Abstract

In this paper we estimate the Italian underground economy using the currency demand approach. Exploiting the characteristics of non stationary time series, we set up a multivariate dynamic specification (VECM) to show how this method can produce results highly misleading, which need to be handled critically. Actually, leaving the standard framework of a single equation specification -up to now used in empirical literature- we demonstrate that the restrictions demanded by this method cause some troubles in identifying the excess of money demand. Besides, even though it is possible to generate time series coherent with Italian official estimates of the underground economy, they are very sensitive to the variable included and to the choice of starting period.