Pedroni test

[petra89]

Hi,

I am trying to run Pedroni cointegration test for a panel with 19 cross sections (countries), 41 years and 3 regressors, total 779 observations, all of them in logs. But when I try to run the test I get the error "Insufficient number of observations", I checked the data and there are not missing values. The thing is that I tried with the same data without the logs and the test works...so, could you help me to understand why?

Thanks.

[EViews Gareth]

Do you have negative values before taking logs?