# Re: st: Krinsky and Robb procedure

 From German Muchnik Izon <[email protected]> To [email protected] Subject Re: st: Krinsky and Robb procedure Date Mon, 25 Aug 2008 08:33:04 -0600

```Dear Martin,

Thank you so much for your suggestion. It works!!
Now I am going to see if I can program a convolution
approach in stata to see if the median WTP I get for the
two goods I included in my survey are statistically
different from each other.

Best,
German.

-----------------------------
Germ�n M. Iz�n
Department of Economics
PhD. Candidate
The University of New Mexico
Ph# (505) 277-1951

--- On Sat, 8/23/08, Martin Weiss
<[email protected]> wrote:

From: Martin Weiss <[email protected]>
Subject: Re: st: Krinsky and Robb procedure
To: [email protected]
Date: Saturday, August 23, 2008, 11:45 AM

Let me try an answer: if you look at the code of
-simulate- in line 57, saving into a -tempfile- with
the option
-saving(`simdat')-. I was able to turn this
-tempfile- into a
permanent one by appending the following after line
134 (and saving

***********
u `simdat', clear
save mysimdat
***********

This will allow you to access the meanwtp and the
medianwtp from each
simulation run
afterwards.

HTH
Martin

Quoting German Muchnik Izon <[email protected]>:

> Hi all,
>
> I have a new question. I am estimating 95%
confidence
> intervals for willingness to pay (WTP) for a good
using a
> command called wtpcikr that implements the Krinsky
and
> Robb procedure, recently developed by P. Wilner
Jeanty for
> stata. Based on a power point presentation given by
Jeanty
> at the 6th North American Stata Users Group Meeting
2007,
> the steps of this procedure is as follows:
>
> 1.Estimate the WTP model of interest
> 2.Obtain the vector of parameter estimates     and
the
> variance-covariance (VCV) matrix
> 3.Calculate the Cholesky decomposition, C, of the
VCV
> matrix such that
> 4.Randomly draw from standard normal distribution a
vector
> x with k independent elements
> 5.Calculate a new parameter
vector Z such that
> 6.Use the new parameter vector Z to calculate the
WTP
> measures of interest
> 7.Repeat steps 4, 5, and 6 N(>=5000) times to
obtain an
> empirical distribution of WTP
> 8.Sort the N values of the WTP function in
ascending order
> 9.Obtain a 95% confidence interval around
mean/median by
> dropping the top and bottom 2.5% of the
observations
>
> The example given in this presentation is:
>
> drop _all
> . set memory 8m
> . use south
> . gen lbid=ln(bid)
> . probit ypay lbid unlimwat govtpur environ
waterbill
> urban
> . wtpcikr lbid unlimwat govtpur environ waterbill
urban,
> reps(50000) meanl expo
>
>
> Is there a way in stata to obtain a vector of the
mean
> WTP's that are estimated in step 6 after each
replication.
>
> Thank you!!
>
> German.
>
> *
> *   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/
>

*
*   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/

*
*   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/
```