Haiming (Richard) Li

Cross-Border Mental Health Effects of Cannabis Legalization Policy: Evidence of Spillovers from New Jersey to Pennsylvania

Does recreational cannabis legalization in one state affect mental health in a neighboring state that has not legalized? I study New Jersey's April 2022 dispensary openings and their consequences for Pennsylvania, combining ZIP code–quarter inpatient discharge records from the Pennsylvania Health Care Cost Containment Council for 2018–2022 with driving times from each Pennsylvania ZIP code to the nearest New Jersey dispensary. Border-area ZIP codes within 120 minutes of a dispensary are matched on observables to distant ZIP codes beyond 180 minutes and compared in a difference-in-differences design, with a continuous-treatment specification in driving time as a companion. Psychiatric admissions in treated ZIP codes rise by 1.06 per 1,000 residents per quarter, roughly 8 percent relative to controls. The increase is concentrated in depression (14.4 percent), alcohol- and tobacco-related disorders (7.2 percent), and organic mental disorders (8.9 percent), while cannabis-specific and psychosis diagnoses are unchanged — a pattern pointing to mood dysregulation and substance co-use rather than acute intoxication. Cost–benefit assessments that omit these cross-border externalities understate the costs borne by non-legalizing neighbors.

Draft available on request.

Fueling Out of Work: Vehicle Market Restructuring and Labor Supply among Car-Dependent Workers

with Yizhao Wang · Under review

The 2011 reform of Corporate Average Fuel Economy standards set footprint-based targets that were less stringent for larger vehicles, and U.S. production shifted away from compact cars toward SUVs and light trucks, thinning the supply of entry-level vehicles. Using CPS ASEC data for 2001–2019 and a rural–urban difference-in-differences design, we find that rural labor force participation and employment fell by about 0.9 percentage points after the reform, with larger declines among low- and middle-income adults, while usual hours rose among those who stayed employed. The pattern is consistent with a fixed-cost channel: where a private vehicle is the only way to reach a job, restructuring the vehicle market shows up in labor supply.

Policy Spillovers Across Substances: E-Cigarette Regulation, Cannabis Legalization, and Health Outcomes

with Yizhao Wang

The Causal Effect of Poor Oral Health on Heart Disease: A GWAS-by-Subtraction Genomic Structural Equation Instrumental Variables Model

with Chad D. Meyerhoefer, Erin Ware, John Moeller, Haiyan Chen, Jessica Faul, and Richard Manski

We estimate the causal effect of total tooth loss on heart disease using genetic and survey data from the Health and Retirement Study, 2006–2020. A GWAS-by-subtraction procedure constructs a polygenic score instrument purged of variants with direct effects on heart disease. The instrumental variables estimates are substantially larger than the corresponding OLS associations and stronger for men, which implies that cost projections for a Medicare dental benefit are overstated when they ignore offsetting reductions in heart disease treatment.

Supported by NIH/NIDCR R03DE034043