ARDL model

[Wendel Ndzana]

How to estimate a ARDL model with interaction variables ?
First, i used the PCA method to generate the interaction variable, is this correct ? When i use these variable, the software answers me:
initial vector: extra parameter xy founded
specify skip option if necessary.
Second when i generate the interaction term by taking the product of the two variables, i get extremely large short-term coefficents. is tis normal?

[Sebastian Kripfganz]

Please show us the command lines you typed into the Stata command window and the corresponding Stata output. Otherwise, it is hard to give a helpful response. Please see point 12 in the FAQs.

[Wendel Ndzana]

Sorry Sébastien... There are the command lines i typed in the first case started from PCA method until xtpmg.

[Wendel Ndzana]

Dans le second cas où j'ai généré les termes d'interaction par le produit de variables, les lignes de commande étaient

[Sebastian Kripfganz]

I am afraid I cannot provide help on xtpmg. Maybe consider the command xtdcce2 as an alternative.

The large short-run coefficients are a bit worrying. There might be a high degree of collinearity among those variables.