Are Workers Less Likely to Report Sexual Harassment When Unemployment and Retaliation Risks Are High?

Dahl and Knepper (2025) examine why workplace sexual harassment is frequently underreported and whether fear of employer retaliation plays a central role. They use administrative data on sexual harassment charges filed with the Equal Employment Opportunity Commission from 1995 to 2016, combined with county-level unemployment data and major reductions in unemployment insurance benefits, especially in North Carolina. They find that weaker labor market conditions discourage reporting. A one-percentage-point increase in local unemployment raises the likelihood that a filed charge is judged meritorious by about 0.3 percent, suggesting greater underreporting. Sharp cuts to unemployment insurance increased the selectivity of reported cases by over 30 percent, showing that reduced economic security silences victims.

Why This Article Was Selected for The Policy Scientist

This article addresses a policy issue of broad and persistent importance: how labor market conditions and economic insecurity influence workers’ willingness to report sexual harassment. The paper is timely given recent labor market volatility and sustained attention to retaliation and whistleblowing. Dahl and Knepper have written extensively in this area, and this study builds on prior research linking reporting behavior to economic incentives. The EEOC administrative data are unusually rich, and the quasi-experimental use of unemployment shocks and unemployment insurance cuts supports causal interpretation.

Full Citation and Link to Article

Dahl, G. B., & Knepper, M. M. (Forthcoming 2026). Why is workplace sexual harassment underreported? The value of outside options amid the threat of retaliation. American Economic Review. https://doi.org/10.1257/aer.20221703 

Central Research Question

The central research question of this article is why workplace sexual harassment is so extensively underreported, despite strong legal prohibitions and substantial evidence that harassment remains widespread. Dahl and Knepper focus specifically on whether fear of employer retaliation—particularly the risk of job loss—discourages victims from reporting harassment when their outside labor market options are weak. Rather than treating underreporting as a fixed behavioral trait, the authors ask whether reporting decisions respond systematically to economic conditions that alter the costs of retaliation. Their core hypothesis is that declines in outside options raise the threshold of harassment severity required for victims to come forward, leading to increased underreporting even if the underlying prevalence of harassment remains constant or rises.

Previous Literature

The paper builds on several strands of prior research. First, it draws on a large interdisciplinary literature documenting the prevalence and consequences of workplace sexual harassment, including evidence from surveys showing substantial effects on mental health, job satisfaction, productivity, and career trajectories. While this literature establishes that harassment is common and harmful, it has been less successful in explaining why formal complaints are so rare relative to self-reported victimization.

Second, the authors engage with work on retaliation and whistleblowing, which emphasizes that employer reprisals can deter reporting of misconduct. Earlier studies rely primarily on surveys or qualitative evidence to document fear of retaliation, but lack credible strategies to measure how reporting behavior changes when retaliation becomes more or less costly. Related economic research on workplace injuries and regulatory complaints shows that workers are less likely to report violations when unemployment is high, suggesting that labor market power affects compliance and enforcement.

Finally, the paper contributes to the growing economics literature inspired by the #MeToo movement, including theoretical and empirical work on reporting incentives, workplace gender composition, and institutional responses to harassment. Compared to this literature, Dahl and Knepper’s contribution is to provide a formal test of retaliation-based underreporting using labor market shocks and policy changes, rather than relying solely on self-reported attitudes or cross-sectional correlations.

Data

The analysis relies primarily on confidential administrative data from the Equal Employment Opportunity Commission covering all sexual harassment charges filed between 1995 and 2016. These data include detailed information on the nature of each charge, demographic characteristics of the complainant, whether retaliation occurred, and whether the EEOC determined the charge to have merit. The merit determination, which reflects either a settlement or a finding of reasonable cause, is used as a proxy for the severity and selectivity of reported cases.

The EEOC data are merged with county-level unemployment data from the Bureau of Labor Statistics to measure local labor market conditions at a monthly frequency. The authors also link a subset of charges to EEO-1 establishment-level data, which provide information on the gender composition of workers and managers. This linkage allows them to explore how power dynamics within firms interact with economic conditions to shape reporting behavior.

In addition, the paper exploits detailed data on unemployment insurance policy changes, focusing especially on North Carolina’s 2013 reform, which sharply reduced benefit levels and durations. These data are combined with Current Population Survey and Department of Labor statistics to measure changes in unemployment insurance recipiency following policy changes.

Methods

The authors employ a quasi-experimental research design grounded in a simple threshold model of reporting behavior. The model predicts that when the expected cost of retaliation rises—due to higher unemployment or weaker unemployment insurance—workers will report only more severe cases of harassment, increasing the average quality of reported charges.

Empirically, the main strategy is to examine how the probability that a sexual harassment charge is deemed meritorious varies with local labor market conditions. County and time fixed effects are used to absorb persistent differences across locations and national trends in enforcement or reporting. By focusing on within-county variation over time, the authors isolate the effect of changes in outside options on reporting selectivity.

To strengthen causal interpretation, the paper also analyzes an exogenous policy shock: North Carolina’s large reduction in unemployment insurance generosity. A difference-in-differences design compares changes in the selectivity of sexual harassment charges in North Carolina to changes in similar Southern states that did not alter their unemployment insurance systems. Event-study analyses are used to assess pre-trends and dynamic effects. Instrumental variables estimates further scale the effects by changes in unemployment insurance recipiency.

While the study does not use randomized controlled trials, it relies on well-established causal inference techniques, including fixed effects, plausibly exogenous policy variation, and robustness checks using alternative labor market definitions.

Findings/Size Effects

The authors find strong and consistent evidence that underreporting of sexual harassment increases when outside options deteriorate. A one percentage point increase in the local unemployment rate raises the probability that a filed sexual harassment charge has merit by approximately 0.3 percent. This increase in selectivity implies that victims are less likely to report marginal or ambiguous cases when labor market conditions worsen.

Importantly, this pattern is unique to sexual harassment. The selectivity of other discrimination charges shows little or no response to unemployment, and other forms of harassment exhibit smaller effects. This differential response is consistent with sexual harassment cases involving higher risks of retaliation and greater power asymmetries.

The unemployment insurance results are even more striking. Following North Carolina’s 2013 reforms, the share of sexual harassment charges deemed meritorious rose by roughly 7 percentage points, representing an increase of more than 30 percent relative to the pre-reform mean. Instrumental variables estimates indicate that each one percentage point increase in unemployment insurance recipiency reduces charge selectivity by about 0.34 percentage points.

The authors also find that underreporting effects are amplified in male-dominated workplaces and in establishments where men disproportionately occupy managerial roles. These findings reinforce the interpretation that power disparities and retaliation risk play a central role in suppressing reporting.

Conclusion

This paper provides compelling evidence that underreporting of workplace sexual harassment is not merely a function of social stigma or individual preferences, but responds systematically to economic incentives and institutional protections. By showing that reporting behavior becomes more selective when outside options weaken, the authors demonstrate that enforcement data can substantially understate true harassment prevalence during economic downturns.

The study makes an important contribution by combining rich administrative data with credible quasi-experimental variation, advancing the literature beyond descriptive accounts of underreporting. While the absence of randomized interventions limits definitive causal claims, the identification strategies are strong and transparent. The findings are plausibly generalizable to other jurisdictions with similar labor market institutions and enforcement regimes.

Overall, the paper reframes sexual harassment underreporting as a labor market and policy problem, highlighting the role of economic security in enabling legal enforcement.