Effect of airline choice and temporality on flight delays

https://doi.org/10.1016/j.jairtraman.2020.101813Get rights and content

Highlights

  • Service failures are rampant in the US civil aviation sector.

  • We analyze delays using a novel service recovery/double deviation framework.

  • We identify consumer-centric factors as predictors of departure and arrival delays.

  • We predict delay likelihoods using mixed-logit and multinomial regression.

Abstract

The civil aviation industry is critical to the economic development of any nation. Being a key infrastructure service, the performance of the civil aviation industry spills over to other segments of the economy. We investigate service failure among US airlines, using departure delay, and model its likelihood using mixed-logit regression. We characterize arrival performance for delayed departures using a service recovery/double deviation framework. We model the likelihood of different arrival possibilities using multinomial logistic regression. We use the chosen airline, temporality, and flight duration as predictors. Our key results include: (i) Afternoon/evening flights, and longer duration flights are more likely to depart late; (ii) Delta Airlines is most likely to depart on time; (iii) Southwest Airlines is best in recovering from delayed departures; (iv) United Airlines is most likely to aggravate delays.

Introduction

Services constitute more than two-thirds of the world's GDP ( World Bank Report, 2019). Improvement in infrastructure services is critical to economic development. The civil aviation sector is vital in propelling economic growth (Baker et al., 2015; Dunbar, 1990; Fung et al., 2006; Weisbrod, 1990). It also contributes significantly to job growth, social welfare, and is thus vital to public policy (Caves, 2003).

The United States is one of the world's largest domestic civil aviation market, worth US$179 billion (Euromonitor, 2018), with the top four airlines- American Airlines, Delta Airlines, Southwest Airlines, and United Airlines accounting for 17.6%, 17.4%, 17.0%, and 14.9% of total domestic revenue passenger miles (US Bureau of Transportation Statistics, 2020). A billion passengers traveled by air in the US in 2019 (Nov 2018–Oct 2019), of which around 80% used domestic scheduled flights (US Bureau of Transportation Statistics, 2020). Mazzeo (2003) emphasizes that on-time performance is a critical part of a consumer's perceived utility. Aggregate indicators of on-time performance (or the lack of it) also reflect the overall efficiency of a nation's civil aviation sector.

The economic impact of aviation delays includes an enhanced direct cost for the airline, lost man-hours for those traveling for work, and opportunity costs of time wastage for leisure (Peterson et al., 2013). A 2010 FAA (Federal Aviation Administration) study estimated the annual economic costs associated with flight delays at $31.2 billion (Ball et al., 2010). Another study by Ferguson et al. (2013) estimates these costs at $40.2 billion. The enormity of these estimates propels us to think of the enormous consequences and motivate the exploration of factors that relate to and can predict delays. We focus on aviation delays in the United States, with particular attention to the top four airlines, which account for around 70% of the domestic aviation market.

Various factors cause flight delays, including those which are outside the purview of airlines such as weather, air traffic control processes, security procedures (Bhat, 1995; Taylor, 1994). At the same time, these external contingencies impinge on each airline. In a competitive market, consumers make comparisons between different airlines, and it is the relative performance that counts (Park et al., 2004, 2006). Kim and Park (2016) establish that airline delays cause negative emotional responses, enhance negative word-of-mouth, and lower repurchase intention. Airline delay also impacts performance dimensions like pricing power, market share, and market growth. Forbes (2008) reports that each minute of delay reduces the ticket price by $1.42 on average, which increases with the degree of competition.

There is a dearth of studies relating airline delays to factors which the consumers can incorporate into their decision of which airline service to buy. This study considers consumer-centric factors such as the choice of airline, the departure day and time, and flight duration, as drivers of flight delays. The availability of heuristics like departure delay probability and arrival performance, as a function of airline and temporality is a primary aid to help consumers in making purchase decision(s). At a macro level, these indicators can guide policy-making, such as incentivizing on-time performance and managing airport congestion. Thus, our study aims to be of use to multiple stakeholders in the civil aviation industry.

We adopt a service failure and recovery framework in our analysis of flight delays. We characterize departure delays as initial service failures. Then we study the arrival performance as a service recovery opportunity. Making up for the initial departure delay by reduced arrival delay is considered a successful recovery. An increase in the eventual arrival delay, relative to the initial departure delay, is termed “double deviation” (Bitner et al., 1990). This analysis, being intuitive and straightforward, is a powerful heuristic that customers can employ while evaluating different airlines. The conceptual model is illustrated in Fig. 1.

Prior research has characterized and empirically assessed departure delay and waiting as service failure (Boshoff, 1997; Taylor, 1994; Wen and Geng-qing Chi, 2013). Thus delays are accepted as an important manifestation of service failure in travel services. However, to the best of our knowledge, no study has assessed departure delay as initial service failure and arrival performance as a service recovery opportunity.

We analyze the likelihood of departure delay and the likelihood of arrival performance, subject to departure delay. Specifically, we address the following research questions:

  • (1)

    How does the probability of departure delay change based on intra-day (AM vs. PM departure) and inter-day temporality (Weekday vs. Weekend departure)?

  • (2)

    How does the probability of departure delay vary for the top four US airlines by market share {American Airlines, Delta Airlines, Southwest Airlines, United Airlines} flying in the US domestic market?

  • (3)

    How does the probability of departure delay change with an increase in the flight duration?

Subsequently, we address the following research questions concerning arrival performance of airlines modeled as the probability of three events- {recovery/no recovery/double deviation}:

  • (4)

    How does the arrival performance change based on intra-day (AM vs. PM departure) and inter-day temporality (Weekday vs. Weekend departure)?

  • (5)

    How does the arrival performance vary for the top four US airlines by market share {American Airlines, Delta Airlines, Southwest Airlines, United Airlines} flying in the US domestic market?

  • (6)

    How does the arrival performance change with an increase in the flight duration?

Following this introduction, the rest of the paper is organized into six sections. We start by reviewing the existing literature. Subsequently, we present the analysis and results, first in terms of departures, and then in terms of arrival performance. Then we state the theoretical and managerial implications. Finally, we highlight the limitations of our work and conclude with future research directions.

Section snippets

Literature review

Delays constitute a significant manifestation of service failure in transportation services (Cheng and Tsai, 2014; Schmöcker et al., 2005). Bamford and Xystouri (2005) report that delay is one of the three major consumer complaint areas. Morrison et al. (1989) find that every one percent increase in delayed flights reduces customers’ willingness to pay by $1.31 per single fare.

Different variables explain airline delays in previous research. In developing an understanding of a phenomenon, it is

Departure delay analysis

We model the probability of a flight departing late in the US aviation market using the well-known mixed logistic regression model (Andrews et al., 2002; Ben-Akiva et al., 2002; Fiebig et al., 2010; Hensher and Greene, 2003; Hess and Polak, 2005; McFadden and Train, 2000). We model the probability of departure delay as a function of airline choice, the flight duration, and two intra-day and inter-day temporal parameters. We account for intra-day effects by segmenting the flight departure time

Arrival performance

We now analyze flight delays on arrival. This analysis complements the earlier assessment of departure delays. We take a novel approach in characterizing arrival performance as a service recovery opportunity, once service failure in terms of departure delay has already occurred.

We divide the arrival performance (contingent on departure delay) into three possibilities:

  • 1.

    Recovery: Once a service failure occurs in terms of departure delay, airlines can and do attempt to make up for delays.

Implications

This paper identifies and analyzes one dimension (delays) of the service failure and recovery likelihoods in the US civil aviation industry, based on extensive empirical analysis. Specifically, the study finds that PM departure flights are more likely to exhibit service failure (delayed departure) but also more likely to show service recovery and less likely to exhibit double deviation. This brings out the fact that despite technological advancements, the US aviation sector is unable to handle

Limitations of the study

On-time performance at arrival is impacted significantly by airline schedules, specifically, the extent of padding. Although we argue that all observed increase(s) in the scheduled duration of flights should not be attributed to padding. However, the possibility of padding cannot be denied. Bhadra (2009) emphasizes that increased block times are a source of inefficiency, and structural reforms are needed to control it. Taking a contrary view, Fan (2019) cautions that padding may be a misnomer

Directions for future research

We restrict this work to the US aviation market. It is desirable to identify customer-centric factors explaining airline delays in other countries, particularly emerging markets. Global airlines need to incorporate cultural factors of the origin and destination country to optimize service delivery and reduce the negative impact of service failures.

This study has analyzed four airlines individually without looking at their operating models. It would be prudent to attempt the same analysis by

CRediT authorship contribution statement

Swapan Deep Arora: Conceptualization, Methodology, Writing - original draft. Sameer Mathur: Resources, Data curation, Writing - review & editing.

Declaration of competing interest

None.

Acknowledgments

Both the authors are thankful to Mr. Aryansh Gupta ([email protected]) for providing valuable research assistance toward this project. Sameer Mathur is grateful to the Indian Institute of Management, Lucknow, for providing financial support in the form of Seed Money Grant SM 260.

Swapan Deep Arora is a doctoral student (FPM), in the Marketing area at the Indian Institute of Management, Lucknow, India. He has a bachelor's degree from Thapar University, India and over 15 years of industry experience.

References (68)

  • S.J. Forbes

    The effect of air traffic delays on airline prices

    Int. J. Ind. Organ.

    (2008)
  • C. Goodwin et al.

    Consumer responses to service failures: influence of procedural and interactional fairness perceptions

    J. Bus. Res.

    (1992)
  • D. Greenfield

    Competition and service quality: new evidence from the airline industry

    Econ. Transport.

    (2014)
  • S. Hess et al.

    Mixed logit modelling of airport choice in multi-airport regions

    J. Air Transport. Manag.

    (2005)
  • M.S. Kim

    Analysis of short-term forecasting for flight arrival time

    J. Air Transport. Manag.

    (2016)
  • N.Y. Kim et al.

    A study on the impact of airline service delays on emotional reactions and customer behavior

    J. Air Transport. Manag.

    (2016)
  • M.C. Lonzius et al.

    Robust scheduling: an empirical study of its impact on air traffic delays

    Transport. Res. E Logist. Transport. Rev.

    (2017)
  • B. Lubbe et al.

    Flight delays: towards measuring the cost to corporations

    J. Air Transport. Manag.

    (2012)
  • J.-W. Park et al.

    The effect of airline service quality on passengers' behavioural intentions: a Korean case study

    J. Air Transport. Manag.

    (2004)
  • G. Santos et al.

    Determinants of delays at European airports

    Transp. Res. Part B Methodol.

    (2010)
  • Y. Suzuki

    The relationship between on-time performance and airline market share: a new approach

    Transport. Res. E Logist. Transport. Rev.

    (2000)
  • J. Yimga et al.

    Airline schedule padding and consumer choice behavior

    J. Air Transport. Manag.

    (2019)
  • S.S. Allan et al.

    Analysis of delay causality at newark international airport

  • R.L. Andrews et al.

    An empirical comparison of logit choice models with discrete versus continuous representations of heterogeneity

    J. Market. Res.

    (2002)
  • M. Ball et al.

    Total Delay Impact Study: A Comprehensive Assessment of the Costs and Impacts of Flight Delay in the United States

    (2010)
  • D. Bamford et al.

    A case study of service failure and recovery within an international airline

    Manag. Serv. Qual.: Int. J.

    (2005)
  • M. Ben-Akiva et al.

    Hybrid choice models: progress and challenges

    Market. Lett.

    (2002)
  • V.N. Bhat

    A multivariate analysis of airline flight delays

    Int. J. Qual. Reliab. Manag.

    (1995)
  • M.J. Bitner et al.

    The service encounter: diagnosing favorable and unfavorable incidents

    J. Market.

    (1990)
  • C. Boshoff

    An experimental study of service recovery options

    Int. J. Serv. Ind. Manag.

    (1997)
  • S. Bratu et al.

    An analysis of passenger delays using flight operations and passenger booking data

    Air Traffic Contr. Q

    (2005)
  • K. Butcher et al.

    Waiting for service: modelling the effectiveness of service interventions

    Serv. Bus.

    (2008)
  • R. Caves

    The social and economic benefits of aviation

    Towards Sustain. Aviat.

    (2003)
  • Y.-H. Cheng et al.

    Train delay and perceived-wait time: passengers' perspective

    Transport Rev.

    (2014)
  • Cited by (11)

    • Flight delay causality: Machine learning technique in conjunction with random parameter statistical analysis

      2023, International Journal of Transportation Science and Technology
      Citation Excerpt :

      Results presented in Table 5 show that when the time period of 0–60 min as the base category is taken into consideration, the probability of on-time arrival increases by 7.92% and 26.21% for the time periods of 60–120 and 120–180; however, it decreases by 3.59% and 11.68% for time periods of 180–240 and over 240 min. This finding is in line with the results of a recent study illustrating the longer flight duration is positively associated with the probability of service recovery in arrival delays (Arora and Mathur, 2020). The reasons underlying the heterogeneous impact of air time on arrival delay may be attributed to the implementation of the distance-based Ground Delay Program (GDP).

    • How do airlines react to slot displacements? Evidence from a major airport

      2023, Journal of Air Transport Management
      Citation Excerpt :

      Most of the first applications in aviation also focused on accidents, e.g. Li and Baker (1995), McFadden (1996, 1997) and Gudmundsson (1999). In more recent times, aviation accidents (and incidents) have continued to be investigated through logistic regression models (Bazargan and Guzhva, 2011; Boyd, 2015; Dai et al., 2021), but several other topics have also been explored, such as air traffic forecasts (Kotegawa et al., 2010; Rajendran et al., 2021), flight cancellations (Xiong and Hansen, 2013), flight rerouting (Evans et al., 2018) and flight delays (Wang and Vaze, 2016; Arora and Mathur, 2020; de Oliveira et al., 2021), among others. In this section, we first summarize essential knowledge on the formulation and interpretation of logistic regression models, and then on its estimation and assessment.

    • Resilient airline scheduling to minimize delay risks

      2022, Transportation Research Part C: Emerging Technologies
      Citation Excerpt :

      Recently, there have been few studies which follow data-driven frameworks focusing on flight delay. Arora and Mathur (2020) analyze the impact of delayed departure of flights on their arrivals by multinomial logistic regression. Lambelho et al. (2020) propose a machine-learning based mechanism to assess the flight schedules of airlines with respect to their cancellation and delay predictions.

    View all citing articles on Scopus

    Swapan Deep Arora is a doctoral student (FPM), in the Marketing area at the Indian Institute of Management, Lucknow, India. He has a bachelor's degree from Thapar University, India and over 15 years of industry experience.

    Prof. Sameer Mathur is a faculty in the Marketing area at the Indian Institute of Management, Lucknow, India. He has a Ph.D. in Marketing from Carnegie Mellon University, USA.

    View full text