Journal Issue

Poverty reduction takes longer when income inequality is greater

International Monetary Fund. External Relations Dept.
Published Date:
June 2003
  • ShareShare
Show Summary Details

A recent empirical study of the relationship between the annual change in countries’ poverty rates and their annual growth rates (see box, page 165, for details about the study) finds, as have many earlier studies, that, on average, growth reduces the poverty rate by the same amount that an economic downturn or crisis increases it. This result is important because it does not support the view that economic downturns hurt the poor more than growth helps them as a result of irreversibilities and the existence of poverty traps. According to this view, getting out of poverty is much more difficult than falling into it because of a potentially irreversible loss of wealth, health, or opportunity. A youth who drops out of school, never goes back, and thus never learns how to read is an example of irreversibility. Similarly, in a poverty trap, regardless of the growth rate, the poor never make it out of poverty because they lack basic skills or opportunities to participate in economic activity.

But even if the empirical evidence suggests that these effects are not strong, the simple relationship between growth and poverty reduction leaves unexplained a large part of the observed changes in poverty rates. How can the picture be completed?

Significance of income inequality

Apart from growth, both the initial level of development and initial income inequality explain a significant part of poverty reduction. This was shown by a test that was set up in which nothing was known about a country but its initial income distribution (measured by the Gini coefficient, a measure of income inequality, where the higher the number, the greater the level of inequality) and level of development (measured by real per capita GDP). Using this information and assuming that income distribution is log-normal (that is, the distribution of income is bell-shaped and has additional statistical characteristics in line with actual income distribution), it is possible to predict approximately how much each percentage point of growth will reduce the poverty rate in a given country.

More specifically, it is predicted that the greater the income inequality or the lower the level of development, the less growth reduces poverty. Is this prediction in line with the data? Indeed, the statistical test showed that when income inequality and the level of development, along with the growth rate, were used to predict actual poverty changes, the predictions were much more accurate than if only the growth rate had been used.

For example, countries that have a low Gini coefficient and relatively high income per capita will need less growth to halve their poverty rate. Thus, countries in the Middle East and North Africa would need only 33 percent growth in real per capita incomes (an annual growth rate of 2.9 percent for 10 years) to halve their poverty rate, whereas sub-Saharan African and Latin American economies would need 50 percent growth (or an annual growth rate of 4 percent for 10 years) to achieve the same result. Despite these disparities, the two regions both have relatively low elasticities of the poverty rate with respect to economic growth—African economies because of their low level of development, and Latin American economies because of their high income inequality.

Is there a trade-off?

If there were a trade-off between growth and poverty reduction, it would be because countries with high growth achieved less pro-poor growth than countries that grew more slowly. But even then, because their growth rates were higher, such countries ultimately achieved as much poverty reduction as the slow-growing countries. Can the existence of such a tradeoff be tested with the sample restricted to countries with positive growth rates?

Two statistical tests of the trade-off hypothesis yielded no clear evidence that countries that achieved a certain measure of poverty reduction for each percentage point of growth (after controlling for their initial income distribution) did so at the expense of growth. If anything, the tests showed the opposite: not only was there no trade-off between growth and poverty reduction, but countries that were most efficient at reducing poverty for each percentage point of growth, given their initial distribution, exhibited higher growth rates as well.

Inflation is bad for the poor

It is commonly recognized that inflation under a certain level does not affect long-term growth but that inflation rates above a certain level impede growth. As far as poverty is concerned, the evidence is that this relationship between growth and inflation means that annual inflation of more than about 10 percent hurts the poor through its negative effects on growth. There are additional channels through which inflation may harm the poor. First, inflation may cut into their real wages because of the rigidity of nominal wages; second, because the poor have limited access to banking services, they cannot insulate the real value of their cash savings from inflation and thus suffer more than wealthier people who earn interest. The data compiled for this study can help measure the additional effect of inflation on the poor.

Tests showed that, when growth was negative, a very high annual rate of inflation (above 80 percent) increased the poverty rate. Not only did economic recession (possibly associated with crisis) increase the poverty rate, but, when it was accompanied by high inflation, it increased the poverty rate even further. But there is no evidence to suggest that when growth is positive and accompanied by inflation, the poverty rate drops by less than when growth is not accompanied by inflation.

It seems reasonable to think that high adult literacy and primary enrollment rates would enhance the link between growth and poverty reduction and that longer life expectancy would also be associated with pro-poor growth. But tests of these hypotheses did not find a significant link between these human development indicators and the pace of poverty reduction. Nor was there evidence that growth is more efficient in reducing poverty in countries with less corruption.

Policy implications

For policymakers, these findings confirm that economic growth is the most important source of poverty reduction. But also countries with more even income distribution have been found to achieve greater poverty reduction through growth than economies with greater income inequality. This means that when countries establish their targets for poverty reduction, they need to take into account not only prospective growth but also the initial level of development and income inequality. Moreover, although inflation above a moderate rate is bad for the poor, keeping inflation low is not enough to strengthen the link between growth and poverty reduction.

Copies of IMF working Paper No. 03/72, “Macroeconomic Performance and Poverty Reduction,” by Anne Epaulard, are available for $15.00 each from IMF Publication Services. See page 166 for ordering details. The full text is also available on the IMF’s website (

Measuring poverty

Poverty can be measured many ways. In this paper, the poor are defined as those who live on less than $2 a day (in 1993 prices), a measure that makes it possible (given World Bank data) to study a large number of countries, including some transition economies. It is important to include transition economies in the data set because they have specific characteristics—high average income and low income inequality (measured by the Gini coefficient)—that might influence the link between growth and poverty reduction.

Indicators for the poverty rate and income distribution are taken from a data set put together by the World Bank. In the sample of 99 episodes of growth or economic downturn in developing and transition economies, the initial poverty rates range from 0.4 percent to 90 percent, with a mean of 42 percent.

When countries establish their targets for poverty reduction, they need to take into account not only prospective growth but also the initial level of development and income inequality.

Other Resources Citing This Publication