ESTIMATING DAILY SEASONALS IN FINANCIAL TIME-SERIES - THE USE OF HIGH-PASS SPECTRAL FILTERS

It is demonstrated that combining time domain and frequency domain ana lyses results in a considerable improvement in modelling seasonal patt erns in daily exchange rate changes. A high-pass filter is used, follo wed by the usual time domain analysis with a GARCH model, to estimate day-of-the-week effects in the spot return on U.S. dollars relative to British pounds and the results are seen to compare favorably with tho se from a pure time domain approach.