WORLD BANK’S NEW POVERTY ESTIMATESEconomic & Political Weekly EPW OCTOBER 25, 200831A Global Perspective on Poverty in India Martin RavallionIn 2005, one in three of the people in the world who consumed less than $ 1.25 a day (at 2005 purchasing power parity) lived in India – more than any other country. They accounted for about 40 per cent of India’s population. Twenty-five years earlier, 60 per cent of India’s population lived below the same real line. While this is clear progress, India’s long-term pace of poverty reduction by this measure is no more than average for the developing world, excluding China. This article first discusses the methodology underlying the World Bank’s recent revised estimates of global poverty and then analyses the Indian numbers.When discussing absolute pover-ty in the world as a whole, there is a case for using a common standard for all countries. It would clearly be questionable to conclude that an Indian living above India’s poverty line, but below (say) Brazil’s, is not poor in a global context, when we count someone at the same standard of living in Brazil as poor. Such international comparisons are never going to be easy as there are inevitably some confounding contextual factors, such as differences in poor peoples’ access to non-market goods and perceptions of relative deprivation in relatively well-off countries. But it is of interest to see what the best available data tell us about the extent of absolute poverty in the world as a whole when assessed by a common standard in terms of command over commodities. And it is hardly surprising that individual countries, including India, are interested in seeing how they are faring in the evolving global picture of absolute poverty. An ongoing research project at the World Bank has tried to offer such a global account of the extent of absolute poverty and to track how the picture evolves over time. The effort has been going on for almost 20 years, starting with a background paper [Ravallion et al 1991] done for the 1990 World Development Report [World Bank 1990]. The latest estimates are re-ported in Chen and Ravallion (2008a).This article summarises the findings and discusses how India fares in this global picture of poverty. The article first explains how we measure global poverty; the box summarises the key steps, which are explained in the text of the article. The article then turns to the implications for India.Prices for Measuring Poverty Given a poverty line that accords with perceptions of what constitutes “poverty” inagiven country, a guiding principle for poverty measurement within that country (comparing different regions say) has been that two people with the same real consumption level should be treated the same way no matter where they live. The real value of the absolute poverty line should be the same in different places. The World Bank’s global poverty meas-ures apply the same principle to the world as a whole. Thus, along with household surveys, a key input to measuring global poverty is data on prices. Only then can we say whether one can buy more with (say) 5 yuan per day in China than Rs 10 per day in India. The exchange rates that equate purchasing power over commodities (both internationally traded and non-traded) are called purchasing power parity rates, or PPPs. The rationale for using a PPP rather than the market exchange rate stems from what is called the Balassa-Samuelson effect in international econom-ics. This recognises that market exchange rates, which tend to equate purchasing power in terms of internationally traded goods, are deceptive for measuring real incomes in developing countries, given that some commodities are not traded; this includes services but also many goods, including some food staples. Furthermore, there is a systematic effect, stemming from the fact that low real wages in developing countries entail that labour-intensive non-traded goods tend to be relatively cheap. Market exchange rates can greatly under-estimate real income in poor countries. The Balassa-Samuelson effect provided the motivation for the International Com-parison Program (ICP), which collects prices from a large sample of outlets in each country in each “benchmark year”. The results of the new 2005 ICP were released this year [World Bank 2008a, b], superseding the 1993ICP, which was the previous benchmark year used for global poverty measurement. The aggregate poverty count will not in general be independent of the reference year, even if the underlying data are the same; this is a widely acknowledged fea-ture of international comparisons, whether poverty or national output. However, this is a moot point given that the data have changed so much from oneICP round to These are the views of the author and need not reflect those of the World Bank or any affiliated organisation. Gaurav Datt, Dominique van de Walle and Tara Vishwanath provided helpful comments on an earlier version of this article.Martin Ravallion (mravallion@worldbank.org) is at the Research Department of the World Bank, Washington.
WORLD BANK’S NEW POVERTY ESTIMATESOCTOBER 25, 2008 EPW Economic & Political Weekly32Microsoft adThis is the first time in 20 years that India has participated in the ICP, so a fairly sizeable revision was possible, and may not be too surprising.1
WORLD BANK’S NEW POVERTY ESTIMATESOCTOBER 25, 2008 EPW Economic & Political Weekly34ICP rounds [World Bank 2008b]. Thus the PPP conversion is only done once for a given country, and all estimates are revised (back to 1981) consistently with the data for that country. So the PPPs serve the role of locating the residents of each country in the “global” distribution, but we do not mix the new PPPs with those from previous ICP rounds. Having converted the international poverty line atPPP to local currency in 2005 we convert it to the prices prevailing at each survey date using the country-specific official consumer price index (CPI). We then apply this poverty line to the survey data; the latest estimates use 670 household surveys for 116 countries. Interpolation methods are used to “line-up” the survey-based estimates at the common “reference years” including 2005 (calendar year).For the world as a whole, we estimate that about 1.4 billion people – one quarter of the population of the developing world– lived below $ 1.25 a day in 2005 [Chen and Ravallion 2008a]. While that is far more poverty than our past work suggested – see for example, Chen and Ravallion(2004) – we find that the deve-loping world is making progress against extreme poverty. Tracking trends over time is not easy, given that survey data availability and (probably) data quality tend to deteriorate as one goes further back in time. Using the best available consumer price indices and household surveys, we estimate that in 1981, 52 per cent of the population, 1.9 bil-lion, lived below $ 1.25. Data are poor for some regions (such as Africa and eastern Europe and central Asia) for the 1980s. Focusing on 1990 as the base date, we find that 42 per cent lived below $ 1.25 per day, 1.8 billion people. The trend over time is similar to past estimates. The trend rate of decline in the percentage living below $1.25 a day is almost exactly one percentage point per year (strictly it is 0.98 percentage points using a regression on time, with a stand-ard error of 0.06 per cent) – up from about 0.8 percentage points per year using our old estimates [Chen and Ravallion 2008a]. The trend is virtually identical if one starts the series in 1990.An important yardstick for assessing the developing world’s performance against poverty has been provided by the first Millennium Development Goal (MDG1), which is to halve the 1990 incidence of ex-treme poverty by 2015. Simply projecting the trend measured by Chen and Ravallion (2008a) forward, the estimated propor-tion of the population living below $ 1.25 a day in 2015 is 16.9 per cent (standard error of 1.5 per cent). Given that the 1990 poverty rate was 41.7 per cent, the developing world as a whole appears to be on track to achievingMDG1. However, the developing world outside China is not on track for reaching the goal, which will require a higher rate of progress [Chen and Raval-lion 2008a]. And, given the lags in survey data availability, these calculations do not allow for the effect of the higher food and fuel prices since 2005, which have probably set back progress by a few years at least.So far the discussion has focused on the “headcount index”, given by the propor-tion of the population of living in house-holds with consumption (or income) per capita below the poverty line. This is the most popular measure in practice, but it is known to have a number of conceptual problems, including the fact that if a poor person becomes poorer then the index does not change. A better measure from this point of view is the “poverty gap (PG) index” defined as the mean distance be-low the poverty line as a proportion of the line where the mean is taken over the whole population, counting the non-poor as having zero poverty gaps. ThePG index for the developing world as a whole 2005 is 7.7 per cent for the $1.25 line [Chen and Ravallion 2008b]. To put this in perspective, world (including in the Organisation for Economic Cooperation and Development countries) GDP per capita in 2005 at 2005PPP was $ 24.58 per day, implying that the global aggregate poverty gap was 0.33 per cent of global GDP using the $ 1.25 line.4 This can be in-terpreted as the minimum cost of elimi-nating poverty using transfers, though the actual cost could be very much higher, given incentive effects and administrative costs of targeting. By contrast, the cost of providing a “basic income” of $ 1.25 in the developing world – by transferring this sum to every person, whether poor or not – would be 4.3 per cent of world GDP. The mean consumption of the world’s poor by the $ 1.25 line in 2005 was $ 0.87, which is about one-thirtieth of globalGDP per capita. Similarly to the headcount index, we find a continual reduction in the globalPG index over time, from 21.7 per cent in 1981 to 14.2 per cent in 1990, then falling to 7.7 per cent in 2005.A Global Perspective on Poverty in IndiaIndia has a long tradition of rigorously measuring absolute consumption poverty, based on the National Sample Surveys (NSS) that started in the 1950s, under the leadership of the distinguished statistician P C Mahalanobis.5 That tradition has been inward looking, in that the aim has been to measure poverty by standards that most Indians would accept as relevant to what “poverty” means in India. That is entirely appropriate. And when discussing policies for reducing poverty in India, one should use poverty measures deemed relevant to India.The use of household surveys, such as the NSS, in measuring poverty has been widespread, but it has been questioned by some observers. It has been claimed by Bhalla (2002) that theNSS underestimates consumption levels, leading to an over-estimation of the level of poverty in India and underestimation of the pace of pover-ty reduction. The main reason given is the large and rising gap between the measure of aggregate household consumption im-plied by the NSS and the estimate of pri-vate consumption that can be derived from India’s National Accounts Statistics (NAS).6 The gap is unusually large for India. TheNSS data suggest a consumption ag-gregate that is not much more than half of the household consumption component of the NAS. Furthermore, the divergence has tended to rise over time, with lower growth rates implied by the NSS [Deaton 2005]. We do not know what role NSS survey methods have played in this divergence fromNAS consumption. National Sample Survey Organisation’s (NSSO) survey methods appear to have changed rather little over many decades. That is probably good news for comparability reasons, al-though it does raise questions about whether their methods are in accord with international best practice. This is some-thing that should be reviewed in the future, in the light of international experience. official urban poverty line to the rural line of 1.51 is also valid for the international line; in other words, the relative cost of living between urban and rural areas is assumed to be the same. Given this ratio and the 2005 consumption PPP for India of Rs 15.60 from World Bank (2008a), and taking account of the sample design of India’s ICP, one can back out unique rupee poverty lines for urban and rural areas corresponding to any given international line.11
he rupee values of the international line of $ 1.25 at PPP are Rs 21.53 and Rs 14.24 per day for urban and rural areas respectively.12
urban and rural India of Rs 17.24 and Rs 11.40 respectively, India’s official
WORLD BANK’S NEW POVERTY ESTIMATESOCTOBER 25, 2008 EPW Economic & Political Weekly36Indeed,India’s share of poverty in the developing world outside China has fallen, but only slightly, from 39 per cent in 1981 to 38 per cent in 2005.Thefalloccurred in the 1980s; the proportion was also 38percentin 1990. Looking across the rest of the developingworld,manycoun-trieshaveclearly not had India’s success againstpoverty. But many have done better too.Judged by the $ 1.25 line, the trend rate of poverty reduction seen in India over 1981-2005 is not sufficient to achieve MDG1; the projected poverty rate for 2015 is 34 per cent, while the target is 26 per cent (half the 1990 rate of 51 per cent). When one uses the $ 1.00 line (close to India’s of-ficial line atPPP), theMDG1 will be reached, though just barely; the projected $ 1.00 a day poverty rate for 2015 is 16 per cent, while the target is 17 per cent. However, the likely impacts on India’s poor of the recent rise in food and fuel prices make it unlikely in my view that India will reach the firstMDG even using the official poverty line.Slow ProgressThe potential for economic growth to re-duce India’s poverty rate is evident from the research results discussed above, com-paring the country’s poverty measures for the $ 1.00 line (close to India’s official line, at 2005PPP) with the new international poverty line of $ 1.25 a day at 2005 PPP. Recall that a large share – 17 per cent! – of India’s population is found in this narrow $ 0.25 a day interval. The other side of the coin to the implied vulnerability of India’s “near poor” to a downturn (as discussed above) is that any growth process that raises all levels of living by the same pro-portion will have a sizeable impact on the poverty count. A 20 per cent increase in mean household consumption without any change in inequality would be equivalent in its effect on India’s $ 1.25 a day poverty rate to lowering the poverty line from $ 1.25 to $ 1.00. Such a “distribution neu-tral” growth process would reduce India’s $ 1.25 a day poverty rate dramatically, from 42 per cent to 24 per cent. Realising that potential is another matter. The overall response of India’s poverty measures to growth in GDP per capita has clearly fallen short of this potential.18 An important factor is that (as noted above) growth in mean consumption as meas-ured by theNSS has been lower than eitherGDP growth or growth in mean consumption as measured by theNAS. The sources of this divergence are not yet well understood. There are two competing interpretations that can be offered based on what we know. On the one hand, it is possible that India’s pace of poverty reduction is being underestimated using the NSS; on the other hand, it is no less likely that the extent of inequality and its increase over time has been underestimat-ed, thus reducing the impact ofGDP growth on poverty in India. There appears to be wide (though not universal) agreement that growth inGDP per capita is necessary for sustained pov-erty reduction; the main differences lie in how much impact is expected from economic growth (depending, in part, on how one interprets the gap betweenNSS andNAS consumption). There is even more debate on the role played by inequality. The data do not suggest that India is a high inequality country. For example, based on the same NSS data used for measuring poverty, the poorest 20 per cent account for about 8 per cent of total household con-sumption; in high inequality countries such as Brazil or South Africa, the poorest 20 per cent account for 4 per cent or less of total consumption [World Bank 2007]. (Indeed, if India had been a high inequality country, we would not have expected to find that much potential for growth in mean con-sumption to reduce poverty.)19 Nonetheless, the extent of inequality and how it evolves during the growth process, matter to India’s progress against poverty. Inequality is relevant in two (con-ceptually distinct) respects.20 First, the way inequality evolves in a growing economy naturally determines how much the poor share (absolutely) in the benefits of that growth. If the bulk of the growth is found in places and/or sectors of the economy where the poor are not concentrated, then poverty will not fall much. There is evidence that the (geographic and sectoral) pattern of growth in India has not been particularly “pro-poor”, which is putting upward pressure on inequality in India.21 Second, high inequalities in certain di-mensions can undermine the growth process itself, and (hence) retard progress against poverty. Specific sorts of inequality in India entail that poor people lack the opportunities others enjoy for taking advantage of new economic opportuni-ties, including those unleashed by market-oriented reforms.22 Unequal access to the advantages of good schooling is an important example; there is evidence that such inequalitiesgreatlyattenuate thepoverty-reducing impact of growth in India’s non-farm economy.23 This can be reinforced by inequalities in access to credit for financing productive investment opportunities. The geography of economic Figure 2: Poverty in India Compared to the Rest of the Developing World Headcount index (% below $1.25 a day at 2005 PPP) | | | | | |1980 1985 1990 19952000 200570604020Developing worldas a wholeDeveloping world less ChinaIndiaInstitute for Social and Economic Change (ISEC)Advertisement No. A/2/2008 dated July 30, 2008With reference to the above advertisement inviting applications for the posts of (1) Associate Professor (CHRD) – Reserved for SC and (2) Assistant Professor – (CSEP) – Reserved for BC, the last date for receipt of applications is extended up to November 25th, 2008.For food, clothing and footwear, 72 per cent of the 717 sampled price outlets for India’s ICP were in urban areas and only 28 per cent were rural, while for other goods the outlets were solely urban. The ICP took simple averages of these prices. It is assumed that goods other than food, clothing and footwear had the same prices in rural and urban areas. Then the implicit urban and rural international poverty lines for India consistent with the 2005 ICP have weights of 0.72 and 0.28 respectively.
even larger adjustment was required for China, which had not officially participated in the ICP before 2005 [World Bank 2008b]. On the implications for global poverty measures for China see Chen and Ravallion (2008b).
Journal of Economic Inequality,
Review of Economics and Statistics,
John Micklewright and Steven Jenkins (eds), , Oxford University Press, Oxford.
orld Bank Economic Review,
).
).
[© by the International Bank for Reconstruction and Development/The World Bank, Washington DC, United States.]
