Problem with garch (tseries)
The Ljung-Box test is susceptible to outliers in this context. I haven't looked at this data at all, but I suspect that a rank Ljung-Box test here would give a dramatically different result. More details are in the Ljung-Box working paper on the Burns Statistics website. Patrick Burns patrick at burns-stat.com +44 (0)20 8525 0696 http://www.burns-stat.com (home of S Poetry and "A Guide for the Unwilling S User")
Hannu Kahra wrote:
Ricardo, have you tested for the ARCH effects?
r <- scan("jpy.txt")
Read 1459 items
a <- r-mean(r) rsq <- a^2 # Testing for ARCH effects Box.test(rsq, lag=5, type = "Ljung-Box")
Box-Ljung test
data: rsq
X-squared = 2.8022, df = 5, p-value = 0.7305
# Look at the acf and pacf of squared returns acf(rsq) pacf(rsq)
It is no wonder that you are unable to fit a regular GARCH model. Regards, Hannu On 8/21/06, Ricardo Zambrano Aguilera <Ricardo.Zambrano at corpbanca.cl> wrote:
Greetings i have this time series, (currency japanese log returns) the data?s iid and it don?t have any serial correlation but it does not fit to a regular garch ... what could it be? best regards RZA <<JPY.txt>> pd:weekens removed
_______________________________________________ R-SIG-Finance at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance
[[alternative HTML version deleted]] ------------------------------------------------------------------------
_______________________________________________ R-SIG-Finance at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance