Assumptions of Panel ARDL

[Hamid muili]

Dear Professors,
I will like to know if there's an assumption concerning the numbers of Cross sectional unit and time dimension for panel ARDL. I have seen it used with different specification both for N<T and N>T. Based on the premise of large N and Large T.

[JanDitzen]

I think it would help if you would clarify which estimator you are using. Since you talk about large N, large T, I am assuming you are referring to the CS-ARDL and CS-DL estimators (Chudik et. al 2016) and implemented by xtdcce2.

For the asymptotic distribution of the MG estimator T is not allowed to be small relative to N and N and T are jointly converging to infinity. What does this mean in practice? T should be large enough so that you could run a time series ARDL on it. N should be large enough that you can run a standard OLS or MG regression on it.