re: st: to use 'fe' and 'cl' options

[Christopher Baum <kit.baum@bc.edu>]

<>
S. said

I am trying to estimate the regression below:
xi: xtivreg  y F.x1 x2 x3  L.x4  i.year if year > 1978, fe cl(id)

but it is impossible to use 'fe' and 'cl' together.
Even I found in the help manuel the statement like this:
vce(cluster id) and use it, I got ''vcetype 'cluster' not allowed'' error message.

How can I use both 'fe' anc cl together?


Use Mark Schaffer's -xtivreg2- from SSC, which allows one- and two-way clustering. It requires that you also install -ivreg2- and -ranktest- from SSC.

Maarten's suggestion that if it isn't in official Stata, it probably is statistically dubious (or at least controversial) is not quite correct. Neither -sureg- nor -reg3- allow for cluster SEs, for instance, although sureg is just GLS, and reg3 can reproduce OLS and 2SLS estimates (but not their standard errors if clustering is used). In fact neither of these official commands even allow for robust SEs, although they appear on almost every other Stata estimation command! I can only conclude that the obstacles are related to development priorities rather than sound statistical issues.

Kit

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/