Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Switching regression on Panel data


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Switching regression on Panel data
Date   Mon, 24 Oct 2011 14:01:14 -0400

Andrea Mannberg <andrea.mannberg@econ.umu.se>:
I doubt your application satisfies the assumptions maintained by
http://www.sciencedirect.com/science/article/pii/030440769401645G
but see also
http://www.ucl.ac.uk/~uctpb21/doc/Econometrics_Journal_10_2007_pp263-293.pdf
https://www.msu.edu/~ec/faculty/wooldridge/current%20research/selectivje.pdf

I cannot see how you ever hope to estimate the effect of gender while
"controlling for individual heterogeneity."

On Mon, Oct 24, 2011 at 8:30 AM, Andrea Mannberg
<andrea.mannberg@econ.umu.se> wrote:
> Dear Statalisters,
>
> I am running a switching regression on a panel dataset.  If I understand it correctly, I cannot simply estimate the inversed mills ratio via xtprobit, insert this in the second step and bootstrap standard errors to correct for the generated regressor. Instead, I should estimate separate probits for each year and the corresponding imr. The second step is then to be estimated by fixed effects (Wooldridge, Journal of Econometrics, 68, 1995).
>
> I face two problems with this approach. First, my analysis is a program evaluation. Our basic hypothesis is that the program affects both the outcome in the first step (probit) and the outcome in the second step (continuous variable). Running separate probits for each time period  removes the possibility to control for individual heterogeneity (since I cannot control for both being in a treated household before treatment and being in a treated household after treatment). Second, since I am interested in the effect of for example gender, I cannot do fixed effects in the second step.
>
> To say the least, I am no expert on this issue, so I am not even sure that I have understood the basic procedure correctly. Any advice is highly appreciated.
>
> Best regards
> Andrea Mannberg

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index