STRAIGH TALK Aid and Growth: The Policy Challenge
We need more than aid to break the cycle of poverty
NOW that developed countries and international financial institutions have committed themselves to writing off the debt of highly indebted poor countries, the challenge will be to convert these resources into actual growth and faster progress toward the Millennium Development Goals. While for some it may seem that the war against poverty can be won simply by getting rich countries to provide more debt relief and aid, the view of experts—including those behind recent reports by the U.K. Commission for Africa and the Millennium Project—is that this is just one of the necessary ingredients. It’s early days yet in the campaign to make poverty history. If it’s to succeed, we have to recognize the failures of the past as well as be open-minded about the solutions for the future. And the first thing to recognize is the chequered history of aid.
Aid and growth
The best way to get the poor in low-income countries out of poverty is to strengthen economic growth in those countries. To the layperson, this may mean just sending these countries more aid. Yet one point about which there is general agreement among economists is that there is little evidence of a robust unconditional effect of aid on growth.
Before going further, let me say that the word “effect” implies causality. This is different from correlation. It’s possible to find in the data a negative correlation between aid and economic growth, but this doesn’t mean that more aid causes less growth. For instance, if aid tends to go to countries that are doing badly, you would get aid and growth being negatively correlated even though aid doesn’t cause poor growth: the direction of causation is the reverse. This is why economists use a technique called instrumental variables analysis to tell causality from simple correlation. In recent papers that I have written with Arvind Subramanian of the IMF’s Research Department, we describe how we found a negative correlation between aid and growth when we didn’t use instrumental variables, but how this essentially disappeared once we used the technique (Rajan and Subramanian, 2005 a and b). This means that aid skeptics may have been mistaken in viewing negative correlations found in the past as supporting their view. But unfortunately, we don’t find a robust, significant positive correlation either.
Does this mean that aid can’t, in any circumstances, boost growth? Of course not! The layperson’s thinking does, of course, have some significant basis. Poor countries are short of resources and ought to be able to put aid inflows to good use. There are case studies of countries that have grown using aid, and specific aid projects that have helped the poor enormously. What we economists haven’t identified is a reliable set of economic circumstances in which we can say that aid has helped countries grow. And this isn’t for want of trying.
For example, an influential study suggested that aid leads to growth, but only in countries that have good governance (Burnside and Dollar, 2000). This certainly seemed a very reasonable conclusion—a necessary condition for aid to help growth is obviously that aid receipts shouldn’t be spirited away to Swiss bank accounts. Unfortunately, however, it doesn’t seem to be a sufficient condition for aid to help growth, since follow-up studies suggest the finding isn’t robust (Easterly, Levine, and Roodman, 2004). It would appear that other levers are needed in addition to reasonable governance for aid to be effective.
A recent study (Clemens, Radelet, and Bhavnani, 2004) takes another crack at parsing the data, working from the assumption that not all aid is alike in its impact on growth. Again, the rationale is plausible. Why, for instance, should we expect humanitarian aid to result in growth, or why should we expect aid devoted to education (children are a long-term project if ever there was one) to produce growth in the short run? The study indeed shows that aid likely to have a shortterm economic impact (for instance, aid used to build roads or support agriculture directly) is positively correlated with short-term growth. Here again, however, I’m not fully persuaded. The authors of this study argue that the reason to focus on short-impact aid is because the literature focuses on country growth rates over four-year periods. So I presume it follows that if one were to depart from the literature and look at long-run growth (say growth over decades, which is what we really care about), economic aid (as contrasted with, say, humanitarian aid) cumulated during the period should have a discernible effect on growth (and there would be no need to separate out short-impact aid from long-impact aid). My work with Subramanian suggests that economic aid doesn’t have a robust positive correlation with long-run growth.
Despite my own convictions about what the past tells us, I will acknowledge that the debate about aid effectiveness is one where little is settled.
Unfortunately, further cross-country research along existing lines may not yield credible answers. We can continue trying to find some variable that will select out those countries that have received aid and also grown (or attempt to find some form of aid that is positively correlated with growth). But what do we conclude once we do that? Put another way, when the same data are pored over many times, there is a danger we will find patterns that are there by accident. This is why many economists have become skeptical that crosscountry studies can tell us much more.
Of course, the layperson would despair of the ability of further econometric analysis to shed light on the debate long before the economist. It should, however, be of concern to the layperson that the best example we have of aid working systematically for a group of countries is the Marshall Plan, whereby the ravaged countries of postwar Western Europe were returned to the ranks of the rich. The reason it worked so well might be that these countries’ institutions, including the education of their people, were probably capable of sustaining much higher per capita GDP than their postwar low. Perhaps this is why one might see a country emerging from conflict experience a substantial period of catch-up growth, when aid is very effective—Mozambique or Uganda might be more recent examples. Nevertheless, it should be sobering that the canonical recent example of a country clawing itself out of poverty into the ranks of the rich is Korea. Korea was indeed ravaged by war, but its spectacular growth started approximately when aid inflows tapered off.
“It’s sobering to think that by constraining the growth of manufacturing, aid inflows may have prevented recipient countries from taking the path to growth followed first by the East Asian tigers and now by China.”
Dodging “Dutch disease”
According to some, there is a better way—to focus on what we know works. Specifically, funding should support microinterventions or programs, validated through evaluations and experimentation, that might be very helpful, say, in furthering education and health care, which undoubtedly lead to growth. Here, we have learned a lot from work by Abhijit Banerjee of the Massachusetts Institute of Technology, Michael Kremer at Harvard, and their students, as well as from the World Bank, including its World Development Report 2004.
We know that providing services to the poor isn’t just about money. One can build spanking new schools and pay teachers a good wage, yet they may not come in to teach. One can provide free drugs to the hospitals, intended for the poor, but the druggist may simply sell them on the black market. This isn’t to say that schools and hospitals aren’t necessary, but bricks and mortar are often the easy part. Policymakers also need to create the right incentives for the service provider and the poor client, as well as the right allocation of power and information between them to ensure that reasonable quality services are provided. And we know that the law of unintended consequences is always at work. This means that few programs ever operate as the designers intended, so we need abundant experimentation, frequent monitoring and evaluation, and a sharing of best practices so that these targeted interventions can have their intended effect.
Unfortunately, I’m not sure that even if each microintervention works well by itself, they will all work well together. Interventions could affect each other and get in each other’s way or vie for the same resources. They could also have adverse spillover effects on the rest of the economy.
The last isn’t just a possibility. Suppose a lot of aid flows in to support interventions in education, health care, and other social services. The recipient country quickly hires many educated workers as teachers, clerks, nurses, foremen (to build the schools), engineers, and government and aid administrators. Because well educated people will be in high demand, their wages will tend to rise and may well go up rapidly. In turn, factories will have to escalate the wages they pay to managers, engineers, and supervisors. Now factories that produce for the domestic market and don’t face competition can pass on their higher costs. But factories that export can’t, so they will cut down on operations and even start shutting down. This is one example of a phenomenon called Dutch disease, which makes aid recipients less competitive. Subramanian and I show that in countries that received more aid in the 1980s and 1990s, the export-oriented, labor-intensive industries not only grew more slowly than other industries—suggesting that aid did in fact create Dutch disease—but the manufacturing sector as a whole also grew more slowly. Again it’s sobering to think that by constraining the growth of manufacturing, aid inflows may have prevented recipient countries from taking the path to growth followed first by the East Asian tigers and now by China.
That said, Dutch disease is not a terminal condition. It can be mitigated through sensible policies. But to do so, one must first acknowledge its existence and its pernicious effects. The same goes for other possible diseases caused by aid.
How about humanitarian aid?
Does this cautious approach toward aid in general mean that the world should hesitate to give humanitarian aid? Absolutely not! But the form of the aid matters. In the midst of a humanitarian disaster, one should concentrate on getting enough relief in kind to the affected area if local production is typically not possible (for example, because the failure of local crops is the proximate cause of the humanitarian disaster). Aid sent in the form of cash may, however, be better if supplies are available but the distressed population doesn’t have the buying power. Aid could then create more local jobs. Also, when the immediate emergency is over, donors should be careful that additional aid doesn’t hamper incentives for local produc-tion—that, for example, donated secondhand clothes don’t kill the business oflocal tailors.
And every so often, donors are confronted with the Good Samaritan’s dilemma. An uncaring local government siphons off a fraction of the humanitarian aid in return for allowing the aid to get through to the starving people. While the aid reduces the immediate suffering of the people, it also entrenches the government, perpetuating the people’s long-term suffering. There are no easy solutions to this dilemma.
There is hope
To ignore the past, or to read only rosy lessons from it, is to condemn oneself to relive it. While it would be churlish to deny that many poor countries have made tremendous progress in creating the conditions for sustained growth, it doesn’t serve the citizens of poor countries either if we say that all the problems of the past are well and truly in the past. While no one has the “magic bullet” for growth, there are some things that do seem important. These include sensible macroeconomic management, with fiscal discipline, moderate inflation, and a reasonably competitive exchange rate; laws and policies that create an environment conducive to private sector activity with low transaction costs; and an economy open for international trade. In addition, investments in health and education—which create a population that not only lives a better life but also sees opportunities in growth and competition—ought to be encouraged.
One way rich countries and international financial institutions can help is by making policies that broadly meet these requirements an essential condition for aid. They should, however, resist micromanaging and overlaying broad economic conditionality with too many detailed economic prescriptions, or with social and political conditionality. Once a country has the necessary broad environment in place, it should have the freedom to chart its own path. After all, the failure of past grand theories of growth should make us wary of becoming overly prescriptive.
Rich countries can also help by reducing the impediments they place in the way of poor country exports, and by coaxing these countries to lower their own trade barriers, including barriers to other poor countries. They can spend more to foster research on drugs and agricultural technologies that would benefit the poorest countries. They can be more active in ensuring that their companies and officials don’t grease the wheels of corruption in poor countries (Birdsall, Rodrik, and Subramanian, 2005, for other suggestions). And they should never hesitate to give humanitarian aid in the face of a disaster (box).
Let us draw hope from the willingness of the outside world to provide more, and better, aid. Ultimately, though, poor countries hold their future in their own hands. It’s only through their own will and actions that the good intentions of the outside world can be used to truly make poverty history.
Raghuram Rajan is Economic Counsellor and Director of the IMF‘s Research Department.
BirdsallNancyDaniRodrik and ArvindSubramanian2005“How to Help Poor Countries,”Foreign AffairsVol. 84No. 4 pp. 136–52.
BurnsideCraig and DavidDollar2000“Aid, Policies, and Growth,”American Economic ReviewVol. 90No. 4 pp. 847–68.
ClemensMichael A. StevenRadelet and RikhilBhavnani2004“Counting Chickens When They Hatch: The Short-Term Effect of Aid on Growth,”CGD Working Paper No. 44 (Washington: Center for Global Development).
EasterlyWilliamRossLevine and DavidRoodman2004“Aid, Policies, and Growth: Comment,”American Economic ReviewVol. 94No.3 pp. 774–80.
RajanRaghuram and ArvindSubramanian2005a“What Undermines Aid’s Impact on Growth,”NBER Working Paper No. 11657 (Cambridge, Massachusetts: National Bureau of Economic Research).
RajanRaghuram and ArvindSubramanian2005b“Aid and Growth: What Does the Cross-Country Evidence Really Show?”IMF Working Paper 05/127 (Washington: International Monetary Fund).