Temporal Aggregation of the Returns of a Stock Index Series股票指数系列报酬的时间聚合.pdfVIP

Temporal Aggregation of the Returns of a Stock Index Series∗ Kurt Br®ann®as Department of Economics, Umea˚University kurt.brannas@econ.umu.se Abstract The effects of temporal aggregation on asymmetry properties and the kurtosis of returns based on the NYSE composite index are studied. There is less asym- metry in responses to shocks for weekly and monthly frequencies than for the daily frequency. Kurtosis is not smaller for the lower frequencies. Key Words: Asymmetric moving average, QGARCH, estimation, kurtosis, Pearson IV, NYSE. JEL Classification: C13, C22, C51, C53, G12, G14. Umea˚Economic Studies 614, 2003 1. Introduction This study focuses on temporal aggregation of the daily returns of the composite index of the New York stock exchange (NYSE) to weekly and monthly frequencies. More than 37 years of daily returns are used for the empirical study. The study is based on the nonlinear asymmetric moving average (asMA) model (Wecker, 1981; Br®ann®as and De Gooijer, 1994) with an asymmetric quadratic GARCH specification (Br®ann®as and De Gooijer, 2003) for the conditional variance. The density is specified as Pearson type IV (Pearson, 1894, 1895; Premaratne and Bera, 2000; Br®ann®as and Nordman, 2003b). For the conditional mean specification, the temporal aggregation results of Brewer (1973) have previously been employed by Br®ann®as and Ohlsson (1999) for the asMA model when extended by an autoregressive component. Drost and Nijman (1993) gave temporal aggregation results for weak GARCH models in combination with ARMA models. They also indicated that the density will