Government Size and Output Growth: the Effects of "Averaging out"

P>Panel data studies typically "average out" the error terms to be five calendar years apart such that they are less influenced by business cycle fluctuations. Using dynamic growth equations over the "globalization years" of 1986-2004, we provide an examination of the role of government expenditures to GDP (G/Y) in long-run growth. While the yearly time span is actually not prone to serious serial correlation problems, more powerful implications follow: We do observe strong negative long-run effects of G/Y on output growth in yearly time spans, while the averaged-out 5-year panels suggest the long-run economic impact of G/Y is muted.