In late August, the world’s most high-profile development project celebrated another milestone. The Millennium Villages Project opened its newest site, in northern Ghana, with newly minted Ghanaian president John Mahama and the UK’s international development secretary on hand. The official interest in the project’s arrival in Ghana only emphasized its unique stature in the typically wonkish world of development economics. Wherever MVP founder, Columbia professor, bestselling author and former World Bank presidential hopeful Jeffrey Sachs and his vision of aggressive, integrated rural development go, money, attention and political approval are almost bound to follow.
The project’s expansion and Sachs’s photo-op in Ghana belies the fact that just a few months ago, the empirical basis of Sachs’s enterprise was revealed to be flimsy. On May 8, the British medical journal Lancet published a paper that was supposed to vindicate the MVP, which has attempted to prove that an intensive and targeted sustainable development program could improve the lives of villagers in 10 African nations—and maybe in the rest of the developing world as well. He sold the MVP as nothing less than a revolution in development policy. The trouble is that the project has had difficulty proving that it was having its intended effect, or even any effect. The Lancet paper was going to change that. Its co-authors included John W. McArthur, the one-time CEO of Millennium Promise, the MVP’s fundraising arm, and Sachs.
Instead, the opposite happened. A letter from four respected development experts—Jesse Bump of Georgetown University, Michael Clemens of the Center for Global Development, Gabriel Demombynes of the World Bank, and Lawrence Haddad of the University of Sussex—demonstrated that one of the paper’s key findings, that child mortality decreased in the Millennium Villages at three times the national rate, had been reached in error. On May 21, the MVP withdrew the claim.
In the summer, I spoke to both Clemens and Bump. As social scientists, they are willing to believe that the MVP’s integrated rural development model is worthy of the political and financial support that’s currently lined up behind it—but only so long as there’s evidence that Sachs’s methods actually work. “The villages have put themselves up as a model, and there’s some burden of proof about demonstrating whether the model works or doesn’t work,” says Bump. “All we’re saying is that the proof isn’t here yet.”
The MVP has exhibited a startling anti-empirical streak that predates the Lancet controversy. When the project was launched in 2005, it included no control group, in this case villages that would not receive development assistance from the project, and whose observed progress could be compared to conditions in the intervention area. Without a control group, the MVP was able to claim that any improvements in the villages were a direct result of the project’s interventions. For instance, in a 2008 report the MVP credited itself for an increase in cell phone usage in the villages—even though cell phone access had been dramatically increasing throughout Africa over the same period.