Assortment and Price Optimization under an Endogenous Context-Dependent Multinomial Logit Model

42 Pages Posted: 9 Aug 2023

See all articles by Yicheng Bai

Yicheng Bai

Cornell University - School of Operations Research and Information Engineering

Omar El Housni

Cornell University - School of Operations Research and Information Engineering

Paat Rusmevichientong

University of Southern California - Marshall School of Business

Huseyin Topaloglu

Cornell University - School of Operations Research and Information Engineering

Date Written: August 8, 2023

Abstract

Motivated by empirical evidence that the utility of each product depends on the assortment of products offered along with it, we propose an endogenous context-dependent multinomial logit model (Context-MNL) under which the utility of each product depends on both the product's intrinsic value and the deviation of the intrinsic value from the expected maximum utility among all the products in the offered assortment. Under the Context-MNL model, an assortment provides a context in which customers evaluate the utility of each product. Our model generalizes the standard multinomial logit model and allows the utility of each product to depend on the offered assortment. The model is parsimonious, requires only one parameter more than the standard multinomial logit model, captures the assortment-dependent effect endogenously, and does~not require the decision-maker to determine in advance the relevant attributes of the assortment that might affect the product utility. The Context-MNL model also admits tractable maximum likelihood estimation and is operationally tractable, with efficient solution methods for solving assortment and price optimization problems. Our numerical study, which is based on data from Expedia, shows that compared to the standard multinomial logit model, the Context-MNL model substantially improves out-of-sample goodness of fit and prediction accuracy.

Keywords: Assortment Optimization, Price Optimization, Context-dependent Multinomial Logit Model

Suggested Citation

Bai, Yicheng and El Housni, Omar and Rusmevichientong, Paat and Topaloglu, Huseyin, Assortment and Price Optimization under an Endogenous Context-Dependent Multinomial Logit Model (August 8, 2023). Available at SSRN: https://ssrn.com/abstract=4534984 or http://dx.doi.org/10.2139/ssrn.4534984

Yicheng Bai

Cornell University - School of Operations Research and Information Engineering ( email )

Ithaca, NY 14853
United States

Omar El Housni (Contact Author)

Cornell University - School of Operations Research and Information Engineering ( email )

2 E Loop Rd
New York, NY 10044
United States

HOME PAGE: http://https://people.orie.cornell.edu/oe46/

Paat Rusmevichientong

University of Southern California - Marshall School of Business ( email )

701 Exposition Blvd
Los Angeles, CA California 90089
United States

Huseyin Topaloglu

Cornell University - School of Operations Research and Information Engineering ( email )

Ithaca, NY
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
119
Abstract Views
423
Rank
432,212
PlumX Metrics