Growth is Good for the Poor

GROWTH IS GOOD FOR THE POOR

D. Dollar & A. Kraay

Journal of Economic Growth, 7, 195-225 (2002)

Principal Research Question and Key Result	Do incomes of the poor rise when the economy as whole experiences growth? What is the elasticity of the incomes of the poorest quintile with respect to the mean income? The data in this paper show that the incomes of the poorest quintile rise linearly with mean incomes such that if mean income grows by 1% the income of the poorest will also grow by 1% i.e. Elasticity of 1.
Theory	The paper does not have a theory section, however as they are looking at the elasticity of the incomes of the poorest quintile with respect to the mean income, they are in effect looking at how income inequality is effected by growth. Thus the paper is related to the literature on growth and inequality.
Motivation	If growth benefits the poorest in society then it should be at the heart of any poverty reduction strategy. The neoclassical growth prescription for growth including macro stability, openness to trade, property rights, small government etc. are often said to be good for growth but bad for the poor, and this needs to be evaluated. In recent years there has been significant effort to make growth pro-poor through education, public spending on health, increasing labour productivity etc. and the effectiveness of these policies in raising incomes of the poor needs to be evaluated.
Data	Mean income is measured as real GDP per capita. Income of the poorest quintile is measured using household surveys in some cases and using Gini calculations in other cases. In other words the data sources are mixed which is far from ideal. It is a highly unbalanced and irregularly spaced panel. Mean of only 4 observations per country. Very few observations for poorest regions e.g. SSA
Strategy	OLS with logs of income of poorest as dependent variable and mean income as independent variable with controls. Coefficient on mean income is therefore the elasticity of the income of the poorest with respect to mean incomes. They do a first difference estimation to dispose of time invariant country specific sources of heterogeneity. They also do IV estimates and a GMM using levels and the differenced estimates. Endogeniety, omitted variable and measurement error problems are addressed with an IV strategy (e.g. Geography) that uses lags of the right hand side variables as instruments. In the first stage there is strong correlation.
Results	They do not reject the null hypothesis that the coefficient is 1. They do not find significant coefficients when they include trade, inflation, financial development etc. but there is a small significant and negative effect of government consumption on the incomes of the poorest. They instrument using lags and the results are similar. They find no evidence that economic integration measured by WTO membership etc. reduces incomes of the poor. They look at other factors that may have direct effects on income of the poor through the income distribution – education, social spending, agricultural productivity and democracy. Of these variables, only education is generally found to increase growth directly, however, they may affect how much the poor participate in any growth. All enter with the expected sign, but only democracy is weakly significant.
Robustness	They include regional dummies to see if growth affects incomes of the poorest differently in different regions. The coefficient is reduced by still statistically not different from 1. Interact mean incomes with GDP to see if different relationship in poor/rich countries – no effect on coefficient. Include dummies by decade, and also for shock periods – no change so growth/inequality relationship is stable over time, and the poor are not more affected by negative growth than mean incomes.
Problems	Using national income is problematic as national income does not necessarily translate into household income/consumption. Also it does not account for the shadow economy etc. There are lots of problems with the data. In general comparability across countries is problematic, but here the data are drawn from different sources which means they are even less comparable. There is likely to be significant measurement error, and this is most likely systematic given that quality data collection will be harder to achieve in the poorest parts of the world. This is further reflected in the fact that there are relatively few observations for SSA, and hence the results could be being driven by the wealthier countries in the sample. All this being said, at the time this was probably the best data available. The inability to find effects of policies that are thought to be pro-poor may be down to measurement error alone. Restricting the analysis to the poorest quintile is somewhat restrictive. It would have been interesting to see results for poorest 10%/5% etc. IV strategy is not convincing as lagged variables are not likely to be exogenous as growth in incomes 5 years ago could be associated with greater installment of productive capacity which then influences growth in incomes today. Same of policy, redistribution etc. etc.
Implications	Average incomes of the poor increase with mean income. The authors claim that the basic package of polices (trade, rights etc.) that are thought to increase growth, also increase the incomes of the poor. This is not a trickle down, but a concurrent happening. Although the results may largely be in line with other studies, no cross country evidence can give guidance as to what mix of growth policies may benefit the poor. For this micro work is probably necessary. There is no claim that growth is all that is needed, and the policies such as health spending may not increase incomes, but they may nevertheless be important to the poor for other reasons not related to income. Given the substantial problems with this paper it would be unwise to make policy prescriptions based on its findings.

 

HALVING GLOBAL POVERTY

HALVING GLOBAL POVERTY

T. Besley & R. Burgess

Journal of Economic Perspectives Vol. 17, No. 3 pp. 3-22 (2003)

A Summary

In a Nutshell

The field of development economics has shifted away from the pure neoclassical model of accumulation in a stable macro environment, to a paradigm that focusses not on redistribution, but rather on growth that expands opportunities for households and holds governments to account. This means more emphasis on specific policies that can be shifted in a pro-poor direction.

Growth alone will probably not suffice to significantly reduce poverty. To the extent that it does reduce poverty the microeconomics of growth should remain a focus i.e. uncovering the specific institutional order that creates growth. However, the growth rates needed to halve poverty are enormous relative to historic rates, so study should be undertaken to uncover policy and institutional changes that directly tackle poverty.

In finding the right mix of policies and reforms it is unlikely that cross country data will be the prime means of analysis. Such data can only provide signposts for more focused [micro] work. E.g. Human Capital, Access to Credit, Property Rights, Governance. Changes in these policies can best be evaluated at a sub-national level. In this regard national governments will be key. Whilst donor support may be important in certain situations, the 0.7% of developed world GDP that it is recommended by the UN be donated to the developing world (a target which is not met) amounts to only $142bn a year whereas to bring everyone living below a dollar a day to an income above that line would cost $443bn a year

Quantifying Global Poverty

The best type of data are based on household surveys that collect information on income and consumption rather than aggregate GDP figures. There has been progress here, and there is now much data that is broadly comparable across countries. However, the revealed picture of global poverty remains at best partial. Based on the $1 day measure poverty has fallen around 6% since 1990, but the absolute number has decreased only from 1.3bn to 1.2bn.

 There is much heterogeneity in poverty. East Asia has dramatically reduced poverty due to performance of China. Sub-Saharan Africa (“SSA”) has stagnated. [This gives cause to question Dollar & Kray as there is no reason to assume that the incomes of the poorest will react similarly across varying contexts.]

Growth and Poverty

Growth can benefit the poor both directly and indirectly. When assessing the preferred measure is national income per capita although this is not available for all lower/middle income countries. There are problems with comparability across countries as rising income per capita does not translate into rising household consumption on a 1:1 basis, and the ratio varies across territories. For example growth in SSA seems to have the lowest impact on poverty. Moreover the rate of growth needed in SSA to successfully halve poverty in line with the MGDs is 28 times the historic average, so even if growth were to increase the incomes of the poorest equiproportionatley it is most likely an impossible task to increase growth by a factor of 28.

Inequality and Poverty

Income growth does not seem to change inequality within nations. However there is a significant association between inequality and the level of poverty. A one standard deviation change in inequality in SSA would in theory lead to a 50%+ reduction in poverty in the region. Thus a focus on inequality is needed. In particular, given that taxation is not a viable means in many low income countries; other means need to be found such as increasing access to credit and infrastructure. This involves political economy analysis of how to orient these policies in a pro-poor direction. Furthermore the distributional impact of growth should be a focus, and drivers of growth that benefit the poor should be found.