Private Choices in Public Health: Endogenous Behavioral Responses to Pandemic Risk

Epidemic control depends not only on policy mandates but also on endogenous behavioral responses--how individuals adjust private preventive actions to evolving risk. Using over 1.1 million county-day observations from 1,206 U.S. counties, we integrate high-frequency mobility data with local COVID-19 mortality to estimate the semi-elasticity of preventive behavior to mortality-driven risk perception. Fixed-effects regressions exploit within-county temporal variation, controlling for day-of-week and month effects, county-specific linear trends, and public health policies. Behavioral elasticity evolved as the pandemic progressed. Early in the pandemic, mortality-driven risk perception had only a limited effect on behavior (3.1% reduction in mobility per additional death per 10,000 residents, 7-day lag), while shelter-in-place mandates drove large declines (-10.16%). As uncertainty declined and individuals learned about transmission risks, behavioral responsiveness strengthened markedly, with mobility falling 7.8% per additional death (7-day lag per 10,000)--indicating that risk-calibrated adaptation supplanted broad precautionary withdrawal. By the post-vaccination stage, the mortality-mobility association was no longer significant, consistent with behavioral desensitization as perceived risk fell. To unpack the selective logic of prevention--when behavior most closely tracked perceived risk--we find that reductions were most pronounced in discretionary settings (-5.2% retail/recreation; -5.3% transit) but statistically insignificant in essential activities (-0.4% grocery/pharmacy) and low-risk outdoor spaces (-1.7% parks). Responsiveness weakened by roughly 0.5% per month, while shelter-in-place and mask mandates amplified mortality-driven behavioral responses by 2.8% and 4.7%. These findings highlight that voluntary, risk-responsive behavior--amplified by policy interventions--was pivotal to epidemic control.