Messing about with Poverty Statistics

Sep 19th  2000

Throughout the 1990s, the results of the NSS "thin samples" on household consumption expenditure generated much interest in both academic and policy circles. These results, which suggested at an all-India level that rural poverty did not show any declining trend over the 1990s despite higher rates of aggregate GDP growth, were seen as important inputs into the ongoing policy debate on the effects of the liberalising economic policies instituted by successive governments over the 1990s.
Those who had been questioning the economic reform package have pointed out that these policies have led to a neglect of rural and agricultural investment, resulting in reduced productive employment opportunities in rural areas and to higher food prices, developments likely to be associated with persistent or even increasing rural poverty. They pointed out that the evidence of stagnation or even decline in rural non-agricultural employment was also in conformity with the evidence of persisting rural poverty.
By contrast, proponents of the economic policies of the 1990s suggested that the "thin samples" simply did not allow for such conclusions, and that nothing could be said about rural consumption or poverty until the next large sample results were available. It was also argued by proponents of the official economic reform strategy, that the NSS consumption expenditure estimates from the thin samples were out of sync with the CSO estimates of GDP. Thus they maintained that the association of higher GDP growth rates with the persistence of rural poverty in particular was not a real fact which needed to be understood and addressed, but more of a failure of the statistical system to capture the actual consumption in rural areas.
With this background, obviously the results of the 55th Round (1999-2000) of the NSS, which is the first "large sample" since 1993-94, have been eagerly awaited. Now that the round is complete and the data due to be released later this year, it has also become evident that this NSS Round is important and interesting for another reason. This is that in the 55th Round, a different technique has been used to establish household consumption levels. The basic change is in terms of a change in the reference period.
Since the 1950s, the NSS consumption surveys have been using an uniform reference period of one month. Since the interviews are evenly spread out over the year, problems of seasonality were ironed out. However, in recent years the NSS has revived the issue of whether a one week reference period is more suitable for determining non-durable consumption than the one month reference period currently used.
This is not a new issue : indeed, it is a question which has been of concern to the NSS since the very inception of the surveys in the early 1950s. In fact, in the formative years of the NSS, considerable attention was paid to the length of reference period suitable for ascertaining the correct level of consumption of different items of goods and services, and a special report was brought out covering the period April 1951 to March 1954 on the suitability of reference period.
Most interestingly, the NSS under guidance of P.C. Mahalanobis had carried out a special investigation into this very issue in March-April 1952, based on 1254 households of 76 villages of West Bengal. The households were divided into two groups. For one group, consumption details were procured by actual weighing of food items (clean rice, pulses, sugar and salt) by field staff. For the other group, data collection was by questioning, and here again the group was divided equally between those for whom the questions pertained to a reference period of one week and those for whom the reference period was one month.

 | 1 | | 3 | 4 | Next Page >>

Site optimised for 800 x 600 and above for Internet Explorer 5 and above