Estimating willingness to pay in model results with interactions

[Samir Thapa]

Hi,

I am trying to estimate mean and median wtp from a logit model that has interaction terms.

My logit model is as below:

logit wtp bid_level i.health_impact##c.edu i.cc_impact##c.firewood_quan i.subsidy_knowhow##c.firewood_quan i.gender##i.collateral ready_save yearly_income

I tried both the available/found wtp commands - wtpcikr (Jeanty, P.W, 2007) and wtp (Hole, A.R., 2007); but the interaction terms are not allowed.

I am using stata 13.1.

Would appreciate your help.

Thank you.

Jeanty, P.W. Constructing krinsky and robb confidence intervals for mean and median willingness to pay (wtp) using stata. in Sixth North American Stata Users’ Group Meeting, Boston, August. 2007.

Hole, A.R., A comparison of approaches to estimating confidence intervals for willingness to pay measures. Health economics, 2007. 16(8): p. 827-840.

[Carlo Lazzaro]

Samir:
if you do not plan to use -margins- after -logit- and the user-written command -wtpcikr- does not support -fvvarlist- notation, you can create interactions by hand and re-run -logit-.

[Samir Thapa]

HI Carlo,

Thank you so much.

I did the manual interactions to get the wtp estimates.

I will check if -wtpcikr- supports fvvarlist.

Samir

[KHALID JOYA]

Hi dear Samir!
could you please let me know how you did that cause I have the same problem and I want to find wtp for interaction terms using stats but when I have run the wtp commond, interactions are not allowed.

thank you

[Carlo Lazzaro]

Khalid:
I think you have to create interactions by hand, as the community-contributed command you're using does not support -fvvarlist- notation.

[KHALID JOYA]

Hi dear sir!
could you please let me know how can I create interactions by hand?
really I couldn't get what do you mean by that.
cause I don't have any knowledge about that.
thank you.

[Carlo Lazzaro]

Khalid:
here you are:

. sysuse auto.dta
(1978 Automobile Data)

. g trunk_weight=trunk*weight

. reg price trunk_weight

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(1, 72)        =     19.76
       Model |   136783548         1   136783548   Prob > F        =    0.0000
    Residual |   498281849        72  6920581.23   R-squared       =    0.2154
-------------+----------------------------------   Adj R-squared   =    0.2045
       Total |   635065396        73  8699525.97   Root MSE        =    2630.7

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
trunk_weight |   .0613037   .0137893     4.45   0.000     .0338153     .088792
       _cons |   3483.685   676.2708     5.15   0.000     2135.564    4831.806
------------------------------------------------------------------------------

.

[KHALID JOYA]

Thank you very much dear sir I did as you said and it worked for me.

As for now, cause I want to calculate the wtp for interactions I am using this formula:

categorical wtp= coefficient of A + coefficient of A * Coeffcient of B / - coefficient of price

cause in the interactions we have two different coefficients (coefficient A * coefficient B) and both of them should be included in calculations.

but stata wtp commond calculate like this: wtp=coefficient of A * coefficient of B/-coefficient of price

due to that the results from my calculations are much different with results from stata.

could you please let me know if there is any other command in stata for interactions WTP or categorical wtp.

your consideration will be highly appreciated
thank you sir

[Carlo Lazzaro]

Kahlid:
not that I know.
Sorry I cannot be more helpful.
As an aside, please call me Carlo, like all on (and many more off) this list do. Thanks.

[Hong Il Yoo]

KHALID JOYA: I would like to recommend the command -singleb- that Alejandro Lopez-Feldman wrote some time ago. You can download it by typing -ssc install singleb- at the command prompt and -help singleb- to find a link to the background paper. In a nutshell, -singleb- allows you to estimate a probit model that is parameterised directly in terms of the mean (equation "Beta" in -singleb- output) and standard deviation (equation "Sigma") of normally distributed WTP. You can interpret "Beta" coefficients directly as the effects of regressors on E(WTP|X).

[Akhmadjon Akhmadjonov]

Good part of the day Carlo. I have been reading your responses for different issues arising from stata and from econometrics itself. I am really grateful for your responses even they weren't for me, they are really helped me to tackle giant number of obstacles which I have faced so far. And finally I need your personal advice and help regarding my dissertation. Could you please help me to choose right economic model and proper econometric approach to finalize my research analysis?
My research question is How FDI,Tradeopenness effect on Economic growth of China and India? Comaprative analysis of economic growth based on Solow growth model. I have run time-series regression from 1982-2018 in order to compare two countries. But I have problem with ommited variables. I am using the following variables FDI,Tradeopenness, Total factor productivity,Capital stock,Labor force particiaption rate, Human Capital, Gross SAvings and Remittance inflow and their effect on GDP per capita.
I have faced with some problems with time-series data and now I thinking to transfer my data to panel. What do you think about this? Could you please suggest me smth as professional in this sphere?
Thank you in advance
Best Regards
Akhmadjon Akhmajonov

[Diriba Abdeta]

Hello dear Statisticians!,
I am also trying to estimate WTP by using 'wtpcikr' command provided by (Jeanty, P.W, 2007), after bivariate probit regression model. However, Stata is replaying with an error message "WTPCIKR finds no estimates, please estimate a model or re-estimate the model". Any help!

Thank you in advance!!