Knowledge that Transforms

In many settings, a policy discontinuity comprises several treatments that cannot be separately identified using a standard regression discontinuity design. I propose a method for identifying distinct treatment components from a single discontinuity by exploiting the asymmetry between entities entering versus exiting treatment. Using data from New York City for 2009–13, I apply my strategy to the discontinuity associated with the introduction of class size caps—a widespread approach for reducing class sizes. I find that class size reductions increase student achievement, although these gains are counteracted by a newly hired teacher effect. The method has broad potential applicability.

This paper provides evidence for why individuals join unions instead of free riding. I model membership as legal insurance. To test the model, I use the incidence of news stories concerning allegations against teachers in the United Kingdom as a plausibly exogenous shock to demand for such insurance. I find that for every five stories occurring in a region, teachers are 2.2 percentage points more likely to be members in the subsequent year. These effects are larger when teachers share characteristics with the news story and can explain 45% of the growth in teacher union membership between 1992 and 2010.

This paper studies the expansion of US child support policies from 1977 to 1992 and its consequences for marriage and fertility decisions. I develop a model showing that child support enforces ex ante commitment from men to provide financial support in the event of a child, which (1) increases premarital sex among couples unlikely to marry and (2) reduces the abortion rate by reducing the cost of child-rearing to single moms. Using variation in the rollout relative to the timing of nonmarital pregnancy, I find that child support policies reduced the likelihood of marriage and reduced the abortion rate.

The Self-Sufficiency Project (SSP) was a well-known welfare-to-work experiment that provided a generous but time-limited financial incentive to leave welfare and enter the workforce. Experimental evidence showed large short-term impacts but no lasting effects. We argue that these conclusions need to be reassessed. Policy changes implemented during the SSP implied that the control group’s behavior did not provide an appropriate counterfactual. We estimate the impacts the financial incentive would have had in an unchanging policy environment. This reassessment leads to significant changes in the lessons previously reached. Our study demonstrates that experimental findings need to be interpreted with care.

This paper reports findings from a randomized evaluation of an intensive tutoring program conducted in underprivileged high schools. Within each school, the intervention targets students identified as having the ability to pursue a college education. The program is designed to strengthen their readiness for higher education. We demonstrate that such an intervention can have negative effects on a large fraction of participants, even though participation is entirely voluntary. This result is consistent with a simple model where time invested in extracurricular programs and time invested in homework represent imperfect substitutes in the education production function.

This paper presents findings from a national experimental evaluation of performance bonuses funded by the Teacher Incentive Fund grant program. The study finds no robust evidence of positive impacts of bonuses on student achievement, although some specifications suggest a small positive effect that compares favorably with other education interventions. When controlling for covariates, we find that offering bonuses of an average yearly cost of $100 per student had a small significant impact of about 0.04 standard deviations. However, these impacts are smaller (0.01 standard deviations) and become insignificant when not controlling for covariates or using an alternative method of inference.

Conscription forces young men to serve in the military, irrespective of their opportunity costs. Using random assignment in the Danish draft lottery, we find a negative average effect of peacetime military service on earnings, an effect varying by ability, with high-ability men facing a $23,000 lifetime earnings penalty but with low-ability men facing none. This gradient is robust to different measures of labor market prospects. Educational disruption is an important mechanism, while service has little effect on health or criminal activity. The opportunity costs of conscription are borne by men with the best labor market prospects.

We present experimental evidence on four US reemployment programs targeting unemployment insurance (UI) recipients during the Great Recession. All programs reduced UI spells, produced UI savings that exceeded program costs, and increased employment rates. The services referral program had the smallest effects, occurring because of voluntary UI exits by participants to avoid requirements. The two programs reviewing participants’ UI eligibility produced greater effects because they induced voluntary exits and disqualified participants who did not meet UI requirements. The program requiring participation in both eligibility review and job-counseling services was the most effective, suggesting that services improved participants’ job search efforts.