Government shares in total output are characterized by significant variation across countries. I noticed a strong negative correlation between government consumption shares and the price of government services in terms of private consumption. Motivated by this empirical observation, I developed a neoclassical growth model with added government that is capable of matching the variation in government shares very closely using only relative prices. In addition, I provide empirical evidence showing that the relative price of government consumption increases with income, which is consistent with distortions prevailing in poor countries. These two observations combined imply that government shares tend to be higher in poorer countries.