Dear list members,
I want to apply AR(1)-GARCH(1,1) model in order to conduct a test of
structural shifts in conditional correlations which I previously estimated.
To be more exact, first, I estimate the conditional correlations using the
DCC-GARCH model. Now I want to check whether these correlations
significantly increased after the crisis began. For that reason I have to
use dummy variable regression. ‘1’ will stand for the turbulent period,
‘0’for the tranquility period.
This is how I programmed it in R:
d <- rep(0,991)
for (i in 814:922)
d[i]<-1;
step1 = arima(DCCrho, order = c(1,0,0), xreg=d, include.mean = TRUE)
step2 = garch (step1$res, order = c(1,1), include.intercept = TRUE)
the observations 814:922 are from the turbulent period that is why for them
the dummy variable takes the value ‘1’. Below, there are 2 steps that
apparently have to be made in order to program AR(1)-GARCH(1,1) because this
cannot be done in one step.
As a result of the first step I get 3 coefficients: ‘ar1’, ‘intercept’ and
‘d’ what seems to be correct. The coefficient ‘d’ represents the structural
shift. So if it increased significantly (I test it using the student’s
t-test) then it means that the conditional correlations increased
significantly after the crisis began.
My question is: do you think everything makes sense here? Or did I make any
unforgivable mistakes that have to be corrected? It is the first time I do
something like that so I need your help in order to be sure I am on the
right path.
Thank you very much in advance for any suggestions.
Greetings
Marcin
[[alternative HTML version deleted]]
