|
Article Excerpt
Abstract
Understanding air travelers' carrier choices is an important issue for airline managers. Using the survey data recently collected in central Iowa, this article examines how travelers' airline choices are affected by their past service failure experiences (e.g., seat denials or considerable flight delays) with the candidate airlines. The effects of service failure experiences on traveler choice probabilities are estimated separately for business and leisure travelers, and by type of service failures (seat denials, flight delays, and baggage mishandling). The results indicate that, for both business and leisure trips, travelers' airline choices may not be affected by any service failure experience. This condition implies that, in each trip occasion, travelers may choose the airlines that maximize their utilities without regard to past service failures. Although this behavioral pattern may seem somewhat counter-intuitive, it is consistent with the random utility theory.
**********
In today's competitive airline environment, understanding travelers' carrier choice behaviors is an important issue for airline managers. If airlines can learn how travelers' airline choices are affected by the airfares and quality of carrier services, they can use such information to make effective decisions on pricing, yield management, and marketing strategies (Proussaloglou and Koppelman, 1995). To date, several studies have modeled the airline choice decisions of travelers by using discrete choice models (see, e.g., Morrison and Winston 1989; Nako 1992; Proussaloglou and Koppelman 1995; Yoo and Ashford 1996; Pels et al. 2001). While these studies all gave important implications to airlines regarding the primary determinants of travelers' airline choices, they have one common limitation.
The limitation is that previous airline choice studies did not consider how the service failure experiences of travelers, such as seat denials (bumping), flight delays, and baggage mishandling (lost, damaged, delayed, or pilfered), would affect their future airline choice decisions. From the airline management perspective, testing this "service-carryover" effect in airline choice models is an important topic, as carriers' service policies (e.g., on-time arrival and overbooking policies) may depend on the significance of this effect. If, for example, it is found that travelers' airline choices are significantly affected by their past seat denial (bumping) experiences, airlines may use such information to adjust their overbooking levels (see, e.g., McGill and Van Ryzin 1999 for a detailed review of airline overbooking studies). By the same token, if it is found that travelers' flight-delay experiences affect their future airline choices significantly, such information may be used by airlines to adjust the on-time arrival targets.
Currently, however, the nature and significance of these service-carryover effects are largely unknown because no empirical work on this topic has been conducted to date. Some airline choice studies (e.g., Morrison and Winston 1989; Nako 1992) have attempted to capture the effect of flight delays on travelers' carrier choices by using the on-time performance data of carriers (publicly available data on percentage of flights arriving on time) as a variable in the utility function. These studies, however, did not capture the service-carryover effect because the published data do not necessarily represent the actual experiences of decision makers.
Theoretically, two interpretations are possible for the expected effect of service failures. The first is that travelers' service failure experiences would negatively affect their future airline choices. If, for example, a traveler was bumped by an airline because of the overbooking at trip occasion t (time t), the traveler's choice probability for this airline at the subsequent trip occasion (time t+1) may become substantially lower than that at trip occasion t. Similarly, if a traveler experienced a considerable flight delay by an airline at time t, the traveler's choice probability for this airline at the subsequent trip occasion (t+1) may become substantially lower than that at time t. The underlying theory for this behavioral pattern is that travelers are loss averse (Tversky and Kahneman 1991) (1) with respect to airline service qualities, so that they are less likely to choose the same airlines after bad experiences (i.e., service failures) than after good experiences.
The second interpretation is that service failures may have no impact on travelers' airline choices. An implicit assumption employed in the previous paragraph is that travelers avoid the airlines with bad service experiences because (after the experiences) travelers will feel a higher risk of being "mistreated" by the carriers (i.e., carriers lost the "goodwill"; see, e.g., Chatwin 1999). It should be noted, however, that in reality the service failure probability of the future is unconditional on (independent from) the service failure outcomes of the past. Thus, if a traveler was bumped by an airline at time t, it does not mean that the traveler is more (or less) likely to be bumped by the airline at the next trip occasion (t+1). This condition implies that if travelers are "rational" decision makers who understand that the probability of experiencing service failures in any trip occasion is unconditional on their past service failure experiences, they may simply choose the airlines that maximize their utilities in each trip occasion, without regard to their past service failure experiences. This behavioral pattern is consistent with the random utility theory. (2)
In this article, we empirically test which of the two mutually exclusive hypotheses mentioned above would hold for air travelers; i.e., the loss aversion hypothesis or the no-service-carryover hypothesis. Using the survey data collected in central Iowa, we estimate a discrete-choice model of airline selection that incorporates the dynamic service-carryover effects. The effects of service failures on traveler choice probabilities are estimated separately for business and leisure travelers, and by type of service failures (seat denials, flight delays, and baggage mishandling). If travelers are loss averse with respect to airline service qualities, we should find that service failure experiences of travelers would negatively affect their future airline choices. If, on the other hand, travelers are "rational" decision makers, their airline choice decisions may not be affected by any of their past service failure experiences.
It should be noted that our intent is not to derive the generalizable implications that apply equally well to all airlines and regions. Since we use the data of only a single region (central Iowa), our results may not apply to the regions (and airlines) that are not considered in this study. Given these conditions, our study should be considered as (1) a case study of a particular region, and (2) an initial attempt to model the dynamic service-carryover effect in the airline industry.
MODEL
We model the airline choice behaviors of travelers by using the multinomial logit model. The logit model is widely used in many fields to analyze consumers' choice behaviors (see, e.g., Ben-Akiva and Lerman, 1985, for...
|
