st: questions about xtabond2

[Samia Tavares <samia.tavares@rit.edu>]

I submitted a question about xtabond2 but got no response, so I'll  
rephrase it (and add a couple of others that came up):

1. I'm a bit confused about the distinction between predetermined and  
strictly exogenous variables. From what I understood, predetermined  
variables go in the gmmstyle option and strictly exogenous ones go in  
the ivstyle one. But what is the difference? I thought it was that if  
a variable becomes endogenous when differenced, it goes in the gmm  
option. But if that is the case, the help file says you can simply  
specify eq(level) in the iv option.

2. Related to this, and just to make sure, you cannot have a variable  
in both the gmm option and the iv option at the same time, right?

3. In one dataset, I have 3 years corresponding to 3 US presidential  
elections--1996, 2000, and 2004. I constructed the lagged dependent  
rather than relying on Stata's time series command (so the lagged  
dependent is for 1992, 1996, and 2000). Data are at the county level  
for one state. My question is about tsset. Would I do tsset county  
year, even though the years are 4-years apart? Or would I create a  
time variable so that 1996=1, 2000=2, and 2004=3 then do tsset county  
time? The answers differ depending on how I do tsset.

4. What if some of your independent variables don't vary with time  
but a dynamic panel model makes sense due to the nature of the  
dependent variable? Can the command still be used?

Thanks in advance.



*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/