Working Progress

• Sounding Out Sleep: Can Audio Messages Shift Children's Bedtime Norms?
  with Osea Giuntella, Rania Gihleb and Wendy Troxell 

Racial disparities in children’s sleep can shape health and human capital trajectories. Using U.S. time-use data and a randomized online experiment with 729 mothers, we show that Black/African-American children go to bed later and exhibit stronger parent-child bedtime correlations than White children, with socioeconomic factors only partially explaining these gaps. In the experiment, culturally informed audio messages varied in framing and narrator ethnicity. Personal narrative messages most strongly increased Black/African-American mothers’ intentions to advance bedtime (+16 pp) and extend sleep duration (+22 pp), while disparity-focused messages mainly increased newsletter engagement (+12–19 pp). Narrator identity had negligible effects, highlighting the importance of content over messenger. Our findings suggest low-cost, culturally targeted interventions can reduce sleep disparities and improve child development, emphasizing the need to consider both intentions and actual behavioral responses.

• Early unemployment and long-term effects on health and risky health behaviors
  with Francesco Principe

This paper examines the long-term effects of youth unemployment (ages 18--25) on adult health behaviors, chronic conditions, and occupational outcomes. Using harmonized SHARE panel data linked to country-specific unemployment rates, we analyze how early labor-market shocks influence smoking, alcohol use, diet, physical activity, chronic disease, and job characteristics. We find that higher youth unemployment raises the probability of daily smoking by 3.25\% and prolongs smoking duration, while reducing heavy drinking by 4.7\%, vigorous physical activity by 1\%, and frequent fruit and vegetable consumption by 0.5\%. Cumulative exposure also increases the likelihood of working in physically demanding jobs and lowers adult income. Effects are strongest among lower-educated individuals, with additional persistence among high-income adults and slightly higher susceptibility among women. These results highlight that early economic adversity shapes adult health and labor outcomes through stress, behavioral, and occupational pathways, with lasting implications for health inequality and long-term economic well-being.

• The Impact of Cannabis Legalization on Sleep 
  with Osea Giuntella and Drake Colman 

This paper provides the first causal evidence on how medical and recreational cannabis legalization (MCL and RCL) affect adult sleep duration in the United States. Leveraging the staggered adoption of cannabis policies across states from 2003 to 2023, we estimate their impacts using a two-way fixed effects Difference-in-Differences framework complemented by the Sun and Abraham (2021) estimator to address treatment-effect heterogeneity. Our primary outcome is the probability of sleeping at least seven hours per night. We find that MCL increases adequate sleep among adults aged 45--60 by approximately 2.5 percentage points (a 4.2\% relative increase). In contrast, RCL decreases adequate sleep among adults aged 30--44 and 45--60 by 2--3 percentage points (roughly a 5\% relative decline), driven primarily by workers facing high psychological demands on the job. These findings demonstrate that cannabis policy effects are highly heterogeneous by policy type and age group, highlighting that medical—but not recreational—legalization can improve sleep duration for specific working-age populations. Our results have important implications for public health, labor productivity, and the design of cannabis legislation.

• AI as Manager and Job Preferences: A Field Experiment
  with Alex Chan and Aspire Institute

This paper examines job seekers’ preferences over job offers that vary in manager characteristics and in the use of algorithmic personnel decision support. In a field experiment, actual job seekers repeatedly choose between job offers that differ in the following attributes: type of work, whether performance evaluations are supported by AI, and wages. We assess how these attributes shape the evaluation of job opportunities. We document the extent to which job seekers are willing to trade off wages to work with gender-concordant managers and in environments without algorithmic oversight. Finally, we examine heterogeneity in preferences by job seeker gender and across traditionally male- and female-dominated sectors.