Richard,
The dataset is the NLSY97, selection just on males, blacks and whites
only, and balanced for the 1997-2002. Here's the entire code, plus
output. I'm going to snip the fixed effects, though.
. xi:xtpoisson rp i.race age age2 hgc hhd1 i.year i.fip, vce(boot);
...
Note: One or more parameters could not be estimated in 7 bootstrap
replicates; standard error
estimates include only complete replications.
. mfx, dydx varlist(_Irace_2 age age2 hgc hhd1);
no observations
r(2000);
I wonder if the vce(boot) and/or the fact that not all replicates
were successful has something to do with it? What happens if you run
it without vce(boot)? What does ereturn list tell you e(N) is?