Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

How to calculate WTP for products with certain attribute levels?

Dear Sawtooth Support-Team,

I would like to calculate the WTP of my CBC participants for 5 products with certain attribute levels. I know, that the calculation is controversial, but I want to try it.

It’s about wines and I have the attributes price (with levels: 2, 5, 8, 11, 14, 17, and 20 €), quality in points (with levels: 76, 81, 86, 91, 96) and grape (with levels: A, B, C, D, E).

I got how I can calculate the price value of an utility point:
The range between my price levels is 18€ and e.g. the utility range for respondent 1 for price is 166,21. Hence, I get approximately 0,10€ per utility point for this respondent.

But how can I calculate the WTP for this respondent for a wine with 76 quality points and grape A, for example? I read other questions about WTP estimation in the forum, but it was mainly about calculating a monetary value for the change from one attribute to another. And as I said, I want to calculate the WTP for 5 products with fixed attribute levels.

Is it possible to multiply the €-value per utility point (0,10€) with the respective utility values of the levels (76 quality points and grape A) and sum them up?
So if 76 quality points have an utility of -58,68 and the grape A 2,47 it would be: -58,68 x 0,10€ + 2,47 x 0,10€ = -5,62€

Thank you for your help in advance.

Best regards,
asked Aug 9, 2019 by Mary (340 points)
edited Aug 9, 2019 by Mary

1 Answer

0 votes

I think just like having a level as reference for computing the WTP for an attribute level, you'll need a product profile to use as reference for calculating the WTP for a product.  So identify a baseline or reference profile and then calculate utility of one of your constructed products relative to that baseline.  

I can imagine situations where your different products might require different reference profiles.

I'd recommend you use the share equalization method for calculating WTP rather than the formula, or that if you do use the formula, calculate WTP at the respondent level and then take the median value across respondents.
answered Aug 9, 2019 by Keith Chrzan Platinum Sawtooth Software, Inc. (95,675 points)
Thank you for your answer, Keith.

Yes, I wanted to calculate the WTP at the respondent level.
I thought that I could use quality level 76 and grape A as base or reference product and calculate the WTP for the other 4 combinations for each respondent. But that won’t work because the respondents differ in that their lowest utility is assigned to different levels, right?

Sorry, what is the share equalization method for calculating WTP?  Can I calculate the WTP for the different products at the respondent level with it?

Thanks in advance.

Do you care about utilities for specific levels of the reference product?  You're going to use the total utility for that product, and heterogeneity among values doesn't matter as much if you're computing WTP at the respondent level and then computing a median, right?

As for share equalization, simulate your reference product and one of the  four test products whose WTP you want to estimate.  Change the price of your test product until its share is equal to the share of the reference product.  This is the equalization price.
I am sorry Keith, I am not sure if I understand everything right and I am quite confused now. Or maybe I didn’t explain my purposes sufficiently?

Maybe you could make an example for me with my data?
Can I upload the Excel file with my data and you show me how to calculate the WTP for one respondent for a certain product?
Sure thing, Mary.  

Send your Excel file to keith@sawtoothsoftware.com.  Please specify the specs of the reference product you want to build and of one of your test products (whose WTP you want to estimate).  If you do this with a formula, you can just use the $/utility (calculated for a respondent) and multiply this by the difference in utility between your reference and test products, right?

Though again, I'd recommend using the simulation-based share equalization rather than the WTP formula.