Analysis of U.S. Airline Passengers' Refund and Exchange Behavior Across Multiple Airlines



Analysis of U.S. Airline Passengers' Refund and Exchange Behavior Across Multiple Airlines

Authors

Dan C Iliescu, Laurie A. Garrow, Georgia Institue of Technology, US; Roger A Parker, Boeing, US

Description

Airline passenger ticketing and exchange behavior is modeled across multiple carriers using hazard models and data from the major U.S. ticketing clearinghouse.

Abstract

The airline industry is very dynamic. Within the last four years, U.S. carriers have seen tremendous pressure to control costs while competing in a low-fare market that is being overtaken by low cost carriers. Multiple factors have contributed to the fact that today more than 50% of the U.S. airline capacity is operating under bankruptcy. While some of the factors leading to bankruptcy are well-recognized and include high fuel costs, high labor costs and increased market penetration of low-cost carriers, other factors, such as the impact of the Internet on purchasing behavior, is less understood. However, as carriers develop business plans for emerging from bankruptcy, it is clear that the emergence of the Internet as a major distribution channel cannot be overlooked. Specifically, Forrester Research estimates that in 2005 more than 32 million households in the U.S. will use the Internet to buy leisure trips, for which they will spend $63.3 billion, and that by 2009 these on-line sales will grow to $111 billion.

The emergence of the Internet as a major distribution channel can influence airline passengers? purchasing behavior in several ways. The Internet has led to a reduction in information search costs and an increase in price transparency for customers. In turn, many researchers have hypothesized that this has resulted in fierce price competition, less brand loyalty, and an increase in cancellation and rebooking rates. Increases in cancellation and rebooking rates would occur when a passenger purchases a ticket (e.g., for $500 with a $50 exchange fee) but keeps looking for cheaper fares up until the time of departure. If a cheaper fare becomes available (e.g., for $400), the passenger ?cancels? the original ticket, pays an exchange fee, and purchases the cheaper fare (e.g., in this case paying a total of $450 vs. $500). The difficulty in analyzing this problem is that, while a carrier can use their own data to examine rebooking behavior on their own flights, they cannot determine how many cancellations are due to passengers rebooking on competitor flights that have lower prices. To answer this question, ticketing and pricing data from multiple airlines is needed.
One source of this information in the U.S. is the Airline Reporting Corporation (ARC). ARC is the ticketing clearinghouse for many airlines in the U.S. and essentially keeps track of purchases, refunds, and exchanges for participating airlines and travel agencies. While not all U.S. carriers or distribution channels are represented in the data, ARC data is unique in the sense that ticketing transactions across multiple airlines can be observed. In addition, pricing, exchange, and partial itinerary information is available, which allows one to determine how often passengers are exchanging tickets for lower priced itineraries and whether these tickets are on the original carrier.

In this paper, we analyze airline passengers? cancellation and rebooking behavior across multiple carriers using 2004 ticketing data from ARC. Multiple origin-destination pairs are selected to examine differences in leisure and business markets as well as differences in market structure (e.g., percentage of low cost carriers, number of itineraries available, direction of travel, distance between origin-destination pairs, etc.). Hazard models are used to predict the probability an individual will ?survive? until the next time period or ?die? due to cancellation or exchange. Unlike many applications of hazard models, this application is unique in that the exact times of birth (ticketing date) and death are known, as are the exact causes of death (cancellation, exchange on different date, exchange to different itinerary, etc.).

In conclusion, this paper is one of (if not) the first to examine passenger cancellation and rebooking behavior across multiple carriers and provides new insights into airline passenger behavior (particularly related to differences between individual and group bookings). To the best of our knowledge, it is also the first published study to be based on ticket clearinghouse data from ARC. This is of particular interest to airlines because as the use of the Internet has increased, traditional data sources (such as Computerized Reservation System, CRS, data on booking reservations made through travel agencies) is becoming less reliable since it represents a smaller percentage of the market. For example, five years ago, CRS data represented about 80% of all booking transactions, while today, in some markets, it represents only about 20%. Moreover, given the backbone of many current revenue management systems is based on CRS booking data, there is increasing interest from airlines in assessing the viability of using alternative data sources, such as the ARC data. Consequently, this study will provide one of the first assessments of the benefits of using ARC data to model passenger ticketing, cancellation, and exchange behavior.

Publisher

Association for European Transport