Response to Comment on Duflo (2001)
Esther Duflo, Nathan Lazarus
, and I reply to the comments of Roodman 2022
on the data and econometric specifications of Duflo (2001)
. We replicate Roodman’s results and agree with his corrections: standard error clustering, data transcription mistakes, and incorporating survey weights. More importantly, Roodman describes bias in Duflo's DiD specifications as a result of Mincer equation–driven wage scale dilation. We run additional specifications that seek to correct for this bias and find mixed results. Finally, we supplement Roodman's proposal to use later followups — by which point the dilation will have dissipated or even reversed — by using a more complete dataset from Hsiao (2022)
and find significant positive effects, supporting Duflo’s findings. We also discuss additional weak instrument corrections. Slides here
, full reply forthcoming.
Do the crowds believe their own wisdom?
With Dean Karlan and coauthors, I study an intervention to increase user trust in reviews of disc golf courses on a popular mobile application. Results forthcoming.
Lead Poisoning in Poor Countries: Causes, Effects, and Solutions
Despite alarmingly high levels of lead poisoning, little is known about the sources of exposure to lead in poor countries, methods to eliminate that exposure, and the gains from doing so. I propose three studies to address these knowledge gaps: (1) An event study using exogenous variation in the timing of leaded gasoline bans across Africa; (2) a randomized experiment that staggers the rollout of a subsidy for lead-free paint; (3) an experiment that cross randomizes publicly available lead-testing equipment with the introduction of a reputable lead-free turmeric vendor. Proposal here
Depression and Loneliness among the Elderly Poor (2022)
Using a new longitudinal study of a large sample of elderly poor in Tamil Nadu, India and comparing it to similar datasets from seven other poor countries and the US, we describe trends in depression among the elderly. We find that depression in poor countries increases sharply with age, which appears to be driven by the increasing probability of living alone — especially among women, who have longer life expectancy and therefore are more likely to be widowed — resulting in loneliness. I did the research for and wrote the literature review, edited much of the rest of the paper, helped design the figures, and did some of the data analysis. Working paper here
. Currently working on revisions to be resubmitted.
Public Information is an Incentive for Politicians: Experimental Evidence from Delhi Elections
Two years prior to elections, two-thirds of Delhi municipal councilors learned that they had been randomly chosen for a pre-election newspaper report card. Treated councilors in high-slum areas increased pro-poor spending, relative both to control counterparts and treated counterparts from low-slum areas, and were more likely to be selected by the party to run in the next election. Treatment also increased the likelihood of the incumbent’s party winning the next election. In contrast, non-public disclosure of similar information to councilors had no effect. I helped edit this paper. Working paper here