Abstract
In this paper we address the Rogoff's (1996) puzzle on the exchange rate excess volatility and the slow convergence to PPP. We argue that the evidence in favour of the latter stylised fact could be biased by the adoption of unreliable measures of the speed of convergence. In fact, adopting persistence profiles no evidence is found in favour of the puzzle: for a set of nine bilateral cases we estimate a relatively fast speed of adjustment. Since the robustness of the inference made on the basis of the persistence profiles depends on the correct identification of the cointegration relationships, a particular attention is devoted to this problem. We identify the cointegration space by imposing two competing sets of overidentifying restrictions: the first one identifies the PPP, the second implies a cointegrating relationship between the real exchange rate and the interest rates. Adopting a dominance criterion we conclude in favour of PPP for most cases. Keywords: purchasing power parity, cointegration, persistence profiles.