Technical Program
Revenue Management Presentations ( Based
on submitted abstracts to date* )
Keynote Presentation
Rick
Zeni
JetBlue Airways
Choice
Architecture: Treating Your Customers like Humans
Rick
Elieson
American Airlines
Customers
have unprecedented access to information, and airlines often have little
control over how that information is conveyed. Because of the
disparate fare and inventory distribution model employed by most airlines
in the world, compounded by the continued disaggregation of fares due to
unbundling initiatives, airlines are at risk of becoming even further removed
from the decision of how to present their unique value proposition to the
customer.
In
his presentation, Rick will discuss how Revenue Management is on a course
to converge with Merchandising as the airline pricing model continues to
evolve. Borrowing examples from the world of behavioral economics, he
will show why it is important that we be very active in that evolution,
particularly in defining the decision sets that are presented to
customers. Distilled into 10 principles of effective choice architecture,
he will illustrate how to simultaneously improve both revenue and customer satisfaction, …but where airlines need a new mindset and
a host of new research.
STATISTICAL
METHODS FOR ESTIMATING SELL-UP RATES FROM HISTORICAL BOOKING DATA
Peter Belobaba and Chris Boyer
MIT International Center for Air
Transportation
Recently developed methods like hybrid forecasting and fare adjustment for
optimization, designed to reverse the revenue losses of spiral down, depend
explicitly on estimates of expected passenger sell-up behavior. We
provide an overview of approaches developed in the MIT PODS Consortium to
estimate these sell-up rates from historical booking data readily available
in RM databases. We report on recent findings with respect to the
performance of these approaches, based on PODS simulations. We then
introduce statistical methods for clustering estimates from different
markets in an effort to overcome the problem of data sparseness which seems
to be inherent in sell-up rate estimation. The results of the
clustering provide insights about the characteristics of “good” estimates
of sell-up rates, as well as intuition about the relationships between
estimated sell-up rates and various market characteristics.
PERFORMANCE
OF DYNAMIC PROGRAMMING IN RM SYSTEMS
Sarvee
Diwan and Peter Belobaba
MIT International Center for Air
Transportation
Dynamic programming offers a theoretically attractive alternative to
traditional RM optimizers because DP models explicitly account for the
pattern of passenger arrivals and calculate dynamic booking limits as a
function of remaining capacity. We first develop a simplified
simulator to evaluate the effects of actual demand variance on the
performance of standard DP on a single flight leg by excluding the effects
of forecast quality and competition. Sensitivity analyses show that the
performance of DP relative to EMSRb depends on
the demand variability, demand factor, fare ratios and passenger arrival
pattern. We also evaluate DP in a larger competitive network using PODS,
the Passenger Origin Destination Simulator. Despite not showing
significant benefits in the simplified simulator, DP can outperform EMSRb as part of a complete RM forecasting and
optimization system in a competitive environment, due to its more aggressive
seat protection and resulting passenger sell-up.
Forecasting and optimization of fare
families in network revenue management -
Policy based Revenue Management applied to fare families
Thomas Fiig, Karl Isler, Craig
Hopperstad, and S. Olsen
SAS,
Swiss, Hopperstad Consulting
The paper
regards the general forecasting and optimization of semi-differentiated
fare structures. In particular we will focus the important problem of fare
families, which we expect will ultimately be the consensus fare structure
amongst airlines.
Forecasting
for fare families is based on a general choice model, which explicitly
represents passenger travel purpose, willingness to pay and product
preference (taking into account sell-up and buy-down between the families).
This allows a forecast to be produced for each policy (set of lowest open
classes by family). In contrast, hybrid forecasting cannot produce policy
forecasts, since it is limited to produce forecasts only for classes that
close in reverse fare order. Hence it is suboptimal.
Policy-based
revenue management (cf. Authors presentation AGIFORS, AMS 2009) offers a
framework for optimizing an arbitrary fare structure by applying a marginal
revenue transformation to the fares and the demand of policy forecasts, converting
the fare structure to an equivalent independent demand model. The resulting
(independent) problem will be solved by DAVN using DP (Dynamical
Programming) on the legs. Benchmark studies are reported for a network
set-up both for monopoly and competitive scenarios.
Choice-Based
Revenue Management
Garrett Van Ryzin
Columbia
University
Over the past decade, using consumer choice models for revenue management
(RM) has transitioned from being just an interesting theoretical concept to
a practical reality. Choice models have many advantages; they can naturally
model important buy-up and diversion phenomenon and can be applied to
newer, undifferentiated low-fare structures and dynamic pricing problems.
They also pose unique challenges in terms of data, forecasting and
optimization. In this talk, we survey work in this area and discuss its
implications for RM practice.
The Top 10 Lessons of PODS
Bill Brunger
and Peter Belobaba
MIT International Center for Air
Transportation
The MIT/PODS Consortium has been sponsoring excellent academic
research for more than 15 years. The author will describe the 10 most
interesting (you be the judge...) findings which have practical application
to airline management.
Best
regards.
OR
Models for Online Group Sales Merchandising
Richard Ratliff
Sabre
This talk
will cover two related topics in online group sales merchandising: 1)
importance of display ordering and 2) dynamic packaging of air and hotel
content. The application of customer choice models to both of these
areas in Sabre’s group e-commerce platform will
be discussed.
Dynamic
models to assess revenue-sharing agreements in airline alliances
Robert
Shumsky, Christopher
Wright, and Harry Groenevelt
Tuck
School of Business at Dartmouth
We examine
dynamic game-theory models of two-partner airline alliances to analyze the
effects of various proration schemes on partner behavior and alliance-wide
revenue. We begin with models that assume complete information: each
airline knows its partner's inventory levels, and both have identical
forecasts of future arrival probabilities and revenue distributions over
the entire alliance network. We then consider alliances that
operate with incomplete information, assuming that partners only know each
others' bid prices. We explore the airlines' equilibrium policies and
compare the overall performance of the alliance to a centrally-controlled
network.
Airline
Alliance Collaboration and Operations
Xing
Hu, Rene Caldentey,
and Gustavo Vulcano
NYU Stern School of Business
We
investigate contractual agreements between multiple airlines operating
within a network alliance. Each airline controls a set of available flights
and its own reservation system. We study contracts that specify how revenue
generated by itineraries involving multiple airlines should be split among
the carriers. We propose a two-step hierarchical approach. First, we
formulate a static problem in which airlines select partitioned booking
limit and show that a simple transfer price mechanism achieves first best.
Next, we study the dynamic problem and prove that the static transfer
prices are asymptotically optimal.
Modeling dependent demand in a
network-based Revenue Opportunity Model
Christian Temath,
Michael Frank, and Stefan Pölt
University of Paderborn
and Lufthansa German Airlines
Techniques for performance
measurement are an integral part of a revenue management (RM) system. The
Revenue Opportunity Model (ROM) is a widely known method to measure revenue
management performance. However, a traditional ROM does not show valid
results for dependent demand structures and for network-based RM controls.
While earlier studies focused on the robustness of the network-based ROM
with independent demand, this presentation describes an extension to
dependent demand structures. Furthermore we analyze the effect of unconstraining and forecast errors on ROM robustness
under different scenarios.
Introduction
of REMATE
Michael Frank, Tobias Schröder,
Catherine Cleophas, Max Gerlach
Deutsche
Lufthansa AG and Berlin
University
In this presentation we introduce a newly
developed decision support and training environment for revenue managers
called “Revenue Management Training for Experts“ (REMATE). It is based on stochastic
simulations and allows the user to create scenarios by specifying supply,
demand and an optimization and forecasting method. REMATE supports Leg, O&D and two hybrid approaches.
The user can enter various influences on the automated
systems, manipulating the forecast and the optimization outcome as well as
pricing and planning. The results serve as decision support with respect to
the impact of specific actions. Another feature of REMATE is game play: a risk-free training environment. It
allows multiple users to play against each other and to try steering
strategies in order to outperform competitors.
Finally we give a short overview of applications
and preliminary results.
The
Strategic Effects of Requiring Minimum Sale Volumes to Trigger Commissions
under Competition
Guillermo
Gallego and Masoud Talebian
Columbia University And University of New Castle
We
consider a game between two providers with fixed capacities that compete
for customers through
a broker. The broker works on commission margins and can influence customer
choices. We study
the effects of requiring minimum sale volumes, or thresholds, to trigger an
increase in commissions
when commission margins are fixed, and also when commission margins are
decision variables.
With fixed commission margins at least one provider has an incentive to impose minimum sale
volumes to trigger commissions, usually at the expense of the broker. In equilibrium, the
provider with a higher total commission on sellable capacity gets priority and there are cases where
the broker is forced to discard items that he cannot sell. If the providers also compete with
commission margins, the equilibrium results in less revenue for the smaller provider and more
revenue for the broker, with the revenue for the larger provider remaining unchanged.
Capacity
Sharing
Darius
Walczak
PROS
Abstract:
Surplus Economy passengers can be upgraded to
Business compartment if the latter is not expected to book full. This
is a simple example of capacity sharing where different classes of
demand that normally book in separate compartments are sometimes directed
to share a portion of airplane’s inventory. We show that sharing
capacity dynamically can provide an incremental revenue increase over typical
static optimization. We present both the exact dynamic solution to
the 2 compartment problem as well as some heuristic approaches to help with
implementation. We describe extensions to the problem: the three
compartment problem and the “moveable curtain” problem which are also of
importance to airlines.
Using
Revenue Management to Enhance the Customer Experience
Brian
Wishlinski
PROS
Revenue Management
Ancillary
revenue is quickly becoming a mechanism for airlines to enhance financial
performance. The revenue streams can be separated into two broad
categories - those revenues that can be attributed to specific customer
actions (baggage fees, preferential seating, lounge membership) and those
that can be attributed to specific flights (duty-free sales, on-board
entertainment, meals-for-purchase). In this presentation, we will
show how traditional revenue management systems can be adapted to
incorporate these streams, thus using revenue management to enhance the
customer experience.
Improving
Overbooking Economics Using E Tools
Andrew
Cusano
Delta
Overbooking
generates significant incremental revenue.
While overbooking models and algorithms have become more
sophisticated over the past years, the practical delivery methods of
customers changing flights and soliciting for volunteers on overbooked
flights has remained somewhat static.
Furthermore, e-commerce tools have been advanced significantly. These tools allow for greater reach to
contact and interact with customers.
This presentation
looks at one airline’s experience in trying to integrate the practical
delivery of overbooking via e-commerce tools in order to improve
overbooking economics. In
particular, the author outlines the use of e-commerce tools for flight
leveling purposes and volunteer solicitation on overbooked flights, the
issues with implementing these methods in e-commerce tools, and the lessons
learned from these implementations.
How
to estimate passenger show rates with a logical and combinatory approach
Karine Lacerte
and Christine Dupuis
Air Canada and Ecole Polytechnique
de Montreal
The logical analysis of data (LAD) consists in an intelligent form of data
mining. In this talk we will present the construction of a model that
improves the accuracy of predictions for show rates, in order to adjust the
overbooking levels. This approach was chosen not only for its originality,
but also for its success in various sectors. It differs from other
conventional data mining methods by its ability to detect combinatory
information about the passengers. The input consists in a number of
observations (passengers), each described by a vector of attributes derived
from characteristics such as booking class, day of the week, departure
time, itinerary origin… The LAD method detects sets of conditions for which
all corresponding passengers have a significantly higher or lower show rate
than the entire population. These are referred to as patterns, and they
consist in a number of bounded attributes. Air Canada’s tool for forecasts
and overbooking is based on historical statistics for each flight. Results
will be presented comparing this system with the LAD. These results show
the LAD to be very competitive and closer to actual show rates. In
addition, some very rich information can be extracted from the patterns and
used for various purposes.
Measuring
Forecast Accuracy. Myth or Reality?
Burak
Ozdaryal
United
Airlines
Lack of
knowledge of true demand poses a unique problem for Revenue Management
practitioners. Quantifying forecast quality is not necessarily intuitive to
an RM scientist focused on revenue improvement. Yet, relevance of forecast
quality to the end user cannot be underestimated. The primary
business benefit of forecast accuracy measurement is critical in the
adoption of a revenue enhancing ideas. The secondary benefit of being able
to identify revenue improvement opportunities cannot be overlooked.
We will
be discussing two alternatives United Air Lines has adopted in recent
years. The simpler method is simulation based and gives high level insights
about forecasting models and their sensitivity to input data accuracy. The
second approach is based on actual observations and stress tests
assumptions about demand which cannot be evaluated in the simulated
environment.
Vendor Presentation
The Evolution of Competitor Fare
Collection Systems
Chris Buckingham
Infare
Competitor fare collection systems
started as simple screen scrapers intended to provide a quick snapshot of competitors fares at a given point in time. These
systems have now evolved to provide business intelligence far beyond the
fares themselves such as availability, inventory steering practices,
schedule changes, and benchmarking. To do this effectively
improvements in these systems had to made to
incorporate disciplined data collection and database practices. Chris
Buckingham from Infare Solutions will
present the evolution of these systems and the current state of the art.
Cargo Presentations ( Based
on submitted abstracts to date* )
Performance
Measurement for Cargo Revenue Management
How can
airlines measure the success of their cargo revenue management system
implementation? What is the value of
cargo revenue management?
Presentations by American Airlines, Air Canada, and Virgin Atlantic
will be followed by a discussion on the topic.
Revenue
Management and Business Models
How can
cargo revenue management support an airline in identifying and implementing
a profitable business model? What
should cargo executives want from their revenue management team? Presentations by Lufthansa Consulting and
Millenium Aviation will be followed by a
discussion on the topic.
Advances
in Cargo Revenue Management
This
session will highlight recent developments in the application of
Operational Research to Air Cargo Revenue Management. Presentations will be made by JDA, PROS,
RTS, and Sabre.
Business
and Revenue Model Optimization (BRMO) in the Airline Industry
Ricardo
V. Pilon
Millennium
Aviation, Inc.
Like a
product has a life-cycle, so does a company’s business model and even
industries face business model life cycles. This presentation addresses the
fundamental core of a business model and its relationship to the scope and
approach of revenue management. It argues that effective and sustainable
growth in profitability as part of revenue management can only be achieved
when methodology and tools are intertwined with strategic (business)
planning and a company’s product/service delivery process. This, it finds,
can only be attained with a refined perspective of a company’s role in the
wider value chain.
Lumpy
and Late!
Setareh
Mardan
PROS
Cargo
industry has unique features that challenge traditional revenue management
solutions. Lumpy, highly volatile demand and very short booking cycles are
some of these features. We discuss these challenges and propose a Dynamic
Programming optimization model that adjusts the cargo acceptance criteria
in real time with the goal of optimizing financial performance. We present
numerical results to illustrate this performance.
Integrated
Network Planning using Air Cargo Traffic
Mariana LaDue
Sabre
mum profitability for your airline and the best value for your
customers start with sophisticated network planning. Competitive network
schedules, profitable fleet assignments and effective aircraft routing need
to be created by integrating air cargo demand and freighter fleet with the
traditional passenger inputs.
Revenue
Integrity - Combining pre-sales revenue optimization and post-sale revenue
retention
Christopher
Schmacke
Lufthansa
Consulting
A highly
complex business like the air cargo business needs a complete and holistic
approach to optimize and secure revenues. Revenue retention combines
analytical methods of revenue management with hands on revenue retention
strategies to improve hub traffics and protect revenue from leaking.
Cargo
Infusion
Mukundh
Parthasarathy & Raunak Singh
Sabre
Airlines that
implement a sound cargo revenue management strategy coupled with the use of
advanced technology and industry best practices will achieve significant
financial gains based on the anticipated rapid growth in the air cargo
sector.
World air
freight is expected to grow more rapidly than mail, averaging annual growth
of 6.5 percent through 2021, and world air cargo traffic is expected to
exceed 464 billion revenue ton kilometers in 2021. Unfortunately, the
increase in air freight cargo traffic as measured in RTKs does not
necessarily mean increase in profit. According to Mercer on Travel and
Transport specialty journal (2009), in the historical period from 1974
to 2008, the average annual increase of 5 percent to 7 percent in air cargo
traffic translated to an annual cargo yields decline of 2 percent to 3
percent.
The
greatest weakness of the cargo industry is not being proactive in terms of
managing challenges and capitalizing on opportunities. Even worse is the
high inertia and slow pace in reacting to challenges, adapting to changes
and adopting new initiatives and technologies.
The key
strength of the air cargo industry comes from two key partners — airlines
and freight forwarders. These players either work together for success or
work separately and fail. This applies to working together on all
initiatives that are put forth by IATA to introduce efficiency, achieve
cost reduction and improve customer service. The next key strength for
cargo carriers comes from being visionary and taking the first step in
improving tools and technologies. These carriers should empower their
employees by providing the best operational and decision-support
capabilities and tools. Leveraging these two key strengths could
potentially transform a business that is struggling into a win-win
situation for freight forwarders and carriers, creating substantial growth
potential and outstanding customer service.
Why is it
difficult to make a profit in the air cargo business? The paper looks at
how principles of revenue management, combined with streamlined business
processes and automation can help alleviate this problem.
|