Fall 2014 Seminar Series

DATE: 10/23/2014
LOCATION: E51-335
TIME: 4:15pm
Reception immediately following

SPEAKER:
Garrett van Ryzin

TITLE
Estimating Willingness-to-Pay and Willingness-to-Sell from Negotiation Data and Implications for Pricing

ABSTRACT
We consider estimating price response for a firm selling differentiated products where the price outcome of a sales transaction is determined through negotiation between a salesperson and a customer. This selling process is common in many industries, including retail automotive, consumer lending, insurance, real estate and many business-to-business markets. We assume that each salesperson has a minimum reservation price that he/she will accept for a deal and each customer has a maximum willingness-to-pay (WTP). If the customer WTP is greater than the reservation price (RP), then a deal takes places at the Nash bargaining equilibrium price; if the WTP is less than the reservation price, then no deal takes place. Given a data set of the outcomes of such negotiations, we develop an expectation maximization (EM) algorithm to estimate the parameters of the underlying distributions of WTP and RP. Our method extends to the case where WTP and RP may be correlated and/or determined by a set of observable covariates. Using a real-world data set, we show that our EM approach produces more accurate out-of-sample estimates of customer-price response than traditional price response models (e.g. logit model). Furthermore, our model provides new information that can be used to set negotiating limits that improve revenues, and we demonstrate the potential of these limits with a counterfactual analysis of our data set.

 

This is joint work with Robert Phillips and A. Serdar Simsek.