Please Help with Panel Data

[HanoWorld]

Dear All,

Hope all are fine. I would to seek your urgent assistance in my panel data and model specification. I'm trying to estimate my model which is: yp= f(kp, op, cfinancing) for 10 countries from 1990-2012.

Where: yp=real GDP per capita (constant 1999USD), kp=physical stock per capita, op=openness to trad, cfinancing= conventional financing.

The thing is that when I tested for co-integration, to my surprise, I could not reject null hypothesis for no co-integration.

I could not figure out where is the problem. Is it my data or my model specification. I have attached my data.

I would really need some help in this. Please advice.

Thank you in advance.

[jason_ll]

Try adding the labour force, maybe? The capital stock isn't the only thing influencing GDP... labour is important too, so adding it might help.

Not sure though - sometimes variables that should be cointegrated in theory fail the cointegration tests.

[HanoWorld]

Try adding the labour force, maybe? The capital stock isn't the only thing influencing GDP... labour is important too, so adding it might help.

Not sure though - sometimes variables that should be cointegrated in theory fail the cointegration tests.

Grateful for the reply. I found the solution.

The data is correct but my choice of which statistic to choose for my decision was the problem. In Pedroni's paper he mentioned that in small sample panel data the panel ADF and group ADF statistics perform well in determining the properties of small sample size (as mine) while v-panel statistic for large samples. Doing so, i found my varaibles co-integrated significantly.

Thanks for the help.