Search

Performance measurements are the set of measures used to determine how well an organisation is doing, giving direction for business improvement and growth. The study draws on existing literature on the adoption, approach and implementation of performance measurements systems within the small and medium-sized independent hotel accommodation. However, existing studies show a gap in understanding what constitutes business performance in SMEs despite its importance. To this end, this research, therefore, explores the perception of owners/ managers of PM concept, approaches and implementation to achieve the overall objective of an organisation. Four research objectives provided guidelines for this exploratory research, using a phenomenological research approach. The study has several conclusions based on documents analysis and 30 semi-structured interviews with owners and managers in the hotel accommodation sector. This research found that different definitions and approaches to performance measurement support the multi-dimensional aspect of performance measurement. The different dimensions of performance measurements were considered paramount to understanding the practicality of the concept. This study argued that small and medium-sized independent hotel accommodations adopt performance measurements in practice in different formats to enhance organisational performance. This research contributes theoretically by extending PM literature to understand the definition, designing and implementing performance measurement in small and medium independent hotels. Also, the phenomenological research method approach offers a key insight into the study through a life experience from owners and managers, providing valuable insight to the study. More so, for practical implication, this research identified and complied the performance indicators used in the industry. This information was used to develop a suitable framework for the small and medium-sized independent accommodations. The Performance Steering-Wheel and the Practical Performance Indicators would help accommodation owners/managers be consistent in their performance measurement approach.

Strategic purchasing behaviour has received growing attention in the field of revenue management. Its occurrence potentially hurts providers' revenue with substantial losses. The need to detect strategic customers and predict their decisions has been highlighted by researchers. Theoretical models rely on assumptions about how customers make decisions and what factors influence them. Conditioned experiments are relatively expensive and not representative of the actual system. By comparison, approaches based on statistics and machine learning from historical data can be relatively cheap, representative of actual conditions, and data-based. However, the widely used approaches pose certain challenges, such as interpretability, overfitting, and stability. These may influence their ability to classify - that is, predict customer types and decisions. We propose a conceptual framework and data linkage for detecting strategic customers and predicting customer decisions. The proposed framework and data linkage were developed based on cancel-rebook behaviour by two customer types, namely strategic and myopic. They were also based on two customer decisions: buy or wait. The evidence showed that the input variables in the framework were good predictors of customer types and decisions. Ultimately, we propose a new approach, namely a hierarchical rule-based inferential modelling and prediction, to integrate statistical analysis, rule-based inference, maximum likelihood prediction, and machine learning in a hierarchical structure. The referential value-based discretisation technique used in this approach can alleviate information loss and distortion as an effect of over-generalisation caused by discretisation. It also captures the structure of the data better than other discretisation technique. We used belief-rule-based inference to analyse the relationship between inputs and outputs. An interdependence index was used to measure the relationship between input variables. The hierarchical structure deals with sparse matrices by decomposing input variables into several groups of evidence. The outputs generated by all groups of evidence are then combined to obtain a final inference. The classifiers, developed based on maximum likelihood evidential reasoning (MAKER) framework and a hierarchical rule-based inferential modelling and prediction, are transparent and interpretable. The classifiers perform better than the majority of alternative classification models for both datasets (customer types and customer decisions). Their performance is similar to that of classification trees.

We consider the problem of finding profit-maximising prices for delivery time slots in the context of attended home delivery. This is a multi-stage optimal control problem that admits a dynamic programming formulation. However, this formulation is intractable for realistic problem sizes due to the so-called "curse of dimensionality''. To facilitate an approximation of this problem, we characterise some mathematical properties of the dynamic program: We show that the underlying Bellman operator has a unique fixed point and provide a closed-form expression for it. Moreover, we show that - under certain technical assumptions - the value function, which has a discrete domain and a continuous codomain, admits a continuous extension, which is a finite-valued, concave function of its state variables, at every time step. Motivated by these results, we develop an approximate dynamic programming algorithm for a more general class of problems, characterised by value functions that are concave extensible and submodular in their state-space. Our new algorithm computes deterministic upper and stochastic lower bounds of the value function in the realm of dual dynamic programming. We show that the proposed algorithm terminates after a finite number of iterations. Moreover, we also derive probabilistic guarantees on the value accumulated under the associated policy for a single realisation of the dynamic program and for its expected value. Furthermore, we compare our algorithm with two other approximate dynamic programming solutions proposed in the literature. Our analysis entails a theoretical examination from a control-theoretical perspective and a parametric numerical case study. Our experiments are based on real-world data, from which we generate multiple scenarios to stress-test the robustness of the pricing policies to errors in model parameter estimates. Our analysis shows that the proposed algorithm dominates the others with respect to computation time and profit-generation capabilities of the delivery slot pricing policies that it generates.

The application of revenue management (RM) is changing more rapidly than ever before, driven as an important factor of the daily operation to keep prices competitive and to create real-time optimal pricing. In the age of the Internet and social media, negotiated fixed rates have become outmoded. Consumers now have access to online rate comparisons and real time reviews. They think more strategically when making purchasing decisions. Thus, they become more demanding. This research provides an empirical study of revenue management and pricing with an emphasis given to the hospitality industry. The aim of this research is to examine the gap between the theoretical approach and the empirical analysis, the rationality between the implementation of dynamic pricing approaches and the impact on the customer. Furthermore, the research examines the perception of consumers’ willingness to pay when using the Name-Your-Own-Price (NYOP) mechanism, which allows customers to have a greater influence on the amount they are prepared to pay. Instead of posting a price, the seller waits for a potential buyer’s offer, which he or she can either accept or reject. Finally, this study examines, whether the use of social media plays a decisive role in the online purchase environment used by the hospitality sector and the effect it has on a consumer’s willingness to pay. Accordingly, hotel revenue managers will be able to use the findings of this study to effectively plan their short-term, and long-term pricing strategies to generate a stronger revenue management performance for their property, namely to increase the RevPAR (revenue per available room). The research can be useful to businesses, as empirical data and tests were employed to determine what kind of impact the different pricing policies have on the long-term profit optimization. These practical and theoretical elements of the field reinforce each other‚ as well as to a large extent, the constructive interplay of theory and practice. The research is twofold, the holistic approach, which discusses the development of the theoretical dimension, is complemented by the practical analysis of the collected data of the surveys. This approach ensures the relevant observation of ‘real-time’ data and the evaluation of the set of hypotheses. The study conducted two large scale interrelated structured surveys. The first structural survey (NYOP) provides a better understanding of the final consumer, by using the name-your-own-price mechanism and by observing the extended role of social media in the booking procedure. Hypotheses were tested and in the second survey in-depth data from revenue managers and executives working across the tourism industry was collected, in an attempt to measure the use of pricing strategies within the industry. The research contributes to the theory by empirical testing how the extended RM objectives influence RM and pricing. It provides a clear picture of the necessary elements for a successful implementation of pricing strategies. Finally, the study has implications for the consumer. Thus, the researcher investigates consumer’s perception to the NYOP model and the expanding role of social media to the consumer-booking pattern.

Resource allocation decisions are central to efficiently managing limited resources. In this dissertation, we design algorithms for three different problems that involve the allocation of reusable resources. The goal of each work is to provide simple-to-use approaches with some measure of performance, either theoretical, empirical, or both. Two of the problems studied focus on healthcare operations and are related to the management of hospital beds. The third problem is related to revenue management and is based on the challenge of finding an ideal price to sell products. The first technical chapter delves into a hospital care unit placement problem. In this setting, patients derive varying rewards from potential care unit placements. We devise an algorithm to learn these rewards while simultaneously making placement decisions based on available capacity. The demand and occupancy time for this problem are not learned but instead assumed always to be unknown. The second technical chapter addresses the issue of learning demand and service rates by considering how a firm can determine the optimal pricing for multiple bundled rental products. A key challenge here is the loss of customers if their desired product is not fully available. We design a dynamic pricing algorithm for this problem that learns about the demand rate and service rates in this stochastic system over time to identify a set of pricing options for every product. The final technical chapter showcases the deployment of a decision-support tool that makes use of inputs from data-driven algorithms. Specifically, we tackle the issue of coordinating how hospital decision-makers can optimally share a pool of available bed resources to make patient placement decisions. We devise a matching algorithm for this problem and implement it in a large academic hospital. A mix of theoretical and empirical analysis supports our work in this dissertation. Each of the technical chapters highlights challenging aspects of managing reusable resources in realtime settings and provides evidence that simple methods have the potential to be effective for complex problems.

Revenue management is the study of models and algorithms that address inventory allocation and pricing decisions in the face of uncertainty and limited capacities. Out of many models studied in revenue management, fluid approximations have been an effective tool to deal with large-scale inventory allocation and pricing problems and develop efficient algorithms; while inventory placement, which determines how to place on-hand products into different fulfillment centers, is an important decision that has a significant impact on subsequent inventory allocation and pricing power; lastly, discrete choice models are widely used to model customer choice behaviour and capture the fact that customers substitute among the offered products, which ultimately help in making better inventory allocation and pricing decisions. In this dissertation, we investigate one new model in each of these areas and develop efficient algorithms with performance guarantees. These new models represent various approaches from different angles we take to aid in improving inventory allocation and pricing decisions. First, based on the fact that high-variance demand occurs in many applications but is not fully addressed by traditional revenue management models, we explore a new revenue management model that incorporates general mean and variance for the number of customer arrivals, with the goal of developing a policy to determine which product to make available to each arriving customer in order to maximize total expected revenue. We devise a fluid approximation corresponding to this model and use it to develop an asymptotically optimal policy. Second, we consider inventory placement, delivery promise and fulfillment decisions faced by an online retailer jointly. We study a two-stage model where in the first stage, we place a set of products with given numbers of units into different fulfillment centers with capacity constraints. Once we make the placement decisions, we enter the second stage where we face random demand for the products from different demand regions. In response to each demand, we pick a delivery promise to offer and choose a fulfillment center to use to serve the demand. The goal is to determine where to place the units in order to maximize the total expected profit from sales over a finite selling horizon. For this problem, we provide a general approximation framework, which leads us to a set of policies. The best policy provides $1/(4+\epsilon)$-approximation for any $\epsilon>0$, and the policy can be computed in polynomial time for each fixed $\epsilon$. Lastly, we study a natural variant of the multinomial logit model by incorporating rank cutoffs, which characterizes the number of products customers will focus on during the choice process. To be more specific, after associating random utilities with all the products and the no-purchase option, a customer with rank cutoff k would ignore all alternatives whose utilities are not within the k largest utilities and choose among the remaining alternatives. We show that the assortment optimization problem under this choice model is NP-hard and propose a polynomial-time approximation scheme. We also run numerical experiments to show that incorporating rank cutoffs can result in better predictions of customer choices and more profitable assortment recommendations.

Over the years, last-minute cancellations and no-shows have presented difficulties for the hospitality sector, negatively affecting hotel operations and revenue. The cancellation policies of hotels have changed significantly as a result. These rules have been modified to act as "price fences" as a result of revenue management strategies used in the hotel industry. This study aims to find out how various cancellation policies impact consumers' perceptions of risk and intent to reserve hotel rooms at various times. The study provides insight into the factors that affect consumer behavior and choices for hospitality goods and services by examining these elements. The results of the study offer some understanding into how hotels and other hospitality businesses can use cancellation policies to customize their marketing strategies to meet consumer needs and preferences while successfully managing revenue and profitability.

This dissertation presents a comprehensive analysis of human behaviors with an emphasis on practical methods that generate a positive societal impact with regards to driving safety and business management decisions. Human behaviors were explored through three major works: (1) automatic clustering of kinematic driving data patterns to enhance the understanding of drivers' normal maneuvering behaviors; (2) detecting distracted driving behaviors to enhance human driving safety through the fusion of kinematic data and driving models; and (3) modeling and predicting consumers' behaviors to assist airline management decisions. The major contributions are summarized as follows. The first work introduces a new multi-model fusion and detection algorithm through effectively synthesizing three different state-space models, each of which represents a driver's typical kinematic motion pattern. The kinematic motion trajectory can be dynamically updated and predicted by using the Kalman filtering method. A normalized likelihood is further proposed for model fusion by weighing the contribution of each kinematic motion model. The results are then used to build two monitoring control charts. Specifically, both an Exponentially Weighted Moving Average (EWMA) control chart and a Cumulative Sum (CUSUM) control chart are developed to automatically detect and assess distracted driving behavior. This work is attractive in that it can be used to generate real-time warning signals during driving to avoid risky distracted driving behaviors. The second major work aims to identify typical drivers’ normal maneuvering behaviors from a collected naturalistic driving dataset. This is done by improving the existing Hierarchical Dirichlet Process Hidden Semi-Markov Model (HDP-HSMM) inference method. In literature, a HDP prior can cause consistency issues, which harms the estimation of duration probabilities, transition probabilities, and total number of states estimated. Many solutions have been proposed to counter this issue; however, none have been from the perspective of HDP-HSMMs. This research presents a robust algorithm called rHDP-HSMM, which adds a state-merging procedure to the inference to avoid redundant states. As a result, the proposed rHDP-HSMM leads to more consistent states, faster convergence, and better parameter estimation than the existing HDP-HSMM method. The proposed rHDP-HSMM model is then used to enhance the understanding and inference of drivers’ maneuvering behaviors from a naturalistic dataset. The results can be served as a surrogated model to generate realistic simulation data for testing autonomous vehicles. The third major work combines concepts from economics and constrained Gaussian Process (CGP) regression to probabilistically describe consumer purchasing behaviors. System-wide aggregation methods are often used to model customer buy-down/sell-up behaviors, however, the resulting estimates become uncharacteristic of the original markets of interest. This work addresses this issue by applying the multitask learning strategy to a CGP model, in which all fare groups' data are shared to estimate individual fare group demand simultaneously. Specifically, adding appropriate airline demand constraints facilitates an inter-group cross-learning capability. The proposed CGP model enables an accurate estimation of customers' purchasing behaviors in fare groups that have limited historical purchasing data. Buy-down and sell-up behaviors are then estimated by decomposing demand into three different states that consider which fare groups were available at the time of purchase. As such, this work offers both modeling contributions and a better understanding of customer purchasing behaviors in airline applications. Collectively, the contributions from this dissertation have advanced the modeling of human behaviors across the transportation and business management fields.

The censored demand estimation problem has been well studied over the past 20 years. Theseminal paper is Talluri and Van Ryzin 2004, which models demand as the product of an unobserved arrival rate and purchase probability, and uses Expectation Maximization (EM) for estimation. In this work, I apply their model to the case of a small hotel with a single product, and adapt the model to use a time-varying arrival rate. Modeling the arrival rate as a step function, the choice of where the steps fall is a crucial hyperparameter, so I propose a heuristic algorithm, the Step Variance Minimization Heuristic (SVMH), to specify the size and location of the steps in the arrival rate function. On both real and simulated hotel data, SVMH is superior in terms of both in-sample and out-of-sample mean squared error when compared to (a) a constant arrival rate and (b) evenly-spaced steps of the same num- ber. To improve the convergence time of EM, I introduce a variant algorithm, Regularized Expectation Maximization (REM), that uses 81% fewer iterations than EM, on average. I show that REM is a Generalized Expectation Maximization (GEM) algorithm, and thus has similar theoretical properties to EM. I also conducted multiple numerical studies that show that models estimated using REM perform similarly in-sample and out-of-sample to their EM counterparts, and produce nearly identical revenue curves. Thus, the convergence improvements offered by REM do not come at a cost to good parameter estimation.

Recent trends point to increasing use of social network information by firms and public agencies for personalized policies. However, the cost of implementation can be high and the use of personal information can reduce satisfaction. The value of such policies depends upon network structures, and may be insignificant for certain classes of large networks. Thus, firms and public agencies may need to be more careful about the design of mechanisms within social networks. In this thesis, we focus on a particular application of mechanism design problems with social network effects, i.e., the pricing problem of a single firm selling a product to consumers in social networks, and study the value of price discrimination in large social networks. Recent trends in industry suggest that increasingly firms are using information about social network to offer personalized prices to individuals based upon their positions in the social network. In the presence of positive network externalities, firms aim to increase their profits by offering discounts to influential individuals that can stimulate consumption by other individuals at a higher price. However, the lack of transparency in discriminative pricing may reduce consumer satisfaction and create mistrust. Recent research has focused on the computation of optimal prices in deterministic networks under positive externalities. We would like to answer the question: how valuable is such discriminative pricing? We find, surprisingly, that the value of such pricing policies (increase in profits due to price discrimination) in very large random networks are often not significant. Particularly, for Erd\"{o}s-Renyi random networks, we provide the exact rates at which this value decays in the size of the networks for different ranges of network densities. Our results show that there is a non-negligible value of price discrimination for a small class of moderate-sized Erd\"{o}s-Renyi random networks. We also present a framework to obtain bounds on the value of price discrimination for random networks with general degree distributions and apply the framework to obtain bounds on the value of price discrimination in power-law networks. Our numerical experiments demonstrate our results and suggest that our results are robust to changes in the model of network externalities.

In this dissertation, we focus on revenue management practices in the settings where the traditional assumptions are challenged because of the rise of social technologies in the new age. We develop theoretically sound and practical policies (pricing, inventory, and information signals) when the firm faces uncertainty, dependency and non-stationarity in market demand. We also provide effective insights into the firm's decision making. The first part of the dissertation is motivated by the challenges of decision making for a new product. We study the interplay between pricing and learning for a monopolist whose objective is to maximize the expected revenue of a new product over a finite time horizon. We consider a setting where a firm can learn by observing sales data at (different) prices over time. Its customers are not forward-looking. To capture the adoption process with demand learning, we develop a continuous-time Markovian Bass model where the adoption rate depends on the selling price and on the past sales. We show that a firm can apply real-time demand data to update the demand parameters and optimize the price accordingly. We propose two simple and computationally tractable pricing policies that are provably near-optimal. Our framework is sufficiently flexible to be applied to real-time control optimization and parameter learning with general network effect in demand process. In the second part of the dissertation, we consider a setting when underlying demand is uncertain and follows a complex stochastic process, which makes pricing problems difficult to solve. In such cases, certainty equivalent (CE) policies, based on solving the deterministic relaxation of a stochastic pricing problem, can be used as practical alternatives. CE policies have lighter computational and informational requirements compared to solving the problem to optimality. This is particularly true when the firm does not have complete information about the underlying demand distribution. While the effectiveness of CE pricing policies has been theoretically studied in some settings (e.g, independent demand), the performance of CE policies are not known in general settings. This paper analyzes the performance of CE policies in a pricing problem (for a given inventory level) where future demand depends on sales and inventory and the firm has limited opportunities to change price. We show that CE policies are asymptotically optimal: as the problem scale (denoted by m) becomes large, the percentage regret decreases at the rate of 1/sqrt{m}. We also extend the result to the joint pricing and (initial) inventory problem. Our numerical results are even more promising. Even in non-asymptotic settings (small scaling factor and a few price changes), CE policies perform well and often result in revenues that are only a few percentage points lower than optimal. The third part of the dissertation is an on-going project which studies the cooperated relationship between a seller and an social influencer where the influencer has a revenue-sharing contract with the seller (so that their incentives are aligned). We study the following questions: (i) for the seller, how to compensate the influencer? (ii) Who is the ideal influencer to post about a given product? (iii) What attributes of an influencer make her more or less valuable to a company? The customers are assumed to be Bayesian learners so they can infer the state of the product from available information.

In online retailing, core operational questions include deciding on the assortment of products that should be displayed to incoming customers and how to adjust these assortments and prices in real-time according to current capacity and observed demand. In this dissertation, we explore three new developments in online retailing and develop models and techniques for revenue management in these settings. The underlying goal in each of the chapters is to develop computationally efficient and provably good algorithms. First, we consider a revenue management model over a single flight leg, where customers are allowed to lock an available fare. The goal is to find a policy to decide which set of fare classes to make available at each time period to maximize the total expected revenue. We develop an approximate policy that is guaranteed to obtain at least half of the optimal total expected revenue. Second, we consider dynamic assortment problems with reusable products, in which each arriving customer chooses a product within an offered assortment, uses the product for a random duration of time, and returns the product back to the firm to be used by other customers. The goal is to find a policy for deciding on the assortment to offer to each customer so that the total expected revenue over a finite selling horizon is maximized. We present an approximate policy that is guaranteed to obtain at least half of the optimal total expected revenue. The policy is based on constructing linear approximations to the optimal value functions. Lastly, we examine the revenue-utility assortment problem with the goal of finding an assortment that maximizes a linear combination of the expected revenue of the firm and the expected utility of the customer. This criterion captures the tradeoff between the firm-centric objective of maximizing the expected revenue and the customer-centric objective of maximizing the expected utility. When customers choose according to the multinomial logit model, and there is a constraint on the offered assortments characterized by a totally unimodular matrix, finding the optimal assortment requires solving a nonconcave optimization problem. We provide an efficient algorithm for finding the welfare optimal assortment by solving a parametric linear program and testing a polynomial size collection of candidate assortments.

This dissertation studies several problems in revenue management involving dynamic pricing, assortment selection, and their joint optimization, through demand learning. The setting in these problems is that customers’ responses to selling prices and product displays are unknown a priori, and the only information the decision maker can observe is sales data. Data-driven optimizing-while-learning algorithms are developed in this thesis for these problems, and the theoretical performances of the algorithms are established. For each algorithm, it is shown that as sales data accumulate, the average revenue achieved by the algorithm converges to the optimal. Chapter 2 studies the problem of context-based dynamic pricing of online products, which have low sales. For these products, existing single-product dynamic pricing algorithms do not work well due to insufficient data samples. To address this challenge, we propose pricing policies that concurrently perform clustering over products and set individual pricing decisions on the fly. By clustering data and identifying products that have similar demand patterns, we utilize sales data from products within the same cluster to improve demand estimation for better pricing decisions. We evaluate the algorithms using regret, and the result shows that when product demand functions come from multiple clusters, our algorithms significantly outperform traditional single-product pricing policies. Simulations with both synthetic and real data from Alibaba show that our algorithm performs very well, and a field experiment at Alibaba shows that our algorithm increased the overall revenue by 10.14%. Chapter 3 investigates an online personalized assortment optimization problem where customers arrive sequentially and make their choices (e.g., click an ad, purchase a product) following the multinomial logit (MNL) model with unknown parameters. We develop several algorithms to tackle this problem where the number of data samples is huge and customers’ data are possibly high dimensional. Theoretical performance for our algorithms in terms of regret are derived, and numerical experiments on a real dataset from Yahoo! on news article recommendation show that our algorithms perform very well compared with benchmarks. Chapter 4 considers a joint assortment optimization and pricing problem where customers arrive sequentially and make purchasing decisions following the multinomial logit (MNL) choice model. Not knowing the customer choice parameters a priori and subjecting to a display capacity constraint, we dynamically determine the subset of products for display and the selling prices to maximize the expected total revenue over a selling horizon. We design a learning algorithm that balances the trade-off between demand learning and revenue extraction, and evaluate the performance of the algorithm using Bayesian regret. This algorithm uses the method of random sampling to simultaneously learn the demand and maximize the revenue on the fly. An instance-independent upper bound for the Bayesian regret of the algorithm is obtained and numerical results show that it performs very well.

A critical aspect of revenue management is a firm's ability to predict future demand. Historically hotels have used pick-up based models owing to the complexities of trying to build casual models of demands. Machine learning approaches are slowly attracting attention owing to their outstanding predicting power and flexibility in modeling relationships. This study provides an overview of approaches to forecasting hospitality demand using machine learning models, including Neural Network, Nearest Neighbors, Tree, and Support Vector Machine. The out-of-sample performances of the above approaches are illustrated by using two sets of data: one from a single hotel with long booking windows up to 12 months, the other from 24 hotels with 14 days advanced bookings and additional information including pricing, location, etc. This research appears to be the first study in academia applying machine learning approaches in hotel demand forecast. The empirical findings prove that machine learning approaches outperform traditional models, especially given long booking history. The proposed models are valuable for practitioners in improving forecast accuracy and optimizing revenue, and lay the groundwork for future research into refining machine learning models in hotel revenue management.

This dissertation consists of two independent essays that share the common goal of predicting and influencing consumer demand: (i) Scheduling Advertising on Television; and (ii) Peer Effects in the Diffusion of Solar Panels: A Dynamic Discrete Choice Approach. These essays are data-driven problems of operations and management science, which we address using optimization and econometrics.

We study three assortment optimization problems that reflect the current retail environment. The goal of these problems is to compute an assortment which maximizes the retailer’s expected revenue or profit, depending on the scenario. For each problem, we present efficient algorithms to compute the optimal assortment, or prove that the problem is NP-hard and provide an approximation algorithm. First, we study assortment optimization for an omnichannel retailer, who operates both an online store and a physical store offline. The retailer selects a subset of products from the online store to offer in the physical store, knowing that customers can visit the physical store to observe products and update preferences for similar products offered online. We introduce a features tree to organize products by their shared features, and describe how customers update their preferences for online products. Second, we study assortment optimization when certain products look more attractive if they are offered alongside in the assortment. For example, by observing a lower quality product, customers may perceive an increase in value for a target product. We call this effect synergy and introduce a synergy graph to describe the synergy structure between products. We classify the difficulty of the synergistic assortment optimization problem based on the synergy graph. Finally, we study assortment optimization with dynamic substitution and inventory stocking costs. This problem considers customers who arrive over a finite selling horizon to make a purchasing decision amongst the products that have not stocked-out. The retailer does not have the ability to modify the assortment seen by each arriving customer. Hence, the assortment that each customer purchases from depends on the initial assortment and stocking decisions of the retailer, as well as the purchasing decisions of the preceding customers.

Travel booking behavior has changed substantially over the past two decades. The emergence of new technology and online intermediaries has provided travelers with the flexibility to book up until the date of stay. This has created a fast-paced, dynamic booking environment that disrupts traditional revenue management strategies focused on pricing and allocating rooms based on the time of purchase. The study explores the joint effects of technology and the economy on booking window lead times. It also evaluates a range of forecasting techniques and the importance of utilizing the booking curve for forecasting in dynamic booking environments.

This dissertation focuses on optimizing inventory and pricing decisions in the online retail industry. Motivated by the importance of great customer service quality in the online retail marketplace, we investigate service-level-constrained inventory control problems in both static and dynamic settings. The first essay studies multi-period production planning problems (with or without pricing options) under stochastic demand. A joint service-level constraint is enforced to restrict the joint probability of having backorders in any period. We use the Sample Average Approximation (SAA) approach to reformulate both chance-constrained models as mixed-integer linear programs (MILPs). Via computations of diverse instances, we demonstrate the effectiveness of the SAA approach, analyze the solution feasibility and objective bounds, and conduct sensitivity analysis. The approaches can be generalized to a wide variety of production planning problems. The second essay investigates the dynamic versions of the service-level-constrained inventory control problems, in which retailers have the flexibility to adjust their inventory policies in each period. We formulate two periodic-review stochastic inventory models (backlogging model and remanufacturing model) via Dynamic Programs (DP), and establish the optimality of generalized base-stock policies. We also propose 2-approximation algorithms for both models, which is computationally more efficient than the brute-force DP. The core concept developed in our algorithms is called the delayed marginal cost, which is proven effective in dealing with service-level-constrained inventory systems. The third essay is motivated by the exploding use of sales rank information in today's internet-based e-commerce marketplace. The sales rank affects consumers' shopping preference and therefore, is critical for retailers to utilize when making pricing decisions. We study periodic-review dynamic pricing problems in presence of sales rank, in which customers' demand is a function of both prices and sales rank. We propose rank-based pricing models and characterize the structure and monotonicity of optimal pricing policies. Our numerical experiments illustrate the potential of revenue increases when strategic cyclic policy is used.

In the past two decades, retailers have witnessed rapid changes in markets due to an increase in competition, the rise of e-commerce, and ever-changing consumer behavior. As a result, retailers have become increasingly aware of the need to better coordinate inventory control with pricing in order to maximize their profitability. This dissertation was motivated by two of such problems facing retailers at the interface between pricing and inventory control. One considers inventory control decisions for settings in which planned prices fluctuate over time, and the other considers pricing of multiple substitutable products for settings in which customers hold inventory as a consequence of stockpiling when promotional prices are offered.In Chapter 1, we provide a brief motivation for each problem. In Chapter 2, we consider optimization of procurement and inventory allocation decisions by a retailer that sells a product with a long production lead time and a short selling season. The retailer orders most products months before the selling season, and places only one order for each product due to short product life cycles and long delivery lead times. Goods are initially stored at the warehouse and then sent to stores over the course of the season. The stores are in high-rent locations, necessitating efficient use of space, so there is no backroom space and it is uneconomical to send goods back to the warehouse; thus, all inventory at each store is available for sale. Due to marketing and logistics considerations, the planned trajectory of prices is determined in advance and may be non-monotonic. Demand is stochastic and price-dependent, and independent across time periods. We begin our analysis with the case of a single store. We first formulate the inventory allocation problem given a fixed initial order quantity with the objective of maximizing expected profit as a dynamic program and explain both technical and computational challenges in identifying the optimal policy. We then present two variants of a heuristic based on the notion of equalizing the marginal value of inventory across the time periods. Results from a numerical study indicate that the more sophisticated variant of the heuristic performs well when compared with both an upper bound and an industry benchmark, and even the simpler variant performs fairly well for realistic settings. We then generalize our approaches to the case of multiple stores, where we allow the stores to have different price trajectories. Our numerical results suggest that the performance of both heuristics is still robust in the multiple store setting, and does not suffer from the same performance deterioration observed for the industry benchmark as the number of stores increases or as price differences increase across stores and time periods. For the pre-season procurement problem, we develop a heuristic based on a generalization of the newsvendor problem that accounts for the two-tiered salvage values in our setting, specifically, a low price during end-of-season markdown periods and a very low or zero salvage value after the season has concluded. Results for numerical examples indicate that our modified newsvendor heuristic provides solutions that are as good as those obtained via grid search.In Chapter 3, we address a retailer's problem of setting prices, including promotional prices, over a multi-period horizon for multiple substitutable products in the same product category. We consider the problem in a setting in which customers anticipate the retailer's pricing strategy and the retailer anticipates the customers' purchasing decisions. We formulate the problem as a two-stage game in which the profit maximizing retailer chooses prices and the utility maximizing customers respond by making explicit decisions regarding purchasing and consumption, and thus also implicit decisions regarding stockpiling. We incorporate a fairly general reference price formation process that allows for cross-product effects of prices on reference prices. We initially focus on a single customer segment. The representative customer's utility function accounts for the value of consumption of the products, psychological benefit (for deal-seekers) from purchasing at a price below his/her reference price but with diminishing marginal returns, costs of purchases, penalties for both shortages and holding inventory, and disutility for deviating from a consumption target in each period (where applicable). We are the first to develop a model that simultaneously accounts for this combination of realistic factors for the customer, and we also separate the customer's purchasing and consumption decisions. We develop a methodology for solving the customer's problem for arbitrary price trajectories based on a linear quadratic control formulation of an approximation of the customer's utility maximization problem. We derive analytical representations for the customer's optimal decisions as simple linear functions of prices, reference prices, inventory levels (as state variables), and the cumulative aggregate consumption level (as a state variable). We then embed the consumer's optimal policy (in analytic form) in the retailer's profit maximization problem and show that the retailer's problem can be reformulated as a quadratic program that can be solved numerically. Despite the additional generality of our problem context, our solution methodology is computationally tractable. Results for numerical examples indicate that the separation of the customer's two types of decisions turns out to be important in inducing realistic purchasing patterns, including demand spikes like those observed in practice and the associated stockpiling, as well as pre- and post-promotion dips, which are also observed in practice. We discuss how our approach can be extended to multiple customer segments, limitations of our approach, and future research directions.

This thesis studies firm’s operational decisions in strategic settings. In particular, it focuses on a firm’s dynamic pricing or assortment optimization problems, in the presence of competitors or customers that respond to the firm’s actions strategically. We study three problems in this area with different features. We optimize over the firm’s strategies, and characterize the outcome of the game between the firm and its strategic counter-parties.

We are experiencing a time in which advances in information technology change everything around us. Almost every aspect of our work and life, from daily travel to grocery shopping, is now or will soon be connected to the internet. This drastic change has led to the emergence of many new products that display network effects, meaning that individual consumers value a product more when more other consumers use the same product. At the same time, development of data storage/processing tools offers a great chance for many firms to collect large amounts of data and understand their customers' purchasing behavior. For these firms, a significant challenge --- as well as opportunity --- is to develop models of customer choice behavior that incorporate network effects and to use such models to make better pricing decisions. This dissertation addresses the development and analysis of some such models. We consider a seller's problem of determining revenue-maximizing prices for an assortment of products that exhibit network effects. Customers make purchase decisions according to a multinomial logit (MNL) choice model, modified --- to incorporate network effects --- so that the utility each individual customer gains from purchasing a particular product depends on the market's total consumption of that product. In the setting of homogeneous products, we show that if the network effect is comparatively weak, then the optimal pricing decision of the seller is to set identical prices for all products. However, if the network effect is strong, then the optimal pricing decision is to set the price of one product low and to set the prices of all other products to a single high value. This pricing scheme boosts the sales of the single low-price product in comparison to the sales of all other products. The analysis is also extended to settings with heterogeneous products, and we show that optimal solutions have a structure similar to that found in the homogeneous case: either maintain a semblance of balance among all products, or boost the sales of just one product. Based on this structure, we propose an effective computational algorithm for such general heterogeneous settings. Subsequently, we study the preceding pricing problem from a robust optimization perspective. Unlike the classical MNL model where products' prices and sales have a one-to-one correspondence, in the MNL model with network effects a fixed set of prices may not uniquely determine sales. This occurs because, for given prices, sales arise as the solution to an equilibrium condition. In some cases, there may be multiple sales levels that satisfy the equilibrium condition. Among those sales equilibria corresponding to a given set of prices, we call the one with the highest revenue the ``optimistic" equilibrium, and the one with the lowest revenue the ``pessimistic" equilibrium. In our initial study mentioned in the previous paragraph, we implicitly took an ``optimistic" approach. We next take the pessimistic attitude and study the revenue-maximizing problem in the pessimistic setting. In the case that there is only one product to sell, the problem has the same pessimistic optimal price as the optimistic one, when the network effect is relatively weak. However, when the network effect is strong, the optimal policy requires the seller to become more conservative and the pessimistic optimal price is lower than the optimistic one. In the case that there are two products, the structure of equations at equilibrium becomes more complicated and we are not able to derive an analytical solution. To numerically solve this problem, two directions for finding the pessimistic optimal solution are proposed: divide and search, and linear relaxation. In addition, our numerical studies show that when the network effect is strong, the revenue from offering exactly one product is almost as good as that from selling an assortment of multiple products. This suggests that a company selling multiple products with strong network effects may be wise to simply offer just one product.

We include three works on resource allocation in this thesis. Proper pricing helps the system to allocate resource efficiently. However, the computational efforts to obtain such prices are not always easy. In airline network revenue setting, high-quality bid prices can be obtained via approximate dynamic programming approach. However, this leads to an exponential-size linear program which is slow to solve. We show that this large linear program is actually equivalent to a compact linear program with appealing interpretation by exploiting the minimum-cost network flow structure of this problem. Computational experiments indicate that our results can speed up the computation for the approximate linear programming approach by a factor ranging between 13 and 135. In the second chapter, we study how to better serve clients by allocating resources to open facilities with right location and type. We extend the classical facility location problem by associating types with the facilities and the clients. The types define a partial ordering and a client can only be served by facilities of equal or higher type. We show how to obtain approximation algorithms for two interesting special cases. We also study the problem embedded in time, which is known as the lot-sizing variant. We give algorithm that finds optimal solution for the variant. In the third chapter, we study how to allocate medical resource in imaging facilities. The most common method to allocate imaging machine time is by a appointment system. However, despite the best efforts by medical staff, the inherently stochastic nature of the duration of various procedures and unexpected disruptions can bring considerate congestion to the system. In the multi-site setting, we seek to complement appointment scheduling with real-time resource sharing. By monitoring system status and making prediction of future, we can identify potential congestions and divert patients accordingly to prevent these congestions from happening. We provide simulation results showing that this method improves patient waiting time and facilities overtime, which are two important quality measurements.

The long-term permanency and profitability of a firm requires decisions to be made wisely and on time. For this purpose, it is essential to consider all aspects of a decision in terms of its impact on revenue, planning, scheduling, and inventory in an integrated framework.In this research paper, an interactive intelligent decision support system for making an integrated decision in the presence of demand uncertainty is proposed. The system operates in a multi-product, multi-period setting, and its objective is to maximize the profit of the firm over time. To achieve its objective, the system first obtains the optimal price and capacity plan for the coming periods. The output of this first step becomes the input for the second step, in which the problem of scheduling is solved. At the end, based on the scheduling step, the optimum inventory policy is determined.To cope with demand uncertainty in the pricing and planning phase, a robust optimization model is proposed in which the demand is considered to belong to an interval and there is no knowledge (such as statistical distribution) associated with thedemand. The robust problem is solved using a metaheuristic.During the scheduling step, a general setting for the problem is considered, in which each product is treated like a project with a flow network. To address the problem of scheduling, a simulation optimization method is applied in which the optimization step determines the dispatch rule of the jobs and the simulation step schedules the dispatched jobs on the production line.During the inventory step, the system obtains the best schedule for ordering and storing the raw material in order to minimize the inventory cost. For this purpose, a mixed integer mathematical model is proposed and a metaheuristic is applied to obtain the best solution.All modules of the proposed decision support system are supported with a database that stores the data obtained from the shop floor and the market. This database is used to assess the costs and parameters in models by applying a cost estimation support system.To evaluate the effectiveness of the proposed decision support system, it has implemented in a small size textile production line. The data generated by the system and its users are analyzed for a period of four months. Four indicators of profit per product, overall equipment effectiveness, percentage of realized schedule and work-in-progress are monitored during these four months and their values are compared against the same time period in previous year. The results show that the system has improved in terms of profitability, equipment effectiveness and production line control. However, the work-in-progress has not improved.

This thesis handles a fundamental problem in retail: given an enormous variety of products which does the retailer display to its customers? This is the assortment planning problem. We solve this problem by developing algorithms that, given input parameters for products, can efficiently return the set of products that should be displayed. To develop these algorithms we use a mathematical model of how customers react to displayed items, a customer choice model. Below we consider two classic customer choice models, the Multinomial Logit model and Nested Logit model. Under each of these customer choice models we develop algorithms that solve the assortment planning problem. Additionally, we consider the constrained assortment planning problem where the retailer must display products to customers but must also satisfy operational constraints.

This dissertation studies several strategic-level pricing decisions of firms that are motivated by recent changes of pricing policies in several service industries. It consists of three essays, each analyzing a different problem of selecting the optimal pricing policy for firms in a certain service industry. All three essays contribute to the arising areas of strategic-level revenue management and consumer-driven operations management. The first essay studies whether preventing resale of tickets benefits the ticket providers for sporting and entertainment events. Different from what common wis- dom may suggest, I find that this event organizers can benefit from reductions in consumers’ (and speculators’) transaction costs of resale in many cases. Further, I propose ticket options (where consumers would initially purchase an option to buy a ticket and then exercise at a later date) as a novel ticket pricing policy, and show that ticket options naturally reduce ticket resale and increase event organizers’ revenues. The second essay studies a conditional upgrade strategy that has recently become common in the travel industry. A consumer can accept a conditional upgrade offer after making a reservation and pay the fee to upgrade at check-in if the higher-quality product type is still available. I identify multiple benefits of conditional upgrades including demand expansion, price correction, and risk management. Moreover, I find that using conditional upgrades can generate higher revenues than having the ability to optimize product prices and use dynamic pricing. The third essay studies the firm’s strategic decision of whether to bundle the an- cillary service (e.g., baggage delivery) into the main service (e.g., air travel) or to unbundle and charge separate prices. I find that a firm that can price-discriminate when selling the main service should unbundle the ancillary service because con- sumers’ likelihoods of purchasing the ancillary service are low, or a large proportion of consumers are myopic instead of forward-looking, or the firm is dependent on inter- mediaries to make sales. I also find that the firm’s ability to price-discriminate when selling the main service reverses some classic bundling results for a uniform-pricing firm.

Predicting the demand for future flights accurately is highly important in order to ensure the survival of any airline. In order to reliably forecast the number of potential passengers Revenue Management forecasting methods need error-free true demand data. The true demand for a booking class is the number of passengers willing to purchase the ticket assuming the booking class is always available for purchase. However, due to the aircraft’s physical capacity limitations and historical Revenue Management decisions, the collected booking data seldom represent true demand. Adding to the true demand modelling complexity, airline customers’ pref- erences and experiences vary significantly. Not accounting for these phenom- ena when estimating the true demand may result in erroneous predictions of the true demand. In addition, with similar options offered by various airlines the substitution patterns manifest strongly in the observed booking numbers. To account for the varying preferences of passengers and for correlations between products we introduce a novel Segment Specific Cross Nested Logit Model with Brand Loyalty. The model explicitly handles substitution pat- terns between all the market options, allows tracking changing customers’ preferences and accounts for their past experience. Using data from a Stated Preference experiment conducted among 360 Australians we showed that our discrete choice model constantly outperformed the more simplistic methods by 0.04 to 0.06 in terms of the adjusted ρ2, attaining results on par with more sophisticated models. At the same time our model was only marginally more computationally expensive than the simpler models. In this thesis we also propose a novel unconstraining model. The Market Spill Recapture Model delivers true demand estimates for all the booking classes offered by all the airlines in a specific market. Using an agent based booking simulator developed for this research, with discrete choice models from the Stated Preference experiment driving virtual passengers decisions, we simulated booking data for six virtual airlines. The Segment Specific Cross Nested Logit Model with Brand Loyalty was used to model demand interactions in our unconstraining model. For the simulated data our model delivered true demand estimates that were between 1% to 10% more accurate than the Multi-Flight Recapture Method.

The objective of this research is to investigate the core financial analyticalcompetencies that hospitality managers must possess. Similar measurement was employed in both the industry practitioners’ and the educators’ survey, as the aim was to explore the importance and use of revenue and cost management competencies among manager participants and compare the findings with those obtained from the educators’ survey. The results of this study showed that cost control, followed by revenue management, budgeting, forecasting, pricing, and asset management, are the essential financial competencies of hotel managers. The perception gap was found between hospitality managers and educators on the importance of these financial competencies. The findings of this study could provide suggestions for hospitality education and industry training programs.

The U.S. hardwood industry has traditionally depended on the domestic demand to sustain levels of production above 14 billion board feet per year. Because of the collapse of the U.S. housing market in 2009 and the economic recession that followed, the industry moved its sight to the international markets, as an opportunity to replace some lost demand, and pursue long term growth. Previous research on international marketing of hardwood products indicates that, there is a growing concern among U.S. companies to understand the main competitiveness factors in key markets such as Asia and Europe. Finding opportunities to add value to U.S. hardwood exports has been the goal of this research project. A case study and survey research were carried out among importers and exporters, whereby it was found that aspects related to price, quality and service, are critical in achieving competitive advantage. This motivated a study in demand and pricing management, which found that these tasks may be subject to innovation through optimization approaches.

This dissertation is comprised of two essays, both of which study how particular market environments affect firms' abilities to price discriminate.In the first chapter, I analyze the pricing decisions of airlines in monopoly markets. Airline markets are noted for having several key features: (1) airlines have limited capacity and limited time to sell, (2) airlines face uncertainty in the popularity of any given flight, and (3) consumers who purchase tickets close to the departure date are less price sensitive than those who buy well in advance. These forces influence the pricing decision - what I call dynamic adjustment to stochastic demand (1 and 2) and intertemporal price discrimination (3). While the previous literature has emphasized each force in isolation, in this chapter, I estimate a model of dynamic airline pricing taking both into account. I use an original data set of daily fares and seat availabilities at the flight level. With model estimates, I disentangle key interactions between the arrival pattern of consumer types and remaining capacity under stochastic demand. I find dynamic adjustment to stochastic demand is particularly important as a means to secure seats for high-valuing consumers who arrive close to the departure date. It leads to substantial revenue gains compared to pricing policies which depend on date of purchase but not remaining capacity. In aggregate consumers benefit, despite facing higher fares on average, as a result of more efficient capacity allocation. Finally, I show an empirical procedure abstracting from stochastic demand will systematically understate the price insensitivity of consumers who search for tickets close to the date of travel.In the second chapter, Brian Adams and I develop an empirical analysis of zone pricing under competition. While monopolists can only increase profits by adopting more granular pricing policies, this is not necessarily the case in markets with competition. Using an original data set for the retail drywall industry, we estimate a structural model of supply and demand. We use the model estimates to calculate equilibrium under alternative pricing policies to quantify the welfare implications of zone pricing. We find consumer surplus decreases substantially but firm profits appear to increase with finer pricing. As firms have not adopted these policies, they must face some additional costs. We call these costs "spatial menu costs," and our analysis finds them to be substantial: at least 22.1% of estimated profits, or 2.2% of revenues. Finally, we show that competitive interaction plays an important role in recovering menu costs - failing to account for competitive effects leads to an overestimate of profit gains and implied menu costs by 32.9%.

It is a common practice for many industries to bolster sales and revenue through dynamic pricing strategies. For instance, markup pricing is especially prevalent in the commercial airline industry. Meanwhile, for products sold on a seasonal basis, many stores generally decrease their prices toward the end of the season. In this thesis, we compare the expected revenue that can be generated from two representative pricing policies: markup only and markdown only, in a continuoustime single product revenue management model over a finite horizon. In our model, the initial inventory is fixed and there is no replenishment opportunity during the selling period. The demand follows a homogeneous Poisson process whose rate is controlled by price only. The decisions are to select a price from a set of predetermined prices at each time during the selling period. We show that the markdown policy is superior to the markup policy when the inventory level is low and the remaining time is ample; when the remaining time is short, the markup policy is superior. Our findings complement the previous studies in each of these policies and provide valuable insights for practitioners to choose among different pricing strategies. We verify our findings through numerical tests.

The dissertation contains two major research projects. In the first project, we first study a multi-period inventory planning problem. In each period, the firm under consideration can source from two possibly unreliable suppliers for a price-dependent demand. Our analysis suggests that the optimal procurement policy is neither a simple reorder-point policy nor a complex one without any structure, as previous studies suggest. Instead, we prove the existence of a reorder point for each supplier. No order is placed to that supplier for any inventory level above the reorder point and a positive order is issued to that supplier for almost every inventory level below the reorder point. We characterize conditions under which the optimal policy reveals monotone response to changes in the inventory level. Furthermore, two special cases of our model are examined in detail to demonstrate how our analysis generalizes a number of well-known results in the literature. In the second project, we study a long-run inventory planning problem in which the retailer can replenish inventory and change price adjustment. We establish that it is optimal to change the price from low to high in each replenishment cycle, the optimal order-up-to level may decrease when the ordering cost increases, and fewer customers are served when the unit cost of procurement increases. Additionally, we provide efficient algorithms to compute the optimal stocking and pricing policies.

This dissertation consists of three essays that analyze stochastic dynamic optimization models and game models in operations management.Markov decision processes (MDPs) and sequential games are good models of many real sequential decision processes. However, in diverse applications in operations research and economics, the state of the MDP is a vector and the curse of dimensionality obstructs analysis and computations. Several veins of research seek to exorcise this curse. One vein, the domain of the first essay, identifies MDPs with myopic optima, namely sequential decision processes that can be solved via a temporal sequence of static problems. The first essay identifies new classes of MDPs with myopic optima and sequential games with myopic equilibrium points.Seasonal fashion goods and seats on specific airline fights exemplify a job lot whose sale as time passes should be managed closely as a deadline approaches. The second essay analyzes a dynamic revenue management model in which firms set prices and hold back inventory for sale later in the season. We concentrate on a model with demand functions that are stochastic, nonstationary, and iso-elastic. If there is only a single firm, the resulting Markov decision process has a myopic optimum that canbe specified nearly explicitly and is easily computed. If there are multiple firms, there is a myopic Markov-perfect equilibrium point that can be computed easily and facilitates the analysis of comparative dynamics.A pricing schedule is nonlinear if the resulting revenue is not a linear function of quantity. The third essay presents a dynamic nonlinear pricing model with stochastic demand. We start with a two-segment model for illustrative purposes and derive properties of the model and its optimum which include the existence of a myopic optimum that can be specified nearly explicitly and is easily computed. Furthermore, the optimal policy is a linear decision rule. The essay ends by extending the two- segment model to an arbitrary number of segments and uses numerical examples to compare models with two and three segments.

In queueing theory, the phenomenon that customers get impatient and renege from the system when the waiting time exceeds their tolerance is called customer abandonments. For those abandoning customers, they may go back to the service system after some time. This is called customer retrials. In the modern design of service systems, the impact of customer abandonments and retrials on the system performance and queueing dynamics has been realized. In this thesis, we aim to characterize the stationary properties for queueing systems with both customer abandonments and retrials. We also apply some of those properties in a revenue management problem for queueing systems with customer abandonments.First, we study a multi-server queueing system with both customer abandonments and retrials. By using RTA (retrials see time average) approximation, we characterize the stationary properties of such a queueing system. We also justify the appropriateness of RTA approximation in this model both theoretically and numerically.Then, we extend the study to incorporate negative customers and unreliable servers. Two different models are considered. One is a multi-server retrial queueing system with both regular customers and negative customers. The other is a single-server retrial queuewith negative customers and server interruptions. Closed-form expressions of some relevant performance measurements for both models are obtained.For the last, we apply some of our findings in queues with abandonments and retrials to a revenue management problem for a queueing system with customer abandonments. By using a mechanism design framework, we identify the optimal price-lead time manus and scheduling policies for different scenarios.

This dissertation addresses three multi-dimensional problems in single-resource revenue management. Many problems in revenue management involve managing a single resource but their dynamic programming formulation still requires using a multi-dimensional vector as the state variable. For instance, singleflight-leg revenue management is an extensively studied problem where airlines use the same seat resource to accommodate different classes of customers. In these problems, due to the multi-dimensional state space, the exact formulation is either intractable, or the resulting optimal policy is inconvenient to implement operationally. We develop approximations for the exact formulations, which generate tractable and operationally attractive policies. The first problem we study is a strategic decision problem about whether or not to discontinue a product sold under warranty, whose failure probabilities are unknown initially and are learnt as sales take place and failure information is accumulated. Since there are multiple types of failures that the product can fail from, we formulate the problem as a multi-dimensional optimal stopping problem with Bayesian learning. Two approximations based on dynamic programming decomposition and deterministic approximation are developed, and insights about the value of learning are extracted from asymptotic analysis. Next we study a dual-channel pricing and capacity allocation problem for hotel revenue management. While one channel is the spot market in which we can adopt dynamic pricing, the other is a conference market with a fixed price offered for conference participants. Remaining rooms in the conference market will be released to the spot market if not booked by a deadline. Tactical decisions on number of rooms to reserve and fixed price to offer for the conference market need to be made at the beginning of the selling horizon. For the operational pricing problem in the spot market, because the two markets will join together in a future time, we need a two-dimensional dynamic program which tracks the remaining capacities in both markets to make optimal pricing decisions in the spot market. We develop a single-dimensional approximation to the exact two-dimensional formulation, which generates a robust and operationally attractive policy. For the tactical problem of finding the optimal capacity allocation between spot and conference markets and choosing the fixed price to charge in the conference market, we construct an asymptotically optimal policy through a deterministic formulation. Finally, we consider a revenue management problem where we sell a product to multiple markets with heterogeneous price sensitivities. We can allocate the capacity to different markets and charge different prices in different markets (separate pricing), in which case we gain pricing flexibility, or we can merge all markets together and serve them with a common price (joint pricing), in which case we obtain capacity flexibility. We study the tradeoff between pricing and capacity flexibilities and establish conditions under which one is more important than the other. For a hybrid model where separate pricing is adopted early in the selling horizon and joint pricing is used towards the end, we develop a single-dimensional approximation which gives rise to a policy with remarkably good performance especially for problem instances with tight capacity, large number of markets, or drastically different price sensitivities.

A simulation model is developed for scheduling elective patients to a hospital facility based on resource availability. The approach takes into account two different types of patients, emergency in which admission occurs randomly, and elective in which demand is random, but patients are only admitted if they have been scheduled. The simulation aims to maximize the hospital's rewards from treating patients, in the presence of penalties for overbooking the facility. Few different heuristics for admission policies are compared and results analyzed. Simulation was performed on data from the Neurosurgery department at the UCLA Ronald Reagan Hospital. Even though the simulation is based on data from a neurosurgery department it can be used when admitting patients to other medical departments as well.

With the success of Revenue Management (RM) techniques over the past three decades in various segments of the service industry, many manufacturing firms have started exploring innovative RM technologies to improve their profits. This dissertation studies RM for make-to-order (MTO) and make-to-stock (MTS) systems. We start with a problem faced by a MTO firm that has the ability to reject or accept the order and set prices and lead-times to influence demands. The firm is confronted with the problem to decide, which orders to accept or reject and trade-off the price, lead-time and potential for increased demand against capacity constraints, in order to maximize the total profits in a finite planning horizon with deterministic demands. We develop a mathematical model for this problem. Through numerical analysis, we present insights regarding the benefits of price customization and lead-time flexibilities in various demand scenarios. However, the demands of MTO firms are always hard to be predicted in most situations. We further study the above problem under the stochastic demands, with the objective to maximize the long-run average profit. We model the problem as a Semi-Markov Decision Problem (SMDP) and develop a reinforcement learning (RL) algorithm-Q-learning algorithm (QLA), in which a decision agent is assigned to the machine and improves the accuracy of its action-selection decisions via a “learning" process. Numerical experiment shows the superior performance of the QLA. Finally, we consider a problem in a MTS production system consists of a single machine in which the demands and the processing times for N types of products are random. The problem is to decide when, what, and how much to produce so that the long-run average profit. We develop a mathematical model and propose two RL algorithms for real-time decision-making. Specifically, one is a Q-learning algorithm for Semi-Markov decision process (QLS) and another is a Q-learning algorithm with a learning-improvement heuristic (QLIH) to further improve the performance of QLS. We compare the performance of QLS and QLIH with a benchmarking Brownian policy and the first-come-first-serve policy. The numerical results show that QLIH outperforms QLS and both benchmarking policies.

Much of the recent revenue management literature takes customer behavior into account. A number of parametric models that incorporate customer choice have been developed. In some cases, these models closely approximate reality and provide high quality solutions. Nevertheless, most studies of revenue management models do not consider the possibility that the model used to generate decisions is different from reality. Such analyses also typically do not address effects of forecasting or how forecasts evolve when the model being used is misspecified. (A model for which there is no parameter setting that makes the model a correct description of reality is called a misspecified model). In this dissertation, we study some models of customer choice in revenue management and test their performance when implemented in settings where their assumptions are violated; i.e., when they are misspecified. First, we study a model based on the notion of "buy-up" that considers the dependency of the customers who are willing to purchase low-fare tickets and those who prefer high-fare tickets. To implement this parametric model, a decision maker (revenue manager) needs to observe some data to estimate its parameters (buy-up rate and demand distributions), and make decisions (booking limits) using the model. Meanwhile, the choices of booking limits will affect customers' behavior and thus affect the following observed data. We study the above dynamics and show the convergence of booking limits when the buy-up model is misspecified and customer arrivals are actually deterministic. Numerical studies are also provided to show the performance of the model. Second, we continue the study of the "buy-up" model and consider more complicated actual customers' behavior in which the numbers of different types of customers are stochastic. We present a general necessary condition for the convergence of booking limits and buy-up rate estimates. We provide sufficient conditions for convergence using two different approaches. Comparisons among the optimal revenue, the revenue associated with convergence in the "buy-up" model, and the revenue obtained from the Littlewood rule are presented in the end. Third, we study the performance of the Littlewood rule when it is used to manage bookings for substitutable fights. We show the convergence of booking limits under some assumptions, and make some conjectures about the limiting behavior of booking limits in other settings. Some numerical studies are performed to enlighten future work.

This thesis explores how revenue management (RM) principles would integrate into a parking system, and how advanced reservation-making, coupled with dynamic pricing (based on booking limits) could be used to maximize parking revenue. Detailed here is a comprehensive RM strategy for the parking industry, and an integer programming formulation that maximizes parking revenue over a system of garages is presented. Furthermore, an intelligent parking reservation model is developed that uses an artificial neural network procedure for online reservation decision-making. Next, the work evaluates whether the implementation of a parking RM system in a dense urban parking district (and thus avoiding "trial-and-error" behaviors exhibited by drivers) mitigates urban congestion levels. In order to test this hypothesis, a parallel modeling structure was developed that uses a real-time decision-making model that either accepts or rejects requests for parking via a back-propagation neural network. Coupled with the real-time decision-making model is a micro-simulation model structure used to evaluate the policy's effects on network performance. It is clear from the results that the rate at which parkers renege is a primary determinant of the value of the implementation of RM. All other things being equal, the RM model in which the majority of parkers is directed to their precise parking spot via the most direct route is much more robust to the random elements within the network that can instigate extreme congestion. The thesis then moves from micro-evaluation to macro-evaluation by measuring the performance of the urban parking system from the perspective of the set of relevant stakeholders using the hyperbolic DEA model within the context of the matrix DEA construct. The stakeholder models, including that of the provider, the user, and the community, have defined inputs/outputs to the hyperbolic DEA model, which allows for the inclusion of undesirable outputs such as network delay and incidence of extreme congestion. Another key contribution of this work is that of identifying design issues for current and future dense urban parking districts. Clearly, reneging rate and the tenacity of perspective parkers is a key consideration in cases where RM policy is not implemented.

This dissertation consists of three essays on supply chain and revenue management. Its main focus is the integration of dynamic production and demand management decisions for multiple products facing uncertain demand. The study considers three problem settings to inquire into flexibility's influence on a dynamic pricing strategy, managing exogenous demand for intermediary products, and allocation of shared resources among multiple products in a make-to-order environment. The first chapter studies a joint mechanism of dynamic pricing and capacity flexibility to mitigate demand and supply mismatches. We consider a firm producing two substitutable products with price-correlated demands utilizing capacitated product-dedicated and flexible resources. We characterize the structure and sensitivity of the optimal production and pricing decisions and find that the existence of a flexible resource in the firm's capacity portfolio helps maintain stable price differences across items over time. This result has favorable ramifications from a marketing standpoint as it suggests that even when a firm applies a dynamic pricing strategy, it may still establish consistent price positioning among multiple products if it can employ a flexible replenishment resource. We also investigate the economic benefits of a joint strategy versus applying each tool individually. The second chapter considers a setting where several intermediate products are assembled into an end product with external demand for both the end and intermediary products. We address the questions on how to decide on the production quantities for each component, when to initiate an assembly operation and how to set admission rules for demands targeted at various products. We characterize the structure and sensitivity of the optimal policy and extend the model to multiple customer classes and partial revenue collecting schemes. We propose a novel heuristic policy that is efficient, easily implementable and robust with respect to the number of intermediate products. Finally, the third chapter studies a basic mass customization setting by focusing on a manufacturing firm that assembles products to customer orders. For a two-stage production-assembly operation, we partially characterize the optimal production and demand admission control decisions. Based on the insights gained, we propose a heuristic algorithm and provide computational results on its performance.

"Uncertainty" is used broadly to refer to things that are unknown or incompletely understood. In operations management, basic sources of uncertainty may include decision uncertainty, model uncertainty, analytical uncertainty, data uncertainty, and so on. Although uncertainty is unavoidable in decision making, different mechanisms can be designed to mitigate the impact of uncertainty. One commonly used strategy is "decision postponement," wherein the decision maker purposefully delays some of the decisions to a time when uncertainty is reduced or resolved. This type of a recourse action provides the decision maker with increased ability to match supply with demand. In this dissertation, we study the value of decision postponement in the context of different settings, including capacity investment, revenue management, and supply chain coordination. These problems share one characteristic in common: decision postponement, and as such, are all modeled as two-stage stochastic programming problems. In the first stage, a set of decisions are made under uncertainty so as to maximize the expected profit or utility. Then in the second stage, all uncertainty is resolved and a deterministic optimization problem is solved to determine the postponed decisions, constrained by the first stage decisions. In capacity investment, we study the capacity, pricing, and production decisions of a monopolist producing substitutable products with flexible or dedicated resources. While the capacity decision needs to be made ex-ante, under demand uncertainty, pricing and production decisions can be postponed until after uncertainty is resolved. We show how key demand parameters (the nature of uncertainty, market size, market risk, and risk attitude) impact the optimal capacity decision under the linear demand function. In particular, we show that if the demand shock is multiplicative, then in terms of the "invest or not" decision, the firm will be immune to forecast errors in parameters of the underlying demand shock distribution. Furthermore, incorrectly modeling the demand shock as additive, when, in fact, it is multiplicative, may lead to overinvestment. On the other hand, while the concept of a growth in market size leads to similar conclusions under both additive and multiplicative demand shocks, how market risk affects the optimal capacity decision depends critically on the form of the demand shock. In addition, the decision-maker's attitude toward risk significantly affects the optimal capacity level, and its impact highly depends on the structure of the resource network. Our analysis provides insights and principles on the optimal capacity investment decision under various settings. In airline revenue management, a well-studied problem is the optimal allocation of seat inventory among different fare-classes, given a capacity for the flight and a demand distribution for each class. In practice, capacity on a flight does not have to be fixed; airlines can exercise some flexibility on the supply side by swapping aircraft of different capacities between flights as partial booking information is gathered. This provides the airline with the capability to more effectively match their supply and demand. In this dissertation, we study the seat inventory control problem considering the aircraft swapping option. Our analytical results demonstrate that booking limits considering the swapping option can be considerably different from those under fixed capacity. We also show that principles on the relationship between the optimal booking limits and demand characteristics (size and risk) developed for the fixed-capacity problem no longer hold when swapping is an option. We develop new principles and insights on how demand characteristics affect the optimal seat allocation under the swapping possibility. We also perform a numerical study, which indicates that the revenue impact of using the "true" optimal booking limits under the swapping possibility can be significant. In supply chain coordination, we consider the influenza vaccine supply chain, which, due to the biological complexity of the production process, has a unique characteristic in that production yield is highly uncertain. Given the market demand and price, a monopolist supplier must decide how much raw material to input into production in the first stage. However, since the yield is unknown and production is costly, it is not necessarily in the supplier's best interest to ensure that all market demand is met. The supplier's input quantity depends on the trade-off between the costs of overproduction and undersupply. This, in fact, is one of the reasons why the influenza vaccine manufacturers in the United States lack motivation to produce sufficient amounts of vaccine to meet all demand [Williams (2005), Chick et al. (2008)]. In operations management, it is a well-known result that decentralized supply chains, where each player is only interested in optimizing her own objective, often lead to poor overall performance for the supply chain. However, a higher efficiency is achievable through contracting on a set of transfer payments [Cachon (2004)]. A "coordinating" contract is referred to as one in which each player's objective is in accordance with the supply chain's objective. Given the fact that influenza vaccine plays an important role in health care industry, it is important to study how different contracts impact the influenza vaccine supply chain, where the uncertainty is on the supply side. We study a game in which the supplier and the retailer are engaged in certain type of contracts that specify how risk is shared between the players. We study both the pre-ordering and the post-ordering settings, which respectively refer to the cases where the retailer orders the vaccine before or after the vaccine production is completed. We show that pre-ordering wholesale price contracts dominate post-ordering wholesale price contracts in terms of the resulting supply chain efficiency, but neither of them are able to fully coordinate the supply chain. We also find that cost-sharing contracts are able to coordinate the supply chain, while payback and advance-ordering wholesale price contracts fail to do so. Finally, we prove that if the unsold vaccine can be salvaged with some positive value, then the supply chain can be easily coordinated with wholesale price contracts. In studying this type of stochastic programming problems, it is not only important to characterize the optimal solution, but also important to gain an understanding of how the optimal solution will be affected by environmental parameters. Since the most inaccurate part in stochastic programming often lies in the parameters of the distribution functions, it is both interesting and meaningful to investigate how the optimal solution varies with the intrinsic nature of the random variables. Consequently, we make use of stochastic order relationships to study the behavior of the optimal solutions when the underlying random variables become either "larger" or "more risky."

Service providers face the risk of losing revenue if physical capacity does not match the demand requiring its use, so operating with the optimal physical supply profile is essential to maximizing revenue. Research on how to determine this physical supply has not always accounted for the space required to house it, and has typically assumed that: the optimal supply mix can be accommodated by the available space, the existing number of inventory units within the space is appropriate, and inventory units are homogeneous in terms of the space they occupy. The research that has addressed the use of space in Revenue Management sometimes incorporates space as a constraint to the problem, but other times uses space as the decision variable. Therefore, testing whether there is a revenue difference between these two space outlooks in situations where these key assumptions do not hold is warranted. A simulation model using data from a casual, full-service restaurant was developed to compare the impact of incorporating space in these two ways into the Revenue Management problem. A full-factorial experimental design created 36 distinct simulation scenarios, with these two space outlooks serving as the primary factor, and three other factors providing a range of operating conditions. For each scenario, all possible table mixes were enumerated, simulated, and ranked according to total revenue. The top revenue-generating table mix under the two space methods were paired at every level of the other factors and revenue differences were analyzed. Results from the simulation experiment did not reveal any systematic revenue difference between the two space methods when the tables used at a restaurant were of standard size or larger. When the tables were smaller than standard size and the restaurant experienced extremely high demand, using space as the decision variable generated a significant revenue benefit over incorporating space as a constraint. To make these findings accessible to practitioners and because published means for determining the optimal physical supply profile do not always recommend table mixes that fit in the available space, table mix heuristics developed by Kimes and Thompson (2005) were modified to account for both methods of incorporating space. The size of the tables used by the restaurant and the seating rules followed affected which heuristic recommended the most lucrative supply mix.

For many products such as cars, electronic appliances and furniture, the price that a customer pays is negotiated from the posted price. In such transactions, negotiation plays an important role in determining whether a customer purchases, and if so, at what price. A retailer may adopt negotiation because negotiation allows retailers to price discriminate among customers with heterogeneous willingness-to-pay. However, negotiating the price with customers often comes at a cost, which offsets the benefit generated from price discrimination. We have seen many retailers actively choose between posted pricing and negotiation. Hence, this dissertation visits two problem domains in operations management, assuming that a retailer can make an active choice about sales formats. We first consider a retailer's revenue management problem in the presence of inventory considerations, assuming that the retailer can negotiate. Second, we consider a capacity-constrained manufacturer's wholesale pricing problem, assuming that the retailer is free to choose sales formats. The first problem considers a stochastic dynamic programming formulation embedded by a negotiation model, which produces a number of interesting analytical and numerical results. As one would expect, the optimal posted price of a negotiating retailer includes a premium on top of the price that a retailer would choose in the take-it-or-leave-it pricing. This price premium helps the retailer extract more revenues from high willingness-to-pay customers, and peaks at moderate inventory levels. In addition, the results show that negotiation acts as a substitute or complement to dynamic pricing. The second problem considers a supply chain has limited capacity and the retailer chooses either posting pricing or negotiation, when selling to customers who are heterogeneous in their willingness-to-pay. We find that the supply chain may lie in one of the three different types of equilibria. Two of these three are cases where the supply chain ends up doing as if the sales format were exogenously determined. The third equilibrium type, however, arises when the manufacturer wants to enforce negotiation, but cannot induce the retailer to do so without sacrificing some of its profit margin. This is an equilibrium where the retailer benefits from its discretion over the sales format.

In this thesis, we develop approximate dynamic programming and stochastic approximation methods for problems in inventory control and revenue management. A unifying feature of the methods we develop is that they exploit the underlying problem structure. By doing so, we are able to establish certain theoretical properties of our methods, make them more computationally efficient and obtain a faster rate of convergence. In the stochastic approximation framework, we develop an algorithm for the monotone estimation problem that uses a projection operator with respect to the max norm onto the order simplex. We show the almost sure convergence of this algorithm and present applications to the Q-learning algorithm and the newsvendor problem with censored demands. Next, we consider a number of inventory control problems for which the so-called base-stock policies are known to be optimal. We propose stochastic approximation methods to compute the optimal base-stock levels. Existing methods in the literature have only local convergence guarantees. In contrast, we show that the iterates of our methods converge to base-stock levels that are globally optimal. Finally, we consider the revenue management problem of optimally allocating seats on a single flight leg to demands from multiple fare classes that arrive sequentially. We propose a stochastic approximation algorithm to compute the optimal protection levels. The novel aspect of our method is that it works with the nonsmooth version of the problem where capacity can only be allocated in integer quantities. We show that the iterates of our algorithm converge to the globally optimal protection levels. In the approximate dynamic programming framework, we use a Lagrangian relaxation strategy to make the inventory control decisions in a distribution system consisting of multiple retailers that face random demand and a warehouse that supplies the retailers. Our method is based on relaxing the constraints that ensure the nonnegativity of the shipments to the retailers by associating Lagrange multipliers to them. We show that our method naturally provides a lower bound on the optimal objective value. Furthermore, a good set of Lagrange multipliers can be obtained by solving a convex optimization problem.

The purpose of the study was to determine customer perceptions of fairness concerning pricing policies charged by the hotel industry, and to examine how different outcomes in pricing policies affect customer perceptions of fairness. Convenience-Interception survey sampling was used to collect 460 sample data at the Dallas Love Field Airport. After analyzing data, one can infer that when revenue management information was provided, customers are satisfied. Further, age, education, Airline FFP enrolled and redeem miles, and pricing based on marketing channels plays an imperative role in this study.

The number of trips on highways and urban networks has significantly increased in the recent decades in many cities across the world. At the same time, the road network capacities have not kept up with this increase in travel demand. Urban road networks in many countries are severely congested, resulting in increased travel times, increased number of stops, unexpected delays, greater travel costs, inconvenience to drivers and passengers, increased air pollution and noise level, and increased number of traffic accidents. Expanding traffic network capacities by building more roads is extremely costly as well as environmentally damaging. More efficient usage of the existing supply is vital in order to sustain the growing travel demand. Travel Demand Management (TDM) techniques involving various strategies that increase the travel choices to the consumers have been proposed by the researchers, planners, and transportation professionals. TDM helps create a well balanced, less automobile dependent transportation system. In the past, several TDM strategies have been proposed and implemented in several cities around the world. All these TDM strategies, with very few exceptions, are static in nature. For example, in the case of congestion pricing, the toll schedules are previously set and are implemented on a daily basis. The amount of toll does not vary dynamically, with time of day and level of traffic on the highway (though the peak period tolls are different from the off-peak tolls, they are still static in the sense that the tolls don't vary continuously with time and level of traffic). The advent of Electronic Payment Systems (EPS), a branch of the Intelligent Transportation Systems (ITS), has made it possible for the planners and researchers to conceive of dynamic TDM strategies. Recently, few congestion pricing projects are beginning to adopt dynamic tolls that vary continuously with the time of day based on the level of traffic (e.g. I-15 value pricing in California). Dynamic TDM is a relatively new and unexplored topic and the future research attempts to provide answers to the following questions: 1) How to propose and model a Dynamic TDM strategy, 2) What are the advantages of Dynamic TDM strategies as compared to their Static counterparts, 3) What are the benefits and costs of implementing such strategies, 4) What are the travel impacts of implementing Dynamic TDM strategies, and 5) How equitable are the Dynamic TDM strategies as compared to their Static counterparts. This dissertation attempts to address question 1 in detail and deal with the remaining questions to the extent possible, as questions 2, 3, 4, and 5, can be best answered only after some real life implementation of the proposed Dynamic TDM strategies. Two novel Dynamic TDM strategies are proposed and modeled in this dissertation -- a) Dynamic Congestion Pricing and b) Dynamic Highway Space Inventory Control System. In the first part, dynamic congestion pricing, a real-time road pricing system in the case of a two-link parallel network is proposed and modeled. The system that is based on a combination of Dynamic Programming and Neural Networks makes "on-line" decisions about road toll values. In the first phase of the proposed model, the best road toll sequences during certain time period are calculated off-line for many different patterns of vehicle arrivals. These toll sequences are computed using Dynamic Programming approach. In the second phase, learning from vehicle arrival patterns and the corresponding optimal toll sequences, neural network is trained. The results obtained during on-line tests are close to the best solution obtained off-line assuming that the arrival pattern is known. Highway Space Inventory Control System (HSICS), a relatively new demand management concept, is proposed and modeled in the second half of this dissertation. The basic idea of HSICS is that all road users have to make reservations in advance to enter the highway. The system allows highway operators to make real-time decisions whether to accept or reject travellers' requests to use the highway system in order to achieve certain system-wide objectives. The proposed HSICS model consists of two modules -- Highway Allocation System (HAS) and the Highway Reservation System (HRS). The HAS is an off-line module and determines the maximum number of trips from each user class (categorized based on time of departure, vehicle type, vehicle occupancy, and trip distance) to be accepted by the system given a pre-defined demand. It develops the optimal highway allocations for different traffic scenarios. The "traffic scenarios-optimal allocations" data obtained in this way enables the development of HRS. The HRS module operates in the on-line mode to determine whether a request to make a trip between certain origin-destination pair in certain time interval is accepted or rejected.

Since the service is an economic good, imposing an admission price often provides an efficient means for matching supply and demand. On the other hand an important dimension of providing a service is the time the customer has to wait, especially in situations where the time of delivery is the single factor apart from price that determines whether the customer will do business with the firm. These issues arise in a variety of situations in practice, ranging from manufacturing, to pure service systems, telecommunications, call centers and high bandwidth lines for data transfer. The problem becomes more involved when the potential customers differ in their sensitivity to waiting. In the first part of this work we classify customers into two classes, based on their sensitivity to delay and study the effects of price discrimination on revenue. Next, we consider a system without the ability to price discriminate. We suggest splitting the available capacity into two servers and advertising different entrance fees. Finally, we propose a mechanism that employs options for service as means to increase system flexibility, customer satisfaction and revenue generation. It is typical in this line of research to assume that, upon arrival to the system the customer is given the opportunity to select the mode of service and the corresponding price without any restrictions, and, in many cases, is informed about the level of congestion in the system before he makes this selection. Although this is beneficial to the customers, it may make it harder for the service provider to plan for the necessary service capacity. There is thus a tradeoff between customer flexibility and service provider capacity planning that is often met in supply chains of physical products. In the last chapter of this thesis we propose a mechanism that addresses this tradeoff by employing options for service. The options are offered at a fixed price per unit. A customer purchases a number of them and for a specified number of subsequent periods, can exercise one option at any period, after observing the system congestion, to obtain expedited service.

In this dissertation, I study two topics in the context of revenue management. The First topic involves building a mathematical model to analyze the competition between many retailers who can change the price of their respective products in real time. I develop a game-theoretic model for the dynamic price competition where each retailer's objective is to maximize its own expected total revenue. I use the Nash equilibrium to predict market equilibrium and provide managerial insights into how each retailer should take into account its competitors' behavior when setting the price. The second topic involves working with Amtrak, the national railroad passenger corporation, to develop a revenue management model. The revenue management department of Amtrak provides the sales data of Auto Train, a service of Amtrak that allows passengers to bring their vehicles on the train. I analyze the demand structure from sales data and build a mathematical model to describe the sales process for Auto Train. I further develop an algorithm to calculate the optimal pricing strategy that yields the maximum revenue. Because of the distinctive service provided by Auto Train, my findings make important contribution to the revenue management literature.

This thesis presents a dynamic pricing model where a seller offers two types of a generic product to a random number of customers. Customers show up sequentially. When a customer arrives, he will ---depending on the prices---either purchase one unit of type 1 product or one unit of type 2 product, or will leave empty-handed. The sale ends either when the entire stock is sold out, or when the customers are exhausted. The seller's task is to post the optimal prices for the two product types to each customer to maximize the expected total revenue. We use dynamic programming to formulate this problem, and derive the optimal policy for special cases. For general cases, we develop an algorithm to approximate the optimal policy and use numerical examples to demonstrate the efficiency of the algorithm. Finally, we apply the results to a continuous-time model where customers arrive according to a Poisson process. We develop a heuristic policy and use numerical examples to show the heuristic policy is very effective.