Abstract
The long-run tendency for government expenditure to grow relative to GNP, Wagner's law, is tested for six European countries using data from around the mid-19th century to 1913. With few exceptions the results suggest that: nominal and real GNP, nominal and real government expenditure, and population were nonstationary in their levels but stationary in first differences; either nominal GNP and nominal government expenditure and/or real GNP and real government expenditure were cointegrated in five of the six countries, and that these variables were cointegrated with population in the remaining country; and Granger-causality was mainly unidirectional from income to government expenditure. Thus, there is considerable support for Wagner's law in 19th century Europe.