Panel Cointegration Testing

[pankaj swain]

Dear Experts,

My panel dataset consists of firm-specific variables. I checked for the presence of the unit root in my variables and found it to be a mix of I(0) and I(1) variables. Now, regarding the cointegration test, I have the following questions.

1. With a mix of I(0) and I(1) variables, can I employ the Pedroni/Kao/Westerlund test implemented by -xtcointtest-, or are they suitable only when all the variables are I(1)?
2. If not, the other suitable option seems to be the ARDL bounds test proposed by Pesaran, Shin and Smith (2001). However, is there any command to perform the bounds test for panel data? The -ardl- command by Kripfganz and Schneider (2016) presumably works for time-series data, as I got the error "sample may not include multiple panels" when I used it with my panel data.
3. In case all my variables are I(1) and I need to find the number of cointegrating equations, can I use the Johansen Cointegration Test for panel data? A search in STATA resources found the -tsvecrank- command, which suggests its suitability for multivariate time series.

I would greatly appreciate any guidance on this matter. Thanks in advance.

Regards
pankaj