st: time-trend in correlation equation

[Bülent Köksal <bkoksal@gmail.com>]

Dear Stata Users,

I am calculating 90-days rolling-window correlations between
conditional volatilities of daily returns for two stocks as follows:


webuse stocks, clear

forvalues i=90(1)180 {
local j=`i'-89
mgarch ccc (toyota nissan= , noconstant) in `j'/`i', arch(1) garch(1) nolog
}


Two questions:

1. How can I accumulate correlations and standart errors in a file?

2. To control for time-effects in variances, we can include a
time-trend to the model:

.mgarch ccc (toyota nissan= L.toyota L.nissan, noconstant), arch(1)
garch(1) het(t)

My second question is, let's say that the correlations calculated
above increases. How can we test whether this increase is
statistically significant? Some papers included a time-trend in the
correlation equation. Is this possible in Stata, or is programming
necessary to achieve this task? Thanks.

bülent

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/