· On VoxDev, Ervin Dervisevic summarizes his recent work with Markus Goldstein on how gender and marriage perceptions affect reporting of labor in Ghana. “In this context, the social norm is for men not to help their wives with family-oriented work such as growing food for the household, meaning that both women and men do not want to report when that does happen… It seems that male workers tend to under-report their own work relative to what their manager-spouse reports… An important finding of our study is that there are significant differences in male workers’ estimated labour productivities using manager and worker reports. Self-reports by male workers suggest underreporting of hours they work on spouses’ plots. This suggests that the prevalent logic assuming self-reporters as more reliable sources may be problematic. Our results show that there might be circumstances where proxy reporters are a better choice, especially if some of the respondents are motivated to misreport their own work based on gender norms or other factors.”
· Planet Money’s weekly newsletter covers a recent paper on what happens to customer support workers when a company adopts ChatGPT: “after the software company adopted AI, the average customer support representative became, on average, 14 percent more productive. They were able to resolve more customer issues per hour. That's huge. The company's workforce is now much faster and more effective. They're also, apparently, happier. Turnover has gone down, especially among new hires. Not only that, the company's customers are more satisfied. They give higher ratings to support staff…. not all employees gained equally from using AI. It turns out that the company's more experienced, highly skilled customer support agents saw little or no benefit from using it. It was mainly the less experienced, lower-skilled customer service reps who saw big gains in their job performance”.
· The latest JEP has several pieces of likely interest to our readers:
o Jishnu Das and Toan Do on what we are learning from 20 years of health insurance in developing countries: “Our conclusion from this review is that health insurance schemes have successfully increased financial protection and utilization, but there is little evidence (yet) of improvements in health outcomes. Further, there has been little demand for health insurance among households, even when it is heavily subsidized. .. we believe that health insurance triggered behavioral response among providers that have systematically undermined the objectives of insurance schemes.” And interesting factoid of the day “.. in Nigeria, 75 percent of healthcare providers see fewer than two patients a day”
o Doug Miller has an introductory guide to Event Study designs, along with supplementary materials that provide lots of example graphs and Stata code for replicating them. “Behind the scenes of the easily digestible event study picture, a researcher needs to make a number of choices….In this essay, I discuss the range of decisions that go into an event study model, and in this way I aim to improve the understanding of these models”. One recommendation I like (and have made before with DiD) is to not just show the treatment effects, but also the raw data so one can see differences in levels and what trends look like.
o Following on Lelys’ post this week on mental health, Banerjee and co-authors have a paper on depression and loneliness among the elderly in developing countries: “Our first key finding is that the prevalence of symptoms of depression among the elderly is much higher in poorer countries than it is in the United States…Second, many of the elderly in low-income countries feel lonely, despite the common presumption that most elderly in these countries live with their family”
· Congrats to our colleague Oyebola Okungobe and her co-author Victor Pouliquen for their paper on technology and taxation in Tajikistan being selected as the AEJ Policy best paper award for this year.
· Looking for a Stata Bayes expert: Rachael Meager, Leo Iacovone, and I are looking for someone who is an expert in Stata’s Bayesian analysis commands to hire as a short-term consultant. We want to see how much of our RStan analysis can be carried over into Stata. Should be 2-3 weeks work for an expert, to start as soon as possible – see details here and let Rachael and I know if you are interested.
· Call for papers: the Liberal Arts Colleges Development conference (LACDEV) will be held at Amherst College on September 29-30.
Join the Conversation