Elsevier

European Economic Review

Volume 46, Issue 8, September 2002, Pages 1475-1500
European Economic Review

Aid allocation and poverty reduction

https://doi.org/10.1016/S0014-2921(01)00187-8Get rights and content

Abstract

This paper derives a poverty-efficient allocation of aid and compares it with actual aid allocations. The allocation of aid that has the maximum effect on poverty depends on the level of poverty and the quality of policies. Using the headcount, poverty-gap, and squared poverty gap measures of poverty, alternatively, all yield similar poverty-efficient allocations. Finally, we find that the actual allocation of aid is radically different from the poverty-efficient allocation. With the present allocation, aid lifts around 10 million people annually out of poverty in our sample of countries. With a poverty-efficient allocation, the productivity of aid would nearly double.

Introduction

The allocation of aid among countries can legitimately reflect multiple objectives. Aid may be used to rebuild post-conflict societies, to meet humanitarian emergencies, or to support the strategic or commercial interests of the aid-giver. However, one core objective most commonly cited to support aid programs is poverty reduction. In this paper, we estimate the allocation of aid that would maximize the reduction in poverty and compare it to actual allocations. Our principal finding is that the poverty impact of aid could be roughly doubled if donors made use of recent research findings on the impact of aid in deciding their aid allocation. To the extent that donors are interested in poverty reduction, the estimated ‘poverty-efficient’ allocation is directly useful for policy-makers. But, even where donors wish to pursue other objectives, this allocation is also useful, because it provides information on the opportunity cost (in terms of poverty reduction) of pursuing other objectives with aid resources.

The recent results that donors need to take account of are that:

  • the impact of aid on growth depends on the quality of economic policies and is subject to diminishing returns (Isham and Kaufmann, 1999; Burnside and Dollar, 2000);

  • there is a wide range of different evidence that the quantity of aid does not systematically affect the quality of policies – even with ‘conditionality’ (Collier, 1997; Williamson, 1994; Rodrik, 1996; Alesina and Dollar, 2000); and

  • aid resources are typically fungible, so that it is difficult for donors to target them to particular groups or use them to alter the distribution of income (Pack and Pack, 1993; Feyzioglu et al., 1998).

In Section 2 we revisit the first result above, using a broad measure of policy and a larger number of countries than covered in previous analyses. We confirm that the marginal efficiency of aid in terms of increases in income depends on the quality of policies and on the amount of aid that a country is receiving (diminishing returns). In Section 3 we consider the donor's optimization problem, if the objective is to reduce poverty and if the donor takes as given the quality of policies and the distribution of income in aid-receiving countries. We derive an algorithm for the ‘poverty-efficient’ allocation of aid among countries that has a simple, intuitive logic: holding the level of poverty constant, aid should increase with policy (because it has a larger growth impact in the better policy environment); and, holding policy constant, it should increase with poverty (because the poverty impact of growth is higher). What defines the poverty-efficient equilibrium is that the marginal impact of an additional million dollars in aid is equalized across aid-receiving countries.

In Section 4 we subject the estimated poverty-efficient allocation to a number of sensitivity analyses. The estimated coefficients from the growth analysis play a role in the algorithm, and we vary these coefficients by a standard deviation in either direction to investigate the practical importance of imprecision in the estimates. In addition, there are a number of different poverty measures that could be targeted by donors (headcount, poverty gap, or squared poverty gap). We examine the impact of shifting from one measure to another, and from a $1 per day poverty line to a $2 per day poverty line. Our benchmark allocation of aid is correlated 0.89 or above with any of the alternative estimates that arise from this sensitivity analysis. The actual allocation of aid is correlated only 0.57 with our benchmark allocation. We conclude from the sensitivity analysis that significant changes in the estimated parameters or in the poverty measure of choice lead to only minor variations in the ‘poverty-efficient’ allocation – variations that are minor especially in comparison to the difference between our benchmark and the actual allocation of aid. In this sense, the poverty-efficient allocation is quite robust.

The fifth section of the paper examines in more detail the extent to which the actual allocation of aid deviates from the poverty-efficient allocation. We show that aid has the ‘wrong’ relationship with policy, after controlling for poverty. Precisely, in the range of policy in which aid becomes increasingly effective in poverty reduction, aid is currently lower the better is policy. In short, aid is being tapered out with reform, when it should be tapered in with reform. We estimate that in our sample of countries aid as currently allocated sustainably lifts 10 million people per year out of poverty. The same volume of assistance, allocated efficiently, would lift an estimated 19 million people out of poverty. Thus, the productivity of aid could be nearly doubled if it were allocated more efficiently.

Section snippets

The mapping from aid to growth

Our objective in this section is to arrive at estimates of the impact of aid on growth for a large number of countries, as a first step toward estimating the impact of aid on poverty reduction. Burnside and Dollar (2000) have shown that the impact of aid on growth depends on the quality of the incentive regime.1

The poverty-efficient allocation of aid

The results above suggest that donors can affect growth through their allocation of aid; growth in turn will typically lead to poverty reduction in low-income countries. Dollar and Kraay (2001) show that on average growth of per capita GDP is translated into proportional growth of income of the poor. Furthermore, the policies that are good for growth (and are measured by the CPIA) are good to the same extent for income of the poor. The intuition of our approach for allocating aid is

Sensitivity analysis

How sensitive is our estimated poverty-efficient allocation of aid to variations in the parameter estimates from our growth regressions and to the choice of poverty measure? These are the questions that we turn to in this section.

We explained in Section 2 that we would investigate four different variants of the estimated impact of aid on growth. It is straight-forward to recalculate the poverty-efficient allocation in Table 3 using the different parameter estimates. Table 4 shows the

Reallocating aid for poverty reduction

We now consider in more detail how the actual allocation of aid compares to the allocation that maximizes poverty reduction and quantify the gains from moving from the current allocation to our benchmark allocation.

In our model of efficient aid, what a country receives relative to GDP should be a monotonic but non-linear increasing function of the headcount index divided by per capita income (which we will denote POV). It should be a monotonic increasing function of policy (CPIA). The actual

Conclusion

Although aid may be allocated coherently, it is allocated inefficiently with respect to poverty reduction. At present, aid is allocated partly as an inducement to policy reform and partly for a variety of historical and strategic reasons. This produces a pattern in which aid is targeted to weak policy environments and to countries which do not have severe poverty problems. The diversion of aid from poverty reduction to policy improvement would be justifiable were there evidence that the offer

Acknowledgements

The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. We thank Aart Kraay, Martin Ravallion, and two anonymous referees for useful advice and Charles Chang and Dennis Tao for excellent research assistance.

References (25)

  • Alesina, A., Dollar, D., 2000. Who gives aid to whom and why? Journal of Economic Growth 5 (March)...
  • Boone, P., 1994. The impact of foreign aid on savings and growth. Mimeo., London School of Economics,...
  • Bourguignon, F., 2000. The pace of economic growth and poverty reduction, Mimeo., DELTA,...
  • F. Bourguignon et al.

    Poverty measures and anti-poverty policy

    Recherche Economiques de Louvain

    (1990)
  • C. Burnside et al.

    Aid, policies, and growth

    American Economic Review

    (2000)
  • Chen, S., Ravallion, M., 2000. How did the World's poorest fare in the 1990s. Policy research working paper no. 2409,...
  • P. Collier

    The failure of conditionality

  • G. Datt et al.

    Regional disparities, targeting, and poverty in India

  • Devarajan, S., Hossain, S.I., 1998. The combined incidence of taxes and public expenditures in the Philippines. World...
  • Dollar, D., Kraay, A., 2001. Growth is good for the poor. Policy research working paper no. 2587, The World Bank,...
  • T. Feyzioglu et al.

    A panel data analysis of the fungibility of foreign aid

    World Bank Economic Review

    (1998)
  • J.A. Frankel et al.

    Does trade cause growth?

    American Economic Review

    (1999)
  • Cited by (602)

    View all citing articles on Scopus
    View full text