· On VoxDev, Erin Kelley, Greg Lane and David Schönholzer look at how providing public information can improve safety in Kenya’s informal public transit sector. “We ran a randomized control trial with more than a thousand passengers at one of Kenya’s busiest bus terminals…We equipped 52 buses across five bus companies operating on a major route with safety tracking devices….We rated the top safety performers each month using the GPS data we collected. We privately provided the information on the safest companies by distributing pamphlets to a randomly selected group of passengers without informing the bus companies of these pamphlets. We then monitored which transit companies consumers chose….In addition to the pamphlets, we erected large signs across the transit station publicizing our tracking efforts. We then monitored consumers’ choice of transit company, and whether transit companies improved their safety performance…we find that passengers do not respond to the safety information they receive privately…passengers did respond to publicly provided information [and] additional publicly available safety information improved transit driver safety among the lowest performing companies.”
· Dan Millimet has a blogpost that looks at how estimators like the least absolute deviation and tobit estimator can be more robust than OLS when there is missing data – useful in cases where you think people drop out of the sample because they would have low values – e.g. people who would score lowly don’t take a college entrance test, or firms that would earn low profits exit. “If one imputes values of the dependent variable in a linear regression model estimated by LAD, the estimates will be consistent for the full population as long as the imputed values are on the correct side of the (conditional) median. For example, if one believes that all students that opt not to take the college entrance exam would have scored poorly had they taken the exam, one can replace the missing scores with zeros and estimate the model using the full sample by LAD. The estimates will be consistent as long as the non-test takers would have scored below the (conditional) median, even if the scores would not have been identically zero. However, OLS requires more. While it does not require the imputations to be correct, it does require the imputation errors to be classical measurement error”
· The Economist summarizes a lot of research on how grandparents affect female labor supply and child outcomes in “the age of the grandparent has arrived”.
· With the job market currently on, here is our advice from last year for those thinking about doing post-docs or hosting them.
· Reminder: the BREAD/IGC Virtual PhD course on firms and development is now underway, with lectures Thursday and Friday. The website has slides and video from the first few talks – next week Chris Woodruff and I are talking on constraints to capital and labor on Thursday 26 January, and Rocco Macchiavello on contracts on Friday 27 January.
· Another good XKCD cartoon on subgroup analysis.
Join the Conversation