Monetary Models of Exchange Rate and the Random Walk: the Italian Case Over the Recent Float

This paper presents new empirical evidence on the Italian Lira - US $ exchange rate over the recent float. A univariate model as simple as the monetary model of exchange rate (MMER) can outperform a benchmark random walk with drift (RW) in out-of-sample forecasting up until the early 90s. This superiority of the MMER with respect to the RW model vanishes toward the end of 1992 (after the departure of the Italian Lira from the European Monetary System - EMS). These findings are most likely dependent on the presence of market imperfections such as capital control and high inflation in the Italian economy. These results are counterintuitive with respect to the theory of MMER which says that these models can be useful in explaining the behaviour of the exchange rate in the presence of perfectly competitive markets. Results are robust to changes in variables and the sample period.