Development economists have always struggled with two questions when it comes to foreign aid: how much to give and how to use it effectively. The questions are, of course, related. If aid is going to be wasted or, more seriously, appropriated by corrupt recipients, there will likely be little support for giving.

The earliest proponents of aid put the “absorptive capacity” of the aid recipient at the heart of the matter. For them the right levels of aid are what I call “demand-determined”: they reflect what a country can usefully absorb and put to constructive purposes.

Today, some others (such as Jeffrey Sachs) go instead by what I call “supply-determined” aid flows: they typically advocate aid targets such as 0.7 percent of the GNP of the donor country, similar to a tithe or zakat. By marrying such targets to the politically derived “millennium development goals” of the United Nations, they endorse giving nearly regardless of absorptive-capacity problems. In fact, they condemn as reactionary anyone who argues that, for instance, in Africa, absorptive capacity (due to civil wars, corruption, and the fragility of regimes) is so low that one cannot step up aid very substantially. Although well-meaning, these economists might have been regarded as unwitting saboteurs by the great development economists (such as Paul Rosenstein-Rodan) who also pioneered the aid programs of the 1950s.

And yet in practice it is hard to find anyone who says that giving aid is simply one’s duty, regardless of consequences. For both camps, the important question is this: assuming that aid can be increased, how can it be used most productively? Abhijit Banerjee believes he has the right formula: perform more controlled experiments. That is how we evaluate new drugs, and, according to Banerjee, we should do the same to assess the efficacy of different aid-financed projects. Doubtless this would help, although what we know works in one village or set of villages may not work in another part of the country or another part of the world.

But the large-scale problems of development—including the improved efficacy of aid—go well beyond the small-scale Banerjee approach, brilliant as it is. Development economists have long been grappling with broader questions of strategy. For example, should aid be spent on immediate distribution to the poor, or should it be used to finance infrastructure investment, which may reduce poverty through the classic “pull-up” mechanism of growth? Is an inward-looking strategy favoring autarky a better route to prosperity than an outward-oriented strategy favoring trade? And, if the latter is preferable, as many postwar empirical studies have shown convincingly, can aid be used to facilitate trade liberalization? To demand that development economists subject questions like these to controlled experiments makes little practical sense.

To return to the question of how more aid can be absorbed productively, I offer an idea of mine that has made some headway. As I suggested earlier, it is hard to think of substantial increases in aid to Africa being spent productively in Africa. But it is not so hard to think of more aid being spent productively elsewhere for Africa. Thus, expenditures could be stepped up vastly in rich countries for the development of new vaccines and cures for crippling diseases afflicting African nations, just as the Institutes for Tropical Medicine in London did in colonial times. And since Africa suffers from enormous shortages of skilled manpower, why not organize a Gray Peace Corps in aging, rich countries where large numbers of doctors, engineers, and scientists could be hired at “tropical premiums”? Meanwhile, we could train more Africans in different scientific fields with aid money spent in the rich countries on fellowships. One could go on. If foreign aid is made to walk on two legs—aid spent in Africa and aid spent elsewhere for Africa—we can increase productively used aid funds far more rapidly than if we make it walk on one leg alone by insisting that aid for Africa be spent in Africa (as Sachs and others often demand).

I believe that exploring these kinds of issues will give us the biggest payoffs in decisions about developmental strategies. There is perhaps a role to be played by the Banerjee approach to micro-level studies in this mammoth task, but it can only be a minor one.