Jumps and time-varying correlations in daily foreign exchange rates

Abstract

This paper extends the multivariate latent factor ARCH model approach of Diebold and Nerlove (Journal of Applied Econometrics 4 (1989) 1) as a parsimonious alternative that pays particular attention to time series properties of daily foreign exchange rates such as jumps and to changing volatilities in both the common and country-specific factors. Using seven major daily dollar exchange rates from January 1 1992 to December 31 1996, this paper finds evidence of significant time-varying correlations and the country-specific variances. Consistent with the finding of Alexius and Sellin (1997) (A latent factor model of European exchange rate risk premia. Manuscript, The Economic Research Institute, Stockholm School of Economics), the two factor model appears to be a reasonable description of the major exchange rates.