Cross-section weights and white period

[Ask_steffensen]

Hi,

My study partner and I want to test the connection between real growth in GDP and venture capital investments in the North (Denmark, Finland, Sweden and Norway). Hence, we use panel data for the four countries with a time period from 1989 to 2009. We use real growth in GDP as dependent variable and venture capital investments, GDP per capital, and stock market as explanatory variables.

We found that all variables were non-stationary, why we have differentiated the variables in order to make them stationary. Furthermore, we have used logarithm for GDP per capital and stock market. Finally, we have lagged the venture capital investments one period.

We use following equation: d(rg) c d(vci(-1)) dlog(gpc) d(sm)

Unfortunately we can only create a significant model where all variables are significant if we use Cross-section weights and White period with no d.f. correction. I have attached a picture of our output.

We are both quite new to econometrics and hence we are not sure if it is wrong to use these options? We know that white period assumes a large number of cross-sections, which we do not have. How big a problem is this? Furthermore, we know that cross-section weights assumes presence of cross-section heteroscedasticity, but how do we test if that is the case?

Finally, it would really be a great help if you could point out anything that looks suspicious?

NB. We use Eviews 6 student version.

Thanks in advance!

Kind regards,
Ask

[Ask_steffensen]

Bump!