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]

st: Multivariate probit with sample selection

From   <[email protected]>
To   <[email protected]>
Subject   st: Multivariate probit with sample selection
Date   Thu, 18 Aug 2011 10:09:52 +0100


Date: Wed, 17 Aug 2011 21:01:23 +0200
From: "Knoller, Christian" <[email protected]>
Subject: st: Multivariate probit with sample selection


I have a dataset that contains insurance choices of an insurance
company's portfolio. Policyholders can decide whether to purchase a base
contract or not. Those that have purchased the base contract can also
purchase two additional coverages. I am interested in what policyholders
purchase which of these two additional contracts. 
The problem is that I do not observe if policyholders that have not
purchased the base contract would have purchased the additional
coverages. So running a bivariate probit regression only using those
policyholders that have purchased the base contract might cause some
sample selection bias. 

Is it possible to estimate a bivariate probit regression with sample
selection in advance, i.e. a model similar to heckprob but with an
additional outcome equation, or in other words a model with one
selection equation and two outcome equations?

Thank you very much,

Yes, it is possible. See the examples, with downloadable code, discussed
by L Cappellari & SP Jenkins, 'Calculation of multivariate normal
probabilities by simulation, with applications to maximum simulated
likelihood estimation'. The Stata Journal, 6 (2), June 2006, 156-189.
[Article is free to download from Stata J website]

Professor Stephen P. Jenkins  <[email protected]>
Department of Social Policy 
London School of Economics and Political Science
Houghton Street, London WC2A 2AE, U.K.
Tel. +44 (0)20 7955 6527
Changing Fortunes: Income Mobility and Poverty Dynamics in Britain, OUP
Survival Analysis using Stata:
Downloadable papers and software:

Please access the attached hyperlink for an important electronic communications disclaimer:

*   For searches and help try:

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