st: Testing parity across two IV regressions

[Eric Stuen <eric.stuen@gmail.com>]

My issue: I want to test for parity of the coefficient of a variable
in two IV regressions.  The sample is the same for both, but the set
of excluded instruments varies.  I have tried using -suest- to
estimate the two as a system with a common vce matrix, but -suest-
apparently doesn't work with IV estimation.

Is there another way to approach this test in STATA, perhaps by
combining the error matrices manually?  Or would -reg3- handle this?
I am using robust clustered standard errors, which seems to be
incompatible with some commands.

The estimation command and syntax used:

xtivreg2 depvar (var1 var2 = var_iv1 ... var_iv10) var3 var4
yeardummies1-43 trends1-123, fe liml robust cluster(group_id)

The second regression is very similar:

xtivreg2 depvar (var1 var2 = var_iv5 ... var_iv14) var3 var4
yeardummies1-43 trends1-123, fe liml robust cluster(group_id)

The object is to test for parity of the coefficient of var2 between
the two regressions.

Thanks,

Eric Stuen

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/