st: xtabond2 memory problems

[Kit Baum <baum@bc.edu>]

Kevin writes

I'm using xtabond2 to run some regressions.  My panel dataset  
contains 1674
observations with data for each state by year and quarter.  Not all  
states
have observations for each quarter.

When I attempt to run the regression:

     xtabond2 Investment l.Investment pi, gmm(l.Investment)

I get an error stating:

".no room to add more variables due to width


The nature of Arellano-Bond estimation is to create a huge number of  
instruments, each of which is a separate variable in Stata terms. Try  
constraining the lags allowed for GMM instruments to, say, lags 3  
through 5. That should drastically reduce the number of instruments  
and make your problem feasible.

A-B is really made for "small T, large N"; you presumably have 50  
states and about 33 quarters per state. That is a huge number in  
terms of A-B, which often is used for T=7 or 8.

Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html


*
*   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/