Spain : Economic, Political, and Social Issues [1 ed.] 9781613241172, 9781606921081

Citation preview

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

SPAIN: ECONOMIC, POLITICAL AND SOCIAL ISSUES

Copyright © 2008. 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.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

SPAIN: ECONOMIC, POLITICAL AND SOCIAL ISSUES

PABLO N. COSTADA

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

EDITOR

Nova Science Publishers, Inc. New York

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

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.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

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. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Spain : economic, political, and social issues / Pablo N. Costada (editor). p. cm. Includes bibliographical references and index. ISBN 978-1-61324-117-2 (eBook) 1. Finance--Spain. 2. Spain--Economic conditions--1975- 3. Spain--Social conditions--1975- I. Costada, Pablo N. HG186.S6S617 2008 320.60946--dc22 2008033041

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

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

CONTENTS Preface Profile

Official Name:Kingdom of Spain

Chapter 1

The Effect of Government Ownership on Bank Profitability and Risk: the Spanish Experiment 9 Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

Chapter 2

Managers’ Discretionary Behavior, Earnings Management and Corporate Governance: An Empirical International Analysis Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

Chapter 3

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

vii

Does One Monetary Policy Fit All? The Determinants of Inflation in EMU Countries Melisso Boschi and Alessandro Girardi

Chapter 4

Banks as Shareholders: The Spanish Model of Corporate Governance Valentín Azofra-Palenzuela, Félix J. López-Iturriaga and Fernando A. Tejerina-Gaite

Chapter 5

How Much do Trade and Financial Linkages Matter for Business Cycle Synchronization? Alicia García Herrero and Juan M. Ruiz

Chapter 6

Secularization and Fertility: Evidence from Spain Pablo Brañas-Garza and Shoshana Neuman

Chapter 7

Sleeping Habits and Reported Sleep Quality Among School Adolescents in Cuenca, Spain M. A. García-Jiménez, F. Salcedo Aguilar, J. Bona García, F. M. Rodríguez Almonacid, E. De las Heras Martínez and R. Sánchez Honrubia

Chapter 8

Autonomy of Schools in Spain Immaculada Egido

Index Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

1

31

51 81

109 141

173

191 201

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

PREFACE Spain is a country whose development arouses particular interest in the world because of the huge increase in Spanish-speaking populations, as well as its economic successes and location between Africa and Europe as well as it rich historical tapestry. This new book presents chapters dealing with current topics of interest in Spain dealing with economic, political and social issues. Chapter 1 - This paper analyzes the effects on Spanish savings banks’ performance and risk when they shed their mutual structure to become government-owned banks. Such a situation arose in 1985 when Spain’s central government allowed regional parliaments to modify savings bank ownership regulations. Regional regulations increased government participation at the expense of depositors’ ownership. This regulatory change constitutes a natural experiment to study the consequences of government ownership on bank behavior. The results of our study suggest that enhanced government ownership leads to an increase in risk. This is particularly marked amongst those savings banks that most increased the weight of local and regional governments on their governance bodies. However, no variation in savings bank performance has occurred. The net result, therefore, is an increase in performance-adjusted risk. Chapter 2 - In the present debate about corporate governance, boards of directors play a more and more relevant role. From this point of view, the board of directors becomes an important mechanism for monitoring managers’ discretionary behavior. Among the several ways for measuring discretionarity, the authorshave focused on accrual decisions in order to manage earnings. This is precisely the main contribution of our paper: the authorshave tested the ability of the board of directors for monitoring by using the discretionary component of earnings management as an indicator of managers’ discretionarity. Using a sample of 450 non-financial companies from 10 OECD countries, the authorshave found empirical evidence showing the link between some features of the board of directors and the discretionary manipulation of financial statements carried out by managers. Specifically, and as most of the literature has pointed, our main result stresses the positive and robust impact of board size on earnings management. As far as other characteristics of boards of directors are concerned, our research does not lend conclusive support to the effect of board composition or meeting frequency on earnings management. Chapter 3 - This chapter aims at assessing the long-run determinants and the short-run dynamics of inflation in each country belonging to the European Monetary Union (EMU). Our work complements the recent literature on this topic for the Euro Area as a whole.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

viii

Pabo N. Costada

Detecting such determinants can be crucial in designing structural reforms acting as aside instruments of monetary policy in maintaining price stability. The empirical methodology consists of a re-interpretation of the structural cointegrating VAR approach, which allows for a structural long-run analysis of inflation determinants along with an accurate assessment of its short-run dynamics. The main conclusion emerging from the estimates is that not only the determinants of inflation differ in the countries belonging to the Euro Area, but also that costpush factors have a considerable role in explaining inflation in most of the countries examined. As a policy implication, a tight monetary policy pursued in those countries whose inflation is mainly driven by costs would result in a contraction of economic activity without exerting relevant effects on price dynamics. Chapter 4 - Compared to the Anglo-Saxon framework on which most of the financial research has focused, in many European corporate systems such as Spain, banks and other financial intermediaries are more important than capital markets. In addition to lending, banks usually own significant proportions of shares and sit at the board of directors. Our results for a sample of 142 non-financial firms between 1999 and 2002 show that banks play an outstanding role in the corporate governance of the Spanish firms. More specifically, due to the banking specialization in monitoring and control, the authorsfind that banks have a dual and very relevant influence on the performance of the firms. For bank-controlled firms, bank shareholdings can enable colluding interactions among banks and, thus, reduce the performance of the firm. On the contrary, for non-bank controlled firms, bank shareholdings improve the corporate governance and increase the performance of the firm. This effect is particularly significant for the firms in which bank scrutiny is more crucial: when the largest shareholder has voting rights in excess of his cash flow rights, and when the control of the largest shareholder can be more contested. Chapter 5 - The authorsestimate a system of equations to analyze whether trade and financial linkages influence business cycle synchronization directly or indirectly. The authorsuse a small, open economy (Spain) as benchmark for the results, instead of the US as generally done in the literature. Neither trade nor financial linkages are found significant in directly influencing business cycle synchronization. Only the similarity in productive structure appears to foster economic integration, after controlling for common policies. Trade linkages are found to increase output synchronization indirectly, by contributing to the similarity of productive structures, which might point to the prevalence of intra-industry trade. The positive influence of financial linkages on output synchronization is even more indirect, by fostering trade integration and, thereby, a more similar productive structure. The net effects of both trade and financial linkages on business cycle synchronization are found statistically significant, but economically very small. Chapter 6 - Since 1950 Spain has shown two parallel trends of dramatic drops in fertility and in religiosity (secularization). This paper explores the relationship between secularization and fertility among Spanish Catholics. The authorsuse a unique, rich, data set which includes various dimensions of religiosity: respondent’s religious affiliation; current church attendance (six levels); current prayer habits (eleven levels); spouse’s religious affiliation; parental religious affiliation; and parental (maternal and paternal) and respondent's church attendance during childhood (nine levels). The multi-facet data on religiosity (rather than a single dichotomous variable) allow for a sophisticated analysis, permitting rigorous conclusions to be drawn. The sample is restricted to married Catholic (female and male) respondents who were raised by Catholic parents and are married to a Catholic spouse in order to obtain a

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Preface

ix

homogenous sample and to focus on the effect of the level (intensity) of religiosity (rather than religious affiliation) on fertility. Fertility is related to the various dimensions of religiosity; first using cross-tabulation and then using OLS regression. Our results are substantive: i) The authorsfind that fertility is not related to the current intensity of religiosity. ii) Exposure to religious activities during childhood has a significant effect on women’s fertility (but not men). Interestingly, a father who rarely attended church services has a negative effect on his daughter’s future fertility (decreasing the number of children by about 0.8), while the mother’s inactive churchgoing has an unexpected positive effect (leading to a increase of one child). iii) The respondent’s own church attendance during childhood does not have any effect on current fertility. In sum, this study demonstrates the significance of childhood experience in shaping one’s 'taste for children'. It also suggests that there is no direct link between the rapid process of secularization occurring in Spain and the decline in birth rates. Chapter 7 - Objective: To describe sleeping habits of school adolescents. To determine the relation between sleep quality and inappropriate sleeping habits. Design: Observational, descriptive, cross-sectional study. Setting: Secondary and primary school of Cuenca, city in Spain. Participants: School children of first and fourth courses of secondary education and sixth course of primary education. Main Measures: Structured, self-administered questionnaire with opened and closed questions on sleeping habits during weekdays and at weekends and sleep disorders, to be answered by the secondary school adolescents and “Children Sleep Habits Questionnaire” (CSHQ) (Owens, 2000) by de primary school children, anonymously and on their own; “Sleep Disturbance Scale for Children” to be answered by primary children parents. Results: The final sample included 1155 secondary students (response rate 89.33%), 537 (45.9%) boys and 618 (54.1%) girls, 14 years old on average (between 11-18 years), 552 primary students (response rate 94.7%), 49.1% boys, 50.8% girls, 11-12 years old and 437 parents. On weekdays, secondary students went to bed at 23.17 h and got up at 7.46 h and at 22.37 h, 8.2 h the primary students (average sleeping time: 8 hours and 18 minutes and 9 h.8 m. respectively). At weekends the secondary adolescents went to bed at 1.02 h. and got up at 10.42 h. (average sleeping time: 9 h.40 m.) and the primary adolescents at 23.27 h., 9.38 h. (average sleeping time: 9 h.44 m.). 45.4% of students said to sleep badly on Sunday night’s. Poor quality sleep was reported by 21.1% secondary students and 8.9% primary students, difficulty in getting to sleep, 23,1% (both groups), 38.2% woke up during the night in secondary students, 6.4% in primary group. 53% of the secondary students and 13.7% of the primary students complained of excessive sleepiness during the day. High school teenagers reported insomnia in 9.9% which DSM-IV criteria. Conclusions: Sleep time decrease which age and bed time is delayed during class days and even more during weekends. Sleep disorders show a high prevalence in our population due to several factors that could lead to a considerable tiredness and hyper somnolence during the day. On weekends a break in sleep habits and insufficient sleep time leads to appearance of insomnia symptoms Chapter 8 - The Spanish educational system traditionally was very centralized. Historically, this centralism gave rise to the existence of educational establishments dependent on the Educational Administration and formally homogeneous, in which the decision making capacity was very limited. Of course, this homogeneity was just from a formal point of view,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

x

Pabo N. Costada

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

because, as educational research has made clear, even in highly centralized educational systems with uniform planning, schools end up being internally different from one another. Despite the same legal regulations, the dynamics of how the school is run often results in profound organizational differences. Nevertheless, and although it is undoubtedly clear that (from a practical point of view) each school has some peculiarities and some ways of functioning, in a centralized system the differences are minimised or simply ignored.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Profile

OFFICIAL NAME:KINGDOM OF SPAIN GEOGRAPHY

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Area: 504,750 sq. km. (194,884 sq. mi.), including the Balearic and Canary Islands; about the size of Arizona and Utah combined. Cities: Capital--Madrid (5.5 million). Other cities-Barcelona (4.9 million), Valencia (2.3 million), Seville (1.8 million), Malaga (1.3 million), Zaragoza (871,000), Bilbao (353,950). Terrain: High plateaus, lowland areas such as narrow coastal plains, and mountainous regions. Climate: Temperate. Summers are hot in the interior and more moderate and cloudy along the coast; winters are cold in interior and partly cloudy and cool along the coast. Time zone: Spanish mainland and Balearic Isles--local time is 1 hour ahead of Greenwich Mean Time (GMT) in winter and 2 hours ahead in summer. Canary Islands are on GMT.

PEOPLE Nationality: Noun--Spaniard(s). Adjective--Spanish. Population (2007 est.): 45,116,894. Annual growth rate (2007 est.): 0.116%. Ethnic groups: Distinct ethnic groups within Spain include the Basques, Catalans, and Galicians. Religion: Predominantly Roman Catholic; Protestant and Islamic faiths also have a significant presence. Languages: Spanish (official) 74%, Catalan-Valenciana 17%, Galician 7%, Basque 2%. Education: Years compulsory--to age 16. Literacy (2003 est.)--98%. Work force (19.2 million): Services--65.1%; agriculture--5.2%; construction--12.5%; industry--17.2% (2005 est.).

GOVERNMENT Type: Constitutional monarchy (Juan Carlos I proclaimed King November 22, 1975). Constitution: 1978. Branches: Executive--president of government nominated by monarch, subject to approval by democratically elected Congress of Deputies. Legislative--bicameral Cortes: a 350-seat Congress of Deputies (elected by the d'Hondt system of proportional representation) and a Senate. Four senators are elected in each of 47 peninsular provinces, 16

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

2

Pablo N. Costada

are elected from the three island provinces, and Ceuta and Melilla elect two each; this accounts for 208 senators. The parliaments of the 17 autonomous regions also elect one senator as well as one additional senator for every 1 million inhabitants within their territory (about 20 senators). Judicial--Constitutional Tribunal has jurisdiction over constitutional issues. Supreme Tribunal heads system comprising territorial, provincial, regional, and municipal courts. Subdivisions: 47 peninsular and three island provinces; two enclaves on the Mediterranean coast of Morocco (Ceuta and Melilla) and three island groups along that coast-Alhucemas, Penon de Velez de la Gomera, and the Chafarinas Islands. Political parties: Spanish Socialist Workers Party (PSOE), Popular Party (PP), and the United Left (IU) coalition. Key regional parties are the Convergence and Union (CIU) in Catalonia and the Basque Nationalist Party (PNV) in the Basque country.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ECONOMY GDP (2006): $1.225 trillion (seventh-largest Organization for Economic Cooperation and Development--OECD--economy). Annual growth rate (2006 est.): 3.9%. Per capita GDP (2006 est.): $27,422. Natural resources: Coal, lignite, iron ore, uranium, mercury, pyrites, fluorspar, gypsum, zinc, lead, tungsten, copper, kaolin, hydroelectric power. Agriculture and fisheries (2.8% of GDP, 2006 est.): Products--grains, vegetables, citrus and deciduous fruits, wine, olives and olive oil, sunflowers, livestock. Industry (15.54% of GDP, 2006 est.): Types--processed foods, textiles, footwear, petrochemicals, steel, automobiles, consumer goods, electronics. Trade (2006): Exports--$211.4 billion: automobiles, fruits, minerals, metals, clothing, footwear, textiles. Major markets--EU 70.35%, U.S. 4.41%. Imports--$289.8 billion: petroleum, oilseeds, aircraft, grains, chemicals, machinery, transportation equipment, fish, consumer goods. Major sources--EU 58.44%, U.S. 3.27%. Average exchange rate (2006): 0.797 euros=U.S.$1.

People Spain's population density, lower than that of most European countries, is roughly equivalent to New England's. In recent years, following a longstanding pattern in the rest of Europe, rural populations are moving to cities. Urban areas are also experiencing a significant increase in immigrant populations, chiefly from North Africa, South America, and Eastern Europe. Spain has no official religion. The constitution of 1978 disestablished the Roman Catholic Church as the official state religion, while recognizing the role it plays in Spanish society. More than 90% of the population is at least nominally Catholic. Among the remaining population, there are about 1.2 million evangelical Christians and other Protestants (2006 est.), 1 million Muslims (2006 est.) and 48,000 Jews (2006 est.).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Official Name:Kingdom of Spain

3

EDUCATIONAL SYSTEM About 70% of Spain's student population attends public schools or universities. The remainder attend private schools or universities, the great majority of which are operated by the Catholic Church. Compulsory education begins with primary school or general basic education for ages 6-14. It is free in public schools and in many private schools, most of which receive government subsidies. Following graduation, students attend either a secondary school offering a general high school diploma or a school of professional education (corresponding to grades 9-12 in the United States) offering a vocational training program. The Spanish university system offers degree and post-graduate programs in all fields--law, sciences, humanities, and medicine--and the superior technical schools offer programs in engineering and architecture.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

History The Iberian Peninsula has been settled for millennia. In fact, some of Europe's most impressive Paleolithic cultural sites are located in Spain, including the famous caves at Altamira that contain spectacular paintings dating from about 15,000 to 25,000 years ago. The Basques, Europe's oldest surviving group, are also the first identifiable people of the peninsula. Beginning in the ninth century BC, Phoenicians, Greeks, Carthaginians, and Celts entered the Iberian Peninsula. The Romans followed in the second century BC and laid the groundwork for Spain's present language, religion, and laws. Although the Visigoths arrived in the fifth century AD, the last Roman strongholds along the southern coast did not fall until the seventh century AD. In 711, North African Moors sailed across the straits, swept into Andalusia, and within a few years, pushed the Visigoths up the peninsula to the Cantabrian Mountains. The Reconquest--efforts to drive out the Moors--lasted until 1492. By 1512, the unification of present-day Spain was complete. During the 16th century, Spain became the most powerful nation in Europe, due to the immense wealth derived from its presence in the Americas. But a series of long, costly wars and revolts, capped by the defeat by the English of the "Invincible Armada" in 1588, began a steady decline of Spanish power in Europe. Controversy over succession to the throne consumed the country during the 18th century, leading to an occupation by France during the Napoleonic era in the early 1800s and a series of armed conflicts throughout much of the 19th century. The 19th century saw the revolt and independence of most of Spain's colonies in the Western Hemisphere; three wars over the succession issue; the brief ousting of the monarchy and establishment of the First Republic (1873-74); and, finally, the Spanish-American War (1898), in which Spain lost Cuba, Puerto Rico, and the Philippines to the United States. A period of dictatorial rule (1923-31) ended with the establishment of the Second Republic. It was dominated by increasing political polarization, culminating in the leftist Popular Front electoral victory in 1936. Pressures from all sides, coupled with growing and unchecked violence, led to the outbreak of the Spanish Civil War in July 1936.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

4

Pablo N. Costada

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Following the victory of his nationalist forces in 1939, General Francisco Franco ruled a nation exhausted politically and economically. Spain was officially neutral during World War II but followed a pro-Axis policy. Therefore, the victorious Allies isolated Spain at the beginning of the postwar period, and the country did not join the United Nations until 1955. In 1959, under an International Monetary Fund stabilization plan, the country began liberalizing trade and capital flows, particularly foreign direct investment. Despite the success of economic liberalization, Spain remained the most closed economy in Western Europe--judged by the small measure of foreign trade to economic activity--and the pace of reform slackened during the 1960s as the state remained committed to "guiding" the economy. Nevertheless, in the 1960s and 1970s, Spain was transformed into a modern industrial economy with a thriving tourism sector. Its economic expansion led to improved income distribution and helped develop a large middle class. Social changes brought about by economic prosperity and the inflow of new ideas helped set the stage for Spain's transition to democracy during the latter half of the 1970s. Upon the death of General Franco in November 1975, Franco's personally designated heir Prince Juan Carlos de Borbon y Borbon assumed the titles of king and chief of state. Dissatisfied with the slow pace of post-Franco liberalization, he replaced Franco's last Prime Minister with Adolfo Suarez in July 1976. Suarez entered office promising that elections would be held within one year, and his government moved to enact a series of laws to liberalize the new regime. Spain's first elections since 1936 to the Cortes (Parliament) were held on June 15, 1977. Prime Minister Suarez's Union of the Democratic Center (UCD), a moderate center-right coalition, won 34% of the vote and the largest bloc of seats in the Cortes. Under Suarez, the new Cortes set about drafting a democratic constitution that was overwhelmingly approved by voters in a national referendum in December 1978.

Government and Political Conditions Parliamentary democracy was restored following the death of General Franco in 1975, who had ruled since the end of the civil war in 1939. The 1978 constitution established Spain as a parliamentary monarchy, with the prime minister responsible to the bicameral Cortes (Congress of Deputies and Senate) elected every 4 years. On February 23, 1981, rebel elements among the security forces seized the Cortes and tried to impose a military-backed government. However, the great majority of the military forces remained loyal to King Juan Carlos, who used his personal authority to put down the bloodless coup attempt. In October 1982, the Spanish Socialist Workers Party (PSOE), led by Felipe Gonzalez, swept both the Congress of Deputies and Senate, winning an absolute majority. Gonzalez and the PSOE ruled for the next 13 years. During that period, Spain joined the North Atlantic Treaty Organization (NATO) and the European Community. In March 1996, Jose Maria Aznar's Popular Party (PP) won a plurality of votes. Aznar moved to decentralize powers to the regions and liberalize the economy, with a program of privatization, labor market reform, and measures designed to increase competition in selected markets. During Aznar's first term, Spain fully integrated into European institutions, qualifying for the European Monetary Union. During this period, Spain participated, along with the United States and other NATO allies, in military operations in the former

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Official Name:Kingdom of Spain

5

Yugoslavia. President Aznar and the PP won reelection in March 2000, obtaining absolute majorities in both houses of parliament. After the terrorist attacks on the U.S. on September 11, 2001, President Aznar became a key ally in the fight against terrorism. Spain backed the military action against the Taliban in Afghanistan and took a leadership role within the European Union (EU) in pushing for increased international cooperation on terrorism. The Aznar government, with a rotating seat on the UN Security Council, supported the intervention in Iraq. Spanish parliamentary elections on March 14, 2004 came only three days after a devastating terrorist attack on Madrid commuter rail lines that killed 191 and wounded over 1,400. With large voter turnout, PSOE won the election and its leader, Jose Luis Rodriguez Zapatero, took office on April 17, 2004. Carrying out campaign promises, the Zapatero government immediately withdrew Spanish forces from Iraq but has continued to support Iraq reconstruction efforts. The Zapatero government has supported coalition efforts in Afghanistan, including maintaining troop support for 2004 and 2005 elections, supported reconstruction efforts in Haiti, sent troops to UNIFIL in Lebanon, and cooperated on counterterrorism issues and many other issues of importance to the U.S.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

LOCAL GOVERNMENT The 1978 constitution authorized the creation of regional autonomous governments. By 1985, 17 regions covering all of peninsular Spain, the Canaries, and the Balearic Islands had negotiated autonomy statutes with the central government. In 1979, the first autonomous elections were held in the Basque and Catalan regions, which have the strongest regional traditions by virtue of their history and separate languages. Since then, autonomous governments have been created in the remainder of the 17 regions. The central government continues to devolve powers to the regional governments, which will eventually have full responsibility for health care and education, as well as other social programs.

TERRORISM The Government of Spain is involved in a long-running campaign against Basque Fatherland and Liberty (ETA), a terrorist organization founded in 1959 and dedicated to promoting Basque independence. ETA targets Spanish security forces, military personnel, Spanish Government officials, and politicians of the Popular Party and the Socialist Party (PSOE.) The group has carried out numerous bombings against Spanish Government facilities and economic targets, including a car bomb assassination attempt on then-opposition leader Aznar in 1995, in which his armored car was destroyed but he was unhurt. The Spanish Government attributes over 800 deaths to ETA terrorism since its campaign of violence began. In recent years, the government has had more success in controlling ETA, due in part to increased security cooperation with French authorities. In November 1999, ETA ended a cease-fire it declared in September 1998. Following the end of that ceasefire, ETA conducted a campaign of violence and has been blamed for the deaths of some 46 Spanish citizens and officials.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

6

Pablo N. Costada

Each attack has been followed by massive anti-ETA demonstrations around the country, clearly demonstrating that the majority of Spaniards, including the majority of Spain's Basque populace, have no tolerance for continued ETA violence. In March 2006, ETA declared another ceasefire, which it ended in June 2007. The government continues to pursue vigorous counterterrorist policy and has worked closely with its international allies to foil several suspected ETA attacks. Spain also contends with a resistance group, commonly known as GRAPO. GRAPO is an urban left-wing terrorist group that seeks to overthrow the Spanish Government and establish a Marxist state. It opposes Spanish participation in NATO and U.S. military presence in Spain and has a long history of assassinations, bombings, and kidnappings mostly against Spanish interests during the 1970s and 1980s. In a June 2000 communiqué following the explosions of two small devices in Barcelona, GRAPO claimed responsibility for several terrorist attacks throughout Spain during the past year. These attacks included two failed armored car robberies, one in which two security officers died, and four bombings of political party offices during the 1999/2000 election campaign. In 2002 and 2003, Spanish and French authorities were successful in hampering the organization's activities through sweeping arrests, including some of the group's leadership. Al Qaeda is known to operate cells in Spain. On March 11, 2004, only three days before national elections, 10 bombs were detonated on crowded commuter trains during rush hour. Three were deactivated by security forces and one was found unexploded. Evidence quickly surfaced that jihadist terrorists with possible ties to the al Qaeda network were responsible for the attack that killed 191 people. Spanish investigative services and the judicial system have aggressively sought to arrest and prosecute suspected al Qaeda members and actively cooperate with foreign governments to diminish the transnational terrorist threat. A Spanish court convicted 18 individuals in September 2005 for their role in supporting al Qaeda, and Spanish police disrupted numerous Islamist extremist cells operating in the country. The trial against 29 people for their alleged participation in the Madrid March 11, 2004 terrorist attack started in February 2007, and was declared ready for judgment at the end of June. One of the 29 was absolved during the trial. The prosecutor asked for sentences as high as 30,000 years of jail for some of them. The court is expected to issue the sentence sometime in October 2007.

PRINCIPAL GOVERNMENT OFFICIALS Chief of State, Commander in Chief of the Armed Forces--King Juan Carlos I President of the Government (Prime Minister)--Jose Luis Rodriguez Zapatero Minister of Foreign Affairs--Miguel Angel Moratinos Ambassador to the United States--Carlos Westendorp y Cabeza Spain maintains an embassy in the United States at 2375 Pennsylvania Avenue NW, Washington, DC 20037 (tel. 202-452-0100) and consulates in many larger U.S. cities.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Official Name:Kingdom of Spain

7

Economy Spain's accession to the European Community--now European Union (EU)--in January 1986 required the country to open its economy to trade and investment, modernize its industrial base, improve infrastructure, and revise economic legislation to conform to EU guidelines. In doing so, Spain increased gross domestic product (GDP) growth, reduced the public debt to GDP ratio, reduced unemployment from 23% in 1986 to 8.47% in first quarter 2007, and reduced inflation to under 3%. The fundamental challenges remaining for Spain include decreasing unemployment further, reforming labor laws lowering inflation, and raising per capita GDP. Following peak growth years in the late 1980s, the Spanish economy entered into recession in mid-1992. The economy recovered during the first Aznar administration (19962000), driven by a return of consumer confidence and increased private consumption, although growth has slowed in recent years. Unemployment remains a problem at 8.47% (2007, first quarter), but this still represents a significant improvement from previous levels. Devaluations of the peseta during the 1990s made Spanish exports more competitive, but the strength of the euro since its adoption has raised recent concerns that Spanish exports are being priced out of the range of foreign buyers.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Foreign Relations After the return of democracy following the death of General Franco in 1975, Spain's foreign policy priorities were to break out of the diplomatic isolation of the Franco years and expand diplomatic relations, enter the European Community, and define security relations with the West. As a member of NATO since 1982, Spain has established itself as a major participant in multilateral international security activities. Spain's EU membership represents an important part of its foreign policy. Even on many international issues beyond Western Europe, Spain prefers to coordinate its efforts with its EU partners through the European political cooperation mechanism. With the normalization of diplomatic relations with Israel and Albania in 1986, Spain virtually completed the process of universalizing its diplomatic relations. The only country with which it now does not have diplomatic relations is North Korea. Spain has maintained its special identification with Latin America. Its policy emphasizes the concept of Hispanidad, a mixture of linguistic, religious, ethnic, cultural, and historical ties binding Spanish-speaking America to Spain. Spain has been an effective example of transition from authoritarianism to democracy, as shown in the many trips that Spain's King and Prime Ministers have made to the region. Spain maintains economic and technical cooperation programs and cultural exchanges with Latin America, both bilaterally and within the EU. Spain also continues to focus attention on North Africa, especially on Morocco. This concern is dictated by geographic proximity and long historical contacts, as well as by the two Spanish enclave cities of Ceuta and Melilla on the northern coast of Africa. While Spain's departure from its former colony of Western Sahara ended direct Spanish participation in Morocco, it maintains an interest in the peaceful resolution of the conflict brought about there

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

8

Pablo N. Costada

by decolonization. These issues were highlighted by a crisis in 2002, when Spanish forces evicted a small contingent of Moroccans from a tiny islet off Morocco's coast following that nation's attempt to assert sovereignty over the island. Meanwhile, Spain has gradually begun to broaden its contacts with Sub-Saharan Africa. It has a particular interest in its former colony of Equatorial Guinea, where it maintains a large aid program. In relations with the Arab world, Spain has sought to promote European-Mediterranean dialogue. Spain strongly supports the EU's "Barcelona Process" which seeks to expand dialogue and trade between Europe and the nations of North Africa and the Middle East, including Israel. It is seen by some as too greatly favoring the European Union. Other proposals are on the table, including one put forward by French President Nicolas Sarkozy in 2007. The latest meeting on the Barcelona initiative was held on November 29, 2005. Spain has been successful in managing its relations with its two European neighbors, France and Portugal. The accession of Spain and Portugal to the EU has helped ease some of their periodic trade frictions by putting these into an EU context. Franco-Spanish bilateral cooperation is enhanced by joint action against Basque ETA terrorism. Ties with the United Kingdom are generally good, although the question of Gibraltar remains a sensitive issue.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

U.S.-Spanish Relations Spain and the United States have a long history of official relations and are closely associated in many fields. In addition to U.S. and Spanish cooperation in NATO, defense and security relations between the two countries are regulated by a 1989 Agreement on Defense Cooperation, revised in 2003. Under this agreement, Spain authorized the United States to use certain facilities at Spanish military installations. The two countries also cooperate in several other important areas. Under a 1964 agreement (currently being renegotiated), the U.S. National Aeronautics and Space Administration (NASA) and the Spanish National Institute of Aerospace Technology (INTA) jointly operate the Madrid Deep Space Communications Complex in support of Earth orbital and solar system exploration missions. The Madrid Complex is one of the three-largest tracking and data acquisition complexes comprising NASA's Deep Space Network. An agreement on cultural and educational cooperation was signed on June 7, 1989. A new element, supported by both the public and private sectors, gives a different dimension to the programs carried out by the joint committee for cultural and educational cooperation. These joint committee activities complement the binational Fulbright program for graduate students, postdoctoral researchers, and visiting professors, which became the largest in the world in 1989. Besides assisting in these exchange endeavors, the U.S. Embassy also conducts a program of official visits between Spain and the United States. Spain and the U.S. are strong allies in the fight against terrorism.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 1

THE EFFECT OF GOVERNMENT OWNERSHIP ON BANK PROFITABILITY AND RISK: THE SPANISH EXPERIMENT Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González University of Oviedo. Financial support provided by the Spanish Science and Technology Ministry and FEDER, Project SEC 2002-04765 is gratefully acknowledged. Fonseca would also like to acknowledge the financial support provided by Herrero Bank Foundation

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

All correspondence to: Francisco González Rodríguez, Dpto. Administración de Empresas, Facultad de CC. Económicas, Avenida del Cristo S/N, Oviedo, 33071, Spain. Tel.: +34-985103698. Fax: +34-985103708. E-mail:[email protected]

ABSTRACT This paper analyzes the effects on Spanish savings banks’ performance and risk when they shed their mutual structure to become government-owned banks. Such a situation arose in 1985 when Spain’s central government allowed regional parliaments to modify savings bank ownership regulations. Regional regulations increased government participation at the expense of depositors’ ownership. This regulatory change constitutes a natural experiment to study the consequences of government ownership on bank behavior. The results of our study suggest that enhanced government ownership leads to an increase in risk. This is particularly marked amongst those savings banks that most increased the weight of local and regional governments on their governance bodies. However, no variation in savings bank performance has occurred. The net result, therefore, is an increase in performance-adjusted risk.

JEL Classification: G21, G32 Keywords: savings banks, government ownership, regulation, performance, risk

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

10

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION This study analyzes the effects on performance and risk-taking that are brought about by increased local and regional government ownership of Spanish savings banks. In Spain, this situation dates back to 1985 and the coming into force of the so-called LORCA law (Ley de regulación de normas básicas sobre órganos rectores de las cajas de ahorros), which laid down the legal framework for savings banks’ governance bodies and established the percentage representation of depositors, employees and public administrations alike in savings bank ownership. Regional regulations subsequently modified savings banks’ governance, increasing the presence of local and regional governments, basically to the detriment of depositors’ representatives. This legislative change constitutes a de facto experiment whereby the consequences on performance and risk-taking when savings banks lose their mutual structure to become government-owned banks can be analyzed. Unlike stock-owned banks, whose governance bodies are made up of a representation of stockholders, the composition of savings banks’ governance bodies is established by law and as such is an exogenous variable. One of the potential backlashes of such exogeneity might be to prevent governance bodies from adapting to the optimum requirements of a competitive market, and systematic differences may exist between the levels of efficiency and risk of differently organized thrift institutions (Demsetz and Lehn, 1985). This paper hopes to contribute to the extensive literature that analyzes the effect of organizational form on bank efficiency and risk-taking by considering a novel facet in the field, the shift from a mutual structure to becoming a government-owned bank. Most of the literature has focused on comparing the efficiency and risk-taking of stockowned and mutual banks, as these are the two prevailing organizational forms in the United States1. Despite the ubiquity of government-owned banks in a raft of countries, as highlighted by La Porta et al (2002), fewer studies have analyzed the impact of state ownership on financial institutions. This is a gap in the scope of research made even more surprising by the fact that savings banks’ governance is a key issue in a number of European countries, where debate rages on the question of whether such state-run institutions should be converted into stock-owned banks. Whilst in Belgium, Denmark, Great Britain, Holland, Ireland, Italy and Sweden savings banks have become stock-owned companies, in other countries such as Germany, Austria, Greece, Portugal, Switzerland and Spain there are savings banks that are partially owned by the state or by local and regional governments. Within this latter group of countries there is open debate nowadays on savings banks’ ownership. Opinions range from extremes advocating conversion into stock-owned thrift institutions to more moderate proposals that defend current structure whilst calling for some modification of the percentage representation of each collective in savings bank governance. International organizations have added their weight to the debate. The International Monetary Fund (IMF), for example, in its 1999 report, and the organization for Economic Cooperation and Development (OECD) in its 1

For a comparative analysis of the efficiency of mutal banks and stock-owned banks, see among others, Altumbas et al., 2001; Blair and Placone, 1988, Cebenoyan et al. 1993; Daniels and Sfiridis, 2001; Mester, 1991,1993; O’Hara, 1981; Valnek, 1999; Verbrugge and Golstein, 1981 and Verbrugge and Jahera, 1981. For an analysis of the differences in risk between mutual al and stock-owned banks see, among others, Hadaway and Hadaway, 1984; Masulis, 1987; Cordell et al., 1993; Esty ,1997a, 1997b; Karels and McClatchey, 1999; Lamm-Tennant and Starks, 1993; O’Hara, 1981; Scharand and Unal, 1998 and Verbrugge and Goldstein, 1981).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The Effect of Government Ownership on Bank Profitability and Risk

11

report on Spain for the year 2000, both include among its recommendations a consideration of a possible shift in Spanish savings banks’ organizational form towards the stock-owned thrift institution. The relatively few papers that analyze the impact of state ownership on banks’ efficiency have yet to come up with conclusive results2. La Porta et al. (2002) analyze data from government–owned banks in 92 countries, concluding that the presence of the state “politicizes” the resource allocation process within financial institutions, since it allows governments to finance investments that may well be politically desirable but are nevertheless inefficient from an economic stance. Sapienza (1999) also concludes that Italian state-owned banks pursue political objectives in their lending policy. Barth et al. (2001) use data on government ownership from Bankscope to report that enhanced government ownership of banks is generally associated with less efficient and less well-developed financial systems. Verbrugge et al. (1999) analyze 65 bank privatizations in 25 countries and document a limited improvement in bank profitability, operating efficiency, and non-interest revenue after privatization. As for Spanish savings banks, Melle and Maroto (1999) not only highlight a positive relationship between public administrations’ representation on boards of directors and the percentage of loans savings banks give the public sector but also point out that this enhanced lending to the public sector induces a negative effect on savings banks’ performance. However, such findings are supported by neither Grifell-Tatjé and Novell (1997) nor Lozano (1998), who claim that government-owned savings banks and stockowned banks in Spain have similar levels of productive efficiency. In contrast to studies suggesting the greater efficiency of stock-owned banks, Altunbas et al. (2001) conclude that government-owned German savings banks are more efficient than their respective private counterparts. As for Belgium, Tulkens (1993) also concludes that public banks’ branches are relatively more efficient than those of private banks. The dearth of studies on the impact of government ownership on banking efficiency is even more exacerbated as far as its influence on financial intermediaries’ risk taking incentives are concerned, as we know of no studies in this field. This paper therefore hopes to fill a knowledge gap by describing not only the impact of government ownership on savings banks’ incentives to take risks but also by analyzing performance change when government ownership replaces a mutual structure. To this end, we use a different approach to the one used in previous studies in that we analyze savings banks’ performance change and risktaking after their governance bodies have been modified by regulations increasing government ownership. The paper is structured as follows: section 2 presents the characteristics of Spanish savings banks’ ownership and the legislative changes introduced since 1985. Section 3 discusses our hypotheses as to the effects of more government ownership on savings bank performance and risk. Section 4 and 5 present the methodology and discuss the empirical results. Finally, section 6 presents the paper’s conclusions.

2

However, there exist in the industrial sector abundant evidence showing that public firms are less efficient than their private counterparts. Boardman and Vining (1989) provide a summary table with the empirical evidence on the relative efficiency of public and private firms.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

12

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2. THE REGULATION AND GOVERNANCE OF SPANISH SAVINGS BANKS There are three types of banks in Spain’s banking sector: commercial stock-owned banks, savings banks and credit cooperatives. All three types of banks compete under equal conditions in the loan, deposit and financial service markets and their accounting practices, external reporting and credit-risk management standards are also the same to all practical intents and purposes. In terms of economic importance, commercial and savings banks hold the lion’s share of the Spanish market. Thus, Spanish stock banks accounted for 61.46% of the banking system’s total balance sheet during the 1984-2000 period, with savings banks chalking up 35.24% and cooperatives holding a mere 3.30%. Table 1 gives percentages of total assets, deposits and loans for each type of bank as part of Spain’s banking system, and also gives the number of offices and employees that they have. Savings banks were initially created in 1835 in Spain as non-profit making organizations. They had a clearly defined social commitment and instead of paying dividends, benefits were allocated to social and cultural activities. Furthermore, they were involved in a different business to commercial stock-owned banks. Differences extended to both geography and client-type, as savings banks basically ran their business each within their particular region, catering for families and small and medium-sized businesses. On the contrary, stock banks were oriented to the national market and industrial firms were more important clients. Two further hallmarks of savings banks were the higher percentage of loans granted to the public sector and the higher percentage of mortgage loans to total private loans. These differences are rooted in state legislation, which banned savings banks from operating beyond their geographic boundaries and obliged them to direct part of their activities towards families. These limitations were withdrawn, however, in 1977, and since 1989 stock banks and savings banks have been subject to the same operative regulations. Nevertheless, despite there no longer being legal differences relating to how each type of bank may operate, time has only moderated their hallmarks, as each continues to operate in the markets where it had its largest market share. At present, the only regulatory difference between stock and savings banks in Spain relates to ownership. The greatest idiosyncrasy of the Spanish savings bank is its peculiar ownership structure, which falls neither into the category of stock-based institutions, nor that of mutuals. Their basic governance mechanisms are the General Assembly and the Board of Directors, made up of representatives from four groups whose percentage representation is established by law.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 1. Importance of each type of bank ownership in Spain

Table 1 reports total assets, deposits and loans for each type of bank as a percentage of the whole Spanish system for the 1984-2000 period. The number of branches and employees is also shown.

Savings Banks % total assets % total deposits % total loans N. branches N. employees Stocks Banks % total assets % total deposits % total loans N. branches N. employees Credit Cooperatives % total assets % total deposit % total loans N. branches N. employees

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Mean

30.14

32.11

34.28

34.82

37.29

38.66

38.42

36.51

35.43

32.50

33.37

33.62

34.72

35.02

36.48

37.26

38.41

35.24

35.36

39.12

42.86

43.49

45.25

45.74

45.30

45.48

46.98

18.31

19.75

49.47

51.70

51.71

50.73

51.31

51.22

42.57

26.33 10,440 69,438

26.55 10,797 71,042

28.28 11,061 72,707

30.40 11,754 74,530

32.58 12,252 78,023

34.07 13,168 83,026

34.41 13,720 84,609

33.13 14,031 83,359

35.51 14,291 82,900

37.78 14,485 82,710

38.09 14,880 83,758

38.65 15,214 84,336

39.02 16,094 87,370

40.06 16,636 90,153

41.50 17,582 93,812

42.17 18,119 97,276

43.26 19,268 101,718

35.40 14,341 83,574

66.73 60.38 70.27 16,412 164,330

64.48 56.07 69.93 16,606 161,621

62.21 52.15 68.18 16,518 157,805

61.74 51.31 66.10 16,498 155,334

59.43 49.62 64.40 16,691 154,696

58.28 49.29 62.97 16,677 155,658

58.48 49.78 62.54 16,917 157,010

60.26 49.49 63.76 17,824 161,987

61.60 47.89 61.10 18,058 159,281

64.62 46.35 58.82 17,636 152,845

63.60 44.62 58.14 17,557 151,174

63.14 44.64 57.22 17,842 148,946

61.85 41.82 56.60 17,674 142,827

61.47 41.33 55.29 17,530 139,198

59.89 43.28 53.74 17,450 135,164

59.12 42.43 52.90 17,140 127,889

57.92 42.46 51.80 15,811 122,374

61.46 47.82 60.81 17,108 126,361

3.13 4.26 3.40 3,315 10,896

3.41 4.81 3.52 3,350 10,823

3.51 4.99 3.54 3,382 10,225

3.44 5.20 3.50 3,248 10,153

3.28 5.13 3.02 3,029 9,674

3.06 4.97 2.96 2,890 9,592

3.10 4.91 3.05 2,919 9,968

3.23 5.04 3.11 3,018 10,643

2.98 5.13 3.40 3,080 11,016

2.88 5.34 3.40 3,072 11,225

3.03 5.63 3.78 3,107 11,195

3.24 5.89 4.13 3,195 11,626

3.42 6.48 4.37 3,311 12,024

3.52 6.96 4.65 3,468 12,804

3.64 5.99 4.76 3,607 13,286

3.62 6.26 4.92 3,697 13,855

3.67 6.32 4.94 3,888 14,495

3.30 5.49 3.79 2,883 10,794

14

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

In 1985, the 31/1985 national law, or LORCA law, unified what until then had been a gamut of differing statute-regulated governance systems established in different Spanish savings banks by establishing the following percentage governance representation: 1) 40% local and regional governments, 2) 44% depositors, 3) 11% founders and 4) 5% employees. Since in some cases the founding members also happened to be local and regional governments, the final percentage of public administrations in savings banks’ General Assemblies and on Boards of Directors can be as high as 51%. In this way, the LORCA law unified the disparity in Spanish savings bank governance, which had previously been established differently depending on each savings bank’s statutes3. With a structure whereby ownership lies with depositors, employees and governments, savings banks may be described as hybrids of mutual, cooperative and a government-owned banks. Table 2. Participation of depositors, employees and regional and local governments in the ownership of savings banks This Table presents the participation percentage that employees, depositors and regional and local governments have in the General Assembly and Board of Directors of the savings banks as established in 1985 by Law 3/81 and in each of the regional laws passed later. The last column shows the total government ownership after adding the ownership governments have directly by law to the one they have being savings bank’s founders. EMPLOYEES

DEPOSITORS

LOCAL AND REGIONAL GOVERNMENTS

FOUNDERS

5

44

40

11

1986 1991 1988 1989 1990 1990 1985 1997

5 7 5 5 5 5 5-10 7

44 41 20 39 26 22 30-40 22

40 42 40 34 44 38 15-25 40

11 10 35 16 10 10 25-35 10

1990 1994 1985 1992 1988 1987 1991 1988 1990

5-10 5 5-15 8 7 5 5 5 11

35-40 44 30-40 28 30 44 41

25-35 40 15-25 32 33 40 32 31 28

5-10 11 25-35 20 30 11 22 33 5

NATIONAL LAW 31/1985

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

REGION

TOTAL LOCAL AND REGIONAL GOVERNMENTS

YEAR REGIONAL LAW PUBLICATION

ANDALUCIA ARAGÓN ASTURIAS BALEARES CANARIAS CANTABRIA CATALUÑA CASTILLA LA MANCHA CASTILLA LEÓN EXTREMADURA GALICIA MADRID MURCIA NAVARRA PAÍS VASCO LA RIOJA VALENCIA

3

OTHER INSTITUTIONS

31 28

6 15 25 21 5-30

12

28

57.78 42 75 45 59.95 66 30.93 70.05 42.08 41.1 31.9 55 63 58.3 53.3 64.4 57.7

Although dispersion would demand an individual analysis of each savings bank, it is nevertheless possible to indicate that savings banks’ structure before 1985 was basically mutual since depositors had ownership percentages of over 50% in most savings banks, and in some cases these reached 83%.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The Effect of Government Ownership on Bank Profitability and Risk

15

However, the LORCA law also empowered Regional Parliaments to modify these representation percentages. These post-LORCA regulations, which are summarized in Table 2, have tended to increase the presence of local and regional governments on the governance bodies of savings banks at the expense of depositors. They also returned minor levels of representation to a range of “other institutions” whose pre-LORCA representation has been annulled by the new law4. According to regional regulations, the group with the greatest representation is public administrations. Whereas depositors’ representation stands at an average of 33%, local and regional government participation ranges between a minimum 30.93% and a maximum 75%, depending on whether the public administrations are also founders or members of other institutions. The final column in Table 2 presents the total percentage that public administrations have on average in savings banks’ ownership, including additional percentages assigned to local or regional bodies by dint of them doubling as savings bank founders. The new percentages established for depositors by regional laws imply an average fall of 11% with respect to the percentage LORCA established for depositors, whilst the maximum percentage public administrations can reach rises from 51% to 75%. The employees’ percentage has remained generally stable at 5% and has increased only in four regions to percentages ranging between 10% and 15%. In short, regional regulations have triggered a decrease in the mutual character of Spanish savings banks, coupled with an increase in state ownership. This process of partial conversion of mutual banks into state banks is atypical and different from the conversion processes analyzed so far in the literature, namely, the conversion in the USA of mutual banks into stock-owned banks5 and the privatization process of public firms in a number of countries around the world6. The latest change in ownership regulations of Spanish savings banks was in 2002, when a “Financing Law” came into force that capped public administration participation in savings banks’ governing bodies at 50%. This meant modifying the structure of the General Assemblies of savings banks in 12 of the 17 self-governing regions of Spain.

3. CHANGES IN SAVINGS BANKS’ OWNERSHIP AND BANKS’ PERFORMANCE AND RISK Firm’s property rights theory suggests that state and mutual enterprises should perform less efficiently and less profitably than private enterprises (Boardman and Vining, 1989). The reason suggested is straightforward: a lack of capital market discipline weakens owners’ control over management, leaving it freer to pursue its own interests and giving it fewer incentives to be efficient. However, the increase in political control brought on by the inception of regional laws fail to affect the non-existent discipline exerted by the capital market on savings banks’ 4

These “other organizations” to whom regional government regulations again give a minority presence on governance bodies of savings banks vary from one region to the next depending on the different charitable and cultural associations and organizations in each region. 5 See, among others, Hadaway and Hadaway (1984), Masulis (1987), Cordell et al. (1993), Esty (1997a, 1997b) and Schrand and Unal (1998). 6 Megginson and Netter (2001) carry out a survey of empirical studies on privatization.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

16

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

managers, merely substituting depositors’ control for political control. Therefore, in order to predict the effects of the change, not only must the percentage representation of each group – public administrations, employees and depositors – be analyzed but also the incentives of each of them to monitor and supervise managers’ decisions. The partial guarantee of deposits and the dispersion of the depositor group have been used as arguments to justify high management discretionality in mutual banks, as the depositors lack the incentives to monitor bank managers (O`Hara, 1981; Rasmunsen, 1988; Dewatripont and Tirole, 1993a, 1993b). Coverage of deposit insurance in Spain is limited to 20,000 euros per depositor. However, history would suggest that the partial guarantee of the deposit insurance is “de facto” an implicitly total guarantee, as historically the banking systems has always guaranteed 100% of the deposits of insolvent banks. The supervisory authority’s reaction is based upon the belief that when a bank is “too big to fail” the social backlash caused by there not being total coverage is enormous. As has been pointed out, there is little incentive for depositors to monitor bank managers. Nor do the structure and characteristics of depositor groups encourage such control and follow-up. Firstly, the representation system is random and disregards the amount deposited by each depositor. Nor can a market system be instigated whereby voting rights can be negotiated freely, as delegating votes is not allowed. Under such circumstances, depositors are unlikely to give the necessary consideration to decisions on issues relevant to their interests. On the one hand, a depositor who is really interested in playing a role in the bank’s management may be deprived of the opportunity by the random nature of the electoral process. On the other hand, those who are elected may lack the incentive to invest resources in obtaining information about the bank because third parties cannot be prevented from benefiting form the efforts that they make. In short, the considerable dispersion of a large number of depositors, the random nature of the election of representatives process, and the high percentage participation of local and regional governments all suggest that depositors have little influence on decision taking in savings banks and that much is left to the discretion of bank managers. However, enhanced state ownership of savings banks at the expense of depositors entails lowering the weight of management objectives and increasing political ones. Both La Porta et al. (2002) and Sapienza (1999) show that one of the spin-offs of politicizing decision-taking in government-owned banks is the pursuit of politically attractive but financially unprofitable projects. Most financial economist see political influence over depositary institutions’ credit allocation as the major reason for the financial crisis of many countries in Latin America and Southeast Asia (Kaufman, 1999). It seems to be the case that politization in decision-taking has a negative effect on performance, which adds to the capital market’s lack of discipline which already characterized the mutual structure. Unlike the consequences on bank performance, there are arguments that suggest both a positive and a negative effect on banks’ risk-taking after the introduction of greater political control. On the one hand, the fact that public administrations are major clients of savings banks may increase the ex-ante bank risk-taking incentives by allowing them to substitute losses from failed risky investment with subsidies (Barth et al. 2001). Furthermore, the subsidies option also encourages the politization of decision taking and the tendency to undertake politically desirable but financially risky projects. In this case, the final result would be an increase in savings banks’ risk-taking.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The Effect of Government Ownership on Bank Profitability and Risk

17

On the other hand, political interest in maintaining savings banks as an instrument by which to fulfill political objectives, as shown by La Porta et al. (2002) and Sapienza (1999), may lead to risk limitation with the joint aims of guaranteeing the continuity of the institution and avoiding a crisis in the savings bank. In other words, politicians may have incentives to limit savings bank risk up to a level that guarantees their solvency so that they do not lose an instrument that may be difficult to substitute. The threat of losing a political instrument would thereby have the effect of discouraging risk, just as the loss of high charter value does for stock-owned banks in regulated environments, as Keeley (1990) was first to point out. Since political participation in savings banks’ decision-taking could favor both higher and lower risk levels, the effect regulatory changes that increase state participation may have on savings banks’ risk level is an empirical question. Analysis of post-legislation savings bank risk variation will highlight which of the two hypotheses prevails. Even though regional government regulations basically consisted of increasing the presence of public administrations in savings banks to the detriment of depositors, they also led to a rise from 5% to 15% employee ownership in four regions of Spain, which means that savings banks share some of the hallmarks of cooperatives. Jensen and Meckling (1979) point out that employee participation in company ownership may lead to penchants for investment projects that recoup investment in a period that is equal to or less than the time the employee will remain in the organization, just as it may also provoke rejection of profitable projects that provide cash flows beyond the term of employment. Moreover, there are other incentives to take decisions that will have negative impacts on savings banks’ profits, such as setting extremely lucrative salaries and other perks. As far as risk is concerned, employees will clearly be inclined towards low-risk investments that will not endanger job stability or salary levels. As is also the case with managers, employees have much of their wealth tied up with the organization they work for, and are therefore more loath to take risks than other stakeholders who can diversify their risk to a greater extent. The logic of these arguments leads to the prediction that greater employee ownership of savings banks should bring about a reduction in both profits and risk.

4. EMPIRICAL ANALYSIS In order to test the effect that different participation of public administrations, employees and depositors has on savings banks’ operative behavior, we compare bank performance and risk before and after the introduction of the above-mentioned regional regulations. Information on the composition of savings banks’ governance bodies was obtained from their annual reports and, failing this, was requested from the thrift institution itself by mail. Savings banks that had been involved in mergers were excluded from the analysis so that other co-existing factors could be isolated7 . This yielded relevant information on 30 savings banks for between 1984 and 1999, 24 of whose boards of directors underwent some change as a result of the new law. 7

Mergers in Spanish savings banks were rife during the period of analysis. In 1984 there were 77 savings banks in Spain, whereas at the end of 1999 numbers had dropped to 50. This merger process was especially intense at the beginning of the 1990s. In 1990, 17 savings banks dissolved and two were absorbed, creating 7 new entities and in 1991, 9 entities dissolved and one was absorbed to create two new savings banks.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

18

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Since savings banks have no market value, it is impossible to use market value-based performance and risk measures. We therefore use two different measures of return on assets to measure performance. Earnings are measured after (ROA1) and before (ROA2) depreciation and provisions for loan losses reserves. The use of these two measures is intended to isolate the impact of possible profit smoothing that can be caused by managerial discretionality regarding the allowances for depreciation and loan losses provisions8. Following Esty (1997a), Williams (1999) and Cebenoyan et al. (1995,1999) we use the time series profit variability, defined as the standard deviation of the two measures of return on assets (RISK1 and RISK2), respectively, as a measure of risk. To analyze the change in the ROA-adjusted risk we use the coefficient of variation, where the standard deviation of the return on assets is divided by the average return on assets over the same period (CV=RISK/ROA). Change in savings banks’ performance and risk-taking is analyzed by comparing the ROA, RISK and CV during the four-year periods before and after the introduction of the regional regulations that modified the composition of the savings banks’ governance bodies in the respective region. The year when the regional law was enforced is omitted in the analysis to better separate the possible effect of the law on savings bank performance and risk. Our analysis covers the period 1984-1999, since in our sample of savings banks the first change in ownership following regional regulations occurred in 1988 and the last in 1995. Differences observed over the pre- and post-regulation periods may be due to economic factors that affect the sector which are unrelated to modifications to savings banks’ governance. In order to overcome this problem, the ROA, the standard deviation of ROA and the coefficient of variation for each savings bank in the four preceding and ensuing years are adjusted by dividing each variable by the respective median of the six savings banks that did not change their ownership or governance bodies between 1984 y 19999. The three performance, risk and risk-adjusted measures are the following: ROAit NCROAt RISKit ARISKit = NCRISKt CVit ACVit = NCCVt

AROAit =

[1] [2] [3]

where AROAit, ARISKit and ACVit are, respectively, the adjusted return on assets, the adjusted risk and the adjusted coefficient of variation of savings bank i in period t. NCROAt, NCRISKt and NCCVt, are, respectively, the median of the return on assets, risk and coefficient of variation of the six savings banks that did not experience any variation in their

8

Beatty et al. (1995) and Scholes et al. (1990), among others, have offered evidence of income smoothing in commercial banks. 9 The six savings banks that did not change their board of directors belong to three different self-governing regions: Andalucía, Castilla La Mancha and Navarra. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

The Effect of Government Ownership on Bank Profitability and Risk

19

governance bodies through out 1984 and 199910. All these variables are estimated for the preregulatory change period and for the post-regulatory change period. A descriptive analysis of each performance and risk measure, together with the percentages of public administrations (GOVERN), depositors (DEP) and employees (EMP) owning savings banks during the period of four years before and four years after the change in the ownership, is shown in Table 3. Table 3. Descriptive statistics This table presents the descriptive statistical tests of the percentages of governments (GOVERN), employees (EMP) and depositors (DEP) in savings banks’ ownership as well as the adjusted return on assets (AROA), the adjusted standard deviation of the return of assets (ARISK) and the adjusted coefficient of variation (ACV) of Spanish savings banks during the period of four years before and four years after the passing of the regional regulation. VARIABLES

GOVERN EMP DEP LN(AT) AROA1 AROA2 ARISK1 ARISK2 ACV1 ACV2 # observations

BEFORE REGIONAL LEGISLATION Mean

Median

25.875 6.208 56.791 12.080 0.858 0.781 1.076 1.165 1.347 1.640 24

28.000 5.000 54.500 12.164 0.863 0.784 0.944 0.996 1.133 1.637

Standard deviation 22.357 4.393 16.699 1.285 0.141 0.308 0.604 0.719 0.696 0.928

AFTER REGIONAL LEGISLATION

Maximun

Mínimum

Mean

Median

68.00 25.00 83.00 15.05 1.20 1.36 2.25 2.61 2.71 3.14

0.00 3.00 22.00 8.46 0.59 0.25 0.20 0.09 0.32 0.15

42.667 8.500 37.292 12.665 0.874 0.884 0.794 1.082 1.206 0.994 24

41.500 8.000 40.000 12.719 0.854 0.809 0.561 0.878 1.080 0.664

Standard deviation 16.523 3.612 6.670 1.306 0.132 0.298 0.718 0.775 0.745 1.063

Máximum

Mínimum

75.00 18.00 45.00 15.80 1.21 1.55 2.88 3.53 3.30 3.94

20.00 5.00 20.00 9.13 0.61 0.41 0.06 0.12 0.17 0.08

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5. RESULTS 5.1. Mean differences Change in savings banks’ risk levels after the increase of government ownership is initially measured through the mean differences between the period of four years before and four years after regional regulation enforcement. Three statistical tests were used to analyze the statistical significance of the change. Together with the parametric mean difference test, we used two other non-parametric tests: the Wilcoxon signed rank test and the sign test. The non-parametric tests do not require any assumptions about the distribution of the variables analyzed and are more adequate in the case of reduced sample sizes11. The Wilcoxon signed rank test analyzes whether the sum of ranks of the positive differences differs significantly from the sum of ranks of the negative differences. The sign test, which compares the number of differences that are positive with the number of differences that are negative, is a less powerful test than Wilcoxon’s since it does not take into account the magnitude of the differences. 10

Besides the estimations in this paper, we also analyzed a different period – three years before and after the passing of the regional regulation. The risk adjustment of the six savings banks that did not change their governance bodies was undertaken by differences instead of by quotient. The results were basically the same and for this reason are not reported in the paper. 11 The Shapiro-Wilk normality test has shown that all the variables meet the normality condition. The statistical test t of means is therefore justified.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

20

The differences in AROA, ARISK, ACV and the ownership percentages of each collective between the preceding and the ensuing period are shown in Table 4. Table 4. Mean differences This table shows the mean difference between the period of four years before and after the modification of savings banks’ ownership introduced by the regional regulation together with the values of the parametric statistical test of mean difference and the non-parametric Wilcoxon and sign tests. GOVERN is the percentage of government ownership, EMP is the percentage of employees’ ownership and DEP is the percentage of depositors’ ownership. AROA1 and AROA2 are the adjusted return on assets before and after, respectively, depreciation and provisions for loan losses reserves. ARISK1 and ARISK2 are the adjusted standard deviation of two previous measures of return on assets. Finally, ACV1 and ACV2 are the adjusted coefficient of variation, defined from the two measures of performance and risk previously indicated.

VARIABLES

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

GOVERN EMP IMPOSIT LN(AT) AROA1 AROA2 ARISK1 ARISK2 ACV1 ACV2

Mean difference Post value previousvalue 16.792 2.292 -19.500 0.575 0.0156 0.1029 -0.014 -0.370 -0.142 -0.646

Test t ShapiroWilk test 0.923*** 0.852*** 0.947*** 0.964*** 0.944*** 0.944*** 0.959*** 0.972*** 0.977*** 0.955***

6.74*** 1.76* -6.39*** 9.35*** 0.917 1.411 -0.09 -1.89* -0.71 -3.02***

Wilcoxon test Sum positive ranks 1.50 31.00 8.00 0.00 190.00 197.00 143.00 91.00 190.00 242.00

Sum negative ranks 274.50 140.00 292.00 300.00 110.00 103.00 157.00 209.00 110.00 58.00

Sign test Value Z

Positive differences

Negative differences

-4.159*** -2.376** -4.059*** -4.286*** -1.143 -1.343 -0.200 -1.69* -1.143 -2.629***

1 3 22*** 0 16 15 12 8 15 17*

22*** 15*** 2 24*** 8 9 12 16 9 7

* Significantly different from zero at the 10% level ** Significantly different from zero at the 5% level *** Significantly different from zero at the 1% level

All three tests indicate that changes in the composition of the savings banks’ governance bodies are statistically significant. There is a 16.8% average increase in local and regional governments’ and a 2.3% increase in employees’ participation at the expense of a 19.5% drop in depositors’ representation by 19.5%. Analysis of the change in savings banks’ size, measured through the natural logarithm of total assets, points to it increasing over both periods of time. The results of the three tests are similar when we analyze performance and risk changes. The return on assets of the savings banks that increased government ownership did not undergo any statistically significant variation between the period preceding and following the change in ownership. However, the reduction in the standard deviation of the return on assets before depreciation and provisions for loan loss reserves (ARISK2) is statistically significant. This risk reduction is also statistically significant after adjusting for performance, as shown by the change in the coefficient of variation (ACV2).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

The Effect of Government Ownership on Bank Profitability and Risk

21

5.2. Regression Analysis As Healy et al. (1992) point out, changes in savings banks’ performance and risk between the periods preceding and following ownership change may also be due to continuity in the savings bank’s trend rather than to the effect of the change in the percentage representation of each of the collectives involved in the savings banks’ governance bodies. The benchmark for post-change performance and post-change risk thus depend on their relation with those of the previous period. If there were no relation between preceding and ensuing values, the appropriate benchmark for performance and risk in the latter period would be zero and the analysis of mean difference previously carried out would offer an adequate measure of the change in both variables. Alternatively, the appropriate benchmark would be the performance (risk) in the preceding period if the savings bank that before the modification in the regional legislation has levels of performance (risk) higher or lower than those of the savings banks that do not vary the composition of their board of directors is likely to realize the same result after the regulatory change in the ownership. In order to correct this problem, the abnormal variation of ROA, RISK and CV is obtained from the intercepts of three regressions in which the adjusted values in the post-change period are used as dependent variables, and the adjusted values of ROA, RISK and CV in the preceding period as independent variables. The OLS-estimated regressions were the following: AROA

i , post

= α 0 + α 1 AROA

i , pre

+ α 2 CSIZE i + ∑ α t Y it + ε i

ARISK i , post = β 0 + β 1 ARISK i , pre + β 2 CSIZEi + ∑ βt Yit + ω i

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ACV i , post = δ 0 + δ 1 ACV i , pre + δ 2 CSIZEi + ∑ δt Yit + ξ i

[4] [5] [6]

where AROAi,post, ARISKi,post, ACVi,post are, the return on assets, risk and the coefficient of variation of savings bank i during the four-year period after the year when the regional legislation related to savings bank i was introduced, divided by the median of each equivalent variable for the six savings banks that do not modify their ownership. AROAi,pre, ARISKi,pre, ACVi,pre are, the return on assets, risk and the coefficient of variation of savings bank i for the four-year period prior to the introduction of regional legislation adjusted by the median of the 6 savings banks. The value of coefficients α1, β1 y γ1 would capture any correlation between the levels of performance, risk and performance-adjusted risk of the preceding and ensuing periods, so that (α1 AROApre,i, β1 ARISKpre,i and γ1 ACVpre,i) measure the effect of pre-regulation changes, respectively, in performance, risk and performance-adjusted risk during the post-regulation period. The intercept of each regression (α0, β0, γ0) would be, respectively, our measure of abnormal savings banks’ performance, risk and performance-adjusted risk originated by ownership change. Pre- and post-regulation changes in size, (CSIZEi), - as measured by the natural logarithm of total assets - are also controlled for in these regressions. Mean difference analysis had revealed an increase in savings banks’ size after regional legislation had been enacted, thus expanding financial intermediaries’ opportunities for diversification and potentially reducing risk levels. Thus, changes in savings bank size must be controlled for before any risk differences can be attributed to ownership change. Finally, since regional legislations were enforced in different years for each region, a set of time dummies is introduced for each year

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

22

in which regional regulations were passed (Y88, Y89, Y90, Y91, Y92, Y93, Y95). Thus, Yit takes the value 1 if savings bank ownership modifications occurred in the year t and takes the value zero otherwise. These variables are intended to control possible time effects derived from the fact that changes in regional regulations occurred in different years. The dummy for 1988 is omitted from the estimations. The results of these regressions for the 24 savings banks that modified the composition of their governance bodies are shown in Table 5. Table 5. Regressions of change in risk This table shows the OLS estimations of the equations [4], [5] and [6]. The dependent variables are the adjusted return on assets (AROA1post,i; AROA2post,i), the adjusted standard deviation of the return on assets (ARISKpost,i: ARISKpost,i) and the adjusted coefficient of variation (ACV1post,i; ACVpost,i) in the period of four years after the change in the regional legislation. Each of the variables is adjusted by quotient by the respective measure in the six savings banks that do not change their governance structure. As independent variables we introduce in each of the regressions the same dependent variable but measured in the period of four years previous to the change in the board of directors. As control variables we introduce the change in the natural logarithm of total assets that each of the savings bank experiences between the previous period and the posterior period (CSIZE) and a set of dummy variables corresponding to each of the years when a change in the regional legislation has occurred, which take the value 1 if the modification of the savings banks’ ownership has taken place in the year t and take the value 0 otherwise. In the estimations the dummy corresponding to the year 1998 is omitted. The values of the t-student test are shown in brackets.

INTERCEPT

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

AROA1pre,i

AROA1post,i (1) 0.295 (1.22) 0.838*** (5.02)

AROA2post,i (2) 1.111*** (3.28)

ARISK1post,i (3) 2.715*** (3.39)

ARISK2post,i (4) 2.478* (2.04)

ACV1post,i (5) 2,77** (2,52)

0.399** (2.72)

AROA2pre,i ARISK1pre,i

0.064 (0.31) 0.167 (0.65)

ARISK2pre,i ACV1pre,i

-0,09 (-0,39)

ACV2pre,i CSIZEi D89 D90 D91 D92

ACV2post,i (6) 1,56 (1,18)

-0.014 (-0.007) -0.130 (-1.26) -0.155* (-1.84) -0.150* (-1.91) -0.093 (-1.08)

-0.112 (-0.26) -0.509* (-1.86) -0.748*** (-3.37) -0.540** (2.73) -0.206 (-0.97)

-2.919** (-2.61) 0.876 (1.05) 0.364 (0.60) -0.362 (-0.66) 0.247 (0.44)

-4.23*** (-3.07) 0.441 (0.50) 0.116 (0.15) 0.494 (0.64) 0.835 (1.08)

-2,86** (-2,02) 1,34 (1,48) 0,42 (0,55) -0,20 (-0,28) 0,68 (0,99)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

0,45** (2,40) -4,08** (-2,50) 0,59 (0,59) 0,70 (0,84) 1,07 (1,42) 0,92 (1,22)

The Effect of Government Ownership on Bank Profitability and Risk

23

Table 5. (Continued) AROA1post, i (1) D93 -0.284** (-2.74) D95 -0.147 (-1.37) # observations 24 Adjusted R2 69.94 F 7.69***

AROA2post, i (2) -0.993*** (3.60) -0.523* (-1.90) 24 58.75 5.09***

ARISK1post ,i (3) -0.771 (-0.99) -0.368 (-0.51) 24 39.12 2.85**

ARISK2pos t,i (4) 1.653* (1.82) 0.108 (0.10) 24 26.78 2.05*

ACV1post,i ACV2pos t,i (5) (6) -0,56 4,24*** (-0,63) (4,22) -0,13 0,88 (-0,15) (0,82) 24 24 28,96% 56,49% 2,17* 4,73***

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

* Significantly different from zero at the 10% level ** Significantly different from zero at the 5% level *** Significantly different from zero at the 1% level

The intercepts of regressions (3) and (4) shown in Table 5 indicate an increase in the standard deviation of both measures of return on assets after controlling for size change and the correlation between pre-and post-regulation change. This increase in volatility goes hand in hand with an increase in the return on assets before depreciation and provisions for loan losses reserves (column 2) but not with the return on assets after depreciation and provisions for loan losses reserves (column 1). The increase in risk but not in performance after depreciation and provisions for loan losses reserves causes the coefficient of variation to also experience a statistically significant increase when performance is measured after depreciation and provisions for loan losses reserves (column 5). Size change has a negative coefficient in the regressions of the standard deviation of return on assets and the coefficient of variation. This latter result is consistent with the diminishing effect of risk that has traditionally been associated with size when size increases the diversification opportunities for thrifts. Thus, the reduction in savings banks’ risk that the mean difference analysis highlighted might be motivated more by the increase in size, than by the change in the savings banks’ ownership, since an increase rather than a reduction in savings banks’ risk is observed together with an increase in government ownership when we correct for size and for the correlation between pre-and post legal change in the regression analysis. Two additional explanatory variables – each savings bank’s change in government ownership (CGOVERNi) and change in employees’ ownership (CEMPi) triggered by regional changes in the law – are also included in the equations for a more in-depth analysis of the impact of ownership change on performance and risk. The squares of these two variables (CGOVERNQi and CEMPQi) are also applied, so as to capture potential non-linear effects, while the change in depositor ownership is omitted to avoid correlation problems. The new estimated models therefore stand as follows: AROA i , post = φ 0 + φ 1 AROA i , pre + φ 2 CSIZEi + φ 3 CGOVERN i + φ 4 CGOVERNQ i + φ 5 CEMP i + φ 6 CEMPQ i + ∑ φt Yit + τ i

ARISKi , post = ϕ 0 + ϕ 1 ARISKi , pre + ϕ 2 CSIZEi + ϕ 3 CGOVERNi + ϕ 4 CGOVERNQi + ϕ 5 CEMPi + ϕ 6 CEMPQi + ∑ ϕt Yit + ψ ACVi , post = γ 0 + γ 1 ACVAi , pre + γ 2 CSIZEi + γ 3 CGOVERNi + γ 4 CGOVERNQi + γ 5 CEMPi + γ 6 CEMPQi + ∑ γt Yit + υ i

i

[7] [8] [9]

Results of the estimations for each of the performance and risk measures are shown in Table 6. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

24

Table 6. Change in board of directors and risk This table shows the OLS estimations of the equations [7], [8] and [9]. The dependent variables are the adjusted return on assets (AROA1post,i; AROA2post,i), the adjusted standard deviation of the return on assets (ARISKpost,i: ARISKpost,i) and the adjusted coefficient of variation (ACV1post,i; ACVpost,i) in the period of four years after the change in the regional legislation. Each of the variables is adjusted by quotient by the respective measure in the six savings banks that do not change their governance structure. As independent variables we introduce in each of the regressions the same dependent variable but measured in the period of four years previous to the change in the ownership. CGOVERN and CEMP are, respectively, the variation in the government ownership and the employees’ ownership. These two variables are also introduced square (CGOVERNQ and CEMPQ). As control variables we introduced the change in the natural logarithm of total assets that experiences each of the savings banks between the previous period and the posterior period (CTA) and a set of dummy variables corresponding to each of the years when a change in the regional legislation has occurred, which take the value 1 if the modification of the composition of the board of directors of the savings bank i has taken place in the year t and take the value 0 otherwise. In the estimations the dummy corresponding to the year 1998 is omitted. The values of the t-student test are shown in brackets.

α0 AROA1pre,i

AROA1post,i (1) 0.585 (1.59) 0.750*** (3.87)

AROA2post,i (2) 0.683 (1.15)

ARISK1post,i (3) 1.321 (1.26)

ACV1post,i (5) 0,77 (0,51)

ARISK1pre,i

-0.143 (-0.99) -0.033 (-0.16)

ARISK2pre,i ACV1pre,i

-0,22 (-1,30)

ACV2pre,i CGOVERNi CGOVERNQi CEMPi CEMPQi CSIZEi D89 D90 D91 D92

ACV2post,i (6) 2,18 (1,10)

0.553*** (3.79)

AROA2pre,i

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ARISK2post,i (4) 2.205 (1.43)

-0.003 (-1.05) 0.00003 (0.51) -0.014 (-0.74) 0.0007 (0.70) -0.014 (-0.07) -0.35 (-1.24) -0.355 (-1.33) -0.320 (-1.34) -0.28 (-1.1)

-0.012* (-2.06) 0.0001 (1.41) 0.003 (0.07) -0.0008 (-0.38) 0.002 (0.007) -0.464 (-0.78) 0.468 (-0.79) -0.193 (0.37) 0.141 (0.26)

-0.017 (-1.60) 0.0008*** (4.40) 0.139 (1.72) -0.004 (-1,12) -1.56* (-2.01) -0.427 (-0.38) 0.747 (0.67) -0.032 (-0.03) 0.144 (0.14)

-0.027 (-1.75) 0.0009*** (3.41) -0.042 (-0.36) 0.802 (0.38) -2.71** (-2.42) -2.062 (-1.28) -0.585 (-0.37) -0.029 (-0.02) -0.127 (-0.09)

-0,01 (-0,68) 0,001*** (3,21) 0,20 (1,79) -0,008 (-1,31) -1,38 (-1,24) 0,24 (0,15) 1,23 (0,78) 0,44 (0,32) 0,89 (0,63)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

0,20 (1,11) -0,03 (-1,73) 0,001** (2,79) -0,07 (-0,46) 0,004 (0,53) -2,85* (-1,93) -2,66 (-1,27) -0,74 (-0,35) 0,05 (0,03) -0,51 (-0,27)

The Effect of Government Ownership on Bank Profitability and Risk

25

Table 6. (Continued)

D93 D95

AROA1post, i (1) -0.492* (-1.91) 0.287 (-1.09)

Turning point of CGOVERNi # observations 24 Ajusted R2 67.33 F 4.95***

AROA2pos t,i (2) -0.773 (-1.35) -0.098 (-0.17)

ARISK1pos t,i (3) 0.179 (0.16) -1.636 (-1.59) 9.78%

ARISK2pos t,i (4) 1.366 (0.89) -1.774 (-1.14) 14.14%

ACV1pos t,i (5) 0,83 (0,55) -1,34 (-0,90) 5.84%

ACV2pos t,i (6) 3,37 (1,67) -1,66 (-0,85) 16.28%

24 67.93 5.06***

24 75.43 6.89***

24 59.39 3.80**

24 63,49% 4,33**

24 68,73% 5,21***

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

* Significantly different from zero at the 10% level ** Significantly different from zero at the 5% level *** Significantly different from zero at the 1% level

The negative coefficient of CGOVERN in regression (2) highlights that savings banks that most increased their government ownership were those that least increased their return on assets before depreciation and provisions for loan losses reserves. Again, we do not observe any relation between ownership change and the return on assets after depreciation and provisions for loan losses reserves in regression (1). In contrast, the positive sign of CGOVERNQ in regressions (3) and (4) indicates that the rise in the standard deviation of both measures of return on assets was greater for those savings banks that most increased government ownership after the introduction of regional regulation. Far from this positive relation between change in government ownership and change in the standard deviation of return on assets being observed for every type of change in government ownership, it is only observed, in fact, for changes in ownership of above 9.78% or 14.14%, depending on the measure of the return on assets used. Furthermore, the positive relation between risk and government ownership holds when we adjust risk by performance in columns (5) and (6). For changes in government ownership higher than 5.84% and 16.28%, in the case of ACV1 and ACV2 respectively, there is a positive relation between the change in the risk adjusted by performance and the variation in government ownership of savings banks. This positive relationship is not consistent with the argument that politicians tend to limit savings banks’ risk to a level that guarantees their solvency if faced with the threat of losing a useful instrument to reach political goals. On the contrary, the relation found between variation in risk and variation in government ownership suggests a positive influence of political presence on savings banks’ risk levels, irrespective of whether it is performance adjusted or not. This result is consistent with the fact that the objective of reaching political goals in decision-taking may not coincide with the objective of economic efficiency and may in fact increase the thrifts’ risk. Similarly, the result is also consistent with the options public administrations enjoy of transferring funds to savings banks through the commercial relationship they maintain with them, and of compensating losses provoked by politically-motivated decisions with fund transfers.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

26

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

Unlike the results for the shift towards government ownership, we do not observe a significant relation between the change in employees’ ownership and the level of postlegislative performance and risk.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

6. CONCLUSIONS Although there are both mutual and government-owned banking institutions, possible differences in performance and risk between them have yet to be analyzed. This paper provides some evidence on this issue by analyzing Spanish savings banks’ change in performance and risk after regional regulations increased political control of decision-taking at the expense of reducing depositors’ ownership. Although mean difference analysis initially seems to indicate that there has been some reduction in savings banks’ risk in the four years after the change in ownership with respect to the four years before that change, regression analysis reveals that an increase in savings banks’ risk level occurs when the correlation between pre- and post-legislative change and the change in the thrifts’ size are controlled. The analysis of mean differences fails to observe significant variations of savings banks’ performance after the ownership whereas the regression analysis reveals an increase in the return on assets before depreciation and provisions for loan losses reserves; this increase is in any case smaller, the larger the increase in government ownership. The smaller performance increase compared to risk increase observed in Spanish savings banks subject to increased government ownership leads to an increase in performanceadjusted risk, i.e. in the coefficient of variation. Furthermore, we observe that the coefficient of variation rises in line with enhanced government ownership for increases in political presence higher than 5.84% and 16.28%, depending on whether performance is measured after or before depreciation and provisions for loan losses reserves. In short, results obtained suggest a positive influence of public administrations on savings banks’ risk level but not on their performance. These findings are consistent with results of studies carried out in other countries indicating that the pursuit of political objectives in decision-taking may facilitate more risky investments, and that the option of fund transfer that is a spin-off of the commercial relationship between savings banks and regional governments may also facilitate this type of heightened risk.

REFERENCES Altunbas, Y., Evans, L. and Molyneux, F. (2001). Bank ownership and efficiency. Journal of Money, Credit and Banking 33, 926-954. Barth, J. R., Caprio G. Jr. and Levine R. (2001). Banking systems around the globe:do regulations and ownership affect performance and stability?. Frederic S. Mishkin, Editor: Prudential supervision: What works and what doesn’t, University of Chicago Press.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The Effect of Government Ownership on Bank Profitability and Risk

27

Beatty, A., Chamberlain, S.L. and Magliolo, J. (1995). Managing financial reports of commercial banks: the influence of taxes, regulatory capital and earnings. Journal of Accounting Research 33, 231-261. Blair, D. and Placone D. (1988). Expense preference behavior, agency cost, and firm organization. Journal of Economics and Business 40, 1-15. Boardman, A. and Vining A. (1989). Ownership and performance in competitive environments: a comparison of the performance of private, mixed and state-owned enterprises. Journal of Law and Economics 32, 1-33. Cebenoyan, A.S., Cooperman, E.S. and Hudgins, S. (1993). The relative efficiency of stock versus mutual savings and loans: a stochastic cost frontier approach. Journal of Financial Services Research 7, 151-170. Cebenoyan, A.S., Cooperman, E.S. and Register, CH.A. (1995). Deregulation, reregulation, equity ownership and S&L risk-taking. Financial Management 24, 6376. Cebenoyan, A.S., Cooperman, E.S. and Register, CH.A. (1999). Ownership structure, charter value and risk-taking behavior for thrifts. Financial Management 28, 43-60Cordell, L.R., Macdonald, G.D. and Wohar M.E., (1993). Corporate ownership and the thrift crisis. Journal of Law and Economics 36, 719-756. Daniels, K.N. and Sfiridis, J.M. (2001). The relative cost efficiency of stock versus mutal thrifts: does organizational form matter?. Paper presented at 2001 FMA Annual Meeting, Toronto. Dewatripont, M. and Tirole J. (1993a). Efficient governance structure: implications for banking regulation. Mayer, C., Vives X., Eds., Capital Markets and Finacial Intermediation. Cambridge University Press, 12-35. Dewatripont, M. and Tirole J. (1993b). The prudential regulation of banks. Editions Payot, Lausanne. Demsetz, H. and Lehn K. (1985). The structure of corporate ownership: causes and consequences. Journal of Political Economy 93, 1155-1177. Esty, B.C. (1997a). Organizational form and risk taking in the savings and loan industry. Journal of Financial Economics 44, 25-55. Esty, B.C. (1997b). A case study of organizational form and risk shifting in the saving and loan industry. Journal of Financial Economics 44, 57-76. Grifell- Tatjé, E. and Lovell C. (1997). The sources of productivity change in Spanish banking. European Journal of Operational Research 98, 364-380. Hadaway, B.L. and Hadaway S.C. (1984). Implications of savings and loan conversions in a deregulated world. Journal of Bank Research 15, 44-55. Hansmann, H. (1988). Ownership of the firm. Journal of Law, Economics and Organization 4, 267-303. Healy, P.M., Palepu K.G. and Ruback R.S. (1992). Does corporate performance improve after mergers?. Journal of Financial Economics 31, 135-175. Jensen, M.C. and Meckling, E.H. (1979). Rights and production functions:an application to labor-managed firms and codetermination. Journal of Business 52, 469-506. Karels G.V. and Mclatchey C. (1999). Deposit Insurance and risk-taking behavior in the credit union industry. Journal of Banking and Finance 23, 105-134.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

28

Ana Isabel Fernández, Ana Rosa Fonseca and Francisco González

Kaufman, G.G. (1999). Helping to prevent banking crisis: Taking the “State” out of State banks. Review of Pacific Basin Financial Markets and Policies 2, 83-98. Keeley, M. C. (1990). Deposit Insurance, risk, and market power in banking. The American Economic Review 80, 1183-1200. La Porta, R., Lopez-de-Silanes F. and Shleifer A. (2002). Government ownership of banks. Journal of Finance 57, 265-301. Lamm-Tennant, J. and Starks L.T. (1993). Stocks vs. mutual ownership structures: the risk implications. Journal of Business 66, 29-46. Lozano, A. (1998). Efficiency and technical change for Spanish banks. Applied Financial Economics 8, 289-300. Masulis, R.W. (1987). Changes in ownership structure: Conversions of mutual saving and loans to stock charter. Journal of Financial Economics 18, 29-59. Megginson, W.L. and Netter J.M. (2001). From state to market: a survey of empirical studies on privatization. Journal of Economic Literature 39, 321-389. Melle, M. and Maroto J.A. (1999). Una aplicación del gobierno de empresas: incidencia de las Administraciones Públicas en las decisiones asignativas de las cajas de ahorros españolas. Revista Europea de Dirección y Economía de la Empresa 8, 9-40. Mester, L. (1991). Agency costs among saving and loans. Journal of Financial Intermediation 1, 257-278 Mester, L.J. (1993). Efficiency in the saving and loan industry. Journal of Banking and Finance 17, 267-288. O’Hara, M. (1981). Property rights and the financial firm. Journal of Law and Economics 29, 317-332. O’Hara, M. and Shaw, W. (1990). Deposit insurance and wealth effects: the value of being “too big to fail”. Journal of Finance 45, 1587-1600. Rasmusen, E. (1988). Mutual banks and stock banks. Journal of Law and Economics 31, 395-421. Sapienza, P. (1999). What do state-owned firms maximize? Evidence from the Italian Banks. Northwestern University Mimeo. Scholes, M.S., Wilson, G.P. and Wolfson, M.A. (1990). Tax planning, regulatory capital planning and financial reporting strategy for commercial banks. The Review of Financial Studies 3, 625-650. Schrand C. and Unal H. (1998). Hedging and coordinated risk management: Evidence from thrift conversions. Journal of Finance 53, 979-1013. Shleifer, A. (1998). State versus private ownership. Journal of Economic Perspectives 12, 133-150. Tulkens, H. (1993). On FDH efficiency analysis: some methodological issues and applications to retail banking, courts and urban transit. Journal of Productivity Analysis 4, 183-210. Valnek, T. (1999). The comparative performance of mutual building societies and stock retail banks. Journal of Banking and Finance 23, 925-938. Verbrugge, J.A., Megginson W.L. and Owens W.L. (1999). State ownership and the financial performance of privatized banks: an empirical analysis. Paper presented at

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

The Effect of Government Ownership on Bank Profitability and Risk

29

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

the World Bank /Federal Reserve bank of Dallas Conference on Banking Privatization, Washington, D.C.. Verbrugge, J.A. and Goldstein S.G. (1981). Risk return and managerial objetives: some evidence from the saving and loan industry. Journal of Financial Research 4, 45-58. Verbrugge, J.A. and Jahera J.S. (1981). Expense preference behavior in the savings and loan industry. Journal of Money, Credit and Banking 13, 405-476. Williams, Z.D. (1999). CEO control and project selection: Evidence from mutual thrift conversions. Working paper, Haas School of Business, University of California, Berkeley.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 2

MANAGERS’ DISCRETIONARY BEHAVIOR, EARNINGS MANAGEMENT AND CORPORATE GOVERNANCE: AN EMPIRICAL INTERNATIONAL ANALYSIS ∗

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga** University of Valladolid, Economics and Business School Avda. Valle del Esgueva 6, E-47011 Valladolid, Spain

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT In the present debate about corporate governance, boards of directors play a more and more relevant role. From this point of view, the board of directors becomes an important mechanism for monitoring managers’ discretionary behavior. Among the several ways for measuring discretionarity, we have focused on accrual decisions in order to manage earnings. This is precisely the main contribution of our paper: we have tested the ability of the board of directors for monitoring by using the discretionary component of earnings management as an indicator of managers’ discretionarity. Using a sample of 450 non-financial companies from 10 OECD countries, we have found empirical evidence showing the link between some features of the board of directors and the discretionary manipulation of financial statements carried out by managers. Specifically, and as most of the literature has pointed, our main result stresses the positive and robust impact of board size on earnings management. As far as other characteristics of boards of directors are concerned, our research does not lend conclusive support to the effect of board composition or meeting frequency on earnings management.

∗

The authors are grateful to Spencer Stuart for providing the data and to L. Caprio, R. Crespí, P. Fuertes and the participants in the European Financial Management Conference held in Lugano and the XI Congreso de ACEDE (Zaragoza). All the remaining errors are our only responsibility. This research has received financial support from the Spanish Dirección General de Enseñanza Superior e Investigación Científica (proyecto PB97-0594). ** Correspondence author. Tel. +34-983-4230 00; Fax +34-983-423899; [email protected]

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

32

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

Keywords: Accruals, boards of directors, corporate governance, earnings management, managers’ discretionarity, OECD.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION The separation between finance and management in the large modern companies grounds the so-called general agency problem of capitalism, “the credibility problem facing entrepreneurs of firms when they seek to convince outside investors to contribute funds” (Berglöf, 1997). The resolution of this problem, most of the times presented in terms of the separation of ownership and control, demands the outline of mechanisms of corporate governance which allow an efficient manager’s control (Shleifer and Vishny, 1996). This issue becomes relevant since, as Macey (1998) underlines, shareholders face the problems of collective action preventing them of monitoring and disciplining managers of the company they are investors of. From this point of view, corporate governance can be defined as the process through which investors try to minimize contractual costs raised by the transactions inside the company. These costs are those related both to the separation of finance and management and to the conflict of interests among all the claimholders of cash flow. This is why, in order to achieve a good corporate governance, boards of directors, by law, are charged with such an important role. Shareholders delegate their decision and control rights in the board and the boards delegate the company running in the managers. Therefore, the board of directors is supposed to monitor the managers of the firm by delegation of shareholders. In recent years, scientific literature has reflected the increasing attention boards of directors are paid and a number of papers have tried to evaluate their influence and the degree to which boards achieve their aim. As synthesized by Hermalin and Weisbach (2000), most of the empirical work can be classified into three main fields: (1) How board characteristics such as size or composition are related to firm performance; (2) How these features impact on board actions; and (3) What factors affect the makeup of boards and how they evolve over time. Although previous research has achieved some evidence such as the negative relationship between board size and firm performance and the indetermination about the link between board composition and firm efficiency, there is still a number of issues demanding further analysis. Our paper adds to the existing body of work in this area by studying the effect of some board characteristics and actions on monitoring and controlling managers discretionarity. The innovation of the paper relies on the use of accruals and earnings management as a measure of discretionary behavior. Accruals are understood as some means of manipulating financial statements for the sake of managers’ profit12. Among the earnings management instruments potentially available to managers, the literature has stressed accruals and accounting method choices (McNichols y Wilson, 1988; Watts y Zimermman, 1990; Peasnell et al., 1999a). As a consequence of it, several models of detecting and identifying earnings management have 12

Of course, accruals are not always used for wrong intentioned manipulations. Since cash flows suffer from timing and matching problems, they might be an inaccurate measure of firm performance. In order to mitigate these problems, accounting rules allow to use accruals to alter the timing of cash flows recognition in earnings (Dechow, 1994).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

33

arisen (Healy, 1985; DeAngelo, 1986; Jones, 1991; Dechow and Sloan, 1991; Dechow et al., 1995 and Peasnell et al., 2000a), having all of them in common the distinction between the discretionary and the non-discretionary component of accruals. The accruals discretionary component is the most affected by the managers’ ability and necessity of modifying the firm accounts and, hence, it is the most dependent on the ability of corporate governance for reducing managers’ discretionary behavior. In this framework, our aim is to analyze the relationship between boards of directors and discretionary accruals for a sample of 450 OECD quoted companies: by cross-section regressions we firstly estimate discretionary accruals and, secondly, we relate these accruals to several characteristics and actions of boards. The most outstanding result reveals that higher accruals are linked to those companies having the least efficient boards of directors in terms of monitoring ability – mainly due to too large boards. Our paper is divided into five sections. Section 2 introduces earnings management and connects this topic with the general problem of corporate governance. Section 3 presents the empirical foundations of the paper, the sample and the methodology, while results are provided in paragraph 4. There is a final section where some concluding remarks are exposed and some directions for future research are suggested.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2. MANAGERS’ DISCRETIONARY BEHAVIOR, EARNINGS MANAGEMENT AND BOARDS OF DIRECTORS As a consequence of the separation of ownership and control, shareholders face collective action problems preventing them from monitoring and disciplining firm managers. Besides, the delegation of decision rights enhances managers’ discretionary decisions. This is why asymmetric information makes shareholders run the risk of adverse selection and moral hazard on managers side. The agency theory can shed some light on this problem and emphasizes the idea that managers’ behavior should be featured as being broadly discretionary. Because of their jobs, some inside information is only available to managers and not to outside investors. Additionally, this information is usually disclosed lately, so that acquiring information is costly either due to the limited rationality hypothesis or because the disclosure could benefit firm competitors. Therefore, information becomes imperfect and enhances managers’ discretionarity to look for their own utility instead of maximizing the value of company owners. The problems shareholder and other stakeholder face for controlling managers’ behavior lead to the lack of reliable information about managerial acts. In turn, some internally generated measures of firm performance to be reported over intervals are specially necessary (Dechow, 1984). Consequently, earnings are used as a summary measure of managers’ efficiency for a wide range of users: for example, in executive compensation plans, in debt covenants, in the reports of firms seeking to go public, etc. Notwithstanding, managers are fully conscientious of the informative content of earnings and try to alter financial statements for the sake of their own utility and prestige. A number of acts, aimed to modify earning reports and, more generally, to vanish managers’ discretionary behavior, arises. Although newspapers and not too specialized journals have often reported

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

34

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

this situation, scientific literature has also proved and confirmed it. Hayn (1995) and Burgstahler and Dichev (1997) provide evidence that firms manage reported earnings to avoid decreases and losses of earnings. At the same time, the incentives to exercise discretion are bigger when the level of earnings is near a round number (Carslaw, 1988; Thomas, 1989). Not only earnings are not uniformly distributed around zero –what would be justified if earnings smoothing took place–, but also there is a strong bias toward positive values. So, whereas it is quite possible that companies have little positive earnings, the proportion of companies having losses on the same amount is considerable smaller. As well as that, as predicted by positive accounting theory (Watts and Zimmerman, 1986), firms approaching covenant violation makes income-increasing accounting choices to loosen their debt constraints (DeFond and Jiambalvo, 1994; DeAngelo et al., 1994). A foregone conclusion is the existence of earnings management, that is to say, a set of accounting choices to modify earnings reports for the sake of managers’ benefits. As shown by previous research (Peasnell et al., 2000b), there is a series of earnings management instruments ranging from asset sales (Bartov, 1993 and Black et al., 1998) or changes in RandD expenditure (Bange and DeBondt, 1998 and Bushee, 1998) to pure financial reporting decisions such as accounting method changes (Watts and Zimmerman, 1986) or accrual choices (McNichols and Wilson, 1988). Accruals choices have been recently paid a lot of attention to and have become the core of a promising research field. Among the causes explaining the increasing importance of accruals in earnings management, according to Peasnell et al. (1999a) their relative low cost and opaque nature should be referred to. Furthermore, accrual-based measures are appealing not only from a pragmatic point of view but also because theoretically they aggregate into a single measure the net effect of several decisions of measurement and recognition (Watts and Zimmerman, 1990). In spite of some authors supporting accruals role as an instrument for improving the ability of earnings to measure firm performance compared to other measures of firm efficiency (Bowen et al., 1987; Dechow, 1994; Subramanyam, 1996), the above mentioned reasons explain the necessity of recognizing the factors affecting earnings. This recognition is more and more necessary given that capital markets and analysts are not always able to identify them (Sloan, 1996; Bradshaw et al., 1999). Keeping this in mind, and in order to find out the accruals-based earnings management, a number of models has been arisen such as the Healy model (1985), the DeAngelo model (1986), the Jones model (1991), the modified Jones model (Dechow et al., 1995), the industry model (Dechow and Sloan, 1991) and the margin model (Peasnell et al., 2000a)13. All of them seem to be well specified and capable of generating relatively powerful tests for economically plausible levels of accruals management. All the accrual-based models for the detection of earnings management have in common the division of accruals into two different components: the non-discretionary accruals and the discretionary ones. While the first ones include those accruals oriented to improve the information content of earnings reports, the discretionary accruals are those related to managers’ concern in their own interests. On the one hand, non-discretionary accruals are likely to depend on the level of business of the firm, so they are usually calculated according 13

An interesting comparison and evaluation of these models for detecting earnings management in terms of specification and power of test statistics can be found in Dechow et al. (1995) and Peasnell et al. (2000).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Managers’ Discretionary Behavior, Earnings Management …

35

to firm sales and Property, Plant and Equipment (hereinafter, PPE). On the other hand, discretionary accruals are affected by managers ability and necessity to manipulate firms accounts, so they are conditioned by corporate governance to limit managers discretionarity. As far as the relationship between the mechanisms of corporate governance and discretionary accruals is concerned, most of the empirical evidence available focuses on the effect of the board of directors. A close relationship between board composition and earnings management has been found, so that the higher the proportion of outside directors the less the likelihood of earnings management (Peasnell et al., 1999a). Moreover, outside directorships are the main factor for reducing managers discretionarity, so that even the presence of an audit committee does not significantly affect the likelihood of the fraud of financial statement (Beasley, 1996). This negative relationship is more evident when managerial stock ownership is relatively low (Peasnell et al., 1999d). Along with board composition, and as other board of directors focused papers have underlined (Yermack, 1996, Eisenberg et al., 1998), the size of the board is supposed to have a negative impact on the quality financial statement and the higher the number of directors, the easier the accounting fraud (Beasley, 1996). However, there are other links between the characteristics of corporate governance and the choices of earnings management. Firstly, managerial stock ownership is positively related to the explanatory power of earnings and, at the same time, reduces discretionary accruals (Warfield et al., 1995). Moreover, it is a nonlinear relationship: as forecast by the entrenchment and the convergence hypothesis, the sign of the relationship for high enough levels of managerial ownership is the opposite to the one for low level of ownership. Secondly, institutional ownership has a noticeable influence on managers’ discretionary behavior. In spite of the traditional view of the institutional investors as transient and too short-termed focused (Porter, 1992), the pressure of institutional ownership reduces discretionary accruals and earnings manipulation (Rajgopal et al., 1999) and makes stock prices tend to reflect a relatively greater proportion of the information in the earnings of future period (Rajgopal and Venkatachalam, 1998).

3. DATA AND METHODOLOGY 3.1. Sample Our sample has been built by combining two data bases: the Spencer Stuart Board Index providing information about the boards of directors for 1996 and the Standard and Poors’ Global Vantage containing economic and financial firm-level data. After dropping out banks and assurance firms, the Spencer Stuart Board Index constrains the available information up to 450 non-financial companies from ten countries. Three out of the ten countries could be classified as an Anglo-Saxon or market-oriented corporate system (Canada, USA and UK), whereas the seven remaining countries belong to the so-called continental or bank-oriented corporate system (Germany, Belgium, Spain, France, Holland, Italy and Switzerland). As reported in table 1, there are big differences in terms of size and number of companies per country. However, there is not any sample bias on our part since the observations correspond to the largest listed companies in each country.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

36

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga Table 1. Number and size of included companies by country # of companies

Belgium Canada France Germany Italy

12 79 42 33 56

Average size (mill $ sales) 4787.37 3270.19 9819.12 7881.16 2531.88

# of companies Average size (mill $ sales) Holland 37 7339.03 Spain 28 1015.68 Switzerland 17 7224.85 United Kingdom 66 13581.53 U.S.A. 80 41456.74 Total 450 12770.90

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The method used by Spencer Stuart allows for homogeneous information on the companies from the ten countries. Notwithstanding, the collected information is not completely equivalent due to the different requirements each country imposes, so that the countries of the Anglo-Saxon model release more information than their counterparts. In short, the Board Index provides information on board size and composition (number of insiders, outsiders and foreign directors), the number of meetings per year and the compensation the members of the board receive. For some countries there is information available about directors' average age, directors' average tenure, the number of committees the board delegates its authority to and, just for a few counties, the kind of committees. On its side, Global Vantage Database provides with market information and financial statement information about firms. This information has been basically used for calculating the total accruals, the explanatory variables and the control variables. The method used by Standard and Poors also allows for homogeneous information on the companies.

3.2. Variables and Methodology As other previous authors, we use a two-stage approach to partition total accruals into their managed and unmanaged components: we first estimate discretionary accruals as the residuals of total accruals regression and subsequently we find out the impact of corporate governance on abnormal accruals. Since firm data are available just for one single year, we are forced to use a cross-sectional approach, which allows us to overcome the survivorship bias problem inherent in the time-series version of some accruals models. Total accruals are defined following the Jones and the modified Jones models we referred to in section 2. Total accruals are calculated as variation of non-cash working capital minus amortization and depreciation of PPE. The set of independent variables that total accruals will be regressed on includes the change in sales (Δ REV) and the gross level of PPE. The intuition underlying the choice of these variables is that they help to control for unmanaged accruals associated with changes in economic activity and the depreciation charge. As far as the forecast sign of these variables is concerned, there is a clear difference: whereas the second one is obviously positive, it is not easy to predict any sign for the change in sales. On the one hand, the higher the sales revenues the higher the receivables but, on the other hand, increases in sales usually imply increases in short-term commercial debt, so the net effect on working capital is uncertain. In any case, the

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

37

specification of the Jones model for calculating total accruals would be (all the variables are scaled by total assets): TAit = α + β1 Δ REVit + β2 PPEit + εit

[1]

This model tends to measure discretionary accruals with error when discretion is exercised over revenues. In order to avoid that shortcoming, in the modified Jones model the change in revenues is adjusted for the change in receivables, implicitly assuming that all changes in credit sales result from earnings management14. Taking into account that all variables must be scaled by total assets, in the modified model, total accruals are estimated as: TAit = α + β1 Δ REVMit + β2 PPEit + εit

[2]

where Δ REVMit is the difference between the change in sales and the change in net receivables. As explained previously, equations [1] and [2] are the first stage for estimating discretionary or abnormal accruals (AA) given that these accruals are defined as the residuals of the estimation. Hence, discretionary accruals are calculated as:

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

AAi = TAi – [a + b1 Δ REVi + b2 PPEi]

[3]

where a, b1, and b2 are the estimates of the parameters α, β1 and β2. Discretionary accruals (AA) proxy managers’ ability to deviate from standard accounting rules for their sake, so they are extremely useful for analyzing the efficiency of the mechanisms of corporate governance. Since our purpose is to evaluate the role of the board of directors in reducing managers’ discretionarity, we should explain abnormal accruals based on the board characteristics and functioning. There is a quite large number of features of boards of directors which could be used for characterizing boards. Recent literature about boards is basically empirical and focuses on three main questions: the size of the board (Eisenberg et al., 1998; Fernández et al, 1998; Huther, 1997; Jensen, 1993 and Yermack, 1996); its composition and independence (Baysinger and Butler, 1985; Bhagat and Black, 1998; Hermalin and Weisbach, 1988 and 1991; Rosenstein and Wyatt, 1990 and 1997; Weisbach, 1988); and, its internal structure and functioning (Klein, 1998). These papers show the negative influence of board size on firm value, the indetermination of the effect of board independence on firm value and a certain endogenous relationship among directors turnover, board composition and some external elements. This paper joins this empirical literature by analyzing the influence of board size, composition and functioning on firm value. Consequently, we should explain how we have dealt with each one of those boards characteristics. Firstly, we have measured board size as the logarithm of the number of directors (TAMC). As reported in table 2, the mean board size is 11.7 directors, which is consistent with the figure of 12 reported by Yermack (1996), Barnhat et al. (1994), Rosenstein and Wyatt (1997) and Klein (1998), and slightly higher than 14

The underlying intuition is that it is easier to manage earnings by exercising discretion over revenue on credit sales than over the revenue on cash sales.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

38

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

Fernández et al. (1997) and Denis and Sarin (1998) data, whose results are around 9 directors per board. The differences in board size are closely related to the firm size and country features. The second characteristic of the board, its composition, is measured by the proportion of outside directors15 (EXTP). This variable has been usually considered to proxy board independence as long as outside directors are not managers. We would have liked to distinguish between grey16 directors and true outside directors, but the available information was not enough to separate both groups. As displayed in table 2, the average value of the percentage of outsiders is high, which means that more than half of the boards were dominated by outsiders (we consider it an outsider-dominated board when OUTSID is higher than 0.7) while insider-dominated boards (INSID) were only 17% of the sample. This is relevant because outsider-dominated boards are a priori more independent and are supposed to monitor more effectively. Nonetheless, the empirical evidence is not conclusive and balanced boards, which benefit both from insider expertise and from outsider independence, seem to be preferable. With reference to the internal functioning of boards, there is an unbalance between the great amount of factors related to it and the scarcity of information available about them. We have chosen the number of board meetings held per year (COMYEA) as an indicator of the extent to which directors feel involved in managers control. Had only board related explanatory variables be included we could have introduced a serious omission bias.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 2. Variables descriptive statistics Total accruals, revenues and PPE have been scaled by total assets. Sales and assets are in million of US dollars. PBV is the equity market-to-book ratio

TA REV PPE TAMC EXTP INSID COMYEA AA PBV ROA ADVERSE ADV97 TASS SALES

no. 387 403 350 419 381 381 408 393 419 418 428 428 406 404

Mean -5,760E-2 6,348E-2 0,5667 11,7470 0,7082 0.2831 6,5944 0,1714 3,3786 5,873E-2 0,1192 9,579E-2 10939,15 9517,57

Std. dev. 7,823E-2 0,1734 0,2891 3,7324 0,1971 0.1959 3,1658 0,1450 3,6581 7,848E-2 0,3244 0,2947 22944,17 16814,08

Minimum -0,52 -0,65 0,00 4,00 0.00 0.00 0,00 -0,3340 0,11 -0,43 0,00 0,00 27,90 0,00

15

Maximum 0,29 0,82 2,67 26,00 1,00 1.00 19,00 0,8648 28,74 0,82 1,00 1,00 243283,0 165370,2

According to most of the previous literature, Spencer Stuart Board Index defines inside directors as those directors having executive responsibilities (managers), whereas outsiders are the non-manager directors. 16 Grey directors are those keeping any special relationship with the firm, either due to professional services, family ties or any other sort of non-managerial links. Nevertheless, it is not the only possible classification and there might be other kinds of special status directors as the blockholders, i.e. those owning a significant proportion of stocks.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

39

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 3. Descriptive statistics by country Dutch boards are a special case since companies are forced by law to have a two-tier board system with a Supervisory Board and a Management Board. These is why the proportion of outside directors is not reported

CAN

no. 69

USA

71

UK

61

SP

22

FR

41

IT

25

GE

30

NETH

36

BE

10

SW

16

Mean σ Mean σ Mean σ Mean σ Mean σ Mean σ Mean σ Mean σ Mean σ Mean σ

AA96 0,2326 0,1659 0,1728 0,1440 0,2356 0,1288 0,0860 0,1638 0,1561 0,0690 0,1441 0,1261 0,0914 0,1180 0,1274 0,1727 0,1474 0,1063 0,1257 0,0791

LNTAMC 2,4981 0,1905 2,5328 0,1670 2,4674 0,2359 2,4966 0,4047 2,5138 0,3090 2,2300 0,2926 2,6910 0,3338 1,8762 0,2862 2,5195 0,4210 2,1620 0,2669

EXTP 74,88% 17,11% 78,07% 11,21% 48,00% 12,19% 74,07% 19,99% 81,19% 15,67% 73,93% 25,21% 59,82% 9,85%

77,93% 9,01% 89,90% 12,04%

ROA 4,18% 7,28% 7,61% 6,53% 10,06% 6,28% 8,39% 7,65% 5,35% 12,81% 3,66% 5,52% 2,91% 2,44% 5,30% 3,93% 5,11% 6,26% 4,40% 6,28%

TASS 3441,4 4556,4 18375,8 15586,0 8201,6 8889,9 3384,8 4607,6 11875,1 13236,1 5047,7 12196,2 12892,5 19085,3 4137,6 6329,7 5748,3 7551,8 5136,0 9391,8

Therefore, we have incorporated a set of variables related to incentives of earnings management and firm performance as firm size, firm profitability and some other control variables. Regarding firm size, two proxy variables have been taken into account: total assets (TASS) and sales (SALES). As regards firm profitability we have chosen ROA as an indicator of firm performance. Later on, we will introduce a set of dummy variables (based either on the SIC classification or on nationality criteria) in order to incorporate both industry and country effects. These last groups of variables enhance a sensitivity analysis to test the robustness of the results to the influence of industry and country effects. Finally, another dummy variable concerning the audit report is also used (ADVERSE). These variables equals 1 when the audit report reveals serious and negative reservations about the quality of the financial statements of the firm or the fulfillment of the standard accounting principles and equals 0 otherwise. The variable is defined for 1997 year too (ADV97). This new variable is interesting not only because of its additional explanatory power but also because it allows the division of the whole sample into two sub-samples, so that differentiated effects can be found. In addition, it would have been very useful to have information available about the existence of the audit committee, since the chances are that such a committee reduces managers’ discretionarity. Unfortunately, our data base does not provide

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

40

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

with those data17. Although the auditor’s opinion allows to overcome partially that missing data, there is no doubt that information about audit committee would have been more accurate. All these variables are displayed in table 2, where mean, minimum and maximum values are reported along with the standard deviation after having dropped out the observations whose data could not be adequately identified. In addition, table 3 reports information, to a country-level, for the variables to be the basis of the estimation. After having calculated the discretionary accruals, the models to be estimated are the following equations: AAit = α+β1 lnTAMCit+β2 EXTPit+β3 COMYEAit+β4 ROAit+β5 SIZEit+εit

[4]

AAit = α+β1 lnTAMCit+β2 EXTPit+β3 COMYEAit+β4 ROAit+β5 SIZEit+β6 COUNTRY+εit

[5]

AAit = α+β1 lnTAMCit+β2 EXTPit+β3 COMYEAit+β4 ROAit+β5 SIZEit+β6 INDUSTRit+εit

[6]

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

4. RESULTS As previously stated, the first stage in the empirical analysis tends to measure the achievement of discretionary accruals as the residuals of the ordinary least squares estimation of equations [1] and [2]. Results reported in table 4 show, as theoretically forecast, a negative impact of PPE on total accruals and an uncertain effect of sales. These results, however, do not deserve further comments and have only instrumental interest as long as they proxy manager’s discretionarity as abnormal accruals. These results are the basis for estimating the influence of the board of directors on managers’ discretionary behavior. To do so, we have regressed discretionary accruals on the characteristics of boards and control variables. Results are displayed in table 5. Some issues should be stressed. The most outstanding and significant result is the role played by the size of the board for limiting managers’ discretionarity. All the estimations run have in common a clear and strong positive relationship between discretionary accruals and the number of directors. This result confirms the theoretical foundations the literature has suggested for justifying the effect of board size. Obviously, the higher the number of directors the more-in-depth the control undertaken by boards since there are more people devoted to monitor managers. However, a high number of directors implies some difficulties due to the poorer coordination, communication or flexibility associated with large boards, so that the benefits of more detailed monitoring seem to be outweighed by these problems. This last kind of relationship is the most frequent and is empirically supported by a number of papers (Lipton and Lorsch, 1992; Jensen, 1993; Yermack, 1996; Huther, 1997; Eisenberg et al., 1998).

17

Spencer Stuart Board Index provides information about the board of directors committees just for four out of the ten countries in the sample, so it would have dramatically reduced the number of observations.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

41

Table 4. Estimation of discretionary accruals Ordinary least squares cross-section estimation. The dependent variable is total accruals (TA). Variables have been scaled by total asset. Table presents the variables coefficients and p-value (in brackets) along with some statistics about the whole estimation. (***) stands for significant to a confidence level higher than 99%, (**) for a level higher than 95% and (*) for a level higher than 90% Model 1 Intercept REV

Model 2 -.002 [.757]

-0.002 [.552] .021 [.685]

REVM PPE R2

-.359 [.000] 12.6 %

Adj.-R2

12.1 %

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

F-Ratio

24.44 [.000]

-.020 [.700] -.349 [.000] 12.4 %

***

***

11.9 % ***

***

24.01 [.000]

Table 5. Discretionary accruals and corporate governance Ordinary least squares cross-section estimation. The dependent variable is discretionary accruals (AA), that is to say, Jones model residuals. Variables have been scaled by total asset. Table presents the variables coefficients and p-value (in brackets) along with some statistics about the whole estimation. (***) stands for significant to a confidence level higher than 99%, (**) for a level higher than 95% and (*) for a level higher than 90%

Intercept BOASIZ(log) EXTP ROA

(1) -0.019 [.793] 0.133 [.016] 0.038 [.476] 0.136 [.012]

**

**

(2) -0.006 [.552] 0.118 [.036] 0.038 [.479] 0.136 [.012]

COMYEA SALES

-0.030 [.583]

TASS R2

3.4 %

0.023 [.678] 3.4 %

**

**

(3) -0.030 [.682] 0.136 [.015] 0.035 [.517] 0.124 [.024] 0.047 [.391] -0.040 [.473]

**

**

3.6 %

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

(4) -0.016 [.829] 0.121 [.031] 0.034 [.521] 0.126 [.022] 0.040 [.464]

0.013 [.822] 3.4 %

**

**

42

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga Table 5. (Continued)

Adj.-R2

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

F-Ratio

(1) 2.3 % 3.027 [.018]

(2) 2.3 % **

3.003 [.019]

**

(3) 2.1 %

(4) 2.0 %

2.501 [.031]

2.417 [.036]

**

Secondly, as far as EXTP is concerned, no significant relationship between managers’ discretionarity and the proportion of outside directors has been found. It is also a result closely related to previous research since, despite the theoretical background supporting outside directorships, no conclusive empirical evidence has been found yet (Hermalin and Weisbach, 2000). Therefore, the benefits tied to outside directors’ independence and monitoring incentives seem to be outweighed by the expertise and better knowledge about the company and the industry held by inside directors. Although this result is not totally proved, we could make out, as previous research has pointed at, the pertinence of a balanced board makeup (Rosenstein and Wyatt, 1997). The third variable related to the board of directors is the number of meetings held per year. As reported in columns 3 and 4, it is not significant at all. It means that board activity -gathering of the board as a whole body- does not seem to be relevant in order to reduce earnings management. Once again, there are theoretical foundations for this result since Vafeas (1999) explains and tests how the limited time outside directors spend together could not be used for the meaningful exchange of ideas but for routine tasks. Along with board size, the other significant variable is firm performance (ROA). Nevertheless, and contrary to other authors’ previous evidence (Peasnell et al., 1999b), firm performance is positively related to discretionary accruals. No doubt it is an outcome that should be elucidated in future research and it is very likely that further refinements of our measures should be required. For instance, instead of using firm proficiency single-handedly, it might be more pertinent to introduce the performance of other firms whose executives the firm managers are going to be compared with. The last control variable is firm size and it has been measured in two different ways: sales and assets (columns 1 and 3, and 2 and 4 respectively). In both cases, the inclusion of this variable has scarcely affected the results but its switching sign (positive for TASS, negative for SALES). In any case, it is not a relevant fact given that this variable is not significant at all, although the switching sign might be partially explained by the opposite sign REV and PPE exhibit in the estimation of total accruals. Before presenting the reporting tables, we summarize the results achieved up to now by stressing the fact that the board size seems to be the only relevant characteristic of the boards affecting managerial discretionarity. In addition, this relationship comes out to be negative. Once identified the effect of the board of directors on abnormal accruals, we test the robustness of the estimations by introducing country and industry effects. This is why, as reported in table 6, the group of explanatory variables has been enlarged with a set of dummy variables reflecting the national identity of each company. Columns 1 and 2 report the coefficients for all the variables, whereas the results of the stepwise estimation are displayed in column 3 –only the last model variables are reported after having taken into account colinearity (tolerance) statistics. Estimations have been reproduced with several alternative

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

43

variables for board composition. Since the results remain basically unchanged, for the sake of brevity they have not been reported.18 Table 6. Discretionary accruals and corporate governance (Country effect) Ordinary least squares cross-section estimation. The dependent variable is discretionary accruals (AA), that is to say, Jones model residuals. Dummy variables stands for Switzerland (SW), United States (USA), France (FR), Spain (SP), Belgium (BE), the Netherlands (NETH), United Kingdom (UK) and Germany (GE)

BOASIZ(log) EXTP ROA SALES

(1) 0.169 [.003] 0.145 [.029] 0.111 [.036] 0.022 [.701]

***

**

**

TASS SW

R2

-0.116 [037] -0.169 [.011] -0.163 [.005] -0.235 [.000] -0.089 [.089] -0.034 [.522] .104 [.136] -.246 [.000] 16.7 %

R2-Aj.

13.7 %

F-Ratio

5.540 [.000]

USA FR

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

SP BE NETH UK GE

**

**

***

***

*

***

(2) 0.155 [.006] 0.147 [.026] 0.110 [.038]

0.071 [.218] -0.120 [.031] -0.189 [.004] -0.171 [.003] -0.233 [.000] -0.089 [.086] -0.036 [.502] .097 [.164] -.250 [.000] 16.9 %

***

5.660 [.000]

***

**

**

**

***

***

***

-0.166 [.001]

***

.151 [.004] -.213 [.000] 11.7 %

***

*

***

13.9 % ***

(3) 0.174 [.003]

***

10.7 % ***

11.254 [.000]

***

In spite of being slightly modified, results are totally consistent with previous ones. The most outstanding issue is board size keeping the positive and significant impact on managers’ 18

All this alternative estimation results are available from the authors upon request.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

44

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

discretionary behavior. The proportion of outside directors is more significant than previously and is positively related to abnormal accruals. Although this last result points at a negative effect of directors independence on firm performance, it is neither robust nor significant in the stepwise estimation, so its impact is not very reliable. The last comment on table 6 concerns R2 and adjusted–R2 coefficients, given that the explanatory power of the model is noticeably enhanced by the introduction of country variables compared to table 5. Table 7. Discretionary accruals and corporate governance (Industry effect) Ordinary least squares cross-section estimation. The dependent variable is discretionary accruals (AA), that is to say, Jones model residuals. Previous regressions are run again and industry effect introduced by a set of dummy variables. Companies have been sorted out based on the SIC code first digit (for instance, from 1000 to 1999: COD1)

Intercept BOASIZ(log) EXTP ROA SALES

(1) -0.022 [.780] 0.102 [.066] 0.019 [.728] 0.091 [.097] -0.032 [.564]

*

*

TASS

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

COD2

R2

0.193 [.101] 0.064 [.579] 0.302 [.004] 0.148 [.123] 0.004 [.945] 0.113 [.105] 0.066 [.277] 8.9 %

Adj.-R2

5.8 %

F-Ratio

2.942 [.001]

COD3 COD4 COD5 COD6 COD7 COD8

***

(2) -0.012 [.871] 0.093 [.095] 0.019 [.722] 0.092 [.094]

0.003 [.955] 0.184 [.119] 0.058 [.616] 0.298 [.004] 0.141 [.139] 0.003 [.955] 0.113 [.106] 0.066 [.278] 8.8 %

*

*

***

2.916 [.001]

**

0.113 [.027]

**

-0.088 [.088] 0.129 [.017]

*

**

6.8 %

5.8 % ***

(3) -0.0314 [.623] 0.114 [.024]

5.9 % ***

6.942 [.000]

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

***

Managers’ Discretionary Behavior, Earnings Management …

45

Closely related comments should be repeated in the face of table 7, where industry classification has been introduced. As stated above, column 3 exhibits stepwise estimation results and only the final model variables are reported. Once again, the board size comes out to be a significant and positive determinant of managers’ discretionarity, whereas board makeup has not any longer significant effect and confirms prior caveats about its role. The explanatory power of the model (R2 and adjusted–R2 coefficients) improves too compared to the non-industry effect model. The last group of results is due to the introduction of the auditor opinion by two dummy variables ADVERSE and ADV97 (table 8). These variables are aimed at controlling for circumstances when the auditor report warns about financial statement not reflecting the real situation of the firm or not meeting the standard accounting principles (columns 1 and 3). Table 8. Discretionary accruals and corporate governance (Auditor’s report) Ordinary least squares cross-section estimation. The dependent variable is discretionary accruals (AA), that is to say, Jones model residuals. ADVERSE is a dummy variable equaling 1 when the auditor’s report is negative (1)

CTE BOASIZ (log) EXTP

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ROA ADVERS E ADV97 SALES R2 Adj.-R2 F-Ratio

-0.015 [.835] 0.131 [.019] 0.042 [.437] 0.131 [.015] -0.054 [.313]

-0.025 [.649] 3.7 % 2.3 % 2.626 [.024]

**

**

**

(2) ADV= 1 -0.008 [.958] 0.086 [.593] 0.017 [.908] 0.429 [.007]

0.179 [.275] 26.8 % 18.2 % 3.116 [.027]

***

**

(3) ADV= 0 -0.013 [.868] 0.132 [.025] 0.038 [.502] 0.110 [.055]

-0.052 [.378] 2.9 % 1.6 % 2.217 [.067]

(4)

**

*

*

-0.015 [.837] 0.130 [.019] 0.040 [.456] 0.134 [.013]

-0.053 [.322] -0.026 [.645] 3.7 % 2.3 % 2.618 [.024]

**

**

**

(5) ADV97 =1 -0.201 [.274] 0.350 [.048] -0.176 [.299] 0.498 [.006]

0.151 [.373] 44.1 % 33.9 % 4.337 [.010]

**

***

***

(6) ADV97= 0 0.0046 [.952] 0.115 [.048] 0.043 [.445] 0.119 [.035]

-0.043 [.462] 2.7 % 1.4 % 2.130 [.077]

**

**

*

Results are quite similar to the previous ones, so no further comments are required and they corroborate the close relationship between board size and discretionary accruals. The ADVERSE variable was also used for separating the firms with a negative auditor report (columns 2 and 5) from the remainder companies in the sample (columns 3 and 6). Broadly speaking, the relevant influence of board size is confirmed once again, although the R2 and adjusted-R2 coefficients are quite different across the sub-samples. The six estimations have been additionally run including the country and industry effect. Since results have not

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

46

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

noticeably been changed –with the only exception of R2, we have just reported the basic estimation for the sake of brevity and clarity.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5. CONCLUSION The separation of company ownership and control grounds a number of conflicts of interest and has encouraged the research for mechanisms mitigating those conflicts. This is why both academia and practitioners have paid an increasing attention to corporate governance. Although both the external and the internal corporate governance mechanisms have been in the core of this attention, the most prominent interest has focused on the boards of directors. In turn, the boards of directors are considered as the main instrument at the shareholders disposal for monitoring and controlling managers’ behavior. It makes then much sense any intent to identify the features and actions of boards in order to know more in depth their ability for reducing the possible managers’ discretionarity. The recent literature suggests that board effectiveness relies on three main issues: board size, composition (in terms of outside and inside directorships) and the structure of internal functioning. Our empirical research, aimed to determine the impact of the board of directors on earnings management, is also founded on these issues. Earnings management is defined as the accounting processes aimed to modify financial reports for the sake of managers. The main contribution of our paper is the introduction of the discretionary component of earnings management as an indicator of managers’ discretionary behavior. The sample includes 450 non-financial companies from ten OECD countries. Three out of them could be classified as Anglo-Saxon or market-oriented corporate systems (Canada, USA and UK), whereas the seven remaining countries belong to the so-called continental or bank-oriented corporate system (Germany, Belgium, Spain, France, Holland, Italy and Switzerland). The availability of data for a number of countries allows us to broaden our analysis to a wider basis than the Anglo-Saxon area to which most of the previous research is linked. As far as the results are concerned, we have found evidence supporting the relationship between some boards features and the managers’ ability to discretionary altering the information stemming from the financial statements. More specifically and according to most of the previous literature, our main result shows a positive and significant impact of board size on earnings management. This result is quite consistent with alternative measures of firm size, or country and industry effect. Consequently, the coordination and communication problems arising in large boards seem to outweigh the possible benefits from a more detailed control by a higher number of directors. Regarding other characteristics of the boards, our research does not support any conclusive evidence either about board composition or about meeting frequency. Although both characteristics are theoretically related to the ability of board monitoring, our results do not lend any support and claims for further research. This is a quite promising direction for future research, so that new ways of identifying directors’ independence are proposed. There are also other directions for future research as the analysis of the boards internal functioning, stressing the committees structure as a meaningful issue that should be taken into account. It would provide a better knowledge of inside vs. outside directorships.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

47

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

REFERENCES Bange, M.M and DeBondt (1998) ‘RandD budgets and corporate earnings targets’, Journal of Corporate Finance, 4: 153-184. Barnhart, S. W.; Marr, M. W. and Rosenstein, S. (1994) ‘Firm performance and board composition some new evidence’, Managerial and Decision Economics, 15: 329-340. Bartov, E. (1993) ‘The timing of asset sales and earnings manipulation’, The Accounting Review, 68: 840-855. Baysinger, R. D. and Butler, H. N. (1985) ‘Corporate governance and the board of directors Performance effects of changes in board composition’, Journal of Law, Economics and Organization, 1: 101-124. Beasley, M.S. (1996) ‘An empirical analysis of the relation between the board of director composition and financial statement fraud’, The Accounting Review, 71(4): 443-465. Berglöf, E. (1997) ‘Reforming corporate governance Redirecting the European agenda’, Economic Policy, 24: 93-123. Bhagat, S. and Black, B., (1998) ‘Board independence and long-term performance’, Working paper no. 143, Columbia Law School, Center for Law and Economics Studies. Black, E.L.; Sellers, K.F. and Sheehy, T. (1998) ‘Earnings manipulation using asset sales An international study of countries allowing noncurrent asset revaluation’, Journal of Business, Finance and Accounting, 25: 1089-1118. Bowen, R.M.; Burghstahler, D. and Daley, L.A. (1987) ‘The incremental information content of accrual versus cash flows’, The Accounting Review, 62: 723-747. Bradshaw, M.T.; Richardson, S.A. and Sloan, R.G. (1999) ‘Earnings quality and financial reporting credibility an empirical investigation’, Mimeograph. Burgstahler, D. and Dichev, I. (1997) ‘Earnings management to avoid earnings decreases and losses’, Journal of Accounting and Economics, 24(1): 99-126. Bushee, B.J. (1998) ‘The influence of institutional investors on myopic RandD investment behaviour’, The Accounting Review, 73(3): 305-333. Carslaw, C.A. (1988) ‘Anomalies in income numbers Evidence of goal oriented behavior’, The Accounting Review, 63: 321.337. DeAngelo, H.; DeAngelo, L. and Skinner, D.J. (1994) ‘Accounting choice in troubled companies’, Journal of Accounting and Economics, 17: 113-143. DeAngelo, L. (1986) ‘Accounting numbers as market valuation substitutes A study of management of public stockholders’, The Accounting Review, 61: 400-420. Dechow, P.M. (1994) ‘Accounting earnings and cash flows as measures of firm performance. The role of accounting accruals’, Journal of Accounting and Economics, 18: 3-42. Dechow, P.M. and Sloan, R.G. (1991) ‘Executive incentives and the horizon problem An empirical investigation’, Journal of Accounting and Economics, 14: 51-89. Dechow, P.M.; Sloan, R.G. and Sweeney, A.P. (1995) ‘Detecting earnings management’, The Accounting Review, 70(2): 193-225. DeFond, M.L. and Jiambalvo, J. (1994) ‘Debt covenant violation and manipulation of accruals’, Journal of Accounting and Economics, 17: 145-176. Denis, D. J. and Sarin, A. (1998)’ Ownership and board structures in publicly traded corporations’, Working paper, Krannert Graduate School of Management, Purdue University.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

48

Pablo de Andrés Alonso, Valentín Azofra Palenzuela and Félix J. López Iturriaga

Eisenberg, T.; Sundgren, S. and Wells, M.T. (1998) ‘Larger board size and decreasing firm value in small firms’, Journal of Financial Economics, 48: 35-54. Fernández Álvarez, A.I.; Gómez Ansón, S. and Fernández Méndez, C. (1998) ‘El papel supervisor del consejo de administración sobre la actuación gerencial. Evidencia para el caso español’, Investigaciones Económicas, 22(3): 501-516. Hayn, C. (1995) ‘The information content of losses’, Journal of Accounting and Economics, 20: 125-153. Healy, P.M. (1985) ‘The effect of bo-nus schemes on accounting decisions’, Journal of Accounting and Economics, 7: 85-107. Hermalin, B.E. and Weisbach, M. S. (1988) ‘The determinants of board composition’, RAND Journal of Economics, 19(4): 589-606. Hermalin, B.E. and Weisbach, M. S. (1991) ‘The effects of board composition and direct incentives on firm performance’, Financial Management, Winter: 101-112. Hermalin, B.E. and Weisbach, M.S. (2000) ‘Boards of directors as an endogeneously determined institution A survey of the economic literature’, Mimeograph Huther, J. (1997) ‘An empirical test of the effect of board size on firm efficiency’, Economics Letters, 54: 259-264. Jensen, M. C. (1993) ‘The modern industrial revolution, exit, and the failure of internal control systems’, Journal of Finance, 483: 831-880. Jones, J.J. (1991) ‘Earnings management during import relief investigations’, Journal of Accounting Research, 29(2): 193-228. Klein, A. (1998) ‘Firm performance and board committee structure’, Journal of Law and Economics, 41: 275-303. Lipton, M. and Lorsch, J. W. (1992) ‘A modest proposal for improved corporate governance’, Business Lawyer, 481: 59-77. Macey, J. R. (1998) ‘Measuring the effectiveness of different corporate governance systems Toward a more scientific approach’, Journal of Applied Corporate Finance, 10(4): 16-25. McNichols. M. and Wilson, G.P. (1988) ‘Earnings management from the provision of bad debts’, Journal of Accounting Research, 26: 1-31. Peasnell, K.; Pope, P.F. and Young, S. (1999a) ‘Outside directors, board effectiveness, and abnormal accruals?’, Working paper 99/006. Lancaster University. Peasnell, K.; Pope, P.F. and Young, S. (1999b) ‘Managerial ownership and the demand for outside directors’, Working paper 99/007. Lancaster University. Peasnell, K.; Pope, P.F. and Young, S. (1999d) ‘Board composition and earnings management Do outside directors constrain abnormal accruals?’, Mimeograph. Peasnell, K.; Pope, P.F. and Young, S. (2000a) ‘Detecting earnings management using crosssectional abnormal accruals models’, Accounting and Business Research, 30(4): 313-326. Peasnell, K.; Pope, P.F. and Young, S. (2000b) ‘Accruals management to meet earnings targets: UK evidence pre- and post-Cadbury’, British Accounting Review, 32(4): 415-445. Porter, M.E. (1992) ‘Capital disadvantage: America’s failing capital investment system’, Harvard Business Review 70(5): 65-82. Rajgopal, S. and Venkatachalam, M. (1998) ‘The role of institutional investors in corporate governance an empirical investigation’, Mimeograph. Rajgopal, S.; Venkatachalam, M. and Jiambalvo, J. (1999) ‘Is institutional ownership associated with earnings management and the extent to which stock prices reflect future earnings?’, Mimeograph.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Managers’ Discretionary Behavior, Earnings Management …

49

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Rosenstein, S. and Wyatt, J.G. (1990) ‘Outside directors, board independence, and shareholder wealth’, Journal of Financial Economics, 26: 175-191. Rosenstein, S. and Wyatt, J.G. (1997) ‘Inside directors, board effectiveness, and shareholder wealth’, Journal of Financial Economics, 44: 229-250 Shleifer, A. and Visnhy, R.W. (1986) ‘Large shareholders and corporate control’, Journal of Political Economy, 94, 3: 461-458. Sloan, R.G. (1996) ‘Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review, 71, 3: 289-315. Subramanyam, K.R. (1996) ‘The pricing of discretionary accruals’, Journal of Accounting and Economics, 22: 249-281 Thomas, J. (1989) ‘Unusual patterns in reported earnings’, The Accounting Review. October: 773-787. Vafeas, N. (1999) ‘Board meeting frequency and firm performance’, Journal of Financial Economics, 53: 113-142. Warfield, T.D.; Wild, J.J. and Wild, K.L. (1995) ‘Managerial ownership, accounting choices, and informativeness of earnings’, Journal of Accounting and Economics, 20: 61-91. Watts, R. and Zimmerman, J.L. (1990) ‘Positive accounting theory a ten year perspective’, The Accounting Review, 65: 131-156. Watts, R.L. and Zimmerman, J.L. (1986) Positive accounting theory. Prentice Hall. Englewood Cliffs. Weisbach, M.S. (1988) ‘Outside directors and CEO turnover’, Journal of Financial Economics, 20: 431-460. Yermack, D. (1996) ‘Higher market valuation of companies with a small board of directors’, Journal of Financial Economics, 40: 185-211.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 3

DOES ONE MONETARY POLICY FIT ALL? THE DETERMINANTS OF INFLATION IN EMU COUNTRIES *

Melisso Boschi University of Essex, UK and Ministry of Economy and Finance, Italy

Alessandro Girardi University of Rome “Tor Vergata” and ISAE, Italy

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT This chapter aims at assessing the long-run determinants and the short-run dynamics of inflation in each country belonging to the European Monetary Union (EMU). Our work complements the recent literature on this topic for the Euro Area as a whole. Detecting such determinants can be crucial in designing structural reforms acting as aside instruments of monetary policy in maintaining price stability. The empirical methodology consists of a re-interpretation of the structural cointegrating VAR approach, which allows for a structural long-run analysis of inflation determinants along with an accurate assessment of its short-run dynamics. The main conclusion emerging from the estimates is that not only the determinants of inflation differ in the countries belonging to the Euro Area, but also that cost-push factors have a considerable role in explaining inflation in most of the countries examined. As a policy implication, a tight monetary policy pursued in those countries whose inflation is mainly driven by costs would result in a contraction of economic activity without exerting relevant effects on price dynamics.

JEL no. C32, E00, E31, E37. Keywords: Inflation, markup, EMU countries, long-run structural VARs, subset VEC models.

*

The views expressed do not necessarily reflect those of the Ministry of Economy and Finance and of the Institute for Economic Studies and Analyses (ISAE) of Italy.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

52

Melisso Boschi and Alessandro Girardi

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION On 1 January 1999 eleven European countries entered the third stage of the European Monetary Union. A new currency – the euro – replaced the national currencies, and a new institution – the Eurosystem, consisting of the European Central Bank (ECB) and the national central banks of those countries that adopted the euro – took on the responsibility for the monetary policy within the Euro Area. The single countries’ governments remained in charge of the fiscal policy under the binding constraints of the Stability and Growth Pact. In the new European institutional architecture designed by the Treaty of Maastricht, the Eurosystem has the main objective of maintaining the price stability in the Euro Area. More precisely, the single monetary policy of the ECB is carried out targeting the price stability of the Euro Area as a whole and, as such, it is common to all countries belonging to the EMU. The price stability is quantitatively defined as an annual inflation rate, referred to the Harmonized Index of Consumer prices (HICP), close to 2% in the medium-run. The ECB’s monetary policy strategy is based on two pillars related to the temporal perspectives relevant for assessing the risks to price stability: the economic analysis aims to determine the short to medium-term determinants of price dynamics, while the monetary analysis focuses on longerterm horizons. The economic analysis focuses on the shocks hitting the European economy and produces projections of the main macroeconomic variables; the monetary analysis, exploiting the long-run link between money and prices, monitors the development of several monetary indicators, including the aggregate M3, its components, and counterparts, in particular the domestic credit and the different measures of excess liquidity (ECB, 2004). By this strategy, the ECB fits one instrument to all countries, disregarding the countryspecific determinants of prices long- and short-run dynamics. Nevertheless, the assessment of such determinants can be crucial in designing structural reforms acting as aside instruments of monetary policy in maintaining price stability. In this respect, a disaggregated analysis, conducted by estimating national models, can shed light on country-specific determinants of inflation and possibly identify those countries where cost-push factors have the most relevant role. This chapter aims at assessing the long-run determinants and the short-run dynamics of inflation in each country belonging to the European Monetary Union (EMU). Our work is a complement of the recent literature on this topic for the Euro Area as a whole (Banerjee and Russel, 2002b; Bowdler and Jansen, 2004; Boschi and Girardi, 2005). With reference to the monetary policy strategy pursued by the ECB, the present analysis may be considered as contributing to the first “pillar,” aiming to uncover the economic determinants of inflation. The empirical methodology consists of a re-interpretation of the structural cointegrating VAR approach recently proposed by Garratt et al. (2003). This allows for a structural longrun analysis of inflation determinants along with an accurate assessment of its short-run dynamics. To this end, the Vector Error Correction (VEC) methodology (Johansen, 1995) is applied to a four-dimensional system including the labor productivity, the real exchange rate, the domestic inflation rate and the ratio between unit labor cost and price level. We use data ranging from the first quarter of 1984 to the last quarter of 1998 for each EMU member country, namely Austria, Belgium, Finland, France, Germany, Ireland, Italy, Netherlands, Portugal, and Spain. Luxembourg and Greece are not included, the first due to its negligible economic dimension, the second because it entered EMU later than the other countries.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

53

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The investigation consists of two stages. First, we disentangle the long-run determinants of inflation in each of the member countries of the EMU within the theoretical framework developed by Boschi and Girardi (2005), which is based on two long-run dynamic relationships (Juselius, 2002) relating inflation to the markup and the output gap respectively. This extends the model introduced by Banerjee et al. (2001). These long-run structural relationships between the variables of the model are embedded in an otherwise unrestricted VAR model and finally tested formally. Next, the estimated models are used to analyze the short-run behavior of inflation, highlighting the factors that drive its dynamics. The main conclusions emerging from the estimates are that: i) the determinants of inflation differ in the countries belonging to the Euro Area; ii) cost-push factors have a considerable role in explaining inflation in most of the countries examined. A tight monetary policy pursued in those countries whose inflation is mainly driven by costs would result in a contraction of economic activity without exerting relevant effects on prices dynamics. This would be the consequence of higher financing costs for firms and a lower aggregate demand determined by an increasing interest rate. These are precisely the effects observed in the most recent years, when most of the European economies have been characterized by close to zero output rates of growth. The paper is articulated as follows: Section 2 discusses some stylized facts on the convergence process of EMU member countries and their structural characteristics. Section 3 introduces the long-run theoretical framework. Section 4 describes the econometric methodology used, illustrating the long-run structural modeling approach and the parsimonious (subset) VEC procedure to analyze short-run dynamics. Section 5 illustrates the tests performed in order to check the statistical validity of the theoretical constraints on the long-run structure and reports both long- and short-run parameters estimates. Conclusion and References follow.

2. STYLIZED FACTS During the second stage of the EMU a progressive homogenization of national economic policies and structural features of several European countries took place, even though the Maastricht criteria were met only partially. Despite very dissimilar conditions in terms of deficit/GDP and debt/GDP ratios across member countries at the time of their entrance in the third stage of the EMU, an almost complete convergence in terms of interest and inflation rates occurred. Figure 1 (dashed line) shows the reduction of the standard deviation of yield differentials between ten-year government bonds issued in the EMU member countries (with the exception of Greece) and the corresponding ten-year government bond issued in Germany, regarded to a risk-free asset (Favero et al., 1997), over the EMS years. During the eighties and the nineties, indeed, Germany was a meta-economic reference point for the other European countries. Although its central role may be less evident in recent years, Germany still weights for roughly one third of the euro area GDP. Moreover, the German monetary and fiscal policy strategies have inspired, to some extent, the EMU’s institutional architecture.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Melisso Boschi and Alessandro Girardi

54

8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1984

1986

1988

1990

1992

Inflation

1994

1996

1998

Yield

Figure 1 – Standard deviation of inflation and long-term interest rate differentials vis-à-vis Germany for EMU countries: 1984-1998. Percentage values.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Analogously, the dispersion of inflation differentials with respect to Germany decreased over time, mainly starting from the first years of the nineties (Figure 1, continuous line), suggesting that to some extent a convergence of price levels has occurred in Europe. This is also shown in Figure 2, illustrating the dispersion of the relative price level at the beginning of the period of analysis (horizontal axis) and of the average inflation differential over the period 1984-1998 (vertical axis) for the EMU member countries with respect to Germany. It should be noted that countries exhibiting higher (lower) price levels as compared to the base country are characterized by lower (higher) inflation rates. 7.0 Portugal

6.0 5.0 4.0

Spain

3.0

Italy

2.0 Finland

1.0

France Austria Belgium Luxembourg Netherlands

Ireland

0.0 -1.0 -2.0 40

50

60

70

80

90

100

110

120

Figure 2 – Relative prices (horizontal axis) in 1984 and 1984-1998 average inflation rate differentials vis-à-vis Germany for EMU countries (vertical axis, percentage values).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

55

The observed nominal convergence occurred in Europe during the mid-eighties and nineties was coupled by real convergence in several economies. Spain, Portugal and Ireland successfully restructured their economic systems through projects mainly financed by Cohesion Funds (see Basile et al., 2001). Interestingly, the huge difference in GDP growth rates across EMU member countries was not accompanied by a relevant differentiation in terms of sector composition of their added value. Table 1 reports the percentage share of each sector’s added value for the ten countries analyzed over the sample period. Each country’s average rate of growth is reported in the fourth column. Table 1 – Distribution and average rate of growth of output (1984q1–1998q4).

Aut

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Bel Fin Fra Ger Ire Ita net Por Spa

PRIMARY SECONDARY TERTIARY 3.26 33.66 63.08 66.73 2.22 31.05

5.80 3.64 1.54 8.57 3.78 4.03 7.06 6.44

33.26 29.02 35.67 35.94 32.76 30.45 31.53 32.44

AV. RATE OF GROWTH 2.40

60.93 67.34 62.79 55.49 63.46 65.52 61.41 61.12

2.33 2.50 2.13 2.11 5.72 2.06 2.91 3.25 2.95

Therefore, it is important to consider the structural features of these economies beyond such stylized facts, in order to uncover the main determinants driving inflation in EMU member countries.

3. THEORETICAL FRAMEWORK Inflation can be defined as the loss of purchasing power undergone by money over time: an increase in the general level of prices increases the number of monetary units required to buy a given good. The rate of growth of the prices level measures the rate of inflation. Inflation can be classified according to its causes: excess demand inflation, cost-pushed inflation, and imported inflation. The first type of inflation occurs when an increase of aggregate demand pushes the price level up because of the presence of an inelastic supply of goods and services, or in the extreme case of full employment. In monetary terms, the supply is higher than the demand for money. The economic agents will try and spend in goods and services the excess money balances, but since all production inputs are full employed, the increased aggregate demand will not result in increased supply but rather in a higher equilibrium level price. This is called excess demand inflation, or, more simply, demand inflation. According to the original cause of the excess money supply, we distinguish between financial and credit inflation. In the first case the excess money supply is caused by seigniorage, i.e. by the government issuing money

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Melisso Boschi and Alessandro Girardi

56

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

in order to finance public expenditure; in the second case the excess money supply is determined by the excessive credit created by the financial system. The second kind of inflation is originated by costs. Economic theory states that under perfect market competition equilibrium prices equal marginal costs. However, in practice a number of markets operate under imperfect or even monopolistic competition. In such market structures prices may well differ from marginal costs. Specifically, it is usually assumed that the mechanism of price formation be based on the markup, i.e. on the application of a proportional marginal profit on production costs. This pricing strategy implies that inflation may be determined by costs. For example, a sudden increase of oil price may be transmitted on prices through the markup. Finally, in open economies, a further cause of inflation, referred to as “imported” inflation, is the nominal exchange rate devaluation. Inflation occurs because, in first instance, the exchange rate devaluation raises the local currency-denominated price of imported consumption goods and production inputs. Subsequently, the change in international relative prices will affect quantities, driving the economic system to a new equilibrium. The main consequences of inflation are related to the change of relative prices. Relative prices change as a consequence of inflation and this causes uncertainty, distortions and income redistribution. An increase of inflation may exert a number of effects on the economy. First, given the nominal interest rate, it will reduce the real interest rate thus stimulating consumption to the detriment of saving and reducing the real burden of debt. Second, a raising inflation may reduce the international competitiveness of exporting firms, thus inducing them to reduce the markup. In this Section the possible theoretical long-run path of the EMU countries’ inflation is presented.19 It consists of two dynamic steady-state relationships extending the scheme proposed by Banerjee et al. (2001) and Banerjee and Russell (2002 a). The starting point of the analysis is the following system:

pt − wt = −ω1 ⋅ OGt − ω2 ⋅ φt − ω3 ⋅ Δpt − ω4 ⋅ t (1) wt − pt = −γ1 ⋅ U t + γ 2 ⋅ φt + γ 3 ⋅ t Δpt = −δ ⋅ U t

(3)

Ut = −ψ ⋅ OGt

(4)

(2)

where pt indicates the logarithm of the price level, wt the logarithm of nominal wages,

OGt an output gap measure, φt the logarithm of productivity, Δ the difference operator, and U t the unemployment rate. The parameters are all positive. As in Banerjee et al. (2001), (1)

19

This Section and the subsequent draw extensively on Boschi and Girardi (2005).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

57

and (2) represent the formulas for markup20 and real wages, respectively, (3) identifies the Phillips curve, and (4) the Okun’s law. The linear trend in (1) captures the possible effects of taxation and other costs (especially raw materials and energy) on the formation mechanism of markup. Analogously, the trend in (2) represents the possible influence of factors such as unemployment benefits and tax rates on the demand for real wages.

3.1

Cost-push Inflation Substituting (4) in (2), OGt can be deleted in (1) and (2) obtaining the relationship

between markup and inflation:

( pt − wt ) =−

ω ⋅γ ⋅ψ (ω ⋅γ ⋅ψ−ω1 ⋅γ3) (ω2 ⋅γ1 ⋅ψ−ω1 ⋅γ2) ⋅φt − 3 1 ⋅ Δpt − 4 1 ⋅t (γ1 ⋅ψ−ω1) (γ1 ⋅ψ−ω1) (γ1 ⋅ψ−ω1)

(5)

In order to assure that labor and firms have stable income shares in the long-run, the coefficient of φt in (5) must be unitary. Assuming that firms maximize profits ( ω2 = 1 ), this condition holds for any values of γ 2 if firms fix prices independently of demand ( ω1 = 0 ) or if linear homogeneity is assumed ( γ 2 = 1 ). Therefore, equation (5) becomes:

pt − ulct = −μ1 ⋅ Δpt − μ 0 ⋅ t

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

where

ulct = wt − φt

(6)

indicates

the

unit

labor

cost

and

where

μ1 = ( ω3 ⋅ γ1 ⋅ ψ ) / ( γ1 ⋅ ψ − ω1 ) and μ0 = ( ω4 ⋅ γ1 ⋅ ψ − ω1 ⋅ γ 3 ) / ( γ1 ⋅ ψ − ω1 ) are nonnegative parameters. If μ1 = 0 , the model (1)-(4) becomes analogous to the standard one proposed, for example, by Layard et al. (1991) and Franz and Gordon (1993), where inflation does not represent a cost to firms. In an open economy framework, equation (6) is modified to take into account the possible relevance of the import price on markup, as in de Brouwer and Ericsson (1998) and Banerjee et al. (2001):

pt − δ ⋅ ulct − (1 − δ) ⋅ pmt = −μ1 ⋅ Δpt − μ 0 ⋅ t or

− st − β0 ⋅ pppt = −β1 ⋅ Δpt − b0 ⋅ t 20

The presence of Δpt in (1) implies that inflation may represent a cost to firms even in the long-run (e.g. because of the difficulties faced by price-setting firms in adjusting prices in an inflationary environment with incomplete information). Thus, an increase in costs may not be fully reflected in higher prices because the markup falls with higher inflation.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Melisso Boschi and Alessandro Girardi

58

where st = ulct − pt = ( wt − pt ) − φt indicates the logarithm of labor income share,

pppt = ( pt* + et − pt ) = ( pmt − pt ) is a competitiveness index, given by the logarithm of the real exchange rate, and β0 = (1 − δ) / δ , β1 = μ1 / δ , b0 = μ 0 / δ . If β0 > 0 , the external sector plays a role in the formation of domestic prices. Adding a stochastic residual, ε mu ,t , we obtain the first long-run condition to test:

− st − β0 ⋅ pppt + β1 ⋅ Δpt + b0 ⋅ t = ε mu ,t (7) where ε mu ,t is supposed to be stationary.

3.2

Demand Inflation

Generally, under this second approach inflation is studied through a relationship between price changes and a cyclical indicator (see, for example, Stock and Watson, 1999). From equations (4) and (3) this relationship can be represented as:

Δpt = β2 ⋅ OG t (8) where β2 = ψ ⋅ δ is a positive parameter. The potential output required to obtain OGt is

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

here estimated by means of a constant returns to scale production function21 of labor ( N t ) and capital stock ( K t ), Yt = F ( K t , At ⋅ N t ) (Binder and Pesaran, 1999) re-written as:

Yt = At ⋅ f ( κt ) Nt where

(9)

f ( κt ) = F ( K t ,1) is a function that satisfies the Inada conditions and

κt = K t / ( At ⋅ N t ) indicates the capital stock per effective labor unit. Assuming that the logarithm of the technological progress index At is given by ln( At ) = ϕ ⋅ t + ut where ut is a mean-zero I (1) process, equation (9) becomes (in logs):

φt = ϕ ⋅ t + ln [ f ( κt ) ] + ut

21

Alternatively, an algorithm for the extraction of trend from actual output or an explicit statistical model can be used (Clark et al., 1996; Harvey and Jaeger, 1993).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

59

Binder and Pesaran (1999) show that the long-run path of productivity is determined mainly by the technological progress, i.e. E [Δφt ] = ϕ . Therefore, the variable OGt is specified with a linear trend as a proxy of GDP and employment growth associated to the technological progress.22 The second long-run condition to test is obtained adding a stochastic residual, ε pc ,t , to (8)

Δpt − β2 ⋅ φt + b1 ⋅ t = ε pc

(10)

where the output gap measure is φt − ϕ ⋅ t , b1 = β2 ⋅ ϕ > 0 and ε pc is supposed to be stationary.

4. ECONOMETRIC METHODOLOGY

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The econometric methodology is based on the VEC methodology (Johansen, 1995). This modeling approach allows to describe in detail both long-run relationships and short-run dynamic interdependencies existing among (a small set of) variables. More specifically, the approach used in this study consists of two steps. In the first step, the empirical investigation is driven by the theoretical specification of the long-run equilibrium paths. This is consistent with the idea that economic theory is able to highlight the long-run equilibrium relationships among variables, but it is less informative about their short-run dynamics (Garratt et al., 2003). In the second step, the dynamic structure of the model is specified according to the statistical properties of the short-run parameters.

4.1

The Structural Cointegrating VAR Model

The long-run relationships presented in Section III are approximated by log-linear equations and embedded in a VEC model: m −1

Δy t = a + ∑ Γ j ⋅ Δy t − j + A ⋅ ε t −1 + Φ ⋅ d t + u t j =1

(11)

ut ~ N ( 0, Σu ) (12) This model allows to take jointly into account both the short-run dynamics among the variables collected in the vector y t =[ st , pppt , Δpt , φt ]′, and the long-run structure represented by the vector of residuals ε t of cointegration relations: 22

Under the assumption that the share of employed workers on population is stationary, as in Garratt et al. (2003), (labor) productivity may represent a measure of per-capita output.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Melisso Boschi and Alessandro Girardi

60

b ⋅ t + B′ ⋅ y t = ε t

(13)

In (11) Γ j ’s are matrices of autoregression coefficients, A is a matrix collecting the adjustment coefficients of short-run dynamics to long-run paths, a is a vector of intercepts,

d t is a vector of dummy variables whose parameters are in matrix Φ , and u t is a vector of residuals distributed according to (12). Equation (13) summarizes the r < k equilibrium relationships that are supposed to hold in the economy: matrix B collects the parameters defined in (7) and (10), vector b contains b0 and b1 (i.e. the slopes for linear deterministic trends – these are restricted to belong to the cointegration space in order to avoid quadratic trends in the level variables), and ε t contains the residuals ε mu ,t and ε pc ,t . All four variables in y t are considered endogenous a priori, while their possible exogeneity will be verified ex post. This strategy can be also justified from a statistical point of view since the presence in the model of variables erroneously treated as exogenous could produce non efficient estimates. In order to exactly-identify the cointegrating matrix B , r contemporaneous restrictions 2

on each cointegration relationship are imposed. Out of these r restrictions, r are normalizations necessary to rotate the cointegration space in the directions represented by the equilibrium conditions. The structural relationships (7) and (10) provide the remaining

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

r 2 − r constraints plus an additional one needed in order to obtain an over-identified model. Thus, the system (13), solved with respect to the parameters collected in b and B , becomes: ⎡b0 ⎤ ⎡ −1 −β0 ⎢ ⎥ ⋅t ⎢ 0 0 ⎣b1 ⎦ + ⎣

β1 1

⎡ ε mu ,t ⎤ 0 ⎤ ⋅ y ⎥ t ⎢ ε pc ,t ⎦⎥ −β2 ⎦⎥ ⎢ ⎣ =

(14)

The above theoretical framework can be verified through a LR test of the overall constraints imposed in (14). If r = 1, the above framework can also serve as a procedure to discriminate among competitive theories. If inflation is interpretable exclusively from a supply-side point of view, imposing an additional constraint to the r2 = 1 exactly-identifying ones (14) becomes

b0 ⋅ t + [ −1 −β0

β1 0] ⋅ y t ε mu ,t =

(15)

From a demand-side perspective, (14) becomes:

b1 ⋅ t + [ 0 0 1 −β2 ] ⋅ y t = ε pc ,t

(16)

with two additional constraints.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

4.2

61

The Subset VEC Model

The short-run dynamics is modeled using a parsimonious (subset) VEC model, obtained dropping those parameters of the matrices A , Γ j and Φ with p-values lower than a threshold, according to the Sequential Elimination of the Regressors Testing Procedure (SER/TP) proposed by Brüggemann and Lütkepohl (2001). Specifically, the statistically significant parameters of A give useful information about how the economy moves around the long-run equilibrium path. Moreover, the rows of A containing only zeroes allow to identify possible (weakly) exogenous variables. This model reduction process has two further implications. Firstly, the impulse response functions (and their confidence intervals) may differ, even markedly, from those derived from an unrestricted model (Brüggemann and Lütkepohl 2001). Secondly, dropping the statistically irrelevant variables can improve the quality of the forecasts generated by the model (Clements and Hendry, 2001, p. 119).

5. THE ESTIMATED STRUCTURAL VEC MODELS

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5.1

Preliminary Analysis

Prior to the estimation of the model (11)-(13), the unit root tests have been performed on the time series over the period 1984q1-1998q4. The sample span refers to a macroeconomic framework characterized by an acceleration of the harmonization process of domestic economic policies towards the EU commitments and a progressive liberalization of capital and trade movements. The first few years of the European Monetary System (EMS) are left out because of the turbulence caused by adjustment to the new monetary system. This choice also allows excluding the absorption process of the oil shocks occurred in the seventies, whose effects were particularly severe for small open economies, heavily dependent on foreign energy net suppliers. In order to avoid an arbitrary distinction between the variables of each country model, as suggested by Sims (1980), all of them are modeled as endogenous a priori, but their possible weak exogeneity is subsequently tested.23

5.1.1 Data Sources and Variables Construction The elaborations have been performed using the econometric packages J–Multi 3.30 for the unit root tests and the construction of the VEC models, and Pc–Fiml 10.3 for the preliminary analysis and the diagnostic tests. Quarterly non-seasonal adjusted data are from OECD (Statistical Compendium CD–Rom, 2004/2) and IMF (IFS CD–Rom, October 2004). The log level price, pt , is the consumer price index, the log real unit labor cost, st , is the ratio of real wages over productivity, the productivity, Φt , is given by the logarithm of the ratio of total output over the total number of employed workers.

23

This decision can also be motivated from a statistical point of view since the erroneous treatment of endogenous variables as weakly exogenous can produce inefficient estimates.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Melisso Boschi and Alessandro Girardi

62

The foreign variables in pppt = et + pt* − pt = (e0t − et ) + pt* − pt are given by: 16

pt* = ∑ wi ⋅ pit i =1

16

and

et* = ∑ wi ⋅ eit i =1

where i denotes the i-th country in the group including the Euro Area member countries, the remaining G7 countries not belonging to the Euro Area, the EMU member countries, and Switzerland. The star, *, denotes the variables of the rest of the world (RoW). The nominal effective exchange rate is given by difference between e0t = log(national currency/US dollar) and et* = log(RoW currency/US dollar). The weights wi , shown in Table 2, are given by the country i’s share of world trade, where the latter is defined as the sum of imports and exports.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5.1.2 Unit Root Tests The employed econometric methodology allows for series integrated at most of order 1. Testing for unit roots is a way of checking for the absence of I(2) variables in our sample, as this might produce poor results when coupled with standard VEC modelling (see Haldrup, 1998). Therefore, we have run ADF tests on each series, both in levels and differenced, with an optimal regression lag determined according to the BIC criterion with a maximum lag of four. The critical values are taken from Davidson and MacKinnon (1993). The results, reported in Table 3, suggest that all variables are indeed integrated of order one. The only exceptions are the levels of Δpt for Italy and the Netherlands, and of st for Spain, for which the null hypothesis of a unit root is rejected respectively at the 10% and the 5% levels of significance, and the first difference of Δφt for Ireland and Spain, for which the null is rejected at the 5% level of significance.

5.1.3 Model Specification and Dummy Variables Table 4 reports a brief description of each model’s main features. The sample period is 1984q1–1998q4 for all models. The number of lags has been selected according to the BIC criterion. In order to obtain a satisfactory fit of the model to the data, especially with regard to the residuals normality, dummy variables dXXY have been introduced. dXXY is a series 0,0,…,1,0,…, where XX indicates the year and Y the quarter. These dummy variables are grouped in three main categories according to the source of shock: idiosyncratic shocks (I), EMS shocks (EMS), and 1995 Mexican crisis shock (M). 5.1.4 Cointegration Rank The cointegration rank r has been determined using the trace test and the maximum eigenvalue test. Table 5 reports eigenvalues as well as the trace (upper part) and the maximum eigenvalue test (lower part) results. Both tests have been corrected for the number of degrees of freedom. Critical values are taken from Osterwald and Lenum (1992).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 2 – Matrix of bilateral flow trade weights. AUS BEL CAN DEN FIN AUS .0000 .0291 .0082 .0104 .0083 BEL .0095 .0000 .0060 .0094 .0072 FIN .0136 .0363 .0086 .0431 .0000 FRA .0131 .1161 .0090 .0118 .0075 GER .0697 .0905 .0091 .0263 .0136 IRE .0057 .0372 .0092 .0120 .0075 ITA .0339 .0544 .0125 .0129 .0073 NET .0146 .1509 .0060 .0172 .0109 POR .0100 .0409 .0048 .0159 .0081 SPA .0118 .0422 .0064 .0097 .0081 Note. Partner countries are reported in columns. The trade is set to zero. Source: OECD.

FRA .0600 .2042 .0589 .0000 .1574 .0814 .1920 .1113 .1520 .2432 weights

GER IRE ITA JAP NET POR SPA SWE .5237 .0050 .1121 .0297 .0394 .0066 .0225 .0198 .2388 .0095 .0580 .0244 .1879 .0078 .0290 .0238 .1852 .0086 .0443 .0528 .0542 .0085 .0258 .1456 .2337 .0122 .1300 .0356 .0658 .0171 .0933 .0180 .0000 .0113 .1117 .0539 .1136 .0139 .0462 .0318 .1318 .0000 .0367 .0468 .0600 .0046 .0223 .0199 .2701 .0094 .0000 .0317 .0614 .0134 .0648 .0164 .3135 .0111 .0566 .0296 .0000 .0084 .0293 .0279 .2029 .0064 .0760 .0202 .0559 .0000 .2215 .0188 .1926 .0093 .1185 .0305 .0506 .0689 .0000 .0148 are averages over the period 1994-1996. The weights sum to unit by

SWI UK USA .0347 .0408 .0496 .0159 .1050 .0635 .0903 .1270 .0973 .0250 .1169 .0952 .0450 .1035 .1025 .0171 .3613 .1466 .0377 .0919 .0900 .0212 .1153 .0762 .0211 .1000 .0456 .0151 .1053 .0730 row. Each country’s own

Table 3 – ADF unit root test. AUS

BEL

FIN

FRA

GER

IRE

ITA

NET

POR

SPA

st

-2.16 (4)

qt

-1.15 (1) *

-1.07 (1) *

-0.86 (1) *

-1.93 (1)**

-2.31 (1)**

-2.40 (1) **

-1.63 (1) *

-1.03 (1) *

-2.42 (1) *

-1.76 (1) *

Δpt

-1.58 (3) *

-2.47 (3) *

-1.72 (2) *

-2.39 (4) *

-2.28 (2) *

-2.78 (3) **

-3.29 (1) **

-2.70 (3) *

-2.34 (3)**

-2.54 (3)**

φt

-1.83 (4)**

-2.45 (4)**

-3.00 (4)**

-1.27 (0)**

-2.67 (2)**

-2.50 (2)**

-1.16 (0)**

-2.82 (3)**

-2.32 (1)**

-1.24 (1)**

Δst

-6.10 (2) *

-6.13 (0) *

-7.72 (0) *

-5.94 (0) *

-4.62 (0) *

-8.99 (0) *

-4.25 (0) *

-8.14 (0) *

-4.22 (1) *

-3.50 (0)**

Δqt

-5.49 (0) *

-5.15 (0) *

-5.06 (0) *

-5.16 (0) *

-5.54 (0) *

-4.99 (0) *

-5.17 (0) *

-6.08 (0) *

-5.18 (0) *

-5.01 (0) *

Δ 2 pt

-13.3 (2) *

-8.70 (0) *

-9.10 (1) *

-8.03 (1) *

-7.82 (2) *

-7.88 (2) *

-8.62 (0) *

-11.5 (2) *

-6.91 (2) *

-8.20 (2) *

Δφt

-4.35 (3) *

-7.99 (0) *

-5.11 (2)**

-7.73 (0) *

-6.25 (0) *

-3.13 (0) *

-3.75 (1) *

-7.25 (0) *

-11.3 (0) *

-2.98 (0) *

*

-1.86 (0)

*

-3.03 (4)

**

-2.40 (1)

**

-1.40 (0)

*

-2.24 (0)

**

-1.79 (2)

*

-1.79 (1)

*

-0.94 (0)

*

-3.62 (4)**

64

Melisso Boschi, Alessandro Girardi Table 4 – Models’ specification.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Country

Lags

Sample period

AUS

4

1985q1–1998q4

BEL

5

1985q2–1998q4

FIN

4

1985q1–1998q4

FRA

2

1984q3–1998q4

GER

2

1984q3–1998q4

IRE

4

1985q1–1998q4

ITA

2

1984q3–1998q4

NET

4

1985q1–1998q4

POR

3

1984q4–1998q4

SPA

2

1984q3–1998q4

Dummy variables d871 d952, d953 d924 d952, d953

shock to the labor market (I) shock to the real exchange rate (M) shock to the real exchange rate (EMS) shock to the real exchange rate (M)

d924

shock to the real exchange rate (EMS)

d924 d952, d953 d861, d983 d862 d871, d872 d911, d912 d923, d924 d952, d953, d961 d924 d972 d923, d924 d951, d952, d953 d972 d924 d951, d952 d973, d981 d844, d851, d911 d924 d953, d954 d924 d953

shock to the real exchange rate (EMS) shock to the real exchange rate (M) shock to the inflation (I) shock to the productivity (I) shock to the productivity (I) shock to the labor market (I) shock to the real exchange rate (EMS) shock to the real exchange rate (M) shock to the real exchange rate (EMS) shock to the labor market (I) shock to the real exchange rate (EMS) shock to the real exchange rate (M) shock to the labor market (I) shock to the real exchange rate (EMS) shock to the real exchange rate (M) shock to the labor market (I) shock to the labor market (I) shock to the real exchange rate (EMS) shock to the real exchange rate (M) shock to the real exchange rate (EMS) shock to the real exchange rate (M)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

AUT 0.67 0.48 0.15 0.07

BEL 0.52 0.30 0.16 0.02

FIN 0.47 0.30 0.24 0.11

FRA 0.54 0.21 0.11 0.02

H0 R=0 R≤1 R≤2 R≤3

H1 R≥1 R≥2 R≥3 R=4

5% 62.99 42.44 25.32 12.25

1% 70.05 48.45 30.45 16.26

AUT 114.8 50.93 13.14 4.04

H0 R=0 R≤1 R≤2 R≤3

H1 R=1 R=2 R=3 R=4

5% 31.46 25.54 18.96 12.25

1% 36.65 30.34 23.65 16.26

AUT 63.81 37.79 9.10 4.04

Table 5 – Cointegration rank. EIGENVALUES GER IRE ITA 0.45 0.44 0.50 0.34 0.35 0.25 0.18 0.19 0.13 0.00 0.08 0.08 TRACE TEST BEL FIN FRA GER 70.52 77.59 65.92 70.35 30.46 41.57 21.06 35.49 10.69 21.66 7.54 11.41 0.88 6.35 0.94 0.06 MAXIMUM EIGENVALUE TEST BEL FIN FRA GER 40.06 36.01 44.86 34.86 19.77 19.92 13.52 24.08 9.82 15.31 6.60 11.36 0.88 6.35 0.94 0.06

NET 0.41 0.36 0.26 0.04

POR 0.48 0.26 0.17 0.06

SPA 0.49 0.27 0.19 0.06

IRE 75.68 41.78 17.06 5.10

ITA 69.71 29.48 12.43 4.55

NET 74.46 44.48 19.38 2.47

POR 69.15 31.72 14.63 3.76

SPA 72.75 33.57 15.37 3.37

IRE 33.89 24.72 11.96 5.10

ITA 40.22 17.05 7.89 4.55

NET 29.98 25.10 16.91 2.47

POR 37.42 17.09 10.87 3.76

SPA 39.19 18.19 12.01 3.37

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 6 – The specification of the long-run structure

Bel Fin Fra Ger Irl Ita Por Spa

Specification

t

(16) (16) (16) (16) (16) (16) (15) (16)

* * * * * * * *

st

qt

Δpt

φt

χ2(gdl)

stat

[prob]

1 1 1 1 1 1

* * * * * *

1

*

(2) (2) (2) (2) (2) (2) (3) (2)

17.26 13.94 32.27 27.25 16.73 23.69 25.11 17.13

[0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00]

–1

Does One Monetary Policy Fit All?

67

The trace test results point to two long-run relationships for Austria and the Netherlands, while only one relationship is detected in all other models. These results are confirmed by the maximum eigenvalue test with the only exception of the Netherlands for which no long-run relationships exist according to the second test. We follow Johansen (1992) in accepting the trace test results in order to avoid inconsistency problems possibly arising with the maximum eigenvalue test. The rest of the analysis thus sets r = 2 in the Netherlands’s model.

5.1.5 The Specification of the Long-run Structure In the models of Austria and Netherlands the two cointegrating relationships seem to be identified by the long-run structural theoretical relationships (15) and (16). In all other countries the prices appear to be determined in the long-run by the equation (15), excepting for the model of Portugal whose inflation is determined exclusively by demand-side factors. For sake of completeness, the last three columns of Table 6 display the LR test results for the specification of the relationship (second column) alternative to the one discussed at length in the following Subsection regarding those countries’ models with rank r = 1 . The central columns of Table 6 report the estimated parameters (indicated by a star, *) of each model’s “best” specification.

5.2 THE ESTIMATED LONG-RUN STRUCTURE For each model, the chosen long-run structural specification is shown in Table 7, where the estimated coefficients and their corresponding standard errors are reported.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The statistics of the LR test for the over-identifying restrictions have a

χ 2 distribution

with a number of degrees of freedom depending on the number of restrictions. The probabilities associated to the statistics’ realizations are reported in square brackets. The signs of the estimated parameters are consistent with the economic theory and statistically significant. Moreover, LR test results do not reject the over-identifying restrictions. To summarize, all countries with high rates of growth are characterized by demand-side inflation (see Table 1). A significant exception is Ireland whose long-run structure is more similar to that of the countries belonging to the “core Europe”, i.e. Germany, Austria and the Netherlands, whose specification of the markup is almost identical. In order to better compare the estimated long-run structure, Table 8 reports each model’s net markup (first three columns) as long as the estimated coefficients μ 0 (fourth column) and

μ1 (fifth column) derived from the estimation of the structure reported in Table 7. Consistently with the economic theory, inflation is an extra cost to firms in all economies whose prices are determined according to the equation (15), excepting for Finland. Unlike other countries, in Austria, Germany, the Netherlands and Ireland, whose economic structure is rather similar, the import price level pmt does not appear to be a determinant of the net markup. Specifically, in Belgium, Finland, and Spain the markup is determined in the same proportion by pmt and ulc, while in Italy and France the unit labor cost has a bigger influence on the markup than the import prices.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 7 – Estimates of equations (15) and (16).

b ⋅ t + B′ ⋅ y t pppt

εt t

st

Δpt

φt

Austria

ε mu ,t ε pc ,t

0.0013

–1.0000

11.4410

(0.0003)

(1.8844)

0.0013

1.0000

(0.0002)

(0.0418) χ2(2) 1.81 [0.41]

ε mu ,t

0.0054

Belgium –0.5515

–1.0000

(0.0012)

(0.1291) 2

χ (1) 2.53 [0.11]

ε mu ,t

0.0012

–0.2507

Finland –1.0606

–1.0000

(0.0005)

(0.0763) 2

χ (2) 5.26 [0.07]

4.2497 (1.6379)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ε mu ,t

0.0014

–1.0000

(0.0003)

France –0.1407

1.9751

(0.0339)

(0.8569)

2

χ (1) 2.82 [0.09] Germany

ε mu ,t

0.0037

–1.0000

12.0290

(0.0010)

(4.3173)

χ2(2) 0.21 [0.90]

ε mu ,t

0

Ireland 0

–1.0000

29.3130

0

(6.6052) 2

χ (3) 7.51 [0.06]

ε mu ,t

Italy –0.6481

–1.0000

(0.0594) 2

χ (2) 1.15 [0.56]

4.2779 (0.6866)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 7 (Continued)

b ⋅ t + B′ ⋅ y t pppt

εt t

st

φt

Δpt

Netherlands

ε mu ,t ε pc ,t

0.0025

–1.0000

12.0390

(0.0006)

(2.2240)

0.0019

1.0000

–1.0000

1.0000

–0.3461

(0.0001) χ2(3)=1.94 [0.58] Portugal

ε pc ,t

0.0020 (0.0006)

(0.1136) 2

χ (2)=4.98[0.08]

ε mu ,t

0.0080 (0.0007)

–1.0000

Spain –1.0589 (0.1085) χ2(1)=0.84 [0.36]

7.8374 (2.1699)

Does One Monetary Policy Fit All?

71

The economic structure of France, in particular, looks very similar to that of the core Europe’s countries being characterized by a proportion of ulct over pmt of 8 as determinants of the markup. The fourth column of Table 8 shows how the exogenous component of inflation, captured by the slope of the linear time trend, is small and ranging between 0% for Ireland and Italy, and 1.5% for Germany and Spain. It is advisable to notice that the exogenous component of German inflation coincides with the minimum inflation rate compatible with the German economic structure benchmarked by the Bundesbank (see Sinn and Reutter, 2000). Finally, the fifth column of Table 8 shows the variation of markup induced by an annual increase of 1% (corresponding to a 0.25% increase on a quarterly basis) in inflation. The highly heterogeneous results can be summarized noticing that inflation represents an high cost to those countries whose prices are determined to a lesser extent by imports prices, i.e. Germany, Austria, Netherlands, and Ireland, where the reduction of markup ranges from 3% to 7%. In all other countries, inflation does not represent such a high cost, being the markup contraction comprised between 0% of Finland and 1% of Spain. The markup/inflation trade-off estimated for France, 0.4%, is quite similar to the 0.7% obtained by Banerjee and Russell (2002), while the values for Germany (1.2%) and Italy (2%) are quite different. This may be due to a number of reasons, including the different sample periods, the treatment of seasonality, the specification of the deterministic part of the cointegration space and the measurement of variables. Table 8 – Net markup, trade-off between markup and inflation, extra-costs.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

NET MARKUP AUT BEL FIN FRA GER IRE ITA NET SPA

pt

ulct

1 1 1 1 1 1 1 1 1

–1 –0.64 –0.48 –0.88 –1 –1 –0.61 –1 –0.49

EXTRA-COSTS

MARKUP/INFLATION

pmt

μ0 ⋅ 400

μ1

–0.36 –0.52 –0.12

0.52% 1.40% 0.24% 0.48% 1.48%

–11.44 –2.74

–0.39 –0.51

1.00% 1.56%

–1.73 –12.03 –29.31 –2.60 –12.04 –3.81

As for the four biggest European economies, namely Germany, France, Italy and Spain whose aggregate GDP is over 80% of the Euro Area’s GDP, and for Belgium and Finland, the absence of cost-push inflation points to an excess of production capacity. Regarding Ireland, the estimated long-run structure is at odds with the sustained economic growth characterizing the economy over the sample period. This may be due to the presence of a number of exportoriented multinational firms. As for the remaining countries, whose long-run structures include the Phillips curve, the results are heterogeneous. Table 9 reports the potential, i.e. not augmenting inflation, annual rate of growth of output and employment derived from equation (16).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

72

Melisso Boschi and Alessandro Girardi Table. 9 – Potential output growth. Austria 2.08 %

Netherlands 0.76 %

Portugal 2.32 %

The lowest value refers to the Netherlands, implying that its economy is characterized by full employment, as also indicated by the estimated value of the productivity coefficient in equation (16). Notice the high value of potential growth of Portugal, possibly due to the slow transition of the Portuguese economy from the prevalence of the primary sector to the secondary and tertiary sectors. This is witnessed by the large share of the agricultural sector on the Portuguese economy (over 7%) when compared to that of the other European countries. The long-run structure of Austria and Netherlands, whose models include two cointegration relationships, may be better described by embedding the second long-run restriction into the first one, thus obtaining a new stationary relationship (since a linear combination of two stationary relationships is stationary itself). This leads to a single longrun relationship between the output gap and the net markup. Table 10 reports the effect on the net markup of an annual one percent increase in the output gap. Table 10 – Markup and output gap

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Austria –0.72%

Netherlands –3.01%

Table 10 clearly shows that the markup is anti-cyclical, consistently with Gali (1994) according to which private firms do set prices depending on demand. This can be justified by the two following argumentations. The first line of reasoning relies on the increase of the marginal costs faced by firms due to the employment of production inputs in a period of economic expansion. The second one considers the pressure exerted on wages and other costs by an increase of aggregate demand.

5.3

THE ESTIMATED SHORT-RUN STRUCTURE

In this Section we discuss the dynamic properties of each country’s estimated VEC model. Table 11 reports the main diagnostic tests for the single equations.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 11 – Diagnostic tests of the dynamic models.

σu

A

AUT

BEL

FIN

FRA

GER

IRE

ITA

NET

POR

SPA

Δst

0.0082

0.0081

0.0172

0.0051

0.0083

0.0080

0.0065

0.0079

0.0120

0.0057

Δqt

0.0201

0.0176

0.0150

0.0155

0.0136

0.0271

0.0167

0.0188

0.0186

0.0166

Δ pt

0.0033

0.0038

0.0039

0.0023

0.0041

0.0039

0.0028

0.0036

0.0064

0.0044

Δφt

0.0081

0.0077

0.0166

0.0052

0.0072

0.0058

0.0061

0.0055

0.0104

0.0028

2

F(4,31)

F(4,27)

F(4,42)

F(4,39)

F(4,36)

F(4,34)

F(4,39)

F(4,29)

F(4,34)

F(4,43)

Δst

1.12

0.86

2.93

3.31

1.83

2.50

1.41

1.13

2.71

0.64

Δqt

1.95

0.32

1.49

1.29

1.37

1.24

1.96

0.68

1.21

0.57

Δ pt

1.37

0.30

2.59

2.58

0.80

0.82

0.41

2.59

0.22

2.66

Δφt

0.26

0.54

3.30

1.97

1.12

1.61

1.18

5.28

5.64

0.63

2

2

N

χ (2)

χ (2)

χ (2)

χ (2)

χ (2)

χ (2)

χ (2)

χ (2)

χ (2)

χ2(2)

Δst

0.17

5.17

1.14

1.43

3.88

7.15

1.93

0.10

4.23

21.09

Δqt

16.49

2.70

7.22

2.88

1.20

9.55

4.19

16.78

5.62

2.01

Δ pt

6.44

2.15

0.18

0.00

0.57

0.81

3.43

3.09

4.77

1.93

Δφt

0.15

1.53

0.42

0.88

5.29

9.48

2.74

1.35

5.95

4.29

2

AS

2

2

2

2

2

2

2

2

F(4,27)

F(4,23)

F(4,38)

F(4,35)

F(4,32)

F(4,30)

F(4,35)

F(4,25)

F(4,30)

F(4,39)

Δst

0.33

0.94

0.47

0.55

0.15

0.49

0.51

0.20

0.27

0.44

Δqt

0.24

0.33

2.73

0.39

0.96

0.19

0.74

0.23

0.25

0.60

Δ pt

0.69

0.17

0.59

0.09

0.98

0.48

2.40

0.78

0.23

1.83

Δφt

0.21

0.36

1.08

0.23

0.27

0.40

0.59

0.08

0.53

0.43

2

Note. Statistics in bold (italics) indicate statistical significance at the 5% (10%) level.

74

Melisso Boschi and Alessandro Girardi

The statistical fit of the inflation equation is satisfactory in almost all models. This supports the choice of the theoretical framework used for the analysis of the price dynamics in the EMU countries. The only exception is Spain whose equation of productivity growth shows the lowest standard error. The residuals do not appear to be serially correlated in 33 (38) equations at the 5% (1%) level of significance. The models of Netherlands and Portugal, whose single equations residuals are serially correlated, do not present autocorrelation at a system-wide level (not reported). This supports the choice of the number of lags for each model. The results of the autocorrelation tests on the squared residuals are even more satisfactory. Autocorrelation is rejected for all equations at the standard 5% level excepting the equation of Δqt in the model of Finland, where autocorrelation is rejected at the 4% level. Finally, the residuals of all equations of 5 out of 9 models appear to be normally distributed, while in the remaining 4 models non-normality emerges only for the equations of the competitiveness index and, in the model of Spain, for the Δst equation. At a system-wide level, only the models of Netherlands and Spain show non-normal residuals. The dynamic properties of the single models are analyzed conditioning on the elimination of the statistically insignificant short-run coefficients through the SER/TP approach developed by Brüggemann and Lütkepohl (2001). The models are estimated with a 3SLS procedure. The parsimonious models are obtained setting a threshold significance level of t = 1.60 for the short-run parameters and following the BIC criterion.1 Table 12 reports the restrictions determined for vector a (column two), matrix A (column three), matrices Γ j (column four), and matrix Φ (column five) for each model (11), as well as the LR test statistics (column seven) which are χ distributed, with the

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2

number of degrees of freedom given by the number of total short-run restrictions (column six). All LR statistics are well below the 5% and 10% level critical values, therefore showing how the data do not reject the imposed restrictions. Table 12 – Short-run restrictions. NUMBER OF RESTRICTIONS

AUT Bel Fin Fra Ger Ire Ita net Por Spa 1

a

A

Γj

Φ

1

3 1 1 2 1 1 1 3 2 0

22 41 18 10 5 21 9 34 13 4

11 11 12 19 30 10 18 21 12 9

2 0 1 2 2 2 1 1 1

χ2

STATISTICS

5% C.V.

10% C.V.

37 55 31 32 38 34 30 59 28 14

36.53 37.53 24.61 18.51 32.72 15.38 13.67 35.60 15.60 7.69

52.19 73.31 44.99 46.19 53.38 48.60 43.77 77.93 41.34 23.68

48.36 68.80 41.42 42.59 49.51 44.90 40.26 73.28 37.92 21.06

The choice is justified by the opinion that it is preferable to maintain the coefficients with uncertain significance rather than deleting them. Therefore, we adopt a “conservative” strategy, in the terminology of Krolzig and Hendry (2001).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

75

Table 13 shows the speed of adjustment coefficients along with the corresponding standard errors of each model. From the analysis of the coefficients of matrix A emerges that 33 out of 48 (lagged) cointegration residuals are statistically significant. This finding points to the existence of a strong adjustment effect running from the ε terms to the first differenced variables. Conversely, an adjustment mechanism for the real exchange rate cannot be detected. This implies that the real exchange rate can be considered as a candidate to be one of the common trends of the system. The weakly exogeneity of the real exchange rate characterizes all models, excepting those of Spain, Austria, and Finland. Specifically, the equations describing inflation dynamics are influenced by all cointegration errors determined in the previous Subsection. Table 13 – Speed of adjustment coefficients and their associated standard errors.

ε t −1 ε mu ,t −1 AUT

Δst

Δpppt

Δ 2 pt

Δφt

0.205

0.146

–0.072

–0.119

(0.046)

(0.045)

(0.011)

(0.048)

–0.529

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ε pc ,t −1 BEL

ε mu ,t −1

FIN

ε mu ,t −1

FRA

ε mu ,t −1

GER

ε mu ,t −1

IRE

ε mu ,t −1

ITA

ε mu ,t −1 ε mu ,t −1

NET

ε pc ,t −1 POR

ε pc ,t −1

SPA

ε mu ,t −1

(0.095) 0.092

–0.055

0.028

(0.019)

(0.012)

(0.005)

0.217

0.327

–0.216

(0.068)

(0.072)

(0.073)

0.166

–0.094

(0.029)

(0.024)

0.021

–0.011

0.015

(0.006)

(0.002)

(0.003)

0.088

–0.014

–0.077

(0.019)

(0.004)

(0.017)

0.172

–0.052

–0.059

(0.028)

(0.009)

(0.024)

0.057

–0.083

(0.017)

(0.007)

–0.548

–0.142

0.382

(0.112)

(0.044)

(0.076)

–0.558

0.408

(0.066)

(0.111)

0.062

0.100

–0.086

0.004

(0.014)

(0.043)

(0.011)

(0.001)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

76

Melisso Boschi and Alessandro Girardi Focusing on supply-side error corrections, which are included in nine out of ten cases, the

absolute value of the speed of adjustment coefficient of the equation of Δ pt is 2

systematically lower than the corresponding coefficient of Δst in all models, excepting those of the Netherlands and Spain. This suggests that if inflation is cost-pushed, the supply-side disequilibrium is corrected mainly through adjustments occurring in the labor market rather than as a consequence of monetary policy decisions.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

6. CONCLUSION By adopting a single currency, the EMU countries waived their monetary policy, which has since been taken on by the European Central Bank. Maintaining price stability within the Euro Area is the main task of the ECB, which has quantified it as an average inflation rate ranging from 0% to 2% in order to minimize the inflation costs related to redistribution effects, uncertainty and market distortions. The main goal of the present study is to assess the validity of the choice of monetary policy as the right instrument to maintain price stability. We consider all main countries belonging to the Euro Area, namely Austria, Belgium, Finland, France, Germany, Ireland, Italy, Netherlands, Portugal, and Spain. The sample period goes form the first quarter of 1984 to the fourth quarter of 1998, including a time horizon characterized by a relatively stable macroeconomic framework where oil shocks are absorbed by the system, the constraints implied by the exchange rate arrangements are binding and the financial system is progressively being liberalized. The econometric strategy, based on the estimation of a VEC model for each country, develops in two successive phases: in the first phase the long-run paths of the EMU countries’ economies are specified; in the second phase the dynamic properties of the single models are analyzed. The long-run structure includes two economic relationships linking, through a dynamic equilibrium, inflation to the markup and the output gap respectively, thus allowing for a distinction between cost-pushed and demand-pushed inflation. The coefficient estimation is conditioned on the execution of unit root tests on all variables. A priori all variables are treated as endogenous in each VAR model, while their weak exogeneity is tested ex post in order to avoid imposing an arbitrary distinction upon them. The cointegration rank is determined according to the trace test and the maximum eigenvalue test. The specification of cost-pushed inflation expresses the markup as a function of the real unit labor cost, the import prices, and the linear trend as a way to capture the influence of national structural factors. The specification of demand-pushed inflation relies on a version of the Phillips curve featuring the inflation rate as a function of the unemployment rate, and the Okun’s law, with the unemployment depending on the output gap. Moreover, the productivity is included as a further explanatory variable, while a linear trend is intended to proxy the growth of output and employment due to technological progress.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Does One Monetary Policy Fit All?

77

Both the cointegration relationships are supported by the data in the models of Austria and the Netherlands, in the model of Portugal only the demand-side long-run relationship holds, while in the rest of the countries the supply-side relationship holds. For those economies characterized by cost-pushed inflation, the net markup is calculated showing that inflation represents a cost to firms due to the corresponding loss of competitiveness. The above loss is higher in those economies where inflation is not affected by import prices. The first important result is that demand-pushed inflation is detected in those countries characterized by a sustained output growth, the only exception being Ireland, whose long-run structure includes only a cost-push determination of inflation. An excess of production capacity, i.e. an insufficient level of aggregate demand, therefore characterizes all other countries, whose long-run structure does not include a demand-pushed inflation equation. The second part of the paper presents a short-run dynamic analysis of the parsimonious VEC models. The statistically insignificant coefficients, i.e. those to which corresponds a value of the t statistics lower than the threshold value of 1.60, are deleted. The LR test results suggest that the data do not reject the parsimonious specification of all models. Almost the 70% of the speed of adjustment coefficients related to the error correction terms are statistically significant, thus confirming the presence of strong feedback mechanisms running from the error correction terms to the first differenced variables. The above results suggest that the ECB should take into account that the actual determinants of inflation differs in the single countries belonging to the Euro Area, and that the best objective of the monetary policy is the only demand-pushed inflation. A tight monetary policy pursued in those countries whose inflation is mainly driven by costs would result in a contraction of economic activity without exerting relevant effects on price dynamics. This would be the consequence of higher financing costs for firms and a lower aggregate demand determined by an increasing interest rate. These are precisely the effects observed in the most recent years, when most of the European economies have been characterized by close to zero output rates of growth. In 2003 the ECB has redesigned its strategy setting a rate of interest of 2% as its main objective, but there is still room for a modification of the monetary policy strategy capable of considering the different factors driving inflation in EMU countries.

REFERENCES Banerjee, A., Cockerell, L., & Russell, B. (2001). An I(2) analysis of inflation and the markup. Journal of Applied Econometrics, 16, 221–240. Banerjee, A., & Russell, B. (2002a). The relationship between the markup and inflation in the G7 economies and Australia. Review of Economics and Statistics, 83, 377-384. Banerjee, A., & Russell, B. (2002b). A markup model for forecasting inflation in the Euro Area. European University Institute Working Paper, 2002/16. Basile, R., de Nardis, S., & Girardi A. (2001). Regional inequalities and cohesion policies in the european union. Documento di Lavoro ISAE, 23. Binder, M., & Pesaran, H. M. (1999). Stochastic growth models and their econometric implications. Journal of Economic Growth, 4, 139-183.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

78

Melisso Boschi and Alessandro Girardi

Boschi, M., & Girardi, A. (2005). Euro Area inflation: Long-run determinants and short-run dynamics, mimeo, ISAE. Bowdler, C., & Jansen, E. S. (2004). A markup model of inflation for the Euro Area. ECB Working Paper, 306. Brüggemann, R., & Lütkepohl, H. (2001). Lag selection in subset VAR models with an application to a U.S. monetary system. In R. Friedmann, L. Knüppel, & L. Lütkepohl (Eds.), Econometric Studies: A Festschrift in Honour of Joachim Frohn. Münster: LIT Verlag. Clark, P., Laxton, D., & Rose, D. (1996). Asymmetry in the U.S. output-inflation nexus. IMF Staff Papers, 43, 216-251. Clements, M. C., & Hendry, D. F. (2001). Forecasting non–stationary economic time series. London: MIT Press. Davidson, R., & Mackinnon, J. (1993). Estimation and inference in econometrics. Oxford: Oxford University. de Brouwer, G., & Ericsson, N. R. (1998). Modeling inflation in Australia. Journal of Business and Economic Statistics, 16, 4, 433-449. European Central Bank (2004). The monetary policy of the ECB. Frankfurt: European Central Bank. Favero, C. A., Giavazzi, F., & Spaventa, L. (1997). High Yields: The spread on German interest rates. Economic Journal, 107, 956-985. Franz, W., & Gordon, R. J. (1993). German and American wage and price dynamics. European Economic Review, 37, 719-762. Gali, J. (1994). Monopolistic competition, business cycles, and the composition of aggregate demand. Journal of Economic Theory, 63, 73–96. Garratt, A., Lee, K., Pesaran, H. M., & Shin, Y. (2003). A long–run structural macroeconometric model of the UK. Economic Journal, 113, 412–455. Haldrup, N. (1998). An econometric analysis of I(2) variables. Journal of Economic Surveys, 12, 595–650. Harvey, A. C. & Jaeger, A. (1993). Detrending, stylized facts and the business cycle. Journal of Applied Econometrics, 8, 231-247. Johansen, S. (1992). Determination of the cointegration rank in the presence of a linear trend. Oxford Bulletin of Economics and Statistics, 54, 383–397. Johansen, S. (1995). Likelihood–based inference in cointegrated vector autoregressive models. Oxford: Oxford University Press. Juselius, K. (2002). Wage, price, and unemployment dynamics and the convergence to purchasing power parity in the Euro Area. Mimeo, University of Copenhagen. Krolzig, H. M., & and Hendry, D. F. (2001). Computer automation of general to specific model selection procedures. Journal of Economic Dynamics and Control, 25, 6-7, 831-866. Layard, R., Nickell, S. J., & Jackman, R. (1991). Unemployment: macroeconomic performance and the labor market. Oxford: Oxford University Press. Osterwald–Lenum, M. (1992). A note with quantiles of the asymptotic distribution of the maximum likelihood cointegration rank test statistics. Oxford Bulletin of Economics and Statistics, 54, 461–472. Sims, C. A. (1980). Macroeconomics and reality. Econometrica, 48, 1–48.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Does One Monetary Policy Fit All?

79

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Sinn, H. W., & Reutter, M. (2000). The minimum inflation rate for Euroland. CESifo Working Paper, 377. Stock, J. H., & Watson, M. W. (1999). Forecasting Inflation. Journal of Monetary Economics, 44, 293–335.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 4

BANKS AS SHAREHOLDERS: THE SPANISH MODEL OF CORPORATE GOVERNANCE Valentín Azofra-Palenzuela, Félix J. López-Iturriaga* and Fernando A. Tejerina-Gaite Department of Financial Economics & Accounting University of Valladolid (Spain)

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Compared to the Anglo-Saxon framework on which most of the financial research has focused, in many European corporate systems such as Spain, banks and other financial intermediaries are more important than capital markets. In addition to lending, banks usually own significant proportions of shares and sit at the board of directors. Our results for a sample of 142 non-financial firms between 1999 and 2002 show that banks play an outstanding role in the corporate governance of the Spanish firms. More specifically, due to the banking specialization in monitoring and control, we find that banks have a dual and very relevant influence on the performance of the firms. For bankcontrolled firms, bank shareholdings can enable colluding interactions among banks and, thus, reduce the performance of the firm. On the contrary, for non-bank controlled firms, bank shareholdings improve the corporate governance and increase the performance of the firm. This effect is particularly significant for the firms in which bank scrutiny is more crucial: when the largest shareholder has voting rights in excess of his cash flow rights, and when the control of the largest shareholder can be more contested.

1. INTRODUCTION The most recent and expanding literature in financial economics emphasizes how the financial system interacts with the mechanisms of corporate governance and with the * Author for correspondence: Félix J. López Iturriaga, Department of Financial Economics & Accounting, Avda. Valle del Esgueva 6, 47011 Valladolid, SPAIN, Tel. +34-983-42 30 00, Fax +34-983-32 38 99, [email protected]

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

82

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

performance of the firms. In this framework the size of the banking sector and the size and liquidity of the stock market are related to the ownership structure of the firms and thus, to the way in which the control of the firms is exerted. Countries have been traditionally classified into two groups on the basis of the orientation of the financial system: bank oriented vs. market oriented system (Allen and Gale, 2001; Beck, Demirgüç-Kunt and Levine, 2001; Demirgüç-Kunt and Levine, 2001a and 2001b). Although the possible superiority of each model has been long debated, we cannot yet conclude that any one of them performs better than its counterpart. On the contrary, each model of financial system has its own advantages. The belief that market-based systems are best has been proved too simplistic and a more nuanced approach is necessary (Allen and Gale, 2001). For example, financial markets may be bad for risk sharing. On the contrary, financial institutions are not simply veils, but are crucial to overcoming market imperfections. An optimal financial system should rely on both financial markets and financial intermediaries. The differences in financial institutions and markets across countries also have implications for corporate governance. While equity markets provide a market for corporate control, the monitoring by the banks could perform the same external oversight role as takeovers. Nevertheless, the differences in the functioning of the mechanisms of corporate governance are so pronounced that one could assert that the central agency problem is different in each framework (Becht and Röell, 1999; Morck et al., 2005). We aim to take into account the complex net of interactions within the financial system and are interested in how banks as shareholders affect the performance of the firms and, consequently, influence on the market value of the owned firms. We focus on the Spanish corporate system, which is clearly bank-oriented and in which financial intermediaries play a prominent role. The analysis of the Spanish case is interesting because, despite the dominance of banks, Spain can be seen as an example of the general tendency of national financial systems to become more market-oriented as they become richer (Schröder and Schrader, 1998). In turn, the Spanish evolution is likely to be replicated by a number of bank-oriented countries. Our results for a sample of 142 Spanish non-financial firms between 1999 and 2002 show that banks play an outstanding role in the corporate governance. Banks do not only act as traditional creditors but also own significant stakes in non-financial firms and sit at the board of directors. More specifically, due to the banking specialization in monitoring and control, we find that banks have a dual and very relevant influence on the performance of the firms. For bank-controlled firms, bank shareholdings can enable colluding interactions among banks and, thus, reduce the performance of the firm. On the contrary, for non-bank controlled firms, bank shareholdings improve the corporate governance and increase the performance of the firm. This effect is particularly significant for the firms in which bank scrutiny is more crucial: when the largest shareholder has voting rights in excess of his cash flow rights, and when the control of the largest shareholder can be more contested. The paper is divided into five sections. Section 2 reviews the foundations of the market vs. bank orientation of financial systems. Section 3 presents the main characteristics of the Spanish financial system and is aimed to frame the problems of corporate governance for Spanish firms in the international arena. Section 4 describes the sample and variables used, and we explain the empirical method. Section 5 shows the empirical results and we discuss

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

83

some implications of the results. In the final section some conclusions are drawn from the most outstanding results.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2. MARKET CONTROLLED AND BANK CONTROLLED FIRMS: THE IMPACT OF BANKS As previously stated, financial systems has been traditionally classified into two main models: the market-oriented or Anglo-Saxon one and the bank-oriented or Continental one1. Although both of them channel household savings to the corporate sector, allocate investment funds among firms, and allow both firms and households to share risks, they perform these tasks in quite different ways. But differences do not only concern the weigh of financial markets and of banks but also the characteristics of financial intermediaries. In the bank oriented countries, on top of banks having a more relevant function, the banking system is more concentrated. This concentration means that banks have a higher market share and provide a wider range of financial services to the corporate sector. While universal banks are common in the Continental European countries, commercial and investment banks are de facto or de iure separated in the USA and in the UK. The international empirical evidence on the differences between both systems is profuse, so we will not focus on these differences (Allen and Gale, 2001; Demirgüç-Kunt and Levine, 2001b; Beck, Demirgüç-Kunt, Levine and Maksimovic, 2001; Domowitz et al., 2001; Schmukler and Vesperoni, 2001). Nevertheless, since Spain is a bank oriented country, we will review how banks perform the intermediation tasks and how this impacts on the corporate governance of non-financial firms. Banks may influence the corporate governance in a number of ways. Let us suggest the three most remarkable ones. Firstly and most evident, they may own a significant proportion of shares. The high ownership concentration in bank-owned firms reduces the free-rider problem to monitor managers but raises an agency problem between large shareholders and minority shareholders. In addition, small shareholders usually deposit their shares in banks, so that banks’ control rights are far above their cash flow rights. In fact, these control rights allow the banks other kinds of influence such as sitting at the boards of directors irrespective of the stake of shares they own. This difference between control rights and cash flow rights is vital to assess the involvement of banks in corporate governance. In the countries with not very developed financial markets, a proportion of control rights far above the cash flow rights allows the large shareholders avoiding the pressure of some mechanisms such as the market for corporate control, and enables them to expropriate rents from the non-controlling shareholders (Grossman and Hart, 1988; Harris and Raviv, 1988; Shleifer and Vishny, 1997). The second issue about the influence of banks on the corporate governance is related to the possible conflict of interests as a consequence of their dual role as creditors and shareholders. Although in most of the countries banks are not usually shareholders at the

1

In recent years a new classification scheme, based on the legal origins of each country, has gained increasing acceptance (La Porta et al., 1997, 1998 and 2000). In spite of its high explanatory power, we will base on the banks vs. markets criterion as a way to underline the role played by banks in corporate governance.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

84

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

same time (La Porta et al., 1999), these last authors show that Spain is an exception and therefore, an interesting environment to test this conflict of interests. The relation between shareholders and lenders raises some agency problems such as the so-called underinvestment problem and the problem of substitution of assets (Jensen and Meckling, 1976; Myers, 1977). In turn, if a bank plays this dual role, it might help to solve these conflicts of interests and, thus, it could have a positive effect over the value of the firm (Prowse, 1990). Banks has additional information over the investment opportunities of the firm and, in fact, banks as shareholders improve the selection of profitable investment projects (Kang et al., 2000; Morck et al., 2000). Besides, firms with banks as shareholders are likely to have better chances of raising external funds (Hoshi et al., 1990) and cheaper funds (Petersen and Rajan, 1994). These ideas are consistent with the evidence outside Spain, showing how financial markets react positively to the participation of banks as shareholders in Continental Europe firms (Pedersen and Thomsen, 2003). The third issue about the influence of banks on corporate governance is the informational content of bank shareholdings, which calls for how the reputation of the bank can affect the market price of the owned firm. Banks are credited as specialists in monitoring and controlling borrowers. Hence, financial markets (especially minority shareholders) trust that banks as shareholders will make more effort in scrutinizing the investment projects of the firms whose shares they own. González (2001) shows that bank shareholdings –irrespective of the conflicts of interests due to the simultaneous role of creditor and shareholder-, discloses information about the good quality of the firm and increases the firm’s market value. Consequently, as a partial abstract, we would like to stress three consequences from the participation of banks in the governance of non-financial firms: 1) The possibility of opportunistic behaviour of banks due to the difference between control rights and cash flow rights; 2) The coincidence of the roles of shareholder and creditor; and 3) The informational content of banks’ stakes.

3. THE SPANISH FINANCIAL SYSTEM 3.1. The Spanish Financial System in the International Arena Spain has been traditionally seen as a country with a bank-oriented financial system. In spite of the general trend among developed countries toward a more market-oriented system, banks and other financial intermediaries are yet the core of the Spanish financial system. In order to provide a broad idea of the Spanish system, we report some descriptive data. Most of the information comes from the comprehensive survey gathered by Beck et al. (2001). The first set of data on which we base our comparison concerns the orientation of the financial system, i.e. the banks vs. markets dichotomy. Consistent with Beck, Demirgüç-Kunt and Levine (2001), we firstly calculate the overall size of the Spanish financial system, both in terms of banks and markets. Our definition of Overall size is the result of scaling by GDP the sum of bank assets and stock market capitalization. To measure the size of the banking sector we have defined Bank assets/GDP as the ratio of the total domestic assets of deposit money banks scaled by GDP. The size of markets is measured by the stock market

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

85

capitalization as a share of GDP2. Based on these measures, we compute the mean value for different groups of countries: the G7, the European countries, the former 15 members of the European Union, and the group of upper middle income countries to which Spain is supposed to belong. 4

3.5

3

Upper middle income G7 Europe EU-15 Spain

2.5

2

1.5

1

0.5

0 Overall size

Bank assets/GDP

Market capitalization/GDP

Bank vs. markets

Source: Demirgüç-Kunt and Levine (2001b) and authors’ calculations

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Figure 1: Size and orientation of the financial system

As shown in Figure 1, Spain can be considered similar to most of the other European countries in terms of the size of the financial system. Although a bit smaller than the most developed G7 countries (1.27 against 1.55), it is larger than the other upper middle income nations. The right hand columns in Table 1 give some clue about the orientation of the financial system, showing a more active role of banks relative to markets. This is consistent with the traditional wisdom of Spain as a bank-oriented country. Nevertheless, the evolution in latest years reported in Figure 2 shows that markets are more and more important relative to banks. With the exception of the 2001-02 break-down of capital markets (due to the crash of the dot-com and technological firms), the ratio of bank assets to market capitalization has fallen from 3.2 to 2.2. This corroborates for Spain (as for other developed countries) the financial disintermediation or the trend to enhance capital markets relative to financial intermediaries. Focusing on the outstanding role of banks, we define three measures of financial intermediaries. First, Liquid liabilities/GDP is defined as the ratio of liquid liabilities of bank and non-bank intermediaries to GDP. It is supposed to be informative about the overall size of financial intermediaries relative to the size of the economy. Second, Claims of deposit money banks/GDP equals deposit money bank credits to the private sector as a share of GDP. This 2

We have also calculated the ratio of bank assets to stock market capitalization as an indicator of bank vs. markets orientation

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

86

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

.

variable is a general indicator of bank activity in the private sector. Finally, Claims of other financial institutions/GDP is the ratio of non-bank credits to GDP and provides a broad measure of non-bank activity in the private sector. 3.50

3.00

2.50

2.00

1.50 1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

Source: Bank of Spain and Madrid Stock Exchange (Bolsa de Madrid) Figure 2: Bank assets-to-market capitalization ratio 2

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1.8 1.6 1.4

Upper middle income G7 Europe EU-15 Spain

1.2 1 0.8 0.6 0.4 0.2 0 Liquid liabilities/GDP

Claims of deposit money banks/GDP

Claims of other financial institutions/GDP

Nonbank vs. banks intermediaries

Source: Demirgüç-Kunt and Levine (2001b) and authors’ calculations Figure 3: Financial intermediaries

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

87

Figure 3 is quite informative about how financial intermediation is performed by banks and other financial institutions. One can see that the size of financial intermediaries in Spain (measured through liquid liabilities) is near the international mean value (0.76 in Spain against 0.88 in the G7 countries and 0.67 in European countries). On the contrary, non-bank intermediaries are much less important in Spain than in the other countries (0.06 against 0.51 and 0.30 respectively). More interestingly, when we compare the size of deposit money banks with the size of other financial intermediaries (right hand columns in Figure 3), we see that the large size of intermediaries in Spain is due basically to the bank sector rather than to other financial institutions. If we compare banks and non-bank intermediaries in terms of assets instead of claims (Figure 4), the results confirm previous ideas: the assets of the Spanish banks are close to other countries but the assets owned by other financial institutions are anomalously low, so that there is an unambiguous leading role of banks as the most important type of financial intermediary in Spain. This is the main reason to provide a more in-depth explanation of the Spanish banking system and the net of relations between non-financial firms and banks in Spain. 2.50

2.00 G7 Europe EU-15 Spain

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1.50

1.00

0.50

0.00 Assets of banks/GDP

Assets of other financial institutions/GDP

Nonbanks vs. banks intermediaries

Source: Demirgüç-Kunt and Levine (2001b) and authors’ calculations Figure 4: Financial intermediaries

3.2. The Spanish Banking System Two of the most meaningful features of the Spanish banking system are the close link between banks and the governance of non-financial firms, and the heterogeneous kind of legal status of banks. Regarding the involvement of banks in the corporate governance, banks own often significant stakes in non-financial firms. This fact is consistent with the highly concentrated

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

88

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

ownership of Spanish firms. In fact, to have a complete view of the Spanish corporate system and how it is related to the international framework, we present some data about the ownership concentration. Recent studies suggest that Berle and Means´ model of widely dispersed corporate ownership is not so common as expected, even in developed countries. In fact, large shareholders control a significant number of firms in many countries. La Porta et al. (1999) traced the control chains of a sample of 30 firms in 27 wealthy economies, including Spain. They found that, except in economies with very good shareholder protection, few of these firms were widely hold. Also, the controlling shareholders typically had power over firms significantly in excess of their cash-flow rights, mainly through the use of pyramids, deviations from one-share-one-vote rule and cross-holdings. Spanish corporations, as most Continental European ones, have a high concentrated ownership structure. Spain belongs to the civil-law legal system, characterized by a lowmedium investors’ protection, which leads to ownership concentration. This fact has been shown both by internacional research (La Porta et al., 1999; Faccio and Lang, 2000; Faccio and Lang, 2002) and domestic evidence (Crespí and García-Cestona, 2001; Galve and Salas, 1996). In particular, Crespí y García-Cestona (2001) show that the sum of direct and indirect voting rights of the largest stake is, on average, 40.09% for non-financial companies. Taking the top three largest shareholders, they get an accumulated holding of 56.59%. Although this study simply adds up direct and indirect control stakes without tracing them to the ultimate owners, the figures are similar with those in Tejerina (2006), where control chains are developed. Regarding the owner identity, the table below compares some of the previously cited studies. The table presents the percentage of firms controlled by different controlling owners at the 10% threshold. Controlling shareholders are classified into five types 1) Family: a family or an individual; 2) Widely held corporation: a non financial firm, widely held at the control threshold; 3) Widely held financial institution: a financial firm that is widely held at the control threshold; 4) State: a national government (domestic or foreign) or government agency; and 5) Miscellaneous: voting trusts, employees or cooperatives. Companies in which no shareholder controls at least 10% of votes are classified as widely held. Table 1: Control of Spanish large publicly traded firms

Family Widely held corporation Widely held financial institution State Miscellaneous Widely Held Total number of firms

La Porta et al. (1999) 1995 25 0 15 45 0 15 20

Faccio and Lang (2000, 2002) 1997 74.07 1.69 11.11 3.61 0.47 9.06 530

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Tejerina (2006) 2002 60.29 6.61 21.31 2.20 0.73 8.82 136

Banks as Shareholders: The Spanish Model of Corporate Governance

89

The differences among the studies shown in Table 1 arises both from the year of reference and from the dissimilar number of firms in each sample. So, at the end of 1995 Spanish firms were still engaged in the transition towards a more international and competitive environment. The State still held important stakes in several of the largest corporations, although privatization was very significant in the next years. In addition, the largest samples include a higher number of small firms, which leads to a more important family control. In spite of that, most of the firms have a control shareholder owning more than 10% of the voting rights. This confirms the high degree of concentration in Spanish companies ownership structure. Another characteristic of the ownership structure of Spanish firms is the relevance of financial institutions as majority shareholders. Tejerina (2006) reports 21.31% of firms controlled by financial intermediaries distributed as follows: 14.70% commercial banks, saving banks and credit cooperatives, 4.41% investment banks, and 2.20% other financial institutions. Due to the important function of banks in the Spanish economic system, banks are not only creditors or shareholders, but also sit at the board of directors. In Table 2 we report some descriptive data about the presence of banks in the corporate governance of non-financial firms (Tejerina, 2006). We would like to stress the growing proportion of firms with at least one bank as shareholder (it increases from 58.1% to 63.2% between 1999 and 2002). This increase is even higher when we exclude investment banks to focus only on commercial banks (38.3% vs. 48.5%). We can infer that almost half of the Spanish quoted firms have a commercial bank as reference shareholder. The proportion of representatives of banks at the boards of directors is quite stable and, jointly, two thirds of Spanish listed firms are somehow under banking influence, either as shareholder or as director.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 2: Banks as participants in the corporate governance 1999 58.1% 28.4% 19.7% 38.3% 38.3% 23.4% 13.9% 63.1%

Bank as shareholder Shareholder and creditor Shareholder but not creditor Commercial bank as shareholder Bank as director Director and creditor Director but not creditor Shareholder and/or director

2002 63.2% 32.3% 30.9% 48.5% 37.5% 19.8% 17.7% 68.4%

Source: Tejerina (2006)

Table 3: Banks as shareholders

Average number of banks as shareholders Mean percentage of shares

1999 1.6 19.6%

2002 2.3 21.1%

Source: Tejerina (2006)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

90

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

In Table 3 we report some data about the characteristics of banks as shareholders. There is an increase in the mean number of banks owning significant proportions of shares (1.6 banks vs. 2.3 banks), although the average stake of each bank remains with small changes. In Table 4 we report some descriptive data about the firms with at least one director representing banks. Although the mean number of bank directors and the proportion on the whole board do not change significantly, there is a different trend conditional on the type of director: while unaffiliated directors are more usual, the proportion of outside directors appointed by banks decreases. Table 4: Banks on the board of directors

Average number of bank directors Proportion of banking directors Proportion of firms with unaffiliated directors Proportion of firms with outside directors Proportion of firms with banks as shareholders

1999 2.3 20.1% 74.1% 46.3% 87%

2002 2.5 21.5% 80.4% 37.2% 86.3%

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Source: Tejerina (2006)

The second remarkable characteristic of the Spanish banking sector is the heterogeneous legal status of banks. In Spain there are three main banking institutions: Commercial banks, Savings banks and Credit cooperatives. While commercial banks are clearly stockholderoriented, savings banks are private foundations, with a board of trustees with representatives from regional authorities, city halls, workers, depositors and the founding entity (Crespí et al., 2004). Saving banks are a quite specific institution in the Spanish financial system with no formal owners and no market for corporate control of them. Credit cooperatives are mainly located in rural areas. In spite of the differences in the legal status among them, they compete under equal conditions in the loan, deposit and financial service markets. There are no operational differences, and regulations are the same for the three types, as well as their accounting practices, external reporting and credit-risk management standards. Nevertheless, there are huge differences in terms of market share: while commercial banks and saving banks account each one for around 45% of the credit market share, credit cooperatives account for only around 7%. Likewise, the deposit market share for saving banks, commercial banks and credit cooperatives are, respectively, 58%, 35% and 7%. Regarding the stake in non-financial firms, there are some common points and some differences among institutions (Casasola et al., 2001; Crespí et al., 2001). Commercial banks, saving banks and credit cooperatives has in common that they do not usually own a too high percentage of shares in order not to internalize the costs of expropriating minority shareholders. In spite of being low, this percentage is high enough to enable banks to control the firms (Casasola et al., 2001). Some differences among the different kinds of banks are shown in Table 5. As one can see, saving banks are more prone to own shares than their counterparts, either as first or second shareholder. Commercial banks and credit cooperatives do not show significant differences among them.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

91

Table 5: Distribution of banks as main shareholders

Commercial banks Saving banks Credit cooperatives Total

1st shareholder 30% 39% 31% 100%

2nd shareholder 24% 49% 27% 100%

Source: Casasola et al. (2001)

Nonetheless, since the three kinds of banks can compete under equal conditions in markets, hereinafter we will include all of them under the general term of banks.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

4. EMPIRICAL DESIGN Our research aims to analyze how banks as shareholders influence the performance of the firms in which they participate. This influence may be conditional upon three issues. Firstly, we have to take into account if the largest shareholder has difference of rights, i.e., if he has decision rights above his cash flow rights. The higher the difference between decision rights and cash flow rights, the more prone he is to extract private benefits, even if this behaviour is harmful for the firm (Grossman and Hart, 1988; Harris and Raviv, 1988; Shleifer and Vishny, 1997). In there are difference of rights, the possible control by other shareholders becomes more relevant. Consequently, banks as shareholders may play a different role depending on the ability to expropriate of the largest shareholder. Secondly, the attitude of the bank may be different if he is the largest shareholder or, by the contrary, there is another larger shareholder. Thirdly, we introduce the involvement of other reference shareholders or how they engage in the control of the firm. The closer these reference shareholders are to the largest shareholder, the higher their involvement in the corporate control is.

4.1. Sources and Data We have built a data base with data from 142 non-financial Spanish firms between 1999 and 2002. Our data are very representative of the Spanish capital markets and they account for more than 90% of the Spanish listed firms. The panel is unbalanced since some new firms have joined the sample but there have been mergers and acquisitions, so some observations have disappeared during our period of study. Nevertheless, the estimations based on unbalanced panels are as reliable as those based on balanced panels (Arellano, 2003). The final number of observations is 561 as shown in Table 5:

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

92

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite Table 6: Distribution of the sample across industries

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Industry Consumer goods Intermediate goods Energy Building Communications Services Technological (New Market) TOTAL

1999 31 37 10 8 2 41 12 141

2000 31 37 10 8 3 41 12 142

2001 31 37 10 8 3 41 12 142

2002 28 36 9 8 3 40 12 136

Data about the ownership structure of the firms are key for our purpose. The analysis of the ownership structure allows determining bank shareholdings, and -more importantlyclarify the possible excess of control rights over cash flow rights. This kind of information is not easy to achieve due to the complex ownership pyramids so common in the Spanish corporate system (Morck et al., 2005). Voting rights and cash flow rights have been calculated following La Porta et al. (1999), Faccio and Lang (2002), and Claessens et al. (2000). The core concept is the chain of control, which is based on the voting rights. It requires defining a threshold proportion of shares so that a shareholder can be considered as controlling shareholder. Consistent with most of the prior literature, we have defined the threshold proportion as 10%.3 A firm has a pyramidal ownership structure if: 1) has an ultimate owner, and 2) there is at least one listed firm between the owned firm and the ultimate owner throughout the chain of control. We consider a firm having a controlling shareholder or an ultimate owner if the direct and indirect voting rights over the firm are, at least, 10%. Otherwise, the firm is supposed to have a disperse ownership structure. The methodology of chains of control relies to allocate voting rights on the weakest link of the chain. For instance, let us suppose that a family owns 15% of the shares of firm A, and firm A owns 20% of the shares of firm B. In this case, the family has 3% of cash flow rights on firm B (0.20·0.15) but 15% of voting rights (the weakest link of the chain of control). Let us present two examples in order to explain the methodology of chains of control. Both of them bases on real data of Spanish firms in 2002: In example 1, an analysis of the ownership structure without the methodology of the last owner and the chains of control would stop at the first stage of shareholders. Thus, we would conclude that 30% of the voting rights over Aldeasa, S.A. are in the hands of Altadis, S.A. Nevertheless, 14.9% of the shares of Altadis, S.A. are owned by the Chase Manhattan Bank. This is the weakest link in the chain of control, so Chase Manhattan Bank has 14.9% of voting rights but just 4.4% (the result of 0.3·0.149) of cash flow rights.

3

We have run a sensitivity analysis with a threshold of 20%. Results remain unaffected and are not reported for brevity.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

93

Example 1.

Aldeasa, S.A. (2002)

30% Altadis, S.A.

5.2%

3.5%

Bestinver Gestion, S.A.

Fidelity International Limited

5% SEPI

14.9% Chase Manhattan Bank

As far as example 2 is concerned, the first stage in the chain of control shows that Renfila, S.A. is controlled by Roviralta y cia, S.A., which owns around 23.8% of shares. Nevertheless, the Lasa family has significant stakes in four shareholders of Renfila, S.A. (Donato Lasa, S.A., Inversiones y Promociones Lasa, S.A., Edificios, Promociones e Inversions de la Construcción, S.A. and Promofila, S.L.) In fact, the Lasa family has 50.6% of voting rights (the sum of the weakest links in the chain of control, i.e., 13.9% + 12.4% + 11.6% + 12.7%) and 30.6% of cash flow rights (13.9%·21.9%·100% + 12.4%·100% + 11.6%·21.4%·100% + 12.7%·100%):

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Example 2. Renfila, S.A.

13.9%

Donato Lasa, S.A.

21.9%

Alquiler de Fincas, S.L.

100% Lasa family

12.4%

Inversiones y Promociones Lasa, S.A.

11.6%

Edificios Promociones e Inv. de la const., S.A.

100%

Lasa family

21.4%

Inversiones y Promociones Lasa, S.A.

12.7%

Promofila S.L.

100%

Lasa family

18.4%

23.8%

Amesepten Förvaltning AB-OM

Roviralta y cia., S.A.

Control

6.6%

6.2%

Fastighet Meteor HD. B.V.

Sveriges F. Bank

Control

Fundación M. Francisca de Roviralta

100%

Lasa family

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Spira Investments

94

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

The Comisión Nacional del Mercado de Valores (CNMV, the Spanish Securities Exchange Commission) provides with information about shareholders who –directly or indirectly- own more than 5% of the shares. This information usually concerns the first or the second stage and does not allow going up in the chain of control. Consequently, when shareholders are firms or institutional investors, we have looked for the ownership structure of the shareholder in INFORMA, a database for more than 880.000 Spanish firms. Replicating this schema all over the chain of control, this procedure allows the calculation of both voting rights and cash flow rights4. The information about the Balance sheet and the State of Income and Expenses has been provided by the CNMV too.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

4.2. Variables We are interested in testing the effect of bank shareholdings on the performance of the firm. Therefore, we use two types of dependent variable, both of them concerning the performance of the firm: a market based measure and an accounting measure. The market based measure (VALUE) is the market-to-book value of the firm (Claessens et al., 2002; La Porta et al., 2002). The market value of the firm is defined as the sum of the market value of equity and the book value of debt. VALUE can be understood as the present value of the possible discretionary behavior of the controlling shareholders (Cronqvist and Nilsson, 2003). The accounting measure of the firm’s performance is ROA, i.e., the return on assets. Most of the explanatory variables concern the presence of banks as shareholders. A summary of them is provided in Table 7. Although we consider as banks both commercial banks and saving banks, we have excluded the investment banks due to their different strategic approach. Investment banks make most of their income from commissions and capital gains. They specialize in holding shares for medium and long term in order to achieve capital gains but, unlike commercial banks, investment banks seldom engage in the corporate governance of the firms in which they participate. In turn, they hardly have any influence on the owned firms. On the contrary, commercial banks are much more active in the control of the firms. There are some other independent variables to introduce the characteristics of the governance and the ownership of the firms potentially affecting the activism of banking shareholders. RIGHTD is a dummy variable which equals 1 when the largest shareholder – irrespective of being a bank or not-bank- has voting rights in excess of the cash flow rights, and equals 0 otherwise. This variable is aimed to identify the firms in which the largest shareholder has more ability to extract private benefits. CONTEST provides a measure of the extent to which the power of the largest shareholder can be contested by other reference shareholders (Maury and Pajuste, 2005). It is defined as the sum of the squares of the differences between the first and the second largest ownership stakes, and the second and third largest ownership stakes5. High values for CONTEST mean a high concentration of control, whereas low values indicate that the largest shareholder faces greater contestability.

4

When the shareholder was a non-Spanish firm we have looked for additional information at the corresponding website or asked for information by e-mail. 5 CONTEST = (C1-C2)2 + (C2-C3)2, in which Ci is the cash flow rights of the first, second and third shareholder. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

95

Banks as Shareholders: The Spanish Model of Corporate Governance

A number of control variables are included too. We have controlled for the size of the firm by LOGTA, the log of the total assets. We have also controlled for the growth of the firm by SALESG, the sales growth (in percentage). We have also added industry and time dummy variables. Table 7: Variables concerning banks as shareholders Variable

BANKOW %BANKOW BB BNB NBB NBNB

Description Dummy variable that equals 1 if there is a bank as shareholder owning at least 3% of the shares and equals 0 otherwise. Percentage of the shares of a non-financial firm owned by a bank. Dummy variable that equals 1 if both the largest and the second largest shareholder are banks, and equals 0 otherwise. Dummy variable that equals 1 if the largest shareholder is a bank and the second largest shareholder is not a bank, and equals 0 otherwise. Dummy variable that equals 1 if the largest shareholder is not a bank and the second largest shareholder is a bank, and equals 0 otherwise. Dummy variable that equals 1 if none of the two largest shareholders are banks and, and equals 0 otherwise.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

4.3. Empirical Method The cross sectional analyses have some shortcomings (Stock and Watson, 2003), so we choose as empirical method the panel data procedure. Panel data combines cross section data with time series. Ours is a micropanel since the number of observations (around 140) is notably higher than periods (4 years). As previously stated, it is an unbalanced panel data but this issue does not effect the consistency of the estimates. Relative to cross sectional analyses, panel data allow relaxing some implicit assumptions in the statistical analysis. One of the main advantages is the control for the so-called constant and unobservable heterogeneity introduce by the fixed-effects term (Arellano, 2003). Each one of the firms in the sample has its own specificity –e.g., the preferences and culture of the managers, the traditional functioning of the mechanisms of corporate governance, etc. This specificity is different from a company to another and likely to be kept throughout the study period. A pooling analysis of all the companies without noticing these particular characteristics could cause an omission bias and distort the results. Another interesting characteristic of panel data is the dynamic dimension since it enhances testing long time adjusting processes and determining the value of the firm when the explanatory variable change. Our model could be expressed as follows where sub-index i stands for the observation and sub-index t stands for time: qit = β0 + PBit·β1 + Cit·β2 + Sit·β3 + Ait·β4 + εit qit is the performance of the firm β0 is the intercept Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

(1)

96

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite PBit is a 1 X 1 dimension vector of the variables concerning banking shareholdings Cit is a 1 X 2 dimension vector with the control variables Sit is a 1 X 6 dimension vector with industry dummy variables Ait is a 1 X 3 dimension vector with the time dummy variables εit is the random effect for each observation and year

Model (1) is a random effect model since εit includes the individual unobservable effects. This random effect term could be split into two effects as follows, where ηi stands for the individual unobservable effect and υit stands for the random effect: εit = ηi + υit

(2)

Thus, model (1) would become model (3):

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

qit = β0 + PBit·β1 + Cit·β2 + Sit·β3 + Ait·β4 + ηi + υit

(3)

A crucial issue in panel data analysis is model specification. It is highly important to know the kind of relation between the fixed-effects term ηi and the other independent variables, given that there will be the same relation between the independent variables and the random component. If there is no correlation, the most suitable method is minimum generalized squares, which provides the linear unbiased estimator with minimal variance. In the opposite case, this estimator becomes inconsistent and individual effects must be removed. Among the alternative ways to remove them, the within-group method is chosen, because it allows keeping as many available periods as possible. Thus, we provide the Hausman test to test the null hypothesis of lack of correlation between the independent variables and the fixed-effects term (Hausman, 1978)6. We are concerned with the possible reverse causality between firm’s value and banking shareholdings. This means that a positive relation between the performance of firms and bank shareholdings could mean that banks choose to buy the shares of the best performing firms instead of banks increasing the performance of the firms that they own. In this case, the presence of banks (bank shareholdings, bank directorships or bank liabilities) should change depending on the value of the firm. Nevertheless, firms are not likely to change their ownership structure, the board of directors or the relation with bank creditors conditional upon contingent over or undervaluation. La Porta et al. (1999) point that ownership structure is stable throughout time. The reverse causality could only bias the results if banks quickly and systematically modified their involvement in most of the non-financial firms according to the valuation of the firm, which is far away from the usual behavior of banks7.

6 7

Hausman test follows a chi-squared distribution with as many degrees of freedom as estimated coefficients. This assertion is consistent with the results of Claessens et al. (2002), who analyze the relation between the value of the firm and the ownership structure.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

97

5. FINDINGS 5.1. Descriptive Analysis We begin our empirical analysis by reporting the basic descriptive statistics of the variables (Table 8). As shown, 43% of the firms in our sample have a bank as reference shareholder. 15% of the firms have a bank as the largest shareholder (the sum of BB and BNB), although in only 3.2% of firms both the largest and the second largest shareholders are a bank. Around 10% of the firms have a bank as second reference shareholder. Importantly, in almost half of the firms (47%) the largest shareholder –irrespective of being a bank or a nonbank- have control rights over his cash flow rights, so that he may have incentives to extract private benefits from other shareholders. Table 8: Descriptive statistics

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Number of observations, mean, standard deviation, maximum and minimum value. VALUE is a proxy of the market-to-book ratio; ROA is the return on assets; BANKOW the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders; RIGHTD is a dummy variable for differences between control and cash flow rights; CONTEST is the contest to the control of the largest shareholder; LOGTA is the log of assets and SALESG is the sales growth.

Variables VALUE ROA BANKOW %BANKOW BB BNB NBB NBNB RIGHTD CONTEST LOGTA SALESG

Obs. 531 546 561 559 561 561 561 561 561 561 548 541

Mean 1.36 0.071 0.43 0.092 0.032 0.117 0.096 0.74 0.47 0.16 5.53 1.46

Std. Dev. 1.31 0.074 0.49 0.16 0.17 0.32 0.29 0.43 0.49 0.23 0.81 28.16

Max. 22.6 0.42 1 0.93 1 1 1 1 1 0.99 7.96 654.3

Min. 0.63 -0.92 0 0 0 0 0 0 0 0 3.59 -1

A preliminary analysis of the influence of banks shareholdings on the performance of the owned firms is given by a test of means comparison. Hence, we test whether VALUE and ROA show statistically significant differences between groups depending on the contest faced by the largest shareholder and on the fact that the largest shareholder has control rights in excess of cash flow rights. The results of the tests of means comparison are reported in Tables 9 and 10. Whereas in Table 9 the sample is divided depending on the extent to which the control of the largest shareholder is contested, in Table 10 we divide the sample on the basis of difference of rights.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

98

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite Table 9: Comparison of means depending on the contest to the largest shareholder Value of the firm (VALUE) and performance of the firm (ROA) according to the contest to the control of the largest shareholder. The sample is divided depending on the median value of CONTEST. p-value is the highest level of significance to reject the null hypothesis of equality of means between groups (the lower the p-value, the more different the mean values are). High contest VALUE

ROA

Low contest VALUE

ROA

Comparison of means (p-value) VALUE

ROA

Panel A: Bank control Bank as 2nd shareholder (BB)

1.134

0.063

1.213

0.077

0.30

0.35

1.279

0.089

1.194

0.083

0.40

0.44

0.33

0.037

0.92

0.27

2.014

0.104

1.596

0.072

0.62

0.06

Non-bank as 2 shareholder (NBNB)

1.301

0.057

1.385

0.075

0.40

0.02

p-value

0.37

0.014

0.48

0.64

Non-bank (BNB)

as

2

nd

shareholder

p-value Panel B: Non-bank control Bank as 2nd shareholder (NBB)

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

nd

Tables 9 and 10 show no differences in VALUE between groups irrespective of the kind of control (bank control vs. non-bank control), of the contest, and of the differences in rights. Nevertheless, there are some differences in the return on assets for the firms with higher contestability and for the firms in which the largest shareholder has control rights in excess of his cash flow rights. In both cases, bank shareholdings have a dual effect. In bank controlled firms, a bank as second shareholder is significantly related to a lower performance of the firm (6.3% against 8.9% in Table 9 and 6.7% against 8.8% in Table 10). On the contrary, for nonbank controlled firms, a bank as second shareholder increases the performance of the firm (10.4% against 5.7% in Table 9 and 8.9% against 6.5% in Table 10). Although quite preliminary, these results suggest a positive impact of bank shareholdings on firms’ performance and, consequently, a positive contribution of banks for the corporate governance of some non-financial firms. When firms are controlled by a bank, there is not such need of the influence of another bank. But if the firm is not controlled by a bank, the monitoring provided by banks as second shareholder means a significantly higher performance. Another interesting piece of evidence provided by Table 9 is the significantly higher ROA for firms with higher contest relative to firms with lower contest in Panel B (non-bank controlled). This result could suggest that the control of non-banking largest shareholders must be more contested that the control of banks, corroborating the positive involvement of banks in the corporate governance of the firms in which they participate. Thus, we could assert that bank shareholdings have a different influence on Spanish firms depending on the characteristics of the owned firms. While banks could collude when both

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

99

Banks as Shareholders: The Spanish Model of Corporate Governance

the largest and the second largest shareholders are banks, they seem to act as effective mechanisms of control when firms are controlled by non-bank shareholders. Table 10: Comparison of means depending on the differences between control rights and cash flow rights Value of the firm (VALUE) and performance of the firm (ROA) according to the fact that the largest shareholder has control rights over his cash flow rights. RIGHTD equals 1 when there are differences in rights and equals 0 otherwise. p-value is the highest level of significance to reject the null hypothesis of equality of means between groups (the lower the p-value, the more different the mean values are). RIGHTD=1 Panel A: Bank control Bank as 2nd shareholder (BB) Non-bank as 2nd shareholder (BNB) p-value Panel B: Non-bank control Bank as 2nd shareholder (NBB) Non-bank as 2nd shareholder (NBNB) p-value

RIGHTD=0

Comparison of means (p-value)

VALUE

ROA

VALUE

ROA

VALUE

ROA

1.160 1.212 0.54

0.067 0.088 0.003

1.134 1.265 0.65

0.067 0.085 0.40

0.76 0.60

0.99 0.74

2.357

0.089

1.252

0.087

0.19

0.90

1.328

0.065

1.354

0.067

0.79

0.74

0.22

0.036

0.61

0.30

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5.2. Explanatory Analysis Our explanatory analysis is built over the estimation of model (3). The results for the baseline estimations are reported in Tables 11 and 12: Table 11 shows the influence of bank shareholdings on the value of the firms and Table 12 the influence on the accounting performance of the firms. The baseline results point that the market value of the firm depends on the nonbank control (NBB and NBNB in Table 11) and that accounting measures of firms’ performance are affected by bank shareholdings (BANKOW in Table 12). Concerning the market value, one can see that seemly there is not a common pattern: sometimes nonbank control positively influences the value of the firm (NBB) and sometimes nonbank control has a negative effect on the value (NBNB). However contradictory it may seem, the results make sense if we consider the monitoring function of banks: when a bank is the second largest shareholder, it monitors the controlling nonbank shareholder and, consequently, improves the value of the firm. On the contrary, if the second largest shareholder is also a nonbank shareholder, both shareholders might collude and extract private benefits in detriment of the other minority shareholders. As far as the return on assets is concerned (Table 12), bank ownership positively impacts on the performance of the firm, although it does not imply any effect on the market value of the firm. As we will expose later, this result could be explained by a certain reputation effect of banks.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

100

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite Table 11: Impact of bank shareholding on the value of the firm Estimated coefficients and (standard deviation) from the panel data estimation of model (3). The dependent variable is always VALUE, a proxy of the market-to-book ratio; BANKOW is a dummy variable for the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders (bank and bank; bank and no-bank; no-bank and bank; nobank and no-bank respectively); LOGTA is the log of assets and SALESG is the sales growth. All the estimations are controlled for industry and time effects (not reported). Hausman test is a test for the correlation between the independent variables and the fixed-effects term. The Ftest and the Wald test allow testing the joint significance of all the explanatory variables. *** for 99% confidence level, ** for 95% and * for 90%.

BANKOW

(1) 0.090 (0.137)

(2)

%BANKOW

(3)

(4)

(5)

(6)

0.091 (0.456)

BB

-0.133 (0.350)

BNB

-0.060 (0.206)

NBB

0.796 (0.194)

***

NBNB LOGTA

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

SALESG Adj.-R2 F test Wald test Hausman test Observations

0.113 (0.105) 0.007 (0.082) 0.124 44.96 8.85 527

0.129 (0.103) 0.008 (0.082) 0.122 ***

44.22 6.68 527

0.141 (0.102) 0.009 (0.082) 0.122 ***

44.39 6.19 527

0.138 (0.101) 0.007 (0.082) 0.122 ***

44.32 6.18 527

0.103 (0.101) 0.002 (0.080) 0.178 ***

61.45 8.65 527

-1.233 (0.263) -1.116 (0.527) -0.001 (0.086) 0.09 6.81

***

18.82

***

**

***

***

527

On top of these basic results, we are concerned with two characteristics of the control of firms by banks potentially affecting the performance of the non-financial firms. These characteristics are the difference between control rights and cash flow rights (and, consequently, their incentives to expropriate wealth from other shareholders), and the extent to which the control of the largest shareholder may be contested by other shareholders. This is why we replicate the above analyses in two scenarios. The first scenario is the firms in which the largest shareholder has control rights in excess of cash flow rights. The second scenario are the firms in which the largest shareholder have less need of other shareholders to control the firm, i.e., when his control is less contested by the other shareholders. The results of these estimates are reported in Tables 13 and 14. In order to clarify the results and to focus on our core variables, we do not report the results for the control variables.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

101

Banks as Shareholders: The Spanish Model of Corporate Governance Table 12: Impact of bank shareholding on the return on assets of the firm Estimated coefficients and (standard deviation) from the panel data estimation of model (3). The dependent variable is always ROA, the return on assets; BANKOW is a dummy variable for the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders (bank and bank; bank and no-bank; no-bank and bank; no-bank and no-bank respectively); LOGTA is the log of assets and SALESG is the sales growth. All the estimations are controlled for industry and time effects (not reported). Hausman test is a test for the correlation between the independent variables and the fixed-effects term. The Wald test allows testing the joint significance of all the explanatory variables. *** for 99% confidence level, ** for 95% and * for 90%. (1) BANKOW

0.043 (0.012)

(2)

(3)

(4)

(5)

(6)

***

%BANKOW

0.092 (0.070

BB

-0.005 (0.026)

BNB

0.001 (0.022)

NBB

0.020 (0.015)

NBNB LOGTA

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

SALESG Adj.-R2 F-test Hausman test Observations

-0.012 (0.030) 0.007 (0.001) 0.125 9.39 23.02 539

***

***

-0.003 (0.030) 0.007 (0.001) 0.100 7.29 19.11 538

***

*** ***

0.005 (0.030) 0.007 (0.001) 0.098 7.11 17.73 539

***

*** ***

0.004 (0.030) 0.007 (0.001) 0.098 7.10 17.93 539

***

*** ***

0.003 (0.030) 0.007 (0.001) 0.102 7.43 17.88 539

***

*** ***

-0.018 (0.015) 0.001 (0.030) 0.007 (0.001) 0.101 7.36 17.72

***

*** ***

539

As shown in Table 13, banks as owners (BANKOW) are positively related to the return on assets when there are no differences between control and cash flow rights. Nevertheless, by and large, the most outstanding results in Table 13 are those concerning the role of banks when there are incentives to expropriate. When a bank is the largest shareholder (both BB and BNB), the ownership structure does not affect the performance of the firm. It could mean that banks are not usually involved in extracting private benefits. Furthermore, when the largest shareholder is not a bank (both NBB and NBNB), banks are key shareholders to monitor the largest shareholder. In fact, non-bank shareholders can collude to expropriate other shareholders (as suggested by the negative coefficient) whereas a bank as second shareholder monitor the largest non-bank shareholder and improves the value of the firm (as suggested by the positive coefficient).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

102

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite Table 13: Impact of bank shareholding depending on the difference in rights Estimated coefficients and (standard deviation) from the panel data estimation of model (3). RIGHTD equals 1 when the control rights of the largest shareholder are above his cash flow rights and equals 0 otherwise. The dependent variables are VALUE, a proxy for the marketto-book ratio, and ROA, the return on assets; BANKOW is a dummy variable for the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders (bank and bank; bank and no-bank; no-bank and bank; no-bank and no-bank respectively). All the estimations are controlled for industry and time effects (not reported). Hausman test is a test for the correlation between the independent variables and the fixedeffects term. The F-test and the Wald test allow testing the joint significance of all the explanatory variables. *** for 99% confidence level, ** for 95% and * for 90%.

BANKOW Adj.-R2 Hausman test F / Wald test Observations %BANKOW

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Adj.-R2 Hausman test F / Wald test Observations BB Adj.-R2 Hausman test F / Wald test Observations BNB Adj.-R2 Hausman test F / Wald test Observations NBB Adj.-R2 Hausman test F / Wald test Observations NBNB Adj.-R2 Hausman test F / Wald test Observations

Firms with differences between control and cash flow rights (RIGHTD=1) VALUE ROA -0.201 -0.003 (0.269) (0.011) 0.166 0.281 9.23 53.54 27.91 *** 8.89 245 253 0.926 0.089 (1.143) (0.081) 0.186 0.303 4.93 58.00 27.97 *** 9.81 245 252 0.074 -0.018 (0.512) (0.021) 0.182 0.283 5.34 51.23 27.31 *** 9.01 245 253 -0.121 0.023 (0.426) (0.018) 0.181 0.287 5.38 51.16 27.36 *** 9.18 245 253 0.677 ** 0.004 (0.300) (0.012) 0.171 0.267 8.69 61.16 33.22 ** 9.71 245 253 -0.514 ** -0.007 (0.272) (0.011) 0.216 0.268 0.55 52.08 31.43 *** 9.79 245 253

*** ***

*** ***

*** ***

*** ***

*** ***

*** ***

Firms without differences between control and cash flow rights (RIGHTD=0) VALUE ROA 0.011 0.030 ** (0.083) (0.012) 0.136 0.202 14.04 ** 4.15 4.95 *** 83.54 *** 282 286 0.617 0.030 (0.537) (0.052) 0.142 0.140 16.19 ** 14.91 5.20 *** 38.52 *** 282 286 -0.067 0.013 (0.270) (0.061) 0.136 0.112 10.68 * 21.44 *** *** 4.96 4.05 *** 282 286 0.125 -0.013 (0.194) (0.045) 0.138 0.113 14.64 ** 35.83 *** *** 5.03 4.06 *** 282 286 -0.121 0.011 (0.139) (0.032) 0.139 0.113 13.68 ** 13.26 ** *** 5.10 4.06 *** 282 286 0.072 -0.007 (0.136) (0.031) 0.137 0.112 11.60 * 13.38 ** 5.00 *** 4.05 *** 282 286

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Banks as Shareholders: The Spanish Model of Corporate Governance

103

Table 14: Impact of bank shareholding depending on the contest to the largest shareholder Estimated coefficients and (standard deviation) from the panel data estimation of model (3). CONTEST is the lack of contest to the control of the largest shareholder. The dependent variables are VALUE, a proxy for the market-to-book ratio, and ROA, the return on assets; BANKOW is a dummy variable for the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders (bank and bank; bank and no-bank; no-bank and bank; no-bank and no-bank respectively). All the estimations are controlled for industry and time effects (not reported). Hausman test is a test for the correlation between the independent variables and the fixed-effects term. The F-test and the Wald test allow testing the joint significance of all the explanatory variables. *** for 99% confidence level, ** for 95% and * for 90%.

BANKOW Adj.-R2 Hausman test F / Wald test Observations %BANKOW

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Adj.-R2 Hausman test F / Wald test Observations BB Adj.-R2 Hausman test F / Wald test Observations BNB Adj.-R2 Hausman test F / Wald test Observations NBB Adj.-R2 Hausman test F / Wald test Observations NBNB Adj.-R2 Hausman test F / Wald test Observations

Firms with high contest VALUE ROA -0.076 0.041 (0.126) (0.014) 0.155 0.126 5.01 6.15 42.74 *** 40.62 269 273 1.093 0.083 (0.731) (0.067) 0.166 0.105 0.91 4.05 44.80 *** 32.55 269 273 -0.120 -0.011 (0.233) (0.028) 0.160 0.100 0.05 2.78 42.53 *** 31.07 269 273 0.103 0.024 (0.204) (0.019) 0.159 0.109 0.87 2.96 42.50 *** 32.74 269 273 0.157 0.035 (0.186) (0.019) 0.173 0.170 14.11 4.93 43.25 *** 34.92 269 273 -0.180 -0.034 (0.200) (0.015) 0.173 0.173 8.42 2.87 43.20 *** 36.81 269 273

***

***

***

***

***

*

***

**

***

Firms with low contest VALUE ROA 0.059 0.029 (0.156) (0.011) 0.296 0.259 2.80 49.44 32.91 *** 10.45 258 266 -0.007 0.116 (0.459) (0.088) 0.291 0.247 1.38 51.05 32.50 *** 9.73 258 266 -0.123 0.032 (0.511) (0.033) 0.293 0.236 1.64 47.47 32.67 *** 9.23 258 266 -0.126 -0.032 (0.269) (0.033) 0.293 0.236 1.15 47.74 32.78 *** 9.23 258 266 0.080 0.001 (0.213) (0.015) 0.297 0.232 3.27 46.67 32.97 *** 9.03 258 266 0.034 -0.001 (0.175) (0.015) 0.290 0.232 1.67 46.47 32.53 *** 9.03 258 266

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

**

*** ***

*** ***

*** ***

*** ***

*** ***

*** ***

104

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite Table 15: Impact of bank shareholding Estimated coefficients and (standard deviation) from the panel data estimation of model (3) for the firms in which the largest shareholder have more control rights than cash flow rights and his control can be more contested. The dependent variables are VALUE, a proxy for the market-to-book ratio, and ROA, the return on assets; BANKOW is a dummy variable for the presence of banks as shareholders; %BANKOW is the proportion of control rights owned by banks; BB, BNB, NBB and NBNB are dummy variables for banks as one of the two largest shareholders (bank and bank; bank and no-bank; no-bank and bank; no-bank and no-bank respectively). All the estimations are controlled for industry and time effects (not reported). Hausman test is a test for the correlation between the independent variables and the fixedeffects term. The F-test and the Wald test allow testing the joint significance of all the explanatory variables. *** for 99% confidence level, ** for 95% and * for 90%.

BANKOW Adj.-R2 Hausman test F / Wald test Observations %BANKOW

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Adj.-R2 Hausman test F / Wald test Observations BB Adj.-R2 Hausman test F / Wald test Observations BNB Adj.-R2 Hausman test F / Wald test Observations NBB Adj.-R2 Hausman test F / Wald test Observations NBNB Adj.-R2 Hausman test F / Wald test Observations

VALUE -0.739 (0.388) 0.248 3.61 24.25 105 4.408 (3.322) 0.216 0.98 22.10 105 -0.318 (0.688) 0.209 0.93 20.29 105 0.281 (0.696) 0.208 0.75 20.21 105 1.124 (0.600) 0.287 5.94 24.34 105 -0.997 (0.577) 0.274 3.25 23.65 105

**

***

**

**

**

**

***

*

**

ROA 0.005 (0.017) 0.212 3.05 14.02 105 0.051 (0.091) 0.190 1.08 14.24 105 -0.048 (0.025) 0.210 1.88 18.01 105 0.066 (0.035) 0.205 137.42 2.29 105 0.065 (0.020) 0.333 8.24 25.21 105 -0.054 (0.032) 0.195 237.8 2.15 105

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

**

*

*

*** **

***

***

*

*** **

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Banks as Shareholders: The Spanish Model of Corporate Governance

105

In other terms, banks have a pivotal function in the Spanish bank-based system and they seem to engage in the governance of the firms whose shares they own. The comparison between the results for VALUE and ROA points at a sort of reputation effect for banks: banks as second shareholders improve the market value of the firm but not the accounting performance of the firm. This could be understood as capital markets considering bank shareholdings as a signal of the effective scrutiny of managers by banks. When we introduce the contest to the largest shareholder as a criterion to split the sample, we see that bank shareholdings positively affect the performance of the firm irrespective of the degree of contestability (columns 2 and 4 in Table 14). Nevertheless, and consistent with the results of the descriptive analysis, the influence of banks on the performance is only significant for the firms in which the largest shareholder can be more challenged (column 2). This result makes sense because the firms with high contest are the firms in which a second shareholder has more ability to influence on the corporate governance. Since both the excess of control rights over cash flow rights, and the contest to the control of the largest shareholder are so critical determinants of the influence of bank shareholdings we run an additional analysis focusing on the firms in which the largest shareholder have more control rights than cash flow rights and faces more contest from other reference shareholders (Table 15). This last set of results is fully consistent with previous ones. Although there are some exceptions, broadly speaking, bank shareholding has a negative influence on the value of the firm but not necessarily on the ROA. In bank controlled firms, the second shareholder being also a bank negatively affects the performance of the firm. This result could suggest a kind of collusion between both shareholders, as shown by Casasola et al. (2001). Nevertheless, this negative impact on the performance of the firm does not mean a lower market value, which could be explained by the reputation of banks as controlling shareholders. Results in Table 15 corroborate the positive influence of banks when firms are controlled by non-bank shareholders both on the value and on the performance of the firms. The underlying intuition is that banks act as effective monitors, so that the negative influence of a non-bank second shareholder in non-bank controlled firms becomes positive when the second largest shareholder is a bank.

6. CONCLUSION Recent literature has emphasized how the functioning of the mechanisms of corporate governance can be affected by the characteristics of the financial system. Spain has a bankoriented financial system, in which banks provide universal financial services and get involved in the governance of firms in three non-mutually excluding ways: as creditors, as shareholders and as directors. To some extent, the undisputable influence of banks can be related to some characteristics of the Spanish corporate system: high ownership concentration, risk of expropriation of minority shareholders by large shareholders, and control rights in excess of the cash flow rights of the main shareholders. The analysis of the Spanish case is interesting because, despite the dominance of banks, Spain can be seen as an example of the general tendency of national financial systems to

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

106

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

become more market-oriented as they become richer. In turn, the Spanish evolution is likely to be replicated by a number of bank-oriented countries. Our research focuses on the impact of bank shareholdings on the performance of nonfinancial firms. In those cases banks are at the core of a net of possible conflicting relationships between managers, creditors, large dominant shareholders and minority shareholders. Our results show that banks play an outstanding role in the corporate governance of the Spanish firms. This influence is particularly significant for the firms in which bank scrutiny is more crucial: when the largest shareholder has voting rights in excess of his cash flow rights, and when the control of the largest shareholder can be more contested. More specifically, due to the banking specialization in monitoring and control, we find that banks have a dual and very relevant influence on the performance of the firms. For bankcontrolled firms, bank shareholdings can enable colluding interactions among banks and, thus, reduce the performance of the firm. In fact, when the two largest shareholders are banks, when they need each other to control the firm, and their voting rights exceed their cash flow rights, we find that banks have incentives to expropriate minority shareholders and, thus, to harm the performance of the firm. On the contrary, for non-bank controlled firms, bank shareholdings improve the corporate governance and increase the performance of the firm. This positive influence of banks takes place when the non-bank dominant shareholder has voting rights in excess of his cash flow rights (i.e., more incentives to expropriate) and faces more contest from other reference shareholders (i.e., the bank influence is more effective). This effect also suggest a reputation effect of bank shareholdings in capital markets since the increase in market value is not necessarily related to improvements in accounting based measures of performance. In turn, we conclude that the impact of bank shareholdings on the corporate governance of Spanish firms partially is conditional on the ownership structure of firms. Banks can have incentives to collude with other banks in order to extract private benefits but can engage in monitoring activities to improve the performance of the firm too. In some way, this view challenges the traditional belief of capital markets as better channels to fund firms.

REFERENCES Allen, F., & Gale, D. (2001). Comparing financial systems. Cambridge:MIT Press. Arellano, M. (2003). Panel data econometrics. Oxford:Oxford University Press. Becht, M., & Röell, A. (1999). Blockholdings in Europe: An international comparison. European Economic Review, 43, 1049-1056. Beck, T., Demirgüç-Kunt, A., & Levine, R. (2001). The financial structure database. In A. Demirgüç-Kunt, & R. Levine (Eds.), Financial structure and economic growth: A crosscountry comparison of banks, markets, and development (17-80). Cambridge: MIT Press. Beck, T., Demirgüç-Kunt, A., Levine, R., & Maksimovic, V. (2001). Financial structure & economic development: Firm, industry, & country evidence. In A. Demirgüç-Kunt, & R. Levine (Eds.), Financial structure and economic growth: A cross-country comparison of banks, markets, and development (189-241). Cambridge: MIT Press. Casasola, M. J., Sanmartín, M. & Tribó, J.A. (2001). La participación bancaria en estructuras con varios grandes accionistas. Economía Industrial, 341, 43-53.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Banks as Shareholders: The Spanish Model of Corporate Governance

107

Casasola, M.J., & Tribó, J.A. (2004). Banks as blockholders. Carlos III University Working Paper Series DEE-2004-01. Claessens, S., Djankov, S., & Lang, L. (2000). The separation of ownership and control in East Asian corporations. Journal of Financial Economics, 58, 81-112. Claessens, S., Djankov, S., Fan, J., & Lang, L. (2002). Disentangling the incentive and entrenchment effects of large shareholdings. Journal of Finance, 57 (6), 2741-2771. Crespí, R., & García-Cestona, M.A. (2001). Ownership & control of Spanish listed firms. In F. Barca, & M. Becht (Eds.), The Control of Corporate Europe (207-227). Oxford: Oxford University Press. Crespí, R., García Cestona, M.A., & Salas, V. (2004). Governance mechanisms in Spanish banks. Does ownership matter?. Journal of Banking & Finance, 28, 2311-2330. Cronqvist, H., & Nilson, M. (2003). Agency costs of controlling minority shareholders. Journal of Financial and Quantitative Analysis, 38 (4), 695-719. Demirgüç-Kunt, A., & Levine, R. (2001a). Financial structure & economic growth: Perspectives & lessons. In A. Demirgüç-Kunt, & R. Levine (Eds.), Financial structure & economic growth: A cross-country comparison of banks, markets, & development (3-14). Cambridge: MIT Press. Demirgüç-Kunt, A., & Levine, R. (2001b). Bank-based and market-based financial systems: Cross-country comparisons. In A. Demirgüç-Kunt, and R. Levine (Eds.), Financial structure and economic growth: A cross-country comparison of banks, markets, & development (81-140). Cambridge: MIT Press. Domowitz, I., Glen, J., & Madhavan, A. (2001). International evidence on aggregate corporate financing decisions. In A. Demirgüç-Kunt, & R. Levine (Eds.), Financial structure and economic growth: A cross-country comparison of banks, markets, and development (263-295). Cambridge: MIT Press. Faccio, M., & Lang, L. (2002). The ultimate ownership of Western European corporations. Journal of Financial Economics, 65, 365-396. Faccio, M., & Lang, Larry H.P. (2000). The Separation of Ownership and Control: An Analysis of Ultimate Ownership in Western European Corporations. EFA 0136. Available at SSRN: http://ssrn.com/abstract=222429. Galve, C., & Salas, V. (1996). Ownership structure and firm performance: Some empirical evidence from Spain. Managerial and Decision Economics, n. 17, pp. 575-586. González, F. (2001). El contenido informativo de las participaciones industriales de la banca. Economía Industrial. 341, 25-33. Grossman, S., & Hart, O. (1988). One share-one vote & the market for corporate control. Journal of Financial Economics, 20, 175-202. Harris, M., & Raviv, A. (1988). Corporate governance: Voting rights and majority rules. Journal of Financial Economics, 20, 203-235. Hausman, J.A. (1978). Specification tests in Econometrics. Econometrica, 46, 1251-1272. Hoshi, T., Kashyap, A., & Scharfstein, D. (1990). Bank monitoring and investment: Evidence from the changing structure of Japanese corporate banking relationships. In R. G. Hubbard (Ed.), Asymmetric information, corporate finance and investment (105-126). Chicago: The University of Chicago Press. Kang, J.K., Shivdasani, A., & Yamada, T. (2000). The effect of bank relations on investment decisions: An investigation of Japanese takeover bids. Journal of Finance, 55 (5), 21972218.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

108

V. Azofra-Palenzuela, F. J. López-Iturriaga and F. A. Tejerina-Gaite

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3 (4), 305-360. La Porta, R., López De Silanes, F., & Shleifer, A. (1999). Corporate ownership around the world. Journal of Finance, 54 (2), 471-517. La Porta, R., López De Silanes, F., Shleifer, A., & Vishny, R. (1997). Legal determinants of external finance. Journal of Finance, 52 (3), 1131-1150. La Porta, R., López De Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of Political Economy. 106, 1113-1155. La Porta, R., López De Silanes, F., Shleifer, A., & Vishny, R. (2000). Investor protection and corporate governance. Journal of Financial Economics, 58, 3-27. La Porta, R., López De Silanes, F., Shleifer, A., & Vishny, R. (2002). Investor protection and corporate valuation. Journal of Finance, 57, 1147-1170. Maury, B., & Pajuste, A. (2005). Multiple large shareholders and firm value. Journal of Banking & Finance, 29, 1813-1834. Morck, R., Nakamura, M., & Shivdasani, A. (2000). Banks, ownership structure, and firm value in Japan. Journal of Business, 73 (4), 539-567. Morck, R., Wolfenzon, D., & Yeung, B. (2005). Corporate governance, economic entrenchment, and growth. Journal of Economic Literature, 43, 655-720. Myers, S. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5, 147-175. Pedersen, T., & Thomsen, S. (2003). Ownership structure and value of the largest European firms: The importance of owner identity. Journal of Management & Governance, 7, 2755. Petersen, M., & Rajan, R. (1994). The benefits of lending relationships: Evidence from small business data. Journal of Finance, 49, 3-37. Prowse, S. (1990). Institutional investment paterns and corporate financial behaviour in the United States and Japan. Journal of Financial Economics, 27, 43-66. Shleifer, A., & Vishny, R.W. (1997). A survey of corporate governance. Journal of Finance, 52 (2), 737-782. Schmukler, S., & Vesperoni, E. (2001). Firms’ financing choices in bank-based and marketbased economies. In A. Demirgüç-Kunt, & R. Levine (Eds.), Financial structure and economic growth: A cross-country comparison of banks, markets, and development (347375). Cambridge: MIT Press. Schröder, U., & Schrader, A. (1998). The changing role of banks and corporate governance in Germany: Evolution towards the market?. In S. W. Black & M. Moersch (Eds.) Competition and convergence in financial markets. The German and Anglo-American models (17-33). Amsterdam: Elsevier. Stock, J., & Watson, M. (2003). Introduction to Econometrics. Boston, Mass.: Addison Wesley. Tejerina, F. (2006). The influence of banks on the corporate governance of non-financial firms (La influencia de las entidades bancarias en el gobierno de las empresas no financieras). PhD dissertation. Valladolid (Spain): University of Valladolid.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 5

HOW MUCH DO TRADE AND FINANCIAL LINKAGES MATTER FOR BUSINESS CYCLE SYNCHRONIZATION? Alicia García Herrero and Juan M. Ruiz1 Department of International Economics Bank of Spain, Madrid, Spain

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT We estimate a system of equations to analyze whether trade and financial linkages influence business cycle synchronization directly or indirectly. We use a small, open economy (Spain) as benchmark for the results, instead of the US as generally done in the literature. Neither trade nor financial linkages are found significant in directly influencing business cycle synchronization. Only the similarity in productive structure appears to foster economic integration, after controlling for common policies. Trade linkages are found to increase output synchronization indirectly, by contributing to the similarity of productive structures, which might point to the prevalence of intra-industry trade. The positive influence of financial linkages on output synchronization is even more indirect, by fostering trade integration and, thereby, a more similar productive structure. The net effects of both trade and financial linkages on business cycle synchronization are found statistically significant, but economically very small.

Keywords: business cycle synchronization, trade linkages, financial linkages, productive structure, integration. JEL classification: E32, F41, F12, E44.

1

Mailing Address: Bank of Spain, Dept. of International Economics (ERI), Alcalá 48, 28014 Madrid, Spain. Authors’ e-mail addresses are alicia.garcia-herrero and jruiz (please add @bde.es at the end to complete the address). We thank Andrew Rose and participants at the 6th ETSG conference for comments. The opinions expressed herein are those of the authors and not necessarily those of the Bank of Spain. Updated versions of this paper can be found at http://www.eco.uc3m.es/jruiz/research.htm

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

110

Alicia García Herrero and Juan M. Ruiz

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1. INTRODUCTION The last few years have witnessed increasing economic globalization stemming from a very rapid growth in trade and financial linkages, among other factors. At least at first sight, one would be tempted to think that tighter trade and financial linkages contribute to the synchronization of business cycles. However, there is neither a clear a priori in the theoretical literature nor a consensus in the empirical work. In fact, they generate both demand and supply reactions, which may counteract each other. In addition, it is not even clear whether business cycle synchronization has increased over time. It very much depends on how synchronization is measured and which countries are considered. The issue is relevant for several reasons. First, if business cycles are more synchronized, the transmission of shocks across countries will be stronger and faster. This could be an important rationale in favor of international policy coordination. Second, business cycles synchronization has profound implications for the design and functioning of common currency areas. Third, if the business cycle in a country is mainly driven by external factors, such as trade and financial linkages, domestic policies aimed at economic stabilization are bound to have a smaller impact. In the same vein, if trade linkages lead to business cycle synchronization, external demand will not manage to dampen economic fluctuations, but quite the opposite. This implies that exchange rate policy will be unlikely to play an important role in boosting demand at times of low economic activity. This paper contributes to the empirical literature mainly in two ways. First, most of the existing studies analyze the issue estimating a reduced-form equation. However, there are a number of interrelations between trade linkages, financial integration and business cycle synchronization, which need to be taken into account so that the results are meaningful. We, therefore, use a system of equations to analyze the issue. Second, many studies suffer from the lack of bilateral data to measure financial linkages and use aggregate financial stocks or flows. This, which measures financial integration with the rest of the world, can hardly explain business cycle co-movements between two countries. Those studies which use bilateral data generally take the US or a group of big economies as a benchmark to measure business cycle synchronization. Such a large economy, or area, influences other countries through many channels other than trade and financial linkages, which is bound to bias the estimated coefficients. To minimize this problem, we use a small open economy, namely Spain, as a benchmark. From our empirical exercise, we obtain several conclusions: First, trade or financial linkages only influence the synchronization of business cycles through their effect on the similarity of economic structure. Second, the synchronization of output increases as economic structures become more similar —suggesting the prevalence of sectoral shocks in the last 15 years—, and as macroeconomic policies become more synchronized. Third, more trade integration increases the similarity of productive structures (which might point to intraindustry trade), and thus leads to higher business cycle synchronization. The total effect of trade integration on the similarity of productive structures turns out to be positive, but economically small. Fourth, the net effect of financial linkages on output synchronization is also indirect, positive, and very small: its fostering of trade linkages is reflected in its positive effect on the similarity of productive structures, and thus on the correlation of cycles.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter…

111

Perhaps the more important conclusion of the exercise, however, is our finding that, even though these indirect effects of trade and financial integration over business cycle synchronization are statistically significant, they are not very relevant economically. The effect of a similar productive structure or synchronized macro policies seems economically more relevant in influencing the synchronization of cycles. The rest of the paper is organized as follows: the next section reviews recent literature on the relationship between trade and financial integration and business cycle synchronization; section 3 outlines the main theoretical predictions and the estimation strategy; section 4 presents the empirical results and section 5 concludes.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2. RELATED LITERATURE Although the synchronization of business cycles has been extensively analyzed in the literature, there is no clear picture of whether it has increased over time, even less so of its determinants. The conflicting evidence on the trend of synchronization over time may be attributed to the country coverage, the sample period and/or the econometric technique applied. On the one hand, Helbling and Bayoumi (2003) find decreasing synchronization between the US and rest of G-7 countries, Heathcote and Perri (2003a,b) report a similar result between the US and an aggregate of Europe, Japan and Canada. On the other hand, Kose et al (2003b) show an increasing co-movement between individual advanced countries and world (G-7) aggregates. With a broader perspective, Bordo and Helbling (2003) find increased synchronization over the last 125 years for 16 industrial countries. In the same vein, using dynamic factor models, Stock and Watson (2003),2 Helbling and Bayoumi (2003) and Lumsdaine and Prasad (2003) show strong evidence of a common factor driving business cycles in advanced countries. However, with a similar methodology but for a sample of sixty countries, Kose, Otrok and Whiteman (2003) find that the common component (the so-called “world factor”) is less important in developing countries. There are also large differences in how synchronization is measured. Kose et al (2003b) use correlations of output and consumption of countries with respect to aggregate consumption and output of G-7 countries. They complement it with dynamic factor models to look for common components and assess whether the importance of the common component has increased over time, signaling a stronger synchronization. Heathcote and Perri (2003b) split the sample in two equal-length periods and measure cross-regional correlations of the log-difference of US GDP with that of an aggregate of Europe, Japan and Canada. They also propose and use a measure of correlation that corrects for the existence of high conditional volatility, based on Loretan and English (2000). Helbling and Bayoumi (2003) employ various indicators of synchronization, including a binary indicator of expansions and recessions; correlation coefficients and detrended series.3 They finally use dynamic factor 2

In particular, they find that find that this common component has become more important to explain G-7 business cycles after 1984 than between 1960 and 1983 3 Detrending is done using Baxter and King (1999) band-pass filter to eliminate low- and high-frequency components to keep business cycle components defined as those between 6 and 32 quarters. An alternative method used is log first differences.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

112

Alicia García Herrero and Juan M. Ruiz

models to assess what is the role of common components on output synchronization. Finally, Imbs (2004b) measures synchronization by using cross-country correlations of band-pass series of quarterly GDP over the last 20 years. Moving to the potential channels of synchronization we focus on this study, namely trade and financial linkages, neither the theoretical nor the empirical literature offer a definitive answer on their impact on synchronization. Regarding trade, Kose and Yi (2001) suggest that higher trade integration might lead to more or less synchronization of cycles, depending on the nature of trade and the type of shocks. Countries will become more synchronized if there is an increase of intra-industry trade and industry-specific shocks are the main drivers of business cycles. However, if there is more inter-industry trade, then industry-specific shocks would reduce the co-movement of output in both countries. Empirical studies find that higher trade integration increases cross-country output correlations, especially among advanced economies [Frankel and Rose (1998), Clark and van Wincoop (2001), Imbs (2004a, 2004b)], possibly reflecting increased intra-industry trade rather than inter-industry trade. Measures of trade linkages also differ across studies. Some of the earlier studies used aggregate measures of trade openness (i.e., trade integration instead of trade linkages between two countries). This is obviously less appropriate to investigate the determinants of business cycle synchronization between two countries. As for bilateral trade relations, some authors have used de jure measures namely restrictions to trade, such as import duties [IMF WEO (2002)]. The most common de facto measure is the sum of exports and imports between two countries, divided by GDP [IMF WEO (2002), Imbs (2004b)], or over the ratio of the product of GDPs divided by world output, to make it independent of country size (Clark and van Wincoop (2001)). Another alternative, non-standard measure is the dispersion between two countries’ goods prices [IMF WEO (2002)]. More details on these measures will be offered in Section 3, since we shall be using them in our study. As for financial linkages, there is some evidence of a positive relationship between financial integration and business cycle co-movements both in output and consumption in the case of advanced economies (Imbs 2004a,b) but not so for developing economies (Kose, Prasad and Terrones (2003b)). In addition, these results are challenged by potential reverse causality. In fact, Heathcote and Perri (2003b) propose that higher financial integration may arise as a result of less correlated real shocks, since the gains from asset trade are bigger. By fostering financial flows, financial integration would dampen GDP correlations more than the reduction implied by the lower correlation of shocks The measures of financial linkages also differ.4 As for trade linkages, earlier studies used aggregate measures rather than bilateral ones (i.e., trade integration instead of linkages). This is even more the case than for trade because of the difficulties in finding bilateral data of financial transactions. Among the aggregate measures, several authors have employed aggregate de jure indicators, namely a global index of capital account restrictions from the IMF Annual Report on Exchange Arrangements and Exchange Restrictions5. Imbs (2004b) uses the sum of these indices in two countries as a bilateral de jure measure of their financial linkages. Another de jure measure of aggregate financial integration is an index of stock market liberalization (Prasad et al (2003)). Among de facto measures, there are quantity and price measures, most of which are aggregate and not bilateral. The most comprehensive 4 5

Edison et al (2002) and Prasad et al (2003) provide surveys of different measures of financial integration. Prasad et al. (2003), IMF (2001b) and IMF (2002).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

How Much do Trade and Financial Linkages Matter…

113

aggregate quantity measure is the sum of stocks of external assets and liabilities of foreign direct investment and portfolio investment6 (IMF WEO (2002), IMF WEO (2001b) , Prasad et al. (2003)7 and Heathcote and Perri (2003b)8).9 Other aggregate measures are total capital flows as a share of GDP, but it suffers from large volatility (Prasad et al (2003)). Others are proxies of risk sharing obtained regressing GDP on disposable income (Kalemli-Ozcan et al (2003)) 10 A bilateral quantity measure (i.e., of financial linkages) is the sum of gross asset positions between two countries, but this is only readily available for the US against the rest of the world (Imbs, 2004b)). An alternative source of bilateral data are equity transaction flows (Portes and Rey (2003)) although it is only available for a few countries, and equity holdings from the Coordinated Portfolio Investment Survey conducted by the IMF in 1997 and 2001. The latter also has geographical limitations, as well as underreporting and a poor collection method (Lane and Milesi-Ferretti (2004)). There are also bilateral price measures, such as differences from covered interest rate parity, but with very limited data availability (Frankel, 1992), and asset price arbitrage (IMF, 2001) based on rolling correlations of stock and bond prices. The latter, though, suffers from potential reverse causality. The methodology generally used in the literature to test for the relevance of trade and financial channels is the estimation of a single equation. The fact that there may be indirect effects going in opposite directions might account for the generally small impact found in studies using single equation regressions. To our knowledge, Imbs (2004b) is the only one who estimates a system of simultaneous equations to take into account direct and indirect effects on synchronization but there are a number of differences between his analysis and ours. First, he does not consider the possible two-way relationship between financial linkages and trade linkages (Aizenman and Noy (2001) or the incentives for financial linkages that might stem from a low correlation of business cycles Heathcote and Perri (2003b). Second, he works with a limited set of 24 countries, with a very high proportion of rich economies in the sample. Having mostly developed countries in the sample might induce a selection bias in the results, as developing countries are likely to be also very poorly linked commercially and financially. Third, his estimated coefficients might be picking up some other channels through which big economies affect other countries’ business cycles. Finally, Imbs (2004b) includes output correlations from the 80s and 90s. However, the existence of a number global common shocks in the 80s (although less prevalent than in the 70s) makes it difficult to identify the source of output co-movements.

6

Bank lending is not included. Prasad et al (2003) also separate financial flows into its main constituents: FDI, bank loans and portfolio flows. 8 Heathcote and Perri (2003b) use, for assets, the sum of FDI plus the equity part of portfolio investment. They also test for separate measures (FDI on one side and equity holdings on the other). 9 The original indices were also constructed by Lane and Milesi-Ferretti (2001) from the accumulation of financial flows and with some valuation adjustments. 10 The idea is that with perfect risk sharing, disposable income should be unrelated to GDP, whereas in the absence of risk sharing, they should be closely related. Kalemli-Ozcan et al (2003) also use measures of consumption risk sharing. Imbs (2004b) uses pair wise sums of this estimate of risk sharing as measure of bilateral financial integration 7

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

114

Alicia García Herrero and Juan M. Ruiz

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

3. ESTIMATION We assess empirically whether trade and financial linkages foster or hinder the synchronization of business cycles, while taking into account other potentially relevant determinants of synchronization. Both in the case of trade and financial linkages, there are arguments for and against their fostering synchronization. Trade linkages should, in principle, lead to more synchronized business cycles as higher investment or consumption in one country implies an increase in imports from trade partners. However, depending on the patterns of trade, larger commercial linkages might increase or decrease synchronization. If both countries develop intra-industry trade, then output should be more synchronized even if shocks are mostly sector-specific. However, trade may also foster specialization in production, thereby reducing business cycle synchronization if shocks are mostly industry-specific. Financial linkages could strengthen or weaken the co-movement of output, depending on its effect on specialization and the nature of shocks. On the one hand, there may be more synchronization if financial linkages allow for spillovers from demand shocks. On the other, there should be less synchronization if financial links lead to the reallocation of capital according to comparative advantage. This should contribute to specialization in production, fostering inter-industry instead of intra-industry trade. The description of the way in which trade and financial linkages may affect synchronization is clearly multi-directional. This implies potential endogeneity problems. Moreover, the different directions of indirect effects might offset each other and lead to very small net effects if we just try to correct the endogeneity problem using instrumental variables in the estimation. We shall, thus, use a system of equations to deal with this issue. We also consider other possible sources of synchronization, namely the convergence of economic policies, which we approximate with the volatility of exchange rates and the differences in inflation rates. Finally, we use bilateral data to account for trade and financial linkages. Data on financial linkages is particularly difficult to find except for the US, which obliges us to focus on one aspect of financial integration for which bilateral data is available, namely FDI. We choose a small open economy as a benchmark country, Spain. This is unlikely to have other channels of influence on other countries, limiting the problem of omitted variables in previous studies with de facto bilateral data of financial linkages.

4. ESTIMATION STRATEGY AND DATA ISSUES The direct and indirect channels through which trade and financial linkages may affect business cycle synchronization can only be taken into account through a system of equations. We, therefore, estimate a system of four equations, in which we test for the determinants of business cycle synchronization (eq. 1), those of trade and financial linkages (eqs. 2 and 3, respectively) and those of the similarity in productive structure (eq. 4). As previously explained, the latter is a key variable both in the cases of trade linkages and also business cycle synchronization.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter… ρi,t = α0 + α1 Ti,t + α2 Si,t + α3 Fi,t + Controls(ρ) + ερ Τi,t

= β0 + β1 Si,t +β2 Fi,t + Controls(T) + εT

(Eq. 2)

Fi,t = δ0 + δ1 ρi,t + δ2 Ti,t + Controls(F) + εF

(Eq. 3)

115

(Eq. 1)

Si,t = γ0 + γ1 Ti,t + γ2 Fi,t + Controls(S) + εS (Eq. 4) where:

ρi,t is the correlation between Spain’s business cycle and country i at time t.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Ti,t is bilateral trade integration between Spain and country i at time t. In principle, the expected sign of its coefficient in Eq. 1 is positive but it could be dampened or even reversed if trade contributed to a high degree of specialization. Si,t is an index of the similarity of economic structure between Spain and country i. This should be closely linked to the share of intra versus inter-industry trade. The more similar the economic structure (i.e., the lower the degree of specialization between two countries), a tighter business cycle synchronization is expected. Fi,t is bilateral financial integration with country i. As for trade, the expected sign of its coefficient in Eq. 1 is ambiguous for the reasons previously mentioned. Although optimally one should conduct a panel data regression with the structure outlined above, given the poor quality of the geographical disaggregation of financial data prior to 1997, we choose to conduct a cross section regression using data for the period 19972003. We, therefore, drop the time subindex for all variables considered. Among several possibilities in the literature, we choose to measure business cycle synchronization (ρI ) as the Pearson correlation of the log difference of annual GDP.11 For trade linkages Ti between Spain and country i , we use the standard bilateral de facto measure, as in Frankel and Rose (1998) as a benchmark, namely the sum of bilateral imports and exports between Spain (ESP) and country i divided by the sum of their respective GDPs. Denoting this measure by T

T 1ESP ,i =

1 ESP ,i

, we have:

X ESP ,i ,t + M ESP ,i ,t 1 ∑ T t GDPESP ,t + GDPi ,t

where XESP,i,t are exports from Spain to country i at time t, MESP,i,t are imports to Spain from country i at time t, and GDPi,t is country i’s GDP at time t.12 Note that we are taking a time average (over the period under study) of this measure.

11

GDP is measured at purchasing power parity and was obtained from the IMF’s World Economic Outlook database. 12 Data for exports and imports is obtained from the IMF’s Direction of Trade Statistics. Data for GDP (at purchasing power parity) is obtained from the IMF’s World Economic Outlook database. All data are annual. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Alicia García Herrero and Juan M. Ruiz

116

As a robustness exercise, we also consider Clark and van Wincoop (2001)’s measure, which is independent of country size (and dependent only on trade barriers). Denoting this alternative measure T

T 2 ESP ,i =

2 ESP ,i

we have:

1 ⎛ X ESP ,i ,t + M ESP ,i ,t ∑⎜ T t ⎝⎜ GDPESP ,t × GDPi ,t

⎞ ⎟⎟ GDPWorld ,t ⎠

2

Taking into account Deardorff (1998)’s, who shows that this measure is equal to one if preferences are homothetic and there are no trade barriers, we not that if we use T

2 ESP ,i

in the

regressions, we can drop GDPWorld,t from the computation of the index. This would just be a scaling factor which will multiply the coefficient of T

2 ESP ,i

but will not change its sign or

significance. All the results presented here are robust to measuring trade linkages in this alternative way. In order to measure financial integration through a bilateral de facto measure, we initially used bilateral FDI flows from and to Spain from the OECD. Although data on stocks of FDI would have been a better indicator, it was not available for Spain. We measure financial integration by taking the sum of inward and outward FDI flows and computing a time average over the period of study:

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

FESP ,i =

1 ∑ I ESP,i,t + I i,ESP,t T t

where Iijt represents financial flows from country i to country j (ESP denotes Spain) at time t. The similarity in productive structure can be measured in several alternative ways. All of them are based on data of shares of each productive sector, and differ in the depth of disaggregation of economic activities and whether or not they concentrate on manufactures (at greater disaggregation13) or on all sectors (at lower disaggregation14). Let sn,i,t be the share of industry n in country i at time t. Then the first measure of economic similarity can be expressed as

S 1ESP ,i = −

N 1 ∑∑ sn,ESP ,t − sn,i ,t T t n =1

where N is the number of sectors. Note that S

1 ESP ,i

represents the time average of

discrepancies in economic structures, as in Imbs (2004b).15 S

1 ESP ,i

might take values

13

Typically, 2- or 3-digit ISIC classification groups. At 1-digit ISIC classification groups. 15 We include a minus sign in front of the definition of structure similarity so that a higher value of S implies more similarity between the productive structures in both countries. This of course only changes the sign of its associated estimated parameter, but neither its size nor its significance. 14

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter…

117

between 0 for identical structures and –2 for disjoint productive structures. Therefore higher values for S

1 ESP ,i

imply more similarity between the Spanish productive structure and that of

country i. Clark and van Wincoop (2001) use a similar concept but taking time averages of structures before computing distances in shares.16 N

1 n =1 T

S 2 ESP ,i = − ∑

∑s

n , ESP ,t

t

− ∑ sn ,i ,t t

Industry shares sn,i,t can be measured using a number of different indicators. The three main indicators are shares in total employment, shares of value added, or shares of

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

production. All the results presented in the next section use the definition S

1 ESP ,i

described

above applied to shares of value added, although the results are robust to using other definitions or data on employment or production, as they are highly correlated. We use data for the industrial sector at the two-digit ISIC level from UNIDO.17 We also use a number of controls in the regressions as suggested by previous work on each subject. One potential source of business cycle synchronization is the similarity of macroeconomic policies and the similarity of productive structures. We therefore include a number of variables to approximate this effect, such as the volatility of the bilateral exchange rate, the average inflation differential and a dummy variable to account for use of the euro as official currency. In the case of trade linkages, a number of studies have suggested that gravity variables play an important role in explaining the importance of trade between two countries. We therefore include distance, sum of land areas, product of populations, product of GDPs, and two dummy variables to account for sovereign access to the sea and a common main language.18 Recent studies19 have suggested that gravity variables also explain bilateral financial linkages. We, thus, include distance, time difference between main financial centers, common language and the sum of per capita GDPs.20 This last variable tries to capture the idea that richer countries tend to generate more financial flows (both inward and outward). Surely the most difficult variable to control is the similarity of productive structure. Following on Imbs and Wacziarg (2003) we use the pair-wise difference of per capita GDPs, based on the idea that rich countries tend to be more diversified and thus possibly more similar, whereas poorer countries tend to be more specialized.

16

Clark and van Wincoop (2001) use a similar concept but taking time averages of structures before computing distances in shares. Imbs (2001) uses the Pearson correlation coefficient between sectoral shares sn,i,t. 17 We could in principle use data at the three-digit ISIC level and increase the desegregations of activities. However, some countries in the sample do not report data at that level of desegregations, and therefore we opted for a lower level of desegregations in order to increase the sample size. 18 Some studies include, instead of common language, a dummy variable capturing past colonial relationship. In the case of Spain both variables coincide. 19 See, for example, Portes and Rey (2003). 20 As the effect of distance on trade and financial integration might not be linear, but stronger for shorter distances (in other words, an increase in distance reduces trade and financial integration, but at a diminishing rate) we also try the log of distance and time differences, instead of its levels.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Alicia García Herrero and Juan M. Ruiz

118

Results As a preliminary step we show some stylized facts of the main variables of interest in this study: business cycle synchronization, trade and FDI linkages. The degree of bilateral business cycle synchronization between Spain and EU countries increased substantially from 1960 to 1995 (figure 1). Since then, it has fallen somewhat and now hovers at 0.6 (in terms of Pearson correlation coefficient of annual growth rates). Bilateral synchronization between Spain and G7 countries also rose fast from 1970 to 1976 but then fell again. Since Spain’s entry in EU in 1986, it has risen at a slower pace than synchronization with EU countries. Business cycles in Spain and in Latin American countries move in opposite directions since the late 1980s. All in all the period of closer synchronization between Spain and other countries was from 1975 to 1985.

Spain: GDP synchronization (ten-year rolling correlation of growth rates) Pearson correlation coefficient

EU

0.8 0.6 0.4 0.2

G-7

0 -0.2 -0.4 -0.6

LATAM-7

-0.8

2000

1995

1990

1985

1980

1975

1970

1965

-1 1960

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1

Ten-years ending in:

EU (14 countries) and G-7 exclude Germany before 1970. LATAM-7: Argentina, Brazil, Mexico, Chile, Colombia, Peru and venezuela. Source: Penn World Tables 6.1 and author's calculations.

Figure 1. Evolution of GDP synchronization between Spain and selected regions.

Trade linkages between Spain and EU countries started to rise already ten years before Spain’s entry into EU but since then the increase has been exponential (Figure 2). In fact the sum of imports from and exports to EU countries has reached 0.002% of those countries’ combined GDP. Trade linkages with G7 countries began to grow later, in the mid 1980s and at a much lower pace, reaching about 0.0007% of their combined GDP as a sum of imports and exports. Trade linkages with Latin American countries haven remained relatively small throughout the period. Spain started to have FDI linkages with EU and G7 countries in the mid-1980s, which increased enormously in the mid-1990s (Figure 3). FDI linkages with Latin American countries also rose then but at a lower pace. In 2000, there was a sharp fall of FDI linkages with all countries but it has recovered again with Latin American countries in the last few

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter…

119

years. Still the size of these FDI linkages is smaller than that with EU and, to a lesser extent, G7 countries. Turning to the estimation of the system of four equations, we first report the results of the estimation of each equation separately, using OLS. Since there are good reasons to suspect endogeneity problems, we complement the estimation of equation 1 (the main equation of interest to us) with the use of suitable instruments for trade and financial linkages (T and F) and similarity of structure S. In order to disentangle the direct and indirect effects of trade and financial linkages on business cycle synchronization, we finally turn to a joint estimation of the whole system of four equations, using three-stage least squares (3SLS).

Spain: Trade linkages 2.5

2.0

EU 1.5

1.0

G-7 0.5

LATAM-7

2000

1995

1990

1985

1980

1975

1970

1965

0.0 1960

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Thousandths of a percentage point

(sum of imports and exports over sum of GDPs)

EU (14 countries) and G-7 exclude Germany before 1970. LATAM-7: Argentina, Brazil, Mexico, Chile, Colombia, Peru and venezuela. Source: IMF Direction of Trade Statistics, Penn World Tables 6.1 and author's calculations.

Figure 2. Evolution of trade linkages between Spain and selected regions.

As regards the determinants of business cycle synchronization, estimated by a single equation (equation 1), trade integration seems significant in explaining the correlation of business cycles (Table 1), although once we control for common policies (the volatility of exchange rates seems particularly significant), this effect vanishes. In these OLS estimations for equation 1, neither financial linkages nor the similarity of productive structure appear significant, However, the endogeneity of trade (T) and financial linkages (F) (measured with FDI only), and the similarity of the productive structure (S) might lead to highly biased coefficients. This problem is tackled later by the use of IV estimation as reported in the lower half of table 1. Before turning to the estimation of equation 1 using instrumental variables, we turn to the OLS estimation of equations 2 to 4.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Alicia García Herrero and Juan M. Ruiz

120

Spain: FDI linkages (sum of FDI inflows and outflows from/to selected regions) 60000

Millions of Euros

50000

40000

30000

20000

EU

G-7

10000

LATAM

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

0

LATAM: Includes Caribbean countries. Source: OECD and author's calculations.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Figure 3. Evolution of FDI linkages between Spain and selected regions.

The estimation of trade linkages (Eq 2) shows that financial linkages, approximated by FDI, affect trade positively (β2>0) and significantly (Table 2). Among the variables included to account for a gravity model, distance to the main city appears as highly significant and with the correct sign. The coefficient of the similarity in productive structure (β1) is not significant. This could be due to endogeneity problems or because of conflicting effects, depending on whether intra or interindustry trade is more prevalent. The coefficient on the product of average GDPs should have a positive sign, although in specification V and VI it is significantly negative. Again this may point to a bias due to the endogeneity of FDI integration, as the problem only appears when F is included in the regression. Financial linkages, estimated by OLS on equation 3 seem to be determined by trade linkages and distance. The only anomaly is in the sign of the time difference between financial centers, which might again point towards and endogeneity problem. The significance of lagged trade linkages might point out to a global effect of trade integration on financial integration, as described by Aizenman and Noy (2004). An alternative and simpler explanation could be the high correlation of trade integration in the 80s and 90s. An OLS regression for the similarity in productive structure (Eq. 4) described in Table 4 points to the difference in percapita GDP as a good explanatory variable, as suggested by the theory. The similarity in productive structure seems to be positively influenced by trade linkages. In other words, trade linkages promote a similar economic structure. Again, all these coefficients might suffer from important biases stemming from the endogeneity of T and F. Given the biases introduced in the estimation of equation 1 due to the endogeneity of T, F, and S, we proceed to estimate equation 1 using appropriate instruments for those

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

How Much do Trade and Financial Linkages Matter…

121

variables.21 We report estimates of instrumental-variable regressions with alternative specifications of equation 1 in the lower half of table 1. The last three regressions include our controls for common policies. Note that, because of the availability of instruments, the number of observations drop to 43. Although coefficients change slightly from the top half of table 1, overall we still see no significant contribution of trade or financial linkages to explain business cycle synchronization, once we control for our proxies for common policies. Estimation of equation 1 by instrumental variables, however, still pools together the direct and indirect effects of trade and financial linkages over business cycles synchronization, for example through their effect over the convergence of productive structures between Spain and the other countries in the sample. If indirect effects through different channels point to opposite directions, the net effect might become small and thus contribute to its statistical insignificance. We therefore conduct a three-stage least-squares regression on the whole system of four equations. Estimating the system of four equations, the results change to a large extent (Table 5a). The most relevant, for the purpose of our study, is that only the similarity in productive structure (S) is found significant in determining output synchronization, after controlling for the effect of common policies. In this regard, exchange rate volatility is found significant while differences in inflation are not. Trade linkages influence output synchronization only indirectly through their effect on the similarity of productive structure. The direct effect of trade on the similarity of productive structures is positive (γ1>0): stronger trade links tend to make productive structures more similar, which might point to intra-industry trade. On the other hand, more trade promotes stronger financial links (δ2>0). The total effect of trade on business cycle synchronization is still positive (γ1α2 >0), in line with previous studies that do not separate the two effects. The influence of financial linkages on output synchronization is also indirect, through its effect on trade. Since financial integration seems to foster trade integration (β2>0), this means a positive indirect effect on the similarity of productive structures and thus on the synchronization of cycles. (α2 γ1 β2>0). The important influence of a similar economic structure on business cycle synchronization is in line with Imbs (2004b) but the relevance of trade and financial linkages is smaller in our case, since he also finds direct effects. This difference might be related to the fact that we use a small open economy as a benchmark, and a wider set of countries, as opposed to Imbs (2004b). The latter may have biased upward the coefficients, as there are other channels of influence of the US economy which are not considered. Another reason, as regard financial linkages, might be the limitation of our data. FDI flows are only one type of financial linkages considered, albeit an important one. There are also other findings from the system of equations, worth mentioning: (i) We did not find a reverse causality from business cycle synchronization to financial linkages, as argued by Heathcote and Perri (2003b); (ii) the model seems to confirm a double causality between trade and financial linkages; (iii) a similar productive structure, apart from contributing to higher output synchronization, also tends to foster trade. Such positive influence should be understood in terms of intra- more than inter-industry patterns of trade in line with the results by Kose and Yi (2001). 21

In order to instrument T, F and S, we use the same independent variables as those in tables 2 to 4.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Alicia García Herrero and Juan M. Ruiz

122

The relations that have been found significant in the system of equations can be summarized in the following diagram. Another important question concerns the economic relevance of the statistically significant effects found in the previous exercise. As described before, the total effect of trade on the synchronization of business cycles is given indirectly through an increase in the similarity of productive structures. Specifically, in our benchmark 3SLS regression in table 5a, the effect of trade linkages on our measure of comovement of output is equal to (γ1α2)= 10727.62, whereas the total effect of financial linkages is given by (α2 γ1 β2)= 3.81 x 10-5. This implies that increasing trade links by one standard deviation starting from its mean (see table 6), increases bilateral cross country correlation of GDP from 0.706 to 0.732. Increasing financial links by one standard deviation increases the correlation of output from 0.706 to 0.737. Common Sectoral Shocks

Financial integration

+ Trade Integration

Common Policy

+

+ + +

More Similar Productive Structure

+

Output Synchronization

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Figure 4. Channels leading to business cycle synchronization found in the empirical exercise.

This is hardly an economically meaningful change, and reflects that fact that business synchronization in the Spanish case for the last 15 years has been influenced more by common policies and presumably by common sectoral shocks. Performing the previous exercise with similarity of productive structure (S) and exchange rate volatility, we find that an increase in these variables by one standard deviation from its mean would imply a change in the degree of GDP correlation from 0.710 to 0.918 and 0.790, respectively, a much stronger effect than that of trade and financial links. A number of additional tests are conducted to test for the robustness of our results. First, we include an alternative hypothesis for the gravity models is that the effect of distance on trade and financial integration might not be linear, but stronger for shorter distances. In other words an increase in distance reduces trade and financial integration, but at a diminishing rate. This hypothesis is captured by including the log of distance and time differences, instead of its levels, and estimating with 3SLS as before. The gravity variables for trade and financial integration become more significant (Table 5b) than in the benchmark case. The significance of the variables of interest, and the channels of influence on business cycle synchronization does not change much. The exception is the bi-directional relationship between trade and the similarity of economic structure. This now becomes only one-way, with trade integration affecting the similarity of productive structure, but not vice-versa. A second robustness exercise aims at tackling the problem of the low number of observations (43), in the system of equations. We extend the number of observations by

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter…

123

imputing the value of zero to the observations where no data on FDI flows is available. The list of countries now included in the regression increases to 104.22 As can be seen from Table 8 in the appendix,23 this is a relatively safe assumption in many cases but not all24. The results are relatively similar to the extent that trade and financial linkages do not seem to affect business cycle synchronization directly but only indirectly through their effect on the similarity of productive structure (Table 5c). Still, there are a number of differences in the results worth mentioning. First, there is now a negative and significant effect from contemporaneous trade linkages to FDI linkages (Eq 3). However, the positive effect from previous trade integration is maintained. Second, the link from the similarity of productive structure to trade linkages also seems to be broken (Eq. 2). Third, FDI linkages appear significant in increasing the similarity of productive structure. This was not the case before, which implied an even more indirect impact of financial linkages on business cycle synchronization. The diagram in the appendix (figure 5) summarizes the relations that have been found significant in this case.

Common Sectoral Shocks

Financial integration

+

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Trade Integration

+

Common Policy

+

+ More Similar Productive Structure

+

Output Synchronization

Figure 5. Channels of effects found in the empirical exercise with the extended set of countries (104).

Finally, in order to control for global shocks, we also introduced a variable to approximate the similarity in the exposure of both economies to oil shocks. For each country, we measure net imports of oil as a percentage of GDP and average that percentage for the period 1990-2002. We then multiply that measure with the equivalent one for Spain, which is positive25. In principle, countries that are more dependent of oil should have a high and positive dependency ratio, whereas oil exporting countries have a highly negative indicator. A high and positive product of both indicators indicates countries that are affected by an oil shock in a similar way as Spain. A highly negative indicator represents countries that would benefit from an increase in the price of oil, as opposed to the Spanish economy. 22

23

24

Consistent with the inclusion of new observations in the estimation of the system of simultaneous equations, the table of cross correlations has been expanded (See Table 7b in Appendix). Correlation coefficients above 0.6 are highlighted.

The table highlights the 44 countries included in the original regression.

The main risk of introducing a bias lies in those countries in Latin America that are summarized in the OECD data, like Peru. 25 Details of the construction and sources used for this oil dependency index can be found in Appendix B. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

124

Alicia García Herrero and Juan M. Ruiz

We introduce this indicator as an explanatory variable for growth correlations. However, it turns out not to be statistically significant26 in any of the specifications tried (OLS, IV or 3SLS estimations). This result could be interpreted as confirmation that in the period of study (1990-2003) oil shocks were not an important factor driving global economic fluctuations, as they were in the 70s or, to a lesser extent, in the 80s.

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

6. CONCLUSIONS This paper assesses what is the role of trade and financial linkages in business cycle synchronization while considering a large number of interrelations between the relevant variables through a system of equations. This allows us to identify direct and indirect effects of trade and financial linkages on output co-movements. While there are number of possible endogeneity problems associated with trade and financial linkages as explanatory variables for output synchronization, in principle one could eliminate those biases by using suitable and readily available instruments. However, the reduced form IV estimates might appear small or not significant because, in theory, direct and indirect effects might run in opposite directions, cancelling each other. We, therefore, conducted the estimation of system of equations in order to separate direct and indirect effects of trade and financial linkages on output synchronization. This approach seems validated by our finding that only indirect effects (through their effect on the similarity of productive structure between the two countries) are significant. The other contribution of the paper is to take a small, open economy as benchmark of the analysis and not the US or a group of rich countries accounting for a big share of world GDP. Business cycle synchronization between small open economies should depend more on trade and financial linkages than on other factors, many of which cannot be explicitly included in the analysis. These have probably biased upward the estimation of the trade and financial coefficients in previous studies. Our finding of no direct influence of trade or financial linkages on cycle synchronization is even more interesting for a small open economy, such as Spain. In addition, the significance of indirect influence justifies the use of a system of equations, instead of a reduced form. Summarizing the results, we find that only the similarity in productive structure (S) is significant in determining output synchronization, after controlling for common policies (exchange rate volatility). Trade and financial linkages appear to increase output synchronization only indirectly, by fostering the specialization of productive structure. While trade and financial integration do lead to increased output synchronization, its indirect influence highlights that a precondition for this effect is the convergence of the productive structure of both countries. In particular, financial or trade liberalization without measures to allow the reallocation of productive resources inside a country might not lead to a correlation of business cycles. Another interesting policy conclusion is to weaken the idea that, with the increasing economic globalization, external demand both for goods and services, but also for financial assets, does not help boost the economy. 26

P-values for a test of significance of this variable are never lower than 0.88 in all specifications.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

APPENDIX A: TABLES

Table 1. Dependent Variable: Growth correlations with Spain, 1990-2003 ( ρ ) OLS Estimation Specification Number of Observations

Ia 162

Trade Linkages 1990-19991 (T) (

28270.24 ** 9326.31 )

FDI Linkages 1991-20002 (F)

IIa 50 (

17911.03 11349.81 )

0.0000373 ( 0.0000558 )

IIIa 49 (

16519.16 * 9885.22 )

IVa 126 (

21551.52 *** 8318.65 )

Va 152 (

0.0000334 ( 0.0000482 )

Similarity in Productive Structure 1980-20003 (S) (

-0.1234 0.2494 )

( Exchange rate volatility 1990-20034

-0.000219 *** 0.00008 )

-0.060645 ** ( 0.0308499 ) 0.07

0.05

VIIa 49 (

2173.28 11045.42 )

0.0000486 ( 0.0000421 ) -0.087102 ( 0.2140615 )

Average Inflation differencial 1990-2003

0.08

1282.891 11538.26 )

0.0102476 ( 0.0783445 ) 0.1048364 ( 0.1042206 )

0.05

(

0.0000558 ( 0.0000439 )

Member of Euro Area (1=yes)

Adjusted R2

14683.55 11181.21 )

VIa 50

0.21

0.087204 ( 0.0971558 )

0.0981344 ( 0.0932183 )

0.0000239 0.000305 )

0.0002579 ( 0.0003062 )

(

-0.183092 *** ( 0.0504493 ) 0.46

-0.169869 *** ( 0.0484815 ) 0.41

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 1. Continued 5

IV Estimation (Two-Stage Least-Squares) Specification Number of Observations

Ib 43

1

Trade Linkages 1990-1999 (T) (

15845.18 ** 6123.641 )

2

FDI Linkages 1991-2000 (F)

IIb 43 (

15396.28 * 7961.088 )

3.64E-06 ( 0.0000405 ) 3

Similarity in Productive Structure 1980-2000 (S)

IIIb 43 (

13571.05 8346.635 )

IVb 43 (

12904.9 * 6903.678 )

2

0.3346314 ( 0.3415994 )

0.07

0.03

0.06

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

5

Instruments used are the same as those used in the three-stage least-squares regression in tables 5a-c.

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

(

9515.291 9760.288 )

0.3226887 ( 0.3265886 )

4

0.89

11035.64 7568.933 )

VIb 43

9.45E-06 ( 0.0000379 )

Average Inflation differencial 1990-2003

Adjusted R

(

-6.28E-06 ( 0.0000426 )

Member of Euro Area (1=yes)

Exchange rate volatility 1990-2003

Vb 43

VIIb 43 (

8618.184 10202.96 )

-5.29E-06 ( 0.0000409 ) 0.4502342 ( 0.3216657 )

0.0290518 ( 0.0726864 )

0.034051 ( 0.0758788 )

(

0.0000563 ( 0.0002409 )

0.0000492 ( 0.0002442 )

2.76E-06 ( 0.0002569 )

-0.102627 ** ( 0.0428706 )

-0.102297 ** ( 0.0431826 )

-0.102971 ** ( 0.0450546 )

0.25

0.24

0.0136597 0.080493 )

0.18

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 2

Dependent Variable: Trade Linkages with Spain 1990-19991 (T) OLS Estimation Specification Number of Observations

I 164

II 50

FDI Linkages 1991-20002 (F) (

2.49E-09 *** 5.81E-10 )

III 49

( Spanish spoken (1=yes)

-2.33E-10 *** 5.50E-11 )

2.35E-09 *** 5.99E-10 )

(

(

3.47E-06 3.90E-06 )

(

(

-2.44E-10 ** 1.04E-10 )

(

-2.48E-10 ** 1.06E-10 )

(

-2.28E-10 *** 5.38E-11 )

VI 49

3.90E-09 *** 7.98E-10 )

(

3.71E-09 *** 8.08E-10 )

)

(

4.87E-06 3.78E-06 )

(

-1.57E-10 1.05E-10 )

(

-1.52E-10 1.06E-10 )

(

1.02E-07 5.85E-07 )

(

-4.21E-07 1.49E-06 )

(

-1.61E-07 1.54E-06 )

(

2.02E-07 5.66E-07 )

(

-1.03E-06 1.44E-06 )

(

-6.86E-07 1.47E-06 )

(

9.61E-07 ** 4.35E-07 )

(

1.61E-06 1.52E-06 )

(

2.14E-06 1.62E-06 )

(

7.94E-07 4.19E-07 )

(

1.74E-06 1.45E-06 )

(

2.49E-06 1.54E-06 )

(

-1.46E-13 1.03E-13 )

(

-1.19E-13 1.45E-13 )

(

-1.57E-13 1.47E-13 )

(

-3.38E-11 4.44E-11 )

(

7.93E-11 6.41E-11 )

(

7.08E-11 6.43E-11 )

(

-2.12E-24 ** 9.72E-25 )

(

-2.03E-24 ** 9.73E-25 )

Access to seacoast (1=yes) Sum of Land Areas (in km2) Product of populations (in billions) Product of average GDPs 1990-2003

( Adjusted R2

V 50

(

Similarity in Productive Structure 1980-20003 (S) Distance to main city (km)

IV 165

0.11

0.37

0.37

1.86E-24 *** 5.00E-25 ) 0.17

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

0.43

0.44

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 3

Dependent Variable: FDI Linkages with Spain 1991-20002 (F) OLS Estimation Specification Number of Observations

I 51

II 50

Trade Linkages 1990-19991 (T) (

9.70E+07 *** 2.73E+07 )

III 44 (

-5.73E+07 7.31E+07 )

Trade Linkages 1980-19891 (lagged T) (

IV 49 (

V 44

9.33E+07 *** 2.85E+07 )

5.17E+08 ** 2.27E+08 )

Similarity in Productive Structure 1980-20003 (S) (

551.822 805.832 )

VI 50 (

(

3.42E+08 *** 9.19E+07 )

(

436.485 824.528 )

( Spanish spoken (1=yes)

(

-66.499 420.718 )

(

-430.519 738.892 )

(

-0.088 * 0.053 )

(

-0.070 0.055 )

(

-0.089 * 0.054 )

(

-0.076 0.055 )

(

-0.088 0.054 )

(

-0.073 0.055 )

275.891 346.251 )

(

198.125 312.821 )

(

16.674 330.851 )

(

243.758 325.407 )

(

106.274 332.493 )

(

195.286 316.938 )

(

16.831 337.539 )

(

377.424 346.876 )

(

93.736 321.498 )

(

94.286 409.611 )

(

162.926 345.193 )

(

125.242 421.415 )

(

95.213 325.340 )

(

82.741 410.638 )

(

113.809 86.830 )

(

128.538 * 78.352 )

(

110.780 80.699 )

(

130.867 * 79.876 )

(

119.524 80.685 )

(

129.090 79.333 )

(

120.118 80.654 )

(

0.023 * 0.013 )

(

(

0.023 * 0.014 )

(

0.027 * 0.014 )

(

0.026 * 0.014 )

Absolute time difference to main financial centre Sum of percapita GDPs (average 1990-2003) ( 2

3.68E+08 *** 9.30E+07 )

( Access to seacoast (1=yes)

Adjusted R

-0.114 ** 0.059 )

9.83E+07 *** 2.89E+07 ) (

Growth correlations with Spain, 1990-2003 (ρ ) Distance to main city (km)

VI 44

0.042 *** 0.013 ) 0.20

(

0.026 ** 0.013 ) 0.37

0.40

0.026 ** 0.013 ) 0.35

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

0.40

0.35

0.40

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 4

Dependent Variable: Similarity in Productive Structure 1980-20003 (S) OLS Estimation Specification Number of Observations Trade Linkages 1990-19991 (T)

I 128

II 50

42218.00 *** ( 9.01E+03 )

FDI Linkages 1991-20002 (F) (

2.60E-05 2.29E-05 )

III 49

IV 128

5.47E+03 5.79E+03 )

(

24043.81 * 9293.399 )

(

0.0000113 2.84E-05 )

(

)

(

-0.000017 7.08E-06 )

(

-9.24E-06 5.83E-06 )

(

-8.10E-06 6.62E-06 )

(

6.94E-06 3.87E-06 )

(

-6.21E-07 2.88E-06 )

(

-5.03E-07 2.96E-06 )

Sum of percapita GDPs (average 1990-2003)

0.14

0.01

0.01

0.26

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

VI 49

(

Absolute difference of percapita GDPs (average 1990-2003)

Adjusted R2

V 50 ( 0.0000275 ( 0.0000255 )

0.02

2323.486 6288.534 )

VI 128 (

28199.85 *** 9079.603 )

(

-2.45E-05 *** 5.77E-06 )

0.0000208 ( 0.0000318 )

0.00

0.24

Alicia García Herrero and Juan M. Ruiz

130 Table 5a

Three-stage Least Square regression on the whole system of four equations 43 Observations Dependent Variable

Output Synchron. (ρ ) (Equation 1)

Trade Linkages 1990-19991 (T) (

Trade Linkages (T) (Equation 2)

7553.61 9082.60 )

FDI Linkages (F) (Equation 3) (

-1.44E+08 1.31E+08 )

Trade Linkages 1980-19891 (lagged T)

(

-2.27E-05 3.69E-05 )

(

3.55E-09 *** 1.22E-09 )

2

FDI Linkages 1981-1990 (Lagged F)

( Similarity in Productive Structure 1980-20003 (S) (

0.7018 *** 0.2826 )

(

0.000032 *** 9.77E-06 )

(

-2.85E-11 1.13E-10 )

(

6.03E-07 1.63E-06 )

(

3.08E-06 2.04E-06 )

Distance to main city (km) Spanish spoken (1=yes) Access to seacoast (1=yes)

Member of Euro Area (1=yes) (

0.0026 0.0706 )

(

-0.0001 0.0002 )

(

-0.0970 *** 0.0397 )

Average Inflation differencial 1990-2003 4

2

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Sum of Land Areas (in km ) (

-5.21E-13 ** 2.52E-13 )

(

8.55E-11 8.55E-11 )

(

-1.49E-24 1.19E-24 )

Product of populations (in billions) Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003)

7.00E-05 5.64E-05 )

-0.056278 ( 0.0497338 ) (

-144.9104 340.891 )

(

99.70097 72.85905 )

0.0250425 * ( 0.0153059 )

Absolute difference of percapita GDPs (average 1990-2003) 2

(

-607.2559 1407.631 )

Absolute time difference to main financial centre

Implicit R

15285.44 ** 7190.144 )

-0.000374 ( 0.0003418 )

Growth correlations with Spain, 1990-2003 (ρ )

Exchange rate volatility 1990-2003

(

8.34E+08 ** 3.61E+08 )

( FDI Linkages 1991-20002 (F)

Similarity in Prod. Struct. (S) (Equation 4)

( 0.16

0.00

-3.17E-07 5.73E-06 )

0.48

-0.04

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

How Much do Trade and Financial Linkages Matter…

131

Table 5b

Three-stage Least Square regression on the whole system of four equations 43 Observations Dependent Variable

Output Synchron. (ρ ) (Equation 1)

Trade Linkages 1990-19991 (T) (

Trade Linkages (T) (Equation 2)

2731.86 9691.41 )

FDI Linkages (F) (Equation 3) (

-1.04E+08 1.28E+08 )

Trade Linkages 1980-19891 (lagged T)

(

0.000024 0.000040 )

(

(

6725.705 7261.269 )

7.28E+08 ** 3.30E+08 )

( FDI Linkages 1991-20002 (F)

Similarity in Prod. Struct. (S) (Equation 4)

5.37E-09 *** 1.44E-09 )

0.0000136 ( 0.0000359 )

FDI Linkages 1981-19902 (Lagged F) Growth correlations with Spain, 1990-2003 (ρ ) Similarity in Productive Structure 1980-20003 (S) (

0.4816 ** 0.2426 )

-359.0764 1291.439 )

(

0.0000198 *** 7.28E-06 )

(

-3.98E-07 7.91E-07 )

(

-119.8954 168.7232 )

(

4.87E-07 1.72E-06 )

(

-73.63136 322.7569 )

(

2.45E-06 1.94E-06 )

Log of Distance to main city (km) Spanish spoken (1=yes) Access to seacoast (1=yes) Log of absolute time difference to main financial centre

(

Member of Euro Area (1=yes) (

0.0347 0.0707 )

(

0.0000 0.0002 )

Average Inflation differencial 1990-2003 Exchange rate volatility 1990-20034 (

73.23183 ** 32.78706 )

-0.0987 *** 0.0389 )

Sum of Land Areas (in km2) Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

(

(

-4.33E-13 * 2.41E-13 )

(

1.28E-10 * 7.21E-11 )

(

-2.05E-24 * 1.32E-24 )

Product of populations (in billions) Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003)

0.0300283 ** ( 0.0146192 )

Absolute difference of percapita GDPs (average 1990-2003) Implicit R2

( 0.31

0.27

-4.24E-06 6.06E-06 )

0.52

0.09

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

132

Alicia García Herrero and Juan M. Ruiz

Table 5c

Three-stage Least Square regression on the whole system of four equations 104 Observations Dependent Variable

Output Synchron. (ρ ) (Equation 1)

Trade Linkages 1990-19991 (T) (

Trade Linkages (T) (Equation 2)

-6733.33 12268.77 )

FDI Linkages (F) (Equation 3) (

-1.82E+08 ** 9.33E+07

Trade Linkages 1980-19891 (lagged T)

(

0.000062 0.000054 )

(

(

-4925.193 12525.26 )

8.62E+08 *** 2.51E+08 )

( FDI Linkages 1991-20002 (F)

Similarity in Prod. Struct. (S) (Equation 4)

7.55E-09 *** 1.16E-09 )

0.0002277 *** ( 0.0000637 )

FDI Linkages 1981-19902 (Lagged F) Growth correlations with Spain, 1990-2003 (ρ ) Similarity in Productive Structure 1980-20003 (S) (

0.2075 ** 0.1019 )

1.45E-06 1.69E-06 )

(

-1.92E-07 4.43E-07 )

(

-32.50794 85.5454 )

(

-2.81E-07 6.77E-07 )

(

-49.95582 115.1807 )

(

-7.38E-08 6.60E-07 ) (

26.34079 * 15.73312 )

Spanish spoken (1=yes) Access to seacoast (1=yes) Log of absolute time difference to main financial centre (

0.0738 0.0827 )

(

0.0006 *** 0.0002 )

(

-0.1461 *** 0.0378 )

Average Inflation differencial 1990-2003 Exchange rate volatility 1990-20034 Sum of Land Areas (in km2)

( Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

415.645 589.1573 )

(

Log of Distance to main city (km)

Member of Euro Area (1=yes)

(

Product of populations (in billions)

-1.96E-13 1.66E-13 )

(

1.80E-10 *** 5.98E-11 )

(

-3.90E-24 *** 1.08E-24 )

Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003)

0.0126983 * ( 0.0073866 )

Absolute difference of percapita GDPs (average 1990-2003) Implicit R2

( 0.19

0.34

-2.85E-05 *** 5.97E-06 )

0.43

0.30

Standard errors in parenthesis 1

Measured as the average over the period of the sum of bilateral exports plus imports over the sum of the respective GDPs

2

Measured as the average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. S may take values between -2 (disjoint structure) and 0 (identical structure)

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

* Significant at 10%, ** Significant at 5%, *** Significant at 1%

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 6

Summary Statistics Variable

No. Observ.

Mean

Std. Dev.

Min

Max

Coeff. of Variation

5%

Percentiles 50%

95%

Growth correlations with Spain, 1990-2003 (ρ )

177

0.9890

0.42

0.0604

0.8339

0.9628

Trade Linkages 1990-19991 (T)

165

0.00000085 0.00000242 0.00000000 0.00001900

2.84

0.00000000

0.00000012

0.00000301

Trade Linkages 1980-19891 (lagged T)

122

0.00000045 0.00000092 0.00000000 0.00000612

2.07

0.00000000

0.00000012

0.00000194

2

FDI Linkages 1991-2000 (F)

0.7063

0.2944

-0.3294

52

397.66

815.66

0.17

3554.15

2.05

0.34

29.44

2333.90

Similarity in Productive Structure 1980-20003 (S) Member of Euro Area (1=yes) Average Inflation differencial 1990-2003

142 199 163

-0.6636 0.080 85.357

0.2964 0.273 336.407

-1.4457 0.000 0.533

-0.1890 1.000 3320.130

0.45 3.39 3.94

-1.1706 0.000 1.561

-0.6534 0.000 5.711

-0.2550 1.000 489.304

Exchange rate volatility 1990-20034 Distance to main city (km) Log of distance to main city Spanish spoken (1=yes) Access to seacoast (1=yes) Absolute time difference to main financial center Log of time difference to financial center

183 199 199 199 199 199 199

0.568 6262 8.517 0.106 0.794 3 -0.49

0.887 3923 0.731 0.308 0.405 3.177945 3.31

0.003 494 6.203 0 0 0 -6.91

5.303 19589 9.883 1 1 1.20E+01 2.48

1.56 0.63 0.09 2.92 0.51 0.95 -6.73

0.075 1282 7.156 0 0 0 -6.91

0.200 6037 8.706 0 1 2 0.69

2.442 15374 9.640 1 1 10 2.30

Sum of Land Areas (in km2) Product of populations (in billions) Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003) Absolute difference of percapita GDPs

199 197 167 167 167

1182581 1145.52 1.E+17 23414 10192

1898689 4490.48 5.E+17 7469 4249

504784 0.70 1.E+14 15554 627

17600000 1.61 48145.25 3.92 5.E+18 3.42 50361 0.3189786 18802 0.4169212

505043 2.56 5.E+14 16493 2095

616872 222.89 1.E+16 20730 11072

3010592 4537.81 7.E+17 38921 14947

1

Average over the period of the sum of bilateral exports plus imports over the sum of GDPs

2

Average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. Higher values imply more similarity.

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

Table 7a

Log of time difference to financial center

1.000

0.251

0.569

0.642

1.000

Similarity in Productive Structure 1980-20003 (S) Member of Euro Area (1=yes) Average Inflation differencial 1990-2003

0.199 0.256 0.244 0.359 0.661 0.567 -0.329 -0.097 -0.105

0.210 0.253 0.058

Exchange rate volatility 1990-20034 Distance to main city (km) Log of distance to main city Spanish spoken (1=yes) Access to seacoast (1=yes) Absolute time difference to main financial centre Log of time difference to financial center

-0.496 -0.099 -0.168 -0.320 0.000 -0.063 -0.201

1.000 0.293 0.272 0.875 0.690

1.000 0.097 0.234 0.253

1.000 0.281 0.457

1.000 0.755

Sum of Land Areas (in km2) Product of populations (in billions) Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003) Absolute difference of percapita GDPs

0.073 -0.191 -0.155 0.202 0.282 0.101 -0.107 -0.116 0.008 0.142 0.190 0.083 0.135 0.602 0.196 0.482 0.321 0.317 0.407 0.179 -0.256 -0.385 -0.271 -0.075 -0.204

1.000

4

Exchange rate volatility 1990-2003

Average Inflation differencial 1990-2003

Member of Euro Area (1=yes)

Absolute difference of percapita GDPs

Absolute time difference to main financial centre

0.940

FDI Linkages 1991-20002 (F)

Sum of percapita GDPs (average 1990-2003)

Access to seacoast (1=yes)

0.345

Product of average GDPs 1990-2003

Spanish spoken (1=yes)

0.342

Trade Linkages 1980-19891 (lagged T)

Product of populations (in billions)

Log of distance to main city

1.000

2

Distance to main city (km) 1.000 0.931 0.223 0.250 0.924 0.665

3

Similarity in Productive Structure 1980-2000 (S

FDI Linkages 1991-20002 (F)

Trade Linkages 1980-19891 (lagged T)

1

Trade Linkages 1990-1999 (T)

Growth correlations with Spain, 1990-2003 (ρ ) Trade Linkages 1990-19991 (T)

Sum of Land Areas (in km )

Cross Correlations (Based on common 44 observations. Boldface: correlations above 0.6) Growth correlations with Spain, 1990-2003 (ρ )

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1.000 0.212 0.338

1.000 -0.476

1.000

1.000

-0.177 -0.454 -0.617 -0.132 0.036 -0.418 -0.485

-0.121 -0.456 -0.573 -0.055 0.036 -0.404 -0.482

-0.026 -0.288 -0.335 -0.052 0.034 -0.180 -0.190

1.000 0.231 1.000 0.101 -0.139 0.043 -0.178 -0.195 -0.241 -0.233 -0.193 -0.344

1

Average over the period of the sum of bilateral exports plus imports over the sum of GDPs

2

Average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. Higher values imply more similarity.

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

-0.214 -0.476 -0.580 -0.182 -0.086 -0.457 -0.422

1.000 0.727 1.000 0.111 0.010 0.166 0.104 0.208 0.023 0.066 0.071 0.019 -0.018 0.129 0.171

-0.245 0.335 0.261 0.306 0.362 -0.145 0.009 0.004 0.177 0.233 -0.053 -0.007 -0.026 0.068 0.126 0.314 -0.217 -0.367 -0.190 -0.259 -0.501 0.055 0.272 0.138 0.273

-0.008 0.145 0.385 0.342 -0.082 0.100 0.247 0.214 -0.108 0.115 0.233 0.141 -0.279 -0.157 -0.145 -0.388 0.065 0.023 0.187 0.216

1.000 0.510 1.000 0.588 0.512 0.003 -0.322 0.374 0.473

Table 7b

1.000 0.263 0.143

1.000 0.719

1.000

4

Average Inflation differencial 1990-2003

Member of Euro Area (1=yes)

-0.237 -0.073 -0.138 -0.029 0.185 0.067 -0.088

Sum of Land Areas (in km2) Product of populations (in billions) Product of average GDPs 1990-2003 Sum of percapita GDPs (average 1990-2003) Absolute difference of percapita GDPs

0.096 -0.022 0.042 0.291 0.347 -0.109 0.153 0.149 0.190 0.173 0.115 0.014 0.027 0.097 0.257 -0.049 -0.005 -0.015 0.118 0.125 0.162 0.222 0.279 0.654 0.351 0.070 -0.009 -0.035 0.024 0.009 0.323 0.470 0.490 0.497 0.598 0.397 -0.100 -0.211 -0.128 -0.281 -0.233 -0.425 -0.390 -0.246 -0.550 -0.419 0.061 0.177 0.055 0.225

-0.112 -0.381 -0.577 -0.120 0.130 -0.269 -0.378

-0.081 -0.396 -0.571 -0.096 0.159 -0.270 -0.409

-0.037 -0.242 -0.336 -0.075 0.100 -0.105 -0.155

0.034 -0.090 -0.222 -0.044 0.300 0.037 -0.138

1

Average over the period of the sum of bilateral exports plus imports over the sum of GDPs

2

Average over the period of bilateral inflows and outflows of FDI to and from Spain

3

Computed from value added from the industrial sector only. Higher values imply more similarity.

4

Coefficient of variation of the bilateral exchange rate with Spain (monthly average).

-0.147 -0.391 -0.514 -0.149 0.063 -0.287 -0.282

Absolute difference of percapita GDPs

Log of time difference to financial center

1.000 0.075 0.359 0.251

Exchange rate volatility 1990-20034 Distance to main city (km) Log of distance to main city Spanish spoken (1=yes) Access to seacoast (1=yes) Absolute time difference to main financial centre Log of time difference to financial center

1.000 0.319 1.000 0.028 -0.073

Sum of percapita GDPs (average 1990-2003)

Absolute time difference to main financial centre

1.000 0.309 0.013 0.767 0.661

0.345 0.324 0.019

Product of average GDPs 1990-2003

Access to seacoast (1=yes)

1.000 0.914 0.260 0.076 0.860 0.630

0.244 0.409 0.452 0.245 0.660 0.575 -0.042 -0.051 -0.055

Product of populations (in billions)

Spanish spoken (1=yes)

1.000 0.075 0.125 0.240 0.017 0.083 0.145

Similarity in Productive Structure 1980-20003 (S) Member of Euro Area (1=yes) Average Inflation differencial 1990-2003

2

Log of distance to main city

1.000

Distance to main city (km)

1.000 0.681

Exchange rate volatility 1990-2003

FDI Linkages 1991-20002 (F)

1.000 0.944 0.629

3

Trade Linkages 1980-19891 (lagged T)

1.000 0.246 0.259 0.184

Similarity in Productive Structure 1980-2000 (S

Trade Linkages 1990-19991 (T)

Growth correlations with Spain, 1990-2003 (ρ ) Trade Linkages 1990-19991 (T) Trade Linkages 1980-19891 (lagged T) FDI Linkages 1991-20002 (F)

Sum of Land Areas (in km )

Table of Cross Correlations - extended set of observations (Based on common 104* observations. Boldface: correlations above 0.6) Growth correlations with Spain, 1990-2003 (ρ )

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

1.000 -0.752

1.000

1.000 0.838 0.110 0.133 0.295 0.076 0.140 0.121

* Includes 44 observations from previous table plus common observations included by setting FDI Linkages equal to zero for missing values.

-0.051 0.104 0.237 0.159 -0.095 0.109 0.175 0.122 -0.104 0.140 0.168 0.061 -0.165 0.198 0.005 -0.208 0.054 -0.297 -0.046 0.068

1.000 0.543 1.000 0.629 0.551 1.000 0.201 -0.056 0.376 0.030 0.128 -0.038

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Table 8

Countries included in the regressions (total=104) ISO code

Country Name

ISO code Country Name

ISO code Country Name

ISO code Country Name

ARG AUS AUT BDI BEN BFA BGD BLZ BOL BRA BRB BWA CAF CAN CHE CHL CHN CIV CMR COG COL CPV CRI CYP DNK DOM

Argentina Australia Austria Burundi Benin Burkina Faso Bangladesh Belize Bolivia Brazil Barbados Bostwana Central African Republic Canada Switzerland Chile China Cote d'Ivoire Cameroon Congo Brazzaville Colombia Cape Verde Costa Rica Cyprus Denmark Dominican Republic

DZA ECU EGY ETH FIN FJI FRA GAB GBR GER GHA GMB GNQ GRC GTM HKG HND HTI HUN IDN IND IRL IRN ISL ISR ITA

JAM JOR JPN KEN KOR LCA LKA LSO MAR MDG MEX MUS MWI MYS NER NGA NIC NLD NOR NPL NZL PAK PAN PER PHL PNG

POL PRT PRY ROU RWA SEN SGP SLE SLV SWE SYC SYR TGO THA TTO TUN TUR TZA UGA URY USA VEN VNM ZAF ZMB ZWE

Algeria Ecuador Egypt Ethiopia Finland Fiji Is. France Gabon UK Germany Ghana Gambia Equatorial Guinea Greece Guatemala Hong Kong Honduras Haiti Hungary Indonesia India Ireland Iran Iceland Israel Italy

Jamaica Jordan Japan Kenya Korea St. Lucia Sri Lanka Lesotho Morocco Madagascar Mexico Mauritius Malawi Malaysia Niger Nigeria Nicaragua Netherlands Norway Nepal New Zealand Pakistan Panama Peru Phillipines Papua New Guinea

Poland Portugal Paraguay Romania Rwanda Senegal Singapore Sierra Leone El Salvador Sweden Seychelles Syria Togo Thailand Trinidad and Tobago Tunisia Turkey Tanzania Uganda Uruguay USA Venezuela Vietnam South Africa Zambia Zimbabwe

In boldface: countries included in the original sample of 44 countries. The rest of countries (60) were added after setting Financial Integration (F) equal to zero for all missing observations of that variable.

How Much do Trade and Financial Linkages Matter…

137

In any event, these results are only preliminary, mainly because of data limitations. In fact, financial integration is only measured through bilateral FDI flows and there is no account of portfolio or other capital flows.58 This might lead to underestimating financial linkages and their effect on business cycle synchronization.

APPENDIX B: DEFINITION OF VARIABLES AND SOURCES Output Synchronization (ρ): Measured as the Pearson correlation between the log differences (growth rates) of annual GDP for Spain and those of a given country. Data for annual GDP at purchasing power parity was taken from the IMF’s World Economic Outlook database. Trade Linkages (T): Measured as the sum of imports and exports between Spain and a given country, over the sum of their respective GDPs. This measure is then averaged over the denoted period of time. That is,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

TESP ,i =

X ESP ,i ,t + M ESP ,i ,t 1 ∑ T t GDPESP ,t + GDPi ,t

Data for exports and imports was obtained from the IMF’s Direction of Trade Statistics. GDP data was taken from the Penn World Tables version 6.1. Financial Linkages (F): Measured as the sum of inflows and outflows of FDI between Spain and a given country. This measure is then averaged over the duration of the period. Data for FDI flows was obtained from the OECD’s International Direct Investment Statistics. Similarity in productive structure (S): Measured as the time average of discrepancies in economic structures. In particular, we take the shares sn,i,t of value added for industrial sector n in country i at time t and construct the following indicator of distance:

S 1ESP ,i = −

N 1 ∑∑ sn,ESP ,t − sn,i ,t T t n =1

For value added, we take industrial sectors at 2-digit ISIC level. Data was obtained from the United Nations Industrial Development Organization (UNIDO). Distance to main city: Computed at the great circle distance (in km) between Madrid (Spain), and the main city of a given country. In general, we take the capital city as the main city, except for the US (New York), Pakistan (Karachi), Brazil (Sao Paulo), China (Shanghai), Canada (Toronto), Switzerland (Zurich), Germany (Frankfurt), Turkey (Istambul), Israel (Tel Aviv), India (Mumbay), Australia (Sydney), Cote d’Ivoire (Abidjan), Kazakhstan (Almaty), Morocco (Casablanca), New Zealand (Auckland), Nigeria (Lagos), South Africa (Johannesburg) and Yemen (Aden). Data was obtained from http://www.indo.com/distance/index.html. 58

New versions of this paper will make use of newly processed data for bilateral financial flows and stocks obtained from the Spanish Balance of Payments.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

138

Alicia García Herrero and Juan M. Ruiz

Spanish spoken: dummy variable which takes value 1 if a given country has Spanish as the main language. Data was elaborated by the authors. Access to seacoast: dummy variable which takes value 1 if a country has sovereign access to the seacoast. Data elaborated by the authors. Absolute time difference to main financial center: Absolute value of the standard time zone difference between the main city used for “distance” and mainland Spain. Source: http://www.timeanddate.com/worldclock/ Member of Euro Area: dummy variable which takes value 1 if a given country has joined the Euro. Data elaborated by the authors. Average Inflation Differential: Computed as the time average over the period referred of the absolute difference of quarterly inflation rates between Spain and a given country. Annual inflation data was obtained from the IMF’s International Financial Statistics. Exchange Rate Volatility: Computed as the standard deviation (over the period referred) of the bilateral nominal exchange rate (monthly average) between Spain and a given country. Monthly exchange rate data was obtained from the IMF’s International Financial Statistics using bilateral exchange rates for both countries vis-à-vis the US dollar. Sum of land areas: Computed as the sum of land areas (in square km) of Spain and a given country. Data for land areas was obtained from http://www.infoplease.com/ipa/ A0004379.html and the CIA World Factbook. Product of Populations: Computed as the product of average populations in both countries for the period chosen (divided by 1012). Data on countries’ population was obtained from the World Bank. Product of Average GDPs: obtained as the product of average annual GDPs measured at PPP. GDP data at PPP was obtained from the Penn World Tables 6.1. Sum of per capita GDPs: time average of the sum of per capita GDP for Spain and a given country. Data was obtained from the Penn World Tables 6.1. Absolute difference of per-capita GDPs: measured as the time average over the referred period. Data was obtained from the Penn World Tables 6.1. Similarity of oil dependency: constructed as the product of average oil dependency in Spain and a given country i:

⎛1 Moili ,t − Xoili ,t ⎜⎜ ∑ GDPi ,t ⎝T t

⎞ ⎛1 MoilESP ,t − XoilESP ,t ⎟⎟ × ⎜⎜ ∑ GDPESP ,t ⎠ ⎝T t

⎞ ⎟⎟ ⎠

where Moili,t and Xoili,t are imports and exports of oil in country i at time t and ESP represents Spain. Data for oil imports and exports as well as nominal GDP (all in current US dollars) was obtained from the World Bank.

REFERENCES Aizenman, Joshua and Ilan Noy (2004): “On the Two Way Feedback Between Financial and Trade Openness,” NBER Working Paper 10496.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

How Much do Trade and Financial Linkages Matter…

139

Baxter, Marianne and Robert King (1999): “Measuring Business Cycles: Approximate BandPass Filters for Economic Time Series,” Review of Economics and Statistics, 81, pp. 575-93. Bordo, Michael and Thomas Helbling (2003): “Have National Business Cycles become More Synchronized?,” NBER Working Paper 10130. Clark, Todd and Eric van Wincoop (2001): “Borders and Business Cycles,” Journal of International Economics, vol. 55, pp. 59-85. Deardorff, Alan (1998): “Determinant of Bilateral Trade: Does Gravity Work in a Neoclassical World?,” in Frankel J. (ed.) The Regionalization of the World Economy, University of Chicago Press. Edison, Hali, Michael Klein, Luca Ricci and Torsten Slok (2002): “Capital Account Liberalization and Economic Performance: Survey and Synthesis,” IMF Working Paper 02/120. Forni, Mario, Marc Hallin, Marco Lippi and Lucrezia Reichlin (2000): “The Generalized Dynamic-Factor Model: Identification and Estimation,” The Review of Economics and Statistics. 82(4), pp. 540-554. Frankel, Jeffrey (1992): “Measuring International Capital Mobility: A Review,” American Economic Review Papers and Proceedings, 82(2) pp. 197-202. Frankel, Jeffrey and Andrew Rose (1998): “The Endogeneity of the Optimum Currency Area Criteria,” Economic Journal 108, pp. 1009-25. Heathcote, Jonathan and Fabrizio Perri (2003a): “Why has the U.S. Economy Become Less Correlated with the Rest of the World?,” American Economic Review Papers and Procceedings,” vol 93, pp. 63-69 Heathcote, Jonathan and Fabrizio Perri (2003b): “Financial Globalization and Real Regionalization,” Working Paper, Georgetown University. Helbling, Thomas and Tamim Bayoumi (2003): “Are they all in the Same Boat? The 2000-01 Growth Slowdown and the G-7 Business Cycle Linkages,” IMF Working Paper 03/46. Imbs, Jean (2003): “Co-Fluctuations,” mimeo. (http://faculty.london.edu/jimbs/Research/ Cofluct2001.pdf) Imbs, Jean (2004a): “The Real Effects of Financial Integration,” Working Paper, London Business School. Imbs, Jean (2004b): “Trade, Finance, Specialization and Synchronization,” Review of Economics and Statistics, forthcoming. Imbs, Jean and Romain Wacziarg (2003): “Stages of Diversification,” American Economic Review, 93(1). International Monetary Fund (1997): World Economic Outlook, May. International Monetary Fund (2001a): “International Linkages: Three Perspectives,” World Economic Outlook, Chapter II, October. International Monetary Fund (2001b): “International Financial Integration and Developing Countries,” World Economic Outlook, Chapter IV, October. International Monetary Fund (2002): “Trade and Financial Integration,” World Economic Outlook, Chapter III, April. Kalemli-Ozcan, Sebnem, Bent Sorensen and Oved Yosha (2003): “Risk Sharing and Industrial Specialization: Regional and International Evidence,” American Economic Review, vol 93, pp. 903-18.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

140

Alicia García Herrero and Juan M. Ruiz

Kose, Ayhan and Kei-MuYi (2001): “International Trade and Business Cycles: Is Vertical Specialization the Missing Link?,” American Economic Review Papers and Proceedings, vol 91. pp 371-75. Kose, Ayhan, Eswar Prasad and Marco Terrones (2003a): “Financial Integration and Macroeconomic Volatility,” IMF Staff Papers, Vol 50, pp. 119-42. Kose, Ayhan, Eswar Prasad and Marco Terrones (2003b): “How Does Globalization Affect the Synchronization of Business Cycles?,” IMF Working Paper 03/27 Lane, Philip and Gian Maria Milesi-Ferretti (2001): “The External Wealth of Nations: Measures of Foreign Assets and Liabilities for Industrial and Developing Countries,” Journal of International Economics 55, pp. 263-294. Lane, Philip and Gian Maria Milesi Ferretti (2004): “International Investment Patterns,” IIIS Discussion Paper 24. Loretan, M. and W. English (2000): “Evaluating ‘correlation breakdowns’ during periods of market volatility,” International Finance Discussion Paper 658, Board of Governors of the Federal Reserve System. Lumsdaine Robin and Eswar Prasad (2003): “Identifying the Common Component of International Economic Fluctuations: A New Approach,” The Economic Journal, 113 (484), pp. 101-127. Nadal-de Simone, Francisco (2002): “Common and Idiosyncratic Components in Real Output: Further International Evidence,” IMF Working Paper 02/229. Portes, Richard and Hélène Rey (2003): “The Determinants of Cross-Border Equity Flows,” mimeo. Prasad, Eswar, Kenneth Rogoff, Shang-Jin Wei and Ayhan Kose (2003): “Effects of Financial Globalization on Developing Countries: Some Empirical Evidence,” mimeo, IMF. Stock, James and Mark Watson (2003): “Understanding Changes in International Business Cycle Dynamics,” NBER Working Paper 9859.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

In: Spain: Economic, Political and Social Issues Editor: Pablo N. Costada

ISBN 978-1-60692-108-1 © 2009 Nova Science Publishers, Inc.

Chapter 6

SECULARIZATION AND FERTILITY: EVIDENCE FROM SPAIN Pablo Brañas-Garza* and Shoshana Neuman Department of Economic Theory and History University of Granada Campus de la Cartuja s/n, 18071 Granada, Spain Department of Economics, Bar-Ilan University 52900, Ramat-Gan, Israel

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Since 1950 Spain has shown two parallel trends of dramatic drops in fertility and in religiosity (secularization). This paper explores the relationship between secularization and fertility among Spanish Catholics. We use a unique, rich, data set which includes various dimensions of religiosity: respondent’s religious affiliation; current church attendance (six levels); current prayer habits (eleven levels); spouse’s religious affiliation; parental religious affiliation; and parental (maternal and paternal) and respondent's church attendance during childhood (nine levels). The multi-facet data on religiosity (rather than a single dichotomous variable) allow for a sophisticated analysis, permitting rigorous conclusions to be drawn. The sample is restricted to married Catholic (female and male) respondents who were raised by Catholic parents and are married to a Catholic spouse in order to obtain a homogenous sample and to focus on the effect of the level (intensity) of religiosity (rather than religious affiliation) on fertility. Fertility is related to the various dimensions of religiosity; first using cross-tabulation and then using OLS regression. Our results are substantive: i) We find that fertility is not related to the current intensity of religiosity. ii) Exposure to religious activities during childhood has a significant effect on women’s fertility (but not men). Interestingly, a father who rarely attended church services has a negative effect on his daughter’s future fertility *

Corresponding author: Email: [email protected].

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

142

Pablo Brañas-Garza and Shoshana Neuman (decreasing the number of children by about 0.8), while the mother’s inactive churchgoing has an unexpected positive effect (leading to a increase of one child). iii) The respondent’s own church attendance during childhood does not have any effect on current fertility. In sum, this study demonstrates the significance of childhood experience in shaping one’s 'taste for children'. It also suggests that there is no direct link between the rapid process of secularization occurring in Spain and the decline in birth rates.

Keywords: fertility, religion, Catholic, church attendance, prayer, parental religiosity, taste for children, Spain.

1. MOTIVATION In the last 50 years, Spain has experienced two well-known phenomena: (1) a dramatic drop in fertility rates from a birth rate of 3 children in the mid 1950s to 2.8 in 1975 and to a mere 1.2 children in recent years. This very sharp drop started in 1975 (at the onset of democracy) and followed a descending trend until the late 1990s when it stabilized at a rate of 1.2 children (Fernández-Cordón, 1998). Although this decrease in birth rates is evident in many other European countries, Spain has undergone the sharpest changes. Fertility in Spain was higher than in most European countries up to the mid 1980s, but from that point on, the Spanish fertility graph fell below the fertility figures of most European countries (McDevitt, 1999, pp: A41-A42)59. Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

8

7

6

5

4

3 30

35

40

45

50

55 MEN

60

65

70

75

80

85

90

WOMEN

Figure 1. Secularization in Spain: Women and Men, since 1930.

59

The fertility rates in other European countries for 1998 are as follows: Iceland-2.0; Denmark-1.7; France-1.7; Sweden-1.8; UK-1.6; Austria-1.5; Belgium-1.5; Croatia-1.5; Hungary-1.5; Ukraine-1.4; Estonia-1.3; Germany- 1.3; Czech Republic-1.2; Bulgaria-1.1; Source: McDevitt (1999).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Secularization and Ferility

143

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

(2) A very significant and rapid process of secularization. Spain, which has always been considered a bastion of Catholicism, has experienced a clear drop in church attendance. This trend is even more remarkable if we consider that part of this process occurred during the Franco regime when church and state were closely linked. Figure 1 (from Brañas-Garza, 2004) shows the path of secularization – vertical axis shows attendance- from 1930 to 1998 (by decades). Observe that the temporal trend for both women and men (and without any gender difference) is clearly decreasing and so we must not expect any sort of “revival” (see Chaves, 1989). Analogously, the survey that will be used for our statistical analysis reveals that about two-thirds of the respondents attended church services on a regular basis when they were 12 years old and only about 10% rarely participated (no more than 2-3 times a year). The remaining one-forth attended occasionally. When these individuals grew up (after one to three decades)60, their church attendance habits changed drastically: less than 15% attend on a regular basis (every week) and close to 70% rarely attend (never, up to 1-2 times a year). In sum, Spain is experiencing a clear process of secularization. The observation of these two parallel phenomena naturally leads us to question and explore the interrelations between fertility and secularization61. This is precisely what we aim to do in the study presented in this paper: explore the effect of religiosity on fertility62 for a representative sample of Spanish Catholics. Our study replicates and expands upon papers by Adserá (2006a) who used two fertility surveys (the 1985 and 1999 Spanish Fertility Surveys- SFS) and Brañas-Garza and Neuman (2007) who used the Unesco-ISSP. Whereas Adserá finds that practicing Catholics exhibit significantly higher fertility rates than non-practicing Catholics, our previous paper supports a stronger effect due to parental transmission, that is, the array of religious and cultural values that parents pass on to their offspring seems to play a crucial role in current fertility decisions. This paper enlarges upon previous research by posing the following question: In what way does the loss of religious values affect current fertility decisions? Following previous papers we use both current (the lack of) religious activity and parental transmission of tastes for secularization. In order to test the effect of religiosity on fertility we use a data set that includes a wealth of information on several dimensions of religiosity. The data set was collected in 1998 by the Centro de Investigaciones Sociológicas (Center for Sociological Research, Spain) under the International Social Survey Program: Religion II funded by the UNESCO. The survey comprises 2488 personal interviews that were conducted throughout the 47 provinces of Spain and contains information on several dimensions of religiosity as follows: (1) The respondent's current religious denomination and religious activity as evidenced by two dimensions of religiosity: church attendance (a public religious activity with utilitarian/social motives- six alternative levels) and prayer (an intimate/private 60

The youngest respondent is 22 years old -10 years above the age of 12, while the oldest is 48 years old - 36 years above the age of 12. 61 Tilley (2003) studies the opposite case, that is, how fertility affects subject’s decisions on religiosity (using a British sample). 62 Throughout the paper we will refer to the (married) respondent’s number of children. Various terms will be used: fertility, fertility rates, birth rates, number of children. All of these terms refer to the same variable.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

144

Pablo Brañas-Garza and Shoshana Neuman

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

religious activity with purely religious motives-11 levels). If the respondent is married- the spouse’s religious affiliation is also taken into account. (2) The individual's exposure to religiosity during childhood including religious denomination and mother’s and father’s church attendance when the respondent was a child (nine alternative levels); and the individual’s church attendance at the age of 12 (nine levels)63. (3) A battery of personal and environmental socio-economic background questions (e.g. number of household members, marital status, age, education, personal income, household income, population size in place of residence, region) is also included64. The economic literature on the relationship between religiosity and fertility65 is mainly empirical and takes two complementary directions. The first set of studies focuses on interdenominational differences in family size. A second set of papers looks at the effect of religious practice within a given denomination (in most studies it is proxied by church attendance) on birth rates. A more detailed discussion follows in the second section. This paper analyzes the impact of individual loss of religious values on fertility, for a sample of young Spanish Catholics who grew up in households headed by two Catholic parents. The novelty of the analysis stems from the highly detailed information on the religiosity of respondents as it includes several indicators and dimensions of religiosity. This leads to a finer analysis, permitting us to distinguish between effects of current religiosity (expressed by church attendance and by prayer habits), of parental religiosity (church attendance when the individual was a child) and of the respondent's exposure during childhood (reflected by church attendance at the age of 12)66. The relationship between religiosity and fertility is examined using two types of statistical analyses. One is a descriptive table that relates parental and current religiosity to the number of children (measured by the mean, median and mode number of children). The second type uses an OLS regression analysis to present 'fertility equations'. In addition to the variables that are the focus of our study (parental religious inputs, exposure during childhood and current religiosity), the estimated equations also include other socio-economic variables that affect fertility in order to control for their effects and to arrive at net effects of religiosity variables. The analysis is carried out separately for each of the genders (allowing for gender differences). We restrict our analysis to married Catholic respondents (with a Catholic spouse) who grew up in households with Catholic parents. This is done in order to form a homogenous 63

Papers such as those by Brañas-Garza and Neuman (2006) or Grotenhuis and Scheeppers (2001) use retrospective individual church attendance at 12-15 years to explain current religiosity. However, very few also account for parental religious activity; what Brañas-Garza and Neuman (2006) call intergenerational transmission of 'religious capital'. Moreover the papers that focus on “transmission” (see Booth and Joo Kee 2006, for instance) do not consider religion as part of the transmitted capital. 64 This unique and rich data set was also used by Iannaccone (2003) as part of an international comparative study, by Brañas-Garza (2004) to study secularization in Spain, by Brañas-Garza and Neuman (2004) to explore determinants of religiosity in Spain and by Brañas-Garza and Neuman (2006) to explore the above-cited transmission of 'religious capital'. 65 Previous literature has also highlighted the relevance of religious affiliation and of religiosity on a wide range of economic and demographic factors such as educational attainment (Chiswick, 1988; Lehrer, 1999), marital stability (Lehrer and Chiswick, 1993) and economic growth (Barro and McCleary, 2003). 66 Unfortunately the data set does not contain information on the spouse's current level of religiosity (as indicated by church attendance and/or prayer) and/or information on parental or childhood prayer habits.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Secularization and Ferility

145

sample where all players belong to the same religion and are subject to the same rules of religious conduct. The respondents vary according to their level of current religious activity and by their exposure to religious practice during childhood. The paper is structured as follows: a short overview of the literature on religiosity and fertility is presented in the next section. The third section outlines background information on the composition of our sample in terms of birth rates and the religious practice of the respondents and their parents. In the following section an empirical analysis of the effect of the diverse dimensions of religiosity on the respondents' fertility rate is presented and discussed. The last section summarizes and concludes.

2. LITERATURE OVERVIEW There is a large body of literature (by demographers, sociologists, anthropologists, psychologists, historians and economists) linking religiosity and fertility (in both directions, see Tilley, 2003 for instance). In our short overview we will only refer to a small number of (mainly economic) studies which use individual religious attributes to explain fertility decisions. Let us now focus on the two main lines of inquiry:

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

2.1. Differences Between Religions (and Fertility) The first line of research on fertility and religiosity relates to inter-denominational differences in birth rates. Goldscheider (1971) offers three kinds of hypotheses on the relationship between religion and fertility: the religious-values hypothesis, the characteristics hypothesis, and the minority-group-status hypothesis. (1) The religious values hypothesis suggests that religious ideology and teachings affect fertility. This is done both directly by rejecting contraceptives and abortions and elaborating rules on the timing and frequency of sexual intercourse, and indirectly by teachings on the desired (large) number of children and on the appropriate roles for men and women. Religious values are more effective when religious institutions have the means to communicate values to their members and to establish mechanisms to promote compliance and to punish non-conformity (McQuillan, 2004); (2) The characteristics hypothesis claims that fertility differences between religions will disappear once the relevant socio-economic factors are controlled for (Anderson, 1986); and (3) in the sociological literature it is argued that minority religious groups (e.g., Jews) tend to adopt practices to reduce family size (Goldscheider, 1967; Knodel, 1974). Many of the empirical studies on this topic analyze US data. In the late 1970s, after years of consensus that Catholics (in the US) have significantly larger families compared to nonCatholics, Westoff and Jones (1979) found that fertility rates of these two groups were rapidly converging due to a sharp decrease in Catholic fertility. This result has been confirmed by Lehrer (1996) for Catholics in the US and by O'Grada and Walsh (1995), for Northern Ireland. Mormons (in the US) have much higher than average fertility rates, while Jews and people with no religious affiliation have birth rates that are below the US average. Fundamentalists are less likely than others to use effective methods of contraception and when faced with an unwanted pregnancy, are more likely to choose adoption over abortion

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

146

Pablo Brañas-Garza and Shoshana Neuman

(Medoff, 1993). Lehrer (1996) too found that Mormons (and to a lesser extent also conservative Protestants and Catholics) have significantly larger families due to the pronatalist orientation of their church. Schellekens and van-Poppel (2006) examine fertility patterns of Catholics, Protestants and Jews in The Hague (Netherlands) from 1860 to 1909 and conclude that religious ideology accounts for the low level of birth control among Catholics. Morgan et al. (2002) offer explanations as to why Moslems have higher fertility rates compared to non-Moslems and provide evidence from four Asian countries. They also present an overview of the literature on the subject. A battery of studies examines differences in fertility between Hindu and Moslem women in India. Some embrace the religious-values hypothesis claiming that the higher total fertility rate of Moslems is the result of the Moslem religion, which rejects family planning and embraces polygyny (e.g., Iyer, 2002). Jeffery and Jeffery (2000) are proponents of the characteristics hypothesis, suggesting that factors that are unrelated (or only tangentially related) to religion explain the different fertility rates of the two religious groups and if these factors (e.g., education, region of residence, economic status) were controlled for, the fertility differential would disappear. Borooah (2004) combines the two views by introducing both religious affiliation (Dalit Hindu, non-Dalit Hindu, Moslem) and socio-economic background variables into his estimated fertility equations. His econometric analysis suggests that over half of the observed birth surplus of Moslems over Hindus stems from differences in their responses toward fertility-influencing factors. Differences in the endowments of these factors explain the other half of the fertility surplus. The economic literature also discusses the effect of the couple’s religious composition on the number of births within marriage. Becker et al (1977) suggest that the lower stability of inter-faith marriages should reduce the number of births (compared to otherwise equal couples where both spouses have the same religious denomination). Lehrer and Chiswick (1993) find strong evidence for the fragility of inter-marriages. The larger the difference in religious background, the more likely conflict will arise. Because couples are aware of the frailty of their union, they desire fewer children. Lehrer (1996) finds a significant negative effect of out-marriage on fertility for Mormon and Catholic women. Grotenhuis and Scheeppers (2001) find that subjects (from a Dutch sample) who form part of a heterogeneous couple are more willing to reduce church attendance and become non-members. Adserá (2006a) also concludes that inter-faith unions in Spain are less fertile than unions of the same religion.

2.2. Differences Within a Given Religion (and Fertility) The effect of the level or intensity of religiosity (of members of a given religion) on fertility is studied in the second line of research. Neuman and Ziderman (1986) analyze a sample of Israeli Jewish male respondents, using a very rich data set on religious activities. They find a significant positive relationship between religiosity (measured by time devoted to religious activities) and fertility67. Several 67 In the estimated ‘fertility equations’ the respondents’ socio-economic background variables are controlled for in order to arrive at net effects of level of religiosity.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Secularization and Ferility

147

papers analyze samples composed of Catholics and present findings that indicate that respondents who attend church more frequently have higher fertility rates (Mosher and Hendershot, 1984; Williams and Zimmer, 1990; Sander, 1992; Adserá, 2006a). Amin et al. (1997) discuss the relationship between religiosity and fertility in Bangladesh. They argue that fertility is related to religiosity via the use of family planning. They found that contraceptive use is mainly influenced by religiosity at the district level and not so much by individual religiosity. In almost all the studies that have been published, religiosity is proxied by the respondent's church attendance (a dichotomous variable). With few exceptions – like Thornton et al. (1992) which focus on marriage rather than fertility – there is no reference to parental religiosity or exposure to religious activities during childhood. As Bisin and Verdier (2000, 2001) suggest, the cultural and religious environment in which children live shapes their future preferences. The formation of ‘social capital’ and of ‘religious capital’ begins early on with childhood experiences (either through training by family members or by watching them practicing social and religious activities). Children's preferences68 (including preferences for children) are acquired through a process of adaptation and imitation that depends on their parents' socialization actions. Individuals who grew up in households with intensively practicing Catholic parents (who highly valued large families and opposed the use of contraceptives and abortion on religious grounds) might internalize these values and opt for high birth rates. Contrary to this process of imitation and adaptation, psychologists also mention a reverse process of negative reaction. Individuals who grew up in a large family with many children and suffered from the cost and implications of having a hard-working mother and many siblings (i.e. little attention by the mother) might react with a desire to have fewer children. In this paper we follow the second line and explore the effect of religiosity on fertility for a homogenous sample of Spanish Catholics. We do not look at inter-denominational comparisons. Such comparisons are more relevant in a pluralistic society. Spain, however, is a country with one major religion and where almost 90% of the population is Catholic (see Brañas-Garza and Neuman, 2004 and 2006, for a more detailed analysis). Moreover, in Spain Catholic affiliation per-se seems to have lost some of its distinctive traits in recent years due to much lower church attendance and the wide use of family planning and contraceptives (Adserá, 2006a). Nevertheless, respondents who were raised by more religious practicing parents, who were exposed to church services during childhood and who are more religious (reflected in church attendance and in prayer habits), might have a taste for more children. This is precisely what we aim to explore.

3. SAMPLE AND DATA We examined a sample of Catholic respondents who were raised in households composed of two Catholic parents in order to have a homogenous sample of members belonging to the same religious denomination. This facilitates our focus on the effect of the intensity of religiosity on demographic behavior. Unlike Adserá (2006a) and Brañas-Garza and Neuman 68

For instance, Grotenhuis and Scheeppers (2001) show that church attendance during childhood (12-15 years old) has a positive and significant effect on current religiosity.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

148

Pablo Brañas-Garza and Shoshana Neuman

(2007), who analyzed only a female sample, we estimate 'fertility equations' for (separate) samples of both female and male respondents69. Data on various aspects of religiosity are used: the respondent's current religious activity as expressed by church attendance (6 levels) and by prayer habits (11 levels); the mother's and the father's church attendance when the individual was a child (9 levels); and the respondent's church attendance at the age of 12 (9 levels). The survey we use (conducted in 1998 by the Center for Sociological Research in Spain, under the International Social Survey Program: Religion II) includes very detailed information on religiosity but does not contain an explicit question on the number of children born. This core variable has been calculated using information on number of household members. Only married respondents are included and the age of the respondents is restricted to 45 for women and to 48 for men. This age limit was set to eliminate the possibility of children who got married and left the parents' household70. Obviously, these restrictions lead to a significant drop in sample sizes, but on the other hand yield a sample of a young cohort that was born in the early 1950s and married around the mid-1970s at the onset of Spanish democracy. The survey also includes information on various socio-economic background variables such as age, schooling, population size in city of residence, region of residence, personal income and family income. The variables will be used in our 'fertility equations' to control for background effects and arrive at net effects of religiosity on fertility. The coefficients of these variables are also informative and add to the understanding of fertility patterns. The region of residence is a key variable for two reasons. First, regions differ in terms of macro-level variables that might affect fertility, namely male and female unemployment rates, employment prospects, availability of more stable public-sector jobs, proportion of tenured jobs and of full-time employment (see Adserá 2006b, for a discussion on the effect of those variables on the number of children Spanish women desire71). Second, public facilities such as education and healthcare also vary among the regions due to differences in public spending by regional governments on health and education. For instance, according to reports by the “Bilbao-Vizcaya-Argentaria Public Capital Stock Series”, in 1998 Navarra ranked first in per capita public spending on health (352 euros, the regional average is 202 euros) and also on education (606 euros, compared to an average of 497). Andalusia ranked last (129 and 430 euros, on health and education, respectively). Obviously, this also means differential health and education facilities for children that translate into different costs of raising children, resulting in a different demand for children. Also, unlike other regions, in Castilla la Mancha and the Basque Country schooling is free at all levels. In Andalusia there are very few public kindergartens, while in the north there are plenty. By including the region as one of the explanatory variables we (partially) control for the different economic, educational and health 69

The decision regarding how many children to have is taken jointly by both the wife and the husband. It would therefore be optimal to have information on the religious activities and the religious background (as well as the socio-economic background) of each of the spouses in order to relate religiosity to fertility. However, as we have detailed data only for one spouse (the respondent), we can still gain insight into this phenomenon by analyzing women and men separately. 70 Children who left the household in order to study are still counted as household members. 71 Adserá (2006b) reports that when women in the 1995 SFS were asked why they restricted their fertility, several of the responses had to do with economic factors such as lack of economic resources (26.1%), the need to work outside the home (13.2%), unemployment of self/spouse (5.4%); small house (4.9%) or lack of childcare centers (2.1%). These factors are most likely region specific.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Secularization and Ferility

149

environments in which our respondents live, operate and make their family-planning decisions. Thirdly, regions also differ in terms of their religious make-up and the prevailing religious attitudes and norms. These differences might affect the desired (and actual) number of births in the region. Amin et al. (1997) claim that religiosity at the district level has a stronger effect on demographic behavior than the individual's own religiosity. In more traditional (and rural) districts there might be greater peer pressure to have larger families combined with group opposition to contraception. This reflects some sort of ‘neighborhood effect’. The coefficients of the region dummies therefore reflect the effects of employment, education and health prospects, as well as the religious attitudes of the inhabitants of the various regions. Let us begin with some descriptive statistics of the fertility variable and of the various dimensions of religiosity. The average and the distribution of the number of children are fairly similar in the female and male samples (see Appendix Table 1). Recall that our samples are composed of married women and men. The women are younger than 46 and the men are younger than 49. Obviously, the samples also include individuals who have not yet reached their desired number of children (in the regressions we control for this factor by including dummy variables for age groups). In the female sample, the number of children ranges from 0 to 7, with an average of 1.894 (SD is 1.379). In the male sample, the number of children is in the 0-6 range, with a mean of 2.013 (SD=1.354)72. About 15% have no children, around one-forth have one child, the mode is 2 children (30%), close to 20% are parents of 3 children, around 10% have 4 children and fewer than 5% are parents of 5 children or more. A summary of the descriptive statistics regarding the various dimensions of religiosity that are employed in our statistical analysis of fertility are presented in Figures 2 and 3. Figure 2 includes information on parental church attendance and the individual's church attendance at the age of 12, while Figure 3 contains information on current church attendance and current prayer habits. For each of the three childhood variables data is given on a scale of 1 to 9 (1- never attended church services to 9- attended several times a week)73. Because the responses to the questions regarding childhood are retrospective and might be inaccurate, we created a variable with three broader categories by combining similar responses. In this way, the 9 original options were reduced to the 3 following categories: (1) Secular=1: For original values of: 1 (never attended); 2 (once a year) and 3 (once or twice a year). Although we find positive levels of church attendance, it should be noted that it is nearly impossible not to go to church in Spain at least a few times per year for weddings, christenings or other social events. (2) Low-practicing=1: For original values of: 4 (attended a few times at year); 5 (once a month) and 6 (two or three times a month). This category includes individuals who

72 The sample average is larger than the birth rate in 1998 (1.2 children) because the sample includes respondents who had children during the last 3 decades when birth rates were higher. 73 Based on question #28 for the mother, question #29 for the father and question #30 for the child: "When you where a child, did your mother/father/yourself attend church services?” The options are: Never (1); once a year (2); once or twice a year (3); a few times a year (4); once a month (5); two or three times a month (6); almost every week (7); every week (8); several times a week (9).

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

150

Pablo Brañas-Garza and Shoshana Neuman attended church with some regularity but are still far from the prescription (once per week). (3) Intensive=1: For original values of: 7 (attended almost every week); 8 (every week); and 9 (several times a week). This is a category that is composed of intensively practicing Catholic mothers/fathers/children.

Given that there are no measurement errors for current religiosity, the original values can be used and will be used in the regression analysis. Additionally, in this case we also define three broader categories for both church attendance and for prayer habits. For church attendance: (1) Secular =1: For original values of: 1 (never attends); 2 (once a year) and 3 (once or twice a year). (2) Low-practicing =1: For original values of: 4 (attends once a month) and 5 (two or three times a month). (3) Intensive =1: For an original value of: 6 (every week). Similarly, we also define three broad categories for prayer habits:

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

(1) Secular =1: For original values of: 1 (never prays); 2 (once a year); 3 (twice a year) and 4 (a few times a year). (2) Low-practicing =1: For original values of: 5 (prays once a month); 6 (two or three times a month); 7 (almost every week); 8 (every week) and 9 (several times a week). (3) Intensive =1: For original values of: 10 (prays every day) and 11 (several times a day). By establishing similar definitions for childhood and for current religiosity measures, it is easier to compare the two. The dummy variables version will also be used in the regression analysis. The use of dummy variables does not require any assumptions on the type of relationship between fertility and religiosity, while the use of continuous variables (for church attendance and for prayer) is based on the assumption of linearity and monotony of the effects. As shown in Figures 2 and 3 (the three categories that represent the complete lack of church-attendance are marked in black, see secular-subjects on previous pages), there is a dramatic decrease in religious practice by offspring as compared to their parents and by the adult respondents as compared to their childhood religious experience. Our sample is restricted to women aged 18-45 and men aged 18-48. This means that the observed large decline took place during a relatively short time period. The youngest respondents are only a few years older than twelve, while about three decades have elapsed since the oldest respondents’ childhood experience. As Figure 2 demonstrates, over 50% of the mothers and over one-third of the fathers attended church services intensively (almost every week; every week; or several times a week). When they were twelve years old the respondents themselves were even more active than their parents- 65% attended church intensively (with their parents or with school). This changed dramatically when the individuals grew up. Currently (in 1998) over twothirds of women and men (66% of women and 73% of men) rarely go to church (never, up to

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

151

Secularization and Ferility

two times a year). Only around 14% of the respondents (both women and men) go to church frequently. This very significant intergenerational change stems from the rapid secularization of Spanish society (Brañas-Garza, 2004) and from the weakening network effect of participation in church services. The social gains from churchgoing diminished significantly at the onset of democracy in 1975, when the intimate link between the State and Catholicism collapsed (Adserá, 2006a). Regional differences can be found in church attendance. A breakdown of the sample by region74 shows considerable variability. According to our survey, Catalonia, the Basque Country, the Canary Islands and Madrid are the most secular regions. Indeed, close to 40% rarely go to church (the figures should be treated with some caution due to the small sizes of the regional samples). In contrast, Castilla la Mancha, La Rioja, Navarra and Castilla Leon show a very low level of secularization as more than 40% attend church services on a regular basis (every week). % 50 40

mot her s

30 20 10 0

% 50 40

f at her s

30

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

20 10 0

% 50 40

at t wel ve

30 20 10 0 never (1)

once a year (2)

2-3 ti mes a year (3) a f ew ti mes a year (4)

once a month (5)

2-3 ti mes a month al most ever y weeks (6)

ever y week (8)

(7)

sever al ti mes a week (9)

Figure 2. Tabulation of Retrospective Data.

As Figure 3 shows, our respondents are somewhat more active at prayer (a more intimate/pure religious activity). Women pray more frequently than men, reflecting gender

74 We broke down the larger sample of 2044 respondents who answered the relevant question (with no restrictions on marital status and age). We are interested in the religious make-up of the region that leads to a ‘neighborhood effect’. Moreover, the restricted samples are too small for a breakdown by region. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

152

Pablo Brañas-Garza and Shoshana Neuman

differences in the taste for religiosity75. Twenty percent of women and only 9% of men pray at least once a day. About 45% of women and two-thirds of men rarely pray. The remaining one-third of women and one-forth of men pray occasionally. Without a doubt, we can say that over half of the male population prays at some point. A breakdown of prayer habits by region reveals similar differences to the regional differences in church attendance: Castilla la Mancha, Castilla Leon and Navarra are at the top with over 40% praying at least once a day. A similar figure is also found for Madrid and Murcia. As the survey does not include questions on parental prayer habits, an intergenerational comparison is therefore impossible. In all likelihood a significant intergenerational change has also occurred in this religious activity. However, many young Spaniards still cherish the religious practice of prayer (see also Branas-Garza and Neuman, 2006, for descriptive statistics on church attendance and on prayer habits for a larger sample of Spaniards). % 50 40

c ur r e nt a t t e nda nc e

women

men

30 20 10 0 never (1)

once a year (2)

1-2 times a year (3)

once a month (4)

2-3 times/ month (5)

ever y week (6)

% 50 40

c ur r e nt pr a y e r ha bi t s

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

30 20 10 0 never (1)

once a year (2) twice a year (3)

f ew times a

once a month

2-3 times a

year (4)

(5)

month (6)

almost ever y ever y week (8) sever al times once a day (10) sever al times week (7)

a week (9)

a day (11)

Figure 3. Tabulation of Current Religious Data.

A description of the female and male samples in terms of socio-economic and demographic characteristics is presented in Appendix Table 2. The samples are representative of the Spanish population for the age group under discussion.

4. SECULARIZATION AND FERTILITY: AN OVERVIEW We will now study the link between the loss of religious values and fertility among young Spaniards, using a cross-tabulation descriptive presentation where each dimension of religiosity is examined separately. 75

See also Brañas-Garza and Neuman (2004) for a more detailed discussion of gender differences.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

153

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Secularization and Ferility

Figure 4 presents the average number of children for women and men, sorted by levels of religiosity (secular, low-practicing and intensively practicing) using all five religiosity variables, namely: individual’s current church attendance; current prayer activity; mother’s church attendance; father’s church attendance and respondent’s church attendance at the age of 12. Starting with the female sample we observe that the loss of religious values (both church attendance and prayer) does not seem to have an effect on fertility, although the effect of praying seems to be larger and negative. Retrospective data support a different view: the number of children seems to be negatively correlated with secularization. In fact, fertility increases with the intensity of the father’s church attendance (1.76, 1.88 and 1.91 children, for respective levels of secular, low and intensive practice) and with the respondent’s church attendance during childhood (1.48, 1.88 and 1.91 children for levels of secular, low and intensive participation in church services at the age of 12). In contrast, the mother’s loss of religious habits seems to be a positive determinant of the number of children that her daughter will have (an average of 2 children if the mother was rarely practicing and 1.73 children if she was intensively practicing). In the male sample, current secularization does not seem to affect the number of children but respondents who never pray have fewer children. Both the father’s and the mother’s secularization seem to have a positive impact on fertility (the number of children drops with an increase in the intensity of parental church attendance), while exposure to church services during childhood does not have a direct effect. Thus, parental religious investment (with the exception of the role of the father for women) could be an explanatory variable of current low levels of fertility. Lower religious investment by the mother and the father affect decisions by their offspring regarding fertility (when adults). In contrast, the role of the father and the respondent at age 12 is more ambiguous. All in all, religious experience during childhood affects later preferences and the 'taste for children' and seems to have a more pronounced effect on fertility than current religious activity. 2.5 pr ayer

mass

mother

f ather

at-12

WOM E N 2

1.5

1

Figure 4 (Continued on next page.)

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

154

Pablo Brañas-Garza and Shoshana Neuman

2.5

prayer

mass

mother

father

at-12 MEN

2

iv e te in

in t ic

lo

w-

pr

ns

g

r la cu

ac

se

iv e te in

t ic

ns

g

r

in

la ac pr wlo

in

cu se

iv e te in

t ic

ns

g

r la pr wlo

in

cu

ac

se

te in

ac lo

w-

pr

ns

g

iv e

r la

t ic

cu se

iv e ns

in ac

te

t ic

cu se

pr wlo

in

la

g

r

1.5

1

Figure 4. Secularization and Fertility.

In the next section we propose a general framework to study the effect of religiosity on fertility and explore a set of ‘fertility equations’ that use the multiple dimensions of religiosity as explanatory variables, including socio-economic determinants of fertility. The coefficient of each variable represents the net effect of that variable. While the first type of statistical analysis is possibly more intuitive, the econometric estimation of fertility equations is a more accurate and sophisticated tool as it reveals the net effects as well as their significance.

5. AN ECONOMETRIC STUDY OF THE EFFECT OF SECULARIZATION ON FERTILITY

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

5.1. A Formal Model Denote by N = {1, 2....n} the group of individuals in our data set and by b , i ∈ N the

i

number of children individual i desires (and has). It is assumed that the desired number of children is determined by four groups of factors. (1) Current Religiosity: This is the core variable of our paper. Our sample includes Catholic individuals. The effect of the Catholic religion on fertility is therefore part of the constant. However, within the Catholic religion there are different levels of religious practice that reflect levels of religiosity. To test the assumption of a positive relationship between religiosity and fertility the regression equation should include levels of current religious practice. These could be included in two alternative fashions, which reflect different underlying assumptions: a) using a continuous variable for levels of religious practice that includes a set of values ranging from the lowest (denoted by 1) to the highest. Using a continuous variable implies that we assume a monotonic, linear relationship with constant differences between any two consecutive values.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

155

Secularization and Ferility

Formally, denote by si the level of secularization76 of individual i. In a pro-natalist religion, such as the Catholic religion, birth rates will be a positive function of the level of religiosity of the individual, i.e.,

∂bi > 0, ∀i ∈ N and therefore si > si′ → bi > bi′ ∂si We may conjecture that if ∀i, j ∈ N is satisfied,

si > s j → bi > b j b) Alternatively, the assumptions of linearity and monotony can be relaxed, and the level of religious practice can be defined using a set of dummy variables which represent broader levels of religious practice. These levels can be referred to as sub-denominations: secular, low-practicing and intensively-practicing Catholics. Each sub-group has a different pronatalist policy and all individuals who belong to a given group follow the rules of conduct of that group. k

Formally, denote by D1, D2, …, Dd the set of different sub-denominations and by ci the birth rate that the Dk sub-denomination claims for any of its members i. If the Dh sub-denomination claims larger families for its believers compared to Dz, then individuals who belong to Dh will have more children than those in Dz,

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

cih > c zj ⇔ bi > b j , ∀i ∈ Dh , ∀j ∈ Dz However, within the same sub-denomination, there are no differences in fertility rates (where everything else is constant), i.e.,

bi = b j , ∀i, j ∈ Dk , k = 1...d (2) Childhood exposure to religiosity shapes the respondent’s values and preferences, including his or her 'taste for children': Preferences are part of the individual's endowments and can therefore be combined with, (3) Socio-economic endowments: Socio-economic factors that affect fertility include education, income, age, city size (urban/rural). Denote the set of socio-economic endowments (including preferences shaped by

{ i1,....., xis }, i ∈ N where x

religiosity during childhood) by X = x

i

ij

is the j-th variable for

individual i and s represents the total number of influencing variables.

76 Obviously we might use religious activity instead of secularization: let us define ai as the level of religiosity of individual I and so on. Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

156

Pablo Brañas-Garza and Shoshana Neuman

The derivative

∂bi represents the marginal effect of this attribute (variable) on fertility ∂xi j

and could be either positive or negative. (4) Regional public policies and institutions:A fourth set of variables that affect birth rates are related to local (national) regulations that include availability of public schools, kindergarten and day-care centers; regional spending and subsidies for health and education; labor laws and regulations (such as minimum wages, unemployment benefits, maternity leave, job security). These are region specific. Let us assume that there are two regions R and R . Ri is the region individual i lives in and pi is the public spending of this region on education, health and other facilities that are used for raising children. For purposes of simplicity we assume that regions differ only in terms of regional government spending. If R =Ri spends more than R =Rj, then,

pi > p j → bi > b j , ∀i ∈ Ri ,∀j ∈ R j And, everything else being constant, the effect of the regional variable is similar for all the inhabitants of the region, such that,

{ }

bi = b j , ∀i, j ∈ Rk , Rk ∈ R, R

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

The regional variable pi, can be extended to a vector of policies of region Ri, i.e., Pi= (pi1,…, pim), where each component of the vector represents a different aspect of public macro-level regional policy.

The 'Fertility Equations': Based on the above formal presentation, we arrive at: (1) (1) bi = F ( si , X i , Pi ), ∀i ∈ N , assuming a continuous effect of current religiosity, or (2) (2) bi = F ( Di , X i , Pi ), ∀i ∈ N , if we assume aggregated effects of subdenominations. Assuming linearity, we arrive at the following equations: f

m

j =1

k =i

bi = ζ si + β j ∑ xij + η k ∑ pik + ε i

(1)

or d

f

m

k =i

j =1

k =i

bi = ∑ δ k Di + β j ∑ xij + η k ∑ pik + ε i Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

(2)

157

Secularization and Ferility

ζ captures differences among subjects with different levels of religious practice (proxied by church attendance and by prayer habits), and δ k reflects the effect of the kwhere

religious sub-denomination (defined using aggregated levels of church attendance and of prayer) and is equal for all the respondents who belong to this sub-denomination; the vector β j captures the effects of the socio-economic endowment variables that have an impact on fertility, including preferences shaped during childhood by exposure to parental and the respondent’s own religious practice. Finally, η k represents the impact of fertility-related public policies and

ε i is the error term.

5.2. Fertility Equations

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

As mentioned above, the descriptive statistics can serve as a basis to gain some intuitive preliminary information on the interrelationship between religiosity and fertility. Regression analysis is a more powerful tool that is capable of yielding net effects when all the explanatory variables are considered. The significance of the effects (coefficients) of the various independent variables is also tested. Robust conclusions can therefore be derived only from the multiple regression analysis that estimates ‘fertility equations’. We use a 'reduced form' model in which fertility decisions are analyzed taking into account the constraints embodied in the outcomes of other previous household decisions (i.e. educational attainment, age at marriage and spouse’s participation in the workforce) as well as exogenous environmental constraints77. Table 1 presents fertility equations based on the formal econometric model. The independent variables include the following sets: Table 1. Fertility Equations: Young Catholic Spanish Women and Men, 1998 Independent Variables

Women Eq. 1

(1) Current religiosity Church attendance

Men Eq. 2

Eq. 1

Eq. 2

Secular

----

0.05 (0.86)

----

-0.20 (0.57)

Low Continuous (1-6) Prayer habits Secular Low Continuous (1-11)

----0.11 (0.08) ------0.32 (0.28)

-0.04 (0.88) ----0.20 (0.42) -0.38 (0.12) ----

---0.04 (0.89) -------0.00 (0.89)

-0.69 (0.13) ---0.08 (0.82) 0.14 (0.73) ----

77 An alternative approach uses a 'structural' model followed by the estimation of a set of simultaneous equations that relate to the variety of decisions made in the household (e.g., fertility, education, labor force participation). The structural model has the advantage of clarifying the interrelated processes that take place within the household. However, it has major statistical identification problems that lead to estimation problems and biases. The coefficients of the 'reduced form’ equation are unbiased estimates of the effects of the explanatory variables (constraints) on fertility.

Costada, Pablo N.. Spain : Economic, Political, and Social Issues, Nova Science Publishers, Incorporated, 2008. ProQuest Ebook Central,

158

Pablo Brañas-Garza and Shoshana Neuman Table 1. (Continued)

Copyright © 2008. Nova Science Publishers, Incorporated. All rights reserved.

Independent Variables

Women Men Eq. 1 Eq. 2 Eq. 1 Eq. 2 (2) Exposure to religiosity during childhood: formation of values and preferences Mother Secular 1.01 (0.00) 1.04 (0.00) 0.56 (0.16) 0.56 (0.16) Low 0.62 (0.01) 0.65 (0.01) 0.06 (0.81) 0.06 (0.81) Father Secular -0.80 (0.01) -0.79 (0.01) -0.14 (0.70) -0.14 (0.70) Low -0.39 (0.08) -0.37 (0.11) 0.14 (0.63) 0.14 (0.63) Child (at 12) Secular -0.53 (0.11) -0.54 (0.11) -0.12 (0.73) -0.12 (0.73) Low -0.04 (0.83) -0.04 (0.86) -0.04 (0.86) -0.04 (0.86) (3) Socio-economic and demographic background years of schooling -0.01 (0.39) -0.02 (0.26) -0.02 (0.28) -0.02 (0.26) age 31-40 0.74 (0.00) 0.70 (0.00) 1.07 (0.01) 0.70 (0.00) age 41-45 (48) 1.10 (0.00) 1.02 (0.00) 1.10 (0.01) 1.02 (0.00) city with 10,000 0.13 (0.50) 0.13 (0.51) 0.05 (0.84) 0.04 (0.84) inhabitants or less (4) Regional public policies and institutions (regions) Andalusia 0.51 (0.16) 0.48 (0.18) 0.58 (0.11) 0.51 (0.16) Aragón -0.41 (0.42) -0.53 (0.31) -0.23 (0.74) -0.34 (0.63) Asturias -0.23 (0.67) -0.21 (0.71) -0.33 (0.60) -0.41 (0.52) Balearic Islands -0.78 (0.23) -0.71 (0.28) 0.04 (0.93) 0.03 (0.95) Canary Islands 0.02 (0.95) -0.03 (0.93) 1.25 (0.07) 1.31 (0.07) Castilla la Mancha 0.75 (0.18) 0.69 (0.22) -1.17 (0.10) -1.34 (0.06) Castilla León -0.03 (0.94) -0.07 (0.87) -0.58 (0.20) -0.66 (0.15) Catalonia -0. 32 (0.37) -0.32 (0.38) -0.56 (0.15) -0.70 (0.08) Valencia -0.20 (0.62) -0.29 (0.47) -0.58 (0.21) -0.67 (0.15) Extremadura -0.07 (0.90) -0.09 (0.87) 0.51 (0.53) 0.34 (0.67) Galicia 0.75 (0.12) 0.71 (0.15) 0.70 (0.25) 0.84 (0.17) Cantabria -1.39 (0.12) -1.57 (0.09) -0.25 (0.69) -0.45 (0.49) Murcia -0.95 (0.06) -0.89 (0.08) 0.22 (0.75) 0.29 (0.68) Navarra 2.73 (0.00) 2.60 (0.00) 1.20 (0.14) 1.01 (0.23) Basque Country 0.06 (0.90) 0.03 (0.94) 0.31 (0.52) 0.13 (0.79) La Rioja 3.56 (0.00) 3.39 (0.00) -2.07 (0.11) -2.20 (0.10) Constant: 1.61(0.00) 1.77 (0.00) 1.18 (0.08) 1.51 (0.05) Sample size: 207 207 157 157 R2 0.39 0.39 0.27 0.28

Notes: The sample includes Catholics (women younger than 46, men younger than 49) raised by two Catholic parents and married to a Catholic spouse. p-values are given in parentheses. The reference groups (for the dummy variables) are as follows: for level of religiosityintensive (for definitions see pages 12 and 13); for age groups: age