Estimate T-GARCH

[stratos]

Hello to the community.

I am a little confused about an estimation of a T-GARCH(1,1,1) model.
So, I have two variables (X,Y) which are stationaty and the ARCH LM test showed that there is arch effect
I have estimated the model and I found that the threshold term is negative. What does this mean?
Any suggestions?

Best regards
stratos

[trubador]

It means, negative shocks decrease the volatility or positive shocks cause more volatility. This may sound counter-intuitive, but it is typically the case in exchange rate returns (in US$ terms), where the negative shocks actually correspond to good news and may lead to appreciation of the currency. The impact of positive shocks (or bad news) is more severe and therefore causes more volatility and may lead to significant depreciation. Recent turmoil in emerging markets is a good (but an unfortunate) example of this...

[stratos]

Thanks for the response.
Actually in my model that results I think it's a little weird since I have estimated the effects
of oil returns to gas returns (GAS_returns=a+b*OIL_returns) and I am confused right now.
So, I suppose that bad news for oil price is a positive shock (I assume that when oil price increases is bad news...Isn't it?)
and it would cause more volatility, therefore the negative threshold term is well-defined coefficient... Am I right?

[trubador]

Commodity price increases are usually considered "bad news" for countries that are dependent on the import of those goods. Oil returns is a controlling variable in your case. Assuming that the residuals from your equation are stationary, impact of a positive shock will cause more volatility in gas returns than that of a negative shock.

[stratos]

Residuals are stationary. It's fine that. The country I examine is an importer , thus it depends from oil. So, as i understand, the negative threshold term is fine in my model, since a positive shock (bad news) will cause more volatility in gas returns