Paradigms in the study of urban labor markets in LDCs : a reassessment in the light of an empirical survey in Bombay City

Paradigms in the Study of Urban Labor Markets in LDCs: A Reassessment in the Light of an Empirical Survey in Bombay City World Bank Staff Working Paper No. 366 SW P366 December 1979 Prepared by: Dipak Mazumdar Development Economics Department Development Policy Staff 79 I.W. 20433, U.S.A. i lterpretations in this document are those of the author ee attributed to the World Bank, to its affiliated to any individual acting in their behalf.   The views and interpretations in this document are those of the 3 1 author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. V4- C ,V WORLD BANK Staff Working Paper No.366 December 1979 PARADIGMS IN THE STUDY OF URBAN LABOR MARKETS IN LDCs: A REASSESSMENT IN THE LIGHT OF AN EMPIRICAL SURVEY IN BOMBAY CITY The paper analyzes the differences in the monthly earnings of workers in three sub-sectors of the Bombay labor market. A hypo- thesis is developed to explain the very large net intersectoral differences observed in the proportions of temporary migrants, and in the supply prices of temporary and permanent migrants. The evidence presented on labor turnover and migration is consis- tent with the hypothesis. It suggests a view of the process of migration and urban wage determination that is different from the current Harris-Todaro paradigm. The study was financed by the Research Committee of the World Bank (RPO 670-245). i1I'NTlATiOEL MC.NTY FUND Prepared by: Dipak Mazumdar Development Economics Department Development Policy Staff Copyright ( 1979 The World Bank 1818 H Street, N.W. Washington, D.C. 20433 U.S.A.   'CONTENTS Page I. THE CURRENT PARADIGM AND ITS ALTERNATIVES ... 1 A. A Revised Hypothesis .................... 7 II. THE BOMBAY LABOR MARKET AND A DESCRIPTION OF THE SURVEY SAMPLE ..................... 10 A. Structure of the Bombay Labor Market ..... 10 B. Changes in the Composition of Employment, 1961-1971 .......... . . . .. . . . . 12 C. The Survey Sample .......*................ 14 D. The Labor Market Problem in Bombay City.. 15 III. EARNINGS DIFFERENCES BETWEEN SECTORS ........ 19 IV. DIFFERENCES IN THE INCIDENCE OF TEMPORARY MIGRATION ........... ........... 25 A. The Importance of Return Migration ...... 29 B. Return Migration, "Graduation", and the Harris-Todaro Paradigm ............ 33 V. "TEMPORARINESS" OF MIGRATION, EARNINGS AND FAMILY SIZE .......................... 37 VI. CONCLUSIONS ........ . . . . ................................ . 43 ANNEX: Determinants of Monthly Earnings of Employees ..................... 45 REFERENCES ........ . . .. . . . .......................... . . . 48 LIST OF TABLES 1 Structure of Employment, Greater Bombay, 1961. 11 Percentage Rise in Employment Between 1961 and 1971 ............................... . 13 3 Percentage of Migrants with Little Waiting Before First Job ....................... 16 4 Distribution of Migrants by Duration of Unemployment .............................. 17  Page 5 Percentage Distribution of Sample Workers by Earnings Group ............... 19 6 Percentage Distribution of Sample By Age Group ....o.o..................... 21 7 Percentage Distribution of Sample By Education ........................... 22 8 Percentage Distribution of Sample By Duration of Residence in the City .... 26 9 Proportions of Child Migrants and Recent Migrants .o.o.*..*.......e..e..... 27 10 Percentage Distribution of Workers By Duration of Current Job .............. 29 11 Migrants: Mean Number of Jobs Changed in Urban Market ......................... 30 12 Percentage Distribution of Migrants by Years of Residence in Greater Bombay .... 32 13 Proportions of Natives and Migrants Among Those with Recently Acquired Jobs.. 35 14 Average Size of Family in Bombay .......... 40 Al Differences in Log of Earnings ............ 47  Chapter I: THE CURRENT PARADIGM AND ITS ALTERNATIVES 1/ In the study of the economic development of LDCs the analysis of the expansion of urban employment and the attendant process of internal migration is of central importance. Economic analysis and policy in the fifties were dominated by the paradigms proposed by Ragnar Nurkse and W. Arthur Lewis which draw largely rosy conclusions from the possibility of an "unlimited' supply of labor from the rural to the urban areas at a more or less constant real wage. Surplus labor in traditional agriculture in "disguised unemployment" was expected to create favorable conditions for the rapid accumulation of capital through the reinvestment of profits in the modern sector of LDCs, if only the state or the "capitalists' took advantage of this source of concealed savings. No prior act of savings was necessary through a lowering of real wages, as the enclosures might conceivably have brought about during the industrial revolution in Britain.Z/The discussions of the 1950s did not pay particular attention to problems which might develop in the urban labor market. The rural-urban wage gap was not expected to be large (Lewis talked of 30% higher real wages in the urban areas as normal), and unemployment or underemployment in the urbaai areas could not be substantial: migrants who in the urban areas could not secure an income at least as much as they enjoyed in the rural areas by sharing in the family pot would return to the farms even if their marginal contribution to the rural family income was very. small. 1/ The views expressed in this paper are of the author and not of the World Bank. My thanks are due to Lalit Deshpande of Bombay Univer- sity who conducted the field survey with the help of a large staff of interviewers over a period of two years. He is, however, not in any way responsible for the conclusions reached in this paper. The full report on the research is expected to be published jointly by the World Bank and Bombay University in due course. An earlier version of this paper was presented at the Forty-Eighth Annual Conference of the Southern Economic Association, Washington D.C., in November 1978. Thanks are due to Deepak Lal, Mark Leiserson, Lyn Squire and participants at a seminar at the Institute of Social Studies at the Hague for helpful comments. 3/ One school of thought contends thAt when the enclQoure imyeRmeit transformed land ownershi from a :ommnal to an tnrtyidual a those with common rights to land had 1ittJL& alternative but to - sell out to their more substantial neigfhbors and join the ranks of the landless.  -2- In the late sixties the paradigm of the urban labor market began to change as observers became conscious of high rates of open unemployment in many urban labor markets and urban poverty showed itself no less oppressive than rural poverty. In the works of Frank (68), Todaro (69) and Harris and Todaro (70), 1/ the idea that rural-urban migration just meets the urban demand for labor was superseded by the concept that migration is excessive. There are three distinct elements in this new paradigm: - There is a large wage gap, maintained by institutional factors, between the rural sector and that part of the urban labor market which is dominated by factories or by public employment (variously termed the "formal", the "organized", or the "ILO sector"). - People migrate in response to expected income, rather than the income they are actually likely to earn, so that they are prepared to accept a period of unemploy- ment in the hope of getting a high wage job. Outside the "formal" sector, urban wages subject to competitive market forces can go down to significantly lower levels, and indeed are pushed down below the supply price of migrants as they crowd in to queue for formal sector jobs. Such urban underemployment exists side by side with open unemployment. I have called this view of migration and the urban labor market a paradigm because although the ideas have been summarized in simple mathematical models, and sometimes the models have pro- duced some predictions about the urban labor market that are not intuitively obvious, 2/ they have not yet been tested empirically 1/ References are liited at the end of the naDer. 2/ Some examples are the following. Urban job creation may actually lead to a higher unemployment rate, cf. Todaro (76), revised by Blomqvist (78). Increases in employment and wages in the formal sector may lead to a decline in average earnings over time in the informal sector - Mazumdar (74). For the way induced migration affects the shadow wage rate in project evaluation, see Harris and Todaro, Mazumdar (76), Selowsky (78).  in any convincing way, and indeed it is likely that the formulations so far devised are too simple to be tested. 1/ Todaro, in his recent survey of the empirical testing of this group of ideas 2/ was able to point only to studies of "macro" migration functions, rather than studies of individual migrants' behavior, as lending support to the idea that migrants respond to expected rather than actual income differentials. Such functions only demonstrate that the rate of migration from a particular area of origin is sensitive not only to the wage level, but also to employment conditions, in the des- tination point 3/--relationships that are intuitively obvious and have been well documented in the inter-regional migration functions of various countries. 4/ Such observations of behavior in the aggregate do little to explain the decision of the individual migrant to accept unemployment, or an economic loss in a job paying less than he could earn at home, with the hope of ultimately breaking into the high wage sector. We shall return to this point later in this paper, but suffice it to say here that the length of time for which an individual migrant is prepared to accept such loss is a crucial variable, to be examined both theoretically and empiri- cally if the idea behind the Harris-Todaro type of migration function is to be properly tested. 1/ Little attempt seems to have been made in most formulations to go beyond highly simplified viewaof the factors influencing migrants' decisions, and particularly their expectations about urban employ- ment. 2/ Todaro (76), pp. 67-74. 3/ As measured, for example, by Sabot, by job openings in a specified time interval as a proportion of the number unemployed. 4/ See a survey of the migration literature by Greenwood (75).  -4- Before one comes to the migration function, and the urban unemployment and/or underemployment it might generate, the initial postulate that wages in the urban formal sector are artificially high must be studied carefully. It is not enough to observe the apparent difference between the going unskilled wage in the factory or public sector and the wages prevalent in urban domestic service or in agriculture. One serious deficiency of both the Lewis and the Harris-Todaro paradigms is that they treat labor supplied to the urban market as of homogeneous quality. Yet peasant workers attracted to cities are likely to differ widely with respect to a whole list of variables--for example whether or not they have families to support in the city, how long they intend to stay in the city, or in a given type of work, their responsiveness to urban work, and many others. The differential in wages one observes between the urban formal and informal sectors may in fact reflect differences in labor quality rather than the effect of institutions. In analyzing these wage differences it is possible, with moderate success, to allow for differences in standard human capital variables like education or experience. Such variables have been shown to have a considerable influence on wage rates. in LDCs, and perhaps a greater influence than in the industrial economies. Other indicators of labor quality, even when measurable, are less easily tractable in earnings function types of analysis, but may be extremely important. We shall discuss this point more concretely later in the paper. Empirically it is difficult to quantify wage differences between sectors, if any, for labor of equivalent quality. But when the apparent differential is large, and institutional factors in wage determination seem to be strong, care should be taken to see if differences in labor quality (other than the education-experience variables which are commonly controlled for) are not responsible for a substantial part of the wage difference. Though wages  - 5 - in the formal sector may appear to be determined by institutions, employers can always, to some extent, adjust the quality of labor they employ in order to minimize labor costs if wages are altered substantially by institutional forces. 1/ Indeed, institutional factors may often do no more than hasten changes in labor quality standards that employers are already introducing. One of the most important distinguishing features among rural-urban migrants in LDCs is the type of migration. 2/ In particular, a distinction has to be made between individual migrants, who live in the urban areas without their families, and migrants who bring up a family in town. Note that the distinction is not between migrants who migrate with or without their families, but between migrants with different family compositions in the urban area during their urban residence. In many LDCs a large proportion of male migrants live on their own, or with similar migrants in all-male households. Such households exist side by side with families. 1/ Upgrading and downgrading jobs in response to changes in labor market conditions, rather than altering the pay associated with the job, has been recognized as a widely practiced method of market adjustment, cf. Reder (55). The increase in labor pro- ductivity in response to minimum wage legislation, without a change in production techniques, is dramatically documented in a study of Puerto Rico by Gregory and Reynolds (65). 2/ The differences between migrant types with respect to the permanence of their attachment to the urban labor market have been stressed for a long time in African labor history. One of the mainsprings of the effort to increase wages in the formal sector of Africa through minimum wage legislation has been the desire to stabilize migrating labor which was said to have been a drag on African productivity.  The importance of this distinction between the two types of migrants is that they differ in their supply prices as well as in their desirability for urban employers. The difference in supply prices means that migrants with families are willing to accept employment only at a higher urban wage than individual migrants. The major reasons for the difference in supply prices of the two types of migrants are: (a) the substantially higher housing costs for the family migrant in town; (b) the lower earning strength of the family in town, because women and children are less easily part of the urban labor force than they are of the rural; and (c) the existence of disguised unemployment on family farms,which could make the marginal contribution of an individual worker low relative to the average income per earner of the family. / The reason for the higher demand price of family migrants is that they are settled in the urban environment, and likely to be more stable; the relationship between stability and productivity of labor is well documented in the literature. 2/ Certain types of employers will pay the higher wage that will attract family migrants, because the higher productivity of such workers will more than offset their higher-wage cost.. 3/ -1/ Th.e family gains anh moiiiR equal to thi 'average ncome -Becaums thb departing migrant no longer shares in th- family pot, and tt loe a smaller amount equal to the'latter's marginal Poduct; .Thz0 ,AX individual wanting to help his family would be- willing to migrAte at a very low supply price,' perhaps one 'tha't will merely cover his subsistence cost in town. This is esp'ecially releVAnt tfor people who migrate for only a part of'the year (seasonal migration} or o part of their working life'(periodic migrationi-, In African a9Zri- culture'the division of labor'between the sexes means that Ab.e- bodied males, who are'responsible'largely for thi heavy work. of felling trees and clearing bushes,needed about every two years, could be absent for two years at a time with6ut affectin5 the productivity of the family farm. 2/ The economic cost of non-stable labor has been widely studied for African urban labor, based as it was on the migrant labor system. An early and careful study of a Natal Dunlop Factory came to the conclusion that, because of the migratory tendencies of native workers, their productivity came to only 29% of that of European workers, while the potential productivity of a stable native worker was as high as 81% of the European worker's--University of Natal(50), Chapter V. 3/ It is necessary fr this argument only that the bulk of stable labor be supplied by family migrants. Some individual migrants may indeed be stable without affecting the reasoning.  The relationship between stability and productivity need not be equally strong in all sectors of the urban labor market. In the formal factory sector we could expect employers to profit considerably from the existence of a stable labor force. But in many parts of the informal sector--in small shops or eating houses for example--productivity increase with stability of labor is likely to be minimal. Under competitive conditions, then, we would expect wage differentials to emerge in the urban labor market with the higher wage sectors attracting more stable family migrants. A. A Revised Hypothesis I would thus like to suggest a revision of the current paradigm of the urban labor market stemming from the Harris-Todaro work, on each of the three points noted earlier: - A wage gap exists between the formal urban sector and the rural areas,with formal urban sector wages reflecting the cost of employing migrant workers accompanied by their families, who, like native urban workers, are more stable and hence more productive than other migrants. Institutional factors affecting wages may push up wages beyond this level, but not necessarily so, and certainly not by as much as a cursory glance at earnings data for individual workers would suggest. - There are,different types of migrants attracted from the rural to the urban areas, and the analysis of equilibrium in the urban labor market will be incomplete if we leave out of account the different supply prices of these dif- ferent types. In particular, the supply price of indivi- dual migrants will be substantially lower than that of family migrants.  - 8 - - The difference in wages between the formal and other sectors of the urban labor market will be related to the difference in supply prices of family and individual migrants. The high-wage formal sector will have a work- force which contains a larger proportion of family migrants, who tend to be more permanent, than the non- formal sectors in which the lower wages will attract predominantly individual migrants of lesser stabilityo The revised hypothesis provides an explanation of wage differentials--between the rural and urban markets and between different sectors of the urban market. It does not--as the Harris- Todaro paradigm purports to do--contain any prediction of excessive migration or urban unemployment. But then we have yet to see empi- rical work which demonstrates that urban unemployment rates are extremely high except for particular groups--mainly educated labor-- for which specific analysis and diagnosis are called for, i/ I have used the framework of this hypothesis, with its stress on different migrant types, as the basis for the analysis of wage determination in the textile industry of Bombay I/ before the Second World War. My attempt was to provide explanations both for the rural- urban wage gap and for earnings differentials between casuals and per- manents employed in the industry. Thus the hypothesis is not so much a new proposal as an attempt to introduce into the analysis of urban labor markets in LDCs earlier ideas which had been inexplicably neg- lected in the current literature. I was fortunate in obtaining the collaboration of Bombay University in undertaking a special survey of the Bombay labor market in the years 1972-74. The Survey answered the need to obtain information on different parts of the labor market--the formal as well as the informal--on a range of common topics. 1/ See, for example, Mark Blaug and Richard Layard, Grad'uate Unemployment in India London, 1970, and my own forEEcoming work on the problem of urban unemployment in Malaysia. 2/ See Mazumdar (59) for an early theoretical suggestion and Mazumdar (73).  -9- The Survey sample is described in Section II, after a brief review of the broad structure of the Bombay labor market. In Section III we begin the analysis of the material obtained from the Survey on the differences in earnings between sectors. Section IV sets out the profile of the workforce in each sector to highlight its composition, distinguishing migrants in terms of their duration of residence in the city and their period of work in their current job. Section V analyzes impermanence of migration in the different sectors in terms of the material presented in the previous sections, the question of waiting before the first job, and job changes in the urban labor market. It then proceeds to discuss family size in Bombay. The latter is a crucial variable in the argument linking the degree of permanence of migration on the one hand, and the supply price of migrants on the other. We conclude with a recapitulation of the main results. We should be careful to note that throughout the analysis all the results refer to male wage earners only. Female workers are in a small minority in the Bombay labor market. Our survey did collect information on females, but an analysis of this important topic is outside the scope of this paper.  - 10 - Chapter II: THE BOMBAY LABOR MARKET AND A DESCRIPTION OF THE SURVEY SAMPLE A. Structure of the Bombay Labor Market Heather and Vijay Joshi 1/ have given a breakdwon of the city's labor force between the organized and the unorganized sectors, classified by industry and the worker type, for the year 1961. They used the employment data collected by the Directorate of Employment and Training (which covered all public sector establishments and privately owned establishments employing 25 or more workers) to map out the limits of the "organized" sector. The difference between this set of figures and the Census returns for the same year gave the structure of employment in the "unorganized" sector. The Joshis note that this classification more or less identifies the organized sector with workers who are effectively covered by labor legislation, although theoretically the Factory Act extends beyond this limit to workers in establishments with no less than 10 employees, if they use electric power in their operations. Table 1 gives 'a breakdown of the employment structure in 1961 by broad categories as calculated by the Joshis. The unorganized sector, which provided about half of the employment in the city, accounted for a disproportionately large share of trade, commerce and services, although it is interesting to note that the 1/ Heather Joshi and Vijay Joshi, Surplus Labour and the City: A Study of Bombay-(Delhi, 1976)  Table 1: STRUCTURE OF EMPLOYMENT, GREATER BOMBAY, 1961 (Percentages of the Total Workforce) Organized Unorganized Sector Industry Employees Total Employees Others Employers Total Manufacturing 28 12 7 5 2 43 (including Primary Production) Construction and Utilities 3 1 * 1 * 4 Trade and Commerce 3 12 5 7 3 18 Transport and Communications 6 5 3 2 * 11 Services 9 14 10 4 * 24 Total 50 45 26 19 5 100 Note: The total workforce of Bombay in 1961 was 1.687 million. An asterisk means less than 1%. Some totals do not add up exactly due to rounding. Source: Joshi and Joshi, op. cit., Table III.2, pp. 54-5.  - 12 - manufacturing unorganized sector was also quite large. A little less than half of the workers in the unorganized sector were non-employees. These were single workers, family workers and workers in household industry. The category of single workers should be explained. The Bombay labor market, then as now, makes use of a large number of workers employed on a daily contract, as and when work becomes available, much like dock labor in industrial countries. These workers are not attached to any particular enterprise but they do show a significant amount of occupational and industrial specialization. There are well known areas in the city, almost like market places, where such workers assemble each day and are hired on a casual day-to-day basis. Such casual labor is a substan- tial part of the non-employee labor force of the unorganized sector; the figure given for 1961 in the table probably under- estimates the share of the unorganized sector in the total employment in Bombay at that time. This is likely because casual laborers working in the manufacturing sector would have been included in the returns on total employment obtained by the Directorate of Industry from the individual enterprises. In the textile industry such labor at this date could have accounted for as much as 20-25% of the work force. B. Changes in the Composition of Employment, 1961-1971 Our survey of the labor market was undertaken in 1973-74. Joshi and Joshi have provided estimates of the growth of employment by sector for the intercensal decade 1961-1971. Their figures are given in Table 2.  - 13 - Table 2: PERCENTAGE RISE IN EMPLOYMENT BETWEEN 1961 AND 1971 Greater Greater Bombay and Bombay Thana Urban Districts Population 43.8 46.1 Workers: Census 30.3 32.8 Workers: Organized 25.8 34.4 Manufacturing: Census 35.0 39.9 Manufacturing: Organized 22.3 35.e Souree: Joshi and Joshi, op.cit., Table III.3, pp. 58-59. It should be noted that the Census of 1971 used a different, and in effect more conservative, definition of workers from that used in 1961. Thus, using the 1961 definition, the Census figure for workers for 1971 is an underestimate; Joshi and Joshi suggest of the order of 4 percent. The "organized" sector figures given in Table 2 were obtained from the returns of the Directorate of Industry and are not affected by this change in definition. Thus, allowing for the change in census definitions, the growth of organized sector employment in Bombay lagged behind that of total employment by perhaps 1 percent per annum; the lag was greater for the manufacturing sector of Greater Bombay. However, Table 2 also shows that if we include the expansion of employment in the satellite towns of Thana urban districts the lag in the organized sector growth rate is substantially less.  - 14 - Another aspect of the organized sector employment growth needs to be noted. It is the major change in the industrial composition of employment in Bombay City during the decade. Employment in cotton mills, which accounted for almost one-half of total manufacturing employment in the organized sector of the city in 1961, remained virtually stagnant throughout the decade. The increase in employment in organized industry came largely from very high rates of growth in newer industries--particularly the more skill- intensive industries like petro-chemicals, pharmaceuticals and electronics. C. The Survey 'Samp'le It was decided to cover three sectors of the market: (a) the single workers and the casuals as we have described them above; (b) the workers in the small-scale sector; and (c) the -workers in factories. We also decided to exclude from the sample owner/workers and family workers who constitute an important proportion of the city's "informal" sector of workers. The study was deliberately limited to wage earners. The sample framework was provided by lists of esta- blishments in the city--for factories, the list maintained by the Chief Inspector of-Factories, and for small establishments, the registers maintained by the Municipal Ward Offices. The sample size of workers to be interviewed was predetermined for each of the three sectors of the labor market, and the total for each sector was distributed among establishments belonging to different industrial groups, so as to ensure that for each sector of the market the sample of establishments reflected the industrial distribution of the wage earners. For  - 15 - each of the factories.selected, the quota of workers to be interviewed was selected at random from the employee rolls, while for the small-scale sector, whose average employment was 2.9 workers per establishment, all the wage earners in each of the sample establishments were surveyed. The size of the sample was about 2,700 workers in factories and 2,000 workers in small-scale units. The casual workers are not attached to any esta- blishments. Thus we do not have a sample frame for an establishment-based survey. It was decided to survey 1,100 casual workers randomly selected from those who gathered for work in the well-known market places for casual workers, but taking care to ensure that the industrial composition of the sample reflected the industrial distribution of the population of single workers in the city, as derived from census data. D. The Labor Market Problem in Bombay City: Open Unemployment vs. Low Income The survey emphasized the differences in earnings between sectors, and inter alia, the causes of low incomes. Those totally unemployed at the date of the survey were not included. The casual workers by definition were not employed on a regular basis, and would be unemployed some days of the month, but the survey found only a small incidence of unemploy- ment among them. The average number of days worked per month by the sample of male casual workers was 23, giving an overall rate of underemployment of about 10 percent (assuming 26 days to be full employment for most workers).  - 16 - An aspect of unemployment captured for the entire sample was the period that migrants waited for their first job in the city. Migrants were asked whether they had a fixed job or firm offer of a job before they came. At another point they were asked about the length of unemployment before their first job in Bombayl at still anothe about who supported them during their period of waiting. The results for the sample are reported in Table 3 by sector. Table 3: PERCENTAGES OF MIGRANTS WITH LITTLE WAITING BEFORE FIRST JOB Migrated with No unemployment No need for support fixed job or before first because of no Sector firm offer job waiting Factory 21.2 32.2 45.2 Small-scale 24.9 36.3 49.8 Casual 13.8 20.0 Not available It is interesting to see how the proportion of migrants who said they had no long wait increased as the questions were phrased more loosely. In both the factory and small-scale sectors, it appears nearly half the sample of migrants more or less sailed into their first urban job. Waiting was clearly more common among casual workers.  - 17 - The distribution of migrants sampled by their reported period of actual unemployment before the first job is given in Table 4. Table 4: DISTRIBUTION OF MIGRANTS BY DWRATION OF UNEMPLOYMENT BEFORE FIRST JOB (percentages) Se /orDa i Up to 11- 31- 91- 181- 1 Set Dy i 10 30 90 180 365 year Factory 32.2 7.6 16.7 14.5 13.1 10.9. 4.8 Small-Scale 36.3 8.8 18.9 15.7 8.2 8.3 3.7 Casual 20.0 28.0 30.2 13.0 6.3 2.5 1.2 It is clear that those who were unemployed for a reasonably long period, say more than 3 months, 1/ were in a minority. It is also interesting to note that this proportion was higher in the factory sector than in the other sectors-- which is consistent with the model of queueing up for high wage jobs. In the absence of a model specified in detail it is difficult to assess the prevalence of long periods of unemploy- ment. Intuitively, a mere 15% of the sample waiting more than 6 months does not appear to be an excessive proportion in a normally functioning labor market,without a preponderance of Harris-Todaro gamblers. None of the migrants financed themselves,from savings or borrowings, while unemployed. Of those who needed help, 50% said they received support from their family members--and rather more did so in the factory sector, perhaps in reflection of the higher income of the relatives who helped them. Even the 1/ We checked to see if these figures were affected by excluding from the sample those who had arrived in the city as children. They were not.  - 18 - two other types of helpers who were mentioned in significant numbers in the response of the workers to the questionnaire-- viz. "caste fellows" and "native persons" (from the same area of origin)-typically did not accept any payment for their support. We conclude that there is little evidence of the type of "investment in unemployment" for the fresh migrant to the urban market which has been stressed in the Harris-Todaro paradigm. Since much of urban unemployment in their view is ascribed to this phenomenon, it would appear that this particular problem is not important in the Bombay labor market. On the other hand, as we shall see0 the differences in earnings in different parts of the market are significant.  - 19 - Chapter III: EARNINGS DIFFERENCES BETWEEN SECTORS We begin our exploration of the working of the Bombay labor market and the characteristics of the three major sectors distinguished by analyzing intersectoral differences in earnings. The distribution of earnings of male workers by broad monthly earnings groups for each sector is given in Table 5. Table 5: PERCENTAGE DISTRIBUTION OF SAMPLE WORKERS BY EARNINGS GROUP / Earnings p Ser tor Casual Ea1- Factory Moknth (RS.)/ Under 100 12.0 2.9 0.1 100 - 200 51.0 36.3 3.0 201 - 300 30.8 31.3 9.8 301 - 400 4.9 14.5 20.4 401 - 500 0.9 7.8 32.6 501 - 600 0.3 2.8 17.7 601 - 700 - 2.2 8.5 701+ 0.1 1.2 7.9 Number of observations 900 1972 2760 Mean Earnings (Rs./m,nth) 183.7 270.3 470.6 1/ Earnings are monthly earnings which take account of kind payments, overtime and bonus as well as the days worked in each sector.  - 20 - The earnings differentials between the three sectors are very large--whether we use the mean, the median or the mode. Factory workers earn 75% more than those in the small-scale sector, and two and a half times as much as those in casual work. Of course, a significant part of the earnings differences will reflect differences in workers' characteristics, including the standard human capital factors. Our task in this section is to assess the extent of the differential after we have controlled for the effects of these variables. This preliminary sorting should also show whether certain types of workers--e.g., of a particular educational range--are concentrated in particular sectors. At this stage the degree of permanence of migration is left out of account. Stepwise multiple regression models were run with earnings as the dependent variable for each sector separately. The explanatory variables used could be giouped as follows: - Standard human capital variables--age; education; occupation; period of training; knowledge of English. - Personal attributes of the worker--migrant or native; marital status; whether the worker was a principal or secondary earner. - Background variables which are said to influence earnings in a segmented labor market--caste; religion; mother tongue; and parental occupation and education. The first group of factors was the most pervasive in its influence. For casual workers education seemed to have rather little influence on earnings, compared to the other factors, but this partly reflects the limited variation of educational levels among workers in this sector. Tables 6 and 7 give the percentage distribution of the sample by age and education in each sector.  - 21 - Table 6: PERCENTAGE DISTRIBUTION OF SAMPLE BY AGE GROUP Age - Casual Small-scale Factory GroJe Sector 19 15.3 14.4 0.8 20-24 30.4 29.2 11.0 25-29 21.7 19.4 15.9 30-39 21.0 19.7 34.2 40-49 8.0 10.5 25.1 50 + 3.'6 v 6.8 13.0 Mean age 27.6 29.2 36.9 Though the age structures of the casual and small- scale sectors are very similar, the working population in the factory sector is substantially older. By contrast, in educational attainment workers in the small-scale sector are very much like those in the factories, but the casual workers have a distinctly lower educational level--concentrated below grade 6, or primary school graduation.  - 22 - Table7 : PERCENTAGE DISTRIBUTION OF SAMPLE BY EDUCkTION Educational Level/Setct Cas.ual Small-scale Factory Illiterate 31.8 14.7 19.2 Grades 1-3* 12.9 6.3 9.1 Grades 4-6 29.3 20.6 23.3 Grades 7-9 15.2 21.7 18.9 Grades 10-11 10.3 27.9 25.3 uiter-schoo1 Certificate, 0.5 8.8 4.2 Dipldas a te t tlet *hic