We evaluate the claim that world consumption poverty has fallen during the 1990s in light of alternative assumptions about the extent of initial poverty and the rate of subsequent poverty reduction in China, India, and the rest of the developing world. We
assess the extent of poverty using two indicators: the aggregate poverty headcount and the poverty headcount ratio, and consider two international poverty lines that are widely used ($1.08/day and $2.15/day 1993 PPP). We find that under some of the assumptions considered, world poverty has risen. We conclude that, because of uncertainties in relation to the extent and trend of poverty in China, India, and the rest of the developing world, world poverty may or may not have increased. The extent of the increase or decrease in world poverty is critically dependent on the assumptions made. Our conclusions suggest the importance of improving the quality of global poverty statistics.
A series of influential studies has advanced the conclusion that world poverty has fallen substantially since the early 1990s (see, for example, Bhalla (2002), Chen and Ravallion (2001, 2004) and others). Furthermore, it is widely thought, on the basis of national poverty estimates and studies based on the international poverty lines, that poverty has fallen in India and China in the 1990s. However, the present extent of poverty and the recent pace of poverty reduction in these two countries with the largest populations in the world are still debated and debatable. Estimates of the extent and trend of poverty in the rest of the developing world are much less favorable and also questionable. This gives rise to unsettling concern as to whether world poverty has actually fallen.
This question takes on special importance in light of the UN’s first Millennium Development Goal, which calls for the halving of the percentage of the developing country population living under the World Bank’s “$1 per day” international poverty line
between 1990 and 2015. Whether this goal is likely to be achieved has been a central concern in recent debates. For example, Bhalla (2002) avers gushingly that “Toward that goal…15 years hence, and already achieved today, resources are used to fight the nonexistent poverty of tomorrow.” (pp. 92-93). Chen and Ravallion (2004) conclude that “if the trends over 1980-2001 continue then the aggregate $1 per day poverty rate for 1990 will be almost halved by 2015, though East and South Asia will be the only regions to more than halve their 1990 poverty rates.” However, they also find a low rate of reduction in the “$1 per day” headcount ratio measure of poverty in Latin America between 1990 and 2001 (from 11.3 percent to 10.5 percent)5 and increases in the same measure of poverty in Sub-Saharan Africa during the period (from 44.6 percent to 45.7 percent).6 A cautious stance is taken by Deaton (2002), who highlights the uncertainties surrounding global poverty estimates.
Recent debates on whether the world is on the right track in regard to poverty have frequently centered on whether the poverty reduction thought to have taken place in China and in India in the 1990s has been sufficiently to caused a decrease in poverty
worldwide, despite the possibly poor record of poverty reduction elsewhere.7 In order to assess this question, we undertake a sensitivity analysis. Specifically, we assess whether the conclusion that world poverty has fallen between 1990 and 2001 is robust to alternative assumptions concerning the extent of initial poverty and the rate of subsequent poverty reduction in China, India, and the developing world outside China and India (henceforth, ‘non-China-India’). For each set of assumptions concerning initial poverty headcounts in China, India, and the remainder of the developing world, and concerning subsequent rates of poverty reduction in China and India, we determine the maximum non-China-India poverty headcount ratio in 2001 that is consistent with the hypothesis that world poverty has not risen between 1990 and 2001. If non-China-India poverty headcount ratios were higher than this maximum at the end of the period then world poverty must be concluded to have increased between 1990 and 2001.
We draw on the literature estimating poverty in China, India and world as a whole to identify alternative estimates of the initial extent of poverty and subsequent rates of poverty reduction. The extent of poverty is assessed using two indicators: the poverty headcount ratio (HCR) and the aggregate poverty headcount (HC), and two widely accepted (if weakly conceptualized) international poverty lines: $1.08/day and $2.15/day of 1993 “international dollars”.8
We find that under at least some plausible assumptions, world poverty must be concluded to have increased in the 1990s. More generally, the analysis draws attention to the need to assess claims regarding the extent and trend of global poverty in light of their robustness to alternative reasonable specifications of underlying assumptions. The assumptions made influence both the extent of poverty reduction each country is thought to have experienced and the relative weight attached to each such experience when determining the aggregate poverty trend.
The paper is structured as follows: in the next section, we briefly discuss the role of the aggregate headcount and the headcount ratio in poverty assessment. In section III we describe the data sources and literature that we employ in constructing the alternative specifications that we examine of the initial extent of poverty in India, China and the non-China-India developing world and of subsequent poverty reduction experiences in China and in India. In section IV we identify for each such specification, the maximum non-China-India poverty headcount ratio in 2001 that is consistent with concluding that world poverty has not increased. Section V concludes.
We would like to thank the Bureau of Development Policy of the United Nations Development Program, and in particular Terry McKinley, for providing essential financial support for this research. We would like also to thank Gaurav Datt for his kind assistance with a query and Tanweer Akram for helpful suggestions.
Dept. of Economics, Barnard College, Columbia University and University Center for Human Values,
Princeton University. Email: . Tel. 212-854-3790, Fax. 212-854-8947
Dept. of Economics, Columbia University. Email:
Updates of this paper will be available on www.columbia.edu/~sr793/sensitivityanalysis.pdf
The regional average reported is for Latin America and the Caribbean.
Chen and Ravallion (2004) find that there has been a sustained increase in the headcount ratio in Sub-
Saharan African countries from 1993 onwards.
See for example Wade and Wolf (2002).
See Chen and Ravallion (2001).