Knowledge that Transforms

Abstract Parole was eliminated for many US offenders by Truth-in-Sentencing (TIS) laws in the 1990s. I exploit the introduction of TIS in Arizona to explore its impact on offenders before, during, and after incarceration. TIS Offenders were assigned significantly shorter sentences, largely eliminating the intended increase in punishment. These offenders reduced their rehabilitative effort while incarcerated, with rule infractions increasing by 22% and education enrollment falling by 24%. Finally, TIS offenders became 23% more likely to return to prison for a new conviction. I argue these effects were driven by TIS removing parole incentives, given that time served remained largely unchanged.

Abstract We use variational inference (VI), a technique from the machine learning literature, to estimate a mortality-based Bayesian model of nursing home quality accounting for selection. We demonstrate how one can use VI to quickly and flexibly estimate a high-dimensional economic model with large datasets. Using our facility quality estimates, we examine the correlates of quality and find that public report cards have near-zero correlation. We then show that in contrast to prior literature, higher quality nursing homes fared better during the pandemic: a one standard deviation increase in quality corresponds to 2.5% fewer Covid-19 cases.

Abstract We evaluate two interventions targeting young women in Nairobi, Kenya. The first is a multifaceted program involving vocational training, in-kind transfers of physical capital, and ongoing mentoring. The second is an unrestricted cash grant. Both interventions shift women into self-employment, and impacts persist after six years. Both programs also increase income in the short-term, but those effects disappear over time. Though the two treatments have similar impacts on labor market outcomes, women in the multifaceted program report significantly higher wellbeing six years after treatment relative to both women in the control group and those who received the grants.

Abstract We study the wellbeing returns from volunteering in England's National Health Service (NHS) Volunteer Responders, set up in response to Covid-19. Using linked survey and administrative data, we exploit the oversubscription of volunteers and the random allocation of tasks via an app to establish causality. Volunteers show stronger wellbeing and feelings of belongingness and connectedness to their local area. Welfare analyses suggest that the benefits of the programme substantially exceeded its costs. We are the first to study the welfare effects of a nationwide volunteering programme. Our findings show that pro-social behaviour improves personal wellbeing as well as social welfare.

Abstract Individuals with extensive debt may be granted debt relief in court. We provide a comprehensive evaluation of the Danish debt relief program with data from court records linked to nationwide register data. Using event-study methods and quasi-random assignment of applicants to court trustees with varying admission rates, we show that debt relief leads to a large increase in earned income, employment, assets, real estate, secured debt, home ownership, and wealth that persists for more than 25 years after a court ruling. The net transition of workers into employment accounts for two thirds of the increase in earned income.

Abstract Changes in the minimum wage may impact college enrollment and educational attainment. Using institutional data on college enrollment and program completion, we find that enrollment falls markedly among students at public two-year institutions in response to increases in the minimum wage. The largest enrollment effects are seen for those students who are enrolled part-time at community colleges. We find little evidence of negative effects on the attainment of certificates or degrees, suggesting that increases in the minimum wage are unlikely to divert students from degree attainment.

Abstract Research and development tax credits often aim to increase investment in experimentation, hoping that firms invent fundamentally new technologies that in turn generate positive spillovers. Since most policies require that companies make profits in order to claim credits, they might also shift investments towards less risky refinement and exploitation. Following the availability of credits, we demonstrate that firms do not experiment but deepen invention in areas of extant expertise. We observe stronger shifts for firms operating in uncertain markets where search failures are more likely to reduce credit eligibility.

Abstract Do more educated workers earn higher wages partly because they have access to high-paying firms and occupations? We rely on linked employer-employee data to combine the estimation of AKM models with the decomposition of the returns to schooling. We exploit exogenous variation in education driven by changes in compulsory education. We show that education provides access to better-paying workplaces and occupations: 30 percent of the overall return to education operates through the workplace channel and 12 percent through the occupation channel. The remainder is associated exclusively with the individual. Match quality plays a modest role in the returns to education.

Abstract India is home to a comprehensive affirmative action program that reserves a fraction of positions at governmental institutions for various disadvantaged groups. While there is a Supreme Court-endorsed mechanism to implement these reservation policies when all positions are identical, courts have refrained from endorsing explicit mechanisms when positions are heterogeneous. This lacuna has resulted in widespread adoption of unconstitutional mechanisms, countless lawsuits, and inconsistent court rulings. By formulating mandates from the landmark Supreme Court judgment Saurav Yadav vs The State of U.P. (2020) as technical axioms, we show that the 2SMH-DA mechanism is uniquely suited to overcome these challenges.

Abstract We derive a new powerful, simple test for mediation. Testing for mediation is empirically important and the no-mediation hypothesis H0:θ1θ2=0 poses a theoretical problem, nonregular at the origin, where standard tests have extremely low rejection probabilities. We prove that a similar test only exists if 1/α∈N. It is unique, but with unfortunate properties. An innovative varying-g method is introduced to derive a nearly similar test, gopt, with power close to the envelope without these undesirable properties, and a second, less similar test, g1.2, with more empirical appeal. Results are illustrated in an educational setting.