-xtabond2- now accepts aweights, pweights, and fweights. fweights must be constant over time. The weights for the differenced (transformed) equation are just copied directly from the weights for the original observations in levels. Thanks to Dieter Urban for suggesting this. The command now also lists the instruments constructed, as part of the output, on a suggestion from Jonathan Temple and an inspiring nudge from -xtdpd-.
-abar- (Arallano-Bond autocorrelation test) now works after weighted regressions. In addition, it now works after the new official Stata command -ivregress-, and after Bartlett/Newey-West regressions in ivreg2 and ivregress, with or without the robust option. The Arellano-Bond test is attractive for its generality, being appropriate for linear GMM, thus 2SLS and OLS, with various kinds of standard errors, and for both panels and (I believe) ergodic time series.
Finally, -cmp- now has an -interactive- option that lets you interrupt it, explore and adjust the interim solution with commands such as -ml plot-, then resume using -ml max-. "cmp" stands for "Conditional Mixed Process". The command is a flexible generalization of ivtobit, ivprobit, bitobit, mvtobit, biprobit, mvprobit, etc., that allows multi-equation models that mix probit, tobit, and regular uncensored, continuous equations. It requires Stata 10. Peter Gottschalk tells me "cmp" is a confusing name! I welcome suggestions for a new one. The winning entry gets a free copy of the program. :-)
To install or update these commands, use "ssc install [command name], replace". Or use -adoupdate- to update. For xtabond2 and cmp, you then need to restart Stata or do "mata: mata: mata mlib index".
--David
David Roodman
Research Fellow
1776 Massachusetts Avenue NW
Washington, DC 20036
Center for Global Development
Independent Research and Practical Ideas for Global Prosperity
�
*
* 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/
