“Biased Forecasts and Voting: The Brexit Referendum Case”

with André Reslow.


This paper explores whether macroeconomic forecasters use their forecasts to influence voting outcomes. We develop a theoretical model in which voters have to choose between remaining in a status quo or leaving it for an alternative. Voters rely on a macroeconomic forecaster that has preferences over a referendum outcome to form beliefs about the future states of the economy. The model yields four main predictions. First, the forecaster releases biased estimates approaching the vote. Second, the forecaster releases biased estimates also after the vote to preserve the credibility of its estimates. Third, the bias in forecasts subject to the alternative state dominates the bias in forecasts subject to the status quo. Fourth, an increase in the forecaster’s stakes in the voting outcome leads to more biased estimates. All the theoretical predictions find empirical support in the context of the Brexit referendum.

Link: Latest version (September 2021); Sveriges Riksbank WP (March 2019); Uppsala University WP (March 2019)

Media: The Telegraph (Full article here); RT UK News (video); Citywire Financial Publishers; Brexit Central