Growth and poverty in rural India

POLICY RESEARCH WORKING PAPER 1405 .'~ g tur Growth and Poverty Higheragcul es -reduced absolute poverty-in in Rural India rural India, born by raising smaliholder productivi and Marti Ravary increasing ,realagricultural Martin Ravallion -; ;--: . G a Da wages. BiUt gains to the poor Gaurav Datt i.-- were far smaller in the short: run-than in the iong run.. Badcground paper for World Daveopment Report 1995S The World Bank Office of the Vice Preident - Deve[opment Economic A - January 1995  POLICY RI!SeARCH WORKING PAPER 1405 Summary findings Unlike most developing countries, consistent poverty yields, which benefited poor people both directly and measures for India can be tracked over a long time. through highcr rcal agricultural wages. And thc benefits Ravallion and Datt used 20 houschold surveys for rural from higher yiclds were not confined to those near the India for the years 1958-90 to measure the effects of poverty line - the poorest also benefited. agricultuml growth on rural poverty and on the rural The process through which India's rural poor labor market and to find out how long it takes for the participate in the gains from agricultural growth takes effects to be felt. time, although about half of the long-run impact comes They found that mcasures of absolute rural poverty within three years. responded elastically to changes in mean consumption. The long-run elasticity of the head-count index to farm But agricultural growth had no discernible impact - yield was over 2 - of which 40 percent came through either positive or negative - on the share of total wages. Short-run elasticities were far smaller. consumption going to the poor. Inflation adversely affected the rural poor by eroding For the rural poor, Ravallion and Datt artribute the their real wages in the short run. long-run gains from growth to higher average farm This paper - a product of the Office of the Vice President, Development Economics - is one in a series of background papexs prepared for WorldDevelopment Report 1995 en labor. Copies of the paperare available free from the World Bank, 1818 H Street NW, Washington, DC 20433. The study was funded by the Baaik's Research Support Budget under the research project "Poverty in India, 1950-90" (RPO 677-82). Plcase contact the World Development Report office, room T7-101, cxtension 31393 (34 pages). January 1995. The Policy Researcb Workixg Paper Series disseminates the findings of work sn prOgress to encourage the exchange of ideas about development isswes An objective of thesries s toget the fiings outquickly. cven if thc presentaionsare less than fdly polished The papers cany the names of the authorsand should be usedandciftdaccordingly. The findings, interpretations, and conclusions are the autbors' own and should not be attributed to the World Bank its Executive Board of Directors, or any of its member countries. Produced by the Policy Research Dissemination Centcr  Growth and Poverty in Rural India Martin Ravallion and Gaurav Daet Policy Research Department, World Bank, 1818 H Street NW, Washington DC. These are the views of the authors, and should not be attributed to the World Bank. The support of the Bank's Research Committee (under RPO 677-82) is gratefully acknowledged. The authors are also grateful to Berk Ozler for help in setting up the data set used here. The comments of James Boyce. Lyn Squire, Dominique van de WaUe, Michael Walton, and seminar participants at Cornell University, the International Food Policy Research Institute and the World Bank are gratefully acknowledged.  1 Introduction The scope for economic growth to make a real difference in the lives of the developing world's poor has been the subject of (often vociferous) debates in both academic and policy circles.' Measuring the long-rm benefits to poor people from economy-wide changes calls for a time series of representative household-level surveys; yet such surveys are sporadic at best for most countries. India is an exception. There one can trace distributional impacts o ier a long period using reasonably comparable and nationally representative surveys of consumption.2 Here we use these surveys to examine how much India's rural poor have benefited from agricultural growth, what role the labor market has played, and whether the impacts were distributionally biased one way or another. We depart from past analyses for India and elsewhere in five main ways: i) The identification of distributional impacts. It is theoretically possible for a growth process to have sufficiently adverse effects on inequality that poverty increases, and some have argued that this is also the reality of India's rural development. Another view-often termed "trickle down"-denies this, but still allows that distribution may worsen even though on balance the poor gain somewhat. By contrast, some growth processes can entail favorable distnbutional I Recent surveys spanning the wide range of views concerning the impacts of growth in farm productivity on rural poverty include Saith (1990), Singh (1990) and Lipton and Ravallion (1994). Differences between countries or time periods may account for some of these differences, but certainly not all. Contrast, for example, Ahluwalia's (1978, p.320) conclusion that "..there is evidence of some trickle down associated with agricultural growth" with Saith's (1981, p.205) claim that "there can be little doubt that current growth processes have served as generators of poverty"; both were using data for the same country (India) over roughly the same period (1957-73). 2 It seems that not even for the U.S.A. can one track poverty measures as well. The were 15 Consumer Expenditure Surveys for the U.S.A. over the same period for which we have 20 National Sample Surveys for India. 1  shifts, and consequently larger gains to the poor than a "trickde down'. Here we propose a method for identifying and testing the distributional impacts of growth on poverty. ii) The role plaved by the labor market. Past work has often ignored or down-played the rural labor market's capacity to transmit the benefits of technical progress to the poor. Marked differences in emphasis can be found in policy-oriented discussions on this point.3 Here we aim to quantify the role played by real wages in distributing the gains from aggregate growth. iii) Allowing for dynmic effects. The dynamics of the distributional impacts of growth are of obvious interest, though the topic has received surprisingly little attention. Past models have analyzed the consumption-based poverty measures within a static framework, despite theory and evidence to the contrary. Stickiness in the adjustment of poverty measures also has an important implication: long-run responses can far exceed those in the short-run. iv) The time period of analsis. Much of the scholarly debate for India has focused (often-though not always-for lack of data) on periods of rather little growth; these data may have low power in testing the effects of growth on poverty (Srinivasan, 1985). We use a new data set embracing a period (since the mid-1970s) of higher agricultural growth. v) The treatment of survey spacing. Unlike past work, we deal consistently with the uneven survey spacing in the estimation. This can matter to estimating dynamic effects. The following section outlines the model we will be using. Section 3 then presents our results, while conclusions can be found in section 4. Appendix 1 describes our data and its sources, while Appendix 2 discusses related work in the literature on rural poverty in India. I Contrast, for cxample, the IFAD (1992) report with World Bank (1990). The former emphasizes the scope for reducing rural poverty by developing smallholder agriculture, and pays little attention to the role of the unskilled labor market; by contrast the World Bank report emphasizes the importance of positive employment and wage effects in achieving pro-poor growth (though not to the exclusion of other channels). 2  2 Modelling bnpats of growth on poverty 2.1 Characterizing alternative growth processes A poverty measure (P) can be written as a non-increasing function of the mean (p), and a vector of parameters $. = (x ,--,) for the Lorenz curve:4 P = P(P, a) (1) Let the Lorenz parameters also vary with the mean 2L(p) (other variables are ignored for now) and consider the effect of an increase in the mean.' Assume that all these functions are differentiable, and let subscripts denote partial derivatives. We can distinguish three cases: i) Immis_rizin m growth; dP/dp = Pp + EP