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, is already up and running.

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

Re: st: RE: marginal effects after Heckman

From   Adri Bestazza <>
Subject   Re: st: RE: marginal effects after Heckman
Date   Tue, 30 Mar 2010 16:37:06 +0100

I was wondering whether there was any built-in command in STATA 10 for
the McDonald and Moffitt decomposition method for the marginal effects
of the Heckman model, for both continuous and binary exogenous

I tried to follow the explanation provided for the Tobit model in: on page 31 & 32,
but there is something which is not clear to me: the mfx command in
Stata returns marginal effects and standard errors. How to get
standard errors when the decomposition is done manually?

Finally, I understand that in the example the decomposition is done at
the mean value of the continuous explanatory variables (the variables
with respect to which the unconditional marginal effects, dE(y)/dx,
are computed): is there a way to derive the mean of the unconditional
marginal effects associated with the values of each explanatory
variable instead?

Thank you in advance for any help!


On Wed, Mar 24, 2010 at 10:14 PM, Maarten buis <> wrote:
> --- On Wed, 24/3/10, Adri Bestazza wrote:
>> In order to get unconditional marginal effects, I used
>> the mfx command. Yet, I noticed that the marginal
>> effects I found are identical to the coefficients for the
>> outcome equation, when the Heckman model is estimated by
>> maximum likelihood.
> Marginal effects are the derivative of your dependent
> variable with respect to your explanatory variables. With
> commands like -heckman- you think of the dependent variable
> in multiple ways, so you need to specify what kind of
> dependent variable you have in mind. If you do not do so
> Stata will take the default prediction, which in case of
> -heckman- is the linear predictor, and the derivative of
> the linear predictor with respect to the explanatory
> variables are of course just the coefficients. To see the
> possible parameterizations of the dependent variable see:
> help heckman postestimation##predict
> Also see -help mfx-, in particular the -predict()- option.
> Hope this helps,
> Maarten
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
> --------------------------
> *
> *   For searches and help try:
> *
> *
> *

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index