Cognitive Dissonance in the Use of Scientific Estimates: Evidence from COVID-19 and Implications for Climate Policy

Abstract

Objective information from credible sources is critical for efficient policy that involves uncertain outcomes. This paper presents evidence that scientific projections are under-utilized in decisions involving dynamic risks. We leverage the COVID-19 pandemic, using plausibly exogenous updates to the dominant model on death projections and show revealed mitigation actions were largely driven by contemporary outcomes over scientific predictions. Further, we document behavior consistent with cognitive dissonance – agents favor scientific forecasts when they predict more optimistic outcomes. When taken to the context of climate policy, we demonstrate that these estimates would imply an undervaluing of carbon costs of 50 percent.