Voting and information : evidence from a field experiment

Biased beliefs affect real-world decisions, including political solutions to societal challenges. One crucial example is environmental policy: people tend to underestimate the incentive effect of Pigouvian policies. Addressing biased beliefs at scale is then paramount. In the days leading up to a ballot initiative in Washington state, we implemented a large-scale field experiment providing information on carbon taxes to over 285,000 individuals. We complemented it with a survey experiment of about 1,000 individuals, with the same treatments as in the field experiment, shedding light on social desirability bias and mechanisms around belief revision. Using data at the voting precinct level, we show that our intervention increases revealed support for carbon taxes, mainly for a treatment centered around earmarking of tax revenue, which was one of the design features of the ballot initiative. We find the effect to be stronger in precincts relatively opposed to the initiative, and less exposed to media coverage of carbon taxes, and more exposed to coverage challenging their effectiveness.