Demonstrated risk preferences and COVID-19 regulations in the United States

Abstract

Research has pointed towards U.S. state officials setting COVID regulations based on their constituents’ political affiliation. But a further explanation is needed as prior to 2020, U.S. voters did not choose their political party in accord with how they thought politicians would act in a pandemic. In contrast, other papers have found that people with higher risk preferences took fewer mitigating actions during COVID. Building on these results and the public choice view that political markets lack a dynamic-feedback process, this paper hypothesizes that upcoming elections incentivized state officials to partially set regulations in congruence with their constituents’ demonstrated risk preferences. The hypothesis is tested with a balanced panel of all U.S. states over seven time periods ranging from April until shortly before the 2020 election. A log-linear hybrid model finds a negative relationship between risky actions and the stringency of COVID regulations at the between-state level. The relationship is statistically and regulatorily significant while controlling for relevant time-varying and time-invariant health, political, and economic measures. Multiple robustness tests confirm these results, including instrumenting people’s risky actions. At the within-state level, regulations only varied with changes in revealed risk preferences when governors faced impending feedback from a reelection contest. Republican governors running for reelection decreased regulations when revealed risk taking increased whereas their Democratic counterparts responded by increasing regulations. In states without a gubernatorial election, regulations show little responsiveness to changes in risk taking, corroborating the public choice viewpoint.