OF the various measures devised to measure the incidence, depth and severity of poverty, the index most often quoted is the "head-count index". Unfortunately, changes in the methodology adopted by the NSS Expenditure Surveys prior to 1972-73 make estimates of the head-count index during the earlier years non-comparable.
However, allowing for this inadequacy, it appears that the following broad conclusions can be arrived at: first, that the incidence of poverty declined between the mid-1950s and early 1960s; second, that starting with the bad harvests of the mid-1960s we see a sharp rise in the incidence of poverty which touched a post-Independence peak by the late 1960s; third, that starting then and especially after 1975, India had witnessed a significant decline in poverty incidence till 1989-90; and finally, that the years of stabilisation and structural adjustment have witnessed an initial sharp rise in poverty till 1992 and a subsequent decline in 1993-94, but that the 1993-94 figure appears higher than in 1989-90, pointing to a reversal of the earlier declining trend.
The literature examining the determinants of these poverty trends has identified three factors: first, agricultural output growth, leading to a rise in agricultural value added per head of the rural population and contributing to an increase in wages; second, price movements, especially movements in the price of food, which determine the value of the inflation-adjusted earnings of the poor; finally, government expenditure, which by directly and indirectly expanding non-agricultural employment, provides an impetus to a rise in money wages. While all three factors played a role in explaining the rise or decline in poverty incidence during different time periods, the relative significance of each and their directional effect varied across time. It has been argued that the decline in poverty incidence during the 1950s was due more to the stability of food prices (ensured in large part with PL 480 imports) and the rise in government expenditure. Inflation and a cutback in government expenditure after the mid-1960s contributed to the subsequent rise in poverty. However, the decline in poverty incidence after the mid-1970s appears to be attributable more to agricultural output growth and increases in state expenditure.
There is broad agreement on these trends and their determinants among economists in India today. However, disagreement increases as the focus of attention shifts to the years of economic reform. The disagreement on trends in poverty is based on the data set chosen. The most quoted estimates of poverty in India are based on data yielded by the "extensive" Household Consumer Expenditure Surveys, which since 1972-73 have been conducted once every five years. The two most recent surveys relate to 1987-88 and 1993-94. However, since 1986-87, the Government has revived an earlier practice of conducting consumer expenditure surveys on a more regular basis, with such surveys being available annually since 1988-89. The difference between these 'annual' surveys and the 'quinquennial' surveys is that the former are characterised by much smaller sample sizes. A smaller sample implies that the certainty with which the poverty ratio can be precisely identified tends to be lower purely because the range within which the actual poverty ratio can be estimated to lie tends to be larger than that for estimates based on bigger samples. Nonetheless, these estimates are of considerable value.
Disregarding this statistical truth, the Government, supported by sections of the media, has been claiming that the so-called 'thin' samples which the National Sample Survey Organisation (NSSO) surveys at substantial cost in the years between its larger quinquennial exercises should now be completely disregarded. The reason for this insistence is not hard to decipher. Using those annual surveys, economists have found that the poverty ratio had risen quite sharply during the early years of liberalisation. For example, even World Bank economists have indicated in a recent study (India: Achievements and Challenges in Reducing Poverty, World Bank, May 27, 1997) that based on the Planning Commission's poverty line (Rs. 49.09 and Rs. 56.64 at 1973-74 prices in rural and urban areas respectively), the poverty ratio which declined from 39.23 per cent in rural India and 36.60 per cent in urban India in 1987-88 to 36.43 and 32.76 per cent respectively in 1990-91, subsequently rose to 37.42 and 33.23 per cent in 1991 and 43.47 per cent and 33.73 per cent in 1992.
The massive increase in rural India suggests that even allowing for the greater variance associated with thin samples, poverty did increase in the early reform years. Further, even though the incidence of poverty fell in 1993-94 to 36.66 and 30.51 per cent, the figure for rural India is still higher than in 1990-91 (36.43 per cent) and 1989-90 (34.30 per cent). Not happy with that trend, the Government has decided to disregard the thin samples.
If we do include the thin samples in the analysis, the main conclusions that follow are: First, the rural poverty ratio at the all-India level, and in most States, fell sharply between 1973-74 and 1986-87 with subsequent years showing no clear trend in either direction. Thus if the liberalisation of the economy is dated back to the mid-1980s, the entire period since then is characterised by a slowdown in the pace of poverty reduction.
Second, there was a marked increase in rural poverty in 1992 as compared with either 1989-90 or 1990-91. This was true of almost every State except Kerala where poverty fell continuously. At the all-India level, the number of poor increased by over 60 million between 1990-91 and 1992, or in the first 18 months of the reform period. Third, there was an appreciable moderation in poverty between 1992 and 1993-94 in both absolute and proportionate terms. Nonetheless, as compared with 1989-90, the rural poverty ratio was higher (36.7 as compared with 34.3 per cent) in 1993-94.