7
$\begingroup$

I was wondering: is there are a package in R for automated GARCH model selection? I'm thinking of something like what the forecast package does for ARIMA models.

If I implement this myself, would it be appropriate to just do a grid search over the possible parameters for the GARCH and ARIMA parts of the model (using the rugarch package), and select the one with the lowest AIC (or BIC)?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ Did you come up with anything interesting in the end? $\endgroup$
    – Richard Hardy
    Commented Mar 5, 2017 at 18:48

1 Answer 1

Reset to default
5
$\begingroup$

My experience with equities suggested that if you are confined to garch(p,q), then garch(1,1) is what you will want. Using a components model (Lee and Engle) is better -- it is sort of like a garch(2,2) but not quite the same.

When modeling multivariate garch (where there was a lot of choice in parameterization), it seemed to be that BIC was defnitely better than AIC. BIC has a larger penalty and so suggests smaller models. It looked like the penalty should be even bigger than in BIC -- that the BIC models were still too big.

Improve this answer
$\endgroup$
6
  • $\begingroup$ Thanks for the info. I'm trying to estimate a model that has an ARIMA and a GARCH component, and I was already thinking of constraining the garch component to (1,1). $\endgroup$
    – Zach
    Commented Jan 4, 2012 at 20:29
  • $\begingroup$ The garch estimation and the ARMA estimation are pretty much independent of each other -- the ARMA estimates you come up with are likely to be very similar whether you estimate with the returns or garch residuals. $\endgroup$
    – Patrick Burns
    Commented Jan 5, 2012 at 9:09
  • $\begingroup$ So could I could use the arima parameters from auto.arima in the forecast package, and just add garch(1,1) errors using rugarch? That would certainly make things simpler. $\endgroup$
    – Zach
    Commented Jan 5, 2012 at 13:01
  • 2
    $\begingroup$ Simpler and very likely to be seriously close to what you would get doing it the "right" way. $\endgroup$
    – Patrick Burns
    Commented Jan 5, 2012 at 17:45
  • $\begingroup$ @PatrickBurns, are you refering to Engle & Lee "A permanent and transitory component model of stock return volatility" University of California at San Diego, Economics Working Paper Series (1993)? I cannot find the paper itself, only references... Could you include a link? $\endgroup$
    – Richard Hardy
    Commented Feb 17, 2015 at 20:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.