AGIFORS Revenue Management 2005 Agenda 

Randy Bell - American Airlines

Larry Weatherford and Aleksey Khokhlov - University of Wyoming

Last year, we presented the theory of using Dynamic Programming (DP) based bid prices as a control mechanism and showed that, due to the “curse of dimensionality,” it took over 35 hours to solve a 2-leg network (30 seats on each leg).  This year, we will show how this curse can be overcome and very complex networks can be solved with DP in seconds.  We will also explore how to incorporate the stochastic nature of demand into the solution.

RM Performance Under Simplified Fare Structures

Peter Belobaba and Maital Dar - MIT International Center for Air Transportation

The recent trend towards simplified fare structures has raised questions about the performance of existing RM forecasting and optimization models. In this presentation, we use PODS simulation results to explore the impacts on traffic and revenues of the introduction of these less restricted fares in different networks. We then examine the performance of traditional RM approaches under alternative fare structures. Some modifications to traditional forecasting and optimization models are proposed, and their revenue impacts are compared to those of existing RM methodologies.

Inter-temporal Valuations, Product Design and Revenue Management

Guillermo Gallego and Ozge Sahin - Columbia University

We study inter-temporal choice models where customer valuations evolve over time. This requires discounts to induce people to book early and therefore leads to a low-to-high pricing scheme, at least in expectation. We show that for a wide range of capacity levels, selling partially refundable fares (or equivalently, selling call options on capacity) can result in significantly higher expected revenues than those obtained using traditional low-to-high revenue management techniques (e.g., booking and overbooking limits). The use of options allows customers to self-select products (including options) without the need of imposing fences like Saturday night stays. As a byproduct, we have observed that increasing the uncertainty of customer valuations benefits providers using options, but may hurt providers using traditional revenue management.

DAVN-MR: A Unified Theory of O&D Optimization in a mixed Network with

restricted and unrestricted fare products

Thomas Fiig (SAS), Karl Isler (Swiss), Craig Hopperstad (Hopperstad

Consulting Inc.), Richard Cléaz-Savoyen (MIT International Center for Air

Transportation)

We propose a new forecasting and optimization methodology DAVN-MR (Displacement Adjusted Virtual Nesting - Marginal Revenue). DAVN-MR extends the O&D network optimization principles for optimizing revenue in a restricted fare structure to an arbitrary mix of restricted and unrestricted fare products in O-D markets sharing the same leg(s) on a network. DAVN-MR is proven to be optimal for deterministic O&D demand. Simulations in PODS (Passenger Origin Destination Simulator) at MIT have shown that the method

shows significant revenue benefits that exceed those of existing RM-methods. Variants of DAVN-MR are in production at Swiss and SAS.

Demand Distributions for Revenue Management

Richard Ratliff - Sabre Holdings

This presentation involves new findings on the use of mixture distributions to reconcile some of the observed differences between the EMSRa and EMSRb revenue management optimization algorithms. Examples of the modified demand CDF's will be explored, with a special emphasis on the application of the Gamma distribution. Also, computational results of using high-order polynomial approximations for estimating traffic resulting from fare class allocation decisions (and its applicability in RM optimization modeling) will be presented.

Modeling in a competitive environment 

Igor Cakulev, Royce Kallesen and P.R. Narayanan - PROS Revenue Management

The proliferation of low fare carriers and the gradual erosion of fare fences and fare rules have caused the demand to buy down. This aspect of purchasing behavior has received much attention in the recent times through modeling of priceable demand to estimate the willingness to pay. Simulation studies performed thus far have shown that adopting hybrid demand model for forecasting and optimization provide positive revenue benefits. We extend these studies further to determine how the models perform in the presence of competition under different control environments. In this presentation, we will discuss new results of such modeling in a competitive environment.

Price Sensitive Train Management

Anand Medepalli and Bob Garner - Manugistics

This talk will focus on the train operator business problem, their issues in maximizing revenues and gaining market share in a competitive environment of low fare carriers and a customer environment of price sensitivity. The solution discussion will focus on price sensitive revenue management as the means to meet train operator objectives.

Integrated Pricing & O&D Revenue Management for restriction-free fare structures

Dieter Westermann - Lufthansa Systems Berlin GmbH

The introduction of restriction-free fare structures primarily by low-fare carriers invalidated important assumptions of commonly used revenue management algorithms. As a consequence these systems are unable to full-fill their task of maximizing revenue. Especially the usage of sophisticated approaches like O&D revenue management are questioned by the airlines.

However the unique features and capabilities of PNR-based O&D systems developed over the previous years offer network carriers a very good option to respond to the new challenge. The presentation goes into some details why traditional revenue management fails. In addition it addresses how an O&D system can become the solution to the complicated problem of a network carrier, which has to deal with both types of business across its network at the same time.

Empirical testability of sell-up effect in Revenue Management theory through organizational issues

Jean Michel Chapuis - University of La Rochelle

The main goal of this paper is to improve the testability of the Revenue Management theory, dealing with the effectiveness of sell-up techniques to show how this can be done. This research identifies three fields, namely the methodological heart, the technical redesign and the organizational border, where the theory can be confronted
with reality. The paper suggests that research methods or empirical tests should be matched with the field under consideration. A questionnaire is used to survey whether sell-up is profitable at the organizational level and how interdependence between departments influences its returns.

Integrating Science and Business Processes to Maximize Revenue

Brian Wishlinski - PROS Revenue Management

Revenue Management departments have relied on scientific solutions to maximize their airlines' revenue performance. Revenue performance can suffer, however, without documented business processes. Many business practices employed today by revenue management departments can circumvent the optimization recommendations, thus causing revenue degradation. This talk will focus on the integration of leading-edge

forecasting and optimization principles with industry best practices in revenue management business processes. A methodology will be explored to document these processes, train the analyst community as well as measure their performance. Case studies will be presented to articulate these points.

Computing virtual nesting controls for network revenue management under customer choice behavior

Gustavo Vulcano (NYU) and Garrett van Ryzin (Columbia University)

We consider a revenue management, network capacity control problem in a setting where heterogeneous customers choose among the various products offered by a firm (e.g., different flight times, fare classes and/or routings). Customers may therefore substitute if their preferred products are not offered. These individual customer choice decisions are modeled as a very general stochastic sequence of customers, each of whom has an ordered list of preferences. Minimal assumptions are made about the statistical properties of this demand sequence. We assume the firm controls the availability of products using a virtual nesting control strategy and would like to optimize the protection levels for its virtual classes accounting for the (potentially quite complex) choice behavior of its customers.

We formulate a continuous demand and capacity approximation for this problem which allows for the partial acceptance of requests for products. The model admits an efficient calculation of the sample path gradient of the network revenue function. This gradient is then used to construct a stochastic steepest ascent algorithm. We show the algorithm converges (a.s.) to a stationary point of the expected revenue function under mild conditions. The algorithm is relatively efficient even on large network problems, and it produces significant revenue increases in our simulation experiments. The examples also provide interesting insights into how protection levels should be adjusted to account for choice behavior. Overall, the results indicate that choice behavior has a significant impact on both capacity control decisions and revenue performance and that our method is a viable approach for addressing the problem.

Behavior of Revenue Management if the Models Ignore Competition

Anton J. Kleywegt (Georgia Institute of Technology) , Tito Homem-de-Mello (Northwestern University), and

Bill Cooper (University of Minnesota)

In many revenue management models, each seller models demand as a function of the prices of that seller only.

That is, the demand models that sellers estimate with data, model the quantity demanded as a function of the

prices of the seller, and these models do not include the prices of other sellers of substitute products. One may expect that in equilibrium, the demand data that are observed by each seller are conditioned on the equilibrium prices of the other sellers, and therefore such models can implicitly capture the effect of the prices of the other sellers. In this research we investigate what a statement such as the one in the previous sentence may mean in settings in which competitors observe a sequence of data, and at each step update their demand models and their prices. We compare the resulting outcomes with the outcomes in other settings, such as the equilibria of omniscient competitors, and of collaborators.

A new framework for Revenue Management and Pricing Optimization based on Bilevel Programming 

Jean-François Pagé (Air Canada), Jean-Philippe Côté (ExPretio Technologies and Université de Montréal), Gilles Savard (École Polytechnique de Montréal)

As traditional RM systems reach their practical limitations, some research and development has recently been done to propose new directions, for instance choice- based models with statistical modeling, rule-based systems, etc. We propose a new framework based on bilevel optimization where both airline and customers objectives are taken into account, as well as network topology and the pricing actions of competitors (market response). Details of the model will be presented along with different contexts (tactical, strategic, joint RM and Pricing Optimization) for which this approach is appropriate. We will also discuss the current state of work and introduce some future research directions.

Yield Management in the New Fare Environment

Krishnan Saranathan and Wen Zhao - United Airlines

Traditional yield management systems have been less effective in the new fare environment with limited fare fences, resulting in significant losses due to revenue dilution. We will present our efforts to develop a new set of models for forecasting, optimization, and inventory control. We will discuss our partnership with PROS Revenue Management to develop and implement a new forecasting model that takes into account of customer buy down behavior. We will also review our new dynamic programming based optimizer that eliminates most of outdated assumptions in EMSRb based model, integrates overbooking and seat allocation decisions. The impact of our efforts to enhance the control mechanism with our reservation system to handle local and connecting traffic differently will also be reviewed. Finally, we will review the results from our simulation analyses and discuss change management issues and challenges

Revenue Maangement of Korean Air

Haewoon Yang - Korean Air

Late 2001 Korean Air started using Revenue Management as a solution of revenue increment with introduction of PROS RM Product. Korean market environment for RM operation is quite different from American market or European, and then the carrier has been experiencing several kinds of trouble while settling the system into our environment and business organization. On the other hand, the Korean carrier has been calculating certain amount of revenue increment by its own simulation (shown in presentation) and wish to show how business structure has been changed and training, evaluation and etc for the smooth operation of RM system. Those carriers which want to operate RM system efficiently in their own business environments may get some help from my presentation, I hope. However, the carrier has also been agonizing over some problems and then wants to hear good advices from the experts or other airline counterparts.

Vendor Presentation

SKYPRICE - NAVITAIRE RM SOLUTION FOR SIMPLIFIED FARE ENVIRONMENT

Petr Kocandrle - Navitaire

AAI - Revenue Integrity Solutions in an ASP Model

Frank Arciuolo - Airline Automation, Inc.

QL2 Software - Airfare Information Service (AIS)

Paul Campbell - QL2 Software, Inc.