Knowledge that Transforms

Empirical peer effects research often employs censored peer data. Individuals may list only a fixed number of links, implying mismeasured peer variables. I first document that censoring is widespread in network data. I then introduce an estimator and characterize its inconsistency analytically; an assumption on the ordering of peers implies that censoring causes attenuated peer effects estimates. Next, I demonstrate the effect of censoring in two data sets, showing that estimates with censored data underestimate peer influence. I discuss interpretation of estimates, propose methods for correction and bounding, and give implications for the design of network surveys.

Recent work situates gender norms as a key driver of the sexual division of labor. But the explanatory power of Becker’s comparative advantage explanation is still not well understood. Drawing on unique data, we test the predictions of a formal Beckerian model. We complement this by proposing and analyzing new measures of specialization. We show that comparative advantage plays little or no role in the sexual division of labor within couple households. Absolute advantage also plays no role in specialization for same-sex couples, and this is not explained by having fewer children.

This paper examines inequalities in the match between student and degree quality using linked administrative data from schools, universities, and tax authorities. We analyze two measures of match at the university subject level: undergraduate enrollment qualifications and graduate earnings. We find for both that disadvantaged students match to lower-quality degrees across the entire distribution of achievement in a setting with uniform fees and a generous financial aid system. While there are negligible gender gaps in academic match, high-attaining women systematically undermatch in terms of expected earnings, driven by subject choice. These inequalities in match are largest among the most undermatched.

Using College and Beyond data and a variant of Dale and Krueger’s matched-applicant approach, we revisit the question of how attending an elite college affects later-life outcomes. We expand the scope by examining additional outcomes and not restricting the sample to full-time workers. For men, controlling for selection eliminates the relationship between college selectivity and earnings; there are also no effects on men’s educational attainment or family outcomes. We find significant effects for women: attending a school with a 100-point-higher average SAT score increases women’s probability of advanced degree attainment and earnings while reducing their likelihood of marriage.

We present the first analysis of the effect of grade retention on adult criminal convictions, exploiting test cutoffs for ninth-grade promotion in Louisiana. Eighth-grade retention increases the likelihood of violent crime conviction by 1.05 percentage points (58.44%) and increases the number of violent crime convictions at first conviction. The effects are likely driven by declines in high school peer quality and reduced educational investments that result in lower noncognitive skill acquisition. Extrapolating effects away from the cutoff shows that our results are generalizable to a larger group of low-performing students and are evident for both property and drug crimes.

We assess the effect of the minimum wage on labor market outcomes. First, we apply modern machine learning tools to predict who is affected by the policy. Second, we implement an event study using 172 prominent minimum wage increases between 1979 and 2019. We find a clear increase in wages of affected workers and no change in employment. Furthermore, minimum wage increases have no effect on the unemployment rate, labor force participation, or labor market transitions. Overall, these findings provide little evidence of changing search effort in response to a minimum wage increase.

The utility of machine learning (ML) for regression-based causal inference is illustrated by using lasso to select control variables for estimates of college characteristics’ wage effects. Post-double-selection lasso offers a path to data-driven sensitivity analysis. ML also seems useful for an instrumental variables (IV) first stage, since two-stage least squares (2SLS) bias reflects overfitting. While ML-based instrument selection can improve on 2SLS, split-sample IV and limited information maximum likelihood do better. Finally, we use ML to choose IV controls. Here, ML creates artificial exclusion restrictions, generating spurious findings. On balance, ML seems ill-suited to IV applications in labor economics.

Parents spend considerable time and resources investing in their children’s development. Given evidence that the Earned Income Tax Credit (EITC) affects maternal labor supply, we investigate how the maximum available EITC amount affects a broad array of time use activities, focusing on the amount and nature of time spent with children. Using 2003–18 time use data, we find that federal and state EITC expansions increase maternal work time, reducing time devoted to home production, leisure, and time with children. However, almost none of the reduction comes from time devoted to “investment” activities, such as active learning and development activities.

This paper examines the evidence from randomized evaluations of sector-focused training programs that target low-wage workers and combine up-front screening, occupational and soft-skills training, and wraparound services. The programs generate substantial and persistent earnings gains (12%–34%) following training. Theoretical mechanisms for program impacts are explored for the WorkAdvance demonstration. Earnings gains are generated by getting participants into higher-wage jobs in higher-earning industries and occupations, not just by raising employment. Training in transferable and certifiable skills (likely underprovided from poaching concerns) and reductions of employment barriers to high-wage sectors for nontraditional workers appear to play key roles.

Female workers earn $0.89 for each male-worker dollar even in a unionized workplace, where tasks, wages, and promotion schedules are identical for men and women by design. Using administrative time-card data on bus and train operators, we show that this earnings gap can be explained by female operators taking fewer hours of overtime and more hours of unpaid time off than male operators. Female operators, especially those with dependents, pursue schedule conventionality, predictability, and controllability more than male operators. While reducing schedule controllability can limit the earnings gap, it can also hurt female workers and their productivity.