Long-haul low cost airlines: Characteristics of the business model and sustainability of its cost advantages

Transportation Research Part A 106 (2017) 215–234

Contents lists available at ScienceDirect

Transportation Research Part A journal homepage: www.elsevier.com/locate/tra

Long-haul low cost airlines: Characteristics of the business model and sustainability of its cost advantages

MARK

⁎

Christian Soyk , Jürgen Ringbeck, Stefan Spinler Chair in Logistics Management, WHU - Otto Beisheim School of Management, Burgplatz 2, 56179 Vallendar, Germany

AR TI CLE I NF O

AB S T R A CT

Keywords: Airlines Business models Long-haul Low cost Point-to-point

Existing academic literature is inconclusive about characteristics and viability of the long-haul low cost airline business model, whereas several airlines of this type are emerging. This article aims to validate its defining characteristics by clustering a sample of 37 transatlantic airlines using principal component and hierarchical cluster analyses along a newly constructed long-haul airline business model framework. To contribute to the evaluation of business model viability, cost differences between clusters are uncovered subsequently followed by a discussion of their sustainability. Key findings include the characterization of the emerging long-haul LCC business model and its significant differences from existing legacy hub and leisure carrier models. On a cluster average, 33% lower unit costs compared to legacy hub carriers were identified, of which 24 percentage points were evaluated as sustainable.

1. Introduction Low cost carriers (LCCs) such as Ryanair and Southwest Airlines have revolutionized short- and medium-haul1 airline markets across the world since the 1990s. The validity of the business model is proven practically through higher profitabilities (cf. Corbo, in press). Additionally, many academic studies of short-/medium-haul LCCs were conducted, which agree on common underlying principles of the business model (cf. Fu et al., 2011; de Wit and Zuidberg, 2012; Cho et al., 2015; Fageda et al., 2015; Fu et al., 2015). LCCs that target long-haul markets (i.e., flights with stage lengths longer than 4000 km) are a more recent phenomenon. Since the 2000s, carriers of this type have emerged in the Asia-Pacific region, foremost AirAsia X in 2003 and Jetstar in 2007. Since 2015, this type of airline has also emerged in the North Atlantic market, for example Eurowings long-haul, Norwegian long-haul, Westjet longhaul and Wow air. Existing academic literature is inconclusive about the defining characteristics of the long-haul LCCs (cf. Morrell, 2008; Daft and Albers, 2012; De Poret et al., 2015; Whyte and Lohmann, 2015). Evenly important, there is no consistent view of the economic viability of long-haul LCCs. For example, Moreira et al. (2011) state that the “viability of long-haul LCC operations must be highliy questionable”. On the other hand, Daft and Albers (2012) note that “regular low cost, long-haul operations are possible if the traditional full-service carrier product is effectively unbundled and suitable trunk routes can be identified.”. To fill this gap in the literature, this paper aims to validate the defining characteristics of the long-haul LCC business model. To this end, we adapt and enhance an existing airline business model framework. This framework builds the foundation for the subsequent clustering of airlines using principal component and hierarchical cluster analyses. Furthermore, to contribute to the discussion on long-haul LCC business model viability, average stage-length adjusted cost differences between clusters are compared. The sources of cost advantages and their sustainability are discussed in further detail.

⁎

Corresponding author. E-mail address: [email protected] (C. Soyk). 1 For the purpose of this paper, short- and medium-haul operations include flights with stage lengths up to 4000 km as defined by Eurocontrol (2005). http://dx.doi.org/10.1016/j.tra.2017.09.023 Received 28 April 2017; Received in revised form 9 August 2017; Accepted 25 September 2017 0965-8564/ © 2017 Elsevier Ltd. All rights reserved.

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Fig. 1. Components and underlying elements of the industry-unspecific business model framework, adapted from Wirtz et al. (2016).

Besides the academic novelty, there is also practical relevance for this study. Airline managers are monitoring the contemporary long-haul LCC development closely, as the entries of LCCs into the short-/medium-haul market in the 1990s and 2000s have lowered profitability of many legacy hub carriers or forced them to restructure (Franke, 2004). Policy makers need to be aware of shifts in long-haul growth away from primary to secondary airports, for example. This knowledge is key to re-evaluate infrastructure policies and allocate resources early on. This paper is structured as follows: Section 2 presents a review of previous studies on airline business model frameworks and on long-haul LCCs. Section 3 reasons the focus on the North Atlantic market, Section 4 introduces the methods used to determine the business model clusters and to evaluate cost differences. Section 5 presents and discusses the results of the two analyses. Section 6 concludes this paper and suggests further research. 2. Literature review This section begins with a comparison of existing airline business model frameworks, followed by a review of previous studies that examined the phenomenon of long-haul LCCs including their cost characteristics. 2.1. Previous studies on airline business model frameworks Research on business models has become an important aspect for both academia and management with the purpose to accurately describe a company’s value generation system with a manageable number of components (Wirtz et al., 2016). In this study, we aim to structurally analyze and cluster different airlines using this business model approach. As we intend to leverage existing airline business model frameworks, we reviewed existing industry-specific frameworks in this section. Furthermore, we related these frameworks to the latest research on industry-unspecific business model research to ensure that the frameworks cover the value generation system exhaustively. Wirtz et al. (2016) summarize the current state of general, industry-unspecific, research on business models and provide a framework summarizing the findings of previous studies. Fig. 1 depicts this framework with strategic components Strategy, Resources, and Cooperation, market components Customer, Market offer, Revenue, and operations components Manufacturing, Procurement, and Financial. Underlying elements of each component are summarized beneath the component name within each of the boxes.2 From an industry-specific perspective, two different airline business model frameworks have been developed and applied or enhanced several times, as indicated in Table 1. Mason and Morrison (2008) developed the Product and organizational architecture framework, which was enhanced by Lohmann and Koo (2013) and Jean and Lohmann (2016). This framework differentiates between the product and the organizational architecture of an airline. The product aspect contains service quality elements that relate the product to consumer preferences, namely connectivity, convenience, and comfort. The organizational architecture describes the vertical structure, production and distribution/sales elements (Mason and Morrison, 2008). The second key airline-related framework stems from Daft and Albers (2013) and was applied and enhanced by Daft and Albers (2015). It differentiates between corporate core logic, configuration of value chain activities, and assets. To evaluate which of the existing frameworks is closest to the industry-unspecific framework, we allocated elements from Mason 2 To harmonize nomenclature from different authors, this work describes over-arching dimensions of the framework as components and underlying, more descriptive dimensions as elements. Individual variables within these elements are referred to as items.

216

Meta-study categorizing existing airline business model frameworks

Examination of hybrid business models established between LCCs and hub carriers

Analysis of external effects on airline business model changes over time

Corbo (in press)

Jean and Lohmann (2016)

Examination of airline business model spectrum

Lohmann and Koo (2013)

Pereira and Caetano (2015)

Development of framework for measuring convergence of airline business models

Daft and Albers (2013)

Empirical analysis of airline business model convergence

Development of business model framework for comparison of low cost airlines

Mason and Morrison (2008)

Daft and Albers (2015)

Study type

Author/Year

Table 1 Previous studies on airline business models.

217

22

10

6: Network, revenue streams, distribution channels, alliances and partnerships, fleet structure, value proposition 6: Revenue, connectivity, convenience, comfort, aircraft, and labor

N/A

36

20

40

37

Items

N/A

8: Basic offering, Internal policy choices, external value network, inbound activities, production, marketing activities, tangible assets, intangible assets

6: Revenue, connectivity, convenience, comfort, aircraft, and labor

7: Internal policy choices, external value network, inbound, production, marketing, tangible assets, intangible assets

12: Profitability, Cost driver, Revenue, Connectivity, Convenience, Comfort, Distribution/sales, Aircraft, Labour, Airports attractiveness; Market structure

Elements

8 US carriers

6 US carriers, 6 European carriers (separately)

N/A

26 European carriers

9 US carriers

5 German carriers

6 European carriers

Application

C. Soyk et al.

Transportation Research Part A 106 (2017) 215–234

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

and Morrison (2008) and Daft and Albers (2015) to the components of the integrated business model framework of Wirtz et al. (2016). The framework of Mason and Morrison (2008), although still very broad, does not fully cover cooperation, manufacturing, and procurement components.3 On the other hand, the framework of Daft and Albers (2015) covers all components from the industryunspecific framework by Wirtz et al. (2016) and was therefore selected as the foundation for this study. 2.2. Previous studies on long-haul LCCs All previous studies that specifically addressed long-haul LCCs are listed in Table 2. To compare studies in detail, 11 characteristics that describe the long-haul LCC business model were extracted from the studies based on their appearance in at least two of the studies. Subsequently, each study was assessed along the 11 characteristics, which are discussed in more detail below. All previous studies mention or agree on one common characteristic: They describe a long-haul LCC model that operates a pointto-point network instead of a hub-and-spoke network. Point-to-point airlines operate a more decentral network and focus on simple operations, whereas hub carriers schedule their flights in and out of their hub to enable passengers to connect between flights. Often 50–60% of the passengers on long-haul flights of hub carriers are connecting passengers, not starting and terminating their travel at the ends of the long-haul flight (Morrell, 2008). Although only Morrell (2008), Wensveen and Leick (2009), and Douglas (2010) actually mention pricing schemes for connecting flights of long-haul LCCs, all three expect the same sum-of-sector pricing concept. This pricing scheme for connecting flights simply adds the prices of two or more flight legs to determine the overall fare for the whole Origin-Destination (OD) journey. Another pricing scheme often applied by hub carriers determines prices individually for each OD, more independent of the legs traveled. In that case, tickets are often priced at marginal costs and well below the sum of the prices of the two individual flight legs (Wittmer et al., 2011, p. 88). This pricing scheme becomes particularly challenging for the overall profitability when a carrier has a large share of connecting passengers priced at this marginal cost level (Tretheway, 2004). Previous studies also agree on target passenger groups, although only three studies discuss this characteristic in particular. The three studies highlight that long-haul LCC models would not only target leisure and Visiting Friends and Family (VFR) groups but also price-sensitive business travelers. The remaining eight characteristics are not described consistently in previous academic studies. Starting with the facilitation of transfers, earlier studies, e.g. Francis et al. (2007), point out the importance of hub feed for long-haul flights but further describe that a long-haul LCC does not rely on any feed, except for self-connecting passengers. On the other hand, De Poret et al. (2015), for example, conclude that a long-haul low cost operating model can be successful on routes that are in high demand or when sufficient feeder traffic can be found at either end of the route. They further state that transfer passengers from short-haul LCCs might be a viable source of demand. Regarding a differentiation of long-haul LCCs from charter airlines, Pels (2008), for example, concludes that charter carriers are different from long-haul LCCs as the former relies on tour operators to fill their aircraft whereas the latter carriers have to sell their tickets directly to passengers. On the other hand, Maertens (2015) identifies major similarities between the business model of Eurowings long-haul and leisure operators. Whereas similarities to leisure carriers can certainly still be a valid characteristic, Maertens (2015) limited his study to Eurowings long-haul, potentially not being representative for all emerging North Atlantic longhaul carriers. Another characteristic of long-haul low cost flying mentioned in previous studies is the unbundling of services. Unbundling refers to a trend introduced by short-/medium-haul LCCs where the ticket is sold separately from any add-ons such as luggage, a seat reservation, or food and drink options. Francis et al. (2007) and Wensveen and Leick (2009) state that cutting these add-on services on long-haul flights is difficult and likely less accepted by passengers. In contrast, more recent studies from De Poret et al. (2015) and Maertens (2015) see the absence of frills in the lowest ticket price as a key characteristic for the long-haul low cost model. In terms of travel classes offered, Moreira et al. (2011) and Daft and Albers (2012) describe long-haul low cost as a single-class model with absence of any premium classes. More recent studies from De Poret et al. (2015) and Maertens (2015) describe the model with a premium class. The usage of cargo space is also discussed with contrary outcomes. Morrell (2008) expects that due to high density single-class economy seating and the associated luggage volume and weight, space or weight for additional cargo is very limited. Daft and Albers (2012) state that additional cargo load can be the decisive differentiation for profitable operations when load factors are low. Particularly earlier studies from Francis et al. (2007), Pels (2008), Morrell (2008), and Douglas (2010) find that very large aircraft such as the Airbus A380 are required to achieve significant unit cost advantages over other carriers and are thus aircraft of choice for long-haul LCCs. More recent studies from Daft and Albers (2012), De Poret et al. (2015), Maertens (2015), and Wilken et al. (2016) identify smaller, efficient aircraft as a key characteristic for this business model, as markets large enough to fill single-class Airbus A380 are rare. Wilken et al. (2016) even expects that smaller 200-seat single-aisle long-haul aircraft such as the Airbus 321 Longrange could be an option for long-haul point-to-point flying. Most studies agree that long-haul point-to-point flying results in only few weekly frequencies being offered per destination, much less than the daily operations of hub carriers on most long-haul routes. The driver for this is the lower point-to-point demand due to the absence of feeder traffic and the usage of large aircraft. Whyte and Lohmann (2015), on the other hand, see potential for more frequencies on a non-stop Australia-Europe route. It is important to note that at the time of the study there was no non-stop route 3

Table A.9 in Appendix A provides the detailed allocation of the relevant airline-specific elements to those of the industry-unspecific components and elements.

218

Applicability of LCC model to long-haul; cost assessment Analysis of competition between airline network types Applicability of LCC cost advantages to long-haul

Definition of 3 business model types and subsumption of long-haul LCC

Evaluation of long-haul LCC models using strategic concepts Cost simulation of long-haul flying

Francis et al. (2007)

Wensveen and Leick (2009)

Douglas (2010)

219

Route profitability simulation of long-haul low cost flying

Route profitability simulation of low-cost longhaul flying

Daft and Albers (2012)

De Poret et al. (2015)

Moreira et al. (2011)

Morrell (2008)

Pels (2008)

Study type

Author/Year

EU-NA

Y

Y

N/A

Y

N/A

Y

Y

EU-NA

N/A

Y

N/A

Y

Y

EU-NA

AS-EU, EU-NA

Pointtopoint

Region

Y

N

N

Y

N

Y

N

N

Enabling transfers

Table 2 Previous studies that analyzed long-haul low cost airlines.

Y

Y

b

Y

b

Y

Y

N

b

Y

Y

Y

b

b

b

b

b

N

Y

N

N

Y

Y

N

b

N

b

Y

Premium cabins

b

b

Leisure and VFR only

b

Y

Y

Y

b

b

Y

b

N

b

Sum-ofsector pricing

Unbundling services

Different from charters

Long-haul low cost characteristicsa

Y

Y

Y

b

Y

N

b

Y

Cargo loading

N

N

b

b

b

b

Y

Y

b

Y

b

b

Y

Few frequencies

Y

Y

Y

Large aircraft

Y

Y

Y

Y

Y

Y

Y

Y

Secondary airports

b

b

10%

26%

20–25%

< 40%

b

20%

Cost advantage

“[…] The viability of long-haul LCC operations must be highly questionable.” “[…] Regular low cost, long-haul operations are possible if […] fullservice carrier product is effectively unbundled and suitable trunk routes […] identified.” “[…] Findings demonstrate how challenging the successful running of a European long-haul low-cost carrier can be.” (continued on next page)

b

“[…] We may see lowcost airlines on thick markets, such as London-New York[…]” “[…] doubt on the widespread establishment of the business model for long-haul flights.” “If long-haul, low-cost carriers are to survive […] in the evolving air transport environment, they must continue to innovate.”

“[…] May not ultimately prove viable […].”

Viability of business model

C. Soyk et al.

Transportation Research Part A 106 (2017) 215–234

Y

Y

AU-EU

AS-EU, EU-NA

LCC cost simulation of a specific long-haul route

Demand analysis for point-to-point flying

Whyte and Lohmann (2015)

Wilken et al. (2016)

Y

N

Y

Enabling transfers

N

b

b

Y

Unbundling services

Y

N

Different from charters

Y: Yes, i.e., study agrees with characteristic. N: No, i.e., study disagrees with characteristic. a Extracted from listed studies when mentioned in 2 different publications. b Characteristic not addressed in this study.

Y

AS-EU, EU-NA

Assessment of 1 new long-haul LCC

Maertens (2015)

Pointtopoint

Region

Study type

Author/Year

Table 2 (continued)

N

b

b

b

b

b

Leisure and VFR only

Sum-ofsector pricing

b

Y

Y

Premium cabins

Long-haul low cost characteristicsa

b

Y

b

Cargo loading

N

N

N

Large aircraft

Y

N

Y

Few frequencies

Y

N

Y

Secondary airports

b

13–17%

b

Cost advantage

b

“[…] It remains to be seen how Eurowings will be able to compete with full service legacy carriers that can […] offer similar or even lower fares and daily frequencies […].” “All of these factors cast some doubt that the long-haul low-cost concept would not threaten FSAs in the same way it has in short-haul markets.”

Viability of business model

C. Soyk et al.

Transportation Research Part A 106 (2017) 215–234

220

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

connecting Australia and Europe. Most studies describe usage of secondary airports as a characteristic of long-haul low cost flying, albeit to different extends. For example, Francis et al. (2007) believe the secondary airport strategy to be less effective due to the absence of facilities required for long-haul flying. Whyte and Lohmann (2015) state that long-haul LCCs will require major airports for passengers to be able to selfconnect from other flights. Summarizing this review, we formulate the following key insight: Insight 1: There is no common definition of the long-haul LCC business model in academic literature and elements of this model are often described with opposite characteristics. In this study, we aim to clearly define the long-haul LCC business model. We can also observe a simultaneous emergence of several North Atlantic carriers, which refer to themselves as long-haul LCCs. Based on these observations we hypothesize the following: Hypothesis 1. North Atlantic long-haul LCCs have a similar business model among each other and are significantly different from other long-haul carriers, thus, forming a new business model cluster in themselves. It is clearly difficult to review business model viability if there is no common definition of the model itself. However, we attempted to summarize previous evaluations in Table 2 by quoting the critical sentences of the respective articles. Whereas, for example, Francis et al. (2007) and Moreira et al. (2011) articulate that the success of long-haul LCC operations is questionable, others, e.g. Pels (2008) and Daft and Albers (2012), state that the model has potential if sufficient demand can be generated. As for the short-/medium-haul LCCs, the success of long-haul LCCs is likely correlated with their ability to lower their operating costs significantly compared to their competitors. Previous studies estimate potential savings between 10% (Moreira et al., 2011) and 40% (Morrell, 2008). Summarizing the previous studies, we can derive the following insight: Insight 2: There is no consistent view in academic literature on long-haul LCC business model viability, driven by different estimates around potential cost advantages. We thus aim to further analyze the actual cost differences between the emerging long-haul low cost and other carriers, leveraging the clear business model definitions to be established in the first part of this study. Again, given the recent emergence and growth of these long-haul LCCs we hypothesize that there must be a significant cost advantage for the carriers to operate profitable: Hypothesis 2. Long-haul LCCs have a sustainable cost advantage over other carrier types. 3. The North Atlantic airline market We are focusing on airlines from one region to ensure geographical comparability. The three reasons for the selection of the North Atlantic market are detailed in this section. In the following, four key groups of North Atlantic carriers are distinguished: Airlines that are part of a joint venture (JV), airlines that are part of an alliance but not part of a JV, leisure carriers and fourth, those newly emerging LCCs of interest in this study. Joint ventures demonstrate the strongest degree of collaboration between airlines and go beyond the ties of airline alliances. For example, since they market certain flights collectively, participation in a JV has significant commercial impact. Thus, these carriers are highlighted here as a separate group. One of three joint ventures was formed between Star Alliance members Air Canada, United Airlines and the Lufthansa Group carriers Austrian Airlines, Brussels Airlines, Lufthansa, and Swiss. Another joint venture was created between the Sky Team members Air France, Alitalia, Delta Airlines, and KLM. The third joint venture was formed between One World members American Airlines, British Airways (including Paris-based carrier Open Skies), Finnair, and Iberia. North Atlantic long-haul routes are the remaining key profit pool for these JV carriers. This profit pool has been maintained mainly through the oligopoly-like structure of the three large joint ventures (Francis et al., 2007). This remaining profit pool is a first key reason for the selection of the North Atlantic market in this study. Air Berlin and TAP are examples of carriers that are part of one of the three alliances but not part of one of the North Atlantic JVs. These carriers benefit from ties, albeit less intensive ties compared to JVs. For example, these airlines benefit through alliance-wide code-sharing agreements. Leisure carriers such as TuiFly or Thomas Cook serve some North Atlantic routes with low frequencies and are focused on leisure passengers, often as the subsidiaries of holiday package companies. Since 2008 and 2009 respectively, the Open Skies agreements between the European Union (EU) and the United States (US) and Canada (CA) allow any EU, US or CA airline to operate commercial flights connecting any point between the relevant territories. These Open Skies agreements significantly lowered barriers to entry (De Poret et al., 2015). Likely motivated by this development, several new North Atlantic long-haul carriers have emerged, for example Eurowings long-haul, Norwegian long-haul, Westjet longhaul, and WOW Air. International Airlines Group (IAG) and Air France/KLM Group have announced the start of their own long-haul low cost subsidiaries. Interestingly, this long-haul low cost market is approached from different directions: Eurowings is a subsidiary of Lufthansa, an existing legacy carrier, while Norwegian is a traditional short-haul LCC entering the long-haul market. The recent liberalization of this market is the second key reason for the selection of this market for our studies. Fig. 2 depicts the North Atlantic seat capacity (directional EU-US/CA) by carrier type over the past four combined flight plan periods. It can be observed that the supply-side market share of the dominating joint venture carriers has declined from 77% in the combined summer and winter flight plan periods 2013/14 to 70% in 2016/17. Therefore, recently, non-JV carriers have grown at a significantly higher pace than the JV airlines. The strongest share of the growth has come from the long-haul LCCs. These carriers 221

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Fig. 2. North Atlantic seating capacity (westbound) along flight plan periods split by carrier type.

only contributed 0.4 M seats in 2013/14 but already 2.3 M in 2016/17, having gained market share significantly, from 1% to 5%. This recent shift in capacity share is the third reason for the selection of the North Atlantic market in this study. 4. Methodology This section comprises three parts. First, the method applied to analyze and delineate long-haul airline business models is outlined. This is followed by the introduction of the method applied to quantify cost advantages between airline business models and to evaluate the sustainability of any advantages. Lastly, the filters used to determine the airline sample are described and the finally selected sample is presented. 4.1. Method to analyze and delineate long-haul airline business models To validate Hypothesis 1, i.e., newly emerging long-haul LCCs exhibit a similar business model among each other and are significantly different from other long-haul models, we clustered long-haul carriers along a business model framework and subsequently compared these clusters. Table 3 summarizes the full business model framework used in this study, including the 17 items, respective scales, further explanation and data sources. Components were named according to the industry-unspecific framework by Wirtz et al. (2016), elements, items and scales were adapted from Daft and Albers (2015). Items that are specifically measuring long-haul relevant aspects of an airline, were marked accordingly with the supplement long-haul. As previously reasoned, the business model framework by Daft and Albers (2015) builds the foundation for the framework used in this analysis. The framework was adjusted and enhanced, partly to account for the application to long-haul carriers and partly due to data availability. While we removed, adjusted, and added items and scales, we ensured that all elements continue to be represented by their respective subordinated items. Most importantly, items basic operations design and basic route design were replaced by network connectivity and network concentration. Network spatial scope (i.e., concentration) and connectivity are key defining characteristics for long-haul airlines. The newly defined items measure strategic network characteristics using continuous scales, while items from Daft and Albers (2015) differentiate these factors with ordinal scales. Additionally, the item national scope, i.e., the share of domestic flights, are nondescript in this trans-continental analysis. Furthermore, the financial component was not covered in this study, as relevant data was not available to us across the full airline sample. However, characteristics of this financial component are also reflected through other components, e.g. the fleet structure or supply management. There are three types of scales in this framework: Continuous scales measure the concrete metric impact of an item. This type of scale is indicated in Table 3 with a # or % sign. For example, fleet inhomogeneity is measured using the number of different aircraft types. The second type are dichotomous ordinal scales, which are indicated with box brackets []. For example, the availability of a frequent flyer program is measured with either 0 or 1, indicating whether or not a program is available. The third type are non222

223

Finance management, infrastructure

Financial

c

b

8

Airline annual reports, latest available in 08’2016

Average age (# of years) of long-haul fleet

Excluded from analysis due to limited data availability across sample.

Fleet age (long-haul)

% of flights served from hub airports

Airport sizes (long-haul)

http://airfleets.net

Diio LLC (2016)

Airline websites

13

Airline websites

17

16

15

14

12

Airline websites

9 10 11

7

Airline websites

Airline websites Airlineequality.com Airline websites

6

Diio LLC (2016); Airline websites

(0: No bundled connecting flights offered, 1: Sum-of-sector pricing, 2: OD pricing)

Focus of the airline on transfer passengers within the network: Discounted OD fares increase transfer passenger volume. Determined by comparing transfer connection fares with the sum of the individual fares of the flight legs. Tested on 3 routes per carrier. A hub airport was defined with > 100,000 long-haul seats/ month (outgoing). Airport not seen as hub if larger airport in the IATA metro region. Age of aircraft since first date of delivery from manufacturer.

4

Diio LLC (2016), http://airbus. com, http://boeing.com Airline websites

5

3

Diio LLC (2016)

2

1

Diio LLC (2016)

Diio LLC (2016)

#

Data Sourcec

Diio LLC (2016)

Services included in lowest base fare. Scale developed by incurred cost for carrier: Bag included rated higher, as bag handling incurs higher costs for carrier than seat reservation.

Seats counted as premium if different seat product other than economy seat installed. FFPs target regular fliers, mainly business travelers. The availability of FFPs is seen as an indication that the airline targets this traveler group. Holiday package companies often have their own airline subsidiary focused on bringing package holiday passengers to their destinations. Individually offered flights are less important. Any use of a global GDS (Global distribution system). Skytrax overall service rating using a scale from 1 to 5. No discounted fares: One-way equals one-way flex fare. Determined by comparing one-way and return prices on 3 routes per carrier. Services included in lowest base fare.

Count of same airline departures between 30 and 180 min after long-haul arrival. Averaged across all transatlantic arrivals. Origin airports are all long-haul airports on continent where airline is registered. Different sizes of one type counted individually, e.g. B7878/B787-9 counted as two types. Cabin area determined by aircraft type. Weighted by # of transatlantic flights per aircraft type. Cooperation analysis based on the transatlantic market.

Explanation

# of weekly flights/ # of city-pairs

[0: No, 1: Yes] Skytrax rating (# of stars) (0: No discounted fares, 1: 100% of discounted return fare, 2: 50% of discounted return fare) [0: Nothing included, 1: Food and drinks included] (0: Nothing included, 1: Seat reservation included, 2: Bag included, 3: Seat reservation and bag included)

[0: Airline is subsidiary of holiday package company, 1: Airline is no subsidiary of holiday package company]

[0: No Frequent Flyer Program (FFP) available, 1: FFP available]

# of different aircraft types serving transatlantic market Aircraft cabin area (m2) averaged by # of flights (0: Interlining only, 1: Code-sharing, 2: Part of global alliance, 3: Part of joint venture) # of premium seats/total # of seats

# of long-haul flights/ # of origin airports

# of potential intra-airline connections per long-haul arrival

Scaleb

Flight frequencies (longhaul) Transfer pricing complexity

Bundling concept on-board (long-haul) Bundling concept other amenities (long-haul)

Use of GDS Service quality One-way fare availability (long-haul)

Focus on leisure passengers

Premium passenger focus (long-haul) Focus on recurring passengers

Cooperation intensity

According to industry-unspecific framework as depicted in Fig. 1. Adapted for long-haul from Daft and Albers (2015) or newly defined. All data bases and websites accessed in 09’2016.

Supply management

Procurement

a

Route network

Manufacturing

Target product-market combination

Customer

Distribution Service product Fare structure

Inter-organizational relationships

Cooperation

Market offer Revenue

Fleet structure

Network concentration (long-haul) Fleet inhomogeneity (longhaul) Aircraft size (long-haul)

Geographic focus

Resources

Network connectivity (long-haul)

Type of air product

Strategy

Itemb

Elementb

Componenta

Table 3 Long-haul airline business model framework.

C. Soyk et al.

Transportation Research Part A 106 (2017) 215–234

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

dichotomous ordinal scales, where there are more than two options to choose from. Values are ordered in an ascending way and only one value can be chosen. These scales are marked using round brackets (). For example, cooperation intensity is measured in this way, where 0 represents the lowest value, interlining only, and 3 the highest value, the airline is part of a joint venture. In this case, being part of a joint venture is rated higher than being part of an alliance or code-sharing, which indeed is a more intensive way of cooperating between airlines. Another important example is the measurement of long-haul one-way fare availability. A rating of 0 highlights that no one-way fares are available, and 2, the highest rating, actually represents the highest value for this item, i.e., the one-way fare is priced at about 50% of the return fare. Table 3 also includes the data sources used to calculate the values for each item. Flight data was sourced from the Diio LLC (2016) scheduling data base. August is the month (jointly with July) with highest North Atlantic traffic volume and was thus selected for this study as we aimed to include as many carriers as possible to arrive at a most comprehensive overview. To rule out discrepancies resulting from seasonality, the individual items were also estimated with November data. The relative changes between carriers were insignificant, which can be reasoned with the long-term focus of most items. Changes to (network) strategy, geographic focus, fleet structure, and route network occur gradually over time and do not change fundamentally with seasons. Changes to flight frequencies were observed, however, insignificant with regards to the relation between the carriers. To cluster the airlines along the defined business model framework, we performed principal component analysis (PCA) and hierarchical cluster analysis (HCA). In transportation research, PCA is used to interpret data sets with a large number of interrelated variables (cf. Hansen et al., 2000; Brons et al., 2009). PCA reduces the dimensionality of the original data to a set of factors, which together account for as much variation of the original data set as possible. The factors are created by finding eigenvectors of the correlation matrix. Variables from the original data set with the highest correlation to the principal component factors are defined as loading variables. Data points with a strong correlation to one of the principal components generally have a strong correlation with the loading variables as well. (Jolliffe, 2002). HCA is a commonly applied cluster method in transportation research (cf. Sarkis and Talluri, 2004; Cabral and Ramos, 2014). This method initially places each data observation into its own cluster. Subsequently, two clusters with highest proximity are merged during each step. As a measure of proximity we used the average linkage algorithm.4 Both PCA and cluster analysis are executed on a PC version of SPSS. We combined PCA and HCA due to several reasons: PCA simplified the ample number of variables within the business model framework by identifying the most relevant items that load the principal components. Through this, individual items driving the differentiation of clusters can be interpreted, which is as an essential part of this study. The visualization of clusters along the principal components provided guidance for the potential number of clusters. On the other hand, through the reduction of variables onto few principal components, part of the variation of the variables was removed. Thus, for the final cluster definition we applied HCA to the original data set. At the same time, the application of two different methodologies to the original data adds credibility to the results. HCA was also applied to the original data set, not to the reduced principal components. 4.2. Method to quantify and evaluate sustainability of cost advantages To validate Hypothesis 2, i.e., emerging North Atlantic long-haul LCCs have a sustainable cost advantage over other North Atlantic carriers, we followed a two-step approach. We quantified potential operating unit cost differences between the defined airline clusters followed by an evaluation of sustainability of these differences. To compare total operating costs across carriers, they were normalized with system-wide available seat kilometers (ASKs) to arrive at operating cost per ASK or CASK, which is a commonly used value in academia and the airline industry (cf. Doganis, 2002; Halloway, 2008). Although operating costs increase with stage-length, CASK decrease, as aircraft and crew are more productive since they spend more time in the air and less on the ground. Furthermore, stage-length independent airport and handling as well as overhead costs are distributed across more kilometers. Different airlines can have fundamentally different system-wide mean stagelengths depending on their network. Carriers with a higher share of short-haul routes have a lower system-wide mean stage-length and vice versa. Thus, CASK needed to be adjusted for the comparison between carriers. In their airline data project, Swelbar and Belobaba (2016) applied a cost-adjusting formula that is commonly used in the airline industry and was thus selected for this analysis. The formula adjusts costs based on the stage-length difference of a carrier to the sample mean according to Eq. (1), where li is the system-wide mean stage-length of carrier i.

CASK adj,i = CASKunadj,i

li 1 n

n

∑i = 1 li

(1)

After this initial CASK comparison, sustainability of individual cost advantages was evaluated. First, unit cost differences caused by varying number of seats per aircraft were quantified using the average number of seats per cabin space. The cabin space Cj per aircraft type j was calculated using the length and width of the cabin, wj and l j , according to Eq. (2). Aircraft dimensions were obtained from the aircraft manufacturer’s websites.

Cj = wj l j

(2)

The yearly cabin space flown CFi per carrier i was calculated using the total number of flights and the average distance flown per 4

The average linkage algorithm defines the proximity between two clusters as the average distance between all possible pairs of items within any two clusters.

224

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

aircraft type j and carrier i,fi,j and di,j , according to Eq. (3). The total seats flown SFi per carrier i were calculated by further using the seats flown per aircraft type j and carrier i,si,j , according to Eq. (4). Flight, seat, and distance data were obtained from Diio LLC (2016).

CFi =

∑

fi,j di,j Cj

(3)

j

SFi =

∑

fi,j si,j

(4)

j

The average number of seats per cabin space, SCi per carrier i was calculated according to Eq. (5). Subsequently, the mean average of seats per cabin space, SCk , was determined for each identified cluster k containing n airlines, according to Eq. (6).

SCi =

SFi CFi

SCk =

(5)

n ∑i

SCi (6)

n

Finally, the seat adjustment factor for a cluster k1, SAk1, was calculated using the SCk of two different clusters and adjusted with the share of long-haul ASKs, ASKlong − haul,k1, according to Eq. (7).

SAk1 =

SCk1 ASKlong − haul,k1 SCk 2

(7)

In the next step, operating costs by item were collected and aggregated into six buckets: Fuel, staff, maintenance, repair and overhaul (MRO), airport, ATC, and handling, leasing and depreciation, and finally, sales, distribution and other costs. All costs were normalized with system-wide ASKs. Subsequently, the mean average per cost item was calculated for each cluster. Cost data was obtained for 2015 as this was the year with the latest full-year data available at the point of analysis. Where publicly available, the data was extracted from annual financial reports of the airlines. For some carriers that were not publishing annual financial reports, operating cost data was obtained from the 60th IATA World Air Transport Statistics (IATA, 2016). Since cost data was only available on a system-wide level, the analyses were based on system-wide cost and ASK data. 4.3. Airline selection criteria and resulting sample Airlines analyzed in this paper were selected along four criteria, as summarized in Table 4. First, all passenger airlines that were marketing flights between US, Canada and the key European markets were selected. Key European markets were defined as countries where an Open Skies agreement was in place, i.e. EU28 member states plus Iceland (IS), Norway (NO), Switzerland (CH). Hereby, flights operated by a subdivision of an airline or by a third party using identical marketing carrier codes were analyzed as separate carriers only when flights were marketed and distributed as different brands. In this analysis Air Canada Rouge and British Airways Open Skies were handled as separate carriers while Lufthansa Cityline and Swiss Edelweiss Air were not. Second, of this sample, only airlines that operated at least 10 North Atlantic flights per month (total for all destinations) were selected. Airlines that operated fewer flights did not provide sufficient data points for the business model framework and were irrelevant compared to the major North Atlantic carriers. Third, airlines that were not headquartered in one of the countries named in criteria (1) were removed from the sample. In some cases airlines operate flights between two countries of which none is the airline’s home country. These flights are operated under the 6th or 7th freedom of the air right, which are rarely granted by countries. Therefore, most other flights of these particular carriers are operated in other regions of the world. Data from these carriers could distort the results and was thus removed from the sample. And fourth, single particular airlines that were not comparable to any other carrier in the sample were excluded. In this analysis, La Compagnie, a French-based carrier was removed, as it operated business class-only flights. With a premium passenger focus of 100%, this airline differs fundamentally from all other carriers and would distort the results. Table 5 lists the resulting sample of 37 airlines, including country of origin and alliance membership. Availability of total operating cost data and split of costs by item are indicated by carrier. 5. Results and discussion This section comprises two parts. First, the results from the long-haul airline business model analysis are outlined and the characteristics of the emerging long-haul carriers are discussed. Second, unit cost differences between clusters are presented prior to a discussion of their sustainability. 5.1. Results from long-haul airline business model analysis 16 out of 17 items from the framework outlined in Section 4.1 were included in both PCA and HCA. Item # 9 was excluded as it did not contain any variance: All airlines in the sample were using global distribution systems (GDS). Two principal components were selected based on Cattell’s scree plot criterion (Jolliffe, 2002). The scree plot can be observed in 225

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Table 4 Filter applied to determine relevant airline sample. Criteria

Reasoning

Comment

(1) All passenger airlines flying across the North Atlantic between US/CAa and Europe (EU28,b CH,a IS,a NOa)

All sizable North Atlantic markets need to be included.

Airline subdivisions with identical marketing carrier but different brands were analyzed separately, e.g. British Airways Open Skies was analyzed separately, while Swiss Edelweiss Air was not.

(2) All airlines operating > 10 frequencies per month (total all destinations)

Excluding irrelevant airlines to increase comparability.

Operating ⩽ 10 frequencies per month equivalent to max. 1 longhaul aircraft.

(3) All airlines headquartered in markets from (1)

Excluding airlines flying under 6th or 7th freedom of the air rights.

Airlines flying across the North Atlantic under 6th or 7th freedom are operating most flights in other regions. Including their data points could distort the result.

(4) Excluding particular carriers

Excluding carriers that cannot be compared.

La Compagnie, FRb operates few business-class only flights and was excluded.

a b

Referring to ISO 3166-1 alpha-2 letter country codes. EU28: European Union comprising 28 member states (as of 2015).

Fig. B.7. The two first eigenvalues explain 55.5% of the total variation. Oblimin factor rotation was applied subsequent to the initial identification of factors to achieve simpler and more interpretable factors. The factor score of the individual items can be obtained from Table 6. Fig. 3 depicts the loadings of principal components 1 (PC1) and 2 (PC2) with the analyzed 16 items. Since all items correlate with one of the PCs with a value higher than 0.5, we retained all 16 variables in the plot. PC1 is positively loaded with cooperation intensity, network connectivity, flight frequencies, network concentration, transfer

Table 5 Airline sample for different analyses Carrier

Carrier name

Country

Alliance

Total cost availablea

Cost split availableb

AA AB AC AC/RV AF AY AZ BA BA/EC BY DE DL DY EI EW FI IB IG KL LH LO LX MT OR OS S4 SE SK SN SS TP TS UA UX VS WS WW

American Airlines Air Berlin Air Canada Air Canada Rouge Air France Finnair Alitalia British Airways Open Skies Thomson Airways Condor Delta Air Lines Norwegian Air Shuttle Aer Lingus Eurowings Icelandair Iberia Meridiana fly KLM Royal Dutch Airlines Lufthansa LOT Polish Airlines SWISS Thomas Cook Airlines TUIfly Netherlands Austrian Airlines SATA International XL Airways France SAS Brussels Airlines CORSAIR TAP Portugal Air Transat United Airlines Air Europa Virgin Atlantic Airways Westjet WOW Air

United States Germany Canada Canada France Finland Italy United Kingdom France United Kingdom Germany United States Norway Ireland Germany Iceland Spain Italy Netherlands Germany Poland Switzerland United Kingdom Netherlands Austria Portugal France Sweden Belgium France Portugal Canada United States Spain United Kingdom Canada Iceland

OneWorld OneWorld StarAlliance StarAlliance SkyTeam OneWorld SkyTeam OneWorld OneWorld None None SkyTeam None None None None OneWorld None SkyTeam StarAlliance StarAlliance StarAlliance None None StarAlliance None None StarAlliance StarAlliance None StarAlliance None StarAlliance SkyTeam None None None

x x x

x x x

x x

x x

x

x

a b

x x x x x x x x x x x x x x

x x x x x x x x x

x

x

x x x x x x x x x x

x

x x x x x

When total operating cost data was available, the respective airline was included in the operating cost benchmark. When operating cost data was available by cost item, the respective airline was included in the sustainability analysis of cost advantages.

226

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Table 6 Rotated factor loadings of PCA.

1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17

Items

Factor 1

Factor 2

Network connectivity Network concentration Fleet inhomogeneity Aircraft size Cooperation intensity Premium passenger focus Focus on recurring passengers Focus on leisure passengers Service quality One-way fare availability Bundling concept on-board Bundling concept other amenities Flight frequencies Transfer pricing complexity Airport sizes Fleet age

0.847 0.687 0.622 0.524 0.840 0.547 0.576 −0.737 0.573 −0.823 0.112 0.386 0.816 0.685 0.585 −0.031

−0.017 0.063 0.100 −0.105 0.225 0.151 −0.312 0.340 0.149 −0.175 0.892 0.714 0.116 −0.097 0.182 0.716

Significant items with factor loading of >‖0.5‖ are marked in bold text. Oblimin factor rotation was applied. Item #9 was excluded due to lack of variance within the variable.

Fig. 3. Factor loadings of principal components.

pricing complexity, airport sizes, service quality, fleet inhomogeneity, premium passenger focus, and focus on recurring passengers. Negative correlation exists with focus on leisure passengers and one-way fare availability. Thus, PC1 is described in positive direction with central network, hubbing, and focus on premium and recurring passengers. Negative PC1 component characteristics are best described with decentral network, no hubbing, and focus on leisure passengers. PC2 is positively loaded with increasing fleet age and higher service levels in lowest long-haul fare classes, both on-board and pre-/post flight. We thus name the positive characteristics of PC2 services included and old fleet. The negative characteristics of PC2 were described with no services included and new fleet. Fig. 4 visualizes the scoring of the airlines within the sample along the principal components PC1 and PC2. Three groups of carriers can be identified based on the visual proximity of airlines in the data plot. A first group of airlines is located in the top right quadrant. These airlines generally have a more central network, focus on connectivity and premium passengers. A second group of airlines can be located in the top left quadrant. According to the principal components, these airlines operate a less central network with low connectivity and simple pricing but have similar service and fleet age levels as cluster 1 airlines. The third cluster of airlines is located in the lower bottom half of the diagram. In terms of network centrality, connectivity and premium focus, this group of carriers is located in between groups one and two. These airlines offer significantly lower service levels in the lowest fare and operate 227

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Fig. 4. Principal component factor scores including cluster allocation determined by hierarchical clustering method. Airline letter codes according to Table 5.

newer aircraft fleet. This group contains airlines DY, EW, WW, and WS, which recently commenced North Atlantic long-haul operations. Based on these visual observations we conclude the existence of three separate clusters. The subsequent HCA validates these visual groupings. The dendogram displaying the hierarchical connections between groups and the formation of clusters is shown in Fig. B.8. The allocation of airlines to the three clusters can also be observed in Fig. 4: The type of marker represents the cluster into which the airline is allocated based on HCA. One difference between the plot of the PC factor scores and the results of the HCA can be observed. According to the former, IG and S4 are counted as part of the second cluster, HCA allocated these two carriers into cluster one. This discrepancy can be explained as both IG and S4 have incorporated leisure-like characteristics but are also operating a rather central hub network. The results from the HCA are selected as the final allocation, as the PCA factor scores do not reflect the total variance of the data set. Based on these two findings, we name cluster one airlines Legacy hub, as it contains carriers such as Lufthansa, Air France, and American Airlines. Cluster two is named Leisure, as it contains leisure or charter carriers such as Thomas Cook and Air Transat. Cluster three is named No frills point-to-point, as it contains the new long-haul carrier types that differentiate through their focus on a point-topoint network and the exclusion of frills or services in the lowest ticket fare. From this point onwards, we refer to no frills point-topoint carriers instead of long-haul LCCs, as the term better reflects the business model. When revisiting Hypothesis 1 formulated in Section 2, we can confirm the thesis that newly emerging North Atlantic long-haul carriers have a similar business model among themselves, significantly different from other leisure or hub models. Characteristics of no frills point-to-point airlines Table 7 summarizes mean normalized values by airline cluster for the 16 (out of 17) analyzed variables. The statistical significance of differences between clusters was tested with the non-parametric Kruskal-Wallis H-test. Except for aircraft size and fleet inhomogeneity, the null hypothesis that there is no significant difference between at least two out of three clusters was rejected at the 5% level for all other items. For fleet inhomogeneity, the Null-Hypothesis was rejected at the 10% level. From a strategic point-of-view, no frills point-to-point carriers operate a network with some connectivity, situated in between the two other business model clusters. They offer connections to long-haul flights leveraging their short-haul network, enabling passengers to connect. This is an important differentiation to the existing leisure business model. Nonetheless, the connectivity of legacy hub carriers is significantly higher, as their networks are focused around connections at hubs. Network concentration of no frills point-to-point carriers is equal to that of leisure carriers, highlighting a similarly decentralized network with a focus on point-to-point routes. With regards to resources and cooperation, cluster three carriers resemble leisure carriers. They operate a more homogeneous long-haul fleet than legacy hub carriers, likely motivated to reduce costs as observed in the short-/medium-haul low cost model. Although we expect no frills point-to-point carriers to operate smaller aircraft compared to other carriers, this thesis is not confirmed as the differences between clusters are statistically insignificant. Furthermore, they cooperate significantly less than legacy hub carriers, likely driven by the lower need for connecting passengers as indicated by the lower network connectivity. From a customer and market perspective, no frills point-to-point airlines carry significantly fewer premium passengers compared to legacy carriers but carry similar volumes compared to leisure carriers. No frills point-to-point carriers focus on recurring passengers using frequent flyer programs, an important distinction to leisure carriers, which rather focus on touristic passengers. Both no 228

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Table 7 Mean values of normalized items for each airline cluster. Items

1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17

Network connectivitya Network concentrationa Fleet inhomogeneity2 Aircraft size3 Cooperation intensitya Premium passenger focusa Focus on recurring passengersa Focus on leisure passengersa Service qualitya One-way fare availabilitya Bundling concept on-boarda Bundling concept other amenitiesa Flight frequenciesa Transfer pricing complexitya Airport sizesa Fleet agea Number of airlines in cluster

Cluster 1 Legacy hub

Cluster 2 Leisure

Cluster 3 No frills p2p

H-test

Kruskal-Wallis p-value

0.47 0.28 0.32 0.53 0.83 0.40 0.96 0.00 0.46 0.19 1.00 0.76 0.59 0.90 0.84 0.47

0.00 0.07 0.11 0.47 0.14 0.18 0.29 0.86 0.00 1.00 1.00 0.48 0.17 0.36 0.59 0.40

0.17 0.07 0.09 0.36 0.17 0.15 1.00 0.00 0.00 1.00 0.00 0.00 0.22 0.75 0.48 0.07

16.3 11.3 6.0 1.4 22.1 10.8 18.9 29.9 7.3 20.8 36.0 17.5 16.7 13.4 7.7 9.0

0.000 0.004 0.051 0.498 0.000 0.005 0.000 0.000 0.026 0.000 0.000 0.000 0.000 0.001 0.021 0.011

26

7

4

All values normalized between 0 and 1 across all clusters: 0 represents lowest value within the sample, 1 represents the highest value Item #9 was excluded due to lack of variance within the variable. a Statistically significant at 5%-level. 2 Statistically significant at 10%-level. 3 Not statistically significant at 10%-level.

frills point-to-point and leisure airlines offer low price one-way fares, cutting complexity from sophisticated revenue management systems that differentiate between short-term business and longer-term touristic passengers. In contrast to other carriers, no frills point-to-point airlines strictly exclude all on-board services such as food and drinks as well as other services such as checked luggage in their lowest base fare on North Atlantic flights. From an operations point-of-view, no frills point-to-point carriers operate fewer flight frequencies per route compared to legacy hub carriers. This difference is a consequence of fewer transfer passengers, which are partly a result of the lower network connectivity. The leg-based pricing schemes also do not particularly target transfer passengers, as very low fares for connecting flights are eliminated. Compared to legacy hub carriers, both no-frills point-to-point and leisure carriers use smaller (non-hub) secondary airports. Furthermore, no-frill point-to-point airlines have a significantly lower average fleet age than both legacy hub and leisure carriers, which points towards a higher fuel efficiency. Concluding this section, we observe that emerging carriers include business model aspects of legacy hub, leisure, as well as shorthaul LCCs, forming it into a coherent business model. Based on our empirical findings, we can describe the emerging business model of no frills point-to-point carriers in the following way: Insight 3: No frills point-to-point carriers operate a decentral point-to-point model while leveraging low-complexity coincidental feeder traffic at existing short-haul bases. These carriers target all passengers groups and have a strong focus on low complexity and low cost. 5.2. Results from cost analysis In this section, the results of the cost analyses between carriers are presented. At first, the normalized total unit operating costs of the carriers are compared on a cluster level. In a second step, unit cost differences between clusters by cost item are presented and their sustainability is discussed. 5.2.1. Total unit operating costs by airline cluster Fig. 5 displays the stage-length adjusted system-wide CASK (left axis) and the stage-length (right axis) of the airlines within the sample. The mean system-wide stage-length across the sample was calculated with 2594 km in 2015. The adjustment of the sample of airlines in this analysis (n = 31) is rooted in the lack of publicly available cost data of some carriers. Table 5 details the data availability by carrier. Cluster 1, i.e., legacy hub carriers have a stage-length adjusted mean CASK of US$c 7.91 (Sample standard variation: σs,1 = 1.14 ), cluster 2 (leisure) carriers of US$c 8.13 (σs,2 = 1.05), and cluster 3 (no frills point-to-point) carriers of US$c 5.27 (σs,3 = 0.76). This equals a 33% lower mean cost base of no frills point-to-point carriers compared to legacy hub carriers, and a 35% lower average cost base versus leisure carriers, on a stage-length adjusted basis. 5.2.2. Sustainability of cost differences To understand the sustainability of the identified cost advantages, individual cost items are compared between clusters. Fig. 6 229

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Fig. 5. Stage-length adjusted system-wide CASK and mean stage-lengths for airlines within the 3 determined clusters.

presents the mean stage-length adjusted unit operating costs of cluster 1 and cluster 3 carriers by cost item. It also highlights the cost differences between carriers by item. The further adjustment of the sample in this analysis (n = 21) is once again rooted in the lack of publicly available data by cost item of some carriers. Table 5 details the data availability by carrier. 13 percentage points of the total 33% cost advantage of no frills point-to-point carriers are attributable to higher seat densities on long-haul aircraft. On average, cluster 3 carriers install 1.19 seats/m2 of long-haul aircraft cabin space, cluster 1 carriers install 0.95 seats/m2 across their fleet. This difference of 25% on long-haul aircraft across the sample translates to a 13% cost reduction on a system-wide level, estimated with a share of 50% long-haul ASKs. The remaining 20 percentage points of the cost advantage are split across the six cost items. Leasing and depreciation costs are actually higher for cluster 3 carriers. Table 8 summarizes the evaluation of sustainability of cost advantages over time. On an item level, cluster 3 carriers have 21% lower expenses per ASK for fuel compared to cluster 1 carriers. Lower expenses for fuel are directly proportional to a lower fuel consumption. Younger, i.e., more modern aircraft, generally consume less fuel. Results in Table 7 indicate that legacy hub airlines operate on average a 6.7-fold older fleet than no frills point-to-point carriers. A recent report has compared fuel efficiency of airline fleets operating over the North Atlantic in 2014 (Kwan and Rutherford, 2015, p. 17). The fleet of Norwegian, the only cluster 3 carrier analyzed in the report, has been calculated with 9–22% higher fuel efficiency than the cluster 1 fleets named in the report. However, while the report thus confirms our findings, this cost advantage will likely diminish over time. Since the long-haul operations of cluster 3 carriers have only recently emerged, their average fleet age will increase over time. To retain this cost advantage, cluster 3

Fig. 6. Differences in operating costs between Lufthansa and Norwegian by cost item.

230

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Table 8 Cost differences by item between cluster 1 and cluster 3 carriers. Cost item

Cost difference (US$c)

Mostly sustainable over time

Higher seat density Fuel Staff MRO Airport, ATC, handling Leasing, depreciation Sales, distribution, other

−0.90 −0.40 −0.51 −0.25 −0.25 +0.07 −0.26

Yes No Yes No Yes No Yes

Total

−2.50

Mostly sustainable Mostly unsustainable

−1.92 −0.58

airlines need to phase-out their aircraft significantly earlier than cluster 1 airlines. Due to the high investment costs of aircraft, this is unlikely to occur. Thus, this source of cost advantage cannot be seen as sustainable. Cluster 3 carriers operate with 27% lower staff cost per ASK than cluster 1 carriers. Lower cost crew contracts, less overseas rest time, as well as less senior crews have been previously identified as potential drivers for lower staff cost in low cost long-haul airlines (De Poret et al., 2015; Whyte and Lohmann, 2015). Whereas the latter advantage will seize as crews mature, a lower crew cost base and more efficient deployment are sustainable cost advantages - as long as other carriers do not apply similar measures. Maintenance, Repair, and Overhaul (MRO) costs are 37% lower for cluster 3 carriers. Younger fleets require fewer of the laborintensive overhaul procedures, which are a major expense. Currently, cluster 3 carriers have a 6.7-fold younger fleet than cluster 1 carriers. However, once the fleet age increases, these costs will also increase for cluster 3 carriers. Thus, the majority of this cost advantage will diminish over time. Cluster 3 carriers pay 16% less for airport, air traffic control (ATC), and handling expenses. While ATC expenses can hardly be influenced by the carrier, airport and handling expenses can be influenced through the choice of airport. Secondary airports generally charge lower landing and passenger-related fees than large hub airports (De Poret et al., 2015). According to our findings in Table 7, no frills point-to-point airlines are using significantly more often smaller and less expensive secondary airports. As a result, this cost advantage is evaluated as mostly sustainable. Leasing and depreciation costs are 10% higher for cluster 3 carriers. The younger average fleet ages once again explain this difference. As with fuel and MRO costs, over time, this cost difference is likely to decrease. Sales, distribution, and other costs of cluster 3 carriers are 22% lower. In extensive hub-and-spoke networks, large volumes of connecting passengers require processes and customer relation structures, for example, to manage irregularities such as missed connections. Complex OD pricing systems need to be in place to optimize revenue streams also from connecting passengers, not just on a single flight basis. Inhomogenous long-haul aircraft fleets result in more complex aircraft and crew rotations as well as crew training. All of these aspects need to be organized and managed in the back-office, often requiring complex systems and infrastructure. The same is true for different service classes: Managing different service classes, for example, four classes at Lufthansa, increases complexity, and thus cost, further. On the other hand, no-frills point-to-point carriers have lower network connectivity and focus on more efficient and less complex point-to-point (long-haul) networks. As a result, the majority of this cost advantage can be claimed to be sustainable, driven by the difference from the business model. When revisiting the second hypothesis formulated in Section 2, we are able to confirm the thesis that North Atlantic no frills pointto-point carriers have a sustainable unit cost advantage over other North Atlantic carriers. Insight 4: No-frills point-to-point airlines have a 2.50 US$c/ASK or 33% unit cost advantage over legacy hub carriers. 1.92 US$c/ASK or 24 percentage points are sustainable, of which 0.90US$c/ASK or 13 percentage points are a result of a higher seat density on long-haul aircraft. 6. Conclusion Previous studies on long-haul LCCs do not agree on either business model characteristics or level and sustainability of cost advantages. Thus, the aim of this paper was to provide clarity regarding the defining business model characteristics and cost advantages of emerging long-haul LCCs. The study was focused on the liberalized North Atlantic airline market. Our empirical results confirmed that the business model of the emerging long-haul LCCs is significantly different from existing legacy hub and leisure carriers and that these airlines form their own cluster. Furthermore, the results showed that this cluster of airlines operates with an average of 33% lower operating costs compared to legacy hub carriers and with 35% lower costs compared to leisure carriers. 24 percentage points of the 33% cost advantage over legacy hub carriers are sustainable, i.e., they can be expected to prevail over time, driven by differences in the business model (11 percentage points) and higher seating densities (13 percentage points). These findings shall build the foundation for further studies. Three interesting questions arise specifically. First, do these cost advantages overcompensate lower revenue levels, which are likely given the smaller share of premium passengers? A study comparing revenues and resulting profitabilities would provide additional insights into the long-term success of the business model. Second, what are the limitations to this business model, i.e., will it continue to take market share from legacy hub and leisure models 231

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

over time or is it limited to highly demanded markets? A thorough demand analysis building on the findings from Wilken et al. (2016) could be enlightening. Third, what is the impact of latest range-extended narrow-body aircraft (e.g., Boeing 737 Max or Airbus 321 Long-range) to the long-haul market? These smaller and lower cost alternatives to existing wide-body aircraft could open additional routes away from hubs, suitable to be served by the decentral long-haul LCC model. A study to estimate their market impact could provide valuable insights for both practitioners and policy makers. Acknowledgments We would like to thank Pierre Galvin from Amadeus IT Group and Marcia Urban from Bauhaus Luftfahrt e.V. for their continuous feedback. Furthermore, we would like to thank the many anonymous reviewers for their valuable comments. All remaining errors are ours. Appendix A Table A.9 Table A.9 Allocation of elements from airline-specific business model frameworks to industry-unspecific elements and components. Industry-unspecific

Airline-specific

Components from Wirtz et al. (2016)

Elements from Wirtz et al. (2016)

Elements from Daft and Albers (2015)

Elements from Mason and Morrison (2008)

Elements in this study

Strategy

Strategic positions and development paths, value propositions (Core) competencies, (core) assets Networks, partnerships

Type of air product, Geographical focus, input factor policy, business policy Fleet structure, human capital, property capital Inter-organizational relationships Target product-market combination, Distribution, Advertising Cabin product, ground product

Profitability index, Connectivity

Type of air product, geographic focus

Aircraft productivity, Labor productivity N/A

Fleet structure

Distribution/sales, Airport attractiveness

Target product-market combination, Distribution Service product

Fare structure

Market structure, convenience, comfort Revenue achievement

Route network

N/A

Route network

Supply management

N/A

Supply management

Finance management, infrastructure

Cost drivers

Excluded

Resources Cooperation Customer

Market offer Revenue Manufacturing

Procurement Financial

Customer relationships/ target groups, channel configuration Competitors, market structure, value offering Revenue streams, revenue differentiation Manufacturing/ operations, value generation Resource acquisition, information Financing, capital allocation, cost structure

Appendix B Figs. B.7 and B.8

Fig. B.7. Principal component analysis: Scree plot of Eigenvalues.

232

Inter-organizational relationships

Fare structure

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

Fig. B.8. Hierarchical cluster analysis: Results of cluster analysis shown in dendogram.

233

Transportation Research Part A 106 (2017) 215–234

C. Soyk et al.

References Brons, M., Givoni, M., Rietveld, P., 2009. Access to railway stations and its potential in increasing rail use. Transport. Res. Part A: Pol. Pract. 43, 136–149. Cabral, A.M.R., Ramos, F.d.S., 2014. Cluster analysis of the competitiveness of container ports in Brazil. Transport. Res. Part A: Pol. Pract. 69, 423–431. Cho, W., Windle, R.J., Dresner, M.E., 2015. The impact of low-cost carriers on airport choice in the US: a case study of the Washington-Baltimore region. Transport. Res. Part E: Logist. Transport. Rev. 81, 141–157. Corbo, L., 2016. In search of business model configurations that work: Lessons from the hybridization of Air Berlin and JetBlue. J. Air Transp. Manage., (in press). Daft, J., Albers, S., 2012. A profitability analysis of low-cost long-haul flight operations. J. Air Transp. Manage. 19, 49–54. Daft, J., Albers, S., 2013. A conceptual framework for measuring airline business model convergence. J. Air Transp. Manage. 28, 47–54. Daft, J., Albers, S., 2015. An empirical analysis of airline business model convergence. J. Air Transp. Manage. 46, 3–11. De Poret, M., O’Connell, J.F., Warnock-Smith, D., 2015. The economic viability of long-haul low cost operations: evidence from the transatlantic market. J. Air Transp. Manage. 42, 272–281. Diio LLC, 2016. Diio Mi - Market Intelligence for the Aviation Industry < http://www.diio.net > . Doganis, R., 2002. Flying Off Course, third ed. Routledge, London. Douglas, I., 2010. Long-haul market entry by value-based airlines: dual business models support product innovation. World Rev. Intermodal Transport. Res. 3, 202–214. Eurocontrol, 2005. Glossary for Flight Statistics & Forecasts. European Organization for the Safety of Air Navigation (EUROCONTROL) < https://www.eurocontrol. int/sites/default/files/article/attachments/eurocontrol-glossary-for-flight-statistics-and-forecasts.pdf > . Fageda, X., Suau-Sanchez, P., Mason, K.J., 2015. The evolving low-cost business model: network implications of fare bundling and connecting flights in Europe. J. Air Transp. Manage. 42, 289–296. Francis, G., Dennis, N., Ison, S., Humphreys, I., 2007. The transferability of the low-cost model to long-haul airline operations. Tour. Manage. 28, 391–398. Franke, M., 2004. Competition between network carriers and low-cost carriers - retreat battle or breakthrough to a new level of efficiency? J. Air Transp. Manage. 10, 15–21. Fu, X., Dresner, M., Oum, T.H., 2011. Effects of transport service differentiation in the US domestic airline market. Transport. Res. Part E: Logist. Transport. Rev. 47, 297–305. Fu, X., Lei, Z., Wang, K., Yan, J., 2015. Low cost carrier competition and route entry in an emerging but regulated aviation market - the case of China. Transport. Res. Part A: Pol. Pract. 79, 3–16. Halloway, S., 2008. Straight and Level: Practical Airline Economics, third ed. Ashgate, Hampshire. Hansen, M.M., Gillen, D., Djafarian-Tehrani, R., 2000. Aviation infrastructure performance and airline cost: a statistical cost estimation approach. Transport. Res. Part E: Logist. Transport. Rev. 37, 1–23. IATA, 2016. 60th World Air Transport Statistics, < http://www.iata.org/publications/store/Pages/world-air-transport-statistics.aspx > . Jean, D.A., Lohmann, G., 2016. Revisiting the airline business model spectrum: the influence of post global financial crisis and airline mergers in the US (2011–2013). Res. Transport. Bus. Manage. 21, 76–83. Jolliffe, I.T., 2002. Principal Component Analysis, second ed. Springer. Kwan, I., Rutherford, D., 2015. Transatlantic Airline Fuel Efficiency Ranking, 2014. Technical Report November. The International Council on Clean Transportation < http://www.theicct.org/sites/default/files/publications/ICCT_transatlantic-airline-ranking-2014.pdf > . Lohmann, G., Koo, T.T.R., 2013. The airline business model spectrum. J. Air Transp. Manage. 31, 7–9. Maertens, S., 2015. The New Eurowings Flights from Germany. German Aerospace Center - Institute for Air Transport and Airport Research. Mason, K.J., Morrison, W.G., 2008. Towards a means of consistently comparing airline business models with an application to the ‘low cost’ airline sector. Res. Transport. Econ. 24, 75–84. Moreira, M.E., O’Connell, J.F., Williams, G., 2011. The viability of long-haul, low cost business models. J. Air Transp. Stud. 2, 69–91. Morrell, P., 2008. Can long-haul low-cost airlines be successful? Res. Transport. Econ. 24, 61–67. Pels, E., 2008. Airline network competition: full-service airlines, low-cost airlines and long-haul markets. Res. Transport. Econ. 24, 68–74. Pereira, B.A., Caetano, M., 2015. A conceptual business model framework applied to air transport. J. Air Transp. Manage. 70–76. Sarkis, J., Talluri, S., 2004. Performance based clustering for benchmarking of US airports. Transport. Res. Part A: Pol. Pract. 38, 329–346. Swelbar, W.S., Belobaba, P.P., 2016. Airline Data Project < http://web.mit.edu/airlinedata/www/default.html > . Tretheway, M.W., 2004. Distortions of airline revenues: why the network airline business model is broken. J. Air Transp. Manage. 10, 3–14. Wensveen, J.G., Leick, R., 2009. The long-haul low-cost carrier: a unique business model. J. Air Transp. Manage. 15, 127–133. Whyte, R., Lohmann, G., 2015. Low-cost long-haul carriers: a hypothetical analysis of a ‘Kangaroo route’. Case Stud. Transp. Policy 3, 159–165. Wilken, D., Berster, P., Gelhausen, M.C., 2016. Analysis of demand structures on intercontinental routes to and from Europe with a view to identifying potential for new low-cost services. J. Air Transp. Manage. Wirtz, B.W., Pistoia, A., Ullrich, S., Göttel, V., 2016. Business models: origin, development and future research perspectives. Long Range Plan. 49, 36–54. de Wit, J.G., Zuidberg, J., 2012. The growth limits of the low cost carrier model. J. Air Transp. Manage. 21, 17–23. Wittmer, A., Bieger, T., Müller, R., 2011. Aviation Systems: Management of the Integrated Aviation Value Chain, first ed. Springer.

234