Risk Perception Through the Lens of Politics in the Time of the COVID-19 Pandemic

Even when-objectively speaking-death is on the line, partisan bias still colors beliefs about facts. We use data on internet searches as well as proprietary data on county-level average daily travel distance and visits to non-essential businesses from a large sample of U.S. smartphones at the daily level and show that the higher the percentage of Trump voters in a county, holding all else equal, the lower the perception of risk associated with the COVID-19 virus and the lower the level of social distancing behavior exhibited. As Trump vote share rises, individuals search less for information on the virus, they search less for information about unemployment benefits, and they exhibit lower reductions in both their daily distance traveled and their visits to non-essential businesses. Risk perceptions in areas with high Trump vote shares increase in these areas only after 3/9/20, when it was announced that COVID-19 had struck the Conservative Political Action Committee meetings and conservative politicians were self-quarantined, suggesting that their risk perceptions are affected not by changes in fundamental underlying risk, but rather by political-related interpretations of the risk. These patterns persist even in the face of state-level mandates to close schools and non-essential businesses and to "stay home-work safe," and reverse only when the White House releases federal social distancing guidelines on March 16th. This differential is present even in the face of similar levels of ability to telework and in the presence of higher levels of older population at risk. Our results suggest that political partisanship may play a role in determining risk perceptions in a pandemic, with potentially significant externalities for public health outcomes. Relying solely on compliance with voluntary suggested measures in the presence of different political views on the crisis may have limited effectiveness; instead, enforcement may be required to successfully flatten the curve.