To chain or not to chain trade-weighted exchange rate indexes

With the advent of chain calculations for the U.S. national income and product accounts, it seems reasonable to contemplate using the chain approach for other indexes, such as trade-weighted exchange rates (TWEXs). A fundamental criticism of measuring the growth of gross domestic product by a fixed-base-year method is that the estimates are highly sensitive, especially when the economy?s structure is changing dramatically, to the arbitrary choice of the base year. Such a criticism can be levied against TWEXs. In fact, even TWEXs constructed using a Paasche index rather than a Laspeyres index have problems related to base periods. We examine theoretically and empirically the use ofa chain TWEX in relation to two well-known TWEX indexes: the Federal Reserve Bank of Atlanta index, which uses a Laspeyres index, and the Federal Reserve Bank of Dallas index, which uses a Paasche index. The choice of base year alters the behavior ofthe dollar in these two indexes. We contrast this result with the behavior of the dollar in comparable chain TWEXs, where the base year sensitivity is absent. Our results indicate that developers of TWEXs, as well as those revising TWEXs, should consider a chain approach. Furthermore, users need to be aware of the sensitivity of TWEXs to changes in either the base period for trade weights or the reference base period for exchange rates