Rural employment and wages with agricultural growth in India : some intertemporal and cross-sectional analysis

RURAL EMPLOYMENT AND WAGES WITH AGRICULTURAL GROWTH IN INDIA: Some Intertemporal and Cross-Sectional Analyses by Kalpana Bardhan Series : Studies in Employment and Rural Development No. 38 Division : Employment and Rural Development Department: Development Economics Development Policy Staff International Bank for Reconstruction and Development The author prepared this research paper in her capacity as consultant to the Bank. This paper forms part of a research study on Structure of Rural Employment, Income and Labor Markets (RPO 671-30) undertaken in this Division. The views expressed represent those of the author and not necessarily those of the Bank. This report may not be published nor may it be quoted as representing the views of the Bank and its affiliated organizations. Washington, D.C., March 1977 TABLE OF CONTENTS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 1 PART I: CHANGES IN AGRICULTURAL EMPLOYMENT AND WAGES OVER TIME . . . . . . . . . . . . . . . . . . . 6 I.A. CHANGES IN FARM-LEVEL USE OF LAOR . . . . . . . . 6 - Data . . . . . . . . . . . . 6 - Differences in Agricultural Situations in the Four Cases . . . . 7 - Changes in Enployment, Real Wage Rate, and Labor's Share in Agricultural Value Added . 14 - Labor Use by Farm Size . . . . . . . . . . . . . . 25 - Mechanization and Labor Use Pattern in Fero zepur . . . . . . . . . . . . . . . . . . . 36 I.B. LABOR HOUSEHOLDS: CHANGE OVER TILE . . . . . . . 41 - Data . . . 41 - The Phases of Chag. 45 - Conclusion . . . . . . . . . . . . . . . . . . . . 56 APPENDIX I. SEASONAL RESPONSE OF WAGE RATE TO EMPLOYMENT . . . 58 PART II. CROSS-SECTIONAL RESPONSE . . . . . . . . . 66 II.A. FARM-LEVEL DEMAND FOR LABOR . . . . . . . . . . . 66 - Variabes . . . . . . . . . . 66 - Results of Inter-Farm Regressions . . . . . . . . 69 II.B. LABOR HOUSEHOLD'S SUPPLY BEAVIOR . . . . . . . . 75 APPENDIX II. TABLES A.1 - A.3 . . . . . . . . . . . . . . . . . 84 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 RURAL EMPLOYMENT AND WAGES WITH AGRICULTURAL GRO,WTH IN INDIA: Some Intertemporal and Cross-sectional Anyses Kalpana Bardhan For evaluating the distributive impact of increasing investment in agriculture within an institutional context of skewed distribution of cultivated land, an important criterion is the nature of associated changes in employment and wages. The Question has two aspects that interest us here. One is the farm management aspect concerning deter- mination of labor demand--the employment effect of a given combination of changes in technique and crop-mix, composition of the employment effect in terms of hired and family labor, and the wages paid by farmers. The second aspect is from the laborers' point of view--the changes in his employment and wages as the combined result of interacting shifts in supply and demand. Various nonmarket factors beyond the supply-demand framework might be important in this context, in ways about which we do not yet have the necessary information. Within this limit, it is worth finding out how far the observed variations between regions and changes over time in rural employment and wages can be explained in labor market terms. The object of this study is to identify the important factors determining labor supply and demand in rural areas. Using farm manage- ment data for cultivators on one band, and labor time disposition data 2 for landless and landmoor rural households on the other, the first part of the study presents two separate sets of intertemoral comparisons in wages and employment in agriculture. The focus is on differences in the regional context within which the change over time takes place. This part also contains a specific study of seasonality in farm employment and wage rates, based on the contrast between a fast- growing wheat-cotton district with large farms and landless laborers, and a less "progressive" paddy-jute district with small farms, cultivating laborers, and proximity to a big city. The second part, using same of the same data sets, attempts a cross- sectional analysis focused on explaining micro-level variations in employ- ment, in terms of demand and supply factors. The demand analysis, re- peated for two different years and for farm-size groups within a district as well as for a few different districts, gives some indication of behavioral differences associated with various institutional and agro- nomic contexts. Since the mid-sixties a number of major changes have taken place in many parts of rural India. There were substantial increases in public investment in supplying seeds, fertilizers, and credit. This was backed up by fiscal -policies that favored agricultural profit and encouraged farmers to invest in land, and the more prosperous ones among them to invest also in machines. It is normally expected that water-seed- fertilizer-based improvements in the pattern of cropping, unless counter- acted by labor-saving mechanization, would generate more employment-- for hired labor in paaticular--and also change the seasonal pattern by acti- vating the traditionally slack agricultural month's. In what ways wage rates and per capita employment would ultimately respond is determined also by * 3 labor supply behavior, which has two components, viz. (a) farm households' supply of labor for use on the farm in relation to their increased income, and (b) shifts in supply of landless and semi-landless rural labor in an area (i) through seasonal migration and (ii) through changes in the size-distribution of cultivated area. Migration (both seasonal and long-term) of usually landless wage labor, from the more depressed rural areas with short and spare cropping pattern and little else into the intensively cropped areas or into the urban area, has always been an important mechanism of labor supply adjust- ment to demand differences. During the recent period of agricultural growth, which has been largely concentrated in certain regions, the significance of rural-to-rural labor flows might have increased. Apart from adjustment of supply to the seasonal changes in demand for farm labor, the wage determination process might also be influenced by the fact that the wage contracts and the modes of mobilization of migrant labor are usually very different from those for local labor. On seasonal migration an. its impact on labor market processes, there is hardly any direct data. This obviously is a major handicap for an integrated analysis of rural wage determination process in the context of regionally varying demand shifts. The effect of any given increase in irrigation and cropping intensity is bound to be influenced by the concurrent changes in mode of production and farm size distribution. Over much of north India in particular, the large landowners, %ho traditionally leased out in small parcels to landless or nearly landless households, were now induced by the increased profitability of farm investment to take over direct management. They now started to operate large farms with hired 4 .,4 labor and increasingly with the use of labor-saving machines that saved management problems apart from wages, and were rendered economical for them by the favorable tax-price-subsidy policies. Ironically enough, the phase of eviction of tenants in the fifties was partly prompted by the tenancy reform legislation itself. Over much of north India the small tenants-at-will now became agricultural laborers. Even in the regions (like western India) where the tenancy legislation was more effectively implemented and many tenants got ownership of small amounts of land, the viability of their self-employment was not sustainable for long without supportive measures to raise yields. With the earlier availability of additional land to lease in now withdrawn from them, and with their lack of resources for improving yield of the tiny holdings, the workers of these households now became increasingly dependent on wage-employment. In growing areas of north and east India, even a reverse process of leasing out from tiny owners to large farmers has been observed in recent years. Over the sixties, as a combined result of changes in production -technology and those in the pattern of leasing in/out (and hence in the size distribution of farms), there was a sharp increase in the proportion of wage laborers relative to the primarily self-employed categories of rural workers. It is in this context of an increased proletarianization of rural working force and possibly also increased migration flows between. back- ward and progressive regions, that one has to evaluate the effect of recent agricultural growth'for rural labor. 1Vyas's (1976) study for Gujarat shows that the peasants who came to own tiny holdings as a result of a relatively well-implemented tenancy legis- lation, and who since then more or less stuck zo their holdings, now depended increasingly--with stagnating yields and growing families--on their workers taking up wage employment on larger farms or migrating. 5 Changes in wage rates and in employment for an individual laborer result from aggregative shifts in the supply as well as the demand side. But, with the sets of data available for the Dresent study, it is not possible to carry out a simultaneous determination exercise. The data required for that purpose must have within an integrated frame- * work both labor time disposition and farm management data, covering all types of rural households in a region. One would also require, apart from the micro-level information about supply/demand behavior, some macro data relating to economic-demographic structure of rural labor force including migration pattern. In the absence of such an ideally coordinated data set, the most that could be done here was to analyze employment from a primarily demand point of new (using farm management data) and then from the laborers' point of view (using time disposi- tion data) vbile trying in each case to identify propositions deductively consistent with the general indications about the other side of the market and the overall situation. In Part I, farm-level and labor household-level comparisons over .time in employment and wages are made for a number of distinct cases with reference to their specific institutional and production conditions. In Part II, household-wise micro data are analyzed cross-sectionally with a view to identifying the response of labor demand and supply to a few selected economic variables. This approach does not imply that the noneconomic or nonmarket factors are any less important. Instead, one's argument is that it would indirectly indicate whether the usual economic variables are of any significance in explaining the observed variations in employment and wages over time and across regions. 6 PART I: CHANGES IN AGRICULTURAL EMPLOYMENT AND WAGES OVER TIME Two sets of intertemporal comparisons are carried out in this part, differentiated by the kind of data that are used and the view of the labor market they offer. Section A describes the changes in employment and wages from a farm management point of view in four cases, as between a normal agricultural year in the mid-fifties and another in the late sixties. The cases are: Ferozepur district of Punjab, Muzaffarnagar of Uttar Pradesh, Hooghly of West Bengal, and Coimbatore of Tamil Nadu. The first two districts and Darticularly Ferozepur, with a great deal of irrigation, large farms, a high.rate of agricultural growth, and advanced farming techniques, forms an interesting contrast with Hooghly's paddy- jute cultivatioh under high rainfall on very small farms with a moderate rate of adoption of seed-fertilizer innovation. They all are very differ- ent from the relatively dry and agriculturally backward Coimbatore. Section B studies the changes over time in rate of employment, wages, and incomes for labor households in the north Indian statesof Punjab/Haryana and Uttar Pradesh, some parts of which have experienced since the mid- sixties considerable degrees of water-seed-fertilizer-based agricultural growth. I.A. CHANGES IN FARM-LEVEL USE OF LABOR Data Farm management data were collected for random samples of 100-150 farms from 10-15 randomly selected villages in a number of districts, with %farious production conditions and institutional characteristics. As the data were collected for several years initially in the mid-fifties and again towards the end of the sixties, it is possible to compare over -4 7 time for the districts common between the two series. Between the two series there are certain differences in sampling design (in the later series the sample was always stratified so as to represent the actual size distribution of farms in a village, while in the earlier series sometimes an eaual number of farms were selected from each size group) and in the items of data collected or available (like the casual-attached distinction in hired labor which is absent from most of the earlier series). With a bit of careful classification by size, on,: can make the comparisons approximately free of the first difference. The second one makes it necessary to compare the average wage rate for casual and attached farm labor combined. Finally, in selecting the two snecific points of time compared, the main criterion that we used was the normalcy of crop/weather condition. Differences in Agricultural Situations in the Four Cases Of the four districts, Ferozepur is the one that experienced most change in its farm economy over the 13-year period between 1956-57 and 1969-70 which are compared here. Though situated in a low-rainfall area (15-20 iiches in an average year), even in the initial year about 70% of the cultivated area of the farms surveyed was irrigated--largely from canals. Over the;riod there was a large increase in tubewell irrigation mostly through farmers' private investment. In the later year more than 90% of total cropped area of surveyed farms was irrigated, about half of it from tubewells. The crop pattern that traditionally consisted of wheat, gram, and fodder during the major season (rabi, from November to April) and cotton and milletts during the secondary kharif season (July to October) had changed tremendously--more in terms 8 of composition than the frequency of cropping. With the controlled irrigation from tubewells, htere was a large-scale shift to the Mexican variety of wheat, while in the kharif season high-yielding IR-8 paddy and hybrid bajra and maize were now raised as cash crops along with, or as an alternative to, long-staple cotton. Over this period, consistently with the general north Indian trend, the extent of tenant cultivation declined in Ferozepur--from 40% of culti- vated area of surveyed farms, the leased in proportion declined to 12%. Partly with eviction of tenants, consolidation of operational holdings, and partly with irrigation-induced reclamation of land, the average farm size went up from 23 to 29 acres over the period. There were now many more of large commercialized farms than before--with mechanized irrigation and ploughing in most of them and even harvesting-threshing in some. Muzaffarnagar is in the fertile Doab area, between Ganga and Jamuna that constitutes the western zone of the vast and heterogeneous state, Uttar Pradesh. With normally adequate rainfall (45-50 inches) and irriga- tion from an old network of canals, it is part of some of the most densely -populated agrarian settlements in the country. Western Uttar Pradesh is also characterized by a much higher landless proportion of labor households and a much more unequal distribution of land than the rest. The district had a relatively developed agriculture, even in the initial year, with about 75% of planted area irrigated and two active crop seasons. Over the period there was a further increase, with 95% of gross cropped area irrigation in 1968-69, about a third of which came from canals, and another third from tubewells and wells with pumps. Sugarcane, always an important cash crop in the area, increased its share of cropped acreage 9 to 377. Wheat and grem remained the major foodgrains in rabi season; paddy, maize, and fodder, the major kharif crops. in 1968-69 at most a quarter each of paddy and wheat acreage was under high-yielding seed varieties. Increases in use of the seeds and also of fertilizers ner acre were very much less here than in Ferozepur up to the end of the sLcties. As tenancy was made illegal in Uttar Pradesh in the early fifties, very little farm area was reported as leased in even in the initial year. Whether tenants were evicted completely or some continued informmlly is not very clear. What does come out clearly from the two surveys of the district is a shar decline in the fragmentation that was severe earlier. From 85 fragments per 100 acres it came down to 16. Ferozepur farms were less fragmented initially: 29 fragments per 100 acres that was brought down to 10 by a notably vigorous consolidation program. The big decline in fragmentation of Muzaffarnagar farms could be mainly indicative of evicticn of informal tenants who had earlier remained unrecorded and mossibly disguised as attached laborers. The mean size of sample farzis -was about 13 acres in the earlier year and 16 in the later one. Ecoghly district, in a fertile densely settled delta area, experi- enced some changes in its agriculture over the period, rather moderate comared with the Ferozepur t7pe "green revolutio 1"at Quite i=ortant in its own text. While the monsoon rainfall is normally abundant enough for its major traditional crops of paddy and jute (both sown in April-May and harvested around September), groundvater irrigation during the winter and early summer months is the major constraint for raising high-yielding varieties of sumer (boro) paddy, wheat, and zotato with fertilizers. Over the period, though the extent of irrigation had 10 risen from about 15% to 55% of gown area of the surveyed fa-=s in the district, most of it still came from undependable sources, including tanks, ponds, rivers, even canals sometimes. In 1970-71 about a third of the irrigated area of sample far-s used canals and not more than 15% tubevells, which, put together, accounted for only 26% of gross planted area. Consequently, use of fertilizers and high-yielding seeds, though more comon nov than before, was still low compared not only with north India but also with some of the advanced paddy districts like Tanjore. Another important difference from Ferozeour is that the frns in Booghly are very small, the median size being about 1.5 acres in both the years and the mean about 3. It is small even by the standards of high-rainfall, densely settled paddy areas of east and south India. Farms of 3 acres or more, which being above the mean size can be called large here, formed about a third of the stratified sample of farms in 1970-71. But they accounted for 70% of total farm area. Thus, a large proportion of farms towards the lower end of the distribution (Table 2) had become too small to provide a viable means of self-employment for workers of the farm. family. Unlike in the other cases studied here, a considerable pro- portion of adult male workers of farm families in Hooghly was engaged primarily outside their farms. Farms of less than .75 acres (which were below mean minus 75% of standard deviation of the size distribution of sample farms .nd provided as much as 15% of them) could engage in the later year only about a third of their adult male family workers. Farms of .75 to 3 acres, constituting over 5G% of the total, could absorb only about half of their male workers. Judging from the composition of income and emloyment of households in these groups, many were primarily labor households with their younger men working as agricultural laborers. A few were mainly engaged in trading,.traditional services like priest- hood, and there were even some government employees and schoolteachers. The households operating more than 3 acres had most of their adult male workers engaged on the farm, although even in their case the importance of other kinds of employment--mostly professional and administrative- was more than in any of the other districts. This phenomenon of farm family workers being engaged mainly outside the farm--because it is too small for their full-time employment or because the richer or upper-caste households are culturally disinclined towards farm work--is virtually unknown in a case like Ferozepur. There, even in the more recent year of farm prosperity, more than 90% of all adult male workers of farm households were primarily employed on the farm, with less than 10% employed outside. This was uniformly true of all the size groups, including the largest onesof more than 45 acres. The contrast between thes.e two types of cases has two major implications from the viewpoint of rural labor market processes. One implication is in terms of supply behavior,of family labor for use on the farm. While work participation of women in farm households is known to be strongly influenced by cultural or extra-economic differ- ences, it is normally assumed that supply behavior of male labor is 2 more comparable across regions. But it seems that even the suply of male family labor for use on the farm differs Quite a lot. In Hooghly and Coimbatore, unlike in Ferozepur and Muzaffarnagar, there is some indication of Dreference for nonfarm occupations and withdrawal of family 2Incidentally, in both Ferozepur and Hooghly, on-farm employment of female family labor is quite negligible--possibly for very different reasons. However, some amount of female wage labor is used in Hooghly for paddy transplantation. 12 labor from farm work with increase in farm size. A given increase in farm demand for labor over time or across size groups tends to be met more by hiring labor in these two cases rather than by engaging more of family labor. In Ferozepur, on the other hand, cultivation is clearly a preferred occupation for even the richest' of farmers. The difference arises largely from varying cultural norms and historical reasons. The second labor market implication arises from regional differences in composition of employment or occupational pattern of workers of households cultivating small farms. Are they peasants, or are they more of laborers? In Hooghly a lot of these households were marginal cultivators by income or by employment criterion and could as well be defined as agricultural labor households with land. In Ferozepur and Muzaffarnagar, on the other hand, small farms in a comparable low end of the distribution seem to have engaged most of the family workers, which implies that agricultural labor households in these cases are mostly landless. This is corroborated by the successive labor household surveys conducted by NSSO. In 1956-57, for example, of the rural households with agricultural labor as main .occupation, about 90% were without land (i.e., noncultivating) in Punjab/ Haryana and about 70% in western Uttar Pradesh.3 In West Bengal this proportion, though about 60% for the state as a whole, was much less for its central plains region that contains Hooghly. Relative to each other, these proportions do not seem to have changed much later, in 1964-65, for example.3 It seems that the situation in a case like Hooghly allows or enables 3NSS Reoort no. 33 on "Wages, Empliyment, Income, and Indebtedness of Agricultural Labor Households in Rural Areas," Tables 1 in Parts I and II. Rural Labor Enouiry Recort, 1963-65. Table 2.2 13 a primarily labor household to carry on without having to sell or lease out the tiny farm which engages its workers only secondarily. Hooghly, being very near a major city and an industrial belt might have something to do with it, by enabling individual members to migrate seasonally. In an arid region or in one with little nonagricultural employment opportunities (like the hill region of UP), either the whole family migrates, in which case the land is sold or leased out, or the young male workers migrate permanently,.in which case for the remaining ones cultivation becomes more of a primary occupation. Our fourth case, Coimbatore, is a relatively dry and agriculturally backward district in the inland region of Tamil Nadu. The coastal delta region of the state is very different, particularly a district like Tanjore. Tanjore, with highly irrigated, commercialized paddy cultivation intensively using high-yielding seeds and fertilizers, provides a conspicuous south Indian counterpart of the northern "green revolution." Since just one series of farm management survey has been done for this district, we cannotuse the data for an intertemporal comparison, though -later on in Part II, Section A we shall study the data for Tanjore cross- sectionally-as a counterpoint of Coimbatore. Although no major agricultural breakthrough was experienced in Coimbatore over the period from 1956-57 to 1971-72, there were some imorovements in terms of a small rise in irrigated area and slight increase in the share of paddy and irrigated cotton in a cropping pattern that still consisted predominantly We may note that most of these marginal cultivators in Hooghly were owneres rather than tenants, which is not unexpected. Demand for leasing in land and also its supply would normally depend on the feasibility of a viable nexus between the farm generating enough residual for the tenant family's subsistence and utilizing its labor. 14 of dry crops like short-staple cotton and milletts. The most important difference between the two samples of Coimbatore farms surveyed in the two years was that the size distribution was more concentrated in the later year (Table 2). Mean farm size was 6.5 acres earlier and 12 later. Changes in Emoloyment, Real Wage Rate, and Labor's Share in Agricultural Value Added For the four distinct agricultural situations, Table 1 indicates change over time in real wage rate and wage bill per acre, and the increase in wage paid for hired labor relative to farmer's net income arising from an acre of land. For computing real wages, we have used two deflators: one is average wholesale price for the major foodgrain prevailing in the villages covered by the survey, the other is agricul- tural laborers' general cost of living index in the state to which the district belongs. Correspondingly, the real wage is in grain equivalent (item 6) and in rupees of constant purchasing power (item 8). Then we measured.the average share of hired labor in net value of crop production (net of material costs) at the two points of time (item 11), and the incremental share in net value of production per acre (item 12). The share of all farm labor, hired and from the family, is calculated in item 13 by evaluating family labor at the same rate at which hired labor is paid, which is what most of the existing estimates of total labor's share (Rao, 1975; Bell, 1972) do. Actually, it is more likely that hired and family labor are not perfect substitutes. If-the implicit supply price of unpaid family labor is different from the market rate for wage labor, or if there are differences in quality or efficiency, then because the hired proportion usually rises with production. 15 Table 1: Changes in hired farm labor's wage, employment, and share in value added. Weighted averages for farms in four districts. Ferozepur Muzaffarnagar Hooghly Coimbatore 56-57 69-70 56-57 68-69 56-57 70-71 56-57 71-72 1.Ave.wage rate pd (Rs/day) 2.52 5.68 1.13 2.59 1.75 3.46 1.04 2.89 2.Labor(days) hired/nsa 7 18 14 15 28 57 14 42 3.Total labor/ nsa 22 33 51 42 64 114 56 51 4.Wage paid/ nsa(Rs) 17.6 102.2 15.8 38.9 49.0 190.3 14.6 121.4 5.Village pricea of major food- .174 .359 .198 .476 .153 .385 .171 .324 grain(Rs/1b) (wheat) (wheat) (paddy) (paddy) 6.Grain ecuiv.: i: Wage rate 14.5 15.8 5.7 5.4 7.6 6.0 4.1 6.o (lbs/day) (wheat) (wheat) (rice) (rice)c ii: Wage bill (lbs/acre) 101 284 80 82 214 330 57 250 7.Ag lab's cost of living index for the state(1960- 61 = 100) 99 196 100 215 100 206 98 187 8. Wage at 1960-51 prices i: Es/day 2.55 2.89 1.13 1.21 1.75 1.68 1.06 1.55 ii: Es/acre 18 52 16 18 49 92 15 65 9.Farm business income(Rs/acre) 85 427 132 940 161 665 70 350 10.Net value of prod/acre (Rs) (4 + 9) 103 529 148 979 210 855 85 471 11.Hired labor's share in net prod(pc) 18 20 11 4 23 22 18 26 12.Incremental share(pc) of hired labor 20 3 22 27 13.Total labor's (imputed) share in net prod (pc) 53 35 39 11 53 42 68 31 Source: Farm management data collected by Ministry of Food and Agriculture. House- hold-level data from Part II tables and field returns. Notes Item 1: Weighted average for all hired labor, casual and attached, male and female, used for crop production on the sample farms. Total wage payment deflated by total hired labor days (in standardized 8-hour days). Item 2 and 4: Total hired labor days and total wage paid, each deflated by total net sovn area of the sample farms. Item 5: (a) Simple average of wholesale prices prevailing during the year in the villagescovered. The major foodgrain is not necessarily the staple one for the laborers or the rural poor. Their purchased consumption and perquisi;es often Table 1, note continued consist of a good dea2 of gram in north India, jowar or ragi in south India, and coarse varieties of rice in east India,. On the other hand, this procedure would be justified to the extent harvesting wages are paid in terms of the grains harvested. Besides, in terns of proportional change, there has not been much of a difference between these and wheat or standard varieties of rice during the sixties in particular. (b) Wheat price figure for 68-69 in this case being unavailable, the 67-68 one is used here. It is likely to be a close approximation, because in terms of weather and farm management conditions there was little difference between the two years in this district. Item 6: Rice equivalent derived by using 2/ris c (c) Th'ough ragi and jowar are more important in Coimbatore than paddy, their price figures were not available for the later year. Items 7 and 8: Using the weighting diagram derived from NSS 11-12th round data on agricultural laborers' consumption pattern, Labor Bureau constructed the agricultural laborers' cost of living index (ACPI) at 1960-61 prices. District level wage figures from farm samples are deflated in (8) by state level ACPI, a procedure that is at best an approximation in the absence of anything closer. It turns out to be reasonably consistent with the change in real wages as indicated by grain equivalent. (d) Agricultural laborers' consumption price index for western region of UP issued by UP Gov't Directorate of Statistics. The base year in this case is 1957-58. Such regional breakdowns are not available for other states. Item 9: Gross value of production minus the cost of all purchased items (hired labor and material costs including land revenue, rural value of leased-in land, depreciation of equipments and farm construction, and interest on working capital). 17 growth and cropping intensity, this procedure would overestimate labor's share in the earlier period. This is the reason why even in the prime "green revolution" case of Ferozepur total labor's share seems to have dropped sharply (item 14 of Table 1; also Rao, 1975). The share of hired labor alone did rise over the period in this case, though not by a very impressive margin. Changes in labor's share, real wages, and hired proportion of labor in Table 1 relate to the average situation for the sample of farms: average for different farm sizes, sum of the effects of crop-specific technological changes, of changes in crop mix, in pattern cf tenancy, and in size distribution of cultivated area. They measure the composite or overall effect of all these changes on labor's share and real vages. As weighted averages they reflect the dominant type of farms in the area and the major elements of change. While the usual exercise5 measuring the effect on labor's share of the shift from traditional to a high-yielding variety of a crop, leaving all other changes aside, is obviously useful for specific policy evaluation, our purpose here is soAewhat different. It is more to evaluate the combined -effect on real wage and employment of hired labor, arising from the major changes i a situation which are often interlinked though not necessarily working in the same direction. The micro estimate might show the effect of an acre shifted from local to high-yielding wheat variety on employment, wage bill, and production to be the same between two regions or two periods, and still the adoption rates can be very 5Apart from Bell's (1972) exercise for Kosi area of Bihar, and one by Parthasarathy and Prasad (1971) for east and west Godavari districts, several others are available frc/m PEO studies,in particular. Mellor and Lele (1973) usefully gathered their results in one place. 18 different in the two situations owing to other factors like the configuration of relative crop prices. The contrast between the two north Indian districts, despite the similarity of large sized commercial-farms, is interesting. In the canal-irrigated areas of both Punjab and western UP, farmers' investment in tubewells and tractors was boosted by a combination of official policies of price protection, supply of seeds and fertilizers, liberal pumping of finance which were preceded in the fifties by consolidation of holdings somewhat resembling the 18th century English enclosures. But the ensuing change was much bigger in Punjab in terms of labor-using innovations in methods of cultivation. Agriculture in western UP with its sugarcane and rotations of paddy-fodder-wheat had already reached a high level of labor intensity from which a further increase would have involved bigger absolute changes in techniques and crop-mix. Actually, the extent of adoption of high-yielding varieties of wheat, paddy, and bajra was much smaller here than in corresponding areas of Punjab and Haryana. By the end of the sixties, when our FMS-based comparison ends, the water-seed-fertilizer -technology had barely started in Muzaffarnagar. As against more than 80% of wheat and paddy acreage in Ferozepur covered by hyv seeds, constituting nearly 4% of gross cropped area in 69-70, the corresponding figures were 30 and less than 10 in Muzaffarnagar in 68-69. Use of fertilizers per acre was no more than a quarter of what was being used by Ferozepur farmers at the time. One reason could be that though the extent of irrigation was about the same, the more controllab)e type of irrigation from tubewell was much less important in Muzaffarnagar. The other reason came from the tremendously profitable alternative-of sugarcane, and the fact that being a 10-12 month crop it substitutes at least two crops in the usual 19 rotations im the area. Over the period, the relative price of sugarcane and raw sugar increased so much that, even with little rise in its yield comnared with the doubling of wheat and paddy yield from the new varieties, the differential of net income between an acre of sugarcane andL an acre of wheat-paddy rotation became wider. For the Muzaffarnagar farms, sugarcane expanded from 29% to 37% of net sown area over the ueriod. Though sugarcane is a traditionally labor-intensive crop, compared with an alternative paddy-wheat rotation using high-7ielding seeds and fertilizers it is not so. Besides, sugarcane in UP is not very intensively cultivated as commared with Maharashtra which is the other major sugar- cane area in the country. It is in this context of increasing farm incomes caused by price rise for the a.jor commercial crop rather than increased intensity of cultivation that one can understand the low and declining share of hired labor in Mazaffarnagar. As can be seen from the following calculations based on the two series of FMIS data for the district: Sugarcane Wheat Paddy Muzaffarnagar a b c d a b c d a b c d Tnitial veriod' (1954-57) 27.8 1040 580 177 102.5 371 156 78 1L5 373 156 89 Later period (1967-69) 37.7 6810 5880 125 206.9 219 1621 52 252 2024 1662 74 a: yield (tonnes per hectare) b: gross value of production (rupees per hectare) c: farm business income (rupees per hectare) d: mandays of labor used (hired + family) per hectare Sugarcane figures are averages for planted (2/3 rds) and ratoon (1/3 rd), weighted according to their relative acreage. 7A rise from 36% to 42% of gross cropped area, counting sugarcane acreage twice. 8In 1969-70, for example, sugarcane area in UP was six tiMes that in irri- gation-constrained Maharastra, but the physical yield per acre was only about 60% of that in the latter. 20 Farm business income (net of all oaid costs of production) per acre rose (item 9) much more in Muzaffarnagar than it did in Ferozepur, with little more than a fraction of the latter's seed-fertilizer use rate. In both cases, money wage rate had approximately doubled. The intensity of labor use per acre rose substantially along with that in Ferozepur, but not in Muzaffarnagar. Accordingly, the incremental share of hired labor in net value of crop production over the period was 20% in the former and only 3% in the latter case (Table 1). In money as well as in real terms, the average wage rate for hired farm labor was the highest in Ferozepur in both the years. In most of the other cases, rise in money wage over the period barely caught up with the rise in grain price and in agricultural laborers' cost of living index, so that the real wage rate hardly rose at all . It rose in Coimbatore just so that its initially depressed level was now on uar with the others (item 6(i)). In terms of grain equivalent, the wage paid per labor day hired in Ferozepur was more than twice the level in the other cases, both in the initial and in the later year. Over the period there was more variation in real wage bill per acre than in real wage rate per day. That is because the use of hired labor per acre (item 2) changed differently. Both rose considerably in Ferozepur, Hooghly, and Coimbatore--in the last case mainly replacing withdrawn family labor; they did not rise in Muzaffarnagar. As a propor- tion of net value of crop production per acre, payment of wages remained more or less stable over the period in the two cases that experienced a 9Defined approximately as gross production minus cost of all purchased nonlabor inputs, including depreciation, interest, rent, and land revenue. See note to Table 1. 21 large direct emDloyment effect of growth (item 3). Its rise in Coimbatore was because of a big change in the composition of farm labor use, even though the per acre level of labor input did not rise. The incremental share of hired labor in net value of production was about a fifth in the two cases of employment-generating agricultural growth. It is interesting that in these two cases where there was a major increase in labor intensity (hired as well as total) of cultivation, the level of real wage rate did not rise very much. Actually, there was a small rise in Ferozeour and a slight decline in Hooghly (items 6(i) and 8(i)). But these were quite minute relative to the big increases in farm employment that were generated in each case over the period. It seems to indicate that either labor supply rose in the market along with the upward shifts in demand, or there was underemployment among rural workers at the wage rates prevailing in the initial period. The supply of wage labor expanded - due to the formerly self-employed within the region now entering the labor market, due to migration of labor from other regions during the peak seasons, possibly both. From available evidence, this .rather than underemployment is the explanation in a case like Ferozepur, typical of the fast-growing rural areas of Punjab and Haryana. The other explanation for the phenomenon of increasing employment at nearly uncharged real wage rate in terms of "institutional" or non- market determination of wages implies underemployment equilibrium at least during the initial period. The evidence of initially high level of real wage rate and employment per worker (for both cultivating and labor households) in Punjab are just not consistent with the hypothesis of underemployment 22 equilibrium at a wage determined by institutionalized subsistence requirements. The case of Hooghly has certain features more likely to fit into this description: there was considerable underemployment among workers of the farm households, the real wage rate was low and, as we will subsequently discuss, the degree of seasonal fluctuation in farm wages was remarkably low. But even in such a situation where agriculture could not employ--particularly in the earlier year--more than half of the farm households' disposable labor time, there can be other factors holding a wage floor apart from farmers' consideration of the wage-efficiency nexus. The proximity of an urban area, with a huge informal sector that serves to absorb unutilized rural labor seasonally floating back and forth, seems likely to have acted in Hooghly as a mechanism to sustain a certain minimum level of agricultural wage rate. In the remaining two cases, where labor intensity of cultivation did not rise over the period as there was no significant seed-fertilizer breakthrough, there still was some change in the composition of labor use. Large increases in farmers' incomes, generated mainly by price rise, -induced some withdrawal of family labor from the field. In this connection, we might note that employment of women family members on the farm was initially a little more important in these two cases than it ever was in the other two cases. Mich of the decline in use of family labor could be accounted for by the commonly observed tendency to withdraw women and children from the labor force with increases in farm size and income. There is also a tendency for the younger men of prosperous, usually upper 23 caste, households with large farms to prefer nonmanual professions to manual farm work, the at farmers of Punjab/Haryana being striking exceptions to this. The second factor in these two cases that served to raise the hired proportion of labor over time was the increased concentration of cultivated area in relatively large fars (Table 2). A closely related change over 10 the period was in the leased in proportion of farm area, a decline in the once imortant category of small tenant cultivators who tend to be 11 heavier users of family labor than even small owner cultivators. In Coimbatore, in particular, the substitution of hired for family labor was large enough to pull the real wage rate up a bit from its earlier miserable level, and to raise the incremental share of hired labor in value of production. But, unlike in Ferozepur, it was not because of any major shift in production functions raising total employment per acre, rather because of organizational changes in units of production and the farm farilies' labor supply behavior partly in response to that. In conclusion, one can summarize as follows the results of this exercise on aggregative changes in hired labor's share and wage rate in real terms. 10 Net area leased in as percentage of net sown area was 33 in Coimbatore in the earlier year and only 6 in the later year, most of the decline in the small and medium-size groups of farms. For Muzaffarnagar, because of the ban on tenancy, the reported data on this are useless. However, as we argued before, the sharp drop in fragmentation of holdings over the period coupled with the increased concentration of operational holdings strongly suggest that eviction of sinall tenants or withdrawal of area leased out to small operators did take place there. See Par- II, Section A on this point. 2o4 (a) in the two cases where substantial increases in farm emxloyment were generated over the period, the incremental share of hired labor in value added was about a fifth. Four-fifths of the increase in net value of crop production thus vent to the farm households. Tncremental share of wage labor in total farm employment was, however, more in Ferozenur (80/ of additional labor was hired there) than in Hcoghly (5h). The fact that incremental share of hired labor in net production was still about the same in the two cases probably indicates that the incremental share of profit was larger in Ferozenur, the high growth case. (b) Labor's average share in value added and the level of real wage rate did not rise very much in these two cases, even though there were large aggregative increases in per-acre use of hired labor. This could have resulted from a process of rising labor supply along with demand, a process indicated by much of the evidence regarding land distribution, leasing pattern and structure of rural work force. (c) Where large increases in farm income came from price rise for cash crops rather than use of seed-fertilizer-based innovations, hired labor's share in net value of production declined in one case; but it rose in the other where size-distribution of operated area got much more con- centrated comared with the initial period and the tendency to withdraw family labor from the field was sharper. (d) Grain equivalent of wage paid for an average 8-hour day in the four situations tended to a closely similar level in the later year, except in Ferozepur where it was much higher even earlier. Over time, increases in money wage rates lagged behind grain price and cost of living in the cases of moderate to low growth, except where the hired proportion of 25 aggregate farm labor rose a lot substituting the use of family labor. But in the case that experienced a large degree of agricultural growth wage did not lag behind price. Level of living for labor households depends, of course, not just on the real wage rate but also on the rate of employment per laborer and the earner-dependent ratio. About these factors we can learn separately from labor inquiry data (Part I, Section B). (e) The increase over time in farm employment of hired labor depends not just on total employrent generated by a given combination of changes in cropping methods, material inputs, and other techniques. It also depends on how the tenancy arrangements and the size-distribution of operated area changed in the meantime, affecting in the process the hired proportion of farm employment and the composition of rural workers as between wage laborers and own-account workers. Incremental wage share for hired labor tends to be pulled down by increased skewness of the operational size-distribution of land which depresses employment, by increased supply of wage laborers (noncultivating or marginally cultivating) in the market which depresses the wage, by lack of growth of the type based on labor-using seed-fertilizer innovation, and by farmers' windfall income from price rise. The negative effect of any one of these may be offset by opposite change in any of the others. The worst possible situation from labor's point of view can be found in an occasional confluence of all the negative changes, as Mizaffarnagar apparently experienced between mid-fifties and mid-sixties. Labor Use by Farm Size One of the prevalent empirical generalizations about Indian agri- culture is that the small farms are more intensively cultivated, 26 12 that labor use per acre and farm size are negatively correlated: either because the small farms are traditionally more fertile and better irrigated, or because they have a labor cost advantage arising from a larger per- acre availability of presumably less expensive family labor, possibly because of both. The studies to this effect were based mostly on the earlierseries of farm management data relating to the mid-fifties. By the time the later series of surveys were done, some of the initial technological and institional conditions had changed. The size distribution of farms itself had changed-- especially in the areas that experienced agricultural investment and growth over the ceriod. The. exnripnce itself had been.during the sixties mostly a large-f armer chenomenon, despite the inherent size-neutrality of the seed-fertilizer innovation. Lastly, unlike in the earlier period, mechanization of farm operations was now becoming a significant influence for land-labor ratio. The changes in distribution of total cultivated and irrigated land among different farm sizes and the asymmetry of technological choices available to them have im-lications for the pattern of labor use by size, affecting their overall relation in an area. Table 2 traces, for the four districts, changes over time in distribution of land, labor use and its composition. In each case, the size-grouping of farms is done with specific reference to the distribution prevailing there (in the later year). Because of wide differences in land quality and in the distribu- tion of cultivated area, this procedure allows more systematic comparison 12Farm Management Reports, earlier series. Sen (1962, 1975). Rudra (1973). Bharadwaj (1974). 27 across the cases than a common set of absolutely defined ranges. Within each case we used a common grouping for comparing over time. Farms lying between the mean size in a case and the mean minus 75% of standard deviation roughly from the medium-size group. Farms below this range of the dist::ibution are defined as small within the specific regional context, and these are placed in the first group. The third group-- covering farms ranging from the mean size to mean Olus 75% of standard deviation--may be defined as relatively large. The fourth group beyond this range covers the "very large" farms. Increase in irrigation, a basic element of change over the period, was clearly more in the case of larger farms. The earlier irrigation advantage of the smaller farms had almost disappeared by the later year. The irrigated proportion of total planted area was now much more equal across the size groups than earlier. If one considered the quality of irrigation, then the larger farms with most of the tubewells would emerge as clearly better endowed in the later year. The changed distribution of irrigation by size would directly modify the earlier inverse relation between size and labor intensity. Presence of certain other factors might counteract this effect. The smaller farms can and sometimes do manage to buy tubevell water; but the exorbitant rates they have to pay seriously limit their use of it. More often, they might be using the highly labor-intensive methods of traditional irrigation from wells and persian wheels. For an otherwise similar tubewell/pumpset irrigated farm the labor saved on irrigation technology would partly offset the employment generated by a more intensive cropping. However, -when the large farmers' tubewell irrigated cropping pattern is pushing up labor demand, the low 28 productivity use of labor on small farms in well irrigation might get increasingly harder to sustain. Besides, over much of north India canal irrigation was already the dominant mode prior to the spate of more labor-using tubewell irrigation. Another counteracting faccor is mechanization, whichis obviously more relevant in the two north Indian cases than in Hooghly, where even the "large" farms are often too small for tractors. Use of tractors and other machines on larger farms along with seed-fertilizer makes the technological change accompanying any given,increase in irrigation qualitatively different in their case in terms of employment effect. Thus, a varying combination of changes over time in the distributions of cultivated area and irrigation-, in source of irrigation and croo- mix by size, and mechanization would affect in different ways the tradi- tionally observed inverse relation between farm size and employment per acre. in the four situations, which experienced different degrees and combinations of change in these elements over the period, what was the net outcome in the later year in terms of intensity of labor use by farm size was not uniform. Considering the two cases where the growth that took place did increase employTmnt per acre, one notices that the share of larger farms in.. aggregate (sample) use of labor (Table 2, last set of columns) had gone up significantly in Ferozepur, but remained unchanged in Hooghly. The difference reflects a rise in their share of total operated area in one case and its constancy in the other.13 While labor use per acre increased for all the size groups in both cases (seventh set of columns), it was still inversely correlated with farm size. This conclusion stands, and is even strengthened when we go beyond the grouped д � } , " ТвЪ1е 2 I,вЬог иве clдarцcCerlstics 1п Гдпгs 1;rouped Ъу relatlve ыiге,в ftour д1пСгlеtв. � .. �� ��� �- � ���� � � � - Мии-дп,уд » дед pcr funn�- Пгоир shure fп «R1гl�-1гьl,.�ине�� Farm етрl% Index а!' - - - _ .• . .._ - ) S1ze ыrpsa Р_с_ о1' Р_с. оГ TrrlKдted Ccopping Fwni]у иrkrвЪ 1'вm.urkre '1'иСд1 1ab. 111гед 1иЪ_ ипgе pdJ 1 1I ! 11 S (иа цеrед) i'nrms пд aren prul�p�lon lntenaity(рс) per ns дсrа (дпуа} {duys)/пда days/гьsц 1jg� Hit•ед 1'оl.в1 (tired'1'оСц1 lгired Fruu.1'и1:п1, 111rr:d 1''шп.ТоLа1 f Т ТТ Т УТ Т IT I !Т I I1 _1__TI' 1 1Т I TI 1 1i. � - -Т А_� - - -- Feroгepur 1956-57 { I ) гидд 196)-7о { ТТ ) -- _ __ �..__.___ �__._-�_ ._�____ I < 13 ��У'2� 21 1о 7 85 96 1.45 148��- �17 .1В ---150 185� 28 7г 3 Э9 7г BD _- гб 2lдб q76 G96-- ю--5-п-.17 1"S 15 15 1ti 13-29 1+7 41 39 28 67 95 1Эо 142 .12 .10 157 2n7 25 30 5 17 11Э 99 95 1ь'!1 347 763 27 52 4h 26 39 3г 29-4G 19 а1 г9 г7 57 91+ 116 1h2 .о7 _nb 16з г2г 19 33 В 2о 93 112 гВ6 67з 720� 1г1о зз 2г 25 г9 г4 г'r > 1+h 8 а7 г2 38 Тl ВВ 1��о 12б .о3 ,о4 233 цг 17 23 11 15 99 1оо 7о7 1i17 937 1481 35 1о 18 3о г2 26 л11 в1Z�� 1аа lогл 1ио 1оо 68 9z зг5 136 .оу .о t62 2i4 2г Эз 7 18 1оо 1оо 16г 5о3 537 969 1оо 1оо 1оо 1.оо 1on 1оа ��` ~�- Мигпi'fцrnдgar 1956-57 (1) дпд 1968-й9 {ТТ)� -i_.__..�.___-_ - __. -�____.___..�_�.________,,__ .{ � 5 - ^г9 25 -1о 8 В5 9Т �� 1- �� ".1+9 .61 1о5 1о3 72 60 13 6 9© 11о 5В З2о -29 29Э�_ -.9_.-]6--1h 'Э 15 11 7.5-15 44 Э1 38 22 6't у7 а-� 1�� .23 .31 167 123 53 44 1!+ 13 1ао 97 15Э 579 146 493 � 37 !+U 3') 19 25 2Э г5-г4 15 г5 в2 г9 79 93 1�� 1�� .г1 .ц 152 138 49 30 17 13 1о5 1о7 з16 944 г+,5 711? 25 ro г1 гб г7 гб > 21i 12 19 30 41 75 1oU J� .13 .12 2Э7 г1.9 1+l+ !+1 15 19 95 Ч'1 l+7г 1441 63+1 1366 29 7.4 26 52 33 4о л1_1 s1Zea 1ои 1оо 1оо 1оо 74 97 12 � 1 8 -23 ,гз i59 143 51 42 11+ 15 1.оо 1аа 1В6 656 г37 G65 1ои 1оо 1.ао 1оо 1по 1оо - � 1looghlY i95G-5'( (Т) вnд 197t)-71 (i1) �_-�._.. __�__._.-.�-_.__�. _�.�_..-,__._._._.-_..,_.._�.__ , г.l�'_м� xt1'~15 3 2 51 58 137 181 � г.15 3.15(•97) 21 67(г18) '(5 г67 2b 59 1ог gg 1з- 37 23 1nг) 3 4� 1+ -г 7~ s -:�5-3 1+'I 51 27 2В 3о 63 126 162 .В91.11(_61) 43 77(1411 "11 127 31 1г3- 9!+ 10Э 48 11В 73 207 29 33 31 22 ho 3) � 3-5 13 1h 1В 1В 2 59 11у 151 .76 ,57(,ЭВ) 46 9В(149) 65 122 2) 65 9о 97 95 217 25 l+70 1.6 16 1G г1 1В 19 > 5 гг 19 5г 5г 8 52 1г2 1h7 .41 .зг(,г5) 02 12г(162) 61 98 г7 58 1о6 1оо 186 Чо5 65 78г 5г 47 !+Э 55 35 45 I л11 дiхеи 1оо 1ои 1оо цю 14 57 123 153 .6В .65(.3В) 53 91(154) 6!г 11h г0 57 1оо 1оо 7В 179 165 3ho 1оо 10о 1оо зоа 1оо 1ио L � _ _ " __.___�___�__._ _�_ - ,- _- -_ ______-_�__.w_� CaimЪatore 1956-57 (Т) апд 1971-7� (11)' < 4- ь� 5.i 2о 25 5 34 5г 1 11+9 х8г 1. оо .73 63 э5 д1 57 18 � 31 � 1 гю 1 оо 53 г4г 93 169 31 37 � 36 h- 15 6 г.б-1г зб 3г3 Э4 28 49 !,8 �! 134и 166 ~� .4В .3Э 94 55• 62 59 Э6 42 1ое 1аг 99 436 Э2о 460 до 45 4h 27 i6 з2 12-19 8 19 20 23 1В 49 11Э 163 .гг -13 1о3 6о 35 51г 1о 46 97 1оо 168 566 G63 779 15 1г 12 25 го 24 > 19 3 1В 16 44 19 !г3 117 159 .1о .о5 1G6 ЭВ 29 44 1Э 42 Bi 97 ЗВ7 886 ].19Э 1246 14 G В h!, ) 38 А11 ;1zes 1оо 1оо 10U 10D 34 4б 131 163 .4U .1В 85 50 56 51 ]4 42 1оо 1о0 9о 359 1+9В Ьг15 100 100 1о0 10о 1о0 100 Виигсе; Frlmuiy дцСа Гог farмn houдehulds, 1�'M.i. - Nules (ц) 1ielдtive f'впп-вizе groups аге деГlие3 и1СЬ refeг•епсе Со 1he нlze-dlвtriUutlun (Ъ) contlnued иГ swnlrYe lьou3ehul.ds frur» и дlдtгlеС 1п Yhe 1пСег уеиг. 'Рlде весоnд К�гоир сап оГ farm 1'umiцes ии t,n LЬе 1niCiпl уевr. F'етпlе uork рдгtlаlрвtlоп 1� neд5ltgiЪle Ъе еиllед тг;дlWп-вIzед, conLalninK tбosa lying Ъеl.иеел menn ппд тедп minus 75х in я11 the cдsos excepi MuxnfCnrnagar, 1'ог нhic4 ие lдпvе included Гетцlе uorkers и1' д4влдаrд devlutlan. 17доsе bеlои nre �rouped цд вта11 fthe flrдt опе . 1`he о1' Гигт С'arwllieu мith а иеlgьс ог -5 7п vieu of ttte ].ои iutensl4y и1' tLeir acLuul ,� fвrms LeLween та;ип цпд теиn ]us 75� оГ s.d_ nra definг:d па "].ягgе" 1и Lhe empluymeut. t}xlyd group; und Lhoдe above thiв ].lrnit ггrе 1rr'the {пЧУ{� Gt•оир uhlch свn Ъе (с) 7`оtцl dttys о1' Гшп11у 1пЬоr enlrnged 1п tlде Гагшв оГ tfte Кгиир divlded Ъу 1:Ъе numbe.r nf сиllед "very ]цrКс_" For compariug 1пt>ur исе, Ll,a вв,пе qroup3ug is use3 for fmnlly uurherc. '1'Ье parentцetical Г1gш-ев Гоr 11ои�1гlу 1п 1970-71 дге nfl:er leпving , tLe iu3tiul уецr вs ие11, even thougFд Сде clze-d1cLrlЪutLon дiд поС remain uuL tlге lцгgе ❑шпЬеrз и1' oft'-fипп иorlгers, most оГ Chem in smu]). fигтв. 1и Lh1в + иисiгипКед uver the peciod except 1л llooghly. case, the fir;ureз outslde ргп•enLheses впд ргеsитпЪlу tlьаде for Ll+е ear•11ег уегзг � (b) Ac1u1L famfly иид•Кегв рг3тцг11у engaged in L1ье Гвгт. In the inillпl уевг, Lhe вlгои Чt1вt Гдгпь employrnonf, иоиlд иork оиС 1o�i,er head оГ а1.1 fпггп Гumily uorkers. диСи ап thiи иге гподtlу иvвilиЪlе 1п Lha Гиrпь иГ вдиlt еигпегв 1п С1де t'umily. {д) TLe averoge ЧиКе rиte рдlд 1эу encG bruup 1а sxpresseil as и percent.uKe оГ tlге ,гипglе � 1и t}ге ригtlсиlиг 11str1cLs, v1tL иtге excepClon, сдыеs иГ adult uorkers ot' Che вvеrи+,е. '1'he пvег`гцSе иnge ruCes (Гоr гil1 1rlred 1дLor for furm ногК) пrе derl.ved fwnll(ев Ъе1n}; prGnurily enUвg.:d in иurk ou6side L1ie1r ('вnп 1s very rnre. '1'he ftaru tlде иаг5е Ъi11 for и graup of Гш•тп Qet'lдtед Ьу Wде toCU1 nuиLcr of• cunverted � exeepLiun I9 31оаКlдl.у, ЧLег-е, in Cbe smuller Гиггп-аlсе groups, qul6e и!'ev пdult (B-ljour) 1аЪоr дпуs 1дlrед Uy Chern: в иelghted иvегпgе нп);е raLe pufd Ъу 1.}ге д;г'оир. Гги»i1y илгпЬеrя пrе nonagrlculLurпl, цurkers. Rur �Uia dletrict, ие huve given � Lии вeLs иГ fiqureп 1п tire 1«1.rг ycur--6he ипе in pnrcutheaes Lused an СвтПу uurkerв 1»•lmnrtly епgдgгд in the Гnг,п; Lt�e oLЬer Luscd un иц udult enrneru дЭ • N l � г 1 л 30 data of Table 2 and disaggregate. in Part V Section A, using farm- level data and controlling for the effect of other independent variables including intervillage differences, labor use per acre is observed as negatively related to farm size in the two cases. What is more, the "partial" inverse relation is stronger in the later year than it was earlier. In both cases, smaller farms managed to increase their use of at least some material inputs, in spite of the prevailing infrastructura-1 bias against them and the larger ones' much greater access to capital. In Ferozepur in particulax, if only because even the small (< 13 acres) and medium (13-29 acres) farms were absolutely large there, the per-acre use of fertilizers and'improved seeds was no less in their case than on.larger farms. Because of this, and also because mechanization was ad.opted more on the larger farms, the increase in labor use per acre was more for the medium and small farms. 'EvenAn n prime 'green revolution, area,_-Ehesmaller faxms without-their earlier irrigation a d vantage w ere still the more labor-intensive ones in per-acre terns, though their share of ,acreage was now dwindling. About the composition of labor by farm gize, a notable common feature of the two cases of growth is that the input of family labor rose along with that of hired labor in all the size groups. The component of family labor, of course, increased more on the relatively small farms, 14 13 The share of size groups 3 and 4 in net sown area (Table 2, second set of columns) went up from 51 to 65 pc in Ferazepur, and stayed at 40 pc in Hooghly. 14 For Hooghly, we should really compare here the medium with the larger size groups. The reason is that many.of the farms in the "smail" group were having nonfarm occupation or hiring out labor on others' farms. They were really marginal cultivators in the sense that it was not the main source of their income or employment. If we want to leave them out, then we have to consider the medium group (.75 to 3 acres) as "small,' 31 mainly because- of its greater availability in per-acre terms in their case. Cross-sectionalthe number of adult farm workers goes up with farm size, but at a much smaller rate, reducing the per-acre figure sharply for large farms. Besides, with increasing yield prospects, the smaller farms have stronger incentive to mobilize more of their family workers' time for farm work by shifting from less Dreferred1 kinds of employment that they might have been obliged to take up earlier.15 On bo-th these counts, one would expect that the larger farms meet a bigger fraction of any increase in labor requirement from the market. In the ,wo cases of growth, the fraction they met by mobilizing family labor was positive Omlike in the other two cases). But there was a difference. In Ferozemur the hired percentage of increase in labor used on a farm (Ag/AT, from the tenth set of columns, Table 2) was not vastly different across the four relative size groups. For the two larger groups put together, 78% of the incrementa1 labor use was hired, whIIch was about the same as in "small" farms and slightly less than in the medium-size groun. In Hoogbly for the "larger' farms of 3 acres or rmore, 70% of addi- tional em=loyment mer farm was hired. For smaller farms of .75 to 3 acres, on the other hand, only 28% of the additional labor was hired. Earlier, these tiny farms with stagnating yields were too inadequate for absorbing much of the family workers' labr . They had (and still 15 in Mzaffarnagar and Coimbatore, in contrast, on-farm employment for the family vorkers declined even in «small farms. It seems that they might have had to turn more to off-far= emloyment as wage laborers, or mi grate, 16For the· "larger" farms of 29 acres or abore, per-farm input of labor rose from 2.68 to 4.44 man-years (counting 300 8-hour days as one man-year). The input of hired labor rose from 1.37 to 2.74 man-years jer such farm, and inuz of family labor from 1.31 to 1.70 man-years, 32 have) a reserve of family labor, often channeled into low paid or otherwise less preferred occupations, which was now tapped by the yield- raising innovations. The "small" farms of Ferozepur were initially utilized family labor supply better. If we look at the rate of self-employment on the farm for an adult worker, we find that in both cases it went up considerably in all size groups (sixth set of columns). Apparently, part of his labor time that earlier went without work or in doing off-farm work during the tradi- tionally longer lean season was now mobilized for the intensified crop- ping pattern. In these two cases, culturally and agronomically very different from each other, there is little indication of income inelasticity of family labor supply for farm work. It is also clear that in the case of Ferozepur, given the operational distribution of land and the structure of households, the limit for further increases in use of family labor was getting rather narrow. In the later year, an average made family worker was putting in more than 200 8-hour days of farm work; and it was about the same in all size groups which indicates that this probably was close to the limit. Beyond this, given the large size of farms, an increase in adoption of the seed-fertilizer innovation would seem likely to generate bigger employ- ment for hired labor and add to their share of value added--unless farmers accelerate mechanization as they have done more recently. In Hooghly, on the other hand, even in the later year and in most size groups, there was more of "reserve" family labor--both in terms of number of workers in farm households and the number of days for which one was engaged on the farm. In this situation, given the stagnation of nonagricultural employment opportunities in West Bengal, increases in use of the seed- fertilizer innovation would seem likely to continue primarily to promote 33 utilization of family labor in small and medium farms. The employment generated for agricultural laborers might be relatively minor, con- centrated in the group of "very large" farms (above 5 acres). Let us now consider the cases which made little headway over the period in terms of using the new seeds and fertilizers. In Muzaffarnagar, as discussed earlier, the initial intensity of cropping was quite high; the level of labor use per acre was higher than in Ferozepur 69-70. A kind of plateau had already been reached, from which shift on to a higher course would involve technological innovation on a more significant scale than what took place over the period. Certain other changes were of outweighing importance: particularly a sharp decline in the initially heavy fragmentation of operational holdings, associated with an increase in concentration of operated area and in average farm size, eviction of small tenants-at-will as part of these processes_, and large increases in farmers' profit coming from price rise. The combined effect of these, uncompensated by lack of increase in seed-fertilizer adoption rate (which itself is explained partly by the relative changes in crop prices in the case--footnote 6), took the form of less use of family labor on farms. An average farm, though 2h% bigger now, used an unchanged 2.22 man-years (300 man-days each) of labor. Of this, the farm household now supplied 640 as against 72% earlier. Most of the decline in family labor use was in the medium and large farms, possibly reflecting, among other things, a stronger culture-specific bias against field work in response to affluence. In the case of households with small farms, the workers might have been obliged to engage more in 3)4 other kinds of rural work. Unlike in Coimbatore, the number of workers in a small cultivating household did not decline in 1fzaffarnagar over the period. That would have indicated increased rmigration on their part. In Coimbatore the difference between the two years in farm size- distribution was even sharper, the large farMs' (> 12 acres) share in cultivated area having almost doubled (from 36 to 67 pc). Labor use per acre went u in the case of the larger farmers, as they monopolized almost all of whatever irrigation and related development that took place over the period. From the grouped data of Table 2, the initially observable inverse relation between farm size and labor intensity had largely disappeared in the later year. In a dry and -poorly irrigated area, relative to one with an accessible groundwater table and canal irrigation, the smaller farms would be more constrained in improving methods of cultivation and composition of crops. Their per- sistent backwardness in this case is reflected in the lowand declining rate of employment per family worker on own farm as compared with the other cases (6th set of colu:mns, Table 2). An inmediate effect of the increased concentration of operated area and irrigation in the larger farms in Coimbatore was the sharp change in composition of labor use--an increase in its hired proportion from one- fourth to more than three-fourths. Between the two years, employment on a large (12-19 acre) farm had gone up from 1.9 to 2.6 man-years out of which the hired part rose from .6 to 2 .2. On a very large farm in the area (>19 acres) nearly all of the labor was now hired, amounting to about 4 man-years as compared with the earlier level of 1.3 man-years out of a total of 3. 35 Withdrawal of family labor from the field was even more here than in Muzaffarnagar. In these two cases, in contrast with Ferozenur and Hooghly, we find an example of one possible cluster of conditions that tends to reduce the com=onent of farm labor supplied by the cultivating households themselves. It seems reasonable to think that in a situation combining increase in concentration of land and/or irrigation in the relatively large farms, absence of any significant breakthrough in terms of seed and fertilizer use, and the major presence of a cash crop with a rising relative price that offers big profits, one is more likely to observe a process of family labor being withdrawn from the field by larger farzmers or pushed out into wage employment and migration at the lower end of size distribution. Our more definitive result is, of course, the non- -withdrawal (even increase) of family labor from farm work with even a moderately broad-based technological change. With any given ineouality of the operational size distribution of land (which renders the smaller farms' share of household workers larger than their share of cultivated acreage), the more concentrated the water-seed-fertilizer use is among larger farms, the more will be the hired proportion of aggregate labor use in the farm sector of the area. By the same logic, a more uneven distribution of this input package will mean f%iller self-employment for the small farms' family workers otherwise obliged to do more of wage labor or wait around in low- income work. In a situation where the distribution of cultivated area getb still more concentrated and the use of yield-raisIng inputs is also similarly concentrated, as was apparently the case in Coimbatore, the resulting increase in farm use of hired labor (or in its share of value 36 added) need not mean that labor has gained there over time: in comparison with either the earlier period or a situation (like Ferozepur or Hooghly) with a different combination of changes in distribution of land and material inputs. The increase in labor hired on larger farms must be set off against the decline or loss of self-employment for small culti- vator-cum-laborers, and against the labor market implications of formerly self-employed rural workers being turned into wage laborers.- Section I.B looks at this issue from the viewpoint of labor households. Mechanization and Labor Use Pattern in FerozeDur Of the four cases we studied here, mechanization was most prevalent in Ferozepur 1969-70, although even there the process was still in its early phase. For the entire sample of 150 farms covering L.4 thousand planted acres, there were 115 tubewells and 39 tractors, that is, approxi- mately 2.6 and 1 respectively per 100 acres on an average. We noted above a significantly negative inverse relation between farm size and labor use per acre in this case, which is explained by the considerable use of hyv seeds and fertilizers on small farms and possibly also by more mechani- zation on larger farms. The second aspect is elaborated upon here. Most of the available micro studies using Indian data relating to the sixties tend to conclude that (a) tubewell/pumpset irrigation has a large positive employment effect compared with just canal irrigation and a smaller positive one in the case of a switch from well/persian wheel irrigation; (b) tractor ploughing, when instrumental to the adoption of a tighter crop rotation, has mixed effects that on balance can be labor using; but (c) when used for mechanizing other operations, like reaping 37 or threshing, tractors are labor-saving. Different combinations of farm mechanization would thus have a varying outcome, depending also on the change in crop-mix and in the level of use of improved seeds and fertilizers. The data required for estimating the employment effects of tubewells and tractors separately are not available for the particular case. Besides, that would be a whol> new study off the main focus of this one. What we will do here is to look at the overall outcome for labor use arising from clusters of related differences between types of farms distinguished by mechanization level. Accordingly, Table 3 below divides the sample farms into four groups depending on whether they have or do not have tubewell and/or tractor, and then shows the differences in terms of farm size, level and composition of labor use and wage payment. Since the available data are in terms of ownership rather than use of tubewell or tractor, the unavoidable assumption in a comparison based on them is that the farms without tubewells are not more than marginal users of irrigation from this source and those not owning tractors do not use them much either. From Table 3 it can be seen that among the farms without tractors, an average tubewell irrigated (group c) and a conventionally irrigated one (group a ) were very similar in size, in per-acre labor use level and the average rate of wage they paid. Both were, on the average, smaller- sized than the tractorized farms. These two categories of nontractorized farms, which in Ferozepur 69-70 constituted nearly 80% of all farms and 60% of net sown area, differed much less in labor use per acre than in its compo-otion. Employment arising from greater use of hyv seeds and fertilizers that the tubewell irrigation made feasible was not entirely offset by its saving of labor compared with traditional modes Table 3 Ferozepur, 1969-70. Labor use characteristics of farms grouped by degree of mechanization. ehanizationMan-days used/nsa Wage Pc share of Level Pc of Av.frm Fertiliz. Fam. Days of Mndy Wage'bill p shar sample size used/nsa farm a Attachd farm empl/ Hired pd/ *pd/ac grp in aggrt farms (nsa) (Rs) wrkrs lab. fam.wrkra Fain. Casual Att. Total dayb (Rs) fired lab NSA Farms Without Tubewells: Irrigated farms without tubewells a. W/o tractor 110 23.1 44 2.10 .72 180 16.) 7.5 11.0 35.0 5.27 98 32 32 b. With tractor 1t 78.8 41 2.83 )1.17 112 11.0 1.7 5.8 14.6 6.59 69 6 11 Farms With Tubewells: Tubewell-irrigated farms . c. w/o tractor 37 22.6 127 1.78 1.014 238 18.4 8.7 10.4 37.5 5.41 103 29 28 d. With tractor 19 43.8 113 2.115 1.76 244 13.6 8.2 12.5 34.3 6.16 128 33 29 nsa: net sown acres (a) adult male worker primarily engaged on the farm. (b) weighted average for casual and attached labor. CO 39 of lifting water. The nontractorized tubewell farms remained on balance a little more labor-intensive. More important differences (Table 3) were 17 that they had fewer family workers, to which they responded by hiring more attached laborers and also by more intensively employing each family worker. The group (b) of tractorized, canal-irrigated farms were very large- sized and labor-saving (with less than half the level of labor use per acre of the others). For their large size, the number of family workers was inevitably too small; besides, they were apparently not available for intensive employment on the farm as in the other groups. The poorer availability of family labor was partially made up for by attached laborers. As the payment for a permanent farm servant in Ferozepur works out to be a slightly higher daily rate than that for casual labor, the average wage rate -caid by these farms for all labor hired was somewhat higher. But the rate of use of hired labor was so low that the wage bill per acre for this group was only about half of what the tractorized tubewell farm was paying. The average tubevell irrigated far, with tractor (group d) was considerably larger-sized than its tractor-less counteroart, which must be nart of the reason behind its tractorization. its labor use 0er acre was marginally lower than in the case of the other type of farm with or without tubewell. But its use of hired labor per acre--particularly of permanent farm servants--was more than in any other group. Partly because of this, the wage bill per acre for faxms in this group was the highest. So was also their profit rate per acre. 17For Ferozepur farm families, the working members were mostly adult male and crimarily engaged in self-employed cultivation. 40 Quite possibly, the tubewell-irrigated tractor farms were at the time using the machine mostly for ploughing and carting and not much for the other more labor-intensive operations compDared to what they might be doing now or likely to do in the future, given the rapid rate of growth of tractors in the region. Although operationwise tractor use data are not available specifically for this case, studies1 more generally about northern India during the sixties do point to this effect. In that case, the observation--based on a situation like Ferozepur 69-70--that tractorized tubewell farms are only marginally less labor-intensive than nontractorized tubewell fa:Lms will not hold in a different situation with so many privately accumulated tractors that they are commonly used for mechanizing a wide variety of operations. 19 The projective estimates showing a negative direct employment effect of tractor technology assume a composition of multiple uses. This indicates the need for an alternative policy environment (or a corrective one) for limiting,the acquisition of tractors to what would make their use beyond seedbed preparation uneconomic. An agrarian trade union can act (such as in Tanjore district) as a pressure group and an enforcement agent to achieve the same. In the particular situation of Ferozepur in 69-70, the tubewell- irrigated non-tractor farms were the most labor-using in per-acre terms-- enormously more than the tractorized canal-irrigated farms and marginally more than farms having both tubewell and tractor as well as those having neither. In terms of hired labor use alone, however, the tractorized tube- well farms had a slight edge over the bullock farms and a big one over the tractorized canal-irrigated farms. 18 See Lockwood (1972), for example. 19 Raj Krishna (197h); Billings and Singh (1971). 41 I.B. LABOR HOUSEHOLDS: CHANGE OVER TIE A commonly observed fact in rural India is that the suppliers of wage labor constitute a very mixed category rather than a pure one. They variously combine self-employment in tiny farms or traditional vocations with wage-emoloyment in farm or nonfarm work. For a household, the occupations are often even more mixed, a typical example of which is one with its young male workers employed as agricultural laborers while the others are mainly self-employed in the farm or other household enterprise. The wide variability of combination of employment categories for the household as an allocative unit makes the use of some cut-off point unavoidable for data collection and analysis. The absence of an unique criterion for setting the cut-off point accounts partly for the changing definition of a labor household used in successive rural labor surveys. Data Nationwide rural labor surveys were conducted by the NSSO for 1950-51, 56-57, 63-64/64-65, and 70-71. The first two related just to agricultural labor households, and the last two covered rural labor households in general. There are some differences in definitions and ivestigation L2 procedures. However, one can derive some approxirmate comparisons over time, provided one has done the necessary sorting and regrouping in order to separate out the more comparable items and household sets. An agricultural labor household was defined, in the 50-51 inauiry, by whether the major part of working time of its earners (or head of household) was accounted for by wage-employment in crop production, and in the 56-57 inquiry, by whether the major part of household income was derived from farm labor. Employment data in the first case were collected only for persons rezorting some wage-employment each month, which would give an overestimate of the rate of such employment for the bulk of usual laborers, and an underestimate for the labor household as .a whole by leaving out the marginal workers. The second inquiry more correctly recorded employment intensities in different activities for all the usually active members. These two major differences interfere less with the comparison of wage rates and emloyment for adult male agricultural laborers than with similar comparisons for other workers and for the labor household as -a whole. The definitions and sampling design used in the 1963-65 surveys were deliberately kept comparable with the 56-57 one, except for making its coverage more general. For 70-71, the 25th round of NSS collected time disrosition, consumption and wage data for two strata of rural households--those cultivating small farms (defined as the lowest decile of farms in a village) and those not cultivating any land but L3 mainly wage-earning, which together were described as constituting the "weaker sections" of zural people. Any household from either strata deriving its major income from trading, manufacture, mechanized transport, professional and salaried employment, and rent was excluded. Thus, the samnle in effect became quite closely representative of rural labor households, which is also corroborated by. the time disosition pattern of their members. A distinction between casual and attached laborers was maintained in the first two inquiries but was abandoned thereafter, because of doubts about whether the dividing line was readily recognizable and unique across regions. This change affects particularly the comnarability of, wage rates. The rate at which a casual laborer on short-ter= or daily contract gets paid has very different implications for labor supply or demand behavior as compared,to what is earned by an attached laborer, who is often not free to take u any other employment during the period of contract and gets paid in various ways ranging from a monthly wage with pEtrouisites to a fixed crop share with or without credit or use of land. In the comparisons over time presented here, we have as far as possible made adjustments to take care of these major differences among the surveys For agricultural labor households, which form a common group for the first three inquiries, the wage rates, consumption level, and demographic structure are more comparable than employment levels. The employment levels are, however, perfectly comparable between 56-57 and 6L-65. As between 64-65 and 70-71, labor households without (i.e., not cultivating) land are more comparable than those cultivating some land. For, the lowest decile of cultivating households in the 70-71 survey, though mostly having a preconderance of wage employment in work time disposition, may not always properly represent the labor households with land (with wage labor as the main source of income). This would be less of a problem in states like Punjab and Haryana, where about 85% of all rural labor households by income criterion do not cultivate any land than in a state like Uttar Pradesh where about half of them were culti- 20 vating. The weight of cultivating labor households being large in the latter case, any possible underrepresentation in 70-71 will affect 21 their comarability with 64-65. But, if we separate out the more developed western zone of Uttar Pradesh from the rest of the state, then the problem becomes manageable, because like Punjab/17aryana rural labor 20 In. Punjab-Haryana about 15% of all rural households were cultivating some land in 64-65. In 70-71 also, the small (lowest decile) cultiva:c? stratum for the comparable area constituted 15% of the two strata of households put together. It seems more likely, from Census working force data as well as NSS landholding data, that the relative weight of culi ig labor households became even smaller in the region over the sixties. ever, their weight being small anyway, the issue of representativeness matters less here. In UP, on the other hand, the weight was large, 5-. of rural labor households cultivating some land in 64-65. If the small cultivator stratum in TO-71 did renresent this group adequately, then -eir proportion would appear to have declined to 36%. Part of the decline might be real. Part of it might be due to underrepresentation of prim=riy labor households in the 9th or even 8th decile of cultivating househol4s. 21 If cultivating labor households were underrepresented in the 25th rc*'4, then compared with 1964-65 it would tend to overestimate the rate of wage employment in their case and possibly also underestimate the rate Cf total emloyment. The reason is that the self-emloyment component goes up very much with rise in far size in the lowest deciles. The wage employment tends to decline, and total employment tends to go up. 22 households in this zcne are overwhelmingly noncultivating or landless. For the rest of the state, at least the landless group is comparable between 70-71 and earlier years. For Punjab-Haryana and western UP, the cultivating labor household group is also comparable nearly as well. 23 Table 4 presents, for the Punjab region and for Uttar Pradesh, some of the changes over 1950-51, 56-57, and 64-65 for agricultural labor house- holds and for the men usually working as agricultural laborers. Table 5 constructs a similar comparison for rural labor households in general between 64-65 and 70-71, which captures the initial phase of north Indian "green revolution." Tables 6 and 7 give, as a proxy for inter- temporal change, cross-sectional comparisons between western UP and the less developed rest of the state. Another difference, as already mentioned, is that in the western zone labor households are mostly landless. For '1970-71, the zonal comparison within UP is based in Table 6 on primary NSS data processed specifically for the purpose. For 56-57, it is based in Table 7 on published data in processed form. Because of this, and also the difference in coverage, we have kept them apart instead of combining for an interzonal comarison of iiterteioral change. The hases of Change First, we observe a clear pattern of change in the estimated number 22 The western zone of UP covers roughly 18-20 districts (out of the state total 54) which are distinguished by their better initial endowment, faster recent growth in agriculture, and more commercialized labor relations com- pared with the rest of UP. About 801o of agricultural labor households here were landless in 50-51 and in 56-57. In 70-71, on the basis of the 25th round NSS data for two "weaker" strata of rural households, the landless proportion works out for 74 in this zone. 23 Successive reorganization of states in the region makes certain adjust- ments necessary for comparisons over time. These are detailed in notes to Tables 5 and 6. 46 of labor households (item 1 of Tables 4 and 5); a pattern consistent with other kinds of evidence about change in the macro structure of rural working force over the period. The large increase in the number of agri- cultural labor households between 1950-51 and 56-57 must be to a consider- able extent a reflection of the initial eviction of small tenants-at-vill, which took place over much of north India mostly during the early fifties. The number was stable for a while in the following period until a whole new set of factors were set in motion or reinforced by "green revolution" in northwest India. Between 64-65 and 70-71, the estimated number of rural labor households without land went up by nearly 50% in Punjab and Haryana combined, and by about 25% in Uttar Pradesh. The Census data. indicate a similar large increase ,in the ratio of primarily wage-employed rural workers to the primarily self-employed ones over the sixties. * NSS landholdings data for 53-54 (8th round).and 70-71 (26th round) provide a corroborative evidence for the entire period, covering both the earlier phase of tenant eviction and the later one with the land poor unable to lease in any more having to depend increasingly on wage employ- ment and even lease out. In 1970-71, the proportion of rural households not owning any land in Punjab-Haryana was only 9%; but the proportion not cultivating any was as high as 54%. In 1953-54, for a roughly comparable 24 area, the respective percentages were about 38 and 32. In Uttar Pradesh the percentage of rural households not owning land was 9.4 in 1953-54 and 4.6 in 1970-71, while the percentage not cultivating went up from 5 to 24. 24 In erstwhile Punjab (including present Haryana),Delhi, and Himachal Pradesh , the percentage not owning land in 1953-54 was 35.5 and those not operating any 27. In PEPSU, these percentages were 42 and 37. 47 .-Table 4: Employment, wages, level of living for agricultural labor households, 1950- 51, 1956-57, and 196L-65 The Punjab Areaa Uttar Pradesh 50-51 56-57 64-65 50-51 56-57 64-65 l.Est. no (000) of 282 371 382 1580 1965 1730 ag. lab hshld (475)b (2046)b i: hshld size 4.9 5.2 5.5 4.2 4.6 4.6 2.No. of wage (M) 1.30 1.30 1.37 1.20 1.19 1.23 earners/hshld (W) .30 .28 .29 .20 .60 .55 3.Rate of empl. (days/wk) per (ag. lab) 3.56 4.40 5.42 4.90 3.35 3.63 male lab from (all sources) 5.44 5.23 5,99 6.17 5.00 5.33 4.Ag.wage enpl. (M) 4.63 5.70 7.40 5.81 3.99 4.46 (days/wk)/hshld (W+C) .46 1.46 1.60 .52 1.56 1.42 5.Ave.wage rate for male ag. lab . 1.84 1.98 1.18 .92 (Rs/day) 1.82e 2.13d .80e 1.10d .6.Ave.annual earning (Rs)/hshld from (M) 539 815 166 255 ag. labor (M+W+C) 418 554 803c 380 250 396c 7.Per capita monthly cons.exp. (Rs) 13.50 17.60 24.4oc 10.85 11.20 17.75c 8.Pc change in cost of living index -2 +34 +50 (ACPI) +25c +33c 9.Pc change in real terms i: wage rate for male ag. lab +10 -17 -13 ii: wage earning/hshld (all). {+34 +20 +25 from ag. labor (M) +34 +17 1+4 iii: per capita cons.exp. +32 +14c +26c Notes (a) To take account of state reorganizations, the comparable region covers in 1950-51 erstwhile Punjab, PEPSU, Delhi, Himachal Pradesh; in 1956-57 and 1964-65 it covers Punjab-Earyana, Delhi, and Himachal Pradesh. All the figures (both aggregates and averages) are adjusted accordingly. (b) The figures within parentheses for 1964-65 give the number of all rural labor households. (c) Consumption expenditure figures in item 7 and pc change in cost of living index in item 8 are with reference to 1963-64 rather than 19614-65. Correspondingly, the percentage change in item 9(iii) is between 1956-57 and 1963-64, (d} In item 5. While comparable wage rates for 50-51 and 56-57 relate to casual agricultural labor, averaged over the various,agricultural operations, for 1964-65 the available figure relates to casual and attached agricultural labor taken together, (e) Estimated wage rate for casual and attached agricultural labor, for comparison with the 64-65 figure. Bardhan (P.1970) derived for Punjab a daily wage figure for attached labor in 56-57--from the attached labor households' income from agricultural wage employment and their amount of such employment. It was then combined with the casual wage rate by using the respective amounts of agricultural wage employment for the two types of households as weights. The figure for UP is estimated here by using the same method. Item 8: Percentage change over each of two intervals. Based on Agricultural Laborers' Consumer (general) Price Index issued by the Labor Bureau. The old series was with 1950-51 as base and the new series with 1960-61 as base year. For UP the ACPI were not issued by the Bureau prior to 1965-66. We have used Jose's (1974) estimate for the last two years. 48 Tables : Elploynent, wages, level of living for rural labor households. 1964-65 and 1970-71. Punjab and Haryana Uttar Pradesh 66-65 70-71 64-65 70-71 1. Est.no.(000) of noncultivating lab hshld 404 597 986 1203 2. No.of working age mena per lab hshld 1.48 1.50 1.30 1.13 i: Av. hshld sizeb 3.51 4.91 3.06 4.00 3. Rate of empl. (days/wk) per f ag. lab. 4.07 3.12 2.88 2,90 man from tall sources 5.87 5.34 5.31 5.66 4. Ag. empl. on - wages/hshld Sm 6.13 4.77 3.80 3.28 (days/wk) v + c 1.26 .78 1.22 1.09 5. Av. wage rate for male ag. lab (Rs/day) 2.13 4.47 1.11 2.19 6. Av. wage earning per hshld (Rs/wk) from male ag. lab 13.14 21.32 4.22 7.18 7. Per capita monthly 25.17C 48.44 17.83c 33.61 cons. exp. (Rs) 8. Pc change in (ACPI) 64-65 to 70-71 +47 +22 cost of living L63-64 to 70-71 +59 +43, +65 9. Pc change in real terms in i: wage rate for male ag. lab. +63 +75 ii: wage earning/ hshld from +15 +49 ag. lab (m) iii: per capita cons. +33c +46c, +24d Sources: For 1964-65, Report on Rural Labor Enquiry. For 1970-71, primary data from 25th round of NSS. Notes While item 1 gives the estimated total number of noncultivating labor households, the other items relate to all rural labor households (cultivating or not) in the respective samples. Item 2(a): In 1964-65, men from 15 to 64 years. In 1970-71, men from 15 to 60 years. (b): Excluding very young children (below 5) and very old persons (above 60 in 70-71, above 64 in 64-65). Item 3: Per working age man belonging to rural labor households. For 1964-65, available tabulation is in terms of employment per usually active men. For comparison with 1970-71, we have converted them here into figures per working age man by using the age-sex-occupational distribution of household members (in Tables 2.4, 3.4, 8.5, and 9.5 of RLE Report). Item 5: For both the years, wage rates, are averages for casual and attached agricultural labor. Item 7: (c) relates to 1963-64. Item 9(iii): (c) pc change as compared with 1963-64. Item 8: Based on the new series (with 1960-61 as base year) of Labor Bureau's index numbers of ar.ricultural laborers' consumption (general) price (ACPI). Items 8 and 9(iii): (d) for UP, an alternative set of agricultural laborers' cost of living index (1957-58 as base) is available, constructed by UP Govt Directorate of Economics and Statistics and published in their Quarterly Bulletin of Statistics. Between 6L-65 and 70-71, the pc change computed from this index is about the same as that computed from Labor Bureau's ACPI. But, the change between 63-611 and 70-71, which is us'ed for deflating con- sumption data over this period, works out as larger from UP DFS index than 49 This increased rural proletarianization in the region is closely linked up with a process of technological change taking place within imperfect land (lease) and credit markets. For the landpoor, unavailability of the traditional means of securing viable self-employment through leasing in additional land, coupled with their disadvantages in the credit market, served as a push factor turning them out into the market as laborers. At the same time, the demand for labor generated by the growing agriculture of the region must have acted as a pull for them as well as for migrant rural labor from depressed areas. From Tables 4 and 5 (item 9), our second observation is that in the period following 64-65, there was a considerable rise in average real wage rate for agricultural labor in both states, in contrast with its decline in the preceding period. Before 64-65, money wage rates lagged behind the rising cost of living for agricultural laborers in the area. Now they were moving ahead. Intensive cultivation centered on use of seed- water-fertilizer raised not only the aggregate annual level of labor demand, but also accentuated the seasonal peaks for employment and activated some of the lean months interspersing the traditional rotations. In the subsequent disuession of seasonality in Ferozepur in contrast with Hooghly, we find that barring the prior presence of substantial underemployment the wage rate would not be insensitive to seasonal peaks in labor demand. In Ferozepur between 56-57 and 69-70, both the increased level of annual labor use and the heightened seasonal peaks contributed to the rise in average farm wage rates. While the real farm wage rate increased in both regions between 64-65 and 70-71, it is quite remarkable that there was no increase in the rate of agricultural wage-employm-nt per laborer (item 3, Table 5). In Punjab it had even declined a bit. This obviously dces not mean that for each individual laborer the rate of farm wage empoloyment failed to rise. It might have risen' for some and declined for others. The labor market is too fragmnted and imperfect to ensure an even spread of either an existing or any additional amount of employrent. All that the data indicate here is that there was a large increase in the number of rural wage laborers and the increase in aggregate farm employment of hired labor was not large enough to raise the level of employment zer laborer. * The phenomenon of rising wage rate along with unchanged or declining average employrmnt rate is exnlained by (a) the sharp rise in supply of laborers in the market and (b) the changes in seasonal pattern of em=loy- ment associated with the increased cropping intensity. Time-bound and .heavier peak operations might be involving a few weeks of concentrated employment for a much larger number of laborers. Aggregate demand in the rest of the year, though on a higher plane compared with the earlier situation, still might not have been high enough to hold average employment per-laborer at the level that prevailed earlier for a much smaller size of rural wage laborers. In other words, if the urward shift in seasonal demand curve for hired labor is asymnetrical (i.e., wider for the peak seasons), and if wage is*not insensitive to that, then the annual average of seasonal wage rates may go up even if the number of laborers has risen enough to meet the higher level of 'demand. Change in wage income, being a product of changes in both vage rate and employment oer laborer (or per labor household), indicates better than either component his (its) absolute gain directly from. the given agricultural growth. Item 9(ii) of Tables h and 5 shows the changes over time in real 51 earnings from agricultural wage employment per labor household. In Punjab, during the period of agricultural growth following 64-65 the extent of its increase was no larger than what it was over the preceding period. This was because while the real wage rate was rising more in the recent period, the rate of farm wage-employment per labor household was not, thus almost reversing the earlier combination. A major reason for the sharp increase in the number of labor households-- the market-dependent proportion of rural -v;rkers--was the shift in organiza- tion of production and distribution of operated area (tenants evicted, peasants turned laborers, capitalist farmers rising). Another reason is migration of labor from the relatively depressed to the growing area. Though its effect was also to pull down the average employment per laborer, it has an obviously different distributive implication. Rising immigration (or declining outmigration) must be regarded as a positive element in counting benefits of regional development for rural labor in general, along with any rise in wage rate and employment per laborer and consumption per capita. On migration of rural labor in response to interregional disparities in agricultural growth, there is little direct data. ovever, some idea of the positive distributive effect of regional growth in the form of absorbing labor inflow (or checking an outflow), and the negative effect of stagnation in the form of pushing labor out, can be derived indirectly from differences in the number of earners and size of an average labor house- hold. An imoortant observation from the data in Tables 4-7 is that there 25 5 With available data, it is not easy to separate out the part of an ob- served increase in wage-laborers among all rural workers that is on account of migration from less developed regions from the part that is due to farmer tenants or potentially viable ones being pushed out into the market. 52 is a certain positive association--both over time and across regions-- between level of development and the average size of labor household and thE number of its male workers in particular. Both have increased since 1964-65; both have been larger in the more progressive Punjab-Haryana area than in Uttar Pradesh, and within UP in 1970-71 (Table 6) larger in the faster-growing western zone compared with the rest. Considering the change in per capita real consumption as an indicator of labor households? level of living, Tables 4 and 5 (item 9(iii)) show that the increase during 1963-64 to 70-71 was more than during the preceding 7-year period in Punjab. Punjab laborers had a similar rise in consumption in the early fifties too. But then it was relative to the near starvation level of the post-war, post-Partition period; and it came simply because after prolonged shortage, several good monsoons had by the mid-fifties brought 'foodgrain prices down from their abnormal heights. Thereafter, the cost of living for agricultural laborers (item 8) kept on rising, particularly in the last period when it ate up half of the rise in money wages and per capita consumption expenditure. Still, the residual increase during 63-64 to 70-71 was more than over the preceding 7 years, indicating on an average a moderate absolute gain for labor from Punjab's growth. The question of relative gain or distribution between farmers and laborers, and also that of distribution among the labor households tiemselves, are quite different. Existing studies indicate unequal distribution of the gains. For Uttar Pradesh, the picture regarding change in laborers' average level of living is less definite. The increase in per capita real consumD- tion level of labor households over the 7-year period following 63-64 appears to be larger than in the preceding period if one uses one cost- 53 26 of-living index, and nearly equal if one uses another. All one can say is that a mild pace of increase was at least maintained over the 1L-year period since 56-57. In Punjab, on the other hand, a slower increase in the first half got stepped up later. The difference is consistent with the larger agricultural growth, the faster pace of adoption of seed-fertilizer innovation in Funjab-Haryana than in Uttar Pradesh since the mid-sixties. In almost all the years we studied here (except the earliest, 50-51), agriculture provided absolutely more wage-employment and earning per laborer and per. labor household in the Punjab region than it did in UP. In 56-57, 64-65, and 70-71, male farm wage-employment per household (item 4, Tables L and 5) was 45-65 p.c. larger in Punjab-Haryana. If we take all (m + v + c) farm wage-employment p.er household, then in 64-65 it was nearly 50% larger, and in both 56-57 and 70-71 about 30% larger. This source also constituted a largpr proport.ion of their total emoloryment in the 'unjab region. If.we look at total employment per laborer (item 3), the difference between Punjab and UP is much less than in the agricultural wage-employnent. Self-employment has always been much smaller for the labor households in Punjab-Harana and western Uttar Pradesh, where they are predominantly landless, than in the other zones of UP. Their greater dependence on the market for employment leaves them more vulnerable to farm mechanization. 26 Agricultural laborers' consumer price (general) index prepared by the Labor Ministry registers a 43% increase for UP between 63-64 and 70-71. An alternative index (see note (d) of Table 6), with a different base year, prepared by Gov't of UP shows a 65% increase. It may also be noted that for the period between 64-65 and 70-71, over which wage comparisons are made, the two sets of index numbers yield about the same increase, thus giving a unique measure of real wage change in the state over the period. Price change from 63-64 to 64-65 seems to be for some reason interpreted differently by the two sets of index numbers. From Table 6, which compares the agriculturally prosperous and commercialized western zone of Uttar Pradeshwith the rest of the state, one observes a consistent pattern of differences quite similar to those between the Punjab region and UP as a whole. Wage rate for agricultural labor (in terms of -both money and grain equivalent--items 7 and 8) was significantly higher in western UP, so was the extent of male agricultural wage-employment per labor household (item 5). Thus, the real wage earning per household from male farm labor was about twice as large there. Almost repeating the contrast with the Punjab region observed in Table ., Table shows that the rate of farm wage-employment oer laborer is not much different in western UP, but the number of male wage earners living in a labor household is larger here than in the migration-proned -depressed rest of the state. The level of employment and wage earning per household were thus much larger in the western zone. Employment of women and children as agricultural laborers, from Tables 5-7, is relatively more important. in the rest of UP than in Punjab-Haryana and its own western zone. It is partly a result of paddy cultivation, with its female labor demand for transplantation, being more important in eastern and central zones. Partly, the phenomenon also reflects the more acute poverty in those areas, because in the cultural context of north India women and children working as wage laborers is an act of desperation. When we add up farm wage-employment for all members of a household, the level in rest of UP is higher than in the more developed western zone. Bu± wage rates for female and child labor being even lower than the low level for male labor in the rest of UP, labor households remain poorer than in the western zone. Table6: Rural Labor Households: Noncultivating (NCL) and Cultivating (CL). Vest UP and rest of UP. 1970-71. Table 7: Agricultural Labor Households in West UP andirest of UP. 1956-57. West* UP Rest of UP NCL CL NCL+CL NCL CL NCL+CL West UP Rest of UPa 1. Est.no. (000) .1. Pc lab. hshld of lab hshld 639 225 864 56h 465 1029 not cultivating 2. No.adult mena any land 70 33 per hshld 1.29 1.21 1.27 .98 1.06 1.02 2. Adult ag. lab. m 1.26 1.17 3 Whld sizeb per hshld Lw .22 .78 (excl.old & babies) 4.31 4.1 4.34 3.115 4.05 3.72 1: Av.bshld size 14.33 4.61 14. Rate of empl. 3. Rate of empl. (days/wk) ag.lab. 3.12 2.06 2.85 3.35 2.49 2.95 (days/wk) per ag. lab. 3.5 3.2 per mana from Ltotal 5.66 6.13 5.77 5.27 5.84 5.51, male lab. Ltotal 5.1 5.0 4. Ag. wage empl. 5. Ag. wage empl. (days/wk) 5 ma 4.02 2.A8 3.62 3.29 2.65 3.00 (m + w) per hshld per hshld tw + C .111 .1114 .1411 1.86 1.35 1.63 (days/wk) 4.9 5.2 6. Wage rate 5. Av. daily wage (R]s.) (Rs/day)- for ag. lab. 2.55 2.70 2.58 1.80 1.77 1.79 for casual ag. lab. 1.04 .82 male nonag. lab. 3.06 2.37 2.93 3.39 2.28 2.96 6. Av. weekly hshld T. Grain equiv. (1bs)c earning (Rs.) from I: ag. wage rate 6.5 6.9 6.6 3.9 3.8 3.9 ag. lab.(m + w + c) '.65 3.89 11: hsWd earning (wkly) 7. Per capital daily from male ag. lab. 26 17 24 13 10 12 cons.(ol.) of cereal 19.5 15.8 Source: Householdwise data from the 25th round of NSS. Sources: Tables by zones from NSS Report no. 33 on "Wages, Employment, Income, and Indebtedness Notes of Agricultural Labor Households in Rural Arcos"; (a) Working age men of 15-60 years. and NSS Report no. 146 on "Consumer Expenditure (b) Working age men and women (15-60 years) and children between 5 and 15 years. of Agricultural Labor Households in Rural Areas." Children below 5 years and old people above 60 are left out of these Notes computations as they are mostly out of the labor force. (a) The figures for "rest of UP" are worked out as weighted (c) Grain equivalents are derived by using the average cereal price deflators averages for ALE zones 1, 2, and I of the state, relevant for agricultural laborers in the two parts of UP. The deflators covering its southern, central, eastern are derived as follows. Average rural retail prices (from Oct. 1970 to regions. The weights are derived from the distributiot. June 1971) for the staple cereals--rice, wheat, and grairi--prevailing of the sample households across these zones. in west UP and r.st of UP (simple average of their prices in east and (b) Including cereal content of perquisites. central zones) are used. The weights of these three cereals in the consumption of agricultural laborers differ between the western zone and the rest of UP. Applying the relevant relative weights of the three cereals for agricultural laborers to their zonal prices, we get the weighted average rural retail prices of cereal (wheat-graln-rice combined) for agricultural laborers in 1970-71: Rs 0.39 and 0.6 per lb. in 'West UP and rest of UP, respectively. *According to the NSS zoning, western Uttar Pradesh includes 18 districts: Saharanpur, Muzaffarnagar, Bijnor, Meeirut, Moradabal,Bilandshahr, Rampur, Bareilly, Pillibhit, Shahjahanpur, Badiiuu, Aligarh, Mathura, Etah, Farukhtbad, Mainpuri, Etawah, Agra. For 56-57, ALE zone 3 has almost the sane coverage with marginal differences. 56 In 1956-57 (Table 7), the differences for labor households between the western zone and others--in terms of per head employment, wages and earnings, and the number of laborers--vere similar in kind but much smaller in extent compared to 1970-71. In the meantime not only had there been more agricultural growth in western UP, but the structure of rural work force had also changed more than in the rest of the state. Over the sixties, according to Census, the number of agricultural laborers more than doubled, in vestern UP as against a less than 50% rise for the rest of the state. In the 71 Cernsus, western UP had 23.5% of the state's agricultural laborers (as against 16.5% in 61 Census). In 70-71 (from 25th round) western UP accounted for 53% of the state's rural labor house- holds without land; in 50-51 (ALE it had 46% of the agricultural labor households without land. The differences between western U? and the rest of the state were less in 1956-57 than what they became in 70-71. The difference in wage rate was more now, reflecting the growth disparity. So were the differences in average.household size and the number of male laborers per household, indicating a recent increase in either outmigration of rural labor from the rest of UP or imigration into krest UP. Conclusion From the above comparisons of labor households over periods of dif- ferent growth experience as well as across regions with varying levels of development, the broad tendencies that enrge quite consistently can be stated as follows. Agricultural growth in the late sixties did not, at least in the north Indian context, result in a higher rate of agricultural wage-employment er laborer. ncreasing egate employment generated by irrigation and technical change was accompanied by sharply increasing numbers of rural wage laborers. The latter arose from changes in operational distribution of land, in the nature and extent of leasing, and rural-to-rural immigration of labor, quite apart from the natural increase. Manifestation of labor's absolute gain from growth was rather in the form of (i) a higher farm wage rate and hence more wage income per laborer, (ii) having to migrate less in search of work, as indicated by a larger number of laborers per house- hold, and (iii) a higher per capita consumption-level. The absolute gain might have, however, been too small for even maintaining a stable distribu- tive share for landless and landpoor laborers relative to farmers. In the backward areas, or over the periods of low growth, even this small absolute gain was often nonexistent. 58 APPENDIX I SEASONAL RESPONSE OF WAGE RATE TO EMPLOYMENT An increased intensity of cultivation based on use of irrigation, better seeds, and fertilizers, unless counteracted too much by mechaniza- tion, generates not only more employment but also modifies its seasonal distribution. Taking the two different cases of labor-using growth discussed earlier, we find certain remarkable differences in the seasonal pattern of farm employment and in the corresponding behavior of wage rates. Although the seasonal data available to me are neither very ade- 27 quate nor quite uniform as between the two cases, ' they still offer a vivid empirical illustration of the different kinds of seasonal wage response to employment variations that one might expect in a high- employment situation as against one with underemployment.* Table 8 and Chart 1 compare between Ferozepur and'Kooghly the seasonal patterns of employment for an average family farm worker (and an attached laborer, in Ferozepur where they are important) and of daily wage rates for casual labor. In the absence of matching time disposition data for wage laborers in these cases, we regard the differ- ence in employment levels for farm workers belonging to cultivating households as indicating in ordinal terms a similar difference in the degree of employment or underemployment for rural workers in general. 27 For Hooghly,70-71,I had access to 10-day periodwise data about wages paid by the farms surveyed and their employment of hired and family labor. For Ferozepur, seasonal breakdown for the sample farms was available regarding employment but not wages. The monthly averages of daily wage rates used in this case are for casual field labor prevailing in the villages covered by the survey and collected from the village records as part of background data. .59 That is, if the workers of farm households in one region have a higher level of employment both during the peak season and for the agricultural year as a whole than those in another, then we infer that the degree of employment of rural workers in general is also higher in the former or that there is more underemployment in the latter. Hooghly and Ferozepur belong to different tyes in terms of pressure of population on land and the level of employment for rural workers, even during the more recent veriod when the gap might have been less wide than it was initially. The comparison between these two cases of employment generating growth shows how the response of wage rate to a given extent of increase in farm labor demand or employment is determined by the level of employment per laborer prevailing in an area, how close it is to full utilization. The behavior of farm wage rate over the years and across lean and peak seasons within a year was responsive to employment in one case and insensitive in the other. A plausible way of looking at the phenomenon would be in terms of the hypothesis that the amplitude of seasonal variation in agricultural wage rate is an ordinal index of 28 regional difference in the degree of underemployment of rural labor force. .n the-case of Ferozepur, from Table 8 , the monthly average of daily wage rates 'or casual male field labor remains at a 2teady level for 9 months of the year and then rises sharply in the major peak season (April and May, when rabi crops are harvested and the soil prepared for kharif crops). art 1 shows-that this was the nattern in both the initial and the more recent year, except for two differences: in the later year the seasonal rise was sh=rer and of slightly longer duration, and there apoeared a 23 See Raj (1959). 60 Table 8 Seasonal variation in man-days of own-account farm employment per adult family farm worker. Hooghly 1970-71 Ferozepur 1969-70 (primarily cultivating hshlds) (a) (b) (a) (a) (c) (d) (e) June 9 10.5 3.05 1.78 July 20 23 5.32 14 16 3.15 2.62 Aug. 21 28 5.28 16 17.5 3.14 2.79 Sept. 19 25 5.65 9.5 10.5 3.34 1.98 Oct. 23 28 6.41 10.5 12 3.37 1.91 Nov. 21 28 6.33 13 14.5 3.41 2.16 Dec. 21 25 5.48 16.5 17.5 3.62 3.12 Jan. 18 25 5.23 12.5 14 3.28 1.36 Feb. 18 24 5.71 14 14.5 3.21 1.47 Mar. 18 27 5.87 13 14.5 3.35 1.26 Apr. 24 30 8.02 10.5 11.5 3.17 .98 May 25 34 9.71 16 17 3.21 2.25 June 19 24 8.11 Year 245 321 6.43 155 168 3.30 23,68 Notes (a) Man-days of employment on farm (connected with production of crops and milk, including the labor given in exchange or gratis) per head of family farm worker (almost entirely adult male in both cases). Self-employment in dairy produc- tion was negligible in Hooghly. (b) Same for a permanent farm servant. (c) Total agricultural employment (own-account -lus hired out on others' farms) per family farm worker. Hiring out was negli- gible in Ferozepur. (d) Monthly average of daily wage rate for casual farm labor. Ruoees per day. For Ferozepur, the figures are monthly averages of operationwise wage rates for male field labor prevailing in the villages covered in the sample (published in FMS Report). For Hooghly, the figures are average daily wage rates paid for casual labor hired in(on very short term, including daily, contract) by the 116 primarily cultivating households in the sample (leaving out the 34 for which cultivation was a secondary occupation by employment and/or income criterion). The monthly average in this case was derived from the primary data, by dividing the farms' wage payment by corresoonding emloyment of this category of labor in a month. (e) Aggregate amount of labor hired in (casual plus aboached) by the 116 farms in Hooghly (see (d) explaining their selection). Figures are in terms of thousands of 8-hour days. 叫 62 milder second Deak in autumn, which is cleaxly a result of rise in irrigated cultivation of khar-if crops like paddy which are harvested at that time. The seasonal Dattern of on-farm em-Qloymen'U* of family farm workers (and also of permanent farm servants) shows a similar stability for 9 months of the year with a major intensification in the early summer months. Ferozepur, in the recent year, had achieved a high level of farm emplcyment of family workers end attached laborers throughout the year, with even fuller em-Qloyment and a sharp rise in casual labor's wage in the peak season. The total number of 8-hour man-days worked by an adult male fiamily farm worker was, on an average, 20.5 days in a month (Table 8). In no month was it less than 18 days. For a permanent farm servant, average monthly employment was 27 days. For the busiest o-oerations during May, an attached laborer's employment reached the level of 34 8-houar days (26 on crop production, 8 on t-ending cattle), and that of an adult male family farm woT-ker about 25 days. In Hooghly, the avetage wage rate paid for casual hired labor by the sample farms remained almost unchanged throughout the year, with a .barely visible mild rise in the peak harvest season in December. Thle pattern was more or less similar in the earlier year too. Unlike -'-n Ferozepur, where the seasonal wage curve and the family labor utilization cu rve move more or less together, in Hooghly the wage curve is insensitive to seasonal variations in farm labor employment. An adult family farm ,worker here was employed on an average for 13 days in a month, with a maximum of 17 days and a minimum of 10, adding up to an average of 155 8-hour days in a year, which, by any standard, was less than full utilization. So was his peak season employment of 17 man-days in a month, compared with 25 in Ferozepur. Even after considering about 15 .... .... ... .. . 63 more man-days that were hired out during the year, the intensity of employment offered by agriculture to family farm workers remained low and yet characterized by a large degree of seasonal fluctuation with sharp dips. In the use of hired labor, the degree of fluctuation was even sharper. The seasonal stability of farm wage rates in Hooghly,* in spite of a considerable degree of seasonal variation in labor use, can be explained in terms of two major features of the situation. One is the long-standing pressure of population on land, the acutely dense agrarian settlement; the other is the alleviating proximity to an overcrowded urban area that offers plenty of low-paid, short-term employment in its 29 huge "informal" sector. The major reason for the lack of any significant wage rise in the peak season is the presence of chronic rural underemployment in contrast with its absence or even a labor shortage in the case of Ferozepur. If the farm employment of rural workers are at a sustained high level most of the time, then for the busiest operation the wage rate would rise sharply, even when moderated by seasonal immigration of labor from outside. The cropping pattern in Hooghly is lower level than in Ferozepur with much less of the land under controlled irrigation and planted with high-yielding varieties. Compared to the earlier period, the extent of chronic rural underemployment might have gone down absolutely with *There is a question about the extent of reverse seasonality in foodgrain price (low following harvests, high rest of the time), which would make the real wage rate fluctuate more than its money value. Price supportive grain 'purchase policies have generally brought down the degree of seasonal price fluctuation (see Krishnaji, 1975). Multiple cropping has also contributed to this. Post- harvest (Nov.-Dec.) paddy price in Hooghly was just 5-7 p.c. lower than the annual average. Besides, here the kind wages are already evaluated at the mean price. Thus, deflating the cash part would make the degree of seasonal rise (relative to mean) go up from 7% to at most2%. Ir Ferozepur 69-70, wheat price in May-June was about 7% lower than the mean. The peak, 39% higher than the mean in money terms, will then be 46% higher in real terms. The major contrast between the two cases would remain just as sharp, 29 64 improved cropping and increased labor use Der acre. For workers in the lowest farm-size group, which contains many primarily labor households, rate of farm employment increased. To a smaller extent it increased for those in the larger size groups too. Still, even in the later year the extent of annual farm employment for an average adult male family farm worker was much smaller in Hooghly than in Ferozepur. The fact that over the period (Chart 1) seasonal wage pattern remained practically unchanged in Hooghly, with very little rise during the busy season, is consistent with the hypothesis that as long as there is any major under- employment, seasonal rise from the floor will be held back. The second aspect of wage behavior in Hooghly is the steadiness of the seasonal floor itself. What keeps the wage rate from declining much in the lean season when farm employment drops quite a lot unlike in Ferozepur (see Chart 1)? The phenomenon is difficult to explain in terms of a orocess of subsistence or nutrition-based -age determination, 30 because most of the usual deductions from this hypothesis do not seem to hold in this case. Engaging permanent farm servants or even informal labor tying was relatively unimportant, and payments in kind no more prevalent than in the other case. Eesides, there was a lot of short- term mobility of labor from the villages of this district to the city. These conditions are contrary to what would be exrected in an environment of nutritional determination of wages. A more plausible determinant of the farm wage floor in Hooghly prevailing in spite of the underemployment, 29 Covering a wide range of manual services, like those of domestic servants, porters, helpers in petty shops and household enterprises, workers in small production and business units particularly of seasonal type. The migrant labor from rural West Bengal to the city of Calcutta tend to be absorbed more in these activities than in the organized industrial sector. Several studies have shown that the manual workers in organized industries consist much more of rural migrants from other states than from within the state. uRodgers (1975). seems to be its location near a metropolitan area. Seasonally complementary nonagricultural employment oportunities in the huge informal tertiary sector of the city might have served to sustain a floor by enabling some workers to float back and forth in and out of the peak seasons. Going back to the difference between the two cases in terms of reak season behavior of wage rate, and in each case the change in this behavior over a period of growth, the evidence sustains the comon sense expectation that the wage rise is determined by the basal degree of underemloyment of available labor. It also indicates that as long as there remains unutilized labor time -per worker (irres-ective of whether -the underemployment is open or disguised in lower-paid or otherwise less preferred work), the wage rate may not rise even with a large increase in farm employment in the peak season. Where the level of labor utilization is high, wage rate is quite responsive to employment changes--both across seasons within a year and over the years. 66 PART II: CROSS-SECTIONAL RESPONSE II.A. FARM-LEVEL DE4AND FOR LABOR In Part I, Section A, we grouped farms by size and analyzed variations in labor use pattern in relation to farm size in several different situations. Although farm size is a major determining variable in this context associated with differences in resource availability, in techniaues used, and organization of production, the operation of other independent variables both across different farms and over time requires one to take account of their separate effects. To "apportion" the observed variation in labor use between a number of variables, or to indicate its ceteris paribus response to each, we have used multiple regressions here. This is more to find the sign and the significance of relationship with each of the factors, while effects of the others are "controlled," than to provide a set of elasticity estimates or response coefficients readily usable for projection. That, even if we wanted to do it, would be of doubtful use, particularly because of the fact that there are various categories of farms and.types of technological conditions prevailing even within a district. To apply a uniform set of response coefficients, however carefully reached, to the varying categories.would be misleading, the main purpose here is rather to identify and interpret the similarities and the dissimilarities in the response of labor use to the selected variables across categories of farms within an area, among different- situations and over time. Variables For explaining inter-farm variations in per-acre use of hired and total labor days, we will consider the folloing variables. 67 (a) farm size (net sown area, to abstract from variations in fallowing practices); (b) the irrigated proportion, or the index of cropping intensity which is largely determined by rainfall and irrigation; (c) degree of use of chemical fertilizers, in the more recent year; (d) proportion of Poss croqed area under the_mQraaabor- intensive crops: the high-yielding varieties; or even with tradi- tional varieties, jute in a paddy area, or cotton and paddy in a millets area; (e) average wage paid per day of farm labor hired; (f) availability of family labor, which depends on their alternative employment opportunities, income level, cost of hiring.labor, and socio-cultural factors that partly determine work participation of women and children of farm households. (g) the extent to which the cultivator is an owner-operator or a tenant. Since the sample farms are from randomly selected villages that even within one district can differ in institutional or soil-climate conditions, we have used dummy variables for the villages in order to isolate approxi- mately the "within village" interfarm relationships from the "between village" vv.riations. Coefficients of the village dummies, which turn out to be.significant in most of the cases, indicate the shifts in intercept of the estimated regression equation. The beta-coefficients then give better approximation of "within villIage" farm-level behavior. The regressions in each case are carried out separately for different farm-size groups. There are two reasons for doing this. Supply behavior of family labor may be very different for small farms than for larger ones. 68 Secondly, for the small farms, which are usually marginal and irregular employers of hired labor, there may not be any systematic relationship between use of hired labor and any of the other "independent" variables. It is likely to be more systematic for the larger farms, and more indicative of demand behavioi in the usual sense. The method used for grouping farms by size for the regressions is roughly the same as that done earlier in Table 2 and explained in its note. 31 For an interfarm cross-sectional analysis, specification of the wage variable and interpretation of its coefficient is more of a problem than for interregional or intertemporal analysis. Given the form in which ?MS data were collected, -wage is the average rate a farmer pays for the labor he actually hires during the agricultural year. Variations in it can arise from several sources. As the farms.are drawn from randomly selected villages, not necessarily contiguous, an important source is intervillage differences in the structure of seasonal or operational wage rates. Secondly, even within a village, the average rate paid by a farmer can vary depending upon the periods and the operations for which he hires labor (which depend on the cr-)p-mix and the supply of family labor), nature and duration of labor contracts and his specific bargaining strength in setting them. In other words, when interpreting any observed cross-sectional rela- tion between the labor days hired by a farm and the average wage he pays, one is dealing with a combination of shifts in the demand curve and movements 31The relative size groups are defined with reference to the operational distribution of land prevailing in a district in the later year. The same grouping is used for the earlier year as well. For Coimbatore, howeve.L, we have reversed the procedure and defined the groups with reference to the earlier yearls distribution so as not to have too few cases in the "small" group for the later year when the concentration was much higher. 69 along a given one. To the extent wage variations across farms arise from those in timing, in the quantum and in the choice of contracts of labor hired even within a market-given wage structure, an observed relation between labor hired and wage paid across farms could be the effect of variations in the mix and the intensity of cropping, that is, an effect of varying demand curves. On the other hand, to the extent it arises from differences in the wage structure, the observed partial coefficient could be viewed partly as variations among otherwise similar farms along a given demand. curve. The relative strength of these two elements is not easily measurable. Results of inter-farm reQressions Tables A.1-A.3 (Appendix II) give the estimated regression equations for * total laoor use per acre, labor hired per acre, and the hired proportion. It is assumed that interfarm variations in actual use of labor reflect those in demand. Demand for hired labor is re-arded as a residual from the total demand after the family decides how much labor it can supply for use on the farm. The latter is assumed to be a function of the number of workers in the farm household and the wage rate, both of which are included as independent variables in the demand functions for total and hired labor. The main results, in terms of the signs and significance of the partial coefficients, are as follows. Farm size. Labor use Der acre is either not very responsive or nega- tively responsive to farm size, after adjustments are made for the effects of the other variables. In the earlier year, it was not significantly responsive ia most of the four districts (except in Ferozepur, where the response was positive). In the more recent year, we find a consistently 70 negative and significant response. It thus appears that the smaller farms tend to use more labor input. Among the nine samples of farms that we have considered here, reflecting more or less different situations, this tendency is observed more in an environment of agricultural growth as in the later set of years relative to the earlier one. Hired labor use per acre, on the other hand, tends to be positively responsive to farm size in the cases where it comes out as a significant variable (Table A.2). The hired proportion shows this tendency more strongly. It is positively responsive to farm size in a sifnificant way in almost all the cases that we studied here (Table A.3). The negative relation between total labor use per acre and size is not inconsistent with the positive one between hired labor use per acre and size. Smaller farms tend to use more labor per acre and hire a smaller proportion of it not only because they tend to have more family workers relative to the land, a factor that is adjusted for in the regressions in Tables A.1-A.3 through variable 7(6 but also because they tend to get an average family worker more intensively employed on the farm (as shown in Table 2 earlier). Irrization or crooping intensity. Increases in irrigation, or multiple cropping as an alternative.indicator, tends to have a significantly positive effect on the use of labor per acre (both total and hired). In some cases, either of them is a significant variable for a particular size group, but not for all farm sizes combined. In other cases, it is the opposite: significant for the groups combined but not separately. In this context, it appears that cropping intensity is a better explanatory variable than irrigation in Hooghly, Tanjore, and in 71 the more recent years for the two north Indian districts. These situations are characterized by high levels of irrigation and/or high rainfall, and thus variations across farms are relatively unimportant in this respect. In the remaining four situations of relatively low irrigation and/or rainfall, variations in irrigation intensity across farms are larger and more meaningful. Substitution of a dry millet crop by a wet paddy crop with irrigation in a case like Coimbatore is a far more significant factor than just the number of times an acre is cropped. In an irrigated or high-rainfall area of vet crops, on the other hand, raising a third crop of paddy or vegetable, instead of letting the land fallow for a period, may be a more relevant factor. A qualitative change in the crop mix would here take the form of improved, fertilizer-responsive varieties. This is treated as a separate variable. The adjusted effect of fertilizer use per acre (or, effect of the propor- ____tion of_croDoed area under fertilizer-using labor-intensive seed varieties) is significantly positive in the later year when their use was considerable. In Coimbatore, these were unimportant; but the share of paddy and cotton -in cropped area seems to have a very similar effect on labor use. Tenancy. As measured by the (net) leased in proportion of farm area, in three cases32 (Hooghly, Tanjore, and Coimbatore), it does not turn out to be a systematically significant variable from the point of view of total labor use per acre. In Tanjore and Coimbatore, it has a negative relationship with the use of hired labor per acre and with the hired proportion. The underlying reason is not immediately obvious. A crop-sharing 320f the other two, in Ferozepur the extent of tenancy declined a lot over the oeriod (from 40% of net sown area for the sample farms to 12%). In Muzaffarnagar, The state ban on tenancy having driven it underground, the reported data mean almost nothing. 72 tenant would normally use less of all current inputs including labor, pro- vided he is the one who decides. If the landlord stipulates--formally or informally--the use of inputs, even labor, or if alternative employment prospects for workers of tenant family are poorer than for otherwise similar owner-operators, then it is possible that the tenant would use less of hired labor and more of family labor on the farm. It all depends on the terms of contract and on alternative employment opportunities. Availability of family labor. As measured approximately by the per- acre number of adult family workers mainly engaged in cultivation, this comes out as an important explanatory variable in many of the cases. The coefficient is significantly negative for hired labor per acre (also, for the hired proportion in total labor use), which is entirely consistent with what one would expect on a priori grounds. For total labor days per acre, its coefficient wherever significant is positive. (It is significant in 5 of the cases.) The hired proportion of total labor use also shows a consistently significant negative response to this variable (with only two exceptions out of the nine samples). Ia some of the cases, number of family workers per acre is a highly significant variable for the medium and the large farms, but insignificant for the small farm. The explanation in the second case could be that the denominator being in any case small and the numerator being an integer, any change in either tends to magnify resulting changes in the ratio and turn them into big erratic fluctuations out of tune with differences in total labor use per acre within the group. 73 When it comes to explaining variations in hired labor input per acre in the small fam-size group, the regression equations with the variables considered naturally give poor fit (Table A.2). They give much better fit in the case of medium and large farms. The small farms' use of hired labor usually,being.supplementary or of contingent nature, variations are much less systematic than for the other groups. Variations in their total labor use per acre are, however, explained better with the particular model (Table A.1). Waae. Average rate of wage paid is a significant explanatory variable for interfarm variations in labor hired per acre, for the later year in Ferozepur, Muzaffarnagar, Hooghly, and also in Tanjore (Table A.2). In Kooghly and Tanjore, the relation is stronger for large and medium farms, though insignificant for the small ones. For the hired proportion of labor too, wage is a significant explanatory variable in these four cases and also in later year, Coimbatore. In the remaining cases, mostly in the earlier year, it is insignificant. For explaining variations :-i hired plus family labor used per acre, wage is a significant variable in only 3 out of the 9 sets: later year Ferozepur, earlier year Muzaffarnagar and Tanjore. Although one cannot generalize on the basis of just nine district- years, it is quite remarkable that the partial relation between vage and labor use per acre was significant in approximately half of the cases; somewhat more often for hired labor than for total labor. The use of village dummies might have reduced the importance of this variable by taking account of at least some of the intervillage differences in wage structure. 74 When significant, the sign of the wage coefficient apears as a negative in some cases and positive in others, It is negative in Tanjore, Muzaffarnagar, Coimbatore, and for just the small farms in Ferozepur, 69-70. It is positive for the medium and/or large farms in Ferozepur and Eooghly during the later year. The latter were cases of high growth; so was Tanjore. It is not very difficult to visualize different sets of technological and institutional conditions affecting the labor market outcome variously in terms of cross-sectional relation between wage and labor use. Under certain conceivable conditions, the farms hiring more labor might be paying a lower average wage rate. If thedistribution of land is highly concentrated and the labor market has too many landless laborers with little alternative opportunities, then the farms hiring more labor might more easily manage cheaper labor tying arrangements. These conditions hold more or less in the cases of Tanjore_and_Coimbatore. Crass-sectionally, the causal relation could be reverse too: lower market wages in some villages33 might encourage more intensive use of labor. Even after adjustments for irrigation and technological differences, the effect might show up in adoption of more intensive weeding and manuring practices. In a different labor market situation, where demand has been growing rapidly and there is not much of readily mobilizable stock of underemployed rural labor, larger use of hired labor may go with the payment of a higher average wage rate. The feasible improvement in crop-mix may be so much more profitable that the farmers' decision to have it is not affected at all by the higher wage involved in securing the required timly availability 33As we have used village dummies for an entire sample, but not for any size group of farms, the effect of intervillage variations in wage structure on their labor use might be reflected more in the subset regressions. 75 and larger amount of labor. In Ferozepur, the bigger users of hired labor were employing a lot more of attached labor in the later year. They were doing so not for saving on wage cost per day (as a matter of fact, translated in per-day terms, the attached laborers were being paid as much or even more than casual labor), but rather to take care of the enlarged and the steadier stream of farm work that the technological change generated' In such a situation, timely availability of labor might be the determining factor in adoption of improved cropping practices. The average wage rate the farmer pays might then be more determined by the level of his labor use and less of a determinant. How do the wage rates, among other'things, affect the supply side of the market under different conditions? The FMS data tell us nothing directly about the suppliers of wage labor. In the next section, we take a look at the supply behavior of labor households in two different situations. II.B. LABOR HOUSEHOLD'S SUPPLY BEHAVIOR Supply of labor in the theoretical (ex ante) sense is not directly measured by actual employment data for the laborers, as demand for labor is not by what the farm actually employs. Still, by relating employment variation across households with some of their conditions one is likely to get some idea of supply behavior; perhaps more so than by asking the laborer the hypothetical question of how much work he is ready to do at various wage rates. Using the 25th round (70-71) NSS data set for Uttar Pradesh, this section attempts to find a cross-sectional view of differences in employ- ment in relation to other factors for the laborers. For that, we first look at grouped data and then do a household-level micro analysis: for the western zone and for the rest of the state separately. In each case, 76 the distinction between landless labor households and cultivating labor households is maintained in view of the usually postulated differences in heir market supply behavior and/or differences in farmers' demand_for labor___ fromthe two categories of households. The reference period for collection of labor time disposition data is a week preceding the date of enumeration. We have used the Central Sample of data relating to a total of 1,003 noncultivating labor households and 1,298 small cultivator households. The sample was spread out over four quarterly subrounds into which the entire year (July TO-June 71) was divided. The subrounds for NSS data collection cannot,.of course, precisely correspond to the seasonal cycle of agriculture in a particular region. Roughly speaking, the major rabi harvest season in Uttar Pradesh is reflected in subround III or IV CApr.-June) and the kharif harvest in_subround_I or I (Oct.-Dec.),depending on whether it is in the east or the west of the state. Table 9 groups the labor households of each category by the rate of employment per head of working age men,who are behaviorally more comparable across regions and types of households. For each labor household category by employment rate, a further breakdown is given by subrounds in order to indicate differences in terms of seasonal behavior. One of the remarkable things, from Table 9 is that there is a considerable section of labor households--28% of the noncultivating and 21% of the cultivating ones-- with their working age men employed for less than a third of the 56 hours in the reference week, On the other hand, there is a much bigger section (1 and 3 in Table 9) in either category of labor households in which the working age men are almost fully employed during the reference period. * Total employment per man, including employment on wages as well as on own account. 77 Table 9. Characteristics of noncultivating labor and small-farm households according to intensity of total employment for men (15-60 yrs.). Uttar Pradesh 1970-71 Ermpl. in 8-hour days. aAll wage Adult males Adult male Per head eml.as Daily wage as prop. of Weekly total empl.as total p3.c. of recd for all all workin2 Household eml.(days)/ p.c.of total empl(days-adult male adult male age adults grp/subrnd adlt male epl.for h.h. for h.h. ltot.em-ol. wage empl. & childena 1. Noncultivating hshld with adult male intensity of employment exceeding 66% (4.6 days in a week): 62% of all noncultivating rural lab hshlds. SR I 6.96 76 2.78 79 2.51 .27 II 6.80 78 2.69 62 2.27 .31 III 6.78 74 2.86 79 2.49 .31 IV 6.74 75 2.66 74 2.50 - .30 2. Noncultivating lab hshld with adult male intensity of employment 33% (2.3 days in a week) or less: 28% of the set. I .47 7 1.44 100 2.44 .21 II .15 3 .88 100 1.26 .20 III .51 7 1.22 99 3.21 .18 IV .56 13 .97 79 2.03 .22 3. Small cultivator households with adult male intensity of employment exceeding 66% (4.6 days in a week): 72% of all cultivating labor hshlds. I 6.81 69 2.63 44 2.16 .27 II 6.69 73 2.73 51 1.93 .30 III 6.90 73 2.67 53 2.31 .28 IV 6.81 75 2.57 48 2.05 .28 k. Small cultivator hshld with adult male intensity of employment 33% (2.3 days in a week) or less: 20% of the set. I .90 9 1.56 91 2.99 .16 II .61 3 1.74 0 - .10 III .29 3 1.31 92 1.27 .15 IV .4T 7 1.22 55 3.15 .18 a. All prime working age adults of 16-60 years, and children of 5-15 years. 78 The high-employment households (1 and 3) have a seasonally sustained high pitch of employment which is in clear contrast with the unstable low level for the low-employment ones (2 and 4). Secondly, although the component of own-account employment was a little larger for the high- employment households, the bulk of adult male employment was on wage in both sections and accounted for most of the difference between them. Thirdly, the rate of wage received for male labor was somewhat higher and a lot more stable seasonally for the high-employment households. Among the latter, it was even higher and more stable for the noncultivating labor households (group 1) than for the cultivating ones (3). Finally, the degree of work participation among women and children was more in the low-employment households, accounting for about 90% of total employment for all the members put together as against 30% in the high-employment section. The men in these households were possibly float- ing workers-being away from home much of the time to work or to look for -work elsewhere. The women and the children had to work more. For most of these low-employment, poorer, labor households, the working age men also constituted a smaller proportion of the hous Lid members, possibly indicating more of individual male migration in theircase. Obviously, regional differences within a state in terms of cropping pattern and intensity and employment,opportunities in general acount for some of the cross-sectional variations in a statewide sample. Besides, since the sample households are randomly distributed over the year, each with a reference week (instead of being a fixed set surveyed throughout the year as FMS data), part of the interhousehold employment variations shown by the data simply reflects their varying placements in the seasonal cycle of agricultural operations. Some art of t.he var'iations and 79 apparent stratification in the distribution of employment among laborers could also have been generated by labor market imperfections. One can easily think of a number of stratifying mechanisms even within a homogeneous reginal context. Long-term contract of an attached farm laborer is a major one. Special skills (of artisans, ploughmen, cotton pickers, cane juice processors, etc.) form another. The question of differences in labor market behavior between landless laborers without any other means for self-employment and those cultivating some bit of land is also quite relevant in this connection. In the following regression analysis, we have tried to control ap-proximately for (a) regional differences, by distinguishing between the western zone and the rest of UP, (b) seasonal variations, by using a dumy variable for each of the four subrounds into which the agricult-ral year was divided for collection of the data, and (c) differ- ences bei;ween the landless and the cultivating labor households, by fitting a regression equation separately for them.' For a given.stratum of labor households within a region, interhousehold variations in supply of male labor for farm wage-employment in a week are sought to be explained by the following factors: (a) the number of working age men (15-60 years) in the household ( M); (b) the wage rate for male agricultural wage-employment for which we used two alternative variants. One is the average rate received ( w) for the household's actual employment of this kind in the reference week. The other is the average rate for agricultural 34Zonal subdivision for the rest of UP is not maintainable without rendering the number of cases too few in each for regression analysis. 80 labor in the village (ve), derived as an average received by the labor households of either stratu (landless or cultivating) 35 as a group. The average wage rate can then be used for each of the households from the village, including those that did not have any male farm wage employment during the reference week. Since the actual wage cannot be computed in the latter case, only the households with nonzero farm wage employment can be included when using the first variant of the wage variable. Besides, one could argue that the village wage is better indicative of the market price at which supply decisions are made. However, the households included being those with at least one working age man (because supply of male labor is to be explained), the differ- ences between actually received wage rates and the village wage turn out to be small in most cases. (c) level of per-capita consumption expenditure (C) of the household. (d) dummy variables (Dl-D4) for the subrounds to which the household's reference period employment data belong. The dependent variable, viz. household supply of adult male agricultural labor, is defined alternatively as AL: man-days actually engaged in agricultural wage-employment during reference week; or 36 ALS: supply of such labor, as AL plus unemployment, assuming that the unemployed labor time would normally be available for employ- ment at the prevailing farm wage rate. 35The households of either strata selected from a village were all surveyed within a subround. Hence averaging of any data for the village is affected by seasonal variations between the subround quarters. It might still be affected by seasonality within a subround. NSS 25th round defines unemployment as the time spent during the reference week not working, but seeking and/or being available for work. In most 81 Table 10 gives the regression estimates, by region and by household stratum. Number of adult male members comes out expectedly as the most important explanatory variable. Consumption level of the housenold has a small negative effect on the employment or supply of male agricultural labor in the western zone, and no effect in the rest of UP. At their generally low level of consumption and income, the negative relation cannot be interpreted as leisure prefer- ence (or income effect). It is far imore likely to reflect a kind of "distress sale" of labor, so as to reach some minimum consumption level, on the part of households with bigger families and/or facing lover than usual wages. In any case, the coefficient is very small. The seasonal or subround effect, apart from wb,t is taken account of through wage rate, is sometimes significant. The kharif peak shows up in rositive coefficients of Dl and D2 in west UP; the rabi peak and the ensuing drop in the positive and negative coefficients of D3 and D4 in the rest of UP. This is all in the case of landless households. There is no subround effect in the case of cultivating labor households. For west UP--characterized by greater irrigation, capitalist farms and also r .e landlessness among rural labor households--the partial relation between supply of male agricultural labor and wage rate is positive. 'The relation is not auite. as, significantin -the r f he_a cases it turns out to be a minute fraction of the reference week. The time spent in leisure, family affairs, domestic work, and the time rep,rted as outside the labor force (i.e., not available for work) are widely regarded as containing elements of unemployment which. however, are not easy to separate out. 82 Table 10 Regression estimates for household supply of male agricultural labor. Uttar Fradesh, 1970-71. Western UP For noncultivating labor households AL = .18 + 2.73*M + .57*VW - .03*C + 1.T6*Di + .60D3 - .26D4 (.84) (.28) (.23) (.01) (.52) (.56) (.62) 2 R .21 F = 21.5 2 7 ALS = 2.34* + 3.20*M + .34**W - .03*C + 1.33*Dl + .91**D2 (.72) (.22) (.17) (.01) (.37) (.44) R = . 41 F = 45.5 e -, 9 AS.-, .1 For cultivating labor households AL = -5.08** + 4.30*M + 1.28**VW - .01C + .79D1 + 2.18D2 - .65D4 (2.32) (.65) (.55) (.03) (1.20). (1.32) (1.35) R2 = .39 F = 9.8 e ALS = 1.04 + 4.46*M + .50W - OlC- .12DI + .27D2 - 1.27D4 (1.97) (.50) (.41) (.03) (1.10) (1.20) (1.28) R2 = .67 F = 15.4 e= .16 Rest of UP For noncultivating labor households AL = 1.43** + 3.0T*M - .1OVW + .0002C + .52D3 - 1.30**D4 (.70) (.36) (.28) (.0005) (.48) (.57) .2 R =.19 F = 16.2 e = -.04 ALS =-2.TT* + 3.4T*M - .10W - .24D2 + .89**D3 + .76D4 (.54) (.25) (.16) (.43) (.42) (.52) R2 = .47 F = 42.4 e = -.03 For cultivating labor households AL = 2.77* + 2.6T*M - .69**VW - 03- .51D1 + .89D2 (.93) (.40) (.29) (.02) (.52) (.55) -R2 = .18 F = 11.9 ez --,-30 ALS = 3.20* + 3.14*M - .18W - .48Dl - .09D3 + .28D4 (.58) (.31) (.10) (.50) (.52) (.55) R2 = .39 F = 22.0 e = -.05 Notes *Coefficient significant at 1% level; **Coefficient significant at 5% level. Variables with F-level or tolerance-level insufficient for computation are not in the equation. NSSO prepared certain multipliers or weighting factors for the 25th round sample households in different stratum and in different regions. The regression estimates here, and the aggregative estimates in Tables 6 and 7 based on this data, are derived by using those multipliers---- e: Wage elasticity of agricultural labor. 83 Cross-sectionally, within the more advanced region, those supplying or taking up more agricultural labor get to do so at a higher average wage rate than those who are marginally involved in the market. This could be interpreted as indicating the "normality" of supply behavior in such a situation. (In the depressed regions of the state, variations in employ- ment or labor supply--arising from demographic differences among labor households--take place at inflexible waqe rates, the low levels of which may even lead to "distress sales" of labor). Even when wage is a significant variable, the wage elasticity of household supply of male agricultural labor (measured at the mean level by e in Table 10) is always less than unity, even .less than .5 most of the time.In the more developed region the supply curve is far from horizontal;it is actually highly inelastic.In the less developed region the wage elasticity of supply is zero 7or a very small negative. TableA..,1<ýPonse or total labor use per nt sown aare to uelel,ed variablea. Five districta. Size-groups. Farm regreusion2. a XI x2 X2' X3 X3' XII Xs X6 P.c.of gca District/ k,,o. Vertilizer in impr. Adult signiifieance or year/urz- Constant Farm size Irrig. Crpping (or F&M) var. or lab- P.c.area Ave.vage tarm fam. Eq. form 2 village intercepts iza r term (nsa) intensity irntensity Use/nsa intens.cropa leasedi TalC id wrkr/acre fittdd R d.t. (du:any variable) ir 1956-57 (-2.331 .15" .27§ .016" .04 .056" Loglinear .66 85 All 9 coet. signif. All sizes (.4-) (.06) (.0) (.006) (.06) (.025) -2-53» -.07 1.12" .012" .03 .062" Loglinear .69 85 8 out or 9 coer.signif. (.41) (.07) (.28) (.005) (.07) (.024) 'm 4.2E4 .06 45 .079* .13 .60* Loglinear .54 23 2.5-15.0 a . (.33) (.18) (.20) (.03) (.16) (.15) 4.16* .03 .31* .03 -.11 .391 Loglinear -52 48 (5-35 a- (.56) (.19) (.08) (.02) (<09) (.08) ed.Carge 7.94 -.01 10.21" .02 -,23 81.01" Linear .39 64 15 ce. ('.32) (.07) (4.01) (.01) (.61) (16.09) !rmur 1969-70 flYV wheat .11 sizes 7.86" -.46" -.12 .04" --35* .01 Loglinear .55 130 6 out or 14 coeff.signi. (.19) (.06) (.14) .(.02) (.i1) (.03) r.all 7.99k -.67" -.58 .14" -.58% .06 Loglinear .47 31 15 ae. (.46) (.30) (.614) (.04) .(.1) (.06) zia -151.6 -10.5 432.2" 296.1 431.3" 29.6 Linear .39 62 L5- , a. (223.8) <14.2) (123.2) (184-3) (182.2) (189.5) -1631 -lo." 556.6& 110.9" 249.3" 20.7 .inar .47 107 a1, w-, (141.A) (2.7) (82.7) (36.0) (130.9) (148.5) _ 19'6-57 -0 All tizes 11.3 -.16 20.77* 8.33 15.27* Linear .62 81 5 out or 9 coeft.signir. (12.1) (.20) (8.66) (5.93) (2.89) 3r.all 3.09 3.25 17.37 11.26 50.48" Linear .41 15 2.5-7.5 ae. (30.77) (3.01) (16.49) (8.15) (18.73) -,ium 43.4" -.92 12.08 -11.09 107.561 Linear .45 37 7.5-15 ac. (17.36) (1.14) (7.40) (6.69) (25.88) 12.57 -.07 25.18" -13.15" 95.25" Linear .42 62 (12.04) (.19) (8.31) (5.12) (20.51) > 7.5 ae. 36.861 -.02 5.18 -9.62 103.86,1 Linear .34 62 (9.Ž4) (.21) (6.35) (5.30) (21.90) zefrerna;r 196-69 All Sizes . 4.93* -16" .65 -.13 .04 Loglinear .56 131 2 out or 14 coerr.signit. (.14) (.04) (.19) (.07) (.04) zr.'11 85.8 1.9 42,5 -9.0 10,7" Linear .26 32 < 7.5 a. (43.7) (8.5) (25.9) (6.4) (4.4) 4.52, -.01 .91ø -.15 .14" Loglinear .31 42 7.5-15.ac. (.28) (.17) (.27) (.08) (.05) 4-50" -.02 .94* -.16 .04 Loglinear .26 108 %ed.&1rge (-15) (.06) (.18) (.09) (.04) >7,5 ae. .2 -.06 .b0* .06" - -.08 Loglinear .33 108 (.15) (.05) (.17) (.02) (.82) Ea.y 1956-57 4.011 .07 1.03w -.03 .07 .11 Loglinear .81 70 5 out or ce ne.g- (-11) (.06) (.10) (.06) (.17) (.07) S4 0'* -.01 1.68* -.05 -.50 -.08 Loglinear .52 48 Ss11 ed. ) (,13) (.13) (.25) (.09) (.32) (.14) 25-01 6.9 48.1* 78.1 " 7.9 3.1 Lincar .53 4 (13.07) (6.>) (11.6) (32.1) (11.5) (2.4) Large 4.06* -.07 1.26* -.26" -.05 .16" Loglinear .94 23 > 3.0 ae. (.18) (.11) (.13) (.li) (.13) (.07) Xl X2 X2 X3 xj' X4 X5 x P.c.o gca LictrI n/ Fertllzar in im.pr. Adult aificance of ytar/.rr- Constant 1arm 1ize Irrig. Cropping (or IILM) var. ar lub- P.c.ure Ave.unue rur et,I. f' torm 2 village iILrepts ize gr.up term (nlu) intensity Intensity Use/nsa intens.crop leaused3In 'yte jald vrkr/acre fitted R d.f. (dum~;y vrible) :::..1r 1970-1 Excl 34 4.121 -.311 .02 .27k Coerc.for --15 .001 Loglinear .53 110 r.arg.farmrs (.19) (.08) (.07) (.03) Improved (.12) (.08) All sizes 3.8 -.30* -,09 .32" .ddy .05 -.07 Lglinear .74 130 3 out ar 14 coeff.signir. n = 150 (.19) (.05) (.07) (.03) (.10) (.06) 5=-1l::.ed, 238.8* -61.8" -34.2 .65" -15.5 9.4 Linear .60 61 < 3.0 c. (140.5) (15.4) (33.8) (.09) (9.3) (11.2) Large 3.84 -.29"' -.001 .34" -.14 -.01 Loglinear .57 43 < 3.0 ac. (.32) (.14) (.07) (.05) (.18) (08) lrhntore 1956-57 12.9 .23 8.09m 1.44 19.25* Linear .83 75 3 out ar 9 coerr.signi. (21.4) (.66) (11.94) (12.06) (5,142) Ul size5 -57.2 -.G2 60.98k -2.15 21.26" Linar .80 75 3 out ar 9 coerf.signir. (31.6) (.70) (9.75) (12.79) (5.82) Medium 85.3 -7.0 87.8* -39.1 21.7 Linear .42 52 X2' in place 1.75-6.5 (50.4) (6.0) (17.3) (24.2) (18.9) a X2 gives sImIlar Ied.narge 36.9 -.05 85.9M -23.7 30.2" Linear .43 76 sult > 1.75 ae. (30.1) (.99) (14.) (19.6) (13.8) Coirbatore 1971-72 58" -1.29 172.5d -2.3 -1.6 Linear .71 129 3 out af 14 caeff.nignif. All sizes (25) (1.48) (17.3) Paddy&cotton (8.5) (2.0) 62" -1.6 86.1 3.2 -2.7 Linar .51 129 4 out at 14 cef.s15nif. (32) (1.9) (32.6) (11.0) (2.6) 2r..l 9.3 --5 21.1 -.5 Linear .50 147 X2' in place 6.5 ae. (53.5) (18.6) (26.5) (15.8) (21.2) a X2 gives S6.5u rge 67.9 -LS 164.ok -7-7 1.5 Linear -56 91 s alur > 6,3 a. (25.1) (1.6) (16.7) (8.9) (1.9) result 1967-68 97,91 .73" 28.5" .30" -3.29" Linear .80 130 All 14 coeff. signit. (11.8) (.32) (4.7) (.05) Hyv paddy (1.3) All sizes 13b.9- -1.3* .49 1.6 -10.t -3.29m Linear .74 130 All coeff. signi. (13.6) (.4) t.05) (10.7) (6.3) (1.51) 60.5 -3.5 8.3 .5* -5.1 Linear .87 18 < 2.25 ac. (60.4) (16.3) (11.8) (.1) (16.4) Mediu:a 79.4ø -3.6 21.9" .5" -12.9"* Linear .45 73 2..5-7 az. (20.5) (2.6) (6.8) (.1) (5.9) 33.7W -.75 26.0" .5" -3.5"" Linear .46 122 d arg (11.2) (.143) (5.4) (.1) (1.6) > 2.25 ac. 76.5" 1.01" .5" -2.6 -3.3t Linear .37 121 (11.1) (.48) (.1) (10.7) (1.8) 9.4 -44 36.9» .5* -2.6 Liner .54 44 (19.6) (.61) (9.4) (.)' (1.7) 76-5" -.52 .6* 8.1 .42.8" -4. Liner .46 43 (1.) (.68) (.1) j (18.7) (17.14) (2.1) _________ icVant ai p.c. level. sig.itican: at 5 p.. level. r.Mifi::ant at 10 p.C. level. f : Lat,,r used ir this and nubaequent tables in meanured In converted Q-hr days e.mployed in alI vork coanected vith crop production. Only for Ferozepur 69-70, labor is measured in hours. This is why the coefficients of linear eauations in this cae are bigger. .,' .р,.. �_,.. ... .�...._...._....._ ._. __..-.._ .__.. _....�.� . _.._ . н ' . W `� . . J 1. ♦ . , . . ` ^ ' • �- - •---- __._...��=�..а-•-�---�- �°_ =.;_"--°-"-`.г �.й � • --У,--__ ¢ . ___._.f.... ' __'' . _- ' -..ia Y::=v iзЫе А•�,..'.гч�освг аГ sLr°д tsSor ц:е 9ег ..ес гоvп асге : ._ sel�cced vzr_зbZ_s. -у� гв3г_s3?дг.. ??че �цsv_:_ta. - � а Х1 � Хг :{2' :C.G ' I • УЗ' ' :44 � :Г � :{5 11I ?.с.дf Scs � B:scric./ .� 'е^:ц_е_ г '_��г. � ;l.iг:1� 5iтui:lcsec_ �f угчr/;sr_- С�_с_с '�:� з___ ir_id. L-дoeia:; (аг ?i.,�) i•гег. �г 1г5� Аа.ггеs (�•г-_.лs�г • .^ _а:. Е'q :о �( .� п11sт_е ссаге_ct si�e :;n С_-- �(гsа} ) 1агг.^s!t•г (д:rаи .г ! � _ `L=_sed �' _ _ _-_:.� � _ _ i 'згiаsз :zz -s. es с1 rdtcбдi_� -:гг 1.3 i {сс.т.f :.гЫе) _- _ _ :°: ->> ; , _ _.,. I -. l � f � .� ! _ _.. .! . I 1 � - i.чJ � _'• . 5., ) ... � �. -� � ' _iпси ; .=J � L': ,..... ,. : � ;: (2.'сё} ; (.0Э; ) (2.�2) I {.ОСS) 1� (.331 � {д.С3} I ; � I 3 сог._. _�i . ц1 s_�гч -,q4 � .35�' � .�4 .07 1i .1'i -.�+'С��''1�е3-) .ач � д � 8 oct � �° (.5$) {� �) ' {.3Z) {.04} I� {.22) � _э � i 5 ' 9 _мгt:.si�c__. 3aaL' (З.до) {.7?) ) (Э.С4•R .CGi ,I -'Э I -�.04 _1�csr + .г7 I 23 1 {, ) � } {.ОС4} ;� {.ау} ! {7.гз) ) ; ;6 ! � Че�!� -3,j4ал i.3��в � .ЗS =i• �. -.U� � - :9тм� й:gSzaг= .32 �:Э (i.3г) � (.S1) � {.z5) {.з5) � (.29) I (.го} -� � ,ч,•д.L�1з=3е -2.s2• .э?' .г9 .13"� I .iЗ 1 - 40+•I[..:у1.iа:з.. .35 1 5�� {-э5) �(.3v1 � {.гз) (.05) � � t.г5) I� {.ьг) � i 1 _?гмге:иг 1оЬ9-%0 ` i а!':Т гhesa I �ь I . ,L`� sizгч i.01" .1г I .5о .05 I _,70У � _,10 I1о�1�гв.;i ,у � 130 � 2 сиг oi 1� меi_.зi�1`. (.S2) {.г2! (I (.>3) (.03) ) � (.цЭ1 i t.S<) �. I ScaL: '+. i?' � 1.07 ! -1.30 .г5вл � _* .29' .ООг ; ..7gii.^..esr I . Э5 11 3? + , {г.ЭSi Е {•Т�) 1i ('-.гб) {-12) (.4п.) {.:.5) � . ;'м_t.i::д -2ЭЭ ) -5•3 t 2S?.5•` 8k.7 i83.5� -='ч8.:' [.lcasr .Э3 ) :2 I (204} {1?.L) I {107,7) (1о0.9) (i59.1) ('-5j.5) г - Lчrze -]�5 �'2.. Е 6г0.1• ц.а`= j 41.3 -1i3ч.д" Lice_ i.99 � 3Э {2Э0) � (3-31 � (ц7.2) ( � (34.1) f ('37.11 � {1L8.7) ' I � '.da1,S1�g� -23: I-7•i ЗТ7.1" Тг.Sмб - I Эг0.7" -3`о�.9' Liae_ 1.3Э 107 i , ;1эз) � {2 5) С'°-.i} (з1.0) f (1е1.г) (.� .г; � 1 i?�.`'_-.С"�г^'-=„ :о7С_.7 е � � � � � � f � ta �L1i s'.гед 1.'2. i гг��` ,(7• � �! -.5о -.Са t,ед.'+�еаг� .��7 1 '�1 �� аиг __ j�пе:'>.'_r?:. (.Э5) (._;} (.17) . (•3�) � (.1г) I 5е�11 2о.,. I -.3 -12.3 г Э.2 -о.д ��'.nesr �.20 � '_5 (1а.41 . {1.а) (э.э) (�.г) (�.z) :!3di;+m .57 '-.12 ё4 _.59 .Ь5 :.a�l:aesrl .13 3Т (г.55) ` i.72) j (•2г) (.5а) (.Эа) � ,�itd..'�1ц•g: ?.�>" I ?� ' .г5 -.62 23 Log1'1е-j ,__ о'2 _ {.59) 1 t.г�}_ i.1г') � (.за) :.зэ) 11 � � 5tu_xfCi_^_sr_ _96'-лЭ � I I � � I � 1{ I 4 f j (�.3 . 5 i i.i7 .Оа � -7.Э3" ( -7.15• LL^asr .51 т31 I 1 оис оГ 14 мei:.s?�L. ( ) � (•'э) � (ь .z4} {.О3) � (г.96) {;..9�=) � � _. ,Lг si�_а t �3.Аб• �0 , в -Э -.13 _.72• Lo-ylicэarl .;3 (_.... 5 r � оР 1'-• сое:_°.зi�.�•' � (.ц1) ( {-rŭ) {-2$) ( (.2S) {•i+) • � 5е31.� 1.цh*+ ` -.:о' �?.=5 1.З��у' -1.�Ь' i:..-yllreei:i .33 32 1 {�5э) ; ; 377 {г.4г) � ' (.5г) (.39) � I .. Чс��п� _7.1 � -'г.$ ;0.0'' .1о. I _ё.2 � _=ч.2* С.1аез.: � ..Э � �-1 � {Э5.5) 1(•:.г) � {19.4) 1(.о4) � f f5.4) � (7.4) � � � i { ! �1 г 1 � !(гi.:3е;ёг -L.3 F .i � li2.Си � г1в � � _i.o'�"' t _2...�"` 'Lisг� ? .77 � _07 г �гг.г) � {.-) ) (3.г} � {,ог) j� � (з.�> � t�.2) �� � � i 4 c�rs- г? � .о � gg. � ,�, _,Э,•..у ; -гз.з� _е_ ..с� � 30 ) !агг � �.sa `_з) 1 t.оз) � 'j 1 (n.кг ! i;.э� � ` � �. гсс 1т !5S5_д7 { ; � ) i ; j .?7.5" �� _,: r 37.5' � � � � о.7 --.5' � :.i.ee.� � ..э � q� � i оис эf Э есе2".в_ __. д1S s1_ {13-г) {_.э} � <9<') � гисе , d3.i) j l_'�, I. � � . ев � ?З,+,м - 1z � -�.Т 4,5" --Э.о" � Т.д"* Е _е.5' .,�==es � ; � 5Э � 3 а�tг af ., сае:;.а , "' � (z0.4} � (1:Oi � :�э.1) � � tza.1) ;s.э) г :з.г) i C:..zi' � �" � Э � :пг'_`.!�ед. -23,3 1 •g: � 36.Э• � г � г.i � -� 0'•" '?cesr � .17 � �9 � {2��.зi � ,s.�) � {1з.ь) i s {.��.5) � {г._} ;� � ) � з.г � -•3� - =г, ! � � ь� 1 -1.-г.�- I г., -• .5] � г.: �а.г3а i. А -У . � � •�.г ! � _�� ___'-__� г XI X2 X21 X3 X -3' XII 11.ri.llizr III Impr. Adull, coliýitutil Faljo si-le I vrig, croplil JIL, (or I,,m4) var. oi- lab- 11. c, a rLu Ave WILUC i,arvi Fulil. Lil. r01,111 2 21:r., gruul, terjø (1W.) Intd11-41ty I[ILLou I ty U4.1j12U I~dd in Le IILLý-d II d. i'. (dul.u,ly vhL-lubl.) Lxcl, J4 07 .2>* 2.51x -.42- "gl1i4eur .47 310 wr4 (.31) (.19) -2y1. I.Ogllu,ýLli' .65 CJLIL Of III CUClY,tilgfLIV. All, iitzeu (.25) .58 36. 53* -llxø 12.92ý -5.99 I.InLar 40 130 1 OUL OC li, coerf-aiwilf. (19.9) (1.60) (12.22) (.05) (1-15) (3.96) ALI -.11 .36 )-1- 2.32 '002 LoElineur .51 61 (.37) - 25) (.08) (-39) (.31) -.15 2. 34 IX 1,0g111wi. .56 43 (.51) (.24) -.02 12.111 -1. 01 -1.55 LIncajI- .611 75 9 Gut. C>f 9 All aizeý) (7.02) (.22) (3.913 (3.95) ( i .,M ~1.112k -.31 Løgllac"r ý66 '75 M.d.&býr&c 23. JN" -.21 fi.ght -10.09 1.01 Ufleut. .1,1 16 (9.68). G3'- (4,70) 01-45) 20.14 3- 120.0¥ -3(;. ø* u 1,6 1,1 tiLar .61 3,18 2 out of 14 coefY.Lilgulf. (1.5) (17-3) (16.8) (8.5) (2.0) All slz,ej 3-7- I22.1t -9 1,11.eur .65 129 2 out or 14 (B-6) (2.0) Bm-11 -15.3 -19. 5 105,3% 21.6 5), ý 8,L Lindar t) (23.9) 3) (3.9.1) Medium -60.6 162.,.- 31. J,. 1.5.9 LInL-i, -65 36 (62.5) IJIA) (24.3) (13.4) (8.3) 11.6 .05 Lineul- -50 gi (27 -5 (1.8) (la.3) (9.8) 2 . 02) .101 190'1-6b ý5.24 .4 20.5k -2. d N -4.2 Liliear 4 130 13 out of 14 All ul,4(:u (.3) (3.8) Go4) (1.1) (4 -3) 20.2M .13p LI'ricar .76 131 13 out ol' 14 (3.6) (9.3) (4.0) 511.7 -11.3 20.-JK9 -6.0 -li. )I Un8u1ý .81 1-1 (11.,l) MLdI ul.ý -2.7 27-7N .3111 (9.9) 00 Md. &Iurgc 9.0 2,,f. 9 .30N -3.2* -4.7 Lificas- 121 (-39) (.05) (8.2) . . . . . . . . . . . 1ýligni VICUuL at 1, p.C. level. liotea - I-abur uuud in lijiu and uuleIýquent Lubl.a Itj mekujuI-cd jil B~lw dya tit 5 p.c. level. etoployLd lix LLI l work connectLil wi-LII exop liroduction. w-- ýýL ZI- UL 10 P.C. level- {, _ ...,---,... �.. . ,...._ .. _ _.. _ . . _ ...-...У... . . _.._�.�. . . „_.. , .»_.... . .....�._-.__. ._ ._.. ._._ . , , . . � � ' �� . . а 88 ' . ` ^ Г ' ` } .. _._..__..-�- ----....`..._ . _ • .. , _.. w�._...._.. _�.�_.�, . . ��..��.�.� � - _ --_.'_ ..�..___• _ � Yab1. fд,�З,'гsgсезе ... �i.г1 ;:7-or_i�a .. £ г _•��, ' . - - ase: Со --'-.ггд vцlshL-_s. у � - j ° � SC � 72 :<' � ( ,_ . з � хз' � :cs z5 � хд �istrfcгl . '.с.оГ �сз езт/tзг- °._�.у_,=;� �!д 1:pr, а � � У соаsгад: � ?1-cz зlее � ._ г; � � . i . ..... __ Ь"u1: - � - I 5i�if.cs: сс о. 111 elte � гау ==Z� onef_3 �(дг ёiЧ) �•:а.г, ог 1sЬ_ � а . -.� �,tие.чь.гп �:i^i : i ^ �^cq :гrt а} i -:easitv 1гг.сь- '=у !С5г/csa 1!а _as { ��s. - aq. :ог_ � � т11:уе :а:е_алдс � _ - о-' I ' ..c�os _ t�7liYЭe_:аьiс '/зс.-а •г� �д 3' � 1,_.� �д.� __" .е^ос оиг 'а ? 1 . . : � � ..�у �.я�• �� А1'. s1_г4 -.�0 009• .2l•S• 1 f - I � ^ f {.121) � (.0е2') � (.�?5)' 1 (.015} 1 �.35�ив ;цае'у• � 50 � ЭЕi � 1д co1t. зг��г. t I _ а I .175) f .. ` а_ I i Med.S1a.-3e (,'у'�Э9) � �G�!• � .1ь1' j i J г) (.c39I . .025 � -1.'_,гп ! �,,,�, �у� { � .j ,, � j ( (.�15) i.3�) { � � !1 -:azenc? _Э.4-"7 � � f � I ' I f + :.L з4с:ч -г.__� { .�З� г .Эо . ! -.аг I I - ,? j 1 {.79) � (•.2) � l•5t} у (.ог) j � -.1�' ( :,д5=+✓.еi:� ..9 L]0 1 аиг ot ?а aoг::.s:ri=. � 3'N viгac ' ' (.23} (.J7} _ � � •Sгл.L1 -3.'�h� { t,г6. -7, 51 ! • t2.ca} � {.sз) 1 (1.о2} (. о> I -.5д -.О6 � са�аавr .3� , з2 f l;ед.l.и.� .Оо 111 .ссо 1! (.3'г) (.1i) � (.ё3) � (.ц'<J I � (.'2) (.279) � .54' -.ё3' S?atsr .1`о �2 м�d.S1i ге й2 I .С5 гI ;аз I(.12) (._8} � I � (._,'S) г t. G?f 1: •,} l:с8г 1 .352� -�2_'1 �:песг I,23 t:ОТ ws.g_- --_ - ; -т} � (.1�т) � (.2R6) � . ' rU.1 3_._i' в.О' '' ff 00$� � .j53' I I J { -s (.-} 1 (.сОг} i(.аЕг) �.asz) (.J20) I:.г,ез� �._ гз. з a�z or 9 гаг,'t.ai- _. 1 51ха гоио- 1 � А 6 � "�л i :ise геgг. .1 � � � 7i._? 3 no I, I I � 3ij..!Л._:9.- _ � � _ �.з�'- .- _63�6а , t 4 { �i2 зiгеле -.Зг' { ;9+ ! ?3• t I ( (•L2) (•17Y (.'•3) � �Э {. 5) (.S3) �'s`+-'.а_г ,53 � 1Э1 { 2 оис at 1д оаеР_.э?.yz=_. s:.ц: .53 � .о9 1о.,г I _ „ _ � � � I (-�21 � (.3Эг (а.и1} f � -2.i8 -.3г� !дgl.:.:e1I ,о'i{ j2' к.-а.ыа-; -1.5?п .ЭЭ• .ТОЗ I i(.431 (.Э5) I � � � r_� , (.:�з) + . ( ) 1 � (.гЭ; ;.z5) I {. (.г7)• .23• "°�`�е-� .Э2 � s09 i :vr;tl 1.о3 .Я1 ,7�. ( U ! t.d7) � 2.� 1 i,27) ( , i -t:�a) i (:�k] �'s�_:r� ''° � бг 1 � ?ос;Ь!'г ±пs �с7 { i --- ,�з. .,zв4 ,2s � .Ог .гз� � -•зг� � .о9 �•�� � -.0 �itear � Tt l._Т} (,.02) (.12) ( А7) i � j) I•i7 70 h out о1 Э сое::.з1~i_*. ,.а (.аг) ! - � oxaLSc.a. .4Т .с$ ..14 �м � (•.-) i (.со) ( (.27} I (,эг.э) � (.iг} -,01 sae_ ' .16 ьЭ , йг1'3е .5Т` � -.Jг- ! .гЭ , � - j (.021 ; _ (,25} (.J31 � .34М. -.4;..� г�..._ц I � � � i (.го) ( (.15) ; L2'f) •9i гЭ 1 с'о `гhlц _ч1д-' � j { . " � i _"_. -�� YxeS. д: -.01Э � .иг3.'^ { а17 Ра.:�7' . i i 1 _ara._п.-s ' '�:5 I -.31z '1 .226+ -.ОЬг I (.01ЯУ (.01о1 � (.Ядо1 з :.гаsц .25 Lo � (.г01) I � s�a.:2s�e3. --.Г-S ; ь . , Е � t (.ог9) (.o4o3I 1 (.5Г9) � (.3°-у} � (.25-ь} � � 1 (.1до') � (.412) �008 f СаSцде=гг! .27 1 6i � 1 г-G к- .557 ! . со8 -.;Яи ( . ]" ) ,( ! „ � I � 1._2• (.г331 4 г.7t?} � i,'Сг} 1 i � .о&7 _.973'� Giar.ц -i9 =3 (.330) �.076) (�15о) C�:.-.bawr- _а5ё-9' ' � 'Э.-' f ' У ( ! I { s�.I.1 3S2_4 `?3. ��.+ � л . 1 (9.Зг) ' С. J 5.�q f -.13 I -д.-7' "�+:х_ 1 5,' � i5 4? еис о: Э сае_=.:i%а±:. цгл,�� � .Э`} ! �^ } Е (5.`r3) 1 ;�.kdl i � . 'гЭ.ь1• { {?.z: � :6.31 � • -9.1а _1�.�о..1 1_r;esr ' .зо 'I Sг ! c,L ._.� ` � { � � (а.г�.} (6.гг) I � ! j Lit aat ( t � � ' � j � г а_л±�: :., �_ тг 1 г � � t � I1 1 '3�• � .о:п ! у � �_. .з ! � `^ �.'.G) 1 (.7У) � .ОЭ�' ...с01 � lfапг j . (.з7) I .5'о :ЭЭ !] оиг а: 1'а _�at_.si� a. � ?1i siгes � г7м � .Я4• � � 'aC�:rac:aa � (-эЭ) (.ООд} { 1{ { : (.1Я1 :.а_} (.0,))• ' (.ОЗ} 1 (.и031' -`аааг 1 ,>7 ! ,2Э i 3 оиг ог' 1� соаЛ:.__� _., Чr1.S1ar3г. v^9 � �с3п { .2l'�• � ? :5а f � ` I1 1 . (.'_с1 t.."�2 � ' (. З} � .`�1J � .w_tг+т .д] Э1 a Ior_ 'ад _.:j � 1 (.041 (.^.0411 i � i ц_ 4iхаз (.:'.1i �.J.гt ` j ,.�,Э7'� { � г , 'i51 � ,��z2� ^ag=lzeвr{ .;ч � .3' , , c�tt ot 1< _,,•г:..,:�:i` .)_о• � [ i {г^/ уздду 1 а; ! - ... 1 (.з5�) i (.��', i � �,G:% f � К.. ') � ь.7г3}1 �iяasr , .� t _Э'_ аис уС i =а±::.�_ .х", ^ :дгдt•:а -.'ч_7� 1 ; l i � , ` 4."_io� i �. �.+ i � � .'�J�) ; ? � 01)" ' - _ ! _ � деггr .33 � -;? , � ��z.i,<: .ьг `' , i � .23' � � . � "'_�Э1 . , .:.х '1� � I �'�� ".` � ^ �°�^� 3-- � .�.�"''�" -- _D;�.ces:. .Э� . ._ _., -� ... ._.�-..-�,,..Т-а+��-�. �..-- _ ...�.�.а-..-�.� - .._..._-. _.. .. . , ..., ....... � , _... ,__�_ ._... .._.__.._�.�-.,..... �-.,F�. г у ^ t �. 89 REFERENCES Bardhan, P. (19T3). "Size, productivity, and returns to scale," Journal of-Political Economy, vol. 81. Bell, C. (19T2). "The acquisition of agricultural technology: its deter- minants and effects," journal of Develo-oment Studies. Billings, 14. H. and Singh, A. (1971). "The effect of technology on farm employment in India," Development Digest. Jose, A. V. (19T4). "Trends, in real wage rates for agriculturall lZborers," Economic and Political Weekly. Krishna, Rai (1974). "Measurement of direct and indirect employment effects of agr i cult ur .a, -rowth with technical change" in Employment in Developing Nations, edited by Edwa:rds, E. 0. Krishnaj-i, N. (1975). Paper published in Social Scientist (journal of Indian School of Social Sciences, Trivandrum). Lele, U. J. and Mellor, J. W. (1973). "Grov"uh linkages of the new foodgrain technology," Indian Journal of Agricultural Economics. Lockwood-, B. (1972). "Pattern of investment in farm machinery and equip- ment," Economic and Political Weekly. Parthasa'rathy, G. and Prasad, D. S. (L974). "Res-ponse to and impact o'L* HYV rice according to land size and tenure in a delta village, Andhra Pradesh," The Develo-cinR Economies. Raj, K. N. (1959). "Droloyment and unemployment in the Indian economy," Economic Develonment and Cultural Chang-e-. Rao, C. H. Hanumantha (1975). Technological Chanae and Distribution of Gains in Indian Asriculture. Rodgers, G. B. (1975). "Nutritionally based wage determination in the low-income labor market," ILO: World Emplonent Pro--rairme,. Rudra, A. (1973). "Marginalist explanation for more intensive labor input in smaller farms: emi:)irical verification," Economic and Political Weekly. Sen, A. K. (1964). "Size of holdings and productivity," Economic Weekly. Vyas, V. S. (1976). "Structural change in agriculture and the small farm sector," Economic and Political Weekly. 90 Re-orts Census of India, 1961, 1971. Census of India (1974). Re-ort on Resurvey of Economic Questions. Directorate of Economics and Statistics (Ministry of Food and Agriculture, G01). Renorts or. Farm Management in India. Reports on First and Second Agricultural Labor Enquiry, a.nd Rural Labor Enauiry. Quarterly Bulletins of Statistics, issued by UP Government. NSS Reports on Landholdings Survey, 8th round and 26th round.