Models for Retail Inventory Management with Demand Learning Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 19, 2019
Creator
  • Wang, Zhe
    • Affiliation: Kenan-Flagler Business School
Abstract
  • Matching supply with demand is key to success in the volatile and competitive retail business. To this end, retailers seek to improve their inventory decisions by learning demand from various sources. More interestingly, retailers' inventory decisions may in turn obscure the demand information they observe. This dissertation examines three problems in retail contexts that involve interactions between inventory management and demand learning. First, motivated by the unprecedented adverse impact of the 2008 financial crisis on retailers, we consider the inventory control problem of a firm experiencing potential demand shifts whose timings are known but whose impacts are not known. We establish structural results about the optimal policies, construct novel cost lower bounds based on particular information relaxations, and propose near-optimal heuristic policies derived from those bounds. We then consider the optimal allocation of a limited inventory for fashion retailers to conduct "merchandise tests" prior to the main selling season as a demand learning approach. We identity a key tradeoff between the quantity and quality of demand observations. We also find that the visibility into the timing of each sales transaction has a pivotal impact on the optimal allocation decisions and the value of merchandise tests. Finally, we consider a retailer selling an experiential product to consumers who learn product quality from reviews generated by previous buyers. The retailer maximizes profit by choosing whether to offer the product for sale to each arriving customer. We characterize the optimal product offering policies to be of threshold type. Interestingly, we find that it can be optimal for the firm to withhold inventory and not to offer the product even if an arriving customer is willing to buy for sure. We numerically demonstrate that personalized offering is most valuable when the price is high and customers are optimistic but uncertain about product quality.
Date of publication
Keyword
Subject
DOI
Identifier
Resource type
Rights statement
  • In Copyright
Advisor
  • Deshpande, Vinayak
  • Chen, Li
  • Mersereau, Adam
  • Ziya, Serhan
  • Sunar, Nur
Degree
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2015
Language
Publisher
Place of publication
  • Chapel Hill, NC
Access
  • There are no restrictions to this item.
Parents:

This work has no parents.

Items