Comovement of foreign exchange and term premia with overlapping and non-overlapping returns
This paper re-examines returns earned on a variety of Euro-deposit positions at three months, six months, and twelve months. We use these to test the restrictions of an intertemporal capital asset pricing model, and we find that we cannot reject a single factor model when the holding period and the maturity are three months. In addition, the fit improves when the holding period and maturity increases to six and twelve months. These excess returns are "overlapping" when the data is measured monthly. We also consider returns defined over the same markets and maturity horizons but calculated without overlap; that is, we vary the investment horizon, but compute the returns when the holding period is only one month. We strongly reject the one factor model with these "non-overlapping" returns. We investigate the possibility that finite sample size and power properties cause our inability to reject a one factor model when we use overlapping returns. We generate term structure data for three countries according to the null hypothesis of a single factor, and determine the empirical distribution of the test statistic. The restrictions imposed by the model are not, in fact, rejected frequently enough with overlapping returns at any of the investment horizons when inference is based on the asymptotic distribution. We also find the power of a one factor model test against a two factor model is larger when the returns are measured to avoid overlap.
Fernald, Julia Dana, "Comovement of foreign exchange and term premia with overlapping and non-overlapping returns" (1992). Dissertations available from ProQuest. AAI9308568.