ARDL model error

[Joshua Kriskinans]

Hi,

I have previously run stationarity tests to ensure the data i have is acceptable for the ARDL model fulfils both I(1) and then I(0) in first difference under the Augmented Dickey Fuller test. However when i run the command

“ardl depvar indepvar1 indepvar2 … , aic”
I get the error term:
likelihood information not found in last estimation results
stata(): 3598 Stata returned error
_ardl_optimlag(): - functioned returned error
_<istmtt>: - functioned returned error

With the error code message saying "You are using a Stata interface function and have asked Stata to perform a task. Stata could not or refused."

Is there a solution to this? Any help would be greatly appreciated.

[Daniel Schneider]

Hi Joshua,

I suspect that this is due to the model you are trying to estimate being incompatible with the data set because of collinearity or insufficient observations. In the following, I assume that the statement you indicated:

ardl depvar indepvar1 indepvar2 ... , aic

corresponds exactly to what you have tried, where "..." stands for additional regressor variables. If you have used other options besides -aic-, please let me know.

Could you please check whether

regress depvar L(1/2).depvar L(0/2).(indepvar1 indepvar2 ...)
estat ic

gives you a valid estimation output table and information criteria. If it does not, then it is ok that -ardl- terminated without producing estimates, but it should in any case have terminated more elegantly and with an informative error message. This will be the case in the next version of the command that we are currently preparing. If the -regress- command above does produce valid results, -ardl- needs are more substantive fix. In that case it would be great if you could make the data set and the exact command statement available to me so I can replicate the problem on my computer.

Greetings,
Daniel