Abstracts

(Contact email addresses mentioned where available.)





Title: Analysis of an integrated pricing model
Authors: Patrice Marcotte, University of Montreal Gilles Savard, Ecole Polytechnique (Montreal)

In many industries, an underlying network structure impacts the pricing process, and provides opportunities for increasing revenues. This is the case not only in transportation (airlines, railways) but also in the realm of supply chain management and telecommunications. In our model, a firm acting in a network-based competitive environment must set prices on the subset of arcs in its control. Each arc is endowed with a number of attributes (price, delay, quality, etc.) and, once prices are set, customers are routed onto paths that minimize their disutility, expressed as a linear combination of the attributes. We assume that the valuation of the attributes is continuously distributed across the user population, and represented by a multivariate probability density function. Further, the network is subject to congestion. In this context, we formulate the model as bilevel program where the leader's revenue maximizing schedule takes into account the customer reaction, which is characterized as the solution of a parametric (infinite-dimensional) variational inequality. We propose for its solution a two-phase algorithm. First, we show how a discretized version of the problem leads to a mixed-integer program that can be addressed for a global maximum by commercial software. Next, we improve upon this solution, using a subgradient approach. Numerical experiments are reported. Finally, we comment on the generalization of the model to stochastic and time-varying environments, and discuss the issue of competitors' reaction.




Title: Revenue Management in the Context of Dynamic Oligopolistic Network Competition
Authors: Terry L. Friesz, Reetabrata Mookherjee and Matthew Rigdon, Penn State University

In this paper we are concerned with devising optimal decision rules as well as computational methods for a normal form differential Stackelberg game in which the leader maximizes its total revenue from spatially dispersed production of a single homogeneous commodity, subject to a lower bound on profit. The followers are Cournot-Nash agents who make time-varying production and distribution decisions relative to the same single homogenous output. The Cournot-Nash followers are described by the dynamic oligopolistic network competition model developed in Friesz et al (2003) and which takes the form of a differential variational inequality. We show that this differential variational inequality is equivalent to a functional nonlinear complementarity problem (NCP). Consequently, the complete Stackelberg game/bilevel model takes the form of a continuous time optimal control problem constrained by a functional nonlinear complementarity problem (FNCP). The specific decision environment we consider is one for which the followers are in Cournot-Nash oligopolistic game theoretic competition according to dynamics that describe the trajectories of inventories and correspond to flow conservation for each firm at each node of the network of interest. The market for physical distribution (delivery) services is at the origin-destination level and is perfectly competitive. That is, the oligopolistic firms, acting as shippers, compete as price takers in the market for physical distribution services. Joining them in this competition for physical distribution services are other very numerous agents who employ the same distribution services to support their activities in markets distinct from that for our homogeneous commodity of interest. Although the oligopolistic producers cannot influence physical distribution prices, they can consider and react to temporal fluctuations in distribution service pricing and to congestion arising in the distribution network. The complete, bilevel Stackelberg determines plant-specific outputs, the allocations of outputs to meet demands and output shipment patterns (all three of which are control variables), as well as output inventories (which are state variables) when there firms compete by constructing temporal as well as spatial strategies for their network activities. We discuss alternative philosophies for solving optimal control problems with FNCP constraints. We provide the details of a discrete time, finite dimensional mathematical programming approximation. We present a numerical example based on this approximation.




Title: Dynamic flow management of on-demand service and production networks
Author: Costis Maglaras, Columbia University,
c.maglaras[at]columbia.edu

This talk overviews some recent results on problems of revenue management and control of stochastic service and production networks. Such problems combine tactical economic decisions of dynamic pricing and capacity control, with operational decisions of order sequencing, prioritization, and leadtime quotation. Focusing on a relatively simple system that is capabale of producing multiple products, we will present some structural results about the degree in which the various types of control interact with each other, their respective time-scales, and their impact on the overall system profitability. Most of these insights remain valid in more complex service and production systems.




Title: Token-Bucket Pricing for Shared Services
Authors: Opher Baron, Rotman School of Management, University of Toronto
Opher.Baron[at]Rotman.Utoronto.Ca
Dirk Beyer, HP Labs, Palo Alto, CA
dirk_beyer[at]hp.com
Gabriel Bitran, Sloan School of Management, Massachusetts Institute of Technology
gbitran[at]mit.edu

This paper presents novel Token-Bucket pricing-schemes for shared-services and a framework for their analysis. One of the major barriers for implementing these schemes is the buyer's challenge of choosing token buckets' parameters. The buyer'sobjective is to minimize expenditures such that the probability of shortages in any period, as dictated by token-bucket admission controls is lower than a threshold. We show that the event of a loss (backlog) in this system is equivalent to the event of a two-(one-) sided regulated-random-walk to reach (cross) a threshold. The buyer's control parameters are the token rate (drift of the random-walk) and the bucket depth (threshold to be crossed). Using the theory of large deviations, we bound the probability that a two-sided regulated-random-walk reaches a threshold and give convex approximations for the buyer's problem. Therefore, under mild assumptions, the buyer's controls can be efficiently approximated. Furthermore, tailored closed-form solutions and approximations for cases of exponential and normal demands are provided. We show empirically that for normal demand the costs of these approximations are typically within 2% of the optimal ones for service levels higher than 80%. The results of this paper support the use of token bucket pricing schemes for shared services.




Title: Modeling extensions for a class of business-to-business revenue management problems
Author: Nicola Secomandi, Graduate School of Industrial Administration, Carnegie Mellon University
ns7[at]andrew.cmu.edu

Recently, the business consulting and operations research communities have shown great interest in porting revenue management (RM) principles from traditional business-to- consumer (B2C) settings to the realm of business-to-business (B2B) e-commerce. The talk discusses how existing B2C RM inventory and pricing control models can be unified and extended to deal with a class of B2B RM problems.



Title: Dynamic Cross-Selling in E-commerce Retailing
Authors: Serguei Netessine, The Wharton School, University of Pennsylvania
netessin[at]wharton.upenn.edu
Sergei Savin, Graduate School of Business, Columbia University
svs30[at]columbia.edu
Wen-Qiang Xiao, Graduate School of Business, Columbia University
WXiao09[at]gsb.columbia.edu

We consider the problem of dynamically cross-selling products (e.g., books) or services (e.g., travel reservations) in e-commerce setting. In particular, a company faces a stream of stochastic customer arrivals and may offer, in addition to the product desired by the customer, a package of this product with another one. We formulate the cross-selling problem as a stochastic dynamic program blended with combinatorial optimization. Given consumer preferences and product inventory, we analyze two related issues: (1) how to dynamically select products to cross-sell and (2) how to dynamically price these product packages.



Title: A Robust Nonlinear Fluid Model of Dynamic Pricing and Inventory Control with no Backorders.
Authors: E. Adida and G. Perakis, MIT
eadida[at]MIT.EDU
georgiap[at]MIT.EDU

We study a fluid model for dynamic pricing and inventory control in a make-to-stock system. First demand is deterministic and subsequently demand is uncertain. We consider a multi-product capacitated, dynamic setting with no backorders. We introduce an algorithm that computes the optimal production and pricing policy on a finite time horizon. When demand is uncertain we consider the additive demand case. We use ideas from robust optimization. We give numerical examples and discuss some insights.




Title: Data-driven revenue management
Authors: Dimitris Bertsimas and Aurelie Thiele, MIT
dbertsim[at]mit.edu
aurelie[at]mit.edu

We present an approach on revenue management problems under uncertainty that directly works with the data, and is based on the belief that what is the primitive in models of practical relevance is data, not distributions. The method leads to insightful solutions, and promising empirical behavior.




Title: Multiperiod Pricing for Perishable Products; a Robust Optimization Approach
Authors: Anshul Sood, (Presenter), and G. Perakis, MIT
anshul[at]MIT.EDU
georgiap[at]MIT.EDU

We propose a model for pricing a single perishable product in a multi-period oligopolistic market with many sellers competing through pricing. We address the competitive aspect of the problem together with demand uncertainty. In particular, we propose a model that uses ideas from robust optimization. We establish existence and uniqueness of pricing equilibrium policies and introduce an iterative learning algorithm for computing the equilibrium policy. We illustrate our results through some numerical examples and discuss some insights.



Title: Online Low Price Guarantees - A Real Options Analysis
Authors: Chris K. Anderson, Ivey School of Business, University of Western Ontario
Ben Marcus, Ivey School of Business, University of Western Ontario
Montgomery Blair, Dollar Thrifty Automotive Group

A common practice among large retailers is the low price guarantee, rebating consumers if they find an identical product cheaper elsewhere, or if the retailer discounts the good within a specified timeframe. This provides consumers with some level of comfort in their purchase decision. A similar low price guarantee is provided by numerous service industries that allow reservation of capacity yet don't penalize the consumer for failure to keep that reservation; examples include hotels and car rental. Given that a consumer is not required to keep the reservation, they may make another reservation either at a competing firm or the same firm if future prices decline (a similar dilution of reservation occurs with corporate clients whom are allowed to also pay regular rates). The increasing availability of pricing information on the Internet affords consumers the opportunity to be more strategic in their behaviors. As a consumer we are able to quickly and easily check prices from numerous service or goods providers. The ease of price information potentially makes these guarantees very costly to the service or good provider. We analyze the implied costs associated with these guarantees by making analogies to financial options. Motivation for this research comes from a large car rental firm, Dollar Thrifty Automotive Group Inc., which is currently considering offering a low price guarantee to all consumers whom book a reservation though their website.




Title: Analysis of a Yield Management Model for On Demand IT Services
Authors: Yezekael Hayel, INRIA-IRISA, Rennes, France,
yhayel[at]irisa.fr
Laura Wynter, IBM Research, Hawthorne, NY,
lwynter[at]us.ibm.com
Parijat Dube, IBM Research, Hawthorne, NY,
pdube[at]us.ibm.com

In [WLD03], a model for performing yield management in the context of IT provisioning was presented. This model is especially valuable in the context of an on demand operating environment. On demand IT services allow users to arrive at will into an IT system, in which scheduled jobs have already reserved some of the resources. Many potential applications of this type of IT infrastructure exist and a few are already in operation. One example of an on demand IT service that exists today is the case of dynamic off-loading of web content. When a customer, such as an online retailer, experiences very heavy web site traffic, that retailer may have its excess traffic automatically redirected to an off-loading service. The process is invisible to end-users of the retailer. The yield management model described in [WLD03] could be used by the off-loading service provider to allocate its own capacity optimally and profitably. Many other potential applications of this technology are on the horizon: application service providers may run software applications on their own cluster of servers and allow customers, for a fee, to use those applications remotely. Yield management in this case sets capacity allocations (server use, storage, and bandwidth) and multiple price points to offer to those customers, depending on the available resource level of the service provider, as well as the market demand. Similarly, computing centers, that rent processing capacity to customers, can operate more profitably and more efficiently through use of a system of yield management. In this presentation, we present a detailed analysis of that model, both in simplified cases where an analytical analysis is possible, and numerically on larger problem instances, and confirm the significant revenue benefit that can accrue through use of yield management in an IT on demand operating environment.



Title: Equilibrium problems with equilibrium constraints: A new modeling paradigm for revenue management
Authors: Houyuan Jiang, Daniel Ralph, Stefan Scholtes, The Judge Institute of Management, University of Cambridge
h.jiang[at]jims.cam.ac.uk
d.ralph[at]jims.cam.ac.uk
s.scholtes[at]jims.cam.ac.uk

Pricing and seat allocation are two classical problems in revenue management. Traditionally, they are studied separately. Recently, some researchers started to consider pricing and seat allocation in an integrated manner in order to achieve higher revenue. At the same time, revenue management under competition has received considerable attention recently. In this paper, we use a framework of equilibrium problems with equilibrium constraints (EPEC) to model both competition between airlines and the integration of pricing and seat allocation for each airline. We review the literature of EPEC and discuss existence and uniqueness issues. We also discuss numerical approaches for solving EPEC models arising from airline revenue management.




Title: Revenue Management Through Cross-Selling
Authors: E. Lerzan Ormeci, Department of Industrial Engineering, Koc University
lormeci[at]ku.edu.tr
O. Zeynep Aksin ,Graduate School of Business, Koc University
ZAKSIN[at]ku.edu.tr
Evrim-Didem Gunes, INSEAD, Boulevard de Constance
evrim-didem.gunes[at]insead.edu

Many firms in mature industries resort to growth by deepening customer relationships and making them more profitable rather than increasing market share. A significant part of this profitability comes from revenues generated by the sale of additional products and services to existing customers. Given the growing dislike among consumers for telemarketing, this type of selling is increasingly being performed via cross-selling and up-selling initiatives. Inbound call centers are an important point of contact with the customer, where this type of selling takes place. We model the cross-selling problem of a call center as a dynamic admission control problem to address the question of when and to whom to cross-sell such that revenue generation is maximized while congestion costs are kept as low as possible. Optimal dynamic cross-selling policies are partially characterized, and certain structural properties are explored. We relate our findings to standard marketing practice in such problems.



Title: Optimal Dynamic Stochastic Network Yield Management
Author: Xiubin Wang, Transportation and Logistics Research Center, University of Wisconsin-Superior
xwang1[at]uwsuper.edu

We discuss the optimal solution to the general dynamic stochastic network yield management in the context of airline operation. Necessary and sufficient conditions are identified for a policy change. A solution scheme is proposed to find the pivotal times for each fare bucket prior to departure. It is a double recursion from the end to start of the sales season and from low capacity to high capacity. The solution presented is general.




Title: Efficiency Loss in Resource Allocation and Supplier Selection Games
Authors: John Tsitsiklis, Ramesh Johari, MIT
jnt[at]mit.edu
rjohari[at]mit.edu

Motivated by certain proposals for network resource allocation, we analyze a simple game in which the users of congested finite-capacity links anticipate the effect of their actions on the link prices. We discuss existence and uniqueness of a Nash equilibrium, and establish that the efficiency of the system drops by no more than 25% relative to the social optimum. We extend the results to a more general resource allocation mechanism in which a set of producers compete for a limited amount of available resources. Motivated by electricity markets, we consider an analogous situation involving a set of competing suppliers who bid in order to satisfy a prespecified demand. We show that if each producer's "bid" consists of a supply function within a certain one-parameter family, the efficiency loss at a resulting Nash equilibrium can be bounded and decreases to zero with the number of suppliers. Finally, we argue that the particular families of demand and supply functions we consider are the only ones that possess certain desirable properties.



Title: Hotel Revenue Management in Competitive Market
Author: Hong Jin, Starwood Hotels and Resorts Worldwide, Inc.
jin[at]alum.mit.edu

This talk addresses the key issues in hotel Revenue Management (RM) in competitive markets, including competitive pricing, channel management, inventory control, market response analysis and etc. The paper start with discuss the key planning and operational issues in hotel RM. Then, it presents a new analytical RM approach developed and applied for a big hotel in a competitive metropolitan area, with comparison to the traditional RM approach currently adopted by the most hotels. To illustrate the contributions of the new approach, the paper shows the improvement results in the key performance metrics, highlighted by a 40% growth in both RevPAR index and total revenues year over year. This analysis is a combination of theory/modeling discussion and operational experiences and best practices in applying the new RM approach in a competitive market. The key contribution of the new approach is making complex optimization/statistics based analytical approach highly operational in effectively managing hotel daily pricing and inventory. The new approach provides some new techniques in solving some key RM issues haunting hotel revenue managers for a long time, e.g. incorporating competitive pricing into RM/Pricing and Inventory Control.




Title: Business Rule Constrained Price Discovery Models in Transportation Procurement Auctions
Authors: Jiongjiong Song, Institute of Transportation Studies, University of California
jiongjis[at]uci.edu
Amelia Regan, Graduate School of Management, University of California
aregan[at]uci.edu

B2B auctions are a dominant price discovery mechanism for transportation service procurement in shipper’s logistics operations. However, shippers are struggling with decisions on how to assign bids to carriers considering both costs and business requirements. In this work, we propose optimization based bid analysis models and examine heuristic algorithms with numerical analysis.



Title: Optimal Price Design for Variable Capacity Outsourcing Contracts
Authors: Chris Kenyon and Giuseppe Paleologo, IBM Research (ZRL, WRL)
chk[at]zurich.ibm.com
gappy[at]us.ibm.com

Outsourcing of information technology (IT) infrastructure and business processes (BP) is a significant aspect of the business landscape with contract signings in the tens of billions each year for the major providers (e.g. Accenture, EDS, IBM). A recent trend emphasizes variable capacity contracts linked to business or IT metrics for medium to long term (5-10 year) deals. This new emphasis on variability is linked to a growth in interest in the accompanying pricing schemes. A basic tension in the design of these pricing schemes occurs between the objectives of the outsourcing provider (hereafter Provider) and the company whose IT or BP are being outsourced (hereafter Company). In other words their utility functions on contract attributes differ. For example the Company may desire utility-style pricing whilst the Provider may be interested in the certainty with which it achieves a given margin. The pricing design tension is accentuated by: history-dependent capacity costs for many contract elements to the Provider; the wide range of capacities of interest to the Company; and the unpredictability of actual events over the contract lifetime. Contract elements typically include dependencies on people, hardware, and software licenses. These dependencies introduce significant delivery cost granularity, dependence on the direction of capacity change, potential delivery lags, and cost non-linearities. In this paper we characterize the price design problem for variable capacity outsourcing contracts and provide a multi-objective utility theory framework for obtaining optimal solutions. Furthermore we show how to construct a pareto-efficient frontier of price designs with axes of Provider-utility and Company-utility by systematically varying contract design parameters. This frontier serves as an appropriate space for contract negotiation.



Title: Dynamic Pricing under Strategic Consumption
Authors: Chunyang Tong and Sriram Dasu, Marshall School of Business, University of Southern California
dasu[at]marshall.usc.edu

We investigate the dynamic pricing of limited capacity when the seller faces strategic buyers, who make their purchase decision based on the currently prevailing price as well as their anticipation of the future price to be posted by the seller. Two pricing schemes are analyzed: upfront pricing scheme in which the seller posts his pricing path at the start of the selling horizon and buyers choose the time for their bidding offer; contingent pricing scheme under which the seller determines the price based on the realized sale. For the above two pricing scheme, consumers’ equilibrium behavior is characterized and optimal pricing-path for the seller is analyzed. Numerical studies are conducted to show the potential gains for the seller if he considers buyer’s strategic behavior. We conclude that buyer’s strategic behavior plays an important role in seller’s optimal pricing scheme and his profitability



Title: An Asymptotically-Optimal Dynamic Admission Policy for a Revenue Management Problem
Authors: Marty Reiman, and Qiong Wang, Bell Labs
marty[at]research.bell-labs.com
qwang[at]research.bell-labs.com

We consider the following canonical revenue management problem, which has been analyzed in the context of airline seat inventory control and has applications to other service industries and supply chain management. There are M resource types (legs) and J customer classes (routes). There is a total of Cm of resource m (seats on leg m), m=1,…,M. Customers of class j arrive in a (possibly nonstationary) stochastic process, require Amj units of resource m, and pay pj if accepted. The aim is to make dynamic accept/reject decisions to maximize the total expected revenue obtained over the finite horizon [0,T] subject to the resource constraints Cm. We introduce a control policy motivated by fluid and diffusion limits (as the Cm’s and arrival rates grow large). Our control policy makes an initial resource allocation decision based on solving a linear program (LP). This leads to a determination of which classes to accept and which to reject. The LP is re-solved (using the remaining capacities) at successive ‘trigger points’ until the end of the finite horizon at T. Our policy preserves desired features of existing approaches, such as nesting, bid-price control, and adaptive optimization, while avoiding known potential pitfalls associated with these schemes. We provide both a theoretical justification as well as numerical comparisons with existing schemes to validate our approach.



Title: Price-Directed Control of a Closed Logistics Queueing Network
Author: Dan Adelman, Graduate School of Business, The University of Chicago
dan.adelman[at]gsb.uchicago.edu

Motivated by one of the leading intermodal logistics suppliers in the United States, we consider an internal pricing mechanism for managing a fleet of service units (shipping containers) flowing in a closed queueing network. Nodes represent geographic locations and arcs represent travel between them. Customer requests for arcs arrive over time, and the problem is to find an accept/reject policy that maximizes the long-run time average reward rate from accepting requests.

We formulate the problem as a semi-Markov decision process, and give a simple linear program that provides an upper bound on the optimal reward rate. Using Palm calculus, we derive a nonlinear program that approximately captures queueing and stockout effects on the network. We show that this nonlinear program provides a new, alternative derivation of an upper bound due to Harel (1988) for the Erlang loss probability. We also interpret the optimality conditions as optimizing the behaviour of decentralized agents assigned to units. Using optimal solution information in an auxiliary pricing problem, we construct an easily computable functional approximation to the dynamic programming value function. The resulting policy is computationally efficient and produces superior economic performance as compared with other policies and the upper bound. Furthermore, it provides a methodologically grounded solution to the firm's internal pricing problem.



Title: Pricing & Elasticity Price on segmented markets under a Conjoint Study
Author: Agustín Cano, Aeromexico
acano[at]aeromexico.com.mx

The objective is finding and understanding the market response to different critical variables in mexican airline markets. Therefore an airline made a marketing study using the concept of Choice Based Conjoint that helped understand the elasticity of the market, and how the Brand can support price differentiation with competitors. The conclusion of this study will be shown.




Title: An MPEC Approach to Dynamic Pricing and Demand Learning
Authors: Soulaymane Kachani (Columbia University, IEOR Department)
sk2267[at]columbia.edu
Georgia Perakis (MIT, Sloan School)
georgiap[at]MIT.EDU
Carine Simon (MIT, Operations Research Center)
casimon[at]mit.edu

In this talk, we present an optimization framework that addresses joint pricing and demand learning in an oligopoly setting in order to maximize expected revenue for perishable products under fixed capacity constraints. The model we consider assumes a parametric family of a company’s demand as a function of its price and other competitors’ prices, the parameters of which are dynamic and are learned over time. We show how the problem reduces to a mathematical program with equilibrium constraints (MPEC) and study the properties of the model. We propose two solution methods. One of these methods solves the lower level using a fictitious play scheme. We illustrate our approach using a 3-company example.



Title: Algorithms for Revenue Management in Unrestricted Fare Markets
Authors: Craig Hopperstad, Hopperstad Consulting
cah[at]gte.net
Peter Belobaba, MIT International Center for Air Transportation
belobaba[at]mit.edu

While the assumption of independent fare class demands was at best questionable even under traditional restricted fare structures, it is completely invalid under the increasingly less restricted fare structures typical of low-fare airline markets. We present some alternative approaches to RM seat inventory control of unrestricted fare classes, including adjustments to EMSRb methods to account for sell-up, willingness-to-pay models, and dynamic programming techniques. We use the Passenger Origin-Destination Simulation (PODS) to assess the performance of these RM methods in a competitive airline market environment.




Title: Dynamic Pricing of Information Goods under Demand Uncertainty
Author: Eric W. Cope, The Sauder School of Business, University of British Columbia
eric.cope[at]sauder.ubc.ca

E-business retailers have used dynamic pricing as a means of estimating aggregate customer demand response to price changes when this response is initially unknown. This study looks at dynamic pricing in the context of learning customer demand for information goods sold on the Internet. Because information goods can be replicated quickly at almost no cost, issues with inventory and production generally do not arise. Rather, the major concern in designing a dynamic pricing strategy in this setting is efficiently managing the exploration-exploitation trade-offs involved in price selection. We introduce a nonparametric Bayesian model of demand uncertainty involving Dirichlet priors, which are very flexible and easy to elicit, and encode many natural assumptions about the dependence among various demand values. We provide both analytic formulas and efficient approximation methods for computing the posterior distributions after sales data have been observed. Within the framework of this Bayesian model, we investigate several "index function" heuristics for sequential pricing. These heuristics are shown to be efficient in terms of both performance and computation, and they are robust to misspecifications of the prior. Finally, we show how to extend these methods to dynamically price multiple "versioned" or "bundled" information goods.



Title: Problems of Contract Pricing
Authors: Ahmet Kuyumcu and Dinesh Mehta, Zilliant, Inc,
ahmet.kuyumcu[at]zilliant.com
dinesh.mehta[at]zilliant.com

Most business transactions are governed by contractual agreements. Companies (buyers, sellers, or both) may commit to terms and conditions of contracts that may be in effect for many years. Profitable contract management involves optimizing these terms and conditions and requires solving different forms of pricing problems such as list pricing, floor pricing, spot pricing, and rebate/incentive optimization. This presentation defines pricing models that arise from contractual agreements and focuses on a specific case study.



Title: Dynamic Pricing for Parallel Flights
Authors: Dan Zhang, Graduate Program in Industrial Engineering, University of Minnesota
William L. Cooper, Graduate Program in Industrial Engineering, University of Minnesota
billcoop[at]me.umn.edu

We describe a Markov decision process model for dynamically pricing multiple flights between a common origin and destination. Consumers decide what, if anything, to purchase based upon the prices of all the flights as well as their own preferences. We describe analytical properties of the problem as well as computational techniques. The work is motivated by a situation faced by low-fare carriers who sell tickets on their own websites over the Internet.



Title: A Value Function Decomposition Approach to Revenue Management Network Problems
Author: Darius Walczak, PROS Revenue Management,
dwalczak[at]prosrm.com

We show an intuitive way to decompose value function of a popular dynamic network problem and how to apply this decomposition to generate approximate dynamic solutions fast. Network problems arise naturally in many RM and Pricing applications with the airline Origin and Destination network (of flights) being a well-known example. While in principle these problems can be solved optimally by recursion, straightforward approaches suffer from the 'curse of dimensionality' and thus are not feasible to implement. We present numerical results for a small (but relevant) network with products that can be offered at a number of price points.



Title: Optimal Pricing and Replenishment in a Single-Product Inventory System
Authors: Hong Chen, Cheung Kong Graduate School of Business
Owen Wu, Sauder School of Business, University of British Columbia,
owen.wu[at]sauder.ubc.ca
David D. Yao, Columbia University

We study an inventory system that supplies price-sensitive demand modeled by Brownian motion, focusing on the optimal pricing and inventory replenishment decisions, under both long-run average and discounted objectives. Analytical solutions are obtained in all cases, and related to or contrasted against previously known results wherever applicable. In addition, we bring out the interplay between the pricing and the replenishment decisions, and the way they react to demand uncertainty. We show that the joint optimization of both decisions results in significant profit improvement over the traditional way of making the decisions separately or sequentially. We also show that changing price over time will only result in a limited profit improvement over a single price (when both are optimally determined). The relative improvement becomes more significant in applications where the profit margin is low. An analytical bound on the profit improvement is derived.



Title: Incentive-Compatible Revenue Management in Queueing Systems: Optimal Strategic Idleness and other Delaying Tactics
Author: Philipp Afeche, Kellogg School of Management,
p-afeche[at]kellogg.northwestern.edu

How should a capacity-constrained firm design an incentive-compatible price-scheduling mechanism to maximize revenues from a heterogeneous pool of time-sensitive customers with private information on their willingness to pay, time-sensitivity and processing requirement? I consider this question in the context of a queueing system that serves two customer types and provide the following insights. First, delay cost-minimization, which plays a prominent role in controlling and pricing queueing systems, should in this setting not be the only criterion ex ante. Second, this problem gives rise to optimal scheduling policies with novel features. Compared to the delay cost-minimizing scheduling policy, these policies increase, decrease or reverse the delay differentiation between customer types. The optimal level of delay differentiation emerges from a trade-off between operational constraints and customer incentives. The approach presented in this talk can be adapted for designing revenue-maximizing and incentive-compatible mechanisms in systems with different properties.



Title: Capacity Allocation to Support Customer Segmentation by Product Preference
Authors: Guillermo Gallego, Columbia University
Ozalp Ozer, Stanford University
Robert Phillips, Nomis Solutions

A capacity-constrained supplier sells into a market with exogenous prices and customers segmented by product preferences. What product should the supplier offer to buyers from each segment as they arrive? How should this decision be updated? This problem is relevant both to traditional revenue management industries such as hotels and cruise lines which offer products of different qualities as well as to contract manufacturers who may differentiate customers by lead time. We present both structural results and numerical simulations.



Title: Optimizing Decisions over the Long-term in the Presence of Uncertain Response
Author: Edward Kambour, PROS Revenue Management,
ekambour[at]prosrm.com

When there is random response to discrete actions, one can utilize expected returns as the objective. This process is complicated when the underlying parameters of the response are not known. In such cases the only way to decrease the uncertainty and become more confident in future decisions is gather information over the decision space. However, this process does not necessarily coincide with that which maximizes expected return in the short run. In this discussion we will examine, in a rigorous mathematical fashion, the effects of uncertainty on the general decision-making process. The goal is to balance the value of gathering additional data versus the implementation of the action that (as of now) has the highest expected return.



Title: Dynamic Revenue Management Games with Forward and Spot Markets
Authors: Guillermo Gallego, Columbia University
Srinivas Krishnamoorthy, Columbia University
sk711[at]columbia.edu
Robert Phillips, Columbia University

We consider a market with two capacity providers – a low cost entrant and an established incumbent - each with fixed capacity, who compete to sell in a spot market and a forward market. Prices are fixed and the providers make strategic capacity allocation decisions. We identify the type of equilibrium behavior we could expect from the two providers under different assumptions about market structure, demand and available capacity. For the infinitely repeated game, we identify the existence of subgame-perfect Nash equilibria and show how the two providers can obtain higher average revenues (than static Nash equilibrium revenues) by implicit collusion. The study has implications for revenue management markets where providers have dynamic competitive interactions rather then a single static interaction.



Title: Price and service competition among service providers operating general queueing facilities
Authors: G.Allon and A.Federgruen, Columbia University

We analyze a general market for an industry of competing service facilities. Other than through intrinsic and unalterable service characteristics, firms differentiate themselves in terms of their price levels and the waiting time customers experience. The expected market share a firm acquires may thus depend on all firms' prices and their stated guarantees (service levels) for the expected waiting time or a given fractile thereof. The equilibrium behavior under price and/ or service competition depends critically on the structure of the capacity costs incurred to meet a given service level under a given demand volume. We show that a large variety of queuing models give rise to a specific class of parameterized capacity cost functions. We characterize the impact of the structure of these capacity cost functions on the equilibrium behavior in the industry, under various types of competition.



Title: Estimating the Impact of Revenue Management Decisions
Author: Warren Lieberman, Veritec Solutions
warren[at]veritecsolutions.com

While it has always been important, estimating the financial impacts of revenue management is receiving increasingly greater attention as revenue management extends its reach to more and more industries. In this presentation, we explore a new methodology for estimating the incremental financial contribution from revenue management decisions. Interestingly, the concepts underlying this methodology may also extend to providing a new basis for revenue management decision-making in some industries.



Title: Dynamic Negotiation Guidelines for Make-To-Order Contracts
Authors: Jeremie Gallien, Yann Le Tallec and Tor Schoenmeyr MIT Sloan School of Management and Operations Research Center
jgallien[at]mit.edu

We seek to identify quantitative, dynamic guidelines for the sales negotiations of make-to-order contracts along the dimensions of price, quantity and delivery lead-time. We study accordingly the dynamic programming formulation of a related admission control problem, and derive a dynamic, computationally efficient policy using a fluid approximation. Simulation experiments reveal that its performance is superior to that of static policies over a important range of market environments. In addition, a comparison with the optimal anticipative policy allows to quantify the value of better forecasting. Finally, we discuss the potential implementation of this work into a decision support system.




Title: Auction Algorithms for Market Equilibrium
Authors: Rahul Garg and Sanjiv Kapoor.
grahul[at]in.ibm.com
kapoor[at]iit.edu

In this paper we study algorithms for computing market equilibrium in markets with linear utility functions. The buyers in the market have an initial endowment given by a portfolio of items. The market equilibrium problem is to compute a price vector which ensures market clearing, i.e. the demand of a good equals its supply, and given the prices, each buyer maximizes his utility. The problem is of considerable interest in Economics. This paper presents a formulation of the market equilibrium problem as a non-linear program. We construct the dual of this non-linear formulation and define conditions under which prices achieve market clearing. These conditions arise naturally from complementary slackness conditions. We then define an auction mechanism which computes prices such that approximate market clearing is achieved. The algorithm we obtain outperforms previously known methods.



Title: On the Choice-Based LP Model for Network RM
Author: Garrett van Ryzin and Qian Liu, Columbia University

Recently, Gallego, Iyunger and Phillips proposed a choice-based LP model for network revenue management which parallels the deterministic linear programming model widely used in current network RM. We build on their work in two ways: First, we characterize the offer sets (sets of available fare products) produced by this model by extending a notion of "efficiency" proposed by Talluri and van Ryzin and show that, asymptotically as demand and capacity are scaled up, only these efficient sets are used in an optimal policy. Second, we propose a practical decomposition heuristic that significantly improves on the performance of the static LP solution. We illustrate the method on several numerical examples and discuss various extensions.




Plenary Talks





Title: How the Internet is Impacting Revenue Management, Pricing, and Distribution
Speaker: E. Andrew Boyd Chief Scientist and Senior Vice President of Science and Research
PROS Revenue Management

The exuberance of investors toward Internet companies in the late 1990’s stemmed from a realization that the Internet was destined to have an enormous impact on business. While most Internet start-ups no longer exist, their vision of how the Internet would change our lives is being realized at a pace even they might not have imagined. Revenue management and dynamic pricing in an e-commerce setting pre-date the Internet by at least two decades, as witnessed by historical developments in the travel and transportation industries. Yet, characteristics unique to the Internet are bringing about tremendous changes to distribution and therefore to the mathematical foundations of revenue management and dynamic pricing. We discuss some of the changes being brought about by the Internet and the research challenges they imply for the near future.



Title: Callable and Flexible Revenue Management
Speaker: Guillermo Gallego (Columbia University )
Joint work with Garud Iyengar, Steven Kou (Columbia University) and Robert Phillips (Stanford University)

Revenue management is utilized in industries where a low-to-high fare booking or sales process culminates with the delivery of a good or a service. The low to high fare structure requires a careful allocation of capacity to balance the cost of high-fare lost sales and unused capacity. Capacity providers can increase profits, service their customers better and improve capacity utilization if they can implement a mechanism to broker the transfer of capacity from low to high valuation customers. In this talk we investigate mechanisms to induce low fare customers to self-select and grant the capacity provider the option to recall their capacity. We discuss both the single-leg model where cash is the inducement and the network model where the inducement is a combination of cash and an alternative product.