Alexandre Mas

Peer effects on worker output in the laboratory generalize to the field.

Comparing lab and field estimates What do Swiss high-school students and Eastern European seasonal laborers have in common? When the former are tasked with stuffing questionnaires into envelopes in a classroom setting and the latter are employed to pick fruit in the United Kingdom, both work harder in the presence of their peers. Herbst and Mas reanalyzed the results of 35 such studies, either experiments carried out under controlled conditions or empirical studies based on data collected in the field (see the Perspective by Charness and Fehr). Encouragingly, they found that the magnitude of the spillover effect—how much harder a worker works when other workers are alongside—was the same. Science, this issue p. 545; see also p. 512 Lab experiments and real-world observations are in agreement that people work harder when they work together. [Also see Perspective by Charness and Fehr] We compare estimates of peer effects on worker output in laboratory experiments and field studies from naturally occurring environments. The mean study-level estimate of a change in a worker’s productivity in response to an increase in a co-worker’s productivity (γ) is γ^ = 0.12 (SE = 0.03, nstudies = 34), with a between-study standard deviation τ = 0.16. The mean estimated γ^-values are close between laboratory and field studies (γ^lab−γ^field = 0.04, P = 0.55, nlab = 11, nfield = 23), as are estimates of between-study variance τ2 (τ^lab2−τ^field2=−0.003, P = 0.89). The small mean difference between laboratory and field estimates holds even after controlling for sample characteristics such as incentive schemes and work complexity (γ^lab−γ^field = 0.03, P = 0.62, nsamples = 46). Laboratory experiments generalize quantitatively in that they provide an accurate description of the mean and variance of productivity spillovers.

Welfare reform, time limits, and infant health

This paper offers evidence that welfare time limits contributed to a deterioration of infant health. We use the fact that the dates at which TANF recipients were first subject to timing out varied by state. We show that by 2000 there was a marked difference in TANF duration spells depending on whether the state employed the 60-month Federally imposed time limit, or a shorter limit, differences that were not present under AFDC. There were significant increases in infant mortality when time limits became binding in a state. These increases occurred primarily among mothers who could have plausibly timed-out of TANF: poorly educated and unmarried women with at least one previous live-birth. There is some evidence that the population of mothers affected by time limits were less likely to seek prenatal care in the first trimester, suggesting a possible role for reduced medical care in explaining the deterioration in infant health.

Potential Unemployment Insurance Duration and Labor Supply: The Individual and Market-Level Response to a Benefit Cut

We examine how a 16-week cut in potential unemployment insurance (UI) duration in Missouri affected search behavior of UI recipients and the aggregate labor market. Using a regression discontinuity design (RDD), we estimate marginal effects of maximum duration on UI and nonemployment spells of 0.45 and 0.25, respectively. We simulate the unemployment rate implied by the RDD estimates assuming no market-level externalities. The implied response closely approximates the decline in the unemployment rate following the benefit cut, suggesting that, even in a period of high unemployment, the labor market absorbed the influx of workers without crowding out other job seekers.

Does Transparency Lead to Pay Compression

This paper asks whether pay disclosure in the public sector changes wage setting at the top of the distribution. I examine a 2010 California mandate that required municipal salaries to be posted online. Among top managers, disclosure led to approximately 7 percent average compensation declines, and a 75 percent increase in their quit rate, relative to managers in cities that had already disclosed salaries. The wage cuts were largely nominal. Wage cuts were larger in cities with higher initial compensation, but not in cities where compensation was initially out of line with (measured) fundamentals. The response is more consistent with public aversion to high compensation than the effects of increased accountability.

Employment Effects of the Affordable Care Act Medicaid Expansions

We examine whether the recent expansions in Medicaid from the Affordable Care Act reduced “employment lock†among childless adults who were previously ineligible for public coverage. We compare employment in states that chose to expand Medicaid versus those that chose not to expand, before and after implementation. We find that although the expansion increased Medicaid coverage by 3.0 percentage points among childless adults, there was no significant impact on employment.

Sources of Displaced Workers’ Long-Term Earnings Losses

We estimate the magnitudes of reduced earnings, work hours, and wage rates of workers displaced during the Great Recession using linked employer-employee panel data from Washington State. Displaced workers’ earnings losses occurred mainly because hourly wage rates dropped at the time of displacement and recovered sluggishly. Lost employer-specific premiums explain only 17 percent of these losses. Fully 70 percent of displaced workers moved to employers paying the same or higher wage premiums than the displacing employers, but these workers nevertheless suffered substantial wage rate losses. Loss of valuable specific worker-employer matches explain more than half of the wage losses.