This paper documents the existence of electoral cycles in GDP growth forecasts released by governments. In a theoretical model of political selection, we show that governments release overly optimistic GDP growth forecasts ahead of elections to increase the reelec-tion probability. The bias arises from lack of commitment if voters are rational and from manipulation of voters' beliefs if they do not expect the incumbent to be biased. Using high-frequency forecaster-level data from the United States, the United Kingdom, and Swe-den, we document that governments overestimate short-term real GDP growth by 0.1-0.3 percentage points.