AGIFORS Reservations
and Yield Management 2003
Technical Program
Presentations
( Based on submitted abstracts to date*
)
“Bid Price Control at US Airways: Benefits and Costs of
O&D Revenue Management”
Rick Zeni, US
Airways
US Airways has been using true
bid price O&D control for the past four years.
During this presentation we will
share our experiences with the benefits and challenges
faced by the revenue management
group at the airline. We
will review the evolution of
the business process as well as
system implementations.
"Revenue Management with Restriction-free Pricing"
Shankar Mishra, Sabre, Inc.
Impacts on RM Performance of Low-Fare Carriers and Alternative
Fare Structures"
Peter Belobaba, Andrew Cusano,
Massachusetts
Institute of Technology
Recent changes in traditional
airline fare structures, brought about by
changing demand characteristics
and increasing low-fare carrier
competition, have raised
questions about the effectiveness of RM systems in
terms of generating incremental
revenues. In this
presentation, we use the
Passenger Origin-Destination
Simulator (PODS) to examine the impacts on the
performance of leg-based and O-D
control RM systems under a variety of
alternative fare structures with
different ratios of full and discount
fares, reduced advance purchase
requirements and less stringent fare
restrictions.
The simulation results focus both on the overall revenue
changes associated with these
alternative fare structures and the
incremental revenue gains of RM
methods given these new structures.
“Optimal Selection of Data
Collection Points”
Edward Kambour. PhD, PROS Revenue Management
A key element of implementing a
Revenue Management system is the selection of the Data Collection
Points (DCPs). The DCPs are the time points during the booking cycle when
bookings are sent to the forecaster.
From a scientific perspective, the DCPs are meant to
partition the booking period into intervals for which the rate is
constant (homogeneous). On
one hand, using many DCPs is likely to achieve homogeneity.
On the other hand, fewer DCPs will lead to greater forecast
stability. This
presentation will describe a method for selecting the optimal number
and placement of DCPs.
"Constrained
Forecasts"
Ronald P. Menich, Ph.D, Manugistics
Reservation
inventory control systems are often quite complex, with different
controls at various aggregation levels, and sometimes built-in
profiling functions for automatically changing these values over
time as bookings accrue and departure nears.
Given this complexity, it is of interest to ascertain how
well recommended inventory controls from a Revenue Management System
will perform. Constrained
forecasts simulate the acceptance and rejection of unconstrained
booking requests subject to recommended controls, and form the basis
for revenue, projected load factor, and other performance metrics.
We will discuss a deterministic simulation for quickly
producing constrained forecasts within a Revenue Management System
"Some Static Network Revenue
Management Models With Asymptotic Property"
Xiubin Wang, University of
Wisconsin-Superior
In the area of airline revenue
management, static network revenue management models that usually
adopt discrete seat allocation method show an advantage of
computational applicability on a large complex service network over
the single-leg based dynamic pricing ones. In this paper, we first
examine a non-linear programming allocation model and show its
asymptotic property. We further present two classes of stochastic
and deterministic models with asymptotic property. In addition, we
show that the early static network yield management model developed
by Curry in 1990 is asymptotically optimal in a stochastic and
dynamic setting. Our result generally depends on a weak assumption
of incremental stochastic demand only.
"Market
Airline Class Revenue Management"
Craig Hopperstad, Hopperstad Consulting, Inc.
In the twenty-odd year history of
airline Revenue Management there has been a relatively steady
progression in the sophistication of forecasting and optimization
methodologies. This
progress has not been matched with improvements in the modeling of
the underlying phenomena (an initial assumption of leg/class
independence followed by an assumption of path/class independence).
This paper represents a step forward in the modeling of the
phenomena, describing an approach to incorporating the assumption of
independent market/airline/class demand.
Results are provided from studies using PODS (Passenger
Origin/Destination Simulator).
8.
"EMSR
vs. EMSU: Revenue
or Utility?"
Professor
Larry Weatherford, University of Wyoming
EMSR
has been used for over a decade as the standard for leg seat
inventory control. But
what if, your airline company is not risk-neutral (as EMSR assumes)?
This presentation will review EMSR calculations and also
introduce Expected Marginal Seat Utility (EMSU), which allows for
the possibility to factor in risk-averseness in making seat
allocation decisions between the various fare classes.
Numerical results are presented.
9.
Modeling Price Elasticity
Katia Frank, Kiran Ravulapati, Wassim Chaar,
Delta Technology,
Atlanta, GA
We study the
relation between price and demand in an airline setting. Airline demand is unique because of the revenue management
controls used for customer segmentation.
This characteristic makes the estimation of true demand at
different price levels very difficult.
We present a model that estimates the demand under certain
conditions and assumptions by eliminating the revenue management
effects from the data. The
resulting demand curve is then used to obtain price elasticity for
airline products at different fare levels.
We used this model for different types of markets and
products with reasonable
10
"O&D
Passenger Demand Forecasting"
Alvin Lim, John Blankenbaker, Wassim Chaar,
Delta Technology
An
airline origin-destination (OD) revenue management software system
accepts or rejects requests for booking on airline seats by
comparing the fare for the request to a minimum acceptable fare
predetermined by the system. In
order to determine the minimum acceptable fare, the system typically
solves an optimization problem that takes as input the airline’s
flight network schedule, fares, forecasted passenger demands, and
available capacity on each of the airline’s scheduled flights.
Apart from the passenger demand forecasts, the airline has
almost full control over the other input components.
It turns out, however, that forecasting demand for airline
products is not a trivial task, both because of the large numbers of
items to be forecasted and the large amount of uncertainty in many
of the data series.
In
this talk, we describe a top-down method for inferring OD passenger
demand forecasts from readily available flight segment passenger
demand forecasts. The novelty of the proposed approach is its
ability to blend historical product usage, leg level forecasts, and
future flight schedules to estimate future OD level demands.
Furthermore, the method provides a natural OD control mechanism that
is lacking from many other OD revenue management systems.
We
will present an overview of the mathematical model as well as
numerical results from actual airline data.
"Adaptive Mechanisms in an OD Demand
Forecast”
Silvia Riedel Lufthansa
Systems Berlin
Adaptivity
is an important feature for revenue management forecast systems. Due
to continuously changing demand caused by seasonality, special
events like holidays or trade shows, changes in the flight schedules
or changes of the political or cultural situation of a
destination, there is a need for robust, adaptive forecasting
techniques able to cope with such changes. In this presentation an
overview of how such changes are handled in the new
forecasting system of Lufthansa Airline is presented. It
describes a general model of adaptation, which is based on abstract
terms of attractivity, attractivity changes and short term
influences.
* Technical
Program is subject to change
Call for papers
Come and share with us your ideas,
thoughts, current trends, philosophies, and latest technological
advances on topics ranging from revenue optimization, O&D
control, demand forecasting and more.
If you are interested in presenting at the study
group meeting, please complete the on-line
registration form and submit it directly via the web
Presentations are
subject to approval and time slots are available on a first come,
first serve basis - so if you are interested, please act now!
The AGIFORS 2003
Reservations and Yield Management Conference technical program is currently
being coordinated, for more information please contact gina.morello@aa.com.
|