Research

This paper models a frictional labor market where employers endogenously discriminate against the long term unemployed. The estimated model replicates recent experimental evidence which documents that interview invitations for observationally equivalent workers fall sharply as unemployment duration progresses. We use the model to quantitatively assess the consequences of such employer behavior for job finding rates and long term unemployment and find only modest effects given the large decline in callbacks. Interviews lost to duration impact individual job-finding rates solely if they would have led to jobs. We show that such instances are rare when firms discriminate in anticipation of an ultimately unsuccessful application. Discrimination in callbacks is thus largely a response to dynamic selection, with limited consequences for structural duration dependence and long term unemployment.

We develop a model where selection into marriage and household search generate a marital wage premium. Beyond selection, married individuals earn higher wages for two reasons. First, income pooling within a joint household raises risk-averse individuals’ reservation wages. Second, married individuals climb the job ladder faster, as they internalize that higher wages increase their partner’s selectivity over offers. Specialization according to comparative advantage in search generates a premium that increases in spousal education, as in the data. Quantitatively, household search explains 16-41% and 20-68% of the premium for males and females respectively, and accounts for its increase with spousal education.

This paper models a frictional labor market where employers endogenously discriminate against the long term unemployed. The estimated model replicates recent experimental evidence which documents that interview invitations for observationally equivalent workers fall sharply as unemployment duration progresses. We use the model to quantitatively assess the consequences of such employer behavior for job finding rates and long term unemployment and find only modest effects given the large decline in callbacks. Interviews lost to duration impact individual job-finding rates solely if they would have led to jobs. We show that such instances are rare when firms discriminate in anticipation of an ultimately unsuccessful application. Discrimination in callbacks is thus largely a response to dynamic selection, with limited consequences for structural duration dependence and long term unemployment.

What are the characteristics of workers in jobs likely to be initially affected by broad social distancing and later by narrower policy tailored to jobs with low risk of disease transmission? We use O*NET to construct a measure of the likelihood that jobs can be conducted from home (a variant of Dingel and Neiman (2020)) and a measure of low physical proximity to others at work. We validate the measures by showing how they relate to similar measures constructed using time use data from ATUS. Our main finding is that workers in low-work-from-home or high-physical-proximity jobs are more economically vulnerable across various measures constructed from the CPS and PSID: they are less educated, of lower income, have fewer liquid assets relative to income, and are more likely renters. We further substantiate the measures with behavior during the epidemic. First, we show that MSAs with less pre-virus employment in work-from-home jobs experienced smaller declines in the incidence of `staying-at-home', as measured using SafeGraph cell phone data. Second, we show that both occupations and types of workers predicted to be employed in low work-from-home jobs experienced greater declines in employment according to the March 2020 CPS. For example, non-college educated workers experienced a 4ppt larger decline in employment relative to those with a college degree.

This paper investigates gender differences in the job search process, both in the field and lab. First, we collect rich information on job offers and acceptances from undergraduates of Boston University’s Questrom School of Business. We document two novel empirical facts: (1) there is a clear gender difference in the timing of job offer acceptance, with women accepting jobs substantially earlier than men, and (2) the gender earnings gap in accepted offers narrows in favor of women over the course of the job search period. To rationalize these patterns, we develop a job search model that incorporates gender differences in risk aversion and overoptimism about prospective offers. We validate the model assumptions and predictions using the survey data, and present empirical evidence that the job search patterns in the field can be partly explained by greater risk aversion displayed by women and the higher levels of overoptimism (and slower belief updating) displayed by men. Next, we replicate the findings from the field in a specially-designed laboratory experiment that features sequential job search, and provide direct evidence on the purported mechanisms. Our findings highlight the importance of risk preferences and beliefs for gender differences in job-finding behavior, and consequently, early career wage gaps among the highly-skilled.

We propose a new, computationally efficient way to approximate the “grouped fixed-effects” (GFE) estimator of Bonhomme and Manresa (2015), which estimates grouped patterns of unobserved heterogeneity. To do so, we generalize the fuzzy C-means objective to regression settings. As the regularization parameter m approaches 1, the fuzzy clustering objective converges to the GFE objective; moreover, we recast this objective as a standard Generalized Method of Moments problem. We replicate the empirical results of Bonhomme and Manresa (2015) and show that our estimator delivers almost identical estimates. In simulations, we show that our approach delivers improvements in terms of bias, classification accuracy and computational speed.

We estimate the unconditional distribution of the marginal propensity to consume (MPC) using clustering regression and the 2008 stimulus payments. By deviating from the standard approach of estimating MPC heterogeneity using interactions with observables, we can recover the full distribution of MPCs. We find households spent between 4 and 133% of the rebate within a quarter, and individual households used rebates for different goods. While many observable characteristics correlate individually with our estimated MPCs, most of these relationships disappear when tested jointly. Notable exceptions are income and the average propensity to consume, which correlate positively with the MPC. Household observables explain only 8% of MPC variation, highlighting the role of latent heterogeneity.

Using detailed micro data, we document that households often use ``stimulus'' checks to pay down debt, especially those with low net wealth-to-income ratios. To rationalize these patterns, we introduce a borrowing price schedule into an otherwise standard incomplete markets model. Because interest rates rise with debt, borrowers have increasingly larger incentives to use an additional dollar to reduce debt service payments rather than consume. Using our calibrated model, we then study whether and how this marginal propensity to repay debt (MPRD) alters the aggregate implications of fiscal transfers. We uncover a trade-off between stimulus and insurance, as high--debt individuals gain considerably from transfers, but consume relatively little immediately. We show how this mechanism can lower short-run fiscal multipliers, but sustain aggregate consumption for longer.

How do inflation expectations affect the job search behavior of workers when wages are set in nominal terms? A canonical job search model incorporating nominal wage rigidities implies that on-the-job search should increase and reservation wages should decrease with expected inflation.  Higher inflation expectations therefore lead to more frequent job-to-job transitions. We show in a novel survey that workers search more under higher values of hypothetical inflation. In the Survey of Consumer Expectations, workers with higher inflation expectations have lower reservation wages and are more likely to search and  to change jobs. The relationship between expected inflation and employer-to-employer transitions also appears in aggregate time series data.

We document the extent to which workers in AI-exposed occupations can successfully retrain for AI-intensive work. We assemble a new workforce development dataset spanning over 1.6 million job training participation spells from all US Workforce Investment and Opportunity Act programs from 2012 – 2023 linked with occupational measures of AI exposure. Using earnings records observed before and after training, we compare high AI exposure trainees to a matched sample of similar workers who only received job search assistance. We find that the average earnings return to training among AI-exposed workers is high, around $1,470 per quarter. Low-exposure trainees capture higher returns, and trainees who target AI-intensive work face a 29% earnings return penalty relative to their high exposure peers who pursue more general training. We estimate that between 25% to 40% of occupations are ``AI retrainable'' as measured by its workers receiving higher pay for moving to more AI-intensive occupations---a large magnitude given the relatively low-income sample of displaced workers. Positive earnings returns in all groups are driven by the most recent years when labor markets were tightest, suggesting training programs may have stronger signal value when firms reach deeper into the skill market.

What are the costs of inflation in the labor market? When wages are set nominally, inflation leads to reduced purchasing power, which prompts workers to search for other jobs to regain it; this search is costly.  We quantify these costs in a model of on-the-job search with nominal rigidities, where search effort responds endogenously to inflation.  Relative to a flexible wage counterfactual, inflation erodes the value of a match to a worker through real wage losses and larger search costs. The real wage loss is absorbed as a benefit to firms, whereas the cost of search is a net aggregate cost of inflation.

This paper investigates self-reported wedges between how much people work and how much they want to work, at their current wage. More than two-thirds of full-time workers in German survey data are overworked---usual hours exceed desired hours. We combine this evidence with a simple model of labor supply to assess the welfare consequences of tighter weekly hours limits via willingness-to-pay calculations. According to counterfactuals, the optimal length of the workweek in Germany is 37 hours. Introducing such a cap would raise welfare by .5-1% of GDP. The gains from a shortened workweek are largest for workers who are married, female, college-educated, white-collar, middle aged, and high income. An extended analysis integrates a non-constant wage-hours relationship, falling capital returns, and a shrinking tax base.

We develop a model of educational investment, marriage, and household labor market search to quantify how changes in incentives to positively sort in marriage - summarized by changes in marital surplus - contributed to the rise in U.S. household income inequality. While there are always positive incentives to sort by skill and education, the former strengthened relative to the latter over time. These changes incentivized further educational attainment, especially for the high skilled. Unlike findings from previous studies, the resulting increase in like-education-skill marriages contributed to a significant rise in income equality.

We investigate the role of information frictions in the US labor market using a new nationally representative panel dataset on individuals' labor market expectations and realizations. We find that expectations about future job offers are, on average, highly predictive of actual outcomes. Despite their predictive power, however, deviations of ex post realizations from ex ante expectations are often sizable. The panel aspect of the data allows us to study how individuals update their labor market expectations in response to such shocks. We find a strong response: an individual who receives a job offer one dollar above her expectation subsequently adjusts her expectations upward by $0.47. We embed the empirical evidence on expectations and learning into a model of search on- and off- the job with learning, and show that it is far better able to fit the data on reservation wages relative to a model that assumes complete information. We use the framework to gauge the welfare costs of information frictions which arise because individuals make uninformed job acceptance decisions and find that the costs due to information frictions are sizable, but mitigated by the presence of learning.

This paper develops a multisector search model in which workers choose their sectors in response to sector-specific and idiosyncratic shocks. Unemployed workers can search for work in their sector of last employment, or become ``move unemployed,'' spending extra time in unemployment to reach a new sector. Sectoral shocks induce net movements of labor into relatively productive sectors, while idiosyncratic worker shocks ensure that gross intersectoral flows through unemployment are always positive, a prediction consistent with the data. I use the model to test the sectoral shifts hypothesis (Lilien (1982)) in the context of the Great Recession in which the construction sector experienced a relatively large shock. In a two-sector calibration of the model to construction and non-construction, I find that sectoral dispersion shocks have no impact on aggregate unemployment. While they increase net mobility, this increase is accomplished through a change in the composition of gross flows rather than their level. The results suggest that, a priori, we should not expect sectoral shocks to generate unemployment fluctuations.