Hi all,

I have a panel data set in long-form, where the dependent variable (Y) contains data of 10 emerging countries and for each country 30 years. The independent variable (X) contains data of the U.S. over 30 years.
I have used the 'xtset country year, yearly' command and now I would like to use the xtreg-command to do an regression of X on Y, however Stata tells me there are 'insufficient observations'. This is because the Y-var contains 300 observations and the X-var 30.

To solve this I copied the data of the U.S. to all observations. So now I have the following:

Year | Country_i | US | Y_value | X_value
1 India US 0.5 5.5
...
30 India US 0.7 6.0

1 China US 1.3 5.5
...
30 China US 1.4 6.0

1 Brazil US 1.1 5.5
...
30 Brazil US 1.6 6.0

etc.


However, when I now use the correlation command I get an unrealistic low correlation. When I test the correlation between X and Y for each country separately, the correlations are much a higher.

My question is whether my solution is valid and if not whether there is a solutions to let Stata know that one variable contains a panel and the other variable not?

Many thanks in advance.
Chris