(How) Do electoral surprises drive business cycles? Evidence from a new dataset

This paper documents that surprise election outcomes – measured as deviations between realised vote shares and expected vote shares based on a newly constructed dataset of opinion polls and party and candidate vote shares close to election day – are causing non-negligible short-term contractions in economic activity. We find that, on average, a percentage point higher surprise is associated with a 0.37 percentage point lower year-on-year growth rate one year after the election. These effects are only present in countries with strong democracies and seem to operate mainly through increased economic policy uncertainty and lower investment growth over a window of up to eight quarters after an election. In addition, surprise performances of left-wing political parties and in elections with transitions to left-wing governments (pre-defined from the Parlgov Database) are associated with the largest effects on the economy.