Document(s) 3 of 16

In Chapter 2 the focus was on households. We looked at trends in household welfare – and the changes in the incidence of poverty at the household level in particular. But household welfare is the result of the working of factor markets, and for households near the poverty line, trends in labor markets are of paramount importance. In this chapter our attention shifts to individuals. We begin to address the question: are the observed developments in the labor market consistent with and illuminate the results obtained in the previous chapter on household-level poverty trends?

Trends in aggregate employment

The Task Force on Employment Opportunities appointed by the Planning Commission reported a sharp decline in the labor-force growth rates between the 1980s (1983 to 1993–1994) and the 1990s (1993–1994 to 1999–2000). Its estimates from the NSS showed that the growth rate fell from 2.05 percent per annum to 1.03 (GOI July 2001). Taken in conjunction with the increase in measured rates of open unemployment, this slow-down has been widely interpreted to have been the result of 'discouragement' of potential workers from entering the labor force. There are, however, two basic problems with this estimate: (i) is the question of using the correct age-structure of the population; and (ii) the question of appropriate employment category on which the estimates are based.

On (i) it has been maintained that the survey-based age-structure, on which the weighted employment rates of the NSS are based is less reliable than the age-structure reported by the nearest Census of population figures; and on (ii) the Planning Commission estimates of employment growth are based on the UPSS figures of the NSS, and does not distinguish adequately the supply side aspects of the labor force from the effects originating from changes in the demand for labor. Research on both these questions has been extensive, and we shall begin by summarizing the major conclusions from this research.

The age-structure issue

Sundaram and Tendulkar (2006) have looked carefully at the NSS-based age distribution and compared it with the data reported by the Censuses of 1981, 1991 and 2001. They reported that

using the survey-based age-distribution results in a sharp slow-down in the growth of prime-age (15–59) population in the nineties…. [But] in the context of the observed slow-down of population growth reflecting the decline in fertility–from 2.09 percent per annum (pcpa) to 1.97 pcpa–over the same period, equally problematic is the acceleration in the growth of population in the 0–9 age group–from 1.17 to 1.33 pcpa–between the two periods that shows up with the use of the survey-based age distribution.

It is generally accepted by researchers that, while surveys like the NSS are more reliable at getting at participation or employment rates for different age–sex groups, the age-structure of the population itself is better measured by the Population Censuses. Sundaram and Tendulkar thus use the age–sex-sector specific worker-population rates produced by the NSS and re-weight them by the appropriate demographic structure obtained from the Population Censuses. This adjustment has the effect of substantially moderating the slow-down in the growth rate of the labor force: from about 2.06 percent in the 1980s to 1.58 percent in the 1990s.¹

The problem of measuring the labor force

The NSS distinguishes those who participate in the labor market on the basis of several criteria. A major difference is the estimate on the basis of Usual Principal Status (UPS–the activity in which the individual spent most of his time in the reference period of the last 365 days) and of the Secondary Status (SS–the activity defined in terms of some part of time spent in the reference year). Usual Principal and Secondary Status (UPSS) workers are then the principal workers as well as part-timers of various kinds. Clearly the number of SS workers could vary with changes in conditions affecting the supply of secondary workers as well as the demand for them. There is no way of judging from the numbers per se if any change observed is due to predominantly supply or predominantly demand conditions. This is particularly true of secondary women workers who form a substantial but varying part of the UPSS labor force reported by the NSS (Rustagi 2005). This important issue will be analyzed in detail in the next chapter.

Trends in employment by industry

Economic Development, in the history of both today's developed economies, and in the recent growth of developing economies, has been associated with a relative increase of employment away from the agricultural sector. This process is associated with an increase in labor productivity because the relative productivity is generally lowest in agriculture. Our first task, then, is to see how far India has been following this traditional pattern of transformation in recent decades.

Multiple occupations

We need to clarify at the outset the issue of multiple occupations in the Indian economy. How do we trace the changing structure of employment by industry or occupation when a significant number of households have members who pursue more than one occupation? There are in fact two distinct aspects of this issue. First, households would contain more than one earner of 'usual principal status' (UPS at the individual level). Second, a 'principal status' earner might have more than one activity. The first possibility creates a difference between the occupational or industrial classification of households (in terms of the activity of the 'main earner', defined as the main contributor to the household pot), and the occupations or industrial classification of individuals. The second point creates a distinction between the occupational classification of individuals based on the UPS and the UPSS status. The issue of occupational distribution by households is of importance when we are considering household income levels in different occupations. Since income (or expenditure) is available for the household as a whole we would need to define the occupation of the household by the activity of the main earner. We have done this in our work on the tertiary sector (Chapter 10). It is seen that the difference in changes in occupational classification over time by the household and the individual definitions is marginal. Here in this chapter we discuss the changes in the distribution for individuals only by the two alternatives of UPS and UPSS.

Trends in the industrial structure of individuals

It is apparent from the data in Table 3.1 that agriculture has indeed been shedding labor and it would appear that the process seems to have accelerated in the post-reform years of the 1990s. It is equally clear that the absorption of labor in manufacturing has been quite slow – even though it might have increased a bit in the nineties – and much of the increase in the labor force has been accounted for by the various types of tertiary activities, as well as construction.

We can infer from the discussion above that the decline in the share of the labor force in agriculture might have been exaggerated with the UPSS definitions because of the inclusion of secondary workers in the count. This category of workers might be disproportionately represented in agriculture which has a larger component of the self-employed. In so far as this reduction is partly due to supply-side changes affecting secondary workers or indeed due to fluctuations in the pace of technology spread in agriculture (see Chapter 7), the basic shift away from agriculture in the nineties might be overestimated. We therefore looked at the industrial distribution for labor defined on the UPS criterion. This might provide an estimate of the lower limit of the shift from agriculture.

A comparison of the two sets reveals that the UPSS definition does show a substantially larger decline in the incremental share of labor absorbed by agriculture. The UPS definition gives the decline in the incremental share in agriculture from 42.07 in the first period to 27.55 in the second. But even confining ourselves to principal workers, as the UPS definition does, the data confirm that there was a significant decline in labor absorption by agriculture in the nineties compared to the eighties. The gain of the tertiary sector in the incremental share under the alternative UPS definition was from 37.49 to 45.41.

Unemployment

What can we say about the level and trends in the rates of open unemployment in the Indian economy? The NSS data can be used to calculate unemployment rates based on either the CDS or UPS status of the labor force. The CDS estimates measure the rates of person-days which are being spent as 'not working but available for work', measured in half-day units over the reference week (see Appendix 2). These rates differ from the unemployment rates based on the UPS counts describing the 'usual status' of workers. (Note that subsidiary workers are by definition employed, still there can be estimate of UPSS rates of unemployment different from the UPS rates.) We can say that the CDS rates capture open underemployment during the week–as distinct from disguised unemployment on family farms or businesses (when some members of the household workforce are 'unproductively' employed but not declaring themselves available for work). The unemployment rate provided by the UPS measure will be necessarily less than the CDS rate since it measures the proportion of the labor force which is 'usually' unemployed during the major part of the year.

The rates of CDS unemployment in 1999–2000 was 7.2 percent in the rural areas and 7.7 percent in the urban. The UPS unemployment rates were, however, only 1.43 and 4.65 respectively (Government of India (2001), Tables 2.7 and 2.10(a)). It is apparent that underemployment, rather than round-the-week open unemployment is the real issue in rural areas. The difference, although significant, is less striking in the urban economy in the post-reform years between 1993 and 1999. It increased substantially in the rural sector–from 5.6 to 7.2. (The increase in the urban sector was hardly apparent–of the order of 0.3 percent.) A part of the increase could be attributed to the increased 'casualisation' of the labor force–the rise in the proportion of casual labor relative to the self-employed. It cannot be maintained that this is necessarily the result of deteriorating labor-market conditions. It might be partly the result of increased commercialization, as marginal farmers shifted more to wage labor. The income levels of the latter are often higher.

In any event the evidence suggests that even CDS unemployment is very much a problem of the youth, perhaps a result of waiting and job searching in the labor market. The unemployment rates for both sectors and for all rounds fall off sharply for age groups 30 and above. The increase of the unemployment rate between the 50th and the 55th rounds, after a fall between the 38th and the 50th in most age groups, perhaps does indicate a slight deterioration of labor-market conditions, but the phenomenon is one of lesser importance than other issues discussed in the book.

Trends in labor productivity by industry

The data given in the last section shows some movement outside agriculture – even though it has not been as fast as in many other Asian countries during the process of their economic transformation.

Another special feature of the changing employment structure in India has been the overwhelming importance of the tertiary sector in the absorption of labor outside agriculture. This at once raises the question: is the transformation of the employment structure – slow as it is – has really been of the type that has increased earnings of labor. A detailed examination of this point is attempted in the later part of this chapter and also in the chapters on individual major sectors in Part III. Here it is sufficient to note the relative mean productivity per worker in the major sectors and their changes over time – based on the figures given in the National Account estimates.

It is clear from the mean value of labor productivity that they are between 2.5 and 3.5 times higher in the manufacturing and tertiary sectors relative to those in agriculture – even if we take the more moderate estimates based on the UPS estimates (i.e., excluding the secondary workers who are relatively more abundant in agriculture), and even if we are looking at the less productive sub-sectors within tertiary activities (Table 3.2). Further, the productivity differential with respect to agriculture seems to have increased over time. This first cut at the data does strongly suggest that the movement of labor away from agriculture – at a slow pace as it has been – has in fact in the direction of enhancing earnings of worker. We will see later in this chapter if this tentative conclusion is borne out by more direct evidence on wages and earnings levels.

Is India out of line with the experience of other Asian economies?

The exceptional nature of the absorption of labor moving out of agriculture in the tertiary rather than the secondary sector is seen to have been a feature of Indian development in recent decades. At the same time we have found that aggregate figures show that the relative income per worker in the tertiary sector is relatively high. Can we learn something from international experience if the Indian pattern of change is out of line with the observed pattern of development and, if so, in what way?

Papola (2005) discussed in detail the theory in the literature about sectoral shift of GDP and employment. Classical economists like Fisher and Clark explain the shift from industry to services by the changing demand patterns predicted by Engel's law. Fisher argued that services are 'luxuries' with more than a unitary elasticity of demand and so at a higher level of income increasing share of expenditure is absorbed by them and thus leads to high share of services in output and labor force. He assumed that the increase in the share of services in final demand proportionately lead to increase in the share of employment. However, Clark attributes the increase in the share of service employment additionally to low relative productivity in services relative to manufacturing. Later economists like Bamoul and Fucho ascribed the rise in the share of service employment primarily to productivity differentials between industry and services resulting from technological, scale and geographical concentration of production in services. Further, increase in the share of service employment is also explained by the increased tendency of industry to outsource intermediate inputs used by industry to the service sector.

Popola refers to the experience of some Asian economies for comparison with India. The data for the shares of both employment and GDP and their change over the second half of the last century are given in Table 3.3.

It can be seen from the table that the share of workforce in industry increased along with is share of GDP in all countries including India, but it produced a much larger share of GDP in all other Asian developing countries other than India. It shows that the relative sectoral productivity of labor in India has been strikingly low by international comparison. By 2002 the tertiary sector in India contributed more than half the GDP in India but its contribution to employment was only 22 percent. It shows that service-sector growth has been productivity led but not employment led, contradicting views of some economists that employment grow in services because of low productivity vis-à-vis industry.²

The picture presented in Figure 3.1 of relative productivity in services vis-à-vis industry in the comparator Asian countries brings out the striking point that it is only in India – among all the countries represented – that the relative productivity in services has increased over the 40-year period. A second important point to note is that – with the exception of Thailand in 1960 when it had hardly any industry at all – the productivity in services exceeds that in industry only in India in both years, and that by a substantial percentage.

It shows that service-sector growth in India has been productivity led and not employment led, contradicting views of some economists that employment grew in services because this sector has been a repository of low income labor 'pushed out' of agriculture. The heart of the employment problem in India would thus seem to be not an excess absorption of labor in the tertiary sector, but the relatively low productivity of the manufacturing sector, and its persistence over time. It is this low performance of manufacturing which has prevented it from being the dynamic sector playing a central role in productivity growth as well as the reallocation of labor as in other countries in the history of successful economic development.

Figure 3.1 Relative productivity in services and industry, various Asian countries 1960 – 2000.

How much of this productivity differential in favor of the tertiary sector is due to the recent developments of the information technology sector? The answer would appear to be not very much. For one thing the productivity differential in favor of the tertiary sector was substantial even in 1960 when the IT sector was non-existent. Second, in terms of the numbers employed the tertiary sub-sector dealing with IT is quite small even in recent years. Table 3.4 gives an estimate of employment in this sector based on enterprise surveys, and Table 3.5 provides the estimate from the household surveys of the NSS. The Manufacturing sub-sector includes hardware, central processing units (CPUs), communications equipment, electronic components and industrial control and supervision equipment manufacturing (not including medical equipment). The tertiary segment includes telecommunications services, computer and related services (IT and software), research and development services and also start-up companies.

The estimates show that the total employment in the IT tertiary sector is of the order of 400,000 to 600,000 (Tables 3.4 and 3.5). Considering that the total employment in the tertiary sector (in the UPS count of the NSS) was around 150 million, the percentage of tertiary employment in the IT sector was at best 0.4 per cent.

It is necessary to turn our attention to the denominator of the ratio and consider the possible reasons for the low labor productivity of the manufacturing vis-à-vis the tertiary sector in India.

Table 3.6 draws attention to the 'dualism' that exists in Indian manufacturing. The household enterprises (not employing any hired labor) contribute more than half of manufacturing employment whereas establishments with 500 and above employees contribute more than two-fifths of gross value added but employ less than one-tenth of employment. Consequently there is a tremendous difference in relative labor productivity between these two size groups and it is this which leads to very low level of labor productivity in the manufacturing sector. Such a situation does not exist in other developing countries in Asia, as will become clear from the evidence presented in Chapter 9. Unless there is substantial growth of small (10 – 100 employees) and medium (100 – 500 employees) that are relatively labor-intensive and have substantially higher labor productivity than household enterprises leading to substantial increase in the share of manufacturing in GDP with some increase in employment share in the Indian economy, we are unlikely to follow the sectoral pattern of growth as other countries experienced in the development process.

Further discussion of this important issue will be found in Chapters 8 and 9 in Part III of this work.

Employment in the organized sector

It might be useful at this point to put the size of the formal or organized sub-sector in manufacturing in the context of total employment in the formal sector. In Chapter 10 we will examine the formal – informal distinction within the tertiary sector in detail. But for the present purposes the official estimates of different types of employment within the formal sector put out by the Ministry of Finance of GOI will suffice. These are given in Table 3.7. The stagnation of manufacturing in the formal sector is apparent from this table, as is the relatively small share of manufacturing in total formal sector employment. The total including all sectors is itself very small in 2001 – only about 7 percent of all employment. The public sector still dominates the scene in formal employment in spite of India having embarked on a process of encouragement of the private sector since the early 1980s.

Patterns of urbanization and the quality of employment

A feature of economic growth has been the increasing absorption of labor in the urban sector. The rate of urbanization has been slow in India – consistent with the slow transformation of the employment structure. There has been some concern in the literature if the reallocation of labor to higher-quality jobs in non-agriculture has been disproportionately achieved only in the urban economy. We can also refer at this point to the finding in Chapter 5 that there have been important changes in recent decades in the size structure of towns, with a redistribution of population to smaller towns. How does in the change in the industrial structure of employment differ between small and large towns – as well as between rural and urban areas? To throw light on this question we present in Table 3.8 the way the incremental flow of labor in each of the three broad sectors was distributed between the rural, and the three classes of towns. The data are presented separately for the 1980s (1983 to 1993/1994) and the nineties (1993/1994 to 1999/2000), but the classification by size of town is not available for the earlier period. Of particular interest is the relative importance of the flows of new employment in the secondary and the tertiary sectors. The importance of the secondary sector even in the most recent period is higher in the urban areas as whole, and in large towns. In 1990 – 2000 the share of manufacturing was 27 percent in large towns compared to 19 percent in small towns and only 7 percent in the rural areas. But it is apparent from the figures on incremental flows in Table 3.8 that a redistribution of employment in the secondary sector has been taking place in the recent period in favor of small towns, and also the shares of the rural and the urban sectors in the new employment has been almost the same. The importance of the small towns in the tertiary sector has, however, been increasing faster. The small towns have clearly witnessed a substantial swing away from employment in the primary sector. The redistribution of employment to small towns, which has been noticed, has been driven by non-agriculture.

Expansion of education and the quality of employment

The expansion of employment outside agriculture – and the concomitant upgrading of jobs – is closely related to the expansion of education. It has been maintained that the lopsided development of education outside the rural sector has in fact hindered the diversion of employment in the rural economy (Chadha and Sahu 2002).

Table 3.9 in particular brings out the point that it is at the education level 'graduate and above' that the urban economy plays an overwhelming role in attracting the educated. But it is also of great importance to note that the major proportionate shifts in the additional flows of educated labor are to be observed in the smaller towns in the post-reform period. These towns have been able to attract a large proportion of educated labor – with secondary as well as college qualification – in major way in the 1993 – 1999 period. This is another interesting part of the increasing role played by the smaller towns in recent years – which had already been noticed in Chapter 2.

India has made rapid progress in upgrading the quality of its labor force. The number of workers with less than five years of education has come down steeply from 80 percent in 1983 to 65.5 in 1999 – 2000. But even then it is significantly behind most of the rapidly developing countries in Asia. The average years of schooling for the population aged 25 and over in China around 2000 were 5.7, and in East Asia 6.5 compared with 3.6 in India. The proportions with no schooling were 20.9, 22.8 and 44.5 respectively. Equally damaging is the low proportion of those with secondary schooling – known to be a critical group in the development of manufacturing and other modern-sector activities. In India it is only 17 percent at this date, much lower than India's income level would predict. It is only half that of China, and the proportion is worse for females (World Bank 2007, Table 1.3). It will be suggested in Chapter 9 that its relative neglect of primary and post-primary education in earlier years might have been a major cause of the persistence of 'dualism' and the slow growth of the dynamic manufacturing sector in India.

Trends in wages and wage inequality

Wages of casual and regular workers

The wage sector in India is substantial – even in the rural areas. Regular workers (those with a more permanent contract for varying periods of time) are more important in the urban areas, and casual wage workers (those hired on day-today contract as work is available) are in a majority in the rural economy.

Regular workers have several days of work during the week – the NSS data show that the average is between 5.6 and 5.9. Casual workers get work for fewer days of the week – generally less than four. Part of the difference in the earnings per worker between the two categories, therefore, reflects the difference in the number of days of work secured in the week of enumeration. For casual workers the seasonal element is likely to be of great importance. When making comparison between the two groups – which could be used to reflect an aspect of the formal – informal dichotomy in the labor market – it is important to be clear as to the objective if the comparison: are we interested in the levels of income of the two classes or in the returns to a unit of work?

Average wages (earnings) per day

The NSS collected data on the earnings of the workers for the preceding week (seven days) of the survey, and it also recorded in the same field the number of person-days the worker was actually at work. These data give us the earnings per day for both casual and regular workers. The casual – regular wage difference varies by rural – urban location, by gender and also by occupation (i.e., manual or non-manual). In the rural areas there were about ten million casual workers according to the 55th round of the NSS (Census adjusted 15 – 59 age group) and only two million regular. The corresponding figures for the urban sector were 1.4 and 2.5 million. Females were a third of the total in the rural casual labor market, but in all other segments, in the regular category and in urban areas, the representation of females is much smaller – of the order of 15 – 20 per cent. It is to be noted that not all regular workers were classified as non-manual. In fact both in the rural and the urban sectors almost half of the regular were manual workers. On the other hand, only 10 – 15 per cent of the casual workers were non-manual.

We examined the data from the NSS showing the difference in mean wages per day for different categories of workers in the 15 – 59 age group for the 55th round 1999 – 2000. For casual workers the manual wage rates are close to the non-manual, for both sexes, in both the rural and the urban areas. On the other hand, the difference between the two categories of workers for the regular wage earners is huge (between twice and three times as high). It shows the importance of human capital attainments in the determination of regular wages.

It is important to note, however, that even for manual workers alone, regulars earn nearly double the amount of casuals – except for females in rural areas, where the differential is more like 50 per cent higher. Since regular workers get a significantly larger number of days of work in the week (also get paid for the whole week), the difference in earnings would be even higher. While a part of this difference – even for manual workers – might reflect measured human capital attributes, a good deal of the differential really pertains to the formal-informal sector dichotomy in the labor market. This differential is partly due to institutional factors (employment in large establishments, or in the public sector) and partly due to the operation of the wage-efficiency relationship for 'established' workers with low turnover.

Distribution of wages

The data on wages from the NSS have been analyzed by Puja Vasudeva-Dutta (2004). Vasudeva noted that the dispersion in wages among casual workers is much smaller than among regular wages. This is confirmed by the graphs in Figure 3.2, which also suggests that the dispersion seems to have increased over time for regular wage workers, but not for the casual. A major reason for the difference is, of course, is that regular wage workers have a much greater variance in human capital attributes, particularly education. There is a big difference between manual and non-manual wage difference for regular workers, but not for the casual, reflecting the dispersion by skill and education for the former category.

Growth rate of wage rates

The figures 3.3a and 3.3b give a picture of the growth rates of real wage rates for different categories of labor namely rural male (RM), rural female (RF), urban male (UM) and urban female (UF).

Figure 3.2 KDF distribution for regular and casual workers for different NSS rounds.

Figure 3.3a Growth rate of wage of regular non-manual wage earners.

The wages of regular non-manual workers in the second period increased at around twice the rate of the pre-liberalization era, but there was little change in the growth rates of casual manual workers. This is an aspect of the increase in inequality in the labor market in the nineties.

Trends in wage inequality

It can be inferred from Figure 3.2 that there is strong suggestion from the KDF` graphs that wage inequality is higher for regular workers and has increased over time. Vasudeva's summary measures for wage inequality indicated that, for regular workers, GE(0) went up from 0.286 to 0.337 between 1983 and 1999, and GE(2) increased from 0.381 in 1983 to 0.430 in 1999. The level of inequality was much less for casual workers and it also declined over time. The values of GE(0) were 0.143 in 1983 and 0.117 in 1999.

Figure 3.3b Growth rate of wage of casual manual wage earners.

It is interesting to note that while inequality among regular workers increased significantly between 1993 and 1999 – and more so at the upper end as evident from the larger increase in the GE(2) measure – the 'between group inequality' for educational groups did not change by all that much. Much the more important part of the inequality increase was accounted for by the 'within group' component. This is in line with the evidence from other countries which have experienced increase in wage inequality in the globalization era. While returns to formal education do increase, it is the differential valuation of the individual worker's non-formal attributes which seem to be more important in the increase in inequality.

Vasudeva has used the regression-based methodology of Fields to study the 'factor inequality shares' of different explanatory variables in the earnings functions estimated separately for the regular and the casual workers (Field 2000). The 'factor inequality share' gives a quantitative estimate of the total inequality in the dependant variable (in this case 'wage earnings') explained by the different explanatory variables in the earnings function.

A semi-logarithmic Mincerian (standard or augmented) wage determining function can be written as:

where a =[β₁.....β_j, 1] and Z = [Z₁.....Z_j, ε] are vectors of coefficients and explanatory variables respectively. An inequality index I can be defined on the vector of wages (w). Applying Shorrocks' theorem the relative factor inequality weights (i.e., the percentage of inequality that is accounted for by the jth factor) can be calculated as follows:

where cov[.] denotes the covariance, cor(.) the correlation coefficient and σ(.) the standard deviation.

The major results from Vasudeva's exercise can be summarized as follows:

1. As far as regular workers are concerned just over half of the variance in the log of wages are explained by the earnings function. The same variables explain much less – a third – of the variance for casual workers.
   
   
2. In terms of the explained part of the variance, human capital variables were most important for regular workers. Age accounted for about a quarter and education a third of the explained variance in 1999. The other important factor in line was industry affiliation – contributing another quarter.
   
   
3. By contrast, human-capital factors were of much less importance for casual workers – only age, and not education having any positive contribution, but at a much lower level of around 7 percent. The single most important explanatory variable was geographical difference – the state of residence contributing no less than 62 percent for casuals as against only 3.5 percent for regulars.
   
   
4. Although for regular workers the wage gap between those with graduate and primary-school qualifications increased between 1983 and 1999 (see 'Rural – urban differences' section below), the share of education in the explanation of the variance declined from 23 to 17 percent. The importance of age increased as did that of industry affiliation. Further, Vasudeva confirms that the increase in the 'contribution of selection coupled with the fall in that of education suggest a rising importance of unobservable for regular workers, possibly linked to the process of trade liberalization'.

Inequality in household welfare

Although substantial wage employment in India is still only a part of total employment, and a good deal of households are outside the wage-sector – mostly self-employed. We would want to know if the experience of non-wage households mirrors that of the wage earners. This section therefore looks at the trends of welfare of all households irrespective of the type of employment. We choose as our measure the average (mean) per capita expenditure (APCE) of the households as recorded by the NSS of successive rounds. Figure 3.4 portrays the movement of the KDF distribution of APCE over time, separately for the rural and the urban areas.

It is apparent that while the modes of the distribution have shifted outward in both sectors, but more so in the urban sector, there has been a more pronounced 'flattening' of the distribution in the urban sector signifying an increased degree of inequality.

Figure 3.4a KDF distribution of APCE, Rural (poverty line: Rs.196.50, at 1993 – 1994 = 100).

Figure 3.4b KDF distribution of APCE, Urban (poverty line: 227.20, at 1993 – 1994 = 100).

Table 3.10 gives the measures for overall inequality. Note the large values for GE(2) which is more sensitive to high incomes. While this measure has decreased substantially in the rural areas it has increased in the urban.

Rural – urban differences

The discussion in the last section has suggested that inequality has increased more strongly in the urban economy, at least in the post-reform era. Thus the disparity in household welfare between the two sectors has increased, we now look a bit more intensively at the rural – urban difference. Has the disparity increased more for some groups rather than others? Can we isolate more concretely the factors responsible for it?

Figure 3.5 brings out clearly the point that the relative difference in household welfare has increased for higher expenditure groups. We can compute the 'Blinder – Oaxaca' decomposition of mean outcome differential between the rural and the urban sectors, The difference between two groups can be decomposed into three parts: i) due to differences in endowment (E); ii) due to differences in coefficients including the intercept (C); and iii) due to interaction between coefficient and endowment (CE). Depending on the model that is assumed to be 'true' model (absence of discrimination), the three-fold decomposition can be used to determine the explained (Q) and unexplained (U). By using the low group (rural APCE) as the no-discrimination base we calculated Q = E and U = C + CE.

Figure 3.5 Urban – rural difference in APCE by percentile.

Table 3.11 summarize the results for the 'Oaxaca decomposition' for APCE between the rural and urban areas. The calculations show that there has been a substantial increase in the 'discrimination' factor for urban households in the post-reform years between the 50th and 55th rounds. The increase in the rural – urban disparity is due not to the better endowments of the urban workers but to the higher returns to the human-capital factors secured by them in the urban economy.

Returns to education

The results discussed above suggest that educational developments have been a major player in the increase in inequality and in the growing rural – urban disparity. The increments to income from successive levels of education could be approximated by the difference in co-efficient to the education dummies in a fitted earnings functions for regular wage earners. These are reported in Table 3.12 and graphed in Figures 3.6a and 3.6b separately for the rural and the urban areas.

The difference between rural and urban economies is brought out dramatically in Figures 3.6a and 3.6b. The lift to the returns to education in the post-reform years occurs at different levels of education in the two areas. In the rural economy the sharp increase occurs at the level of secondary education, while in the urban sector the lift is observed at the college graduate level. The curves for the successive rounds, however, intersect at lower education levels in both sectors, showing that at levels less than middle, the returns to education are in fact depressed for the later years. They are nearly at the same level for middle-school leavers in the rural sector, and for secondary-school leavers in the urban. All this is consistent with expectations about what might happen with the increase in the supply of educated labor. Demand outstrips supply in the post-reform period in the rural areas for secondary-school leavers, and for college graduates in the urban.

Analysis of the returns to education by age-groups revealed an interesting finding about the urban labor market. For the 20 – 29 group the marginal returns to secondary education actually fell in the 50th and the 55th rounds while those for college graduates showed a sharp upward movement. By contrast for the older 30 – 39 group there was a milder increase for both the secondary and college graduates in both the 50th and the 55th rounds. It is clear that the demand for the more educated has been soaring in recent years and has affected the new entrants to the urban labor market more strongly.

The literature has drawn attention to the increased demand for more educated labor in the era of globalization in a number of countries and has stressed the importance of skill-intensive technical change in manufacturing in particular, and/or the importance of more skill-intensive manufactured goods in international trade (see Chapter 1 above). But we have seen that the employment expansion in post-reform India has been concentrated not so much in skill-intensive manufacturing as in the tertiary sector. It is therefore useful to ask the question: is the expansion of demand for educated labor which we witness particularly in the market for college graduates in the urban areas originating mostly in the tertiary sector? Table 3.13 gives the distribution of the addition to the UPS workforce by industry for different levels of educational attainments.

Figure 3.6a Private return to different levels of education (urban).

Note
1, 2, 3, 4 and 5 denote the levels of education namely literate, primary, middle, secondary and graduate respectively.

Figure 3.6b Private return to different levels of education (rural).

Note
1, 2, 3, 4 and 5 denote the levels of education namely literate, primary, middle, secondary and graduate respectively.

Figure 3.7 Returns to education in urban areas by age-groups.

It is apparent that the market for college graduates in particular has expanded relatively more in the tertiary sector. A somewhat unexpected finding is that this trend had been going on since the eighties. Equally revealing is the finding that labor with less than middle level of schooling is now almost entirely absorbed in the primary sector. The difference between NO_ED and BASE_ED education stressed in the Introduction to the book would seem to be drawn at the boundary of primary education in the Indian labor market in the late nineties. Entry into the non-primary sector would now seem to require post-primary education.

We will discuss in Chapter 10 in particular that the bias in Indian policies towards tertiary education has encouraged the growth of skill-intensive industries. It is seen from the evidence presented here that the demand for labor with college education seems to be outrunning the supply with this pattern of development – even with the historical bias in education policies. It is very likely that the return to college education has continued to increase in the years since 1999–2000.

Appendix

Employment estimates based on current daily status (CDS)

Current daily status of all individuals above the age of five is coded in the NSS for each half-day over the seven days preceding the survey. The activity of each half-day could be classified as (i) employed (ii) unemployed or (iii) out of labor force. Even if an individual is not classified as unemployed under the usual status in the week, some half-days of unemployment are possible if they are available for work for those units of time. Thus apart from the 'usual' unemployed the unemployment days would be contributed by casual workers; the self-employed who are working generally; and even by those 'usually' outside the labor force. The CDS unemployment rate is calculated by adding up these person-days of unemployment as a proportion of all days of employment plus unemployment.

There has been a school of thought in the Planning Commission and other GOI circles which has favored the use of CDS for the estimates of employment. This generally produces estimates which are much lower for 1999 – 2000 than those calculated on the UPS or UPSS basis. Thus the growth rate of employment shows a significantly higher rate of decline than the other estimates. For example, Srinivavsan (2005, Table 4) gives the growth rate of employment in the 1987 – 1999 period (between the 43rd and 55th rounds) in the rural areas at –4.59 percent on the CDS basis against –1.48 percent on the UPSS basis. But Srinivasan comments:

The total number of person-days of employment is not the same as the total number of employed persons. The reason is that a given total number of person-days of employment could be distributed among the same number of persons in many ways so as to lead to different numbers of persons employed. For example, consider a four person economy in which all four participate in the work force and together they were employed for ten person-days in the week. This yields a person-day rate of employment of 10 out of 28 or 36%. If the ten person-days are distributed in a way that one person is employed for seven days, another for three days and the remaining two are unemployed, then person-rate of employment is two out of four or 50%. On the other hand, if it is distributed in a way that three persons work for three days each and one person works for just a day, the person rate of employment is four out of four or 100%, given the priority given to the status of employment!

We know that the NSS estimates of the numbers of persons in different demographic groups are underestimated, so we have to get the population figures from the Census counts. It is inappropriate to apply the employment rates based on person-days to the count of persons obtained from the Census of Population to arrive at the total number of employed persons.

Document(s) 3 of 16