Testing Wagner’s hypothesis using disaggregated data: evidence from a global panel

The paper examines the relationship between aggregate and disaggregated public spending and GDP in a panel of 50 developed, developing and transition economies over the 1981–2012 period. The paper specifically concerns the direction of causality between the two variables: from output to spending (as per Wagner’s hypothesis), or the reverse causality (based on Keynesian theorizing). Methodologically, we conduct panel unit root and cointegration tests, as well as three types of causality tests (Toda–Yamamoto, Dumitrescu–Hurlin, and Juodis–Karavias–Sarafidis). The tests reveal the weak evidence of cointegration, and the absence of Keyensian-type causality in all expenditure categories. Wagner’s hypothesis generally found support in all specifications and categories, although the mutual feedback between expenditure and GDP was also observed in a number of cases.