Imagine I have a series of observations over a number of time
periods(T), in multiple regions (I).  I have region and time specific
explanatory variables (xit), but I also have a number of national
variables (xt), for which there is not a regional disaggregation.  I
want to include the national data as a variable for all regions, but
if i simply repeat the national data for each region (in a stacked
model), and estimate a single B for that variable, it seems to me that
the t values will be wrngly calculated, since I don't have
'independent' observations for each region.  Is that the case, and if
so do you have a method of avoiding the problem?