How real is the secular decline in rural poverty

Abstract

here is a consensus that poverty inIndia has declined since the mid-1970s. The recent annual assess-ment of the World Bank on India com-mends that “India scored great success inparing down the incidence of rural povertyfrom 55 per cent in the early 1970s to lessthan 35 per cent by the late 80s” [WorldBank, 1999: p vii]. This consensus is based onestimates of headcount measure of povertydefined with reference to a calorie normand obtained using the National SampleSurvey (NSS) data on household consumerexpenditure distribution. Broadly, all thestudies agree that poverty, rural poverty inparticular, has finally declined since the1970s: (i) Rural poverty estimates declinedfrom about 55 per cent in the 1950s and1960s to about 35 per cent in the 1980sas a result of agricultural growth and publicintervention; (ii) fluctuations, whatsoever,in rural poverty estimates are due to fluc-tuations in agricultural performance causedby droughts, etc (Figure 1).This paper examines this consensus, first,on methodological grounds and then onsubstantial considerations. Further, not-withstanding several advances in measure-ment, the bulk of the literature representsa mechanical exercise without adequatelyappreciating how the data peculiarities andstructural changes in the rural economycould have affected measurement of pri-vate household consumption and its valu-ation. This naturally raises the question:How valid are the findings on reductionin rural poverty in India? How do weexplain the fluctuations in rural povertyestimates? Does the observed decline inrural poverty estimates really represent agrowth in real consumption as (implicitly)hypothesised by the various studies? Doesthe data underlying the summary measuresof poverty reflect any growth in real con-sumption measured in physical quantitiesof cereals? If not, why? Anticipating theresults reported in this paper, we concludethat the widely believed poverty reductionrepresents largely a reduction in over-estimation of poverty for the initial years,and increasing underestimation of povertyfor the recent years. This is corroboratedby the fact that the fall in poverty has notmeant a proportionate and secular rise inper capita consumption in real terms. Evenmore significant is the fact that during thisperiod (1972-94), the proportion of caloriedeficient rural population has increasedfrom 65 per cent to 75 per cent.The paper is organised as follows: Sec-tion I spells out the notion of poverty andthe implicit assumptions involved inmaking poverty estimates at a point of timeand over time. Section II describes the NSSdatabase and how far it meets the require-ments of the methodology being followedin poverty estimation and their implica-tions. This section covers the NSS featuresthat would explain the observed periodicfluctuation/reduction in poverty estimates.It deals with the concept of consumerexpenditure and the accounting procedure;with the survey schedules; and with themethods of valuation with special empha-sis on dual valuation of consumption andits implications for poverty estimates.Section III attempts to empirically verifysome of the statistical implicationsdiscussed in the previous section. The finalsection sums up the paper.