WDP33 September 1988 3 3 U World Bank Discussion Papers Land and Labor iu South Asia Inderjit Singh . d us: , w ..- ;fi-L CX#' rFe ...- RECENT WORLD BANK DISCUSSION PAPERS No. 1. Public Enterprises in Sub-Saharan Africa. John R. Nellis No. 2. Kaising School Quality in Developing Countries: What Investments Boost Learning? Bruce Fuller No. 3. A System for Evaluating the Performance of Government-Invested Enterprises in the Republic of Korea. Young C. Park No. 4. Country Commitment to Development Projects. Richard Heaver and Arturo Israel No. 5. Public Expenditure in Latin America: Effects on Poverty. Guy P. Pfeffermann No. 6. Community Participation in Development Projects: The World Bank Experience. Samuel Paul No. 7. International Financial Flows to Brazil since the Late 1960s: An Analysis of Debt Expansion and Payments Problems. Paulo Nogueira Batista, Jr. No. 8. Macroeconomic Policies, Debt Accumulation, and Adjustment in Brazil, 1965-84. Celso L. Martone No. 9. The Safe Motherhood Initiative: Proposals for Action. Barbara Herz and Anthony R. Measham [Also available in French (9F) and Spanish (9S)1 No. 10. Improving Urban Employment and Labor Productivity. Friedrich Kahnert No. 11. Divestiture in Developing Countries. Elliot Berg and Mary M. Shirley No. 12. Economic Growth and the Returns to Investment. Dennis Anderson No. 13. Institutional Development and Technical Assistance in Macroeconomic Policy Formulation: A Case Study of Togo. Sven B. Kjellstrom and Ayite-Fily d'Almeida No. 14. Managing Economic Policy Change: Institutional Dimensions. Geoffrey Lamb No. 15. Dairy Development and Milk Cooperatives: The Effects of a Dairy Project in India. George Mergos and Roger Slade No. 16. Macroeconomic Policies and Adjustment in Yugoslavia: Some Counterfactual Simulations. Fahrettin Yagci and Steven Kamin No. 17. Private Enterprise in Africa: Creating a Better Environment. Keith Marsden and Therese Belot No. 18. Rural Water Supply and Sanitation: Time for a Change. Anthony A. Churchill, with the assistance of David de Ferranti, Robert Roche, Carolyn Tager, Alan A. Walters, and Anthony Yazer (Continued on the inside back cover.) 33 ~~~ World Bank Discussion Papers Land and Labor 'n South Asila Inderjit Singh The World Bank Washington, D.C. Copyright (© 1988 The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing September 1988 Discussion Papers are not formal publications of the World Bank. They present preliminary and unpolished results of country analysis or research that is circulated to encourage discussion and comment; citation and the use of such a paper should take account of its provisional character. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Any maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning the delimitation of its boundaries or its national affiliation. Because of the informality and to present the results of research with the least possible delay, the typescript has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to Director, Publications Department at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to photocopy portions for classroom use is not required, though notification of such use having been made will be appreciated. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list and indexes of subjects, authors, and countries and regions; it is of value principally to libraries and institutional purchasers. The latest edition of each of these is available free of charge from Publications Sales Unit, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'Iena, 75116 Paris, France. Inderjit Singh is the principal economist in the China Department of the World Bank. Library of Congress Cataloging-in-Publication Data Singh, Inderjit, 1941- Land and labor in South Asia / Inderjit Singh. p. cm. -- (WorLd Bank discussion papers ; 33) BibLiography: p. ISBN 0-8213-1129-8 1. Land tenure--South Asia. 2. AgricuLtural laborers--South Asia. 3. Farms. Small--South Asia. 4. RuraL poor--South Asia. 5. South Asia--Rural conditions. I. Title. II. Series. HD860.3.Z63S559 1988 333.3'0954--dcl9 88-27647 CIP -~~~~~~~~~~~~~~~~~~~~~ ii LAND AND LABOR IN SOUTH ASIA Controversy is as alive in South Asia as elsewhere. The "Structuralists" take the view that the root causes of backwardness and poverty are found in the structure of the agrarian economy and that their remedy lies in its radical alteration. The technologists contend that the main causes of backwardness lie in the low productivity inherent in agriculture and that the remedy lies mainly in new more productive technologies and more profitable inputs. The advent of new technologies associated with the "Green Revolution" and its rapid success in some and not in other areas has only heightened this controversy, and makes an examination of changes in agrarian structure even more relevant. This paper explores the relationship of land and labor to rural poverty in South Asia. The first half discusses the unequal distribution of land among the rural population revealing how large proportions of those living in rural areas of India, Bangladesh, and Pakistan either have very limited access to land (in terms of the size of holding they owned or operate). It further elaborate on the misused and abused concept of "Landlessness" and its relationship to rural Land and labor. The second section discusses the imbalance between available employment opportunities and the rural wage labor force. People who rely mainly on wage employment and unemployment in rural areas are examined and conclusions drawn as to the different types of occupational opportunities -- agricultural and non-agricultural -- available to wage laborers. These findings have been selected from Inderjit Singh's background research on the subcontinent's rural poor. For a fuller examination, see his forchcoming publication "The Creat Ascent: The Rural Poor of South Asia". The book expands on the theme that technological change in agriculture and the increase in factor productivity that comes from it are a major positive forces for the alleviation of poverty in South Asia. * I am indebted to Tribhuwan Narain who provided invaluable assistance in the research for this paper. -v - CONTENTS PAGE Access to Land: Patterns of Land Holding and Landlessness.....1 Two Theories on Agrarian Change ............................... 2 A. Structure of Ownership Holdings . . .... 8 India .................................................. *..8 Bangladesh ............................................. 15 West Bengal and Bangladesh ..............................21 Pakistan ............................................... .22 East and West Punjabs ................................ ..24 Overstatement of Land Inequality ......................... 37 B. Structure of Operational Holdings . . 25 India ..................................... ................ 25 Bangladesh ............................................... 30 Bengal and Bangladesh . . .................................. 32 Pakistan ................................................. 34 Overstatement of Land Inequality .................... ..... 37 C. Landlessness Conceptual Confusion: Who are the "Landless" .... ... 38 India ................................................... ..... 42 Bangladesh ............................................... 47 Pakistan . ................................................ 53 Access to Jobs: Rural Wage Labor and Employment .............. 56 A. Rural Wage Labor Force . . ... 56 Magnitudes and Trends .................................... 57 India .................................................... 59 Some characteristic of Indian Rural Labor . ........ ........ 63 Bangladesh ............................................... 67 Pakistan ......................................... ......70 B. Rural Unemployment .... ......... ........ . ........................ .73 India ............... ..........................o.....*..*......74 Bangladesh ............................................... 85 Pakistan . .. . .... . 87 Conclusions . . ........ ......... . 89 - vi - C. The Occupational Distribution of Rural Wage Employment...91 Non-Crop and Non-Farm Employment .. 92 India .................................................... 93 Pakistan and Bangladesh .................................. 98 D. Some Final Numbers ...................................... 101 E. Conclusion .............................................. 104 Endnotes ...... 105 Annexes ...... 124 1 - Grouping of Indian States by Regions ................ 124 2 - Data Assumptions ................................... 126 3 - Rural Unemployment: Some Concepts and Definitions..129 5 - Annex Tables ..... 136 Bibliography ................................................ 185 - vii - List of Tables Page 1. India: Cumulative Distribution of Rural Households and Land Ownership (1961-62 and 1971-72) ....... ......... 10 2. India: Percentage Change in Number of Ownership Holdings and Area Owned by Different Size Groups of Holdings .12 3. Bangladesh: Percentage Distribution of Total Owned Land in Rural Area by Size of Holdings .17 4. Pakistan: Land Ownership by Size Groups of Land Owned - 1950s and 1970s .23 5. India: Cumulative Distribution of Rural Households and Operated Areas - 1961-62 and 1971-72 .26 6. Bangladesh: Size Distribution of Operated Holdings 1960-61 and 1967-68 .31 7. Bangladesh: Distribution of Agricultural Holdings, 1960-1984 .33 8. Pakistan: Distribution of Operational Holdings and Area by Farm Size in Pakistan and Provinces, 1960 and 1972 .36 9. India: Agrarian Profile and Landlessness in Rural India .43 10. India: Percentage of Landless Households Not Owning Nor Operating Land, Select States, 1954-55, 1961-62 and 1971-72 .46 11. Total Landlessness and Near-Landlessness Among Households in Rural India According to National Sample Survey Data, 1954/55 - 1971/72 .48 12. Landlessness in Rural Bangladesh .51 13. Pakistan: Rural Households, Landownership, Tenancy and Estinates of Landlessness (1961-1972) .54 14. South Asia: Composition of Rural Labor Force by Usual Activity Status, 1972-78. 58 15. India: Changes in Rural Labor, 1964-65, 1974-75 .60 16. India: Employment Conditions of Agricultural Labor Households - A Comparison of Rural Labor Surveys . 64 - viii - 17. India: Composition of the Rural Male Work Force in Indian States: 1072-73 and 1982-83 ..................... 66 18. Bangladesh: Rural Labor Households in Bangladesh ....... 69 19. Rural Laborers in Pakistan ............................. 71 20. India: Rural Labor Force--Percentage Distribution by Usual Activity, Current Activity and Sex ............... 75 21. India: Grouping of Indian States by Level and Variability of Unemployment Rates (% of the Labor Force) by Daily Status, Rural Persons 1972/73 .......... 79 22. India: Poverty and Unemployment, 1972-73 (Incidence of Unemployment by Monthly per capita Expenditure Groups in Rural India) ................................. 81 23. India: Rural Agricultural and Non-Agricultural Incomes in 1970 ........................................ 94 24. South Asia: Percentage Composition of Rural Labor Force and Employment in the Agricultural and Non-Agricultural Sectors ............................... 95 25. India: Rural Per Capita Incomes by Farm Size Group and Sources, 1970-71 ................................... 96 26. Bangladesh: Farm and Other Income, by Size Group of Holdings, 1965-66 ................................... 99 27. Pakistan: Contribution of Nonfarm to Total Income of Farm Families, by Size of Holding, 1968 ............. 99 28. South Asia: Estimates of Small Farms and Landless Households and Population in 1980 ..................... 103 ACCESS TO LAND: PATTERNS OF LAND HOLDING AND LANDLESSNESS To evaluate the prospects and problems of small farmers and the landless in South Asia,l/ one has to understand the agrarian structure within which they seek their livelihood and the ways in which this structure influences the distribution of the benefits of economic progress. This influence is exerted in three ways. (i) The agrarian structure determines the ownership of rural assets by different groups: these assets in turn form the basis for the way in which rural people derive their livelihood, whether from land, from other assets, or from labor. (ii) It shapes the nature and pace of technological changes, and the ways in which the benefits (and costs) of these changes are distributed. (iii) It determines the hierarchy of rural social and political relationships, which closely follow control over land. Broadly speaking, the term "agrarian structure" encompasses all the institutions relating to land, labor and the productive assets by which rural people earn a living, together with the network of economic, social and political relationships linking different rural groups. This paper will, however, focus more narrowly on two of the central features of South Asia's agrarian structure: the pattern of land holdings, both owned and operated, and the access to employment in agricultural and non- agricultural rural sectors.2/ In rural South Asia, as in other developing areas, disparities in land holdings produce disparities in incomes, and control of land usually coincides with control of local institutions. Access to other rural factors of production is largely determined by the size of land holdings. Variables such as the quality of land, the nature of tenurial arrangements, and the intensity with which land and technology are used in different areas, or by different groups in a particular area, may offset (or strengthen) che effects of the size of hoLdings. Nonetheless, the size distribution of land holdings remains the single most important determinant of agrarian relationships. I start this paper with a brief discussion of two alternative explanations (more accurately, theories) that purport to explain the trends over time in the patterns of land holding and the changing agrarian structure in South Asia. Two Theories of Agrarian Change In recent years the rural economy of South Asia has been characterized by high population growth rates in all areas and rapid changes in production technology in many regions. These factors form the starting points for two alternative but not mutually exclusive explanations of the patterns of landholding and landlessness in the subcontinent, theories that can be respectively characterized as the Malthusian or population pressure theory and the Marxian or class-polarization theory. The population pressure theory is based on the observation that the combination of rising rural population and a limited land base reduces the average size of owned holdings as they are apporLioned among heirs, adding to the problem of rural poverty. If the average rate of population increase is the same among large and small landholders, families with initially large holdings will be better off than others, but the size of all holdings will fall in absolute terms; there will be no change in the pattern of concentration) (If wealthier families with larger holdings have more surviving sons whose off-farm incomes can be invested in additonal land, or if they divide their land less often than other families, they may be able to compensate for the process of partitioning, or actually to enlarge their holdings.) If the rich wish to buy land, the poorest may be obliged to sell: partitioning by the smallest holders tends to produce holdings too small to withstand crises such as crop failures, or the death of draught animals. Lack of new land prevents expansion of the cultivable area as the labor force grows, forcing more of its -3- members to rely on wage employment. The class-polarization theory starts by observing that technological changes leads to the commercialization of agriculture, and to increases in the use of purchased inputs, in the proportion of total output sold for cash, and in the use of hired labor. As commercialization proceeds, the control of essential resources, including land, becomes more concentrated. Small farmers (especially tenants) facing competition from commercialized agriculture run increasingly into debt. In an economic system based on private property and with a high incidence of tenancy, the demand for credit leads to mortgages, foreclosures and the alienation of land from small cultivators to big landlords and money lenders. Disparities in incomes and in the ownership of land and other assets increase as small cultivators are driven into the classes of sharecroppers and landless laborers, while the land they formerly owned becomes concentrated in fewer hands. Much of the literature on agrarian change in South Asia subscribes to one or other of these theories.3/ Attwood (1979) demonstrates that their relative merits cannot easily be tested on the basis of available data, since most of the testable hypotheses that can be derived from each theory are common to both. Thus both predict that: - The proportion of landless villagers will increase. - This increase will be due mainly to downward mobility 4/ among cultivators, especially small holders. - Most families will move downwards, very few upwards. - Big landholders will make a disproportionately large share of the land purchases while small holders will make a disproportionately large share of the sales. - The mean or median size of holdings will decline. 5/ But there are three predictions on which they differ: - The Marxian theory predicts that ownership holdings are bound to become more concentrated, while the Malthusian theory suggests that although holdings of all sizes are -4- likely to shrink, there may be little or no change in the overall degree of concentration. - The Marxian theory suggests that the main cause of downward mobility is land sales by smaller to larger owners, while the Malthusian theory suggests that the main cause is the partitioning of holdings among multiple heirs (sales may, but need not necessarily, take place). - The Marxian theory suggests that big holdings will get bigger and small ones smaller, so that their heterogeneity (standard deviation from the mean) will increase, while the Malthusian theory suggests that, as partitioning will reduce most holdings to a fraction of their original size, heterogeneity will decrease. These last two hypotheses cannot be tested with aggregate data: it is necessary to trace what happened to particular landowning households over time. One useful analysis of this type is Attwood's (1980) study of the land holdings of the same households and their heirs over 50 years (1920-1970) in a village in Maharashtra. He found both upward and downward mobility, and found that (in terms of size of land holdings) some of the poor had become richer and some of the rich had become poorer. Land had not become concentrated in only a few hands. Nearly 25 percent of the landless in 1920 had land in 1970 (a change predicted by neither theory) while 44 percent of the originally landed had lost their land. Those with the smallest holdings in 1920 had enlarged them while the largest landowners of 1920 had lost most of their holdings. Contrary to the Marxian theory, most land purchases were by small holders, and most sales were by large owners. Consistent with the population-pressure theory, repeated partitioning was an important cause of downward mobility. This leaves the hypothesis that the distribution of ownership holdings will (or will not) become more concentrated. The available evidence on trends in ownership in India and Pakistan suggests that the distribution has become less concentrated over time, a trend that is not consistent with the Marxian perspective. - 5 - Vyas (1976), showed that, over a forty-year period in Gujerat, market transactions in land were few and most changes in holdings were made in response to land reform legislation. Landless households acquired land much more frequently than small households lost their land. [Evidence of the landless acquiring small plots of land is also given for Gujerat by Dantwala and Shah (1971) and for Haryana by Sheila Bhalla (1977). 6/] Most of the additions to the ranks of agricultural laborers came from small farm and artisan households. Finally, Vyas noted the strong possibility that many workers reported themselves as agricultural laborers less for lack of land than for lack of enough land to provide a living. He found no evidence of large-scale displacement of small farmers by larger farmers.7/ As Vyas points out in another (1979) study, a decline in the concentration of ownership holdings can be caused by (1) market processes, whereby marginal and small farmers may have purchased land from larger holders, (2) institutional processes, whereby larger holdings, broken up by land reform, may have been allocated to smaller holders, and (3) demographic processes, whereby holdings have been subdivided in response to increases in population and the size of rural households. The evidence suggests that all three processes have been at work in India, and that by and large land markets have worked in favor of small and marginal farmers rather than against them. Though there were some evidence in the 1950s of larger holders purchasing land from smaller owners, this process came to a halt during the 1960s. Fear of the imposition of ceilings on land holdings, opportunities for profitable intensification of land use, and tenancy legislation that made it easier for tenants to acquire land in a number of states, all made it less desirable and more difficult for larger owners to enlarge their holdings. Even where technological change has been rapid, the "proletarianization" thesis cannot be accepted, although the superficial evidence--an increase in the "agricultural laborer" category of the rural work force and a downward shift in the - 6 - structure of farm holdings--seems to support it. Sheila Bhalla's studies (1976, 1977) of changing agrarian structure and relations in Haryana during the 1960s, when the green revolution was in full spate, found that "the more-than-doubling of the number of male agricultural laborers from 1961 to 1971 ... cannot be attributed to the loss of land previously owned or leased in by households which are now landless labor households," and that "no household cultivating under 15 acres today, and no landless labor household, lost any land by sale to pay off debt, or sale for any other reason. None lost land through foreclosure of mortgages and none leased out land previously self-cultivated." 8/ The same analyst also found that the decline in the average size of small holdings in Haryana was caused by the subdivision of plots among families as farming became more profitable.9/ Other microeconomic evidence, however, suggests that big farmers in Punjab-Haryana may have been acquiring more land for cultivation, though from whom is not certain. 10/ Where technologies have changed rapidly--East Punjab, Haryana, West Punjab, and Tamil Nadu--the processes predicted under both theories have accelerated, while operational land holdings have definitely become more concentrated. Further displacement of tenants, a sharp increase in the use of hired labor and, among larger holders, diversification of assets away from land, have all been observed, especially in areas with assured irrigation; these trends lend some credibility to the Marxian perspective. Similar changes may be expected to accompany agricultural innovations in other areas. Finally, government policies, notably in India, where they have been of special help to "middle peasants", may have limited the process of polarization that might otherwise have taken place. It thus appears that both population pressures and technological change have helped to alter the agrarian structure, but that each of these determinants explains only certain aspects of the transformation of the pattern of South Asian land holdings. But what have been the changes in the patterns of land holdings and numbers of the landless in South Asia? Despite the enormous importance of and interest shown in questions relating to land distribution in South Asia, the available data are notoriously unreliable. The main sources of information are national agricultural censuses carried out every decade. These data suffer from at least four shortcomings: (a) they are not based on cadastral surveys but depend on land records, which are often unreliable and open to manipulation; (b) they relate to holdings and not necessarily to households, and although it is generally assumed that there is "one holding per household" this remains no more than an assumption; (c) they often relate to operated holdings 11/ rather than ownership holdings, which may remain unrecorded; and (d) definitions of size categories of holdings and of the principal occupations of households are rarely comparable from one census to another. Alternative data sources that allow meaningful cross- checks of the reliability of census data exist only for India. These are the National Sample Surveys (NSS) carried out from time to time. Unfortunately, however, the NSS data do not agree with the census data even for the same year, nor do they agree with supplementary sources. For example, the NSS for 1971-72 shows that 24 percent of all Indian cultivating households held their land on some form of tenancy basis, but comparable census data give what appears to be an unduly low figure of 9 percent. Such inconsistencies make it very hard to reconcile and interpret the available information, quite apart from calling its basic reliability into question. At best only broad magnitudes can be given; sometimes even these give irreconcilable results, and a considerable degree of interpretive judgment has to be used. Cross-country comparisons are even more shaky because definitions, size categories, relative reliability, and years for which data are available differ considerably. Although these - 8 - problems are widely known, they bear repeating because data are often used uncritically. The numbers given in this chapter are the best available; where there is a conflict, the more "reliable" data are presented. Finally, it is important to keep in mind the subcontinent's enormous diversity.' Aggregate data at the country level, particularly for India, are not very illuminating; anyone acquainted with South Asia knows that West Bengal is more like Bangladesh than like East Punjab, which in turn is more like Pakistan than like Tamil Nadu. To give data for each of the Indian states along with those for Pakistan and Bangladesh would make this book unmanageably long; on the other hand some disaggregation of the discussion and analyses that follow is essential if misleading generalizations are to be avoided. In the discussion that follows, we have therefore grouped the data for the Indian states into four broad regions--the North, East, West, and South. 12/ A. The Structure of Ownership Holdings Some analysts have argued that the ownership of agricultural land in South Asia has become more concentrated over time, as small owners have been bought out by larger ones and have been pushed into the ranks of the landless in a process of "proletarianization of the rural masses" (see below, section D). But the available data do not fully support this view. It seems that both the numbers of marginal and small owners and their share of the total area of owned land have increased; that the average size of their holdings has risen in many areas; that many rural households that previously owned no land have acquired some; and finally that, although land ownership remains highly unequal, this inequality has generally diminished. The subsections that follow examine the detailed evidence on trends in land ownership in three countries of the subcontinent. -9-_ India. The data on ownership holdings shown in Table 1 have been compiled from the NSS 17th (1961-62) and 26th (1971-72) Rounds.13/ Whereas the census data for these years are only for "ownership" of "cultivated holdings," the NSS provides information on the incidence of land ownership among all rural households (including those who own no land). 14/ Some interesting findings emerge. -- Over 9 percent of all rural households in India in 1971- 72 owned no land at all, not even enough on which to build a homestead. The regional variations in this percentage are quite marked: 4 percent in the North, 11 percent in the West, 14 percent in the South, and 8 percent in the East owned no land whatsoever. 15/ -- The percentage of all rural households owning no land fell in the 1960s in India as a whole and in nearly all regions. The fall was sharpest in the West, followed by the South and the East (there was an absolute but not a percentage decline in the North); it occurred despite an 8 percent rise in the absolute number of rural households (from 72.4m in 1961/62 to 77.8m in 1971/72), suggesting that many of those who had previously owned no land acquired some during the 1960s. - Holdings of owned land remain highly concentrated, however. In 1971-72, 63 percent of all rural households in India owned less than 1 hectare (2.5 acres) and they accounted for only about 13 percent of the area owned; at the other end of the scale, the 2 percent of all rural households that owned more than 10 hectares (nearly 25 acres) accounted for nearly 15 percent of the area owned.16/ This unequal pattern of land ownership prevails throughout India, though with some regional variation; the percentage of large holdings is highest in the South. What have been the changes in the size distribution of ownership holdings and area owned? Has ownership become more concentrated? What has happened to marginal and small ownership Table 1: IhDIA: CUMIUATIVE DISTRIBUTIONI OF RURAL HOUSEHOLDS AND LAND OWNERSHIP (1961-62 and 1971-72) Nlorth India West India South Indle East India All Indl. Size of lloldlb 1961 - 62 F97T - 72 1961 - 62 1971 - 72 1941 - *2 1971 - 72 1961 - 62 1971 - 72 1961 - 62 19t1 - 72 (is me) 1 A U A N A N A N A N A N A N A N A N A Ownership "oldist (2) Zero 4 0 A 0 13 0 It 0 16 0 14 0 12 0 6 0 12 0 9 0 Ze-r10 s6 It 64 16 44 7 44 9 70 to 70 12 61 14 1 22 60 9 13 1I Z.ro-2.0 76 30 11 49 57 19 60 38 6I 21 62 26 64 37 as 54 75 24 1e 26 Zero-4.0 94 65 93 73 so 46 76 59 91 43 92 49 95 64 96 77 90 52 90 62 Zero-100 99 65 99 92 93 10 94 63 96 73 96 76 100 9 t00 95 91 19 9 s All Uoldieas 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 t00 100 100 100 Total No: (Nine) 16.6 17.4 17.0 16.6 19.6 20.6 19.3 20.6 12.5 77.6 Total Avert (11in.o NI) 24.2 21.2 56.2 56.6 26.8 27.1 21.0 19.0 126.2 126.1 0 M: Percentage of omnership holdlags; A: Percentage of Arseo oeod. Serce: Ca lled frm the ISS 11th (161-62) and 26th (1971-72) *omae. The pereentegee have bees rounded off to the neirest whole nmuber. - 11 - holdings? These issues have been studied in some detail by a number of authors, using data from the NSS 8th (1953-54), 16th (1960-61) and 26th (1971-72) Rounds. 17/ Vyas (1979) has carefully documented changes in the proportions of all owned land (defined in terms of numbers of holdings and area of owned land) held by five groups of owners, (defined in terms of the size of their holdings) 18/ during the 1950s and 1960s. Table 2 is compiled from his data, and shows that the shares of marginal, small and medium sized ownership holdings grew over time as proportions of both total numbers of holdings and total area under ownership; marginal holdings registered the largest proportionate rise in numbers and small holdings showed the highest percentage increases in their share of total area. Larger holdings declined in importance; in 1953-54, they had made up 10 percent of all holdings and 53 percent of the area owned, but by 1971-72 they constituted only 5 percent of all holdings and 38 percent of the area owned. The average size of holdings fell for all size groups, though only very slightly for small and medium-sized holdings; the declines were much more marked among the ownership groups with very large and marginal holdings (particularly the latter). The distribution of ownership holdings became less skewed (the Cini coefficient fell from 0.525 to 0.512) between 1961-62 and 1971-72, a period of rapid technological change in many areas. 19/ All-India data tend to hide more than they reveal, but these broad findings have been substantiated in a state-by-state study by Sirohi, Ram and Singh (1976), reproduced in Table A.1 in the statistical appendix. These findings show that between 1961- 62 and 1971-72: -- The number of marginal ownership holdings of less than 1 hectare increased in all states except Rajasthan, Punjab and Kerala, with the largest increases occuring in Orissa, Gujerat and Karnataka. -- The total area covered by marginal holdings (less than 1 hectare) increased (by a larger percentage than their Table 2: inu: FaEWrCm cmA in ENt= u or ammAuslp uoai s - AK OUN0 *EFFUEU SIZE YWS f iiiCs (ALL IMI A) Average Percentage Change Average Percentage Change Average Size of oVnerehip Area (Acres) 1961-62 Over 1953-54 Area (Acres) 1971-72 Over 1961-62 Area (Acres) Holding (1953-54) Number Area (1961-62) Number Area (1971-1972) Marginal (Leon I Acre) 0.27 53.5S 21.5Z 0.21 51.2S -3.02 0.14 Smil (1-4.9e) 2.62 28.82 28.0 2.60 22.02 21.71 2.60 Medium (5-14.9) 8.54 16.7S 15.2Z 8.44 4.32 2.5Z 8.29 Big (15-49.9) 24.60 4.82 2.3Z 24.20 -10.2Z -11.52 23.87 Large (50 and more) 88.71 -37.7Z -34.02 80.96 -20.02 -27.02 73.87 TOTAL 6.25 31.02 4.072 4.97 26.62 -2.12 3.64 Source: NSS No. 36, 6th. Rounds for 1953-54; NSS No. 144, 17th. Round for 1961-62, and NSS no. 215, 26th Round for 1971-72. Tables on Land Holdin&o. Compiled from Vyas (1979). - 13 - increase in number, implying an expansion in the average size of holding). This was also true of small holdings (2.5 to 5 acres), except in Andhra Pradesh, Bihar, Maharashtra, Orissa, Rajasthan, West Bengal and Kerala. -- The large increases in the numbers of marginal and small holdings and the total land area they cover have generally been accompanied by a decrease in the number and area of medium and large holdings (i.e., those of 2.1-10 hectare and 10.1-20 hectare respectively). The decline of large farms is most noticeable in Kerala, Assam, West Bengal, and Bihar. -- The Gini coefficient of land distribution fell almost everywhere in India during the 1960s, except in Punjab. A detailed analysis by Sanyal (1977) of changes in ownership holdings in the 1950s and 1960s in six states -- Punjab, Haryana, West Bengal, Bihar, Gujerat and Andhra Pradesh -- also shows that the proportion of rural households owning no land has either shrunk, as in Punjab (from 37 percent in 1953-54 to 12 percent in 1961-62 and to 9 percent in 1970-71), in West Bengal and even Bihar, (from 17 percent to 9 percent to 4 percent), or has remained constant, as in Andhra Pradesh and Gujerat. Sanyal attributes the growth of the smallest land-owning group both to the subdivision of larger holdings and to the redistribution of surplus land, which was acquired by those who had previously owned none. Figure 1 and Annex Table 2 gives the percentage share of various size groups in the total number and area of ownership holdings. The medium ownership holdings above 4 hectares account for 54 percent of the total area in 1971, while the shares of this size group in the total number and area of ownership holdings in 1981 declined to 8 and 48 percent respectively. The proportion of marginal holdings increased from 63 in 1971 to 67 percent in 1981 and the proportion of area under marginal holdings rose from 10 in 1971 to 12 percent in 1981. This evidence contradicts the general impression that land ownership is becoming more unequal. - 14 - Figure 1. INDIA: Percentage Distribution of Ownership Holdings and Area Operated Over 5 Broad Categories of Ownership Holdings in 3 Major States and All India for 1960/61, 1970/71, and 1981/82 Marginal Small Semi medium Medium Large lessthan 101-202 405-1012 4.05-10.12 10.13hs. 1,01 ha. ha. ha. ha. and above House Area House- Area House- Areo Hous Area House AMec State Year hold owned hold owned hold owned hold owned hold owned 1971-72 Haryana 1982 1 1961-62 968412 39 4 0 Punjab 1971-72 57 L 7L 4 9~~~~~~~4 1982 10 1 1 1961-62 8 17 6 10 29 4 24 O WestBenoCl 1971-72 13 2 2 1 9 1982 61 1 2 1961-62 15 1 3 21 9 31 3 2 All Ini 1971-72 6721 2 1982 12 15 Source of 1971-72 etimcrtes. Saivokshana. Vol 5. Nos. 3 and 4. Source d 1961-62 stimatee NSS Report No. 144. - 15 - To conclude,the broad evidence on land ownership in rural India up to the early 1970s shows that the unequal distribution of land ownership following Independence still persists, despite a number of movements in the direction of a more even distribution. Nevertheless, the number of marginal and small owners as well as the total area they own and the average size of their holdings appears to have increased, while the importance of large farms has decreased; meanwhile, the number and proportion of rural households owning no land has declined significantly in many parts of India. Furthermore, it is generally agreed that "the subdivision of land holdings and the sale and purchase of land seem to have been more important in bringing about structural redistribution than the redistribution of land through tenancy reforms and legal ceilings on land holdings" (C.H.H. Rao, 1976). This is quite contrary to the uninformed rhetoric on the subject that suggests a growing concentration of land ownership and a slow proletarianization of the small peasantry, via gains by large owners who have bought them off (see Annex 2). Bangladesh "The unequal distribution of land ownership, aggravating the existing poverty in rural areas, continues to increase." Bangladesh is "a clear illustration of the trend towards land concentration, accompanied by a movement downward of most of the households." "The percentage of rural households (land owning and landless rural families) without land other than homestead land increased from 17.2% in 1961 to 31.8% in 1977".20/ "The process of increased proletarianization was brought about by the conversion of families owning small amounts of land into households of landless workers... the distribution of land among the remaining landowners became less equal". 21/ "The data indicate that the degree of concentration of ownership is increasing". 22/ Despite the wide currency of statements such as,these, there are no reliable data on land ownership in Bangladesh as a whole and certainly none that reveal unmistakable trends. - 16 - The only data on land ownership are from the Land Occupancy Surveys (LOS) carried out in 1977 and 1978 under the auspices of USAID. 23/ These surveys covered only some 4,100 households in detail. Although efforts were made to design a careful sampling frame, the sample size is to small to represent Bangladesh's 12 million rural households. Close examination of the data (see below) confirms the unreliability of many of the numbers derived from the LOS. In addition to inconsistencies in the data, the various very large changes recorded in a single year suggest that the results from the two annual surveys may not be strictly comparable. Thus, while it may be possible to accept data for a given year, provided that they are backed by information from other sources, it is not appropriate to compare LOS data from one year to the next; consequently, no trends should be extropolated from the LOS data unless they are substantiated by independent time series. The distribution of total owned land (including homestead land) 24/ presented in Table 3. It appears that ownership is even more skewed in Bangladesh than in India. 25/ In 1978, nearly 15 percent of rural households owned no land at all; 90 percent of rural households owned less than 5 acres, and together comprised only 52 percent of the total area under ownership. At the other end of the scale, under 3 percent of households owned more than 10 acres each and 25 percent of the area. 26/ (See Annex Table A.4). A comparison of the data for 1978 with those for 1977 suggests that very large changes took place, in both the numbers of households in various size groups and their share of total land owned, during just one year. There is much room for scepticism about whether such changes in fact occurred. In India, even in areas with comparable land ownership patterns and demographic pressures, shifts of this magnitude have not been recorded over periods of one or two decades, (see Annex Table A.1). For example, Bangladesh LOS data for the percentage increase in landless rural households and the rise in the total number of households combine arithmetically to give an incerase of over 455 - 17 - Table 3 LUb3L&D&SR PU=NAG DISTIMIMUTOU OF TOMAL MWM LAW IN FIA.L ABU n SIZE Of ELDIE (1977 AM 1971) Size aroup of Numbsr of Households (N) Total Are& Ovned (A) Percentage Absolute Holding 1977 1978 1977 1978 ChNang for Group (Acres) N A Zero 11.07 14.69 - - 34.8 - 0-1 47.44 44.68 9.30 8.33 (4.4) 33.3 1-2 16.43 15.21 14.43 12.78 (6.0) (4.8) 2-3 8.91 8.69 13.18 12.28 (1.0) 0.2 3-4 5.27 5.16 11.13 10.29 (0.5) (0.5) 4-5 3.29 3.08 9.00 7.93 (4.6) (5.2) 5-6 2.09 2.11 6.90 6.61 2.4 3.1 6-7 1.43 1.44 5.69 5.40 2.4 2.0 7-6 1.02 0.92 4.65 3.98 (7.5) (7.9) 8-9 0.69 0.79 3.60 3.86 15.9 15.3 9-10 0.42 0.56 2.46 3.06 34.0 33.7 10-11 0.34 0.51 2.15 3.05 52.3 52.7 11-12 0.29 0.32 2.03 2.14 14.7 13.1 12-13 0.16 0.30 1.18 2.17 100.0 97.6 13-14 0.22 0.22 1.82 1.69 0.0 (0.2) 14-15 0.13 0.16 1.19 1.37 25.0 23.6 over 15 0.80 1.16 11.29 15.07 0.0 43.7 All 100.00 100.00 100.00 100.00 1.5 7.6 Totals - alas 11.85 12.03 1.5 - Ulu acres 19.35 20.81 7.6 of which theset -under 5 acres 10.95 11.01 10.54 10.74 0.6 1.9 (including landless) -O to 5 acres 9.64 9.24 10.54 10.74 (4.2) 1.9 (excluding landless) -over 5 acres 0.90 1.02 8.81 10.07 13.3 14.3 Overall Cinl Coefficient of 0.6577 0.6842 Concentratlos Source:Compiled from F.T. Januzzi and J.T. Peach, 1980, "The Agrarian Structure of Bangladesh: An Impediment to Development , Westview Press, (1980), Appendix Tables D.I. and E.I. - 18 - thousand in the number of landless households between 1977 and 1978. Such an increase, amounting to nearly 2.5 million persons, seems incredibly large for one year; interestingly, it almost exactly matches the fall in the number of households in the 0.0 - 1.5 acre ownership category, suggesting that the change represents households in this group that have been reclassified as "landless" over the period. Other oddities emerge from close scrutiny of the data. At the higher end of the scale, households with holdings of 12-13 acres apparently nearly doubled both their numbers and their proportion of total area owned in a single year. Meanwhile, the data imply that the number of small ownership holdings (between 0.1 and 5 acres) declined, but that the average size of ownership holdings increased in small size categories. Does this mean that larger holdings have become progressively smaller, as in India? If so, the increase in the average size of small holdings should have been accompanied by a decrease in the size of larger holdings; in fact, calculations based on the data show a rise in the average size of holdings of 15 or more acres--from nearly 23 to nearly 33 acres! The impression of a high concentration of ownership is further weakened by the LOS data on average household size given in Annex Tables A.5 and A.6. These show that household size is positively correlated with size of holding, so that the area owned per capita is more evenly distributed than the data on ownership by households imply. 27/ Nonetheless, a striking feature is the small amount of land owned per capita in all size classes. The average amount owned per household is 1.6 to 1.7 acres (compared with around 4 acres in India); the per capita figure is 0.2 to 0.3 acres. The data on ownership of cultivable land (that is, excluding homestead land) present an even grimmer picture. First, Annex Table A.3 shows that nearly a third of all rural households owned no land other than homestead land in 1978. Second, the ownership distribution of cultivable land is much more - 19 - concentrated than that of total land. Over 60 percent of all households owned less than 1 acre of land other than homestead; on the other hand, 2.5 percent of households, each owning more than 10 acres, account for over 22 percent of the non-homestead land. Looking at the differences in landlessness between 1977 and 1978, Table A.3 suggests that there was a fall of over 400,000 households (about 2.3 million people) in the category of those owning no land except homestead land in a year. If we try to combine this decrease in households owning no land other than homestead with the large increase in households owning no land at all (Annex Table A.7), we see that nearly 800,000 rural households (a population of over 4.7 million or 7 percent of the total rural population) must have lost their homesteads in a single year. 28/ Evidence collected on land sales lends some support to the idea of a general trend toward landlessness. Writing of the 1969-73 period, A. R. Khan (1977) argues that "to survive, most of the small farmers (perhaps up to 50 percent of the total if 2 acres is accepted as the minimum needed for self-sufficiency) must have been forced to borrow and to sell assets. Since most of them were already indebted, distress sales of assets (including land) must have occurred very frequently." He presents data showing that "those who owned less than one acre (accounting for about a quarter of the farmers) sold well over half of their remaining land each year. The higher farmers in contrast sold progressively less." Other evidence of "distress sales" of land during and following the 1974 floods is cited by Stepanek (1979), 29/ who also describes how the slow process of land alienation has taken place in Bangladesh: "...their land holdings had been diminishing for a number of years. Floods, poor crops, and high prices had forced them to sell land in bits and pieces. In some cases they sold their land in one year, and, in succeeding years, their bullocks, their homes and finally, their utensils and farm implements. The transition has been from medium to small - 20 - plots and then to sharecropping and landlessness. The final blow was the flood of 1974, which reduced these people to destitution." (Op.cit., p. 98). A similar process is described by M. Alamgir (1978) with data from a survey carried out during the famine of 1974. 30/ This showed that 44 percent of all rural households sold some type of asset in 1974, including 37 percent who reported "distress sales." Land sales rose sharply that year; nearly three-quarters of all sellers were households with less than 5 acres, while the major purchasers were "large farmer, medium farmer and business groups" (p. 104-107). Islam (1979) comments that rich farmers and others have invested their surpluses in land, which provides a high and secure yield compared to other available options. Alamgir (1978) and others have argued that the transfer of assets in rural areas, particularly of land, can be attributed to exploitative production relations. Other causes may be at work, however; a major cyclone, a brutal civil war, and a severe famine together with constantly rising population pressure, make it hardly surprising that small owners were forced to sell their assets. In this respect, Bangladesh's experience has been uniquely difficult. To sum up, the evidence on land ownership in Bangladesh suggests that: -- The distribution of land ownership is very unequal, and may have been worsening. -- Nearly a third of all rural households own no land other than homestead land, and 15 percent do not even own a homestead. -- Very small ownership holdings (less than 2 acres) constitute nearly two-thirds of all ownership holdings and nearly three-quarters of all owned holdings excluding homestead land. -- The ownership of land on a per capita basis is less concentrated than that on a per household basis, but the inequality is still signficant. - 21 - -- It appears that small ownership holdings have become smaller, and that a larger number of very small owners have sold their land to larger owners and joined the ranks of those owning no land. This tendency was especially evident during and after the famine of 1974. West Bengal and Bangladesh Because of the similar agrarian conditions in Bangladesh and West Bengal (India) it is worth comparing the structure of land ownership in the two areas. Annex Table A.8 presents the data. If anything, land ownership was more egalitarian in Bangladesh in 1977 than in West Bengal in 1971; the latter seems to have had an even heavier concentration ownership into a large number of very small holdings. (Since the West Bengal data are for 1971, this difference may have become even more pronounced.) In both regions, over 50 percent of all holdings are smaller than 1 acre, although those owning no land at all seem larger in Bangladesh. Given the similarity between the two areas and the relatively greater reliability of the data for West Bengal (especially with regard to changes over time), it may be possible to use the latter as a rough guide to the kinds of changes that may have occurred in Bangladesh in the past two decades. In West Bengal: (i) the average size of holdings fell, for all size groups; (ii) the number of marginal (less than 1 acre) and small (1-5 acres) holdings rose, as did their share of the total owned area; (iii) the share of medium (5-10 acres) and large holdings (10-15 acres and over) fell, in terms of both numbers and share of total area; and (iv) the concentration of land holdings consequently fell, as did the percentage of rural households owning no land. Similar trends would in principle be likely to have occurred in Bangladesh. On the other hand, the relatively greater success of land reform efforts in West Bengal and the devastating impact of civil war, floods, and famine in Bangladesh, are likely to have raised the proportion of those owing no land and the - 22 - concentration of ownership, as the LOS data seem to suggest. These issues must remain unresolved pending better survey data (preferably cadastral) on land ownership in Bangladesh. Pakistan 31/ Land distribution in Pakistan is the most unequal in the subcontinent. 32/ The earliest, though incomplete, evidence comes from the West Pakistan Land Reform Commission. This shows that land ownership was heavily concentrated in Pakistan at the time of Independence 33/ though apparently less so in the Punjab than in Sind, 34/ where about 8 percent of all owners owned very large holdings (over 100 acres each), comprising 55 percent of the owned area; a little over three thousand landlords commanded a third of all the land. (Annex Table A.9). Data on land ownership are not available again until the 1970s. Table 4 shows that land ownership was apparently still very unequal in 1972, although the area in very large holdings had been reduced. More dramatic changes occurred in the four years following the Land Reforms of 1972 than in the preceding two decades. By 1976, 71 percent of all owners owned less than 6.25 acres, and accounted for 25 percent of the land, while the 6 percent owning more than 25 acres accounted for only 36 percent of the land. The greatest changes affected large owners (those with holdings of over 50 acres) whose number fell from 3 percent to 1 percent of all owners and whose share of the total cultivated area dropped from 29 percent to 22 percent. Comparing the Punjab and Sind in this period, small holdings of less than 6.25 acres increased both as a proportion of the total number of holdings and as a proportion of all land owned in both provinces, although by more in the Punjab. Farms of over 50 acres accounted for 8 percent of the farms and over 40 percent of the land owned in Sind; in the Punjab they accounted for only 1 percent of owners and only 18-20 percent of area owned (see Annex Table A.10). The Gini coefficient of land ownership in Pakistan fell from 0.64 in the 1950s to 0.55 by 1976, mostly reflecting a Table 4 PAKISTAN: LAND OWNERSHIP RY SIZE GROUPS OF LAND OWNED - 1950's and 1970's 1950. 1972 1976 No. of Area No. of Area No. of Area Slze Group Owners Owned Size Group Owners Owned Size Group Owners Owned (Acres) (Cu 2) (Cu Z) (Acres) (Cu 2) (Cu 2) (Acres) (Cu 2) (Cu 2) 0 - 5 64.4 15.3 0 - 5 66.5 17.6 0.0 - 6.25 70.8 24.9 5 - 25 93.1 47.0 5 - 25 94.5 56.7 6.25 - 12.50 88.3 46.2 25 - 100 98.8 68.8 25 - 100 99.0 82.8 12.50 - 25.0 95.9 64.3 W 100 - 500 99.9 84.6 100 - 150 99.4 89.1 25.0 - 50.0 98.5 77.5 Over 500 100.0 100.0 Over 150 100.0 100.0 Over 50.0 100.0 100.0 5.07 48.64 10.06 67.64 10.3 60.13 ln *ln Acres min *ln Acres aln min Acres Compiles From: 1) H. H. Khan (1981) compiled from his Landownership Table 3.1 -3.5 citing COP Land Reform Comission reports and personal co mmnications. 2) GOP 'Agriculture Census 1972,' Annex. - 24 - decrease in inequality in the Punjab. 35/ At least part of this fall can be attributed to benami (hidden) transfers within families of owners. Nonetheless, one may conclude that: -- Land ownership in Pakistan has become more equal since the land reform of 1972. -- Pakistan's concentration of land ownership continues to be the most extreme in the subcontinent; this is expecially true in Sind. -- Small owners (with holdings of less than 5 acres) are still the dominant group and both their numbers and their share of the total area owned are increasing. -- Some small owners have increased the size of their holdings, especially in the Punjab. This, combined with a decline in the concentration of land holdings at the top end of the scale, has led to a preponderance of medium sized holdings, of 15-50 acres. East and West Punjabs Contary to accepted notions, there are many basic similarities in the ownership patterns in West Punjab (Pakistan) and East Punjab (India) 36/. As Annex Table A.11 shows, nearly two-thirds of all owners or land owning households in both areas own less than 5 acres; the number and total area of these holdings has been increasing while their average size has declined or remained constant; the number of larger owned holdings has declined, but their total area has decreased less or increased. Thus the average size of medium (5-25 acres), large (25-50 acres) and very large (over 50 acres) households (East Punjab) or individual (West Punjab) ownership holdings has increased, with the largest increase occuring in the medium size groups. The main differences betwen the two regions lie in the average size of farms. Very large farms still exist Pakistan, where in West Punjab in 1976 over 10,000 individuals owned more than 150 acres; in the Indian Punjab fewer than 8,000 households owned holdings of more than 50 acres. 37/ Ownership holdings of - 25 - under 5-6.6 acres make up about two-thirds of all holdings in both Punjabs, but in West Punjab their average size is 2-1/2 times larger. Differences in the importance of large owners and the average size of small ownership holdings have major consequences for agrarian structure and for the policies followed in the two areas, as will be shown later. B. The Structure of Operational Holdings The distribution of land ownership provides a useful guide to the ownership of assets in rural areas, but it does not tell us how these assets are used. Since land rental markets are a prominent feature of rural life in the subcontinent, one must look beyond the distribution of owned land to the distribution of operational holdings (i.e., those cultivated either by owners or by tenants), in order to determine who has access to land, and consequently who can benefit from strategies to make land more productive. 5i) India Table 5 shows the cumulative size distributions of operational holdings by regions in India for 1961-62 and 1971-72 (including households that operate no land). 38/ For the latter year several features stand out. -- Between one-quarter and one-third of all rural households in the four regions operate no land. (As we shall see later, not all of these households depend wholly on wage employment, but many do.) -- Another 20-25 percent of all rural households operate marginal holdings of less than 2.5 acres, and a further 15-20 percent of all rural households (20-25 percent of cultivating households) operate small holdings of between 2.5 and 5 acres. Holdings are especially small in the East and South. -- There are very few medium to large holdings: only 2-8 percent of all rural households (5-12 percent of Table 5 INDIA: CUMULATIVE DISTRIBUTION OF RURAL HOUSEHOLDS AND OPERATED AREAS - 1961-62 and 1972-72 North India West India South India East India All India Size of Holding 1961 - 62 1971 - 72 1961 - 62 1971 - 72 1961 - 62 1971 - 72 1961 - 62 1971 - 72 1961 - 62 1971 - 72 (In Ila) N A N A N A N A N A N A N A N A N A N A Zero 23 0 26 0 21 0 25 0 33 0 35 0 29 0 24 0 27 0 28 0 Zero-l. 56 10 61 13 42 3 43 3 67 9 67 11 63 13 68 20 56 7 61 9 Zero-2.0 76 29 80 42 56 10 59 18 79 21 82 30 82 38 87 55 74 21 83 30 Zero-4.0 90 56 93 68 74 25 78 37 91 42 92 59 94 66 96 80 86 41 90 53 Zero-12.0 99 91 100 94 94 66 96 75 89 76 99 84 100 93 100 97 98 77 99 84 All Holdings 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Total No: (Nkn) 11.6 12.7 11.9 14.1 13.7 14.9 13.5 15.3 50.8 57.1 Total Area (Nn Ha) 25.4 22.7 57.9 58.1 28.0 26.4 22.1 18.5 133.4 125.7 N: Cumulative percentages of numbers of operated holdtngs; A: Cumulative percentages of areas operated. 0 Sources: Compiled using data from the NSS 17th (1961-62) and 26th (1971-72) Rounds; V.N. Dandekar and N. Rath, Poverty in India, the study by Sirohi, Ram and Singh (1976) and Research lank of India 'All India Debt and Investment Survey'. N.8.: Since 1971-72 statevize distributions were unavailable. These data wer compiled by applying the changes in the distribution of operational holding, of various rates given In Strahi, Ram and Singh to the 1961-62 distributions given in the NSS and Sendekar and lath. The number of households were grouped by various region. using the RSI data. These data ware then consolidated to get the all-Indta distribution. - 27 - cultivating households) operate holdings greater than 10 acres, except in Western India where 23 percent (30 percent of cultivating households) operate holdings larger than this. -- Marginal and small farmers account for three quarters of all cultivated holdings but only a quarter of the cultivated land. By contrast, the 10 percent of cultivated holdings larger than 10 acres account for nearly 54 percent of the operated area. One might think that the size distribution of operational holdings would be less skewed than that of ownership holdings, on the assumption that larger owners generally rent out land to smaller ones. The need to use land intensively, together with the diseconomies of scale associated with the management of labor, also favor relatively small holdings and promote transfers of land in this direction. This presumed flow of land underlies the comonly held belief that the agrarian structure consists of large landlords at one end and small tenants on the other, with owner- cultivators lying in between. The reality is far more complex, however. The percentage of rural households that operate no land is much larger (25-35 percent, depending on the region) than the percentage owning no land (4-13 percent--see Table 1). Further, though quite a large proportion of rural households (12-30 percent) own very small holdings (under 0.5 acres) a much smaller proportion (only 3-11 percent) operate holdings of this size. It seems likely that owners of very small holdings, finding them uneconomic to run, rent them out to larger operators, and choose to earn their livelihood in some way other than farming. 39/ Thus, 'landlords' and 'tenants' cannot be identified as mutually exclusive social classes--a point that is often missed. Many of those with relatively large operational holdings are tenants of small owners. The analysis by Vyas (1979) of changes in the distribution of operational holdings tends to support this interpretation. His study reveals that the distribution of - 28 - operated holdings in the 1960s was less skewed than that of ownership holdings in only five states--Assam, Bihar, Orissa, Uttar Pradesh and West Bengal--all in Eastern India. It was more skewed in four states--Tamil Nadu, Karnataka, Punjab, and Harayana--all of which except Tamil Nadu experienced high rates of agricultural growth during the same period. No clear conclusion emerges for the remaining states. What have been the changes in the distribution of operational holdings over time? Three trends are evident from the all-India data (given in Annex Tables A.12 and A.13: -- Although operated land is still very unequally distributed, inequality has been reduced somewhat over time. -- The number of small and very small holdings increased during the 1950s and 1960s, as a result of subdivision and the growth of population, but the average size of marginal holdings (below 2.5 acres) also increased. -- .Larger holdings (above 15 acres) fell proportionately (in terms of both numbers and area), but their average size did not decline. 40/ Figure 2 gives more recent data and shows that the area operated under marginal holdings in 1971 has risen from 9.8 percent to 12.2 percent by 1982. The area under large holdings has dropped significantly from 29 to 18 oercent from 1961 - 1982 (see Annex Table A.14). The view that small operators are losing their land to larger ones as the latter buy them out is widely held among proponents of the "class polarization" theory of changing agrarian structure , and may be difficult to dislodge. It provides a convenient explanation for the rise in the number and proportion of agricultural laborers in the rural work force--an increase too large to be accounted for solely by the growing numbers of people in agricultural labor households. Nevertheless, the evidence presented so far on the changes in the size distribution of holdings in India is not consistent with this perception. - 29 - Figure 2 Percentage Distribution of Operational Holdings and Area Operated over 5 Broad Categories Holdings in 3 Major States and All India for 1960/61, 1970/71, and 1981/82 States 'eofar marginal Small Semi-Med!um MecltlJr sarge No A:ea No Alea No Area NC Area NC Area '461-62 12 W S 24 Punjab 1971-72 7.1 L 32 . 23 WestBengal 1971-72 25 23 2 1 3 98a1-32 29 28lL i1 ; 2 2 231 20 2 A;! rndta 5?.222 i 0 ,Q84-82 Source N S.S Pounrds - 30 - Changes in Punjab and Harayana have run contrary to the general findings noted above that (1) marginal and small holdings have increased in numbers and area; (2) large holdings have declined in numbers and area, and (3) inequality has declined somewhat over time (see Annex Table A.15). In this region, both the numbers and area of smaller ownership and operated holdings declined in the 1960s, and inequalities as measured by Gini coefficients increased. 41/ The number of very small holdings of less than 2.5 acres rose (while their combined area declined), and the number of holdings in all other size classes decreased as did their average area, the decline being greatest in the large size groups, according to the NSS data. A tendency for both the numbers and the share of area accounted for by medium-sized holdings (5-15 acres) to increase in Punjab is supported by the NSS data as analyzed by Sanyal (1977). Bangladesh Reliable data on the size distribution of operational holdings in Bangladesh are only available for the 1960s. They do not permit systematic comparisons between the size distributions of ownership and operational holdings for the same year, nor do they show what proportion of rural households operated no land. The size distributions of operated holdings in Bangladesh in 1960-61 and 1967-68 are shown in Table 6. By 1967-68, nearly 57 percent of these holdings were less than 2.5 acres in size, accounting for 22 percent of the cultivated area. Farms of less than 1 acre accounted for 25 percent of all holdings and under 5 percent of the total cultivated area. By comparison with 1960-61, the distribution of operational holdings had become a little less unequal. Evidence for later years suggests a dramatic reversal. Early tabulations of the 1977 Agricultural Census show that the size distribution of operational holdings has shifted quite dramatically against farms of less than 2.5 acres and in favor of "$medium" farmers (Table A.16)--reversing earlier trends during the sixties, when both the number and even devoted to these farms had Tab le 6 MISGADUDSU: SIZEN DlSmISTION Of OPRTRED HOLOINGS 1980-61 AM 1967-4 1960-61 1967-68 1977 Size Group Number Area Number Area Number Area 1960 to 1967 1967-1977 Average Size of farm As 2 of As X of As 2 of As 2 of As X of As X of Percentage Increase Percentage Increases (Acres) (acres) Total Total Total Total Total Total In Number In Area In Number In Area 1960 1967 1977 Under - 0.5 13.1 0.9 12.3 1.2 5.5 0.5 5 22 -60 -60 0.24 0.31 0.32 0.5 - 1.0 11.2 2.3 12.7 3.1 10.4 2.1 26 33 -25 -31 0.73 0.76 0.73 1.0 - 2.5 27.3 13.0 31.6 17.1 33.9 16.1 30 30 -3 -4 1.68 1.69 1.67 2.5 - 5.0 26.3 26.4 26.3 30.0 29.2 29.2 12 13 0 -l 3.55 3.57 3.5 3.0 - 7.5 11.4 19.3 9.2 17.8 11.6 19.7 -9 -9 15 13 5.98 6.06 6.0 7.5 - 12.5 7.2 19.1 5.2 15.5 6.6 17.4 -18 -19 Is 15 9.38 9.32 9.31 12.5 - 25.0 3.1 14.1 2.2 11.0 2.5 11.6 -21 -23 3 7 16.01 15.60 16.25 25.0 + 0.4 4.8 0.5 4.5 0.4 3.4 40 -7 1 1 33.9 28.17 32.08 Total 100 100 100 100 100 100 12 * 3.53 3.13 3.5 (6.14 (21.7 (6.87 (21.5 (6.25 (21.96 min.) mIn.acres) mn.) m1n.acres) mn.) min.acres) Notes: (I) Data is for 'farms' and not for 'households.' (2) Figures may not add to total due to rounding. (3) * Indicates negligible. Source: (l) Government of Pakistan, Agricuiltural Census Organilzation, `1960 Pakistan Census of Agriculture: A Summary of East Pakistan Data," P.C. b M.C. Press, Lahore. Pg. 52, Table 3. (2) Government of Bangladesh, Bangladesh Buireau of Statistics "Master Survey of Agriculture in Bangladesh, 1967/68 (Seventh Round, Second Phase)". Dacca, 1972. Page 9, Table 1. (3) Bangladesh Bureau of Statisties, Agriculture Census, 1977. - 32 - actually increased while that of larger farms had decreased. Combined with an increase in the average size of larger farms and a growth in their number, this suggests that many smaller farmers have either sold out their farms and become landless or stopped operating. Nearly 0.82 million holdings have dropped out of these categories--their operators probably joining the class of the landless in Bangladesh. Taking the data from the 1977 Agricultural Census together with the 1977 LOS data on ownership holdings cited earlier, one finds that the distribution of operated holdings is more skewed than that of ownership, suggesting that, in Bangladesh as in India, very small owners are probably net lessors of land.42 (Table 7). The gaps in the data are so wide, and the available information so generally unreliable, that any number of conjectures are possible. It seems likely, however, that the share of land operated in small holdings has declined, and that land markets have worked against small holders, pushing them into the rural proletariat. This is supported by fragmentary evidence on rural land-sales and purchases in the early seventies. (see Annex Table A.17). West Bengal and Bangladesh Although the Eastern parts of India particularly West Bengal in the 1960's experienced changes similar to those occuring in Bangladesh, the recent changes in Bangladesh have been quite dramatic. As Table A.18 in the annex shows, although both West Bengal and Bangladesh in the beginning of the 60's had a fairly similar distribution of operated holdings by 1977 the number of smaller holdings had decreased while those in West Bengal continued to increase. Furthermore while the area and number of medium size holdings in West Bengal continued to decrease, those in Bangladesh have increased. The distribution of operated holdings in Bangladesh has become more skewed and concentrated over time. - 33 - Table 7 Bangladesh: Distribution of Agricultural Holdings, 1960-1984 Percent of all* Change in Size of Farms Percent of all Holdings Land Operated Total Area (acres) 1960 1977 1984 1960 1977 1984 1977-1984 less than 1.0 24.3 15.9 40.5 3.2 2.6 7.8 +205 1.0 - 2.5 27.3 33.9 29.9 13.0 16.1 21.2 + 36 2.5 - 5.0 26.3 29.2 17.9 26.4 29.2 27.4 - 3 5.0 - 7.5 11.4 11.6 6.7 19.3 19.7 17.6 - a 7.5 and above 10.7 9.5 4.9 38.1 32.4 25.9 - 17 All* oldings 100.0 100.0 100.0 100.0 100.0 100.0 semorandu Items dumber of Farms (million) 6.139 6.257 10.045 Cultivated Area (m acres) 21.726 21.95 22.678 Population (s) 50.0 81.4 97.1 Source: Bangladesh Census of Agriculture and Livestock, 1983-84, BBS, May 1986, and earlier censuses, in "Bangladesh: Promoting Higher Growth and Human Development". Washington, D.C.: World Bank Country Study, October 1983, p. 145. - 34 - As a result, one consequence is that the average size of operated holdings continues to be much larger in Bangladesh than in West Bengal: by 37 percent (in 1960-61) and 84 percent (Bangladesh in 1977 compared with West Bengal in 1970-71). This runs contrary to the general view that it is Bangladesh which has the smallest holdings in the subcontinent; although it may have experienced the severest of demographic pressure the concentration of land holdings have become even more pronounced. West Bengal has seen a much greater long-term decline in the average size of holding. It is one of the most densely populated and longest settled regions in the subcontinent. It has not experienced the social and economic dislocations experienced by Bangladesh in the last decade which has made Bangladesh's agrarian structure more inequitable. Pakistan The only data on the size distribution of operational holdings in Pakistan come from the 1960 and 1972 agricultural censuses. These have been carefully examined by M. H. Khan (1981), some of whose main findings are reproduced in Table 8 and Tables A.19 and A.20 in the annex. The data show that, in 1972: -- In sharp contrast to Bangladesh and most of India, only a third of all operated holdings were of less than 5 acres. Together they accounted for 5 percent of the area in 1972, and their numbers and area had declined rapidly nationwide since 1960. The decline was especially noticeable in the Punjab, where all other size groups increased in terms of both number of holdings and in percentage of total cultivated area. -- The numbers and combined area of farms over 50 acres have increased in Pakistan as a whole, but not in Sind, where they have fallen slightly. Medium sized farms--the 12.5 to 50 acre group--increased in number and combined area, most noticeably in the Punjab: this category also declined in Sind. - 35 - -- The average size of holdings fell in all size groups except the 5-12.5 acre group. Holdings of this size, small by Pakistan's standards, dramatically increased in number and area covered, especially in Sind. 43/ The conclusion is clear: especially in the Punjab, the size distribution of operated holdings in Pakistan has become more skewed, favoring medium (12.5-50 acres) and large farms (over 50 acres) at the expense of small ones. A comparison of the distribution of ownership (Table 4) with that of operational holdings (Table 8) shows that though the Punjab has the large farms, it is Sind that has the large landlords. In the Punjab, a large proportion of marginal and small owners (less than 12.5 acres) rent their land to owners of medium and large holdings. In Sind, by contrast, the owners of large and very large holdings rent land to small and sometimes medium sized sharecroppers. Medium and larger operational holdings predominate in the Punjab, and their operators are more likely to be owners who have rented-in more land to enlarge their holdings. In Sind, small operational holdings are mostly farmed by sharecroppers (haris), who rent from larger owners. East and West Punjab The Indian and Pakistani Punjabs share two basic similarities (see Table A.21 in the annex]: (1) a fall in the share of operational holdings of less than 5 acres in the total number of holdings, combined with an increase in their average size, and (2) a rise in the share in total numbers and combined area of medium size holdings (12.5-50 acres). As with ownership holdings, however, operational holdings are typically much larger in Pakistan than in India. In West Punjab, nearly 50 percent of the operated area is in holdings between 12.5 and 50 acres, with another 15 percent in even larger holdings; in East Punjab, by contrast, nearly 80 percent of the operated area is in holdings of under 25 acres. Thus, the total area of operated holdings of less than 5 acres increased in Punjab-Haryana (India) but declined in West Punjab, Pakistan, and the number and combined area of T4ble 8 PAKISTAN: DISTRIBUTION OF OPERATIONAL HOLDINGS AND AREA BY FARM SIZE IN PAKISTAN AND PROVINCES, 1960 AND 1972 Pakiatan Punjab Sind Farm Size Percent Percent Percent Percent Percent Percent (acres) Year of Farms of Area of Farms of Area of Farms of Area Under 5.0 1960 49.5 9.4 51.6 10.9 25.6 5.5 1972 28.2 5.2 26.1 4.8 19.0 4.5 5.1 - <12.5 1960 27.6 22.3 26.9 24.9 38.9 22.5 1972 39.9 25.2 39.0 24.6 51.7 34.5 12.51 - <25.0 1960 15.0 25.6 14.7 28.5 23.3 28.6 1972 21.1 26.6 23.1 28.8 21.9 29.2 2Z.1 - <50.0 1960 5.9 19.4 5.4 20.2 9.0 21.1 1972 7.6 18.8 8.8 21.3 5.2 13.2 50.1 - <150.0 1960 1.8 13.4 1.3 10.6 2.1 14.2 1972 2.7 15.1 2.7 14.7 1.7 10.7 150.1 4 over 1960 0.3 10.0 0.2 4.9 0.3 8.2 1972 0.4 9.1 0.3 5.8 0.4 7.0 Total 1960 4.86 48.93 3.33 29.21 0.68 (Farms In mlillons and operated area In million acres) 1972 3.76 49.06 2.38 31.03 0.75 9.46 Source: Pakitatn, Census of Agriculture. 1960 and 1972 as cited In N.H. Khan (1981), Table 3.17. - 37 - holdings over 25 acres rose substantially in the latter, but fell quite dramatically in the former. Whereas both regions have seen a dramatic shift towards "medium" size holdings, in the West Punjab this group of farms have been in the 20-50 acre size while in the East Punjab they have been in the 5-15 acre size. In addition the West Punjab has a substantial number of large farms above 50 acres--nearly 65,000 compared to 11,000 in E. Punjab and Haryana. These differences in the size of holdings are an important factor in explaining the patterns of technological change that have characterized the rapid growth in these two regions. These trends have been accentuated in the eighties with important implications for both the types of mechanization and the rural employment found in the two regions. Overstatement of Land Inequality The data on the concentration of land holdings in South Asia do not present a true picture of the extent of inequality in the ownership and use of land assets in rural areas. The existing unequal distributions of ownership and operated holdings are both ameliorated by two facts: First, the distribution of holdings per households (or per farm) overstates the inequality, because households with smaller holdings also tend to be households with fewer family members. In Bangladesh, for example, households owning less than 1 acre of land had only 4.7 members on the average compared to 11.2 members per family for those owning over 15 acres. This negative relationship, i.e., between size of holding and number of family members in the household dependent on that holding, pertains throughout the sub-continent. As a consequence the distribution of per capita holdings are much more equitable than the data on holdings per households suggest. Second, the size of holding is not even a close proxy for the income potential of the land held, as land quality and productivity very significantly. Smaller size holdings are generally farmed more intensely and have higher percentages of area under irrigation almost everywhere in South Asia. This reduces the inequality in income potential between small and large - 38 - holdings within a given region. Further, given the enormous diversity in land quality, irrigation, and cropping intensities in different regions, comparisons by size of holdings across regions are inevitablty misleading. More arid areas and those with low land quality and lower productivity--e.g., Sind and Rajasthan-- generally have larger holdings on average than areas with assured irrigation--the Punjabs, Tamil Nadu--or adequate rainfall--West Bengal. Thus the raw data on the size distribution of holdings both within and across regions greatly overstate the real inequality in income potential between small and large holdings. A proper comparison would require adjustments by household size and land qualities. Available data are inadequate to allow such adjustments to be made, however, at national or state levels. If adjustments were made to bring out the real differences in the net returns per hectare on farms of different size in different regions, the resulting distribution of the per capita income streams from different size holdings would be much more evenly distributed. Although these facts are widely recognized, uncritical use of the data on land holdings continues unabated. C. Landlessness Concern over the problem of "the landless" has been growing in recent years. This is evidenced by a rash of recent publications on the subject 44/ so much so that a certain cynicism has set in. 45/ The major concern is that by identifying the rural poor with the "small farmer", many policies and programs currently being undertaken by both national governments and international agencies may be irrelevant or even counter productive. In particular it is widely believed that "the landless" (1) comprise a majority of the poor; (ii) their numbers and proportions are increasing dramatically; (iii) they are being bypassed by programs designed mainly to benefit the landed; (iv) the benefits of growth do not "trickle down" to them; (v) few if any programs have proved successful in increasing their incomes; - 39 - and (vi) apart from a radical program of land redistribution, little can be done to improve their prospects. Some go so far as to argue that growth has actually increased the poverty of the landless. Esman (1978) has succintly summarized this concern: 'One of the principal fallacies in the discussion of rural poverty in the third world is to regard the rural poor as an undifferentiated mass of "small farmers" ...In some countries there are many small farm households...but they are seldom the majority of rural households and they are certainly not the poorest. Below them in status, influence and material welfare are landless workers, tenants and sharecroppers and marginal farmers whose holdings are so small, often so fragmented, and of such poor quality that they cannot provide a livelihood from their holdings and must therefore deploy a large proportion of their family labor supply off the farm. Conceiving the rural poor casually as "small farmers" contributes to the continued neglect of those in the lower strata who are much poorer and in many countries far more numerous.' The extent to which so-called "small farmer strategies" will benefit different sections of the rural population depends mainly but not solely on the distribution of land ownership and use. 46/ Conceptual Confusion: Who are the "Landless" The problem of "landlessness" is grossly exaggerated. Part of the problem is conceptual, and another part reflects inadequate or poor data. Some of it is due to misleading and sloppy interpretation of the data--and some may even be politically motivated, i.e., by a desire to exaggerate the problems of poverty. At least three alternative definitions of the "landless" in rural areas are tenable: (i) those who own no land; (ii) those who operate no land; and (ii) those whose major source of income is wage employment. Each definition includes different but not mutually exclusive subsets of the rural population (typically these subsets overlap) and has different implications with regard - 40 - to control over rural and other assets, and how incomes are derived from them. The three are often hopelessly confused. Data on "landlessness" per se is meaningless without adequate means to differentiate between these categories. The data are perforce not up to the task. The first definition, which equates the rural landless with those who own no land in rural areas, is concerned with the ownership of productive rural assets. As we noted at the beginning of this chapter, the ownership of assets (and especially of land) helps to explain both the main sources of rural incomes and the hierarchies of rural wealth, status, and power. Those without land, lacking the main source of income, end up at the bottom of these hierarchies. Land accounts for most of the value of rural assets in the subcontinent. In India, it accounted for two-thirds of the value of all assets in rural households in 1971-72, while houses and buildings accounted for another one-fifth (Table A.22 in the annex). Most assets other than land also end up belonging to those who own land. 47/ Cultivating households (i.e., those who own and operate land) own over 93 percent of all rural assets; the value of their assets per household is 10-15 times greater on average than that of the assets of agricultural labor households, i.e., those who rely mainly on agricultural or other wage employment for their livelihood. Furthermore the distribution of assets is even more skewed than that of land ownership, and this extreme inequality is evident to each of the Indian states. Overall, nearly 75% of all rural assets in India were owned by the top 25% of the households while the bottom 25% owned less than 2% (see Tables A.23 to A.24 in the annex). Clearly, ownership of land is the most direct and effective proxy for the ownership of assets in rural areas in India. Although comparable data are unavailable for Bangladesh and Pakistan, a similar relationship between land and total assets is likely to prevail. The lack of assets is a major handicap for those who own little or no land. Their lack of collateral denies - 41 - them access to credit and thus, in effect, access to inputs. More generally, because social and economic stratification based on land ownership pervades all aspects of rural life, individuals' opportunities are initially defined and subsequently moulded by their access to productive assets, especially land. The second definition equates the landless with those who do not operate any land; in this case, the concern is with access to land as a major source of rural employment and income. The distinction is between cultivators and non-cultivators, and between different conditions of cultivation. The importance of this definition lies in the belief that those who do not operate "'enough" land--either as owners, or as mixed or pure tenants-- cannot earn a stable or adequate livelihood, and must therefore depend to some degree upon other sources of income and employment. Here it should be noted that although there is a widespread assumption that rural per capita incomes are deter- mined by size of holding, 48/ land is certainly not the only source of rural incomes, 49/ nor is there any identifiable direct relationship between the size of land holding operated and poverty. 50/ Household incomes can be critically affected by other factors -- including the quality of land, whether it is irrigated (and if so, how), the intensity with which it is cultivated, and the size of non-farm wage incomes and remittances; per capita income levels are affected by the size and age/sex distribution of the household, the ratio of dependents to workers, and the extent to which household members participate in the labor force. Finally, the third definition equates the landless with rural laborers, or with agricultural laborers. To quantify these groups, one must use data on occupations, not on land ownership or operation. This is because those who neither own nor operate rural land include many artisans and other workers in non-farm (and non-wage) employement. The sub-class of agricultural laborers within this third definition of landlessness is of special concern because its members typically earn less than the - 42 - already low average for rural incomes, depend mainly on unskilled jobs for a living, and have a relatively high level of indebtedness. 51/ They come mainly from a subset of castes (euphemistically called "backward classes" or the "weaker sections" or "scheduled castes" in India) that constitute the poorest element in rural society and typically face social and political deprivation and caste prejudice. These three definitions are not mutually exclusive. Those who own no land may operate some, and those who operate and even own land may still rely significantly or mainly on wage employment for their incomes. Any measures of "landlessness", however defined, must thus remain imprecise at best. The best way to look at the available data is to construct a four-way classification on the basis of the distributions of both owners and operators, made up of the following groups: (a) those who own and operate land; (i.e., landlord and mixed tenant cultivators); (b) those who own land but do not operate it (i.e., non- cultivating landlords); (c) those who own no land but operate some, (i.e., pure tenants); and (d) those who neither own nor operate any land (the purely landless). The subsections which follow attempt to apply this classification to those who live and work in rural areas in the subcontinent. As usual, however, data problems make it much harder to arrive at a clear picture in Bangladesh and Pakistan that in India. India A classification of the kind outlined above was first made by Minhas (1970) for India, using the 1960-61 NSS data. I have constructed Table 9 on a similar basis by incorporating the 1970-71 NSS data and extrapolating for 1980-81, on the assumption that past trends are likely to continue. 52/ The number of rural households "owning no land" is given by the column sum of groups C - 43 - Table 9: IlLA: AGRAIIE PROFnIL AND LIANDLSSNKSS IN RURAL INDIA (HOUSEHOLD NUMIES IN NILLIONS) Owning Not Owning Raw Total Land Land Operatins Land (a) NSS: 1960-61 51.81 (71.5) 1.62 (2.2) 53.4 (73.7) (b) NSS: 1970-71 54.7 (69.8) 2.2 (2.8) 36.9 (72.6) (c) (Est.) 1980-81* 64.38 (67.0) 3.33 (3.47) 67.7 (70.5) (Group A) (Group C) (A + C) Not Operating Land (a) NSS: 1960-61 12.2 (16.8) 6.8 (9.4) 19.0 (26.3) 0) NSS: 1970-72 16.1 (20.5) 5.4 (6.9) 21.5 (27.4) (c) (Est.) 1980-81* 23.6 (24.56) 4.78 (4.97) 26.38 (29.5) (Group 8) (Group D) (S + D) TOTAL (a) NSS: 1960-61 64.0 (88.3) 8.46 (11.7) 72.46 (100.0) (b) NSS: 1970-71 70.8 (90.3) 7.6 (9.7) 78.4 (100.0) (c) (Est.) 1980-81* 87.98 (91.6) 8.11 (8.4) 96.1 (100.0) (A + B) (C + D) (A + I + C + D) Note: Figures in brackets tive percentage of total in each year. B. Minhas: (1970): N5S * 16th Round, 1960-61; NSS: 25th Round 1970-71. - 44 - and D in the Table, while those "operating no land" are shown by the row sum of groups B + D. The following interesting facts emerge. First, the landless, defined as those who own no land, have declined as a proportion of all rural households: by 1970/71, only around 10 percent of all rural households were in this category. 53/ But by 1980-81, even if the proportionate decline had continued, the absolute number of such households would have risen by about half a million. 54/ This represents a rural population of over 38 million owning no land. 55/ Second, the landless, defined as those who operate no land, have increased relatively and absolutely. Nearly a third of all rural households -- over 21 million -- were in this category by 1970-71. If the proportional trends have continued, over 28 million households or 134 million people would belong in this category by 1980-81. A substantial proportion of those who do not operate land do, however, own some. There has been a significant proportional increase in the class of "owners who do not operate their land" -- many of whom may own plots that are too small to farm -- while the proportion of owner-cultivators in the ownership group (though not theitrabsolute number) has fallen. Third, if landlessness is defined in terms of those who neither own nor operate land, there has been a decline in both proportions and absolute numbers. Only 7 percent of all rural households were in this category in 1970-71. By 1980-81, some 5 million households (roughly 5 percent of the total), with a population of 23 million, would be in this group if past trends were to have continued. Overall, lack of access to land appears to have declined in rural India. The percentages of rural households owning no land and of those neither owning nor operating land have declined. The rise in the total number and proportion of rural households not operating land was accounted for by non-operating owners of land; these households did have access to land -- however small their plots -- but chose to lease it out. 56/ - 45 - Overall, more than 93 percent of all rural households in India derived some income directly from land in 1970-71, either as owner-operators, as mixed or full tenants, or as owners who rented out their land; this proportion changed only slightly over the decade for which survey data are available. The all-India data as usual hide marked regional differences. Compared with the Indian average of about 10 percent, the percentages of households not owning land were low in the North (4 percent) and the East (8 percent) in 1971/72. During the 1960s the precentage of "those not operating any land" increased somewhat in the North, West and South (as in the country as a whole), but fell in the East (see Tables 1 and 5). I do not have data for all states on those who neither owned nor operated any land, but Sanyal (1976) has analysed the NSS data for six Indian states; the resulting picture is quite revealing, and is given in Table 10. These data suggest how ambiguous and misleading the concept of "landlessness" can be if it is not carefully qualified. For example, "landlessness" in terms of "those not operating any land" has increased significantly in many agriculturally dynamic states (Punjab, Haryana, Gujerat). But should this be a cause for alarm as it is often made out to be? As development occurs, one would expect the proportion of rural households not operating land to increase in a dynamic and growing agricultural sector. This is the result of a larger proportion of the rural population moving away from agricultural occupations and into a growing and dynamic non-farm sector. This is clearly borne out by data on occupational classifications as I will show later. Contrast this with the data for the same category in two agriculturally stagnant states--West Bengal and Bihar. Here one could argue that increasing demographic pressures coupled with agricultural stagnation could have combined to force smallholders to sell their land and join the ranks of rural labor households, thus increasing the proportion of households not operating land. But in these states this type of "landlessness" has actually declined! Furthermore Table 10 INDIAM PflCIACU OF AJILNSS NOUSKNOLUS MDT OUfNC. AND NOT OPFRA IOG LAWN, SRIACT STAYRS, 1954-55, 1961-62 Awn 1971-72 State Percentage of Houiaeholds not Percentage of Nouseholds not Percentage of Nouqeholds Percentage of Notsqeholds Percentage of Owning Land OperAtivn Land Owning hit not Operating, Neither owninR nor Landless House- Operating hold leasing In land Round Round Routnd 2/ Rouind 2/ Round 8th 17th 26th Rth I/ 17th 26th 17th 26th 17th 26th 26th Punjah 36.8f6 12.33 7.14 36.92 19.n9 58.61 3n.51 s2.Qn 6.SR 5.71) ) …--fi6fi.Rl 1/ Haryana - - 11.69 - - 46R.n - 41.n5 - 6.Q4) Gularat - 14.74 13.44 - 25.41 13.75 11.7R 25.47 11.61 R.78 S.qA P. Andhra Pradesh 3n.12 6.61, 6.95 42.R6 37.9S 36.05 12.03 2Q.66 5.92 6.17 R1.5R 8thar 16.56 R.61 4.14 23.64 21.71 20.65 15.26 17.52 6.41 1.11 R6.56 Ueqt Bengal 20.54 12.56 9.7R 24.3n 31.R8 30.q4 24.21 23.0q 9.67 7.R5 66.65 Notes: 1/ CGves X of non agrtcultuiral holdings deemed comparshle hy Sanval to 17th & 26th round 2/ Rth rouind eRtimates are not avallahle. 3/ Includes Naryana. Source: NSS data reported by Sr Sanyal (1977) In Sarvekshana. - 47 - the percentage of households neither owning nor operating any land has also declined in all the states for which data are given. Mead Cain (1981) in a recent paper suggests that much of the NSS data has an upward bias because it shows many households "owning land" when most likely they own only non-arable land-- probably homestead plots. Although even these plots are used to grow crops in some cases, an adjustment in justified and Table 11 gives his adjusted figure for India. The evidence clearly suggests that "landlessness" even in terms of "those who own to land" has been declining in spite of the oft repeated claims to the contrary. 57/ These examples clearly demonstrate the ambuiguity of the concept of "landlessness"; it fails to distinguish the consequences of dynamic from stagnant processes that have the same outcome--fewer people owning or operating land. Bangladesh What little has been written on "landlessness" in Bangladesh is replete with confusion about definitions and data sources. Nonetheless it tends to be taken as axiomatic that the situation is worsening. What is the evidence? The Esman (1978) study cited earlier states that 75 percent of all rural households are "landless or near-landless." As already noted, the definitions used in that study present major problems. On the basis of LOS data, 11 percent of rural households "owned no land" in 1977, while 33 percent "owned no land except homestead land"; by 1978, the proportion of owners of no land apparently rose to 15 percent, while that of non-owners excluding homestead land apparently fell to 29 percent. These figures should not, however, be taken as indicative of a general trend; the year-on-year differences may well be due to sampling error. Unfortunately there are no other survey data that permit any deductions about trends in landlessness using a definition based on ownership. About 30 percent of all rural households in Bangladesh "'operated no land" in 1977, according to the LOS data in Jannuzi Table 11 TOTAL LANDLESSNESS AND NEAR-LANDLESSNESS ANONC HOUSEIIOLDS IN RURAL INDIA ACCORDING TO NATIONAL SANPLE SURVEY DATA, 1954/55 - 1971/72 Size of Ownership holding 8th Round 1/ (1954-55) 17th Round 2/ (1961-62) 26th Round 3/ (1971-72) (acres) 2 AdJ. Cumulative X Adj. 4/ X Adj. Cumulative Adj.4/ X Adj. Cumulative S Adj.4/ 0.00 21.09 (30.8) 23.09 (30.8) 11.68 (27.5) 11.68 (27.5) 9.64 (25.6) 9.64 (25.6) 0.01-0.49 18.01 (8.0) 41.10 (38.8) 26.23 (10.4) 37.91 (37.9) 27.78 (11.8) 37.42 (37.4) 0.050-0.99 6.16 (6.2) 47.26 (45.0) 5.31 (6.3) 44.2 (44.2) 7.45 (7.4) 44.87 (44.8) Notes: I For the 8th Round, ownership was defined as right of permanent and heritable poasebslon; landholdings was inclusive of all laud, regardless of purpose to which put. 2/ For the 17th Round, ownership was defined to Include ownership-like possessions. 3/ For the 26th Round, ownership end landholdings were defined as for 17th round. 4/ (AdJ.) gives the adjusted figures after correcting for errors in NSS data by excluding hosestead land and leaving the proportion of households with arable land in each ownership category. Sources: 8th: NSS Report No. 36, Appendix 3, Table 15, p.71 and Table 8, p.64. 17th: NSS Report No. 144, Table 3(1), p.7 and Appendix 1, Table 2, p.13. oo 26th: NSS Report No. 215, Table 2, p.67. Tables taken from N. Cain (1981) 'Landlessness In India and Bangladesh: A Critical Review of Data Sources', The Population Council, New York, (Working Paper No. 71, May 1981). - 49 - and Peach (1980). 58/ Again there are no comparable data from which to calculate trends. Jabbar (1978), reviewing the evidence available from various sources, cites the following figures: - The 1960 Census of Agriculture shows 17.5 percent of the total civilian labor force as "landless agricultural laborers." - The 1968 Master Survey of Agriculture reports 20 percent [of the rural labor force] as "landless agricultural laborers." - An IRDP Survey in 1974 of Phulpur and Thakuragaon districts (small samples of 118 and 114 [households/workers] respectively) gives 24.4 percent and 9.7 percent as the proportions of "landless workers." 59/ - Another IRDP Survey of some 7700 households in 12 districts reported an average of 37.6 percent of all households as "landless", i.e. (i) having no farm land and (ii) having homestead land not exceeding 33 decimals." - The 1974 Census of Population reported 25 percent of the rural population as "agricultural laborers." In addition: - The 1977 LOS, which shows 11 percent of households as having no land and 33 percent as having no land but homestead land. - A survey of 10 districts carried out by the Bangladesh Agricultural Research Council (BARC) on the incidence of landlessness in 1978, which reports that "landless labor households" made up 26-33 percent of "agricultural households" in the various districts. 60/ There was enormous variation by thanas (districts), with the maximum percentages of landless labor households reported in the range of 40-60 percent. A. R. Khan (1977) has attempted to reconcile some of these disparate sources in order to look at trends. Any such - 50 - effort is full of difficulties; 61/ nevertheless, his figures, are reproduced in Table 12, despite the fact that they pertain to "landless laborers" (an additional category of the poor, data on which are probably better examined along with data on labor households). Between 1951 and the late 1960s, landless laborers increased substantially as a proportion of all cultivators. Khan states that "the increase in absolute numbers of landless laborers was staggering; in one and a half decades since 1951 they increased by two and a quarter times, an annual compound rate of 5 1/4 percent. This process of increased proleterianization was brought about by the conversion of families owning small amounts of land into households of landless workers." (p. 155-156). A study by Adnan et al. (1978) has attempted a much longer view; its findings from various sources are also shown in Table 2.11. Of these, two sets of figures are comparable -- those from Ishaque (1944-45) and the LOS for 1977 -- because both define the "landless" as those with no land except homestead land. They show a rise of only two percentage points in the proportion of rural households who are "landless." Adnan et al. caution against "drawing any glib conclusions regarding a polarization thesis,'' suggesting that "the socio-economic matrix of rural Bengal contained levelling mechanisms counter-balancing the centrifugal tendencies of an increasing land-man ratio" (p.22-23). Landlessness almost certainly increased significantly after the Great Bengal Famine of 1942-44. Of course it is possible that landlessness declined following partition and the Zamindari Abolition Act of 1950, when land was transferred from Hindu zamindars (landlords) to Muslim peasants. Landlessness could subsequently have risen again following the famine of 1973- 74, so that "the percentage levels of landlessness in 1977 have already crossed those prevailing in the '40s. One may expect an acceleration in the rate of [increase in] the proportion of landless households in rural Bangladesh in the coming decades, ceteris paribus" (p. 25). - 51 - Tab le 12 LAXDLESSNUSS IN RUUaL RaNCLADES! (a) A.R. Khan (1977) Landless Labourers Number of 'Landless as Percentage of Labourers in Mil- Sources Year Total Cultivators lions (Ag. Census) 1951 14.3 1.51 (Ag. Census) 1961 17.5 2.47 (MSA) 1963-64 17.8 2.71 (MSA) 1964-65 17.5 2.75 (MSA) 1967-68 19.8 3.4 (b) Adnan et. al. (1978) Agricultural La- Agricultural Household borers plus Laborers with no Sharecroppers as as Percentage Land Other Percontage of of Total Than Sources Year Rural Household Cultivators Household (Flond Commission) 1938-40 30.61 _ _ (Mukherje: 6 Villages) 1944-45 35S - (Ishaque: 77 Villages) 1944-45 - - 29.91 (Ag. Census) 1951 - 15.32 - (Ag. Census) 1961 - 17.5 - (IRDP: 14 Villages) 1973-74 - 382 (BIDS: 8 Villages) 1974 26S - 381 (LOS: 400 Villages) 1977 - 32S - 52 - The Adnan survey also summarizes the findings of a number of studies 62/ that looked at "how long the landless had been landless...Of those who are landless at present (1978), most have been landless for two generations, a smaller proportion have been landless for three generations and the smallest proportion have been landless only in their own generation." This suggests that the growth in landlessness is not a recent phenomenon and that the landless have very limited opportunities for upward mobility (p.31). Abdullah et al. (1976) used 1961 census data and the 1973-74 IRDP Benchmark survey, and included sharecroppers and full time tenants to estimate the percentages of "landless" households, by district, for 1961. 63/ These are not estimates of those "not owning land", because some of the households included may have some land of their own. The figures, reproduced in Table A.26(a) in the annex, show that landlessness (as defined in the study) was extensive in many districts and had increased over the 13 year period in almost every district. Data from the 1978 BARC study are reproduced in Table A.26(b) in the annex. They also show that in 1973-74 "landlessness" was extensive in almost all districts. In this study the "landless" are defined as "those who live mainly by working on other farmers" fields. They may own homestead land, but they do not own farm land." The district figures are very close to the 33 percent for Bangladesh as a whole suggested for the same category in the 1977 LOS survey. What can be concluded from this welter of confusing and at times conflicting evidence? First, it is safe to say that no consistent and reliable set of data exists that allows us to say anything about trends in landlessness, in terms of either those not owning land or those not operating it (or their intersection), although trends have in fact been inferred from a large set of disparate and non-comparable studies. Second, what little is firmly known suggests that (a) "landlessness" in terms of non- ownership of land is significant (29 percent of all rural - 53 - households), (b) the percentage of "landless labor" to total rural households is large--around 30 percent, and (c) this latter category has been growing in both relative and absolute terms--and its growth may have accelerated in recent years. As Cain (1981) points out "we have some confidence that the percentage of rural landless households stood at 29 in 1978, but beyond that little can be said". Pakistan Data on "landlessness" in Pakistan are practically nonexistent. This is a serious lacuna in the agrarian data base. Information can only be derived by subtracting the number of "farm households" in the agricultural censuses from the number of rural households. (Even the number of rural households must first be calculated by dividing the figure for the total rural population by the assumed average size of rural households.) This exercise is clearly approximate and imprecise, but by using an average household size of 5.43 for 1960-61 and 5.8 for 1972, 64/ and available data on the rural population, one can arrive at approximate estimates that are shown in Table 13. From these one can infer the following: - In 1960-61, out of a total of some 6.5 million rural households, 4.9 million were "farm households;" the remainder--some 24.6 percent of all rural households-- were "non-operating" or "non-farming" households. - By 1972, there were some 8.2 million rural households, of which only 4.0 million were "farm households." This means that over 51 percent of all rural households were in the "non-farming" category. (A similar exercise using provincial data yields 58 percent for Sind and 50 percent for the Punjab). - Thus the proportion of rural households "not operating" land may have increased from around 25 percent to 50 percent between 1961 and 1972. - Between 1951 and 1961, the number of "landless laborer" households increased from 0.14 million to 0.61 million - 54 - Table 13: PAKISTAN: RURAL HOUSEHOLDS, LANDOWNERSHIP, NaNCY AND ESTIMATES CF LANDLESSNESS (1961-19"2) 1961 1972 1) Rural Population (millions) 35.8 47.4 2) Rural Household Size (persons) 5.5 5.8 3) Rural Households (millions) 6.S 8.2 4) No. of Land Owners (million persons, 1950s) 5.1 10.1 5) Total Area Ovned (million acres, 19509) 48.6 67.6 6) No. of Farm Households (millions) 4.9 4.0 of which 2 operating as owner cultivators (X) 41.1 41.7 owner/tenant cultivators (2) 17.2 23.8 tenant cultivators (2) 41.7 34.4 7) Area Oporated by Farm Households (oln A) 48.9 49.1 of which 2 operated by owner cuitivators (2) 38.3 39.5 owner/tenant cultivators (2) 22.5 30.9 tenant cultivators (2) 39.2 29.6 8) Proportion of Farm Households Cultivating Area other that that Owned (2) 58.9 58.2 9) Proportion of Area Operated by Farm Households which are not owned by the operating household 50.5 46.2 10) Estimated No. of Rural Households not Operating Land (millions) 1.6 4.2 11) Estimated Landless Agricultural Labor Households (million) 0.6 1.6* Note: * Estimated using 1961 proportion ((11)/ (11) + (10)) Source: 1) World Bank Country Ewn Report No. 2860 pg. 74; 2) GOP "FAO/WCAARD Country Paper of Pakistan' pg 39 and Afzal, H (1974) pg. 47; 3) Computed from 1) and 2) (Note: Various household sizes and estimates of rural households are available, the chosen ones are deemed most reliable); 4) three 9) from M.H. Khan (1980) citing censuses; 10) by mare subtraction 3)-6); 11) from M.H. Khan citing Census of Population, 1961 Table 51 wherein approximately 2.0 million tenant farmer-vorkers' confirmed by 6) justified estimating 565.435 landless laborers to belong to I household each. - 55 - according to the population census--a 350 percent increase in the decade. - "Landless agricultural labor" households may have increased from 10 percent to around 20 percent of all rural households between 1961 and 1972. One cannot draw conclusions about magnitudes or trends in "landlessness" with any degree of certainty from the available data. This type of indirect and incomplete evidence has, however, been cited by Naseem (1977) as providing "ample justification for the conjecture, supported by some village surveys, that the number of landless workers is increasing as in the past." The village survey in question is a study by J. B. Eckert (1972) carried out in the West Punjab, showing that "permanent labor plus temporary labor" (which Naseem translates into "landless workers") accounted for 12 percent of the population. As the 1961 data showed this category to be 9 percent of the whole rural population, and "as the figure for Punjab would be lower than the average figure for Pakistan," Naseem argues that the proportion of landless workers must have increased. The 0.61 million "landless laborers" become 0.61 landless labor households in 1961, on the assumption that one landless worker represents one household. (I have retained this heroic assumption in Table 13 to show exactly how shaky such an exercise can be.) A large proportion of the "landless" labor force may be working as pure tenants, who operate nearly 30 percent of all farms. But in order to enumerate the people who own no land, or those who neither own nor operate any land, I would need to look at data on the rural labor force by occupations. These data as we shall see, are also inadequate, so that the numbers, occupational status and other characteristics of the "landless" in Pakistan will remain unknown until new surveys are carried out. Conclusion This discussion of "landlessness" in the subcontinent has tried to demonstrate, first, that the available data on the incidence of landlessness are extremely unreliable and, second, - 56 - that trends over time in the extent of "landlessness" (however defined) cannot be accurately identified. More generally, even relatively accurate information on landlessness as such can tell us relatively little about changes in the living and working conditions of the rural poor. The notion of "landlessness" is vague and subject to misinterpretation; though the notion is of some value, it should not be of primary interest. Nor do trends in "landlessness" signify much in and of themselves. They need to be examined along with other evidence on the changing agrarian structure. Instead one should concentrate directly (i) on the conditions of cultivation for those who operate land--the size of holdings, its productivity and tenurial status--and (ii) on the conditions of wage employment for those who depend primarily upon wage labor in rural areas for their incomes. For this purpose, a close look at the occupational distribution of rural labor households is more helpful. This I do in the next section. ACCESS TO JOBS: RURAL WAGE LABOR AND EMPLOYMENT This section begins with a brief overview of the numbers of people who rely mainly on wage employment, it then examines the changing conditions of employment and unemployment in rural areas, and concludes with a discussion of the different types of occupational opportunities--agricultural and non-agricultural-- available to wage laborers. A. The Rural Wage Labor Force Data on rural labor are sparse, except in India. Nevertheless, a common trend is observable throughout the subcontinent: a rising proportion of all rural households depends partly or primarily upon wage employment, or on occupations other than operating farms as owners or tenants. As a corrollary, the proportion of those dependent primarily on farming as a major source of income is declining. It has already been suggested that this trend could reflect either or both of two contrary and - 57* - opposing processes at work in South Asia: (1) the rapid growth and transformation of agriculture, which is diversifying the structure of rural occupations towards more productive jobs in the service and distribution sectors in rural areas, and/or (2) agricultural stagnation combined with rapid population growth and relatively fixed land resources, which is forcing more and more people to rely on low productivity wage employment in service or other activities in rural areas. Magnitudes and Trends The subcontinent's rural labor force was an estimated 260 million by 1978. As Table 14 shows, the countries of South Asia differ significantly in the percentages of their populations in the rural labor force and the distribution of rural workers among broad categories of activities. 65/ For example: - In India nearly half of the rural population was classified as rural labor, by comparison with under one- third in Bangladesh and Pakistan. - Only a quarter of India's and a third of Pakistan's rural labor forces were working on their own farms as their usual activity; in Bangladesh, nearly half fell into this group. - About 30 percent of the rural labor force in Pakistan was not working in agriculture, compared with about 15 percent in India (comparable data for Bangladesh are unavailable). 66/ - The proportion of the rural labor force accounted for by those described as casual wage earners or agricultural laborers was nearly two-fifths in India, about a quarter in Bangladesh, and a mere 6 percent in Pakistan. In this chapter I looking more closely at these latter two categories, especially the last (casual/agricultural laborers). Altogether, over 100 million people in rural South Asia fall into this group. Who are these people, and what are the trends in their numbers? Table 14: SWUN ASU: 'in _IuI n _L L rAM N w U L. wriVT s3mW 1972rig Pakistan India Bangladesh South Asia 1974 est. 1985 1972-73 Ent. 1985 1974 Est. 1985 (est. 1985) Rural Population (Hlns) 50 69 455 573 65 86 728 Rural Labor Force (Mlns) 15 (30.0)* 20.7 200 (44.0)* 252 16 (24.6)* 26.5 (30.8)* 299 (41.5))* Proportion of Rural Labor (100) (100) (100) (100) -Working on own farm (Hin) 5.7 (37) 7.5 50.6 (25) 63.0- 7.5 (48) 12.7 83.3 (28) -Working as Casual Wage Earners/Agr. Labor (Nln) 0.9 (6) 1.2 78 (39) 98.3 4 (25) 6.6 106.1(35) -Working as Helpers (Cln) 4.6 (30) 6.2 45.6 (23) 58.0 4.3 (27) 7.2 71.4 (24) -Working In Non- agricultural Sector and/or Seeking Work 4.3 (28) 5.6 25.1 (13) 32.8 ** ** 38.4 (13) Notes: In Bangladesh only the Agricultural Labor force was considered. While those working on own farms Include part owners generally, a pure share cropper is also Included here. In India the available breakdown was for farm and non-fare work combined, It was assumed that 742 of the labor was for agricultural activities. The numbers reduced are shown and the last category Increased accordingly. For Pakistan the 1972 Agricultural Census gives higher estimates for labor force than- the Labor Force Survey used here. 1978 esttiates assume there is no change In the labor structure from that which is known. * Percentage of total rural population. 0* Included In other categories. Sources: 1) Government of Bangladesh, 'Agrarian Structure and Change: Rural Development Experience and Policies In Bangladesh' Dacca, May 1978, pg. 64. 2) World Bank, -Economic Situatton and Prospects for India,' Report No. 2431, April 1979. Table 1.8. 3) Government ot India fDraft Five Tear Plans Y115-W3. Vol. IIPlanning Comissilon, New Delhi, 1978. Table 1, pg.. 100- 3, 127. 4) Covernment of Pakistan, Ministry of Finance, Planning and EcOnOmic Affairs, 'Labor Force Survey, 1974-75 pg. XIX, 67 and 81. - 59 - India In 1972-73, India had a rural labor force of some 200 million. Growth at 2.5 percent a year has added nearly 52 million persons by 1985, 67/ and the increment between 1978 and 1983 is likely to have been even greater, at some 34 million. 68/ The Rural Labor Enquiries of 1964-65 and 1974-75 distinguish two categories of rural wage labor: "rural labor households", whose major source of income is "wage-paid manual labor", that is, those for whom "wage employment for manual labor contributed more towards ... income in the 365 days preceding the date of the survey than other two sources taken individually", 69/ and "agricultural labor households", a subset of rural labor households whose labor is mainly concentrated on specifically agricultural activities. The trends in the rural labor force and its composition are harder to quantify due to lack of comparable data. The best comparable data come from the 1964-65 and 1974-75 Rural Labor Enquiries and are summarized in Table 15. 70/ Several features of the 1974-75 data stand out: - of nearly 82 million rural Indian households in India, only a third (some 30 million households) could be classified as rural labor households--that is, households primarily dependent on wage labor;71/ - of these 30 million, over 25 million (84 percent of rural labor households, but only a quarter of all rural households) could be classified as agriculture labor households--that is, households primarily on wage employment in agriculture and related activities; - there is considerable regional diversity, with a much higher proportion of rural households counting as labor households in the East (35 percent) and South (41 percent), than in the North (19 percent) and West (25 percent) in India; 72/ - of all rural labor households, a little more than half (51 percent) were "without land"--that is cultivated Table 15: IDIA: CAUlSl IH RURALA1KE, 1964-65, 1974-75 Rural Labor Households Agricultural Labor Households Rural With Without With Without House- As 2 of All Land Land As 2 of All Land Land hold (1) (Millions) (1) (Htillions) 1964-65 North 16.4 (15.2) 2.5 -1.1 1.4 (13.4) 2.2 1.0 1.2 West 17.5 (22.3) 3.9 1.4 2.5 (18.9) 3.3 1.3 2.0 South 18.3 (33.3) 6.1 2.4 3.7 (30.0) 5.3 2.0 3.3 East 18.2 (29.7) 5.4 2.8 2.5 (24.2) 4.4 -2.4 2.0 All India 70.4 (25.4) 17.9 7.7 10.1 (26.6) 15.2 6.7 8.5 1974-75 North 19.6 (18.9) 3.7 1.7 2.0 (14.8) 2.9 1.4 1.5 West 19.7 (25.4) 5.0 2.3 2.7 (21.3) 4.2 2.0 2.2 South 21.1 (40.8) 8.6 3.9 4.7 (34.6) 7.3 3.3 4.0 East 21.7 (35.0) 7.6 4.2 3.4 (29.5) 6.4 3.6 2.8 All India 82.1 (30.3) 24.9 12.1 12.8 (25.3) 20.8 10.3 10.5 Decade Z Change (Increases) ° North 20 48 55 43 32 40 25 West 13 28 64 8 27 54 10 South 15 41 63 27 38 65 21 East 19 41 50 36 45 50 40 All India 17 39 57 27 37 54 24 (Figures In brackets give rural labor and agricultural labor households as I of all rural households) Source G.O.I., Ministry of Labor, Rural Labor Enquiry 1974-75" Note: Agricultural Labor (drawing over 1/2 of their incomes as wages for agricultural work) households are 852 of the rural labor (drawing wages for labor In rural areas) household and 222 of all rural households. - 61 - neither owned or leased-in land--and hence were totally dependent on manual wage labor, but these constitute only 15.6 percent of all rural households. Again these proportions are higher in the South (22 percent) and lower in the North (12.2 percent) and West (13.7 percent): thus, while a third of all rural households are primarily dependent on wage employment, only a ninth are totally dependent on it. These are the rural landless labor households. Their absolute numbers increased by nearly 29 percent in the decade since 1964- 65; - rural labor households have increased at a faster rate than rural households--this growth has been in excess of 4 percent per annum in all regions of India except the West: agricultural labor households have increased in a similar way, but at a slower rate of around 3.5 percent per annum; - the numbers of rural labor households "with land" have increased at the highest rate and their proportion has also grown significantly; - the average size of rural and agricultural labor households increased from 4.51 to 4.7 persons for the former and from 4.47 to 4.76 for the latter during the decade. Later comparable data are unavailable: with the growth of population, however, the growth of rural and agricultural wage labor is likely to have continued. The dynamics of the rural labor force in India are clear. While the absolute and relative numbers of those dependent primarily on manual wage labor have increased, the proportions of those dependent totally on wage labor have gone down, (although their absolute numbers have not). Thus of 30 percent of the rural households primarily dependent on wage employment, about half are landless laborers, but the other half are small and marginal farmers. This means that the significant increases in the ranks - 62 - of agricultural labor have occurred through a process where an increasing number of those with small or marginal holdings have been pushed increasingly to rely upon wage employment, until it has become a primary source of income for them. I have already documented the main processes that have contributed to this outcome: (i) the subdivision of holdings into smaller and smaller cultivating units via inheritance and in the face of increasing population pressure on land; (ii) the repossession of land held by tenants; (iii) some distress sales of land by poorer farmers and to these may be added; (iv) a rapid rise in the cost of living, which may have forced small farmers to supplement their incomes via wage employment so that labor displaced cultivation as a principal source of their income. 73/ In practically every state and zone in India, the 1960s saw increases in the proportion of the rural labor force that depended on agricultural wage labor (See the state-level data presented in Table A.27 in the annex). K. Sundaram's (1977) analysis of census data shows that for Uttar Pradesh, West Bengal, Punjab, Rajasthan, Andhra Pradesh, Karnataka, Tamil Nadu, Gujerat and Maharashtra, these increases were associated with rising proportions of non-cultivating households. Thus, those who cease to be cultivators in agriculture are not being fully absorbed into other sectors of the economy, as one might expect in a process of economic transition; instead, they are joining the ranks of agricultural labor. In other states, whose proportion of non- cultivating households fell -- Assam, Bihar, Orissa -- households with very small holdings also turned increasingly to agricultural wage labor. The smaller the size of holding, the greater this dependence on wage labor (see Table A.28 in the annex). As Bardhan (1973) has pointed out, the decline in the proportion of cultivators and the increase in the proportion of agricultural laborers in the rural work force were strongly correlated with rising rural poverty. For rural laborers, wage employment opportunities (as measured by the average number of days of wage employment per - 63 - year) have been falling, particularly since the mid-1960s, (see Table A.29 in the annex). 74/ Though the decline in wage employment in agriculture was partially offset by increased self- employment, a decline in the total number of full days of rural employment is evident. As Raj Krishna (1976) has shown, for India as a whole "all unemployment estimates point to a serious worsening of the rural employment situation in the early seventies in comparison with the sixties, owing to the stagnation of agricultural output and a series of bad harvests." 75/ These trends will be explored further in the section which deals which deals with rural unemployment. Nor have real wage incomes improved significantly. Although nominal average daily earnings rose sharply between the mid-1960s and the mid-1970s (see Table A.30 in the annex), the figures need to be adjusted for the decline in hours worked (10-15 percent) and the rise in the cost of living (well over 100 percent) during the decade. In fact, data from rural labor surveys over extended periods since the fifties show that real incomes from agricultural wage employment stagnated during this period, at least on an all-India basis (See Table 16). Some Characteristics of Indian Rural Labor "Small farmers" and "agricultural laborers" in India do not form separate and discrete classes. They differ only in the degree to which they rely on wage incomes and participate in labor markets. Nearly half of India's rural households (Table 15) both cultivate some land and depend primarily upon wage labor Nearly one-fifth of those with some land operate holdings of over two acres (Table A.28 in the annex). This fact, and the diverse sources of income for both agricultural labor and marginal farm households (See Tables A.31 and A.32), emphasize how misleading it can be to assign individuals to simplistic categories -- farmers versus laborers, or agricultural versus non-agricultural workers. To assess either the levels of living of the rural poor or policies to improve them, one must first understand that they derive their employment opportunities and their meager incomes - 64 - Table 16 INDIA: 1/ EMPLOYMENT CONDITIONS OF AGRICULTURAL LABOR HOUSEHOLDS- A COMPARISON OF RURAL LABOR SURVEYS 1950/51 1956/57 1964/65 1974/75 Est. Number of Agricultural Labor Households (millions) 17.9 16.3 15.3 20.8 Agricultural labor House- holds as a proportion of Total Rural Households (2) 30.4 24.3 21.8 25.3 Days of Wage-paid Employment Agriculture Men 189 194 217 193 women 120 131 149 138 Average Number of Earners per household 2.00 2.2 2.01 2.34 ,rage Annual Earnings per Household (Rupees, current prices) 288 277 452 1024 (Rupees, constant prices) a/ 606 767 634 595 (661) _b/ Real Earnings, per earner 303 349 315 254 (282) 1/ Data on non-agricultural labor households was only collecced beginning with 1964/65 survey so that no trend can be established. a/ The appropriate deflator for agricultural laborers does not extend back earlier than 1960/61. An approximation was made using the wholesale price index for food articles, which do at any rate constitute the major portion of the agricultural laborers expenditure. For the period between 1964/65 when both indices are available the increase in the food articles index was 412, in the agricultural laborers general index 133Z. b/ The figure in parentheses indicates the constant price value for 1974/75 using the price deflator for a three year period centered on 1974/75 as foodgrain prices in that year were extraordinarily higher than elther the preceding or subsequent years. Source: Rural Labor Enquiry data as conpiled in Grawe at. al (1979). - 65 - from many sources. (This topic will be examined more fully in Section C below.) It is an old adage in India that rural laborers are "born in debt, live in debt and die in debt." Even after a decade of remedial measures, they seem to have become more indebted and more dependent on informal (and costly) sources of credit, rather than institutional sources. 76/ Two thirds of agricultural labor households were in debt in 1974-75 (up from 61 percent in 1964- 65). The average debt per household in 1974-75 was Rs 584, or 139 percent of that in 1964-65. "Money lenders" were owed nearly half of the total debt, compared with under a third in 1964-65; 77/ another quarter came from relatives and friends; less than 5 percent from cooperative societies; and another 4 percent from banks. Rural wage employment is predominantly casual. Data from the NSS 27th Round indicate that in 1972-73 a large proportion of those who worked in rural areas worked either as "farm helpers" or casual wage labor. Together these categories accounted for 44.5 percent of the total rural labor force of 83 million. In India as a whole in 1972-73, casual laborers alone made up as much as 27 percent of the rural labor force. This proportion was particularly high in some states (Maharashtra, Andhra Pradesh and Kerala, for example) while in the North it was relativel low (Table A.27 in the annex). Parthasarathy (1987) confirms the increase in the overall percentage of wage labor and the growing casualisation of wage laborers, which in 1983 formed 72.8 percent of total wage laborers. (Table 17). Wage employment is more likely to be casual for women than for men. Of the women in the rural labor force in 1972-73, 70 percent were farm helpers or casual wage workers. The labor force participation rates for rural women have been fairly constant (at around 29 percent from 1959-79), but longer-term trends show that womens' participation in household industries, trade and commerce -- that is, opportunities in the modernizing sectors -- have been significantly diminished, throwing women - 66 - Table 17 India: Composition of the Rural Male Work Force in Indian States: 1972-73 and 1982-83. twmai vep labmuw as a aew b _ *A mn u pNe w' i972.73 196283 1972n73 196243 (1) (2) (3) (4) (5) 1. AdhPdmb .. .. 400 46-6 68-6 74-2 2. Am .. .. .. 23-8 35-9 40-0 4-60 3.& Dhi .. .. .. 39-5 42 8 60*9 U-2 4. GtmW ..at .. .. 34-4 40-4 64-6 81.5 5. } ..... .. .. .. 24-0 30.0 40-3 52 3 6. J a a KMbmit .. .. 9-4 26-8 20-9 697 7. MmU. ... .. .. 37-6 413 72n 1 7.3 8 gh. .. .. .. 3 5345 72 1 76.7 9. b Puid.h .. .. 27-3 3.7 56-8 n-0 lo. ashus .. .. .. 53-5 7.4 70-3 71-5 It. O. . .. .. 39-9 43.7 6s.3 76-8 12. ab . .. .. 30-7 33-6 52 3 61*7 13. R*,h . .. .. 10-2 20.5 53-5 63-1 14. Tam NAu 42.4 54-2 60-0 75-1I 15. UtaPftndh.. 216 24-1 66.7 71-6 I w. WUsevpl .. 4..-1 50*4 .146 75 AU-In" ..... ._ _ 34.1 3-G* 4 f nle Sw: 1. NSS 27h Rami, Uh Oc'otr 1977. 2. N- 3th Round, Rvort No. 315, c3d byji (1965). Source: Parthasarathy, G. "Changes in the Incidence of Rural Poverty and Recent Trends in some aspects of the Agrarian Economy", Indian Journal of Agricultural Economics, 42 January - March 1987. Table III, p.8. - 67 - who cannot afford not to earn wages back into the agricultural sector as laborers. 78/ As social and family barriers undoubtedly discourage it, women's employment as casual agricultural wage labor may be taken as indicative of the distress in which many families of marginal farmers and landless laborers find themselves. Rural labor is unorganized. The Rural Labor Enquiry of 1974-75 revealed that only 1 percent of all agricultural laborers belonged to trade unions. Except in one state -- Kerala -- where the Communist Party is active, there were no organized unions in rural areas, and only 2 percent of laborers were even aware of the Minimum Wages Act or the minimum wages fixed under this Act in various states. 79/ One may conclude that Indian rural wage labor households: - are increasing at a rapid rate of 3.5-5 percent a year; - form a growing proportion of all rural households; - face falling employment opportunities and real incomes; - rely upon a variety of sources of income, including non- agricultural work and (if they have any land) earnings from cultivation, to make ends meet; - rely to a large extent on casual work; - are increasingly indebted, and - are unorganized. Without assets, generally illiterate and unskilled, and without political power or resources, these households are at the bottom of the ladder socially, economically, and politically. Bangladesh In 1977 the agricultural labor force in Bangladesh was estimated at around 21 million out of a total rural population of some 76 million. Available estimates, which are shaky and incomplete, put the average annual rate of growth in the labor force growth at around 2.1 percent and the growth of landless households at around 4-5 percent a year. 80/ Although the use of wage labor is fairly widespread in rural Bangladesh, the data on rural labor households are - 68 - sparse.81/ Those available are recent and are summarized in Table 18. In 1978, about 5.3 million households, or 40 percent of all rural households, were classified as rural labor households -- i.e. ones that depended primarily upon manual wage employment. About 75 percent of these households depended primarily upon manual wage labor in agriculture. Again in 1978, about a fourth of all rural households neither owned nor operated any land, and can have been presumed to have been wholly dependent upon manual wage employment; in that year this category would account for a population of over 15 million, 82/ while those dependent primarily upon manual wage labor in rural areas would be around 25 million. The rural labor force in Bangladesh is similar in many ways to that in India -- growing rapdily in absolute and relative numbers, indebted, unorganized, employed in casual work and dependent for its livelihood upon a variety of sources, including farming. 83/ Landless laborers known as kamlas work mostly on a daily wage basis for landowners (or even for sharecroppers), and are entirely dependent upon their employers for work, wages and credit. Studies have established that patterns of sharecropping and hiring labor to some extent follow kinship lines, and that ownership and non-ownership of land, and whether one hires or is hired as a manual laborer, are important factors in one's social status. 841 Those who depend on manual labor for their income are at the bottom of the socio-economic scale as elsewhere in the subcontinent. Proportionately fewer rural labor household in Bangladesh operate land than in India. Of over 2,300 rural laborers who participated in the Food for Work program in 1976, 57 percent had no land; three-quarters of the remainder, 29 percent of the total, operated less than one acre. The average cropped area of those who operated some land was about half an acre -- a very meager holding. 85/ These laborers derived a third of their incomes from sources other than farming and manual labor; nearly 42 percent of the households had a member who held some salaried job, and another 25 percent sold home-produced products.861 - 69 - Table 18: BANGLADESH: RURAL LABOR HOUSEHOLDS IN BANGLADESH % Change 1973-74 1973-74 1977 1978 to1978 (1) Rural Population (mlns) 67.6 76.0 78.0 15.4% (2) Rural Households (mlns) 11.1 11.8 12.0 19.8% (3) Households not owing (a) and not operating any land (mil) 3.7 3.3 3.1 -17.2% (%) (33.3) (27.9) (23.3) (4) Estimated Rural Labor Households (mil.) 4.4 4.7 5.3 +20.4% (%) (39.6) (39.8) (40.0) (5) Estimated Agricultural Labor Households (mil.) 2.8 3.4 4.0 70.0% (%) (25.2) (28.8) (29.2) (6) Average Household size - 6.4(b) 5.8(b) - (Figures in brackets give percentage of all rural households in the category) (a) those who owned no land other than homestead land; (b) estimated simply by dividing the population by number of rural households. Sources: (1) GOP and W.B. population estimates. (2) Jannuzzi and Peach (1977, 1978). (3) G.O.B. Agr. Census (1960) and Januzzi and Peach (1978). (4) Master Survey of Agriculture (1967-68) cited in WCAARD and Robinson (1969). (5) Household Expenditure Survey data cited in Jabbar (1978) and assumptions on labor force in WB Economic Reports, specially Bangladesh: Recent Economic Trends and Medium Term Development Issues, March 4, 1983 and Bangladesh Food Policy Issues, Dec. 19, 1979. - 70 - Real wages for manual labor in Bangladesh declined dramatically in the late 1970s. Aggregate data on days of employment are unavailable, but a study from a sample of 120 labor households in Phulpur in 1969-70 showed that a third of the labor force was employed less than 100 days a year, another 40 percent between 100 and 200 days, and, surprisingly, over 20 percent were employed for over 200 days per year. 87/ It is not clear whether employment opportunities for rural labor, as measured by days worked per year, have fallen over time as in India; no time series data are available, but a number of studies seem to imply a decline. 88/ Pakistan Data on rural labor in Pakistan are the scantiest of all. Estimates residually derived from the Agricultural Census, together with data on rural population and some direct estimates of agricultural labor from the Population Census, are given in Table 19. The estimates depend crucially on the assumption made about the average size of rural households. More recent data are unavailable. Pakistan differs from India in that the numbers of agricultural laborers have been rising far more slowly than the rural population at large. Thus, although the absolute total of agricultural laborers increased by some 28 percent between 1961 and 1972,, their proportion of the rural population fell, from around 18 percent to 15 percent. While the number of "rural households not operating land" increased by some 150 percent in the decade, the number of agricultural labor households" rose by only 20 percent --suggesting a massive diversion of household employment into other occupations. Which occupations? The 1972 Agricultural Census shows that out of a total of 5.5 million "agricultural households", 4 million operated farms, but 1.5 million were classified as "livestock holders". Naseem (1979) argues that during the 1960s a large number of small and marginal farmers were forced to sell or leave their land and eke out their existence with livestock - 71 - Table 19: RURAL LABORERS IN PAKISTAN 1961 1972 1977-78 (1) Rural Population (mins) 42.9 65.3 52.2 - Punjab 25.6 37.8 47.7 - Sind 8.5 14.2 67.1 (2) Rural Household Size 5.5 5.8 (3) Rural Households (mlns) (a) 7.8 11.3 44.9 (4) Farm Households (mlns) 4.9 4.0 -18.4 (5) Rural Households not operating land (mins) (b) 2.9 (37.2) 7.3 (64.6) 150.7 (6) Agricultural Laborers (mlns) 7.6 (17.7) 9.7 (14.8) 27.6 - Punjab 4.8 5.8 20.8 - Sind 1.6 2.4 50.0 (7) Agricultural labor (c) Households (mlns) 1.4 1.7 19.5 Notes: (a) Derived by dividing population by household size (b) Residually derived (3-4) (c) 6 + 2 (Figures in Parenthesis are percentages of total rural households) Sources: (I) Census of Pakistan, 1961, vol. 3 (II) Pakistan Economic Survey, 1977-78 (III) Pakistan Housing Economic and Hemographic Survey Vol. 2 (IV) Pirt 1, 3 and 5 (V) All (I) - (IV) cited in M.H. Khan (1979) and G.O P / F.A.O. "WCARRD Country Paper for Pakistan (pg. 29 for household size) (VI) M. Afzal (1974) - 72 - (buffaloes for milk) and especially small livestock (such as goats). Some of the landless work as shopkeepers or artisans (such as blacksmiths, carpenters, cloth weavers, potters, leather workers) -- supplying services or goods throughout the year for payment at harvest time. Relatively skilled landless artisans have in some cases done better than people with land. Eckert's (1972) detailed study from a survey of 40 villages in the Punjab in 1971 is the only one that focuses on rural labor and employment in Pakistan. Sixty-nine percent of the households he studied farmed their own or leased-in land. Of the 31 percent who were non-farming households, some 13 percent were classified as labor households (3 percent permanent and 10 percent temporary) and the remainder worked as artisans or shopkeepers. These proportions differ from those given by Naseem (1979), who estimated that in 1970-71 there were some 3.2 million laborers and over 5.2 million artisans and shopkeepers. Though any general statement is at best conjectural, it seems likely that employment in non-wage occupations has absorbed much of the increase in Pakistan's non-farming population, particularly in the Punjab. Such a change would be somewhat like that in East Punjab (India), where non-farming households have also increased dramatically as a share of the rural population, and a high proportion of them are engaged in non-wage employment. It is the households of temporarily hired landless laborers, those who are without skills and are employed for less than half a month on average, that are akin to agricultural labor households elsewhere in the subcontinent. In the Pakistan Punjab, they account for around only about 8 percent of all rural households. Naseem estimated that in 1970-71 perhaps 60 percent of hired laborers were only temporarily employed, worked one-third as many days as permanent workers (11.3 days/month against 29 days) and faced a constant struggle to find work. Temporary workers were employed for harvesting by almost all farms, while permanent workers were employed mainly on large farms. He - 73 - estimated that "more than 2 million Punjabis in the landless labor class of rural residents lived at the level of half a rupee per day per person." Recent data on the overlap between marginal farmers and agricultural laborers are unavailable, but data from the 1961 Census show that out of a total of 3.2 million agricultural laborers in Pakistan, only half a million were "landless" -- the rest rented in or owned some land. Wages have generally risen in the Pakistan Punjab. Nominal wages for peak-time agricultural work rose by 52 percent in the 5 years from 1965-66 to 1970-71 (Eckert, 1972). Artisans' incomes rose in real terms. In 1970-71 the per capita incomes of permanent laborers were not much lower than tenant farmers', but those of temporary workers were at least 10 percent lower. 89/ On the basis of the admittedly partial and limited data cited, it seems reasonable to conclude that, although the absolute number of rural labor households has been rising in Pakistan, the proportion of the rural population relying wholly on agricultural wage labor is relatively small and declining; instead, increasing numbers of households rely wholly or significantly on earnings from livestock and artisanal or trade-related activities. This is in sharp contrast to the rest of the subcontinent except Northwest India -- Punjab and Haryana -- where similar conditions prevail. B. Rural Unemployment Despite the serious limitations of the available data, it seems reasonable to conclude from the previous section that more and more people in rural South Asia are coming to seek their livelihood through wage employment, rather than through the cultivation of land they themselves hold as owners or tenants. During recent years, however, the rate of growth of employment opportunities in the rural economies of the subcontinent has failed to keep pace with the growth in the numbers of those seeking jobs. Estimates of the imbalance between work and workers in rural areas vary, depending on how employment and unemployment - 74 - are defined and how the available quantitative estimates are interpreted. 90/ In this section, we are concerned with two issues: first, we want to establish the broad magnitude of the "unemployment" problem (and hence the scale of the efforts that might be needed to tackle it); second, we want to understand the forms unemployment takes and the characteristics of the unemployed, so as to be able to evaluate the kinds of measures that might be appropriate to raise employment levels. As Sen (1975) has pointed out, more has been written on unemployment in India than on unemployment in any other country. Nevertheless, the estimates of rural unemployment in India vary widely, as do the figures for Pakistan and Bangladesh 91/-- largely because of differences in the concepts used and in the measurement and estimation procedures followed. A great deal of the confusion surrounding the figures can be avoided if some of these conceptual issues are clarified at the start. Annex 3 attempts to clarify some of the common concepts and definitions used in measuring rural unemployment. It should be referred to in interpreting the data that follow. Definition aside, given the extent of self-employment in rural areas (in farm or non-farm enterprises), and the fact that so much of the work done consists of household tasks shared among members, even an extended concept of what it means to be "employed" or "unemployed" or to be "seeking employment" must be interpreted with care. Most poor people -- and those who rely mainly on rural wage employment fall in this category -- try to engage in whatever gainful work they can find, often on a temporary or casual basis. The amount of work they do, the period over which they work, and the income they earn from work are all difficult to define and measure. We need to keep these caveats in mind when examining the available evidence. India The most widely available data on rural labor and unemployment rates are for India, and are based on the National Sample Surveys (NSS). The figures used here are taken from the - 75 - Table 20: INDIA: RURAL LABOR FORCE - PERCENTAGE DISTRIBUTION BY USUAL ACTIVITY CURRENT ACTIVITY AND SEX Activity Status 27th Round (U) 32nd Round (W) 32nd Round (D) Category 1972-73 1977-78 1977-78 M F T M F T M F T Self Employed + Helpers In Agr. 35.0 19.4 54.4 34.5 14.7 49.2 35.5 13.4 48.9 Self Employed + Helpers in Non-Agr. 7.0 3.4 10.4 7.7 2.7 10.4 7.8 2.4 10.2 Regular/Salaried Farm Employee 4.2 0.7 4.9 2.9 0.4 3.3 3.0 0.4 3.4 Regular/Salaried Non-Farm Employee 3.5 0.7 4.2 4.6 0.7 5.3 4.9 0.7 5.6 All Casual Wage Earners 14.1 11.1 25.2 17.2 10.3 27.5 15.3 8.5 23.8 Unemployed 0.7 0.1 0.8 2.7 1.3 4.0 5.4 2.7 8.1 Total Percentages: 64.8 35.2 100.0 6919 30.1 100.0 71.9 28.1 100.0 (Millions) 128.7 71.0 199.7 116.3 50.2 166.5 113.4 44.2 157.6 Source: 27th round data from Sarvekshana (1977), 32nd round from Sarvekshana (1979). 1972-73 structure is based on rural labor force 5 years and over while the 1977- 78 sample used 15-19 years. 27th round is based on usual activity (U) while 32nd uses weekly (W) and daily criterion (D). - 76 - NSS 27th (1972-73) and 32nd (1977-78) Rounds. 92/ The broad findings for 1972-73 and 1977-78 are shown in Table 20. The 1972- 1973 data are given in terms of "usual" (long term stock rate) unemployment. The 1977-78 figures show "current" (short term stock rate) and "daily" (short term flow rate) unemployment. (Annex Table A.33). I can draw several broad conclusions from the data. First, the long term stock rate of unemployment is very low; in 1972-73 less than 1 percent of the rural labor force reported cheir usual status as unemployed. This group, however, covers only those "who had practically no work continuously over the long period and who were seeking or available for work throughout." Since virtually everyone is employed at agricultural peak seasons, such chronic unemployment is likely to be voluntary. Other data show that nearly two-thirds of these chronically unemployed have more than primary education. Although the rural status of "chronic" unemployment rate has more than doubled (from 0.4 to 1 percent) and the numbers more than quadrupled (from 0.5 to 2 million) between 1972 and 1978, these are not the illiterate, unskilled wage laborers one has in mind when one is talking about rural unemployment. Second, rural unemployment rates measured on "current" (stock rate) or "daily activity" (flow rate) basis are not exceptionally high, at 4 percent and 8 percent respectively in 1977-78.93/ The latter rate can be interpreted to mean that 8 percent of all rural workforce person-days during the reference week were spent in seeking or being available for work; it suggests an upper limit to the quantum of work (in person-days) that could be "mopped up" by employment-creating public works programs or employment guarantee schemes. But though the rates are low, the numbers they imply are very large: the "daily activity" figures suggest that, on an average day in 1977-78, over 12 million rural person-days were spent in idleness. Third, roughly two-thirds of the labor force is male and one-third is female. Employment by type of work is not - 77 - distributed according to these proportions, however. Proportionately many more men are in regular work (whether farm or non-farm), while proportionately rather more women (many more on the "usual" definition) are casual workers. State-by-state data on participation rates, the incidence of casual labor and person-days of unemployment have been sumnarized by Grawe et al. (1979). The figures (see Table A.34 in the annex) show that unemployment is a highly regional phenomenon. Six states -- Bihar and West Bengal in the East, Andhra Pradesh, Tamil Nadu and Kerala in the South, and Maharashtra in the West -- accounted for two-thirds of all jobless person-days in rural India in 1972-73. Since these states accounted for only 44 percent of the labor force, their rate of unemployment was well above the national average -- 11.3 percent compared with an all-India figure of 7.8 percent. By 1977-78, according to the NSS 32nd Round, rural unemployment in Kerala, Tamil Nadu, Andhra and West Bengal continued to be far above the national average, either on a "current" basis (the percentage of individuals without work in the reference week) or in terms of "daily activity" (the percentage of person-days spent jobless and seeking work). These four states also accounted for nearly 1/2 of the all India rural and urban unemployment combined (see Tables A.35, A.36 and A.37 in the annex). Given the seasonal peaks and troughs of farm work, the main source of rural jobs, unemployment levels might also be expected to fluctuate sharply during the course of any given year--and in fact data for person-days of unemployment at different times in 1977-78 show that jobless rates are highest during the slack season between April and June (10.5 percent in 1977-78, up from 9.3 percent in 1972-73) and lowest during the peak season between October and December (9.3 percent in 1977-78, up from 6.8 percent in 1972-73). The estimates show that women appear to be overrepresented among the unemployed during the slack season. The interseasonal spread in labor force participation rates for women in 1977-78 (4.2 percent) was higher - 78 - than that for men (1.8 percent), even though there were twice as many men as women in the labor force. Moreover, about 2 percent of the males and about 4 percent of the females who report that they are in the labor force in the busy season drop out of it altogether during the slack season. Seasonality of employment, and its particular bias against female employment, needs to be kept in mind in designing remedial policies. Grawe et al. have grouped the Indian states on the basis of both the level and variability of their unemployment rates (Table 21). The data show that the degree of seasonality of employment varies from state to state. Interestingly, three of the four states in the last group in Table 21 (those with both low unemployment and low seasonal variations in employment) are among those that have experienced the fastest rates of agricultural growth. 94/ This suggests that, at a very aggregative level at least, agricultural growth may directly or indirectly improve employment prospects. 95/ The low seasonal variation in employment in certain states may also be specifically linked to the adoption of HYV technologies and the relatively heavy use of irrigation, both of which increase cropping intensities and labor inputs per hectare and reduce interseasonal differences in labor use. Given these variations in the level and seasonality of joblessness, strategies to increase productive employment will need to differ from region to region. "In general, the unemployed in regions with high seasonality are most likely to benefit from the spread of irrigation and high yielding varieties. On the other hand, regions which have high rates of unemployment but low seasonal variability might require relatively greater emphasis on programs and policies designed to increase non-farm opportunities." 96/ Data on unemployment levels disaggregated by landholding categories suggest that reported unemployment is heavily concentrated among those in the poorest deciles of the rural population, and especially in landless and rural labor - 79 - Table 21: INDIA GROUPING OF INDIAN STATES BY LEVEL AND VARIABILITY OF UNEMPLOYMENT RATES (Z OF THE LABOR FORCE) BY DAILY STATUS, RURAL PERSONS 1972/73 Index of Sub-round Groups and States Unemployment Variation Rate Max.-Min. (%) Min Group 1: High Unemployment - High Variation Andhra Pradesh 11.23 87 Jammu and Kashmir 8.58 595 Orissa 10.18 105 Tamil Nadu 10.45 134 West Bengal 10.66 109 Group 2: High Unemployment - Low Variation Bihar 10.02 31 Karnataka 8.55 24 Kerala 23.49 18 Maharashtra 9.43 46 Group 3: Low Unemployment - High Variation Assam 1.89 352 Madhya Pradesh 3.41 128 Rajasthan 3.25 426 Group 4: Low Unemployment - Low Variation Gujarat 5.43 30 Haryana 2.93 52 Punjab 3.94 48 Uttar Pradesh 3.38 30 Source: National Sample Survey, 27th. Round, as reported in J. Krishnamurty, The Employment Situation in India, World Bank, draft report, 1978. - 80 - households. Table A.38 in the annex gives the rural unemployment rates on a daily status basis by household types, as estimated by the Planning Commission for 1980. These show clearly that unemployment rates among agricultural and other labor households (12-15 percent) are twice those of all rural households and four to five times the unemployment rates for self employed (including cultivating) households in rural areas. Table A.39 in the annex presents additional data from the NSS 25th Round (1970-71) on seasonal unemployment levels for the lowest decile of cultivating households, numbering 3.7 million (of whom an overwhelming 88 percent had holdings much smaller than 2.5 acres), and for agricultural labor households. It shows that the incidence of unemployment (current status) was higher in landless than in farm households. Not surprisingly, unemployment hits its seasonal peak in the April-June period when most of the harvesting of the rabi (winter) and planting for the Kharif (summer) crop is over. The relatively high incidence of unemployment among small cultivators and wage earning families is confirmed by Visaria and Visaria (1973). Visaria (1980) demonstrates a clear inverse association between poverty (indicated by monthly per capita household expenditure) and the incidence of unemployment, measured in terms of (a) person days of unemployment and (b) the underemployment reported by the "currently employed." Table 22 presents his findings for rural India. They clearly show that in the bottom income decile of households, unemployment/ underemployment rates range from 14-22 percent for the peak season to 16-26 percent for the slack season. This evidence is contrary to the widely accepted view that the poor are too poor to be unemployed. 97/ Research on the incidence of unemployment by occupational groups suggests that it is particularly high among casual laborers. In a careful analysis of the "current" employment status of workers in Cujerat and Maharashtra, Visaria (1980) found that unemployment rates were highest for casual laborers, who were also heavily represented in the lowest deciles in the expenditure - 81 - Table 22: INDIA: POVERTY AND UNEMPLOYMENT, 1972-73 (INCIDENCE OF UNEMPLOYMENT* BY MONTHLY PER CAPITA EXPENDITURE GROUPS IN RURAL INDIA Oct.- Jan.- Apr.- July- 1972- % Share of Households Dec. Mar. June Sept. Sept. Oct. 72 - Sept. 73 MCPCE Group (Rs.) 1972 1973 1973. 1973 1973 MALES up to - 11.00 15.3 19.5 23.6 11.7 17.5 11.00 - 20.99 9.2 12.8 13.7 11.9 11.9 21.00 - 33.99 6.9 8.6 9.8 7.4 8.2 N. 34.00 - 54.99 5.2 6.3 7.2 5.5 6.1 A. 55.00 - 99.99 4.2 4.8 4.5 4.5 100.00 & above 3.4 2.9 3.0 2.8 3.0 All 6.0 8.7 7.9 6.2 7.2 FEMALES up to - 11.00 33.8 26.8 29.9 27.1 29.4 11.00 - 20.99 14.8 18.6 20.2 17.2 17.7 21.00 - 33.99 9.8 13.1 17.4 12.0 13.1 N. 34.00 - 54.99 7.8 9.7 12.4 8.4 9.6 A. 55.00 - 99.99 6.2 7.4 8.0 6.3 7.0 100.00 & above 3.6 2.0 7.1 3.4 3.3 All 9.3 11.5 14.2 10.0 11.3 PERSONS up to - 11.00 23.9 22.1 26.1 17.6 22.4 0.7 11.00 - 20.99 11.3 15.0 16.1 14.0 14.1 9.8 21.00 - 33.99 7.9 10.1 12.2 8.9 9.8 30.1 34.00 - 54.99 6.0 7.3 8.7 6.4 7.1 35.2 55.00 - 99.99 4.8 5.5 5.5 5.0 5.2 19.0 100.00 & above 3.4 2.6 3.9 2.9 3.2 5.1 All 7.1 8.6 9.8 7.4 8.2 100.0 Source: NSS 27th Round data as compiled by P. Visaria (1980) p. 10. * In terms of person days unemployed as a percentage of person days in the labour force. - 82- distribution. Further, casual laborers were much worse off than other workers in terms of "unemployed person-days" (the "daily activity" flow rate of unemployment, which incorporates the notion of underemployment) with rates of close to 20 percent. Casual laborers accounted for half to three quarters of the total unemployed person-days in rural areas of the two states. These individuals had little education and were younger than other workers; most were illiterate and from agricultural labor household with very little or not land. In most of India, unemployment rates are higher where casual workers make up a large proportion of the labor force. Visaria found high coefficients of correlation between the proportion of casual laborers and the incidence of unemployment (in terms of person days) in rural areas in 17 states; the association held for males, females and all persons. Casual laborers clearly need to be given high priority in any program to alleviate poverty and reduce unemployment. Data reported by Visaria help to explain why unemployment surveys in South Asia provide what seem to be implausibly low estimates of rural unemployment. Apparently farmers and their families may be significantly unemployed or underemployed, but because they are unwilling to work off their farms , because they lack information on job opportunities, and/or because social convention and/or caste render them immobile, they are much less likely to seek employment or report their willingness for and/or availability to work. 98/ As has already been shown (Table A.34), women have low overall labor force participation rates; nevertheless, they carry out a significant proportion of certain tasks--rice transplanting, weeding, harvesting, and threshing and winnowing. Most of these are peak season tasks and require hard physical labor. Added to household work and the care of animals--also the traditional sphere of women--this specialization tends to impose a heavy burden of labor during peak seasons on women (and children) in rural areas. Data on overall unemployment levels by sex - 83 - nevertheless show that women in the labor fo:ce have jobless rates two to three times higher than their male counterparts in most household expenditure groups (Table 22 and Table A.40 in the annex). Finally, many of those who report that they are employed are nevertheless still "seeking work or available for additional work." - of 18 million (male and female) wage laborers on farms or in non-farm enterprises, some 2.2 million (12 percent) fell in this category; - of 73 million workers reported as working in households, on their own farms, or in non-farm enterprises, some 14 million (19 percent) fell in this category. - of some 50 million workers who gave their usual activity as casual labor, some 30 million (60 percent) fell in this category. 99/ In addition there were 56 million farm helpers or helpers in non- farm household enterprises for whom no data are reported. But if we use the same proportion for these as for their employers (19 percent, surely an underestimate), some 11 million of them would also be seeking or available for additional work. When all these categories are added up, some 56.2 million of the 198 million shown as gainfully employed in rural India in 1972-73 (Table 20), or 28.7 percent of the total, may have been seeking or available for additional work. This is a stock rate; if I add it to the 8 percent of the labor force reporting themselves as unemployed on a "flow rate" basis, over a third of the rural labor force in India may be either unemployed or seeking and/or available for additional work. 100/ These figures suggest the extent to which labor may be under-utilized and hence the size of the pool of labor that might be available for additional employment. The words "might be" needs emphasis because we cannot tell whether those classified as already employed but available for more work would in fact take up additional employment. The jobs offered might be unacceptable; - 84 - the wages offered might be too low once the costs of transport, search, dislocation and other factors are taken into account; new employment might conflict with current work during the busy season or might interfere in other ways with a currently employed individual's work routine or income stream from his/her present occupationt. An analysis by Visaria (1970) of the 16th (1960-61) and 17th (1971-72) NSS Rounds data suggests that by no means all of those who say they are available for work would actually respond. This study found that although those gainfully employed in rural areas worked an average of 47 hours per week, nearly 12 percent of the respondents reported working 28 hours or less. Of this latter group, nearly two-thirds reported that they were not available for additional work; nearly 80 percent of cultivators in this category, but only one-third of agricultural laborers, reported their unavailability. Further, Visaria and Visaria (1973) found that 37-60 percent of small cultivator and agricultural wage households reported that no one was willing to take up regular full-time employment. Even lower percentages were willing to take up employment outside their villages. Households are generally loath to move into unfamiliar surroundings in which job prospects may be uncertain. The guarantee of a job appeared, however, to induce a significant number of households (ranging between 11-27 percent among the small cultivators and 14-47 percent among the wage earners) to send at least one member outside the village. In many states, over 30 percent of adult males from the households in the "weaker sections" were ready to take up employment outside their villages; 101/ guaranteed full- time jobs might attract more rural people if these jobs were available. Nevertheless, additional work within or near their villages is likely to remain the goal for the vast majority of those reporting themselves as seeking work. Any employment schemes have to be based on this fact. The lack of comparable data precludes the possibility of drawing any definite conclusions about trends in unemployment in - 85 - India. Raj Krishna notes scathingly that: 'The distinguished statisticians and demographers who have been "improving" employment concepts used in successive surveys in India have seen to it that no economist should be able to compare with strictly logical justification, the numbers generated after every "improvement" with previously collected numbers. Nor can anyone carry out any time-series analysis of rural employment- unemployment.' Nevertheless, he notes that "all unemployment estimates point to a serious worsening of the rural employment situation in the early seventies in comparison with the sixties", due to the stagnation of agricultural output and a series of bad harvests. (1976, p. 27 and 37). Table A.41 in the annex gives state-by-state rural unemployment rates expressed in percentages of total labor-force person days for 1973 and 1977. (Unfortunately more recent data are unavailable.) Although the rates fell in seven states, they rose in nine; in seven of these nine, the rise was above the nationwide 10 percent increase. The sharp increases in unemployment in Haryana and Punjab are of special interest, in view of the vigorous agricultural growth in these states. The relationship between employment and agricultural growth at the grass roots level is complex, and is influenced by institutional and technologial factors that are not. readily captured in aggregate data. Generally, the incidence of unemployment declines as the irrigated area rises, but employment seems to stagnate once irrigation potential is reached. Thus an employment strategy based simply on the rapid expansion of irrigation seems to be inherently limited. Bangladesh Bangladesh's 1979-80 agricultural labor force was about 28.4 million or 77 percent of the labor force as a whole. Agricultural labor is estimated to have been rising by 600,000 people (or about 2.2 percent) a year, while the labor force expanded at 3.2 percent per year. Another survey has estimated that over the past 10-12 years, aggregate demand for labor has - 86 - increased by only 1.5 percent. 1C2/ About 59 percent of all rural households are engaged in farming, another 24 percent are classified as wage laborers, and the rest are involved in trade, business, transport and salaried occupations. 103/ Most of the labor force is male; female labor force participation in this largely Muslim society was as low as 4 percent in 1974. Real wages of agricultural laborers have been declining; in 1979/80, they were estimated to be 23 percent below their 1967/70 level. 104/ The only study for Bangladesh that attempts to measure rural unemployment rates based on some notion of the "willingness" criterion is by I. Ahmed (1978). On the basis of a survey of about 300 rural households in Comilla District in 1975, he estimates "involuntary unemployment" at about 8 percent, made up of those "who would take up jobs created through rural works programs"; this represents "the maximum to which public works programs can contribute in generating employment." 105/ Other available estimates, based on "time" or "productivity" measures of underemployment and surplus labor in rural areas 106/, range from 20 to 60 percent depending upon how "labor supply" is calculated and what is used as the "norm" against which unemployment is measured. 107/ Some of these figures give an exaggerated picture of rural unemployment in Bangladesh, but for what they are worth, the available figures are summarized in Table A.42 in the annex. Rahman (1978), using 1973-74 data from an intensive village survey and taking 300 days of work in a year as "the full norm,"' found that an average rural worker was "unemployed" for about 20 percent of the time; a third were jobless for more than 33 percent of the work year. At the same time, however, his concrete findings about actual days worked showed that wage laborers and the self-employed averaged about 240 days of work a year, and that over two-thirds of the labor force worked for more than 200 days a year. A. R. Khan (1981) in a detailed study of two villages also found landless households working around 285-295 days per person per year--more than those households with land. - 87 - This suggests a flow rate of unemployment in the range of 20-25 percent only (see Table A.43(a) and (b) in the annex). These findings are remarkable in view of the impression of severe under- employment/ unemployment associated with Bangladesh. What is evident is that rural laborers in Bangladesh seem to work more days in a year on the average than those in India (see Table 16), but at lower wage rates and lower resulting incomes. Not only do unemployment estimates vary, but the results often convey little or only add to the confusion. Alamgir (1980), for example, using data from a small survey of langarkhanas 108/ and taking a time criterion and minimum norms of 343 and 290 man days, estimates "underemployment" in rural Bangladesh at 34 and 20 percent respectively in 1973-74. At the same time, he finds that the working adults in 62 percent of households were "unemployed" because they worked less than the "desired minimun of days/year/ working adult; those in 81 percent of households were counted as ''unemployed" because their incomes fell short of "the desired minimum"; and 91 percent were "unemployed" on the basis of a combination of time and income criteria. Estimates of this kind strip the concepts of unemployment and underemployment of any practical meaning. Alamgir notes that the majority of the severely underemployed -- the roughly 10 percent of his sample who worked less than 100 days a year -- came from farming households with very small holdings. 109/ He also finds that underemployment (on the basis of the time criterion) is heaviest among owner-cum- tenant and tenant farmers. Recalling the very small holdings of these groups, whose major occupation is really wage labor, this is not surprising. According to Alamgir, landless laborers have a lower incidence of employment than the mixed or pure tenant group, but their employment is highly seasonal, ranging from an average of 11 days in September to 25 in June-July. Pakistan As with Bangladesh, available data are too sparse to provide anything other than a highly generalized and - 88 - impressionistic picture of rural unemployment and its characteristics in Pakistan. The Labor Force Survey (LFS) of 1975-76 estimated Pakistan's workforce at 20.5 million, with a participation rate of under 20 percent. Over three-quarters of the workforce lived and worked in rural areas, and over half was directly employed in agriculture. 110/ The rural labor force was dominated by the self-employed and "family helper" categories, which together accounted for 90 percent of the labor employed in agriculture and allied activities. 111/ In the 1950s and 1960s, Pakistan's labor force had grown at 2.7 percent a year while employment rose by about 3.4 percent a year; during 1970-75 the comparable rates were 3 and 4 percent. These rates suggest that there must have been an absolute decline in unemployment, 112/ even before migration to the Middle East reached its peak in the late 1970s. Indeed, most surveys undertaken in the 1970s indicate that "open unemployment" 113/ is very low, about 1.5-2.0 percent of the labor force, although some data suggest that the proportion of those in rural areas who are "without any work and actively seeking it" may be much higher -- roughly 15 percent. 114/ The figure, however, seems to be mainly made up of school leavers who are new entrants to the rural labor force. 115/ Few micro surveys of occupational or employment status are available. One carried out in 1971 in 40 villages in West Punjab by Eckert et al. (1972) showed that nearly 16 percent of rural households were in the permanent and temporary labor category while another 20 percent worked as artisans and shopkeepers. 116/ The same study found a high rate of unemployment among casual labor -- an apparent parallel with India, although comparable rates cannot be calculated. While permanently employed labor (4.5 percent of the rural population) "enjoyed a very steady employment averaging 29 days monthly," temporary hired laborers (some 8 percent of the rural population) worked an average of only 11 days per month. "Even during their months of fullest employment (October-November) they were idle nearly fourteen days 89 - 'ut of thirt). 117/ Another survey by Robins( nd Abassi (1979) uses an arbitrary time criterion for unautremployment, defining as underemployed "those who work less than 35 hours per week." On the basis of labor Force Survey data, they estimate that between 8 and 10 percent of the employed in rural areas were underemployed between 1968-69 and 1971-72. This ratio had fallen to around 5 percent by 1974-75. In that year, only two fifths of the underemployed indicated that they did "not have enough work," and almost half said that they did not want to work more. Of those who worked more than 35 hours -- 90 percent of the sample -- the average number of hours worked exceeded 50 per person per week. Thus, even on the "time criterion," these data hardly suggest that the rural labor force in Pakistan is suffering from under- employment; most workers are employed for long hours and do not want more work. (This finding is, incidentally, consistent with the evidence Singh, 1988 on rising real wage rates in rural Pakistan.) To summarize, the evidence suggests that rural unemployment and underemployment in Pakistan are relatively insignificant and are probably confined mainly to landless casual or temporary laborers. This group, together with tenant farmers, is poorly endowed with skills and education, has a relatively low propensity to migrate, and is most clearly in need of new job opportunities. Most other underemployed workers work as artisans or as unpaid family helpers on farms; in their case, the prime need is for some extra work to supplement their existing incomes, within reasonable commuting distance of their villages. With the large migration to the middle East from rural areas, even these rates of unemployment must have fallen. Although no data exist to substantiate it, there may be growing seasonal shortages of labor in rural Pakistan today. Conclusions The available evidence from India shows that - "usual" or chronic unemployment is very low -- around 1 - 90 - percent -- suggesting that nearly everyone finds some work during the year; - unemployment measured on a "current" or "person-day" basis is not exceptionally high -- around 4-8 percent -- but the numbers involved are large -- around 12 million workers are without jobs on any given day; - a third of those who are employed report themselves available for additional work; - there are large regional variations in joblessness, with high unemployment (measured in person-days) concentrated in six states -- Bihar and West Bengal in the East, Tamil Nadu, Andhra Pradesh, and Kerala in the South, and Maharashtra in the West; - underemployment is a highly seasonal phenomenon, and the extent of slack period unemployment also varies by states. In general, states that have experienced high agricultural growth have both low unemployment rates and low seasonal variations in their rates; - unemployment is heavily concentrated in landless households: in some states, casual laborers account for half to three quarters of all person-days of rural unemployment; - women are a particularly vulnerable group, with unemployment rates two to three times higher than those of men. Data from Bangladesh are much less reliable. Open unemployment seems to be low -- around 10 percent -- but large proportions of the population appear to work long hours for small returns and in uncertain, seasonal and casual jobs. Over half of the self employed or those working as wage earners worked less than 250 days a year, and 10-15 percent worked less than 150 days a year. The data from Pakistan are also inadequate, but such figures as are available suggest that the rural unemployment rate is very low -- about 2 percent of the labor force -- and mainly - 91 - concentrated among landless casual laborers. This is consistent with reports in recent years of labor shortages and rising real wage rates, as many workers from rural areas have migrated to the Middle East. In aggregate terms, rates of unemployment in South Asia are low by traditional standards, but a large proportion of the population works in marginal, low productivity activities. Whether or not this is called "disguised unemployment" is less important than the fact that a high proportion of those gainfully employed are available for additional work and meanwhile work long hours with little remuneration. The problem is one of inadequate growth in labor demand; its solution lies in increasing the opportunities for productive employment of the wage labor force, a group that includes many cultivators who own or lease small holdings. The final section of this chapter briefly outlines past and current trends in the occupational distribution of these opportunities. C. The Occupational Distribution of Rural Wage Employment There are three main categories of work that can help raise the incomes of poor rural households that depend primarily on wage earnings for their livelihood. First, on-farm employment in crop production may increase as a result of multiple cropping, higher output, and shifts to labor intensive and higher income crops like vegetables and fruits. Second, non-crop agricultural activites (such as dairying or poultry farming, fisheries, and forestry) may offer new income-producing opportunities. Third, non-agricultural rural activities (for example, small-scale manufacturing, processing, transportation, repair, marketing and other services, or rural public works programs) may provide additional jobs and wage earnings. The extent to which increased crop production raises the demand for wage labor depends on the size distribution of holdings, the cropping patterns adopted, and, most critically, on farm technologies. 118/ In some areas of rapid agricultural - 92 - growth (the Indian and Pakistani Punjabs, for example), much of the potential stimulus to employment has been offset by farm mechanization (Singh, 1988, Chapter 4). Greater use of irrigation and higher cropping intensities should on balance increase the demand for farm labor (although changes from persian wheel to tubewell irrigation may not), but how much of this demand will be met by family labor rather than hired workers is not clear. In general, it is unlikely that any achievable rate of growth in crop production over the next decade will be able to generate enough new demand for rural labor to absorb the two main components of potential supply -- the currently unemployed and under-employed in the rural labor force, together with the anticipated increments in that labor force over the years. Consequently, the rapid expansion of non-crop agricultural and non-agricultural rural activities is crucial for the prospects of the poorest in rural areas. 119/ This section briefly outlines the extent to which these activities are already contributing to the incomes of rural labor households. 120/ Noncrop and Nonfarm Employment Non-crop and non-farm sources of employment and incomes involve a wide range of activities. Because many of these activities are undertaken in small scale, highly dispersed, informal enterprises that are often overlooked in surveys and censuses, data on their extent and nature can be hard to come by. A recent World Bank review, 121/which evaluated the available evidence from some 15 developing countries, found that (1) non- farm activities 122/in rural areas are a primary source of employment and earnings for approximately one quarter of the rural labor force in most developing countries (one third if the labor force in rural towns is included; 123/ (2) they are a significant source of secondary earnings for small farmers and the landless in the slack season (when farm and non-farm employment move in opposite directions; (3) their total contribution to employment in rural areas is growing; 124/ (4) they include a wide variety of activities in manufacturing (20-30 percent), services (20-35 93 - percent), commerce (15-30 percent), construction (5-15 percent), transport and processing (5-15 percent); 125/ (5) they are generally carried out on a very small scale, rely on simple labor- intensive methods, require little capital and skills (other than those provided indigenously via apprenticeship) and are widely dispersed spatially (at least until they become concentrated in rural towns as infrastructure improves and markets grow; and (6) most of their demand is generated locally in rural areas and is linked closely to the growth of agricultural and rural incomes. (ii) India In India, 39 percent of the non-farm employment (NFE) is estimated to be in manufacturing, 14 percent in construction, 14 percent in trade and commerce, 24 percent in services and 9 percent in transport and 'other' categories. 126/ The limited available evidence shows that 'non-crop' incomes account for nearly half of the incomes of all rural households (Table 23). The proportions are smaller but still substantial for cultivating households and depend upon whether or not HYVs are planted. As Table 24 shows, the 'non-agricultural' sector accounts for nearly 15 percent of the rural labor force in India and Bangladesh and about 30 percent in Pakistan. 127/ Table 25 presents data on rural per capita incomes in India, compiled from an extensive series of surveys of 4,100 rural households carried out by the National Council for Applied Economic Research (NCAER) between 1969 and 1971. The figures show that the proportion of income derived from crops rises with the size of holding, and that non-agricultural wages and non-farm earnings provide nearly half of the incomes of landless and marginal farmers (those with holdings of less than 1 ha.) and between a fifth and a quarter of the incomes of small farmers (those with holdings of up to 5 ha.). 128/ All these groups have very low per capita incomes; only their non-crop and non-farm earnings permit the very poorest to match the near-destitution incomes of the small farmers. The sources of incomes for the landless are of special - 94 - Table 23: INDIA: RURAL AGRICULTURAL AND NON-AGRICULTURAL INCOMES IN 1970 ('000 Rs) Total Average Income From: Non-Crop Income All Sources Crops as % of Total All Households 2.65 1.38 48 Cultivating Households 3.19 2.28 28 - using modern varieties 4.97 4.21 15 - using traditional varieties 2.60 1.65 36 Non-Cultivating Households 1.82 - 100 Source: Computed from National Council of Applied Economic Research, "Changes in Rural Income in India: 1968-69 to 1970-71," Tables on pp. 76 and 77. Table 24: SOUTH ASIA: PERCENTAGE COMPOSITION OF RURAL LABOR FORCE AND EMPLOYMENT IN THE AGRICULTURAL AND NON-AGRICULTURAL SECTORS Rural Labor Force Country Year Share in Total Agricultural Non-Agricultural Labor Force Sector Sector (As % of Rural Labor Force) Bangladesh 1951 69.8 82.8 14.0 (14) 1961 96.6 84.7 11.9 (14) India 1961 86.0 70.5 15.5 (18) 1971 82.5 70.3 12.2 (15) Pakistan 1951 86.8 63.1 23.7 (27) 1961 84.5 58.3 26.2 (31) Source: Compiled from "United Nations, Agriculture, Industry and Services in the Urban and Rural Labor Force," ESA/P/WP.57. New York, September 1975 (mimeographed) cited in Naiken, L. (1977), p. 5. Data based on country censuses. Table 25 INDIA: RURAL PER CAPITA INCOMES BY FARM SIZE CROUP AND SOURCES, 1970-71 Size Group of Est. Total per Proportion of Income Derived from: Cult. Land X Capita In- Crops Ancilliary Agricultural Non-agro Non-Farm and (Acres) HH come (Rs) Activities Wages Wages Other Sources (A) (B) (C) (D) (E) Landless * 28 540 * 26 26 5 43 0 - 1.0 9 544 33 7 16 3 41 1.0 - 2.5 10 457 50 10 14 2 - 24 2.5 - 5.0 15 566 67 7 7 1 18 5.0 - 7.5 9 672 78 5 3 1 13 7.5 - 10.0 8 822 82 4 2 1 11 10.0 - 15.0 8 905 86 4 1 * 9 15.0 - 25.0 8 1040 92 2 1 * 55 25.0 + 4 1409 98 * * 0 1 All Sizes: 100 677 59 9 9 2 21 (n = 4100) * All non-cultivating households. Note: (A) and (B) are gross values of output from crops and livestock type farm activities less their operating expenses. (C) includes "cash and kind" wages as deos (D), where workers are laborers and not salaried. (E) includes salaries, remittances, property rental income, and incomes from employment and enterprises in rural areas. Ancillary activities include livestock, poultry, dairying, forestry and on-farm processing, while non- farm sources include incomes from rural manufactures, handicrafts, repair and construction, transportation, trade and commercial activities. Source: NCAER, AKIS Survey Data. - 96 - interest. It is often thought that they depend heavily on agricultural wage employment for their livelihood. For this reason, much work has gone into estimating on-farm employment elasticities (Singh, 1988, Chapters 3 and 4), and into efforts to analyze how these elasticities are translated into jobs and income for the landless. But the NCAER data reveal that only a quarter of the incomes of the landless come from agricultural wages. The rest was accounted for by non-crop and non-farm activities, of which employment in construction, non-farm activities such as manufacturing, rural services, and trades represented by far the largest proportion. Vyas and Mathai (1978) have noted that comparisons of India's rural occupational structure over time show hardly any change in the proportion of the labor force working in agriculture; the figures for 1957, 1961, and 1971 were 72.1, 71.2 and 72.1 percent respectively. There has thus been little occupational diversification in the rural economy. In fact, the proportion of the rural work force employed in industrial activities has actually fallen (from 18 to 16 percent between 1961 and 1971). Vyas and Mathai attribute these findings to the weakness of the linkages between agricultural and non-agricultural activities in the rural economy. Nevertheless, non-agricultural activities are important sources of rural employment in India, accounting for 15-20 percent of the labor force in nearly all states (see Table A.44 in the annex). If non-crop agriculture (livestock, forestry, fishing, etc.) is included, this proportion rises to 20-25 percent. Table A.45 provides data from selected Indian states on the sources of rural per capita income by size of holding. Even if non-crop income represents a small proportion of the incomes of the total rural labor force, it is a significant source of earnings for small farmers and the landless throughout India -- in rice and wheat zones, arid and irrigated areas, and high and low growth states. 129/ Data from the NSS 25th Round (1970-71) showed that a significant proportion of poor households (35 percent of - 97 - small cultivators and 43 percent of non-cultivating wage earners) reported that they could supplement their incomes in non-crop agricultural activities. 130/ Evidence from farm management surveys in India indicates that off-farm employment accounted for 50 percent of small farmers' labor and somewhat less of their incomes in the 1950's (Bharadwaj, 1974). By the early 1970s, non- crop activities had assumed a greater significance; many small farmers earned much more from occupations such as dairying than they did from work associated with foodcrop cultivation (Mavinkurve, 1975). In Tamil Nadu, for instance, male agricultural laborers' average annual earnings from non- agricultural occupations have increased from 10 percent to 38 percent of their total incomes over the last 20 years (ILO, 1977). Microeconomic studies of small farmers and agricultural laborers carried out by Agro-Economic Research Centers confirm this trend; they also suggest that members of agricultural labor households who engage in non-crop activities have marginally higher incomes than those who cultivate very small (less than 0.5 ha.) owned and tenanted holdings. 131/ Pakistan and Bangladesh The data for Pakistan and Bangladesh are sparse, but they confirm the growing importance of non-farm sources of incomes in rural areas, especially for those with small holdings (Tables 26 and 27). A rising share of the rural labor force appears to be engaged in non-farm work, and an increasing share of rural incomes comes from non-farm sources. In 1970-71, 19 percent of rural employment in the Pakistani Punjab was in non-farm activities; 48 percent of farm families in Pakistan relied on income from non- farm activities, which contributed 23 percent of their total income. Nearly 70 percent of small farmers and the landless undertook non-farm work which accounted for nearly 39 percent of their income. 132/ In the Pakistani Punjab, A.S. Haider (1977) has documented the rapid shift into non-farm employment produced by the green revolution. New job openings have emerged in - 98 - Table 26: BANGLADESH: FARM AND OTHER INCOME, BY SIZE GROUP OF HOLDING, 1965-66 Average Size Household Incomes (Rs) Size Group of of Holding in % of Households Other Non-farm Land Holding Size Group in the Sample Farm Sources as a % (Acres) (Acres) of Total 0 - 3 1.82 45 584 362 38% 3 - 10 4.80 48 858 163 16% 10 + 13.51 6 2307 1220 34% Source: Khan, T.M. and Bose, S.R. (1968) citing data from "Survey of Employment, Income and Expenditure of Rural Households in East Pakistan, 1965-66," Rajshahi University. Note: The major source of other income of small cultivators is wage labor on and off the farm, while for the medium cultivators, it is petty trade and in some cases remittances from family members working elsewhere. The bulk of the incomes (non- farm) of large cultivators is from trade, non-farm work and additional remittances. Table 27: PAKISTAN: CONTRIBUTION OF NON-FARM TO TOTAL INCOME OF FARM FAMILIES, BY SIZE OF HOLDING, 1968 Size of Holding Non-Farm/Total Income (Percentage) 0 - 6.25 acres 39 6.25 - 12.49 acres 22 12.5 - 18.74 22 18.75 - 24.99 acres 8 Over 25 acres 6 All Classes 23 Source: David B. Steele, "Rural Industrialization and Employment," I.L.O., November 1975 (mimeo) based on F. Kuhuen, Agriculture and Beginning Industrialization: West Pakistan, Opladen: C.W. Leshe Verley, 1968, Table 83, 85 and 86. - 99 - distribution, repair and maintenance services (especially for farm machinery), rice milling, small en,ineering workshops, transportation, construction, marketing agencies, and financial institutions. Meanwhile, however, some of the traditional activities of village artisans -- blacksmiths, carpenters, weavers, potters, cobblers, traders, millers -- have declined in importance or have been transformed and displaced to small farms. Many artisans stay in villages but commute to work elsewhere. Another feature of the rural economy in Pakistan has been extensive out migration. Most of the migrants are from land owning groups, but artisans represent a significant percentage of the total. 133/ Small towns dependent largely on rural migrants have grown rapidly as -new non-agricultural occupations have flourished. Rural migrants have become a major source of remittances back into the rural economy, thus raising both rural incomes and the demand for non-farm work. Farm labor is becoming scarce as higher returns in non-agricultural occupations raises rural wage rates. 134/ Rural infrastructure has been extensively developed, the new roads and electrification schemes have made it easier to set up small engineering shops and manufacturing facilities. D.A. Khan (1978) estimates that, between 1960 and 1972, incremental employment in non-farm occupations exceeded that in on-farm employment (which had been considerable owing to the green revolution). 135/ In Bangladesh, both the 1974 Census and the 1973-74 Household Survey showed that nearly 30 percent of all rural households can be classified in non-agricultural occupations. Most of this activity consisted of rural trade and services, organized around small rural towns that serve as marketing centers. Small-scale and cottage industries in rural areae accounted for less than 4 percent of the total labor force in Bangladesh. A survey by Ali (1980) documented the extent to which farmers in Bangladesh are taking up non-farm employment to obtain - 100 - additional income. Interviews with over 2,140 farm families in 1980 showed that 84 percent of all farm operators reported off- farm employment of over 100 days per year, but that those with holdings of more than 5 acres did not work off the farm. The percentage reporting off-farm work of 100 days or more had risen steadily, from 8.4 percent in 1951 to 30 percent in 1960 to 51 percent in 1970 to 84 percent in 1980 -- a tenfold increase in three decades. Only 10 percent of those engaged in part-time farming produced enough on their land to meet their yearly food needs. More than 53 percent worked off their holdings for more than 9 months (270 days), another 23 percent did so for 6 to 9 months. D. Some Final Numbers Despite the diversity of South Asia and the extreme difficulty with data comparability, we have been able to classify all rural households who depend primarily on wage incomes into three mutually exclusive categories: (a) Landless Rural Labor Households Those to "who do not operate any land" and who have to rely mainly on wage employment, often as casual agricultural laborers, for their livelihood-- but partly also on employment in marginal activities in the rural non-farm sector. Raising farm productivity can provide benefits to them only indirectly via an increased demand for labor, where and when it is forthcoming. In 1980 there were some 20 million households in this category and they accounted for 17% of all rural households and 13% of the rural population. (b) Near Landless Households Those with less than one acre (0.4 hectares) of operated area. The present holdings of this group of farmers are too small to provide a subsistence standard of living, even allowing for likely productivity increases. These households are akin to the landless in their dependence on rural wage incomes as a major source of livelihood. They can also supplement their wage incomes - 101 - through a variety of on-farm ancillary activities such as dairying and poultry. In 1980, the near landless accounted for 13% of all rural household and 12% of the rural population in South Asia; they also accounted for 22% of all cultivated holdings and 4% of the total cultivated area. (c) Marginal Farmers Those with between 1 and 2.5 acrea (0.4-1 hectares) of operated area, whose holdings are too small to provide an adequate standard of living at present levels of productivity, but whose per capita incomes could be improved substantially by future productivity increases. Nonetheless they still depend primarily on wage earnings to supplement their incomes from farming which provides them with only below subsistence living standards. In 1980 marginal farmers accounted for 17% of all rural households and 16% of the rural population in South Asia; they also accounted for 25% of the cultivated holdings and 8% of the cultivated area. The data are given in Table 28. 136/ These three groups together accounted for 47% of all rural households and 41% of the rural population in South Asia in 1980. 137/ These groups accounted for some 55 million rural households with a total population of 272 millions in 1980. In addition there are some 19 million small farmers-- those households with between 2.5 and 5 acres (1-2 hectares) of operated areas whose holdings at present levels of productivity provide a standard of living close to the margin of subsistence. Though they do not depend primarily on wage incomes, these households participate in wage employment to supplement their meager farm earnings. Although future farm productivity increases could definitely provide them with an adequate standard of living, they will continue to seek supplementary wage employment. In 1980 these households accounted for an additional 16% of all rural households and 17% of the rural population in South Asia; they also accounted for 21% of the cultivated holdings and 14% of the cultivated area. Table 28: SOUTH ASIA: ESTIMATES OF SMALL FARM AND LANDLESS OITSErHOLDS AND POPIULATTON IN 1980 (BASED ON 1970-80 POP11LATION CHANCES AND 1970'S PROPORTIONS) Pakistan India Bangladesh TOTAL SOUTH ASIA H (mln) FS P(Mln) H (mln) FS P (mln) U (mnn) FS P (mln) H (min) P (min) Landless Rural Labor -(non cultivators with income from wages over half of total income) 2.0 4.5 9.0 15.0 4.4 66.0 3.0 4.6 13.8 20.n 88.8 (19.8) (15.4) (16.0) (12.6) (21.0) (16.7) (17.0) (13.3) Near Landless -(cultivators operating less than 0.4 ha) 0.2 4.9 1.n 14.0 5.0 70.0 1.1 5.1 5.6 15.3 76.6 (1.9) (1.7) (15.0) (13.3) (7.7) (6.8) (13.0) (11.5) Marginal Farmers0 -(cultivators operating 0.4 to 1.0 ha) 0.5 5.0 2.6 16.9 5.3 R9.6 2.3 6.2 14.3 19.7 106.5 (5.0) (4.4) (18.1) (17.1) (16.1) (17.2) (17.0) (16.0) Small Farmers -(cultivators operating 1.0 to 2.0 ha) 0.7 5.3 3.8 15.9 6.1 97.0 2.0 7.1 14.2 18.6 119.0 (6.9) (6.5) (17.0) (18.5) (14.n) (17.1) (15.8) (17.3) Sub-Totals 3.4 - 16.4 61.R - 322.6 R.4 - 47.9 73.6 386.9 (33.7) (28.0) (66.O) (61.5) (58.7) (57.R) (62.4) (58.1) All Rural Households 10.1 5.8 58.6 93.6 5.6 524.2 14.3 5.8 82.9 11R.0 665.7 (100.0) (100.0) (100.0) (100.0) (10n.n) (I1n.n) (10n.n) (100.0) Sources: Note: H, FS and P refer to households, family size and estimated population respecttvely Figures In brackets give percentages. - 103 - E. Conclusion This section has dealt with rural wage labor. On the basis of such limited evidence as is available, we have seen that large numbers of those living in rural areas in South Asia are unable to support themselves by farming, because their land holdings are very small or nonexistent. Employment opportunities for these people who make up the rural wage labor force fall short of the demand for jobs, and that those who do find work as farm laborers seem increasingly to suffer from underemployment and from erosion of the real value of their earnings. Finally, I have shown that earnings from non-farm agricultural activities and from non-agricultural rural jobs are of special importance for marginal farmers and the landless, but any dramatic expansion of these income sources in recent years has been confined to those areas where agricultural growth has been high. This link between agricultural growth and the growth of non-farm employment opportunities is critical to any evaluation of employment prospects for the rural poor in the future. In this paper I have discussed some of the characteristics of the rural poor and shown how two factors which both define these people and determine their economic status -- their access to land and to jobs. Clearly, any coherent policy to alleviate conditions must base itself on new approaches to the problems summarized in this paper. The small land holdings which characterize most rural South Asian households and the lack of adequate income-earning opportunities for those who lack even the minimal security of an economically viable plot of land must be addressed. - 104 - ENDNOTES 11 "South Asia" here refers to Bangladesh, India, and Pakistan. 2/ See "The Great Ascent: The Rural Poor in South Asia" by the same author. 3/ The Marxian perspective seems to dominate the literature on rural South Asia. Among others see Alamgir (1978), A. R. Kahn (1977), A. Rudra (1978a, 1978b), H. Alavi (1976), F. Frankel (1971), U. Patnaik (1976-1977), and R. Sau (1973). For the Malthusian perspective, see Vyas (1979), V. M. Rao (1972), Ray and Ray (1973) E. Stokes (1975), D. Kumar (1975), and A. Chakravarti (1975). 4/ Attwood defines a family's upward mobility as an expansion in its land holding and downward mobility as a contraction of that holding. 5/ Attwood (1979) p. 496-497. 6/ Between 1961 and 1971 some 38,000 formerly landless households in two relatively poorer regions in Haryana purchased land. 7/ The fact that an overwhelming ntumber of small farmers have been able to retain their holdings does not mean that they may not have become poorer. As Vyas (1979) has pointed out, immiserization can occur with or without proletarianization. As family size increases, previous levels of production may become more and more inadequate, forcing households to rely more and more on agricultural labor and other supplementary occupations. Under these circumstances, and in the absence of productivity increases or easy access to off-farm employment, per capita incomes will fall. 8/ Sheila Bhalla (1977), p. A-4 "Only 5.3 percent of today's landless agricultural labor households cultivated any land at all in 1962. All of them were pure tenants" (p. A-12). 9/ Nearly 75 percent of those who "lost" land on balance were transferring it to other family members: Sheila Bhalla (1977). - 105 - 10/ See Rudra, Maji,4 and Talib's (1969) study of 261 farms above 20 acres in 1968-69. It is hard to reconcile this with the NSS data, but since Rudra et al. collected their data from the Block Development Officer at the local level, one tends to give their findings greater credence. The official NSS data reveal the de jure structure; it is doubtful whether the de factor nature of land ownership will ever be known. 11/ An individual may operate a holding as a pure tenant (i.e., he rents all of the land he operates), as a part-tenant (i.e., he owns some of the land he operates and rents the remainder from a landlord) or as a landlord (i.e., he owns all the land he operates). Alternatively land owners may rent out all their land to other cultivators. Thus ownership holdings cannot be assumed to coincide with operated holdings. Conceptually separate from all these groups (whose members have "access" to land, and are discussed in this section and the following section) is the group of landless wage-earners (who neither own land nor operate it as tenants, and are discussed in the last sections of this chapter). 12/ The North comprises Jammu and Kashmir, Himachal Pradesh, Punjab, Haryana and Uttar Pradesh; the West, Marharashtra, Gujerat, Rajasthan and Madhya Pradesh; the South, Tamil Nadu, Kerala, Andhra Pradesh and Karnataka; and the East, West Bengal, Bihar, Orissa, Assam and Arunachal Pradesh. (See Annex 1 at the end of this paper for a discussion of the rationale for this subdivision). 13/ The term "Round" applies to a particular iteration of the NSS. 14 The differences between the two sources are significant both for the all-India aggregates (the reported number of households is similar, but not the reported area) and for various states, where the differences on holdings and area are often as large as 30 percent. See NSS 26th Round, Sanyal (1976), and Lakshiminarayan and Tyagi (1976). There - 106 - was a small change in the definition of "ownership" between the NSS 8th Round (1954-55) and the subsequent NSS 16th and 17th Rounds (1960-62) and 26th Round (1971-72), but data for the latter Rounds are fully comparable. 15/ According to the Rural Bank of India's All India Debt and Investment Survey for 1971-72, the percentage of all rural households owning no land was 3 percent in Bihar and Uttar Pradesh, 4 percent in Rajasthan, 5 percent in Punjab, 7 percent in Haryana and Orissa and 8 percent in Madhya Pradesh, but was 10 percent in Andhra Pradesh, 11 percent in West Bengal, 12 percent in Kerala, 16 percent in Gujerat and Maharashtra, 17 percent in Karnataka and Tamil Nadu and over 19 percent in Assam. 16/ Published data in South Asia are provided on both an acre and hectare basis (1 Hectare = 2.471 acres). Where possible data have been converted to a hectare basis, but in many cases the only published data are on an acre basis and cannot be easily converted to a hectare basis. The reader will have to accept this inconvenience. 17/ These include the studies by Vyas (1979), S. K. Sanyal (1976) (1977) (1979), H. Singh (1976), A. S. Sirohi, C. S. Ram and C. B. Singh (1976), S. K. Sinha (1976) and P. C. Joshi (1979) among others. Also see the special Conference Number (July-September, 1976) of the Indian Journal of Agricultural Economics. 18/ What is 'marginal' or 'small' is always relative and varies from region to region. In order to retain some consistency the following approximate definitions have been used throughout: marginal (less than 1 ha or 2.5 acres), small (between 1.1 to 2 ha or to 5 acres); medium (between 2.1-10 ha or 5.1-25 acres), large (between 10.1-20 ha or 25.1-50 acres) and very large (over 20 ha or 50 acres). 19/ Cini coefficients give different results depending on how they are calculated. Those calculated by Harpal Singh (1976) for ownership holdings (including landless - 107 - households) show a worsening distribution (0.8375 for 1953- 54, 0.8596 for 1961-62 and 0.8738 for 1971-72), but Sanyal (1979) has pointed out that these, being based on only five size classes, are misleadingly high when compared to those calculated using 12 size classes. The latter are closer to Vyas's findings. 20/ Bangladesh: Rural Development in Four Thanas in Kushtia District (1978, p.16). 21/ A. R. Khan (1977, p. 155). 22/ Jannuzi and Peach (1980, p. 18). 23/ Earlier censuses and surveys were confined to operational holdings, which are discussed later. The results from the LOS have been variously reported by Jannuzi (1976), and by Jannuzi and Peach (1976, 1977, and 1979), but are all reported again in Jannuzi and Peach (1980). Since the various reports contain slightly different figures, we have used the last as the final and correct version. 24/ "Homestead land (owned) is the area of land owned and occupied by the dwelling units(s) of the household and the immediate area surrounding the dwelling units." (Januzzs and Peach, 1980, p. 131). Data on land ownership excluding homestead land, which by definition is not even potentially cultivable, is given in Appendix Table A.3. 25/ These data are not strictly comparable with Indian data, partly because they are based on claims rather than records and partly because they exclude urban owners of rural land. "Total land owned refers to the total of all land claimed by respondents and does not necessarily coincide with total land area in Bangladesh" [Jannuzi and Peach (1980), p. 181 and ..."these data do not include rural land owned by urban residents... Land owned by non-resident land- owners is therefore not necessarily included... Therefore it must be stated categorically that the amount of land owned by urban residents in rural areas is not known" (ibid, pp. 18-20). - 108 - 26/ Based on these data, the Gini coefficients of ownership holdings are 0.6577 and 0.6842 for 1977 and 1978. These may be compared with 0.6635 for all India, and the following selected Indian states in 1971-72 - Bihar (0.606), West Bengal (0.5554), Kerala (0.4953), Assam (0.5416), Karnataka (0.6039), Orissa (0.5666) and Tamil Nadu (0.6429). Only two Indian states -- Andhra Pradesh (0.6859) and Punjab (0.7346) -- had a higher concentration of ownership holdings [see Sirohi, Ram and Singh (1976)]. 27/ Again, how reliable are these data? Between 1977 and 1978, land owned per capita apparently increased in almost all size classes. Table A2.3 shows an average of 16.25 persons per household owning over 15 acres. One year earlier, the average household size in this class was 11.95 persons -- another anomaly left unexplained. 28/ Total homestead land declined from 1.17 mln. acres in 1977 to 1.15 mln. acres in 1978. By changing the definition of the "size of holding" and hence the distribution, Jannuzi and Peach make it impossible to calculate how much land area is used for homesteads vs. cultivation in each size class. 29/ The evidence is from Ali Akbar's (1975) study of the 1974 near-famine conditions in Bangladesh. 30/ This survey by the Bangladesh Insitutute of Development Studies covered 1774 village households and 787 inmates of langarkhanas or gruel kitchens where people rendered destitute by famine received free meals. 31/ This section is based on M. H. Khan (1979) who has done the first thorough analysis of the available data. 32/ Extreme care has to be taken in interpreting ownership data for Pakistan, especially when comparing them to data from India or Bangladesh, for two reasons. First, they pertain to "persons owning land", or individual "owners", rather than households. The number of owners far exceeds the number of households. Second, they do not cover those households which own no land. Both of these characteristics make the Cini coefficient higher than it would otherwise be. - 109 - 33/ The relevant data are shown in Annex Table A.9. 34/ These two provinces dominate Pakistan' s economy. In 1975-76, they accounted for 79 percent of all land owners and 85 percent of the owned area excluding Baluchistan. They also contributed 85 percent of the cultivated area, 92 percent of the irrigated area and nearly 90 percent of Pakistan's output of major crops and foodgrains. M. H. Khan (1981, Ch. 1). 35/ M. H. Khan, (1981) Table 3.8. 36/ The Indian data are for both East Punjab and Haryana and have been adjusted to exclude the category of households owning no land (a category not given in the Pakistan data). The figures are still not strictly comparable because, as noted earliers, the Pakistan data are for individual "owners" while the Indian data are for "owning households". 37/ Although there must be large holdings hidden under bogus transfers in both regions, the extent of this concealment is unlikely to differ between them. 38/ A more detailed breakdown is given in Table A.11 in the annex. 39/ Small owners who rent out their land do not necessarily join the ranks of the "landless" or "agricultural wage earners." The class of "non-operating owners" covers people engaged in occupations other than agriculture, including workers in cities. 40/ These broad conclusions are supported by the analyses by K. N. Raj (1976) and Harpal Singh (1976) and by the analysis of state data by Sirohi, Ram and Singh (1976), except that the latter find a dramatic decline in the average size of marginal holdings between 1960-61 and 1970-71, even in Pun jab. 41/ See Sanyal (1977) and Sirohi, Ram and Singh (1976). 42/ According to the LOS, 72 percent of all households that owned land were in the 0.1-2 acre category, and accounted - 110 - for 24 percent of the owned land in 1977; according to the 1976 Pilot Census, 48 percent of all cultivators were in the 0.1-2.5 acre category, accounting for only 18 percent of the area cultivated. 43/ The 5 acre cut-off may represent a "subsistence" holding below which farmers in Pakistan cannot make an adequate living. Those with holdings smaller than this may rent them out or sell them to take up other occupations. 44/ See for example I.L.O. Poverty and Landlessness in Rural Asia, Geneva, 1977 and M.J. Esman Landlessness and Near- Landlessness in Developing Countries, Cornell University, 1978. 45/ "The very recent fad about and display of interest in the landless is less due to a charitable concern on part of the established officialdom and academia than due to a very real bout of enlightened self interest arising out of the threat of disintegration to the established order in the face of the growing trend of landlessness and agricultural stagnation." S. Adnan et.al., (1978). 46/ There is a danger that those who oppose one simplistic definition of the rural poor as "small farmers," may end up espousing another--a category of poor people whom they label the "landless and near landless". In fact, it is precisely the heteregeneity of the rural poor--in terms of such variables as their degree of command over land and other assets, how they use these assets, how productive these assets are, their other sources of income, how they share incomes and in what proportions--that makes the problem of rural poverty so intractible. Unwarranted generalizations that identify the rural poor either with "small farmers" or with only "the landless and near landless", both confuse and oversimplify the issue. 47/ See Davatia (1976), H. Lakshminarayan (1979), Reserve Bank of India (1976), G. Kelkar and S. Subramaniam (1977) and N.S. Krishnamurti and S. Subramaniam (1977). The most - 111 - important source on the pattern of asset holdings in rural India is the information collected y the Reserve Bank of India in the All India Debt and Investment Survey (1971). 48/ This assumption pervades most of the literature on poverty and landlessness, starting with Dandekar and Rath (1981) through I.L.O. (1977) and the Esman study (1978). 49/ As Adnan et al. (1976) st'ate: "the literal meaning of the term "landless" is far too imprecise and heterogeneous for any serious undertaking to identify the poorest target groups ... The moral from this would be that from the policy maker's viewpoint, of identifying the landless and assetless "ttarget groups," mere landholding-stratification is not enough, given the significance of other forms or means of production in the non-agricultural sector" (p.28). 50/ For example, in a recent study using 1972-73 NSS data, Visaria (1980) showed that the amount of land operated per capita explained no more than 4-10 percent of the variance in the per capita expenditures of rural households in Gujerat and Maharashtra. The association between the size of land holding cultivated and the incidence of poverty was found to be extremely weak. 51/ See the NSS survey on "Weaker Sections," 25th Round, 1976, Number 233. 52/ This is not necessarily a safe assumption, because changes in the 1970s in response to the green revolution are likely to have differed from those in the 1960s. 53/ NSS-based findings are substantiated by evidence from the Reserve Bank of India's All India Debt and Income Survey, which shows that the proportion of "poor" households (households with assets of up to Rs. 1,000 in 1961, and of up to Rs. 2,500 in 1971), reporting land as an asset increased from 27.5 percent to 30.5 percent between 1961- 1971. At the same time, however, the number of households reporting no land as an asset also rose, from 14.9 m to 18.8m. - 112 - 54/ Comparable data on the percentage of households "owning no land or land under 0.05 acres" show a similar decline, from 23 percent in 1953-54 to 11.7 percent in 1961-62, and to 9.6 percent by 1971-72. (NSS: 8th, 17th and 26th Rounds). 55/ Assuming that the average Indian agricultural household has 4.74 people, as reported by the Second Rural Labor Enquiry for 1974-75. 56/ It is worth recalling that nearly a third of all leased-out land in India comes from holdings of less than 2.02 hectares. This proportion is as high as 50-60 percent [of the leased-out total] in many states. 57/ Esman (1978) for example. He also gives figures for the "landless and near landless" in India of 53 percent and for Bangladesh at 75 percent. In an earlier version of his study he gave figures of 79 percent for India and 88 percent for Bangladesh. 58/ Out of a total of 11.85 million rural households, 8.19 million were shown to be cultivating as owners, or as mixed or pure tenants. 59/ The evidence is from M. Hossain's (1976) study of tenancy in Bangladesh. 60/ "Landless are those who live by working on other farmer's field. They may own their own homestead land but they do not own any farm land." This somewhat ambiguous definition does not cover sharecroppers, who are treated as a separate category. See BARC (1978). 61/ "It is very difficult to construct a time series of the number of landless laborers, and the figures in the table must be treated as approximate. Unfortunately the definitions change from one survey to another. The decennial censuses of 1951 and 1961 enumerated the landless agricultural laborers as a distinct category. But the various rounds of the Master Survey of Agriculture used different and more detailed categories. The estimates have been built up from these categories as carefully as possible - 113 - in the absence of a detailed explanation of some of ti important terms (emphasis added). Moreover, the methodology employed was designed to ensure that, if anything, the estimates for the latter years understate the extent of landlessness in rural Bangladesh" (p. 156). Khan (1976), p. 156. 62/ Quadir (1960), S. Adnan and H. Z. Rahman (1978) and H. Z. Rahman and A. Rahman (1978). 63/ This widely quoted study states that samples were 'admittedly neither sufficiently large nor scientifically drawn -- I carried out a complete enumeration of all households in 96 villages or paras which were picked "intuitively at ramdom" -- i.e. by grabbing any bundle of schedules that looked neither "too large nor too small"' (p. 212). The 1961 census gives data on "landless agricultural laborers" as a percentage of the "civilian labor force." By assuming similar participation rates and household sizes for all categories of the total agricultural labor force, Abdullah obtains the proportion of "landless labor households." To these he adds households (i) renting [in] all land, (ii) renting land tilled and also working for hire and (iii) sharecroppers, to get the extent of landlessness. 64/ Various estimates of household size and the number of rural households are available; those chosen are the ones deemed most reliable. 65/ Differences in definitions and in the quality of the data make cross-country comparisons of this kind approximate and indicative. 66/ Jodha, N.S., 1972 and 1979. 67/ Based on age-adjusted participation rates from the 27th Round of the NSS and "best guess" population estimates. See Grawe, Krishnamurti and Baah-Dwomoh (1979). 68/ Assuming that the rural labor force continues to account for 83 percent of the total labor force, as it did in 1972-73. - 114 - 69/ Rural Labor Enquiry -- 1974-75: Summary Report. The emphasis is on "manual" and on "wage payment" -- in cash or kind; the definition broadly identifies typical occupations of those without skills or other assets. 70/ The figures differ from Table 14, both because they are for different years and because Table 14 refers to the rural population, while Table 15 refers to rural households. 71/ This is lower than the proportion of the rural labor force, some 39 percent, classified as 'casual or agricultural laborers' in 1972-73. The discrepancy can arise because individual members of households otherwise not classified as "rural labor households" could still be seeking wage employment as casual or agricultural laborer. 72/ This is the same regional classification as that used in the earlier section on land. See Annex 1. The proportion of all rural households in India in each region was: North (22 percent), East (27 percent), West (24 percent) and South (27 percent). 73/ Some of the households listed as rural labor in 1974-75 may have been tenant farmers whom their landlords listed as laborers to conceal their tenancies. The extent of this bias is not known. 74/ The figures do not imply that these workers were unemployed during the rest of each year; many agricultural labor households also cultivate their own land. Separate data on days worked as laborers for those with and those without land would be more revealing but are not available. 75/ More recent harvests may have ameliorated this trend partially. 76/ Rural Labor Enquiry 1974-75 op. cit. p. 18-21. See also Chapter 7 for a discussion of small farmer credit sources. 77/ Many of these money lenders are also the employers of their debtors, and thus exert considerable power over them. See Prasad (1976). The problems of interlinked factor and product markets is also discussed in Chapter 6, in Singh, I.J. "The Great Ascent: The Rural Poor in South Asia". 1988. - 115 - 78/ In India in 1961, over 18 percent of the (rural) female labor force was employed in industry (including household industries), trade and commerce, with 71 percent in agriculture. By 1971 the ratios were 9 percent and 80 percent respectively. See A. Mitra (1979). 79/ op. cit. p. 20-21. 80/ E. Clay and S. Khan (1977). 81/ A 1974 survey found that 80 percent of cultivators employed wage labor and that 33 percent of the total labor used for farm production was hired, the majority being employed on a casual basis. According to the 1960 Census of Agriculture, about 51 percent of farms reported the use of hired permanent labor. M. Hossain (1978) 82/ Using an average household size of 4.72 as estimated for those households who owned no land other than homestead land in 1978. Jannuzi and Peach (1978) p. 108, Table E.2. 83/ Some of these findings are based only on small studies. See particularly ILO, Dacca (1977). 84/ Chowdhury (1978), Bertocci (1972) and Wood (1976). 85/ University of Dacca (1978). 86/ University of Dacca (1978). 87/ M. Masum (1976). 88/ Alamgir (1976), Clay and Khan (1977). 89/ The per capita annual incomes cited for 1970-71 are: large farmer, Rs 1102; small farmer, Rs 318; tenant farmer, Rs 200; permanent laborer, Rs 192; and temporary worker, Rs 173 [Eckert (1972) p. 571. 90/ The conceptual and measurement problems in estimating unemployment are discussed in some detail in Raj Krishna (1976), A. Sen (1975) and K.N. Raj (1976). 91/ The Indian debate on these issues can be found in Raj Krishna (1973a, 1973b, 1976), Dantwala (1970, 1973), Visaria (1972) and K.N. Raj (1976). There are only a few studies available for the other two countries: for Bangladesh, see Ahmed (1978), Clay and Khan (1977) and Clay (1978), and for Pakistan, see J. Eckert et al. (1973). - 116 - 92/ Grawe, Krishnamurty and Baah-Dwomoh (1979) have taken a thorough look at the various sources of data. Much of this section is based on thier findings. The definitions used in the NSS data are also discussed in Annex 1. 93/ Both the daily and weekly status unemployment rates have not changed much between 1972-73 and 1977-78. The data show a slight decline in both rates in this period. Urban rates are also higher than rural rates. See Table A.33 in the annex. 94/ Between 1956 and 1970, the most rapid output growth occurred in Punjab, Haryana, Gujerat, Karnatka, Rajasthan, and Uttar Pradesh. See Bhalla and Alagh (1978). 95/ An association between high agricultural growth rates, greater than average increases in labor demand, and high rates of growth in labor productivity has been suggested elsewhere; see IBRD: Economic Situation and Prospects of India, April 17, 1978, p. 75. 96/ IBRD 1978, op. cit. 97/ See Berry and Sabot (1978) for a survey of views on labor markets in LDCs. Also see Ridker (1971) and P. Bardhan (1978), who contend that the poor cannot afford to be unemployed and that they take whatever work they and find. 98/ See P. Visaria (1970 and 1975). 99/ See Tables 2 and 3, Sarvekshana, Vol. II, No. 2, October 1978. 100/ These rates are comparable with, though not as high as, frequently cited rates of "underemployment," based on arbitrary norms involving time or productivity/income criteria. See the review by Brannon and Jessee (1977) which uses standard definitions of underemployment in terms of time or productivity "norms". 101/ Since many of those willing to work may already have left, this sample of those who have stayed behind may be biased. 102/ The rural labor force is estimated to have increased by 14 percent between 1963-64 and 1969-70 and another 16 percent - 117 - between 1973-74 and 1979-80. See IBRD, Bangladesh: Economic Developments and Second Plan Review, Jan. 1981, Annex V. See also E.J. Clay and J.S. Khan (1976, 1978). 103/ Data from the Bangladesh Institute of Development Studies (BIDS) Survey of 1,774 village households and 787 inmates of langarkhanas (soup kitchens where famine stricken people were provided free meals) in 1974. Cited in Alamgir (1978). 104/ IBRD op. cit. 105/ Unfortunately even this rate involves the use of an arbitrary "full employment norm" because the "total potential labor days available" are obtained by multiplying "potential workers" by this norm. The unemployment rates are then calculated by comparing "labor need" against this arbitrary potential. 106/ See Annex 1 for definitions of these measures. 107/ Even the World Bank uses the arbitrary time criterion. "For the crop year 1977/78 labor requirements in agriculture were estimated at... 13.2 million person years of 250 days/year... (sic) is estimated at 43 percent of the agricultural labor force." IBRD, Bangladesh: Economic Developments and Second Plan Review, Annex IV, p. 6, Jan. 1981. 108/ Soup kitchens at which famine victims were given free meals. 109/ Masum (1979), examining a sample of 127 farms in Phulpur in 1979/70, found that 5 percent of the labor force worked less than 100 days/year. 110/ GOP: Labor Force Survey, 1974-75. 111/ GOP; Fifth Five Year Plan (1978-83), Planning Commission, June 1978, p. 17. Participation rates are around 58 percent for men, but very low -- around 8 percent -- for women in rural areas, reflecting the structure of Muslim society. 112/ Ridker (1971). 113/ A concept comparable to the current (or weekly status) rate in Indian NSS data. 114/ J. Eckert et al. (1973). 118 - 115/ In rural areas only 0.8 percent of the illiterate are shown as "unemployed," compared to 8.4 percent of all matriculates and 14.9 percent of degree holders. The 1974-75 Labor Force Survey gives total rural "unemployment" as 1.3 percent. 116/ Of the total rural population in the Punjab, they estimated that some 12 percent fell in the permanent and temporary labor categories. The majority of hired laborers come from kamee households, who are neither land owners nor tenants but landless laborers or village craftsmen. Hirashima (1973). 117/ Although this sample predated the period of extensive outmigration from rural West Pakistan, Eckert estimated that over a million laborers had left Punjabi villages for work elsewhere before 1971. Subsequently, a growing share of the labor force has migrated from rural areas to the cities, or to jobs in the Middle East. Between 0.5 and 1.0 million Pakistanis (roughly 5 percent of the labor force) were employed in the Middle East in 1981. Roughly 43 percent of the outmigrants are unskilled and 12 percent of these come directly from the agricultural sector. One consequence of heavy outmigration has been a reverse flow into rural areas of remittances from migrants, leading to a rise in rural land values and reported shortages of labor in rural areas. Having had an average income of Rs. 9,000 a year prior to migration, these migrants each remit Rs. 28,000 annually to Pakistan for a period of 3-4 years. In 1980/81 total remittances via official channels were estimated at $1.96. A significant proportion (5 percent) of their total remittances has gone into agricultural land and investments. Rural wages have risen significantly in the past two years. See I.B.R.D. Pakistan: Economic Development and Prospects, Feb. 23, 1981, Annex 1. 118/ The issues discussed in this paragraph are examined further in Chapters 4, 5 of Singh, I.J. "The Great Ascent: The Rural - 119 - Poor in South Asia," 1988, which analyse the prospects of productivity and output growth in the small farm foodcrop sector. Similarly, Chapter 6 provides a detailed treatment of rural employment and income opportunities in non-crop agricultural activities (livestock, etc.). 119/ Migration represents another option for the rural poor, but its potential as a solution to the oversupply of labor in rural areas is strictly limited -- notably by constraints on the mobility of potential migrants and on the absorbtive capacity of receiving areas and employers. Rural-urban migration is a major subject of study in its own right and is not discussed comprehensively in this book for reasons of space. There have been a number of useful studies on the dynamics of migration in South Asia; the interested reader might begin by examining Connell et al. (1976). 120/ See Singh, I.J. "The Great Ascent: The Rural Poor in South Asia," 1988, Chapters 4, 5 and 6 examine the prospects for growth in these activites in some detail; here I only summarize recent trends in the employment and earnings they provide. 121/ I.B.R.D., Rural Enterprise and Non-Farm Employment, Jan. 1978 and a shorter paper by Anderson and Leiserson (1980) based on it. Also see E. Chuta and C. Laidholm (1979) for a review of current understanding of the role of non-farm employment. 122/ These are major non-agricultural rural activities other than crop produc-tion, livestock, dairy or poultry farming, and forestry. They include manufacturing, repair, processing, construction, handicrafts and artisanal work, transportation, trade and commerce -- many of which are, of course, ancillary to agricultural activities and needs. 123/ Estimates for some countries are: 20-24 percent in India, around 50 percent in Taiwan, 30-40 percent in the Phillipines, 20-25 percent in South Korea, 30-40 percent in West Malaris and around 30 percent in Indonesia. - 120 - 124/ See also Chuta and Leidholm (1979) p. 18. 125/ Manufacturing includes a wide variety of activities in food processing including oil pressing, milling and husking, handicrafts, leather works, textiles, wood products including sawmilling, furniture and carpentry, metal working including blacksmithing, welding, fabrication and assembly work for building machine tools and equipment (in Pakistan and E. Punjab and Tamil Nadu it includes small scale manufacture and repairs of diesel and electric tubewells); construction includes that of dwellings and farm buildings, roads and civil works; in commerce, retail trade accounts for three quarters of total employment, the remainder being made up of marketing and financial services; other services include small petty trading, business and repair and artisanal work. 126/ Anderson and Leiserson (1980) p. 245. 127/ This proportion has declined by 3 percentage points in India and increased by 4 points in Pakistan in the intercensus decade, but this could be due to differences in what is included in the 'non-agricultural' category. 128/ These data partly overstate the importance of non-farm incomes as they include pensions, remittances and rental incomes. Rent is, of course, an unlikely source of income for the landless or those with small holdings if they are also cultivating land, but could be a significant part of non-farm income for those with larger holdings. Remittances on the other hand could be a significant source of income for the poorest groups; the data needed to establish their importance are unfortunately not available. 129/ Biases resulting from the inclusion of remittances, pensions and other transfer incomes in non-farm incomes should be kept in mind, however. 130/ Visaria and Visaria (1973). These include livestock and dairying, poultry, forestry, and fishery activities. 121 - 131/ See AERC-Waltair (1973, 1975, 1976), and AERC-Madras (1976), for example. But agricultural laborers are also worse off in some states; see Chauhan et al. (1973) and J. Kaul and S.S. [ahlon (1971). 132/ I.B.R.D. op. cit. (1978). 133/ "Over half the annual migrants from rural Punjab is derived from land owning households of small or large farmers. An additional 10 percent comes from farm tenant households. Fifteen percent are classified as unskilled laborers while 17.5 percent bear with them the diverse skills of their respective trades,," J. Eckert and D.A. Khan (1977). 134/ Even village artisans have incomes that exceed those of permanent hired laborers by 20 to as much as 115 percent for such well paid trades as blacksmiths, barbers and weavers. See J. Eckert (1972) and D.A. Khan (1978). Between 1964-65 and 1976-77, in-kind and wage payments for farm work by permanent laborers rose by 149 percent in nominal terms; non-farm occupations matched these increases. 135/ Khan (1978, p. 935-36) estimates that on-farm employment on small farms increased by some 0.2 m. man years and on medium and large farms by 0.99 m. man years. He reports that non- farm employment increased by 1.5 m. years. These estimates are, however, based on the use of simple fixed I/O labor coefficients to crop averages and indirect employment, is assumed to increase by 1.23 man years for every man year of farm employment -- an estimate based on an Indian study. 136/ The 1980 projections are based on the agrarian structures prevalent in the 1970s and on estimated population changes between 1970-80. 137/ Recall that family size is positively related to size of holding, so that the proportion of households in larger than the proportion of population accounted for by poorer households. We are also aware that land size per se (i.e., without reference to productivity) means little. Still these are fa;-'- meaningful grouping. - 122 - 138/ A regional district-wide study by Y. Alagh and C. S. Bhalla (1979) was used to fix productivity ranges here. - 123 - Annex 1 Grouping of Indian States by Regions The four broad regional groups used for disaggregation of country-wide data on India are: -The North, comprising the states of Jamnu and Kashmir, Himachal, Pradesh, Punjab, Haryana and Uttar Pradesh, in 1971 accounting for 22 percent of all rural households and 18 percent of total cultivated area; -the East, comprising the states of Assam, Bihar, West Bengal, Orissa, Nagaland, Mizoram, and Arunachal Pradesh, accounting in 1971 for 27 percent of all rural households and 15 percent of total cultivated area; -the West, comprising the states of Rajasthan, Maharashtra, Gujerat, and Madhya Pradesh, accounting in 1971 for 24 percent of all rural households and 46 percent of total cultivated area; and -the South, comprising the states of Karnataka, Kerala, Andhra Pradesh, and Tamil Nadu, accounting in 1971 for 27 percent of the population and 21 percent of total cultivated area. Each of these regions has a relatively similar rural population and cultivated area, and can be distinguished on the basis of the characteristics listed below. It should nonetheless be borne in mind that each region includes an enormous diversity of conditions. (a) Agro-climatic zones: the North and East are characterized by mainly alluvial soils, the West by black soils and the South by red and lateritic soils. In addition, a combination of low rainfall and high seasonal temperatures gives the North a relatively low moisture index (between -20 and -80), similar to the West. The East has a very high index (-20 to above +100) while the South has a moderate moisture index (-40 to -60). (b) Agricultural growth: the Northern states have had rates of growth of 2-6 percent a year; the Eastern states have had - 124 - negative to low growth rates never exceeding 4 percent; in Western states, rates have ranged from negative to 2 percent a year; the South has had rates ranging from negative to 6 percent a year. Four of the eight states with per capita annual growth in excess of 2 percent are in the North; three of the six states with negative per capita growth rates are in the East. (c) Main food crops: the North mainly grows wheat and maize, the East and South mainly rice, and the West maize and pulses with some wheat. (d) Foodgrain yields: 138/ yields in the North are valued at 800-1200+ Rs. per hectare, in the South at 1000-1200+ Rs. per hectare, in the West under 600-800 Rs. per hectare, and in the East at 600-1200 Rs. per hectare. (e) Irrigation: The irrigated proportion of the total cultivated area is about two-thirds in the North, about a fifth in the South and East, and only about a tenth in the West. (f) Demographic pressures: the East and West have had population rates of growth in excess of 2 percent, while the South has generally had less than 2 percent growth; experience in the North has been mixed. Finally, the Indian North is similar to Pakistan in many ways, while the Indian East is closer to Bangladesh. - 125 - Annex 2 The data on the size distribution of land holdings and various measures "landlessness" in South Asia presents a daunting challenge to anyone who wants to reconcile conflicting sources of data on the number of rural households, land areas, ownership and operated holdings and provide consistent numbers. Although aggregate national level data were available for India on a consistent basis over some years state level NSS data were not available in the public domain. Data for Bangladesh and Pakistan over any time period were either unavailable or were difficult to reconcile from one source to another. Where originally published data were available I have used these--where these were unavailable we had to reconcile different data sources to arrive at some consistent estimates. This was particularly so in getting data by states to aggregate into regional groupings--North, East, South, West--for India. Certain assumptions also had to be made in obtaining consistent estimates for Bangladesh and Pakistan for various key categories of the agrarian structure. This Annex presents some of the key data sources and assumptions used in constructing some of the tables in this paper. (1) Data on Rural Households Whereas the Indian data provides reasonable aggregate estimates for the number of all households in rural areas, data from Pakistan and Bangladesh provide estimates on only "cultivating households" or "rural labor households". Estimates for all rural househoLds had to derived from rural population and household size figures. (a) For Pakistan it was even difficult to obtain a reasonable estimate on the average rural household size. I settled on a figure of 5.8 after reconciling the evidence provided in [11, [5], [7] and [12] with data on the distribution of rural households by provinces and for the country as a whole were obtained by dividing the rural population estimates - 126 - given in [17]. (b) For Bangladesh I assumed an average household size of 5.8 given in [9] and using estimates of the rural population for 1977 given in [8] and [16] estimated the number of rural households in a similar manner. (c) For India I had the national level NSS data, but in order to calculate the number of all rural households by states and to aggregate these into regional totals I took the data from [13] and reconciled the state level totals from 11] to ensure comparability. (2) Data on Ownership Holdings These were the most difficult to reconcile because data on land ownership in the Indian surveys and the latest Bangladesh survey are associated with "ownership holdings" while those in Pakistan are associated with "owners". This distinction should be kept clearly in mind when comparing data across countries. (a) For Pakistan I used the data as reported by M. H. Khan [10] because of the great care he has taken in reconciling sources. (b) For Bangladesh the data on ownership holdings were taken directly from [9]. The number of ownership and cultivating households given in this source could not be easily reconciled with those given in [2] and [3]. These data are presented separately without any attempt at reconciliation in Tables 3 and 6. (c) For India, the NSS state level data on the size distribution of ownership holdings were not available. Data were contained from [9], [11] and [15] to obtain consistent estimates for each state and these were then aggregated into regional totals. The resulting estimates differ slightly from the regional distributions presented by some researchers (e.g. Sanyal [14] who had access to NSS state level tabulations. - 127 - (3) Data on Operated Holdings These data were more readily available on a consistent basis. (a) For Pakistan the data available in [2] and other sources has been carefully analysed by M. M. Khan [10] and we report his figures here. (b) For Bangladesh, I used the data reported in [2], [3] and [9]. The data in [91 is given in terms of areas owned and leased by size of holdings and these have been added together to get total operated areas. (c) For India, the NSS state level data on the size distribution of operated holdings were not available. Again data were combined from [9], [11] and [15] to obtain consistent estimates. The methodology is described in the footnote to Table 5. (4) Data on "Landlessness" Data on landlessness are the most difficult to obtain even once the conceptual confusion about what is meant by the term are resolved. The only consistent and even partially reliable data are from the NSS in India. The other data are less unreliable and usually inconsitent. (a) For Pakistan the data had to be constructed from various sources. There sources and assumptions are clearly spelled out in Table 8. In order to arrive at the number of "non- owning" and non-operating" households the data on "owners" and "operators" provided in [5] and [12] are subtracted from the estimates of the total number of rural households obtained earlier. (b) For Bangladesh the data are available from a number of sources, are inconsistent, used varying definitions and cannot be easily reconciled. (c) For India the data are from the NSS [11] at the all-India level. State level data were not available except for selected states in Sanyal [14]. Other supplementary tables used in this paper are provided in Annex Tables A.3. - 128 - Annex 3 Rural Unemployment: Concepts and Definitions Raj Krishna (1973) has outlined four broad criteria for measuring unemployment, based on different but widely used concepts. Depending on the criterion used, a person may be counted as "unemployed" or "underemployed" if he/she: (1) "Is gainfully occupied during the year for a number of hours (or days) less than some normal or optimal hours (or days) defined as full employment hours (or days)" -- the time criterion; (2) "Earns an income per year less than some desirable minimum" -- the income criterion; (3) "Is willing to do more work than he/she is doing at present: he/she may be actively searching for more work or be available for more work if it is offered on terms to which he/she is accustomed" -- the willingness criterion; (4) "Is removable from his/her present employment in the sense that his/her contribution to output is less than some normal productivity level and therefore his/her removal would not reduce the output or productivity of the remaining workers and/or organization" -- the productivity criterion,. In the present study I am primarily interested in the extent to which people would be able to get work (or more work than they currently have) at the going or customary wage, if additional employment opportunities were available. From this point of view, data based on the willingness criterion are the most relevant, because it explicitly identifies those who are available and willing to do more work than they are presently doing. How do we measure unemployment rates based on the "willingness" criterion? First, I need to distinguish between different ways of defining the incidence of unemployment, (even after deciding upon the criterion for judging someone to be unemployed). The two most common measures of the incidence of - 129 - joblessness are known as the stock rate (also called the person rate) and the flow rate (also called the time rate) of unemployment. The first measures the number of unemployed persons at a given point in time, as a percentage of the number of persons in the labor force; the second measures the units of time (in hours, days or weeks) over a given period (a week, a month or a year) of unemployed labor as a proportion of the total time units of labor supplied at the going or customary wage. The use of these different concepts will give different percentage estimates of unemployment Second, I need to distinguish between the type of unemployment in terms of the time period over which its incidence is measured. Again, there are two commonly used time bases for estimating employement status, known as current and usual (or chronic) rates respectively. (See below for the distinction between "current"1 and "usual" unemployment in the statistics for India). Finally, given the extent of self-employment in rural areas (in farm or non-farm enterprises), and the fact that so much of the work done consists of household tasks shared among members, even extended concepts of what it means to be "employed" or "'unemployed" or to be "seeking employment" must be interpreted with care. Most poor people -- and those who rely mainly on rural wage employment fall in this category -- try to engage in whatever gainful work they can find, often on a temporary or casual basis. The amount of work they do, the period over which they work, and the income they earn from work are all difficult to define and measure. I have kept these caveats in mind when examining the available evidence. The NSS in India measures unemployment based on three concepts: (a) The usual status concept, which is meant to determine the usual activity status--employed, unemployed or outside the labor force--of the respondents covered by the survey with reference to a given time period-- usually confined to the - 130 - previous 365 days. This gives data on those who are continuously unemployed and seeking work over a longer period. These are also often refered to as the usually or chronically unemployed. (b) The weekly status concept, which uses a reference period of the preceeding seven days, and according to which a person who reports as having worked for at least one hour during the reference period of one week while pursing a gainful occupation is deemed to be employed. A person who did not work even for one hour during the reference period but was seeking or was available for work is deemed to be unemployed. (c) The daily status concept, which records the activity status of a person for each of the preceeding seven days. A person who worked for at least one hour but less than four hours was considered to have worked for half a day while one who worked for four hours or more was considered to have been employed for the whole day. All three measures are stock rates, which measure the number of people in the labor force whose current or usual status is to be out of work. The usual status measure is more appropriate for those in search of regular employment who may not accept casual work, while the weekly and daily status unemployment estimates bring out the seasonal and part time aspects of unemployment more effectively. The longer the reference period used, the greater the likelihood of being employed "for each one hour or one day" and the lower the unemployment rates. Accordingly the daily status unemployment rates exceed the weekly status rates, which in turn exceed the usual status rates. Many consider the daily status unemployment rate as the most inclusive and significant indicator of the magnitude of unemployment. Finally, the NSS collects time disposition data on employment (by half days) during each day of the reference week; this yields a flow rate calculated in person days that is a more comprehensive measure of unemployment and underemployment of the "currently employed". Measures of unemployment based on the "current" (weekly) and "usual" (chronic) stock rate status use - 131 - very restrictive definitions by comparison with the "daily status" flow rates which actually try to measure the number of days employed, although in discrete half day units. As indicated earlier the "current" (weekly) status classifies a person as "unemployed" if he had no work (not even one hour) during the week preceding the interview and is willing to work and seeking a job, while the "usual" status classifies a person as unemployed if he was "usually" unemployed during the last year. Thus the usual status rate < current or weekly status rate