The RATS Software Forum

Is there any code available for GARCH models in which the variance equation is affected by surprises? Examples of such modified GARCH models are provided in the following papers:

Brenner, Pasquariello and Subrahmanyam (2009) "On the volatility and comovement of U.S. financial markets around macroeconomic news annoucements", Journal of Financial and Quatitative Analaysis, 44, No. 6, 1265-1289
and
Bomfim (2003) "Pre-annoucement effects, news effects and volatility: Monetary policy and the stock market", Journal of Banking and Finance, 27, 133-151.

The authors of these 2 papers introduce dummy variables interacted with a certain macroeconomic suprise into the variance. However, I don't think this can accomplished with the regressors or xregressors option in RATS. Do such models need to be estimated using "maximize"?

XREGRESSORS does allow you to include variables (which could be dummy variables coding the "surprises") in the variance equation.

Why do you think this would not work for what you want to do?

Regards,
Tom Maycock

moderator wrote:Why do you think this would not work for what you want to do?

The GARCH models estimated in these two papers introduce the "surprises" as multiplicative functions of volatility. Would the xregressors option on GARCH be able to estimate such a model?

Regards,

No. To get the multiplicative shifts, you would need to set it up with MAXIMIZE.