Weekly links September 22: B+ research designs, Mexican cartel employment, detecting discrimination, using evidence at USAID, and more…


This page in:

·       David Deming discusses addressing big economic questions with B+ research designs, and gives as an example work by Hendricks and Schoellman on human capital and cross-country income differences. He also has a post on how decision-making skill is increasingly valued in the workplace, and an assignment game that can be used to try to measure this skill.

·       Noah Smith on industrial policy and why part of the difficulty in assessing it is that it can encompass many different things – and he argues economists have not studied it enough recently but are starting to again.

·       On VoxDev, Duflo, Keniston, Suri and Zipfel explain how looking within villages at treatment-control comparisons of a training program for coffee farmers in Rwanda would show positive effects, but these seem driven by the control group doing worse the higher the proportion of other farmers in the village treated – which they attribute to negative spillovers in the form of crowding out of inputs in short supply.

·       Russ Roberts interviews Roland Fryer on racial differences, detecting discrimination, affirmative action and more. A nice comment on what the null should be when thinking about discrimination in different settings “Life is all about who gets to claim the null hypothesis.”…and how his grandmother viewed his results “For my grandmother when I would share with her my results, my results were always put into two categories: obvious and wrong.” And on diversity training “It was research I was describing, that a set of sociologists have looked at--over a thousand corporate trainings, the results of them, even in the best conditions in terms of research design. And, what they found is that the average impact is zero for corporate trainings. And ones that are mandated, the average impact is actually negative on the future hiring and promotion of certain minority populations within those companies”

·       On Vox, Dylan Matthews interviews Dean Karlan about how to use evidence and impact evaluation at USAID and the need to make it easy for people to find evidence and use it through producing more “how-to” guides, and synthesis pieces – as well as when an impact evaluation is probably not needed.

·       On VoxDev, Manaswini Rao summarizes her work in India showing how increasing judicial staffing lowers the backlog of cases, allowing banks to recover debts stuck in debt recovery cases, and they then increase lending to local firms as a result.

·       In Science, a paper estimates that Cartels in Mexico employ around 175,000 people, about as many as Oxxo, the largest corner shop chain, and equivalent to being the fifth largest employer in the country. The paper estimates this number from a network-based model that models the change in cartel size as a function of new recruitments, incapacitation of existing members, conflict with other cartels, and a saturation factor that generates instability and internal factions as cartels grow. Lots of assumptions are needed with very imperfect data. One interesting estimate is that a decade after joining, they estimate that 17% of members are dead, and a further 20% incapacitated. The result is that cartels need to continuously recruit lots of new members, and they recommend that programs to reduce new recruitment will be more effective than increasing incarceration of existing members. Caulkins et al. offer some commentary and critique of the assumptions.


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation