Wald test, Fama Mac beth and standard errors of a sum

[giotira]

Hi to all,

I have an unbalanced panel data of 15k cross section over 10 periods.
I want to run a Fama-Macbeth regression to it. Given that Eviews does not perform it, I am regressing the sample by years (10 regression of 15k crossections) and then averaging the results.
The final coefficients will be the simple average of the coefficients of the 10 periods.
What about the p-values?
I am thinking to run a wald test of the jointly significance of the coefficients,i.e.:
these are the 10 regressions
y1=c+x1+...
y2=c+x2+...
...
y10=c+x10+...
the wald test will be
H0: (x1+x2+...+x10)=0

From what I know, Eviews does not perform this kind of test (I need 10 different databases to perform this regression).
I am trying to do it manually and use the definition:
(theta-theta0)/SE(theta)

I am able to do the first part, but what about the stadnard errors? are they simply the sum of the standard errors of the 10 regressions? are they the average?