[Sam Helen]

Excellent, much agreed.

[Sam Helen]

Sebastian Kripfganz - I have been following your responses to estimating panel ARDL models, and have another outstanding question:

You explain that your ardl Stata command does not allow for estimating panel error correction models, nor does it allow for panel bounds testing.

In this case, what would be the sensible options for estimating an ardl model in error correction form on panel data with the option of bounds testing? I can of course think of running individual models by country and then testing each model separately. However this does not appear ideal - we would want the country-level data to be able to interact with on another in a single, larger model (increased efficiency and other benefits).

What are your thoughts on how I could rearrange my dataset to solve this issue?

Much obliged, thanks,

Sam

[Sebastian Kripfganz]

As far as I know, there is no bounds test for panel data. Apart from that, the community-contributed xtdcce2 command might be useful to estimate such models.

[Sam Helen]

This is what I suspected. I appreciate the lead Sebastian!

[John Costopoulos]

Dear Sam,

You can also check the paper below to see if it could provide to you any new ideas. As stated in the Abstract, they use the "asymmetric panel ARDL" method.

Asteriou, Pilbeam and Pratiwi (2020). Public debt and economic growth: panel data evidence for Asian countries. Journal of Economics and Finance (https://doi.org/10.1007/s12197-020-09515-7).