Poverty and unemployment in India : an analysis of recent evidence

Poverty and Unemiployment in India: An Analysis of Recent Evidence SWP417 World Bank Staff Working Paper No. 417 October 1980 E1J PUB HG \ , 3881.5 .W57 W67 no. 417   ihe views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. WORLD BANK Staff Working Paper No. 417 October 1980 POVERTY AND UNEMPLOYMENT IN INDIA: AN ANALYSIS OF RECENT EVIDENCE This paper examines the association between poverty and unemployment using the data collected by the Indian National Sample Survey from October 1972 to September 1973. Contrary to the widely accepted view, supported even in the ILO reports, that the poor are too poor to remain unemployed, this paper high- lights a clear association between unemployment and poverty (as measured by the per capita expenditure of households). Poverty is, however, more wide- spread than unemployment. A careful analysis of the data for two states of India (Gujarat and Maharashtra) shows that according to both the conventional definition and a more comprehensive measure (designed to capture involuntary idleness within the reference week) , the inc:idence of unemployment was markedly higher among those whose "usual activity" was casual labour. The self-employed (including employers) and family helpers reported a much less-than-average level of unemplovment or underemployment. The casual labourers were over-represented among the bottom deciles in terms of their per capita expenditure; they reported a lower-than- average level of literacy and education, and were generally younger than other workers. Those in rural areas were predominantly agricultural labourers with no land or very small land holdings. Two distinct groups among urban casual labourers were agricultural labourers and miscellaneous workers in the "not elsewhere classified" category. The incidence of unemployment (in terms of persondays) in 17 major states of India shows a statistically significant, positive association with the percentage of casual labourers in the rural labour force and in the urban male labour force. In ot:her words, the importance of casual labourers in the labour force seems to explain a good part of the interstate variance in unemploy- ment rates. It is suggested that such differences in the structure or composition of the labour force might: explain some of the differences in unemployment rates reported for different countries. Also, with an increase in the share of casual labour in the labour force, unemployment rates estimated through surveys can be expected to rise in India. Prepared by: Pravin Visaria Development: Research Center Development Policy Staff  Copyright 0 1980 International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. First printing October 1980 Second printing June 1984 All rights reserved Manufactured in the United States of America  (i) Abstract This paper examines the association between poverty and unemployment using the data collected by the Indian National Sample Survey from October 1972-September 1973. Contrary to the widely accepted view, supported even in the ILO reports, that the poor are too poor to remain unemployed, this paper highlights a clear association between unemployment and poverty (as measured by the per capita expenditure of households). Poverty is, however, more widespread than unemployment. A careful analysis of the data for two states of India (Gujarat and Maharashtra) shows that according to both the conventional definition and a more comprehensive measure (designed to capture invol- untary idleness within the reference week), the incidence of unemploy- ment was markedly higher among those whose "usual activity" was casual labour. The self-employed (including employers) and family helpers reported a much less-than-average level of unemployment or underemploy- ment. The casual laborers were over-represented among the bottom de- ciles in terms of their per capita expenditure; they reported a lower- than-average level of literacy and education, and were generally younger than other workers. Those in rural areas were predominantly agricul- tural labourers with no land or very small land holdings. Two distinct  (ii) groups among urban casual labourers were agricultural labourers and miscellaneous workers in the "not elsewhere classified" category. The incidence of unemployment (in terms of persondays) in 17 major states of India shows a statistically significant, positive association with the percentage of casual labourers in the rural labour force and in the urban male labour force. In other words, the impor- tance of casual labourers in the labour force seems to explain a good part of the interstate variance in unemployment rates. It is suggested that such differences in the structure or composition of the labour force might explain some of the differences in unemployment rates re- ported for different countries. Also, with an increase in the share of casual labour in the labour force, unemployment rates estimated through surveys can be expected to rise in India.  (iii) Acknowledgements Thanks are due to Professor V. M. Dandekar, Chairman of the Governing CounciL of the National Sample Survey Organization, and to P. B. Buch, M. A. Telang and S. M. Vidwans, former or current Direc- tors of the Bureaus of Economics and Statistics in Gujarat and Maharashtra, for providing me with an opportunity to study the 27th Round state sample data for the two states analysed in this paper. Several basic tables used for the preparat:ion of this paper were compiled at the Gujarat computer Centre in Gandhinagar, India, with the cooperation of a large staff working under the leadership of P. B. Buch and K. B. Trivedi. Others who have helped in the statistical work include R. Murti Pemmarazu, Shyamalendu Pal, Robert E. Sterrett, Jr. and Cynthia Hwa. I have greatly benefitted from many discussions with Montek S. Ahluwalia, B. S. Minhas and T. N. Srinivasan at various stages of the preparation of this paper. Others, whose advice was particularly useful during the earlier stages of preparation of data for analysis, include Professor M. L. Dantwala, my teacher and colleague at the University of Bombay, and Sudhir Bhattacharyya of the N.S.S. Organization. Thanks are due also to Balu Bumb, Roger Grawe, R. K. Hazari, Mark Leiserson, Dipak Mazumdar, V. N. Rajgopalan, V. V. Bhanoji Rao and Suresh Tendulkar for their comments and suggestions on an earlier draft of this paper. None of them bears any responsibility for the errors that might remain.   Introduction The extent of association between poverty and unemployment in the developing countries is often a subject of considerable debate. The origin of the debate can be traced to a scepticism regarding the validity of the low levels of unemployment reported by successive labour force surveys in developing countries. Some analysts argue that the poor are too poor to remain unemployed and that their low levels of income compel them to work irrespective of the level of reward.-/ Fur- thermore, many of those few who get classified as unemployed in the labour force surveys are believed to be the better-off. This view questions the validity oE the now-conventional concepts of unemployment and labour force that have been developed in the western countries after the Depression of the 1930's and have since been adopted and propagated by international agencies like the International Labour Office in developing countries. This paper seeks to demonstrate that with some adaptation, the conventional approach to the measurement of unemployment can provide better estimates of the underutilization of labour time available in a country; with proper tabulation and analysis, the data show a clear association between poverty and unemployment in India, although poverty 1/ For example, a recent survey article reports that ". . . the openly unemployed are not primarily drawn from . . . low-income groups; on the contrary they are disproportionately from relatively high income families, and are well educated." See A. Berry and R. H. Sabot, "Labour market performance in developing countries: A survey," World Development, vol. 6, p. 1211.  - 2 - is certainly more widespread than unemployment. Empirical evidence is also provided supporting an earlier hypothesis that an important explanation for low levels of unemployment reported in the labour force surveys of several developing countries (such as India) is the status distribution of workers, particularly the predominance of the self-employed and family helpers among workers. The first part of this paper briefly outlines the main features of the approach to the measurement of unemployment adopted in India during the past decade in two nationwide surveys which were conducted in 1972-73 (from October to September) and 1977-78 (from July to June) as part of the 27th and 32nd Rounds of the National Sample Survey. Since very little of the information gathered in the 32nd Round has been released so far, the concepts and definitions used in the 27th Round provide the basis for discussion. The main body of the paper is devoted to an examination of the 27th Round data on unemployment for the "state samples"-/ canvas- sed by the states of Gujarat and Maharashtra in Western India. These data are used to examine not only (i) the extent of association between poverty and unemployment, but also (ii) the "usual" labour force and employment characteristics of those who were unemployed during the entire 1/ The state samples are selected and interviewed by investigators employed by the State Governments. The sampling and other proce- dures followed by them are the same as those used for the "central sample" selected and canvassed under the auspices of the National Sample Survey Organization.  - 3 - reference week (i.e., the week preceding the day of interview),-L/ and (iii) the share of different categories of workers in the total unutilized labour time (or persondays of unemployment). Finally, the last part of the paper examines the likely wider validity of the data from the states for the rest of India as well as for other countries of the developing world. 1/ The "usual" labour force and employment characteristics take into consideration: (a) whether or not the respondents were "usually" in the labour force; (b) if they were usually in the labour force, whether or not they were usually employed; and (c) if they were usually employed, which broad sector (farm or non-farm) of employ- ment, and the status or class of worker (i.e., self-employed, a regular wage or salaried employer, a regular wage or salaried employee, a casual employee or family helper). The usual activity categories which are used to classify the employed have been shown in several tables in the text.  - 4 - Alternate Methods of Measuring Unemployment The 27th and 32nd Rounds have utilized simultaneously three alternative methods of measuring unemployment. These methods include the definition of unemployment used in many countries of the world which classifies as unemployed a person without any work and seeking (or avail- able for) it during the entire reference week (i.e, the week preceding the day of interview).- The activity of the reference week is described as the current activity and those unemployed during the reference week are called the "currently unemployed." This method continues the approach used in the earlier rounds of the Indian National Sample Survey (since 1958-59) and permits comparison with past data as well as with other countries of the world. The resulting estimates can be called the incidence of "week-long" unemployment. In addition to the data on current activities, the respon- dents were also asked about their "usual" activities. The original intention of the group that designed the schedules was to identify usual activities as the "principal activity of the year preceding the day of interview" on the ground that the living standards of the household would be conditioned more by the "usual" activities of the previous year rather than by the activity of the reference week (which may be a transient seasonal activity).-/ The instructions to investigators 1/ Some persons may not actively seek work because of their impression or knowledge that work opportunities are not available. Therefore, the definition of unemployed has been broadened to include those who do not actively seek work but are available for it (at wage rates and under working conditions prevalent in the local situation). 2/ Some earlier rounds of the NSS, notably those conducted during 1958- 59 and 1959-60, had highlighted rather large seasonal variations in the level and patterns of economic activities.  - 5 - used a rather vague definition of a usual activity as one indicated by the "normal working pattern" of respondents, pursued "over a long period in the past" anct "likely to continue in the future.""I Essen- tially, this was an approximation of the "gainful worker" approach to the measurement of the economically active population. Persons reported to be unemployed in terms of their usual activities were likely to be "chronically unemployed," and an estimation of their number was consid- ered important.2/ Data on "usual" activities of respondents also provide some interesting information on what the currently unemplcyed persons usually do. A third estimate of unemployment was obtained by asking persons classified as "'employed" during the reference week (i.e., cur- rently employed) about their activities during each day of the reference week. These data often described as the time disposition data for the reference week, were recorded in units of half-days rather than hours worked.-/ They permit a. more comprehensive measure of unemployment, 1/ See: NSS Organization, National Sample Survey, 27th Round, October 1972-September 1973, Instructions to Field Staff, Vol. 1, Design, Concepts, Definitions and Procedures (New Delhi: September 1972), Para. 7.0.12 and 7.7.2. 2/ Questions on the duration of unemployment, if asked of those classified as "currently unemployed," can be used to identify the chronically unemployed. In the 27th Round, only the "usually unemployed" were asked about the duration of their unemployment. 3/ There is widespread scepticism among persons associated with the NSS about information on hours worked. The respondents generally report the average number of hours worked during the reference week (usually with a preference for even digits, e.g., two to four hours); the investigator selects a number and multiplies it by six or seven (depending on the reported days of work).  - 6 - including the reported underemployment of the "currently employed." Once again, we can estimate the persondays of unemployment according to the various usual activity categories of the labour force. A fourth method, that can be used only when the time dis- position data for the reference week are available, would abandon the priority rule used under the current activity approach which classified as employed a person who worked for even one day. This method, which I would call the major activity approach, would classify a person as unemployed, if the unemployed days in the reference week exceed the days when he is employed./ In discussing the data for Gujarat and Maharashtra, this fourth approach will be illustrated. 1/ Some marginal cases can arise if the number of days when a person is employed in a week equal the days when he is unemployed. In such cases, I would classify them as employed, although an opposite rule can also be used on the plausible ground that any implicit overstatement of the incidence of unemployment would only highlight a serious problem. The labour force status of an individual is uniform according to the priority criterion as well as the major activity criterion. Even if a person is not in the labour force for a major part of the week, the priority rule ensures his inclusion in the labour force.  - 7 - Estimates of IJnemployment According to Alternate Criteria For rural and urban areas of India as a whole, Table 1 shows the 27th Round 1'1972-73) estimates of the incidence of unemploy- ment according to three alternate approaches, as well as the labour force participation rates. The participation rates are almost the same according to both usual and current activity criteria but unemployment rates differ significantly. In rural areas, the usual activity criterion shows a very low incidence of unemployment, significantly elow the estimate based on the current activity criterion; in urban areas the differences between the two sets of estimates are relatively small.-/ It is not possible to estimate the participation rates in terms of person- days from the data published so far, but they would presumably be lower than those based on the other two criteria, insofar as many persons (particularly females) report themselves as outside the labour force on some days within the reference week. The incidence of unemployment is, however, the highest in terms of the persondays criterion; once again, the difference:; are larger for rural areas where one would a priori expect higher underemployment among agricultural workers than in urban areas. 1/ The reported incidence of usual and current unemployment overlaps to a considerable extent, particularly in urban areas of India. Table 1 in Annex L shows the distribution of the usually unemployed according to the duration of unemployment. Evidently, about 15 to 20 percent of i:hose reported as usually unemployed had been unem- ployed for up to t:hree months. They might be new entrants into the labour force. It is debatable whether they should be considered 'chronically unemployed" (as is often done in India.)  - 8 - Table 1 INCIDENCE OF UNEMPLOYMENT IN INDIA, ACCORDING TO ALTERNATIVE CRITERIA, 27TH ROUND OF THE NSS, OCTOBER 1972-SEPTEMBER 1973 UsuaZ Activity Current Activity Persondays Criterion Criterion Criterion Rural Urban Rural Urban Rural Urban Sex India India India India India India (A) Incidence of UneMpZoyment Males 1.16 4.79 3.03 5.97 6.75 7.99 Females 0.48 6.05 5.51 9.78 7.83 12.58 Persons 0.91 5.03 3.87 6.55 7.83 8.85 (B) Percentaqe of Unen7pZoyed* Males 0.75 2.87 1.94 3.55 - - Females 0.18 1.00 1.90 1.44 - - Persons 0.47 1.99 1.92 2.56 - - (C) Labour Force Participation Rates* Males 64.59 59.96 64.07 59.46 NA NA Females 37.71 16.53 34.50 15.68 NA NA Persons 51.40 39.60 49.58 39.07 NA NA Source: NSS, Draft Report No. 255/10, pp. 36, 129. * Figures are arithmetic averages of the estimates for four sub-rounds. They relate to persons aged five years and over. Percentages of unemployed according to usual and current activity criteria show unemployed as per- centage of population aged five years and over. Note: The 17th Round estimates of the incidence of unemployment (according to the current activity criterion) during July 1961-June 1962 were as follows: Males Females Persons Rural India 3.7 8.5 5.1 Urban India 3.0 3.3 3.1 NA: Not available - : Not applicable  - 9 - AsSccLaL3.3 retween Poverty and Unemployment The widely used index of poverty in India is the monthly per capita expenditure (MPCE) of a household, a proxy for income. The 27th Round data on unemployment have been tabulated separately for six discrete MPCE intervals; the lowest of these includes a very small per- centage of total househloids, but because of the very large size of the sample, the estimates for even this bottom group may be reasonably stable. Table 2 shows the incidence of unemployment in terms of the persondays approach for each MPCE group by subround. Figure 1 shows the same data graphically. A clear inverse association between MPCE and the incidence of unemployment is evident in each subround in both rural and urban areas of the country. For Gujarat and Maharashtra, Table 3 shows the incidence of unemployment in terms of usual and current activity data as well as persondays for different deciles of households.!/ While estimating unemployment in terms of persondays, the intermediate six deciles were condensed into three quintiles. 1/ The deciles are deciles of households (not of population), arrayed in an ascending order of their monthly per capita consumption expen- diture. Given the inverse association between average household size and the MPCE of households, the bottom deciles of households include more than 10 percent of the population, whereas the reverse is true of the upper deciles. Deciles of population would have different cut-off points; but the pattern(s) evident in the deciles of households would be seen in the deciles of population as well. Our data are based on the "state" samples. The broad comparability of these data with the "central" samples has been discussed at length and confirmed in the author's "Living Standards, Employment and Education in Western India, 1972--73," mimeographed, 1977.  - 10 - Table 2 INClOEfC' OF UP"YuL0Y=NT (?CRSO::0AYS LN'VLOYED -AS !EFCZNT OF ?ERSON:0AYS ".: LAl1OR rORC-) SY PER Ct2-.\ ('?CT-) ?.U-AL .:,D 1!:: t A:n '- 27' Ro ':D OF .,IE :;S.; OCTOBa^ L,72 - SF?v-:'I, '9,3 rRursL zrrda d80 UI Ir.i3 7 Oct. Z Share ot Oct. Z Snare oc Oct.- Jan.- Apr.- July- 1972- Households Cct.- Jan.- Apr.- July- 1972- Households Dec. Hlar. June Sept. Sept. Oct. 72- Dec. Mar. June Sept. SQpt. Oct. 72- HPC! Grouo (Rs.) 1972 1973 1973 1973 1973 Senc. 73 1972 1973 1973 1973 1973 Sent. 73 MALES =ES up to - 11.00 15.3 19.5 23.6 11.7 17.5 14.7 52.8 22.9 38.0 32.1 11.00 - 20.99 9.2 12.8 13.7 11.9 Ll.9 13.7 12.8 18.6 16.3 15.5 21.00 - 33.99 6.9 8.6 9.8 7.4 8.2 5. 9.5 10.6 10.8 12.5 10.9 -. 34.00 - 54.99 5.2 6.3 7.2 5.5 6.1 A. 8.0 9.7 9.4 9.9 9.3 A. 55.00 - 99.99 4.2 4.8 4.5 4.5 4.5 7.1 6.2 7.0 7.1. 6.9 100.00 & above 3.4 2.9 3.0 2.8 3.0 4.1 3.7 4.6 4.4 4.2 All 6.0 8.7 7.9 6.2 7.2 7.3 7.9 8.1 8.5 . 8.0 FFM4LES FMfAL0S up to - 11.00 33.8 26.8 29.9 27.1 29.4 17.5 30.6 22.2 32.9 25.8 11.00 - 20.99 14.8 18.6 20.2 17.2 17.7 18.8 17.6 17.9 13.7 17.0 21.00 - 33.99 9.8 13.1 17.4 12.0 13.1 S. 13.0 14.2 18.0 15.2 15.1 5. 34.00 - 54.99 7.8 9.7 12.4 8.4 9.6 A. 11.9 11.7 15.0 12.9 12.9 A. 55.00 - 99.99 6.2 7.4 8.0 6.3 7.0 12.6 13.3 15.7 12.7 13.6 100.00 & above 3.6 2.0 7.1 3.4 3.3 11.2 8.8 11.5 9.8 10.3 All 9.3 11.5 14.2 10.0 11.3 12.6 12.8 15.5 13.0 13.' PEPSONS PEAM(SIS up to - 11.00 23.9 22.1 26.1 17.6 22.4 0.7 15.8 42.5 22.7 35.8 29.2 0.3 11.00 - 20.99 11.3 15.0 16.1 14.0 14.1 9.8 15.1 14.4 18.3 15.6 15.9 4.5 21.00 - 33.99 7.9 10.1 12.2 8.9 9.8 30.1 10.4 11.5 12.5 13.2 11.9 19.2 34.00 - 54.99 6.0 7.3 8.7 6.4 7.1 35.2 8.8 10.1 10.5 10.5 10.0 27.4 55.00 - 99.99 4.8 5.5 5.5 5.0 5.2 19.0 7.9 7.2 8.2 S.2 7.9 28.3 100.00 S above 3.4 2.6 3.9 2.9 3.2 5.1 5.0 4.3 5.4 5.0 4.9 20.2 All 7.1 8.6 9.8 7.4 8.2 100.0 8.4 8.7 9.4 9.3 9.0 100.0 Note: The figures for the entire Round are un*eighted averages of the estinates for four subrounds. As a result. Chey are sli6hcly diffore-,t fron che widely quoted escinates reporting the incidence of une-ploynent in terms of persondays as 7.8 percent in rural India and 8.8 percent in urban India.  - 11 - Figure 1 INCIDENCE. OF UNEMPLOYMENT IN TERMS OF PERSONDAYS AMD MONTIHLY PER CAPITA EXPENDITURE OF HOUSEHOLDS RURAL AND URBAN INDIA, 1972-73 30 * to RURAL INDIA I20 I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 110 40 URBAN INDIA 30 & 20 120 <11.00 I.02-20.99 21.00-33.99 34.00-54.99 50.00-99.99 >100.00 _w- =___ SLs *3 thlt PU CoIu EIpdjtute 4.I AuaI R *....n....*4 Frm. -_ PAFeI  - 12 - Tabhle 3 IRCII)PENC D0V 1`Irrlfv1,rST ACCPrI'Ir TO AlTI:It NhTxVE CRITERIrA, nY SEX AND oF MONITlY 1.H CAPYTA FXN.N1Fl)HTPEz (nrctr), Ca2A11A1 AND I'M:Itt, Iq77-3 __________G_________ Ctp,-rrat Thavri7 Polrn'anlztsn Urban Maharanhtyra Thtra2 C!Li'arat2: LPr?'ns C,.t I rn .rftrom Uba.#Jn1f HPCR Docile htlos Fe.aien t'ernonu Hales Femaler lersons Malsc9 FcamJes 2'crasne Vales Fcmales Persoins (A). Usual Activity Data (Ln terms of personls) 1 0.9 0.5 0.7 3.6 3.1 3.5 1.1 - 0.6 6.1 1.2 4.6 2 0.4 0.9 0.6 5.0 4.2 4.8 0.5 0.1 0.3 4.5 1.7 3.8 3 0.9 0.4 0.7 5.4 - 4.4 0.3 - 0.1 5.0 3.2 4.5 4 0.9 0.2 0.6 4.1 4.1 4.1 0.5 - 0.3 5.1 3.5 4.8 5 1.0 - 0.6 3.2 3.2 3.2 1.2 - 0.6 6.2 7.2 6.3 6 0.2 0.5 0.3 3.6 5.2 3.9 0.6 - 0.3 4.6 0.9 5.1 7 0.7 - 0.4 3.1 4.9 3.3 0.7 0.3 0.5 3.9 4.2 3.9 t 0.6 0.6 0.6 3.7 2.8 3.6 1.3 - 0.7 3.4 8.2 3.9 9 0.7 - 0.4 3.6 4.6 3.7 0.4 - 0.2 2.8 4.7 3.0 10 1.1 0.4 0.9 2.6 19.4 4.7 0.3 0.4 0.4 0.9 4.1 1.4 All 0.7 0.4 0.6 3.8 4.3 3.9 0.7 0.1 0.4 4.4 3.8 4.3 (B) Current Activity DataC (in terms of persons) 1 3.6 4.1 3.8 5.7 6.6 5.9 4.3 10.9 7.2 8.6 6.0 7.8 2 2.2 3.3 2.6 6.0 4.7 5.7 3.2 6.2 4.5 5.8 3.1 5.2 3 2.0 2.4 2.2 6.3 7.7 6.6 2.3 5.3 3.7 5.4 5.2 3.4 4 1.9 2.7 1.3 .4,5 6.7 4.9 2.1 6.3 4,0 5.7 5.9 5.7 5 2.8 3.2 2.9 3.5 5.8 3.9 4.5 2.9 3.8 6.6 7.4 6.7 6 0.7 1.9 1.2 3.9 12.6 5.1 2.2 5.4 3.6 4.8 10.6 5.6 7 1.1 1.6 1.3 3.9 6.7 4.3 1.5 2.3 1.8 4.3 6.4 4.7 8 0.9 1.4 1.1 4.3 4.9 4.4 M32 3.5 3.3 3.5 7.4 3.9 9 1.2 2.1 1.5 4.2 5.8 4.4 2.2 3.0 2.5 2.9 6.6 3.3 10 2.6 1.4 2.2 2.7 17.4 4.5 1.5 1.5 1.5 1.1 4.5 1.6 All 1.9 2.5 2.2 4.6 7.3 5.1 2.7 5.0 3.7 5.0 6.0 5.2 (C) Time Disposition Data (in terms of perasn- days) 1 10.1 12.8 11.1 9.9 13.3 10.8 13.1 20.6 16.3 13.5 13.7 13.5 2 7.3 11.0 8.8 10.2 10.9 10.3 11.1 15.7 13.1 9.0 8.2 8.8 3-4 5.9 8.7 7.0 7.6 12.5 8.5 7.7 14.1 10.5 7.9 10.4 8.4 5-6 5.3 6.9 5.9 5.5 12.1 6.3 9.2 12.3 10.5 6.7 12.2 7.5 7-8 3.5 5.2 4.1 6.0 6.7 6.1 6.8 9.2 7.8 4.7 10.6 5.3 9 3.4 4.6 3.8 6.0 7.1 6.1 5.8 9.3 7.1 3.5 7.7 4.0 tO 3.9 4.2 4.0 3.3 21.5 5.3 3.2 5.9 4.0 1.3 4.6 1.8 All 5.5 7.9 6.4 6.9 11.7 7.7 8.2 12.8 10.0 6.8 10.5 7.4 In estimating these figures, priority is given to a person's classiflcation as employed; work for even a balf day during the refereice week is considered adequace to classify a person as employed. Persons classified as unemployed had not worked at all during the reference week; they could have been outside the labour forte during a part of the week.  - 13 - While the data on the incidence of unemployment in terms of usual or current activity do not show a clear consistent relationship with MPCE decile of households,l/ a more or less steady inverse relation- ship is evident between MPCE decile and the incidence of unemployment in terms of persondays. Except for some erratic deviations, the labour force in the bottom deciles of households clearly suffers from a signif- icantly higher incidence of unemployment and/or underemployment. The inter-decile variations in unemployment are, of course, much smaller than the differentiaAs in per capita expenditure. And although poverty is more widespread than unemployment, there is a clear association be- tween the two: The poor did report non-availability of opportunities for work to a considerably greater extent than the average level in the two states.-/ 1/ A statistically significant inverse relationship between average MPCE in different ,ieciles and the incidence of unemployment in terms of current activity criterion is evident for females in rural Gujarat, males as well as females in urban Gujarat, females in rural Maharashtra and males in urban Maharashtra. 2/ These findings differ from the earlier indications of little associ- ation between poverty and unemployment or even a direct relationship between MPCE and the incidence of unemployment. Among the nation- wide surveys, the ninth Round of the NSS, from subsamples 3 and 4 canvassed during August-November 1955 (a sample of about 8,250 households), had indicated a direct (or positive) relationship be- tween the incidence of unemployment and the MPCE, both in rural and urban areas of the country. However the number of per capita expenditure classes in the ninth Round was only four, and effectively three, because the top class accounted for less than one percent of the rural population and only three percent of the urban popula- tion. Further, labour force and employment status was determined in terms of the "unual status" of the respondents, which would show only chronic unemployment, and which fails to show a clear relation- ship between MPCE and the incidence of unemployment even in the data for Gujarat and Maharashtra in Table 3. See: Pravin Visaria, "Labour Force, Unemployment and Underemployment in India: Retro- spect and Prospect," in The Indian Economy: Performance and Prospects, ed. J. C. Sandesara (Bombay: University of Bombay, 1974). Prior to the 27th Round, only in the ninth Round were labour force charac- teristics tabulatecl according to per capita expenditure of the household.  - 14 - It has been argued that unemployment (again in terms of persondays in 1972-73), in different regions of rural India does not seem to be related to the productivity of land.l/ Indeed, in 56 regions of rural India, for which data on the unemployment rate during 1972- 73 are available, a positive correlation of 0.30 (significant at the five percent level) is observed between unemployment and average agricul- tural output per hectare (in rupees, 1970-71 to 1972-73).2/ Apparently, the value of agricultural output per hectare is not a good index of the extent of poverty in a region. Land productivity necessarily de- pends on factors such as fertility of the soil, rainfall and irrigation, frequency and pattern of cropping, etc. Also, the regions with high agricultural productivity may attract unemployed labour from neighbouring areas. Unemployed and Unemployment According to Usual Activity To understand the correlates of poverty and unemployment more clearly, we shall examine the "usual" activities of persons clas- sified as "currently unemployed" and of those reporting unemployed per- sondays. The incidence of unemployment according to different criteria will also be estimated separately for each usual activity category. As noted earlier, these categories take into account the main sector of employment (farm or non-farm) as well as the status or class of worker (self-employed, employees and family helpers). Among the employees, 1/ D. T. Lakdawala, "Growth, Unemployment and Poverty," Presidential Address delivered at the All India Labour Economics Conference, Tirupati, December 31, 1977. 2/ The total number of rural "regions" in India in the 27th Round of the NSS was 65. The agricultural output data are based on a study undertaken by the Perspective Planning Division of the Planning Commission and the Jawaharlal Nehru University. According to Lakdawala's address cited above, the agricultural output data were not available for 9 of the 65 regions.  - 15 - a distinction was madea between "a regular salaried employee/wage labourer" and "a casual employee/wage labourer." There may be some scepticism about the extent to which it is possible to idenitify some of these usual activity groups, partic- ularly the "casual labourers," because of the high probability that individuals move across these categories during different periods of the year or even over a short span of time. However, persons experienced in conducting field work in India do not envisage any difficulty in identifying households or workers whose principal activity during the year (or a reference week) in terms of labour time disposition (or source of income) is working as an employee or casual labourer.1/ Table 4 shows the usual activity distribution of persons classified as "current:ly unemployed" according to the conventional prior- ity criterion as well as our major activity criterion, the incidence of unemployment, and the participation rates for Gujarat and Maharashtra.- 1/ According to a personal communication from Sudhir Bhattacharyya, the NSS official in charge of the design of the 27th Round Survey of employment and unemployment, the interviewers reported no diffi- culty in identifying or classifying the "casual wage labourers" although the Instructions to Field Staff did not define the term "'casual labourer."' As shown later, a large majority of casual labour- ers in Gujarat and Maharashtra were agricultural labourers or workers in non-classifiable occupations. These groups are indeed likely to suffer from instability of employment; and their classification as 'casual' labourers is quite plausible. 2/ The participation rates shown in Table 4 are based on the current activity clata. The small differences between the participation rates based on the priority criterion and those based on the major activity criterion are due to some editing of the data during the time interval between their estima- tion and some rounding of the multipliers during later data processing.  - 16 - !eble 4 VIVAL m?TIVTISzS o? P0?SnS C'IS ID A5 i D StlTC 3t9 S tX2 VW UIIDD! ALTRNATv .TC.RIT . ...IARAT A5D SWAKASITkA. 1972-73 tb.m1a,ed During ueew1eyad TJrtes 09mpIlaye Dwee Unemployed During the Referle the RnereoY ce Week the iefetee Meek tloe Refereee Week (Prior,tv Criterion) (Nator Activity Crtt.rten) (Priorir Ctiterton) (Kaier Activty Criterion) VAuMAl AiV etty Males Femles Perose X FavaJjj enrse. Mg * P J g F. .1*