Official Intervention in the Foreign Exchange Market

Abstract

In an article in this Journal, Dean Taylor (1982) proposes a simple statistical test of the significance of central bank losses in the exchange market: Would random intervention be likely to result in as large a loss? This comment shows that, as it stands, his test statistic does not allow properly for the effect of the mean level of intervention. There is in fact an element of ambiguity in the question Taylor poses, and this is illustrated with examples taken from his empirical work. The connection between the Taylor test and regression analysis is noted and explored.