Abstract
This study uses the periodic GARCH (P-GARCH) model of Bollerslev and Ghysels (Citation1996) to capture the irregularly repetitive seasonal variation in the volatility of 15-minute NTD/USD exchange rate changes. The specification of state variables enables us to test the microstructure hypotheses in the FX market.
Notes
1 For a detailed discussion on estimating the P-GARCH model, see Bollerslev and Ghysels (Citation1996). As discussed in Bollerslev and Ghysels (Citation1996), for large sample sizes the QMLE-based inference procedure is generally reliable.
2 To de-seasonalize returns for intraday pattern, Andersen and Bollerslev (Citation1997) advocate flexible Fourier form (FFF). Martens et al. (Citation2002) use set of multiplicative factors to model the intraday seasonality. They estimate seasonal multiplier after averaging sums of squared returns across similar time periods and then obtain a de-seasonalized return series by dividing each return by its seasonal multiplier.
3 The final model is chosen on the basis of the AIC, SIC, and MSE criteria. Because of space limitations, the results of all estimated models are not shown. They are available upon request from the authors.