Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Smith-Blundell test (1986) for effects of endogeneity after probit, tobit

From   Jeffrey Wooldridge <>
Subject   Re: st: Smith-Blundell test (1986) for effects of endogeneity after probit, tobit
Date   Sun, 13 May 2012 17:05:20 -0400

test is the inverse Mills ratio (defined for both the y2 = 0 and y2 =
1 outcomes). This assumes that the reduced form for y2 is probit. The
same approach works when y1 is probit or tobit.

I have a fairly recent paper that derives the variable addition tests
in this case. If you email me at I can send it to
you. I've been bad about posting my recent working papers.


On Fri, May 11, 2012 at 10:32 AM, Danny Dan <> wrote:
> Hello Friends,
> I have a question related to the -probexog- and the -tobexog- commands
> in Stata (written by Prof. Christopher F Baum, Boston College, USA).
> -probexog- and the -tobexog- calculate the "Smith-Blundell test (1986)
> for effects of endogeneity after probit, tobit". These tests are based
> on the two-step procedure (analogous to Rivers-Vuong method for binary
> response models) as proposed by Smith and Blundell (1986).
> I know that Smith-Blundell test for tobit model can be used with
> continuous endogenous explanatory variable (Econometric Analysis of
> Cross Section and Panel data, Wooldridge 2002, page 531). Could you
> please let me know whether these tests (-probexog- and -tobexog-) can
> also be used when the endogenous explanatory variable is either binary
> or censored?
> Please let me know if my question needs further clarification.
> Thank you for your time.
> Please help.
> Regards,
> Dan
> *
> *   For searches and help try:
> *
> *
> *

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index