Knowledge that Transforms

The primary motivation behind quantitative work in international trade and many other fields is to shed light on the economic consequences of policy changes and other shocks. To help assess and potentially strengthen the credibility of such quantitative predictions, we introduce an IV-based goodness-of-fit measure that provides the basis for testing causal predictions in arbitrary general equilibrium environments as well as for estimating the average misspecification in these predictions. As an illustration of how to use the measure in practice, we revisit the welfare consequences of the U.S.-China trade war predicted by Fajgelbaum et al. (2020).

We assess the impact of discrimination on Black individuals’ job networks across the United States using a two-stage field experiment with 400+ fictitious LinkedIn profiles. In the first stage, we vary race via AI-generated images only and find that Black profiles’ connection requests are 13% less likely to be accepted. Based on users’ CVs, we find widespread discrimination across social groups. In the second stage, we exogenously endow Black and white profiles with the same networks and ask connected users for career advice. We find no evidence of direct discrimination in information provision. However, when taking into account differences in the composition and size of networks, Black profiles receive substantially fewer replies. Our findings suggest that gatekeeping is a key driver of Black–white disparities.

This article studies the unexpected retraction of a U.S. federal policy in 2016 that would have more than doubled the “overtime exemption threshold” from $455 to $913 per week and thereby grant overtime protection to an additional 20% of salaried workers. Although the policy was blocked by a federal court injunction a week before it was supposed to take effect, I show that it nevertheless had a persistent positive impact on workers’ earnings. Leveraging a bunching design with administrative payroll data from ADP, I find that employers raised workers’ salaries to the $913 threshold even after the policy was repealed. Over the next 18 months, difference-in-difference estimates reveal that employers did not slow the wage growth of workers affected by the policy relative to those already earning above $913 per week, nor did they hire new employees at a lower pay rate. Real wages remained persistently elevated relative to what they would have been absent the policy and separation rates decreased among workers bunched at the $913 threshold. Comparing highly exposed firms to unaffected firms, I find an increase in employers’ wage bills but no change in aggregate employment. Taken together, the results indicate that temporary policies impacting wage levels can have permanent effects on the labor market. Survey responses collected by the Department of Labor suggest that morale concerns play a key role in driving the wage hysteresis.

We study the staggered introduction of a generative AI–based conversational assistant using data from 5,172 customer-support agents. Access to AI assistance increases worker productivity, as measured by issues resolved per hour, by 15% on average, with substantial heterogeneity across workers. The effects vary significantly across different agents. Less experienced and lower-skilled workers improve both the speed and quality of their output, while the most experienced and highest-skilled workers see small gains in speed and small declines in quality. We also find evidence that AI assistance facilitates worker learning and improves English fluency, particularly among international agents. While AI systems improve with more training data, we find that the gains from AI adoption are largest for moderately rare problems, where human agents have less baseline experience but the system still has adequate training data. Finally, we provide evidence that AI assistance improves the experience of work along several dimensions: customers are more polite and less likely to ask to speak to a manager.

We study optimal nonlinear taxation of single and married households. Taxes on couples depend on the earnings of both spouses and are an example of multidimensional tax schedules. We develop novel analytical techniques to study properties of such taxes. We show that the optimal marginal taxes for married individuals are generally lower than for single individuals because resource sharing in couples provides socially valuable redistribution. Under realistic assumptions, the optimal tax rates for married individuals increase with the correlation of spousal earnings, the marginal tax rates for one spouse increase (decrease) in the earnings of the other if both spouses have low (high) earnings, and the primary earner faces lower marginal taxes than the secondary earner.

What causes adverse policing outcomes, such as excessive uses of force and unnecessary arrests? Prevailing explanations focus on problematic officers or deficient regulations and oversight. We introduce an overlooked perspective. We suggest that the cognitive demands inherent in policing can undermine officer decision making. Unless officers are prepared for these demands, they may jump to conclusions too quickly without fully considering alternative ways of seeing a situation. This can lead to adverse policing outcomes. To test this perspective, we created a new training that teaches officers to consider different ways of interpreting the situations they encounter. We evaluated this training using a randomized controlled trial with 2,070 officers from the Chicago Police Department. In a series of lab assessments, we find that treated officers were significantly more likely to consider a wider range of evidence and develop more explanations for subjects’ actions. Critically, we also find that training affected officer performance in the field, leading to reductions in uses of force, discretionary arrests, and arrests of Black civilians. Meanwhile, officer activity levels remained unchanged, and trained officers were less likely to be injured on duty. Our results highlight the value of considering the cognitive aspects of policing and demonstrate the power of using behaviorally informed approaches to improve officer decision making and policing outcomes.

Moral foundations theory is an influential empirical description of moral perception. According to this theory, individuals make moral judgments based on five distinct “moral foundations”: care, fairness, loyalty, authority, and sanctity. We provide a theory that explores the claimed evolutionary basis for these moral foundations. The theory conceptualizes these five moral foundations as specific modifications of fitness payoffs in a 2 × 2 game. We find that the five foundations are distinguishable from each other and evolutionarily stable. However, they are not a minimal set: strict subsets of the foundations suffice to describe all preferences that are evolutionarily stable. Not all evolutionarily stable foundations deliver social fitness improvement over the Nash equilibrium in the fitness game: we characterize which do. Finally, we study moral overdrive, that is, the situation in which the moral component of preferences totally dominates fitness payoffs and drives decision making entirely. While every one of the five foundations is compatible with moral overdrive in at least one fitness game, there is no fitness game in which moral overdrive is compatible with social fitness improvement. These results are partially extended to n × n games. We derive two testable implications from the theory and find empirical support for them.

We develop a tractable quantitative framework for modeling the rich patterns of spatial mobility observed in smartphone data. We show that travel is frequently undertaken as part of a travel itinerary, defined as a journey starting and ending at home that can include more than one intermediate stop on a given day. We show that these travel itineraries provide microfoundations for consumption externalities and generate both complementarity and substitutability between locations. We show that the consumption externalities implied by travel itineraries are central to matching quasi-experimental evidence from the shift to working from home. We find that these consumption externalities are key drivers of the agglomeration of economic activity in central cities and shape the relative welfare gains from alternative transport improvements in favor of investments in central cities.

When people receive new information, sometimes they revise their beliefs too much, and sometimes too little. We show that a key driver of whether people overinfer or underinfer is the strength of the information. Based on a model in which people know which direction to update in, but not exactly how much to update, we hypothesize that people will overinfer from weak signals and underinfer from strong signals. We then test this hypothesis across four different environments: abstract experiments, a naturalistic experiment, sports betting markets, and financial markets. In each environment, our consistent and robust finding is overinference from weak signals and underinference from strong signals. Our framework and findings can help harmonize apparently contradictory results from the experimental and empirical literatures.

By augmenting the continuous-time specification of Harris and Laibson (2013) with the assumption that hard borrowing constraints do not bind in equilibrium, present bias can be tractably incorporated into rich consumption-saving models featuring stochastic income, risky and illiquid assets, and costly borrowing. I present closed-form expressions characterizing how present bias affects consumption, illiquid asset demand, and welfare. This welfare analysis specifies the channels through which present bias can matter for policy and uncovers the present-bias dilemma: present bias can have large welfare costs, but individuals have little ability to alleviate these costs using financial commitment devices like illiquid assets.