A Nobel prize for using RCTs?

Economics as a discipline doesn’t always connect with the real world, but at least the Nobel laureates with an interest in development economics are working on something that matters.  The new laureates have apparently been selected on the basis of their ‘experimental’ approach.  I gave the subject a brief mention in my forthcoming book, The Poverty of Nations, but you’ll have to wait to next March before you can read it.  Here, as a spoiler, are the sentences in the typescript I gave over to their evidence on Randomised Control Trials.  The reference is to A Banerjee, E Duflo, 2011, Poor economics, published by Penguin.

Banerji and Duflo advocate a greater use of RCTs, but their own examples show cases where this fails. In one study they mention, it appeared that textbooks did not help education; that was misleading. In another, the evaluation was supposed to identify the influence of contraception on family size; it overlooked the importance for parental decisions of the  prospects of children surviving to adulthood. Experiments and RCTs work by screening out extraneous information; the gaps that are left are only to be expected.


Easterly argues that the Washington Consensus worked: post hoc, propter hoc?

Development economist Bill Easterly  has posted a new paper arguing that the “Washington Consensus” and structural adjustment might have worked after all.  These were the basis for the liberal market policies forced on developing countries by the IMF and the World Bank in the 80s and 90s.   The argument is that although most of the measures failed to show any consistent benefits at the time, subsequent improvements in development might not have happened without it.

There are three core problems with that position.  The first problem is evidential: showing that something happens over a long period of time does not show that a policy near the beginning is what started it.  If structural adjustment really did work, there should be evidence of it starting to work at the time, and evidence that countries which did it more faithfully had better results.  There really isn’t.  Second, the ‘policy outcomes’ Easterley uses as a test – currency value, inflation rates, trade shares and so on – are not necessarily the outcomes of policy at all; they are indicators that economies have avoided some of the problems that impede growth.  Third, over that length of time, there have been lots of other influences.  The massive improvements in recent years might just be attributable to poverty reduction strategies, the growth of democracy, improved governance, basic health care, the internet and the cellphone, the advancement of education, cash transfers, women’s rights and many other things.  The more influence we attribute to any of those – and I’d argue that they all matter – the less we attribute to structural adjustment.