# WTP in ACBC study

Dear All,

We conducted a ACBC summed price study for a product with a base price exp. 300k Euro, additional features (and levels) vary  from +20k to even +150k Euro. There are about 10 attributes + price. There was no competition for the product in the study (but in the market there are similar products but not with our tested features)- so in our case : just product & its features. Im trying to calculate WTP for the first attribute&levels but encounter a strange outputs - denying intuition.
This is an example for the first attribute:
We used a BYO section where  we priced levels for Attribute 1:
Level 1   0Euro
Level 2   +20k Euro
Level 3   +100k Euro

Average utilities for Attribute 1  (Hierarchical Bayesian):
Level 1   -50,1
Level 2    15,6
Level 3    34,5

Utilities for price were constrained negatively and coded piecewise (7 levels).

I set a scenario (1) with NONE where:
Product 1 :   Attribute 1 (Level 1) + Other Attributes (lowest level) + base price
Product 2 :   Attribute 1 (Level 2)  + Other Attributes (lowest level) + base price +premium price X

and were changing 'premium price' until both product got the same share (exp. 14%; 14%; 72% NONE). So far it's straightforward: my WTP for Level 2 is this premium price X.
Just wanted to make a note that if the base price is increased  to exp. 800k, the premium price where shares of preference are equal decreases and NONE increases - that makes sense.

The problem is with Level 3

Similar scenario (2)
Product 1 :   Attribute 1 (Level 1) + Other Attributes (lowest level) +  base price
Product 2 :   Attribute 1 (Level 3)  + Other Attributes (lowest level) + base price +premium price X

Although average utilities show that Level 3 is the most preferred, to get equal shares of preference my premium price must be minus 10k.
Product 1 (L1): exp. 300k; 14%
Product 2 (L3):  290k ; 14%
NONE: 72%

It's denying intuition as Level 3 contains everything what is in Level 1 and Level 2 + additonal benefits and its average utility is the highest. The only difference is that in the BYO part we showed this Level as the most expensive. Maybe I should somehow imply these BYO prices ? But how? and from the other hand I think HB separates utilities for price and attributes so I'm lost.
On the individual level  out of 60 respondents: level 1 is preferred by 11 resp, level 2 by 19, level 3 by 30.

I've read both articles reg WTP: Assessing the Monetary Value (2001) and Becoming an Expert (2017) but couldn't find an answer to my problem.

robson
asked Jun 27, 2019
edited Jun 27, 2019
Can you clarify for me, when you say, "On the individual level  out of 60 respondents: level 1 is preferred by 11 resp, level 2 by 19, level 3 by 30." are you doing this analysis by looking at the individual-level HB utilities?  Or, are you looking at the raw responses to the BYO question?

You didn't say how much independent variation (random shock) you allowed in the summed pricing experiment.  Did you use the default +30% to -30% random shock?

Indeed, if you use enough random shock to summed price and do not use very many attribute prohibitions of combinations in the experimental design, the the final HB utilities should partial out the effect of attributes separate from the price slope.

Also, n=60 is very small for most purposes.  Your confidence intervals on estimates such as WTP and simulated share of preference will not be nearly as tight as when using n=500 or more.
Hi Bryan,

Reg. "Out of 60 respondents.." - yes, I'm looking at the individual -level HB utilities.
Random shock -20%/+40%
No attribute prohibitions, we are not able to get many respondents , the market we investigate is very specific.
Thank you,
robson