MacroScan (Printable Version) - Rural Employment Trends from the Census

It is already common knowledge that employment growth has been a major casualty of the decade of liberalising reform. The inadequate pace of employment generation has been so evident that even the Finance Ministry’s annual Economic Survey, which has rarely devoted much attention to the issue in the recent past, has been forced to take note.

Of course, official sources have attributed the slowdown in employment generation to the slowdown in the rate of increase in the labour force, which they claim is a combination of lower population growth and greater involvement in education on the part of the 15-19 age cohort. Both of these are obviously to be welcomed. However, it was already clear from the NSS survey data that these two factors are not adequate to explain the slowdown in aggregate employment generation, especially in the rural areas.

In addition, because these assessments of employment generation were based only on the National Sample Surveys, many had argued that they were not sufficient to indicate the real trends over the 1990s. For this reason, the results of the 2001 Census have been eagerly awaited.

The first estimates from the 2001 Census relating to employment, work participation and even employment by industrial category are now available. They do indicate a slightly different pattern in terms of employment growth, from the trends emerging from the NSS. However, the pattern that emerges may be even more worrying from the point of view of macroeconomic strategy.

In this piece we consider only rural trends; the evidence pertaining to urban employment will be considered in a later edition of Macroscan. Chart 1 describes the trends in worker-population ratios in the rural areas at an All-India level. It is immediately apparent that there is no drop in this ratio in 2001, unlike the 1999-2000 NSS which showed a decline in this ratio.



Thus, according to the NSS, rural male employment fell from 55.3 per cent of the rural male population in 1990-91 to 53.1 per cent in 1999-2000, while that for females remained broadly the same at around 29 per cent. The Census data show apparent stability in male worker-population rates over this period, especially when compared to earlier decades. And the ratio for rural females has gone up quite significantly.

This would appear to bely the more pessimistic conclusions about rural employment generation that had emerged from the NSS data. However, it turns out that the aggregate data refers to both main and marginal workers, and a disaggregated look provides a very different analysis. If only main workers are considered, the decade 1991-2001 has witnessed a very sharp decline in the proportion of main workers to total population. As Chart 2 indicates, this is especially marked for male main workers, whose share has fallen by nearly 7 percentage points.



The Census category “main workers” refers to those who had worked in some economic activity for the major part of the year, that is for a period of six months (183 days) or more. Work of course, is defined as participation in any economically productive activity, but this still excludes a range of unpaid household work. “Marginal workers” refers to those who had worked for some time during the previous year, but not for the major part, i.e. less than 183 days. They are therefore mutually exclusive categories, analagous but not identical to the NSS categories of “principal” and “subsidiary” occupations.

It is not known whether the definition of work has been used more flexibly in the 2001 Census to incorporate some forms of unpaid labour, which were previously not included. But even if this has occurred, it has clearly not been sufficient to increase the worker population ratios significantly in the rural areas. What is clear is that there is a substantial slowdown in the generation of employment that would qualify for “main work”, in other words that there has not been an increase in the availability of employment that would keep people productively occupied for half a year or more.

This is quite compatible with the NSS evidence on decline in terms of the usual status definition of employment, as shown in Chart 3. Indeed, the Census data suggest an even sharper downward shift, especially for rural males.

However, if does suggest a different picture from that mentioned in the Census of India’s own description, which argues that there is a “substantial increase in female work participation rate”. As we have seen, the increase is actually quite small, and in any case is composed entirely of an increase in the proportion of marginal workers. Main workers have actually gone down as a share of population even in the case of females.

The pattern is repeated even with a disaggregated analysis of rural employment in the states, described in Table 1 (total worker-population ratios) and Table 2 (main worker-population ratios). For the male population in most states, total worker to population ratios remained broadly the same over the decade, with the exception of a substantial increase in Kerala and marginal increases in Karnataka, Tamil Nadu and West Bengal. There were declines even in total worker-population ratios in Orissa, Punjab and Utter Pradesh.



However, as Table 2 shows, the decline in main worker-population ratios was spread uniformly across all the states. Every single state showed a decline in this ratio by at least 3-4 percentage points for rural males. In some states, such as Orissa and Uttar Pradesh, the decline was especially sharp and amounted to as much as 10 or more percentage points, or more than 20 per cent of the proportion of those in such employment.



For rural women, the picture was only slightly more complicated. In terms of total worker-population ratios, there were increases in most states barring Kerala, and huge increases in Punjab and Haryana as well as to a lesser extent in Rajasthan. It is not clear to what extent such increases reflect better recording and recognition of women’s work in these northern states. However, once again the overall pattern in the states reflects essentially the increase in only marginal work even for rural women. Main worker-population ratios declined for rural women in every state except for the three northern states of Punjab, Haryana and Rajasthan, where as mentioned above, improved recording of women’s work may have been the decisive factor.

This fact of falling main worker-population rates has an important bearing on the subsequent analysis with respect to the type of economic activity that the rural workforce is engaged in. One feature which always excites much discussion is the degree of diversification of the rural workforce, or the proportion of agriculture in total rural employment. This has been one feature which also appears stubbornly resistant to change according to Census results, compared with the NSS data which has shown greater variation over time in this regard. Thus, Chart 4 shows that the extent of decline in agriculture’s share of total rural employment has been gentle and less than moderate, and remained very high at around four-fifths of the rural workforce.



Note that this refers only to main workers, and indeed in all the previous Censuses, the industrial classification of workers has been presented only for main workers. However, for the 2001 Census the data presented so far relates only to main plus marginal workers together. This renders the latest data completely non-comparable with the earlier series. It is not known why the Census of India decided to present the data in this manner, which does not allow for comparison or estimates of trends over time.

This is especially unfortunate as the latest Census shows the total employment (main plus marginal) in agriculture to be fairly low. There are indicated in Charts 5 and 6. However, because they relate to all workers, and because as we have seen, there is a very substantial shift in favour of marginal workers and a decline in the share of main workers, this cannot at all be compared with the classifications of earlier Census data based on main workers only.





This has not prevented the organisation from making statements which are not justified by the data that it has presented. According to the Census of India’s website, “The results from the 2001 Census clearly suggest a shift in the composition of the labour force from a predominantly agriculture to a moderately non-agriculture sector”. Such a conclusion would only be justified if it could be seen that the share of agriculture has fallen for main workers only, or that the share of agriculture was higher for both main and marginal workers together in the 1991 Census.

Further, since information relating to type of employment (self-employment, regular or casual work) is not part of this dataset, we cannot tell what form even the new marginal work in non-agriculture appears to be taking. The important point to note is that we do not have adequate evidence to declare that there is actually a diversification of rural employment away from agriculture, and certainly cannot make the further judgement that such diversification is of the progressive variety associated with a dynamic economy.

In sum, it is clear that employment generation in the rural sector has been much less than adequate even after all the increases in marginal workers are accounted for. Charts 7 and 8 indicate the absolute increase in number of main and marginal rural workers respectively. As Chart 9 shows, the number of male main workers increased very little over the decade, by just above 5 million, while the increase in female main workers was less than 3 million. By contrast, the number of marginal workers increased by 26 million for rural men and 27 million for rural women.







This is confirmed by the annual growth rates of employment reflected in the absolute increases, as described in Table 3. While the aggregate employment growth appear to be slightly better than described through the NSS Surveys, at 1.7 per cent for males and 3.2 per cent for females, the increase in main employment is much lower than even the increase in usual status employment indicated by the NSS. Indeed, it is less than half of one per cent per annum for both men and women. So the story of collapse of rural employment generation in the 1990s, which had emerged from the NSS Surveys, appears to be largely corroborated by the latest Census data as well.



Given the overall stagnation in worker-population ratios discussed above, what this suggests is that the vast bulk of additional jobs generated in the countryside over this period have not provided employment for even half a year to rural workers. Obviously, many aspects remain to be explored and it is necessary to await the further and more detailed results of the Census 2001 for proper analysis. But one thing that the results released thus far show very clearly is an intensification of the process of marginalisation of the rural workforce.



Themes > Features	11.04.2002


Rural Employment Trends from the Census

It is already common knowledge that employment growth has been a major casualty of the decade of liberalising reform. The inadequate pace of employment generation has been so evident that even the Finance Ministry’s annual Economic Survey, which has rarely devoted much attention to the issue in the recent past, has been forced to take note. Of course, official sources have attributed the slowdown in employment generation to the slowdown in the rate of increase in the labour force, which they claim is a combination of lower population growth and greater involvement in education on the part of the 15-19 age cohort. Both of these are obviously to be welcomed. However, it was already clear from the NSS survey data that these two factors are not adequate to explain the slowdown in aggregate employment generation, especially in the rural areas. In addition, because these assessments of employment generation were based only on the National Sample Surveys, many had argued that they were not sufficient to indicate the real trends over the 1990s. For this reason, the results of the 2001 Census have been eagerly awaited. The first estimates from the 2001 Census relating to employment, work participation and even employment by industrial category are now available. They do indicate a slightly different pattern in terms of employment growth, from the trends emerging from the NSS. However, the pattern that emerges may be even more worrying from the point of view of macroeconomic strategy. In this piece we consider only rural trends; the evidence pertaining to urban employment will be considered in a later edition of Macroscan. Chart 1 describes the trends in worker-population ratios in the rural areas at an All-India level. It is immediately apparent that there is no drop in this ratio in 2001, unlike the 1999-2000 NSS which showed a decline in this ratio. Thus, according to the NSS, rural male employment fell from 55.3 per cent of the rural male population in 1990-91 to 53.1 per cent in 1999-2000, while that for females remained broadly the same at around 29 per cent. The Census data show apparent stability in male worker-population rates over this period, especially when compared to earlier decades. And the ratio for rural females has gone up quite significantly. This would appear to bely the more pessimistic conclusions about rural employment generation that had emerged from the NSS data. However, it turns out that the aggregate data refers to both main and marginal workers, and a disaggregated look provides a very different analysis. If only main workers are considered, the decade 1991-2001 has witnessed a very sharp decline in the proportion of main workers to total population. As Chart 2 indicates, this is especially marked for male main workers, whose share has fallen by nearly 7 percentage points. The Census category “main workers” refers to those who had worked in some economic activity for the major part of the year, that is for a period of six months (183 days) or more. Work of course, is defined as participation in any economically productive activity, but this still excludes a range of unpaid household work. “Marginal workers” refers to those who had worked for some time during the previous year, but not for the major part, i.e. less than 183 days. They are therefore mutually exclusive categories, analagous but not identical to the NSS categories of “principal” and “subsidiary” occupations. It is not known whether the definition of work has been used more flexibly in the 2001 Census to incorporate some forms of unpaid labour, which were previously not included. But even if this has occurred, it has clearly not been sufficient to increase the worker population ratios significantly in the rural areas. What is clear is that there is a substantial slowdown in the generation of employment that would qualify for “main work”, in other words that there has not been an increase in the availability of employment that would keep people productively occupied for half a year or more. This is quite compatible with the NSS evidence on decline in terms of the usual status definition of employment, as shown in Chart 3. Indeed, the Census data suggest an even sharper downward shift, especially for rural males. However, if does suggest a different picture from that mentioned in the Census of India’s own description, which argues that there is a “substantial increase in female work participation rate”. As we have seen, the increase is actually quite small, and in any case is composed entirely of an increase in the proportion of marginal workers. Main workers have actually gone down as a share of population even in the case of females. The pattern is repeated even with a disaggregated analysis of rural employment in the states, described in Table 1 (total worker-population ratios) and Table 2 (main worker-population ratios). For the male population in most states, total worker to population ratios remained broadly the same over the decade, with the exception of a substantial increase in Kerala and marginal increases in Karnataka, Tamil Nadu and West Bengal. There were declines even in total worker-population ratios in Orissa, Punjab and Utter Pradesh. However, as Table 2 shows, the decline in main worker-population ratios was spread uniformly across all the states. Every single state showed a decline in this ratio by at least 3-4 percentage points for rural males. In some states, such as Orissa and Uttar Pradesh, the decline was especially sharp and amounted to as much as 10 or more percentage points, or more than 20 per cent of the proportion of those in such employment. For rural women, the picture was only slightly more complicated. In terms of total worker-population ratios, there were increases in most states barring Kerala, and huge increases in Punjab and Haryana as well as to a lesser extent in Rajasthan. It is not clear to what extent such increases reflect better recording and recognition of women’s work in these northern states. However, once again the overall pattern in the states reflects essentially the increase in only marginal work even for rural women. Main worker-population ratios declined for rural women in every state except for the three northern states of Punjab, Haryana and Rajasthan, where as mentioned above, improved recording of women’s work may have been the decisive factor. This fact of falling main worker-population rates has an important bearing on the subsequent analysis with respect to the type of economic activity that the rural workforce is engaged in. One feature which always excites much discussion is the degree of diversification of the rural workforce, or the proportion of agriculture in total rural employment. This has been one feature which also appears stubbornly resistant to change according to Census results, compared with the NSS data which has shown greater variation over time in this regard. Thus, Chart 4 shows that the extent of decline in agriculture’s share of total rural employment has been gentle and less than moderate, and remained very high at around four-fifths of the rural workforce. Note that this refers only to main workers, and indeed in all the previous Censuses, the industrial classification of workers has been presented only for main workers. However, for the 2001 Census the data presented so far relates only to main plus marginal workers together. This renders the latest data completely non-comparable with the earlier series. It is not known why the Census of India decided to present the data in this manner, which does not allow for comparison or estimates of trends over time. This is especially unfortunate as the latest Census shows the total employment (main plus marginal) in agriculture to be fairly low. There are indicated in Charts 5 and 6. However, because they relate to all workers, and because as we have seen, there is a very substantial shift in favour of marginal workers and a decline in the share of main workers, this cannot at all be compared with the classifications of earlier Census data based on main workers only. This has not prevented the organisation from making statements which are not justified by the data that it has presented. According to the Census of India’s website, “The results from the 2001 Census clearly suggest a shift in the composition of the labour force from a predominantly agriculture to a moderately non-agriculture sector”. Such a conclusion would only be justified if it could be seen that the share of agriculture has fallen for main workers only, or that the share of agriculture was higher for both main and marginal workers together in the 1991 Census. Further, since information relating to type of employment (self-employment, regular or casual work) is not part of this dataset, we cannot tell what form even the new marginal work in non-agriculture appears to be taking. The important point to note is that we do not have adequate evidence to declare that there is actually a diversification of rural employment away from agriculture, and certainly cannot make the further judgement that such diversification is of the progressive variety associated with a dynamic economy. In sum, it is clear that employment generation in the rural sector has been much less than adequate even after all the increases in marginal workers are accounted for. Charts 7 and 8 indicate the absolute increase in number of main and marginal rural workers respectively. As Chart 9 shows, the number of male main workers increased very little over the decade, by just above 5 million, while the increase in female main workers was less than 3 million. By contrast, the number of marginal workers increased by 26 million for rural men and 27 million for rural women. This is confirmed by the annual growth rates of employment reflected in the absolute increases, as described in Table 3. While the aggregate employment growth appear to be slightly better than described through the NSS Surveys, at 1.7 per cent for males and 3.2 per cent for females, the increase in main employment is much lower than even the increase in usual status employment indicated by the NSS. Indeed, it is less than half of one per cent per annum for both men and women. So the story of collapse of rural employment generation in the 1990s, which had emerged from the NSS Surveys, appears to be largely corroborated by the latest Census data as well. Given the overall stagnation in worker-population ratios discussed above, what this suggests is that the vast bulk of additional jobs generated in the countryside over this period have not provided employment for even half a year to rural workers. Obviously, many aspects remain to be explored and it is necessary to await the further and more detailed results of the Census 2001 for proper analysis. But one thing that the results released thus far show very clearly is an intensification of the process of marginalisation of the rural workforce.



© MACROSCAN 2002