with André Reslow.
This paper investigates whether governments release over-optimistic forecasts for GDP growth approaching elections. High-frequency data at the forecaster level from the United States, Sweden, and the United Kingdom document that governments overestimate short-term GDP growth by 0.16–0.31 percentage points in the months approaching a vote. Consistent with a model of political selection, in which the incumbent government releases optimistic forecasts to signal high ability in order to increase re-election probability, we find that the bias is larger when the incumbent government is not term-limited or constrained by a parliament led by the opposition. Moreover, the bias is allocated across different forecast horizons based on election seasonality. We also show that a reform that outsourced the main forecasting function from the HM Treasury in the United Kingdom to a newly formed government agency reduced the overall bias of forecasts, but not the electoral cyclicality.