I fitted a GARCH(1,1) model to a data set using Rmetrics, and then check the
summary of the fit, I found the test stat for Shapiro-Wilk test on the
residuals is 0.9953833 while the p-value is 0.003871092. When I fit the same
data set to the same GARCH model using S-Plus, the Shapiro-Wilk test on the
residuals give a test stat of 0.9904, but the p-value is 0.9524.

So which one is right?

The output from R and S-Plus are below. Rmetrics version is 260.72

Cheers,
Weimin


# Begin R Output#
> summary(ewk.fit)

Title:
 GARCH Modelling

Call:
 garchFit(formula = seriesData(ewk.ret) ~ garch(1, 1), data = ewk.ret,
    leverage = TRUE, trace = FALSE)

Mean and Variance Equation:
 ~arma(0, 0) + ~garch(1, 1)

Standadized Residuals Tests:
                                Statistic p-Value
 Jarque-Bera Test   R    Chi^2  17.95285  0.0001263537
 Shapiro-Wilk Test  R    W      0.9953833 0.003871092

# End R Output#


# Begin S-Plus Output #

> summary( ewk.fit)

Call:

garch(formula.mean =  ~ arma(0, 0), formula.var =  ~ garch(1, 1), series =
ewk.ret,

leverage = T)

Mean Equation: structure(.Data =  ~ arma(0, 0), class = "formula")
Conditional Variance Equation: structure(.Data =  ~ garch(1, 1) , class =
"formula")
Conditional Distribution:  gaussian

Normality Test:
--------------------------------------------------------------
Jarque-Bera   P-value Shapiro-Wilk P-value

      17.99 0.0001243       0.9904  0.9524
 # End S-Plus Output#

	[[alternative HTML version deleted]]