Is money informative? Evidence from a large model used for policy analysis

In this paper we assess whether monetary variables convey marginal information on the state of the Italian economy, taking as a benchmark the forecasting errors generated by the quarterly model used by the Bank of Italy in the 1990s. We follow two alternative approaches. First we map monetary surprises into estimates of the structural disturbances using a Kalman filter approach, in order to improve the forecasts. Then we look at the sample correlations among forecasting errors in monetary and real variables. We find that bank interest rates have a strong information content. Monetary aggregates play no role according to the first approach; according to the second approach they do, but the sign of their effect is counterintuitive. All in all, the results highlight the role of financial prices and quantities as indicators of the state of the economy but they do not imply a mechanical policy reaction to this information, as both the strength and the sign of the relationship between the surprises in monetary and real variables depend on the source of the shocks.