- Why the World Bank needs a research department: Penny Goldberg offers a strong rationale on Let’s Talk Development
- On VoxDev, Battaglia, Gulesci and Madestam summarize their work on flexible credit contracts, which is one my favorite recent papers – they worked with BRAC in Bangladesh to offer borrowers a 12 month loan, with borrowers having the option to delay up to two monthly repayments at any time during the loan cycle. This appears to be a win-win, with the borrowers being more likely to grow their firms, and the bank experiencing lower default and higher client retention. However, although the post doesn’t discuss it, the product seemed less successful in helping larger SMEs.
- Political business cycles in Africa – Rachel Strohm notes a Quartz Africa story on a phenomenon that has held up a number of my impact evaluations – “Having contracts stalled and major projects abandoned is “very common”... The uncertainty is also magnified because newly-elected administrations could take months to form a cabinet and appoint heads of key agencies... as a bulk of voters travel to their ancestral homes to cast their ballot, businesses are forced to shutter or maintain skeletal operations... [this] has even made phrases like “after elections” a colloquial mainstay”.
- The JDE interviews Eric Edmonds about his experience with the registered report process: “I thought I wrote really good pre-analysis plans and then I saw the template and realized, no, I write really bad pre-analysis plans too. I think just the act of providing that template to give some kind of standardization, is a great service to the profession... I think we need to be in a place where we have pre-analysis plans and we review them, and when we choose to deviate from them in our analysis, we're just able to be clear and to talk about why that is.” (h/t Ryan Edwards)
Evaluating Infrastructure Development
Investment in infrastructure is a key lever for economic growth in developing countries; to this end, World Bank financing for infrastructure is roughly 40% of its total commitments. Knowing the impact of these investments is therefore crucial for policy, but estimating the impact of these investments is difficult: Infrastructure is frequently targeted towards regions where growth is anticipated and coupled with complementary investments. Therefore, separating the impacts of any one investment from others or even from pre-existing growth trends is hard. This explains why development economists are pretty obsessed with finding ways to estimate the causal impact of infrastructure projects, which has led to many creative solutions. One possible option is to use spatial jumps.
Last January, I decided to start signing my referee reports and wrote a blog post about it. Partly because it felt like something I should do and partly because it was a commitment device to try to useful but critical referee reports without sounding mean. Economics suffers from many ills that it has been trying to address, and while mean and overreaching referee reports are not at the top of the list, they are something everyone has experienced and complained about at least once…So, now that I have been signing referee reports for about 15 months, how has it gone?
For quite a few reasons, many researchers have become increasingly skeptical of a lot of attempts to use instrumental variables for causal estimation. However, one type of instrument that has enjoyed a surge in popularity is what is known as the “judge leniency” design. It has particularly caught my attention recently through a couple of applications where the judges are not actually court judges, and it seems like there could be quite a few other applications out there. I therefore thought I’d summarize this design, these recent applications, and key things to watch out for.
The basic judge leniency set-up.
This design appears to have gained first prominence through studies which look at the impact of different types of experience with the criminal legal system. A classic example is Kling (2006, AER), who wants to look at the impact of incarceration length (S) on subsequent labor outcomes (Y). That is, he would like to estimate an equation like:
Y(i) = a + bS(i) + c’X(i)+ e(i)
The concern, of course, is that even controlling for observable differences X(i), people who get longer prison sentences might be different from those who get given shorter sentences, in ways that matter for future labor earnings.
- The St Louis Fed has a new “Women in Economics” podcast series, highlighting not just academics, but also economists in the private sector and thinktanks – so far the only person among their 14 episodes who has worked on development economics is Una Osili - who shares how her work on development in turn led to an interest in the economics of Philanthropy.
- In the IZA World of Labor, Ach Adhvaryu gives a nice summary of the emerging literature on the link between managerial quality and worker productivity in developing countries.
- EGAP’s 10 things to know about survey implementation: lots of nitty-gritty advice on things like training, using PDAs, and budgeting.
- On Let’s Talk Development, Caio Piza, Astrid Zwager and Isabela Furtado summarize results of an impact evaluation they did of productive alliances for farmers in Brazil. They find positive impacts of improving linkages between smallholder producers, buyers and the public sector through financing things like packinghouses, transport and logistics improvements, and constructing processing structures.
"In 2016, 61 million children of primary school age...were not in school, along with 202 million children of secondary school age." That's a tragic number, and it's also a concrete image. While we may have trouble envisioning 61 million children, we have a clear picture in our heads as to what a child not in school looks like, and we have a picture of what it looks like to have a child formerly not in school now in school.
But what about learning? What does improved learning look like? There are lots of studies that examine how to improve learning in low- and middle-income countries. Some report striking learning gains: A technology-aided instruction program in India finds that participation for 90 days would increase math scores by ... 0.6 standard deviations. For the vast majority of people in the world, the first response to that would be, "What's a standard deviation?" Even for educationists and economists, it's hard to envision the difference between the child with and without 0.6 standard deviations additional mathematical learning. (FYI, 0.6 standard deviations is a big learning gain.)