Publications

Microcredit programs increasingly pair loans with training in hard skills, like financial literacy, or soft skills, like aspirations and self-efficacy. We test both in three small randomized trials in India, Peru, and Uganda, assigning microentrepreneurs to a mobile financial-literacy app, a documentary video designed to raise aspirations, both, or neither. The app raised financial literacy by about 0.28 standard deviations, alone or combined. The video raised an aspirational-hope index by 0.33 SD alone and 0.30 SD combined, driven by higher aspirations, a stronger sense of agency, and clearer pathways out of poverty.

This paper asks whether practicing financial decisions can change the economic opportunities people perceive, and whether discussing aspirations can change their financial practices. A four-arm experiment among 348 women in 34 self-help groups in rural India compares a farm-finance game, an aspirations program, their combination, and control. At endline in the final intervention week, baseline-adjusted estimates for the game are 0.694 standard deviations for hope and 0.517 for perceived pathways toward goals; the aspirations program yields a 0.513-standard-deviation estimate for reported financial behavior. The point estimates cross domains: the game has larger hope and pathways estimates, while the aspirations program has a larger behavior estimate. Agency estimates are similar. This pattern motivates a distinction between wanting a better outcome and seeing how financial decisions can help attain it. A simulated farm makes the sequence from capital to productive assets and income concrete, potentially connecting financial practice to perceived opportunity. The arm rankings remain imprecise, and family-adjusted inference is strongest for aspirations and reported behavior. Short follow-up and selective outcome availability limit the conclusions. The evidence motivates evaluating financial education through both the skills it teaches and the possibilities it makes visible.

What scientists choose not to publish can be as important as what they do publish. This paper measures missing research results at the level of individual pre-registered hypotheses and uses this approach to study under-reporting in economics. We develop a standardized procedure for encoding hypotheses from registry entries and pre-analysis plans and apply it to 317 studies registered in the American Economic Association RCT Registry between 2015 and 2017, yielding 5,057 pre-registered hypotheses. Eight to nine years after registration, we search published papers, working papers, reports and other public materials to document whether a result is available for each hypothesis and, if so, whether it is statistically significant. We benchmark these patterns against expert forecasts. We then run a pre-registered randomized controlled trial with study authors to test which barriers keep results from becoming public: awareness of the value of reporting, the effort of documenting results (addressed through direct engagement with a pre-filled report of their hypotheses and our findings), or a lack of research resources. Finally, we introduce the Results Report, a simple, standardized format for reporting results of pre-registered hypotheses that registries and journals could adopt.

Do parents’ stated fertility preferences respond to the children they actually have? Ideal family size and desired sex composition are widely used to measure unwanted fertility and son preference, on the assumption that they are formed before births occur. We test that assumption using the sex of a first birth as a natural experiment. Drawing on survey data from more than 90,000 first-time mothers and 30,000 first-time fathers across many countries, we estimate how a first-born daughter shifts parents’ desired number of daughters, sons, and children, and compare how mothers and fathers update.