Wagner's Law, the proposition that there is a long-run tendency for th
e public sector to grow relative to national income, has become a styl
ized fact in public finance. Numerous empirical studies have found str
ong support for the Law, particularly in a time-series framework. In t
his paper it is argued that these findings are likely to be spurious,
because they have been performed on non-stationary variables that are
likely not to be cointegrated. In the paper Swedish data for the perio
d 1861-1990 are used.