Undernutrition, Agriculture and Public Provisioning: The Impact on Women and Children in India 9780367361723, 9780429344299

Table of contents :
Cover
Half Title
Series Page
Title Page
Copyright Page
Contents
List of figures
List of tables
List of contributors
Foreword
Preface
1 Introduction: nutritional well-being of women and children in India
2 Child underweight and agricultural land productivity
3 Child nutrition: linkages to agriculture
4 Women’s BMI among farm and non-farm households in rural India
5 Child health opportunity index: a regional analysis
6 Nutritional status of women across social groups
7 Access to milk and milk products and child undernutrition
8 Conclusions: are we on the right path to achieve better nutrition?
Index

Citation preview

Undernutrition, Agriculture and Public Provisioning

Using quantitative techniques, this volume provides empirical evidence on the crucial role of public provisioning of food, water, sanitation and health care in reducing undernutrition among women and children in India. The linkages are cogently explored and connected to the sustainable development goals. Key data comes from recent large secondary sources at district, household and individual levels and the econometric methodologies are clearly explained. Taken as a whole, it highlights the effects of public provisioning on malnutrition and identifies the relative importance of agricultural growth in resolving the nutrition problems in rural and semi-urban areas of India. This edited volume will be valuable reading for advanced graduate students, researchers and practitioners in development economics, development studies, and nutrition and public health. Swarna Sadasivam Vepa is a research consultant and visiting professor at the Madras School of Economics (MSE), India. She teaches graduate courses in Indian economic development and development economics. She has taught agricultural economics, microeconomics and macroeconomics courses to undergraduate and graduate students. Recently, she worked as a full-time research consultant on social inclusion at the Centre for Economic and Social Studies, Hyderabad, and as a full-time consultant on the Leveraging Agriculture for Nutrition in South Asia project of UK AID. Her research interests include dryland agriculture, food security, nutrition, social inclusion and gender issues. Brinda Viswanathan, PhD, is Professor at the Madras School of Economics, Chennai, and teaches courses in Indian economic development, development economics and quantitative economics for post-graduate students. Her research interest is in the broad area of development economics and applied econometrics. She regularly contributes as a resource person for workshops and training programmes for college teachers, PhD students and government officials on statistical and econometric techniques and evidence-based policy-making for India’s development.

Routledge Studies in Development Economics

Institutions, Technology and Development in Africa Jeffrey James Urban Policy in Latin America Towards the Sustainable Development Goals? Edited by Michael Cohen, Maria Carrizosa and Margarita Gutman Unlocking SME Finance in Asia Roles of Credit Rating and Credit Guarantee Schemes Edited by Naoyuki Yoshino and Farhad Taghizadeh-Hesary Full and Productive Employment in Developing Economies Towards the Sustainable Development Goals Rizwanul Islam Information and Communication Technologies for Development Evaluation Edited by Oscar A. García and Prashanth Kotturi Rent Seeking and Development The Political Economy of Industrialization in Vietnam Authored by Christine Ngoc Ngo Capital Mobility and Distributional Conflict in Chile, South Korea, and Turkey Kurtuluş Gemici Cooperation Networks and Economic Development Cuba’s High-Tech Potential Andrés Cárdenas O’Farrill Undernutrition, Agriculture and Public Provisioning The Impact on Women and Children in India Edited by Swarna Sadasivam Vepa and Brinda Viswanathan For more information about this series, please visit: www.routledge.com/series/SE0266

Undernutrition, Agriculture and Public Provisioning The Impact on Women and Children in India

Edited by Swarna Sadasivam Vepa and Brinda Viswanathan

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 selection and editorial matter, Swarna Sadasivam Vepa and Brinda Viswanathan; individual chapters, the contributors The right of Swarna Sadasivam Vepa and Brinda Viswanathan to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book has been requested ISBN: 978-0-367-36172-3 (hbk) ISBN: 978-0-429-34429-9 (ebk) Typeset in Times New Roman by Apex CoVantage, LLC

Contents

List of figures List of tables List of contributors Foreword

vii ix xiii xv

R . R A D H A K R I SHNA

Preface

xvii

J . V. M E E N A K S H I

1

Introduction: nutritional well-being of women and children in India

1

S WA R N A S A D A S I VAM VE PA AND BRI NDA VI S WAN ATH A N

2

Child underweight and agricultural land productivity

13

S WA R N A S A D A S I VAM VE PA, ROHI T PARAS AR A N D B R I N D A V I S WANAT HAN

3

Child nutrition: linkages to agriculture

49

A N U S H A G A N A PAT I AND S WARNA S ADAS I VAM VEPA

4

Women’s BMI among farm and non-farm households in rural India

82

B R I N D A V I S WA NAT HAN AND GE T S I E I MMANUE L

5

Child health opportunity index: a regional analysis

115

G . N A L I N E A N D BRI NDA VI S WANAT HAN

6

Nutritional status of women across social groups

151

S WA R N A S A D A S I VAM VE PA AND ROHI T PARAS AR

7

Access to milk and milk products and child undernutrition R O H I T PA R A S A R, R.V. BHAVANI AND S . RAJU

201

vi

Contents

8

Conclusions: are we on the right path to achieve better nutrition?

230

S WA R N A S A DA S I VAM VE PA

Index

235

Figures

1.1 3.1 3.2 3.3 3.4 3A.1 3A.2 5.1 5.2 5.3 5.4 5.5a 5.5b 5.6a 5.6b 5.7 5.8 6.1 6.2 6.3 6.4 6.5

Theory of change diagram: Effect of agriculture, public provisioning and social opportunity on undernutrition Percentage share of GDP in districts having less than 80% rural population Percentage share of GDP in districts having more than 80% rural population Per worker productivity (Rs. ’000s) in districts with less than 80% rural population Per worker productivity (Rs. ’000s) in districts with more than 80% rural population Scatter plot with per hectare productivity and percentage of underweight children using DLHS-2 and NFHS-4 Scatter plot with per worker productivity and percentage of underweight children using DLHS-2 and NFHS-4 Indian states with high neonatal mortality rates and their PCGSDP Rural-urban differentials in high NMR states and percentage share of NMR in IMR Indian states with high percentage of stunting Indian states with high percentage of stunting by residence Percentage of ICDS supplementary nutrition beneficiaries (six months to six years), 2016–17 Percentage of stunted children (birth to five years), 2015–16 Percentage of ICDS beneficiaries (two to six years), 2016–17 Percentage of stunted children (birth to five years) in Tamil Nadu, 2015–16 Cumulative distribution plot of HAZ for ranges of mother’s height Geographical zone-wise box plot of access to ICDS with height-for-age z-score Mean BMI of women across social groups in Andhra Pradesh Mean BMI of men across social groups in Andhra Pradesh Mean BMI of women across social groups in Telangana Mean BMI of men across social groups in Telangana Mean heights of women across social groups in Andhra Pradesh

4 59 60 61 61 78 78 119 120 121 122 133 134 135 136 143 143 183 183 183 184 184

viii Figures 6.6 6.7 6.8 6B.1 7.1 7.2

Mean heights of men across social groups in Andhra Pradesh Mean heights of women across social groups in Telangana Mean heights of men across social groups in Telangana The successor states of Andhra Pradesh and Telangana Milk consumption in children across wealth categories, Odisha (%) Percentage of children who consume powdered/tinned/fresh milk more than once a day across social categories in urban and rural regions, Odisha 7.3 Percentage of children who consume powdered/tinned/fresh milk more than once a day across wealth categories in rural and urban regions, Odisha 7.4 District map of Odisha 7.5 Underweight children (6–59 months) by social groups, Odisha (%) 7.6 Underweight children (6–59 months) by wealth class, Odisha (%) 7.7 Underweight in children (6–59 months) who consume/do not consume milk (%) 7.8 Association of prevalence of underweight children (6–59 months) with percentage of children who consume milk across districts of Odisha 7.9 Frequency of milk consumption among children by age category 7.10 Percentage of children who consume milk daily by household type 7.11 Consumption of commercial child food by children by age category (%) 7.12 Consumption of commercial child food by children across household type (%)

184 185 185 200 207 208 210 212 213 214 215 215 220 221 223 223

Tables

2.1 2.2 2.3 2.4 2.5 2A.1 2A.2 2A.3 2A.4 2A.5 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8

Value of agricultural output growth (2004–05 prices) Sectoral GDP of districts with less than 80% rural population Sectoral GDP of districts with more than 80% rural population Estimated coefficients of agricultural land productivity across districts Estimated coefficients of child underweight across districts Child underweight rates from various sources (2002–04 to 2012–13) Descriptive statistics (level form) Descriptive statistics of variables (in logarithmic form) Correlation matrix for stunting, underweight and wasting, NFHS-4 Correlation matrix of variables used in the child nourishment equation Sectoral GDP of urbanised districts (districts with less than 80% rural population in 2001) Sectoral GDP of predominantly rural districts (districts having more than 80% rural population in 2001) Sectoral GDP of urbanised districts (districts having less than 80% rural population in 2011) Sectoral GDP of rural districts (districts having more than 80% rural population in 2011) Pooled (NFHS-4, DLHS-2) OLS estimation of impact of agriculture on child underweight Pooled (NFHS-4, DLHS-2) OLS estimation of impact of agriculture on child underweight (log form) Mixed effects ML estimation of the impact of agricultural land productivity on underweight children at the level of districts and year Mixed effects ML estimation of the impact of agricultural worker productivity on underweight children at the level of districts and year

18 31 32 33 34 43 44 45 45 46 62 62 62 63 64 65 66 66

x

Tables 3.9

Mixed effects ML estimation of the impact of agricultural land productivity on underweight children at the level of districts and states 3.10 Mixed effects ML estimation of the impact of agricultural worker productivity on underweight children at the level of districts and states 3.11 Fixed effects panel data estimation of the impact of agriculture on underweight children 3.12 Fixed effects panel data estimation of the impact of agriculture on underweight (log form) 3A.1 Descriptive statistics 3A.2 Correlation matrix of DLHS-2 and NFHS-4 variables 3B.1 Value of agricultural output growth (2004–05 prices) 3B.2 Percentage of underweight children below the ages of five and six 3B.3 Description of variables and data sources 3B.4 OLS estimation of the impact of agriculture on underweight children (NFHS-4) 4.1 Distribution of women aged 20 to 45 years, CED rates and mean BMI across major sources of income in rural areas 4.2 Mean differences in BMI and CED rates across major sources of income 4.3 Percentage of women with CED across income, consumption and asset quintile: comparing non-farm and farm households in rural areas 4.4 Distribution of women and CED rates across states of India (%) 4.5 Descriptive statistics of variables 4.6 Estimates from probit model for CED (2005 and 2011) 4A.1 Brief summary of the explanatory variables 5.1 Comparison of child opportunities of outcomes with selected circumstances 5.2 Index of inequality of opportunity (D) and childhood opportunity index (COI) for six zones of India and for each set of circumstances (2005–06) 6.1 Sector-wise shares of gross value added and workforce 6.2 Poverty in Andhra by social groups 6.3 Poverty in Telangana by social groups 6.4 Percentage of women (15–49) by BMI level 6.5 Chronic energy deficiency among adults above the age of 20 years 6.6 Access to drinking water and toilets within the dwelling 6.7 Factors influencing body mass index of males and females in successor states after bifurcation (OLS) 6.8 Factors influencing body mass index in the composite state before bifurcation (OLS)

67 68 69 70 75 77 79 79 80 81 95 97 99 100 102 103 113 138 141 154 154 155 162 163 163 165 166

Tables 6.9 6.10 6.11 6.12 6.13 6.14 6.15 6.16 6.17 6.18 6.19 6.20 6.21 6.22 6A.1 6A.2 6A.3 6A.4 6B.1 6B.2 6B.3 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9

Factors influencing BMI quantiles (0.20) of males and females in Andhra and Telangana Factors influencing BMI quantiles (0.40) of males and females in Andhra and Telangana Factors influencing BMI quantiles (0.60) of males and females in Andhra and Telangana Factors influencing BMI quantiles (0.80) of males and females in Andhra and Telangana Factors influencing height of males and females in successor states after bifurcation (OLS) Factors influencing height in the composite state before bifurcation (OLS) Factors influencing height quantiles (0.20) of males and females in Andhra Pradesh and Telangana Factors influencing height quantiles (0.40) of males and females in Andhra Pradesh and Telangana Factors influencing height quantiles (0.60) of males and females in Andhra Pradesh and Telangana Factors influencing height quantiles (0.80) of males and females in Andhra Pradesh and Telangana BMI and social groups in Telangana Height and social groups in Telangana BMI and social group in Andhra Pradesh Height and social group in Andhra Pradesh Toilet index (index varies from one to five) Drinking water supply index (varies between one and five) Cooking fuel index Housing index Distribution of SC and ST population in Andhra Distribution of SC and ST population in Telangana Well-being across social groups Nutrition status of children in Odisha (%) Distribution of sample households by district Distribution of sample households by location Distribution of sample households across the three survey districts in Odisha Children (6–59 months) who consume milk by social group (%) Frequency of consumption of milk per day by children (6–59 months) Frequency of consumption of milk by children (6–59 months) by social group District-wise percentage of children (6–59 months) who consume milk Prevalence of underweight (%) in children and milk consumption

xi 167 168 169 170 175 176 177 178 179 181 186 187 188 189 195 196 197 197 198 199 199 204 204 204 205 206 208 209 211 215

xii Tables 7.10 Association of child underweight with milk consumption and wealth class 7.11 Association of child underweight with milk consumption, wealth and social group 7.12 Cross tabulation of households having electricity and as per household type 7.13 Distribution of households having electricity and as per household type 7.14 Frequency of consumption of milk by children across urban and rural regions 7.15 Frequency of consumption of milk by children across different settings 7.16 District-wise frequency of milk consumption by children 7.17 Consumption of various commercial child foods by children 7.18 Consumption of commercial child foods by children across different settings 7A.1 Milk consumption frequency by children in urban and rural Odisha (wealth category) 7A.2 Milk consumption frequency by children across household type 7A.3 Consumption of commercial child food by children across household type 7A.4 Consumption frequency of millets by children 7A.5 Consumption frequency of green leafy vegetables by children

216 217 218 218 219 219 220 222 222 227 227 228 229 229

Contributors

R.V. Bhavani is Director of the Agriculture, Nutrition, Health Programme at the M S Swaminathan Research Foundation (MSSRF), Chennai. Rural development, food and nutrition security and leveraging agriculture for nutrition are the focus areas of her work. Between 2004 and 2006, she was on deputation to the National Commission on Farmers, Government of India, as officer on special duty to the chairman. Before joining MSSRF in 2000, Bhavani worked as an officer in State Bank of India for a decade. She has an MPhil degree in planning and development from Indian Institute of Technology Bombay and a PhD in economics from Madras University. Anusha Ganapati was an MSE student of Batch 2018. Anusha is currently working as a senior strategic analyst at an international bank. During her graduation at MSE, she was particularly interested in working on development economics and public policy and participated in related projects through conferences and internships. She is also interested in the subjects of innovation and data science. Getsie Immanuel currently works as a research analyst at the International Food Policy Research Institute (IFPRI) in the department of Poverty Health and Nutrition. Prior to IFPRI, she has experience working with the government and other research institutes. She completed her Master’s in applied economics from the Madras School of Economics, Chennai. G. Naline is currently a research scholar under the UGC-Junior Research Fellowship at the Madras School of Economics working broadly on the topic of child health. Before this, she did her post-graduation of M.A. economics at Loyola College, Chennai, and worked on the dissertation topic of socio-economic determinants of salt workers in Villupuram and also worked as the project coordinator at the Centre of Labour Education and Development from 2011 to 2012. Rohit Parasar is a consultant (statistical work) at the Ministry of Women and Child Development, Government of India. A post-graduate in economics from the Madras School of Economics, he worked as a research fellow at the M S Swaminathan Research Foundation prior to this. His research interests include child nutrition, gender, and food consumption and agri-food markets in India.

xiv Contributors S. Raju is a scientist (social science) in the Agriculture, Nutrition, Health Programme at the M S Swaminathan Research Foundation (MSSRF), Chennai. He holds a Master’s in economics from Bharadhidasan University, Trichy, and has 13 years of experience in working on various projects in the area of development research. Raju is adept at designing survey tools and conduct of primary surveys, managing databases, and analysis of both primary and secondary data sets. He is experienced in decoding and processing unit record data of the National Sample Survey Organisation (NSSO) and the National Family Health Survey (NFHS) and use of statistical tools. Swarna Sadasivam Vepa is a research consultant and visiting professor at the Madras School of Economics (MSE), India, involved in consultancy, teaching and guiding dissertations. She teaches development economics, agricultural economics and macroeconomics courses to graduate students. She was a fulltime research consultant on social inclusion at the Centre for Economic and Social Studies, Hyderabad, and a full-time consultant on Leveraging Agriculture for Nutrition in South Asia project of UK AID. She has many publications in international and Indian journals, and has published a book, Bearing the Brunt, on women in agriculture, and several book chapters, in edited volumes. Research interests include dryland agriculture, food security, social inclusion and gender issues. She received her PhD from the Department of Economics, Delhi School of Economics, India. Brinda Viswanathan is a professor at the Madras School of Economics, Chennai, and teaches courses in Indian economic development, development economics and quantitative economics for post-graduate students. Her research interest is in the broad area of development economics and applied econometrics. She has an edited book on agriculture and institutions and contributed chapters in several edited books on nutrition security, green economy, migration, poverty and vulnerability. She has published in national and international journals on undernutrition among children and women, gender and the labour market with some of this research funded by international organisations and government agencies. She regularly contributes as a resource person for workshops and training programs for college teachers, PhD students and government officials on statistical and econometric techniques and evidence-based policy-making for India’s development.

Foreword

India has achieved an impressive economic transition since the 1990s due to major economic reforms initiated in 1991. Despite high population growth, the per capita income has been raising the overall living standards of the masses. The literacy rate and life expectancy levels have also improved with expansion of the education and health facilities. Because of significant investment in agriculture and irrigation development programmes, and the spread of Green Revolution technologies in the pre-transition period, India was able to achieve food self-sufficiency. India is now able to export food products and maintain a large stock of foodgrains and implement a nation-wide public distribution system. Though economic growth resulted in a substantial decline of poverty and improved food supply situation, it has not translated into commensurate improvements in food energy intake or well-being. Calorie intake of persons including the poor leveled off even before calorie needs were met. The consumption basket of the poor is getting diversified; shifting away from cereals to non-cereals within the food group; and from coarse cereals to fine cereals within the cereals group. Given the ongoing changes in consumer preference, a substantial expansion of incomes of the poor is essential for tackling the food gap of the poor. While foodbased interventions play a supplementary role, pro-poor growth alone can eliminate chronic food insecurity in the long run. It is of interest to know whether the diversification of consumption basket reduces the micronutrient deficiency. So long as the changes in dietary preferences improve the nutrition and health status, even though they may not add calories, it should not be a cause of concern. India faces the problem of child malnutrition despite progress in food production, disease control, and economic and social development. Recent estimates show about 40% of the children under five years of age were suffering from stunting. It means that not only do they not achieve full genetic growth potential but also are exposed to a greater risk of child mortality. Malnutrition, by virtue of its synergistic relationship with infectious diseases has a powerful impact on child mortality in India. Stunted children end up as adults with small body size and reduced life expectation. Further, such adults have low labour productivity. Studies identify malnutrition as a major cause of death in developing countries. Apart from inadequate food consumption, the other important causes of malnutrition are high incidence of gastro-intestinal and respiratory infections and behavioural

xvi

Foreword

factors such as faulty breastfeeding, weaning practices, etc. These factors contribute to the low absorption of nutrients from the food consumed. Professor Sukhatme explained that the conversion of food intake into energy depends on access to safe drinking water, health care and environmental hygiene. Econometric analysis of National Family Health Survey household data showed standard of living, proxy for household income, reduces the risk of malnutrition. Mother’s health status and education level have a favourable effect on the nutrition status of a child. Institutional deliveries also reduce the risk of malnutrition. The risk of child malnutrition is higher when the mother is working. Perhaps, childcare suffers when mother is working, particularly in the case of poor households. In the case of working women, provision of childcare during her absence at the time of work may partly overcome the problem. Since a large number of the poor depend on agriculture for their livelihood, achieving the goal of poverty reduction as well as inclusive growth depends on the improvement of agricultural productivity and processes that facilitate migration of agricultural workers to the rural non-farm sector by diversifying the sector. These will contribute to the diversification of employment opportunities as well as household income. This has been the process of transition towards an industrial economy like many East and Southeast Asian countries, which experienced a sharp reduction in poverty. India needs an agricultural growth rate of 4.0 to 4.5% to reduce poverty and food insecurity. Agricultural development would diversify into dairying, animal husbandry, fisheries, floriculture, horticulture and other areas to achieve such growth rate. These activities are likely to be labour intensive. If flexibility on the supply side is facilitated, production will adjust to the market forces and generate higher incomes in the rural areas. This would also spur the growth of agro-processing industries in rural areas. Overall, there would be a substantial increase in the incomes of the poor which improve the entitlement of food. Diversification of production systems, enhancement of land productivity, improvement of market efficiency, and market supply of diversified foods at affordable prices are crucial for elimination of rural poverty and undernutrition. Mechanisms involved in reducing undernourishment are contextual and nuanced. This book explores the agricultural and undernourishment link in three chapters. Other chapters study the mechanisms of improving the nutritional status of women and children in varying contexts. R. Radhakrishna Chairman, Centre of Economic and Social Studies, Hyderabad, India

Preface

The second sustainable development goal (SDG-2) is to “end hunger, achieve food security and improved nutrition and promote sustainable agriculture.” While it has long been recognised that agriculture and nutrition are inextricably linked through complex pathways, the empirical evidence in India has often suggested otherwise, leading some to label it the agriculture-nutrition “disconnect.” Adapting a conceptual framework developed by UNICEF several decades ago, research has identified several pathways that mediate the relationships between agriculture and nutrition. For example, agriculture is a source of food and of income and livelihoods; improved productivity therefore is expected to translate into better nutritional outcomes. Transmission via this channel depends, however, on how food and income are allocated within the household across age (children versus adults) and gender. This allocation, and the extent to which it results in equitable and healthy outcomes, is also influenced by women’s decision-making abilities and their socio-economic status within the household. Further, a woman’s own nutritional status is a function of her access to a diverse diet, either directly from on-farm production, or through the market at affordable prices. In turn, healthier adults are also more productive in the labour market. Since the agricultural sector is also the single largest employer of women, their participation in the labour force is expected to feed into a virtuous cycle of higher incomes, agricultural productivity and family health, provided children have competent substitute caregivers. Disentangling the extent and strength of these causal pathways in empirical analyses is a formidable challenge. The chapters in this timely book contribute to our understanding of several of these pathways. They rely on both aggregate (district-level) as well as householdand individual-level data to illuminate key relationships. Women’s education and public provisioning emerge as key to ensuring better health outcomes; findings that are robust across data sets and chapters. While I do not propose to further summarise their results, it is important to highlight that the ambit of the book is also wider than what is commonly seen in the literature. Two features stand out: first, rather than treat rural and urban as dichotomies, the analysis treats these as a continuum and documents the spillovers that occur in both directions in the form of peri-urban poultry and dairy markets

xviii Preface and non-farm employment, and the opportunities these represent for better transmission of linkages. A second feature is the focus on inequalities, particularly in access to health opportunities, and how these are circumscribed by social conditions on the one hand, and enhanced by public provisioning on the other. In both these respects, this book breaks new ground. As India rapidly undergoes a nutrition transition and the policy and research focus shifts from concerns of undernutrition to those related to overweight, obesity and associated non-communicable diseases, it is easy to forget that the burden of childhood stunting and wasting, and chronic energy deficiency and anaemia among women, particularly in rural areas, remains large. The research presented in this book not only represents a welcome addition to the literature but provides concrete action points, backed up by evidence and analysis, for the design of appropriate policy interventions. I commend the authors and editors for this work. J.V. Meenakshi, Professor, Department of Economics, Delhi School of Economics, University of Delhi, Delhi

1

Introduction Nutritional well-being of women and children in India Swarna Sadasivam Vepa and Brinda Viswanathan

The pace of economic growth for over two decades in India has not resulted in economic development in a broad sense. The 2019 Global Hunger Index (GHI) ranked India at 102 out of 132 countries, with several African and Asian countries ranked higher than India. In comparison to this, India ranked 77 among 190 countries in the ease of doing business (DB) index by the World Bank. The former report pertains to measures that assess very basic aspects of sustenance and life while the latter report measures country-level features that will enable economic growth and prosperity. The DB report includes the wealthy nations as well, and among them India ranked high while the GHI report has developing and less developed nations and yet, India’s ranking is low. Consequently, several studies that focus on understanding the causes behind this disconnect between different aspects of development have emerged. The global evidence in general is that pro-poor growth and growth of agriculture reduce poverty with substantial reduction in mortality and undernutrition among children. However, if economic growth is attributable to a small sphere of economic activity like the modern services sector as in the Indian context, then it results in regional imbalances in economic development. The regions and population dependent on agriculture have been adversely affected the most, the women and children. Agricultural growth slowed down long back and then followed by slowing down of manufacturing growth in the more recent decade, resulting in limited opportunities for migration from the traditional to the modern sector. Despite the public policy discourse being dominated with the view that high and sustained economic growth results in overall economic development, strong evidence supports that a better educated and a healthier population results in a higher rate of economic growth. The chapters in this book provide more evidence for pathways to nutrition security with focus on the agricultural linkage and other supporting environment. More importantly, the discussions emphasise the role of public provisioning of the goods and services that ensure that socially disadvantaged segments of the population including women have a nutrition secure and healthy life for themselves and their children. Well-being means we must make sure that everyone is healthy and happy. Happiness is subjective well-being and health is physical well-being. One must earn enough to afford a decent living to be happy and healthy. If anyone is sick, poor, lives in dirty surroundings, is not treated fairly, or is caught in dispute, that person will be unhappy. Having access to decent livelihood, eating enough in terms of

2 Swarna Sadasivam Vepa and Brinda Viswanathan quantity and quality and living in a clean environment, not facing unfair social, locational and other circumstances will keep one healthy. Sustainable development is all about making sure that everyone could afford a healthy living. Sustainable development goals (SDGs) include dimensions that concern living well now without destroying our capacity to live well tomorrow—the essence of sustainability. The goals encompass several aspects, including subjective happiness and reducing inequality, which were not part of millennium development goals. Each goal has specific targets and a set of measurable indicators to track progress. Sustainable development goal 2 is about ending hunger—ending deficient calorie intake by men, women and children. It is about achieving food security—making everyone have enough to eat today, tomorrow and in the foreseeable future. It is about producing enough diversified food at affordable prices today without degrading the environment. The three main targets set by the United Nations for 2030 are as follows: 1 2 3

End hunger. Provide safe, nutritious and adequate food for all. End all forms of malnutrition, among children, adolescents, women and old. Double agricultural productivity and the farm incomes of small producers.

The target set for child nutrition is to reduce the current level of stunted and wasted children by half before 2025. Ending hunger also means removing energy deficiency among women, men and children, and ensuring nutritional improvements to achieve the intergenerational growth potential. In other words, it is about improved nutrition—being well nourished and not face growth faltering by children and growth disorders by adolescent and adults. Children need to grow well, gain enough weight for age, be tall enough for their age, and not be too thin or too fat. Similarly, adolescents and adults and especially women must eat adequate diversified food, grow as tall as one could grow, and not be too thin or too fat. Doubling of agricultural productivity ensures better livelihoods for those engaged in agriculture, helps non-farm economy through its spillover effects, reduces poverty and improves affordability of nutritious food. Sustainable development goal 1 is about eradicating poverty and implementing social protection measures. Sustainable development goal 10 is about reducing inequality, aims at increasing the growth rates of the lowest 40% and promoting economic, social and political inclusion. Agriculture provides food and livelihood and is known to reduce poverty (De Janvry and Ravallion, 2009). Slow growth or agricultural stagnation, deceleration in poverty reduction, social exclusion in terms of access to opportunity limit progress in nutritional outcomes. Agriculture nutrition linkages are at the heart of zero hunger goal.

1.1 Past performance and future expectations under SDGs To reach the goals, we should know how far away we are from the targets. Data can tell us about this. But to make sure that we get to the goal, we must understand linkages and mechanisms that help us to get there. We should also know hurdles

Introduction 3 that prevent us from moving forward. We should pause to look back to see how we performed, to assess our capacity to move forward. Looking back tells us the policy atmosphere in which we were operating and the policy atmosphere desirable for accelerating the progress. According to United Nations Development Program, 195 million, constituting about 14% of the population, are undernourished in India. They include women and children. Every third child in India is stunted. About 53% of the women are anaemic. About 80% of women and children do not get enough dietary diversity. About 40% of India’s population depends upon agriculture. Poverty declined, but about 25% in the rural areas are still poor. However, undernourishment is a bigger problem than poverty and encompasses both rural and urban areas; children and women face the problem more than men. We are expected to reduce poverty to lift everyone above the level of US$1.25 a day. We are expected to reduce by half our child stunting rates to a figure below 20%. An even more difficult task is to reverse the present trends of wasting among children under five, which has been increasing in the past two decades from 15.5% to 21% at present. Then we need to reduce wasting in children to 10%. The neo-natal (below the age of 28 days) mortality rates that stand at 24 per thousand should be halved. All these SDG targets will have to be realised between 2025 and 2030.

1.2 Theory of change Theory of change is the roadmap to reach a desired goal. A theory of change is more relevant for an impact evaluation study. If there is an agreement about the current situation and the desired situation, theory of change is about how to reach the goal. It is about the limitations faced at present and how a policy or program change or behavioural change can potentially remove the hurdles (Rogers, 2014). Several organisations and authors who work in the area of development give a simplified diagram to indicate interlinkages between various aspects and the desired outcomes. It is not easy to simplify the complex nexus between nutrition outcomes and other proximate and distant causes. All the same a simplified diagram, followed by a simple explanation of the theme of the book makes it easier to comprehend the context. More importantly, it has the advantage of displaying clarity of thought and purpose of the exercise. Different contexts group together the very same aspects differently. The UNICEF terminology differentiates nutrition specific factors and nutrition sensitive factors. Nutrition specific factors are related to food and nutrient intake of children, adolescents, expectant and lactating mothers, patients and adults. Nutrition sensitive factors, are those aspects that indirectly influence nutrition, such as agriculture, water, sanitation and health, social exclusion, poverty, illiteracy and so on (Unicef, South Asia, 2016). Our terminology, approach and priorities are different. We found evidence that some factors are crucial for reducing undernutrition. These factors are already recognised in the literature as important. The book is building

4

Swarna Sadasivam Vepa and Brinda Viswanathan

upon the existing framework and goes one step further to add to the evidence. For the sake of simplicity, we broadly categorise factors influencing undernutrition into four broad groups: household specific innate capacity, agriculture, public provisioning and social opportunity. An intuitive simple thematic diagram, in Figure 1.1, illustrates the influence of agriculture, public provisioning and social opportunity on child and adult nutrition. Household’s innate capacity to deal with deficient diets and growth disorders determines the overall undernutrition size. The other three factors, agriculture, public provisioning and social opportunities, help and enhance capability of the households. Then, households make sure that children grow well with the right height and weight for their age and adolescents and adults grow as tall as they could, and not too thin and not too fat. The higher the effect size of each of these three factors the lower will be inadequacy of diets and growth disorders. Each generation will be taller than the previous one. In Figure 1.1, three concentric circles—one inside the other—stand for agriculture, public provisioning and social opportunity. The small inner circle means smaller influence and the big outer circle means bigger influence. They surround the middle oval shapes one inside the other, which stand for diet deficiencies and growth disorders together named as undernutrition, the bigger oval is the existing level and the small inner oval is the desired goal. When the influence of agriculture, public provisioning and social opportunity is small they touch the outer oval shape. As the effect size of these three aspects increases, the circles become bigger and they push the oval shape, making it shrink to the small oval shape in the centre. Agricultural productivity, public provisioning and social

Agriculture

Households Undernutrition Und dernutri de ri on rition

Public provisioning

Social Opportunity

Figure 1.1 Theory of change diagram: Effect of agriculture, public provisioning and social opportunity on undernutrition

Introduction 5 opportunity show negative relationship with growth faltering. These factors essentially build the capacity of the households to eliminate undernutrition at its core. The representation of households as overlapping circles conveys the idea that it is related to the other three factors. The policy should strive to enhance capacity of households to deal with undernutrition through interventions that increase the effect size of agriculture, public provisioning and social opportunity to reach the SDG goals. The book puts together chapters written as stand-alone papers, but they do have the same theme of undernutrition among children and women in India. The book assesses the problem of deficient diets and growth faltering and the factors that influence and help reduce them. Leaving introduction and conclusion chapters, there are six analytical chapters in the book. Out of these six chapters, three chapters (2, 3 and 4) broadly explore the influence of agriculture on nutrition of children and women. They also bring out the importance of women’s education and public provisioning. The other three chapters look at the influence of social opportunity, opportunities missed, disadvantages experienced due to being in caste groups, income groups or a remote location, though the focus of the chapters differ. All the chapters pinpoint the lacunae in public provisioning as one of the limitations to reduce undernutrition. Chapters based on household surveys look at the household specific aspects that determine the child and adult undernutrition.

1.3 Agriculture Agriculture provides livelihoods to rural as well as semi-urban population dependent on farm as well as non-farm activities, through its spillover effects. To capture the effect of diversified agriculture, the book defines agriculture as crop plus livestock production (dairy, poultry and other livestock products such as meat). Nutrient-dense foods are important for reducing malnutrition among children and adolescents. Past literature has shown that agricultural productivity and rural economic growth reduce poverty (Datt and Ravallion, 1998, 2002, Ravallion and Datt, 1996). We go one step further and ascertain that agricultural productivity reduces child underweight and prevents chronic energy deficiency among women. Apparently, it seems to refute the findings in the literature that shows those dependent on agriculture get lower levels of income and consumption (Chand et al., 2017) and higher levels of undernutrition (Bhagowalia et al., 2012). The fact is, agricultural productivity happens to be low in the areas of low income (consumption) and high undernutrition. As we have shown in Figure 1.1 the effect size of agriculture is small. When the effect size grows big those in agriculture benefit, as witnessed in Kerala and Punjab. In these states according to the national sample survey data 2013, in rural areas, those self-employed in agriculture show higher average income and consumption than those in non-farm activities (Radhakrishna and Raju, 2015). Impact of agriculture on child nutrition is significant but not enough to reduce it without the presence of other factors. Existing capacity of the households and

6 Swarna Sadasivam Vepa and Brinda Viswanathan the capacity enhancement through public provisioning and social opportunity will make a bigger impact. Both in Kerala and Punjab the child undernutrition is low, probably due to the synergy created by various factors. The first two chapters make two important points. First, in the twenty-first century when the contribution of agriculture to total GDP has fallen drastically, and when about 40% of the rural families derive income from non-farm sources, rural is not synonymous with agriculture and semi-urban is not synonymous with nonagriculture. Non-crop agricultural production such as poultry and dairy takes place in semi-urban areas. Second, diversified agriculture has large spillover effects on rural as well as semi-urban areas with opportunities created in the non-farm sector. Agricultural productivity growth reduces underweight children across the whole district. If we can harness the enhanced productivity of agriculture for non-farm livelihoods in processing, undernourishment would decline. Chapter 2 shows with the help of data that agricultural land productivity depends on rainfall, irrigation, higher levels of commercialisation, surplus foodgrain production and education levels of population. Agro-climatic conditions also influence agricultural land productivity. Chapters 2 and 3 provide evidence that districts with higher agricultural land productivity and agricultural worker productivity show lower underweight rates. Chapter 4 finds that in rural areas of India, nutrition insecurity as measured by low values (18.5 and below) of body mass index (BMI) in cultivator households is lower (higher) than those in wage labour households in 2005. However, in 2011 we observe that this has changed. The CED rates between cultivator and agricultural labour households and between agriculture and non-agricultural labour households are not different. Chapter 6 shows that in the age group of 20 years and above the BMI differences are not significant across social groups for men and women, after controlling for household amenities, land ownership category and other household characteristics in the agriculturally prosperous state of Andhra Pradesh as per the DLHS-4 data. Chapter 7 shows that production and consumption of milk, crucial for child growth, is low in the state of Odisha. This again reiterates our earlier conclusion that higher levels of local agricultural production leads to higher levels of food intake by all in the households.

1.4 Public provisioning Public provisioning is crucial for the reduction of undernutrition. When the households cannot get clean water, sanitation and health care facilities, referred to as WASH by themselves, public provisioning is useful. Private sector on occasions provides essential services as well. The service quality in the private sector may differ from one area to another; public services assume importance when it is a monopoly service to a majority of population such as water supply and supply of electricity, sanitary services, sewage disposal and so on, or a targeted service to the deprived sections of population. Targeted public provisioning takes the form of asset distribution (land and housing), food distribution, free services related to health, conditional and unconditional cash transfers and so on. The coverage of

Introduction 7 population and quality of services differ widely between states and districts within the states, based on the administrative efficiency. Studies on nutrition and health invariably include either coverage of the public sector provision or an indirect outcome of effective provisioning. All the six chapters use public provisioning related variables to explain nutrition outcomes. Chapter 2 shows that the off take of foodgrains from the public distribution system by the state government, percentage of children receiving complete immunisation, and the existence of any government health facility have a significant negative association with the percentage of underweight children in the district. This is a clear indication of importance of public provisioning. Chapter 5 developed a child health opportunity index across different geographical regions which is a measure of the coverage rate of an opportunity discounted by inequality in its distribution caused by various circumstances. The study concluded that 83% of the children are stunted because they were denied the opportunity of integrated child development services. Chapter 6 looks at the body mass index of adult women and men and its determinants across social groups in Andhra Pradesh and Telangana, which were bifurcated in 2014, after following the same policies of public provisioning for decades. Health insurance coverage seems to have a positive impact on women’s body mass index in India, with many households deprived in some aspect or other depend upon public provisioning. Hence, it is imperative to ensure coverage of good quality services to reach the SDG goals.

1.5

Social opportunity

The options that are available to a person depend upon social circumstances, public policy and institutions (Dreze and Sen, 1999). The opportunities which are influenced by social circumstances need attention. The book considered some of them in course of the analysis. Ability of households and individuals to access an opportunity available could be constrained by the social circumstances such as social group, economic group, literacy group, age group, land possession group, gender, illness and so on. The influence of these constraints in achieving the desired nutrition outcome has been the focus. For example, third chapter found the percentage of children who reported diarrhoea in the district increased the number of underweight children and the percentage of women who achieved education above secondary level significantly reduced the underweight rates. Elasticity of child underweight with respect to women’s education was quite high at 0.45, which means a 1% improvement in the proportion of educated women (15–49) would reduce the child underweight by 0.45%. Similarly, if the circumstances of incidence of diarrhoea can be prevented, child nutrition improves. A 1% increase in diarrhoea incidence decreases the proportion of underweight children by 0.9%. In terms of numbers, the effect is bigger. Specific interventions enhance access. For example, providing incentives for doctors to work in rural areas may improve the availability of health care in rural areas and the primary health centres. The entire Chapter 5 is about the inequality in such social opportunities across the states. The chapter looked at several sets of circumstances that differ across

8 Swarna Sadasivam Vepa and Brinda Viswanathan households and found that the inequality of opportunity is quite low in the southern states. Chapters 6 and 7 specifically look at the circumstances of belonging to a social group and the consequent loss of social opportunity with respect to nutrition. Chapter 6 shows the examples where such social group disadvantage was overcome with respect to men’s nutrition, but women’s nutrition gets affected in some states. Similarly, unequal intergenerational improvement in heights shows missed social opportunity of scheduled tribes. Constraints to social opportunity can be removed by conscious efforts. Chapter 7 shows the lost social opportunity of scheduled caste population to feed more milk to their children despite the importance of milk consumption for underweight children. In general, the intersection of caste, class, gender and location increases the vulnerability, and policy should target reduction of such inequality to reach the SDG targets. It is different from public provisioning, as the effort here is to equalise the opportunity rather than simple provisioning.

1.6 Overview of the book1 Having discussed the general theme of the book, the book does not claim to have understood or explained mechanisms through which the process works. It is because the bottom line is to enable the households’ capability to reduce undernutrition. It can happen in different ways under different circumstances. The purpose is to provide information in a systematic and thought-provoking manner to the research scholars. The aim is to tilt the attention of the researchers to less explored areas in nutrition research. We hope that the book would help the research scholars in the area of agriculture or nutrition to design their research agenda with reference to the gaps that exist in public policy. Chapter 2 of the book introduces the topic of agriculture nutrition linkages and provides empirical evidence for the same using 2 SLS methodology. The nutrition data and agricultural data of the District Level Health Survey (DLHS-2 for 2002–04) forms the basis of the analysis. Relationship between agricultural productivity and child underweight or this period was shown at the state level by other studies (Headey et al., 2011; Gulati et al., 2012) in simple cross section analysis of the states. Our earlier state-level study using panel data analysis (Vepa et al., 2014) has also established such a relationship. There was no research beyond state level due to paucity of data. The household-level surveys did not indicate any link earlier. This chapter probably is one of the very few studies that established such a significant link at the district level. Chapter 3 of the book builds upon Chapter 2, combining DLHS-2 data with more recent NFHS-4, GDP data from Indicus and agricultural data for two periods. Three estimations, viz., of single period cross section, mixed effects panel data maximum likelihood estimation and panel data fixed effects estimation, assess the relationship of child undernutrition with agricultural land productivity and agricultural labour productivity and find them significant. The elasticity of agricultural productivity with respect to child underweight seem to be low, but the elasticity of women’s education is high. Even low elasticity can make a big impact at margins.

Introduction 9 Chapter 4 explores the agricultural and non-agricultural pathways for improving women’s nutrition outcome of low BMI. Women’s status in general and mother’s nutritional status are among several factors that have been shown to affect child undernutrition. In this context it would be useful to document empirical evidence that analyse women’s or for that matter adult nutritional status. Arresting intergenerational transmission of undernutrition would contribute to a quicker pace of reduction of child undernutrition which is notoriously at a high level for India. Due to the availability of country-wide data in India for a fairly long period of time on nutritional intakes, empirical studies on trends and patterns in nutrition intakes, and the factors determining its variations have been studied more extensively (Viswanathan and Meenakshi, 2008; Sharma et al., 2015). A major difference between intake and outcome indicators in India is that the former is usually based on household-level information while the latter is for individuals. Anthropometric indicators like height (for age), weight (for age) and body mass index (BMI) are common indicators of nutritional outcomes at the individual level. In most of the studies exploring the role of farming or agriculture in improving nutritional status, nutritional outcome based on young children is more common, compared to older children, or adolescents or adults (Carletto et al., 2015; Ruel et al., 2018). Among adult nutrition outcome measures, BMI is a useful indicator of wellbeing as there are scientifically given normative thresholds that enable a person to be classified as undernourished (BMI below 18.5) referred to as chronically energy deficient (CED). By studying women’s BMI, we can focus on the individual well being of women and women’s empowerment, an area of immense importance in the South Asian context. The individual-level variables relating to empowerment, education and health seeking behaviour play an important role in reducing the risk to CED. Environmental factors like sanitation, drinking water quality and smokefree kitchen atmosphere are also enhancers of nutrition outcome. A household’s economic status continues to play a role in reducing CED risk. The role of public policy lies in ensuring that the basic amenities are accessible to the poor, including health system delivery, while creating awareness on empowerment, though a challenging and slow process, must be systematically undertaken. Chapter 5 provides a quantitative assessment of the gap in two different aspects of child health outcomes using the National Family Health Survey, 2005–06. The empirical approach is based on measuring the overall coverage of an opportunity after discounting for the inequality in its coverage and is the same as Human Opportunity Index (Barros et al., 2009) but is referred to here as Child Health Opportunity Index. These indices are then compared across six geographical regions of India. The two child health opportunities chosen for analysis here are neonatal survival and attainment of age-appropriate height among children aged five years or below. The concept behind the opportunity index was suggested by Roemer (1998) wherein the differential circumstances that children face leads to the inequality in opportunity for a neonate to survive or for a young child to not miss out on a minimum linear growth. Circumstances are classified into four levels—child, maternal, household and public policy or state intervention. The child level variables are gender, birth order,

10 Swarna Sadasivam Vepa and Brinda Viswanathan birth size and birth interval; maternal variables are height and body mass index, education, breastfeeding practices, domestic violence faced by her, dietary diversity and mother lost a child; public policy or state program are tetanus shots, antenatal care, consumption of iron folic acids and place of delivery, Integrated Child Development Scheme; and household-level variables are wealth, religion, use of clean cooking fuel, safe drinking water and prevalence of open defecation and rural/urban place of residence. The analysis in this chapter uses the inequality of opportunity index to show the regional differences in child health for India. With NFHS-3 data set, the result shows that inequality exists for stunting and neonatal mortality rates across the broad geographic zones of India. For the neonatal mortality, not much variation is observed across states or across circumstances. Central zone, comprising of states like Uttar Pradesh, Chhattisgarh and Madhya Pradesh, exhibits the largest inequality compared to other zones and south zone shows the best opportunities and least inequality. Chapter 6 of the book analyses the body mass index and heights of adult women and men above the age of 20 years, across social groups, after controlling for household amenities index, land ownership, land categories, age, education etc. with the help of District Level Health Survey-4 for 2012–13, for the states of Andhra Pradesh and Telangana in India. The results show that the impact of social group on body mass index of men and women has been muted. Social group has no influence on men’s BMI both in OLS as well as quantile regressions, without an exception after controlling for household and individual characteristics. This is probably due to a more effective public distribution system which provides basic calories to the poor at affordable prices. BMI of women is higher in some social groups compared to the base category of other castes in Andhra Pradesh, pointing to the problem of obesity among women. BMI quantiles for women show a significant negative influence of social groups on BMI, though it was not apparent in OLS. Social group being insignificant for men and significant for women is a pointer to intra-household discrimination of women in poor households. Land possession levels adversely affect women’s BMI but there is no influence on men’s BMI, further pointing to a possible intra-household discrimination. Discrimination by caste is quite apparent in heights of adults. All the same inter-generational catching up is apparent in all quantiles, more so among men. Nutritional outcomes are determined by several factors, but household amenities appear to be an important factor that could equalise the nutritional outcomes. In contrast to the DLHS-4 data analysis for adults above the age of 20, that shows social group as insignificant, the analysis of the NFHS-4 data for the age group 15–49, shows that social groups negatively influenced heights and BMI of men and women in Andhra, but influenced only women’s heights and body mass index in Telangana. This points to the differences in the caste composition of the two states after bifurcation. Chapter 7 of the book looks at the consumption of milk and milk products. Milk and milk products are important for children’s growth (Choudhury and Headey, 2018). Odisha is a state deficit in production and per capita availability of milk. Unit level data was extracted from the NFHS-4 (2015–16) round and the diets

Introduction 11 of children in the age group of 6 to 59 months of age and their nutrition status examined with reference to consumption of milk. To further understand the consumption of milk and milk products by children in the age group of 6 to 59 months in poor households, a survey of 400 households was conducted in three districts. The analysis of NFHS-4 data shows an association between consumption of milk and social groups and wealth classes and the consumption of milk and prevalence of undernutrition in the state. It was found that the social group having the lowest percentage of children who consumed milk was the Scheduled Tribe (ST) in rural Odisha at 11%, followed by Scheduled Caste (SC) at 17%. In urban areas also milk consumption was lowest among the ST and SC groups (23.5% and 22.8% respectively). A significant positive association was found between consumption of milk and underweight z-score of children. Analysis of primary survey data revealed that more than half the households did not feed children with milk or milk products; only 34.5% fed children with milk daily. Variation was found across social groups and households’ standard of living, in line with secondary data. Overall, this book has put together evidence that highlights the negligence of potential livelihoods from agriculture, weakening public provisioning and widening inequality of opportunity that is perhaps preventing India from achieving a faster decline in undernutrition, despite faster economic growth. In the six chapters, starting from second through seventh, the findings highlight that agriculture has links to nutrition, and that social circumstances and social group could influence nutrition outcomes wherein public provisioning becomes very crucial to redress social disadvantages. While some of the links are intuitive, it is hard to find empirical evidence and conclusive proof of these linkages. It is even harder to explain the mechanism through which they operate to influence the nutrition outcomes. The chapters do not evaluate the public programs and policies per se but look at policy implications arising out of the empirical evidence. It is not a simple story telling, but a difficult effort to produce convincing evidence.

Note 1 The editors would like to acknowledge the useful suggestions given by several scholars during their earlier publications on the topic of linkages between agriculture and nutrition. We are particularly thankful to Professor M.S. Swaminathan, (late) Dr. Prakash Shetty, Professor A. Vaidyanthan and Professor R. Radhakrishna for the support and encouragement given to us from time to time. We are also thankful to Madras School of Economics for providing the facilities to complete the publication.

References Barros, R. P., Francisco, H. G., Vega, J. R. M., and Chanduvi, J. S. 2009, Measuring Inequality of Opportunities in Latin America and the Caribbean, Washington, DC: The World Bank. Bhagowalia, P., Headey, D., and Kadiyala, S. 2012, Agriculture, Income, and Nutrition Linkages in India. Insights from a Nationally Representative Survey, IFPRI Discussion Paper 01195.

12 Swarna Sadasivam Vepa and Brinda Viswanathan Carletto, G., Ruel, M., Winters, P., and Zezza, A. 2015, Farm-level pathways to improved nutritional status: Introduction to the special issue. Journal of Development Studies, 51, pp. 945–957, doi:10.1080/00220388.2015.1018908 Chand, R., Srivastava, S. K., and Singh, J. 2017, Changing Structure of Rural Economy of India, Implications to Employment and Growth, Discussion Paper, NITI Aayog, November. Choudhury, S. and Headey, D. D. 2018, Household dairy production and child growth: Evidence from Bangladesh. Economics& Human Biology, 30, pp. 150–161. Datt, G. and Ravallion, M. 1998, Farm productivity and rural poverty. Journal of Development Studies, 34(4). Datt, G. and Ravallion, M. 2002, Is India’s economic growth leaving the poor behind? Journal of Economic Perspectives, 16(3), pp. 89–108. De Janvry, A. and Sadoulet, E. 2009, Agricultural growth and poverty reduction: Additional evidence. The World Social Bank Research Observer, 25(1), pp. 1–20. Dreze, J. and Sen, A. K. 1999, India, Economic Development and Social Opportunity, New Delhi, India: Oxford University Press. Gibson, J., Gaurav, D., Rinku, M., and Martin, R. 2017, For India’s Rural Poor, Growing Towns Matter More than Growing Cities. Policy Research Working Paper; No. 7994. Washington, DC: World Bank. Gulati, A., Kumar, A. G., Shreedhar, G., and Nandakumar, T. 2012, Agriculture and malnutrition in India. Food and Nutrition Bulletin, 33(1), pp. 74–86. Headey, D., Chiu, A., and Kadiyala, S. 2011, Agriculture’s Role in the Indian Enigma: Help or Hindrance to the Undernutrition Crisis? Discussion Paper 01085, Washington, DC: International Food Policy Research Institute. Radhakrishna, R. and Raju, S. 2015, Well-being of agricultural households in post-reform period. In C. Ramaswamy and K. Ashok (eds.), Fast Growing Economy: Challenges, Strategies and Way Forward (pp. 151–174), New Delhi: Academic Foundation. Ravallion, M. and Datt, G. 1996, How important is to India’s poor is the sectoral composition of economic growth? The World Bank Economic Review, 25(1), pp. 1–25. Rogers, P. 2014, Theory of Change, Methodological Brief-2 Impact Evaluation, Florence: UNICEF Office of Research. Roemer, J. E. 1998, Theories of Distributive Justice, Cambridge, MA: Harvard University Press. Ruel, M. T., Quisumbing, A. R., and Balagamwala, M. 2018, Nutrition-sensitive agriculture: What have we learned so far? Global Food Security, 17, pp. 128–153. Sharma, A., Yadav, A., Baig, V., Swarnkar, M., Singh, R., and Kumar, S. 2015, Malnutrition & associated risk factors among under five children. Indian Journal of Community Health, 27(3), pp. 311–319. Vepa, S. S., Uma Shankar, V., Bhavani, R. V., and Parasar, R. 2014, Agriculture and Child Under-Nutrition in India: A State Level Analysis. MSE Working Paper 86/2014, Chennai: Madras School of Economics. Viswanathan, B. and Meenakshi, J. V. 2008, Changing pattern of undernutrition in India: A comparative analysis across regions. In S. S. Acharya, B. Davis, and B. Guha-Khasnobis (eds.), Food Security: Indicators, Measurement, and the Impact of Trade and Openness, UNU-WIDER Studies in Development Economics, New Delhi: Oxford University Press. UNICEF, Stop stunting in South Asia. 2016, http://stopstunting.org/wp-content/uploads/ 2016/05/StopStuntinginSouthAsia-ACommonNarrativeonMaternalandChildNutrition_ UNICEF.pdf

2

Child underweight and agricultural land productivity Swarna Sadasivam Vepa, Rohit Parasar and Brinda Viswanathan

2.1

Introduction

The topic of child nutrition assumes importance and the need to reduce malnutrition assumes urgency in view of the sustainable development framework adopted in July 2017 by the United Nations and targets set for the year 2030 worldwide. In this context, data on 44 child-related indicators identified have been analysed in the report titled “Progress for Every Child in the SDG Era,” released in March 2018 (United Nations 2018) The revised data brochure in 2019 looks at the progress made across nations, within the limitations of available data. As per the report “the world will meet neither the 2030 SDG Target to halve the number of stunted children nor the 2025 World Health Assembly target to reduce the prevalence of low birthweight by 30 percent.” Child malnutrition has been a globally well-researched area. Over years, the key areas of reducing child malnutrition were identified. Diversified optimal diet, nutrition knowledge, and feeding practices are intuitively important for child nutrition. “Malnutrition has many different causes working at different levels. Access to water, sanitation and hygiene, income, education and quality health services are all important” (Global Nutrition Report 2018). Child undernutrition is a larger problem than poverty. Poor nutrition in the first 1,000 days of a child’s life from conception can lead to underweight and stunted growth. Maternal health of expectant and lactating mothers plays an important role. Stunting is associated with impaired cognitive ability and reduced school and work performance in later life. Underweight in children under the age of five substantially increases risk of mortality (WHO 2010a). Waterborne and airborne infections can create illness and deteriorating nutritional status (WHO 2010b). As pointed out by 2018 Nutrition report, improved data has not only made us recognise the location in which prevalence of child undernutrition is the highest, but also improved the understanding of the intervention about what works and what does not. Agriculture provides food for all, and livelihoods to a majority of the population in developing countries, including women. Agricultural growth was recognised as the single most important factor that has contributed to rural and urban poverty reduction, in the twentieth century (DFID 2004). Poverty reduction and higher incomes contribute to child nutrition. Thus, agriculture-nutrition link found a place in the development discourse.

14 Swarna Sadasivam Vepa et al. Several papers were published since 2000 on agriculture-nutrition link. The World Bank report in 2007 suggested five pathways that connect agriculture to nutrition. Four of them are related to agriculture and one is about women’s empowerment as agents of household food security. First, self-consumption of the agricultural produce becomes important for nutrition, when markets are less developed. Second, when most of the poor farmers and agricultural labour are net consumers, the nutrition benefit of agriculture comes through high earnings (wages) and low food prices. Third, low food price typically helps all poor households, rural or urban. Fourth, indirect contribution of agricultural productivity growth lies in spillover effects that generate incomes (and improve nutrition) in non-farm activities related to agriculture. Food and agriculture gained attention after the 2008 experience of depleted world food stocks and price hikes. The immediate thrust was on the importance of agricultural production by small farmers (Wolfenson 2013), and special safety nets (HLPE-4 2012) to ward off undernourishment and undernutrition. The significance of agriculture and nutrition was aptly captured by sustainable development goals (SDGs). The target is to end hunger in the world and all forms of malnutrition by 2030. The target is to reduce by half the number of stunted and wasted children under five by 2025. SDG-2 goal of zero hunger seeks doubling of agricultural productivity and incomes of small-scale food producers. This goal also seeks to enhance the opportunities of value addition and non-farm employment (United Nations 2018). In short, doubling of agricultural productivity and generation of farm and non-farm incomes should result in reducing stunting and wasting in children by half, if the link between agriculture and nutrition is strong. To achieve the set targets, elasticity of agricultural land productivity with respect to incomes and income elasticity of nutrition outcome should be high. Poverty reduction should also improve access to diversified food, clean water, sanitation, hygiene and health. Threshold level of income that improves access to household amenities and access to better health care tends to be higher than the income level set for poverty. Public provisioning of food and other public services could fill the gap. Women’s education at secondary level and above tends to be more effective in reducing child undernutrition. Understanding the mechanism through which agricultural productivity translates into reduced child undernutrition is fraught with data gaps and research gaps. From this perspective, agriculture-nutrition link remains a less explored area of research. Papers that reviewed the research on agriculture-nutrition links in Africa and Asia concluded that the evidence of significant impact of agricultural interventions on specific nutrition outcomes were disappointingly few. Lack of evidence does not mean lack of impact (Kadiyala et al., 2014). One of the problems appears to be in the design of the studies that try to capture the effect in randomised control trials (Webb 2014). International funding organisations, including Bill & Melinda Gates Foundation, US Aid and UK aid, financed research projects on agriculture-nutrition linkages in Africa and Asia since 2010 or earlier. As pointed out by Webb (2014), most of the attention went to bio-fortification and food supply chains and women’s nutrition. The focus was to improve nutrition directly through self-consumption.

Child underweight and land productivity 15 Hence the studies that explored the impact of agricultural productivity on child undernutrition, via poverty reduction, are few. Systematic review of research done with respect to India concluded that the evidence of impact of agriculture on child undernutrition was mixed and the link appears to be weak. Data gaps were mentioned as a major hurdle (Gillespie et al., 2012). This chapter attempts to take a fresh look at agricultural productivity and child underweight in India at the district level in the first quinquennium of the twentyfirst century. Child underweight data are available for all districts in India from District Level Health Survey-2 (DLHS-2). National Family Health Survey-3 (NFHS-3) conducted in 2005–06 did not provide district level data on child nutrition. (Chapter 3 considers the recent data from NFHS-4 conducted in 2015.) At the beginning of the twenty-first century, there was a marked deceleration in growth rate of agriculture in crop sector as well as livestock sector. It was the beginning of deceleration in yield growth of major staple crops. Poverty continued to decline, and food prices were at their lowest ever in the first quinquennium of the twentyfirst century (elaborated in the next section). Child undernutrition recorded very little decline up to 2005–06. For children below the age of three, underweight rate fell from 49.4% in 1992–93 to 42.82% in 1998–99 and to 40.86% in 2005–06. Corresponding figures for stunting rates were 52.42% in 1992–93, 50.65% in 1998–99 and 44.73% in 2005–06. Reduction in percentage of underweight children below the age of three was lower between 1998–99 to 2005–06 compared to the previous period (Subramanyam et al. 2010). It resulted in widespread criticism of India being worse than sub-Saharan Africa, Asian enigma and Indian enigma being used to describe the situation (Ramalingaswami et al. 1996; Headey et al. 2011). As figures show, India in 2015 still has a higher stunting rate, at 38% (NFHS4), than the average for sub-Saharan Africa that stood at about 35% (World bank SDG 2.2 Atlas 2018). India also has the largest number of stunted children in the world (Global Nutrition Report 2018). Under age five stunting rates started falling in India after 2005, while they stagnated in sub-Saharan Africa, reducing the gap. Population growth is also high for Africa, compared to India. India has higher stunting rates among children, compared to many developing countries with more equitable distribution of assets and substantial investment in human capital. The relevant takeaway lesson from the global trends and comparison with similar child undernutrition rates is that reversal of the progress or slowing down of the progress is imminent, if conditions worsen. Economic slow-down, high fiscal deficits, withdrawal or dilution of welfare programmes, slowdown in poverty reduction, and social unrest and the like are potential threats that could put India off track, with respect to child nutrition SDG targets. Hence looking back to envisage a worse-case scenario for future is always helpful. It avoids the pitfall of dismissing the importance of agriculture and public provisioning for reducing child undernutrition. This chapter revisits the empirical relationship between agriculture and child undernutrition in India using district-level data for child underweight rates, agricultural land productivity and provisioning of public services. This chapter examines determinants of both agricultural productivity and child underweight

16

Swarna Sadasivam Vepa et al.

to find the links at the district level. One of the reasons for the limited number of empirical studies that try to understand agriculture-nutrition linkages is the lack of agriculture-related information in most household-level survey data, otherwise rich in details pertaining to nutrition outcomes. District-level analysis attempted in this chapter fills this gap and enables combining of data from different sources. Further, the spillover effects of agricultural prosperity or backwardness on nonfarm households, and the overall child nutrition scenario can be captured better at the district level rather than at the household level. District level analysis is seen in the literature of agricultural productivity (Chand et al., 2009) as well as child undernutrition (IFPRI 2018). The approach in this chapter differs from some of the other district level studies in three respects. First, this chapter estimates both agricultural productivity and child underweight at district level. Second, the analysis does not consider urban rural distinction for semi-urban districts. One hundred per cent urban districts are excluded from the study. It enables us to catch spillover effects of agriculture on child nutrition, among non-agricultural households. Third, land productivity calculations include both extensive and intensive use of land on one hand and crop and livestock production on the other. Unlike other studies that consider only the net area sown, this study uses cultivable land. It includes fallow lands and cultivable waste, with shrubs, bushes, trees, water puddles etc. that help agricultural eco systems. Fallows and cultivable waste are useful for grazing livestock and collecting fodder. This chapter also examines models that include agro-climatic aspects of district. The other variables apart from agricultural productivity in the nutrition equation are pretty much the same as in many other studies. The results of the agricultural land productivity equation clearly show that rainfall, irrigation, percentage of non-food crops, food abundance in terms of triennium average food production per capita, and percentage of women education above secondary level significantly influence agricultural land productivity per hectare at constant prices in value terms at the district level. Land operational inequality has a negative influence on land productivity but it is not significant. Rainfall, per capita foodgrain production, and women’s education above secondary level have higher elasticity with respect to land productivity. The agro-climatic zonal dummies are all significant when we take central plateau hill region with lowest productivity as the reference zone. The second equation explains the percentage of children with normal status with respect to weight for age, below the age of six years (the remaining children after deducting the percentage of underweight children from 100). Agricultural land productivity per hectare, percentage of women with secondary education show significant positive influence on percentage of children who are not underweight. Prevalence of anaemia among pregnant women, households without access to toilets, and prevalence of childhood diarrhoea among children below three years had significant negative influence on percentage of normal status of children. Services mostly provided by public authorities that influence percentage of fully vaccinated children, administration of oral rehydration salts for children with diarrhoea, population with access to any government health facility, and offtake of

Child underweight and land productivity 17 foodgrain for public distribution among the poor have all shown significant positive influence on children with normal weight for age status. All the three aspects, viz., agricultural productivity, maternal education and health, and provision of public services of health care, seem to significantly influence child underweight rates across districts, even in the period of slow growth in agriculture and slow progress in nutritional improvements. This chapter has four subsections including Introduction. The second section gives an exposition to the possible logical links and mechanism through which agricultural productivity could reduce poverty and improve child nutrition. This section also presents the past evidence on these links. The third section discusses conceptual issues, data and methodology. Section four presents the results and gives the conclusions.

2.2 Agricultural productivity, poverty reduction and child nutrition Agriculture has multiple links to child nutrition. Some are proximate, such as poverty reduction and food intake. Others are remote. Agricultural prosperity of the region may induce women’s education and demand for better health services, and sanitation from public as well as private sectors. The mechanism through which agricultural prosperity translates into better child nutrition for agricultural and non-agricultural households is by no means simple. Yet it is easy to trace some of the obvious pathways. 2.2.1 Agricultural productivity and poverty reduction The relationship between agricultural worker productivity and agricultural land productivity and the. possibility of poverty reduction depends upon a variety of factors. Land productivity in terms of value of output per hectare gets more attention than the value of output per worker, since yield per hectare captures the technological improvement. However, value of output per worker captures the welfare aspect better. Decomposition of output per worker yields two components: output per unit of land and land per unit of labour: Output/Worker = Output/Land × Land/Worker (Gollin et al., 2014). If the land labour ratio is unchanged, then land productivity entirely results in worker productivity pushing up wages. If the land labour ratio worsens, the worker productivity turns low, based on the relative changes in the three ratios. Raising land productivity is the only way to raise the labour productivity in a situation of high population pressure on land and constant land per worker (Timmer 1988). When land per worker cannot be increased, land productivity should keep pace with population growth to just maintain the labour productivity. As Timmer (1988, page 306) puts it, “running fast technologically to standstill economically.” Hence agricultural workers remain relatively poor, despite technological progress due to population pressure on land. In developing countries, the land per worker falls, resulting in small farm size and a large number of landless agricultural workers.

18 Swarna Sadasivam Vepa et al. Land productivity improves with technology as it happened in most of South Asia during the green revolution (Timmer 1988). Hayami and Ruttan (1985) have shown that despite falling land per worker, labour productivity gains were possible for a variety of reasons, such as scale neutral technology, technical inputs, investment in human capital and so on. Increase in land productivity per hectare could be large enough to compensate for the fall in land per worker1 and results in increased per worker productivity. Declining population growth, as well as a shift of workers out of agriculture, could raise worker productivity, as the number of workers fall for the same output. In India there is a close correlation between land productivity and labour productivity in agriculture at the district level (Chand et al., 2009). Agricultural land productivity and proportion of agricultural workers in the rural workforce (proxy for agricultural land per worker) would explain the inter-state variations in wage earnings per worker and poverty (Radhakrishna and Raju 2015). India in the first quinquennium at the turn of the century from 2001 to 2005 experienced low growth in agriculture. The growth rates of output of cereal crops decelerated from a high of 3.5% in the decade ending in the 1990s to a mere 1% in the first five years of the twenty-first century. The crop sector growth decelerated to 2.1% from 3.1% in the mid-1990s. Livestock sector growth in terms of the value of output decelerated to 3.6% from a high of 4.8% in the decade ending in the 1990s. The growth rates of fruits and vegetables also slowed down. Only cotton witnessed impressive double-digit growth rate. Growth revived in all the crops from 2007 onwards, as can be seen from Table 2.1. Diversification of agriculture into dairying, meat and eggs contributed to enhancement of agricultural productivity. Investment elasticity of output in livestock is believed to be double that of investment in crop land. Land productivity and labour productivity growth was at the lowest ever at 1.8% and capital productivity growth was negative (Planning Commission 2012; Uma Kapila 2013) between 2000 and 2005. Agricultural wage growth decelerated Table 2.1 Value of agricultural output growth (2004–05 prices)

1 2 3 4 5 6 7 8 9 10

Item

2002–03 to 2006–07

2007–08 to 2011–12

Cereals Pulses Oilseeds Sugar Fibres Non-horticultural crops Horticulture All crops Livestock Crops & livestock

1.0 1.8 7.4 1.7 15.1 2.1 2.6 2.1 3.6 2.5

3.0 4.2 4.5 2.2 10.7 2.8 4.7 3.4 4.8 3.8

Source: 12th Five-Year Plan, Chapter 6, 2012, Reprinted in Uma Kapila, 2013

Child underweight and land productivity 19 both for men and women between 2000 and 2005 to 1.4% and 1.1% per annum respectively compared to 2.78% and 2.74% between 1993–94 and 1999–2000. (Sundaram 2007). During 2000 and 2005, K. Sundaram has shown that the number of working poor declined in rural India and increased in urban India, but the workforce above poverty line increased both in rural and urban India showing better quality employment. Higher land productivity in India led to both absolute and relative income gains to the poor, mainly through wages and prices. Poor gained also through increased labour demand. The long-run elasticity of land productivity to poverty reduction was high (Datt and Ravallion 1998; Ravallion and Datt 2002). Improvement in agricultural wages contributed to reduction in poverty among agricultural labour households in 2011–12 (Radhakrishna and Raju 2015). Food-price inflation of primary food articles was at its lowest at the turn of the century between 2000 and 2005, at about 3%, much lower than all commodity prices. The inflation of protein rich foods was also at its lowest during this period. Cereal prices and non-cereal food prices had welfare implications for both rural areas and urban areas. Non-cereal food prices were as important as cereal prices (Radhakrishna 2015). Despite deceleration in agricultural growth between 2000 and 2005, wage growth and food price decline helped the poor and that explains the poverty reduction. The annual rate of reduction in poverty in 1993–94 and 2004–2005 was 0.75% per annum in rural areas and 0.55% per annum in urban areas. Level of poverty reduction rate was higher after 2004–05 at 2.32% and 1.69% per annum for rural and urban areas respectively between, 2004–05 and 2011–12 (Rangarajan Committee Report on Poverty 2014).2 K. Sundaram (2007) has shown that poverty declined at the rate of about 0.75% per annum between 1994 and 2000 and at the rate of 0.9% per annum between 2000 and 2005. Despite deceleration in agricultural growth, poverty reduction accelerated between 2000 and 2005, probably owning to increased wages and decreased food prices. Importance of agriculture for poverty reduction across the states was established for India (Datt and Ravallion 1998; Ravallion and Datt 2002). Poverty reduction has a positive elasticity with respect farm yields. Areas with higher land productivity had higher poverty reduction. Non-farm growth was related to higher farm productivity and it was more pro poor, in states where landlessness was low and rural living standards and literacy rates were better (Ravallion and Datt 2002). Agricultural growth thus has spillover effects that accrue to non-farm populations via employment generation and to urban poor via low food prices. Earlier estimates of Rangarajan (1951–76) show that 10% increase in agricultural output would increase industrial output by 5% in India. Urban workers would benefit both from industrial employment and price deflation (Rangarajan 2004). The link may have weakened in recent years, but it still exists. Based on the analysis of satellite data from 1992 to 2012 (Gibson et al. 2017) found that in India’s current stage of development, urban growth in secondary towns rather than in big cities appears to reduce poverty. Pace of urbanisation between 2001 and 2011 was slower than expected. Population growth rate of class

20 Swarna Sadasivam Vepa et al. one cities decelerated and that of other urban areas consisting of towns accelerated between 2001 and 2011 (India Office of the Registrar General and Census Commissioner 2001, 2011). Poverty reduction in non-agricultural sector was driven by agricultural prosperity in India. Poverty-reducing effect of urbanisation is primarily explained by increased demand for local agricultural products (Calì and Menon 2013). Indirect effect of agriculture on poverty reduction in non-farm sector has been illustrated in other contests as well. While the direct growth effect of agriculture on poverty reduction is likely to be smaller than that of non-agriculture (though not because of inherently inferior productivity growth), the indirect growth effect of agriculture (through its linkages with non-agriculture) appears substantial and at least as large as the reverse feedback effect. (Christiaensen et al. 2006) Agricultural land productivity potentially enhances incomes, sometimes increases local food availability, leads to higher real wages and lower prices of food. This results in reduced poverty. The next step is the potential influence of poverty reduction on child nutrition improvement. The risk of child undernutrition reduces with increased household income but does not ensure the elimination of undernutrition outcomes for children. Middle income states, with better social policy rather than the rich states, had lower incidence of child undernutrition (Sambi Reddy et al. 2004). 2.2.2 Agriculture and child nutrition Direct link of agriculture to child nutrition via food availability and food shortages is an obvious one. Child mortality increasing in the years agricultural production deficiency (Rose 1999) and child undernutrition doubling in the years of adverse agricultural conditions (Robert Jensen 2000) have been well documented in the literature. The evidence from literature in India on the influence of food and agriculture on child undernutrition rates is varied. Bhagowalia et al. (2012), using India Human Development Survey (IHDS) data, showed that agricultural income did not have any positive impact on poverty reduction or reduction in underweight and stunting. On the other hand, non-agricultural income was associated in rural areas with better nutritional outcomes. Similar conclusions were drawn by some other studies. Galab and Reddy (2012) have indicated that households who sell their produce in the market rather than those who predominantly consume from the market had lower underweight rates among children. Part of the problem could be lack of access to quality health care in predominantly rural districts. The state-level study using cross-section data concluded that both overall agricultural growth and foodgrain production growth are not necessary conditions for nutritional improvement in India (Headey et al. 2011). The study however found that, at the state level, agricultural Gross Domestic Product (GDP) per worker and

Child underweight and land productivity 21 non-agricultural GDP per worker have significant negative associations with stunting but not with underweight. A panel data fixed effects model at state level using data for two time periods (1998–99 and 2005–06) shows significant negative association of agricultural worker productivity as well as agricultural land productivity with the proportion of stunted children and the proportion of underweight children in the rural areas. Production of foodgrains per capita did not show any significant association with stunting rates but shows significant association with underweight rates. On the other hand, agricultural growth in the preceding five years shows a significant negative association with stunting rates but not with underweight rates (Vepa et al. 2014). Agricultural land productivity influences child underweight rates at the district level in all quantiles, while worker productivity influence is not significant in the quantiles (Vepa et al. 2015). The evidence of specific interventions in agriculture meant to enhance nutritional outcomes of adults and children is largely lacking with a few exceptions, primarily due to poor evaluation (Ruel et al. 2013). The only success has been that of orange-fleshed sweet potato in improving vitamin A levels in Africa (Nestel et al. 2006). Studies that tried to link food production through interventions at the household level with the nutritional outcome of women and children in the developing world did not find a convincing link due to methodological limitations (Girard et al. 2012). However, using IHDS household level data, for India, Viswanathan (2015) has shown that agricultural production diversity could positively impact women’s BMI at the household level by improving dietary diversity. 2.2.3

Public provisioning and child undernutrition

Past literature has highlighted the benefits of public provisioning of food (Khera 2009), work (Jha et al. 2011), water (Fink et al. 2011), sanitation (Spears 2013) and health care (Lim et al. 2010) for nutrition improvements. The district as an administrative unit of public provisioning and service delivery assumes importance. The demand for public services depends on the relative prosperity and poverty of the region. For example, if people are relatively well off, they may use better quality private health facilities that are more expensive than the facilities provided by the government. In general, public services are subsidised and meant for the relatively poor. Sometimes, if the public provisioning system works well, or if there is no alternative available (such as drinking water supply), all sections of the population use the public services. Effectiveness of the public services depends on the nature of supply of the services on one hand and the actual use of the services by the people, on the other. Underutilisation of public services occurs due to unreliable supply of the public service (in terms of time, frequency and quality) or due to lack of knowledge. Sen (1992) also points to the reduction of the gap between the availability of the public service and its utilisation by the deprived sections of the population, with an increase in public awareness and increased educational levels. Economically backward regions also tend to have a concentration of less-educated people and lower levels of utilisation. Sometimes public services may create an awareness about health and nutrition due to advertising,

22

Swarna Sadasivam Vepa et al.

though the service delivery is poor. States such as Kerala in India provide better quality public services and hence the child mortality rates, and child underweight and stunting rates are the lowest (Drèze and Sen 2013; Government of India 2005; NFHS-4 2015–16). 2.2.3.1

Public provisioning of food and nutrition outcomes

Public provisioning of food is expected to improve calorie intake by the poor especially expectant and lactating mothers and children below the age of five and help address child undernutrition; it is however difficult to capture the impact when studied across all income classes. We often find higher levels of undernutrition among users of the public distribution system (PDS), who are essentially poor. Kaushal and Muchomba (2013) found that increase in food subsidy after expansion of the targeted public distribution system (TPDS) in 2002 had a negligible to negative effect on calorie and protein intake and no statistically significant effect on fat intake. An earlier study by Kochar (2005) has shown a small effect of PDS subsidy on calorie intake. Some other studies have concluded that PDS did not have much of an impact on calorie consumption. A study on review of existing literature on food-based interventions observed that limited benefits accrued to younger children in the households exposed to PDS services for longer durations (Ruel et al. 2013). The study opined that poor service quality and weak goal setting probably explain the lack of overall nutritional benefits. However, after the food price crisis in 2008, and across the board reduction in issue price of rice and wheat in many states, PDS seems to have positive impact on food consumption. Based on a primary survey in three states, Jha et al. (2011) concluded that calorie intake and other nutrient intakes improve with participation in the PDS. Participation in the PDS has been instrumented with other variables satisfying the exclusivity criteria. Khera (2009) found that better access to subsidised rice from fair price shops resulted in major reduction in self-reported measures of hunger in Chhattisgarh. Kaul (2013) estimated that an increase in the value of the PDS subsidy by 1% increases caloric intake by 0.144%, while an increase of 1% in income (expenditure) is associated with an increase of 0.4% in caloric intake. Her projections suggest that implementation of the National Food Security Act will increase the per person caloric intake of the present beneficiaries of the programme by 72 calories per day in urban areas and 66 calories per day in rural areas. It has been realised that calorie-income elasticity varies over a large range ─ from negligible in Behrman and Deolalikar (1987) to 0.30–0.50 in Subramanian and Deaton (1996). Even if calorie-income elasticity is low, the effect on undernourishment may be large if the density of people is high in the neighbourhood of the calorie requirement norm (Ravallion 1990). The suggestion is that, perhaps, the elasticity changes, depending on grain prices and the level of deprivation. Even if calorie-income elasticity is low, there are grounds for optimism about the prospect of eliminating nutritional deprivation by raising incomes of the poor through subsidised PDS grain (Jha et al. 2011). Studies have shown that subsidised grain and effective income transfers could translate into additional nutrient intake. Kishore

Child underweight and land productivity 23 and Chakrabarti (2015) noted that households in Chhattisgarh used money saved from access to PDS rice to spend more on pulses, edible oil, vegetables, sugar and non-food items. In this chapter, effective PDS implementation by state governments is an independent variable influencing child underweight. This chapter considers percentage of offtake (grain lifted by the state government from the storage houses of the central government) as a percentage of the total foodgrain allotment made by the centre to the state as an indication of the commitment of the state government to distribute the available grain to its poor. With the introduction of the Targeted Public Distribution System (TPDS), where, subsidised foodgrains were offered only to the ration card holders below poverty line. offtake was higher for the Below Poverty Line category. The introduction of TPDS has also led to perverse outcome where allotment of foodgrains was increased to states with weak delivery systems and reduced to those with greater administrative efficiency, since the allotment was based on the prevailing poverty levels in the state. In some states like Bihar, the offtake was very low at 50% of allotment, while in Andhra Pradesh it was close to 100%. Across states and over time, the percentage of allotment to offtake varies (GOI, Ministry of Consumer Affairs, Food and Public Distribution). 2.2.3.2

Public provisioning of water, sanitation and health (WASH)

Much of the undernutrition currently prevalent in the children of developing countries is attributable to conditioned malnutrition arising from infections (Gopalan 2014). Access to sanitation and safe drinking water has been identified by researchers as a key factor in reducing stunting and underweight (Bhagowalia et al. 2012; Spears 2013). Hammer and Spears (2013) have shown that a gain of approximately 1.3 cm in height is possible in a four-year-old child with the provision of safe sanitation in the child’s immediate environment. Fink et al. 2011 have shown that improvement in water and sanitation lowers the incidence of diarrhoea, by 7–17% and reduces the risk of under-five child mortality by about 50%. As per the UNICEF/WHO report of 2009 on diarrhoea, the food absorption capacity of children declines with diarrhoeal infections. This may result in undernourishment and underweight. Remedial measures such as administration of oral rehydration salts can substantially prevent survival risks. Access to health facilities such as primary health centres, doctors and hospitals as well as compulsory maternal and child referral programmes go a long way in improving child nutrition. Studies evaluating the National Rural Health Mission (NRHM) felt that it has been ineffective in improving the health service delivery in India (Paul et al. 2011). A critical review of the primary health care system in rural India describes it as dysfunctional (Antony 2014). Another evaluation study of the national rural health programme has noted that lack of functioning primary health centres in the villages as also the distance involved in reaching the centres for referral are deterrents for many women and children to attend the pre-natal and post-natal referral programmes that could improve child health. The presence of an accredited social health activist (ASHA) and Integrated Child Development

24 Swarna Sadasivam Vepa et al. Services (anganwadi) workers in the villages have been effective in spreading health awareness among women about maternal and childcare, vaccination, use of oral rehydration salts, etc. (GoI 2011). Using two rounds DLHS-2 and DLHS-3 data another study concluded that the safe motherhood programme (Janani Suraksha Yojana) that gave cash transfers, conditional on compulsory attendance at the referrals, reduced maternal mortality in India (Lim et al. 2010). Antenatal care was found to be an important factor in reducing child malnutrition by recent (IFPRI 2018) study using NFHS-4 data. Adverse influence of maternal ill health (such as iron and folic acid deficiencies), especially that of expectant and lactating mothers, on child undernutrition has been evident for a long time now (Viteri 1994; Ravi and Radhakrishna 2004). 2.2.4 Women’s education, work, maternal and child health There is a large body of evidence on the impact of women’s education and employment on the nutritional status of children (Mishra and Retherford 2000; Maitra 2004; Ravi and Radhakrishna 2004). Using NFHS-4 district level data and regression decomposition methods for India, the International Food Policy Research Institute (IFPRI 2018) found that maternal education, low BMI of women, age at marriage, and antenatal care are some of the factors that accounted for the difference between districts with high and low burden of stunting. While mother’s education improves child nutrition, mother’s work especially in agriculture has an adverse effect on child nutrition (Berman 1997; Ravi and Radhakrishna 2004; Bhalotra 2010). Probably the type of work influences maternal health and childcare. An interaction term of percentage of women with education higher than secondary level, with percentage of women working, shows significant effect of reducing the percentage of underweight children at the district level (Vepa et al. 2016). Child undernutrition rates vary across states and across districts within the state. Based on a Child Development Index for the districts of India wherein child underweight rates from DLHS-2 (GoI 2005) are a component, Dreze and Khera (2012) found that though most of the high values are concentrated in central and eastern India, there were also pockets of lower underweight rates within these regions. Interestingly, adjacent districts belonging to different states showed sudden changes in the underweight values in a further assessment of this study by Viswanathan (2014). This lends one to infer that perhaps state-level characteristics like public provisioning of basic services and agricultural prosperity may be influencing factors. Hence, this chapter considers the state fixed effects in explaining the variations in child underweight.

2.3 Conceptual issues, data and methodology The motivation for using the district as the unit rather than separating it into rural and urban areas is based on several aspects of the current scenario in India relating to agriculture and public services. As agricultural transformation takes place and processing of agricultural products, both food and non-food, gain importance,

Child underweight and land productivity 25 the traditional break-up into agricultural and non-agricultural, or rural and urban, loses its relevance (Timmer and Akkus 2008). The structural transformation of agriculture in India seems to have spillover effects, which are important drivers of the rural non-farm sector in India (Binswanger-Mkhize 2013). Secondly, due to better connectivity, small towns and semi-urban centres attract trading hubs, processing units, poultry and dairy units and peri-urban agriculture, and they are very much a part of agricultural production and processing. A district is a contiguous agro-climatic region, and the characteristics related to rainfall, water availability, soil type, population density, etc., tend to outline a unique development path to both rural and semi-urban areas. Since our methodology aims to estimate both agricultural productivity and child undernutrition, district as unit provides more information. Finally, the public services, especially the fair price shops for foodgrains, the primary health centres, government hospitals, referral services, etc., may be located in small towns (mostly at the district headquarters) and as awareness and affordability and connectivity increase, used by the rural population, though facilities within the village are important for the deprived sections of society. Hence, there is merit in keeping the district as the unit for the analysis. However, all the 100% urban districts are excluded from the study. It is also found that districts which are less urbanised have lower mean agricultural worker productivity, than more urbanised districts, clearly pointing to the fact that higher agricultural productivity contributes to non-agricultural activity related to agriculture. 2.3.1

Concept of land productivity

In this chapter, land productivity has been used to capture the role of agriculture in reducing undernutrition. The motivation for using the value of land productivity per hectare of cultivable land to represent relative agricultural prosperity is to include the possibility of intensive as well as extensive use of land resources for dairy, poultry, livestock grazing, fodder collection, etc., in addition to crop production. Ideally, inland fisheries will have to be included. However, data was not amenable for such analysis.3 Production either for self-consumption or for the market, whether monoculture or a diversified cropping system, benefits producers and agricultural wage labour, in addition to non-agricultural workers linked to agriculture through backward and forward linkages (Timmer and Akkus 2008). The same families derive income from agricultural as well as non-agricultural sources. This chapter tries to capture the agricultural system for nutrition as well as for enhanced income in a single indicator, viz., the value of land productivity per hectare at constant prices. Land productivity in this study is the three-year average value (at constant prices) of agricultural gross domestic product per hectare of cultivable land in the district.4 The value of non-food crop production, food crop production and livestock production (including dairy and poultry) is part of District Gross Domestic Product from agriculture. Land productivity depends on major agriculture-related aspects, such as rainfall, irrigation, land inequality, food and

26 Swarna Sadasivam Vepa et al. non-food production. The study estimates land productivity as a function of factors that influence agriculture.5 The focus is on modeling the rate of children with normal weight-for-age, across districts, while allowing for the assumption of possible endogeneity of land productivity, one of the explanatory variables in this model. Land productivity, a proxy for agricultural prosperity, is conceptually endogenous as it influences prosperity of the households, directly or indirectly, which in turn would influence underweight. The suitable econometric technique to estimate this empirical relationship is discussed in the section on data and methodology. 2.3.2

Choice of child undernutrition indicator

Long-term nutritional deprivation causes stunting (height for age) in children under five. It may lead to delayed mental development, poor schooling performance and reduced intellectual capacity (WHO 2010a). Wasting (weight for height) is a result of acute undernutrition, caused by deficient food intake or illness. It has high risk of mortality. Underweight in children (weight for age) can reflect wasting (i.e. low weight for height), indicating acute weight loss, or stunting (low height for age), or both. “Underweight” is a composite indicator (WHO 2010a). Since underweight can occur either due to stunting or wasting or both and difficult to say which of them caused underweight. Evidence has shown that the mortality risk of children who are even mildly underweight increased, and severely underweight children are at even greater risk (WHO 2010a). Since it is a composite measure, it reflects one or more deprivations and risks associated with them. Of the three indicators of child malnutrition, viz., stunting, underweight and wasting, stunting and underweight are known to be closely corelated. Underweight is also known to be closely correlated with wasting. This is also seen in NFHS-4 data for selected two states as well (Appendix 2A and Table 2A.4). A specific relationship cannot be predicted between the various parameters of child nutrition. Less than 2% of obese stunted children are found globally as per the 2018 global nutrition report. SDG indicators include stunting and wasting and obesity in children under five, but not underweight. However earlier underweight was used extensively in the international child nutrition metrics. Data are available at the district level from DLHS-2 (2002–04). None of the other surveys provide representative child nutrition data at the district level for the first quinquennium at the turn of the century in India. 2.3.3

Data and methodology

As mentioned earlier, the unit of analysis is the district, with focus on the proportion of children who are not underweight (the main dependent variable) in the age group of six years or below, from the second round of District Level Household Facility Survey (DLHS-2) for the year 2002–2004.6 The data of DLHS-2 pertains to children below the age of six, while most surveys (e.g., IHDS and NFHS) give data on children below the age of five. At the state level, the difference between

Child underweight and land productivity 27 underweight rates for children below five years and below six years in DLHS-2 data was minimal. The data extracted from unit level records for children below the age of five has been presented in Table 2A.1. In this chapter we preferred to use the data for children below the age of six. As is found in respect of many cross-country studies elsewhere, the data for the variables used in the analysis of this chapter pertain to different years, though not to different time periods. This cannot be helped in a situation where surveys are not conducted in the same year. However, all the data pertains to the years falling between 2002 and 2004, except the land inequality data that pertains to 2006. The assumption is that year-on-year variations in agricultural data are not important as we average the agricultural data over three years. Land inequality is generally stable for longer time periods and available only from agricultural census conducted once in five years. The data on underweight and public provisioning of services pertain to the same year. Though the common practice is to use underweight rates directly in the analysis, in this study we have preferred to use the percentage of children above 2 standard deviations in z-score. This has been referred to in the study as children’s normal nutritional status or CNS for short. However, we caution the reader that this normal nutritional status is only with respect to weight and that the nutritional status of the child with respect to stunting, wasting, anaemia, etc., may or may not be normal. Better child weight status helps to reduce the incidence of stunting and wasting. Although this district-level information dates back to more than a decade, it is the only data set that provides information on CNS across districts that also covers large parts of India at the turn of the century. NFHS-3 for 2004–05 unfortunately does not provide district level data. Chapter 3 analyses the more recent NFHS-4 data. The relevance of this study remains in documenting empirical evidence that establishes a firm association between agriculture and undernutrition; this has been ambiguous for India in earlier studies based on other data sets on individual children from household surveys for this period. Though undernutrition rates have shown a slower rate of decline in rural areas and show a higher prevalence rate, we focus on combined rural and urban rates of undernutrition, as improvements in the performance. Agriculture is known to have economy-wide benefits in India. A brief discussion about the data sources and the nature of these variables is in order. The reference period for the data is 2002–2004 or any year close to it when data were not available for the relevant years. Most of the data for the child nutrition equation is from District Level Health Survey-2 with the reference period 2002–04. Agricultural variables are either from the Ministry of Agriculture and Co-operation for the years 2002, 2003 and 2004, or from the Agricultural Census of 2005–06. The data for all the other variables in the equation, except the variable related to agriculture and the public distribution system, are from the DLHS-2. The covariates used to explain the variations in CNS are concerned with: •

Agricultural prosperity represented by (agricultural land productivity) The three-year average (of 2002, 2003 and 2004) value of land productivity at constant prices per hectare of cultivable area in the district represents

28 Swarna Sadasivam Vepa et al. agricultural prosperity of the district. The three-year average value of district GDP from crops and livestock products divided by the three-year average of cultivable area (consisting of net area sown, all fallow land and area under miscellaneous crops not included in the net area sown, cultivable waste) gives the land productivity per hectare. District GDP data for three years at constant prices are from Indicus Analytics that compiles district data from the figures made available by the Central Statistical Organization. Land data for three years are from the Ministry of Agriculture and Cooperation. Appendix 2B gives the details of the variables. •

Indicators of maternal health and child health (percentage of pregnant women with anaemia and incidence rates of diarrhoea in children below three) The percentage of pregnant women reported as moderate to severely anaemic in the district captures mothers’ health. The percentage of children (under three) with diarrhoea to the total number of children captures child health status. The prevalence of diarrhoea is known to reduce nutrition absorption and hence lower the nutritional outcome. Since younger children are more prone to diarrhoea, the age of children to assess diarrhoea rates is restricted to below three years (rather than up to six years, the age of the children considered for CNS in this study). We expect both these health-related variables to have a negative impact on inter-district variations in CNS.

•

Access to basic amenities of sanitation and water supply (percentage of the rural population with access to toilets and water supply) Provisioning of basic amenities like sanitation and water reflects not only the economic prosperity of the districts, as better-off households usually have better quality amenities, but also captures the role of the individual state and its quality of governance in making this accessible to a larger section of the population. On the one hand, the level of private household incomes plays a significant role in accessing these facilities, but if certain districts have better administration for the provision of these facilities, then less prosperous households in that region would also benefit. The percentage of population without toilet facilities (open defecation) and the percentage of population with access to public toilets in the districts are the two variables that capture the quality of sanitation facilities. They are important in explaining the variations in CNS. The percentage of population with access to safe water supply is another variable that could explain the variations in CNS. In India, officially, the term “safe water supply” only means piped water or well water mostly supplied by the public authorities, but the definitions vary from state to state; sometimes they include wells and sometimes they do not. Safe water supply does not refer only to the water treated for safety but also includes untreated water sources. To make the variable uniform, all wells (both public and private) as well as piped water from the government and private sources have been included and referred to as non-natural sources of water.

Child underweight and land productivity 29 •

Access of the rural population to any one of the government facilities within the village (percentage of rural population with access to any one of the seven specified government facilities within the village) Any government health facility located in the village provides timely health care to the deprived sections. Any government health facility in the DLHS-2 data refers to the presence of one of the following centres within the village: primary health centre, sub-centre, community health centre, referral hospital, government hospital, government dispensary and anganwadi centre. Population with access to any one of these facilities as a percentage of total population has been taken as an explanatory variable that could improve child nutrition. These facilities provide timely care in disease prevention or its timely cure, and hence have an impact on CNS.

•

Adoption rates of recommended child health practices (percentage of children who received full vaccination and percentage of children administered with oral rehydration salts for diarrhoea) As mentioned in the objective, the intention of this study is also to understand the impact of certain intervention policies on CNS captured by the use or access to a government facility or a desirable action promoted by the state. One such variable is the interaction between percentage of children administered oral rehydration salts (ORS) solutions and the rate of prevalence of diarrhoea in that district. The interaction of these two variables is considered as the need to use ORS (which would arise when the child is affected), so that if the interaction coefficient is statistically significant with a positive sign, then it implies that the use of ORS reduces the impact of diarrhoea on CNS after controlling for diarrhoea rate. The rate of incidence of anaemia among expectant mothers or use of ORS (or the lack of it) for a diarrhoea-affected child, once again occurs as a combination of the use of either a private facility or a public service. Income of the households and their education level, on the one hand, and the awareness created by state agencies in making the households aware about these aspects, on the other hand, result in a positive impact on child nutrition. This changes the situation for the better. Thus, a less-economically prosperous district could have lower prevalence rate of anaemia among pregnant women as well as higher use of ORS (in the presence of diarrhoea) due to the intervention of government awareness programmes through, for instance, ASHA workers or anganwadi workers.

•

Implementation effectiveness and commitment of the state governments to public distribution of the foodgrains to the poor (percentage of offtake to allocation) Better functioning of PDS in a district implies regular access to subsidised foodgrains through fair price shops. On the one hand, this would impact protein-energy malnutrition due to regular consumption of cereals, pulses

30 Swarna Sadasivam Vepa et al. and oil as there is access to food at a subsidised price. On the other hand, there is an implicit income transfer to the household due to the subsidised price of these commodities, due to which the households can spend on other essential food and non-food commodities that may also add to the reduction of CNS. However, district-level information on the proportion of households consuming from PDS is not available, though the proportion of households holding a ration card is available from DLHS-3. We preferred not to use access to BPL cards, as that need not reflect purchase of commodities. Instead, we used the percentage of offtake of cereals from the central pool as a proportion of the allocation to the state in which these districts are located, so that one would expect larger distribution across the districts. The data pertaining to the year 2004–05 on the proportion of offtake to allocation is from the Ministry of Consumer Affairs, Food and Public Distribution.7 •

Women’s education at secondary level or above that helps knowledge of childcare as well as utilisation of public services available (percentage of women with education at secondary level or above) The proportion of women who have completed education up to secondary level or above captures the role of women in ensuring better child health outcomes in terms of child nutrition status. This represents women’s positive agency effect on child nutrition. Lower levels of literacy also help child nutrition. However higher levels of education may not create adverse effect on childcare, when women are working. This may improve the women’s agency effect substantially, as they also have access to income to spend on nutritious food, childcare services, get maternity leave and other benefits in regular employment. In that case income effect is combined with education effect.

Due to the potential endogeneity of the agriculture variable, we consider estimating the CNS model using three stage least squares (3SLS) that allows endogeneity. The agriculture equation or the first stage equation has six variables as elaborated in Appendix 2B and 13 agro-climatic regional dummy variables. The empirical relationship between nutrition and agriculture has been estimated using a two-equation framework as given here: CNSi = α1 + βAi + γX + e1i, (2) Ai = α2 + δZ + e2i

(1)

Where CNSi = percentage of children (0–59 months) with normal weight for age in the ith district. Ai = agricultural variable captured by land productivity in the ith district. X and Z are the vector of other covariates that influence CNS and A, respectively, and have been discussed earlier in this section. All the variables are in the logarithmic form so that the estimated coefficients from the regression are interpreted as the elasticity of that variable to CNS (or land productivity). In equation (1), CNS is influenced by agriculture (Ai) and due to its possible endogeneity, equation (2)

Child underweight and land productivity 31 captures the variables that determine Ai i.e., land productivity. This two-equation model is estimated using 3SLS approach. The 3SLS method accounts for the covariance across equation disturbances while providing the instrumental variable estimates. This modeling framework not only tries to capture the nature of relationship between child normal weight rates and land productivity at the district level after controlling for other variables, but also informs us on some possible factors that influence land productivity. Further, all the variables that are included in the agricultural equation, except for women’s education, act on the child nutrition equation only through land productivity of the agriculture equation. Even if efficiency of the district governance is an important input, governance related to agriculture will act only through land productivity. The aim is to study the association of land productivity (assumed to be a proxy for agricultural prosperity) and public provisioning of services with child underweight rates at the district level. The next section discusses the main findings of this empirical analysis.

2.4

Results and conclusions

Sector-wise contributions to the total district GDP in the predominantly rural districts with more than 80% of rural population and other districts with less than 80% rural population show that the share of agriculture is not very high. Mean proportion of rural population for predominantly rural districts was 88% and the mean rural population of the other districts was 60% in 2001. The mean share of agriculture in the district GDP for predominantly rural districts was about 22% and that of other districts was about 15%. Tables 2.2 and 2.3 show per worker productivity at constant prices, from agriculture and other sectors. Higher share of agriculture does not mean higher worker productivity in agriculture. Higher agricultural worker productivity is seen in the districts with lower share of agriculture. With the decline in the share of agriculture in the overall GDP, service sector

Table 2.2 Sectoral GDP of districts with less than 80% rural population 2001—other districts

Obs

Mean

Std. dev

Min

Max

Per worker GDP (primary) Per worker GDP (Agriculture) Per worker GDP (secondary Per worker GDP (tertiary) Share of primary sector (%) Share of agricultural sector (%) Share of secondary sector (%) Share of tertiary sector (%) Rural population (proportion)

293 290 293 293 293 290 293 293 295

41.866 28.959 106.83 112.08 27.884 15.800 24.433 47.747 0.6073

76.776 28.354 115.92 44.651 14.603 10.087 12.57 13.586 0.1829

0.1018 0.1018 0.1087 0.1125 0.000 0.00 8.00 12.00 0.00

1012.3 234.74 1064.7 259.76 72.00 48.00 80.00 84.00 0.7995

Source: Authors calculations from district GDP data Note: Per worker GDP is in rupees at constant prices

32 Swarna Sadasivam Vepa et al. Table 2.3 Sectoral GDP of districts with more than 80% rural population 2001—Predominantly rural districts

Obs

Mean

Std. dev

Min

Max

Per worker GDP (primary) Per worker GDP (agriculture) Per worker GDP (secondary) Per worker GDP (tertiary) Share of primary sector Share of agricultural sector Share of secondary sector Share of tertiary sector Rural population (proportion)

323 322 323 323 323 323 323 323 323

22.244 15.293 68.63 84.782 41.7 22.7 17.09 41.257 0.8809

40.247 26.006 72.912 40.086 12.743 10.291 10.631 9.9844 0.0493

0.0954 0.0945 0.1041 0.1062 8.00 1.00 4.00 13.00 80.00

487.85 436.5 454.8 201.31 82.00 51.00 63.00 71.00 100.00

Source: Authors calculations from district GDP data Note: Per worker GDP is in rupees at constant prices

contributed more to the GDP in all districts. Thus, rural is not synonymous with agriculture and urban is not synonymous with secondary sector. The rural urban distinction has also blurred at the turn of the century. Appendix Tables 2A.2 and 2A.3 present the descriptive statistics of the variables in the level form and its logarithmic transformation (as used in the regression models), respectively. Table 2.A5 presents the correlation between the variables used in the model. The magnitude of the correlation rarely exceeds 0.5 and hence the problem of multi-co-linearity is not an issue among the explanatory variables. If the correlation between the CNS rate and other explanatory variables is gleaned (as in the first column), then we notice that districts with lower share of agricultural GDP have a higher percentage of CNS while regions with higher average land productivity have higher CNS. Most of the explanatory variables considered for the analysis have statistically significant pair-wise correlation with CNS. The results of the agricultural equation given in Table 2.4 show that the estimated coefficients behave as expected. Rainfall and irrigation (after controlling for the other one) has an elasticity of about 0.36 and 0.13, respectively. The signs and magnitude in Table 2.4 indicate that most areas are still rain-fed and, among them, those with higher rainfall have higher elasticity towards land productivity. As for irrigation, once rainfall is controlled for, the elasticity is almost one-third of that for rainfall. Though inequality in land holding is insignificant, it has a negative relationship with land productivity as may be expected. The proportion of area under non-food crop production has an elasticity of 0.12 after controlling for other variables. In comparison to this, the triennium average of per capita foodgrain production has a large elasticity of 0.389, implying that productivity gains are higher in regions with higher per capita foodgrain availability; it implies higher yields, which can lead to enhancing income-generating economic activities. Finally, the elasticity of women’s education, which also promotes technology

Child underweight and land productivity 33 Table 2.4 Estimated coefficients of agricultural land productivity across districts Variable

Coefficient

p-value

Triennium average annual rainfall (in mm) Percentage of area irrigated to total cropped area Gini coefficient of operational land holdings Percentage of area under non-food crops to total cropped area Triennium average annual foodgrain production per capita (tonnes/capita) Percentage of women with secondary education and above East Coast Plains and Hills Region Eastern Himalayan Region Eastern Plateau Hills Region Gujarat Hills and Plains Region Middle Gangetic Plain Region Southern Plateau Hills Region Trans-Gangetic Plain Region Upper Gangetic Plain Region Western Dry Region Western Himalayan Region Western Plains and Ghats Region Western Plateau Hills Region R2

0.3658*** 0.1354*** −0.0459 0.1238*** 0.3899***

0.000 0.000 0.754 0.000 0.000

0.3719*** 0.5671*** 0.7753*** 0.5526*** 0.6815*** 1.0040*** 0.3691*** 0.8263*** 0.9756*** −0.5271*** 0.7546*** 0.4193*** 0.2242*** 0.5986

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.000 0.008 0.031

Note: logarithm of triennium average land productivity for 2001–02, 2002–03 and 2003–04 is used as dependent variable. *** denotes significance at 1% level

adoption and skill acquisition for expanding agriculture, is significant and positive, and high, at 0.37. This co-efficient being a proxy for education of men (as the areas with higher education among women would also tend to have a high level of men’s education) also captures men’s education. The coefficients of the zonal dummy variable are largely significant, keeping the Central Plateau Hills Region with lowest productivity as the base reference. About 60% of the variation in land productivity is accounted for by the variables chosen. Land productivity in value terms has higher elasticity with respect to rainfall, foodgrain abundance, and education levels. Table 2.5 reports the results for the second equation, explaining the variations in proportion of normal children. Land productivity is a significant variable in explaining CNS with an elasticity of 0.08; 1% increase in land productivity increases the percentage of nourished children below six year by about 0.08%. Higher level of women’s education has a positive impact while the higher rate of anaemia among pregnant women has an adverse impact on CNS, as may be expected. Higher rates of open defecation contribute to a (negative) elasticity of −0.07, while the coefficient of the percentage of shared toilets is not significant.

34 Swarna Sadasivam Vepa et al. Table 2.5 Estimated coefficients of child underweight across districts Variable

Coefficient

p-value

Land productivity per hectare (value in Rs./hectare) Women with secondary education and above (percent) Prevalence of anaemia among pregnant women (percent) Households with non-natural water (percent) Households without access to any toilet facility (percent) Households with access to shared (public) toilet facility (%) Prevalence of childhood (under three years) diarrhoea percent) Interaction between ORS and diarrhoea Children fully vaccinated (%) Population with access to any government facility State’s PDS grain offtake to allocation (%) Intercept R2

0.0822*** 0.0815* −0.0553** −0.0117 −0.0740*** −0.0073 −0.1211*** 0.0852*** 0.0622** 0.0720** 0.0019** 3.1928*** 0.2911

0.005 0.065 0.037 0.859 0.001 0.588 0.001 0.003 0.012 0.021 0.028 0.000

Note: *p < 0.10, **p < 0.05, *** p < 0.01

Since the data on quality of sanitation has been categorised into open defecation, shared toilets and private toilets (which is not included to avoid perfect colinearity), the insignificance of the coefficient of shared toilets shows that even that could be better in comparison to open defecation. Source of drinking water did matter once other variables are controlled for. It may have been expected that diarrhoea prevalence rate (among three-year-olds and below) is higher in regions with larger rates of open defecation, so that it may not be significant. However, this is not the case and the magnitude of this elasticity is highest after controlling for other variables. What is relevant from a policy perspective is that if there is higher rate of use of oral rehydration salts in the presence of diarrhoea (this is an interaction variable), then it improves the child’s nutritional status. Access to government health facilities also contributes to the reduction in underweight rates with an elasticity of 0.07. Districts in the states with higher PDS offtake also show a better rate of CNS. Thus, once again, it is substantiated that if policy implementation is better at a sub-national level, it has a clear impact in reducing underweight rates. In the CNS equation, the state dummies were not included as the PDS offtake is at the state level and causes perfect co-linearity. However, when agro-climatic zonal dummy variables were included, most of the coefficients in CNS—except education, anaemia and any government health facility variable, were insignificant and the overall goodness of fit of the model also improved. This clearly indicates that regional aspects encompass a large information base that captures institutional features and economic prosperity. Despite its better explanatory power, we do not prefer this model as it provides less information as to what features we could focus on in improving service delivery or programme implementation so that CNS is higher or underweight rates are lower.

Child underweight and land productivity 35

Conclusions Most of the literature on child underweight in India concentrates on childcare by women, health aspects of women and children, and education of women. The level of sanitation and availability of safe water have also received attention in explaining child undernutrition. Given a predominantly agricultural setting, and that undernutrition rates are more persistent in rural areas, this study examined the impact of agricultural land productivity in explaining child underweight rates in India. Alongside this, the study also focused on public provisioning of amenities like water, sanitation, and health care facilities as well as the foodgrains provisioning through PDS. It is observed that the variations in access and the quality of services depend primarily on governance and efficiency of the administrative system. A district-level study as opposed to a household-level study helps capture spillover effects of agriculture and policy issues of public provisioning in a better manner. At the district level, it has been possible to combine the data sets from different surveys pertaining to agriculture, which is not possible at the household level. This study, using district-level variations in underweight rates, has been able to establish the beneficial impact of agricultural land productivity on children’s underweight rates across districts of India for the first quinquennium of the twentyfirst century. This period was known of deceleration of agricultural production growth both in crop and livestock sector. Despite population pressure on land, deceleration in agricultural activity, agricultural land productivity was able to reduce poverty, mostly due to growth in real wages as well as low food prices that have welfare implications to the poor. The link from poverty reduction to decline in child undernutrition is not straightforward, since access to certain crucial services such as education, health, water and sanitation determine the child nutrition out comes. Only in cases where these enabling conditions are satisfied child underweights declines. Hence, the elasticity of agricultural land productivity with respect to child nutrition was small but significant. The major contribution of this chapter is in establishing the child-nutrition-toagriculture linkage in the case of India, which was suspected to be very weak. It shows that an increase of 1% in agricultural land productivity with its spillover effects could result in 0.08% improvement in child nutrition, after controlling for factors related to public provisioning, women’s health and children’s health, and women’s education. It appears that the association of land productivity to child nutrition is more obvious in the district context than in the household context. This is because it may be possible to identify convergence of various factors that interact with one another to produce better outcomes at the district level. As expected, variables in the land productivity equation show large elasticity estimates. As has been explained in Appendix 2B, the variables capture both direct and indirect effects. In this sense, agriculture has an indirect effect largely coming from improvements in net agricultural income or agricultural wages and allied activities due to improved productivity. Economists have already observed the phenomenon in India in the faster growth of rural non-farm employment. The

36 Swarna Sadasivam Vepa et al. increased productivity would also improve the supply of food so that access to nutrition-dense food such as milk and poultry increases for all. The results seem to suggest that foodgrain abundance and low prices would benefit all, along with wage and non-wage spillover benefits from food and non-food crops and forward and backward linkages of agriculture to non-farm employment. At least the 2002–04 data supports this view, though it is not possible to hypothesise that land productivity is the key variable that can capture the agriculture— and-child-nutrition association effectively. More research is needed in this area. The relevance of public provisioning and the quality of the implementation of services that have a direct impact on children’s nutritional status comes out again. The study clearly shows the contribution and convergence of several factors to bring about better outcomes. The results emphasise the points that creation of public awareness in terms of full dose of vaccinations, oral rehydration in emergencies, etc., contribute to better child nutrition. Public health facilities situated within the villages can go a long way in improving child nutrition in the district. Public provisioning is more important for the rural poor and hence the National Health Mission should be taken more seriously. A state’s commitment to public distribution of food to the poor has also been found to be relevant for child nutrition. Considering that the explanatory variables in the second equation do have a significant influence on the proportion of normal children in the district, even lower levels of elasticity—as low as 0.08 and 0.07—may mean larger gains over time, if the influence is at the margins where the deprivation is high. For example, any government facility in rural areas has an elasticity of 0.072, full vaccination elasticity of 0.06, and use of oral rehydration salts in diarrhoea incidence an elasticity of 0.085. Public provisioning of food, health, water, and sanitation services along with land productivity could bring about substantial improvements in child nutrition, since these variables affect marginal groups with less income and poor health status. The results of the study also emphasise the need to enhance agricultural land and worker productivity (achieve SDG-2 goal of doubling farm incomes). Spillover effect of agriculture creates service sector employment in rural and semi-urban areas, if infrastructure develops. Simultaneous efforts that strengthen public provisioning ultimately help to achieve the target of reducing child undernutrition by half. States in India are very large and there is substantial inter-district variation within a given state in terms of agro-climatic conditions, level of prosperity of agriculture, urbanisation and socio-cultural features, thereby contributing to large variations in economic and human development between districts. Consequently, the role of the district administration is increasingly becoming important in the implementation of both central and state schemes that focus either on agriculture or on undernutrition. There is need to pay more attention in linking agricultural development efforts to public provisioning of food, health care, sanitation and water, especially in rural areas where undernutrition rates are more persistent. Improvement of water supply and sanitation remains a major concern at the district level, as pointed out by other studies. It should be possible to use child-specific and region-specific information from anganwadis and other health centres to have a better-targeted intervention in

Child underweight and land productivity 37 pockets of high undernutrition. Regional-level analysis is also useful in policy interventions that focus on improving land and labour productivity, as farming households depend on regional soil and weather conditions. Thus, the results of the study demonstrate the possibility of big gains in child nutrition, where agricultural development efforts converge with public provisioning efforts.

Notes 1 The agrarian structure changed in terms of workforce composition. As per the census, total agricultural labour increased from 22.7% in 1981 to 40.3% in 1991, to 45.6% in 2001, and further to 54.9% in 2011. Absolute number of cultivators declined between 2001 and 2011, from 27.3 to 118.7 million and the number of agricultural labourers increased from 106.8 to 144.3 million for the same period (M.V. Nadkarni 2016). 2 1999–2000 consumption expenditure data problems of recall period and subsequent methodology changes and poverty line for rural India, call for dropping this year for long-term measurement purposes. 3 Land productivity, taken as an indicator to represent agricultural prosperity is not amenable to the inclusion of inland fisheries. In the context of farming system for nutrition, inland fisheries are more relevant. The value of inland fisheries and marine fisheries are not available separately from the data sources for our purpose. Further, the inland water bodies are not included in the land area 4 Cultivable area includes net sown area, cultivable waste (permanent and currant fallows) and area under miscellaneous crops not included under the net sown area. This pretty much includes all land used to produce crops and livestock products, as well as the ecosystem resources such as grazing lands green fodder for collection, vegetative cover, trees and small puddles of water and so on. 5 Appendix 2B elaborates the variables included in the land productivity equation. 6 A child is referred to as low-weight-for-age if its weight is below -2 standard deviations of the reference weight for age as specified by the World Health Organisation (WHO). Thus, for a given district the percentage of children whose z-score is below -2 are defined as moderately and severely undernourished. In DLHS-2, children below the age of six years had been considered for assessing the undernutrition rates among children. World Health Organization defines the three parameters of child malnutrition as follows: Underweight: weight for age < –2 standard deviations (SD) of the WHO Child Growth Standards median Stunting: height for age < –2 SD of the WHO Child Growth Standards median Wasting: weight for height < –2 SD of the WHO Child Growth Standards median 7 http://dfpd.nic.in/allocation-offtake.htm

References Antony, J. (2014) ‘Primary health care system in rural India: A critical review’, Journal of Research in Medical Education & Ethics, December, Vol. 4(1), pp. 23–28. Berman, P., Zeitlin, J., Roy, P., and Khumtakar, S. (1997) ‘Does maternal employment augment spending for children’s health care? A test from Haryana, India’, Health Transition Review, pp. 187–204. Behrman, J. R. and Deolalikar, A. B. (1987) ‘Will developing country nutrition improve with income? A case study for rural South India’, The Journal of Political Economy, June, Vol. 35(3), pp. 492–507.

38 Swarna Sadasivam Vepa et al. Bhagowalia, P., Purnima, M., Quisumbing, A. R., and Vidhya, S. (2012) ‘What Dimensions of Women’s Empowerment Matter Most for Child Nutrition? Evidence Using Nationally Representative Data from Bangladesh’, IFPRI Discussion Paper, No. 01192, International Food Policy Research Institute, Washington, DC. Bhalotra, S. (2010) ‘Fatal fluctuations? Cyclicality in infant mortality in India’, Journal of Development Economics, Vol. 93(1), pp. 7–19. Binswanger-Mkhize, H. P. (2013) ‘The stunted structural transformation of the Indian economy’, Economic and Political Weekly, June, Vol. 48(26–27), pp. 5–13. Calì, M. and Menon, C. (2013) ‘Does urbanization affect rural poverty? Evidence from Indian districts’, The World Bank Economic Review, Vol. 27(2), pp. 171–201. Published by: Oxford University Press. Chand, R., Gargand, S., and Pande, L. (2009) ‘Regional Variations in Agricultural Productivity: A District Level Study’, Discussion Paper NPP 01, National Centre for Agricultural Economics and Policy Research. Christiaensen, L., Demery, L., and Kuhl, J. (2006) ‘The Role of Agriculture in Poverty Reduction: An Empirical Perspective’, Policy Research Working Paper No. 4013, World Bank, Washington, DC. © World Bank, https://openknowledge.worldbank.org/handle/ 10986/9306 License: CC BY 3.0 IGO. Datt, G. and Ravallion, M. (1998) ‘Farm productivity and rural poverty’, Journal of Development Studies, Vol. 34(4). Department of International Development (DFID) (2004) Agricultural Growth and Poverty Reduction, Oxford: Policy Management Oxford, October. Dreze, J. and Khera, R. (2012) ‘Regional patterns of human and child deprivation in India’, Economic and Political Weekly, September, Vol. 47. Drèze, J. and Sen, A. (2013) An Uncertain Glory: India and Its Contradictions, Princeton, NJ: Princeton University Press. Fink, G., Gunther, I., and Hill, K. (2011) ‘The effect of water and sanitation on child health: Evidence from the demographic and health surveys 1986–2007’, International Journal of Epidemiology, October, Vol. 40(5), pp. 1196–1204. Galab, S. and Reddy, P. (2012) ‘Agriculture-Nutrition Linkages among Preschool Children in Andhra Pradesh’ (unpublished). Gibson, J., Datt, G., Murgai, R., and Ravallion, M. (2017) ‘For India’s Rural Poor, Growing Towns Matter More Than Growing Cities’, Policy Research Working Paper No. 7994, World Bank, Washington, DC. © World Bank, https://openknowledge.worldbank.org/ handle/10986/26244 License: CC BY 3.0 IGO. Gillespie, S., Harris, J., and Kadiyala, S. (2012) ‘The Agriculture-Nutrition Disconnect in India, What Do We Know?’ Discussion Paper 01187, June 2012, International Food Policy Research Institute (IFPRI). Girard, A. W., et al. (2012) ‘The effects of household food production strategies on the health and nutrition outcomes of women and young children: A systematic review’, Pediatric and Perinatal Epidemiology, Vol. 26(Suppl. 1), pp. 205–222. Global Nutrition Report (2018) ‘Development Initiatives, 2018 Global Nutrition Report: Shining a Light to Spur Action on Nutrition’, Development Initiatives, Bristol, UK, https://globalnutritionreport.org›reports›global-nutrition-report-2018 Gollin, D., Lagakos, D., and Waugh, M. E. (2014) ‘Agricultural productivity differences across countries’, American Economic Review. Gopalan, C. (2014) ‘The changing nutrition scenario’, Nutrition Foundation of India Bulletin, January, Vol. 35(1), pp. 1–4, http://nutritionfoundationofindia.org/pdfs/Bulletin Article/NFI_bull_04-2014.pdf

Child underweight and land productivity 39 Government of India (GoI) (2005) District Level Household Survey-2 (2002–04), New Delhi: Ministry of Health and Family Welfare. Government of India (GoI) (2011) Evaluation Study of National Rural Health Mission (NRHM) in 7 States, Programme Evaluation Organization, Planning Commission, Government of India Yojanan Bhavan, New Delhi-110001. Government of India (GOI), Ministry of Consumer Affairs, Departmrnt of Food and Public Distribution Annual reports. Government of India, (GOI), Planning Commission (2012) Twelfth Fiver Plan - 2012–2017, Chapter Two. Yojana Bhavan, New Delhi-110001. Hammer, J. and Dean, S. (2013) ‘Village Sanitation Externalities and Children’s Human Capital: Evidence from Randomized Experiment by the Maharashtra Government’, Policy Research Working Paper 6580, The World Bank Sustainable Development Network, http://riceinstitute.org/wordpress/wp-content/uploads/downloads/2013/02/Hammerand-Spears-2013.pdf Hayami, Y. and Ruttan, V. W. (1985) Agricultural Development: An International Perspective, Baltimore, MD: Johns Hopkins University Press. Headey, D., Chiu, A., and Kadiyala, S. (2011) ‘Agriculture’s Role in the Indian Enigma-Help or Hindrance to the Undernutrition Crisis?’, IFPRI Discussion Paper 01085, International Food Policy Research Institute, Washington, DC. HLPE Report-4 (2012) ‘Social Protection for Food Security – A Report by the High Level Panel of Expers on Food Security and Nutrition’, FAO, Rome, http://www.fao.org/cfs/ cfs-hlpe/en/ India Office of the Registrar General and Census Commissioner (2001), Census of India, New Delhi: Ministry of Home Affairs, Government of India. India Office of the Registrar General and Census Commissioner (2001), Census of India, New Delhi: Ministry of Home Affairs, Government of India. Yojana Bhavan, New Delhi-110002. International Food Policy Research Institute (IFPRI) (2018) ‘Status of Child Stunting in India’, Media fact sheet, May 25. Jensen, R. (2000) ‘Agricultural volatility and investments in children’, American Economic Review, Vol. 90(2), pp. 399–404. Jha, R., Sambit, B., and Raghav, G. (2011) ‘Social safety nets and nutrient deprivation: An analysis of the national rural employment guarantee program and the public distribution system in India’, Journal of Asian Economics, April, Vol. 22(2), pp. 189–201. Kadiyala, S., Harris, J., Headey, D., Yosef, S., and Gillespie, S. (2014) ‘Agriculture and nutrition in India: Mapping evidence to pathways’, Annals of the Newyork Academy of Sciences, Vol. 1331(1), pp. 43–56. Kaul, T. (2013) Household Responses to Food Subsidies: Evidence from India, College Park, MD, US: University of Maryland. Kaushal, N. and Muchomba, F. (2013) ‘How Consumer Price Subsidies Affect Nutrition’, NBER Working Paper 19404, National Bureau of Economics Research, Cambridge, MA, US. Khera, R. (2009) ‘Right to Food Act: Beyond Cheap Promises’, Economic and Political Weekly, July, Vol. 44(29), pp. 40–44. Kishore, A. and Chakrabarti, S. (2015) ‘Is More Inclusive More Effective? The “NewStyle” Public Distribution System in India’, Discussion Paper 01421, International Food Policy Research Institute, (IFPRI), Washington, DC. Kochar, A. (2005) ‘Can targeted food programs improve nutrition? An empirical analysis of India’s public distribution system’, Economic Development and Cultural Change, October, Vol. 54(1), pp. 203–235.

40 Swarna Sadasivam Vepa et al. Lim, S. S., Dandona, L., Hoisington, J. A., James, S. L., Hogan, M. C., and Gakidou, E. (2010) ‘India’s Janani Suraksha Yojana, a conditional cash transfer programme to increase births in health facilities: An impact evaluation’, The Lancet, June, Vol. 375(9730), pp. 2009–2023. Maitra, P. (2004) ‘Parental bargaining, health inputs and child mortality in India’, Journal of Health Economics, March, Vol. 23(2), pp. 259–291. Mishra, V. K. and Retherford, R. D. (2000) ‘Women’s Education Can Improve Child Nutrition in India’, NFHS Bulletin No. 15 (1–4). Nadkarni, M. V. (2016) ‘Indian agriculture through challenges: Old and new, Chapter 2’, in Vicissitudes of Agriculture in the Fast-Growing Indian Economy: Challenges, Strategies and the Way Forward, edited by C. Ramasamy and K. R. Ashok, New Delhi: Academic Foundation, pp. 71–86. Nestel, P., Bouis, H. E., Meenakshi, J. V., and Pfeiffer, W. (2006) ‘Biofortification of staple food crops’, The Journal of Nutrition, April, Vol. 136(4), pp. 1064–1067. NFHS-4 (2015–16) ‘India National Family Health Survey - 4’, International Institute for Population Studies, Bombay, Ministry of Health and Family Welfare, Government of India. Panda, A., Sharma, U., Ninan, K. N., and Patt, A. (2013) ‘Adaptive capacity contributing to improved agricultural productivity at the household level: Empirical findings highlighting the importance of crop insurance’, Global Environmental Change, August, Vol. 23(4), pp. 782–790. Paul, V. K., Sachdev, H. S., Mavalankar, D., Ramachandran, P., Sankar, M. J., Bhandari, N., and Kirkwood, B. (2011) ‘Reproductive health, and child health and nutrition in India: Meeting the challenge’, The Lancet, January, Vol. 377(9762), pp. 332–349. Radhakrishna, R. (2015) ‘Wellbeing, inequality, poverty and pathways out of poverty’, Economic and Political Weekly, October, Vol. 50(41). Radhakrishna, R. and Raju, S. (2015) ‘Well-being of agricultural households in post-reform period’, in Fast Growing Economy: Challenges, Strategies and Way Forward, edited by C. Ramaswamy and K. Ashok, New Delhi: Academic Foundation, pp. 151–174. Ramalingaswami, V., Jonsson, U., and Rohde, J. (1996) ‘The Asian Enigma’, in Progress of Nations (pp. 10–17), New York: UNICEF. Rangarajan, C. (2004) Agricultural Growth and Industrial Performance in India - In Selected Essays on Indian Economy (Vol. 1), New Delhi: Academic Foundation. Ravallion, M. (1990) ‘Income effects on undernutrition’, Economic Development and Cultural Change, April, Vol. 38(3), pp. 489–515. Ravallion, M. and Datt, G. (2002) ‘Why has economic growth been more pro-poor in some states of India than others?’, Journal of Development Economics, Vol. 68(2), pp. 381–400. Ravi, C. and Radhakrishna, R. (2004) ‘Malnutrition in India, Trends and determinants’, Economic and Political Weekly, February 14. Rose, E. (1999) ‘Consumption smoothing and excess female mortality in rural India’, Review of Economics and Statistics, Vol. 81(1), pp. 41–49. Ruel, M. T., Alderman, H., and Maternal and Child Nutrition Study Group (2013) ‘Nutritionsensitive interventions and programmes: How can they help to accelerate progress in improving maternal and child nutrition?’, The Lancet, August, Vol. 382(9891), pp. 536–551. Sambi Reddy, B., Ravi, C., Hanumanth Rao, C., and Radhakrishna, R. (2004) ‘Chronic poverty and malnutrition in the nineties’, Economic and Political Weekly, July 10, Vol. 39(28).

Child underweight and land productivity 41 Sen, A. K. (1992) ‘The Political Economy of Targeting’, Annual Bank Conference on Development Economics, World Bank, Washington, DC Spears, D. (2013) ‘How Much International Variation in Child Height Can Sanitation Explain?’, Policy Research Working Paper 6351 Water and Sanitation Program, Sustainable Development Network, World Bank, Washington, DC. Subramanian, S. and Deaton, A. (1996) ‘The demand for food and calories’, The Journal of Political Economy, Vol. 104(1), pp. 133–162. Subramanyam, M. A., Kawachi, I., Berkman, L. F., and Subramanian, S. V. (2010) ‘Socioeconomic inequalities in childhood undernutrition in India: Analyzing trends between 1992 and 2005’, PLoS One, Vol. 5(6), p. e11392, doi: 10.1371/journal. pone.0011392. Sundaram, K. (2007) ‘Employment and poverty in India, 2000–2005’, Economic and Political Weekly, pp. 3121–3131. Timmer, C. P. (1988) ‘The Agricultural Transformation, Chapter 8, Handbook of Development Economics’, in Handbook of Development Economics (Vol. 4) edited by H. B. Chenery, T. N. Srinivasan, T. P. Schultz, J. R. Behrman, J. Strauss, D. Rodrik, and M. R. Rosenzweig, North-Holland: Elsevier. Timmer, C. P. and Akkus, S. (2008) ‘The Structural Transformation as a Pathway Out of Poverty: Analytics, Empirics and Politics’, Working Paper Number 150, The Center for Global Development, Washington, DC. UNICEF (2018) ‘Progress for Every Child in the SDG Era’, www.unicef.org/media/48066/ file/Progress_for_Every_Child_in_the_SDG_Era.pdf. UNICEF/WHO (2009) ‘Diarrhoea: Why Children Are Still Dying and What Can Be Done?’, http://apps.who.int/iris/bitstream/10665/44174/1/9789241598415_eng.pdf. United Nations (2018) ‘Sustainable Development Goals’, www.un.org/sustainable development. Uma Kapila (edited) (2013) Indian Economy Since Independence, New Delhi: Academic Foundation. Vepa, S. S., Umashankar, V., Bhavani, R. V., and Parasar, R. (2014) ‘Agriculture and Child Under-Nutrition in India: A State Level Analysis’, MSE Working Paper 86/2014, Madras School of Economics, Chennai. Vepa, S. S., Viswanathan, B., Bhavani, R. V., and Parasar, R. (2015) ‘Child underweight and agricultural productivity in India: Implications to public provisioning and women’s agency’, Review of Radical Political Economy, December, Vol. 47(4), pp. 579–587, http://rrp.sagepub.com/content/47/4/579. Vepa, S. S., Viswanathan, B., Dhas, S., Umashankar, V., and Bhavani, R. V. (2016) ‘Child underweight and agriculture’, in Undernutrition in India and Public Policy, edited by S. Desai, A. Thorat, D. Chopra, and L. Haddad, New Delhi: Routledge. Viswanathan, B. (2014) ‘Prevalence of Undernutrition and Evidence on Interventions: Challenges for India’, Monograph No. 29, Madras School of Economics, Chennai, www. mse.ac.in/pub/Monograph-29.pdf. Viswanathan, B. (2015) ‘Farm Production Diversity, Household Dietary Diversity and Women’s BMI: A Study of Rural Indian Farm Households’, Monograph No. 34, Madras School of Economics Chennai, www.mse.ac.in/pub/Monograph-34.pdf. Viteri, F. E. (1994) ‘The consequences of Iron deficiency and anaemia in pregnancy on maternal health, the foetus and the infant’, SCN News, Vol. 11, pp. 14–18. Vollrath, D. (2007) ‘Land distribution and international agricultural productivity’, American Journal of Agricultural Economics, February, Vol. 89(1), pp. 202–216.

42 Swarna Sadasivam Vepa et al. Webb, P. and Kennedy, E. (2014) ‘Impact of agriculture on nutrition-nature of evidence and research gaps’, Food and Nutrition Bulletin Food and Nutrition Bulletin, Vol. 35(1) © 2014, The Nevin Scrimshaw International Nutrition Foundation. Wolfenson, K. D. M. (2013) ‘Coping with the Food and Agriculture Challenge: Smallholders’ Agenda’, Preparations and Outcomes of the 2012 United Nations Conference on Sustainable Development (Rio + 20), Natural Resources Management and Environment Department, FAO, Rome. World Bank (2007) ‘Agriculture and Rural Development Department, from Agriculture to Nutrition, Pathways Synergies and Outcomes’, Report no. 40196-GLB. World Bank (2018) ‘Atlas of Sustainable Development Goals from World Development Indicators’, https://optefau.files.wordpress.com/2018/06/9781464812507.pdf. World Health Organization (WHO) (2010a) Nutrition Landscape Information System: Country Profile Indicator Interpretation Guide. https://www.who.int/nutrition/nlis_ interpretationguide_isbn9789241599955/en/ World Health Organization (WHO) (2010b) ‘Water Sanitation and Hygiene’, www.who.int/ water_sanitation_health/en/.

Appendix 2A Data sources, variables and correlations Table 2A.1 Child underweight rates from various sources (2002–04 to 2012–13) State

RSOC$

DLHS-4$

2013–14 2011–13 Andhra Pradesh Arunachal Assam Bihar Chhattisgarh Goa Gujarat Haryana Himachal Jammu and Kashmir Jharkhand Karnataka Kerala Madhya Pradesh Maharashtra Manipur Meghalaya Mizoram Nagaland Odisha Pondicherry Punjab Rajasthan Sikkim Tamil Nadu Tripura Uttaranchal Uttar Pradesh West Bengal

22.3 24.6 22.2 37.1 47.1 16.2 33.6 22.7 19.5 15.4 42.1 28.9 18.5 36.1 25.2 14.1 30.9 14.8 19.5 34.3 16.0 31.2 15.8 23.3 30.5 20.6 34.3 30.0

27.3 27.3

29.5 NA 36.2 28.5 NA 29.7 20.9 38.7 27.4 30.5 27.2 25.5 23.8 25.2 23.6 32.5 27.7 37.4

AHS$

NFHS-3$ DLHS-2$ DLHS-2#

2012–13 2005–06

@

30.8 40.3 39.4 NA NA 45.7 40.6

38.9 36.6

28.0 44.9

32.5 32.5 35.4 55.9 33.9 25.0 44.6 39.6 36.5 25.6 56.4 37.6 22.9 60.0 37.0 22.1 48.8 19.9 25.2 40.1 24.9 19.7 29.8 39.0 38.0 42.4 38.7

2002–04

2002–04

43.5 22.2 33.2 54.3 48.2 28.4 44.9 37.2 37.4 29.3 52.1 44.1 35.2 53.9 46.4 15.0 36.4 16.8 22.8 43.0 26.8 41.5 61.4 12.9 39.9 33.2 59.5 57.2 43.6

42.3 20.3 32.2 54.6 47.4 30.0 46.0 35.6 36.4 22.6 52.2 44.8 35.8 55.4 47.7 12.6 34.9 15.2 21.4 42.8 26.8 40.0 58.3 9.7 38.3 30.2 55.3 52.6 44.9

Source: $: 0–59 months; #: 0–71 months; @: refers to Andhra Pradesh Residual; (1) RSOC: Rapid Survey of Children for 2013–14 [www.wcd.nic.in/]; (2) DLHS-4: District Level Household and Facility Survey for 2012–13 [http://rchiips.org/DLHS-4.html]; (3) AHS: [www.censusindia. gov.in/2011census/hh-series/cab.html]; (4) NFHS-3: National Family Health Survey for 2004– 05 [http://rchiips.org/nfhs/report.shtml]; (5) DLHS-2: Level Household and Facility Survey for 2002–04 [http://rchiips.org/PRCH-2.html]; (6) for DLHS-2 the underweight rates for 0–59 months was calculated from raw data and for 0–71 months was taken from the published reports; (7) NA: Not Available. The numbers for these two states were not reported by either of the two agencies in the respective website.

Table 2A.2 Descriptive statistics (level form) List of variables

Sample Mean size

Percentage of normal weight children Percentage of agricultural in district domestic product Triennium land productivity (Rs./hectare) Triennium agricultural DDP/worker (Rs./person) Triennium average annual rainfall (mm) Percentage of area under irrigation to GCA Gini coefficient of operational land Percentage of area under non-food crops to GCA Triennium average foodgrain production tonnes/capita Triennium average DDP from livestock Percentage of women with secondary and above education Percentage of rural women with secondary and above education Percentage of pregnant moderate and severe anaemia Percentage of households without access to any toilet facility Households with access to public toilets (%) Households with non-natural source of water (%) Children fully vaccinated (12–23) months (%) Percentage of under age three children with diarrhoea Percentage of children using oral rehydration salts (ORS) Interaction between ORS and diarrhoea Percentage of population with access to anganwadis Percentage of rural population with access to any one of the seven Off take as a percentage to its total allocation (%)

524 524

53.78 18.86

14.465 10.327

4.00 0.65

98.20 46.90

519

24.97

19.092

0.90

146.00

524

22.65

28.454

1.00

474.00

524

Std. dev

Min.

Max

1070.92 685.320 137.45 4869.75

522

41.36

28.994

0.00

104.56

523 523

0.48 20.25

0.078 18.939

0.10 0.14

0.83 86.43

522

224.74 223.114

2.53 1683.19

524 524

0.00 30.42

0.001 13.964

0.00 5.84

0.01 91.83

524

21.74

14.977

1.79

91.48

517

41.84

19.703

0.00

100.00

524

58.73

22.295

0.00

94.12

524

2.55

4.362

0.00

51.66

524

94.82

10.516

3.55

100.01

524

46.60

26.044

0.00

98.80

524

13.69

7.888

0.00

42.07

523

34.12

19.690

0.00

100.00

523 438

449.92 368.120 79.72 19.552

0.00 2648.67 2.10 100.00

438

48.25

19.154

5.00

100.00

524

44.12

19.234

5.88

96.15

Table 2A.3 Descriptive statistics of variables (in logarithmic form) Variables

Mean Std. dev Min.

Percentage of normal weight children 3.94 0.302 Triennium land productivity (Rs./hectare) 2.96 0.737 Triennium average annual rainfall (mm) 6.81 0.583 Percentage of area under irrigation to GCA 3.36 1.065 Gini coefficient of operational land −0.75 0.181 Percentage of area under non-food crops to GCA 2.07 1.586 Triennium average foodgrain production per capita 5.10 0.840 (tonnes/capita) Percentage of women with secondary and above education 3.32 0.437 Percentage of pregnant women with moderate/severe anaemia 3.63 0.599 Percentage of households without any toilet facility 3.88 0.915 Percentage of households with access to public toilets 0.32 1.099 Percentage of households with non-natural water sources 4.54 0.211 Percentage of children fully vaccinated (12–23 months) 3.63 0.756 Percentage of under age three children with diarrhoea 2.39 0.776 Interaction between diarrhoea and ORS use 5.81 0.894 Percentage of rural population with access to any one of 3.78 0.466 the seven health facilities Percentage of state’s offtake to PDS allocation 44.12 19.234

Max.

1.39 4.59 −0.11 4.98 4.92 8.49 −5.24 4.65 −2.30 −0.19 −4.40 4.52 0.93 7.43 1.76 0.96 −2.81 −3.51 1.27 −0.36 −1.24 2.75 1.61

4.52 4.61 4.54 3.94 4.61 4.59 3.74 7.88 4.61

5.88 96.15

Table 2A.4 Correlation matrix for stunting, underweight and wasting, NFHS-4 Pairwise correlation between stunting, underweight and wasting Composite Andhra Pradesh Stunting z-score Underweight z-score Wasting z-score

Stunting z-score

Underweight z-score

Wasting z-score

1 0.7084*** −0.0503***

1 0.655***

1

Pairwise correlation between stunting, underweight and wasting (Telangana) Stunting z-score Underweight z-score Wasting z-score

Stunting z-score

Underweight z-score

Wasting z-score

1 0.7187*** −0.0365

1 0.6536***

1

Pairwise correlation between stunting, underweight and wasting (Andhra Pradesh Residual) Stunting z-score Underweight z-score Wasting z-score

Stunting z-score

Underweight z-score

Wasting z-score

1 0.7003*** −0.0605***

1 0.6566***

1

Source: Authors’ calculation using NFHS-4 data for the states of Andhra Pradesh and Telangana Note: ***p< 0.001

Child nourishment rate (%) Women with secondary education and above (%) Prevalence of anaemia among pregnant women (%) Households without access to any toilet facility (%) Households with access to shared (public) toilet facility (%) Households with non-natural water (%) Children fully vaccinated (%) Prevalence of childhood (under three years) diarrhoea (%) Interaction between ORS and diarrhoea Prevalence rate for access to any type of health care (%) PDS offtake of the state as percentage of state’s allocation (%) 0.296* 0.126* 0.198*

0.031

1

5

1

6

7

0.016 0.089

8

1

0.017

−0.326*

0.1138* 0.670* 0.1134* −0.068

1 0.214* 1 0.095 −0.0818

0.248* −0.170*

−0.142* −0.066 −0.252* 0.157*

0.264* −0.008 0.199* −0.041 0.073 −0.153*

−0.354* −0.207*

0.093 −0.02 0.021 0.224* 0.153* −0.060

0.258*

4

0.139* −0.152* −0.308*

−0.103* −0.084 0.199* 0.423* −0.205* −0.208*

0.135*

0.331*

−0.355* −0.337*

1

3

1

1 0.363*

2

−0.254* −0.115*

1

Table 2A.5 Correlation matrix of variables used in the child nourishment equation 9

0.049

1 0.154*

11

−0.002 1

1

10

Appendix 2B Description of variables in the estimation of land productivity

Estimates of land productivity Three-year average value of district gross domestic product from crops and livestock sources at constant prices per hectare of cultivable land: This includes net areas sown, current fallows and permanent fallows, and area under miscellaneous crops not included in the net area sown area. The district GDP data is from Indicus Analytics. The data pertains to 2002, 2003 and 2004, to correspond to the DLHS-2 reference years of 2002–2004. Land productivity has been estimated as a function of six variables in addition to 13 agro-climatic dummy variables. Fourteen major agro-climatic zones have been taken into consideration. The Central Plateau and hilly region with lowest land productivity has been used as the base and 13 dummy variables have been created. The other six variables included in the equation are as follows: 1

2

Three-year average (2002, 2003 and 2004) of annual rainfall in millimetres at the district level: This is an important variable for any study related to agriculture. High rainfall regions have a better potential of land productivity in general. Further, the level of rainfall determines the crop pattern. Some of the low rainfall regions achieve agricultural growth by growing high value crops such as cotton and oilseeds. Rainfall levels also influence groundwater recharge and surface water collected in ponds and tanks. The data have been taken from the India Meteorology Department and expressed in millimetres for all the months in the year for 2002, 2003 and 2004. The annual rainfall figures have been averaged to arrive at the annual average for three years, using the 2001 district list. Unfortunately, 2003 rainfall data was not complete for all the districts and hence only 2002 and 2004 data had to be averaged. In any case, averaging the annual rainfall data gives a better idea of rainfall received for the reference period. Three-year average (2002, 2003 and 2004) of gross irrigated area as a percentage of three-year average of total cropped area in the district: Irrigation influences land productivity, in conjunction with rainfall. Canal irrigation can help even when rainfall is deficient. Canal irrigation can also recharge groundwater to some extent in the winter season and summer season, when canal water is not available.

48 Swarna Sadasivam Vepa et al.

3

4

5

6

The data on area irrigated as well as the total cropped area for each year have been taken from the Ministry of Agriculture and Co-operation. Land operational inequality at the district level represented by Gini ratio: Unequal distribution of operated land in agriculture seems to reduce overall agricultural land productivity. One standard deviation reduction in land operational inequality seems to bring about an increase of 8.5% in land productivity (Vollrath 2007). Hence, it has been considered as an important variable for the determination of land productivity. The distribution of operated land in various size classes for the year 2004–05 for each district is given by the agricultural census data collected by the Ministry of Agriculture and Co-operation. Gini ratio has been calculated from the data for each district. Three-year average of area under non-food crops as a percentage of total cropped area in 2004–05: Non-food crops are important as some of them are high value crops and help the prosperity of low rainfall regions. Reliable data on all crops and total cropped area was available from census at the district level and hence a single year data for 2004–05 has been used. Non-food crops’ contribution to land productivity also captures the spillover effects of agricultural prosperity to non-agricultural employment in the district. Higher land productivity from non-food crops may mean more jobs in the non-crop sector and more incomes and higher prosperity. Per capita foodgrain production in the district in tonnes per person per annum: This variable basically captures the foodgrain abundance in the district. Though the low percentage of non-food crops means high percentage of food crops, it may not amount to high land productivity from food-grains. Foodgrain abundance means higher yields and more land productivity. Further, food-abundance districts in contrast to food-deficit districts also have spillover effects of non-farm employment in trade, transport and processing. The contribution to land productivity from abundance of food-grain production would mean cheaper local grain to the population and better calorie consumption. Three-year average (2002, 2003 and 2004) of food-grain production at the district level has been taken from the Ministry of Agriculture and Co-operation and divided by the projected district population for the year 2004, to get the per capita annual food-grain production in the district as tonnes/per capita. Women’s education is a control variable in the land productivity equation to capture awareness and adoption of technology, capability of acquiring skills relevant for crop and animal production, awareness of bank credit and the ability to take advantage of a number of schemes that the government offers. It is a proxy for all education, including that of men (as men’s education is closely correlated with women’s education), in the land productivity equation. Higher levels of education help skill acquisition and improve technology adoption in agriculture and result in higher land productivity (Chaudhri 1968; Panda et al. 2013). Reliable district data on the percentage of women who received education above secondary level was available in the DLHS data.

3

Child nutrition Linkages to agriculture Anusha Ganapati and Swarna Sadasivam Vepa

3.1

Introduction

The fourth round of the National Family Health Survey (NFHS-4) in India indicated a decline in the under-five mortality rate in the country, to 50 deaths per 1,000 in 2015–16 from 74 in 2005–06. According to a recent UNICEF report for 2019, under-five mortality (deaths) rates stood at 43 per 1,000 live births, while neonatal mortality (deaths in the first 28 days after birth) rates stood at 24 per 1,000 live births. Under-five stunting rates stood at 38% and wasting rates at 21%. India is expected to be on track to achieve 2030 sustainable development goals, with respect to under-five mortality rates and under-five child obesity rate that stands at 2% (while the target is high at 5%). It is likely to miss the 2030 child nutrition targets of stunting and wasting and neonatal mortality. Hence India needs efforts for accelerated reduction (UNICEF, India Country profile 2019). Malnutrition among children under five, causes a reduction in productivity through weak physical status and lowers cognitive development of children (WHO 2010a). India’s future progress, and educational attainment of children, depends upon India’s ability to reduce child undernutrition. Both India report as well as UNICEF (2018) earlier report on Millennium Development goals agree that India could reach the first sub-goal of MDG1 to “Reduce by half the number of people who live on less than a dollar a day,” but not at all, on target to meet the second sub-goal to “Reduce by half the number of people who suffer from hunger (UNDP 2015). A recent report of FAO, State of Food Security in the World 2018, puts India’s undernourished at 195 million people. It includes children (FAO 2019). The urgency of reducing child undernutrition in India cannot be emphasised more. The factors that contribute to the reduction in child undernutrition in children under five is the focus of this chapter. The reason for focussing on agriculture is that about 44% of the population depends upon Agriculture in India. Agriculture is diversifying and undergoing change. High value agriculture is growing faster than the foodgrains-based agriculture. Though transformation of agriculture into high productivity sector is not anywhere in sight. It is of interest to examine if there is a link between agricultural prosperity and child undernutrition at the district level.

50 Anusha. G and Swarna Sadasivam Vepa Two of the three child undernutrition metrics of under-five children declined over time (NFHS-4, India fact sheet). Stunting rate (height for age below 2 standard deviations from the WHO norm) fell to 38% in 2015–16 from 48% in 2005– 06. Underweight rate (weight for age below 2 standard deviations from the WHO norm) of under-five children fell to 38.4% in 2015–16 from 43% in 2005–06. However, the wasting rates (weight for height below 3 standard deviations from WHO norm) in children under five increased to 21.0% in 20015–16, from 19.8% in 2005–06 (NFHS-4, India fact sheet). As stunting and underweight rates are falling, wasting rate is increasing in India. Underweight shows a strong correlation with stunting as well as wasting and it is a composite measure (WHO 2010a). There is interesting emerging evidence that weight-for-age (underweight) in addition to mid-upper arm circumference (MUAC) is the most reliable way of detecting children who are at a risk of mortality (Global Nutrition Report 2018). Mortality risk associated with underweight, in children under five seem to be high (WHO 2010a). Underweight in children under five thus reflects stunting, wasting and the risk of morality. Poverty reducing effect of agricultural growth has been established in the context of many developing countries, including India (World Bank 2007; de Janvry and Sadoulet 2010; Ravallion and Datt 1996, Datt and Ravallion 2002). About 44% of the population depends upon agriculture in India in 2017 (Goldar et al. 2017). Though the share of agriculture in the overall GDP has come down (15%),1 wellbeing of the rural and semi-urban population, depends upon agricultural prosperity. Agriculture’s ability to reduce child nutrition via growth and poverty reduction depends upon the importance of agriculture in the overall economy, the stage of structural transformation and other challenges (Pingali, 2015). Other challenges include women’s literacy and effective public service delivery of health, sanitation and safe water supply. Structural transformation of the Indian economy has been incomplete and stunted, as India failed to shift its workforce from agriculture to labour intensive manufacturing as it happened in many east Asian and south east Asian economies. Stage of agricultural transformation assumes importance for two reasons. First, on the demand side, the food basket gets diversified as the country achieves poverty reduction (Pingali et al. 2017). Second, income earning opportunities get diversified into agricultural output based non-farm activities (Hans P. BinswangerMkhize 2013). Local agricultural production link to rural poverty, and urbanisation seem to exist in India for a long time. Massimilano Calì and Menon (2013) show that in the case of India, between 1983–1999, urbanisation helped to reduce rural poverty. The rural poverty reduction effect of urbanisation is primarily due to increased demand for local agricultural products rather than urban-rural remittances. Prior to 1991, models that linked agriculture and industry have shown that a 10% increase in agricultural output would increase industrial output by 5% and urban workers would benefit by both industrial employment and price deflation (Rangarajan 1982; de Janvry and Subbarao 1986; Radhakrishna and Sarma 1976; Radhakrishna and Murthy 1982). Agriculture industry link may have become weak, but still exists in India.

Child nutrition 51 Higher farm productivity reduced rural poverty through wages and prices (Datt and Ravallion 1998). During the green revolution period in the 1980s and mid1990s, agricultural growth in India was driven by foodgrain production and commercial activity related to surplus production for the market. Since the 1990s, Indian agricultural growth was driven by diversification into high-value products such as fruits, vegetables, cotton, dairy, poultry, meat and fisheries. Diversification and commercial activity create spillover income effect in trade transport storage processing etc., spread across rural and semi-urban areas, blurring the rural urban divide (Timmer and Akkus 2008). A recent study using cross-section regression analysis for India has shown that monthly per capita expenditure and incidence of poverty among self-employed households are positively affected by agricultural productivity per net area sown, extent of irrigation, urbanisation and road density (Radhakrishna 2002). Radhakrishna and Raju (2015). Rural non-farm sector, which is predominantly informal, grew at the rate of 7.1% between 1993–2004 in terms of GDP. The employment growth in this sector was 3.7% per annum (Ramesh Chand 2017). Between 2004–05 to 2011–12, non-agricultural GDP growth in rural India, accelerated to 9.21% per annum but employment growth slightly decelerated to 3.65% per annum. As pointed out by Ramesh Chand, this happened in the face of overall decline in employment in rural sector by 0.28 percent and a decline in agricultural employment by about 2% per annum (Ramesh Chand 2017). Diversification of agriculture which picked up after 2007, helped growth of agriculture in value terms (Appendix 3A2, Table 3B.1). Fall in workers in agriculture by about 10% (National Sample survey 2011–12) after 2011, increased per worker productivity of agriculture.2 Though employment generation capacity of agriculture declined between 2004–05 and 2011–12, agriculture by itself grew at an accelerated rate of 4.27% per annum, mostly due to high value components of agriculture and increased agricultural prices. Non-farm employment generation and income generation driven by agricultural production ought to be higher, if we add non-farm activities driven by agriculture in semi-urban areas to the rural non-farm employment. Economic activities related to dairy products, poultry products, vegetables and fruits that need cold storages and processing units typically occur in semi-urban districts. Even if service sector contribution to district GDP is higher, most of the activities are linked to agriculture. Secondary sector is small and some of the processing activity is again related agriculture. Hence, we ignore the sectors other than agriculture in this chapter for semi-urban districts of India. The pace of urbanisation has been slow, at about 9% between 2001 and 2011. The pattern of urbanisation has shown accelerated growth in the population of towns (class 2 and class 3 cities) and a deceleration of population growth of class one cities, with a population of a million or more. Population increase in mega cities of more than 10 million decelerated to 12.05% between 2001 and 2011 compared to 30.47% between 1991 and 2001 (India Office of the Registrar General and Census Commissioner 2011). Big villages become small towns and small towns become big towns over time. Population growth of rural areas decelerates. Agricultural trade, and processing occur typically in small towns rather than in big

52 Anusha. G and Swarna Sadasivam Vepa cities. In addition, agricultural prosperity benefits the rural as well as urban poor via low food prices. Treating agricultural economy synonymous to rural economy may lead to a possible underestimation of the indirect impact of agriculture on poverty and child nutrition. This is the main reason why we considered all districts which are partly rural and excluded all 100% urban districts to examine the link of agricultural productivity to child nutrition, at the district level. The transmission process of agricultural prosperity into poverty reduction and improvement in child nutrition may also differ from one period to the other, depending upon the stage and type of structural transformation under way. This chapter explores agricultural-nutrition linkage, in the context of rural urban overlap at the district level. It is obvious that agricultural prosperity per se cannot reduce child malnutrition without the existence of other enabling factors. Maternal health care and child health care, sanitation, and safe water reduce illness due to diseases such as pneumonia and diarrhoea and preventable deaths. Water and sanitation are linked to Child diarrhoea. Jyostan Jalan and Martin Ravallion (2001) using propensity score matching method have shown that the prevalence and duration of diarrhoea among children under five are significantly less among the piped water users compared to those who do not. Women’s education promotes child nutrition. Benefits of women’s education accrue from delayed marriage, spacing of children, lower fertility rates, better feeding practices, neo natal care, timely immunisation, better health care such as use of oral re-hydration (ORS) during the incidence of diarrhoea and so on. Nutrition awareness and health seeking behaviour can be attributed to women’s education. This chapter considers the district as the primary unit of observation. It takes a refreshed look at the linkages of agricultural productivity, child health and women’s education to understand the convergence of these factors in affecting child underweight status. A district level study gives a broader picture which is not captured at the household level. Districts are administrative units, allowing for the confluence of other effects, interactions and externalities. This makes the study relevant for policy implications. Moreover, agriculture-related information is easily available at a district level. Another significance of the study is that agriculture gross domestic product per worker and per hectare include both crop production and livestock production. Land includes all cultivable land available, and just not net area sown. District level data, unlike unit level, also allows for a panel analysis of the district observations across time-periods. This chapter seeks to answer the following question: 1

Is there an effect of agricultural land productivity and agricultural worker productivity on child underweight rates at the district level?

This chapter has four sections including this introductory section. The second section gives a literature review of the links of child undernutrition to agricultural productivity on one hand and health care and women’s education on the other. The

Child nutrition 53 third section gives the details of data and methodology, and the fourth section gives the results and the conclusions.

3.2

Child nutrition links to agriculture, health and women’s education

Various interrelated factors influence child nutrition. This section reviews past literature on these issues. The most common approach explains child undernutrition in terms of maternal health, child health, sanitation, mother’s education status and income status of the household. Some studies link child nutrition to rainfall, agricultural production, productivity, land ownership, homegrown produce and so on. The study of IFPRI (2019) using NFHS-4 data at the district level reveals the importance of women’s education. The study explained 70% of the differences between districts with high burden of stunting of more than 40% and the others. Women’s education contributed to 12% of the difference. The contributors to the difference were body mass index (19%) children’s adequate diet (9%), assets (7%), open defecation (7%), age at marriage (7%), antenatal care (6%), and household size (5%) as per the study. Lokshin and others (2005) analysed the first two rounds of National Family Health Surveys (NFHS-1 and NFHS-2) to identify trends in child malnutrition between 1992 and 1998 and assessed the impact of the Integrated Child Development Services programme. They used principle component analysis to condense the asset ownership to construct a linear index to determine economic status. The results reveal a high incidence of underweight amongst children in higher socio-economic groups—suggesting that exposure to disease cannot be attributed solely to poverty. Pathak and Singh (2011) used three rounds of NFHS data to construct wealth indices to ascertain the effect of economic status on the likelihood of child malnutrition over time. The results reveal the heterogeneity in malnutrition and economic status across different states. The study identifies differences in the rate of change and confirms that the disparities in average malnutrition between the rich and the poor have widened over the period 1992–2006. In addition to consideration of economic factors, some papers have also highlighted the need to consider biophysical and geographic variables in affecting under-five child mortality. Public health research focused on understanding the health of population in different geographical regions or spaces— using geospatial analysis. These geographical and biophysical variables include rainfall, temperature, productivity of agricultural lands, distance to urban areas, malaria endemicity, frequency of drought, road networks etc. Kumar et al. (2012) in their paper used these variables alongside a “Coverage-Gap Index” which was constructed on the basis reproductive and health care variables. Controlling for biophysical and geographic variables, the spatial regression model revealed that level of urbanisation, female literacy, number of new-borns cared for, in Primary Health Centres (PHCs), appeared negatively correlated to under-five mortality. Low economic status (represented by BPL cardholders), and coverage gap in health services were positively correlated to under-five mortality. Mosley

54 Anusha. G and Swarna Sadasivam Vepa and Chen’s framework (Mosley and Chen 1984) discusses five broad groups of “proximal determinants” affecting disease incidence. These determinants include maternal factors, nutrient deficiency, environmental contamination, injury and personal illness control—characterised by the availability of health services and the capacity to use them. Cavatorta et al. (2015) utilise data from NFHS 3 and consider one state—Tamil Nadu—as a benchmark for nutritional outcomes and compare the performance of height-for-age of other states that perform poorly on the height for age ‘Z’ score (HAZ) outcomes. To compare the differences in outcomes, the paper uses the counterfactual decomposition method. Counterfactual decomposition used in this paper assumes “ignorability,” or that un-observables are independent of the treatment conditional on observed covariates. For example—“nutrition consciousness” may be one of the unobservable factors in this paper—however, when the equation controls for mother’s education, the distribution of the unobservable variable is not systematically related to the state under consideration. The results of the paper revealed that the HAZ differences between Tamil Nadu and other states were significantly different, and there were also differences in terms of key covariates. Mother’s education, endowment of assets and likelihood of mother’s being anaemic all indicate that TN indicators are better than that of the other states included for comparison. Only in agricultural assets, TN has worse endowments than the comparators. Among agricultural assets, land ownership presented large and consistent relationships with HAZ across states. The other variables including those on sanitation displayed less consistent associations with HAZ. This paper also reveals the importance of holding land and agricultural property—and the differing agricultural endowments help explain child nutrition. The Indian Household Demographic Surveys (IHDS) contain disaggregated data on land ownership, farming practises, and crop diversity in addition to anthropometric measurements of children, expenditure on food consumption and household income. The study by Bhagowalia et al. (2012) uses IHDS data on height-for-age scores as a measure of chronic malnutrition, and attempts to explain it using variables on food production (to establish dietary diversity), caregiver resources (maternal indicators of health, education), health environment of the household (including health seeking behaviour), demographic characteristics and income quintiles (to allow for non-linearity in relationships). The nutrition-income relationships are disaggregated by rural and urban households, and agricultural and non-agricultural households. The results indicate that income has a significant effect on HAZ, but the non-linearity of the relationship implies that income growth might have little effect on stunting unless it involves moving to a much higher level of income. The paper also looks at region-wise differences in HAZ— which indicates that HAZ is higher in non-agricultural households in all regions except the northeast. However, the regional dummies are not significant, and the paper attributes it to the relatively little spatial clustering of nutrition. In terms of dietary diversity—region, religion and non-agricultural versus agricultural differences exist in consumption. As assumed, higher income quantiles have greater dietary diversity relative to the poorest income quantiles. Moreover, ownership

Child nutrition 55 of agricultural assets also has impact on dietary diversity—through increased consumption due to self-production, or due to increased purchasing power. It is essential to note that undernutrition is greater in regions where there is lesser access to basic amenities—which are also the regions that are largely dependent on agriculture. Hagos et al. (2014) incorporate panel data of 43 administrative zones in Ethiopia to understand the relationship between climate change, crop production and child undernutrition. Frequent rain failures put Ethiopia under recurrent famines—the irregularity of rainfall thus has a direct impact on food availability in the region. This chapter demonstrates the association between amount of rainfall and stunting in the region and the significant impact of the agro-climatic zones on child underweight. A global study on maternal and child undernutrition (Black E. Robert et al. 2008) reveals the consequence of suboptimal breast-feeding practises and vitamin deficiencies as a major cause of severe wasting, stunting—in turn leading to high under-five mortality rates in developing countries. Mahmud and Mbuya’s (2016) World Bank study on Bangladesh discusses the need for a multi-sector approach to nutritional enhancement through improvement in water and sanitation, to limit transmission of infections. Mishra and Retherford (2000) found that more educated women had significantly healthier children even after controlling for demographic and socio-economic variables. A paper by E. Liu et al. (2016) studies unit-level data from IHDS which indicates that access to water increases the children’s likelihood of having normal weight status by 2%, highlighting the importance of access to clean water. An earlier paper by Vepa et al. (2015, 2016) and Chapter 2 of this book use district-level analysis to better capture the spillover effects of agricultural development, public provisioning and agro-climatic conditions on child underweight rates. Using a three-stage least squares model, which indicates a positive relationship between agricultural land productivity and child underweight rates, the chapter shows a significant influence of public health facilities, maternal health and education on the child underweight. The convergence of factors at a district level allows for the interaction of the factors and explain that a 1% increase in agricultural land productivity would result in a 0.08% improvement in child nutrition given the public provisioning of services, women and children’s health and education—indicating that this may have an impact at the margins. This chapter uses agricultural land productivity and agricultural labour productivity directly as independent variable at the district level to explain child underweight. In contrast to the normal children (non-underweight children) considered in Chapter 2, this chapter considers the impact agricultural productivity on the percentage of children under weight for age. This chapter looks at the pooled panel data from DLHS-2 (2002–04) and NFHS-4 (2015–16). The analysis uses a limited number of variables which are available in most of the districts in the surveys. Since the data sources are different, definitions do not match. Data gaps were also large for many variables.

56 Anusha. G and Swarna Sadasivam Vepa

3.3 Hypothesis, data, concepts and methods We hypothesise an association between agricultural land productivity/agricultural worker productivity, with the percentage of underweight children under the age of five years or six years, at the district level. Higher land productivity and worker productivity lead to poverty reduction and lower percentage of underweight children, given the enabling conditions of child health and maternal education. Land productivity enhancement has spillover effects outside agriculture. The land ownership and operational inequality may result in unequal gains to those dependent on agriculture, especially keeping the landless at the bottom. Yet, it is reasonable to believe that trickle-down effect as well as spillover effect would be substantial to reduce poverty and child undernutrition both in rural areas and semi-urban areas. Jose (2013) has shown that between 2004–05 and 2009–10, the index of rural real wage increased from 110.88 to 132.01 for males and from 108.82 to 138.55 for females. Diversification of agriculture (horticulture and livestock sectors) speeds up the commercialisation and spillover effects. Poverty reduction coupled with increased years of schooling of young women and better health care for children, would result in reduced incidence of child underweight. 3.3.1

Sources of data and limitations

Data to test the hypothesis has several limitations. Ideally, we need time series district data on agriculture as well as child nutrition variables. In the absence of such data, compilation of panel data poses several problems. For example, splitting several data sets across districts that have been bifurcated or trifurcated since 2001 was necessary for the present study to make the 2001 district list consistent with the 2011 district list. National Family Health Surveys (NFHS) are not strictly comparable to District Level Health Surveys (DLHS). District level NFHS data are available only for 1998, and 2015, but the further back in time we go, more adjustments and consolidation become necessary for district level data. It may result in loss of information and affects the more recently bifurcated states. One to one correspondence of the reference period for all variables is not possible, as sources differ. Each data source has a different reference period. Definitions also vary. It is not possible to make a panel with two surveys of DLHS or two surveys of NFHS in recent times.3 Only DLHS-2 and NFHS-4 provide district level data on one aspect of child nutrition. While DLHS-2 survey was spread over 2002–04, the NFHS-4 survey refers to 2015–16, financial year. District GDP data has comparability issues over period, beyond 2014. Land utilisation statistics used to calculate land productivity had many gaps. The estimations in this chapter use percentage of underweight children4 below 2 standard deviation of the WHO norm as dependent variable. Explanatory variables consist of agricultural land/worker productivity, percentage of women with ten or more years of education and incidence of diarrhoea among children. The dependent variable and the explanatory variable, apart from agricultural productivity come from the District Level Health Survey 2 (2002–04) and National Family

Child nutrition 57 Health Survey 4 (2015–16). Indicus analytics district GDP data set provides information on agricultural GDP (crop and livestock) and requisite information to calculate triennium average per worker GDP at constant prices in thousand rupees per worker per annum. District agricultural GDP and per worker agricultural GDP at constant prices have been averaged over period one (2001–02, 2002–03, 2003–04) and period two (2011–12, 2012–13, 2013–14)5 to make a panel. We divided the three-year average agricultural GDP for 2001–04 with three-year average cultivable area for those three years, to get agricultural land productivity for the first period. Cultivable area consists of net area sown, current and permanent fallows, and cultivable waste. Land utilisation statistics of the Ministry of Agriculture and Co-operation, for the years 2001–02, 2002–03 and 2003–04, form the basis of cultivable area for period one. For period two, we could not get land utilisation statistics and hence used the one period cultivable land data with similar definition available from and agricultural census 2011. The rationale is that cultivable area does not change much over a three-year period. Changes in net sown area lead to corresponding opposite change in fallow land and cultivable waste. The data sets and the years of reference are not strictly comparable between the periods chosen to make a panel. Despite gaps, we assumed that short-term changes are normally small and negligible in the absence of a major disasters. Averaging the agricultural data over three years irons out a few fluctuations that may remain. Lack of data on cultivable area beyond 2011, and lack of comparable district GDP data beyond 2014 have been major limitations. Appendix 3A2, Appendix 3B, table 3B.3 gives the details of data sources. Based on the dimensions discussed in the papers mentioned in the literature review, the three explanatory variables chosen could cover multitude of dimensions to help explain child underweight in the country. However, the number of explanatory variables from DLHS-2 and NFHS-4 that could be included in the panel data analysis were limited. Many variables were also closely correlated with each other. Many districts did not have information on anaemia of children. Many variables had different definitions, which makes matching rather difficult. Sometimes data reported coverage for variables such as immunisation at a high and unrealistic level, rendering it less useful as a variable. Further there was no clue as to the quality of service available for example for water or toilets. Contamination of water and pollution of air with faecal matter is always a threat in rural and semi-urban areas. Availability of toilets—even if they are flush toilets, they could be unhygienic, if they lack 24 × 7 water supply and are not connected to a proper drainage and treatment and disposal system far away from the dwellings. Shallow hand pumps closer to the dwellings with open disposal of wastewater are likely to provide contaminated water. Many districts did not report the data on some variable such as anaemia prevalence among children. Hence only two key variables that could broadly proxy for others seem to be percentage of women with schooling of more than ten years, and percentage of children suffering from diarrhoea. Normally educated women take care of immunisation, feeding practices, and display health seeking behaviour for the children and themselves. Educated women will also be more aware of the government advertising on nutrition and cleanliness.

58 Anusha. G and Swarna Sadasivam Vepa They also utilise the public services better than the less educated. In the districts with large percentage of educated women, men’s educational levels will also be high. Their incomes could also be higher. This is a sort of catch-all variable. Percentage of children who reported to have an incidence of diarrhoea is another key variable. NFHS-4 gives the information for children below five, while DLHS-2 gives information on incidence of diarrhoea for children below the age of three. This variable reflects the “state of affairs” with sanitation, safe drinking water, effectiveness of health care provisioning etc. This is a proxy for WASH (water sanitation and health). Panel data analysis includes the following variables: 1 2

Dependent variable: Percentage of Underweight children under five/six below 2 standards of deviation of the WHO norm (DLHS-2 and NFHS-4). Explanatory variables: a

b c

Triennium average Agricultural (Crop + Livestock) Land Productivity per hectare of cultivable Land at constant prices/Triennium average Agricultural (crop + livestock) worker Productivity at constant prices (Triennium averages of GDP are over 2001–04, and 2011–14 compiled by Indicus analytics based on Central Statistical Organisation data) Percentage of women with ten years of schooling or more (DLHS-2 and NFHS-4) Percentage of children below the age of five/three years who had incidence of diarrhoea (DLHS-2 and NFHS-4)

District matching exercise and data gaps reduce the number of observations. The data for the two time periods across a variety of variables differ in the number of observations. NFHS-4 data covered about 670 districts, but data were not available for all districts. DLHS-2 data covers more than 580 districts. Depending upon the data used, the number of observations differ. Agricultural land utilisation data as well as GDP data has gaps. Compilation of data for two time periods uses 2011 census district list both for DLHS-2 and NFHS-4. In all, using more than 1,000 observations over two time periods at the district level, this section attempts to capture the association between agricultural productivity and child underweight in India. 3.3.2

Methods of analysis

The base model can be stated as follows: UWi = α1 + βAi + γX + e1i,

(1)

Where, UWi = percentage of underweight children in the ith district; Ai = agricultural (crop + livestock) productivity (worker productivity/land productivity) in the ith district; X = the vector of covariates that influence underweight children below the age of five/six

Child nutrition 59 The vector of covariates may vary between panel data models and single period models. Variables both in non-log form as well as log form enter models based on the choice. The major focus of the chapter is on panel data estimation. Pooled panel data ordinary least squares, mixed effects maximum likelihood estimation and fixed effects panel data estimation, test the association of child underweight rates with agricultural land productivity or agricultural worker productivity.

3.4

Results

This chapter considers district data, 10 to 15 years apart, it would therefore be relevant to study the sectoral changes in district gross domestic product across predominantly rural and more urbanised districts. Sectoral shares and per worker GDP capture the dynamics of change, in the context of falling share of agriculture in the total district income. Higher level of commercialisation in the more semiurban districts drive the dynamics of change. 3.4.1

Semi-urban districts and predominantly rural districts

The districts have been divided in to two groups as the predominantly rural districts with more than 80% rural population and more urbanised or semi-urban districts with less than 80% of the rural population. The average level of rural population in the semi-urban districts was 60%. Between 2001 and 2011, the average share of agricultural GDP in the predominantly rural districts declined from 22.7% to 17.9% of the district GDP. In more urbanised districts the average share of agriculture fell from 15.8% to 13.9% of the district GDP. Another important fact is the average share of secondary sector in both semi-urban and rural districts declined over time from 2001 to 2011, as the share of tertiary sector went up in both categories of districts (Figures 3.1 and 3.2). The driving force behind the 80 70 60 50 40 30 20 10 0

2001 Primary Sector

2011 Agriculture

Secondary

Tertiary

Figure 3.1 Percentage share of GDP in districts having less than 80% rural population

60 Anusha. G and Swarna Sadasivam Vepa 80 70 60 50 40 30 20 10 0 2001 Primary Sector

2011 Agriculture

Secondary

Tertiary

Figure 3.2 Percentage share of GDP in districts having more than 80% rural population

share of service sector, most of which is informal, is more likely to be agriculture rather than secondary sector. Secondary sector unrelated to agriculture is more likely to be dynamic in big cities and urban districts than in semi-urban districts. However, average per worker GDP in agriculture at constant prices was not only higher in urbanised districts compared to rural district both in 2001 and 2011 but also recorded a proportional increase over the decade (Figures 3.3 and 3.4). Average annual per worker GDP in agriculture at constant prices increased from Rs. 28,959 in 2001 to Rs. 65,670 in 2011 in semi-urban districts. The corresponding average annual per worker GDP in predominantly rural districts changed from Rs. 15,293 to Rs. 32,835. Obviously commercialised agriculture helps income generation and it is the main driving force in all the districts. Higher agricultural productivity also induces urbanisation due to trade, storage, processing, export and transport activities. The district GDP data reiterated the point that share of district GDP is not reflective of its worker productivity. Semi-urban districts with lower share of agricultural GDP had higher per worker agricultural GDP, while predominantly rural districts with higher shares of agricultural GDP show lower levels of agricultural worker productivity. The tertiary and secondary sectors have shown higher per worker agricultural GDP across districts. The district GDP data gives support to our assumption that the urban rural distinctions are getting blurred and that rural is not synonymous to agriculture and urban is not synonymous to non-agriculture. Agricultural activities, other than sowing of staple crops and commercial crops on a large scale, occur both in rural and urban areas. Dairy, poultry, meat, fisheries6 and horticultural production occur in peri-urban areas, rather than rural areas. Similarly, non-agricultural opportunities exist both in rural and urban areas. As revealed by national sample survey data, more than 40% of the people derive non-farm income in rural India. One should follow agriculture

Child nutrition 61 160 140 120 100 80 60 40 20 0 2001 Primary Sector

2011 Agriculture

Secondary

Tertiary

Figure 3.3 Per worker productivity (Rs. ’000s) in districts with less than 80% rural population

160 140 120 100 80 60 40 20 0

2001 Primary Sector

2011 Agriculture

Secondary

Tertiary

Figure 3.4 Per worker productivity (Rs. ’000s) in districts with more than 80% rural population

non-agriculture distinction and not urban rural distinction, especially when the share of agriculture is falling in the GDP of all areas. The GDP data in a way vindicates our stand that using district as a unit of observation makes more sense. Further including cultivable area to calculate per hectare productivity rather than net sown area is justified as dairy, poultry, livestock and inland fisheries make intensive use of land. Tables 3.1 to 3.4 show that predominantly rural districts had a higher percentage of undernourished children in 2002–04 as per DLHS-2 data as well as in 2015–16 as per the NFHS-4 data, although there was a decline in the incidence. Access to

Table 3.1 Sectoral GDP of urbanised districts (districts with less than 80% rural population in 2001) 2001 (80% rural)

Obs

Mean

Std. dev

Min

Max

Per worker GDP in primary sector Per worker GDP in agriculture Per worker GDP in secondary sector Per worker tertiary Share of primary (%) Share of agriculture (%) Share of secondary (%) Share of tertiary (%) Underweight children (%) Proportion of rural (%)

323 322 323 323 323 323 323 323

22.244 15.293 68.630 84.782 41.700 22.700 17.090 41.257 46.26195 88.810

40.247 26.006 72.912 40.086 12.743 10.291 10.631 9.984

0.095 0.095 0.104 0.106 8.000 1.000 4.000 13.000

487.846 436.501 454.795 201.313 82.000 51.000 63.000 71.000

0.049

80.000

1.000

323

Source: Calculated from district GDP data Note: Per worker GDP is at constant prices in thousand rupees per annum

Table 3.3 Sectoral GDP of urbanised districts (districts having less than 80% rural population in 2011) 2011 ( chi2 = 0.0000

Underweight

Coef.

Std. err.

z

P>z

Ln (production/worker) −0.83083 Women with > 10 Yr. Sch. (%) −0.40958 Children with diarrhoea (%) 0.272837 Constant 52.95435

0.3892451 0.0257756 0.0533529 3.385636

−2.13 −15.89 5.11 15.64

0.033 0.000 0.000 0.000

Random-effects parameters

Std. err.

Estimate

year: Identity Var (_cons) 18.99131 19.27182 2.598847 138.7807 Var (Residual) 122.2104 5.343914 112.1728 133.1462 LR test vs. linear model: chibar2(01) = 118.82 Prob >= chibar2 = 0.0000

Table 3.9 Mixed effects ML estimation of the impact of agricultural land productivity on underweight children at the level of districts and states DLHS-2—NFHS-4 panel data Mixed-effects ML regression Group variable:

State

Wald chi2(3) = 241.61 Log likelihood = −3906.35

Number of obs. = 1040 Number of groups = 30 Obs per group: min = 2 avg = 34.7 max = 140 Prob > chi2 = 0.0000

underweight

Coef.

Std. err.

z

P>z

Ln (prod/hectare) Women with > sec. edu (%) Children with diarrhoea (%) Constant

−2.27996 −0.374557 0.3554671 51.06922

0.4954014 0.0340826 0.0529091 2.6698

−4.6 −10.99 6.72 19.13

0 0 0 0

Random-effects parameters

Estimate

Std. err.

State code: Identity Var (_cons) 81.63177 Var (Residual) 98.24935 LR test vs. linear model: chibar2(01) = 253.56

23.66013 4.376489 Prob >= chibar2 = 0.0000

68 Anusha. G and Swarna Sadasivam Vepa Table 3.10 Mixed effects ML estimation of the impact of agricultural worker productivity on underweight children at the level of districts and states DLHS-2—NFHS-4 panel data Mixed-effects ML regression Group variable:

State

Wald chi2(3) = 231.80 Log likelihood = −3940.0214

Number of obs = 1048 Number of groups = 30 Obs per group: min = 2 avg = 34.9 max = 140 Prob > chi2 = 0.0000

Underweight

Coef.

Std. err.

z

P>z

Ln (Production/ worker) Women with > 10 yr. Sch. (%) Children with diarrhoea (%) Constant

−1.52768 −0.37487 0.341698 48.6414

0.3822958 0.0346857 0.0527065 2.412276

−4.00 −10.81 6.48 20.16

0 0 0 0

Random-effects parameters

Estimate

Std. err.

State code: Identity Var (cons) 79.47605 23.04341 Var (Residual) 99.06696 4.39524 LR test vs. linear model: chibar2(01) = 258.12 Prob >= chibar2 = 0.0000

results show that the three chosen predictors of percentage of underweight children, viz., agricultural land/worker productivity, percentage of women who achieved educational qualification of secondary level or above and percentage of children reporting diarrhoea have been significant with expected sign in all the regressions. Higher agricultural land/worker productivity and larger percentage of women with educational achievement above secondary levels are likely to reduce the proportion of underweight children in the district. Higher percentage of children with diarrhoea in the districts is likely to increase the proportion of underweight children in the district. The random effects parameters indicate that the contribution of “between the years variance” to total variance was lower, at about 13–14%, and the rest of the variance was between the district variance. On the other hand, the contribution of inter-state variance to the total variance was high at 39%. It means that state and district specific factors contributed to variance in underweight. 3.4.4

Fixed effects panel data estimation

We have estimated the fixed effects model to take care of the possible endogeneity of agricultural productivity variable. The Hausman test applied to the data

Child nutrition 69 confirmed the suitability of fixed effects model against random effects model. Four variation of the model alternately including either land productivity or worker productivity in level form and log form show that agricultural land productivity and agricultural worker productivity influence the percentage of underweight children at the district level (Tables 3.11 and 3.12). The only insignificant result was with respect to land productivity in the model using all variable in log form. Perhaps it turned insignificant due to higher correlation of agricultural land productivity with percentage of women with education above secondary level. The correlation in logs is 0.33 and significant. As we can see, districts with higher land productivity per hectare have less than 80% of rural population and more urbanised and they will have more women with higher educational achievements. Percentage of children with incidence of diarrhoea may indicate unhygienic surroundings or limited health care availability and lack of knowledge of handling diarrhoea with oral rehydration etc. However, per worker agricultural productivity is not correlated with women’s education and turn out as a significant explanatory variable in all estimates. Per worker agricultural productivity is a better indicator of well-being. One per cent increase in the average worker productivity reduces child underweight rate by 0.09%. The elasticity is low but effective and perhaps the only way to reduce child underweight, rates, in a limited opportunity economy. The elasticity of child underweight with respect to women’s education is fairly, high at 0.40%. The elasticity of child underweight rate with respect to the incidence Table 3.11 Fixed effects panel data estimation of the impact of agriculture on underweight children Fixed-effects panel data regressions: DLHS-2(2002–04) NFHS-4(2015–16) Dependent variable

Model 1

Model 2

Explanatory variables

Underweight children (%)

Underweight children (%)

co-efficient

co-efficient

t

−4.193 −0.541 0.460 65.16

−5.01** −7.24** 6.25** 18.77** 1048 524 0.30 0.28 0.33 0.000

Ln (Production/hectare) Ln (Production/worker) Women with > 10 Yr. Sch. (%) Children with diarrhoea (%) Constant No. of observations Number of groups R2: Overall within between F test: Prob > F

t

−2.316

−2.61**

−0.582 0.523 60.46

−7.72** 6.91** 15.89** 1040 525 0.32 0.25 0.36 0.000

Note: *p-value < 0.05, **p-value < 0.01

70 Anusha. G and Swarna Sadasivam Vepa Table 3.12 Fixed effects panel data estimation of the impact of agriculture on underweight (log form) Fixed-effects panel data regressions: DLHS-2(2002–04) NFHS-4(2015–16) Dependent variable

Model 1

Model 2

Explanatory variables

Ln (Underweight children %)

Ln (Underweight children %)

co-efficient

co-efficient

ln (Production/hectare) ln (Production/worker) ln (% Women with > 10 years of school) ln (% Children with diarrhoea) Constant No. of observations Number of groups R2: Overall within between F test: Prob > F

t

−0.036

−1.35

−0.469

−6.87**

0.098 5.098

4.25** 20.27** 1037 525 0.27 0.16 0.32 0.00

t

−0.095 −0.431

−3.82** −6.35**

0.085 5.174

3.82** 21.49** 1045 524 0.26 0.17 0.30 0.00

of diarrhoea is low but significant. Increase in 1% incidence of diarrhoea increases the child underweight by 0.085%. The logarithms of the percentages render the elasticity low, but in practice the effect could be large at the margins.

3.5 Conclusions The results indicate an association between underweight children and agricultural land productivity and agricultural worker productivity defined more broadly at the district level. First, agricultural productivity measured as the value of crop and livestock output averaged over three years probably captured the effect of agriculture better. Probably the trickle down as well as spillover effects of agricultural prosperity could be captured more comprehensively in the district analysis. Despite severe data limitations, the effect of agricultural productivity turned out significant on child underweight. Second, single period analysis with NFHS-4 underweight data, pooled OLS, and linear mixed effects likelihood estimation and fixed effects panel data estimation conclusively establish the influence of district agricultural productivity on child underweight rates. The direction of the variables remained as expected and significant despite their weakness in explaining only one third of the variation in underweight rates along with chosen other two key variables. Due to data

Child nutrition 71 limitations and possible co-linearity issues, we have deliberately kept the number of explanatory variables to just three. As our panel data state level analysis for rural India has shown (Vepa et al. 2014), R-square value can be enhanced with inclusion of other state specific structural variables. Instead we preferred to capture them in mixed effects estimation. Third, the elasticity of child underweight rates with respect to agricultural productivity is small. This is also clear from the scatter diagram (Appendix 3A1, Figures 3A.1 and 3A.2). This is due to the existence of other more direct factors that influence child nutrition. Agricultural productivity influences underweight among children indirectly through poverty reduction, diversified food intake, food prices, and affordability of better sanitation, health care, and education. The pathway is more nuanced, but it happens only because of productivity enhancement and hence the effect is small. Alternatively, it indicates the weak database used. We could not use contemporary data or time series. Fourth, the impact of agricultural productivity on child nutrition probably occurs with a time lag and probably needs long-term time series database to capture the effect, in an appropriate manner. Fifth, it is not clear as to who is benefitting from agricultural productivity. Child underweight rates may fall among households engaged farm as well as nonfarm activities. These non-farm activities are related to agricultural productivity increase. Agricultural productivity is low in predominantly rural districts, where child underweight rates are high. Sixth, agricultural land and worker productivity improvements benefit nutrition outcomes via income enhancement, as well as affordable prices. Policy should make diversified livelihoods available to the rural poor and diversified food affordable to the rural poor. Food prices, both cereal and non-cereal food prices enhance diversified food intake and improve the nutrition of the women and children, in rural and urban areas. Despite agricultural production diversification and increase in the share of high value items since 2007–08, food prices of both cereals and protein rich foods have been high. This may have been dampening the positive impact of agricultural productivity on nutrition. Last, while the positive significant impact of agricultural productivity on poverty reduction and child nutrition is not in doubt, policy environment to leverage and harness the benefits is absent. Further, economic growth has been slow since 2010. Rural farm wages remained lower than that of non-farm and urban wages, even though the gap declined. Agricultural prosperity should typically benefit the rural poor, if they could access livelihoods, public services of clean water sanitation and health care. Lack of enabling factors could diminish the positive impact of agricultural productivity.

Notes 1 The GDP calculations changed after 2015. Hence it is not comparable to previous calculation of 12% of GDP in 2014 given in the previous issues of economic survey brought out by the government.

72 Anusha. G and Swarna Sadasivam Vepa 2 Higher per worker productivity in agriculture has welfare implications. Land productivity increased in agriculture at the rate of 3.1% and labour productivity increased at the rate of 4.8% between 2007–08 and 2011–12 (12th five-year plan, volume 2, chapter 6) 3 NFHS-3 (2005–06) did not provide district level information. DLHS-3 did not collect child nutrition information. DLHS-4 covered only 20 states. Many states with high burden of child undernutrition were left out. 4 DLHS-2 underweight data pertains to children under six years. After extracting data for under five children, we found that there is very little difference between under five and under six underweight rates (Appendix 3A, Table 3A.2). Hence, we used under six underweight data from DLHS-2 and under five underweight data from NFHS-4. 5 District GDP data beyond 2014 is not comparable to the earlier period as the methodology of calculation changed after 2014 and new series recalculated comparable data back in time only up to 2011–12. NFHS-4 data pertains to the year 2015. We assumed that the child underweight rates and other explanatory variables of 2015 also reflect the 2011–14 situation, though it is not a realistic assumption. 6 Port towns rather than coastal villages are now the hubs of marine fisheries with big trawlers. Prawn fishponds created with ground water/canal water also tend to be closer to the towns and become peri-urban in nature. 7 Agricultural worker productivity as well as agricultural land productivity turn out to be significant in the single period OLS estimate with NFHS-4 underweight data (Appendix 3B, Table 3B.4).

References Bhagowalia, P., Headey, D. and Kadiyala, S., 2012, Agriculture, Income, and Nutrition Linkages in India. Insights from a Nationally Representative Survey, IFPRI Discussion Paper 01195. Binswanger-Mkhize, H. P., 2013, The Stunted Structural Transformation of Indian Economy, Agriculture, Manufacturing and Rural Non-Farm Sector. Economic and Political Weekly, June 29, Vol. 48, Nos. 26 & 27. Calì, M. and Menon, C., 2013, Does Urbanization Affect Rural Poverty? Evidence from Indian Districts. The World Bank Economic Review, Vol. 27, No. 2, pp. 171–201. Published by: Oxford University Press. Cavatorta, E., Shankar, B. and Florence Martinez. A., 2015, Explaining Cross-State Disparities in Child Nutrition in Rural India. World Development, Vol. 76, pp. 216–237. Datt, G. and Ravallion, M., 2002, Is India’s Economic Growth Leaving the Poor Behind? Journal of Economic Perspectives, Vol. 16, No. 3, pp. 89–108. De Janvry, A. and Sadoulet, E., 2009, Agricultural Growth and Poverty Reduction: Additional Evidence. The World Bank Research Observer, Vol. 25, No. 1, pp. 1–20. De Janvry, A. and Sadoulet, E., 2010, Agriculture for Development in Sub-Saharan Africa: An Update. African Journal of Agricultural and Resource Economics, Vol. 5, No. 311– 2016–5537), pp. 194–204. De Janvry, A. and Subbarao, K., 1986, Agricultural Price Policy and Income Distribution in India (No. 43), Oxford, USA: Oxford University Press. FAO, IFAD, UNICEF, WFP and WHO, 2019, The State of Food Security and Nutrition in the World 2019. Safeguarding Against Economic Slowdowns and Downturns, Rome, FAO. Global Nutrition Report, 2018, Development Initiatives, 2018 Global Nutrition Report: Shining a Light to Spur Action on Nutrition, Bristol, UK: Development Initiatives.

Child nutrition 73 Goldar, B., Krishna, K. L., Agrawal, S. C., Das, D. K., Erumban, A. A. and Das, P. C., 2017, Productivity Growth in India Since the 1980s: The KLEM Approach. Indian Economic Review, Vol. 52, Nos. 1–2, pp. 37–71. https://doi.org/10.1007/s41775-017-0002-y. Government of India (GoI), 2004, District Level Household Survey-2 (2002–04), Ministry of Health and Family Welfare, New Delhi. Government of India (GoI), 2013, District Level Household Survey-4 (2012–13), Ministry of Health and Family Welfare, New Delhi. Government of India (GoI), 2015, National Family Health Survey (2014–15), Ministry of Health and Family Welfare, New Delhi. Hagos, S., Lunde, T., Mariam, D. H., Woldehanna, T. and Lindtjørn, B., 2014, Climate Change, Crop Production and Child Under Nutrition in Ethiopia: A Longitudinal Panel Study. BMC Public Health, Vol. 14, No. 1, p. 884. India Office of the Registrar General and Census Commissioner (2011), Census of India, New Delhi: Ministry of Home Affairs, Government of India. International Food Policy Research Institute (IFPRI), 2019, Status of Child Stunting in India, www.ifpri.org/news-release/new-study-status-child-stunting-india Jalan, J. and Ravallion, M., 2001, Does Piped Water Reduce Diarrhea in Rural India? World Bank Policy Research Working Paper No. 2664. Jose, A. V. 2013, Changes in Wages and Earnings of Rural Labourers. Economic and Political Weekly, Vol. 48, Nos. 26–27, pp. 107–114. Kumar, C., Singh, P. K. and Rai, R. K., 2012, Under-Five Mortality in High Focus States in India: A District Level Geospatial Analysis. PloS One, Vol. 7, No. 5, p. e37515. doi:10.1371/journal.pone.0037515 Liu, E., Balasubramaniam, D. and Hunt, A. F., 2016, Does Access to Water Matter? A Study on the Nutritional Status of Primary-Aged Children in India. Journal of Public Health, Vol. 38, No. 4, pp. 419–424. Lokshin, M., 2005, Improving Child Nutrition? The Integrated Child Development Services in India, Institute of Social Studies, Development and Change, Vol. 36, No. 4, pp. 613–640. Mahmud, I. and Mbuya, N., 2016, Water, Sanitation, Hygiene, and Nutrition in Bangladesh: Can Building Toilets Affect Children’s Growth?, World Bank, Washington DC. Mishra, V. K. and Retherford, R. D., 2000, Women’s Education Can Improve Child Nutrition in India, East West Centre, Asia Pacific Issues, Hawai.edu. Mosley, W. H. and Chen, L. C., 1984, An Analytical Framework for the Study of Child Survival in Developing Countries. Population and Development Review, Vol. 10, pp. 25–45. doi:10.2307/2807954 National Sample Survey, 2014, Employment and Unemployment Situation in India, 68th Round, 2011–12. Report No. 554. NFHS-4 India Factsheet, International Institute for Population Sciences, Govandi Station Road, Deonar Mumbai-400088 (India), www.mohfw.nic.in. Pathak, P. K. and Singh, A. 2011, Trends in Malnutrition among Children in India: Growing Inequalities across Different Economic Groups. Social Science & Medicine, Vol. 73, pp. 576–585. Pingali, P., 2015, Agricultural Policy and Nutrition Outcomes: Getting Beyond the Preoccupation with Staple Grains. Food Security, Vol. 7, pp. 583–591, http://dx.doi.org/10. Pingali, P., Mitra, B. and Rehman, A., 2017, The Bumpy Road From Food to Nutrition Security, Slow Evolution of India’s Food Policy, Global Food Security, Vol. 15, pp. 77–84.

74 Anusha. G and Swarna Sadasivam Vepa Radhakrishna, R., 2002, Agricultural Growth, Employment and Poverty: A Policy Perspective. Economic and Political Weekly, Vol. 37, No. 3, pp. 243–250. Radhakrishna, R. and Murthy, K. N., 1982, Agricultural Prices, Income Distribution and Demand Patterns in a Low-Income Country, Agricultural Information Systems, Washington, DC: Agris, FAO. Radhakrishna, R. and Raju, D. S. R., 2015, Well-Being of Agricultural Households in Post Reform India. The Indian Economic Journal, Vol. 63, No. 3, pp. 329–347. https://doi. org/10.1177/0019466220150301 Radhakrishna, R. and Sarma, A., 1976, Inflation and Disparities in the Level of Living. Indian Economic Journal, Vol. 23, No. 4, pp. 364–374. Ramesh Chand, S., Srivastava, K. and Singh, J., 2017, Changing Structure of Rural Economy of India, Implications to Employment and Growth. Discussion Paper, NITI Aayog, November. Rangarajan, C., 1982, Agricultural Growth and Industrial Performance in India. Research Report No. 33, Washington, DC: International Food Policy Research Institute. Ravallion, M. and Datt, G., 1996, How Important to India’s Poor Is the Sectoral Composition of Economic Growth? World Bank Economic Review, Vol. 10, No. 1. Robert, B. E., Allen, L. H., Bhutta, Z. A., Caulfield, L. E., De Onis, M., Ezzati, M., Mathers, C. and Rivera, J., 2008, Maternal and Child Undernutrition: Global and Regional Exposures and Health Consequences. LANCET, January 19, Vol. 371. Timmer, C. P. and Akkus, S., 2008, The Structural Transformation as a Pathway Out of Poverty: Analytics, Empirics and Politics. Working Paper Number 150, The Center for Global Development, Washington, DC. UNDP, 2015, United Nations. The Millennium Development Goals Report, 2015 (p. 3). New York: United Nations Development Programme. UNICEF, 2018, Progress for Every Child in the SDG Era, http://unicef.in/WhoWeAre/ MDGs. UNICEF, 2019, State of the World’s children 2019, Children, food and nutrition – Growing well in a changing world, UNICEF, New York, https://www.unicef.org/media/63016/file/ SOWC-2019.pdf Vepa, S. S., Uma Shankar, V., Bhavani, R. V. and Parasar, R., 2014, Agriculture and Child Under-Nutrition in India: A State Level Analysis. MSE Working Paper 86/2014, Madras School of Economics, Chennai. Vepa, S. S., Viswanathan, B., Parasar, R. and Bhavani, R. V., 2015, Child Underweight and Agricultural Productivity in India: Implications for Public Provisioning. Review of Radical Political Economy, December, Vol. 47, No. 4. Sage USA. Vepa, S. S., Viswanathan, B., Parasar, R. and Bhavani, R. V., 2016, Child Underweight, Land Productivity and Public Services: A District-level Analysis for India. Working Paper No. 6, Lansa Working Paper Series, Volume 2016. World Bank, 2007, Agriculture and Rural Development Department, from Agriculture to Nutrition, Pathways Synergies and Outcomes. Report No. 40196-GLB. World Health Organization (WHO), 2010a, Nutrition Landscape Information SystemCountry Profile Indicator Interpretation Guide, Geneva: World Health Organization.

Obs 621 610 626 626 Obs 523 523 284 523 521 521 523

Census, agricultural census, indicus variables (2011–14)

Per worker productivity in agriculture (’000s) Per hectare productivity (’000s) Female literacy rate Percentage rural population

NFHS-4 variables (2011–14)

Percentage children underweight Percentage women with more than ten years of education Percentage anaemic children Percentage under three with diarrhoea Percentage children fully vaccinated Percentage BCG vaccinated Percentage access to clean water

Table 3A.1 Descriptive statistics

34.32 34.18 21.28 8.68 63.02 92.19 88.94

Mean

50.03 35.45 0.55 0.74

Mean

Descriptive statistics & correlation matrix of variables

Appendix 3A

66.90 86.30 82.30 29.10 100.00 100.00 100.00

5.80 9.00 2.40 0.00 7.10 37.20 33.90 11.04 14.52 14.19 4.79 16.78 8.70 11.76

(Continued )

Max

2396.29 194.64 0.89 1.00

Max

Min

0.04 0.44 0.24 0.00

Min

Std. dev

150.18 27.10 0.12 0.21

Std. dev

Max 96.60 85.70 60.00 100.00 88.00 66.20 99.80 97.80 100.00 93.80 73.70 100.00 100.00 100.00 100.00 100.00

Min 0.40 0.50 0.30 30.00 5.30 0.10 17.50 2.40 32.30 6.90 0.50 6.70 8.30 0.00 40.80 37.80

Std. dev 13.48 9.15 6.96 14.22 13.63 6.13 19.47 19.76 12.53 15.00 5.11 20.39 23.30 18.28 12.42 13.25

Mean 68.20 10.83 6.48 61.89 58.33 6.03 69.94 45.37 91.63 36.82 5.03 58.30 71.88 62.77 89.40 78.79

Obs

96.00 91.83 84.10 42.07 98.80 100.00 100.00 94.40

219 208 218 214 225 223 269 269 269 269 262 243 219 267 267 267

1.80 5.84 0.50 0.00 0.70 14.40 3.12 18.60

DLHS-4 variables (2011–14)

14.49 14.13 16.02 7.90 26.13 17.60 16.46 15.13

Percentage anaemic children Percentage severely anaemic children Percentage severely anaemic girls Percentage anaemic girls Percentage anaemic women Percentage severely anaemic women Percentage with toilets at home Percentage using cooking fuel Percentage access to clean water Percentage women with more than ten years of education Percentage under three with diarrhoea Percentage ORS Percentage with access to primary health centre Percentage children fully vaccinated Percentage BCG vaccinated Percentage DPT3 vaccinated

Max

46.25 30.51 42.73 13.65 46.87 80.29 82.23 52.43

Min

525 525 519 525 524 456 525 525

Std. dev

Percentage children underweight Percentage women with more than ten years of education Percentage anaemic children Percentage under three with diarrhoea Percentage children fully vaccinated Percentage BCG vaccinated Percentage access to clean water Female literacy rate

Mean

Obs

DLHS-2 variables (2001–2004)

Table 3A.1 (Continued)

–0.31 –0.16 –0.22 –0.14 –0.11

3 Percentage children underweight

4 Percentage rural population

5 Female literacy rate

0.16 0.17

10 Percentage BCG vaccinated

11 Percentage access to clean water

0.25

8 Percentage under three or five with diarrhoea

9 Percentage children fully vaccinated

0.84

1.00

4

0.33

0.35 0.34

0.55

1.00

6

0.08 –0.35 –0.43 –0.23 0.05 –0.03 –0.27 –0.25

0.24

1.00

7

1.00

8

9

10

11

0.05 –0.07 –0.02 0.24 0.20 1.00

0.36 –0.23 –0.34 0.81 1.00

0.44 –0.19 –0.21 1.00

0.31 –0.21

0.42 –0.13

0.09

1.00

5

0.14 –0.35 –0.41 –0.13

–0.14 –0.08

7 Percentage anaemic children

0.28

0.42

1.00

3

0.20 –0.50 –0.19

0.38 –0.20 –0.16

6 Percentage women with more than ten years of education

–0.09

0.32

1.00

1.00

2

2 Per worker agricultural productivity (’000s)

1

1 Per hectare agricultural productivity (’000s)

NFHS-4 DLHS-2 variables (Government of India 2015)

Table 3A.2 Correlation matrix of DLHS-2 and NFHS-4 variables

200

Per hectare productivity 000s

175 150 125 100 75 50 25 0 0

20

40 60 % underweight

80

100

Figure 3A.1 Scatter plot with per hectare productivity and percentage of underweight children using DLHS-2 and NFHS-4

1750

Per worker productivity 000s

1500 1250 1000 750 500 250 0 0

20

40 60 % underweight

80

100

Figure 3A.2 Scatter plot with per worker productivity and percentage of underweight children using DLHS-2 and NFHS-4

Appendix 3B Agriculture and under-nutrition variables

Table 3B.1 Value of agricultural output growth (2004–05 prices)

1 2 3 4 5 6 7 8 9 10

Item

2002–03 to 2006–07

2007–08 to 2011–12

Cereals Pulses Oilseeds Sugar Fibers Non-horticultural crops Horticulture All crops Livestock Crops & livestock

1.0 1.8 7.4 1.7 15.1 2.1 2.6 2.1 3.6 2.5

3 4.2 4.5 2.2 10.7 2.8 4.7 3.4 4.8 3.8

Source: 12th Five-Year Plan, Chapter 6, 2012, reprinted in Uma Kapila, 2013

Table 3B.2 Percentage of underweight children below the ages of five and six States DLHS-2

0–59 months

0–71 months

Andhra Pradesh Arunachal Assam Bihar Chhattisgarh Goa Gujarat Haryana Himachal

2002–04 43.5 22.2 33.2 54.3 48.2 28.4 44.9 37.2

2002–04 42.3 20.3 32.2 54.6 47.4 30.0 46.0 35.6 (Continued )

Table 3B.2 (Continued) States DLHS-2

0–59 months

0–71 months

Jammu and Kashmir Jharkhand Karnataka Kerala Madhya Pradesh Maharashtra Manipur Meghalaya Mizoram Nagaland Odisha Pondicherry Punjab Rajasthan Sikkim Tamil Nadu Tripura Uttaranchal Uttar Pradesh West Bengal All India

37.4 29.3 52.1 44.1 35.2 53.9 46.4 15.0 36.4 16.8 22.8 43.0 26.8 41.5 61.4 12.9 39.9 33.2 59.5 57.2 43.6

36.4 22.6 52.2 44.8 35.8 55.4 47.7 12.6 34.9 15.2 21.4 42.8 26.8 40.0 58.3 9.7 38.3 30.2 55.3 52.6 44.9

Source: 0–59 calculated from raw data and 0–70 taken from the published report at the state level

Table 3B.3 Description of variables and data sources Period I = 2001–2004 S.No. 1 2 3 4 5 6

Period 2 = 2011–2015

Data

Variable used

Period

Source

Year

Percentage of underweight children below six years Percentage of underweight children below five years Percentage of women having more than ten years of schooling Percentage of women having more than ten years of schooling Percentage of children with diarrhoea under the age of three Percentage of children with diarrhoea under the age of five

1

DLHS-2

2002–04

2

NFHS-4

2015–16

1

DLHS-2

2002–04

2

NFHS-4

2015–16

1

DLHS-2

2002–04

2

NFHS-4

2015–16

Period I = 2001–2004 S.No. 7

8

9 10

Period 2 = 2011–2015

Data

Variable used

Period

Source

Year

Agricultural Productivity (constant prices)/hectare a. Agricultural (crop + livestock) GDP (const. prices) three-year average b. Cultivable area (NAS + fallows + cultivable waste) three-year average Agricultural productivity (constant prices)/hectare a. Agricultural (crop + livestock) GDP (const. prices) three-year average b. Cultivable area (NAS + fallows + cultivable waste) Agricultural productivity (constant prices)/worker Agricultural productivity (constant prices)/worker

1 1

Indicus

2001–04

1

Land utilisation sts.

2 2

Indicus

2011–14

2

Ag.Census

2011

1

Indicus

2001–04

2

Indicus

2011–14

Table 3B.4 OLS estimation of the impact of agriculture on underweight children (NFHS-4) OLS estimation of impact of agriculture on underweight children Ln (% underweight children)

Model 1

Model 2

Explanatory variables Ln (ag. productivity/hectare) Ln (ag. productivity/worker) Ln (women with ten years of sch. %) Ln (children under five with diarrhoea %) Constant No. of observations R2 Prob > F * p-value < 0.05 ** p-value < 0.01 Source: Based on NFHS-4

−0.064** −0.420** 0.129** 4.86** 519 0.363 0.000

−0.043** −0.430** 0.128** 4.823** 523 0.354 0.000

4

Women’s BMI among farm and non-farm households in rural India Brinda Viswanathan and Getsie Immanuel

4.1 Introduction Nutrition is an important component of a person’s well-being. It is even more important in the case of women, since an undernourished woman gives birth to an undernourished child, contributing to a vicious circle. Therefore, arresting the problem of undernourishment at different stages of the lifecycle is essential to reduce the huge burden of undernutrition observed in several parts of the world including India. A country like India contributes substantially to this burden but also stands out in terms of several puzzles. Firstly, there has been faster economic growth in last two decades but undernourished children and women are still very large in number (von Grebmer et al. 2016). With modest decline in poverty rates, the undernourishment rates have shown slower rates of decline particularly in rural areas. Secondly, during this period of rapid overall growth, agricultural growth has been rather tame while some regions in India have shown good progress. Thirdly, the share of agriculture in value addition has been declining at a continuous pace but the share of agriculture in employment still remains large (Binswanger-Mkhize 2012). These three structural features seem to have resulted in the people dependent on agriculture continuing to be more disadvantaged from persistent undernutrition. South Asia has a large population dependent on agriculture and a significantly large proportion of them are undernourished. Added to this is poor service delivery of basic amenities and a multilayered discrimination. Social discrimination has been observed to be more persistent in South Asia including India than in any other part of the world. There are many discriminatory practices prevalent in these regions that lead to a skewed distribution of resources between the less- and the more-disadvantaged social/religious group as well as between members of a family. All of this contributes to large gaps in nutritional achievements among segments of population. Under these circumstances, direct state intervention with assistance from non-governmental intermediaries becomes essential to loosen up the persistent rigidities. Timely access to nutritious food and awareness about care practices are also known to reduce severe undernutrition particularly among children. More importantly, countries like India have several food and nutritional security policies as well as agricultural and social policies to address many of these concerns. Yet, there has been limited reduction in undernutrition.

Women’s BMI in rural India

83

This chapter provides an empirical analysis of nutrition insecurity among women in farm and non-farm households in rural India with women’s body mass index (BMI) as the indicator of nutritional outcome. The analysis is carried out keeping in mind the fours dimensions that constitute food and nutrition security, namely, affordability, availability, awareness and absorption using data for two years, 2005–06 and 2012–13. The agricultural pathway that enhances the scope to address nutritional insecurity comes from the following two aspects. Households that produce certain crops and have access to markets impacts agricultural income and hence affordability while crop diversity, livestock rearing and poultry could influence dietary quantity and quality from home consumption. Thus, support to agriculture that improves productivity and preserves agricultural diversity and addresses market imperfections become relevant in order to have a widespread impact on nutritional outcomes (Per Pinstrup-Andersen 2013). Certain malaise associated with deep-rooted sociocultural practices lead to discrimination and hence limited access to resources that would otherwise enhance nutritional outcomes. Empowerment in general and for women in particular strengthens the awareness of one’s rights and needs towards nutritionally secure outcome (van den Bold et al. 2013; Jayachandran and Pande 2015). Absorption of nutrients is crucial for maintaining nutritionally secure outcomes and this is enabled by hygienic and clean environmental conditions as well as timely access to quality health care (Duflo et al. 2008; Spears and Thorat 2015). The evidence on the link between agriculture and nutrition has so far been tenuous (Kadiyala et al. 2012; Das et al. 2017; Ruel et al. 2018). Most studies use child nutrition indicators to study the linkage and very few studies analyse the role of agriculture and other enabling factors on nutrition indicators for adults like women’s short-term nutritional status like BMI (Rao and Pingali 2018; Rao and Raju 2019). With greater emphasis being laid on evidence-based policy-making, more empirical evidence to understand the linkage between nutrition outcomes, agricultural prosperity and other factors become essential (Malhotra 2014). Focussing on a large country like India is essential to reduce the world’s overall burden of undernutrition. Many of the less developed countries have one or several of these issues leading to undernourishment and a study of India with its diverse economic, socio-demographic and democratic polity could serve as important lessons for other regions as well. The focus of this study is to not only understand how certain limited features of agriculture play a role in explaining variations in women’s BMI in India and also to analyse other relevant factors that could be equally important for this. This chapter tries to look at the change from 2005 to 2011, based on two waves of India Human Development Survey (IHDS-I to IHDS-II). Among the indicators of nutrition security, nutrition outcome indicators particularly for children based on anthropometric indicators have been the key variables for assessment. Data availability in this context indicates that the problem of undernutrition among children is more severe and persistent for a country like India. Among several factors that have been shown to affect child undernutrition a recurrent factor relates to women’s status in general and mother’s nutritional status in particular. In this context it would be useful to document empirical evidence that

84 Brinda Viswanathan and Getsie Immanuel analyse women’s or for that matter adult nutritional status. Limited numbers of studies exist for India that link agricultural economy to women’s anthropometric indicators. Such an analysis would be useful for informed policy-making as intergenerational transmission of undernutrition is a key factor in reducing undernutrition and the role played by agricultural sector in this context. There have been several discussions on a conceptual basis of the agriculturenutrition linkage and the pathways through which this linkage could be better harnessed to reduce undernutrition.

4.2 BMI as nutrition outcome indicator Due to the availability of country-wide data in India for a fairly long period of time on nutritional intakes, empirical studies on trends and patterns in nutrition intakes, and the factors determining its variations have been studied more extensively (Viswanathan and Meenakshi 2008; Sharma 2015). A major difference between intake and outcome indicators in India is that the former is usually based on household level information while the latter is for individuals. Anthropometric indicators like height (for age), weight (for age) and body mass index (BMI) are common indicators of nutritional outcomes at the individual level. In most of the studies exploring the role of farming or agriculture in improving nutritional status nutritional outcome based on young children is more common compared to older children, or adolescents or adults (Carletto et al. 2015; Ruel et al. 2018). BMI is a person’s weight in kilograms divided by square of height in meters. For women, height is stable between 20 and 45 years—large part of linear growth is attained by menarche, but studies usually consider 20 years as the starting age to analyse adult women’s heights and the ending age is usually taken as 45 years after which stooping may set in. Given that height is not expected to vary in this age group, the changes in BMI are effectively changes in body weight, not affected by any biological changes but mainly due to various social, economic and environmental factors. BMI is a useful indicator of well-being as there are scientifically given normative thresholds that enable a person to be classified as under-nourished (BMI below 18.5), well-nourished (BMI between 18.5 and 25) or malnourished (BMI above 25). An adult who has a BMI value below 18.5 is also referred as chronically energy deficient (CED) and hence the discussions here would focus on this indicator of nutrition insecurity. By studying women’s BMI, we are able to focus on the individual as the final unit of analysis and also women’s empowerment, an area of immense importance in the South Asian context.

4.3 Women’s BMI in India: an overview Indian women show the largest impact of a nutrition transition due to changes in lifestyle, physical activity and dietary habits. The has resulted in declining rate of undernutrition and increasing levels of overnutrition, so much so that the undernutrition rates (22.9% for women and 20.8% for men) and overnutrition rates (20.7% for women and 18% for women) are nearly similar in 2015–16 among women aged

Women’s BMI in rural India

85

15–49 years while the undernutrition rates were far higher (35.6%) and overnutrition rates were far lower (12.7%) a decade back in 2005–06. Further, it can also be seen that the double burden afflicts women more than men in that the CED rates and overweight and obesity rates are persistently higher among women than men (Dang and Meenakshi 2017). Geographically, the country is getting divided into underweight-overweight regions with the central and eastern regions showing a far higher burden of underweight and the northern and southern regions showing a higher burden of overweight and obesity (Bhattacharya 2017). The central and eastern parts of the country are more dependent on agriculture as the main source of income with lower levels of land and labour productivity (Vepa et al. 2016). Countries with the largest number of obese and severely obese people have changed over 1975–2014, with more middle-income countries joining the USA, especially for women (Di Cesare et al. 2016). This study further shows that India has maintained its position among countries with largest number of underweight men and women and at the same time has jumped to a higher position among countries with largest increase in the proportion of overweight men and women. India is changing fast economically and socially and hence household’s economic well-being and the changing status of women in society can play major role in improving women’s health. Subramanian and Smith (2006) used India National Family Health Survey for 1998–99 and found that undernutrition was most prevalent among women belonging to lowest quintile of standard of living and over-nutrition was observed among top-most quintile of standard of living. As the standard of living became better, the risk of being undernourished declined with it systematically. Women from lower economic status are more likely to be thin while obesity is prevalent among the well-off as well as more developed southern states. Navaneetham and Jose (2008), based on the National Family Health Survey for 2005–06 (NFHS-3), show that around 40% of women in rural India are CED. This is 15 percentage points higher than the incidence among urban women (Viswanathan and Sharma 2009).1 The gender gap in BMI is not significant when compared to a huge gender gap in anemia at the all India level (Jose 2012). However, the gap between male and female rural CED rates are marginally higher than urban and is relatively higher among the lower wealth categories and among scheduled tribes when compared to other caste groups and only very few states like Bihar exhibit huge gender gaps in CED rates.

4.4

Enablers of nutrition security

In order to maintain good health nutritional security is of utmost importance. Several people around the world are nutritionally insecure because their diets are not adequate in quantity and quality and a poor health status drains results in low assimilation of nutrients into the body. There are several aspects to why individuals or members of a household are nutritionally insecure and this section reviews some of the recent findings in this context.

86 Brinda Viswanathan and Getsie Immanuel 4.4.1 Affordability Resources expended per person is the revealed standard of living. Apart from reflecting the affordability and preference for goods and services particularly in remote rural areas it is also reflective of availability and accessibility to goods and services. Poor households are usually self-employed and in the informal sector and collecting precise data on income becomes difficult from such households. However, as consumption is smoothened so the inequality as measured using mpce is lower than that measured using income or wealth data. Per capita income will take into effect the resources available per person and reflects the ex-ante ability of the household to be able to spend on goods and services that would allow them to attain a healthy level of BMI. These two measures of affordability are flow variables while wealth (asset index) captures the stock aspect of affordability and is usually estimated as an index of durable goods capturing the relative economic status across households. IHDS data for the first time collected information on all these dimensions and give scope to analyse the differences in nutrition indicators across these three different measures of affordability. Viswanathan (2015) compared the CED rates across quintiles of monthly per capita consumption expenditure, per capita income, and household wealth index. These three differ in their overall distribution and as a result the CED rates vary across the quintile classes of these three indicators of affordability. The proportion of CED women across the different types of quintiles varies indicating that the distributions of these variables are different in that these three different economic status variables are not very strongly associated. The households are further segmented into non-farm and farm households within each of these economic status variables. CED rates are not significantly different between non-farm and farm households and more so among the lower quintiles of the three affordability variables. However, if mean BMI is compared then women in farm households have a lower value when compared to non-farm households in the bottom two quintiles but the mean value is well above 18.5 threshold. In this chapter we extend this analysis to a more recent IHDS data and make comparison on the changes. Results based on econometric models that control for other variables also show that each of these variables influence women’s BMI (Dahiya and Viswanathan 2015). Since agriculture in India is largely dependent on natural conditions particularly rainfall, there is more uncertainty in the earnings from such activity. If households have other sources of income then it gives them the ability to cope with sudden shocks or prolonged distress in agriculture which is a common feature of Indian agriculture (Barrett et al. 2001; Himanshu et al. 2013). The larger the income diversity, the income earnings are expected to be smoothened through the year so that a lean agricultural period is less likely to affect food consumption as well as other expenditures including that on health. Thus, agricultural households may engage in other non-agricultural activities to smoothen their income flow throughout the year.

Women’s BMI in rural India

87

CED rates are on an average lower among households that have diversified income and that crop diversity contributes significantly to this, apart from nonfarm sources of income (Viswanathan 2015). The share of income from livestock rearing reduced the likelihood of CED substantially and so do other farming activities. In rural areas, for households that had a larger share of net agricultural incomes, the likelihood of CED was lower in comparison to those with a higher share of non-agricultural labour income (Viswanathan et al., 2015). The findings also show that CED rates are highest for women in non-agricultural wage labour perhaps because they have less scope for diversifying within agricultural activities and between agricultural and non-farm activities. If we approach from a sectoral perspective at a macro level then BMI improves from low values for women when agricultural GDP per worker increases (Headey et al. 2011). This study finds that men involved in agricultural work had lower average BMI while for women there was no difference in low BMI across the agricultural or unskilled non-agricultural work. Gulati et al. (2012) show that the level of agricultural income measured as agricultural GDP as a proportion of rural population has a strong and significant negative relationship with a composite index of undernutrition among adult men and women. In a more detailed analysis of women in rural areas the likelihood of CED was found to be less likely among households that depend primarily on cultivation in rural areas (Viswanathan et al. 2015). If we consider agriculture as an occupation, Singh et al. (2011) find that the probability of finding a woman with low BMI is highest for those whose husband’s primary occupation is agriculture and allied sectors when compared to other occupations. Dahiya and Viswanathan (2015) find that women who participate in agricultural work have lower average BMI compared to those who do not work in the labor market when both rural and urban women are considered. However, when the analysis considers only rural women, then those involved with farming activity have a higher average BMI compared to agricultural labour women. Further, women involved with agricultural labour alone or any type of agricultural work were better off with average BMI higher than those involved with non-agricultural labor (Viswanathan et al. 2015). Similarly, Rao and Pingali (2018) find that growth in income from agricultural activities among farming households had an impact on the increase in average BMI but this was not the case with growth in non-agricultural incomes among rural households. 4.4.2 Availability Availability or accessibility of resources that enhance nutrition security is another key factor. Dietary diversity is one component that can be enhanced wither through production diversity in agriculture and allied activities. Public policies like the public distribution system (PDS) allows the regular purchase of cereals at

88 Brinda Viswanathan and Getsie Immanuel a subsidised price while NREGA gives scope for access to cash incomes in lean season so that would enable smoothening of food consumption including a limited compromise in dietary diversity. While considering the influence of production diversity on dietary diversity the role of agriculture as a provider of diverse nutrient is taken into consideration. This makes it a unique economic activity compared to all others creating the possibility of increased access to diverse diet from home consumption while it is also a source of income (Kadiyala et al. 2012; Ruel and Alderman 2013). The latter feature enables the households to supplement the diets through purchase of food items not grown by the household or which need further processing prior to consumption, provided markets are well developed (Hoddinot et al. 2014). Evidence from India further show that foods like pulses that would need further processing are based on market consumption and access to markets is crucial while for cereals market and own-production have a role to play in improving BMI (Rao and Pingali 2018). In the absence of well-integrated markets, households will have to produce foods necessary for a diverse diet and in this sense the production and consumption decisions are also inseparable (Singh et al. 1986). The dietary diversity index reduces the likelihood of CED while the expenditure share of cereals in total food expenditure, which is a measure of low diversity in food consumption, increases the likelihood of CED (Viswanathan et al. 2015). On the other hand, in the same study, changing the empirical specification to a two equation model with an endogenous dietary diversity variable at the households level and a quantile regression model for individual women’s BMI as the second (stage) equation provides some useful insight connecting agriculture and nutrition. The results show that both home-production and cash income from the sale of crop in the market improves dietary diversity of the household. The second stage estimation of the quantile regression of women’s BMI shows that bottom 20% and top 10% of the BMI quantile is not influenced by dietary diversity but the role of dietary diversity is highest at 0.20–0.40 quantile. It could be that at the lowest level of BMI there is perhaps less scope for diversification due to limited resources of land and credit but once these resource constraints are not that binding the effect shows up. 4.4.3 Absorption In the context of nutrition the metabolic rate influenced by the physical activity carried out by the individual affects the energy expenditure and hence the BMI. Similarly, poor health condition affects the absorption of nutrients by the human body so that there is weight loss and hence reduction in BMI. Dang et al. (2019) find that increase in overweight and obesity can be in large part attributed to women’s withdrawal from labour market. It is quite often difficult to collect information on individual dietary habits and physical activity data from secondary data sources. Time-use surveys based on

Women’s BMI in rural India

89

primary survey that accounts for the daily routine activity and combined with 24-hour dietary recall will enable a better assessment of women in agriculture. Rao and Raju (2019) attempt to do so based on a very small sample of women in two different agro-climatic zones of India. Better and safer access to basic infrastructure like water, electricity, sanitation and cooking fuel has a direct impact on the individual’s health status by preventing frequent (infectious) illnesses. For instance, Rao et al. (2007) show that the collection of drinking water and fuels, which is primarily carried out by women, has considerable adverse impact on women’s nutritional reserves. Use of fuels like firewood, crop residue or dung etc. exposes women to harmful smoke while cooking, leading to a higher incidence of respiratory illnesses among women and children (Parikh et al. 1999; Duflo et al. 2008). Jose and Navaneetham (2010) show that CED rates for women in India in 2005– 06 was higher by about 15–20 percentage points when there was either no access to toilets, or when water had to be fetched from outside household premises, or the cooking fuel used polluted the air indoor. Even after wealth effects, rural/urban residence and other socio-demographic variables were controlled for, each of these variables were individually relevant in explaining the presence or absence of CED among women in India. The results of a study may vary if a mean regression is used compared to one across the distribution of BMI. Thus, the quantile regression enables in getting a better perspective across the wide range of BMI unlike the single (mean) regression model. Using this econometric technique, Viswanathan et. al (2015) show that the impact of water purification is higher for the lowest BMI quintile than for the higher BMI quintiles. On the other hand, having access to regular piped water compared to those who have to fetch water from a longer distance, has a significant positive impact with the effect more visible for the higher BMI quintiles and similarly the use of clean cooking fuel has an impact only for higher BMI quintiles. 4.4.4 Awareness When we consider health and nutrition, awareness about the quantity and quality of dietary intake becomes very crucial. A diverse diet helps in building immunity and at the same time a nutritious diet helps in recuperating faster from ill health. General awareness plays an important role in periodic health check-ups, healthy diet and timely immunisation, feeding, and care practices for children. Level of education plays a direct role while women’s empowerment would play an indirect role in improving awareness. An educated woman has better abilities for the control of physical and financial assets and is motivated to eat a healthy diet and feed their babies and children foods that meet their special nutritional requirements. Women’s education and paid work have been shown to be associated with overall well-being of a household or a region, perhaps due to the autonomy and agency effects (Dreze and Sen

90 Brinda Viswanathan and Getsie Immanuel 1995; Agarwal 1997; Sen 1999). Average CED rates in rural India in 2005 is as high as 30% among illiterate women with an average BMI value of 20 and there is a two percentage point decline in CED rates as level of education increases to primary plus middle and then six percentage point decline each for the secondary and higher secondary levels of education (Viswanathan et al., 2015). The graduate women have about 10 percentage point lower CED rates than the illiterates. Thus there is a huge gradient for the influence of education on nutrition insecurity. Many women in developing countries cultivate, purchase and prepare much of the food eaten by their families, but they often have limited access to information about nutrition. Since women’s status is a latent variable and is multidimensional in nature, most studies use proxy measures consisting of several indicators that depict sources of power such as education or age at marriage. Hindin (2005), in a study of malnutrition in Zimbabwe, Malawi and Zambia, suggests that women who have lower levels of autonomy and status within household are more likely to experience undernutrition. Bhagowalia et al. (2012) find that women’s empowerment, which includes her mobility, decision-making power, and attitudes toward verbal and physical abuse, is positively associated with her nutritional outcome as well as that of her children. Similar findings of higher CED rates—a 10% point difference—is noted among women with autonomy in decision-making related to daily purchases, major purchases, health care, or mobility (visits to relatives and friends) compared to the less “luckier” in this context (Jose 2012). Empowerment variables noticeably influence the likelihood of low BMI than quantiles of BMI. Curtailed freedom in terms of not being allowed to go to the market to purchase groceries and socio-cultural restrictions in terms of men and women not eating the meals together and the custom of purdah (covering the head with a veil) increases the risk of CED among women (Viswanathan et al. 2015). Inclusion of education or religion or caste or state level dummy variables (which capture regional differences in these restrictions) results in either some of the coefficients becoming insignificant or the magnitude declines. Thus, indicating that behavioural changes in these regressive aspects empower women from being less discriminated making them capable to take decisions regarding their own well-being as well as their family members, particularly, the children. Government plays a major role in providing basic amenities (like piped water, sanitation facilities, clean cooking fuel and education) to people and hence could also reflect regional effectiveness of the provisioning and maintenance of essential public goods.

4.5 Methodology The focus of this chapter is on understanding the factors associated with risk towards CED among women in rural India. As explained earlier, different ranges of BMI refer to different states of nourishment: chronically energy deficient (or CED) relates to undernutrition when BMI is less than 18.5; normal health when BMI is above 18.5 but below 25; overweight when BMI is between 25 and 30; and obese when BMI is 30 or above. For a developing country like India, a concern for policy intervention is

Women’s BMI in rural India

91

the issue of undernutrition than malnutrition due to high levels of BMI, even though latter is also a becoming concern for public health. The inputs required for policy intervention in rural area would be an assessment of the role of agriculture related variables in explaining nutrition insecurity. The explanatory variables relate to economic status, access to basic amenities like water, sanitation, health and clean cooking fuel, women’s empowerment and education, social status and religious affiliation. The model is estimated using a probit model where the dependent variable takes a value of 1 if the woman has BMI below 18.5 and 0 otherwise. Pr(Yi=1|X) = Φ(Xi β), t

With Φ(t ) =

1

∫

2˜

−∞

(1) exp(−

z2 )dz, 2

X is the vector of explanatory variables and the following likelihood function is maximised to obtain the estimators of the unknown parameters β. L(˜ | Y , X) =

∏ Pr( y = 1| X ) ∏ Pr( y = 0 | X ) i

i

y=1

=

i

i

(2)

y=0

∏[Φ(X ˜)] ∏[1−Φ(X ˜)] i

i

y=1

= y=0

ln L(˜ | Y , X) = ∑ [Φ(Xi ˜)] ∑ [1−Φ(Xi ˜)] y=1

(3)

y=0

Two different variants of this model are estimated to capture different aspects of agriculture as explanatory variables in the model. 1

2

In the first variant, categorical variables representing the major source of household income is used. Cultivation or managing the livestock or agricultural wage labour as the major source of income represents an agricultural household. A statistically significant coefficient with a positive sign for a particular source of income would imply that women in such a household is more likely to have CED when compared to a reference household say, whose major income source is “other source.” More details on the types of income sources as available in the data is discussed in the section on data and subsequently in the section on results. In the second variant, share of income from farming, livestock rearing, agricultural property and agricultural wages and also share of non-agricultural wages are included as separate variables. The significance of any or all the first three variables in the model would capture the differences in BMI between agriculture and non-agriculture household. The positive coefficient would then indicate agricultural households to be more likely to have CED and vice versa for a negative coefficient.

92 Brinda Viswanathan and Getsie Immanuel

4.6 Data In this study the focus is on nutrition insecurity as captured by CED among rural women aged 20–45 years in India. The chapter compares the changes in two years 2005–06 and 2011–12. The analysis has been carried out using data from the two waves of India Human Development Survey conducted jointly by the University of Maryland and the National Council of Applied Economic Research (Desai et al. 2007; Desai and Vanneman 2017). It contains information from 42,152 households in 1,503 villages and 971 urban neighbourhoods across all the states and union territories of India in 2011–12, with re-interviews of households interviewed for IHDS-I in 2004–05. The data set provides information on variables related to health, education, employment, economic status, marriage, fertility and social capital. The data provides regional segregated information and offers gender centric and institutional information. The survey included two questionnaires, one for the household head and the other for the eligible women, who is typically the wife of the household head. The questions related to household wealth, income and expenditure were asked from the household head while the questions related to health, education and some other social indicators were administered to women. 4.6.1  Factors    influencing BMI Given the objective of this study, the interest is to understand the influence of lifestyle linked to farming that directly or indirectly captures affordability, accessibility, awareness and absorption towards the risk of CED. These agriculturally relevant factors are agriculture as a main source of income (variant 1) or share of agricultural income in total household income (variant 2), and women’s own labor contribution to farming activities. There are six categories referring to the employment status and occupation of women, with five dummy variables for the employment categories and a reference category for those not actively engaged in the labour market: 1 2 3 4 5 6

Self-employed only in agriculture comprising of farm work and animal rearing Self-employed in agriculture and sometimes engaged as agricultural wage labor Only as agricultural wage labour Non-agricultural wage labour Salaried work or in business Not actively engaged in any economic activity in the labour market

The reference category chosen in the econometric model is the last one consisting of women who do not actively participate in the labour market and are mainly involved in domestic work. For affordability we consider household’s wealth status based on quintiles of asset index.

Women’s BMI in rural India

93

Water, sanitation, electricity and cooking fuel signify some basic amenities of civilised life. More importantly, they have strong influences on health in terms of causing short-term illnesses like diarrhoea and fever or long-term impacts on respiratory and immune systems, which in turn affect the absorption of the nutrient intakes. Hence, limited or no access to these basic amenities is considered to be associated with the risk of CED. Women’s employment status would reflect the additional physical activity alongside household chores and hence type of employment including those not involved in labour market is used here to assess the nature of energy expenditure. Short-term morbidity status, pregnancy status, access and use of health care facilities during short-term illness and antenatal care during pregnancy are considered essential to maintain a good health. A pregnant woman should, on an average, have a higher BMI compared to others and hence this has been used as a control variable. Though one could have dropped the group of women who are pregnant, the intention is to also assess how access to antenatal care impacts their BMI after controlling for other variables. So the focus is more on the possible effect of access to health care and its impact on BMI and how the magnitude of this coefficient varies across quintiles of BMI. To capture the variations in access to different sources of health care, the women who report pregnant have been classified into: (1) pregnant but no access to antenatal care; (2) pregnant with access to antenatal care and visiting doctor or nurse; (3) pregnant with access to antenatal care and visiting dai or others. Short-term morbidity status was captured by the number of days ill with fever etc. a month prior to the survey so that we expect that this could have resulted in loss of body weight and hence a lower average BMI even after controlling for other variables. Further, this variable is interacted with the type of medical care facility used, if at all used, when reported sick. The medical care sought is classified as: (1) ill but did not seek treatment; (2) ill and visits public doctor; (3) ill and visits private doctor; (4) ill and seeks traditional help. Awareness is captured by women’s own education level and based on the empowerment variable. Education is a categorical variable represented by five groups who are (1) either not literate, or (2) have finished primary, or (3) secondary, or (4) higher secondary or (5) college-level education. Women’s own status within the household is also considered as an important contributor towards better BMI through better bargaining power and to be able to negotiate timely access to resources. In this study we have used three different binary variables to capture women’s status in a household: whether women have permission to go to the grocery store; whether women in the household follow the practice of wearing purdah and whether the women eat along with men or after them. These variables are a reflection of both societal and familial practices that could indicate a woman’s ability to influence her own well-being as well as those of her family members. Indian society has been divided, since ancient times, into various castes (varna system of the Hindus) on the basis of their occupation. Relegation of menial jobs to some social groups with limited or no access to productive resources and the

94 Brinda Viswanathan and Getsie Immanuel subsequent persistent discrimination in several other domains of social and economic status has created high socio-economic disparity among these groups in the Indian society. Five major castes: upper caste Hindus, Other Backward Caste, Scheduled Castes, Scheduled Tribes (reference category) and Other Castes are included as dummy variables on nutritional outcomes. Household composition is expected to capture the impact on the effort of women due to childcare so that women in households with a higher proportion of children than adults would expend more energy and hence have a lower BMI on average after controlling for other factors. The household composition variables relate to the proportion of people in the following age groups in a household: birth to four years of age, five to 14 years of age, 15 to 60 years of age and above 60 years of age. Since these proportions add up to one, the first group is excluded and is taken as equivalent to the reference category as would be in the case of categorical variables. Table 4A.1 defines all the explanatory variables used in the analysis.

4.7 Empirical evidence: preliminary analysis Apart from discussing the nature of different variables used in the analysis, this section presents cross-tabulation of the covariates and CED rates. This enables us to understand the nature of association of the regressors with BMI. 4.7.1   CED rates and mean BMI We observe that most rural households in India depend on agriculture as their major source of income and that many agricultural households have diversified sources of income and this trend has been similar in both 2005 and 2011. Among the rural women aged 20–45 years, 36% of them belong to households that report cultivation as the major source of income in both the rounds, while about 21% and 15% in IHDS-I and IHDS-II belong to households that report agricultural wages as the main source of income (Table 4.1). There has been a decline in the share of agricultural labour households and increase in non-agricultural labour households in 2011 compared to 2005. As for undernourished (or CED) women, we observe that those in wage labour households are the worse off with non-agricultural labour being worst off than all others, and women in households reporting salaried and professionals are the better off; women in cultivator households are somewhat in the middle. Similar ranking is observed for the two years but more importantly there has been a decline in CED rates and more so among the labour households. It is, however, observed that the mean BMI of women in all these households is well above 18.5 with lower standard deviation among wage labour households indicating a narrower distribution around the mean compared to women in other types of households. The increase in BMI is marginal while the decline in CED rates among the women in agricultural households is a healthy sign of improvement unlike among the salaried where the mean BMI has increased further (with a lower CED rate) indicating possibilities of higher overweight and obesity rates.

36.0 20.7 17.6 4.9 6.7 11.0 3.2 100

35.9 15.3 24.5 1.4 8.0 10.1 4.9 100

62.8 12.7 11.0 8.8 9.1 13.8 18.1 30.2

49.6 66.1 22.9 16.7 15.4 17.0 16.4 32.3

2011

2005

2005

2011

Average share of farm Income (%)

Distribution (%)

27.6 30.6 34.7 22.7 22.9 20.7 24.3 28.1

2005

CED (%)

26.6 28.5 29.4 14.4 21.3 18.0 24.0 26.0

2011 20.5 20.1 19.9 21.3 21.3 21.4 21.3 20.5

2005

Mean BMI

20.9 20.6 20.7 21.7 21.8 22.2 21.6 21.0

2011

3.21 2.98 3.00 3.67 3.61 3.30 3.44 3.24

2005

3.61 3.51 3.56 3.62 4.08 4.01 4.14 3.73

2011

Standard deviation of BMI

Note: (1) Distribution of women across households with different major sources of income; (2) Share of farm income is in total income; (3) CED or chronic energy deficiency is BMI below 18.5; (4) The unit for mean and standard deviation of BMI is kg/m2.

Cultivation Agricultural Wage Labour Non-agricultural Wage Labour Artisan Trade & Business Salaried & Professionals Others Total

Major source of income

Table 4.1 Distribution of women aged 20 to 45 years, CED rates and mean BMI across major sources of income in rural areas

96 Brinda Viswanathan and Getsie Immanuel In order to assess if risk for CED rates and mean BMI are statistically different across women in households with different sources of income, a regression model is estimated with dummy variables representing the different sources of income. Yi=β1D1i+β2D2i+β3D3i+β4D4i+β5D5i+β6D6i+β7D7i+ui

(4)

D1i=1 if the ith woman belongs to a household with cultivation as a major source of income; =0 otherwise. D2i=1 if the ith woman belongs to a household with agricultural wage labour as the major source of income; =0 otherwise. D3i=1 if the ith woman belongs to a household with non-agricultural wage labour as the major source of income; =0 otherwise. D4i=1 if the ith woman belongs to a household with artisanal work as the major source of income; =0 otherwise. D5i=1 if the ith woman belongs to a household with trade and business as the major source of income; =0 otherwise. D6i=1 if the ith woman belongs to a household with regular salary or profession as the major source of income; =0 otherwise. D7i=1 if the ith woman belongs to a household with others sources like rents, pensions etc. as the major source of income; =0 otherwise. The dependent variable Yi in equation (4) above takes the value 1 if the woman is CED and 0 otherwise, then the expression E(Yi|D1i=1, Dji=0, ∀ j≠1)=β1 implies that the estimated coefficient βˆ 1 from the data is the mean CED rate of the women who are from households with cultivation (D1i=1) as the major source of income. The estimated coefficient values are multiplied by 100 to show the rates in percentages in Table 4.2. The lower part of Table 4.2 presents the results of the test of hypothesis: H0: βi=βj, H1: βi≠βj; i and j here refer to two different major sources of income. We also estimate another model with Yi as the woman’s BMI. In this model, E(Yi|D1i=1, Dji=0, ∀ j≠1)=β1 implies that the estimated coefficient βˆ 1 from the data is the mean BMI of the women who are from households with cultivation (D1i=1) as the major source of income. Similarly, each of the estimated coefficients is the

F-statistic 11.82*** 14.33*** 28.35*** 0.26 38.66***

Tests of hypothesis

H0: β1=β2, H1: β1≠β2 H0: β2=β3, H1: β2≠β3 H0: β1=β2=β3, H1: At least one is different H0: β4=β5=β7, H1: At least one is different H0: β1=β6, H1: β1≠β6

0.000 0.000 0.000 0.768 0.000

p-value

0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.96 0.36 2.94* 3.75** 40.70***

F-statistic

26.6*** 28.5*** 29.4*** 14.4*** 21.3*** 18.0*** 24.0*** 0.162 0.547 0.053 0.024 0.000

p-value

0.000 0.000 0.000 0.000 0.000 0.000 0.000

p-value

29.39*** 8.31*** 38.47*** 0.03 127.54***

F-statistic

20.5*** 20.1*** 19.9*** 21.3*** 21.3*** 21.4*** 21.3***

Coefficients

2005

BMI (kg/m2)

0.000 0.000 0.000 0.973 0.000

p-value

0.000 0.000 0.000 0.000 0.000 0.000 0.000

p-value

6.64** 0.44 4.29** 0.22 99.13***

F-statistic

20.9*** 20.6*** 20.7*** 21.7*** 21.8*** 22.2*** 21.6***

Coefficients

2011

0.010 0.506 0.014 0.804 0.000

p-value

0.000 0.000 0.000 0.000 0.000 0.000 0.000

p-value

Note: (1) The above coefficient estimates are obtained from equation (1) for two types of dependent variable as in (2a) and (2b) respectively. (2) * p-value < 0.10; ** p-value < 0.05; *** p-value < 0.01 and this applies henceforth for all of the following tables.

27.7*** 30.7*** 34.7*** 22.7*** 22.9*** 20.7*** 24.3***

Coefficients

Cultivation (β1) Agricultural wage labour (β2) Non-agricultural wage labour (β3) Artisan (β4) Trade & business (β5) Salaried & professionals (β6) Others (β7)

Coefficients

2005

Major source of income p-value

2011

CED (%)

Estimated coefficients

Table 4.2 Mean differences in BMI and CED rates across major sources of income

98 Brinda Viswanathan and Getsie Immanuel mean BMI of the women from households with different major sources of income. Thus, the null of testing for equality of some or all the coefficients in the two estimated models above is equivalent to testing for the CED rates or mean BMI to be same for women from households with different sources of income. The results based on the F-tests show that the CED rates (or mean BMI) of women in cultivator households is lower (higher) than those in wage labour households in 2005. However, in 2011 we observe that this has changed. The CED rates between cultivator and agricultural labour households and between agriculture and non-agricultural labour households are not different. In other words, there has been an overall decline in CED rates and the labour income households seem to have caught up. However, it would be difficult to infer on the pathway for the same. Further, women in households with either “trade & business” or “salaried & professionals” or “others” as major source of income have similar mean BMI and CED rates as the nulls of equality of those coefficients cannot be rejected based on the p-value of the test statistic. In this context, similar pattern is observed in both the survey years. IHDS provides information on possession of assets (durable goods) in a household which can significantly ease the work of women and hence save a lot of energy. Also, income is a short-term (transitory in nature) representation of economic well-being. The assets/wealth is a long-term counterpart as households accumulate assets over a period of time after saving from the income earned. The asset or wealth index is estimated using principal component analysis by taking into account the basic household amenities like owned house, cycle, motor, sewing machine, wall clock, cot, chair, fan etc. The quality of the house such as pucca wall, roof etc. is also taken into consideration. The first principle component forms the index which is then categorised into five groups comprising the five quintiles and referred to as the poorest, poor, middle, rich and richest asset groups (from lowest to highest quintile). The bottom most quintile is taken as the reference category for the econometric analysis and we expect the coefficients to be positive and significant with respect to this omitted category. Table 4.3 shows the CED rates across quintile classifications for three different variables based respectively on per capita income, monthly per capita consumption and wealth index. The households are further segmented into non-farm and farm households within each of these economic status variables. In 2005, the proportion of CED women across the different types of quintiles varies indicating that the distributions of these variables are different in that these three different economic status variables are not very strongly associated. For instance, the bottom most quintile has lower CED rates based on income when compared to that based on assets. Consequently, the gap in CED rates between bottom most and top most quintiles based on asset index are substantially higher. CED rates are by and large higher for women among farm households than that of non-farm households. In 2011, the distribution of CED rates is similar not only among farm and non-farm households for bottom two quintiles but the overall distributions have also become similar. What is more important to note is that among non-farm households the rate of decline of CED is steeper compared to farm households and that for the

2011

32.2 35.0 25.8 25.5 16.0 27.0

30.8 36.0 31.0 27.2 18.2 28.9

33.9 29.2 24.0 22.5 13.3 25.6

33.0 28.4 25.5 23.1 18.4 26.6

32.6 31.9 25.5 24.5 18.4 27.0

36.0 33.4 28.4 26.1 19.8 28.9

32.4 28.9 24.7 19.8 16.8 25.7

33.8 28.2 26.8 23.8 17.7 26.6

38.1 31.6 28.5 24.0 13.5 27.0

40.5 34.9 29.1 24.7 18.6 28.9

36.1 27.0 22.9 19.1 12.1 25.6

37.5 28.5 26.1 21.6 15.8 26.6

Note: The income quintile on the bases of per capita household income, MPCE quintiles on the bases of monthly per capita consumption expenditure (MPCE) and asset quintiles are based on quintiles of asset index.

Q1 Q2 Q3 Q4 Q5 All

2005

2005

2011

Asset quintile

MPCE quintile

Non-farm Farm Non-farm Farm Non-farm Farm Non-farm Farm Non-farm Farm Non-farm Farm household household household household household household household household household household household household

Quintile Per capita income quintile classes 2005 2011

Table 4.3 Percentage of women with CED across income, consumption and asset quintile: comparing non-farm and farm households in rural areas

100 Brinda Viswanathan and Getsie Immanuel upper quintiles the CED rates are higher among farm households than non-farm. These findings indicate an improvement in health status among rural women in farm households and the somewhat higher CED rate in the upper quintiles could be a reflection of higher physical activity due to farm work among these women and that may have to be explored subsequently. 4.7.2

State level variations

State level policies or the quality of service delivery of the central (welfare) schemes as well as other economic, social and cultural features that could systematically differ across states of India can have differential impacts on the average BMI across states. Table 4.4 shows that a larger share of women are from states like Uttar Pradesh, Bihar, Rajasthan, Madhya Pradesh, West Bengal, Andhra Pradesh, Orissa. Noteworthy is that in many states, CED rates are lower among women in farm households than in non-farm household except for states like Rajasthan, Orissa and Gujarat in 2005. In 2011, there seems to be a convergence with lesser variability in CED rates for both non-farm and farm households. Table 4.4 Distribution of women and CED rates across states of India (%) States

Distribution of women

CED rates

2005

2005

2011

2011

Non-farm Farm Non-farm Farm Non-farm Farm Non-farm Farm Jammu & Kashmir 0.5 Himachal Pradesh 0.2 Punjab 3.1 Uttarakhand 1.1 Haryana 2.6 Rajasthan 3.1 Uttar Pradesh 9.4 Bihar 7.3 Assam 3.6 West Bengal 10.2 Jharkhand 3.9 Orissa 3.3 Chhattisgarh 1.5 Madhya Pradesh 4.0 Gujarat 4.1 Maharashtra 7.1 Andhra Pradesh 15.0 Karnataka 4.3 Kerala 4.3 Tamil Nadu 9.8

1.6 1.2 1.2 2.4 1.5 6.4 17.4 8.6 2.5 7.3 4.7 6.0 4.4 5.7 4.4 4.4 4.4 4.5 1.7 2.3

1.4 1.0 2.3 1.4 2.1 5.2 15.7 8.5 3.7 10.3 5.1 4.5 2.2 4.1 3.2 5.7 8.6 4.0 2.9 6.4

1.1 0.5 1.7 1.9 1.8 6.5 16.1 8.6 3.4 5.1 3.4 4.8 4.5 7.3 5.5 11.8 6.6 5.6 0.5 2.1

17.0 20.5 12.8 32.9 22.5 32.4 32.2 30.7 11.8 18.6 37.0 24.2 43.9 35.8 34.7 29.9 20.2 22.5 7.0 15.8

11.7 16.5 12.1 28.5 20.3 30.0 27.1 23.7 15.9 12.1 23.4 28.0 43.4 40.4 33.2 29.8 21.5 21.2 10.1 25.6

17.1 22.9 10.3 45.1 16.7 18.9 36.2 29.2 2.5 33.8 31.5 33.7 37.7 30.3 31.8 31.4 27.2 36.8 8.1 22.6

13.3 20.6 8.3 37.4 15.7 25.6 32.1 25.9 2.3 28.8 24.5 38.2 35.2 28.9 34.5 36.9 24.4 35.6 8.4 23.4

Women’s BMI in rural India 4.7.3

101

Factors associated with risk to CED

In order to understand what are the factors associated with nutritional insecurity as assessed by low BMI values, a probit model is estimated. The binary dependent variable is whether the woman is CED (BMI less than 18.5) or not. In this analysis the focus is on comparing farm and non-farm households with following two variants of the model that differ in how agriculture/non-agriculture related aspects feature in the model: (1) In variant 1 has household’s major source of income as categorical variable, (2) Variant 2 has share of income from various economic activities. Other demographic, economic, environmental and social variables are the same in both the variants of the model. The list of factors considered to be affecting low BMI values are mentioned in appendix Table A4.1. As indicated earlier these factors capture aspects of affordability, availability, awareness, and absorption. The descriptive statistics in Table 4.5 shows that overall the CED proportions are marginally lower in 2011 compared to 2005 and the average share of income from agriculture is a little above two-thirds in both the years. The demographic composition has shifted towards the middle aged adults with their share in 2011 close to 70% compared to 60% in 2005. The shift has come from younger age group of children. The absorption or amenities related variables like electricity, clean cooking fuel, and open defecation are all showing changes for the better. Similarly, health-seeking behaviour with respect to short-term illness seems to have shifted marginally towards private doctors rather than not seeking any help. The demographic shares of religious and caste groups show a small change towards the minority groups (Muslims and Scheduled castes and tribes) and this could possibly be because the other groups could have lower family size compared to these economically weaker and socially disadvantaged groups. The awareness variables linked to education and empowerment also show improvements. The education levels have improved in all levels and thereby a more than 10 percentage point decline in illiterates, with about 5% of graduate women in 2011. Among the empowerment variables, more families report men and women having their meals together and the women in purdah seem have also declined marginally but the mobility variable seem to have declined shares. Women’s participation in the labour market has decline substantially in 2011 compared to 2005. Compared to 26% of women who were not active (working for more than 240 hours) in the labour market, about 46% are out of the labour market. This seems to have largely come from a declined in share of self-employed women from 57% to 6% but agricultural labour combined with self-employment work or agricultural labour alone has increased from 16% to 32% and 2% to 10% respectively. The results in Table 4.6 report the estimated coefficients and the p-values for the probit models for the two years separately. CED was less likely in 2005 among households that depended primarily on cultivation or agriculture in general when compared to “other” households after controlling for economic status and other variables. In 2011, source of income in general seem to have a less influence after controlling for other variables with fewer coefficients that are statistically significant. This is so in both variant 1 and variant 2 estimates but in variant 2 the

0.28 8.42 0.66 0.12 0.24 0.60 0.06 32.20 0.15 0.22 0.21 0.20 0.22 0.86 0.09 0.01 0.04 0.05 0.44 0.17 0.10 0.23 0.34 0.40 0.34

Hindus Islam Christianity Others Upper caste Hindus Other backward classes

Scheduled Caste

Scheduled tribes

Other castes Allowed to purchase groceries Eat with family members Do not practice Purdah

0.22 0.29 0.53 0.33

0.11

0.23

0.84 0.10 0.02 0.04 0.04 0.40

0.26 9.49 0.64 0.16 0.21 0.69 0.07 32.38 0.19 0.23 0.23 0.19 0.15

0.423 0.473 0.489 0.473

0.306

0.377

0.346 0.287 0.113 0.186 0.217 0.497

0.317 0.938 0.359 0.247 0.193 0.204 0.098 7.297 0.362 0.414 0.404 0.400 0.414

2005

2005

2011

Std. dev

Mean

CED proportion Log of household income per person (INR) Share of total agricultural income Share of non-agricultural wage income Share of members in age group 4–14 years Share of members in age group 15–60 years Share of members above the age of 60 years Age Asset quintile 1 or ‘poorest’ Asset quintile 2 or ‘poor’ Asset quintile 3 or ‘middle’ Asset Quintile 4 or ‘Rich’ Asset quintile 5 or ‘richest’

Variable

Table 4.5 Descriptive statistics of variables

0.412 0.455 0.499 0.471

0.314

0.420

0.371 0.302 0.147 0.199 0.195 0.490

0.436 0.964 0.335 0.941 0.200 0.211 0.117 7.508 0.393 0.423 0.423 0.395 0.357

2011 Has access to electricity Use clean cooking fuel Open defecation Traditional latrine Improved pit latrine Flush toilet Not literate Primary or middle Secondary Higher secondary Graduate and above Self-employed in agriculture Self-employed in agriculture and agricultural labour Agricultural labour Non-agricultural labour Not in labour force Not pregnant Pregnant but no antenatal access Pregnant with antenatal access and going to doctor or nurse Pregnant with antenatal access and going to dai or others No illness (short-term morbidity) reported in the last 30 days Ill but did not seek treatment Ill and goes to public doctor Ill and goes to private doctor Ill and seeks traditional help

Variable

0.01 0.02 0.07 0.00

0.89

0.00

0.02 0.01 0.24 0.95 0.02 0.03

0.61 0.13 0.75 0.10 0.04 0.11 0.54 0.31 0.10 0.03 0.02 0.57 0.16

2005

Mean

0.01 0.04 0.13 0.02

0.81

0.00

0.10 0.06 0.46 0.95 0.02 0.02

0.84 0.34 0.56 0.15 0.23 0.05 0.41 0.33 0.15 0.06 0.05 0.06 0.32

2011

0.094 0.143 0.261 0.057

0.308

0.059

0.134 0.071 0.425 0.209 0.126 0.159

0.488 0.333 0.433 0.300 0.203 0.308 0.498 0.462 0.306 0.180 0.122 0.495 0.363

2005

Std. dev

0.093 0.190 0.335 0.122

0.392

0.048

0.305 0.229 0.000 0.214 0.149 0.151

0.368 0.474 0.497 0.353 0.418 0.209 0.492 0.470 0.356 0.245 0.212 0.246 0.465

2011

2005 −0.197** −0.237** −0.004 −0.174 −0.180* −0.144

0.027 0.011 0.964 0.119 0.071 0.133

0.018 −0.035 −0.062 −0.252** −0.075 −0.138**

2011 0.715 0.529 0.236 0.020 0.208 0.014

2005

−0.028 0.035

Quintile 2 or ‘poor’ Quintile 3 or ‘middle’ Quintile 4 or ‘rich’ Quintile 5 or ‘richest’ Allowed to purchase groceries (Yes=1, No=1) Eat with family members (Yes=1, No=0) Do not practice Purdah (Yes=1, No=0)

−0.028 −0.146*** −0.216*** −.365*** −0.0859** −0.0294 −0.14***

Asset quintile groups [quintile 1 or ‘poorest’=reference]

Share of income from cultivation Share of income from livestock rearing Share of agricultural property income Share of agricultural wage income Share of non-agricultural wage income Logarithm of per capita income Logarithm of monthly per capita consumption expenditure 0.562 0.006 0.002 0.000 0.014 0.377 0.000

0.205 0.318

−0.105*** −0.171*** −0.202*** −0.363*** −0.016 −0.030 −0.136***

−0.043*** −0.043**

0.000 0.000 0.000 0.000 0.459 0.149 0.000

0.001 0.046

−0.031 −0.145*** −0.213*** −0.359*** −0.081** −0.0319 −0.135***

−0.143** −0.254** −0.175 −0.114* 0.13* −0.03 0.065*

Household income diversity [other income shares except from agriculture and non-agricultural wage=reference]

Agriculture and allied activities Agricultural labour Non-agricultural labour Artisans Petty trade and business Salaried and professionals

p-value

Coefficient

Coefficient

Coefficient

p-value

Variant 2

Variant 1

Household’s major source of income [other sources=reference]

Variables

Table 4.6 Estimates from probit model for CED (2005 and 2011)

0.519 0.006 0.002 0.000 0.018 0.339 0.001

0.016 0.015 0.254 0.079 0.063 0.174 0.067

p-value

0.000 0.000 0.000 0.000 0.455 0.145 0.000

0.003 0.793 0.125 0.430 0.614 0.001 0.040

p-value

(Continued)

−0.104*** −0.173*** −0.207*** −0.374*** −0.016 −0.030 −0.137***

0.076*** −0.003 0.105 0.019 0.011 −0.042*** −0.044**

2011

Coefficient

p-value

−0.283*** −0.176*** −0.0433 −0.15**

−0.0421 −0.278** −0.256***

2005

0.003 0.002 0.470 0.015

0.459 0.017 0.002 −0.246*** −0.158*** −0.179*** −0.253***

−0.160*** −0.380*** −0.166***

2011

0.000 0.000 0.000 0.000

0.000 0.000 0.003 −0.286*** −0.179*** −0.0651 −0.142**

−0.0615 −0.259** −0.255***

2005

Coefficient

Coefficient

Coefficient

p-value

Variant 2

Variant 1

0.003 0.001 0.269 0.021

0.279 0.026 0.002

p-value

−0.211 −0.399*** −0.577*** −0.0153** −0.219***

Traditional latrine VIP latrine Flush toilet Use clean cooking fuel (Yes=1, No=0) House has access to electricity (Yes=1, No=0) Age

−0.198*** −0.197** −0.181*** 0.0138 −0.0165 −0.012***

Sanitation facility [none, open fields=reference]

Share of members in age group 4–14 years Share of members in age group 15–60 years Share of members above the age of 60 years Household size Drinking water is treated in some form (Yes=1, No=0) 0.000 0.020 0.001 0.809 0.691 0.000

0.102 0.007 0.004 0.027 0.000

−0.148*** −0.103*** −0.098* −0.106*** 0.041 −0.019***

−0.169** −0.195*** −0.175** −0.003 −0.053*

0.000 0.000 0.080 0.000 0.161 0.000

0.016 0.006 0.050 0.492 0.073

−0.201*** −0.194** −0.181*** 0.00264 −0.022 −.0118***

−0.223* −0.419** −0.536*** −0.0136** −0.225***

0.000 0.021 0.001 0.963 0.593 0.000

0.085 0.004 0.007 0.039 0.000

Household composition, share of members in different age group [children in age group birth to four years=reference]

Upper caste Hindus Other backward classes Scheduled caste Other castes

Caste [scheduled tribe=reference]

Islam Christianity Other religions

Religion [Hindus=reference]

Variables

Table 4.6 (Continued)

−0.151*** −0.105*** −0.103* −0.113*** 0.040 −0.019***

−0.171** −0.186*** −0.142 −0.001 −0.059**

−0.247*** −0.156*** −0.186*** −0.248***

−0.172*** −0.388*** −0.166***

2011

Coefficient

0.000 0.000 0.065 0.000 0.166 0.000

0.015 0.009 0.110 0.773 0.044

0.000 0.000 0.000 0.000

0.000 0.000 0.003

p-value

2005 −0.0474 −0.0178 0.0288 0.00673

0.004 0.001 0.001

−1.13***

0.572 0.885

0.007 0.052

−0.455*** −0.348***

0.0332 −0.014

0.131*** 0.11*

0.216 0.758 0.783 0.961

−0.549**

−0.469*** −0.410***

0.102*** −0.002

0.015 0.035

−0.116** −0.194*** −0.018 −0.119**

0.167 0.0419 0.225*** −0.0351 0.881**

Note: *p-valuet

adjusted R2 =.00235

Female

166 Swarna Sadasivam Vepa and Rohit Parasar Table 6.8 Factors influencing body mass index in the composite state before bifurcation (OLS) No. of Obs. M = 33608: F = 38212

Males

Females

BMI

Adj R2 =.001221

Adj R2 =.005285

Variables

Coefficient

Coefficient

Household amenities index Social group: others = base Scheduled castes 1 Scheduled tribes 2 Other backward castes 3 Years of schooling Health scheme/insurance Land owned by the household (hec.) Irrigated land owned by household (hec.) Any treatment of drinking water Female/male ratio of the household Age in months Land-location category State: AP= 28 TE= 36

P>t

P>t

0.5690

0.026

1.2053

0.000

0.4434 0.1395 0.0827 0.0734 0.3350 −0.0037 0.0661 −0.8789 0.7953 0.0020 −0.0881 −0.0855

0.240 0.834 0.776 0.020 0.257 0.732 0.438 0.030 0.419 0.019 0.634 0.022

0.1427 0.9160 0.3306 0.0427 −0.0003 −0.0203 0.1262 −0.6951 0.5641 0.0049 0.2795 −0.1483

0.659 0.137 0.225 0.217 0.999 0.045 0.310 0.099 0.559 0.000 0.127 0.000

Source: Regresion results are based on DLHS-4 (2012–13); highlight indicates significance

As BMI quantiles arrange individual in the ascending order of BMI, the men and women with chronic energy deficiency are in the lowest quantile the 20th. Less than 10% among men and about 10% among women, in the lowest quantile fall in the category of chronic energy deficiency. They are undernourished. Normal weight will be in the next two quantiles, viz., 40th, and 60th. Women and men in the 80th quantiles are likely to be overweight irrespective of which socio-economic group they belong. Despite this, in all the quantiles the household amenities show significant larger influence on body mass index both for men and women (Tables 6.9 to Table 6.12). In the first quantile, social group has no influence on the BMI of men, but it influences females—other caste females having higher BMI and those in the scheduled and scheduled tribes showing lower BMI. BMI is influenced by caste as expected only for females up to the 80th quantile. For men in all the quantiles there was no influence of caste. It shows positive association with higher BMI. For men, caste has no influence in all the quantiles. Ownership of land area and irrigated area do not influence the BMI of men in any of the quantiles except for irrigated land area in the 20th quantile for men. We can view it in two ways. First since ownership of dry land is not very useful as an asset, land has no influence on BMI. For a predominantly agricultural state with abundant foodgrain availability and a good public distribution system ownership

Household amenities index 0.6740 Social group −0.0122 Years of schooling 0.0642 Health scheme/insurance 0.1491 Land owned by the household 0.0000 Irrigated land owned by household 0.0290 Any treatment of drinking water −0.2425 Female/male ratio of the Household 1.5643 Age in months 0.0024 Land-location category 0.0570 _cons 16.3480 AP No. of Obs. Male = 15543: Female = 15621 TE No. of Obs. Male = 13124: Female = 13267

0.000 0.710 0.000 0.083 0.996 0.080 0.019 0.000 0.000 0.173 0.000

P>t

Coef.

0.0199 Males

Telangana

0.7748 0.000 0.355 0.0873 0.000 −0.003 0.0391 0.000 0.045 0.0919 0.269 0.468 −0.0026 0.835 −0.029 −0.0094 0.765 0.020 −0.2869 0.004 −0.464 −1.7180 0.000 0.984 0.0036 0.000 0.002 0.1721 0.001 0.345 16.6847 0.000 16.384 for all quantiles from 0.20 to 0.80 for all quantiles from 0.20 to 0.80

Coef.

q20

Coef.

0.0265 Males

.20 Pseudo R2 = BMI P>t

0.0347 Females

Andhra Pradesh

State

Table 6.9 Factors influencing BMI quantiles (0.20) of males and females in Andhra and Telangana

0.000 0.940 0.000 0.000 0.312 0.520 0.000 0.000 0.000 0.000 0.000

P>t

0.4990 0.0712 0.0003 0.1813 −0.0619 0.0602 −0.5731 −0.5528 0.0031 0.3698 16.4011

Coef.

0.0236 Females

0.000 0.038 0.978 0.063 0.108 0.211 0.000 0.059 0.000 0.000 0.000

P>t

0.7531 −0.0213 0.0740 0.2338 0.0027 0.0087 −0.2509 1.7818 0.0032 0.1064 17.4819

0.000 0.469 0.000 0.005 0.633 0.568 0.019 0.000 0.000 0.028 0.000

0.9871 0.0771 0.0374 0.1336 −0.0055 −0.0093 −0.2421 −1.6071 0.0053 0.2748 17.2470

Coef.

q40 Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

Coef.

0.0341 Males

.40 Pseudo R2 = BMI P>t

0.0454 Females

Andhra Pradesh

State

0.000 0.002 0.000 0.169 0.725 0.666 0.054 0.000 0.000 0.000 0.000

P>t 0.4012 0.0093 0.0614 0.3778 −0.0229 −0.0019 −0.3176 1.1651 0.0030 0.2834 17.8267

Coef.

0.0218 Males

Telangana

Table 6.10 Factors influencing BMI quantiles (0.40) of males and females in Andhra and Telangana

0.000 0.820 0.000 0.000 0.241 0.947 0.001 0.000 0.000 0.000 0.000

P>t

0.6112 0.0742 0.0105 0.3481 −0.0167 −0.0068 −0.5667 −0.6399 0.0042 0.3691 17.3723

Coef.

0.0302 Females

0.000 0.051 0.177 0.000 0.418 0.795 0.000 0.004 0.000 0.000 0.000

P>t

0.9434 −0.0141 0.0777 0.2923 −0.0018 −0.0072 −0.4734 1.9010 0.0037 0.1766 18.7209

0.000 0.657 0.000 0.001 0.767 0.637 0.000 0.000 0.000 0.002 0.000

1.2262 0.0467 0.0263 0.3142 −0.0094 −0.0244 −0.3935 −2.1610 0.0064 0.3292 18.6773

Coef.

q60 Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

Coef.

0.0374 Males

.60 Pseudo R2 = BMI P>t

0.0535 Females

Andhra Pradesh

State

0.000 0.084 0.032 0.003 0.524 0.226 0.007 0.000 0.000 0.000 0.000

P>t 0.5086 −0.0032 0.0687 0.3511 −0.0442 0.0253 −0.1756 1.4946 0.0034 0.3363 18.7373

Coef.

0.0230 Males

Telangana

Table 6.11 Factors influencing BMI quantiles (0.60) of males and females in Andhra and Telangana

0.000 0.937 0.000 0.000 0.022 0.423 0.129 0.000 0.000 0.000 0.000

P>t

0.7913 0.0985 0.0139 0.2351 −0.0299 0.0027 −0.6688 −0.7445 0.0049 0.3458 18.8884

Coef.

0.0322 Females

0.000 0.031 0.175 0.016 0.109 0.945 0.000 0.007 0.000 0.000 0.000

P>t

1.1446 0.0221 0.0639 0.4313 −0.0069 −0.0324 −0.3265 2.2990 0.0041 0.1516 20.0870

0.000 0.611 0.000 0.000 0.400 0.175 0.018 0.000 0.000 0.046 0.000

1.5269 0.0125 0.0171 0.5754 −0.0078 −0.0271 −0.3528 −2.1271 0.0079 0.4177 19.6920

Coef.

q80 Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

Coef.

0.0311 Males

.80 Pseudo R2 = BMI P>t

0.0531 Females

Andhra Pradesh

State

0.000 0.763 0.198 0.000 0.598 0.432 0.043 0.000 0.000 0.000 0.000

P>t 0.8089 0.0257 0.0714 0.3178 −0.0725 0.0232 −0.2582 2.5724 0.0037 0.3314 20.0057

Coef.

0.0229 Males

Telangana

Table 6.12 Factors influencing BMI quantiles (0.80) of males and females in Andhra and Telangana

0.000 0.623 0.000 0.022 0.024 0.526 0.113 0.000 0.000 0.002 0.000

P>t

1.1577 0.2121 0.0129 0.1602 −0.0950 0.0696 −0.6461 −1.3488 0.0057 0.4393 20.1579

Coef.

0.0325 Females

0.000 0.001 0.386 0.261 0.003 0.181 0.001 0.001 0.000 0.000 0.000

P>t

Nutritional status across social groups

171

of irrigated land does not determine adult BMI. However, in all the quantiles except for men in the 20th quantile, land location category turns out significant both for men and women. Compared to land less labour category, others show higher BMI. Years of schooling and age show positive significant influence on body mass index of males and females in Andhra. Age has a positive influence on BMI of both men and women, in the sense, older persons show significantly higher BMI than younger persons both in OLS and all quantiles. Any treatment to make water safe appears to have significant influence on the BMI of both men and women in all the BMI quantiles while the variable is insignificant in the OLS both for men and women. Similarly, participation in a health scheme or insurance for anyone in the household has an insignificant influence at the average level, but positive influence on BMI in all quantiles for men and women except in the lower quantiles (20th and 40th) for women. The variable on health scheme or insurance only indicates the awareness of health issues rather than the actual use of the scheme or health insurance as it need not pertain to the individual whose BMI is measured. An interesting observation is the female to male ratio in the household, which uniformly has a positive relationship with male BMI and a negative relationship female BMI, implying that in a household with more females than males, males have higher BMI and females have lower BMI, indicating gender bias. Gender bias is also clear in the fact that caste has no influence on males but effects females. The vulnerability at the intersection of caste, class and gender is obvious. Intra household discrimination in food distribution in the households in quantity as well as quality assume importance. It is of concern in lower BMI quantiles for some social groups. The study brings out the importance of household amenities such as sanitation, drinking water, cooking fuel etc. will have to be of better standards. Caste becomes insignificant after controlling for household amenities. Hence the discrimination in access to household amenities is the key factor.

6.4.3   Body mass index and social group in Telangana  (ordinary least squares) The Ordinary Least Square results show that caste is not significant in explaining the body mass index of men and women after controlling for other characteristics with one exception. In respect of schedule caste males, BMI was significantly lower compared to other castes but not for females (Table 6.7). The reason is not clear. Poverty levels of urban scheduled caste population is highest in Telangana (Table 6.3). Probably, men may have issues of heavy physical work, and older men may be experiencing discrimination in food intake in the face of poverty among scheduled tribes. Compared to other men from upper castes in Telangana,

172 Swarna Sadasivam Vepa and Rohit Parasar scheduled tribe men have lower BMI, while compared women from other upper castes, scheduled tribe women do not show lower BMI. The simple regression results also show that land ownership has significant negative relationship on women’s BMI but not on men’s BMI. Area of irrigated land owned has no influence on the BMI of both men and women, which is understandable in Telangana, which is predominantly arid, and irrigation essentially depends upon rainfall. Land location category also did not show any influence on body mass index of men and women—viz., those owning land in rural areas, those working in urban areas and those owning urban land do not show higher BMI than the rural land less as in the case of Andhra. It is because wages for manual work in agriculture tend to be very high and cultivation of dry tracts of land is not necessarily more profitable. Poor landowners are on par with land less labour. Further, seasonal migration of members of the household for work in urban Telangana makes the land categories less meaningful. Treatment of drinking water turns out to be significant point to the fact that those who do not treat water show lower BMI. Health insurance schemes do not show any influence on BMI, as most schemes at the average level, do not provide adequate health cover. Education turns out to be an insignificant factor for BMI in Telangana, while it was significant in Andhra. Age is significant for women’s BMI but not for men, meaning older men are not fatter than younger men, while older women are fatter than younger women in Telangana. More numbers of females in the households indicates higher BMI for males showing positive discrimination, but it does not show negative discrimination of low BMI for women as in Andhra (Table 6.7).    mass index and social groups in Telangana  6.4.4  Body (results of quantile regressions) The quantile regressions show that caste is not significant in all the quantiles for men and significant for women. The results for Telangana and Andhra are similar for all the quantiles except for the 80th quantile. For Telangana even in the 80th quantile, social group influences women but not men, once we control for other aspects, upper caste urban older women being obese compared to other caste categories. This is an indication of gender discrimination within the household. Men eat well irrespective which social group they belong to but women of SC, ST and OBC groups are have lower BMI than the women in the “others” category. Overall, the implication is that BMI differential and chronic energy deficiency can be eliminated through measures that improve other aspects of life. Caste differentials disappear with improvement in other factors, though special attention is needed to eliminate vulnerability at caste/ class and gender intersection. The age, as expected, has a positive significant influence on the BMI of both for men and women in the quantile results. Male BMI shows insignificant influence of age, in OLS, though male quantile regressions show positive

Nutritional status across social groups

173

influence. The reason for the insignificance of age for male in OLS and not in quantiles could be due to an improvement of BMI at younger ages for males, having both younger and older men of similar BMI. When we consider various segment and control for other variable the relationship comes out as expected in quantiles. Untreated water shows significant negative influence on BMI in all quantiles for men and women. Holding health insurance has no influence on BMI. Further, a household with more females than males has a positive significant influence on Male BMI in OLS and in all quantile regressions. In OLS, the influence of households with more females is insignificant but all quantile regressions show a negative significant influence on female BMI. What it means is that if the household has more females than males, they feed the male members well at the cost of female members and hence males in these households have better BMI and the women of such households have lower BMI. This is a clear indication of gender discrimination in BMI in the households with more females than males. Some of them could be female-headed households. Normally female-headed households have more females as able bodied males migrate, leaving other males and children behind. In Telangana, household amenities have positive and significant influence on female BMI and not for male, BMI in OLS. All the same, household amenities indicate significant influence on BMI, across all quantiles both for men and women. We may conclude that household amenities influence body mass index and a key factor in improving BMI and eliminating chronic energy deficiency. It is this variable that renders the caste insignificant. What it means is if there is no discrimination in access to household amenities; social group is inconsequential. But if there is a difference in access based on caste, public provisioning should be better geared to eliminate the discrimination. Land ownership in number of acres seems to have an adverse negative influence on women’s BMI not men’s BMI in the OLS in Telangana, but quantile regression analysis shows insignificant influence of land ownership on BMI of men and women except in a few instances, such as negative influence on men and women’s BMI in 80th quantile and men’s BMI in the 60th quantile. Irrigated land ownership has insignificant influence on the BMI of men and women. This adverse impact reflects ownership of arid land and unreliable irrigation sources, which yields very little and make the people who are dependent on such lands vulnerable. Land-based urban rural categories are not significant for BMI of males and females in the OLS. All the same, compared to rural landless, rural landed, urban landless and urban landed households have a significantly higher BMI in all the quantiles for males and females without exception. This clearly reflects the vulnerability of the rural land less dependent on agriculture in Telangana. This has a policy implication for Telangana that it should reduce chronic energy deficiency through shift of both men and women on arid lands to non-agricultural occupations. Overall, the implication is that BMI differential and chronic energy deficiency can be eliminated through measures that improve other aspects of life. Caste

174 Swarna Sadasivam Vepa and Rohit Parasar differentials disappear with improvement in other factors most notable among them are household amenities, land location category. Our results show that equalising social opportunity will equalise caste bias. Though both the states had similar policy atmosphere, the BMI outcomes are different in the two states. For example, nutritional transition of lower social groups having higher BMI is not apparent in Telangana. Age and education have no influence on the BMI of Telangana for men and education has no influence on women’s BMI, but age influenced BMI of women, older women being fatter than younger women in Telangana. In contrast, Andhra shows clear influence of both age and education on BMI. Lower educational levels in Telangana are probably the reason for insignificance (Tables 6.9 to 6.12).

   and social group Andhra and Telangana 6.4.5  Heights Heights of men and women depend up on better nutrition in terms of both quality and adequacy, health care and environmental hygiene in the childhood as well as adolescence. Height of a person also has an intergenerational nutrition improvement effect, meaning, the younger generation is taller than the older generation, within the same household. Catching up in height as adults (above the age of 20) is difficult though not impossible (Deaton 2008). We have only considered adults above the age of 20 years in the DLHS-4 analysis. It means most of the potential is already reached for many men and women. Social group is a significant factor determining the height of a person for both males and females, in both the states as well as the composite state of Andhra Pradesh, showing that past discrimination in nutritional intake and access to amenities matter. Past conditions of morbidity also matter. Compared to those belonging to the other castes, males and females of scheduled tribes, scheduled castes and other backward classes, in both the states, are shorter, even after controlling for all the factors that could influence the nutritional outcome. The reason for social group becoming insignificant for BMI and becoming significant for heights is due to the past conditions of living. Social group is also a significant factor in all the quantile regressions for both the states, without any exception. Men and women of “Other” social group were taller than the men and women belonging scheduled castes, scheduled tribes and other backward classes. Significance of caste did not render the household amenities insignificant in explaining heights of men and women in both the states. Household amenities always turned out significant in all the regressions. It is because the index is constructed in such a way that it adequately represents the quality of the living conditions, which are important for nutrition. Possibly, present conditions are also correlated to past conditions. Genetics also play a part in the height reached by an individual. But the potential to become tall exists for all, with nutrition improvement in the childhood and adolescence, irrespective of social group and genetic background. Tables 6.13 and 6.14 present the results of ordinary least squares and quantile regressions.

0.001 0.000 0.000 0.000 0.000 0.505 0.082 0.004 0.000 0.034 0.005 0.000 0.000

0.0037 −0.0163 −0.0194 −0.0087 0.0012 0.0012 0.0001 0.0011 −0.0076 0.0124 0.0000 0.0036 1.5982

−0.0137 −0.0172 −0.0099 0.0008 −0.0045 0.0001 −0.0005 −0.0004 −0.0058 0.0000 0.0011 1.5234

0.0057 0.000 0.000 0.000 0.000 0.007 0.120 0.384 0.807 0.239 0.000 0.217 0.000

0.0 −0.0272 −0.0309 −0.0118 0.0010 0.0020 −0.0003 0.0012 0.0024 0.0066 0.0000 0.0041 1.5499

0.0081

coefficient

0.000 0.000 0.001 0.000 0.324 0.536 0.074 0.374 0.263 0.454 0.002 0.000

0.000

P>t

adj. R2 = 0.024298

Male

Telangana (new state)

Source: Regression results are based on DLHS-4 unit level data (2012–13); highlight indicates insignificance

Household amenities index Social group others = base Scheduled castes 1 Scheduled tribes 2 Other backward castes 3 Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

coefficient

Variables

P>t

adj. R2 = 0.01445

adj. R2 = 0.019188

Height in meters coefficient

Female

Male

TE: No. of Obs. M = 14625: F = 15834

P>t

Andhra Pradesh (residual)

AP: No. of Obs. M = 19051: F = 22445

Table 6.13 Factors influencing height of males and females in successor states after bifurcation (OLS)

−0.0204 −0.0202 −0.0109 0.0008 0.0022 0.0000 0.0000 0.0103 −0.0127 0.0000 0.0050 1.4934

0.0051

coefficient

0.000 0.000 0.001 0.000 0.219 0.939 0.939 0.000 0.020 0.010 0.000 0.000

0.000

P>t

adj. R2 = 0.018872

Female

176 Swarna Sadasivam Vepa and Rohit Parasar Table 6.14 Factors influencing height in the composite state before bifurcation (OLS) No. of Obs. M = 33676: F = 38280

Males

Females

Height in Meters

Adj R2 = 0.025519

Adj R2 = 0.01580

Variables

Coefficient

Coefficient

Household amenities index Social group: others = base Scheduled castes 1 Scheduled tribes 2 Other backward castes 3 Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location Category State: AP= 28 TE= 36 _cons

P>t

P>t

0.0057

0.000

0.0053

0.000

−0.0201 −0.0236 −0.0088 0.0012 0.0013 0.0001 0.0009 −0.0034 0.0096 0.0000 0.0038 −0.0018 1.6341

0.000 0.000 0.000 0.000 0.337 0.095 0.002 0.037 0.020 0.135 0.000 0.000 0.000

−0.0163 −0.0178 −0.0097 0.0008 −0.0015 0.0001 −0.0004 0.0044 −0.0091 −0.000020 0.0028 0.0002 1.5035

0.000 0.000 0.000 0.000 0.225 0.129 0.311 0.002 0.013 0.000 0.000 0.157 0.000

Source: Regression results are based on DLHS-4 (2012–13); highlight indicates insignificance

Tables 6.15 to 6.18 present the results of simultaneous quantile regressions for Andhra and Telangana for men and women. 6.4.6  Factors    influencing heights of men and women Age has significant negative influence on the heights of both males and females in Andhra. Younger males and females were taller than the older males and females in Andhra. In OLS and in all the quantiles both for men and women, age is significant and negative, as expected. It indicates intergenerational improvement. Age is not related to heights of men in Telangana in OLS and all the quantiles, but age is negatively related to heights for women in OLS and the lower quantiles. Younger men are not taller than older men in Telangana, but younger women are taller than older women. Education has a positive impact on the heights of both males and females in both the states. OLS as well as all the quantile regressions show positive significant relationship between years of schooling and the height in meters in both the states. As pointed out by many, height can represent a stock of human capital. Cognitive skills improve with better nutrition and they also improve the height of a person.

0.005 0.000 0.000 0.162 0.321 0.013 0.001 0.101 0.006 0.000 0.000

Source: Based on DLHS-4 unit level data; highlight indicates insignificance

Household amenities index 0.0031 Social group 0.0030 Years of schooling 0.0011 Health scheme/insurance 0.0031 Land owned by the household 0.0001 Irrigated land owned by household 0.0007 Any treatment of drinking water −0.0073 Female/male ratio of the household 0.0092 Age in months 0.0000 Land-location category 0.0051 _cons 1.5245 AP No. of Obs. Male = 15593: Female =15666 TE No. of Obs. Male = 13132: Female = 13270

P>t

Coef.

0.0176 Males

Telangana

0.0064 0.000 0.00708 0.0029 0.000 0.00703 0.0010 0.000 0.00107 −0.0011 0.500 0.00750 0.0000 0.849 −0.00069 0.0000 0.902 0.00240 0.0014 0.471 0.00023 −0.0056 0.261 0.01195 0.0000 0.000 0.00000 0.0017 0.124 0.00403 1.4536 0.000 1.44795 for all quantiles from 0.20 to 0.80 for all quantiles from 0.20 to 0.80

Coef.

q20

Coef.

0.0119 Males

.20 Pseudo R2 = BMI P>t

0.0154 Females

Andhra Pradesh

State

Table 6.15 Factors influencing height quantiles (0.20) of males and females in Andhra Pradesh and Telangana

0.000 0.000 0.000 0.002 0.282 0.001 0.937 0.106 0.701 0.010 0.000

P>t

0.0064 0.0042 0.0011 0.0033 −0.0001 0.0006 0.0069 −0.0074 0.0000 0.0068 1.4104

Coef.

0.0213 Females

0.000 0.000 0.000 0.146 0.864 0.449 0.001 0.172 0.000 0.000 0.000

P>t

0.005 0.000 0.000 0.006 0.780 0.035 0.036 0.162 0.001 0.000 0.000

Source: Based on DLHS-4 unit level data; highlight indicates insignificance

Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

0.0025 0.0024 0.0012 0.0040 0.0001 0.0011 −0.0040 0.0052 0.0000 0.0032 1.5681

0.0051 0.0021 0.0009 −0.0004 0.0000 0.0002 0.0001 −0.0059 0.0000 0.0017 1.4915

Coef.

q40

Coef.

0.0130 Males

.40 Pseudo R2 = BMI P>t

0.0179 Females

Andhra Pradesh

State

0.000 0.000 0.000 0.766 0.900 0.398 0.947 0.077 0.000 0.031 0.000

P>t 0.0060 0.0060 0.0012 0.0031 0.0000 0.0008 0.0013 0.0002 0.0000 0.0045 1.5149

Coef.

0.0173 Males

Telangana

0.000 0.000 0.000 0.114 0.963 0.134 0.612 0.980 0.758 0.001 0.000

P>t

Table 6.16 Factors influencing height quantiles (0.40) of males and females in Andhra Pradesh and Telangana

0.0042 0.0051 0.0008 0.0014 0.0000 0.0007 0.0097 −0.0005 0.0000 0.0031 1.4539

Coef.

0.0143 Females

0.000 0.000 0.000 0.287 0.981 0.215 0.000 0.886 0.003 0.000 0.000

P>t

0.000 0.000 0.000 0.000 0.556 0.018 0.007 0.108 0.000 0.000 0.000

0.00641 0.00295 0.00055 −0.00041 0.00007 −0.00011 −0.00054 −0.00039 −0.00002 0.00106 1.51891

Source: Based on DLHS-4 unit level data; highlight indicates insignificance

Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

0.0038 0.0040 0.0017 0.0058 0.0002 0.0011 −0.0056 0.0063 0.0000 0.0042 1.5903

Coef.

q60

Coef.

0.0246 Males

.60 Pseudo R = BMI P>t

0.0132 Females

Andhra Pradesh

2

State

0.000 0.000 0.000 0.767 0.616 0.765 0.694 0.906 0.000 0.251 0.000

P>t 0.0078 0.0069 0.0011 0.0038 −0.0001 0.0013 −0.0022 0.0027 0.0000 0.0043 1.5491

Coef.

0.0252 Males

Telangana

0.000 0.000 0.000 0.051 0.799 0.032 0.454 0.677 0.685 0.000 0.000

P>t

Table 6.17 Factors influencing height quantiles (0.60) of males and females in Andhra Pradesh and Telangana

0.0054 0.0055 0.0008 −0.0023 0.0000 0.0006 0.0098 −0.0093 0.0000 0.0042 1.4872

Coef.

0.0151 Females

0.000 0.000 0.000 0.153 0.962 0.394 0.000 0.018 0.615 0.000 0.000

P>t

180 Swarna Sadasivam Vepa and Rohit Parasar None of the regressions including quantiles for men and women indicate any influence of landownership on their heights. Interestingly, ownership of irrigated agricultural land had positive influence on men’s heights in OLS, in both the states and the composite state, but not on women’s heights. All quantiles except the top one show positive significant influence of irrigated land on men’s heights but not on female heights in Andhra Pradesh. For Telangana, in the quantile regressions, 20th quantile and 60th quantile show positive impact of irrigated land on heights of men but on the heights of women in Telangana. This aspect needs more research, if the ownership of irrigated land remained with certain communities. Land location variable with landless rural household as the base category has significant influence on the heights of men and women in OLS as well as most quantiles in both the states and in the composite state. It means compared to the male and female members of the household belonging to land less rural category, men and women of other categories are taller. More than the land area owned, and irrigated area possessed, probably, land location category distinguishes the agrarian structure better. Ratio of females to males in the household shows significant positive influence on men’s heights, but there was no influence on female heights. The quantile regressions indicate insignificance of this variable in explaining heights.

6.5 Conclusions and policy implications The well-being in terms of BMI and heights of men and women above the age of 20 years using DLHS-4 data for Andhra Pradesh and Telangana brings out some interesting issues. Social group has no significant influence on body mass index for men and women but has a significant influence on the heights after controlling for other individual and household characteristics. Compared to the base category of other castes, scheduled castes, scheduled tribes and other backward castes show significantly lower heights in both the states. Household with more females than males influencing men’s BMI positively and women’s BMI negatively is another case of gender bias. Such household could be female headed and likely to be poorer. This calls for special program of nutrition awareness and gender sensitivity and equity in food intake for women in SC, ST and OBCs, and especially in female headed and households with more females than males. Age turns out to be a significant factor in explaining the height, younger men and women being taller than older men and women. It indicates intergenerational catch up in both the states in all the quantiles. Compared to OC men and women, SC, ST and OBC, men and women are shorter. Household amenities improve the BMI and heights of both men and women in both the states. Household amenities Index turns out to be significant in all the regression, both in quantile and OLS, for both men and women. It is because, the index not only takes into consideration the presence and absence of an amenity but also considers the type of facility available, the index gradation indirectly includes the quality of the amenity being used. Scheduled caste and scheduled tribe and other backward classes will reach the BMI levels of other

0.000 0.000 0.000 0.013 0.750 0.101 0.002 0.259 0.000 0.000 0.000

Source: Based on DLHS-4 unit level data; highlight indicates insignificance

Household amenities index Social group Years of schooling Health scheme/insurance Land owned by the household Irrigated land owned by household Any treatment of drinking water Female/male ratio of the household Age in months Land-location category _cons

0.0044 0.0036 0.0015 0.0046 0.0001 0.0008 −0.0057 0.0053 0.0000 0.0039 1.6412

0.0010 0.0004 0.0001 0.0016 0.0003 0.0004 0.0014 0.0039 0.0000 0.0010 0.0054

Coef.

q80

Coef.

0.0239 Males

.80 Pseudo R = BMI P>t

0.0099 Females

Andhra Pradesh

2

State

0.000 0.000 0.074 0.738 0.974 0.622 0.260 0.727 0.028 0.067 0.000

P>t 0.0089 0.0070 0.0012 −0.0005 0.0013 −0.0003 −0.0009 0.0000 0.0000 0.0007 1.6103

Coef.

0.0210 Males

Telangana

Table 6.18 Factors influencing height quantiles (0.80) of males and females in Andhra Pradesh and Telangana

0.000 0.000 0.000 0.796 0.065 0.745 0.678 0.996 0.106 0.535 0.000

P>t

0.0057 0.0041 0.0008 −0.0015 0.0004 −0.0005 0.0131 −0.0090 0.0000 0.0027 1.5264

Coef.

0.0134 Females

0.000 0.000 0.000 0.437 0.502 0.433 0.000 0.062 0.911 0.017 0.000

P>t

182 Swarna Sadasivam Vepa and Rohit Parasar castes if only they enjoy the same quality of household amenities, education and ownership of irrigated land. However, household amenities of today cannot equalise the differences of heights, though they substantially contribute to the heights of men and women in future. Hence the policy implication is that all social groups and especially the scheduled and tribes and scheduled caste should have household amenities such as uninterrupted clean piped water supply, flush toilets, drainage connections, electricity, permanent housing and clean cooking fuel. The discrimination in access to these amenities result differences in BMI and heights. It is easier to equalise body mass index across social groups, but it is harder to equalise heights across the social groups, in the short run. Continuous and dedicated efforts in improving household amenities alone can make a difference to the heights in the next generation. Quality of food intake, rich in proteins in childhood and adolescence, improves the height of an individual. Food intake in excess, after the age of 20, leads to overweight, which seem to be the issue with men and women especially in urban areas. While chronic energy deficiency is falling, obesity is on the increase among the adult population above the age of 20 years. At present there is no policy guidance or clarity on addressing the issue. Improvement in educational levels and awareness creation would help.

6.6 Body mass index, heights and social group in the age cohort of 15–49 (NFHS-4) Generally, the age group of 15–49, rather than the age group of 20 years and above assumes importance. It is because this group has adolescents, women prior to conception, expectant and lactating mothers. The nutrition improvements achieved by this age group are transmitted to the next generation. Based on the recent NHHS-4 data, this subsection analyses the influence of social group on BMI and heights of this age group after controlling for household and individual characteristics. At the average level, the body mass index of scheduled tribe women and men in the age group of 15–49 are the lowest compared to the other social groups in both the states. As far as the heights are concerned the scheduled castes men of Telangana have lower average heights compared to scheduled tribe men. In Andhra scheduled caste men are the shortest. Overall, it appears that scheduled caste men and women have better average nutrition out comes in Andhra, compared to the scheduled caste men of Telangana. However, the body mass index and heights of scheduled castes men and women are lower than the average for all social groups. The men and women of the other castes are the tallest and fatter than the remaining groups on the average, in both the states. The problem with NFHS-4 appears to be a large percentage of population not reporting caste of any category and they are kept separate. In Andhra Pradesh and Telangana, scheduled caste, scheduled tribe and other backward class categorisation is officially assigned across the religious groups. Figures 6.1 to 6.8 give the average body mass index and heights across social groups.

25.0

24.4

Andhra Pradesh All Group Average

24.5 24.0 23.5

23.0

23.0 22.5 22.0

23.0

23.1

22.7

21.3

21.5 21.0 20.5 20.0 19.5

Scheduled Caste Scheduled Tribe Other Backward Class

Others

Don’t Know

Figure 6.1 Mean BMI of women across social groups in Andhra Pradesh Source: NFHS-4

25.0

Andhra Pradesh All Group Average

24.4 23.9

24.0 23.2

23.2

22.7

23.0 22.0

21.1

21.0 20.0 19.0

Scheduled Caste Scheduled Tribe

Other Backward Class

Others

Don't Know

Figure 6.2 Mean BMI of men across social groups in Andhra Pradesh Source: NFHS-4

24.5

Telangana All Group Average

24.0

23.9

23.5 23.0

22.6

22.5

22.2

22.0

22.2

21.6

21.5

21.2

21.0 20.5 20.0 19.5

Scheduled Caste Scheduled Tribe Other Backward Class

Others

Figure 6.3 Mean BMI of women across social groups in Telangana Source: NFHS-4

Don't Know

24.0

Telangana All Group Average

23.5

23.4

23.0 22.5 22.0

22.0

21.7

21.5

22.0

20.9

21.0

20.4

20.5 20.0 19.5 19.0 18.5

Scheduled Caste Scheduled Tribe Other Backward Class

Others

Don’t Know

Figure 6.4 Mean BMI of men across social groups in Telangana Source: NFHS-4 1535 1525

1520.8

1520 1515 1510

1531.6

Andhra Pradesh All Group Average

1530

1519.8 1509.7

1505

1509.1 1500.3

1500 1495 1490 1485 1480

Scheduled Caste Scheduled Tribe Other Backward Class

Others

Don’t Know

Figure 6.5 Mean heights of women across social groups in Andhra Pradesh Source: NFHS-4 1670

Andhra Pradesh All Group Average

1660 1650

1641.3

1640 1630

1662.6

1642.0

1625.7 1618.6

1617.4

1620 1610 1600 1590

Scheduled Caste Scheduled Tribe

Other Backward Class

Others

Figure 6.6 Mean heights of men across social groups in Andhra Pradesh Source: NFHS-4

Don’t Know

Nutritional status across social groups 1535 1525

1520.8

1520 1515 1510

1531.6

Telangana All Group Average

1530

185

1519.8 1509.7

1505

1509.1 1500.3

1500 1495 1490 1485 1480

Scheduled Caste Scheduled Tribe

Other Backward Class

Others

Don’t Know

Figure 6.7 Mean heights of women across social groups in Telangana Source: NFHS-4

1670 1660

1651.5

1650

1620

1648.0 1646.1

1636.2

1640 1630

1664.7

Telangana All Group Average

1620.8

1610 1600 1590

Scheduled Caste Scheduled Tribe

Other Backward Class

Others

Don’t Know

Figure 6.8 Mean heights of men across social groups in Telangana Source: NFHS-4

6.6.1   Body mass index and social group in the age cohort  of 15–49 (NFHS-4) According to NFHS-4, chronic energy deficiency prevalence is high in rural Telangana (29%) compared to Andhra Pradesh (20.3%). For rural men it stands at 24.6% for Telangana and 16.5% for Andhra. Basically, it reflects the better nutritional

186 Swarna Sadasivam Vepa and Rohit Parasar status for rural Andhra compared to rural Telangana, which is having rain fed agriculture in more areas than Andhra. Obese women and men in urban Andhra are as high as 45.6 and 44.4% respectively. Telangana has 40% obese women and 31.7% obese men in urban Telangana. Tables 6.19 to 6.22 give the results of factors influencing body mass index and heights of women and men in Andhra Pradesh and Telangana.

Table 6.19 BMI and social groups in Telangana Female BMI

Male BMI

R2 = 0.1306

Number of obs = 7137

R2 = 0.0980

Number of obs = 1056

Coef.

P>t

Coef.

P>t

−69.3*** −84.2*** −76.5*** −75.7

0.002 0.001 0.000 0.320

−29.4 −59.0 −42.1 −203.2*

0.550 0.254 0.305 0.072

29.7 111.4*** 236.2*** 411.2***

0.124 0.000 0.000 0.000

46.5 122.2** 157.7** 331.7***

0.322 0.031 0.014 0.000

0.082 0.000 0.000 0.000

26.8 −71.4** −52.5 −390.5***

0.558 0.026 0.208 0.000

Social groups (base caste = others) SC ST OBC Don’t know Wealth index Poorer Middle Richer Richest

Education (base no education/pre-school) Primary Secondary Higher Don’t know

−31.9* −159.2*** −229.2*** −283.9***

Covered by health insurance/scheme Yes Don’t know Ownership of house

−42.7*** −369.0** −35.1***

0.001 0.031 0.001

−34.9

0.215

−28.4

0.283

−14.7 −303.4*** 46.8*** 2109.4***

0.483 0.003 0.000 0.000

−46.2

0.310

Has bank account Yes Don’t know Living Standard Score Constant

32.3 2107.1***

Source: Authors’ calculation based on NFHS-4 Note: Age Group =15–49 Years; Significance level: *>=90, **>=95%, *** >=99%

0.048 0.000

Nutritional status across social groups

187

Table 6.20 Height and social groups in Telangana Female Height

Male Height

R2 = 0.0674

Number of obs = 7145

R2 = 0.1281

Number of obs = 1056

Coef.

P>t

Coef.

P>t

Social groups (base group) = others) SC ST OBC Don’t know

−23.2*** −7.8** −6.2*** −13.8*

0.000 0.020 0.009 0.099

−20.5** 2.7 1.7 3.0

0.010 0.751 0.787 0.876

8.5** 12.0*** 18.2*** 27.0***

0.012 0.001 0.000 0.000

13.8 26.8*** 38.0*** 50.2***

0.117 0.007 0.000 0.000

0.999 0.000 0.000 0.266

8.0 22.8*** 28.6*** −21.3**

0.340 0.000 0.000 0.013

Wealth index Poorer Middle Richer Richest

Education (base no education/pre-school) Primary Secondary Higher Don’t know

0.0 12.4*** 20.8*** 16.0

Covered by health insurance/scheme Yes Don’t know Ownership of house

−1.6 0.6 −1.3

0.339 0.981 0.361

−2.4

0.633

0.6

0.882

−3.7 −14.7 0.4 1509.2***

0.201 0.447 0.650 0.000

−4.4

0.497

2.5 1597.9***

0.350 0.000

Has bank account Yes Don’t know Living standard score Constant

Source: Authors’ calculation based on NFHS-4 Note: Age Group =15–49 Years; Significance level: *>=90, **>=95%, *** >=99%

In Andhra Pradesh, in the age group of 15–49, BMI of scheduled caste women is not significantly different from other caste women after controlling for wealth index, living standard index and education, but both the men and women of all the other social groups have significantly lower BMI than the other castes. It is in contrast, to the findings of the women and men in the age group of 20 and above

188 Swarna Sadasivam Vepa and Rohit Parasar Table 6.21 BMI and social group in Andhra Pradesh Female BMI

Male BMI

R2 = 0.1256

Number of obs = 9782

R2 = 0.1039

Number of obs = 1436

Coef.

P>t

Coef.

P>t

−15.6 −83.3*** −53.7*** −55.6

0.328 0.000 0.000 0.487

−76.7* −128.1** −55.2* −47.2

0.061 0.011 0.094 0.689

85.4*** 138.5*** 303.2*** 470.3***

0.000 0.000 0.000 0.000

Social groups (base caste = others) SC ST OBC Don’t know Wealth index Poorer Middle Richer Richest

−21.9 77.2 189.9*** 274.9***

0.673 0.176 0.003 0.000

Education (base no education/pre-school) Primary Secondary Higher Don’t know

44.4*** −86.7*** −131.2*** −77.8

0.002 0.000 0.000 0.530

−12.1 −23.5 23.1 −138.6**

0.745 0.464 0.573 0.012

0.000 0.753 0.000

−35.9 −127.9 9.2

0.217 0.395 0.754

−5.8

0.819

35.8*** 2113.2***

0.000 0.000

118.7** −141.4** 31.3** 2038.8***

0.017 0.037 0.036 0.000

Covered by health insurance/scheme Yes Don’t know Ownership of house

−45.2*** −28.5 −50.5***

Has bank account Yes Don’t know Living standard score Constant

Source: Authors’ calculation based on NFHS-4 Note: Age Group =15–49 Years; Significance level: *>=90, **>=95%, *** >=99%

as per the DLHS-4 data, where social group has no influence on the BMI of men and women, after controlling for other factors. The impact of caste/ social group on BMI in the age group of 15–49, after controlling for other factors (living standard index, wealth index and education) is significantly lower for scheduled tribe women, scheduled caste women and the other backward caste women, compared to other castes women, in Telangana.

Nutritional status across social groups

189

Table 6.22 Height and social group in Andhra Pradesh Female height

Male height

R2 = 0.0438

Number of obs = 9786

R2 = 0.0825

Number of obs = 1436

Coef.

P>t

Coef.

P>t

Social groups (base others) SC ST OBC Don’t know

−15.9*** −21.0*** −6.7*** −15.5

0.000 0.000 0.000 0.106

−27.7*** −29.4*** −13.7*** −50.8**

0.000 0.001 0.004 0.047

8.2** 9.9*** 12.2*** 15.0***

0.010 0.002 0.001 0.000

−13.2 −15.0 −5.9 15.0

0.238 0.187 0.625 0.256

1.2 7.8*** 17.7*** 1.3

0.476 0.000 0.000 0.956

−6.9 4.7 10.8* −27.9***

0.276 0.340 0.082 0.003

−3.3** 6.2 6.7***

0.018 0.519 0.000

−7.0 −12.5 4.8

0.104 0.237 0.311

−1.3

0.663

1.7** 1503.8***

0.014 0.000

Wealth index Poorer Middle Richer Richest

Education (base no education/pre-School) Primary Secondary Higher Don’t know Covered by health insurance/scheme Yes Don’t know Ownership of house Has bank account Yes Don’t know Living standard score Constant

0.6 10.3 2.7 1649.5***

0.946 0.468 0.223 0.000

Source: Authors’ calculation based on NFHS-4 Note: Age Group =15–49 Years; Significance level: *>=90, **>=95%, *** >=99%

However, for men in Telangana, caste has no influence on the BMI, after controlling for other factors. Wealth index and living standard index show positive influence on BMI of men and women, both in Andhra Pradesh and Telangana. Wealth index has influence on body mass index of men only in the rich and richest categories in Andhra, compared to the poor for men’s BMI.

190 Swarna Sadasivam Vepa and Rohit Parasar In Andhra, education has negative influence on the BMI of women after secondary education and insignificant for men’s BMI. In Telangana, education has negative influence on women’s BMI, and insignificant for men except for a negative influence at the secondary level. This is an unexpected response. Education is proxying for the age, in this age cohort of 15–49. Younger women have higher education achievements compared to older women, who are not educated and tend to be obese. The regressions do not include age. Health insurance or coverage by any health scheme is again negative for women and insignificant for men both in Andhra and Telangana. This variable is capturing the poverty aspect. Health insurance or health scheme cover is given to the poorer sections of the society by the government. Non-poor in these states go to private hospitals, which mostly deal with cash. Many households do not take an insurance as reimbursement process in India is prolonged process, involving documentation. It has negative influence on women’s BMI and insignificant for men’s BMI in both the states. 6.6.2

Height and social group in the age cohort of 15–49 (NFHS-4)

In Andhra Pradesh social group influences the heights of women and men in the age group of 15–49. Only women’s height but not men’s heights are influenced by social group in Telangana after controlling for wealth Index and living standard index and education level. The reasons are not clear. In the age group 20 years above in the analysis of DLHS-4 data the heights of both men and women are influenced by social group in Telangana. Social group has no influence on both body mass index and heights of men and women in Telangana, in contrast to Andhra Pradesh in the NFHS-4 data for the age group of 15–49. In the case of women of Telangana, the differences across social groups are significant. The conclusions one could draw is that there is gender discrimination in food intake over time in the social groups which are poorer. Boys and men get enough nutrition and gain weight and height, but girls and women of the poorer social groups remain shorter. Another reason for the insignificance of social group for men’s height in NFHS-4 data could be the non-reporting of caste by a large percentage of the sample. While more research is required into this aspect. there are reasons to believe that the upper caste (that constitute less than 10% of the population but rich) may have been missed by the sample. In Andhra they constitute more than 22% of the population and may have been well represented in the sample. Living standard score and wealth Index had no influence on men’s heights in Andhra but influence women’s heights, poorer women and women in households with low standard of living index are shorter. In Telangana, wealth index influences the heights of both women and men. Lower the wealth index, shorter the men and women. Living standard score has no influence on the heights of men and women in Telangana, unlike the amenities index. This is

Nutritional status across social groups

191

probably due to the inability of the living index score to capture the quality of life. Education above secondary level has positive influence on the heights of women and men in Telangana and heights of women in Andhra. Men’s heights in Andhra are influenced only at the level of higher education.

6.7

Conclusion

NFHS-4 data for the age group of 15–49 indicates mixed influence of caste on body mass index and heights. Caste did not influence male height and body mass index but made a difference to women’s heights and body mass index, pointing to gender discrimination in Telangana. In Andhra, caste has significant influence on heights of men and women and body mass index of women. Men’s body mass index is not influenced by social group in Andhra. Education influenced heights positively both in Andhra and Telangana. Two aspects are striking. Despite similar policy atmosphere for decades in both the states, the contrast in nutritional outcomes and the differential influence of caste on heights and body mass index is obvious. The regional factors appear to have strong influence on heights and body mass index of men and women. Second government policies and general prosperity over decades did not benefit women as much as it did to men.

Notes 1 The authors gratefully acknowledge the opportunity given by Centre for Economic and Social Studies, Hyderabad, to undertake this study. 2 Population of upper caste plus backward castes was the residual population after deducting the Muslim and other religious groups including those without religion and the SC and ST population from the total population. The BC proportion of NSS was applied to census population to derive BC population. The estimated BC population was deducted from upper caste plus backward castes population to get the upper caste population in the state. 3 Only in case of Khammam district of Telangana, which was part of East Godavari district of Andhra Pradesh till 1959, bifurcation in 2014, facilitated transfer of seven blocks (Mandals) back to East Godavari and West Godavari districts of Andhra Pradesh, as they will be submerged during the construction of mega irrigation project on the river Godavari in Andhra Pradesh. We have essentially ignored since these blocks are very small in size, though the impact of displacement of tribal population will be very large. The extent of displacement is not known, though lands are officially acquired, but not yet submerged as the project is delayed. 4 The insignificance of caste does not change even if the household and individual characteristics considered are changed.

References Balagopal, K. (2005). Dalits among Dalits, the Ghosts Resurface., Economic and Political Weekly, 40 (29).

192 Swarna Sadasivam Vepa and Rohit Parasar Banerjee, Rahul (2007, September). “Adivasis and Unjust Laws.” Economic and Political Weekly. Bhukya, Bhangya and Ramaswamy, Gita (2002). “Lambadas: Changing Cultural Patterns.” Economic and Political Weekly, 37(16), April 20. Centre for Economic and Social Studies, Hyderabad (CESS) (2017a). MDG Millinnuium Development Goals Report for Andhra Pradesh (Unpublished). Centre for Economic and Social Studies, Hyderabad (CESS) (2017b). MDG (Millinnium Development Goals Report for Telangana (Unpublished). Centre for Economic and Social Studies, Hyderabad (CESS) (2017c). Wellbeing Gaps across Social Groups in Andhra Pradesh: A Status Report of the Scheduled Communities (Mimeograph Report). Centre for Economic and Social Studies, Hyderabad (CESS) (2017d). Wellbeing Gaps across Social Groups in Telangana: A Status Report of the Scheduled Communities (Mimeograph Report). Chancel, Lucas, and Thomas Piketty. “Indian Income Inequality, 1922–2015: From British Raj to Billionaire Raj?” Review of Income and Wealth, 65 (2019): S33-S62. Clarke, Philippa, O’Malley, Patrick M., Johnston, Lloyd D. and John E. Schulenberg (2009). “Social Disparities in BMI Trajectories Across Adulthood by Gender, Race/ Ethnicity and Lifetime Socio-Economic Position: 1986–2004.” International Journal of Epidemiology 38(2), 499–509. Deaton, Angus (2007, August 14). “Height, Health, and Development.” Proceedings of the National Academy of Sciences of the United States of America, 104(33), 13232–13237, Published by: National Academy of Sciences. Deaton, Angus (2008, May). “Height, Health and Inequality, the Distribution of Adult Heights in India.” American Economic Review, 98(2), 468–474. Deaton, Angus (2010, Summer). “Understanding the Mechanisms of Economic Development.” Journal of Economic Perspectives, 24(3). Desai, Sonalde and Dubey, Amaresh (2012). “Caste in 21st Century India: Competing Narratives.” Economic and Political Weekly, 46(11), 40. Deshpande, Ashwini (2010). “Merits, Mobility, and Modernism: Caste Discrimination in Contemporary Indian Labour Markets.” Indian Journal of Labour Economics, 53(3). Deshpande, Ashwini (2011). The Grammar of Caste: Economic Discrimination in Contemporary India, New Delhi: Oxford University Press. Deshpande, Ashwini and Sharma, Smriti (2013). “Entrepreneurship or Survival? Caste and Gender in Small Business in India.” Delhi School of Economics, Working Paper No. 228, Centre for Development Economics, Department of Economics, Delhi university, Delhi. Dreze, Jean and Sen, Amartya (1999). India, Economic Development and Social Opportunity, New Delhi, India: Oxford University Press. Gouda, Jitendra, and Prusty, Ranjan Kumar (2014). “Overweight and Obesity Among Women by Economic Stratum in Urban India.” Journal of Health, Population, and Nutrition, 32(1), 79. Government of Andhra Pradesh, Planning Department, Socioeconomic Survey Report 2015–2016. https://www.ap.gov.in/wp-content/uploads/2016/03/AP-Socio_Eco.-201516_Final.pdf Government of Telangana, Planning Department, Socioeconomic Outlook Report 2016. www.telangana.gov.in/PDFDocuments/Socio-Economic-Outlook-2016.pdf Guha, Ramachandra (2007). “Adivasis, Naxalites and Indian Democracy.” Economic and Political Weekly, 42(32), 3305–3312.

Nutritional status across social groups

193

Ramuth, Harris, Hunma, Sadhna, Ramessur, Vinaysing, Ramuth, Magalutcheemee, Monnard, Cathriona, Montani, Jean-Pierre, Schutz, Yves, Joonas, Noorjehan, and Dulloo, Abdul G. (2020). “Body composition-derived BMI cut-offs for overweight and obesity in ethnic Indian and Creole urban children of Mauritius.” British Journal of Nutrition, 1–33. Jose, Sunny (2011, July 16–22). “Adult Under-Nutrition in India: Is There a Huge Gender Gap?” Economic and Political Weekly, 46(29), 95–102. Jose, Sunny, and Kannan Navaneetham. “A factsheet on women’s malnutrition in India.” Economic and Political Weekly, (2008): 61–67. Kannabiran, Kalpana, Ramdas, Sagari R., Madhusudan, N., Ashalatha, S. and Pavan Kumar, M. (2010, March 27). “On the Telangana Trail.” Economic and Political Weekly, 45(13). Koenker, Roger (2005). Quantile Regression, Econometrics Society Monograph, Cambridge: Cambridge University Press. Kumar, Arjun (2014). “Access to Basic Amenities: Aspects of Caste, Ethnicity and Poverty in Rural and Urban India—1993 to 2008–2009.” Journal of Land and Rural Studies, 2(1), 127–148. Maity, Bipasha (2017). “Comparing Health Outcomes across Scheduled Tribes and Castes in India.” World Development, 96, 163–181. Murdia, Ratna (1979, August 9). “Land Allotment and Land Alienation: Policies and Programmes for Scheduled Tribes and Castes.” Economic and Political Weekly. Neuman, Melissa, Finlay, Jocelyn E., Smith, George Davey, and Subramanian, S. V. (2011). “The Poor Stay Thinner: Stable Socioeconomic Gradients in BMI among Women in Lower-and Middle-Income Countries.” The American Journal of Clinical Nutrition, 94(5), 1348–1357. Radhakrishna, Rokkam (2015, October 10). “Wellbeing, Inequality, Poverty and Pathways Out of Poverty.” Economic and Political Weekly, 50(41). Ramaiah, Avatthi (2015). Health Status of Dalits in India, Economic and Political Weekly, 50(43), 70–74. Sambaiah, Gundimeda (2009). “Prajarajyam Party and Caste Politics in Andhra Pradesh.” Economic and Political Weekly, 44(21). Sarkar, Sandip, Mishra, Sunil, Dayal, Harishwar and Nathan, Dev (2006, November 18–24). “Development and Deprivation of Scheduled Tribes.” Economic and Political Weekly, 41(46), 4824–4827. Sharma, H. R. (2007, October 13–19). “Land Distribution and Tenancy Among Social Groups.” Economic and Political Weekly, 42(41). Srinivasa Reddy, G. and James, P. A. (1979, June 30). “Commisioners for Scheduled castes and Scheduled Tribes.” Economic and Political Weekly, 14(26). Stiglitz, Joseph E., Sen, Amartya, and Fitoussi, Jean-Paul (2009). “Report by the commission on the measurement of economic performance and social progress.” www.stiglitzsen-fitoussi.fr/en/index.htm. Subramanian, S. V., Perkins, Jessica M. and Khan, Kashif T. (2009). “Do Burdens of Underweight and Overweight Coexist among Lower Socioeconomic Groups in India?” The American Journal of Clinical Nutrition, 90(2), 369–376. Subramanian, S. V., Corsi, Daniel J., Subramanyam, Malavika A., and Smith, George Davey (2013). “Jumping the Gun: The Problematic Discourse on Socioeconomic Status and Cardiovascular Health in India.” International Journal of Epidemiology 42(5), 1410–1426. Sundaram, K. and Tendulkar, Suresh D. (2003, December 13–19). “Poverty Among Social and Economic Groups in India in 1990s.” Economic and Political Weekly, 38(50), 5263–5276.

194 Swarna Sadasivam Vepa and Rohit Parasar Thorat, Amit (2010, December 18–24). “Ethnicity, Caste and Religion: Implications for Poverty Outcomes.” Economic and Political Weekly, 45(51), 47–53. Xaxa, Virginius (2005, March 26). “Politics of Religion Language and Identity.” The Tribes of India, Economic and Political Weekly, 40(13). Xaxa, Virginius (2016). “Tribes and Indian National Identity: Location of Exclusion and Marginality.” The Brown Journal of World Affairs, 23, 223.

Appendix 6A Construction of household amenities index

This index has four components that indicate the quality of life. 1 2 3 4

Sanitation (type of toilet used) Drinking water Cooking fuel Housing

The index varies between one and five. It is a continuous variable, where the household may have any fraction as the index. This note elaborates the procedure adopted for construction of index. The note also presents the original classification of amenities available in the DLHS-4 data. The original categories are collapsed into fewer categories for the purpose of index.

Table 6A.1 Toilet index (index varies from one to five) Toilet facility mainly used (DLHS-4) Freq. Flush-to-piped-sewer-system Flush-to-septic-tank Flush-to-pit-latrine Flush-to-somewhere Flush-don’t know-where Pit-ventilated-improved-vip-bio-gas-latrines Pit-latrine-with-slab Pit-latrine-without-slab-open-pit Twin-pit-composting-toilet Dry-service-latrine No-facility-uses-open-space-or-field-jungle Other

10,652 36,761 8,625 818 582 2,240 22,671 3,034 122 435 46,933 107

(%)

Cum.

8.01 27.64 6.49 0.62 0.44 1.68 17.05 2.28 0.09 0.33 35.29 0.08

8.01 35.65 42.14 42.76 43.19 44.88 61.93 64.21 64.3 64.63 99.92 100

196 Swarna Sadasivam Vepa and Rohit Parasar The above categories are recoded as follows between one and three. 1 2 3

Any flush toilet (first five categories) = 3 Pit toilets (six–nine categories) = 2 Dry toilets and no toilets = 1 Share this toilet facility with other households Yes-community-toilet Shared-toilet No

Freq.

(%)

Cum.

5,675 8,208 72,419

6.58 9.51 83.91

6.58 16.09 100

By combining the toilet type and the own or share category, the toilets are graded into five categories as follows: Own toilet =2 Shared toilet =1 Index of flush toilet + owned toilets = 3 + 2 = 5 (top of the scale) Index of flush toilet + shared toilet = 3 + 1 = 4 Index of pit toilet + owned = 2 + 2 = 4 Index of pit toilet shared 2 + 1 = 3 Index of no toilet or dry toilet = 1 (Bottom of the scale) Table 6A.2 Drinking water supply index (varies between one and five) Main source of drinking water for household member Piped-water-into-dwelling-yard plot Public-tap-stand-pipe Hand-pump Tubewell-or-borehole Protected-dugwell Unprotected-dug-well Protected spring Unprotected spring Rainwater collection Tanker truck Cart with small-tank-drum Surface water Packaged-bottled-water Other

Freq.

(%)

Cum.

15,834 68,016 15,618 7,159 2,228 1,617 412 253 16 2,665 982 970 16,939 278

11.91 51.14 11.74 5.38 1.68 1.22 0.31 0.19 0.01 2 0.74 0.73 12.74 0.21

11.91 63.05 74.8 80.18 81.85 83.07 83.38 83.57 83.58 85.59 86.32 87.05 99.79 100

Piped water into the dwelling or yard or plot = 5 = Drinking water into the dwelling Tanker, cart with small tank drum/packaged bottled water = 4 = Transported water Public tap/hand pump/tube well or bore-well = 3 = Public water supply Dug wells (protected and unprotected wells) = 2 = Well water Natural sources (spring, surface water, others) = 1 Natural source

Nutritional status across social groups

197

Table 6A.3 Cooking fuel index Type of fuel household mainly use for cooking

Firewood Crop-residue Cow-dung-cake Coal-lignite-charcoal Kerosene lpg-npg Electricity Biogas Nocooking Other

Freq.

(%)

Cum.

49,270 810 128 177 832 80,116 844 648 142 13

37.05 0.61 0.1 0.13 0.63 60.25 0.63 0.49 0.11 0.01

37.05 37.66 37.76 37.89 38.51 98.76 99.4 99.88 99.99 100

Cooking fuel on a scale of one to five Electricity/LPG/NPG, = 5 = Clean fuel Biogas = 4 Clean fuel but needs raw material Kerosene = 3 = Kerosene = 3 = Less emission Coal lignite charcoal = 2 Coal-less emission Firewood, crop residue, cow dung = 1 fuel with harmful emissions = 1

Table 6A.4 Housing index Type of house (DLHS-4)

Puccka Semipuccka Kachha Other

Freq.

(%)

Cum.

82,718 41,033 9,074 141

62.21 30.86 6.82 0.11

62.21 93.07 99.89 100

Pucca (permanent structures) = 4 Semi pucca (semi-permanent structures) = 3 Kachha (temporary structures) = 2 Other =1

Average household amenities index is an addition of the scores divided by four for four amenities Top of the scale is 20/4 = 5 Bottom of the scale = 4/4 =1 Household amenities index varies between 1 and 5 It is a continuous variable

Appendix 6B Facts about SC and ST population and the Map of Andhra Pradesh and Telangana

Table 6B.1 Distribution of SC and ST population in Andhra District

Srikakulam Vizianagaram Visakahapatnam East Godavari West Godavari Krishna Guntur Prakasam SPS Nellore YSR Cuddapah Kurnool Ananthapuramu Chittoor Andhra Pradesh Source: Census 2011

% to District population

District share to State

SC

ST

SC & ST

SC

ST

9.50 10.60 7.70 18.30 20.60 19.30 19.60 23.20 22.50 16.20 18.20 14.30 18.80 17.1

6.10 10.00 14.40 4.10 2.80 2.90 5.10 4.40 9.70 2.60 2.00 3.80 3.80 5.3

15.60 20.60 22.10 22.40 23.40 22.20 24.70 27.60 32.20 18.80 20.20 18.10 22.60 22.40

3.00 2.90 3.90 11.20 9.60 10.30 11.30 9.30 7.90 5.50 8.70 6.90 9.30 100

6.30 9.00 23.50 8.10 4.10 5.00 9.40 5.70 10.90 2.90 3.10 5.90 6.00 100

Table 6B.2 Distribution of SC and ST population in Telangana District

% to district population

District share to state

SC

ST

SC & ST

SC

ST

Adilabad Nizamabad

17.82 14.54

18.05 7.56

35.87 22.10

9.03 6.86

15.08 5.87

Karimnagar

18.80

2.83

21.63

13.12

3.25

Medak

17.73

5.27

23.00

9.95

5.14

Hyderabad

6.29

1.24

7.53

4.58

1.49

Rangareddy

12.31

4.13

16.44

12.06

6.66

Mahabub Nagar

17.44

26.43

13.11

11.08

Nalgonda

18.27

11.3

8.99

29.57

11.78

12.00

Warangal

17.54

15.11

32.65

11.39

16.14

Khammam

16.84

25.18

42.02

Telangana

15.45

9.08

24.53

8.12

23.29

100

100

Source: Census 2011

Table 6B.3 Well-being across social groups Social group

Poverty employment and literacy (% population)

State

Poverty

Service sector employment

Literacy

AP

TL

AP

TL

AP

TL

ST SC

31 14

14 17

3 7

13 25

71 89

58 54

BC

11

7

14

29

88

70

OC

6

5

17

29

94

87

Source: CESS MDG Reports (NSS 68th Round 2011–12)

Figure 6B.1 The successor states of Andhra Pradesh and Telangana

7

Access to milk and milk products and child undernutrition Rohit Parasar, R.V. Bhavani and S. Raju1

7.1

Introduction

India houses a large population of undernourished people. Women and children in particular, are more affected. Undernutrition in early childhood affects cognitive abilities and subsequently productivity in adults and affects overall growth and development of a society. The first thousand days of the child from the time of conception to two years of age are regarded as the most crucial period of growth of a human being. Adequate nutrition of the right kind is most important, to ensure proper growth of a child. According to the UN World Health Organisation (WHO), undernutrition is estimated to be associated with 2.7 million child deaths annually or 45% of all child deaths. Infant and young child feeding is a key area to improve child survival and promote healthy growth and development. The first two years of a child’s life are particularly important, as optimal nutrition during this period lowers morbidity and mortality, reduces the risk of chronic disease, and fosters better development overall.2 7.1.1

Consumption of milk for nutrition

Milk and milk products are recommended as protein and micronutrient rich foods for consumption, for both pregnant and lactating women and infants and young children six months of age and more. Several baby foods for complementary feeding of infants over six months of age are fortified milk-based products. Milk and milk products are recommended as a component of infant and young child feeding (IYCF) practices for a balanced diet by the Indian Paediatric Society. Milk is a particularly important form of animal-source food, since it is intended for nurturing the young, a population group at high risk of nutritional deficiencies in many lowincome countries (FAO 2013). Milk promotes growth by providing energy, protein and micronutrients and by stimulating growth factors (Drorr and Allen 2011). Several studies have established the association between milk consumption and nutrition status of children (Iannotti et al. 2013). A systematic review and metaanalysis of 12 studies in the nineteenth and twentieth centuries concluded that there is “moderate quality evidence that dairy products supplementation stimulate linear growth” (de Beer 2012). A study of milk consumption by children aged

202 Rohit Parasar, R.V. Bhavani and S. Raju seven to eight years in rural Vietnam concluded that milk consumption benefitted the children, including lowering the occurrence of underweight and stunting, improving micronutrients status and better learning indicators as well as improving the quality of life (Lien et al. 2009). A large-scale study in Southeast Asia with nationally representative data of 12,376 children in Indonesia, Malaysia, Thailand and Vietnam aged between 1 and 12 years, examined the associations between dairy consumption and nutritional status. The prevalence of stunting and underweight was found to be lower in children who consumed dairy every day (10% and 12%, respectively) compared to children who did not use dairy (21.4% and 18.0%, respectively) (p < 0.05). The prevalence of vitamin A deficiency and vitamin D insufficiency was lower in the group of dairy users (3.9% and 39.4%, respectively) compared to non-dairy consumers (7.5% and 53.8%, respectively) (p < 0.05); leading to the conclusion that dairy as part of a daily diet plays an important role in growth of the child and intake of vitamins A and D (Nguyen Bao et al. 2018). A study among pre-school children (24–59 months) representing three large ethnic groups (Whites, Blacks, Mexican Americans) in the US examined the hypothesis that children who drank more milk, measured either by 24-hour intake or frequency of consumption over the past 30 days, will have greater height than those who drank milk less frequently or in smaller amounts and found that children with the highest level of milk consumption had greater height compared to children with lower intake of milk. No such association was found in the case of other dairy products (Wiley 2009). Examining data from rural Bangladesh, Choudhury and Headey (2018) concluded that increasing access to dairy products can be extremely beneficial to children’s nutrition but may need to be accompanied by efforts to improve nutritional knowledge and appropriate breastfeeding practices. The chapter examines the association of complementary feeding of milk and milk products with poor nutrition status of children in India using data from the National Family Health Survey (NFHS) fourth round of 2015–16. This is supplemented by primary data on diets of children including milk consumption. A review of secondary data revealed that eastern India is deficit in milk production, making it an important market for dairy products. A quantitative survey of 400 households was undertaken across three districts in the state of Odisha in east India in mid2016, to understand the dietary pattern of children between 6 to 59 months of age, and intake of milk and milk products. The chapter is organised as follows: following a discussion on the nutrition status of children and milk production and availability in India and Odisha, the methodology of the study is described. Section 7.3.1 examines the findings from secondary data and section 7.3.2, the findings from the primary survey. Section 7.4 discusses the findings and section 7.5 concludes 7.1.2   The scenario in India The nutrition status of population in India is rather grim. With regard to children, the NFHS-4 (2015–16) reports that 38% of children under age five years are stunted (short for their age); 21% are wasted (thin for their height) and 36%

Milk products and child undernutrition 203 are underweight (thin for their age); 58% of children age 6–59 months have anaemia (IIPS and ICF 2017a). The NFHS also reports on adequacy of diet, including intake of tinned, powdered or fresh milk, in children 6–23 months of age.3 As per NFHS-4 (2015–16), only 42.7% of children six to eight months old were reported to be receiving solid or semi solid food and breast milk; a mere 8.7% breastfeeding children aged 6–23 months were reported to be receiving adequate diet; the figure was only slightly higher at 14.3% for nonbreastfeeding children. In Odisha, 34% children under five years are stunted and underweight and 20.4% wasted; 44.6% children are reported anaemic. In terms of complementary feeding, 54.9% of children six to eight months were reported to be receiving solid and semi-solid food and breastmilk; 8.9% children 6–23 months receive adequate diet and only 5% of non-breastfeeding children receive an adequate diet (IIPS and ICF 2017b). In terms of milk production and availability, India is the largest producer of milk in the world, but both production and per capita consumption varies widely across the country. According to the Economic Survey for 2019, India ranks first in milk production with a production of 176.3 million tonnes in 2017–18, accounting for 20% of world production. But there exists wide inter-state variability in milk production. The per capita availability of milk is determined by the production of milk in the state. While the all India per capita availability of milk is 375 grams per day, it varies between 71 grams per day in Assam to 1120 grams per day in Punjab. In Odisha it is 132 grams per day (GoI 2019).

7.2 7.2.1

Methodology Secondary data analysis

Unit level data was extracted from the NFHS-4 (2015–16) round and the diets of children in the age group of 6 to 59 months and their nutrition status examined, with special reference to consumption of milk. 7.2.2

Primary survey4

Taking Odisha as a state deficit in milk production, one district each in central (Khorda), north (Mayurbhanj) and south (Koraput) Odisha were purposefully selected to produce a sample distribution that covered rural, urban and tribal households. The objective was to examine the diets of children 6 to 59 months of age with reference to consumption of milk and milk products. Key indicators of nutrition status of children in the three districts are given in Table 7.1. The proportion of population for sample calculation from each district and of urban, rural and tribal was based on the Census (GoI 2011)—see Tables 7.2 and 7.3. At the next level, the distribution of the sample across urban (slum and nonslum), rural and tribal was undertaken. The samples for tribal households were

204 Rohit Parasar, R.V. Bhavani and S. Raju Table 7.1 Nutrition status of children in Odisha (%) Parameters

Mayurbhanj

Koraput

Khordha

Odisha

Children under five (stunted) Children under five (wasted) Children under five (underweight) Anaemic children (6–59 months)

43.5 17.2 43.8 34.5

40.3 28.5 44.4 71.4

24.7 13.8 20.3 19.0

34.1 20.4 34.4 44.6

Source: NFHS-4 (District and State Fact Sheets)

Table 7.2 Distribution of sample households by district District

Number of households

Percentage

Khorda Koraput Mayurbhanj Total

120 137 143 400

30.0 34.3 35.7 100.0

Source: Authors’ own calculation, based on the Population Census (GoI 2011)

Table 7.3 Distribution of sample households by location Location

Number of households

Percentage

Urban (slum) Urban (non-slum) Urban total Rural Tribal Total

17 54 71 242 87 400

04.3 13.5 17.8 60.5 21.7 100.0

Source: Authors’ own calculation, based on the Population Census (GoI 2011)

selected only from the tribal population-dominated districts of Mayurbhanj and Koraput; urban households were randomly selected from slums and identified low-income pockets in the districts. Table 7.3 gives the details. The cross-tabulation of sample households across urban/rural/tribal in the three districts is given in Table 7.4. Almost equal proportion of tribal households was taken as sample from Koraput and Mayurbhanj. Khorda in central Odisha does not have a tribal setting and so the sample for tribal households was not drawn from this district, even though some of the households are from the scheduled tribe (ST) community. Following similar logic, sample households from slums were drawn from Khorda district only. Both the secondary and primary data were analysed using STATA statistical package.

Milk products and child undernutrition 205 Table 7.4 Distribution of sample households across the three survey districts in Odisha Settings

Urban-Slum

Urban Non-Slum

Rural

Tribal

Total

District

No.

%

No.

%

No.

%

No.

%

No.

%

Khorda Koraput Mayurbhanj Total

17 – – 17

4.3 – – 4.3

17 18 19 54

4.3 4.5 4.8 13.5

86 76 80 242

21.5 19.0 20.0 60.5

– 43 44 87

– 10.8 11.0 21.8

120 137 143 400

30 34.3 35.8 100

Source: Authors’ own, based on the Population Census (GoI 2011)

7.3 7.3.1

Results (Insights from secondary data) Consumption of milk by children in Odisha: secondary data analysis

Data from the NFHS-4 shows that proportion of children consuming milk in Odisha is low, with the proportion not consuming being higher in rural areas as compared to urban (Table 7.5). Moving on to examine the consumption trend across social groups, the legal and policy framework in India discriminates positively towards scheduled caste (SC) and the scheduled tribe (ST) communities, to improve their socioeconomic status so that people from these communities can overcome the social deprivation historically faced by them. The “other” and “other backward caste (OBC)” categories are the social groups that are better off than the SCs and STs; this is also reflected by better development indicators seen among these social groups. In terms of milk consumption, the group having lowest percentage of children who consumed milk was ST in rural regions at 11% followed by SC at 17%. This may be indicative of more vulnerable communities facing a higher level of deprivation, compared to OBC and the “other” caste categories at 22% and 30% respectively. In urban areas also milk consumption was lowest among the ST and SC groups; the consumption was highest in the OBC and other caste categories at around 33%. This means that every third child that belonged to OBC or the “other” caste category in the urban region consumed milk, as against 2 in 10 ST children in urban areas and 1 in 10 in rural tribal areas. The stark difference highlights the significance of regional and social grouping while examining consumption of milk by children. Odisha, with about 23% population belonging to tribal communities, accounts for 9% of tribal population of the country; on an average, only about 17.9% of children in rural Odisha were found to be consuming milk, whereas the percentage was about 28.9% in urban areas, showing a considerable difference in pattern of consumption of milk between the two regions. This is in line with the national

206 Rohit Parasar, R.V. Bhavani and S. Raju Table 7.5 Children (6–59 months) who consume milk* by social group (%) Social group

Scheduled caste Scheduled tribe Other backward caste Other caste Don’t know Total

Rural

Urban

Yes

No

Total no.

Yes

No

Total no.

17.36 11.22 22.45 29.57 11.54 17.85

82.64 88.78 77.55 70.43 88.46 82.15

1060 1729 1470 460 26 4745

22.80 23.46 33.59 33.65 16.67 28.99

77.20 76.54 66.41 66.35 83.33 71.01

193 162 256 211 6 828

*Powdered/tinned/fresh Source: NFHS-4 (raw data)

trend; the national average also shows that the consumption of milk by children in the tribal population is least. It is also important to note that across all social groups, the percentage of children who consume milk is higher in urban areas than in rural areas. In urban Odisha, the percentage of children who consume milk is about 10% higher than in rural Odisha. Similar difference between urban and rural regions can be seen among the ST and OBC groups, while for “others” and “SC,” the difference is less than 10%. Percentage of children consuming milk—a high value commodity—can have association with the wealth of the household. This was examined next. Figure 7.1 shows the percentage of children who consume milk across the five wealth categories in urban and rural Odisha. The graph depicts a secular trend in rural Odisha; that is as wealth level increases, the percentage of children having milk also rises. Only about 9.3% of children in the lowest wealth category have milk; the percentage in the highest wealth category is 35%, almost four times the number in the poorest category. Unlike in rural Odisha, in the urban region the percentage of children who consume milk does not increase in a linear fashion with wealth category. There is a sudden dip in children consuming milk in the middle-income category to 14% from 25.8% observed in the poor category. Barring the middle category being an exception to the rise in percentage of children who consume milk, the trend can be seen in all other categories. If we compare the urban and the rural scenarios, except in the middle category, the proportion of children who consume milk in urban regions in all other wealth categories is higher than in rural regions. The urban-rural gap between the poorest and the richest categories is least as compared to other categories, a possible indicator of limited accessibility to nutritious food for children of poor households, irrespective of being rural or urban.

Milk products and child undernutrition 207 37.5

40 31.4

Percentage

35 30 25

20.8

20 15

0

14.0

11.4

10 5

35.0

25.8 24.5

18.2

14.0 9.3 Poorest

Poor

29.0

Middle Rich Wealth Categories

Rural

Urban

Richest

Total

Figure 7.1 Milk consumption in children across wealth categories, Odisha (%) Source: NFHS-4 (raw data)

However, it is also important to note that even in the richest category, the percentage of children who consume milk in Odisha is only 35% and 37% in rural and urban regions respectively; implying that even among the richest households majority of children—around 63% to 65%—do not consume milk. The consumption in middle income category shows a counterintuitive pattern and the percentage of children consuming milk in the urban middle-income group could be equivalent to the “poor” counterpart in the rural region. The preceding analysis was based on 24 hours recall of the respondents, showing the percentage of children who had consumed milk the day before the survey, irrespective of the number of times milk was consumed by the child. Information on frequency of consumption of milk in a day by children who consume it can give some more insights. Table 7.6 gives this information. It is seen from Table 7.6 that there is not a large difference between the frequency of consumption of milk between urban and rural children. However, it is important to keep in mind the percentages shown here are out of the children who consume milk if at all, which was about 18% in the rural regions and 29% in the urban regions. Of the children who consume milk, a vast majority of them, about 84%, consumed milk more than once in a day. We can also see the frequency of consumption pattern across social groups in urban and rural settings. We have already seen that ST community in rural regions has the least percentage (around 11%) of children who consume milk while highest percentage of children of urban regions (around 33%) in the OBC and Other categories consume milk (see Table 7.4). Figure 7.2 shows the pattern of consumption of milk more than once a day by children across social groups. It is seen from Figure 7.2 that other than in the SC category, in all other categories the percentage of children who consume milk more than once a day is

208 Rohit Parasar, R.V. Bhavani and S. Raju Table 7.6 Frequency of consumption of milk* per day by children (6–59 months) Frequency of consumption/day

Rural

Urban

Total

Once More than once Twice Thrice Four times Five times Six times Seven or more times Do not know Total

15.5 84.5 46.2 23.3 7.4 3.4 1.5 1.8 0.8 100.0

17.1 82.9 48.2 19.2 11.4 2.0 0.0 0.8 1.2 100.0

15.9 84.1 46.7 22.4 8.3 3.1 1.2 1.6 0.9 100.0

* Powdered/Tinned/Fresh Source: NFHS-4 (raw data)

95

Percentage

90

93.18

Rural

Urban

87.5 84.55

85

84.56

80.41 80

78.95

82.56

80.28

75 70

SC

ST

OBC Social Groups

Others

Figure 7.2 Percentage of children who consume powdered/tinned/fresh milk more than once a day across social categories in urban and rural regions, Odisha Source: NFHS-4 (raw data)

more in the rural regions than in the urban regions. Further, all social categories have lesser proportion of children who consume milk more than once in a day than the SC category. Table 7.7 shows details of pattern of frequency of milk consumption among children across social categories in urban and rural regions of Odisha. It is important to bear in mind that though the percentage of children who consume milk more one than once in a day is highest in the SC category in the urban

Milk products and child undernutrition 209 Table 7.7 Frequency of consumption of milk* by children (6–59 months) by social group Social group

SC

No. of times R consumed in a day

ST U

R

OBC U

R

Others U

R

Total U

R

U

Once 12.5 6.8 19.6 21.1 15.5 17.4 15.4 19.7 15.7 16.7 More than once 87.5 93.2 80.4 79.0 84.6 82.6 84.6 80.3 84.3 83.3 Twice 46.2 52.3 43.3 47.4 46.4 46.5 47.1 47.9 45.8 48.3 Thrice 26.1 31.8 21.7 10.5 25.2 22.1 21.3 14.1 23.9 19.6 Four times 7.6 6.8 7.7 13.2 5.8 11.6 8.1 14.1 7.1 11.7 Five times 1.6 2.3 3.1 2.6 5.2 2.3 2.2 1.4 3.4 2.1 Six times 2.2 0.0 2.1 0.0 0.3 0.0 2.9 0.0 1.5 0.0 Seven or more times 2.7 0.0 2.6 0.0 0.6 0.0 2.2 1.4 1.8 0.4 Do not know 1.1 0.0 0.0 5.3 1.2 0.0 0.7 1.4 0.8 1.3 Total 100 100 100 100 100 100 100 100 100 100 Source: NFHS-4 (raw data) Note: *Powdered/Tinned/Fresh; SC = scheduled caste, ST = scheduled tribe, OBC = other backward caste, R = rural, U = urban

regions, the percentage of children in this category who consume milk at all was only 22.8% against the average of 28.9% in urban Odisha. Other than the SC category the variation in frequency of consumption is not large; about 79% of children of the 23% children in ST category who consume milk in urban regions, consume it more than once while the about 80.4% of the 11% who consume milk in the rural regions consume it more than once. On the whole, seeing the limited variation in terms of frequency of consumption of milk across social categories, it can be said that it is important to increase the number of children who consume milk. Observing frequency pattern of consumption without reference to the percentage of children who consume milk at all will not give the complete picture, given the fact that there exists vast difference across social groups and between the urban and rural regions of Odisha. We will examine frequency of consumption of milk by children across the five wealth categories in urban and rural regions of Odisha (see Figure 7.3 and Appendix Table 7A.1). Figure 7.3 highlights that except for the rich category, there is no major difference in terms of percentage of children who consume milk more than once a day between urban and rural regions across different wealth categories. Out of the 35% and 37% of children who consume milk in rural and urban regions respectively as seen in Figure 7.1, about 75% of them have milk more than once a day. In the poorest wealth class, similar to the richest class, the percentage of

210 Rohit Parasar, R.V. Bhavani and S. Raju 100

Percentage

90 80 70 60

75.0 73.0

80.0 76.5

80.2 72.0

40

75.4

68.3

50

63.3 Rural

Poorest

Poor

Middle Wealth Category

74.7

Rich

Urban Richest

Figure 7.3 Percentage of children who consume powdered/tinned/fresh milk more than once a day across wealth categories in rural and urban regions, Odisha Source: NFHS-4 (raw data)

children who consume milk more than once is 75% in urban and 73% in rural regions; however, it is important to keep in mind that the percentage of children who consume milk at all is just 11% and 9% in the urban and rural regions for the poorest category. 7.3.2  Intra-state    variation in percentage of children who  consume milk We have seen the variation in percentage of children who consume milk across social groups and wealth categories in rural and urban Odisha. The analysis gives us an idea about the vulnerability of children in terms of having access to milk in certain social and wealth categories, and especially in rural areas. For policy intervention, it is also important to have a geographical understanding. Table 7.8 gives information on the consumption of milk by children across districts with rural/urban breakup. Table 7.8 shows a huge variation in term of percentage of children who consume milk in Odisha, varying from 6.8% in Kandhamal to 45.5% in Jagatsinghpur. The districts with low level of consumption of milk are scattered across Odisha, except the districts in the eastern-coastal region. Districts that have more than 30% of children consuming milk are Kendrapara, Ganjam, Nayagarh, Bhadrak, Khordha, Puri and Jagatsinghpur; these seven districts are in the eastern coastal region of Odisha. Barring these districts, every district

Milk products and child undernutrition 211 Table 7.8 District-wise percentage of children (6–59 months) who consume milk Districts

Rural

Urban

Total

Kandhamal Balangir Debagarh Koraput Rayagada Baudh Nuapada Mayurbhanj Nabarangapur Sambalpur Kendujhar Subarnapur Kalahandi Malkangiri Dhenkanal Sundargarh Anugul Gajapati Bargarh Baleshwar Jajapur Jharsuguda Cuttack Kendrapara Ganjam Nayagarh Bhadrak Khordha Puri Jagatsinghpur Odisha

7.3 6.8 8.5 5.8 10.2 9.8 7.8 12.1 9.2 8.7 16.0 13.8 16.3 15.7 17.9 15.2 18.7 20.0 21.9 25.0 24.2 23.1 25.6 31.9 31.2 34.1 36.3 30.7 43.1 46.4 18.2

0.0 9.1 66.7 32.3 5.0 0.0 36.4 0.0 55.6 22.9 19.1 42.9 16.7 27.8 15.0 24.2 28.6 25.0 35.7 22.2 44.4 30.5 38.5 22.2 37.1 22.2 23.8 42.0 27.3 36.4 29.0

6.8 7.1 9.5 9.6 9.7 9.7 10.5 11.6 12.1 12.3 16.3 16.4 16.4 16.4 17.5 19.1 19.8 20.4 23.2 24.8 26.4 27.1 28.5 31.4 32.5 33.3 34.7 35.2 40.6 45.5 19.8

Source: NFHS-4

has less than one-third children consuming milk. Figures showing higher consumption under urban in some of the districts like Koraput and Nabrangpur can be misleading because the urban population in these districts is very low as seen in the overall average for these districts. Figure 7.4 shows the map of Odisha with districts.

Figure 7.4 District map of Odisha

Milk products and child undernutrition 213 7.3.3 Undernutrition in Odisha Underweight or weight being lower than the norm as per the age—is an important indicator of undernutrition in children. In Odisha, about 36.8% of children less than five years of age are underweight. We first examine the prevalence of undernutrition across social groups and wealth classes, before undertaking an exercise to see if there is any association between undernutrition and consumption of milk. Figure 7.5 shows the prevalence of underweight among children across social groups. The highest prevalence of underweight children is in the ST category followed by SC, other backward caste with the category “others” showing the least prevalence. The prevalence of underweight in children of ST is more than double than that in the “others” category. In the SC and ST categories, the percentage of underweight children is higher than the state average. Crosstabulation between underweight and wealth category can show us the association of wealth and undernutrition in the state. Figure 7.6 shows the variation in underweight among children across the five wealth categories. The prevalence of undernutrition, as expected, is highest in the poorest category and lowest in the richest category and reduces as the wealth increases. An expected negative relationship between wealth and the prevalence of undernutrition can be observed and is shown in Figure 7.6. The prevalence of underweight in the middle wealth

Social Groups Percentage of Underweight Children Social Groups Average Percentage of Underweight Children (Odisha) 60 50

Percentage

48.5 40 30

34.4 35.3

36.8

29.7

20

20.6

10 0 Scheduled Caste

Scheduled Tribe

OBC

Others

Don't Know

Social Groups

Figure 7.5 Underweight children (6–59 months) by social groups, Odisha (%) Source: NFHS-4

214 Rohit Parasar, R.V. Bhavani and S. Raju

Percentage

Wealth Class Percentage of Underweight Children Wealth Class Average Percentage of Underweight Children (Odisha) 50 45 40 35 30 25 20 15 10 5 0

46.8 41.6

34.4

35.0 27.3

16.9

Poorest

Poorer

Middle Wealth Class

Richer

Richest

Figure 7.6 Underweight children (6–59 months) by wealth class, Odisha (%) Source: NFHS-4

category—35%—is closest to the state average of 34.4%. The variation in prevalence of underweight is more pronounced among the wealth categories than across social groups seen in Figure 7.5; the proportion of underweight children in the poorest category is close to thrice the prevalence in the richest category. 7.3.4 Association of milk consumption and undernutrition in Odisha Given the evidence of association between milk consumption and undernutrition as discussed in the first section, we examine this aspect in the case of Odisha. Table 7.9 shows the comparison of prevalence of underweight among children who consume milk and those who do not consume milk across urban and rural Odisha, and the state. It is evident from the Table 7.9 that the percentage of children with normal weight is higher in the group that consumes milk across rural and urban Odisha and the state as a whole. Figure 7.7 gives a graphical representation of Table 7.9. The prevalence of underweight and consumption pattern of milk at district level is shown in Figure 7.8; this also reveals lower prevalence of underweight children where consumption of milk by children is higher. Districts in Figure 7.8 have been arranged in ascending order of percentage of children who consume milk and underweight prevalence has been plotted against the respective district. Although the association of percentage of children who consume milk and prevalence of underweight is not consistently negative, a decline in

Table 7.9 Prevalence of underweight (%) in children and milk consumption Nutritional status

Rural

Normal Underweight Total

Urban

Total

Milk

No milk

Milk

No milk

Milk

No milk

72.0 28.0 100

63.6 36.4 100

72.7 27.3 100

71.3 28.7 100

72.2 27.8 100

64.6 35.4 100

Source: NFHS-4 No milk

40 35

36.4

30 Percentage

Milk 35.4

28.0

25

28.7

27.8

27.3

20 15 10 5 0

Rural

Urban Region

Total

Figure 7.7 Underweight in children (6–59 months) who consume/do not consume milk (%) Source: NFHS-4 (raw data) 60

% of children who consume milk

Percentage

50

Underweight

40 30 20

0

kandhamal balangir debagarh koraput rayagada baudh nuapada mayurbhanj nabarangapur sambalpur kendujhar subarnapur kalahandi malkangiri dhenkanal sundargarh anugul gajapati bargarh baleshwar jajapur jharsuguda cuttack kendrapara ganjam nayagarh bhadrak khordha puri jagatsinghapur

10

Districts of Odisha

Figure 7.8 Association of prevalence of underweight children (6–59 months) with percentage of children who consume milk across districts of Odisha Source: NFHS-4

216 Rohit Parasar, R.V. Bhavani and S. Raju rate of undernutrition can be seen with rise in percentage of children who consume milk across districts. Further analysis at the household level can help us understand the association of milk consumption with weight of the children. For this analysis, we use underweight z-score that is an internationally developed score based on the weight and age of the child. We try to find out if a child who consumes milk has higher or better weight score than the child who does not consume milk. Table 7.10 shows the regression analysis for the same. The advantage that we have with the regression analysis is that we can control for household factors like wealth and social groups and then see if a child who consumes milk has better z-score. Table 7.10 shows that there is a significant positive association between consumption of milk and underweight z-score of children. The association holds true even when we control for the wealth index of the household. In the wealth group, the poorest category has been kept as the base category. The coefficient for other categories is significant and increases with increase in wealth group. This implies that the underweight z-score of children is significantly better in the richer wealth class as also seen in Figure 7.6 earlier. Adding another important control variable, social category of the household to which a child belongs, we find that even after controlling for both wealth and social groups, consumption of milk by children has a significant association with better z-score for children (see Table 7.11). Moreover, it can also be observed that in comparison to the ST group, all other categories have significantly better underweight z-score. Also, value of the coefficient increases from SC to OBC to the “other” categories, highlighting that there are significant differences among the social groups and that the children in the ST category are most vulnerable. Table 7.10 Association of child underweight with milk consumption and wealth class Number of obs = 4930 R2 = 0.0803 Underweight z-score Milk consumption***

Coefficient

P>|t|

0.160

0

0.197 0.364 0.581 1.003 −1.891

0 0 0 0 0

Wealth class (poorest base category) Poor*** Middle*** Rich*** Richest*** Constant*** ** Significant at 5% *** Significant at 1% Source: NFHS-4 (authors’ calculation)

Milk products and child undernutrition 217 Table 7.11 Association of child underweight with milk consumption, wealth and social group Number of obs = 4848 R2 = 0.0977 Underweight z-score Milk Consumption**

Coefficient

P>|t|

0.126

0.005

0.154 0.267 0.437 0.804

0.002 0 0 0

0.214 0.229 0.579 0.250 −1.991

0 0 0 0.393 0

Wealth class (poorest base category) Poor** Middle*** Rich*** Richest*** Social groups (ST base category) Schedule caste*** OBC*** Others*** Don’t know Constant*** **Significant at 5% *** Significant at 1% Source: NFHS-4 (authors’ calculation)

7.4

Results (Insights from primary survey data)

The primary survey of 400 households across three districts collected basic socioeconomic information about the households and the diet of children 0–59 months of age in the household. 7.4.1

Standard of living of sample households

The standard of living of the surveyed households was assessed using house construction type and the availability of electricity in the house as parameters, in order to have a kind of ranking as a proxy for wealth ranking of the household. A cross tabulation using both parameters is given in Table 7.12. Table 7.13 categorises households into six groups with three groups in construct type—Kuchha, Semi-Pucca and Pucca; and two being availability and nonavailability of electric connection. There is no pucca household that is without electricity connection. Therefore, there are five categories of households; for ease of understanding, Table 7.13 shows the distribution of households across these five categories.

218 Rohit Parasar, R.V. Bhavani and S. Raju Table 7.12 Cross tabulation of households having electricity and as per household type Electricity

Households with electricity

Households without electricity

Total

House type

No.

Percent

No.

Percent

No.

Percent

Kuchha Semi-Pucca Pucca Total

69 173 89 331

17.3 43.3 22.3 82.8

42 27 0 69

10.5 6.8 0 17.3

111 200 89 400

27.8 50.0 22.3 100

Source: Primary survey

Table 7.13 Distribution of households having electricity and as per household type Standard of living

Numbers

Percent

Kutcha house without electricity Kutcha house with electricity Semi-pucca house without electricity Semi-pucca house with electricity Pucca house with electricity Total

42 69 27 173 89 400

10.5 17.3 6.8 43.3 22.3 100

Source: Primary survey Note: There are no Pucca households without electricity

The children in households surveyed were divided into two categories—children less than two years of age and children aged two years and above but below six years. Out of the 400 households surveyed, 173 households (43.3%) had children below two years of age and 227 households (56.7%) had children aged between two and six years. We first analyse the consumption of milk in the surveyed households on the lines of the secondary data analysed in the previous section. This is followed by an examination of consumption of other baby food products including milk powder. 7.4.2

Consumption of milk by children in surveyed households

Table 7.14 highlights consumption of milk in urban and rural regions; it is seen that almost equal percentage (34–35%) of households feed their children with milk daily. However, the percentage of children who have never consumed milk is higher in the rural regions at 54.1% as against 42.3% in urban regions. If we observe the bifurcation of the urban and rural regions (Table 7.15), again the percentage of children who have never consumed milk is highest in the ST category at 69%, in line with the secondary data.

Milk products and child undernutrition 219 Table 7.14 Frequency of consumption of milk by children across urban and rural regions Consumption of milk

Daily More than once in a week Once a week Rarely Never Total

Urban (slum + non-slum)

Rural (tribal + non-tribal)

Total

No.

%

No.

%

No.

%

25 6 1 9 30 71

35.2 8.5 1.4 12.7 42.3 100

113 16 3 19 178 329

34.3 4.9 0.9 5.8 54.1 100

138 22 4 28 208 400

34.5 5.5 1.0 7.0 52.0 100

Source: Primary survey

Table 7.15 Frequency of consumption of milk by children across different settings Milk

Urban slum

Urban non-slum

Rural

Tribal

Total

No. %

No.

No. %

No. %

No. %

Daily 7 More than once in a week 1 Once a week 1 Rarely 4 Never 4 Total 17

41.2 18 5.9 5 5.9 0 23.5 5 23.5 26 100 54

%

33.3 101 41.7 9.3 11 4.6 0.0 3 1.2 9.3 9 3.7 48.2 118 48.8 100 242 100

12 5 0 10 60 87

13.8 138 34.5 5.8 22 5.5 0.0 4 1.0 11.5 28 7.0 69.0 208 52.0 100 400 100

Source: Primary survey

A point to be made here is that as a cultural practice many tribal households do not consume milk. There were cases where surveyed households reared cattle but did not consume milk. Cattle (both cows and bulls) are used as farm animals and for other purposes including use of dung as fertiliser and cooking fuel. The cow’s milk is fed to the calf (both male and female) and to have milk of the cow is seen as the right of the calf by these households. Across districts, the central district Khorda has highest percentage of children who consume milk every day; Koraput and Mayurbhanj, both districts with high ST population have large percentage of children who have never consumed milk (Table 7.16). On an average, one in every three children surveyed was found to consume milk daily. It is pertinent to recall here that the nutrition status of children in Khorda is far better than in Koraput and Mayurbhanj districts as seen in Table 7.16, counterintuitively pointing to the association seen in the previous section using unit level NFHS-4 data between consumption of milk and prevalence of undernutrition.

220 Rohit Parasar, R.V. Bhavani and S. Raju Table 7.16 District-wise frequency of milk consumption by children Consumption of milk

Khurda No.

Daily More than once in a week Once a week Rarely Never Total

%

Koraput

Mayurbhanj

Total

No.

No.

No.

%

%

%

64 6

53.3 5.0

31 10

22.6 7.3

43 6

30.1 4.2

138 22

34.5 5.5

4 8 38 120

3.3 6.7 31.7 100

0 11 85 137

0.0 8.0 62.0 100

0 9 85 143

0.0 6.3 59.4 100

4 28 208 400

1.0 7.0 52.0 100

Source: Primary source

Daily 27.8

39.7

3.5 4.6 0.6

Percentage

100 90 80 70 60 50 40 30 20 10 0

7.1 1.3 8.8

63.6

43.2

More than Once in a Week Once a Week Rarely Never