Essays on Revenue Management and Price Competition

Abstract

This dissertation consists of three main chapters, each focusing on revenue management problems and an issue that arises in structural estimation. In Chapter 2, we consider a dynamic pricing problem in which a firm manages a limited inventory of reusable products and operates a reservation system for customers. Customers arrive randomly over time. Each arriving customer requests to use a product over a certain deterministic time. If this request is accepted, the customer pays an upfront rental fee and uses the product over the specified time. Upon the end of the rental, the product is returned and is available to serve customers in the future. The firm’s objective is to maximize the expected total revenue over a finite selling horizon by dynamically adjusting upfront rental fees. Finding the optimal policy is notoriously difficult due to the well-known curse of dimensionality. Hence, we focus on devising a heuristic pricing policy. We provide analyses of theoretical and numerical performances of the heuristic policy. In Chapter 3, we consider a well-known network revenue management problem with reusable resources. The resources occupied by a customer will be available to serve requests from other customers after a random duration of time. The goal is to find an admission policy that maximizes the expected total revenue over a finite selling horizon. Finding the optimal policy is notoriously difficult due to the well-known curse of dimensionality. Hence, we focus on devising a heuristic admission policy. We provide a performance guarantee result for the heuristic policy. We also provide results that are useful for devising a heuristic policy with a stronger performance guarantee. In Chapter 4, we address an issue that may arise when validating estimated econometric models of Bertrand competition. The issue is that a high measure of goodness-of-fit of the estimated models does not necessarily indicate a close correspondence between price data and a pure strategy Nash equilibrium. Thus, without resolving this issue, it is difficult to validate the estimated models based on goodness-of-fit tests. Despite the extensive empirical applications of Bertrand competition over the last three decades, it remains a daunting task for researchers to resolve the issue. In this chapter, we attempt to fill this gap, identifying a set of easy-to-verify and easy-to-interpret conditions under which one can validate the estimated models using goodness-of-fit tests.