part of the military subsample. The majority of the military subsample of the NLSY79 was eliminated in 1985. The NLSY79 attempts to reconnect with individuals that missed an interview in the previous year.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Unemployment Insurance: Factors Associated with Benefit Receipt

259

[15] For one example of an economic model of how the receipt of UI benefits can affect the expected length of unemployment by affecting a person’s reservation wage, see Mortensen. [16] See E. K. Berndt, B. H. Hall, R. E. Hall, and J. A. Hausman, “Estimation and Inference in Nonlinear Structural Models,” Annals of Economic and Social Measurement, vol. 3, no. 4 (1974). The BHHH algorithm is a quasi-Newton method for finding maximums. [17] In addition to being a highly nonlinear model, the data were all normalized to help the convergence of the parameter estimates. [18] The survivor function at time t for an event is the probability of not having experienced that event prior to time t. The survivor function is mathematically related to the escape rate (hazard rate). [19] We chose median rather than mean because of the skewed nature of our unemployment duration data. [20] For our simulations, if we used only those individuals that reported losing a job from a specific industry, as opposed to using all individuals, it is likely that a portion of the differences we would observe in the likelihood of UI benefit receipt and unemployment duration would be due to differences in other observable factors between the individuals from the different industry groups. For example, it may be that professional services workers have higher average earnings than agricultural workers, which would be earnings effect, not an industry-specific effect. [21] We also tried running a specification of the model that included these interactions for both industry and occupation. The parameter estimates and simulations were generally similar to those for the two separate specifications, but much of the statistical significance for individual parameters was lost due to the correlation between industry and occupation. However, a likelihood ratio test of the joint hypothesis that the interaction terms for both industry and occupation are all equal to zero is rejected at the 95 percent confidence level, suggesting that there are both industry-specific and occupation-specific differences in the effects of past unemployment and past UI receipt on the likelihood of current UI receipt and current unemployment duration. [22] A statistically insignificant result indicates that the effect of a characteristic could not be precisely estimated using the sample data, and does not necessarily prove that the characteristic is unimportant. [23] Note that the average simulated likelihood of UI receipt for first-time unemployed workers is 33 percent. [24] See Brian P. McCall, “Repeat Use of Unemployment Insurance,” in Laurie J. Bassi and Stephen A. Woodbury, editors, Long-Term Unemployment and Reemployment Policies (Stamford, Connecticut: JAI Press, Inc., 2000). [25] As stated above, the occurrence dependence in this case relates to the fact that an individual who does not receive UI benefits during unemployment becomes less likely to receive them during future unemployment. [26] Although the results for some industries were not individually statistically significant, a likelihood ratio test of the joint hypothesis that all of the interaction terms between industry and past unemployment experience are equal to zero is rejected at the 95 percent confidence level.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

260

United States Government Accountability Office

[27] As stated earlier, occurrence dependence relating to previous UI receipt means that an individual who receives UI benefits during unemployment becomes more likely to receive them during future unemployment. [28] However, a likelihood ratio test of the joint hypothesis that all of the interaction terms between industry and past UI receipt experience are equal to zero is rejected at the 95 percent confidence level. [29] Other researchers have found that the weekly benefit amount does not affect UI receipt. See Gritz and MaCurdy. [30] See McCall. [31] Changes in the escape rate over an unemployment period are also affected by the other time-interaction effects included in the specification. However, these other effects do not affect the general shape of this overall trend. [32] However, a likelihood ratio test of the joint hypotheses that all of the interaction terms between industry and past unemployment experience are equal to zero is rejected at the 95 percent confidence level. A similar test of the joint hypothesis that all of the interaction terms between industry and past UI receipt experience are equal to zero could not be rejected at the 95 percent confidence level. [33] In alternative specifications we explored whether an individual’s likelihood of UI benefit receipt and unemployment duration were affected by spousal income in the previous year. We found that spousal income had no statistically significant effect on an individual’s likelihood of UI benefit receipt, but did slightly increased the duration of unemployment. [34] Recall that each measure was broken into earnings brackets and entered into the equation as a categorical variable. See tables 8,9,10, or 11 for the brackets used. The omitted category for BPE is $30,000 and above and the omitted category for high quarter earnings is $9,000 and above. [35] For comparability, the simulations in figure 12 hold the ratio of HQE to BPE as closely as possible to 25 percent. [36] As stated above, the occurrence dependence in this case relates to the fact that an individual who does not receive UI benefits during unemployment becomes less likely to receive them during future unemployment. [37] Although the results for some occupations were not individually statistically significant, a likelihood ratio test of the joint hypothesis that all of the interaction terms between occupation and past unemployment experience are equal to zero is rejected at the 95 percent confidence level. [38] As stated earlier, occurrence dependence relating to previous UI receipt means that an individual who receives UI benefits during unemployment becomes more likely to receive them during future unemployment. [39] However, a likelihood ratio test of the joint hypothesis that all of the interaction terms between occupation and past UI receipt experience are equal to zero is rejected at the 95 percent confidence level. [40] However, a likelihood ratio test of the joint hypothesis that all of the interaction terms between occupation and past unemployment experience are equal to zero could not be rejected at the 95 percent confidence level.

Unemployment Insurance: Factors Associated with Benefit Receipt

261

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

[41] In addition, a likelihood ratio test of the joint hypotheses that all of the interaction terms between occupation and past UI receipt experience are equal to zero could not be rejected at the 95 percent confidence level. [42] Experience rating describes the practice of making a firm’s UI tax rate a function of the amount of UI benefits paid to its former employees. [43] See Martin Feldstein, “Temporary Layoffs in the Theory of Unemployment,” The Journal of Political Economy, vol. 84, no. 5 (1976).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

In: Economics of Employment and Unemployment Editor: Manon C. Fourier and Chloe S. Mercer

ISBN: 978-1-60456-741-0 ©2009 Nova Science Publishers Inc.

Chapter 9

INFLATION, UNEMPLOYMENT, AND LABOR FORCE CHANGE IN EUROPEAN COUNTIES Ivan O. Kitov Institute for the Dynamics of the Geospheres, Russian Academy of Sciences, Moscow, Russia

ABSTRACT

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Linear relationships among inflation, unemployment, and labor force are obtained for two European countries — Austria and France. The best fit models of inflation as a linear and lagged function of labor force change rate and unemployment explain more than 90% of observed variation (R2>0.9). Labor force projections for Austria provide a forecast of decreasing inflation for the next ten years. In France, inflation lags by four years behind labor force change and unemployment allowing for an exact prediction at a four-year horizon. Standard error of such a prediction is lower than 1%. The results confirm those obtained for the USA and Japan and provide strong evidence in favor of the concept of labor force growth as the only driving force behind unemployment and inflation.

INTRODUCTION Current discussions around the Phillips curve are even more active and extensive than 30 years ago, with a full set of models exploring various assumptions on the real forces behind inflation. There is no unique and comprehensive model, however, which is able to explain all observations relevant to inflation in developed countries. There are three principal ways to follow in the discussion on sources of inflation. The first way is to continue the investigation of inflation in the framework of the Phillips curve (PC). The second is to admit that there is no real driving force behind inflation except unpredictable exogenous shocks of unknown origin such as productivity or supply shocks in modern real business cycle (RBC) models. The third is to abolish the current paradigm and to use a different mechanism driving inflation and unemployment together, which is based on

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

264

Ivan O. Kitov

natural first principles (theoretical foundations), and validated by observations (empirical foundations). This paper adds to the development of the third concept using labor force change as the driving force behind both inflation and unemployment. Conventional economists running along the first wide avenue are numerous and represent a good part of the theoretical power elaborating monetary policies of central banks in developed countries. In fact, the Phillips curve allows for a feasible monetary policy due to the assumption that there is an interaction between monetary controllable impulses or exogenous shocks and variables describing real economy such as real GDP, output gap, marginal cost, labor cost share, etc. (Unemployment is missing in this list of the variables associated with real economy because, according to our concept, it does not belong). In the absence of such an interaction, no monetary policy is necessary with inflation completely reflecting money growth in developed economies, as mentioned in the Robert Lucas’ Nobel Prize Lecture (Lucas, 1995). The money supply is an arbitrary choice of central banks, which does not influence any real economic variable. In the framework of the conventional Phillips curves, however, inflation is not neutral. relative to the performance of real economies and central banks that have to balance smoothing of price fluctuations and losses in real economic growth. These are only assumptions, however, not confirmed by empirical evidences to the extent adopted in hard sciences. Statistical inferences supporting the PC assumptions are not objective links or trade-offs between involved economic variables but non-zero correlation. See, for example, Ang et al. (2005), Ball (2000), Ball and Mankiw (2002), Ball et al. (2005), Stock and Watson (1999, 2002a, 2002b, 2003, 2005), Gali and Gertler (1999), Gali, Gertler, and Lopez-Salido (2001, 2005), Sbordone (2002, 2005), Rasche and Williams (2005), Piger and Rasche (2006), among others, where the statistical character of the links between inflation and many other economic and financial parameters is the primary objective. These authors have successfully found that functional dependencies between inflation and studied parameters unpredictably vary through time. Despite similar outcomes sought under the PC approach one can distinguish several “schools of thought” elaborating various approaches both empirical and theoretical. There is a large group of economists who adopted numerous techniques of econometrics, which link inflation to their own lagged values and some measures of real activity, which differ from unemployment as originally introduced by A.W. Phillips. In the simplest approximation, a NAIRU concept has been elaborated by Gordon (1988, 1998), Steiger, Stock, and Watson (1997a, 1997b), Ball and Mankiw (2002), among many others, in order to improve the original model. More complicated econometric PC models include hundreds of variables related to real activity aggregated in few indices, as presented by Marcellino et al. (2001), Stock and Watson (1999, 2002a, 2002b, 2003), Ang et al. (2005), Canova (2002), Hubrich (2005). Another conventional approach is associated with the accelerationist or “expectation augmented” Phillips curve allowing only for backward-looking expectations (Friedman 1968, Phelps 1967). Despite the Lucas (1976) and Sargent (1971) critique and failure to predict actual observations in the USA and other developed countries during the 1970s and 1980s, the model has survived and is often used by central bankers in the elaboration of actual monetary policy (Rudd and Whelan, 2005).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Inflation, Unemployment, Labor Force Change in European Counties

265

Fast growing in number and evolving in theoretical diversity is the group related to the New Keynesian Phillips Curve (NKPC) based on rational expectations not on lagged inflation. The expectations are usually modeled by a random price adjustment process, and thus intrinsically related to real marginal cost. In the most recent models developed by Gali and Gertler (1999), Gali, Gertler, and Lopez-Salido (2001, 2005), Sbordone (2002, 2005), among others, unit labor marginal cost is used as a marginal cost proxy. A hybrid model including lagged and future inflation values, various parameters related to real activity, and exogenous shocks, monetary and price ones, is also considered as an alternative to the pure cases of conventional PC or NKPC models with various degree of success (Rudd and Whelan, 2005). One can also distinguish a group of economists applying a modern behavioral approach in order to explain the price adjustment process - Akerlof (2002), Mankiw (2001), Mankiw and Reis (2002), Ball et al. (2005), among others. In this framework, sticky prices used by the NKPC group are replaced with “sticky” information. This makes individual decisions on price change, i.e. on overall inflation when aggregated over the whole economy, to be imperfect due to imperfection in processing of available information. Effectively, it means that the inflation expectations resulted from the imperfect information processing are not “rational” and do not meet axiomatic requirements of rational expectations used by the NKPC. In practice, the conventional explanation of the price inflation lacks empirical justification extended beyond autoregressive properties of inflation itself, and is also theoretically challenged by modern growth models insisting on independence of real economic performance on monetary issues, as introduced by Kydland and Prescott (1982). The real business cycle theory implies that variations in real economies are almost completely described by exogenous shocks in productivity and supply. Money is absent in RBC models or artificially introduced in some of them-Gavine and Kydland (1996) and Prescott (2004). Numerous econometric studies confirm the RBC assumption on money neutrality by statistical inferences; Atkeson and Ohanian (2002), Piger and Rasche (2005), Rasche and Williams (2005), among many others, have found that AR models explain evolution of inflation almost completely, with a marginal improvement from usage of real economic variables being only a statistical and transient one. A study of inflation and unemployment as economic variables driven solely by labor force change has been carried out by Kitov (2006a, 2006b, 2006c) for the two largest economies – the USA and Japan. The study has revealed linear relationships between inflation, unemployment and labor force. In the USA, the linear relationships are also characterized by time lags with the change in labor force leading inflation and unemployment by two and five years, respectively. In Japan, labor force change, unemployment and inflation evolve synchronously. The revealed linear link allows a partial inflation control and provides clear foundations for a reasonable economic policy related to inflation and unemployment. In this paper, the same approach linking inflation and unemployment to labor force change is applied to Austria and France. The reminder of the paper is organized in four sections. Section 1 briefly presents data sources and the model. Data on inflation, unemployment, and labor force for European countries is available from various sources. This diversity creates a number of problems but allows for an indirect estimation of the uncertainty related to various data series.

266

Ivan O. Kitov

Section 2 is devoted to Austria as a country with elaborated statistics providing a long time series with changing definitions and procedures. The changes are well documented and clear in corresponding curves. The importance of information on definitions and procedures for a successful modelling is illustrated and discussed. Inflation and unemployment in France are considered in Section 3. The country represents an economy with a size in between those of the USA and Austria. The case of France is of a large importance for our concept because of the outstanding changes related to the rules of the European Monetary Union fixing allowed inflation to figures near 2%. The limitation violates the partition of labor force change into inflation and unemployment, which was natural for France and observed since the 1960s. An elevated unemployment is observed as a response to the fast growth in labor force started in 1996 and the fixed inflation. Section 4 discusses principal findings of the study and concludes.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

DATA SOURCES AND THE MODEL The principal source of information relevant to the study is the OECD database (http://www.oecd.org/scripts/cde) which provides comprehensive data sets on labor force, unemployment, working age population, and participation rate. National statistical sources are used for obtaining original data on inflation (CPI and GDP deflator) and corroborative data on unemployment and labor force. As a rule, the data are available at the Eurostat web-site (http://epp.eurostat.cec.eu.int). An extended set of data on economic and population variables in Austria is obtained by the courtesy of Austrian national Bank employees1. In some cases, readings associated with the same variable but obtained from different sources do not coincide. This is due to different approaches and definitions applied by corresponding agencies. Diversity of definitions is accompanied by a degree of uncertainty related to corresponding measurements. For example, figures related to labor force are usually obtained in surveys covering population samples of various sizes: from 0.2 per cent to 3.3 per cent of total population (Eurostat, 2002). The uncertainty associated with such measurements cannot be easily estimated but certainly affects reliability of the inflation/labor force linear relationship (Kitov, 2006a, 2006c). When using the term “accuracy” we refer not to the absolute difference between measured and actual values but to some estimated uncertainty of measurements. This uncertainty might be roughly approximated by variations in a given parameter between consequent surveys or between different agencies. For example, the US Census Bureau (2002) gives a very low measurement related uncertainty for the annual population estimates. At the same time, some micro-surveys conducted after decennial censuses indicate the presence of deviations from the census enumerated values as large as 5 per cent in some age groups (West and Robinson, 1999). Such errors are far above those guarantied by pure statistical approach used in the evaluation of survey/census results. Therefore, one can consider the uncertainty of several percent as the one characterizing the population estimates during and between censuses, at least in some age groups. Survey reported uncertainties are 1

The author thanks Dr. Gnan from the OeNB for providing an extensive data set for Austria.

Inflation, Unemployment, Labor Force Change in European Counties

267

just a formal statistical estimate of the internal consistency of the measurements. (It is worth noting that population related variables could be potentially measured exactly because they are countable not measurable). In any case, the discrepancy between model predicted values and corresponding measurements has to be considered in the framework of measurements uncertainty. The model, which we test in the study, links inflation and unemployment to labor force change rate. It is important to use the rate of growth not increment as a predictor in order to match dimension of inflation and unemployment, which are defined as rates as well. An implicit assumption of the model is that inflation and unemployment do not depend directly on parameters describing real economic activity (Kitov, 2006a). Moreover, inflation does not depend on its own previous and/or future values because it is completely controlled by a variable of different nature. As defined in Kitov (2006a), inflation and unemployment are linear and potentially lagged functions of labor force:

π(t)=A1dLF(t-t1)/LF(t-t1)+A2 (1)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

UE(t)=B1dLF(t-t2)/LF(t-t2)+B2

(2)

where π(t) is the inflation at time t (represented by some standard measure such as GDP deflator or CPI), UE(t) is the unemployment at time t (which is also potentially represented by various measures), LF(t) is the labor force at time t, t1 and t2 are the time lags between the inflation, unemployment, and labor force, respectively, A1, B1, A2, and B2 are country specific coefficients, which have to be determined empirically. The coefficients may vary through time for a given country as different measures (or definitions) of the studied variables are used. Linear relationships (1) and (2) define inflation and unemployment separately. These variables are two indivisible features of a unique process, however. The process is the labor force growth, which is accommodated in real economies though two channels. The first channel is the increase in employment and corresponding change in personal income distribution (PID). All persons obtaining new paid jobs or their equivalents presumably change their incomes to some higher levels. There is an ultimate empirical fact, however, that the US PID does not change with time in relative terms, i.e. when normalized to the total population and total income (Kitov, 2005b). The increasing number of people at higher income levels, as related to the new paid jobs, leads to a certain disturbance in the PID. This over-concentration (or over-pressure) of population in some income bins above its neutral value must be compensated by such an extension in corresponding income scale, which returns the PID to its original density. Related stretching of the income scale is called inflation (Kitov, 2006a). The mechanism responsible for the compensation and the income scale stretching, obviously, has some positive relaxation time, which effectively separates in time the source of inflation, i.e. the labor force change, and the reaction, i.e. the inflation. The second channel is related to those persons in the labor force who failed to obtain a new paid job. These people do not leave the labor force but join unemployment. Supposedly, they do not change corresponding PID because they do not change their incomes. Therefore, total labor force change equals unemployment change plus employment change, the latter process expressed through lagged inflation. In the case of a "natural" behavior of an economic

268

Ivan O. Kitov

system, which is defined as a stable balance of socio-economic forces in corresponding society, the partition of labor force growth between unemployment and inflation is retained through time and the linear relationships hold separately. There is always a possibility, however, to fix one of the two dependent variables. For example, central banks are able to fix inflation rate by monetary means. Such a violation of the natural economic behavior would undoubtedly distort the partition of the labor force change – the portion previously accommodated by inflation would be redirected to unemployment. To account for this effect one should to use a generalized relationship as represented by the sum of relationships (1) and (2):

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

π(t)+UE(t)= A1dLF(t-t1)/LF(t-t1)+B1dLF(t-t2)/LF(t-t2)+A2+B2 (3) Equation (3) balances labor force change, inflation and unemployment, the latter two variables potentially lagging by different times behind the labor force change. The importance of this generalized relationship is demonstrated in this paper on the example of France. For the USA, there has been no need so far to apply relationship (3) because corresponding monetary policies and other potential sources of disturbance do not change the natural partition of labor force change, as observed since the late 1950s. Coefficients in relationships (1) and (2) specific for the USA are as follows: A1=4, A2=-0.03, t1=2 years (GDP deflator as a measure of inflation), B1=2.1, B2=-0.023, t2=5 years. For Japan, A1=1.77, A2=-0.003, t1=0 years (GDP deflator as a measure of inflation) (Kitov, 2006b). The labor force change rate measured in Japan is negative since 1999 and corresponding measures of inflation, GDP deflator and CPI, are negative as well. There is no indication of any recovery to positive figures any time soon if to consider the decrease in working age population and participation rate as observed in Japan from 1999. The formal statistical assessment of the linear relationships carried out by Kitov (2006d) for the USA indicates that root mean square forecasting error (RMSFE) at a two-year horizon for the period between 1965 and 2002 is only 0.8%. This value is superior to that obtained with any other inflation model by almost a factor of 2, as presented by Stocks and Watson (1999, 2005), Atkeson and Ohanian (2001), Ang et al. (2005), Marcellino et al. (2005). When the entire period is split into two segments before and after 1983, the forecasting superiority is retained with RMSFE of 1.0% for the first (1965-1983) and 0.5% for the second (1983-2002) sub- period. In a majority of inflation models, the turning point in 1983 is dictated by inability to describe inflation process with one set of defining parameters. Therefore, special discussions are devoted to statistical, economic, and/or financial justification of the split and the change in parameters (see Stock and Watson, 2005). Our model denies the existence of any change in the US inflation behavior around 1983 or in any other point after 1960. Every inflation reading is completely defined by the labor force change occurred two years before. The linear relationships between inflation, unemployment, and labor force change perform excellent for the two largest world economies during a long period. These relationships are expected to be successful for other developed economies with similar socioeconomic organization. European countries provide a variety of features related to inflation and unemployment as one can conclude from the economic statistics provided by OCED and Eurostat. This diversity includes periods of very high inflation accompanied by high unemployment, periods of low inflation and unemployment, and other combinations

Inflation, Unemployment, Labor Force Change in European Counties

269

complicated by transition periods. It is a big challenge for any theory of inflation to explain these empirical facts. Currently, the diversity resulted in a well-recognized and thoroughly discussed failure of conventional economics to provide a consistent and reliable description covering the past 50 years and all developed countries. As a consequence, the current monetary policy of the European Central Bank is based mainly on invalidated assumptions and subjective opinions of economists and central bankers, but not on a robust model predicting inflation behavior under different conditions. In the USA, the current (and historical!) practice aimed at inflation control, as implemented by the Federal Open Market Committee, definitely, has no visible influence on the observed inflation, if labor force change is the driving force.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

AUSTRIA The first country to examine is Austria. It provides an example of a small economy in terms of working age population. At the same time, the Austrian economy is characterized by a long history of measurements and availability of time series and descriptive information relevant to the concept under study. Austria has been demonstrating an excellent economic performance since 1950 and is characterized by an average per capita GDP annual increment of $467 (Geary-Khamis PPP The Groningen Growth and Development Center and Conference Board, 2006) for the period between 1950 and 2005. This value is very close to that for the USA ($480) and Japan ($485) (Kitov, 2006e). Such a good performance distinguishes Austria from a raw of relatively weak performances of larger European economies such as France ($406), the UK ($378), Italy ($405), and Sweden ($381) during the same period. It was discussed in Kitov (2006a, 2006b, 2006d) that data quality is the principal characteristic defining the success of any attempt of modelling inflation and unemployment as a function of labor force change. There are two main sources of uncertainty in the data related to our study. The first source is associated with measurement errors. It is a more important issue for the accuracy of labor force surveys, which usually provide original data on unemployment and labor force. In the surveys, measurement accuracy depends on sampling and nonsampling errors. The former is estimated using population coverage and some standard statistical principles, and the latter is more difficult to evaluate (CB, 2002). The second source of uncertainty is important for both labor force, including unemployment as a constituent part, and inflation measurements and is associated with variations in definitions given to these economic variables. The definitions are often revised and modified, sometimes dramatically, as one can judge from the description given by the OECD (2005). When applied to labor force, such revisions introduce severe breaks in corresponding time series associated with the change in units of measurements. (In physics, it would have been practically impossible to obtain any reliable empirical relationship if measurement units had varied in such uncontrollable way as in economics.) Moreover, European countries have implemented the changes at different times creating asynchronous breaks. Modifications of methodologies and procedures related to inflation measurements are accompanied by introduction of new measures such as harmonized index of consumer prices

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

270

Ivan O. Kitov

(Eurostat, 2006a). The latter index has replaced the old CPI definition in official statistics of European countries. Therefore, we start with a detailed description of the data obtained for Austria. We use six sources providing annual readings for CPI, GDP deflator, population estimates, unemployment rate, participation rate, and labor force level: Eurostat, OECD, AMS (Arbeitsmarktservice) Österreich (http://www.ams.or.at), HSV (Hauptverband der Sozialversicherungtraeger) Österreich (http://www.hsv.or.at), Statistik Austria (http://www.statistik.at), and the Österreichische Nationalbank (ÖNB – http://www.oenb.at). These sources estimate the same variables in different ways. Comparison of equivalent (by title) time series allows a quantitative evaluation of differences between them. The main purpose of such a cross-examination is twofold: 1) demonstration of the discrepancy between the series as a quantitative measure of the uncertainty in corresponding parameters and 2) determination of the degree of similarity between the series. The estimated uncertainty puts a strong constraint on the level of confidence related to statistical estimates using the data sets. One cannot trust any statistical inference with a confidence level higher than allowed by the uncertainty. On the other hand, equivalent time series obtained according to various definitions (procedures, methodologies, samples, etc.) of the same parameter represent different portions of some actual value of the parameter. For example, various definitions of employment are aimed at obtaining the number of those persons who work for pay or profit. The persons are the only source of goods and services sold for money. The definitions are designed in a way for corresponding estimates to approach the actual value. If consistent and successful, the definitions always provide close to constant and different estimates of the portions of the actual value. Thus, the estimates are scalable - one can easily compute values according to all definitions having only one of them. In this sense, various definitions and related estimates are exchangeable in the framework of the linear relationship between inflation, unemployment, and labor force. Three different definitions of inflation rate are presented in Figure 1: CPI and GDP deflator as obtained using prices expressed in national currency (national accounts -NAC), and GDP deflator estimated using the Austrian shilling/Euro exchange rate (Euro accounts EUR). The latter variable is characterized by the largest variations. The curves corresponding to the inflation measurements represented by the NAC CPI and NAC GDP deflator are closer (correlation coefficient of 0.92 for the period between 1961 and 2004), but differ in amplitude and timing of principal changes. There are periods of an almost total coincidence, however. The EUR GDP deflator series is characterized by correlation coefficients 0.86 and 0.82 as obtained for the NAC GDP deflator and CPI, respectively. Therefore, one can expect a better exchangeability between the NAC CPI and NAC GDP deflator than that in the two other combinations. Since the middle 1970s, inflation in Austria has a definition-independent tendency to decrease. The last 25 years are characterized by annual inflation rates below 5% for the NAC representations. Standard labor force surveys conducted in Europe cover small portions of total population (Eurostat, 2006b). Levels of labor force and unemployment are estimated using specific weights (population controls) for every person in the survey to compute the portion of population with the same characteristics as the person has. Population controls or population portions in predefined age-sex-race bins are primarily obtained during censuses, which theoretically cover entire population.

Inflation, Unemployment, Labor Force Change in European Counties

271

0.25 CPI (NAC) GDP deflator (NAC) GDP deflator (EUR)

0.20

inflation

0.15 0.10 0.05 0.00 1955

1965

1975

1985

1995

2005

-0.05 calendar year

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 1. Comparison of three variables representing inflation in Austria: GDP deflator determined using national currency (NAC) and Euro (EUR), and CPI determined by using national currency. The GDP deflator curves coincide since 2000. Inflation volatility is much lower when it is represented in national currency. Correlation coefficients for the period between 1961 and 2004: CPI NAC/GDP deflator NAC - 0.92; CPI NAC/GDP deflator EUR - 0.82; GDP deflator NAC/GDP deflator EUR 0.86.

Between censuses, i.e. during postcensal periods, estimated figures are used as obtained by the population components change: births, deaths, net migration, as, for example, reported by the US Census Bureau (2002). Because of low accuracy of postcensal estimates, every new census reveals some “error of the closure”, i.e. the difference between pre-estimated and census enumerated values. To adjust to new population figures, the difference is proportionally distributed over the years between the censuses; similar to the procedures applied by the US Census Bureau (2004). Such population revisions may be as large as several percent. Thus, when using some current figures of labor force and unemployment, one has to bear in mind that the figures are prone to further revisions according to the censuses to come. Figure 2 illustrates the differences in population revision procedures between OECD and Statistik Austria (NAC): two curves represent the rate of change in the population of 15 years of age and over in Austria. Between 1960 and 1983, the curves coincide since OECD uses the national definition. After 1983, the curves diverge, with the OECD curve being almost everywhere above that corresponding to the national approach. There are three distinct spikes in the OECD curve: between 1990 and 1993 and in 2002, which are related to population revisions. As explained by OECD (2005), "From 1992, data are annual averages. Prior to 1992, data are mid-year estimates obtained by averaging official estimates at 31 December for two consecutive years". And - "From 2002, data are in line with the 2001 census". The 2002 revision impulsively compensates the difference between OECD and Statistik Austria accumulated during the previous 20 years: the populations in 1982 and 2002 coincide. Such step adjustments are observed in the USA population data as well (Kitov, 2006a). They introduce a significant deterioration in statistical estimates, but are easily removed by a

Ivan O. Kitov

272

simple redistribution as demonstrated by Kitov (2006d). Sometimes such step adjustments are confused with actual changes in the economic variables under stud. One has to be careful to distinguish between actual changes and artificial corrections usually associated with the years of census or large revisions in definitions. 1.5 WAP (OECD)

change rate, %

1

WAP (NAC)

0.5 0 1955

1965

1975

1985

1995

2005

-0.5 -1 -1.5 calendar year

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 2. Comparison of the rate of change in working age population (aged 15 and over) in Austria as determined by the OECD and national statistics (NAC). Notice the spikes in the OECD curve related to step adjustments according to population surveys.

The national estimates in Figure 2 are visually smoother indicating some measures applied to distribute the errors of the closure and other adjustments over the entire period. In average, the population over 15 years of age in Austria has been changing slowly so far – at an annual rate below 0.5% - with occasional jumps to 0.7% - 1.0%. Such weak but steady growth supports, however, a gradual increase in labor force and prevents deflationary periods. The level of labor force can be represented as a product of total population and corresponding participation rate (LFPR) both taken in some predefined age range. There is no conventional definition concerning the age range, however. Popular is an open range above 15 years of age and that between 15 and 64 years. The OECD series using the former definition is presented in Figure 3. OeNB (2005) provides another measure of LFPR - "the fraction of the working-age population that is employed or seeking employment", also presented in Figure 3. The curves have been evolving more or less synchronously, with the OECD curve well above that reported by the OeNB. The LFPR is responsible for a substantial part of the labor force total change: ~ 8% increase from 1976 to 1996, i.e. 0.4% per year. The current LFPR value of about 59%, as reported by the OECD, is historically high. One can hardly expect a further increase in LFPR. A decrease is more probable, as some other developed countries demonstrate. The rate of labor force growth was very low in Austria during the last 10 years, as Figure 4 demonstrates. There are three labor force time series displayed, as estimated by the OECD, Eurostat, and NAC. The Eurostat series is represented by civilian labor force. Prior to 1994, armed forces were included in the civilian labor force (CLF), in services. The NAC readings

Inflation, Unemployment, Labor Force Change in European Counties

273

include the estimates of employment made according to the HSV definition and those of unemployment level made by AMS (Statistik Austria, 2005). Both agencies base their estimates on administrative records. Thus, their approach has been undergoing weaker changes in definitions and procedures since the 1960s compared to that adopted by the OECD and Eurostat. 65 LFPR (OECD) 60

LFPR (OeNB)

55

%

50 45 40 35 30 1950

1960

1970

1980

1990

2000

2010

calendar year

Figure 3. Labor force participation rate (LFPR) in Austria as determined by OECD and obtained from the OeNB. A weak tendency to growth was observed in the beginning of the 2000s.

0.06

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

change rate

0.04

0.02

0.00 1955

1965

1975

1985

1995

2005

dLF/LF (OECD) dLF/LF (NAC) CLF (Eurostat)

-0.02

-0.04 calendar year

Figure 4. Comparison of labor force change rate estimates as reported by OECD, NAC, and Eurostat. Notice the smoothness of the NAC curve.

The curves in Figure 4 have inherited the features, which are demonstrated by corresponding working age populations in Figure 2. The OECD curve is characterized by several spikes of artificial character, as discussed above. The Eurostat curve is similar to that

Ivan O. Kitov

274

reported by the OECD with minor deviations probably associated with differences between LF and CLF. The NAC LF curve is smoother. It demonstrates a period of a slow growth with a high volatility in the 1970s, a period with an elevated growth with a high volatility between 1981 and 1995, and again a slow growth period with a low volatility during the last ten years (from 1995 to 2005). The second period is characterized by significant changes in the labor force definition - both for employment and unemployment (OECD, 2005): • •



Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.



"In 1982, re-weighting of the sample was made, due to an underestimation of persons aged 15 to 29 years. In 1984, the sample was revised and a change occurred in the classification of women on maternity leave: they were classified as unemployed before 1984 and as employed thereafter. In 1987, a change occurred in the definition of the unemployed where non-registered jobseekers were classified as unemployed if they had been seeking work in the last four weeks and if they were available for work within four weeks. In previous surveys, the unemployment concept excluded most unemployed persons not previously employed and most persons re-entering the labor market. Employment data from 1994 are compatible with ILO guidelines and the time criterion applied to classify persons as employed is reduced to 1 hour. "

Therefore, one can expect some measurable changes in the units of the labor force measurements during the period between 1982 and 1987 and in 1994. The latter change is potentially the largest since the time criterion dropped from 13 hours, as had been defined in 1974, to 1 hour. For the sake of consistency in definitions and procedures, the NAC labor force is used as a predictor in this study. The OECD labor force time series is also used in few cases to illustrate that the definitions provide similar results. For the labor force series, quantitative statistical estimates of similarity (such as correlation) are worthless due to the spikes in the OECD time series. There are three curves associated with unemployment estimates for Austria shown in Figure 5, as defined by the national statistics approach (AMS), Eurostat, and OECD. It is illustrative to trace changes in the definitions used by the institutions over time. Currently, OECD and Eurostat use very similar approaches. There was a period between 1977 and 1983 when OECD adopted the national definition, which was different from the one used by Eurostat. During a short period between 1973 and 1977, the three time series were very close to each other. A major change in all three series occurred between 1982 1987 according to the changes in definitions, as described above. Therefore, the unemployment curves in Figure 5 are characterized by two distinct branches: a low (~2%) unemployment period between 1960 and 1982 and a period of an elevated unemployment (~4% for the OECD and Eurostat, and ~6.5% for the AMS) since 1983.

Inflation, Unemployment, Labor Force Change in European Counties

275

0.1 UE (AMS) 0.08

UE (Eurostat) UE (OECD)

UE

0.06

0.04

0.02

0 1955

1965

1975

1985

1995

2005

calendar year

Figure 5. Estimates of unemployment rate in Austria according to definitions given by the AMS, Eurostat, and OECD.

The switches between various definitions, as adopted by the OECD, also do not facilitate obtaining of a unique relationship between labor force change and unemployment. The AMS definition based on administrative records might be the most consistent among the three, but it definitely differs from the definition recommended by the International Labor Organization, as adopted in European countries (Statistik Austria, 2005). We use the national and OECD time series to represent unemployment in the linear relationship linking it to labor force.

0.08

UE

0.06

0.04

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.02 UE (AMS) 0.7*dLF/LF+0.0705 (NAC) 0.35*dLF/LF+0.026 (NAC) 0.00 1955

1965

1975

1985

1995

2005

calendar year

Figure 6. Comparison of the observed (AMS) and predicted by the linear relationships (shown in lower right corner of the panel) using the NAC (AMS+HSV) labor force and the AMS unemployment rate. Changes in the unemployment and labor force definitions between 1983 and 1987 make it impossible to fit the unemployment curve during this period. Otherwise, the predicted curve is in a good agreement with the measured one.

Ivan O. Kitov

276

The above discussion explains why one cannot model the whole period by a unique linear relationship. There was a period of substantial changes in units of measurement between 1982 and 1987. Therefore, we model the Austrian unemployment (UE) during the periods before 1982 and after 1986 separately. The period between 1982 and 1987 is hardly to be matched by a linear relationship. Results of the modeling are presented in Figure 6, where the AMS unemployment curve is matched by the following relationships:

UE(t)=0.35*dLF(t)/LF(t)+0.0260 (t1986)

(5)

0.05

0.04

UE

0.03

0.02

UE (OECD) 0.35*dLF/LF+0.0405 (OECD) 0.3*dLF/LF+0.02 (NAC)

0.01

0.00 1955

1965

1975

1985

1995

2005

1995

2005

calendar year

1.2

cumulative UE

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1.0

UE (OECD) 0.35*dLF/LF+0.0405 (OECD) 0.3*dLF/LF+0.02 (NAC)

0.8 0.6 0.4 0.2 0.0 1955

1965

1975

1985

calendar year

Figure 7. Comparison of the observed (OECD) and predicted (AMS before 1980 and OECD after 1980) unemployment rate in Austria. The upper frame displays annual readings and the lower one – cumulative unemployment since 1968. Notice a major change in unemployment definition between 1981 and 1984 (OECD, 2005).

Inflation, Unemployment, Labor Force Change in European Counties

277

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

The NAC labor force time series is used for the prediction with no time lead ahead of the unemployment. The absence of any lag might be presumed as a natural behavior of labor force and unemployment as one of the labor force components, but labor force change in the US leads unemployment by 5 years. Hence, processes behind labor force change and unemployment growth are different. Coefficients in relationships (4) and (5) provide the best visible fit between the observed and predicted curves. From the Figure and the relationships, one can conclude that there was a step change in the unemployment average level from approximately 0.03 during the years before 1982 to 0.07 for the period after 1986. In addition, the linear coefficient has doubled indicating a higher sensitivity of the unemployment to the labor force change under the new definitions introduced between 1982 and 1987. The annual OECD unemployment readings presented in Figure 7 vary by less than 1%, if to exclude a short period between 1980 and 1983, when changes in definitions resulted in a step-like unemployment increase. Duration of this period of changing definitions is different from that related to the NAC unemployment according to the timing of the changes as adopted by AMS and OECD. This jump in the unemployment rate from 2% to 4% during the two years between 1981 and 1983 is not well modeled. Otherwise, the following relationships are used to match the observed unemployment readings:

UE(t)=0.35*dLF(t)/LF(t)+0.0405 (t≥1983)

(6)

UE(t)=0.30*dLF(t)/LF(t)+0.020 (t≤1980)

(7)

For the period before 1980, the NAC labor force readings are used, and the OECD labor force is used after 1981. We combined the labor force data sets in order to demonstrate their exchangeability in the description of the unemployment. Cumulative curves in the lower panel of Figure 7 illustrate the quality of the overall match between the measured and predicted values. The cumulative curves are very sensitive to the intercepts in relationships (6) and (7) as they are summed through time. Therefore, the intercepts 0.0405 and 0.020 are significant to the last digits. Potential variation in the linear coefficients in (6) and (7) is not so well resolved. Amplitude of the variations in the unemployment during the entire period except the short period between 1980 and 1983 is so low that makes the prediction according to (6) and (7) of a limited reliability. To obtain a more reliable prediction, the unemployment has to undergo an actual (not definition related) change at an annual rate of several percent, what would have been a big surprise for Austria with its stable socio-economic conditions and demographic structure. The agreement observed between the cumulative curves also is not statistically significant since it just reflects the unchanging unemployment and labor force growth rates during the two separately modeled periods. These results can be interpreted, however, as an indication of a weak dependence of the unemployment on the labor force change. The latter is transmitted only by one third into the unemployment as the linear coefficients 0.30 and 0.35 indicate. These transmission coefficients are an order of magnitude smaller than that for the USA (Kitov, 2006a). The difference is of a potential importance because labor force participation rate and unemployment in both countries are close.

Ivan O. Kitov

278

Table 1 consistently lists results of linear regression analysis carried out in the study for various measures of unemployment and inflation with labor force as a predictor, as obtained for Austria. First row of the Table presents standard deviation (stdev) as obtained for the OECD readings of unemployment in Austria during the period between 1983 and 2003. Table 1. Results of linear regression analysis for Austria Period

Predictor

1983-2003

annual UE (OECD)

1983-2003

annual UE (OECD) cumulative UE (OECD)

annual dLF(t)/LF(t) (OECD) cumulative dLF(t)/LF(t) (OECD)

1960-2003

annual GDP deflator (NAC) annual GDP deflator (NAC) annual GDP deflator (NAC) 2-year moving average GDP deflator (NAC) cumulative GDP deflator (NAC)

annual dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC) cumulative dLF(t)/LF(t) (NAC)

1965-2003

annual CPI (NAC)

1983-2003 1965-2003 1965-2003 1965-2003 1965-2003

1965-2003

annual CPI (NAC)

1965-2003

annual CPI (NAC) 2-year moving average CPI (NAC) annual GDP deflator (Eurostat) annual GDP deflator (Eurostat) annual GDP deflator (Eurostat) 2-year moving average GDP deflator (Eurostat)

1965-2003 1965-2003 1965-2003 1965-2003

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Dependent variable

1965-2003

1965-2003

annual GDP deflator (NAC) annual GDP deflator (NAC)

1965-2003

annual GDP deflator (NAC)

1965-2003

2-year moving average GDP deflator (NAC)

1965-2003

A

B

R2

stdev 0.0036

1.03 (0.020) 1.00 (0.006)

0.026 (0.007) 0.010 (0.003)

0.11 0.99 9

0.0035 0.007 0.022

0.880 (007) 0.95 (0.07) 0.93 (0.06) 1.03 (0.004)

0.005 (0.003) 0.003 (0.003) 0.003 (0.002) 0.003 (0.005)

0.81

0.010

0.85

0.009

0.88 0.99 9

0.007 0.011 0.022

annual dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC)

0.76 (0.10) 0.85 (0.10) 0.83 (0.09)

0.010 (0.004) 0.006 (0.004) 0.007 (0.004)

0.60

0.014

0.64

0.013

0.72

0.011 0.046

annual dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC) 2-year moving average dLF(t)/LF(t) (NAC)

0.66

0.027

0.88 (0.10)

0.010 (0.007) 0.008 (0.007)

0.68

0.027

0.87 (0.07)

0.008 (0.005)

0.78

0.02

0.89 (0.06)

0.04 (0.03)

0.86

0.008

0.89 (0.06)

0.004 (0.003)

0.86

0.008

0.91 (0.05)

0.003 (0.002)

0.91

0.007

0.82 (0.10)

0.022 annual dLF(t)/LF(t)UE(t) (NAC) 2-year moving average dLF(t)/LF(t)-UE(t) (NAC) 2-year moving average dLF(t)/LF(t)-UE(t) (NAC)

Inflation, Unemployment, Labor Force Change in European Counties

279

The standard deviation is 0.0036. Second and third rows present regression coefficients with their standard errors, R2, and stdev as obtained for the OECD unemployment between 1983 and 2003 with a predictor computed by relationship (6) with the OECD labor force readings. (A linear regression analysis for the whole period between 1969 and 2003 would be meaningless because of the artificial change in the predicted curve around 1982.) For the annual UE readings after 1983, R2 is very low (0.11) and stdev=0.0035, i.e. marginally lower than stdev for the UE series itself. For the cumulative curves during the same period, R2=0.999 and stdev=0.007. Therefore, relationships (4) through (7) are accurate one but not reliable. In fact, only large and synchronized in time and amplitude actual changes can provide a more reliable evidence for the model. Inflation in Austria provides a variable with higher fluctuations to predict. 0.12 GDP deflator (NAC) 1986-2003 1965-1986

inflation

0.08

0.04

0.00 1955

1965

1975

1985

1995

2005

-0.04 calendar year

2.5

cumulative inflation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

2.0

GDP deflator (NAC) 1986-2003 1965-1986

1.5

1.0

0.5

0.0 1955

1965

1975 1985 calendar year

1995

2005

Figure 8. Comparison of the observed (NAC GDP deflator) and predicted inflation in Austria. The upper frame displays annual readings and the lower one – cumulative inflation since 1960. Notice a major change in labor force definition between 1981 and 1987 (OECD, 2005). The periods before and after 1986 are described separately.

280

Ivan O. Kitov

Figure 8 depicts observed and predicted, annual and cumulative, inflation values in Austria for the period between 1960 and 2003. As mentioned above, there was a significant change in the labor force (employment and unemployment separately) statistics in the 1980s. Thus, the two different periods are described by two different linear relationships without any time lag between variables. The GDP deflator, as determined by the national statistics approach, represents inflation. Labor force is also taken according to the NAC (AMS+HSV) definition. The relationships predicting inflation are as follows:

π(t)=2.0*dLF(t)/LF(t)+0.033 (1960≤ t ≤ 1985)(8)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

π(t)=1.25*dLF(t)/LF(t)+0.0075 ( t ≥ 1986)

(9)

Coefficients in the relationships are obtained by fitting the cumulative curves over the entire period, with 1986 being the point where relationship (8) is replaced by relationship (9). Ratio of the linear coefficients in (8) and (9) is 2/1.25=1.6 and the intercept dropped from 0.033 to 0.0075. The change in the linear coefficients is consistent with the changes in the definition of labor force in between 1982 and 1987 – gradually more and more persons were counted in as employed and unemployed with a substantial increase in the labor force level. The increase resulted in corresponding growth in annual increments and the decrease in the linear coefficient (or sensitivity) in relationship (9). Thus, the sensitivity of inflation to the new measure of labor force (or new units of measurement) in Austria has decreased. This does not mean that the observed inflation path has changed, but, if to use relationship (8) for the second period, the inflation would be overestimated, as shown in Figure 8. The deviation between the two predicted curves after 1986 demonstrates the importance of the changes in definition for quantitative modeling of economic parameters. The two predicted curves are in a good agreement with the actual inflation readings within relevant periods. A prominent feature is an almost complete coincidence between 1968 and 1975, when the highest changes in the inflation rate were observed: from 0.027 in 1968 to 0.095 in 1973, and back to 0.056 in 1975. Conventional inflation models, including the Phillips curve, the NKPC or any other model using autoregressive properties of inflation, fail to describe such a dynamic behavior as a rule. They require introduction of some artificial, i.e. based on various invalidated assumptions, features such as structural breaks. Another opportunity used in conventional models is to split corresponding time series into two segments before and after such inflation peak, as was observed in Austria in 1973. Our model describes the whole period without any difficulty and the best description of the inflation is achieved during the period of the largest changes. This provides the best evidence of an adequate modeling by relationship (8). Similar conclusion is valid for the period after 1987, where an excellent timing and amplitude correspondence is observed between the measured inflation and that predicted according relationship (9). In addition, there is a transition period between 1982 and 1987, where neither of relationships (8) and (9) is expected to be accurate due to the reported changes in the labor force definition. A quantitative measure of the agreement between the observed and predicted curves is provided by a linear regression analysis. Table 1 lists standard deviation for the NAC GDP deflator time series between 1965 and 2003, stdev=0.022 (2.2%). The inflation computed according to (8) and (9) is used as a predictor and results in R2= 0.81 and stdev=0.01 (1%).

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Inflation, Unemployment, Labor Force Change in European Counties

281

Hence, the prediction based on the labor force explains 81% of variation in the original inflation series. Standard deviation could be considered as an equivalent of root mean square forecasting error (RMSFE) – for “in-sample” forecasts in the case of Austria. For the USA, R2=0.62 and stdev=0.014 for the original annual readings of GDP deflator and labor force covering the same period (Kitov, 2006d). Perhaps, the Austrian labor force and inflation measurements are characterized by a higher accuracy. A number of simple measures is proposed by Kitov (2006d) in order to improve the quality of labor force measurements and to obtain more reliable statistical estimates. Due to the lack of information on quantitative characteristics of the revisions applied to the Austrian labor force series, similar to that available for the USA, we cannot correct for probable step revisions. Thus, a natural next step is to apply a moving average technique. A two-year moving average suppresses the noise associated with the labor force measurements and also removes the shift in timing between the inflation and labor force readings - by definition, annual values of labor force correspond rather to July than to December. Averaging over two years effectively moves the center of the measurement period to December. Table 1 represents the results of a linear regression when two-year moving average is applied to the labor force and inflation. Averaging of the labor force solely before usage in relationships (8) and (9), results in R2=0.85 and stdev=0.009. When both variables are averaged in two-year windows, R2=0.88 and stdev=0.007. These results quantitatively evidence an excellent predictive power of relationship (8) and (9) over the entire period between 1965 and 2003. If to recall that the period between 1983 and 1986 is poorly modeled due to the turbulence in the labor force definitions, one can expect that further improvements in the accuracy of the labor force measurements are possible, which might lead to a higher confidence as presented by statistical estimates. Regression of the cumulative curves is characterized by R2=0.999 and stdev=0.0011. Thus, one can precisely replace the inflation cumulative curve or, in other words, inflation index with that obtained from the labor force measurements. This substitution is a reciprocal one– it is possible to exactly estimate the total increase in the labor force between 1965 and 2003 by measuring the GDP inflation. Currently, inflation is Austria, as represented by the NAC GDP deflator, is close to 2%, as explicitly defined by the monetary policy adopted by the European System of Central Banks (ECB, 2004) and correspondingly by the Austrian National Bank (OeNB, 2005). The inflation obeys the revealed dependence on the labor force change as well. Hence, the new monetary policy oriented to price stability does not disturb the relationship describing the last 40 years of the Austrian inflation. Linear relationship (9) obtained for the current period implies that one per cent of the labor force change produces inflation of 2%=1.25%+0.75%, where 0.75% is the persistent inflation level, i.e. the inflation existing even when no labor force change is observed. Thus, an annual change in labor force of +1% produces the OeNB’s target inflation. Obviously, labor force change in Austria is affected not only by the OeNB's monetary policy. There are demographic, social, political, economic processes behind the change. Therefore, it is probable that the labor force will change in future in a way not matching the target inflation. In the case of a decrease in the labor force, a deflationary period is probable starting from -0.6% annual labor force change rate, as relationship (9) defines: 1.25*(-0.006) +0.0075=0.

Ivan O. Kitov

282

Labor force participation rate is stable in Austria during the last ten years and close to 59% (the OECD definition). If this tendency holds in future, the labor force will be defined by the level of the population of 15 years of age and above. Statistics Austria (2006) provides a good population projection and corresponding approximation for this variable as a sum of the population aged between 15 and 60 years and that above 60 years as presented separately: Year 2004 2010 2015

From 15 to 60 years of age 5059 5112 5120

>60 years of age 1789 1928 2053

Total 6848 7040 7173

0.12 CPI (NAC) dLF/LF (predicted)

inflation

0.08

0.04

0.00 1955

1965

1975

1985

1995

2005

calendar year

1.8

cumulative inflation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1.6

CPI (NAC) dLF/LF (predicted)

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1955

1965

1975 1985 calendar year

1995

2005

Figure 9. Comparison of the observed (NAC CPI) and predicted inflation in Austria. The upper frame displays annual readings and the lower one – cumulative inflation since 1960. Notice a major change in labor force definition between 1981 and 1987 (OECD, 2005). The periods before and after 1986 are described separately.

Inflation, Unemployment, Labor Force Change in European Counties

283

The population above 15 years of age will grow by 2.8% between 2004 and 2010 and by another 1.9% during the following five years. The mean growth rate of 0.4% per year provides a 1.2% inflation growth rate during the next ten years. The value is below the 2% target and the Austrian monetary authorities have to provide an approximately 0.8% average annual growth in the participation rate, i.e. from 59% in 2005 to 67% in 2015. Otherwise, the target inflation rate will not be matched.

0.24 GDP deflator (EUR) 0.20

dLF/LF (predicted)

0.16

inflation

0.12 0.08 0.04 0.00 1955 -0.04

1965

1975

1985

1995

2005

2015

-0.08 calendar year

2.5 GDP deflator (EUR) dLF/LF (predicted) cumulative inflation

2.0

1.5

1.0

0.5

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.0 1955

1965

1975 1985 calendar year

1995

2005

Figure 10. Comparison of the observed (EUR GDP deflator) and predicted inflation in Austria. The upper frame displays annual readings and the lower one – cumulative inflation since 1965. Notice a major change in labor force definition between 1981 and 1987 (OECD, 2005). The periods before and after 1986 are described separately.

Figures 9 and 10 show the results of a similar analysis for the other two measures of inflation: the NAC CPI and the GDP deflator calculated at the exchange rate to Euro. The NAC CPI readings are very close to those obtained for the NAC GDP deflator. Therefore, coefficients in relationship (1) are also close: A1=2, A2=0.0315 before 1986, A1=1.35, A2=0.0095 after 1986. The linear relationships for the EUR GDP deflator readings are

Ivan O. Kitov

284

characterized by larger coefficients: A1=4, A2=0.047 before 1986, A1=2.5, A2=0.00 after 1986. Results of the regression analysis are presented in Table 1. The CPI time series is characterized by stdev=0.022 for the period between 1965 and 2003, which is equal to the standard deviation related to the NAC GDP deflator series. At the same time, a linear regression of the CPI NAC against the predicted inflation results in a lower R2=0.60 and larger stdev=0.014. Therefore, even small differences between the GDP deflator and CPI, as defined by correlation coefficient 0.92, result in a large difference in statistical estimates. The Eurostat GDP deflator demonstrates a higher scattering: stdev=0.046 for the period between 1965 and 2003. Correspondingly, R2=0.66 and stdev=0.027, i.e. much poorer than the results shown by the NAC GDP deflator. Especially, it concerns the high standard deviation, which is by a factor of 2.5 larger than that for the NAC GDP deflator. However, if normalized to standard deviation of corresponding inflation series, i.e. to 0.014/0.022=0.64 and 0.027/0.046=0.59, the relative volatility does not differ much in the cases of the NAC and Eurostat GDP deflators. The two-year moving average technique provides a gradual improvement on the results of the regression of the annual values, as presented in Table 1. It is confirmed above that both inflation and unemployment in Austria are linear functions of labor force change rate with no time lag. There is no need to apply generalized relationship (3) to the data in order to balance some potential disturbances, which might be induced by the ESCB fixed inflation rate. Relationships (1) and (2) work excellent separately and its sum should also work well. There is another issue associated with usage of (3), however. Measurement errors make prediction of the annual time series unreliable during the periods of weak changes in defining parameters, i.e. when the change in labor force is lower than the accuracy of the labor force measurements. In such a situation, the observed change is statistically insignificant, as we have obtained for the unemployment. Relationship (3) provides a potential way to improve the match. All the involved variables have almost independent measurement errors. Thus, one can expect an additional destructive interference of the errors when the variables are used together, such as relationship (3) defines. 0.12 GDP deflator (NAC) 1.2*dLF/LF+0.066-UE

inflation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.08

0.04

0.00 1955

1965

1975

1985

calendar year

Figure 11. (Continues)

1995

2005

Inflation, Unemployment, Labor Force Change in European Counties

285

2.0

cumulative inflation

1.6

1.2

0.8

GDP deflator (NAC) 1.2*dLF/LF+0.066-UE (NAC) (AMS) 0.9*dLF/LF+0.074-UE (NAC) (AMS)

0.4

0.0 1955

1965

1975 1985 calendar year

1995

2005

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 11. Comparison of the observed (NAC GDP deflator) and predicted inflation in Austria. The upper frame displays annual readings and the lower one – cumulative inflation since 1960. The predicted inflation is a linear function of the labor force change and unemployment as defined by relationship (3). Notice the absence of the major change in 1986 due to effective compensation of the labor force change by the unemployment. There is a slight discrepancy started in 1994 with corresponding change in linear coefficient and intercept, as described by the relationships in the lower right corner of the lower frame.

Figure 11 displays the observed and predicted inflation. The former is presented by the NAC GDP deflator. The latter is obtained using relationship (3) with coefficients computed for the case of the predictor based on the NAC (AMS+HSV) labor force and the AMS unemployment. This representation of inflation is less sensitive to the changes in the unemployment and labor force definitions. In fact, the unemployment is a part of the labor force and any change in unemployment is automatically included into the labor force change, but the changes in the unemployment and employment definitions are not synchronized. The latter observation makes the changes in the labor force and unemployment also to be asynchronous. In any case, the agreement between the predicted and observed curves is remarkable over the whole interval between 1965 and 2003. There is a small deviation starting in 1994, however, as the cumulative curves in Figure 11 show. One can explain the discrepancy as associated with the change in the employment definition in 1994 - the time criterion was decreased to 1 hour, as mentioned above. Obviously, the change resulted in the increase of the overall labor force level and corresponding change rate. In addition, the labor force survey procedures, including population coverage and timing, were changed and Statistik Austria became responsible for the labor force estimates in line with the Eurostat and ILO definitions since 1994 (Statistik Austria, 2004). These modifications could result in the observed change of the inflation sensitivity to the labor force change due to the introduction of new units of measurements. So far, the inflation in Austria (in all the three representations) was modeled for the period after 1986 separately. The difference between units of measurement in the 7-year long interval between 1987 and 1994 and during the nine years after 1994 was so weak that is could not be resolved using the short intervals. The difference was balanced in (9), i.e. a small overestimation of inflation in the first interval was compensated by a small underestimation

Ivan O. Kitov

286

during the second period. The generalized approach has a higher resolution because of longer baselines: 29 years between 1965 and 1994 and 9 years between 1994 and 2003. Therefore, the deviation between two branches has been revealed and successfully modeled by the introduction of new coefficients in the generalized linear relationships after 1994:

π(t)=1.2*dLF(t)/LF(t)-UE(t)+0.066 (1965≤ t ≤ 1994) (10) π(t)=0.9*dLF(t)/LF(t)-UE(t)+0.0074 ( t ≥ 1995)

(11)

The predicted values of inflation according to relationships (10) and (11) with the NAC labor force and the AMS unemployment are used as a predictor for a linear regression of the NAC GDP readings. For the annual readings between 1965 and 2003, Table 1 lists the following values: R2=0.86 and stdev=0.008. This is an outstanding result considering the uncertainty associated with the measurement of the inflation, labor force, and unemployment. The predictor explains 86% of inflation variation including the periods of high and low inflation, and the periods of intensive growth and decrease of the inflation, as presented in Figure 11. The choice of 1965 is arbitrary and an extension of the period to 1960 does not change R2 much - it drops to 0.84. Standard error of the regression is only 0.008. The slight improvement in statistical description related to usage of (3) instead of (1), as expressed by R2 increase from 0.81 to 0.86 for the annual readings, is apparently related to a stabilizing role of the unemployment readings. Averaging in two-year moving windows provides almost no additional improvement in statistical estimates. When the predicted values are averaged, R2=0.87 and stdev=0.008. When both observed and predicted readings are averaged, R2=0.91 and stdev=0.007. In any case, generalized relationship (3) provides a very accurate description of inflation in Austria between 1960 and present. In this Section, we have scrupulously considered details of the procedures related to measurements in order to obtain the best agreement between the observed and predicted values. As a result we have obtained a very accurate, in statistical sense, description of unemployment and inflation in Austria during the last 45 years. In addition, a prediction of inflation for the next ten years has been computed using population projections provided by Statistik Austria. We have also learned several important lessons for future investigations:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

• •



• •

Data related to labor force and unemployment needs special consideration because of numerous revisions of definitions and procedures. There is not break or any other discontinuity in inflation behavior around its peak and trough values. Linear dependence of inflation and unemployment on labor force change is very consistent and reliable over time. The larger is the amplitude of inflation (unemployment) change the better is its prediction based on labor force change. An alternative opportunity to increase resolution is to improve accuracy of corresponding measurements. The GDP deflator is the best representation of inflation, at least in Austria and the USA. The generalized linear relationship linking together inflation, unemployment, and labor force potentially provides an additional improvement in prediction of inflation.

Inflation, Unemployment, Labor Force Change in European Counties •

287

Quantitatively, the best fit model of inflation in Austria is characterized by R2=0.86 and RMSFE=0.008, as obtained for the period between 1965 and 2003.

Concluding this Section, it is worth noting that Austria provides a good opportunity not only to model the dependence between inflation, unemployment, and labor force change, but also evaluate consistency of various definitions of the studied variables. Despite the documented changes in units of measurements, the variables do not lose their intrinsic links persistent through the last 45 years. There is no reason to think that these bounds will disappear in the near future.

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

FRANCE France is characterized by an outstanding productivity and has the largest GDP per working hour among large developed economies, as presented by the Groningen Growth and Development Center and Conference Board (2006). At the same time, real economic performance in France is far from a stellar one during the last twenty-five years with the mean annual real GDP growth of 2%. Therefore, France is an example of an economy different in many aspects from those in the USA, Japan, and Austria. This is especially important for the concept we examine. Linear relationships (1) and (2) with country specific coefficients are supposed to be intrinsic ones to any developed economy and to express deep socio-economic bounds between people. In turn, the linear relationship for inflation does not depend on such parameters of real economy as output gap, marginal labor cost, and so on. OECD (2005) provides relatively long time series for the variables involved in the study: GDP deflator (between 1971 and 2004), CPI based on the national currency (between 1956 and 2004), labor force level (between 1956 and 2004), unemployment rate (between 1960 and 2004), working age population (between 1960 and 2004), and labor force participation rate (from 1960 to 2004). Inflation estimates are also available at the web-sites of Eurostat- the Euro based CPI between 1979 and 2005, and at the National Institute for Statistics and Economic Studies (INSEE)- http://www.incee.fr. There are three different measures of inflation in France shown in Figure 12: the OECD CPI, the CPI based on the Euro, and the OECD GDP deflator. The time series for CPI and GDP deflator published by the INSEE (2006) almost coincide with those provided by OECD and Eurostat and start from 1983 as a rule. Therefore, they are not presented in the Figure. The OECD GDP deflator and CPI inflation are very similar with only relatively small discrepancies during some short intervals. These curves show a high inflation rate between 1975 and 1985 and a gradual decrease to the current level close to 2%. Only two measures of inflation from the three available are modeled in the study. The Eurostat CPI based on the Euro is limited in time and volatile due to the exchange rate fluctuations. So, this time series is neglected. GDP deflator is probably the best variable reflecting inherent links between inflation and labor force change, as found for the USA, Japan, and Austria. So, our primary goal is to model the GDP deflator provided by the OECD. The OECD CPI time series is also predicted for a comparison. CPI is of a lower interest for our study because it hardly represents a valid economic parameter to model in our framework.

Ivan O. Kitov

288 0.20

GDP deflator (OECD) CPI NAC (OECD)

0.16

CPI EURO (Eurostat) inflation

0.12 0.08 0.04 0.00 1955

1965

1975

1985

1995

2005

-0.04 calendar year

Figure 12. Comparison of various measures of inflation in France. There are three time series: GDP deflator and CPI based on national currency obtained from the OECD web-site and CPI inflation based on the exchange rate to Euro, as given by Eurostat. The GDP deflator and CPI NAC time series start from 1971 and 1956, respectively. The CPI EURO starts from 1979.

0.020 0.015 0.010

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

dLF/LF

0.005 0.000 1950 -0.005 -0.010 -0.015 -0.020

1960

1970 1980 dLF/LF (OECD) 0.0091

1990

2000

2010

0.0048 0.0084 dLF/LF (Eurostat) calendar year

Figure 13. Labor force change rate in France as given by the OECD and Eurostat. The OECD time series starts from 1956 and the Eurostat’s one - in 1983. The latter curve is characterized by higher fluctuations. The mean growth rates of the OECD labor force are also shown for three different periods as defined in the text. Notice a period of strong growth started in 1996.

Inflation, Unemployment, Labor Force Change in European Counties

289

Figure 13 displays the principal variable of the model – labor force change rate, dLF/LF, in France for the period between 1956 and 2004. The Eurostat web-site also publishes time series for the number of unemployed (1983 through 2004) and employed (1978 through 2004) separately. The sum of the two series gives a labor force estimate between 1983 and 2004 also presented in Figure 13. Because of the limited interval spanned by the Eurostat labor force series and its high volatility of unknown origin only the OECD labor force readings are used to predict unemployment and inflation rate. The OECD labor force series can be split into several distinct periods. From 1958 to 1963, a very low and even negative change rate was observed, which is potentially associated with statistical definitions or methodology of measurements in the past. From 1963 through 1981, a strong labor force growth was measured with the mean annual rate of +0.94%. A relatively slow growth between 1982 and 1995 with the mean annual rate of +0.48% is followed by a new period of a strong growth started in 1996 with the mean annual rate of +0.84%. According to the linear relationships under study, inflation and unemployment have to evolve in the same way. It is interesting that the recent increase in the labor force has not been accompanied by any visible change in the inflation, as Figure 12 evidences. 57 LFPR (OECD) 56.5

LFPR, %

56 55.5 55 54.5 54 1965

1970

1975

1980

1985

1990

1995

2000

2005

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

calendar year

Figure 14. Labor force participation rate in France as defined by OECD for the population above 15 years of age. There was a long period of a gradual decrease in LFPR between 1975 and 1995 when the lowermost level was measured -54.4%. In 1996, a period of strong growth started with the average annual increment of ~0.2%. In 2004, the LFPR reached 55.7%.

Taking into consideration a gradual decrease in the rate of working-age population growth in France (OECD, 2006), one can expect an intensive growth of labor force participation rate (LFPR) started in 1996 to be responsible for the rapid increase in the labor force. Figure 14 proves that the expected strong growth in the LFPR has been an actual and consistent one since 1996. During the previous forty years, the participation rate in France was as low as 55% compared to 59% in the USA and above 60% in Japan. So, it is natural that the participation rate in France has started to grow at some point.

Ivan O. Kitov

290

0.20 0.15

UE

0.10 0.05 0.00 1965

1970

1975

1980

1985

1990

1995

2000

2005

UE (OECD) 0.165-13*dLF/LF (OECD) 0.195-11*dLF/LF (OECD)

-0.05 -0.10

calendar year

3.0

cumulative UE

2.5

UE (OECD) 0.165-13*dLF/LF (OECD) 0.195-11*dLF/LF (OECD)

2.0 1.5 1.0 0.5

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.0 1965

1970

1975

1980

1985 1990 calendar year

1995

2000

2005

Figure 15. Comparison of the observed and predicted unemployment in France: the upper frame for the annual readings and the lower for the cumulative values of the unemployment since 1970. There is no time lag between the unemployment and labor force change. Notice the discrepancy started in 1996 – the year when the labor force participation rate started to grow fast, and two years after the Banque de France obtained a new status and introduced a new monetary policy - price stability. The predicted unemployment is about twice as low as the observed one, as presented in the upper panel. The period after 1996 can be described by a different dependence of the unemployment on the labor force with a higher intercept (0.195) and a lower (in absolute value) linear coefficient (-11), as given in the legend. Results of corresponding regression analysis are given in Table 2.

The current period of the labor force growth almost coincides with the establishment of a new entity of the French national bank, Banque de France, as an independent monetary authority having a fixed target value of inflation rate. In 1993, the European System of Central Banks (ESCB) cardinally changed its approach to inflation managing – the main target is currently to reach price stability at a level near 2% of annual growth (ECB, 2004). Whatever reasons are put forth to justify the new approach they are not theoretically and empirically sound, i.e. there are no reliable evidences for the assumptions underlying the

Inflation, Unemployment, Labor Force Change in European Counties

291

current concepts of inflation to be valid. The most recent models rely on exogenous shocks as the driving force behind inflation (Rudd and Whelan, 2005; GG (1999); Gali at al., 2002, 2005; Hall, 2005). Such shocks are inherently unpredictable and uncontrollable in time and amplitude. So, the approach based on an aggregated opinion of central bankers and economists is barely valid in view of unpredictable exogenous shocks. Our concept provides a clear understanding of the nature of these exogenous forces and thus a control over unemployment and inflation. For France, as for the US, Japan, and Austria we use the same procedure to fit annual and cumulative inflation and unemployment readings by linear functions of labor force change rate. The most sensitive to coefficients in relationship (1) is a cumulative curve. Even a small systematic error in predicted amplitude cumulates to a high value when aggregated over thirty-five years. Predicted and measured annual and cumulative curves for the OECD unemployment rate between 1970 and 2004 are presented in Figure 15. The predicted curve in Figure 15a is obtained from the OECD labor force change rate and shows large-amplitude fluctuations around the measured unemployment curve. This is a result of a very large coefficient in the relationship between UE(t) and dLF(t)/LF(t):

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

UE(t)=0.165-13*dLF(t)/LF(t)

(12)

Linear coefficient in (12) amplifies labor force change and any measurement error in the labor force by a factor of 13. This coefficient is also a negative one, i.e. any increase in labor force is converted in a synchronized (no time lag between the labor force and the unemployment change) and 13-time amplified drop of the unemployment rate in France. On the other hand, in the absence of any growth in the labor force the unemployment rate reaches a 16.5% level. (The high sensitivity of the unemployment to the labor force change provides a good opportunity to control the unemployment through a reasonable labor market policy. At the same time, the high sensitivity demands any such a policy to be thoroughly and deeply discussed before implementation.) From 1970 through 1995, there is a good agreement between the observed and predicted curves. The period before 1970 is neglected in the study. As we have learned from the case of Austria, the earlier period is characterized by some changes in the methodology of labor force survey and/or the definitions of labor force itself. The model period after 1970 is also in line with many other studies devoted to the modeling of various Phillips curves in European countries, where the period before 1970 is rarely covered (see Angelini et al. (2001); Canova, F., (2002), Cristadoro et al. (2001); Espasa et al. (2002); Gali et al. (2001), Ihrig and Marquez (2003); Marcellino et al. (2001); Hubrich (2005), among others). The observed unemployment curve gradually elevates from 3% in 1970 to almost 10% in 2004, with the predicted curve fluctuating around the observed one with an amplitude reaching 0.1. In 1996, a sudden drop in the predicted curve started a major deviation from the measured curve. The predicted curve falls from 10% in 1996 to 4% in 2003. It is possible to compute the total number of unemployed people who could get paid jobs under the theoretical curve in excess of the measured number: 4%*27,000,000~1,000,000 per year. Thus, approximately one million less than expected persons have job in France every year since 1996.

Ivan O. Kitov

292

There are three potential explanations of the deviation. The first one is associated with a probable change in unit of measurements, as has been found for Austria. There is no documented change in the labor force and unemployment definitions in the 1990s in France, however. Therefore, this explanation is not working for France. The second possibility is that coefficients in relationship (12) were changed in 1996 by some external forces to new values, but the linear link to labor force is retained. We have discussed such a situation is Section 1 and suggested that generalized relationship (3) has to replace individual relationships (1) and (2). We will examine this assumption in detail later on. The third explanation is that there is no linear relationship between unemployment, inflation, and labor force and the deviation started in 1996 is unpredictable and spontaneous. A standard linear regression analysis is carried out for the period between 1970 and 1995. The OECD unemployment rate is a dependent variable and the theoretical curve is used as a predictor. Table 2 lists some results of the analysis. The measured time series is characterized by stdev=0.032. As expected from the high volatility in the annual readings of the predictor (see Figure 15a) corresponding regression gives R2=0.48 with stdev=0.023. Hence, the annual time series is poorly predicted. Figure 15b represents a cumulative view on the predicted and observed unemployment in France. This view emphasizes the deviation started in 1996. The cumulative curves provide a good way to demonstrate that the oscillations in the predicted curve are induced by some uncorrelated measurement errors, not by actual change. At the same time, the curves definitely show some problematic years in the beginning of the period. Overall, the curves almost coincide and confirm the reliability of the linear relationship between UE(t) and dLF(t)/LF(t). A linear regression of the cumulative curves gives R2=0.998 and stdev=0.028.

0.20

0.15

UE (OECD) 0.167-13*dLF/LF (OECD) 0.205-11*dLF/LF (OECD)

UE

0.10

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.05

0.00 1965

1970

1975

1980

1985

1990

1995

2000

2005

-0.05 calendar year

Figure 16. Same as in Figure 15a, but with the predicted curve smoothed by a 2-year moving average. There is a better agreement between the observed and predicted time series, especially between 1978 and 1995. Notice a slightly higher intercept 0.167 instead of 0.165 for the annual readings in Figure 15.

Inflation, Unemployment, Labor Force Change in European Counties

293

Table 2. Results of linear regression analysis for France Period 1970-1995 1970-1995 1970-1995 1970-1995 1970-1995 1971-1999 1971-1999 1971-1999 1971-1999 1971-1999 1977-1999 1970-1999 1970-1999 1970-1999 1970-1999 1977-1999

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

1971-1999

Dependent variable

R2

Predictor

A

B

annual unemployment (OECD) annual unemployment (OECD) annual unemployment (OECD) annual unemployment (OECD) cumulative unemployment (OECD)

annual dLF(t)/LF(t) (OECD) 2-year moving average dLF(t)/LF(t) (OECD) 5-year moving average dLF(t)/LF(t) (OECD) cumulative dLF(t)/LF(t) (OECD)

0.45 (0.10) 0.71 (0.08) 1.00 (0.07) 1.01 (0.009)

0.04 (0.008) 0.02 (0.006) 0.000 (0.005) 0.04 (0.001)

GDP deflator (OECD) annual GDP deflator (OECD) annual GDP deflator (OECD) annual GDP deflator (OECD) annual GDP deflator (OECD) 7-year moving average GDP deflator (OECD)

annual dLF(t-4)/LF(t-4) (OECD) 2-year moving average dLF(t-4)/LF(t-4) (OECD) 3-year moving average dLF(t-4)/LF(t-4) (OECD) 7-year moving average dLF(t-4)/LF(t-4) (OECD) 7-year moving average dLF(t-4)/LF(t-4) (OECD)

0.48 (010) 0.74 (0.08) 0.94 (0.06) 1.09 (0.07) 0.97 (0.03)

0.03 (0.008) 0.01 (0.006) 0.001 (0.004) 0.01 (0.005) 0.001 (0.003)

CPI inflation (OECD) annual CPI inflation (OECD) annual CPI inflation (OECD) annual CPI inflation (OECD) annual CPI inflation (OECD)

annual dLF(t-4)/LF(t-4) (OECD) 2-year moving average dLF(t-4)/LF(t-4) (OECD) 3-year moving average dLF(t-4)/LF(t-4) (OECD) 7-year moving average dLF(t-4)/LF(t-4) (OECD)

0.50 (010) 0.81 (0.09) 1.00 (0.08) 1.15 (0.09)

0.03 (0.008) 0.01 (0.007) 0.000 (0.006) 0.01 (0.007)

annual dLF(t-4)/LF(t-4)UE(t-4) (OECD) 2-year moving average dLF(t-4)/LF(t-4)-UE(t-4) (OECD) 3-year moving average dLF(t-4)/LF(t-4)-UE(t-4) (OECD) 7-year moving average dLF(t-4)/LF(t-4)-UE(t-4) (OECD) 7-year moving average dLF(t-4)/LF(t-4)-UE(t-4) (OECD)

0.89 (0.06)

0.004 (0.004)

0.88

0.014

0.91 (0.06)

0.003 (0.004)

0.87

0.015

0.97 (0.05)

0.000 (0.003)

0.93

0.011

1.03 (0.05)

0.003 (0.004)

0.93

0.011

0.99 (0.02)

0.000 (0.001)

0.99

0.004

1971-2004

GDP deflator (OECD) annual GDP deflator (OECD)

1971-2004

annual GDP deflator (OECD)

1971-2004

annual GDP deflator (OECD)

1971-2004

annual GDP deflator (OECD)

1977-2004

7-year moving average GDP deflator (OECD)

stdev 0.032

0.48

0.023

0.75

0.016

0.90 0.99 8

0.010 0.028 0.042

0.47

0.031

0.74

0.022

0.91

0.013

0.89

0.014

0.97

0.006 0.043

0.48

0.031

0.74

0.022

0.85

0.017

0.83

0.018 0.042

Moving average is thoroughly used in this study in order to obtain a better agreement between the observed and predicted curves. This technique effectively suppresses the noise associated with measurement errors. Figure 16 displays the annual measured curve and that obtained by a 2-year moving average as applied to the predictor. There is a significant improvement in the predictive power of relationship (12), especially between 1978 and 1995 -

Ivan O. Kitov

294

the curves practically coincide. The improved overall agreement is also reflected in a higher R2=0.75 and lower stdev=0.016, as presented in Table 2. When a 5-year moving average is applied to the predictor, R2 increases to 0.90 and stdev falls to 0.010. Hence, moving average is very efficient in noise suppression and provides an explanation of about 90% of variation in the unemployment rate. One can not expect any further improvement beyond the level associated with some intrinsic measurement uncertainty, however. More accurate measurements of the labor force are necessary for obtaining a higher correlation between the observed and predicted time series. According to relationship (2), inflation is also a linear function of labor force change. Figure 17 illustrates the fit between observed (the OECD GDP deflator) and predicted inflation. Figure 17a compares the measured annual values to those obtained according to the following relationship:

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

π(t)=17*dLF(t-4)/LF(t-4)-0.063

(13)

where π(t) is the inflation at time t, LF(t-4) is the labor force four years before. Thus, there is a four years lag in France between the labor force change and corresponding reaction of the inflation. The linear coefficient 17 indicates that the inflation is aslo very sensitive to the labor forced change. The intercept -0.063 means that a positive labor force change rate has to be retained in order to avoid deflation. The threshold for a deflationary period is a labor force change rate of 0.0037(=0.063/17) per year. Actual change rate was consistently higher than the threshold value over the studied period, as Figure 13 demonstrates. The predicted inflation has been rapidly increasing since 2000 according to the labor force increase started in 1996 and the four-year lag. The observed inflation has been fluctuating near 2% since 1995, however. This inflation rate is the one defined by the ECB (2004) and Banque de France (2005) as the target of monetary policy. Therefore, one might suppose that the observed inflation is fixed by some special measures applied by the ESCB such as a monetary supply constrained to real GDP growth plus 2%. The effect of the inflation rate fixed by force is expressed in the observed deviation of the predicted unemployment and inflation from those measured in France. The unemployment reacts immediately to the labor force increase started in 1996. The inflation reacts four years later. In the absence of the fixed inflation rate or price stability, the observed inflation and unemployment would follow their predicted paths: in 2004, 9% inflation would be accompanied by 4% unemployment. Since the discrepancy between the observed and measured inflation starts in 2000, a linear regression analysis is carried out for the period between 1971 and 1999. The GDP deflator is a dependent variable and a predictor is obtained according to relationship (13). Some results of the analysis are presented in Table 2. Standard deviation of the actual time series for the studied period is 0.042. The regression of the annual readings is characterized by R2=0.47 and stdev=0.031. R2 is a low one and close to that obtained for the unemployment. In both cases, the reason for the low correlation is low accuracy of labor force measurements accompanied by the high sensitivity of the predicted values to the labor force change rate. Moving average provides a more accurate representation of the labor force change rate. For the four-year lag, as observed in France, even a 7-year moving window applied to the predictor does not include the labor force readings contemporaneous to the predicted

Inflation, Unemployment, Labor Force Change in European Counties

295

inflation. Therefore, the lag guarantees a natural "out-of-sample" inflation forecast at various time horizons - from 1 year to 4 years. Table 2 lists standard errors (deviations) and R2, which are obtained by linear regressions with various moving averages. Obviously, the larger is forecasting horizon, i.e. the shorter is corresponding averaging window, the larger is the forecast uncertainty. On the other hand, there must be some optimal width of moving windows. For a very wide window, the readings at the left (early) side of the window introduce some additional noise rather than improve the modeled leading value. In fact, for a 2-year moving average applied to the predicted inflation R2=0.74 and stdev=0.022, for a 3year window R2=0.91 and stdev=0.013, and for a 7-year window R2=0.89 with stdev=0.014. So, the best result is obtained for the 3-year moving average, which explains 91% of variation in the original inflation time for the period between 1971 and 1999. Figure 17b demonstrates the outstanding predictive power of the 3-year moving average. 0.20 0.15

inflation

0.10 0.05 0.00 1965 -0.05

1970

1975

1980

1985

1990

1995

2000

2005

GDP deflator (OECD) 17*dLF(t-4)/LF(t-4)-0.063 (OECD) 9*dLF(t-4)/LF(t-4)-0.060 (OECD)

-0.10 -0.15

calendar year

(a) 0.15 GDP deflator (OECD) 3-year average (predicted)

inflation

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

0.10

0.05

0.00 1965

1970

1975

1980

1985

1990

1995

2000

2005

-0.05 calendar year

(b) Figure 17. Continues on next page.

Ivan O. Kitov

296

2.5

cumulative inflation

2

GDP deflator (OECD) 17*dLF(t-4)/LF(t-4)-0.063 (OECD) 9*dLF(t-4)/LF(t-4)-0.060 (OECD)

1.5

1

0.5

0 1965

1970

1975

1980

1985

1990

1995

2000

2005

calendar year

(c)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 17. Comparison of the observed and predicted inflation, as defined by the relationship given in the text and in the legend (OECD GDP deflator) in France: a) annual readings, b) real annual readings and predicted readings smoothed by a 3-year moving average, c) cumulative inflation since 1970. The inflation lags by four years behind the labor force change. Notice the discrepancy started in 2000 – four years after the start of the labor force. The predicted inflation oscillates around 10% after 2000. The period after 1999 can be described by a different dependence of the GDP deflator on the labor force with a slightly larger intercept (-0.060 instead of -0.063) and a much lower linear coefficient (9 instead of 17), as given in the legend.

One can potentially reach an additional improvement on the results obtained with the 3year moving average by using more powerful techniques for noise suppression. This is not the purpose of this study, however. We just reveal inherent links between unemployment, inflation, and labor force at a high level of confidence, as represented by R2. Further improvements in R2 related to the annual readings above 0.91 hardly deserve any additional effort and potentially fall into a conflict with the level of uncertainty in the inflation and labor force measurements. In our framework, the residual difference between the observed and predicted readings is related solely to measurement errors. In France, labor force is measured with an uncertainty, which is not appropriate to the modeling of the more accurately measured unemployment and inflation. One-year long measuring baseline is not enough for obtaining a reliable estimate of labor force change rate. Moving average takes an advantage of a longer baseline for the calculation of the change rate and provides a substantial increase in the predictive power of relationships (12) and (13). Therefore, a longer basic time unit will potentially result in a higher accuracy of corresponding measurements and in a better correlation between the modeled variables. Table 2 supports this assumption by an example of a regression of 7-year moving averages of the observed and predicted inflation: R2=0.97 and stdev=0.006. Hence, if to replace the current one-year basic interval with a seven-year long one, the inflation prediction would be as accurate as 0.006 for the period between 1971 and 1999. The same effect might be obtained by improvements in the current measuring procedures, however. There is a direct trade-off between the efforts invested in such improvements and the

Inflation, Unemployment, Labor Force Change in European Counties

297

accuracy of predicted inflation and unemployment. Since the problem of low measurement accuracy is a resolvable one we leave it to appropriate agencies. Figure 17c compares two cumulative curves as obtained for the measured and predicted inflation. There is a good agreement during the years between 1971 and 1999. We do not provide in Table 2 statistical estimates for the cumulative curves of inflation in France. Obviously, R2 has to be very close to 1.0 and standard deviation is similar to that for the case of the annual readings. The cumulative curves evidence that the labor force cumulative change provides a precise measure of the inflation index growth and vice versa. 0.25 CPI (OECD) 16*dLF(t-4)/LF(t-4)-0.054 (OECD)

0.20 0.15

inflation

0.10 0.05 0.00 1965 -0.05

1970

1975

1980

1985

1990

1995

2000

2005

-0.10 -0.15 calendar year

(a) 0.15 CPI (OECD) 3-year average (predicted)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

inflation

0.10

0.05

0.00 1965

1970

1975

1980

1985

1990

1995

2000

2005

-0.05 calendar year

(b) Figure 18. Continues on next page.

Ivan O. Kitov

298 2.5 CPI (OECD) cumulative inflation

2.0

16*dLF(t-4)/LF(t-4)-0.054 (OECD)

1.5

1.0

0.5

0.0 1965

1975

1985

1995

2005

calendar year

(c)

Copyright © 2009. Nova Science Publishers, Incorporated. All rights reserved.

Figure 18. Same as in Figure 17, for the observed inflation expressed by the OECD CPI.

Figure 18 and Table 2 represent results of a similar analysis as applied to the OECD CPI inflation. The actual time series is characterized by standard deviation of 0.043 for the period between 1971 and 1999, which is just marginally higher than that for the OECD deflator during the same period. The best predictor for the annual readings is also obtained with a 3year moving average: R2=0.85 and stdev=0.017. These values indicate a slightly lower predictive power of the labor force change rate compared to that obtained for the GDP deflator. This is a common situation for the countries studied so far. GDP deflator is a consistently better measure of inflation as related to labor force change rate. Caveats in CPI definition and measuring procedures are well known and have been actively discussed since the Boskin’s report (1998). Obviously, the problems associated with the uncertainty in CPI measurement lead to the poorer performance of the labor force as a predictor. Having discussed the potentially resolvable problems associated with the uncertainty in labor force measurements, we start to tackle the problem associated with the discrepancy between the observed and predicted curves. This problem is a critical one for the concept. Potentially, the discrepancy is associated with the new monetary policy first applied by the Banque de France in the beginning of the 1990s. The policy of a constrained money supply, if applied, could obviously disturb relationships (12) and (13). New coefficients in the linear relationships are computed and presented in relevant Figures for the periods after 1995 for the unemployment after 1999 for the inflation, respectively. The coefficients are unreliable, however, due to the shortness of observations, but definitely different from the old ones. Probably, one could conclude that the Banque de France has created some new links between the unemployment, inflation, and labor force. Our assumption is a different one. Money supply in excess of that related to real GDP growth is completely controlled by the demand of growing labor force because the excess is always accommodated in a developed economy through employment growth, which causes inflation. The latter serves as a mechanism which effectively returns personal income distribution (normalized to total population and nominal GDP growth) in the economy to its original shape (Kitov, 2006a,d). The relative amount of money that the economy needs to accommodate a given relative labor force increase through employment is constant through

Inflation, Unemployment, Labor Force Change in European Counties

299

time in corresponding country but varies among developed countries. This amount has to be supplied to the economy, however. Central banks are responsible for this process. In the USA and Japan, central banks provide adequate procedures for money supply and individual dependence on labor force change does not vary with time both for inflation and unemployment. The ESCB limits money supply to achieve price stability. In Austria, it does not affect the individual linear relationships because actual money supply almost equals the amount required by the observed labor force growth. For France, the labor force growth is so intensive that demands a much larger money input for creation of an appropriate number of new jobs. The 2% artificial constraint on inflation (and thus money supply) disturbs relationships (12) and (13). The labor force growth induces only an increase in employment, which accommodates the given 2% inflation instead of the 9% predicted inflation. Those people who enter the labor force in France in excess of that allowed by the target inflation have no choice except to join "the army of unemployed". Hence, when inflation is fixed, the difference between observed and predicted change in the inflation must be completely compensated by an equivalent change in unemployment in excess of the predicted one. Generalized relationship (3) mathematically describes this assumption. For France, generalized relationship is obtained as a sum of (12) and (13), which gives the following equation:

π(t)= 4*dLF(t-4)/LF(t-4)-UE(t-4)+0.095 (1971