I am trying to estimate an ARCH(1) model to capture volatility and I am using daily return.
Now, I want to see if there is a particular behavior in a specific period of time and I use a dummy variable (dumr).
I formulated my ARCH in this way but I am not sure if I have used the
right procedure:
arch r dumr, het(dumr) ar(1/4) arch(1)
ARCH family regression -- ARMA disturbances and mult. heteroskedasticity
Sample: 04jan2000 - 28sep2012, but with gaps Number of obs   =      3059
Distribution: Gaussian Wald chi2(5)    =     39.76
Log likelihood = -5803.352 Prob > chi2     =    0.0000
OPG
r       Coef.   Std. Err.      z P>z     [95% Conf. Interval]
r            
dumr    -.045654    .064568    -0.71 0.480     -.172205    .0808969
_cons    .0293792   .0334821     0.88 0.380    -.0362445    .0950028
ARMA         
ar 
L1.   -.0533468   .0257392    -2.07 0.038    -.1037947   -.0028989
L2.   -.0077409   .0190056    -0.41 0.684    -.0449912    .0295095
L3.    .0975302   .0164617     5.92 0.000     .0652657    .1297946
L4.    .0232769   .0188409     1.24 0.217    -.0136506    .0602043
HET          
dumr    .8356495   .0253349    32.98 0.000     .7859941    .8853049
_cons    .4657993   .0250687    18.58 0.000     .4166655     .514933
ARCH         
arch 
L1.    .1138918   .0145224     7.84 0.000     .0854284    .1423552