Growth elasticity of poverty: direct estimates from recent data

Abstract

Using recent poverty data for developing countries, elasticity of $2 poverty headcount with respect to the growth of Gross Domestic Product (GDP) per capita is directly calculated for the 1990s and the 2000s. The global estimate is around −0.84, which is much smaller in absolute magnitude than what some highly influential studies have reported or used. For the poverty-dense South Asia region, the elasticity is of the order of −0.22, which is a dramatic contrast from the previous estimates. For India, where the number of poor people is by far the largest of any country, and where the poverty rate is higher than even in Sub-Saharan Africa (SSA), the elasticity is of the order of −0.13, which is an even bigger contrast from earlier studies. Policymakers and researchers are urged to keep these estimates in mind while judging the likely effect of income growth on poverty, and to discount poverty-reduction claims based on higher elasticity estimates that have been reported or used in many influential studies. In particular, the elasticities shown in this study for South Asia and India seem to be a grim reminder of how unrealistic the existing estimates might be.

Acknowledgement

Useful comments from a perceptive and thorough referee are gratefully acknowledged.

Notes

¹ In a related research, Enders and Hoover (2003) studied the effect of robust growth on poverty in the United States. Somewhat similarly, Arvin and Barillas (2002) investigated causal order across foreign aid and poverty, and Hajro and Joyce (2009) considered the impact of International Monetary Fund (IMF) programs on poverty.

² Ram (2006) explored the position for $1 poverty rate.

³ Their numbers for poverty are almost identical with those in World Bank (2007a, p. 63).

⁴ Considerable cross-country heterogeneity does indeed underlie the estimates for all developing countries in Table 1. The main merit of these estimates is their direct comparability with the numbers reported or used in many earlier studies, as noted in the text. The underlying heterogeneity is shown by the much smaller estimates for South Asia and India in that table. Also, based on data from World Bank (2006, p. 8) and Chen and Ravallion (2007, p. 22), $2 poverty elasticities for 1990–1999 are 0.60 and 0.25 for East Asia and Pacific (EAP) and Latin America and Caribbean (LAC) regions, respectively. These provide further indication of the heterogeneity and show the responsiveness of $2 poverty to income to be quite low. The corresponding numbers could not be calculated for East Europe and Central Asia (ECA) and SSA because growth of GDP per capita in these regions during the period was negative. The calculation could not be done for Middle East and North Africa (MENA) because of increase in poverty rate despite an increase in GDP per capita.

⁵ Chen and Ravallion (2007, p. 22) show the following regional count of persons below $2 poverty line in 2004 and the corresponding poverty rates:

Table

⁶ Preliminary calculations indicate the pattern to be very similar to that in Table 1 if one uses the more recent poverty estimates reported by Chen and Ravallion (2008) that are based on 2005 International Comparison Project (ICP) report. For example, $2 elasticity for 1990–1999 appears to be even weaker at −0.73 relative to all developing countries. The estimates of −0.21 for South Asia and −0.15 for India are essentially the same as in Table 1.