The Green Revolution Reconsidered OTHER BOOKS PUBLISHED IN COOPERATION WITH THE INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE Agricultural Change and Rural Poverty: Variations on a Theme by Dharm Narain Edited by John W. Mellor and Gunvant M. Desai Crop Insurance for Agricultural Development: Issues and Experience Edited by Peter B. R. Hazell, Carlos Pomareda, and Alberto Valdes Accelerating Food Production in Sub-Saharan Africa Edited by John W. Mellor, Christopher L. Delgado, and Malcolm J. Blackie Agricultural Price Policy for Developing Countries Edited by John W. Mellor and Raisuddin Ahmed Food Subsidies in Developing Countries: Costs, Benefits, and Policy Options Edited by Per Pinstrup-Andersen Variability in Grain Yields: Implications for Agricultural Research and Policy in Developing Countries Edited by Jock R. Anderson and Peter B. R. Hazell Seasonal Variability in Third World Agriculture: The Consequences for Food Security Edited by David E. Sahn The Green Revolution Reconsidered The Impact of High-Yielding Rice Varieties in South India PETER B. R. HAZELL C. RAMASAMY with contributions by P. K. Aiyasamy, Neal Bliven, Barbara Harriss, John Harriss, Mauricio Jaramillo, Per Pinstrup-Andersen, V. Rajagopalan, and Sudhir Wanmali Published for the International Food Policy Research Institute THE JOHNS HOPKINS UNIVERSITY PRESS Baltimore and London © 1991 The International Food Policy Research Institute All rights reserved Printed in the United States of America The Johns Hopkins University Press 701 West 40th Street Baltimore, Maryland 21211-2190 The Johns Hopkins Press Ltd., London (»)The paper used in this book meets the minimum requirements of American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984. Library of Congress Cataloging-in-Publication Data Hazell, P. B. R. The Green Revolution reconsidered: the impact of high-yielding rice varieties in South India / Peter B.R. Hazell, C. Ramasamy: with contributions by P.K. Aiyasamy . . . [et al.]. p. cm. "Published for the International Food Policy Research Institute." Includes bibliographical references and index. ISBN 0-8018-4185-2 1. Green Revolution—India—North Arcot. 2. Rice—India—North Arcot. 3. Farmers—India—North Arcot. 4. Rural poor—India—North Arcot. I. Ramasamy, C., 1947- . II. Aiyasamy, P. K. HI. International Food Policy Research Institute. IV. Title. HD2075.N56G74 1991. 330.954'8205—dc20 90-26234 Contents List of Tables and Figures vii Preface xiii 1 Introduction Peter B. R. Hazell and C. Ramasamy PART I: THE DIRECT EFFECTS 2 North Arcot and the Green Revolution C. Ramasamy, Peter B. R. Hazell, and P. K. Aiyasamy 11 3 Economic Changes among Village Households Peter B. R. Hazell, C. Ramasamy, V. Rajagopalan, P. K. Aiyasamy, and Neat Bliven 29 4 The Green Revolution in North Arcot: Economic Trends, Household Mobility, and the Politics of an "Awkward Class" John Harriss 57 5 The Impact of Technological Change in Rice Production on Food Consumption and Nutrition Per Pinstrup-Andersen and Mauricio Jaramillo 85 6 Population, Employment, and Wages: A Comparative Study of North Arcot Villages, 1973-1983 John Harriss 105 vi CONTENTS PART II: THE INDIRECT EFFECTS 7 A Social Accounting Matrix of the Regional Economy, 1982/83 Peter B. R. Hazell, C. Ramasamy, V. Rajagopalan, and Ned Bliven 127 8 An Analysis of the Indirect Effects of Agricultural Growth on the Regional Economy Peter B. R. Hazell, C. Ramasamy, and V. Rajagopalan 153 9 The Arni Studies: Changes in the Private Sector of a Market Town, 1973-1983 Barbara Harriss 181 10 Changes in the Provision and Use of Services in the North Arcot Region Sudhir Wanmali 213 11 Conclusions and Policy Implications Peter B. R. Hazell and C. Ramasamy 238 Appendix A: Sources of Growth in the Region's Paddy Production 254 Appendix B: Survey Design 262 References 271 Contributors 277 Index 279 Tables and Figures Tables 2.1 Structure of regional production, North Arcot district, 1980/81 12 2.2 Annual rainfall and area, yield, and production of paddy and groundnuts, North Arcot district 14 2.3 Area under HYV paddy, North Arcot district 18 2.4 Costs and returns from improved local varieties of paddy 19 2.5 Costs and returns from HYV paddy 20 2.6 Irrigation facilities in rural study villages, 1982 28 3.1 Average cropped area, yield, and production of paddy and groundnuts by farm size group 32 3.2 Cropping patterns by farm size group 34 3.3 Paddy farm incomes 36 3.4 Adult employment per paddy farm in crop production by type of labor 38 3.5 Agricultural wages by operation 40 3.6 Agricultural wage transactions by size of farm, re- survey villages 41 3.7 Changes in household incomes 42 3.8 Composition of family income, small paddy farms 43 3.9 Composition of family income, large paddy farms 43 vii viii TABLES AND FIGURES 3.10 Composition of family income, nonpaddy farmers 44 3.11 Composition of family income, landless agricultural workers 44 3.12 Composition of family income, nonagricultural households 45 3.13 Changes in family expenditures 46 3.14 Consolidated statement of income, expenditure, and savings by household type, resurvey villages 47 3.15 Average budget shares for household expenditure, resurvey villages 48 3.16 Indices of interhousehold distribution of income and consumption expenditure 49 3.17 Average land area owned by quartile, cultivator households 50 3.18 Average land area operated by quartile, cultivator households 52 3.19 Average farm sizes by quartile for rich and poor villages 54 3.20 Gini coefficients for land area owned and operated 55 4.1 Agricultural wages 60 4.2 Land prices 61 4.3 Principal occupations of households, Randam 62 4.4 Occupational structure of the labor force, Randam 62 4.5 Occupational structure of the labor force, Veerasambanur, Vinayagapuram, Duli, and Dusi 64 4.6 Changes in distribution of landownership 68 4.7 Changes in area of land owned from inheritance to 1984 by size group 69 4.8 Gains and losses of land from inheritance to 1984, Randam 70 4.9 Gains and losses of land from inheritance to 1984, Veerasambanur, Vinayagapuram, Duli, and Dusi 70 4.10 Class mobility, Randam 73 TABLES AND FIGURES ix 4.11 Structure of outstanding credit by purpose of loan, 1984 76 4.12 Source of outstanding credit, 1984 78 5.1 Characteristics of study households 87 5.2 Total annual consumption expenditures and incomes 88 5.3 Food expenditures 88 5.4 Rice prices and the calorie cost of the total diet 89 5.5 Daily energy and protein consumption 91 5.6 Daily energy obtained from rice consumption 92 5.7 Mean daily energy consumption, resurvey villages 93 5.8 Total food expenditure, calorie consumption, and rice consumption obtained from own production or in-kind earnings 94 5.9 Households consuming below recommended daily allowance for energy 94 5.10 Income and price parameters and other coefficients estimated from consumption functions, paddy-farm households 98 5.11 Income and price parameters and other coefficients estimated from consumption functions, landless households 100 5.12 Relationship among income elasticities 102 5.13 Sources of change in calorie consumption, 1973/74 to 1983/84, resurvey villages 103 6.1 Village populations 108 6.2 Number of agricultural laborers by village 111 6.3 Expansion of groundwater irrigation, 1973-83 112 6.4 Cropping indices, 1982/83 113 6.5 Household labor use 115 6.6 Samba season wages 116 6.7 Harvesting and threshing in-kind wages 117 6.8 Total paddy-farm labor use, 1982/83 119 x TABLES AND FIGURES 6.9 Farm employment, landless households, 1982/83 120 7.1 Schematic social accounting matrix for North Arcot 128 7.2 Structure of commodity transactions, 1982/83 SAM 132 7.3 Structure of private-sector production, 1982/83 SAM 138 7.4 Structure of government-sector production, 1982/83 SAM 142 7.5 Sources of household income, 1982/83 SAM 144 7.6 Summary of 1982/83 SAM 146 7.7 Sources of household outlays, 1982/83 SAM 150 7.8 Rural household incomes, 1982/83 IFPRI/TNAU sur- vey and 1982/83 SAM 151 8.1 Schematic version of the SAM for North Arcot 156 8.2 Production-sector results from the regional model 166 8.3 Results from the 1982/83 regional model with normal- ized with- and without-green revolution paddy and groundnut production levels 176 8.4 Changes in household incomes, regional model and survey results 178 8.5 Components of change in household incomes as a result of the green revolution, regional model 179 9.1 Index of accounting heads 182 9.2 Private firms, Ami 184 9.3 Financial characteristics of sectors of Ami business economy, 1973 188 9.4 Financial characteristics of sectors of Ami business economy, 1983 190 9.5 Frequency of investments by type of firm, Ami 194 9.6 Labor and employment details in Ami business economy 198 9.7 Average urban wages, Arni 201 9.8 Rural and urban wages, Arni and region, 1983 201 TABLES AND FIGURES xi 9.9 Per capita income as multiple of poverty line, Arni and region, 1983 203 9.10 Commodity flow accounts, Arni, 1973 206 9.11 Commodity flow accounts, Arni, 1983 208 10.1 Occurrence, ranking, thresholds, and weights of services, 1983 215 10.2 Centrality of service provision and distribution of settlements, 1983 219 10.3 Spatial features of middle-order service centers, 1983 220 10.4 Spatial features of high-order service centers, 1983 223 10.5 Centrality scores of service provision, 1973 and 1983 224 10.6 Spatial features of middle-order service centers, 1973 and 1983 226 10.7 Sample villages and their service centers 227 10.8 Number of services used within and outside of sample villages 227 10.9 Definition of service groups 228 10.10 Independent variables in regression 230 10.11 Average and marginal budget shares for sample households 231 10.12 Effects of household characteristics and distance on average expenditures by service group 232 10.13 Estimated input demand equations 234 10.14 Mean distances by service category 235 11.1 Changes in the structure of regional employment 246 A.I Area, yield, and production elasticities for paddy, North Arcot district 256 A.2 Sources of change in area and yield in time-series model 259 A.3 Changes in mean values of paddy variables, North Arcot district 260 xii TABLES AND FIGURES A.4 Decomposition of sources of change in area, yield, and production of paddy, North Arcot district 261 B.I Urban villages in study region 265 B.2 Urban towns in study region 266 B.3 Sample sizes for usable monthly income and expendi- ture data, rural surveys 268 B.4 Sample sizes for monthly income and expenditure data, urban survey 269 Figures 2.1 Area and yield of rice and groundnuts 17 2.2 Gross margins per hectare of paddy 22 2.3 Study villages and towns 24 5.1 Mean energy consumption in resurveyed villages 90 5.2 Households consuming less than 80 percent of energy RDA, resurvey villages, small paddy farmers 95 5.3 Households consuming less than 80 percent of energy RDA, resurvey villages, large paddy farmers 96 5.4 Households consuming less than 80 percent of energy RDA, resurvey villages, landless laborers 97 6.1 Schematic classification of North Arcot villages, 1980s 121 10.1 North Arcot study region, middle-order service centers 221 10.2 North Arcot study region, high-order service centers 222 Preface THE "GREEN REVOLUTION"—a term used for rapid increases in wheat and rice yields in developing countries brought about by im- proved varieties combined with the expanded use of fertilizers and other chemical inputs—has had an important impact on incomes and food supplies in many developing countries. It has also spawned a lively controversy over its impact on the poor, with some critics claiming that inequality, and perhaps even absolute poverty, has increased in rural areas as a consequence of the green revolution. Given the importance of future rounds of yield-increasing technol- ogies for fostering economic development and feeding growing popu- lations in most developing countries, it is imperative that the economic and social forces released by these technologies be better understood so that they can be harnessed to achieve the twin goals of growth and equity. To this end, the International Food Policy Research Institute (IFPRI) embarked, in the early 1980s, on a series of in-depth case studies of the impact of technological change in agriculture. This study of the North Arcot district in South India is the first in that series, and it was undertaken in close collaboration with the Tamil Nadu Agri- cultural University (TNAU) at Coimbatore. A companion study has also been undertaken in the Eastern Province of Zambia. A unique feature of these studies lies in the emphasis given to the growth linkage effects of agricultural growth on the rural nonfarm econ- omy. Inspired by the earlier work of John Mellor and associates at Cornell University, it was hypothesized that the rural poor may obtain significant indirect benefits from agricultural growth because of in- creases in income-earning opportunities that arise in the local nonfarm economy. Moreover, this potential has not been adequately addressed in previous studies of the green revolution. Initial funding for this study was generously provided by the Ford Foundation, New Delhi, and the Overseas Development Administra- tion of the United Kingdom. The project ran into financial distress xin xiv PREFACE when a severe drought in the study region undermined the value of the household surveys conducted in 1982/83, and the need arose to repeat the surveys in the following year. The funds required to complete the study were provided by the Swiss Development Cooperation and Hu- manitarian Aid as part of its support of the companion study in Zambia. The Swiss Development Cooperation and Humanitarian Aid also funded a workshop held at Ootacamund, Tamil Nadu, in February 1986 at which preliminary results from the study were presented to an in- ternational group of scholars and Indian government officials. The final product benefited enormously from the open and frank discussions held at that workshop. Many individuals have contributed to the successful completion of this study. We are grateful to them all. A special note of thanks is due B. H. Farmer, Robert Chambers, Nanjamma Chinnappa, and John and Barbara Harriss, who, as members of the Cambridge and Madras universities team that surveyed the North Arcot region in 1973/74, not only made their earlier data fully available to us for comparative analysis but also assisted greatly in the design and implementation of our own surveys to enhance their comparability with the 1973/74 survey. Mr. A. Venkataraman and Professor V. Rajagopalan first directed our attention to the North Arcot region and, as successive vice-chan- cellors of Tamil Nadu Agricultural University (TNAU), were instru- mental in forging and sustaining the administrative arrangements that made this study possible. Nor could the surveys have been undertaken without the enthusiastic assistance of Professor P. K. Aiyasamy (then head of the Department of Agricultural Economics at TNAU), in de- signing the survey instruments and in recruiting and training the field team. Professor Sundaresan, head of the Poultry Research and De- velopment Centre, also provided vital support to the field team at its Vellore base, and Dr. Radhakrishnan, Management Information Ser- vices, Madras, supervised the entry and processing of the survey data. But the real heroes of the survey were the enumerators who, despite the unusually harsh conditions of the 1982/83 drought, diligently served at their posts and maintained high professional standards. They are as follows: S. AkbarBatcha, A. Alagesan, M. Arumugam, S. R. Asokan, M. Bhoopalan, M. Chandrasekaran, M. Dhamodharan, K. Dasara- than, V. Gunasekaran, P. Jayabalan, G. Jayaraman, U. Jayaraman, D. Kandaswamy, K. Mani, S. Marudhachalam, S. Radhakrishnan, and V. Subramanian. Finally, we are grateful to Jock Anderson, Randy Barker, Robert Chambers, Dana Dalrymple, B. H. Farmer, Marco Ferroni, Barbara and John Harriss, and Michael Lipton for comments on parts of earlier drafts of this study, though we absolve them of responsibility for the final product. The Green Revolution Reconsidered CHAPTER 1 Introduction Peter B. R. Hazell and C. Ramasamy AGRICULTURAL TECHNOLOGIES OF the "green revolution" type have brought substantial direct benefits to many developing coun- tries. Prominent among these has been increased food output, some- times even in excess of the increasing food demands of a growing pop- ulation. This has enabled food prices to decline in some countries, while in others prices have not risen as fast as they would have without the green revolution. One of the attractions of the green revolution technologies is that they are, in principle, scale neutral, and can raise yields and incomes for both small- and large-scale farmers. Yet a number of early studies of the impact of the green revolution concluded that the rural poor did not receive a fair share of the benefits generated. It was argued that large farmers were the main adopters of the new technology, and smaller farmers were either unaffected or adversely affected because the green revolution resulted in lower product prices, higher input prices, efforts by large farmers to increase rents or force tenants off the land, and attempts by larger farmers to increase landholdings by purchasing smaller farms, thus forcing those farmers into landlessness. It was also argued that the green revolution encouraged unnecessary mechanization, with a resulting reduction in rural employment (Cleaver 1972; Griffin 1974). The net result, as argued by some, was a rapid increase in the inequality of income and asset distribution, and a wors- ening of absolute poverty in areas affected by the green revolution (e.g., Griffin 1972, 1974; Fraenkel 1976; Harriss 1977; Hewitt de Ala- cantara 1976; ILO 1977; Pearse 1980). These conclusions have not proved valid when subjected to the scrutiny of more recent evidence (Blyn 1983; Pinstrup-Andersen and Hazell 1985; Lipton 1989). Ahluwalia (1985) provides evidence that the incidence of rural poverty in India declined almost steadily between 1967/68 and 1977/78. This is contrary to the earlier findings of Griffin 2 THE GREEN REVOLUTION RECONSIDERED and Ghose (1979), who analyzed comparable data for the period 1960/61 to 1973/74. Ahluwalia (1977,1985) and Rao (1985) found that the incidence of rural poverty is negatively related to agricultural output levels per head. Bell, Hazell, and Slade (1982) provide evidence that agricultural technology can help alleviate absolute rural poverty. They studied the combined impact of an irrigation project and high-yielding varieties (HYVs) of rice in the Muda River region of Malaysia over the period 1967-74. The average per capita income of the population living in the project area increased by 70 percent when measured in constant prices. Landowning households gained relatively most, but landless paddy workers also increased their real per capita incomes by 97 percent, despite a wholesale shift to tractor mechanization for land preparation (Bell, Hazell, and Slade 1982, Table 7.7). Using farm-level data from a number of Asian countries, Barker and Herdt (1978) found that while small farmers reported greater dif- ficulty in obtaining some inputs, such as credit and fertilizers, differ- ences in the rate of adoption of new varieties between small and large farmers were not significant, even in villages with marked inequality in land distribution. In a study of the impact of the green revolution in the Indian Punjab, Blyn (1983) concluded that (1) real income from family resources increased relatively more for families with smaller holdings, thereby reducing inequality, and (2) the total employment of hired labor increased while real wages remained constant, and this led to a clear gain for labor. Why did the earlier studies err? Pinstrup-Andersen and Hazell (1985) offer four possible reasons. First, the studies were conducted too soon after the release of the green revolution technologies. While it was true that early adopters were primarily larger farmers, the studies failed to recognize that small farmers would follow quickly once they observed the success of their larger-scale brethren. (See, for example, Byerlee and Harrington 1983; Chaudhry 1982; Pinstrup-Andersen 1982; Blyn 1983; Herdt and Capule 1983; and Prahladachar 1983.) This pat- tern may also have been reinforced—perhaps as a result of the initial criticisms—by the later release of plant varieties that were better suited to small-farm needs than the initial HYVs (Lipton 1989), and by im- provements in the provision of services—especially credit, input sup- plies, marketing, and extension—for small farmers (Griffin 1988). Second, the benefits to the poor, as consumers of rice and wheat, through lower prices were largely overlooked. Empirical evidence of consumer gains from technological change in developing-country ag- riculture is plentiful (e.g., Akino and Hayami 1975; Mellor 1975; Even- son and Flores 1978; Scobie and Posada 1978; Pinstrup-Andersen 1979; Introduction 3 and Scobie 1979). The consumer gains come about because food prices are lower than they would have been in the absence of the production increases induced by technological change. Population growth, import substitution, export growth, and domestic price policies can dampen the price reduction. In fact, price and foreign trade policies have been used extensively to strike a more desirable balance between the harmful effects of price decreases on farmers and future food production, and the beneficial effects on consumers. Since the green revolution gen- erates an economic surplus by more efficient use of resources and re- duced unit costs, consumer gains need not imply producer losses. Both may gain. Third, little or no attention was given to indirect growth linkages of the green revolution with the rural nonfarm economy and the re- sulting impact on the incomes of the poor. Johnston and Kilby (1975), Mellor (1976), and Mellor and Johnston (1984) have argued that ag- ricultural growth focused on small- and medium-sized farms generates rapid, equitable, and geographically dispersed growth because of sub- stantial labor-intensive linkages with the rural nonfarm economy. Accumulating empirical evidence from Asia confirms that these in- direct effects are nearly as important for rural areas as the direct effects of technological change (Gibb 1974; Bell, Hazell, and Slade 1982; Haggblade and Hazell 1989). The indirect benefits, however, are not restricted to the poor. They also increase the earnings of skilled workers as well as providing lucrative returns to capital and managerial skills. In the Muda study, for example, Bell, Hazell, and Slade found that the indirect benefits of the project were skewed in favor of the nonfarm households in the region, many of which were already relatively well off. They also found that even among agricultural households, the landed households fared better than the landless. The point to be made is that although the indirect effects of agricultural growth are unlikely to improve the relative distribution of income within rural areas, they can still have wide-reaching effects in alleviating absolute poverty. Fourth, the impact of the green revolution was frequently confused with the impact of population growth, or with institutional arrange- ments, agricultural policies, and labor-saving mechanization. Such con- fusion leads to incorrect identification of the causes of rural poverty, and thus to inappropriate recommendations for action to reduce such poverty. It also leads to a failure to appreciate the extent to which poverty and malnutrition would have been worse today without the additional food bestowed by the green revolution. It also seems likely that too much was concluded from a limited number of case studies. Given the vastness of the South Asian sub- continent and the diversity of natural and social environments that it 4 THE GREEN REVOLUTION RECONSIDERED contains, as Farmer (1986) observes, "It is prima facie not to be ex- pected that 'the new technology' would operate in the same way or have the same social and economic effects all over South Asia, or even all over any one of its countries." To understand more fully both the short- and long-term impacts of technological change on rural welfare, and in order to assist in the design of appropriate technologies, policies, and institutional change to enhance the poverty-reducing role of technological change, the In- ternational Food Policy Research Institute and the Tamil Nadu Agri- cultural University (IFPRI-TNAU) collaborated in an in-depth study of the changes induced by technological change in a rice-growing region in South India. The selected area in the North Arcot district offered several advan- tages. First, it is an important rice-growing region that has benefited from the high-yielding varieties developed in the late 1960s. As in many other green revolution areas, there has also been an accompanying increase in irrigation and the use of other modern inputs, especially fertilizers. Paddy yields grew at nearly 3 percent per year over the period 1950/51 to 1984/85, with most of the increase occurring since the late 1960s. These gains are modest compared with the more dramatic changes observed in Punjab and Haryana, but the North Arcot region usefully typifies the more common experience of other rice-growing areas in India. Second, the region is dominated by small-scale farms; in 1983 the average farm size was 1.2 hectares. Given also that about one-third of the rural households are landless agricultural laborers, the equity issue is an important one for the region. Third, the region is removed from any major urban or industrial center, and so agricultural growth is the main driving force in the local economy. This facilitates analysis of the growth linkage effects of ag- ricultural growth on the nonfarm economy. Last, but by no means least, the region was studied in 1973/74 by a team from Cambridge and Madras universities (Farmer 1977). This study involved the collection of monthly household survey data for one year covering detailed aspects of farm management, employment, sources of income, household assets, food consumption, and expen- diture patterns. The household survey included farm and nonfarm households in the rural areas. In addition, a survey of employment and trade in one of the local towns was conducted. An important finding of the Cambridge and Madras universities study was that only about 13 percent of the paddy area was planted to HYVs despite official statistics claiming that 39 percent of the area was so planted (Chinnappa 1977). However, in a postscript study based on Introduction 5 a return visit in 1976, John Harriss (1977) found that HYVs had by then been adopted much more widely. This is also supported by avail- able statistics on rice production for North Arcot. Consequently, by conducting a similar set of surveys 10 years later, it was hoped to learn much about the impact of the HYVs over a crucial period of techno- logical change. The book is organized in two parts. In Part 1, Chapter 2 provides essential background information on the study region and its economy, the changes in agricultural technology and output that occurred over the period of study, and the surveys and the database used in the research. Two features of the IFPRI-TNAU surveys described in Chap- ter 2 deserve special mention. First, although the individual households selected in the Cambridge and Madras universities and IFPRI-TNAU rural surveys were necessarily different, both surveys were conducted in the same 11 villages. These villages were originally selected as a representative sample for the study region. Second, the IFPRI-TNAU surveys were initially undertaken in 1982/83, but because this turned out to be a severe drought year, the survey was repeated in 1983/84 using a subsample of households from the previous year. The discussion of agricultural growth also includes an analysis of the sources of growth; this analysis is complemented by Appendix A, where formal decom- position methods are applied in an attempt to unravel the separate contributions of increases in HYVs, fertilizer, and irrigation to increas- ing rice production. The remaining chapters of Part 1 (Chapters 3 to 6) are concerned with the socioeconomic changes that occurred in the 11 sample villages between the two surveys. Chapter 3 is concerned with changes in farm production, farm income, employment, wages, family income, con- sumption expenditure, and the distribution of land, and uses the survey data to analyze these changes at a pooled village level. In Chapter 4, an anthropologist (John Harriss) provides an independent but parallel analysis, based on his own fieldwork in 1972/73 and 1983/84. His analysis largely corroborates the findings in Chapter 3 and provides additional insights into some of the causal factors at work. Harriss also addresses the important question of whether the green revolution has tended to polarize class and political alliances, particularly between the rich and poor, that might lead to the kind of political unrest anticipated by some of the more radical critics of the green revolution. In Chapter 5, Per Pinstrup-Andersen and Mauricio Jaramillo deal with observed changes in food consumption and nutrition among the sample households, and develop an analytical framework for measuring the nutritional impact of the green revolution. To conclude Part 1, Chapter 6 examines aspects of the intervillage 6 THE GREEN REVOLUTION RECONSIDERED variation in the changes that have occurred. Building on earlier work by Chambers and Harriss (1977), John Harriss seeks to classify the villages according to underlying differences in their resource endow- ments, location, and caste and class structure, as a basis for better understanding the patterns of change induced by the green revolution. Part 2 is concerned with the indirect benefits of agricultural growth to the region's nonfarm economy. We begin in Chapter 7 with the construction of a social accounting matrix (SAM) that provides a de- tailed description of the structure of the regional economy in 1982/83. The SAM, which is based largely on the IFPRI-TNAU 1982/83 surveys, features 59 production sector accounts, 134 commodity accounts, 8 factor accounts, 10 household accounts, 14 government accounts, a capital account, and a rest-of-world account. It is one of the most detailed sets of accounts that have ever been compiled for a rural region, and it provides insights into the linkages between different parts of the regional economy. In Chapter 8, the SAM becomes the database for a model of the regional economy. This model is an extended input-output model in which production, household consumption, savings, and some govern- ment activities are endogenized; the exogenous variables are exports, investments, and remaining government activity. Once validated, the model provides estimates of the income multiplier arising in the non- farm economy, given a unit increase in agricultural income. The model is also used to estimate what the regional economy would have been like in the early 1980s had the agricultural growth of the previous decade not occurred. The model is particularly attractive for these purposes because (1) it provides detailed results by production activity and house- hold type, and (2) it enables the impact of agricultural growth to be separated from other autonomous sources of growth that occurred in the regional economy. Chapter 9 provides a descriptive analysis of the changes that oc- curred between 1973 and 1983 in the private sector of Arni, one of the market towns located in the study region. The analysis is based on survey data that the author, Barbara Harriss, collected during her own fieldwork in 1973 and 1983. The analysis includes a description of the changes that occurred in the number and types of firms, and in firm assets, output, income, and employment. There is also an analysis of the changes in wages and total employment that occurred in Arni be- tween 1973 and 1983, and in Ami's trading patterns in relation to the study region and the rest of the world. If agricultural growth is to stimulate the development of the region's nonfarm economy, then the provision of many key services, such as credit, agroprocessing, marketing, health, education, transport, com- Introduction 1 munication, and retail and personal services, must keep pace with the growth in demand. In Chapter 10, Sudhir Wanmali provides a detailed description of the spatial patterns of provision and use of services in the study region and how these changed during the period 1973-83. He also uses the IFPRI-TNAU household survey data to analyze the determinants of rural household demand for services, and especially the role of distance. Convenient access to services is clearly as important as their existence and cost. An important subsidiary outcome of the analysis in Chapter 10 is the support it provides for the definition of the study region as a meaningful unit of analysis for the growth linkages work reported in Chapters 7 and 8. The book concludes with a synthesis of the research findings and a discussion of their implications for agricultural research and policy. Parti The Direct Effects CHAPTER 2 North Arcot and the Green Revolution C. Ramasamy, Peter B. R. Hazell, and P. K. Aiyasamy The North Arcot Economy ITH ARCOT DISTRICT, which embraces the study region, lies in the northwest of Tamil Nadu state. It is a relatively densely populated region; in 1981 the population density was 357 persons per square kilometer of land. It is also a relatively poor region within India. For example, in 1980/81 the district's net domestic product (NDP) at factor cost was Rs 3,285 million, or Rs 750 (US$95) per capita. This compared with a national average in 1983 of US$260 per capita (World Bank 1986). Agriculture is the predominant activity in the region, accounting for 40 percent of NDP (Table 2.1). Within the agricultural sector, paddy, groundnuts, and sugarcane are the predominant sources of income. These crops also support a downstream agroprocessing industry that is an important part of the manufacturing sector. In 1981 there were 1,825 paddy hullers, 542 groundnut decorticators, 850 oil mills, and three major sugar factories in the district. Milk production is also important, and a sizable herd of milk and draft animals helps support about 300 tanneries in the district, as well as numerous butchers and dairy- processing and retail shops. Manufacturing accounts for 20 percent of the region's NDP (Table 2.1). Apart from agroprocessing and tanneries, the main manufacturing activities are silk and cotton textiles, an array of cottage industries, and chemical and metal manufacturing. According to the 1981 census, the agricultural sector employed 1.16 million full-time workers, or 68 percent of the region's work force. Of these, 606,000 were cultivators and 552,000 were agricultural laborers. A further 111,000 workers were employed in household industry, and 437,000 were employed in other activities. The largest formal employer is the government. In 1981, the combined employment of central, state, 11 12 THE GREEN REVOLUTION RECONSIDERED TABLE 2.1 Structure of Regional Production, North Arcot District, 1980/81 Sector Agriculture & allied activities Manufacturing Trade, hotels, & restaurants Construction Transport & storage (other than railways) Real estate Banking & insurance Public administration Electricity, gas, & water Communications Railways Mining & quarrying All other sectors Net Domestic Product (million Rs) 1,329.8 649.2 406.4 188.7 151.0 86.2 69.5 64.7 41.1 36.5 25.2 19.1 217.5 Percent of Total NDP 40.5 19.8 12.4 5.7 4.6 2.6 2.1 2.0 1.3 1.1 0.8 0.6 6.6 Total 3,284.9 100.0 Source: Assistant director of statistics, Vellore. and district government, quasi-government organizations, municipali- ties, and block development offices was 73,000 jobs, or 5 percent of the region's full-time work force. There are about 23 urban centers in the district with populations of 8,000 or more, and 13 of these are taluk headquarters. Vellore is the district capital and has a population of about 250,000. Of a total pop- ulation of 4.4 million people in North Arcot district, only 1.0 million, or 23 percent, live in urban areas. The urban population increased at an average rate of 2.6 percent per year between 1971 and 1981, com- pared with a 1.3 percent growth rate for the rural population. The district is blessed with a relatively good infrastructure. A dense network of roads extending over 8,800 kilometers connects all 2,049 rural villages in the district. A railway line also passes through the district and connects all the important towns. There are about 1,037 post offices and 126 telegraph offices. Almost all the villages have electricity. Every village with a population of 300 or more has a primary school, and a high school is generally available within a radius of five miles. There are 536 hospitals, 18 blood banks, and about 300 child welfare centers. There is also a wide network of banking facilities, with 281 commercial bank branches servicing the district. North Arcot and the Green Revolution 13 Agriculture Paddy and groundnuts are the major crops, and these are grown pri- marily in the eastern part of the district. They each account for about one-third of the total cultivated area in the district. The western part of North Arcot is more diversified and produces sugarcane, bananas, horticultural crops, and coconut. Cattle provide the main source of draft power for crop production and also by-products such as milk, carves, and manure. The district enjoys two monsoons: the southwest monsoon from June to September and the northeast monsoon from October to De- cember. The northeast monsoon is the most important and provides about 60 percent of the total annual rainfall of 972 millimeters. In harmony with these rainfall patterns, paddy has traditionally been grown in three well-defined seasons, namely samba, navarai, and sor- navari. The samba crop is the main rainy-season crop. It is sown in July or August and harvested in December or January. The navarai crop coincides with the dry season and depends entirely on irrigation. It stretches from December or January to May. The sornavari crop extends from May or June to September and encompasses the light, southwest monsoon. Millets and sorghum are grown as rainfed crops from June-July to October-November, and as irrigated crops from February-March to June-July. The main cropping season for pulses (red gram, black gram, and green gram) is from June-July to December-January. For ground- nuts, the rainfed season is from June-July to September-October, and the irrigated crop is grown between December-January and March- April. Sugarcane and bananas are planted in January and harvested in December. Of a total gross area of about 690,000 hectares planted to crops in the district each year, 400,000 hectares (58 percent) are irrigated. The net irrigated area is about 250,000 hectares. Water is supplied by canals (7 percent of the net irrigated area), tanks (33 percent), and wells (60 percent). Unlike tubewells, the wells in North Arcot are large, open wells sunk in the regolith to tap groundwater supplies in the crystalline rock beneath. There are about 290,000 irrigation wells in the district, or one for every 1.81 hectares of net sown area. This is the highest ratio of all the districts in the state of Tamil Nadu. Rural electrification has had a strong influence on the expansion of well irrigation; of a total of 160,000 pumpsets in 1982/83, 140,000 were electric and only 20,000 were diesel powered. Irrigation allows almost continuous cropping of the land throughout TABLE 2.2 Annual Rainfall and Area, Yield, and Production of Paddy and Groundnuts, North Arcot District Year 1961/62 1962/63 1963/64 1964/65 1965/66 1966/67 1967/68 1968/69 1969/70 1970/71 1971/72 1972/73 1973/74 1974/75 1975/76 1976/77 1977/78 1978/79 1979/80 1980/81 1981/82 1982/83 1983/84 1984/85 Growth rate (%) Coefficient of variation (%) Average: 1963/64-1965/66 1977/78-1979/80 Percent change Area (thous ha) 259 279 293 305 275 301 278 170 251 294 274 290 269 233 241 276 316 295 307 136 167 118 265 255 -1.47 19.79 291 306 5.2 Paddy Yield (kg/ha) 1,493 1,440 1,438 1,570 1,397 1,320 1,180 1,224 1,540 2,143 2,064 1,906 1,858 1,729 2,116 2,073 2,335 2,179 2,182 1,844 2,345 2,452 2,615 2,694 2.94 11.37 1,468 2,232 52.0 Groundnuts Production (thous t) 387 402 422 480 384 397 329 208 387 631 566 554 500 404 511 572 737 642 671 250 391 290 693 687 1.47 28.58 429 683 59.2 Area (thous ha) 185 206 201 198 200 202 220 201 189 181 227 223 246 228 229 226 222 210 212 200 265 230 305 208 0.96 10.12 200 215 7.5 Yield (kg/ha) 1,232 1,189 1,214 1,020 715 960 805 796 825 1,122 1,044 812 1,024 908 1,131 823 1,243 814 1,052 650 1,264 1,291 1,000 1,076 0.07 18.83 983 1,036 5.4 Production (thous t) 228 245 244 202 143 194 177 160 156 203 237 181 252 207 259 186 276 171 223 130 335 297 305 224 1.04 22.52 196 223 13.8 Annual Rainfall (mm) 1,045 1,351 1,198 993 1,131 1,239 745 741 1,033 811 1,075 1,034 732 896 997 1,283 1,472 1,192 1,048 570 1,062 751 1,272 1,076 n.a. 21.58 1,107 1,237 11.7 Note: The coefficients of variation were calculated after removing trend. North Arcot and the Green Revolution 15 the year. However, since tanks and wells need adequate rain to re- plenish water reserves each year, they provide only limited insurance against drought. This is particularly troublesome because the region experiences wide variations in annual rainfall; coefficients of variation range from 18 to 31 percent among the 13 taluks in the district. During a severe drought in 1982/83, for example, the gross paddy area planted fell 40 percent below trend (Table 2.2). Paddy production is particularly affected by variations in annual rainfall; the coefficient of variation (cv) around trend was 29 percent during the period 1961/62 to 1984/85 (Table 2.2). Yields are less affected by rainfall (cv = 11 percent) than the area planted (cv = 20 percent), suggesting that farmers adjust the area of paddy grown to fit available water reserves each year. Groundnut production is slightly less variable than paddy; the cv around trend was 23 percent during 1961/62 to 1984/85. Unlike paddy, groundnut yields are less stable than the area planted. This is because only part of the crop is irrigated. Small farmers are prevalent in North Arcot. In 1979 there were 574,000 holdings and the average size was 1 hectare. About 68 percent of the farms were 1 hectare or less, and about 86 percent were 2 hectares or less. North Arcot is also dominated by owner-operated farms. Pure ten- ant farms are scarce, and most land-leasing arrangements involve farm- ers who already own some land of their own. Rents are paid in cash or kind, but they usually involve fixed rents. Sharecropping is rare. The Green Revolution in North Arcot Growth in Agricultural Output Paddy and groundnuts are not only the predominant crops in the region's agriculture; they have also been the major sources of growth in agricultural output in recent decades. However, as shown in Table 2.2, to designate this growth as a revolution appears, at least at first blush, to be a bit of a misnomer; the average annual growth rates of paddy and groundnut production over the period 1961/62 to 1984/85 were only 1.47 and 1.04 percent, respectively. This growth was obtained almost entirely from area expansion in the case of groundnuts, and while expansion can be partly attributed to increased investments in irrigation, there was very little change in groundnut technology. Indeed, the predominant varieties, TMV2 and 16 THE GREEN REVOLUTION RECONSIDERED TMV7, which are of the bunch type, were grown throughout the period of study. In contrast, the growth in paddy production was technologically driven; yields increased by nearly 3 percent per year between 1961/62 and 1984/85, while the area grown actually declined (by 1.5 percent per year). Most of this yield increase has occurred since the late 1960s (Figure 2.1) and can be attributed to green revolution inputs such as the high-yielding varieties (HYVs) and fertilizers (see Appendix A). But average growth rates do not adequately capture the discontinuities associated with abrupt changes. Comparison of three-year average yields for 1963/64-1965/66 and 1977/78-1979/80 (periods of relatively normal rainfall) shows that paddy yields jumped about 50 percent be- tween these periods (Table 2.2). Paddy production also increased, by 60 percent, while the paddy area remained virtually constant. These changes are more impressive in size and, given their technological or- igin, can be labeled a green revolution within the spirit of the widespread usage of this term. Changes in Paddy Technology An analysis of the sources of growth (Appendix A) shows that nearly all the growth in the region's paddy production since 1950/51 can be attributed to varietal improvement and the more intensive use of ni- trogen and irrigation water. Other changes in the region's paddy tech- nology involved the mechanization of water lifting and the use of power sprayers and threshers. VARIETIES. One of the reasons that the green revolution did not have a more dramatic impact in North Arcot is that there had been a long and successful tradition of improving paddy varieties at local re- search stations, and some of the features of HYVs that account for their higher productivity had already been incorporated into improved local varieties. For example, TKM6, which was later to become one of the parents of IR20, was developed and released in the region as far back as 1952. This variety is photoperiod insensitive and can be grown all year round. It is also a short-duration variety, with a growing period of only 110-15 days. The first HYV, Taichung Native 1, was introduced in North Arcot in 1965 from Taiwan. As with all subsequent HYVs, the main advan- tages over existing improved local varieties lay in their short stiff-straw and their higher responsiveness to nitrogen, especially during the dry navarai season. The early HYVs proved susceptible to major rice pests and diseases North Arcot and the Green Revolution 17 Rice 1961/62 Area 350 300 250 200 150 100 50 0 1961/62 Yield 3,000 2,500 2,000 1966/67 1971/72 Groundnuts 1976/77 1981/82 V Area (1,000 ha) Yield (kg/ha) -L-4- J I L J I I I | L Yield 1,400 1,200 1,000 800 600 400 200 1966/67 1971/72 1976/77 1981/82 Fig. 2.1. Area and yield of rice and groundnuts. and were not widely accepted by farmers. The major break came with the release of IRS (developed by the International Rice Research In- stitute, IRRI) in the late 1960s. This was widely adopted (Table 2.3) but was subsequently displaced by other IRRI varieties such as IR20, IR36, and IR50 that were better suited to local growing conditions. 18 THE GREEN REVOLUTION RECONSIDERED TABLE 2.3 Area Under HYV Paddy, North Arcot District Year 1950/51 1960/61 1966/67 1970/71 1975/76 1980/81 1982/83 1983/84 Area under Paddy (ha) 117,387 251,766 301,107 294,428 241,298 135,825 118,280 265,015 Area under HYVs (ha) 10,268 60,917 112,541 121,482 108,297 247,206 Percent of Paddy Area under HYVs 3.41 20.69 46.64 89.44 91.56 93.28 Source: Joint director of agriculture, Vellore. During the 1970s, national and state programs began to release HYVs of their own, many of which were based on crosses using IRRI plant material. Of the 38 paddy varieties developed and released in Tamil Nadu during the decade beginning in the mid-1970s, 23 of them had IRRI varieties in their parentage. IRRIGATION. The adoption of HYVs coincided with a rapid expan- sion in the number of irrigation wells in the region, from 179,232 in 1965/66 to 301,116 in 1983/84, This increase facilitated the year-round growing of paddy and freed up land during the main rainy season (samba) to enable an expansion in the area of groundnuts grown (Table 2.2). The number of mechanized wells—electric and oil pumpsets— also doubled over this period, and by the early 1980s over half the wells were mechanized. FERTILIZER. The consumption of chemical fertilizer within the re- gion increased sixfold between 1965/66 and 1984/85, from 5,177 to 30,024 metric tons of nutrients (Fertilizer Association of India, various issues). Nitrogen consumption increased from 3,198 to 17,032 metric tons. Data from the Cost of Cultivation of Principal Crops (CCPC) sur- veys conducted by TNAU for the Ministry of Agriculture show that fertilizer is used more intensively on HYVs than on improved local varieties (Tables 2.4 and 2.5). It is also used most intensively during the irrigated navarai season. Nearly all paddy receives an application of basal fertilizer at trans- planting, but subsequent nitrogen applications (topdressings) are done sequentially, and depend on the health of the crop, the availability of TABLE 2.4 Costs and Returns from Improved Local Varieties of Paddy (1973/74 prices) Yield (kg/ha) Price (Rs/kg) Value output (Rs/ha) Variable costs (Rs/ha) Seed Manures Fertilizers Pesticides Hired labor Hired bullocks Hired machines Other Gross margin (Rs/ha) Total labor (hours/ha) 1972/73 2,042 1.05 2,148 948 119 64 126 4 400 46 168 21 1,200 1,824 1973174 2,267 0.95 2,158 582 84 56 99 —297 21 1 24 1,576 2,129 1974175 2,941 1.22 3,592 723 124 33 169 3 349 21 11 13 2,869 2,081 1975176 2,763 0.89 2,467 769 145 43 233 8 290 28 2 20 1,698 2,046 1976177 3,148 1.01 3,172 1,175 148 219 261 12 439 44 11 41 1,997 2,263 1977178 2,537 0.95 2,406 787 103 83 159 13 358 37 6 28 1,619 1,507 1978/79 2,364 0.97 2,281 1,024 127 55 233 17 446 38 92 16 1,257 1,703 1979/80 2,793 0.91 2,527 811 122 8 157 — 464 47 — 13 1,716 1,973 1980/81 2,368 1.07 2,529 902 93 66 210 6 487 21 3 16 1,627 1,676 1981/82 3,364 0.82 2,767 1,138 216 25 325 13 389 41 122 7 1,629 1,820 1982/83 3,009 0.90 2,722 664 132 14 163 10 316 17 11 1 2,058 1,557 Source: Cost of Cultivation of Principal Crops data, TNAU. Note: Costs and returns based on planted area and averaged over seasons. to o TABLE 2.5 Costs and Returns from HYV Paddy (1973/74 prices) Yield (kg/ha) Price (Rs/kg) Value output (Rs/ha) Variable costs (Rs/ha) Seed Manures Fertilizers Pesticides Hired labor Hired bullocks Hired machines Other Gross margin (Rs/ha) Total labor (hours/ha) 1972/73 2,588 1.02 2,647 1,179 113 73 242 14 483 52 182 20 1,468 1,969 1973/74 2,747 0.94 2,581 817 90 66 184 12 401 25 15 24 1,764 2,338 1974175 3,637 1.21 4,389 845 103 38 219 15 409 37 8 16 3,544 1,955 1975/76 3,239 1.02 3,292 1,067 126 88 340 29 417 40 9 18 2,225 2,226 1976/77 3,746 1.02 3,805 1,986 203 153 600 55 578 43 318 36 1,819 2,295 1977/78 3,022 1.02 3,101 1,175 118 116 284 22 486 41 85 23 1,926 1,891 1978/79 2,772 1.06 2,941 1,240 133 104 325 24 447 34 157 16 1,701 1,816 1979/80 2,835 0.99 2,805 969 114 90 199 15 451 36 50 14 1,836 2,092 1980/81 3,234 1.07 3,453 1,114 89 85 347 23 506 25 19 20 2,339 1,787 1981/82 3,249 0.90 2,908 1,246 138 50 463 33 399 10 149 4 1,662 1,692 1982/83 3,035 1.04 3,168 1,068 139 31 384 7 460 30 14 3 2,100 1,899 Source: Cost of Cultivation of Principal Crops data, TNAU. Note: Costs and returns based on planted area, and averaged over seasons. North Arcot and the Green Revolution 21 water, and so on. For this reason there is a noticeable variation in the amounts of nitrogen used from year to year (Tables 2.4 and 2.5). MECHANIZATION. In addition to an increase in the mechanization of water lifting, the use of power sprayers and power-operated threshers has also expanded. There were, respectively, 925 and 228 such machines in 1982, compared with zero in 1966. A new set of entrepreneurs who own these machines has emerged in the region, and they hire out their services at fixed rates. Land preparation is, with few exceptions, still performed with labor and bullock power. However, there were 529 four-wheel tractors in the study region in 1982, compared with 114 in 1966. Their continued spread does not seem likely, given the predominance of small-scale farmers. Mechanization has led to a modest trend decline in total labor use per hectare of paddy, for both HYVs and improved local varieties (Tables 2.4 and 2.5). But on average, HYVs use about 5 to 10 percent more labor per hectare. Changes in the Profitability of Paddy Production The changes that took place in paddy technology have potentially broader implications for farm incomes than the ensuing changes in per hectare costs and returns. For example, the combination of increased irrigation and the availability of quicker-maturing varieties enabled farmers to crop a larger gross area, the increase in which was not all necessarily devoted to paddy. In this section we shall be concerned only with per hectare profitability; the larger issues of changes in total farm production and incomes are taken up in Chapter 3. YIELDS. As we saw earlier, the region's average paddy yield has grown at about 3 percent per year since the early 1960s, with a sharp jump in the 1970s (Figure 2.1). The CCPC data in Tables 2.4 and 2.5 show that the HYVs were distinctly higher yielding than the improved local varieties when first widely adopted in the early 1970s (about 20 percent higher), but their yields have not increased much since then. Moreover, the yield differential between HYVs and improved local varieties diminished over the years as local research stations incorpo- rated additional features of the HYVs into their own genetic material. COSTS. HYVs are more input intensive than local varieties, with total variable costs averaging about 20 to 25 percent higher per hectare (Tables 2.4 and 2.5). These higher costs are attributable to the more intensive use of fertilizers, pesticides, and hired labor. Total variable 22 THE GREEN REVOLUTION RECONSIDERED costs in constant prices show a modest trend increase over the years for both HYVs and improved local varieties. GROSS MARGINS. While there is considerable variation between years, paddy gross margins (gross revenue less variable costs) show little trend over the years when measured in constant prices (Figure 2.2). Paddy prices barely kept pace with inflation, and the costs of production, particularly fertilizer, increased sufficiently to offset the gains from increased yields (Tables 2.4 and 2.5). The HYVs have gen- erally proved more profitable than the improved local varieties on a per hectare basis (Figure 2.2). Primary Data Sources The research in this study is predominantly based on household and firm-level surveys undertaken at different points in time. In this section Rs/ha (1973/74 prices) 4,000 - 3,000 - HYVs Improved varieties 2,000 - 1,000 1972/73 1976/77 1980/81 Fig. 2.2. Gross margins per hectare of paddy. Source: Cost of Cultivation of Principal Crops data, TNAU. North Arcot and the Green Revolution 23 we briefly review the scope of these surveys, in terms of both their geographical coverage and the kinds of variables that were monitored. Additional details about the surveys are to be found in Appendix B. The Study Region The study region adopted in our research is identical to the one defined by the earlier team from Cambridge and Madras universities (Farmer 1977). It consists of a contiguous area of six eastern taluks (Arkonam, Cheyyar, Wandiwash, Arni, Polur, and Tiruvannamalai) that lie east of the Javadi hills and south of the sandy belt along the Palar River (Figure 2.3). This area produces about three-quarters of North Arcot district's total paddy production; hence in terms of studying the impact of the green revolution, the chosen study region facilitated the efficient concentration of survey resources. A potential drawback is that the district's headquarter town of Vel- lore is not included in the study region. Given that Vellore is the largest urban center in the district with a population of 250,000, its inclusion might seem essential for any analysis that purports to trace the growth linkages from agriculture. However, it turns out that the study region is well serviced by a hierarchy of smaller towns and urban villages, and the trading links with Vellore are concentrated on relatively few, higher- order goods and services (e.g., automobile repair, selected durables, and hospital treatment) that are not widely available elsewhere (Chap- ter 10). In essence, the study region encompasses most of the places where the day-to-day transactions of the region's households are un- dertaken, and as such it defines the kind of economic watershed re- quired from a growth linkages analysis (Bell, Hazell, and Slade 1982; Hazell and Roell 1983). Agriculturally, the study region is more specialized than North Arcot district as a whole. It is primarily a rice- and groundnut-growing area with relatively small amounts of millets, sorghum, and pulses. Its man- ufacturing base is also more specialized into agroprocessing and textiles. A detailed analysis of the region's economy is to be found in Chap- ter 7. The Surveys The first set of survey data available was collected by a team from Cambridge and Madras universities in 1973/74. Despite expectations, the team found that only about 13 percent of the paddy area was planted to HYVs at the time, so the survey really approximated a pre- or early- green revolution situation. A second team from IFPRI and TNAU 24 THE GREEN REVOLUTION RECONSIDERED ANDHRA PRADESH MYSORE egamangalam •*J Vinayagapuram Veerasambanur District Boundary Taluk Boundary Roads Railway line Study area \~^) Sample Village O Fig. 2.3. Study villages and towns. Source: B. H. Farmer, Green Revolution?, p. 8. © 1977 by The Macmillan Press Ltd. undertook similar surveys in 1982/83, by which time over 90 percent of the paddy area was planted to HYVs. This was clearly a post-green revolution situation. Both surveys included a representative sample of all rural house- holds (farmers, landless farm workers, and nonagriculturalists) living in the same 11 villages. The villages were selected through sampling North Arcot and the Green Revolution 25 procedures to be representative of all the rural villages (those with populations of less than 5,000 people) in the study region (see Appendix B). These villages are Vegamangalam, Sirungathur, Duli, Vengodu, Vinayagapuram, Amudhur, Nesal (or Randam, as John Harriss prefers to call it in Chapters 4 and 6), Kalpattu, Veerasambanur, Meppathurai, and Vayalur.1 Their locations are shown in Figure 2.3. The Cambridge-Madras universities survey in the 11 villages had several components, each involving different samples and question- naires (see Chambers et al. 1977). But the data used in this study were taken almost exclusively from two components. The first was a sample of 161 paddy-farm households that participated in a detailed farm man- agement survey for three consecutive seasons ending with the 1974 sornavari crop. The second component was a household sample of 57 paddy farmers, 3 nonpaddy farmers, and 77 noncultivators who par- ticipated in a monthly income and expenditure survey between April 1973 and May 1974. The 57 paddy-farm households were a subsample of the 161 paddy farmers included in the farm management survey. Between them these surveys provided detailed information on most aspects of farm management, employment, sources of income, house- holds assets, food consumption, and household expenditure patterns. The IFPRI-TNAU survey in the rural villages covered a sample of 345 households that participated in a monthly income, expenditure, and farm management survey from March 1982 to April 1983. The sample contained 160 paddy cultivators, 25 nonpaddy cultivators, and 160 noncultivating households (of which about three-quarters were landless laborers). While the survey was conducted in the same 11 villages as the Cambridge-Madras survey, it was not possible to use the same sample of households without losing representation in the post- green revolution situation. To enhance the comparability of the rural household data between the two surveys, the same household and variable definitions were used wherever possible. For example, a cultivator was defined as a farmer operating more than one-fourth acre and a paddy farm as a holding of one-fourth acre or larger on which paddy was or could be grown. Parts of the 1973/74 questionnaires were also used in 1982/83, although they were preceded to take advantage of interim advances in data-processing technology. Members of the earlier Cambridge-Madras universities team also provided advice and visited several of the villages while the 1982/83 survey was ongoing. 1. The Cambridge-Madras team surveyed an additional village, Dusi, which was selected purposively and not as part of the random sample. Apart from chapters 4 and 6, the Dusi data were not used in this study, and they are excluded from the description of the survey procedures and sample sizes. 26 THE GREEN REVOLUTION RECONSIDERED A potential hazard with repeat surveys of this kind is that weather conditions, which remain largely unknown until after a survey has begun, may not prove comparable between years. If they are not, then serious problems can arise in determining how much of the observed changes in the survey data are attributable to the green revolution and how much is simply the effect of different weather conditions. As shown in Table 2.2, annual rainfall was only 732 millimeters in 1973/74, or 35 percent below average. But because rainfall in the pre- ceding two years had been quite normal, there were sufficient tank and groundwater reserves that aggregate paddy area and production de- clined only marginally. At 751 millimeters, annual rainfall was almost identical in 1982/83. However, this time the region was still recovering from the effects of a severe drought in 1980/81 and below-average rainfall in 1981/82, which together had depleted the water reserves available at the beginning of the 1982/83 agricultural year. As a result, paddy area and production fell to nearly half their normal levels in 1982/83, and the region entered a state of economic distress. In fact the situation deteriorated sufficiently that government relief schemes, such as the National Employment Program, were activated in the region during the period of survey. Given the obvious difficulties in comparing survey data between 1973/74 and 1982/83, an additional survey was undertaken in 1983/84. This proved to be an above-average year for rainfall (1,272 mm), and paddy area and production recovered to more normal levels (Table 2.2). But available resources for the 1983/84 survey were very limited, and it proved necessary to limit the survey to those villages surveyed in the previous year that had been most affected by the drought. These villages are Duli, Vayalur, Veerasambanur, Meppathurai, and Amu- dhur. Not surprisingly, they are the villages with the poorest and least reliable supplies of irrigation water (see next section and Chapter 6). Within these villages, half of the 1982/83 sample of paddy cultivators and landless laborers were selected at random for resurvey, and all of the 1982/83 sample of nonpaddy cultivators and nonagriculturalists. The same monthly questionnaire was used as in 1982/83, spanning the period September 1983 to June 1984. In order to obtain information for the complete agricultural year, households were also asked to recall infor- mation for July and August when first interviewed in September. The Cambridge-Madras universities study was less focused on ag- ricultural growth linkages than the present study and, apart from a survey of small businesses in the single town of Ami (see Chapter 9), surveys of the nonrural economy were not undertaken in 1973/74. In contrast, a major effort to study the nonfarm economy was undertaken in 1982/83 that included a monthly income and expenditure survey of North Arcot and the Green Revolution 27 320 urban households, a survey of 1,500 nonfarm businesses located in urban areas, and a survey of the patterns of service provision and use in all the villages in the study region that had populations of more than 750 persons (see Appendix B and Chapter 10 for details). Additionally, the monthly questionnaire for the rural household survey included de- tails about any nonfarm business activities that the sample households undertook, and a repeat survey of small businesses in Ami was un- dertaken (see Chapter 9). Characteristics of the Sampled Rural Villages There are considerable differences among the 11 sampled villages, particularly with respect to population, land and water resources, eco- nomic activities, infrastructure, labor, and social relations in produc- tion. A detailed analysis of the intervillage variations is offered by John Harriss in Chapter 6. This section presents a very brief account of the major features of the 11 villages. In 1982/83, the 11 villages had an average population of 959, ranging from 538 in Duli to 1,487 in Nesal. Nesal, Kalpattu, Vengodu, Vega- mangalam, and Sirungathur are the largest villages, with populations in excess of 1,000. The major castes are Vanniyas, Pillai, Naidus, Mu- daliars, Yadavas, and Harijans. All the villages have a primary school, and Amudhur has a high school. Unlike the other villages, Meppathurai and Vinayagapuram do not have a bus service, but one is available within three kilometers. All the villages have electricity and, apart from Vinayagapuram, are con- nected by surfaced roads. A detailed account of the infrastructure fa- cilities available in each village is to be found in Chapter 10. As in the region generally, tanks and wells are the principal sources of irrigation in the study villages (Table 2.6). Kalpattu and Vegaman- galam are unique in not having tanks. Kalpattu is surrounded by hills that recharge its wells with groundwater all year round. Because of this feature, the village is able to grow crops continuously and its cropping pattern is the most diversified; it includes paddy, banana, turmeric, sugarcane, groundnut, and horticultural crops. Vegamangalam village is supplied with water from a natural spring and also enjoys year-round irrigation. Because of good irrigation resources, Kalpattu, Vegaman- galam, and Nesal are comparatively prosperous villages and are less prone to drought. Duli, Vayalur, Veerasambanur, Meppathurai, and Amudhur have the least reliable sources of irrigation water, and they were severely affected by drought in 1982/83. The sample villages use labor from both within and outside the village. Sirungathur, Veerasambanur, Vengodu, and Amudhur are 28 THE GREEN REVOLUTION RECONSIDERED TABLE 2.6 Irrigation Facilities in Rural Study Villages, 1982 Village Kalpattu Meppathurai Vayalur Veerasambanur Vinayagapuram Nesal Amudhur Vengodu Duli Sirungathur Vegamangalam No. of Tanks 0 1 1 3 2 3 2 3 1 2 0 No. of Wells3 194 159 87 130 109 227 86 134 38 98 69 No. of Pumpsetf 124 69 37 41 73 161 59 75 23 75 60 Average Depth of Wells' (meters) 15.67 11.00 12.18 12.18 15.84 14.21 9.32 12.81 12.54 12.75 10.21 Percent of Households with Access to Irrigation Wells 100.0 95.0 100.0 82.0 100.0 73.7 87.5 94.1 58.3 100.0 50.0 Source: Information collected from village-level development workers and village administrative officers. "Some wells were not in use at the time of the survey. bWells without pumpsets generally have poor water supplies and the water is lifted by mhote. 'Depth of wells was determined from a random sample of nine wells in each of the villages. labor-surplus villages, whereas Kalpattu, Nesal, and Vegamangalam are labor-deficit villages. Some sharecropping is found in Vinayaga- puram and Vegamangalam, but it is unimportant in the other villages. All the study villages have cooperative credit societies that provide crop loans. The sample villages also benefit from the presence of gov- ernment-run fair-price shops, which provide rice, vegetable oils, sugar, and kerosene at subsidized prices. Various state-run developmental programs also benefit the study villages, for example, the Noon Meal Scheme, Integrated Rural De- velopment Programs, and Training and Visit Extension. Village pan- chayats are responsible for local water supply, road maintenance, and health programs. Milk producers' cooperative societies also function in the study villages. CHAPTER 3 Economic Changes among Village Households Peter B. R. Hazell, C. Ramasamy, V. Rajagopalan, P. K. Aiyasamy, and Neal Bliven IN THIS CHAPTER we use the village household survey data to quantify the effects of the green revolution on farm production, income, and employment; the changes in family income and consumption of farm and nonfarm households; and the changes in the distribution of land. There are four problems with the data set that complicate our task. First, 1973/74, the year of the Cambridge-Madras universities (CMU) survey, was not a true pre-green revolution year. Official gov- ernment data show that about 40 percent of the paddy area was planted to high-yielding varieties (HYVs) that year. The CMU survey found a considerably lower adoption rate (13 percent of the cropped area), but most farmers were growing locally improved varieties that already had some of the key features of HYVs (see Chapter 2). In the absence of a base year in which only long-strawed, traditional varieties are grown, the prospective gains from the green revolution to be observed in the survey data are bound to be muted. Second, our survey data for 1973/74, 1982/83, and 1983/84 are not strictly comparable as far as rainfall and irrigation water reserves are concerned. Rainfall was similar in 1973/74 and 1982/83 (about 35 per- cent below average), but since water reserves in the tanks and ground- water were much lower in 1982/83 because of an extended drought, the impact on paddy production was much greater (see Chapter 2). Re- gional paddy production was 40 percent lower in 1982/83 than in 1973/ 74, so it is difficult to say much about the impact of the green revolution between these two years. On the other hand, rainfall was 15 percent above average in 1983/84, and regional paddy production was 40 percent larger than in 1973/74. A simple comparison of 1973/74 and 1983/84 may overstate the effects of the green revolution between these two years. To compound these weather-related problems, we have access to 29 30 THE GREEN REVOLUTION RECONSIDERED regionally representative household data only for 1973/74 and 1982/83. In 1983/84, the survey was confined to a subsample of households lo- cated in villages with the poorest water resources. These villages suf- fered the most during the drought of 1982/83 and, most likely, also in 1973/74. Because of their more limited access to irrigation water, they also are likely to have benefited the least from the interim changes in paddy technology. Nevertheless, comparisons between 1973/74 and 1983/84 in these "resurvey" villages provide our best basis for measuring changes in the economic welfare of the rural households. Third, there is considerable variation in the economic conditions among the 11 sample villages. Some have only limited access to irri- gation water, and supplies are unreliable (e.g., Duli). Others are blessed with generous and stable supplies of water, even in drought years (e.g., Kalpattu). This not only leads to important differences in the potential benefits obtainable from improved paddy varieties, but also determines the very economic and social fabric of the villages and the types of growth that are possible. Poorly endowed villages tend to be less equitable to begin with, and technical change is likely to induce less equitable growth there than in better-endowed villages. In this chapter we exploit the statistical representation of the sample data to analyze changes in the average welfare of different types of households at a pooled village level. We leave it to John Harriss in Chapters 4 and 6 to analyze the changes by type of village, and to relate these changes to the underlying water resource endowments. Fourth, as shown below, the 1973/74 farm sample has a much smaller percentage of farms larger than 1 hectare in the resurvey villages than do the 1982/83 and 1983/84 samples: 1973174 1982/83 1983184 Resurvey villages Nonresurvey villages All villages 25.0 45.9 38.6 63.6 49.5 55.2 66.7 n.a. n.a. These figures are not consistent with other data on changes in the farm size distribution (see Chapter 4 and later sections of this chapter). Nor are they consistent with the village listing (census) data collected by CMU in 1973 and by IFPRI-TNAU in 1982; these put the percentage of farms greater than 1 hectare at 42 and 48 percent, respectively, for the resurvey villages. Since the sampling design did not involve a strat- ification by farm size, the problem seems to be one of unlucky samples, a not uncommon occurrence when working with relatively small sam- ples. But an immediate consequence is that uncorrected sample means for the resurvey villages are biased toward small farms in 1973/74 and Economic Changes among Village Households 31 toward large farms in 1982/83 and 1983/84. For the most part we resolve the problem by reporting separate results for small (1 hectare or less) and large (greater than 1 hectare) farms. Where pooled estimates are reported for the resurvey villages, they are weighted means using the farm size shares observed in the 1973 and 1982 village listings (i.e., large farm weights of 0.42 for the 1973/74 survey and 0.48 for the 19821 83 and 1983/84 surveys). Of the four problems discussed above, only the first two raise un- resolved difficulties for our data analysis: the lack of a true before- green revolution sample, and uncorrected differences in rainfall and water reserves between years. In Chapter 8 we develop a regional model of the study region and use it to simulate the impact of the green revolution under normal weather conditions. Because the model can correct for weather conditions, as well as simulate with- and without- green revolution situations, it enables us to overcome the major limi- tations of our survey data analysis. However, as with any model, its construction requires a healthy dose of assumptions about the way in which the regional economy works. We shall therefore use the survey and model results to provide a check on each other, drawing comfort from instances where the two tell a consistent story. Further checks on the reliability of our findings are provided in Chapter 4, where John Harriss uses his own independently collected data to examine changes in some of the same village and household variables. Changes in Paddy and Groundnut Production Our analysis of the impact of the green revolution begins with the increases in farm production. Since most farmers engage in mixed crop- ping, our analysis must go beyond simple changes in paddy area and yield to encompass any induced changes in the production of other crops. These changes might arise from crop substitution (e.g., more paddy at the expense of other crops, or vice versa), or from the more intensive cropping of land throughout the year (e.g., HYVs and in- creased irrigation permit a greater cropped area during the navarai season). Table 3.1 shows the changes in paddy and groundnut production between the survey years. Paddy production declined between 1973/74 and 1982/83, by 5 percent on small paddy farms and by 33 percent on large paddy farms. This is less than the 42 percent drop recorded at the district level between these two years (Table 2.2), but differences are to be expected since (1) the survey data are based on different procedures for estimating output than those used by district officials, to TABLE 3.1 Average Cropped Area, Yield, and Production of Paddy and Groundnuts by Farm Size Group Area (ha) Yield (kg/ha) Production (kg) 1973174 1982183 1983/84 1973174 1982183 1983184 1973174 1982183 1983184 All villages Paddy Small farms Large farms Groundnuts Small farms Large farms Resurvey villages Paddy Small farms Large farms Groundnuts Small farms Large farms 0.53 1.41 0.30 1.19 0.55 0.75 0.48 1.08 0.35 0.89 0.33 1.21 0.31 0.79 0.28 1.10 n.a. n.a. n.a. n.a. 0.64 2.11 0.15 0.86 2,123 2,854 1,280 1,495 1,773 2,524 1,073 1,227 3,043 3,045 897 969 2,826 2,430 782 914 n.a. n.a. n.a. n.a. 2,777 2,176 1,760 1,309 1,125 4,024 384 1,779 975 1,893 515 1,325 1,065 2,710 296 1,172 876 1,920 219 1,005 n.a. n.a. n.a. n.a. 1,777 4,592 264 1,126 Economic Changes among Village Households 33 (2) they pertain to different 12-month periods, (3) they exclude part of the district's production, and (4) they are, of course, subject to sampling error. The decline was more muted in the resurvey villages, probably because they were more affected by the 1973/74 drought than the non- resurvey villages. The decline in paddy production was entirely due to a loss in the paddy area grown (Table 3.1). Yields were actually higher in 1982/83, despite the drought. Groundnut production also declined between 1973/74 and 1982/83 in the sample villages, by 23 percent for small farms and by 34 percent for large farms. This decline is nearly all attributable to lower yields, since the planted area changed little. Unlike paddy, most groundnuts are not irrigated, and hence yields are more affected by variations in rainfall (see also the coefficient of variation calculations in Chapter 2). Paddy production recovered dramatically in 1983/84, more than dou- bling in the resurvey villages from the low levels of the 1982/83 drought. The recovery was entirely due to an increase in the paddy area. Yields actually declined a little, probably because late rains during the 1983/ 84 samba season caused some damage to the harvest. Groundnut production increased little between 1982/83 and 1983/ 84. While yields recovered after the drought, the area planted to groundnuts declined sharply on both small and large farms. This area decline was necessary to facilitate the expansion of the paddy area, as confirmed by the cropping pattern data in Table 3.2. The sharp increase in the paddy area also displaced sorghum and millets. These relatively drought-resistant crops were widely grown during 1982/83, particularly in the resurvey villages (Table 3.2). Comparing 1973/74 with 1983/84, paddy production increased by 82 percent on small farms and by 143 percent on large farms. On large farms these increases were entirely due to an expansion of the paddy area, but on small farms they were predominantly due to yield increases. Large farms had already widely adopted the HYVs by 1973/74 (Chin- nappa 1977), and their yields were already one-third higher than small- farm yields (Table 3.1). While large-farm yields were still the same in 1983/84, small farmers adopted HYVs during the interim and closed the yield gap. In fact small-farm paddy yields were actually 27 percent higher than large-farm yields in 1983/84 in the resurvey villages (Table 3.1). Increases in the paddy crop area between 1973/74 and 1983/84 were achieved largely at the expense of groundnuts (Table 3.2). There was no accompanying increase in the cropping intensity index, which in 1983/84 stood at 1.62 for small farms and 1.32 for large farms. Part of the increase in the paddy area on large farms was also attributable to a 63 percent increase in their operated farm size. Small farms, on the TABLE 3.2 Cropping Patterns by Farm Size Group (ha) All villages Paddy Groundnuts Sorghum/millets Other crops Total Operated farm size Crop intensity index Resurvey villages Paddy Groundnuts Sorghum/millets Other crops Total Operated farm size Crop intensity index 1973174 0.53 0.30 0.05 0.05 0.93 0.59 1.57 0.55 0.48 0.01 0.03 1.07 0.64 1.67 Small Farms 1982/83 0.35 0.33 0.06 0.09 0.83 0.61 1.36 0.31 0.28 0.14 0.06 0.80 0.66 1.21 1983184 n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0.64 0.15 0.08 0.02 0.89 0.55 1.62 1973/74 1.41 1.19 0.10 0.05 2.74 2.42 1.13 0.75 1.08 0.24 — 2.08 1.48 1.41 Large Farms 1982/83 0.89 1.21 0.34 0.54 2.99 2.68 1.11 0.79 1.10 0.45 0.33 2.68 2.59 1.04 1983184 n.a. n.a. n.a. n.a. n.a. n.a. n.a. 2.11 0.86 0.01 0.20 3.18 2.41 1.32 Economic Changes among Village Households 35 other hand, shrank by 14 percent in size. These results pertain only to the resurvey villages and should not be extrapolated to the entire study region (see a later section in this chapter on changes in the farm size distribution). What, then, can we conclude about the productivity effects of the green revolution between 1973/74 and 1983/84? Paddy production did increase sharply, but this seems to have been more the result of sub- stituting paddy for groundnuts in the cropping pattern. Yields increased on small farms, and this reflects the fact that they adopted HYVs during the period studied. But most large farms had already adopted HYVs by 1973/74, hence their yield increases are not captured in the survey data. Despite the expansion of irrigation between 1973/74 and 1982/83, there is no evidence that this led to any increase in annual cropping intensity. However, our 1983/84 results are confined to the resurvey villages, and these villages participated the least in the general expan- sion in irrigation. The substitution between paddy and groundnuts could also be partly driven by technological change, especially if improved paddy varieties have lower costs and this increases the relative profitability of paddy. On the other hand, part of the substitution appears to be related to weather conditions. Groundnuts need less water than paddy and hence, along with millets and sorghum, are more favored by farmers during drought years. Irrespective of cause, the observed changes in the cropping pattern are considerable, and they have important implications for changes in farm income and employment. Changes in Farm Income The value of farm output (gross output) is dominated by paddy and groundnuts (Table 3.3), hence changes in their production and prices largely determine the changes in total gross output. We have already seen that paddy and groundnut production changed markedly from one survey year to another (Table 3.1). However, these changes were par- tially buffered by opposing price movements, as shown below (in Rs/t, constant 1973/74 prices): Paddy Groundnut 1973/74 865 1,080 1982/83 922 1,547 1983/84 755 1,613 TABLE 3.3 Paddy Farm Incomes (1973/74 prices) All Villages 1973/74 Gross output8 Paddy Groundnuts Other crops Livestock products Bullock & machine rent Total gross output Costs Hired labor Bullock & machine hire Fertilizers Other Total costs Net farm income Small Farms 954 431 757 23 — 2,165 235 27 71 660 992 1,173 Large Farms 3,581 1,803 1,153 88 — 6,625 598 99 303 1,546 2,546 4,078 1982183 Small Farms 982 458 162 129 34 1,765 176 33 103 222 533 1,232 Large Farms 2,499 1,812 264 200 41 4,816 814 96 351 632 1,893 2,923 1973174 Small Farms 808 574 39 5 — 1,426 213 23 35 429 700 726 Large Farms 1,546 994 1,223 91 — 3,854 424 76 75 959 1,534 2,320 Resurvey Villages 1982183 Small Farms 808 339 40 70 18 1,275 84 8 27 110 228 1,047 Large Farms 1,771 1,554 102 204 4 3,635 707 83 236 444 1,470 2,165 1983/84 Small Farms 1,341 426 64 179 3 2,013 276 20 227 385 908 1,105 Large Farms 3,464 1,816 176 820 4 6,280 1,331 48 809 1,208 3,396 2,884 "Includes home-consumed foods valued at farm-gate prices. Economic Changes among Village Households 37 Both paddy and groundnut production fell sharply between 1973/ 74 and 1982/83 (Table 3.1), but these losses were partially offset by higher prices. On the other hand, the sharp increase in paddy output between 1982/83 and 1983/84 was tempered by a drop in price, while the limited growth in groundnut output was supplemented by a price increase. As a result, changes in paddy and groundnut gross output are less dramatic than the changes in production. For example, small and large farms in the resurvey villages increased their paddy production by 103 percent and 139 percent, respectively, between 1982/83 and 1983/ 84, but paddy gross output increased by only two-thirds as much—by 66 and 95 percent, respectively. Changes in farm costs also acted to buffer the changes in production (Table 3.3). Costs were lowest during the 1982/83 drought and increased substantially in 1983/84. These buffering effects led to surprisingly stable farm incomes from one year to another. In the resurvey villages, small- farm incomes were only 5 percent lower during the 1982/83 drought than in 1983/84, while large-farm incomes were 25 percent lower. Taking 1973/74 and 1983/84 as our relevant comparison for evalu- ating the green revolution, Table 3.3 shows that small-farm incomes increased by 52 percent and large-farm incomes by 24 percent. The increases would have been greater but for the 13 percent decline in the paddy price and the sharp increase in labor and fertilizer costs, espe- cially for large farms. Changes in Employment and Wages Total labor use in crop production declined sharply during the 1982/83 drought on both large and small farms, but it rebounded in 1983/84 (Table 3.4). Between 1973/74 and 1983/84, total crop employment in the resurvey villages increased by 15 percent on large farms but declined by 20 percent on small farms. However, since operated farm sizes also changed (Table 3.2), a more useful measure is labor use per hectare of operated land. This declined by 7 percent for small farms and by 30 percent for large farms. Similar declines occurred in the labor use per gross cropped hectare (Table 3.4). Note that the small farms use con- siderably more labor than large farms on an operated area basis, but they use about the same amount per hectare of gross cropped land. This simply reflects their more intensive cropping of land (Table 3.2). Overall, these results suggest that the green revolution did little to increase total crop employment, despite the fact that HYV paddy uses about 5 to 10 percent more labor per hectare than locally improved varieties (Tables 2.4 and 2.5). Using the farm size weights discussed in oo TABLE 3.4 Adult Employment per Paddy Farm in Crop Production by Type of Labor (days) All Villages 1973174 Family labor Male Female Total Attached labor Male Female Total Exchange labor Male Female Total Hired labor Male Female Total Total labor Male Female Total Total labor/ha Operated area Gross cropped area Total hired labor/ha Operated area Gross cropped area Small Farms 95.4 34.3 129.7 5.6 0.4 6.0 3.9 17.4 21.3 39.2 98.9 138.1 144.1 151.0 295.1 500 317 234 148 Large Farms 148.5 58.5 207.0 46.6 — 46.6 12.8 37.3 50.1 125.5 234.7 360.2 333.4 330.5 663.9 274 242 149 131 1982/83 Small Farms 52.2 11.9 64.1 3.0 — 3.0 0.1 0.4 0.5 19.2 57.0 76.2 74.5 69.3 143.8 236 173 125 92 Large Farms 96.7 18.3 115.0 31.1 0.4 31.5 0.6 3.2 3.8 72.5 215.0 287.5 200.9 236.9 437.8 163 146 107 96 1973174 Small Farms 94.4 37.3 131.7 10.8 — 10.8 2.5 7.8 10.3 40.6 89.0 129.6 148.3 134.1 282.4 441 264 203 121 Large Farms 177.4 86.6 264.0 — — — 9.0 24.7 33.7 97.3 170.3 267.6 283.7 281.6 565.3 382 272 181 129 Resurvey Villages 1982183 Small Farms 54.1 10.4 64.5 — — — — — — 8.4 33.6 42.0 62.5 44.0 106.5 161 133 64 53 Large Farms 88.5 15.3 103.8 21.7 0.3 22.0 0.2 1.7 1.9 63.8 154.6 218.4 174.2 171.9 346.1 134 129 84 81 1983184 Small Farms 104.8 43.0 147.8 — — — 0.2 0.3 0.5 28.7 48.6 77.3 133.7 91.9 225.6 410 253 141 87 Large Farms 191.9 111.1 303.0 68.3 15.3 83.6 13.7 3.3 17.0 100.6 143.7 244.3 374.5 273.4 647.9 269 204 101 77 Economic Changes among Village Households 39 the introduction to this chapter, total labor use in crop production declined by 4 percent (from 401 to 386 days) on the average paddy farm between 1973/74 and 1983/84, and by 14 percent per hectare of operated land. These changes reflect the increased mechanization of irrigation pumping and paddy threshing, changes that were not ade- quately offset by the substitution of paddy for groundnuts in the crop- ping pattern (groundnuts use about half to two-thirds as much labor per hectare). Male and female labor were about equally employed in crop pro- duction in 1973/74 and 1982/83, but males dominated in 1983/84 (Table 3.4). Hired labor was also about as important as family labor in 1973/ 74, but its share in total labor use fell in 1983/84, from 47 percent to 34 percent on small farms, and from 46 to 38 percent on large farms. The total use of hired labor fell by 25 percent on the average farm, or by 11 percent if attached labor is included. Although total crop employment did not increase between 1973/74 and 1983/84, real wage rates did (Table 3.5). Wage rates differ by sex and task, but of the more important tasks, wages for males increased by about 20 percent in real terms, while wages for females increased by only half as much. These increases are observed for both cash and kind wages. They are also generally consistent with the changes re- ported by John Harriss in Chapter 4. Real wage rates could only have increased in the face of an 11 percent decline in the use of hired and attached labor on the average farm if the supply of labor had also contracted. Table 3.6 shows that, at least in the resurvey villages, this is exactly what happened. Large- farm households sharply curtailed their participation in the agricultural labor market, and their wage earnings (in constant prices) fell from Rs 228 in 1973/74 to Rs 47 in 1983/84. Given also a sharp increase in wage payments on these farms, net wage earnings (earnings less payments) fell from Rs -196 to Rs -1,284. Small-farm households met part of the increase in the labor deficit on large farms (their net wage earnings increased from Rs 176 to Rs 464), but this was insufficient to fill the gap. Taking a weighted average of small and large farms, net wage earnings fell from Rs 20 in 1973/74 to Rs -375 in 1983/84. As we shall see in the next section, the gap was filled by landless laborers and nonpaddy farmers. As discussed by John Harriss in Chapter 4, other factors that helped to increase wages were a growing demand for labor in rural nonfarm activities; an expansion of local milk production under Operation Flood; and an outmigration of workers that helped contain the growth in the number of landless laborers. TABLE 3.5 Agricultural Wages by Operation Plowing (Men) Village Kalpattu Nesal Vegamangalam Vinayagapuram Sirungathur Vayalur Vengodu Meppathurai Duli Veerasambanur Amudhur Average 1973/74 (Rsl day) 3.50 1.50 3.00 2.00 2.50 2.50 2.00 2.80 2.00 0.75 2.00 2.23 1982183 (Rsl day) 6.00 6.00 5.00 5.00 5.00 5.00 5.00 5.00 6.00 5.00 3.00 5.10 Deflated 1982/83 Wage3 3.25 3.25 2.71 2.71 2.71 2.71 2.71 2.71 3.25 2.71 1.63 2.76 Percent Change -7.0 116.7 -9.7 35.5 8.4 8.4 35.5 -3.2 62.5 261.3 -18.5 23.8 1973174 (Rsl day) 1.00 1.30 1.00 0.70 1.00 1.00 1.50 1.30 1.50 1.25 1.30 1.17 Weeding (Women) 1982183 Deflated (Rsl 1982/83 day) Wage1 2.00 3.00 3.00 2.50 2.00 2.50 2.00 2.00 3.00 2.50 2.50 .09 .63 .63 .36 .09 .36 .09 .09 .63 .36 .36 2.45 1.33 Harvesting (Women) Percent Change 9.0 25.4 63.0 94.3 9.0 36.0 -27.3 -16.2 8.7 8.8 4.6 13.7 1973174 (kg/paddy /day) 5.00 4.25 5.60 3.00 5.60 3.50 5.00 4.25 5.00 4.25 3.50 4.45 1982/83 (kg/paddy /day) 5.60 5:00 5.60 4.00 5.60 4.00 5.00 5.00 5.60 4.25 4.80 4.95 Percent Change 12 18 — 33 — 14 — 18 12 — 37 11 Threshing (Men) 1973174 (kg/paddy /day) 5.00 8.50 5.60 3.25 5.60 7.00 4.25 6.25 1.40 8.50 5.00 5.49 1982/83 (kg/paddy /day) 8.50 10.00 5.60 4.00 5.60 7.00 5.00 7.00 5.60 8.50 6.40 6.65 Percent Change 70 18 — 23 — — 18 12 400 — 28 21 Source: Cambridge-Madras and IFPRI-TNAU surveys. 'Deflated to 1973/74 prices with a rural cost-of-living index for Kunnathur village in Chingleput (Tamil Nadu—An Economic Appraisal, various volumes, Finance Department, Government of Tamil Nadu). Economic Changes among Village Households 41 TABLE 3.6 Agricultural Wage Transactions by Size of Farm, Resurvey Villages (1973/74 Rs) Wage earnings Wage payments Net earnings 1973174 389 213 176 Small Farms 1982/83 200 84 116 1983/84 740 276 464 1973174 228 424 -196 Large Farms 1982/83 26 707 -681 1983/84 47 1,331 -1,284 Changes in Family Income The total family income of each household type is summarized by year in Table 3.7. The detrimental impact of the 1982/83 drought is again apparent. Most households were actually worse off in 1982/83 than in 1973/74, despite the intervening changes in irrigation and paddy tech- nology. There was less change in the resurvey villages between these two years, but family incomes in these villages were already well below average in 1973/74. The favorable rains of 1983/84 led to substantial increases in family incomes in the resurvey villages; in many cases incomes more than doubled. Clearly, the income of rural households in North Arcot is very vulnerable to fluctuations in rainfall. Family incomes improved substantially between 1973/74 and 1983/ 84 in the resurvey villages. It increased by 18 percent for large paddy farms and by 90 percent for small paddy farms. The landless laborers more than doubled their income (up 125 percent), bringing their aver- age income almost up to the level of small paddy farmers. The nonpaddy farmers and the nonagricultural households increased their incomes by about 17 and 55 percent, respectively, even though they were not di- rectly involved in the changes that occurred in paddy technology. These changes are more substantial than the income increases ob- tained from the regional model in Chapter 8 (see Table 8.4 and the accompanying discussion). When normalized for rainfall, the model predicts that the green revolution increased the incomes of farmers and landless laborers by about 30 percent and of nonagriculturalists by 20 percent. It would seem that the survey results were affected by the 1973/74 drought, especially as our 1973/74 to 1983/84 comparison is limited to the poorer and more drought-prone resurvey villages, and this has exaggerated the income increases. However, both the model and the survey are consistent in showing that (1) small farmers increased their incomes in at least the same proportion as large farmers, (2) landless laborers gained favorably compared with any other group, and (3) nonagricultural households became relatively poorer. The details of the composition of family incomes are reported in 42 THE GREEN REVOLUTION RECONSIDERED TABLE 3.7 Changes in Household Incomes (1973/74 Rs) Type of Household Small paddy farms Large paddy farms Nonpaddy farms Landless laborers Nonagricultural households All Villages 1973/74 2,028 (100) 4,553 (100) 1,927 (100) 988 (100) 1,240 (100) 1982/83 1,866 (92) 3,778 (83) 1,135 (59) 807 (82) 1,480 (119) Resurvey Villages 1973174 1,199 (100) 2,764 (100) 1,732 (100) 935 (100) 1,187 (100) 1982/83 1,605 (134) 2,809 (102) 845 (49) 912 (98) 792 (67) 1983/84 2,286 (190) 3,268 (118) 2,032 (117) 2,102 (225) 1,837 (155) Note: Figures in parentheses are income indices with 1973/74 = 100. Tables 3.8 to 3.12. Farm income accounts for only about 50-60 percent of the family income of small-paddy-farm households (Table 3.8). Wage earnings, particularly in agriculture, account for another third, though these were considerably reduced during the drought year. Increased wage earnings also accounted for 45 percent of the increase in family income between 1973/74 and 1983/84 for small paddy farms in the re- survey villages. Large-sized paddy farms receive over 80 percent of their income from farming (Table 3.9), hence the changes in net farm income dis- cussed earlier explain most of the changes in their family income. Nonpaddy farmers have become more dependent on wage earnings since 1973/74 (Table 3.10). Their farm income was almost nonexistent in the resurvey villages during the 1982/83 drought. Agricultural wage earnings were the most important source of income in 1983/84, but nonfarm wage earnings were more important during the drought. Self- employment in nonfarm business activities has also become more im- portant since 1973/74; it accounted for 18 percent of family income in 1983/84. The landless workers are almost totally dependent on wage earnings, particularly in agriculture (Table 3.11). They suffered from the con- traction in agricultural employment during the drought but, more gen- erally, benefited substantially from the increase in the demand for hired labor that has occurred since 1973/74 (see previous sections). The nonagricultural households have the most diverse sources of income (Table 3.12). Own nonfarm business income is most important, and although this virtually disappeared during the 1982/83 drought, it was the prime source of increase in family income between 1973/74 and 1983/84 in the resurvey villages; it increased from Rs 493 per household Economic Changes among Village Households 43 TABLE 3.8 Composition of Family Income, Small Paddy Farms (1973/74 Rs) All Villages Source of Income Net farm income Agric. wage earnings White-collar earnings Other wage earnings Nonfarm business income Rents & money lending Other unearned income Total family income Sample size Composition 1973174 1,173 410 91 36 78 4 236 2,028 35 of Family 1982183 1,232 205 42 123 159 25 80 1,866 72 TABLE 3.9 Income, Large All Villages Source of Income Net farm income Agric. wage earnings White-collar earnings Other wage earnings Nonfarm business income Rents & money lending Other unearned income Total family income Sample size 1973174 4,078 180 138 28 — — 129 4,553 22 1982/83 2,923 79 251 70 214 51 190 3,778 88 Resurvey Villages 1973/74 726 389 24 9 — 51 1,199 15 Paddy Farms 1982183 1,047 200 2 196 60 — 100 1,605 23 (1973/74 Rs) 1983/84 1,105 740 157 20 162 102 2,286 11 Resurvey Villages 1973174 2,320 228 216 — — — 2,764 5 1982/83 2,165 26 239 44 -12 3 344 2,809 41 1983/84 2,884 47 156 — 141 40 3,268 22 44 THE GREEN REVOLUTION RECONSIDERED TABLE 3.10 Composition of Family Income, Nonpaddy Farmers (1973/74 Rs) Source of Income Net farm income Agric. wage earnings White-collar earnings Other wage earnings Nonfarm business income Rents & money lending Other unearned income Total family income Sample size Composition Source of Income Net farm income Agric. wage earnings White-collar earnings Other wage earnings Nonfarm business income Rents & money lending Other unearned income Total family income Sam