Job Market Paper
Temporary Shocks, Permanent Impacts: The Effects of Liquidity on Job Search and Reemployment
(with Nick Flamang)
[Click to View Abstract | Draft]
How do differences in job seekers' liquidity during unemployment affect their reemployment and long-term earnings trajectories? We provide new evidence by examining delays in unemployment insurance (UI) benefit payments, which create high-frequency variation in claimants' cash-on-hand: delays shift the timing of benefits, but not the total amount of benefits received. We utilize administrative delays resulting from a 2013 California system glitch that affected UI benefit payment timing, but only for a subset of active claimants. Our research design compares claimants plausibly randomly delayed during the outage to unaffected claimants after matching on a rich set of demographics, earnings histories, and pre-outage claim histories to minimize pre-treatment differences. The mean delayed claimant had $825 in UI benefits delayed (2.6 weeks of benefits) during the glitch, and waited an average of 34 days before being fully compensated. In the short run, claimants with delayed payments exit UI earlier, are employed faster, and find better reemployment firm matches. Moreover, these effects are highly persistent over time: even five years after the system glitch, delayed claimants have higher employment and earnings. We find novel evidence that these effects are largest among claimants affected early into their unemployment spell. Taken together, these results are consistent with a model of job search incorporating duration dependence, where UI induces longer unemployment spells that reduce reemployment rates and future wages.
Other Working Papers
Understanding Capital Gains Responses to Taxes using Transaction-Level Data
(with Alisa Tazhitdinova)
[Click to View Abstract | Draft]
We study how individuals’ trading behavior responds to tax incentives using administrative transaction-level data on all taxable sales of broker-traded, directly held financial assets between 2011 and 2019. Our empirical design leverages a simple, salient, timing-based tax notch: in the U.S., assets held beyond one year qualify for a 10-20 percentage point reduction in capital gains rates. The size and granularity of the data allow us to study how this capital gains tax rate differentiation shapes individuals’ trading behaviors across narrowly defined demographic and income groups. We find that: (1) retiming responses around the tax rate notch are weak in general; (2) individuals make clear misoptimization errors by realizing gains just before the notch; and (3) this pattern can be explained by both heterogeneous capital gains responses by asset type combined with rigidities in individual trading styles. Finally, we use our empirical results to show theoretically that the weak deferral elasticities imply that a revenue-maximizing government would eliminate short- vs long-term tax differentiation.
Unemployment Insurance as a Financial Stabilizer: Evidence from Large Benefit Expansions
(with Nick Flamang)
[Click to View Abstract | Draft]
To what extent does unemployment insurance (UI) attenuate aggregate financial responses to unemployment shocks? We answer this question using administrative credit bureau records and the unprecedented changes in unemployment and UI generosity during the Covid-19 pandemic. We first find that aggregate sensitivity to the unemployment rate decreased by 50% for auto loans and 66% for credit cards between January 2017 and March 2021. To isolate the effect of UI from other contemporaneous policies shifting unemployment shock responsiveness, we employ a staggered event study design around state-level withdrawals from federal UI programs in late 2021. We find that almost all of the pandemic sensitivity drop is attributable to UI expansions. Our two designs are qualitatively robust to placebo tests on plausibly unaffected credit types, potential demand-side responses for increased credit, and alternate estimation specifications. In a back-of-the-envelope calculation, we calculate that UI expansions prevented about 59% of total potential delinquency-months. Taken together, these results imply that federal UI expansions have had a substantially stabilizing effect during the Covid-19 pandemic. Our findings thus provide powerful empirical support for a largely theoretical body of research on the role of UI as an automatic stabilizer of aggregate economic conditions.
Selected Work in Progress
The Scope, Causes, and Consequences of Worker Misclassification: Evidence from Randomized Tax Audits
(with David Coyne and Ithai Lurie)
Approved project, US Treasury Office of Tax Analysis
[Click to View Abstract]
Worker misclassification—wherein firms erroneously represent their workers’ employment status for tax purposes, typically listing wage employees as independent contractors—is a large and growing problem in employment tax compliance, with meaningful costs to both individual workers and the tax system. Misclassified workers are thought to have lower earnings, rates of health insurance coverage, and retirement contributions (Jackson et al, 2017). At the same time, only about 55% of independent contractors correctly remit self-employment taxes, which meaningfully contributes to the employment tax gap. The problem is prevalent and growing: in 2000, about 30% of audited firms misclassified at least one worker, and the fraction of misclassified workers has increased by about 300% between 2000 and 2019 (GAO 2009; ETA Reports 581). Using randomized audit data from the National Research Program’s Employment Tax Study (ETS), we examine four main questions: (1) how often do audits uncover worker misclassification, and how does this vary by firm or worker characteristics; (2) what is the short-term compliance effect of audits on affected workers’ classification; (3) do the audits have long-term impacts on worker classification and outcomes; and (4) does the risk of being audited impact the behavior of non-audited firms?
Understanding Existing Insurance from Job Loss in the United States
(with Carl McPherson and Matt Unrath)
Approved project, US Census Bureau
[Click to View Abstract]
We study the joint effects of the tax system, government transfers, and self-insurance in mitigating earnings losses from job loss. Transfer payments imply smaller income losses relative to wage losses; behavioral responses (asset liquidation, self-employment, income misreporting, family members’ labor supply) mitigate household income losses relative to worker income losses; and progressive tax rates imply smaller after-tax income losses relative to pre-tax income losses. We leverage a novel merge between social safety net benefit takeup in 42 US states to population tax returns and earnings data to estimate these various margins of response in a single dataset. In doing so, we will be able to integrate the tax system, transfer payments, and household responses into a net measure of household and government insurance against job loss.