Document(s) 13 of 16

The analysis presented in much of the book made use of the detailed unit-level data available from the 'thick' rounds of the NSS. The last of these quinquennial rounds which were available to us was for the year 1999–2000 (the 55th round). Thus we were unable to cover the current trends in the early years of the century. Another 'thick' round of the Survey was completed by the NSS for 2005–2006 (the 61st round). At the time of completing this manuscript (April 2007) the unit-level data sets from this survey are available only in part. But the NSS has released two reports on the some basic tables both for the Employment and Unemployment (EUS) part and the Consumer Expenditure (CES) part of its survey. In what follow we have made use of these published reports to discern some of the trends emerging in the most recent years and see how they fit into the picture emerging from our detailed work for the period ending 1999–2000.

Trends in employment and earnings

India's growth rate accelerated towards the end of the period between the 55th and the 61st rounds. While the average growth rate over the decade 1992–1993 to 2002–2003 was 5.9 per cent, it touched 8.5 per cent in 2003–2004 and 7.5 in 2004–2005 (World Bank 2006, Table 1.2, p. 14). As in the late nineties the service sector continued to lead the growth process. Also the lagging growth in the agricultural sector persisted. If anything the growth in the agricultural sector decelerated from 3.2 percent in the period 1980–1992, to 2.4 per cent in 1992/1993 to 2002/2003 and even lower in the last few years. However, industry recovered somewhat from the slump of the second half of the 1990s. Manufacturing grew at the rate in excess of 6 percent per annum. But the basic structural problems of manufacturing, identified in the book continued to be significant, posing serious problems for sustained growth. These include the persistence of 'dualism' and employment stagnation in the formal (organized) sub-sector.

Recovery of participation rates

The new evidence suggests a major development on the labor-market front between the 55th and the 61st rounds. This is the recovery of participation rates which affect the trends in employment. The slow-down in employment growth in the 1990s discussed in Part I of the book was traced to: (i) a withdrawal of labor of principal status in the young-age groups because of expanded schooling; but (ii) more importantly to the withdrawal of female labor of subsidiary status, particularly in the rural labor market. A good deal of attention was paid to the possible reasons for this withdrawal (Chapter 4). While no definitive conclusion could be reached, there was a strong suggestion that this decline in PRs was due to a shift in demand conditions in the agricultural economy. The 1980s had seen an upsurge in the demand for labor due to the second phase of the green revolution spreading to the eastern states. Prime-age females were pulled into the labor market due to a sudden upsurge of labor. With the passing of this wave of technological change the labor market adjusted to a more stable pattern: the subsidiary workers who met the upsurge in labor demand seem to have been substituted by more permanent labor arrangements, such that while there was a pronounced fall in PRs of subsidiary females, the principal status females increased their participation to some extent.

What caused the recovery in participation in the latest period between the 'thick' rounds? An overview of the changes in broad categories of labor is presented in Figure 13.1. A perusal of the bar chart and the figures attached to it suggest that, while all four groups of labor distinguished increased their participation, the bigger contribution has been made by females – and more so in the rural areas. The increase in female participation has been both in the principal and subsidiary categories. Subsidiary females played a bigger role in the increment to the participation rate in the rural sector but in the urban areas it is the female principal workers who are more important in the change. Subsidiary females are, however, a substantial part of the change in the urban areas as well. The following graphs depict the age profiles of participation rates of females of subsidiary status for all three of the last quinquennial rounds of the NSS.

Figure 13.1 Labor-force participation rates, 1999–2000 and 2004–2005 (source: NSS Report number 515, 2007).

We see that there is an interesting difference between the changes in the profiles in the rural and the urban areas. In the rural sector the increase in the subsidiary participation in the 61st round is concentrated in the prime-age group, and the recovery brings the profile up but not as far as the level of the 50th round (1993–1994). Figure 13.2 depicts it. In the urban areas by contrast the increase in participation is stronger among female subsidiaries of the younger age groups, and in fact their PRs in the 61st round exceed what had been achieved in 1993–1994 (Figure 13.3). Generally, the age-profile of female participants of subsidiary status has shifted to the left in the 61st round. It suggests that in the urban sector young females are tending to enter the job market more strongly than in the earlier rounds.

It is not possible with the summary data available so far to examine the proximate causes of the change described above. It is doubtful if the shift in the demand for labor in agriculture, which apparently explained a significant part of the decline in female participation in the 1990s, could be the cause of the recent recovery. As we have already pointed out the agricultural sector has continued to suffer from stagnation and slow growth. Is it possible that the strong growth in the non-agricultural, and particularly the service sector, which has propelled the demand for female labor inducing an increase in PRs? The fact that females of principal status and of the younger age groups are important in the increase in participation in the urban economy might give some strength to this hypothesis, since non-agriculture is of much greater importance as a source of job growth in the latter.

Figure 13.2 Female subsidiary labor-force participation rate (in %) across age groups, rural areas (source: Same as Figure 13.1).

Figure 13.3 Female subsidiary labor force participation rate (in %) across age groups, urban areas (source: Same as Figure 13.1).

The change in participation rates is of course not the same as the change in employment rates. The gap between the two is accounted for by any change in unemployment rates. Here the evidence as far as males are concerned is that there has been little change in unemployment rates in the UPS count but that the unemployment rate in the CDS count has increased only in the rural sector. Much the more important change has been in the female labor market. The rates for females have gone up significantly. This is true of both the rural and urban sectors, and also for the longer-term unemployed (as measured by the UPS count) and the shorter-term ones (the CDS count). The figures are given in the Table 13.1.

It seems very likely that the increase in female PRs and the increase in unemployment rates are parts of the same phenomenon. As women job seekers enter the labor market their job search pushes up the unemployment rate. The increase in the unemployment rate is, however, less than the increase in participation rate.

As discussed at length in Chapter 3, this development might be due to supply- or demand-side changes, or to both. The problem now, however, is to explain the change in the direction opposite to what was discussed in the chapter. The movement in participation is in the upward direction – which might be caused by the supply curve shifting downwards, or the demand curve shifting upwards, or to both. No conclusion on this issue is possible without detailed work with unit-level data. We will now discuss the changes in employment structure, wage rates and household welfare levels to come up with some plausible hypotheses for future research.

While the period of the 1980s and the 1990s had seen an increase of the share of casual workers – more so in the rural areas – the latest period saw a decline in this share in both sectors, and both for males and females (Table 13.4). In fact, the decline in the share of casuals in female employment has been going on in the urban sector for a long time, and gathered momentum in the latest period. What is a new development is the decline of the casual share among rural females and among males in both sectors. What type of employment did increase to compensate for the decline in the casual share? Females in the urban labor market increased the share of regular wage workers – reinforcing again the trend which had already been under way since the 1980s (Table 13.3). All other groups – males in both rural and urban areas, and females in rural areas – increased their share in self-employment (Table 13.2).

While the movement from casual to regular workers among the body of female urban job seekers signifies an improvement of labor-market conditions, the same cannot be said unambiguously about the increase in the share of self-employment without further information of incomes earned (Table 13.2). It might represent the growth of a dynamic informal sector, or alternatively could signify accumulation of job seekers at the low end of the earnings spectrum.

Changes in the structure of employment by industry

It was seen in earlier chapters that in Indian economic development the tertiary sector had led the way in absorbing labor which had been increasingly diverted from the agricultural sector. Further, even the limited amount of labor finding employment in the secondary sector was mostly absorbed in its informal sub-sector. What is the evidence in the recent period while the post-reform changes had some time to take root? The relevant data are presented in Table 13.5. It is seen that the increase in employment share in the secondary sector in the recent period has been higher than in the tertiary. However, the secondary sector in the table includes construction – and this sub-sector accounted for fully half of the incremental share of employment. Further, the dominance of the informal sector in the growth of manufacturing employment has persisted.

Table 13.6 gives the figures given by the DGE&T (Labor Bureau) for the latest years available. They show an absolute decline in employment in the formal sector as a whole – even taking the private sector on its own. The percentage decline in manufacturing – including the private part of it – is much higher, two-and-half times as much.

Educational upgrading: the role of college graduates

We would expect significant upgrading over time of the labor force in terms of formal education. This is indeed what we see in Table 13.7. But a striking feature of the table is that the expansion is relatively more pronounced for college graduates. There is indeed a curious dent in the middle level of education. There is a decline in the share of workers with middle to higher secondary education in most of the groups distinguished in the table. Even the rural females, the sole group to register an increase in its share in the recent period, did so only to a minuscule percentage. By contrast the increase in the share of workers who are college graduates is large and impressive – particularly for urban females. This is an important development in the labor market, and obviously related to the strong growth of skilled tertiary-sector jobs, particularly in the urban economy (and more for females).

Wage trends

Consistent with the above developments in job trends – and the shifts in the demand for labor – we find that the wage structure in the recent period has tilted in favor of the relatively well educated. Their higher returns to education – which has already been noticed in the 1993–1999 period (see Chapter 3 above in particular), would of course affect the regular wage workers, who also show diversity in educational attainment. The data presented in Table 13.8 bear this out. It suggests an increase in wage inequality in the recent period, reinforcing the trend in the previous period analyzed in Chapter 3.

A rather striking and perhaps disturbing result in the table is the decline overall of real wages. The more educated groups have indeed had a positive trend but it is quite low. The growth rate of mean earnings for the illiterate and those with less than middle level education has been substantially negative. This trend in mean earnings does not necessarily imply that there has been absolute decline in all wage groups. We do not have the distribution of wages within each educational category. The decline in mean earnings would be consistent with a shift in the distribution to the lower end – as with the development of a bi-modal structure with relatively more jobs being created at the low and the upper end of the distribution. Further examination of these possibilities must await the availability of unit-level data which will enable us to study the shapes of the wage distribution within skill or education groups.

Table 13.9 presents the data on trends for casual wage earners. The rate of growth although positive only in the rural areas, shows a marked deceleration relative to the previous period and indeed relative to the 1980s as well (see Chapter 3).

Interpreting the wage–employment trends

The clear evidence about the deceleration of the wage rate at the same time when participation and employment rate increased is a puzzle which needs careful analysis. It contrasts dramatically with the experience of the previous period between the 50th and the 55th rounds, when decline in employment rates was associated with a higher rate of wage growth. At first sight both these phenomenon might suggest significant shifts in the supply curve of labor. The observed trends are consistent with a shift upwards of the supply curve in the 1993–1999 period followed by a reverse movement of the function upwards in the 1999–2004 period. Quite apart from the fact that these opposite movements of the supply curve need explanation, two points need to be emphasized which need to form part of any composite hypothesis:

1 The cyclical behavior of the demand for labor. Evidence has been produced in Chapter 3 that the 1993–1999 period saw a contraction in the demand for labor, particularly in the rural sector. It followed the passing of the second wave of the green revolution in agriculture, particularly in the rice-growing states, and was also accompanied by a shift to the use of labor of principal rather than subsidiary status. We have now seen in this Epilogue that there is strong suggestion in the data that the latest period between 1999/2000 and 2004/2005 has witnessed another upsurge in the demand for labor – this time most likely caused by the growth of the non-farm sector both in the rural and urban labor markets. There has also been a significant change in the composition of labor demand – with a marked shift to regular and more educated labor, and away from market for casual labor.

2 It is difficult to distinguish the movements along a fixed supply curve of labor and a movement of the supply curve itself when demand is fluctuating so much in a cyclical pattern. This is because labor is offered in the market – particularly by female workers who have alternative use of their time within the household – not only by changes in the wage rate (movements along the supply curve) but also by changing expectations about employment in the labor market (movements of the supply function itself). Demand and supply curves are interrelated. It is possible that stronger expectations about the labor market might cause rightward shift in the supply curve of labor which might compensate partly for the upward pressure on wage rates.

3 The last point is related to the wider issue of the response of labor supply to the period of time over which wage earnings are being considered. An important statistical issue needs to be stressed in this connection. The daily-wage rate in casual labor market reported by the NSS (and used in this work) is really the derived daily earnings calculated by dividing the total labor earnings over a longer period of time (a week), and hence depends on the number of person-days of employment obtained during the week. Thus the data might show higher growth rate of weekly earnings than that of daily earnings (or what has been called the wage rate in some parts of the discussion). The supply function of labor might be related to earnings over a longer period if time (week, month or season) than the average day.

The upshot of this discussion is that the rate of growth of earning per week or season might show a different trend than that of average daily earnings, and indeed might have been higher in the recent 1999–2004 period. This should be reflected in higher incomes over the month for agricultural laborers. The NSS does not collect data on income. But the trends might be reflected in the data on expenditure per household collected by the NSS. Since agricultural laborers are at the bottom of the income ladder and suffer disproportionately from the incidence of poverty, a relevant variable to examine is the trends in poverty as calculated from the NSS data from the Consumer Expenditure Survey (CES). This is the topic to which we now turn.

Trends in household expenditure, poverty and inequality

The summary report of the NSS, giving selected tables from the Household Expenditure Survey, has given a rough idea of changes in household welfare (mean expenditure per capita) for different parts of the distribution.

Table 13.10 reveals that for the five-year period between the 55th and the 61st rounds there has been an increase in MPCE for all percentile groups, but that the increase is equalizing in the rural sector, but non-equalizing in the urban. In the rural areas the percentage increase is highest in the poorest slab, and declines gradually from 13 to 14 percent increase in the '5 percent or less' group to 6 percent or less in the 80–90 and the 90–95 groups (although the highest slab shows an increase of more than 10 percent). By contrast in the urban areas the percentage increase in MPCE is low in the bottom slab. In fact it is only 2–3 percent for the poorest group (based on column 7 of the table) for the lowest five groups rising to 7–9 percentage for the 80–90 and 90–95 slabs.

The overall impression from these figures and graphs is that the trend observed in the 1990s between the 50th and the 55th rounds is continuing steadily, with a decline in low incomes and the incidence of poverty, but an increase in inequality in the urban areas.

Mahendra Dev and Ravi (2007) in fact have used the 61st round to calculate poverty ratios over the two periods, using the official poverty line and mixed-reference periods (Table 13.11).1 The data bears out the surmise mentioned at the end of the last section. There has indeed been a reduction in poverty ratios, and it is seen that at least one estimate from the new data suggest that there has been an increase in the rate of poverty reduction in the latest period in both sectors, but more so in the rural areas.

The results given above are based on the consumer expenditures based on the 'mixed-reference period', There is, however, the presumption that for an assessment of trends over longer periods of time, there is some case for ignoring the change in the reference period introduced for the first time in the 55th round. Sticking to the original NSS practice of the 30-day reference period it is possible to compare the trends over two decades – the first over the period 1983 and 1993 (between the 43rd and the 50th rounds) and the second between 1993 and 2004 (the 50th and the 61st rounds). Along with many other authors Mahendra Dev considers the first period to be the 'pre-reform' one, and the second the 'post-reform' years. Although based on simple applications of the official poverty lines and on data published by the NSS rather than the original analysis of unit-level data the differences in the trends between the two decades is intriguing. Mahenrda Dev's figures are reproduced in Table 13.12.

The difference between the rural and the urban sectors is confirmed in this table. The reduction in the rate of poverty decline (normalized by base-year values) increased marginally in the rural sector in the post-reform decade, but it was distinctly reduced in the urban. Consistent with results obtained in the book, the increase in inequality was much more pronounced in the urban sector in the post-reform decade (ibid., table 3, p. 510).

Further confirmation of this and other suggestions emerging from the summary tables must await detailed analysis of the unit-level data from the 61st round of the NSS.

Document(s) 13 of 16