Development Economics and the World Bank President

I am very bust this week what with the start of the Fall term and all, so my blogging has been sparse and that theme will continue the rest of the week I am afraid.  Today I am going to pass on something that Dani Rodrik wrote over at his blog:

World Bank president Bob Zoellick gave an interesting speech today at Georgetown. I wasn't there, but have seen his prepared remarks, which I find both forthright and courageous.

From my perspective, the speech hits all the right notes: the need for economists to demonstrate humility, eschew blueprints, search for differentiated solutions suited to context, learn from the actual policies of successful emerging economies, focus on evaluation but not at the expense of the big questions.

Zoellick, admittedly not an economist himself, goes further and takes on the economics profession. He has many valid points. He wades in on some of the key debates in the profession and doesn’t mince his words. Particularly telling are his surprisingly frank criticism of the current fad with randomized evaluations, which he decries as being too narrow and too unconcerned with scalability, and his open-mindedness towards policies that promote industry through various forms of incentives.

The main theme of his speech is “democratizing development economics.” I like this as a slogan, but fear that it may end up another gimmick. Zoellick offers no new ideas on the governance and internal organization of the Bank. And without changes in these, the bulk of the Bank’s research will continue to be done in Washington, DC by economists from advanced nations.

The criticism of the randomized trial fad is welcome, in my view, not because they are not worthwhile, but the proponents of randomized trials are far too ardent in their dismissal of all other approaches and far to optimistic about the value of randomized trials to solve the world's problems. The external validity (or scalability) problem is very serious one - easily as serious as the problems with isolating causal links in real world data.