Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Ellen Verhofstadt <Ellen.Verhofstadt@ees.kuleuven.be> |
To | "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |
Subject | st: ivprobit two-step estimator and marginal effects |
Date | Mon, 6 Dec 2010 12:44:52 +0100 |
Hello, I have a question with regard to the estimation of marginal effects after an ivprobit two-step estimation. My question is related to a problem discussed by Ramani Gunatilaka and Brian Poi (2007) -and getting residuals in tobit estimation as discussed by Monica and Scott (2004)-. Mister Poi proposes to use the method that can be found in Wooldridge (2002, Econometric Analysis of Cross Section and Panel Data) pp. 472-477 to obtain average partial effects after the two-step estimator. However, after reading the proposed method I'm still doubting how to implement this in STATA. y1 = x1 + x2 + x3 + y2 + y3 with y2 and y3 two endogenous regressors and y3 is censored. This is how I would work in order to obtain the average partial effects. regress y2 x1 x2 x3 z1 z2 z3 predict v2hat, resid tobit y3 x1 x2 x3 z1 z2 z3 , ll(0) ul(1) predict xb if e(sample), xb gen v3hat = `depvar' - xb if e(sample) probit y1 x1 x2 x3 y2 y3 v2hat v3hat mfx Will this method give me the -average partial- effects I'm looking for? Thanks for your help, Ellen * * 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/