Essays on Market Segmentation and Retailers' Competing Strategies

Fei Qin, Purdue University

Abstract

This dissertation focuses on exploring U.S. food retailers’ strategic interactions and the impacts on consumers. Specifically, I examine food retailers’ strategies on segmenting consumers, conducting price discrimination, and designing their product portfolio in the context of the U.S. yogurt market. The first essay examines the segmentation strategies employed by food retailers, with a focus on the use of advanced machine learning techniques (i.e., K-means clustering) to group consumers based on various characteristics, including demographics and purchase history. The second essay applies the data-driven market segmentation obtained in the first essay to a second-degree price discrimination model. The third essay relaxes the implicit assumption made in the first two essays that consumers’ choice set is fixed, and studies a non-price strategy, namely, adjusting assortment, that is adopted by food retailers in response to regulations. By analyzing the retailers’ strategies on market segmentation and responses to regulations, this dissertation aims to shed light on the strategic interactions of food retailers and consumers, and the competitive landscape of food market in general.Understanding the strategies employed by food retailers is of utmost importance in agricultural and food economics as it directly influences consumers and their purchasing decisions. The food retail industry in the U.S. is highly competitive, with retailers continuously devising tactics to attract and retain customers. Dimensions of competition such as pricing strategies, product assortment, promotional activities, and customer service can significantly impact consumers’ choices and behaviors. Investigating the strategies employed by food retailers not only provides insights into their business operations but also sheds light on how these strategies affect consumers.The first essay explores the application of machine learning methods in consumer segmentation under different information environments. Machine learning methods become popular in economic and marketing research, partly because of their flexibility in application. Although recent studies apply these advanced methods to various topics including water, housing, health, and food markets, much is less known about using machine learning methods to facilitate firms’ market segmentation decisions. Using Nielsen Consumer Panel data, I show that K-means clustering, one of the unsupervised learning methods, can be applied to conduct market segmentation. From the retailers’ perspective, incorporating more consumer information (i.e., purchase history) leads to the change in segments consumers belong to.The second essay assesses the effectiveness of data-driven market segmentation in enhancing price discrimination models. Price discrimination models are commonly adopted by firms to optimize revenue and profitability by customizing prices to different customer segments. Existing studies often rely on exogenous assumptions for consumer segmentation, which may or may not be applicable in practice. This study advances the existing literature by replacing the consumer segment assumption with data-driven market segmentation obtained through K-means clustering. The results are then applied to the second-degree price discrimination model to analyze how sensitive the firms optimal profits are under different consumer information environments. The findings reveal that adding consumer information to consumer segment leads to a more inelastic demand for the consumer segments and an increase in firm’s profits.

Degree

Ph.D.

Advisors

Martin, Purdue University.

Subject Area

Behavioral psychology|Food Science|Psychology

COinS