There exists a persistent disagreement in the literature over the effect of business cycles on economic growth. This paper offers a solution to this disagreement, suggesting that volatility carries a positive direct effect, but also a negative indirect effect, operating through the insurance mechanism of government size. Theoretically, the net growth effect of volatility is then ambiguous. The paper reveals the underlying endogeneity of government size in a balanced panel of 95 countries from 1961 - 2010. In practice, the negative indirect channel dominates in democracies, but with less power to choose public services in autocratic regimes the positive direct effect takes over. Consequently, volatile growth rates are detrimental to growth in democracies, but beneficial to growth in autocracies. The empirical results suggest that a one standard deviation increase of volatility lowers growth by up to 0.57 percentage points in a democracy, but raises growth by 1.74 percentage points in a total autocracy. These findings point to a crucial intermediating role of governments in the relationship between volatility and growth. Both the size of the public sector and the regime form assume key roles.