· Karthik Muralidharan and Abhijeet Singh have a nice policy forum piece in last week’s Science discussing India’s new National Education Policy and what existing research tells us about the challenges the education system in India faces, and where there are promising avenues for policy to help improve learning.
· BU’s The Brink on Mario Kart as an analogy for what social and economic programs need to do “the idea is a lot like the way that Mario Kart gives players falling behind in the race the best power-ups, designed to bump them towards the front of the pack and keep them in the race. Meanwhile, faster players in the front don’t get these same boosts, and instead typically get weaker powers, such as banana peels to trip up a racer behind them or an ink splat to disrupt the other players’ screens. This boosting principle is called “rubber banding,” and it’s what keeps the game fun and interesting, Bell says, since there is always a chance for you to get ahead. “And that’s exactly what we want to do in development,” he says. “And it is really, really difficult to do.””
· Interesting discussion of the difficulties of using iterative pre-analysis plans in a complicated study: section VIII of Bidwell, Casey and Glennerster (JPE, 2020) notes “In this study, we sought to achieve both prespecification and iterative analysis by prespecifying a sequence of analytical steps across our three experiments. We planned breaks to incorporate lessons learned into a revised plan in a transparent way. In theory, this is unequivocally advantageous: the dynamic adaptation does not constitute data mining because the revised hypotheses are tested on a new sample and, in our case, a new exogenous source of variation. In practice, we encountered challenges that reveal how economics currently has neither the infrastructure nor the peer review system necessary to make iterative prespecification effective….unlike FDA trials, economics has no system of data gatekeepers that authors can use to credibly lock away some data while they finalize analysis plans. In our case, we planned to use results from one experiment to refine our analysis plan for a second, concurrent experiment. However, without an approved gatekeeper, we could not demonstrate that the changes to our preanalysis plan covering the second experiment were truly prespecified…Prespecification is useful in reducing p-hacking only if readers can check that the final analysis matches the preanalysis plan. This accountability is undermined by complicated or iterative plans. Even reading our red-lined, date-stamped iterative plan became, in the words of one of our reviewers, “almost unbearable.” We admitted defeat, dropped the idea of prespecified iteration, and reverted to the original preanalysis plan as first lodged…Our conclusion is that while preanalysis plans can be useful, they are a blunt instrument. They are most effective for simpler experiments. As the number of experimental layers or arms rises, the cost of giving up the ability to iterate increases, and the transparency of the preanalysis plan falls.” (Thanks to Rachel Glennerster for tweeting about this – another case of something useful in that is missed if you only see the working paper).
· The demand for deferred payments: Jason Kerwin blogs on his forthcoming AER paper that examines the impact of a voluntary program for workers in a tea company in Malawi that enabled them to opt to have some of their pay withheld each payday and paid in a lump-sum at the end of the harvest season. “About half of workers chose to sign up for the product; this choice was actually implemented at random for half of the workers who signed up. Workers who signed up saved 14% of their income in the scheme and increased their net savings by 23%. The savings product has lasting effects on wealth. Workers spent a large fraction of their savings on durable goods, especially goods used for home improvements.”
· Scott Cunningham has an explainer on the Sun and Abraham 2020 Journal of Econometrics paper on “Estimating Dynamic Treatment Effects in Event Studies with Heterogenous Treatment Effects” including a nice interview with Liyang Sun that discusses the background of how she got into this work – in part from a throwaway comment from a professor in a class lecture.
· The latest CSWEP fireside chat featured Oyebola Okunogbe and Danila Serra in a nice discussion with Ben Olken about publishing in the AEJ Applied. The video is up on the CSWEP website. It is well worth a watch/listen to get Ben’s views on what constitutes “general interest” (will it be on a syllabus? Will it be a paper that others working on this topic will need to cite?), key mistakes to avoid/small improvements that can help (making sure tables and figures are self-contained, working hard on nice figures to show results, especially for event study/RD/DiD work) and on what he looks for in a referee report and how they use them.
· Alexandra Cirone provides examples of instruments used in historical political economy research, and how to go about finding them.
· Funding opportunity: Request for Proposals: BRAC Institute for Governance and Development’s Women’s Economic Empowerment and Digital Finance (WEE-DiFine), a research initiative aimed at generating evidence on the causal impact of digital financial services on women’s economic empowerment in South Asia and Sub-Saharan Africa, returns with a second funding call for small grants up to USD 50,000 dedicated to rigorous research on pilot, measurement and qualitative studies. The deadline to submit proposals is May 31, 2021.
Join the Conversation