Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Kit Baum <kitbaum@me.com> |
| To | statalist@hsphsun2.harvard.edu |
| Subject | re: st: NW st. errors with -ivreg2- |
| Date | Sun, 27 Mar 2011 19:44:13 -0400 |
<> I run the following 2 regressions: *reg y x1 x2 x3 x4 x5 *predict fitted_y *newey z fitted_y, lag(12) and *ivreg2 z (y = x1 x2 x3 x4 x5), bw(13) robust The problem is that HAC s.e.'s of the slope coefficient are appreciably different in the two cases. These are different models in terms of how the VCE is calculated. The SEs from the two stage 'by hand' will not be correct in any case, as they are calculated with the wrong residuals. Also note that your -ivreg2- (from SSC) has a 'small' option that will yield different standard errors than the large-sample standard errors you will get from the above command. Kit Baum kitbaum@me.com * * 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/