09 November 2019

Effective policy needs experiments

Long-term, remote goals, such as universal literacy, the ending of war, or the elimination of poverty in developing countries, will require a range of solutions, varying over time and space, for their achievement. Conventional approaches to large-scale, complex problems rarely rely on hard evidence. Too often we read about the supposed 'root causes' of, for instance, violence being poverty and violence, with little supporting evidence. Sure, in some circumstances, tackling such alleged root causes might be the most efficient way of solving a problem. But in others such faith might be misplaced, or even counterproductive. Even Social Impact Bonds are limited when it comes to large-scale, remote goals, because they are inherently short term in nature, and don't allow for new entrants into the problem-solving services. (See here and here for why I am ambivalent about SIBs.)
 
Social Policy Bonds are different. Because they're tradeable, they can target long-term goals whose achievement might take years beyond the planning horizon of  investors. This allows bondholders to benefit from copious research aimed at finding, and funding, only the most efficient initiatives.

This sort of research and experimentation in matters of policy is rarely encouraged or rewarded, so I'm pleased that this year's Nobel prize for economics was awarded to three economists, Abhijit Banerjee, Esther Duflo and Michael Kremer, who used experiments to look at approaches to health care, education and entrepreneurship.

In the mid-1990s Mr Kremer ...began studying poverty with methods more commonly associated with chemists and biologists: randomised trials. If human capital—health, education, skills and so forth—is essential for development, then economists had better make sure they understand where it comes from. In Kenya he conducted field experiments in which schools were randomly divided into groups, some subject to a policy intervention and others not. He tested, among other things, additional textbooks, deworming treatments and financial incentives for teachers linked to their pupils’ progress. .... Educational resources—textbooks, say—turned out to do little for learning outcomes. Making pupils healthier improved their attendance, but did not necessarily mean they learned more. The experiments had a larger result, however: they taught the economics profession that randomised trials could work in the field. A Nobel economics prize goes to pioneers in understanding poverty, the 'Economist', 17 October
Right: they can work if the incentives are there to conduct them. The governments of all too many countries seldom think of offering such incentives. Philanthropists sometimes do, though (see here for example). They could take the lead in issuing Social Policy Bonds for the big, urgent goals that existing bodies are too small, too selfish or too incompetent to target effectively. See here for my piece on Social Policy Bonds published by a journal for philanthropists.

No comments: