[Limdep Nlogit List] Comparing WTP values

Gregory Lee Poe glp2 at cornell.edu
Mon Mar 8 04:25:38 EST 2010


Dear Bakti Hasan Basri and Mikolaj Czajkowski:

Please find below part of a communication (between ****'s) from Christian Vossler (Econ. University of Tenn.) that refers to a limdep program, data sets and output that uses the much more facile combinatorial alternative to the empirical convolutions approach discussed in a follow up paper:  

Poe, G.L., K.L. Giraud and J.B. Loomis, 2005. “Computational Methods for Measuring the Difference of Empirical Distributions.” American Journal of Agricultural Economics 87(2):353-365.

This program builds upon a helpful suggestion from Prof. Greene that efficiently keeps track of the negative tail of the distribution of the difference of two distributions.

I have attached the programs etc. The significance numbers referred to correspond to Table 5, p. 362 of the above-mentioned paper.

I hope you find this useful.

Best regards,

Greg Poe

****
Everything matches perfectly: find four files attached. 
"LinearLogit.xls" contains the MSO and 62T&E draws from the logit model. 
"Kristrom.xls" contains the MSO and 62T&E draws from the Kristrom model. 
"Complete combinatorial approach.lim" is a file you can execute in LIMDEP -- it will give you the output of both significance tests. Before you run the program simply put the two excel files on your c:\ drive (do not put in a folder). If you do this you should get the exact results found in the "output.txt" file.

Note that I added a calculation -- 1 minus Greene's "H0" statistic, which is the p-value. As hoped, it came out 7.6926% for the linear logit and 3.0586% for the nonparametric model.
****
-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au [mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of Mikolaj Czajkowski
Sent: Sunday, March 07, 2010 10:54 AM
To: Limdep and Nlogit Mailing List
Subject: Re: [Limdep Nlogit List] Comparing WTP values

On 2010-03-07 11:53, Bakti Hasan Basri wrote:
> Hi,
> I need a help for ideas  to do a test of comparing wtp values. The method uses the convolution approach and it was proposed by Poe et al. (Poe, Severance-Lossin and Welsh, 1994). Your kindness is much appreciated.
>
> Poe, G., E. Severance-Lossin, and M. Welsh. “Measuring the Difference (X-Y) of
> Simulated Distributions: A Convolutions Approach.” Am. J. of Agr. Econ. 76(4 1994):
> 904-15.
> .
>    


Dear Bakti Hasan Basri,

Here is what I used for this:

procedure=conv-test$
calculate;
len1=row(CONV1); ? get length of CONV1 and CONV2
len2=row(CONV2);
n0sum=0$ ? set up a scalar for counting negative outcomes
matrix; V2ones=init(len2,1,1)$ ? create 1s vector of the CONV2 length
DO FOR ; labCONV1 ; cont=1,len1$
matrix; DCONV=CONV1(cont,1)*V2ones-CONV2$
create; VDCONV=DCONV$
create; SUMtmp=VDCONV<=0$
calc; n0sum=n0sum+sum(SUMtmp)$
ENDDO ; labCONV1 $
calc; list; n0sum/(len1*len2)$ ? gives p-value that CONV1> CONV2

? delete; SUMtmp,VDCONV$
? matrix; delete; DCONV$

endproc; conv-test$

execute conv-test$



Best regards,

Mikolaj Czajkowski
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: output.txt
Url: http://limdep.itls.usyd.edu.au/pipermail/limdep/attachments/20100307/df182377/attachment.txt 


More information about the Limdep mailing list