[Limdep Nlogit List] Computing the selection variables from the bivariate probit model.

[William Greene]

The expressions for the "corrections" appear on page E31-52 in the limdep 9.0
manual. To account for whether the variables are 1 or 0, multiply the "wa" 
and "wb" in the expressions by (2ya-1) and (2yb-1) respectively, and multiply
the correlation by (2ya-1)(2yb-1). This is done internally by the selection
routine, so in fact, you don't have to do this yourself.
/B. Greene

----- Original Message -----
From: "Sami Haile" <samex_zegr8 at yahoo.com>
To: limdep at limdep.itls.usyd.edu.au
Sent: Friday, November 14, 2008 6:41:50 PM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] Computing the selection variables from the bivariate probit model.


Dear limde usrs

I want to Compute the selection variables from the bivariate probit model, i.e. the inverse mills ratio, Which i want to use it as instrumental variable in the next step weighted 2step regression. If I follow the normal formula for invers mills, then I will end up with four invers mills ratio based on the four possibilities of the predicted dependent variable, i.e p11 p10 p01 p00. 

How can I calculate invers mills ratio for the two dependent variables y1 y2, which iI can use them in the second two indpendent  weighted 2step regression.In this case My dependent variables are y1=sellers ify1=1 or 0 otherwise; y2=buyers=1, if y2 =1 or 0 otherwise.  In the next step which is weighted 2step regression, i have dependent variable z1= quantity sold and z2= quantity bought.

Thanks for your time.



      

_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au