Knowledge that Transforms

In a recent article in the American Economic Review, Dahl and Lochner (2012) use changes in the Earned Income Tax Credit to estimate the causal effect of family income on child achievement. Their instrumental variable (IV ) estimates imply that a $1,000 increase in income raises combined math and reading test scores by about 6 percent of a standard deviation. I document a variable coding error. Correcting this error reduces the IV estimates by 32 percent; correcting this error does not change the qualitative conclusions of the study.

We study the evolution of trade liberalization's effects on Brazilian local labor markets. Regions facing larger tariff cuts experienced prolonged declines in formal sector employment and earnings relative to other regions. The impact of tariff changes on regional earnings 20 years after liberalization was three times the effect after 10 years. These increasing effects on regional earnings are inconsistent with conventional spatial equilibrium models, which predict declining effects due to spatial arbitrage. We investigate potential mechanisms, finding empirical support for a mechanism involving imperfect interregional labor mobility and dynamics in labor demand, driven by slow capital adjustment and agglomeration economies. This mechanism gradually amplifies the effects of liberalization, explaining the slow adjustment path of regional earnings and quantitatively accounting for the magnitude of the long-run effects.

Rothstein (2017) successfully replicates Chetty, Friedman, and Rockoff's (2014a, b)—henceforth, CFR's—results using data from North Carolina, but raises concerns about CFR's methods. We show that Rothstein's methodological critiques are invalid by presenting simulations and supplementary empirical evidence which show that (i) his preferred imputation of missing data generates bias; (ii) his “placebo test” rejects valid research designs; and (iii) his method of controlling for covariates yields inconsistent estimates of teachers' long-term effects. Consistent with the conclusions of Bacher-Hicks, Kane, and Staiger (2016) using data from Los Angeles, we conclude that Rothstein's replication study ultimately reinforces CFR's methods and results.

Oliver Hart delivered his Prize Lecture on 8 December 2016 at the Aula Magna, Stockholm University.

This paper shows how to identify and estimate, using standard semi-parametric techniques, a class of dynamic games with perfect monitoring, that have been at the frontier of recent research in political economy. The empirical analysis provides novel quantitative insights into the trade-off that voters face between ideology and ability, the differences in ability and ideology among parties and states, and the differences in preferences between political candidates and voters. We analyze the consequences of term limits and quantify their relative importance. Specifically, we characterize conditions under which term limits improve voters' welfare.

Contrary to standard search models predictions, past studies have not found a positive effect of unemployment insurance (UI) on reemployment wages. We estimate a positive UI wage effect exploiting an age-based regression discontinuity design in Austria. A search model incorporating duration dependence predicts two countervailing forces: UI induces workers to seek higher-wage jobs, but reduces wages by lengthening unemployment. Matching-function heterogeneity plausibly generates a negative relationship between the UI unemployment-duration and wage effects, which holds empirically in our sample and across studies, reconciling disparate wage-effect estimates. Empirically, UI raises wages by improving reemployment firm quality and attenuating wage drops.

This paper establishes a new fact about the compositional changes in the pool of unemployed over the US business cycle. Using microdata from the Current Population Survey for the years 1962–2012, it documents that in recessions the pool of unemployed shifts toward workers with high wages in their previous job and that these shifts are driven by the high cyclicality of separations for high-wage workers. The paper finds that standard theories of wage setting and unemployment have difficulty in explaining these patterns and evaluates a number of alternative theories that do better in accounting for the new fact.

A strategy is obviously dominant if, for any deviation, at any information set where both strategies first diverge, the best outcome under the deviation is no better than the worst outcome under the dominant strategy. A mechanism is obviously strategy-proof (OSP) if it has an equilibrium in obviously dominant strategies. This has a behavioral interpretation: a strategy is obviously dominant if and only if a cognitively limited agent can recognize it as weakly dominant. It also has a classical interpretation: a choice rule is OSP-implementable if and only if it can be carried out by a social planner under a particular regime of partial commitment.

We develop a version of Afriat's theorem that is applicable in a variety of choice environments beyond the setting of classical consumer theory. This allows us to devise tests for rationalizability in environments where the set of alternatives is not the positive orthant of a Euclidean space and where the rationalizing utility function is required to satisfy properties appropriate to that environment. We show that our results are applicable, amongst others, to choice data on lotteries, contingent consumption, and intertemporal consumption.

Chetty, Friedman, and Rockoff (2014a, b) study value-added (VA) measures of teacher effectiveness. CFR (2014a) exploits teacher switching as a quasi-experiment, concluding that student sorting creates negligible bias in VA scores. CFR (2014b) finds VA scores are useful proxies for teachers' effects on students' long-run outcomes. I successfully reproduce each in North Carolina data. But I find that the quasi-experiment is invalid, as teacher switching is correlated with changes in student preparedness. Adjusting for this, I find moderate bias in VA scores, perhaps 10–35 percent as large, in variance terms, as teachers' causal effects. Long-run results are sensitive to controls and cannot support strong conclusions.