There is a positive and robust partial correlation between openness, as measured by the share of trade in GDP, and the scope of government, as measured by the share of government expenditure in GDP. The correlation is robust in the sense that (a) it is unaffected by the inclusion of other control variables, (b) it exists for measures of government spending drawn from all available data sets, (c) it prevails for both low- and high-income countries, and (d ) it is not an artifact created by outliers. In addition, openness in the early 1960s is a statistically significant predictor of the expansion of government consumption over the subsequent three decades. The explanation that best fits the evidence is one that focuses on the role of external risk. Societies seem to demand (and receive) an expanded government role as the price for accepting larger doses of external risk. In other words, government spending appears to provide social insurance in economies subject to external shocks. The central evidence in favor of this explanation comes from regres- sions in which openness is interacted with two measures of external risk, volatility of the terms of trade and the product concentration of exports. In each case, the interaction term is strongly significant (and the fit of the regression improves), whereas the coefficient on openness per se becomes insignificant or negative. The same result is confirmed in panel regressions with fixed effects for time periods and countries. Hence, unlike other explanations for the correlation between openness and government size, this one receives considerable support.
