Convergence in the ERM and Declining Numbers of Common Stochastic Trends

In this paper, we use recursive and rolling cointegration methods to test for a system of several exchange rates being within the process of convergence. We use the methods to analyze how the convergence of five ERM exchange rates has developed during the ERM period. We find that the number of cointegration vectors in the system of ERM exchange rates increases as the sample period is extended, and interpret this as a sign of increased convergence of ERM exchange rates. In particular, we find no evidence of convergence in the first years of the ERM and strong evidence of convergence in the last years of the ERM. In the analyses we acknowledge that managed exchange rates, such as exchange rates in ERM target zones, can be misaligned at their observed values as compared to their fundamental free-float values. For this reason, we also study convergence of filtered shadow exchange rates. We use two filters to extract the shadow exchange rates: a linear filter and a non-linear filter.