Optimal number of dummy's in VECM?

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

j235
Posts: 14
Joined: Mon Oct 24, 2016 9:24 am

Optimal number of dummy's in VECM?

Postby j235 » Tue May 09, 2017 7:02 am

I am performing VECM with around 60-variables(quarterly data) with three variables. I use around 11 dummy and I got VECM which is positive of all test ( sigh of coefficient,autocorrelation, heteroscedasticity, normality of residuals etc.)

So my question does this 11 dummy's is normal for VECM? Or how many dummy variable is optimal for using in VECM?

Frencholivier
Posts: 4
Joined: Sat May 20, 2017 10:25 am

Re: Optimal number of dummy's in VECM?

Postby Frencholivier » Sat May 20, 2017 11:43 am

Dummies are very important for the VECM (=cointegrated VAR). Do you have 60 variables or 3 variables? If you mean "60 observations for each 3 time series" then 11 dummies is acceptable. There is no optimal number. The rule of thumb promoted by Katarina Juselius is to use a dummy (1) for each spike in the residuals greater than 3 times their standard deviation and (2) if that spike corresponds to an exogenous event (not endogenous) --this requires you to know the history of events on time period/country you are investigating.


Return to “Econometric Discussions”

Who is online

Users browsing this forum: No registered users and 25 guests