I revisit the relationship between trade and government expenditures in an attempt to extend the work of Rodrik (J Polit Econ 106(5):997–1032, 1998). Several different tests are conducted. The first is a simple replication of Rodrik’s benchmark model on an extended dataset. The results here are not conclusive but suggest that the causation may run from government spending to trade. I then use bilateral trade in observations and the augmented form of the classic gravity equation used by trade economists. This analysis uses time series changes in government spending and trade. Test for Granger style causality are performed. Again little evidence is found for Rodrik’s hypothesis that trade causes government spending. And some weak evidence exists for the alternative hypothesis. I then extend the analysis to trade tax revenues and I find little evidence over short run periods but over the long run there does seem to be an association between high trade tax revenues in one period and high government expenditures in the next period. Finally I look at a direct test of Rodrik’s hypothesis that trade creates volatility in income which then induces government to provide social insurance. I calculate a measure of trade volatility by looking at bilateral trading partners and find that trade volatility is associated with income volatility. The results of this test are ambiguous because changes in government expenditures are negative related to income volatility but at the same time positively related to trade volatility.