Published on Development Impact

Weekly links March 11: Digital credit, falsifying IV assumptions, the value of research departments, and more…

This page in:


·       With the great title of Festivus Miracle,  Jonathan Roth blogs about how, in the case of heterogeneous treatment effects, you can actually falsify some of the IV assumptions in certain cases – you can test inequalities that need to hold if the instrument is valid and there are more compliers than defiers.

·       Scott Cunningham discusses consideration set IV designs, especially as used in industry applications. The application is with platforms that attempt to match two sides of the market (e.g. Yelp with restaurants and customers; VRBO or AirBNB with accommodation owners and customers; Talkspace with mental health specialists and those seeking care; etc.). The idea is that you can’t randomize which restaurant, accommodation owner, or therapist someone gets, but company’s do control the “consideration set” – the set of choices that get presented to a potential customer. There may be random variation in the characteristics of this set that could be used as an instrument.  For example, Jaffe et al. explain this variation with respect to AirBNB “Frequent bookings, changes in availability, and ongoing search experiments mean that guests searching for the same market and check-in date on the same day are shown different search results. Their choice for where to stay is affected by these results”.

·       The Richmond Fed Econ Focus interview for the first quarter 2022 is with Penny Goldberg. Her view on trade and globalization “I don't think that Africa can play the same role in the future that China or Vietnam or Korea played in the past. The conditions for such export-led growth are not there anymore. I cannot imagine the United States opening its borders these days to an influx of imports from low-wage countries in Africa. So I don't think this model is viable anymore.”. She also discusses her work with Tristan Reed on why deaths from COVID-19 were lower in developing countries than the U.S. And on whether it is useful for the World Bank and Fed to have their own research departments, and what some of the differences are from academia – “People who are close to policy tend to have a much better sense, at least in the short run, of what the important questions are. So in terms of coming up with and framing questions, they're often ahead of academia. I think that's an important service that research departments in these institutions provide”.

·       On Ideas for India, Jean Drèze discusses some perils of embedded experiments in the context of an experiment done to test a new financial management system for MNREGA in Bihar, which the authors disseminate as a case of reducing corruption, but which he notes was an operational disaster and resulting in a lot of delays in payments “The picture that emerges from this case study is a little unsettling. A counter-productive experiment, aborted within a few months, ended up being perceived as a model exercise in evidence-based policy.” He uses this to raise his concerns with embedded experiments more generally “‘embedded experiments’ – at-scale experiments conducted by a research team that attaches itself to a government department. Without denying their possible value, it seems to me that these experiments are fraught with dangers. These include imbibing the perspective of the partner government, distorting its priorities, diverting senior officials from their primary responsibilities, conceding some control over research findings, exercising power without accountability, overestimating government capacity, underestimating State oppression, inappropriate randomisation, violation of consent norms, misuse of privileged access, conflicts of interest, revolving-door effects, distracting incentives, and contamination.”

·       I found this substack by Abraham Thomas on the economics of data businesses interesting. A useful discussion of all the ways data companies collect and collate data, the business model, and this factoid “Abraham Lincoln, Ulysses S. Grant, Grover Cleveland and William McKinley all worked for the same data company” – one which is now 180 years old and still around today.

·       The World Bank has recently updated its gender data portal.

·       On VoxEU, Valentina Brailovskaya, Pascaline Dupas, and Jonathan Robinson review studies on the impact of digital credit in Malawi, Kenya and Nigeria. “To date, the evidence suggests that effects are modest and meet neither the worst fears discussed above nor the transformative effects touted by proponents. While the evidence base is still scant and comes from only a handful of completed studies, no study finds substantial negative effects, while one study finds improvements in household risk-coping ability and another finds modest effects on self-reported financial well-being. There remains serious reason for concern, however. In Malawi, we find that consumers are uninformed about loan terms and regularly pay back late, incurring late fees. This lack of knowledge creates a clear opportunity for exploitative lending.”

·       On the JPAL blog, Vianney Fernández summarizes evidence on what works to reduce intimate partner violence in Latin America.

·       Call for applications for the Robert McNamara Fellowship program: The World Bank Robert S. McNamara Fellowships Program (RSMFP) matches aspiring development economics researchers from developing countries with World Bank research economists, creating unique opportunities for the fellows to participate in rigorous policy-relevant research in the World Bank’s Development Economics Vice Presidency (DEC). Fellows will be hosted at the World Bank in Washington, D.C. for 8 months (September to May each year) and work under the supervision of researchers in the World Bank’s Development Impact Evaluation (DIME) and Development Research Group departments, engaging in high-quality and policy-relevant research projects. Applications due by April 4.


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000