By advocating the use of randomized trials to evaluate development aid, Abhijit Banerjee seeks to repel criticisms from two camps: those who are skeptical of the way aid is spent and those who stress the fragile scientific foundations that justify its distribution.

Policymakers crowd the ranks of the first camp. Rather than viewing aid as a transfer between the rich and the poor, they view it as a shift from rich taxpayers in the North to poor governments in the South. They view governments as inefficient and corrupt—as part of the problem, not part of the solution. Those who subscribe to this view demand proof that the taxpayers’ money is being well spent. In the face of this attack, Banerjee proposes a methodology designed to provide rigorous measurements of the impact of aid programs. By isolating the features that make the greatest impact, it provides greater assurance that the resources transferred to developing nations will get the biggest bang for the buck. This proposal should be welcomed by the first camp.

Scholars (such as William Easterly) fill the ranks of the second camp. They believe that development aid lacks adequate justification. By this they do not mean moral justification; even skeptics of aid are moved by the magnitude of the disparity between rich and poor nations. Rather, they mean the underlying science guiding aid giving, or rather the lack of it.

In truth, there is no theory of development that is logically compelling and demonstrably valid. One good indicator of this deficiency is the very abundance of theories, some pointing to the importance of capital, others to the role of technical change, and still others to the significance of political institutions. The lack of rigorous foundations is also betrayed by the way that scholarly viewpoints change: not when they are proved false but when people rally around new ones. When Robert McNamara headed the World Bank during the Johnson administration, development specialists called for a “war on poverty”; when Barber Conable headed it during the Reagan administration, they sought to “get the prices right.” The field of development responds less to evidence than to political fashion.

Development aid is thus criticized by both practitioners and scholars, and Banerjee provides a rejoinder to both in the form of empirics. Were development like, say, nutrition, then randomized experiments should indeed expose bad development ideas as surely as they expose faddish diets. I fear, however, that in crafting his defense, Banerjee may have so narrowed the focus of the debate as to lose sight of crucial concerns.

If Banerjee’s empirical approach to development aid became dominant, it would transform the field from a search for the underlying forces of development into a form of policy analysis. It might achieve greater certainty about the impact of particular policy features, but it would do so at the expense of larger and possibly more important matters.

This is a familiar complaint when a field of knowledge becomes a formal science. In many instances, it is best ignored: real progress is often made through precision rather than speculation. But in this instance focusing on the impact of policies may cause us to lose sight of the framework within which they are chosen, and to miss out on one of the most important consequences of aid.

To illustrate: after the failure of Kenyan President Daniel Arap Moi to abide by the conditions attached to international loans his country was given in the late 1990s, the IMF and World Bank suspended further lending to Kenya. In December 2002 the opposition seized the presidency, and the new president, Mwai Kibaki, began to attack the problems that Moi had ignored. After a decent interval, the international financial community issued Kenya a clean bill of health, and aid once again began to flow.

While many rejoiced, members of the Kenyan National Assembly did not. Uhuru Kenyatta, himself an unsuccessful presidential candidate, declaimed from the floor that the resumption of aid had hindered the National Assembly’s influence over Kibaki, who had once been forced to come to the National Assembly for funds. With the resumption of international assistance, he did not. Kenyatta argued that it was now more in Moi’s interest to cater to bankers and policymakers in Paris, London, and Washington than to Kenyans.

Better methodologies may enable technocrats to design better public policies. But, as Kenyatta noted, governments are less accountable to those at home when public revenues flow into a country from without. With foreign donors to support them, they need not bargain with their own people or exchange good policies for tax payments. The manner in which policies are financed may thus be as significant as the content of the policies themselves.

Such judgments are not readily subject to measurement. But any reckoning of the merits of development aid will surely have to take them into account.