Dear statalisters,
I am currently doing an event study to test the effect of terrorists events on the volatility of stock markets using GARCH(1,1) . However my knowledge of stata is small and I am knew to these kind of studies and so not sure how to proceed.
My main equation and variance equation are:
(1) (Rit = a +bRt*+epsilon)
Rit is the country i's return of stock index and R* is return of a global stock index.
(2) hit = y0 +y1hit-1 +y2 epsilon^2it-1
The GARCH (1,1) model is used to obtain the volatility for the period prior to the event. window = (n1,n2)
I have to obtain the multiplicative effect of the event on the volatility in order to formulate the cumulative abnormal volatility (CAV = SUM(Mt) -(n2-n1+1)). Which is formulated by:
Where Mt is the multiplicative effect of the event, N is the number of events in the sample, epsilon is obtained from the mean model. And E[hit|omegat*] is the forcasted volatility conditional to information set omega at time t. and E[hit|omegat*] ~ N(ARt,Mt*E[hit|omegat*]). ARt is the abnormal return.
As I said my stata knowledge is small, just as my experience with GARCH models and could use some help in order to proceed.
Do I get the Mt by predicting residuals (epsilon) and variance (E[hit|omegat*]) and filling these in the equation and formulate it in stata? Or would that be wrong? Or should I do a second regression in which I use the event-window (period of 10 days) instead of the estimation-window (500 pre-event trading days)?
Thank you in advance,
Rosa
I am currently doing an event study to test the effect of terrorists events on the volatility of stock markets using GARCH(1,1) . However my knowledge of stata is small and I am knew to these kind of studies and so not sure how to proceed.
My main equation and variance equation are:
(1) (Rit = a +bRt*+epsilon)
Rit is the country i's return of stock index and R* is return of a global stock index.
(2) hit = y0 +y1hit-1 +y2 epsilon^2it-1
The GARCH (1,1) model is used to obtain the volatility for the period prior to the event. window = (n1,n2)
I have to obtain the multiplicative effect of the event on the volatility in order to formulate the cumulative abnormal volatility (CAV = SUM(Mt) -(n2-n1+1)). Which is formulated by:
Where Mt is the multiplicative effect of the event, N is the number of events in the sample, epsilon is obtained from the mean model. And E[hit|omegat*] is the forcasted volatility conditional to information set omega at time t. and E[hit|omegat*] ~ N(ARt,Mt*E[hit|omegat*]). ARt is the abnormal return.
As I said my stata knowledge is small, just as my experience with GARCH models and could use some help in order to proceed.
Do I get the Mt by predicting residuals (epsilon) and variance (E[hit|omegat*]) and filling these in the equation and formulate it in stata? Or would that be wrong? Or should I do a second regression in which I use the event-window (period of 10 days) instead of the estimation-window (500 pre-event trading days)?
Thank you in advance,
Rosa
