The Spatial Configuration Of Airline Networks In Europe
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View The Spatial Configuration Of Airline Networks In Europe as PDF for free.
More details
- Words: 7,158
- Pages: 15
ARTICLE IN PRESS
Journal of Air Transport Management 9 (2003) 309–323
The spatial configuration of airline networks in Europe Guillaume Burghouwta, Jacco Hakfoortb,*, Jan Ritsema van Eckc a
Faculty of Geographical Sciences, Utrecht University, The Netherlands b Ministry of Economic Affairs, The Hague, The Netherlands c Netherlands Institute for Spatial Research, The Hague, The Netherlands
Abstract The deregulation of US domestic passenger aviation in 1978 resulted in the reconfiguration of airline networks into radial route systems, spatially concentrated around a small number of central airports or ‘hubs’. This paper investigates whether a similar spatial concentration trend can be observed in the European aviation network after deregulation at the airline network level. Using the network concentration index, it is demonstrated that European ‘flag carriers’ already showed a very high traffic concentration rate at the beginning of deregulation. Between 1990 and 1999, the distribution of European traffic of these carriers remained remarkably stable according to the network concentration index. A spatial concentration trend of European traffic on a small number of hubs can only be observed for some regional airlines. r 2003 Elsevier Ltd. All rights reserved. Keywords: Airlines; Airports; Aviation networks; Concentration index
1. Introduction The deregulation of US aviation in 1978 resulted in the reconfiguration of airline route networks. Flights were concentrated on a small number of central airports or ‘hubs’ through which a carrier operates a number of daily ‘waves’ of flights. A hub-and-spoke network requires a concentration of traffic in both space and time (Reynolds-Feighan, 2001). Subsequently, an airline network configuration can be described along two basic dimensions: the spatial and temporal distribution of traffic in the network. Between 1987 and 1997, the European aviation market has gradually been deregulated. Yet, with respect to the spatial and temporal dimension of airline networks, there is still no comprehensive picture how European airline networks have changed after deregulation. This paper investigates how the spatial dimension of European airline networks has changed between 1990 and 1999. By spatial dimension we mean the geographical structure of airline networks. Future research should deal with the temporal dimension of airline *Corresponding author. E-mail address: [email protected] (J. Hakfoort). 0969-6997/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0969-6997(03)00039-5
networks (indirect connecting opportunities at hubs). We use data for all national, regional, low-cost and extra-EU carriers operating at least one intra-EU service a week. We consider intra-European scheduled services. The network concentration (NC) index, derived from the Gini-index, is used to measure traffic distributions in the network.
2. Background The European aviation market has gradually been deregulated by means of three ‘packages’ of deregulation measures in 1987, 1989 and 1992. (Button et al., 1998; Hakfoort, 1999). As a result of deregulation, the balance of power in the European air transport regime has shifted from the governments towards the European airlines. After the deregulation of the passenger aviation market in the US in 1978, airlines took advantage of the possibilities of the liberalised market and reorganised their networks. A number of ‘trunkline’-carriers reorganised their networks from ‘point-to point’ into ‘hub-and-spoke’ networks (Viscusi et al., 1998). This reorganisation took place between 1978 and 1985, according to Reynolds-Feighan (2001). Flights between
ARTICLE IN PRESS 310
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
medium airports were increasingly replaced by indirect flights via central airports or ‘hubs’ (see, for example, Viscusi et al., 1998). These hub-and-spoke networks allow airlines to benefit from certain cost and demand advantages, extensively described elsewhere (e.g., Brueckner and Spiller, 1994; Pels, 2001). Overall, spatial concentration and temporal concentration are the two main features of the hub-andspoke network. The hubbing carrier concentrates its network around one or a small number of transfer hubs through which the airline operates synchronized, daily waves of flights (Graham, 1995; Reynolds-Feighan, 2000). On the other hand, some new or recently started US airlines continued operating ‘point-to-point’ networks on a low-cost, no-frill, low-price basis. Low-cost carriers do not need the cost advantages of hub-and-spoke networks because they have low marginal costs per passenger. This is mainly the result of operating high density routes with high utilisation rates, high density seating, electronic ticketing, standardisation of aircraft types and maintenance, low levels of on-board service, use of under-utilised secondary airports and flexible labour contracts (Dempsey and Gesell, 1997; Williams, 2001). However, the resulting network configuration of both full-service and low-cost carriers will not only depend on the cost structure of the airline. Other factors include the size of the origin–destination market, historical background of the carrier and its network, number of stations in the network, intercontinental versus continental orientation, fleet composition, strategic airline management, hub capacity and the average stage length. In contrast to the large amount of empirical studies regarding the changes in airline network structures in the deregulated US air transport market, the number of empirical studies with respect to changing airline network configurations in Europe is still somewhat limited. More knowledge of airline network behaviour in a deregulated European aviation regime is important because of a number of reasons. First of all, the structure of airline networks and the distribution of traffic over these networks affect airport planning and development. Effects include issues in relation to peaking problems, uncertainty in airport traffic forecasting, runway construction plans, terminal lay-outs, regional economic effects, noise effects and accessibility (Caves and Gosling, 1999). Secondly, it can be expected that the effects of deregulation on the European airline network configurations will be different from the US aviation network because of the differences in the geographical, economic and political structure of both markets (see Bootsma, 1997). This paper tries to reduce an apparent gap in the literature. The limited number of existing empirical
studies on European network configurations take the airport as the object of analysis instead of the airline (e.g., Graham, 1997; Reynolds-Feighan, 1995; De Wit et al., 1999; Veldhuis and Kroes, 2002). However, in deregulated markets changes in airport connectivity are primarily the consequence of changes in airline network behaviour. Therefore, we use the airline perspective for the analysis of the spatial dimension of European airline network configurations during the period 1990–1999. Moreover, we introduce the NC-index in order to measure changes in airline network configurations. The NC-index is a modified Gini-index. The Gini-methodology was introduced in air transport analysis by Reynolds-Feighan (1998, 2001) for the measurement of spatial or market concentration in airline networks.
3. Methodology According to Reynolds-Feighan, the hub-and-spoke network requires a concentration of air traffic in both space and time. Hence, the network configuration is defined as the level of spatial and temporal concentration of traffic flows in a given network. The focus here is on the spatial concentration of the airline network. To measure the level of spatial concentration of a network, different concentration measures can be used, such as the coefficient of variance, the Herfindahl-index, Theil’s entropy measure, the C4-firm concentration ratio or the Gini-index. ReynoldsFeighan suggests the Gini-index as the most appropriate concentration measure for airline or airport networks. Allison (1978) and Sen (1976) examined the properties of income inequality measures and proposed a series of criteria that indices should possess. The C4-index only reacts to changes in the traffic distribution in an airport population when the four largest airports are involved. The Herfindahl-index is only sensitive for changes in the extremes of the population. The coefficient of variance on the other hand, reacts well to changes in the population but is extremely sensitive to the underlying distribution. The Gini-index was the only index to satisfy all of the criteria. The Gini-index is not sensitive to the distribution of the population and reacts quite well to changes in all parts of a given population. Regarding spatial concentration, the Gini-index can be defined as G¼
1 XX jyi yj j 2n2 y% i j
ð1Þ
where y is the air traffic at airport i or j; defined as the total number of supplied seats per week; n is the number of airports in the airline network.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
311
Network concentration level
1 0.9
Small radial
Large radial
Small linear
Large linear
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
20
40
60
80
100
120
140
160
Number of airports
Fig. 1. Spatial network configuration types.
If it is assumed that the total incoming air traffic at each airport to be approximately equal to the total outgoing air traffic at the same airport, no airport will command more than half the total air traffic. Therefore, the Gini-index cannot reach its theoretical maximum value of 1. The maximum Gini score increases with the number of airports in an airline network (n) and can be computed as follows:
command equal shares of the total traffic will result in an NC of 0. This includes a multitude of possible configurations, from fully connected networks at one extreme, to collections of isolated routes at the other extreme. An increase in NC indicates a more unequal spread of traffic over the network. This may be caused by *
Gmax
2 ¼1 n
ð2Þ
This maximum Gini-index can be observed in a single hub-network where traffic flows are concentrated on one hub-spoke route. Theoretically, this makes sense since a single hub system is most efficient from the airline viewpoint (Dennis, 2001; O’Kelly, 1998) and the concentration of traffic on one route allows for further advantages of scale. In this study we will correct for the size of the airline network (number of airports) when measuring NC. We define the level of NC as NC ¼
G Gmax
ð3Þ
where G is the observed Gini-index in a network and Gmax the maximum Gini-index given the number of airports in the network. In contrast to the use of the standard Gini-index, our NC-index makes it possible to compare the spatial structure of airline networks independent from network size. NC varies between 0 and 1. A NC of 1 corresponds to a single hub-network where traffic flows are concentrated on one hub-spoke route. Visual inspection of network configurations indicates that NC values of dual or triple hub networks generally range between 0.7 and 0.8. This makes sense, since integrated dual or triple hub-networks require concentration of traffic on the routes between the hubs. A single hub-network with traffic divided equally over all hub-spoke routes has an NC of 0.5. Any network where all airports
*
*
A more unequal distribution of seat capacity over the routes in a single or multiple hub-radial network. This is generally the consequence of route-specific differences in the growth of passenger demand. The removal of routes between smaller airports in the network. The airline concentrates its network on one or a few primary airports. This can be the consequence of hub-and-spoke strategy in order to maximise connecting opportunities. It can also be the consequence of the outsourcing of ‘hub-bypass’ routes to subsidiaries of the airline or regional alliance partners. The transformation of an airline network from a single hub to a multiple hub network.
In order to characterise the (changing) spatial network configuration of a carrier, we distinguish between two dimensions of the airline network. On the one hand the scope of the network using n; the number of airports. On the other hand we use the level of network concentration (NC), independent from n; expressed as NC (3). The level of NC labels a carrier as a radial or linear carrier, with regard to the spatial organisation of its network. Combining n and NC, four ideal network configurations can be distinguished (Fig. 1): (1) the small radial network, (2) the large radial network, (3) the small linear network and (4) the large linear network. To structure the discussion, four ideal dynamics in network configurations are distinguished: *
Concentrated network builders: airlines with both an increase in network size and NC-index.
ARTICLE IN PRESS *
*
*
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
Deconcentrated network rationalisers: Airlines with both a decline in network size and NC-index. Concentrated network rationalisers: Airlines with an increasing spatial concentration of traffic but a shrinking network. Deconcentrated network builders: Airlines with a growing network but a decline in NC.
With the static and dynamic typology of airline networks, insight can be given into the spatial network strategies of airlines in Europe. We will use the static (Fig. 1) and dynamic typologies to describe airline network configurations in Europe for different airline types.
40
30
Number of carriers
312
20
10 Std. Dev = 0.19 Mean = 0.52 N = 162.00
0 0.00
0.13 0.25 0.38 0.50 0.63 0.75 0.88 0.06 0.19 0.31 0.44 0.56 0.69 0.81 0.94
Network concentration for intra-EU traffic
4. Data and selection of carriers 40
*
*
*
*
National airlines: Carriers (formerly) designated as the flag carrier of a member state of the Common European Aviation Area (e.g. KLM, Air France). Low-cost, no-frills airlines: All European-based carriers operating on a low-cost scheduled or mixed (low-cost scheduled/charter) basis (e.g. Ryanair, easyJet). See for a detailed description: Williams (2001). Regional airlines: All other airlines registered in the EU. Our database included 124 regional carriers in 1999 (e.g. Braathens, British Midland). Ten regional airlines were selected to illustrate network developments. These airlines were taken from the top twenty of largest regional carriers in terms of intra-EU seat capacity in 1999. Extra-EU airlines: All airlines with intra-EU services but registered in a non-EU country (e.g. TWA, Cathay Pacific).
5. Analysis of airline network configurations: a first look In order to evaluate the spatial dimension of European airline network configurations, the NC scores and Gini indices of all carriers operating on intra-EU routes were computed. Fig. 2 shows the frequency
30
Number of carriers
The data set used consists of OAG/ABC data for the years 1990–1999. The OAG/ABC data set contains variables based on published information on scheduled flights. Variables include departure airport, destination airport, flight frequency, aircraft type and seat capacity. The data are based on a representative week in July of each year. We refer to Burghouwt and Hakfoort (2001) for a description of the data. In order to analyse the changing network configurations of different types of European carriers as discussed in Section 1, the European airlines were categorised into four groups
20
10 Std. Dev = 0.16 Mean = 0.61 N = 122.00
0 0.00
0.13 0.25 0.38 0.50 0.63 0.75 0.88 0.06 0.19 0.31 0.44 0.56 0.69 0.81 0.94
Network concentration for intra-EU traffic
Source: OAG/ABC data Fig. 2. Frequency distribution of NC values in 1990 (above) and 1999 (below).
distribution of NC both in 1990 and 1999. It can be concluded that between 1990 and 1999 more and more airlines adopted radial networks with high NC levels. Both mean and median moved towards a higher NC rate. Linear networks with low NC values were less frequent in 1999 compared to 1990. Fig. 3 shows the evolution of NC between 1990 and 1999 according to carrier type. The national carrier network configurations have overall higher NCindices than other airline types. This is a plausible outcome. It reflects the national carriers orientation towards their national hubs. Because of the asymmetric deregulation of the European aviation market, carriers are still bounded to their country of registration for their intercontinental services. These intercontinental ‘chains’ have resulted in a stable NC between 1990 and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323 low-cost
regional
national
extra-EU
type of carrier regional
extra-EU
0.9
national
low-cost
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Source: OAG/ABC data Fig. 3. Evolution of NC levels for different airline types, 1990–1999.
Network concentration for intra-EU traffic
1.0
0.8 Network concentration
313
0.8
0.6
0.4
0.2
0.0 0
Extra-EU airlines have the lowest NC scores. This is largely caused by the nature of air services of these carriers within Europe: when they operate intra-EU services, their network is linear, based on fifth freedom rights (Weber and Dinwoodie, 2000).
6. National airlines National airline networks can be characterized as medium or large radial networks (Fig. 4). Differences between the national carriers in terms of concentration indices for intra-EU traffic are small. They varied between 0.63 (Austrian) and 0.8 (Aer Lingus) in 1999. The NC levels of national European carriers have been quite stable between 1990 and 1999 (Table 1). Ten out of sixteen nationals faced an increase in the number of destinations. From a dynamic point of view, four types of national carrier network strategies can be distinguished (Figs. 5–8): 1. Concentrated network builders: Carriers such as British Airways, Finnair and SAS have not only enlarged the scope of their network but also focused this growth on their major hubs and/or a limited number of intra-EU destinations. In the case of SAS, the carrier focused its network growth both in terms of intra-EU capacity and number of routes on its primary airports Copenhagen, Stockholm and Oslo. The three primary airports in the SAS network experienced above average growth levels. The result was a higher NC-index. 2. Deconcentrated network builders: These carriers distributed seat capacity more equally over a larger
20
40
60
80
100
120
number of intra-EU airports
Source: OAG/ABC data Fig. 4. Number of airports and NC levels for different airline types in 1999.
Table 1 Number of airlines, average network concentration level per airline, average Gini scores per airline, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, national airlines, 1990–1999
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Network concentration
Gini index
No. Airports
Capacity %
16 16 16 16 16 16 16 16 16 16
0.744 0.739 0.730 0.730 0.729 0.728 0.728 0.730 0.726 0.729
0.705 0.702 0.694 0.695 0.693 0.693 0.694 0.695 0.691 0.695
44.8 46.4 46.4 46.6 45.9 47.1 48.8 49.1 51.6 53.4
3.1 3.2 3.4 3.5 3.5 3.5 4.2 4.4 5.1 5.5
Source: OAG/ABC; own calculations.
number of airports. The KLM network demonstrates a clear and growing radial network structure focused on Amsterdam Schiphol. Nevertheless, the carrier faced a small decrease in NC values. This is largely caused by the more equal capacity distribution of flow sizes over the routes originating in Amsterdam in 1999 compared to 1990. 3. Concentrated network rationalisers: Some flag carriers limited the number of intra-EU destinations
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
TAP Air Potugal
Swissair
Sabena
SAS
Austrian
Olympic Airways
Lufthansa
KLM
Iberia
Icelandair
Aerlingus
British Airways
Alitalia
Finnair
Luxair
1999
1990
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Air France
Network concentration for intra-EU traffic
314
Fig. 5. NC indices for national airlines, 1990 and 1999. Source: OAG/ABC data
1999
120 100 80 60 40 20 TAP Air Potugal
Swissair
Sabena
SAS
Austrian
Olympic Airways
Lufthansa
Luxair
KLM
Iberia
Icelandair
Aerlingus
British Airways
Alitalia
Finnair
0 Air France
Number of airports served
1990
Fig. 6. Number of EU airports served, national airlines, 1990 and 1999. Source: OAG/ABC data
between 1990 and 1999 but faced an increase in the NC-index. TAP Air Portugal, for example, concentrated the growth of seat capacity on Lisbon and Oporto during the period of analysis. At the same time, the airline rationalised its network by cutting the number of direct destinations from Lisbon, Oporto and Faro. 4. Deconcentrated network rationalisers: A small group of nationals not only faced a decline in network size but also deconcentrated the capacity of its network. Swissair, for example, outsourced a large number of its intra-EU Geneva destinations to its regional daughter Crossair but maintained most of its services
out of Zurich. . The network moved from a dual to a single hub-structure resulting in a decrease in the NC-index (Appendix B).
7. Regional carriers Regional carriers have much smaller networks and lower concentration rates (NC) than the national carriers (Fig. 4, Table 2). In 1999 they accounted for about 26 percent of total intra-EU capacity. The group of regional carriers is diverse: the group ranges from single city-pair carriers to regional carriers operating at
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
a pan-European level. The Swiss airline Crossair for example served more intra-EU destinations in 1999 than most of the flag carriers (Figs. 4 and 8). In 1999, most of the regional carriers had less concentrated networks than the nationals. But also with respect to NC, a large variation in seat capacity distributions can be observed (Figs. 4 and 9): the regional network configurations range from typical linear networks (Wideroe’s, Suckling Airways), radial networks combined with a large amount of point-topoint routes (Skyways, Regional Airlines) to concentrated radial networks (Crossair, Air Liberte! , Braathens). Yet, the radial network dominated the scene in 1999. Only Wideroe’s, Suckling Airways and Oltostfriesische Lufttransport had considerable linear networks in 1999.
315
Regional carrier network configurations have been less stable than those of the national carriers between 1990 and 1999, both with respect to the network size as to the level of concentration (Table 2). While changes of NC of national carriers did not exceed 0.1 points between 1990 and 1999, NC levels of a number of regional carriers have changed dramatically (Fig. 10). Table 4 and Fig. 10 show the different network strategies of a number of regional carriers. Most of the regional airlines concentrated their network in some extent around one or two central hub airports. This can be an independent, mini-hub expansion strategy in the case of the French carrier Regional Airlines at Clermont-Ferrand (Appendix B). However, the carrier had a relatively low NC score because of the large amount of point-to-point routes it still operated in
Change in network concentration, 1990-1999
0.08 0.06 AY
TP
BA
0.04 SK 0.02
FI
0.00
IB
OA
EI
LG AZ
AF
-0.02
SN
-0.04 -0.06
LH
OS
KL
-0.08 SR -0.10 -20
Table 2 Number of airlines, average network concentration level per airline, average Gini scores per airline, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, regional airlines, 1990–1999
-10
0
10
20
30
40
Change in number of airports, 1990-1999
Network concentration
Gini index
No. Airports
Capacity %
124 110 120 116 121 120 116 112 116 124
0.560 0.576 0.581 0.578 0.582 0.569 0.558 0.589 0.563 0.614
0.299 0.284 0.296 0.298 0.306 0.305 0.322 0.322 0.307 0.321
8.3 8.4 8.5 8.5 8.4 8.7 9.4 9.7 9.3 9.3
0.2 0.2 0.2 0.3 0.3 0.3 0.3 0.3 0.3 0.3
Source: OAG/ABC; own calculations.
1990
1999
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 IT IS H BR M AA ID LA TH N EN D S S. A. AI F. R E LI TT O R AL AI R LI BE R TE C R O SS AI EU R R O W IN G S R KL EG M IO U N K AL TY R A IR O W LE LI ID N AN ER ES O AI E' R S W FL AY YV S ES EL SK AP
0
BR
Network concentration for intra-EU traffic
Fig. 7. Change in network size and NC for national carriers, 1990–1999 (see Appendix A for airline codes). Source: OAG/ABC data
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Fig. 8. NC indices for selected regional carriers, 1990 and 1999. Source: OAG/ABC data
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
1990
70 60 50 40 30 20 10 0
1999
BR IT IS H BR M AA ID LA TH N EN D S S. A. AI F. E R LI TT O R AL AI R LI BE R TE C R O SS AI R EU R O W IN G S R KL EG M IO U N K AL TY AI R O R W LE LI ID N AN ER ES AI O E' R W S AY FL YV S ES EL SK AP
Number of airports
316
Fig. 9. Number of airports for selected regional carriers, 1990 and 1999. Source: OAG/ABC data
Table 3 Network strategies of regional airlines
Network growth
Network decline
Concentration
Deconcentration
Concentrated network builders: British Midland, Braathens S.A.F.E., Deutsche BA, Aurigny Air Services, Hahn Air, Air Libert!e, Augsburg Airways, Jersey European Airways, Skyways, Air Botnia, Crossair, Eurowings, Olt Ostfriesische Lufttransport, KLM UK, Regional Airlines, Wideroe’s, Aero Lloyd, Monarch Airlines, Muk Air Concentrated network rationalizers: Air Littoral, Manx Airlines
Deconcentrated network builders: Suckling Airways, Maersk Air, Sata Air Acores, Tyrolean Airways
Deconcentrated network rationalizers: Air Sicilia
Source: OAG/ABC; own calculations.
0.4 YQ
change in NC, 1990-1999
0.3
0.2
UK IJ JY KF
DI FU
-0.0
JZ
IQ
0.1
BU
ZB
BD
LX YP
WF JE ZR OL
BM
VM
EW
VO
DM
SP IG
-0.1
-0.2
CB
-0.3 -20
-10
0
10
20
30
40
change in number of airports, 1990-1999 Fig. 10. Change in network size and NC for regional carriers having scheduled service both in 1990 and 1999, 1990–1999 (see Appendix A for airline codes). Source: OAG/ABC data
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
1999. It can also be the consequence of alliances between national and regional carriers, in which the regional carriers play a hub-feeding role (Graham, 1997). For example, Eurowings and Air UK centered their network on KLM’s hub Amsterdam Schiphol between 1990 and 1999. Even the Norwegian carrier Wideroe’s-probably the best example of (PSO subsidized) linear carrier in Europe-focused part of its network around Oslo during the period of analysis (Appendix B). Regional linear strategies among regional carriers are rare in Europe. It underlines the conclusion of various authors that spatial concentration of traffic at a limited number of airports has cost advantages, even without the transfer of passengers from one flight to another. These cost advantages include the spread of fixed costs over more passengers as well as easy crew rotation and aircraft maintenance.
Table 4 Number of airlines, average network concentration level per airline, average Gini scores, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, low-cost airlines, 1990–1999 Network concentration
Gini index
No. Airports
Capacity %
2 2 3 3 5 6 8 9 9 9
0.654 0.742 0.763 0.657 0.660 0.611 0.667 0.630 0.647 0.617
0.280 0.389 0.566 0.498 0.524 0.484 0.518 0.543 0.561 0.551
8.0 6.5 8.5 9.3 15.2 13.7 13.5 19.9 21.3 24.1
0.2 0.2 0.3 0.2 0.3 0.6 0.6 1.0 1.1 1.3
8. Low-cost carriers The entrance of low-cost carriers to the European aviation market was one of the most profound effects of deregulation. In the US, a number of low-cost carriers, such as Southwest, operate linear networks with low levels of NC. In Europe, the low-cost carrier networks seem to be as concentrated as the networks of national carriers (Table 4, Figs. 3, 11 and 13). All the low-cost carriers operated out of central airports. Ryanair, for example, used Dublin and Stansted as its central airports in 1999. The only exception is Virgin Express (former Eurobelgian). Besides its Brussels-network, Virgin Express also operated services between Madrid–Barcelona–Rome and Shannon–Stansted in 1999. The difference in the network configurations between low-cost and national carriers can mainly be found in network size (Fig. 12). This is because low-cost carriers have been operating air services only for a short period. Therefore, it makes little sense to categorise the carriers according to their network strategies. However, the described networks of the low-cost carriers are about to change. The low-cost carriers experienced fast growth after 1999. Low-cost carriers did not suffer as much from the crisis in the air transport industry after September 11th because of the low fare levels and the fact that low-cost carriers transport relatively few passengers from the American continent
base year
AIR EUROPA
EASYJET
SPANAIR
HAPAG LLOYD FLUGGESSELSCHAF
RYANAIR
CONDOR FLUGDIENST
1999
VIRGIN EXPRESS
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
DEBONAIR AIRWAYS
Network concentration level
Source: OAG/ABC; own calculations.
From Fig. 10 it can be concluded that the only carrier with a significant, regional, linear strategy is Suckling Airways. In 1990 the carrier operated a Cambridge– Manchester and Cambridge–Amsterdam route. In 1999 the airline had services to 11 destinations but did not operate out of a central airport in the UK.
TRANSAVIA AIRLINES
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
317
Source: OAG/ABC data. Base year for low-cost carriers: Debonair: 1996; Virgin: 1995; Condor: 1993; Ryanair: 1990; Hapag Lloyd: 1997; Transavia: 1991; Spanair: 1994; easyJet: 1996; Air Europa: 1994 Fig. 11. NC indices for low-cost carriers, base year and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323 50 45 40 35 30 25 20 15 10 5 0
base year
AIR EUROPA
EASYJET
SPANAIR
TRANSAVIA AIRLINES
HAPAG LLOYD FLUGGESSELSCHAF
RYANAIR
CONDOR FLUGDIENST
VIRGIN EXPRESS
1999
DEBONAIR AIRWAYS
Number of intra-EU airports
318
Source: OAG/ABC data. Base year for low-cost carriers: Debonair: 1996; Virgin: 1995; Condor: 1993; Ryanair: 1990; Hapag Lloyd: 1997; Transavia: 1991; Spanair: 1994; easyJet: 1996; Air Europa: 1994 Fig. 12. Number of intra-EU airports served by low-cost carriers in base year and 1999.
Network concentration level
0.9 0.8 0.7 0.6 0.5 0.4 0.3 1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
JK
U2
UX
Year 2G
BQ/TV
DE
FR
HF
HV
Source: OAG/ABC data Fig. 13. Change NC for low-cost carriers 1990–1999 (see Appendix A for airline codes).
compared to the national airlines. New low-cost subsidiaries have entered the market (Buzz, Go). British low-cost carriers have expanded their services on the European continent (Ryanair, easyJet). Ryanair, for example, opened new base airports at Frankfurt Hahn and Brussels Charleroi. In March 2003, Ryanair served about 65 destinations in contrast to 33 in 1999. Based on the flight schedule available on the website of Ryanair, we have computed the NC-index for Ryanair in 2003. Ryanair had an NC value of 0.7 in March 2003 (against 0.68 in 1999). Hence, the network of Ryanair still centred on multiple central airports (London Stansted, Dublin, Brussels Charleroi, Frankfurt Hahn, Glasgow Prestwick, Stockholm Skavsta and Shannon). Ryanair does not have a criss-cross or linear network such as the network of low-cost carrier South-
west in the US with a NC value of 0.47 in 1999. However, further expansion of Ryanair at the European continent and the increase in the number of home bases may eventually result in a lower NC value for the network.
9. Extra-EU airline networks For a long time, the use of 5th freedom rights by international airlines was the only possibility to create world-embracing network. The 5th freedom is ‘the right of an airline of one country to carry traffic between countries outside of its own country of registry as long as the flights originates or terminates in its own country of registry’ (Button and Taylor, 2000). This traffic right
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
was defined at the Chicago Conference of 1944 and could be agreed upon in air service agreements between states. It enables international airlines to operate routes that would not be economically viable as non-stop destination. Overall, the 5th freedom networks of extra-EU airlines have a linear character with very low NC values compared to the national and regional airline networks. However, a significant decline of NC values together with a decline in capacity share/number of airports served took place during the period of analysis. This development is in line with the study of Weber and Dinwoodie (2000) on 5th freedom traffic operations in the Single European Aviation market. They state that, in the early 1990s, 5th freedom traffic was the only opportunity for extra-EU airlines to develop a European network with destinations that would not be
Table 5 Number of airlines, average network concentration level per airline, average Gini scores, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, extra-EU airlines, 1990–1999
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Network concentration
Gini index
No. Airports
Capacity %
72 66 67 68 37 30 41 34 31 34
0.362 0.349 0.301 0.337 0.403 0.423 0.353 0.435 0.373 0.302
0.159 0.136 0.123 0.131 0.108 0.118 0.079 0.078 0.064 0.034
4.8 4.5 4.6 4.5 3.6 3.9 2.9 2.7 2.5 2.2
0.08 0.06 0.06 0.06 0.05 0.10 0.03 0.02 0.02 0.02
viable as non-stop routes. Some airlines, such as Delta (the airline acquired the extensive 5th freedom network of Pan Am), TWA and United Airlines set up mini-hub operations within Europe based on 5th freedom rights. These kind of operations resulted in relatively high NC values in 1990 compared to 1999. Weber and Dinwoodie (2000) identify a number of factors that limited the 5th freedom traffic in Europe during the period of analysis: *
*
*
*
Source: OAG/ABC; own calculations.
1990
0.9
1995
Global airline alliances: Alliances form a substitute or even a refinement of 5th freedom traffic: European partner airlines can offer better connectivity to more destinations. In 1991, for example, the US carrier United Airlines, used London Heathrow as its European hub, connecting its intercontinental services with smaller Boeing 727s to Brussels, Amsterdam and Paris Charles de Gaulle. Gradually, the carrier shifted its European gateway to Frankfurt, anticipating to the alliance-agreement with Lufthansa in 1993. Now, the European partners of the Staralliance (e.g. Lufthansa, SAS) provide the intra-EU services. Technological developments: Introduction of the Airbus A340 and Boeing 777 has enabled airlines to operate economically thin long-haul routes incapable of supporting Boeing 747–400 s. Product life cycle of international routes: A number of 5th freedom routes has reached maturity and have sufficient demand to justify non-stop service. Competition: Competition of European carriers with higher corporate strength in the markets served makes it less attractive to operate 5th freedom routes. The implementation of the ‘Third Package’ in 1992 resulted in a significant decrease of 5th freedom operations of extra-EU carriers (Table 5).
1999
Network concentration level
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 AC
CX
DL
GA
KU
319
PA
QF
SQ
TG
TW
UA
Source OAG/ABC data Fig. 14. NC values for selected extra-EU airlines (see Appendix A for carrier codes).
ARTICLE IN PRESS 320
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
These developments resulted in a decline in market share of extra-EU airlines and lower NC values. A number of US carriers and Asian carriers gave up most of their fifth freedom operations (Fig. 14).
10. Conclusion Both the spatial concentration of seat capacity on a limited number of hubs and the temporal concentration of flights around a number of synchronized waves of flights per day, are essential features of a hub-and-spoke system. This paper has presented an overview of the changes in the spatial concentration of airline networks in Europe during the period 1990–1999. Spatially concentrated networks possess a radial route structure whereas deconcentrated networks possess a linear or criss-cross route structure. The NC-index, derived from the Gini-index, has been used to measure the spatial distribution of traffic flows in airline networks. Combined with the number of airports served, the NC-index makes it easy to detect changes in the structure of airline networks. We have applied the NC to answer the question to what extent airlines in Europe operate radial or linear networks and how these networks have developed over time. Taken together, the networks of European airlines are increasingly concentrated. Radial networks have been developed between 1990 and 1999. Yet, large differences exist between the various airline types. The national carrier networks can be characterised as large radial networks. The ‘flag’ carriers are still bounded to their national airports because of bilateral regulation of intercontinental services. This implies the operation of radial network structures. Moreover, large networks are complex networks and have many dimensions to absorb new developments (Swan, 2002). Therefore, changes in the NC level of national carriers have been quite small. Radical geographical restructuring of networks of major carriers, as in the US after deregulation, cannot be observed in Europe. Nevertheless, at a smaller scale, radial strategies can be observed among regional carriers. Taking advantage of the liberalised market, some regional airlines restructured their networks from linear into radial networks. Others started from a small traffic base at a central airport and, due to alliances with national carriers, benefited from traffic feed of large intercontinental carriers. Some of the regional carriers, such as Crossair, had even larger European networks than most of the national airlines. Large linear networks were an exceptional phenomenon in Europe at the end of the nineties. Regional airlines such as Suckling Airways, the PSO subsidized Norwegian carrier Wideroe’s and the German Oltfrie-
sische Lufttransport operated some kind of linear network in 1999. Only Suckling Airways intensified its linear network structure. In contrast to the US, European low-cost carriers focused their service on a limited number of airports until 1999. Yet, signs of change are already there: very fast growth, inauguration of new routes, entrance of British low-cost carriers at the continent and use of secondary airports might be the first signs of linear/crisscross networks of low-cost carriers. The evolution of the intra-European aviation network can be characterised by both a spatial deconcentration trend at the airport level and a spatial concentration trend at the airline network level. Intra-EU traffic is spreading over more and more airports (Burghouwt and Hakfoort, 2001). Although we cannot draw any firm conclusions about the factors decisive for the network developments observed based on our data, deregulation seems to have given regional airlines and airports the opportunity to play an important role in the European aviation scene. Economic growth and lower prices generated sufficient demand for new services, not only from primary but also from smaller airports. At the same time, a number of regional carriers concentrated their networks around one or two central airports during the period of analysis. National carrier networks were already focused on key nodes at the beginning of deregulation. In the planning process of these central airports, airport authorities and governments will have to cope with periods of very fast traffic growth at hub airports. This may result in increasing noise pollution in airport regions, peaking problems and new terminal layout requirements. Airports forming part of the network of hubbing carriers, will be more and more dependent of network decisions of these airlines and their alliances for their future network position.
Appendix A. Carrier codes National carriers AF AIR FRANCE AY FINNAIR AZ ALITALIA BA BRITISH AIRWAYS EI AER LINGUS FI ICELANDAIR IB IBERIA KL KLM-ROYAL DUTCH AIRLINES LG LUXAIR LH LUFTHANSA GERMAN AIRLINES OA OLYMPIC AIRWAYS OS AUSTRIAN AIRLINES SK SAS SCANDINAVIAN AIRLINES
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
SN SR TP
SABENA SWISSAIR TAP AIR PORTUGAL
Low cost carriers 2G DEBONAIR AIRWAYS TV VIRGIN EXPRESS DE CONDOR FLUGDIENST FR RYANAIR HF HAPAG LLOYD FLUGGESSELSCHAF HV TRANSAVIA AIRLINES JK SPANAIR U2 EASYJET UX AIR EUROPA Regional carriers BD BRITISH MIDLAND BM AIR SICILIA BU BRAATHENS S.A.F.E CB SUCKLING AIRWAYS DM MAERSK AIR EW EUROWINGS FU AIR LITTORAL IG MERIDIANA IJ AIR LIBERTE! IQ AUGSBURG AIRWAYS JE MANX AIRLINES JY JERSEY EUROPEAN AIRWAYS JZ SKYWAYS
KF SP UK VM VO WF YP YQ ZB ZR Extra-EU AC CX DL GA KU PA QF SQ TG TW UA
321
AIR BOTNIA SATA AIR ACORES KLM UK REGIONAL AIRLINES TYROLEAN AIRWAYS WIDEROE’S FLYVESELSKAP AERO LLOYD HELIKOPTERSERVICE EURO AIR MONARCH AIRLINES MUK AIR carriers AIR CANADA CATHAY PACIFIC AIRWAYS DELTA AIR LINES GARUDA INDONESIA KUWAIT AIRWAYS PAN AMERICAN WORLD AIRWAYS QANTAS AIRWAYS SINGAPORE AIRLINES THAI AIRWAYS INTERNATIONAL TRANS WORLD AIRLINES UNITED AIRLINES
Appendix B Mini-hub expansion strategy in the case of the French carrier Regional Airlines at Clermont-Ferrand:
ARTICLE IN PRESS 322
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
Development of the network of Swissair between 1990 and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
References Allison, P.D., 1978. Measures of inequality. American Sociological Review 43, 865–880. Bootsma, P.D., 1997. Airline Flight Schedule Development; Analysis and Design Tools for European Hinterland Hubs. University of Twente, Utrecht. Brueckner, J.K., Spiller, P.T., 1994. Economies of traffic density in the deregulated airline industry. Journal of Law and Economics 37, 379–415. Burghouwt, G., Hakfoort, J.R., 2001. The European aviation network, 1990–1998. Journal of Air Transport Management 7, 311–318. Button, K.J., Taylor, S.Y., 2000. International air transportation and economic development. Journal of Air Transport Management 6, 209–222. Button, K.J., Haynes, K., Stough, R., 1998. Flying into the Future. Air Transport Policy in the European Union. Edward Elgar, Cheltenham. Caves, R.E., Gosling, G.D., 1999. Strategic Airport Planning. Pergamon, Amsterdam. Dempsey, P.S., Gesell, L.E., 1997. Airline Management. Strategies for the 21st century. Coast Aire Publications, Chandler. Dennis, N.P., 2001. Developments of hubbing at European airports. Air and Space Europe 3, 51–55. de Wit, J., Uittenbogaart, P., Wei-Yun, T., 1999. Hubbing and hubbypassing. Network developments in a deregulated European airport system. ATRG Conference, Hong Kong. Graham, B., 1995. Geography and Air Transport. Wiley, Chichester. Graham, B., 1997. Regional airline services in the liberalized European Union single aviation market. Journal of Air Transport Management 3, 227–238.
323
Hakfoort, J.R., 1999. The deregulation of European air transport: a dream come true? TESG 90, 226–233. O’Kelly, M.E., 1998. A geographer’s analysis of hub-and-spoke networks. Journal of Transport Geography 6, 171–186. Pels, E., 2001. A note on airline alliances. Journal of Air Transport Management 7, 3–7. Reynolds-Feighan, A., 1995. European and American approaches to air transport liberalisation: some implications for small communties. Transportation Research-A 29, 467–483. Reynolds-Feighan, A., 1998. The impact of US airline deregulation on airport traffic patterns. Geographical Analysis 30, 234–253. Reynolds-Feighan, A., 2000. The US airport hierarchy and implications for small communities. Urban Studies 37, 557–577. Reynolds-Feighan, A., 2001. Traffic distribution in low-cost and fullservice carrier networks in the US air transport market. Journal of Air Transport Management 7, 265–275. Sen, A., 1976. Poverty: an ordinal approach to measurement. Econometrica 44, 219–231. Swan, W.M., 2002. Airline route developments: a review of history. Journal of Air Transport Management 8, 349–353. Veldhuis, J., Kroes, E., 2002. Dynamics in Relative Network Performance of the Main European Hub Airports. European Transport Conference, Cambridge. Viscusi, W.K., Vernon, J.M., Harrington, J.E., 1998. Economics of Regulation and Antithrust. MIT Press, Cambridge. Weber, M., Dinwoodie, J., 2000. Fifth freedoms and airline alliances. The role of fifth freedom traffic in an understanding of airline alliances. Journal of Air Transport Management 6, 51–60. Williams, G., 2001. Will Europe’s charter carriers be replaced by ‘‘nofrills’’ scheduled airlines? Journal of Air Transport Management 7, 277–286.
Journal of Air Transport Management 9 (2003) 309–323
The spatial configuration of airline networks in Europe Guillaume Burghouwta, Jacco Hakfoortb,*, Jan Ritsema van Eckc a
Faculty of Geographical Sciences, Utrecht University, The Netherlands b Ministry of Economic Affairs, The Hague, The Netherlands c Netherlands Institute for Spatial Research, The Hague, The Netherlands
Abstract The deregulation of US domestic passenger aviation in 1978 resulted in the reconfiguration of airline networks into radial route systems, spatially concentrated around a small number of central airports or ‘hubs’. This paper investigates whether a similar spatial concentration trend can be observed in the European aviation network after deregulation at the airline network level. Using the network concentration index, it is demonstrated that European ‘flag carriers’ already showed a very high traffic concentration rate at the beginning of deregulation. Between 1990 and 1999, the distribution of European traffic of these carriers remained remarkably stable according to the network concentration index. A spatial concentration trend of European traffic on a small number of hubs can only be observed for some regional airlines. r 2003 Elsevier Ltd. All rights reserved. Keywords: Airlines; Airports; Aviation networks; Concentration index
1. Introduction The deregulation of US aviation in 1978 resulted in the reconfiguration of airline route networks. Flights were concentrated on a small number of central airports or ‘hubs’ through which a carrier operates a number of daily ‘waves’ of flights. A hub-and-spoke network requires a concentration of traffic in both space and time (Reynolds-Feighan, 2001). Subsequently, an airline network configuration can be described along two basic dimensions: the spatial and temporal distribution of traffic in the network. Between 1987 and 1997, the European aviation market has gradually been deregulated. Yet, with respect to the spatial and temporal dimension of airline networks, there is still no comprehensive picture how European airline networks have changed after deregulation. This paper investigates how the spatial dimension of European airline networks has changed between 1990 and 1999. By spatial dimension we mean the geographical structure of airline networks. Future research should deal with the temporal dimension of airline *Corresponding author. E-mail address: [email protected] (J. Hakfoort). 0969-6997/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0969-6997(03)00039-5
networks (indirect connecting opportunities at hubs). We use data for all national, regional, low-cost and extra-EU carriers operating at least one intra-EU service a week. We consider intra-European scheduled services. The network concentration (NC) index, derived from the Gini-index, is used to measure traffic distributions in the network.
2. Background The European aviation market has gradually been deregulated by means of three ‘packages’ of deregulation measures in 1987, 1989 and 1992. (Button et al., 1998; Hakfoort, 1999). As a result of deregulation, the balance of power in the European air transport regime has shifted from the governments towards the European airlines. After the deregulation of the passenger aviation market in the US in 1978, airlines took advantage of the possibilities of the liberalised market and reorganised their networks. A number of ‘trunkline’-carriers reorganised their networks from ‘point-to point’ into ‘hub-and-spoke’ networks (Viscusi et al., 1998). This reorganisation took place between 1978 and 1985, according to Reynolds-Feighan (2001). Flights between
ARTICLE IN PRESS 310
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
medium airports were increasingly replaced by indirect flights via central airports or ‘hubs’ (see, for example, Viscusi et al., 1998). These hub-and-spoke networks allow airlines to benefit from certain cost and demand advantages, extensively described elsewhere (e.g., Brueckner and Spiller, 1994; Pels, 2001). Overall, spatial concentration and temporal concentration are the two main features of the hub-andspoke network. The hubbing carrier concentrates its network around one or a small number of transfer hubs through which the airline operates synchronized, daily waves of flights (Graham, 1995; Reynolds-Feighan, 2000). On the other hand, some new or recently started US airlines continued operating ‘point-to-point’ networks on a low-cost, no-frill, low-price basis. Low-cost carriers do not need the cost advantages of hub-and-spoke networks because they have low marginal costs per passenger. This is mainly the result of operating high density routes with high utilisation rates, high density seating, electronic ticketing, standardisation of aircraft types and maintenance, low levels of on-board service, use of under-utilised secondary airports and flexible labour contracts (Dempsey and Gesell, 1997; Williams, 2001). However, the resulting network configuration of both full-service and low-cost carriers will not only depend on the cost structure of the airline. Other factors include the size of the origin–destination market, historical background of the carrier and its network, number of stations in the network, intercontinental versus continental orientation, fleet composition, strategic airline management, hub capacity and the average stage length. In contrast to the large amount of empirical studies regarding the changes in airline network structures in the deregulated US air transport market, the number of empirical studies with respect to changing airline network configurations in Europe is still somewhat limited. More knowledge of airline network behaviour in a deregulated European aviation regime is important because of a number of reasons. First of all, the structure of airline networks and the distribution of traffic over these networks affect airport planning and development. Effects include issues in relation to peaking problems, uncertainty in airport traffic forecasting, runway construction plans, terminal lay-outs, regional economic effects, noise effects and accessibility (Caves and Gosling, 1999). Secondly, it can be expected that the effects of deregulation on the European airline network configurations will be different from the US aviation network because of the differences in the geographical, economic and political structure of both markets (see Bootsma, 1997). This paper tries to reduce an apparent gap in the literature. The limited number of existing empirical
studies on European network configurations take the airport as the object of analysis instead of the airline (e.g., Graham, 1997; Reynolds-Feighan, 1995; De Wit et al., 1999; Veldhuis and Kroes, 2002). However, in deregulated markets changes in airport connectivity are primarily the consequence of changes in airline network behaviour. Therefore, we use the airline perspective for the analysis of the spatial dimension of European airline network configurations during the period 1990–1999. Moreover, we introduce the NC-index in order to measure changes in airline network configurations. The NC-index is a modified Gini-index. The Gini-methodology was introduced in air transport analysis by Reynolds-Feighan (1998, 2001) for the measurement of spatial or market concentration in airline networks.
3. Methodology According to Reynolds-Feighan, the hub-and-spoke network requires a concentration of air traffic in both space and time. Hence, the network configuration is defined as the level of spatial and temporal concentration of traffic flows in a given network. The focus here is on the spatial concentration of the airline network. To measure the level of spatial concentration of a network, different concentration measures can be used, such as the coefficient of variance, the Herfindahl-index, Theil’s entropy measure, the C4-firm concentration ratio or the Gini-index. ReynoldsFeighan suggests the Gini-index as the most appropriate concentration measure for airline or airport networks. Allison (1978) and Sen (1976) examined the properties of income inequality measures and proposed a series of criteria that indices should possess. The C4-index only reacts to changes in the traffic distribution in an airport population when the four largest airports are involved. The Herfindahl-index is only sensitive for changes in the extremes of the population. The coefficient of variance on the other hand, reacts well to changes in the population but is extremely sensitive to the underlying distribution. The Gini-index was the only index to satisfy all of the criteria. The Gini-index is not sensitive to the distribution of the population and reacts quite well to changes in all parts of a given population. Regarding spatial concentration, the Gini-index can be defined as G¼
1 XX jyi yj j 2n2 y% i j
ð1Þ
where y is the air traffic at airport i or j; defined as the total number of supplied seats per week; n is the number of airports in the airline network.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
311
Network concentration level
1 0.9
Small radial
Large radial
Small linear
Large linear
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
20
40
60
80
100
120
140
160
Number of airports
Fig. 1. Spatial network configuration types.
If it is assumed that the total incoming air traffic at each airport to be approximately equal to the total outgoing air traffic at the same airport, no airport will command more than half the total air traffic. Therefore, the Gini-index cannot reach its theoretical maximum value of 1. The maximum Gini score increases with the number of airports in an airline network (n) and can be computed as follows:
command equal shares of the total traffic will result in an NC of 0. This includes a multitude of possible configurations, from fully connected networks at one extreme, to collections of isolated routes at the other extreme. An increase in NC indicates a more unequal spread of traffic over the network. This may be caused by *
Gmax
2 ¼1 n
ð2Þ
This maximum Gini-index can be observed in a single hub-network where traffic flows are concentrated on one hub-spoke route. Theoretically, this makes sense since a single hub system is most efficient from the airline viewpoint (Dennis, 2001; O’Kelly, 1998) and the concentration of traffic on one route allows for further advantages of scale. In this study we will correct for the size of the airline network (number of airports) when measuring NC. We define the level of NC as NC ¼
G Gmax
ð3Þ
where G is the observed Gini-index in a network and Gmax the maximum Gini-index given the number of airports in the network. In contrast to the use of the standard Gini-index, our NC-index makes it possible to compare the spatial structure of airline networks independent from network size. NC varies between 0 and 1. A NC of 1 corresponds to a single hub-network where traffic flows are concentrated on one hub-spoke route. Visual inspection of network configurations indicates that NC values of dual or triple hub networks generally range between 0.7 and 0.8. This makes sense, since integrated dual or triple hub-networks require concentration of traffic on the routes between the hubs. A single hub-network with traffic divided equally over all hub-spoke routes has an NC of 0.5. Any network where all airports
*
*
A more unequal distribution of seat capacity over the routes in a single or multiple hub-radial network. This is generally the consequence of route-specific differences in the growth of passenger demand. The removal of routes between smaller airports in the network. The airline concentrates its network on one or a few primary airports. This can be the consequence of hub-and-spoke strategy in order to maximise connecting opportunities. It can also be the consequence of the outsourcing of ‘hub-bypass’ routes to subsidiaries of the airline or regional alliance partners. The transformation of an airline network from a single hub to a multiple hub network.
In order to characterise the (changing) spatial network configuration of a carrier, we distinguish between two dimensions of the airline network. On the one hand the scope of the network using n; the number of airports. On the other hand we use the level of network concentration (NC), independent from n; expressed as NC (3). The level of NC labels a carrier as a radial or linear carrier, with regard to the spatial organisation of its network. Combining n and NC, four ideal network configurations can be distinguished (Fig. 1): (1) the small radial network, (2) the large radial network, (3) the small linear network and (4) the large linear network. To structure the discussion, four ideal dynamics in network configurations are distinguished: *
Concentrated network builders: airlines with both an increase in network size and NC-index.
ARTICLE IN PRESS *
*
*
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
Deconcentrated network rationalisers: Airlines with both a decline in network size and NC-index. Concentrated network rationalisers: Airlines with an increasing spatial concentration of traffic but a shrinking network. Deconcentrated network builders: Airlines with a growing network but a decline in NC.
With the static and dynamic typology of airline networks, insight can be given into the spatial network strategies of airlines in Europe. We will use the static (Fig. 1) and dynamic typologies to describe airline network configurations in Europe for different airline types.
40
30
Number of carriers
312
20
10 Std. Dev = 0.19 Mean = 0.52 N = 162.00
0 0.00
0.13 0.25 0.38 0.50 0.63 0.75 0.88 0.06 0.19 0.31 0.44 0.56 0.69 0.81 0.94
Network concentration for intra-EU traffic
4. Data and selection of carriers 40
*
*
*
*
National airlines: Carriers (formerly) designated as the flag carrier of a member state of the Common European Aviation Area (e.g. KLM, Air France). Low-cost, no-frills airlines: All European-based carriers operating on a low-cost scheduled or mixed (low-cost scheduled/charter) basis (e.g. Ryanair, easyJet). See for a detailed description: Williams (2001). Regional airlines: All other airlines registered in the EU. Our database included 124 regional carriers in 1999 (e.g. Braathens, British Midland). Ten regional airlines were selected to illustrate network developments. These airlines were taken from the top twenty of largest regional carriers in terms of intra-EU seat capacity in 1999. Extra-EU airlines: All airlines with intra-EU services but registered in a non-EU country (e.g. TWA, Cathay Pacific).
5. Analysis of airline network configurations: a first look In order to evaluate the spatial dimension of European airline network configurations, the NC scores and Gini indices of all carriers operating on intra-EU routes were computed. Fig. 2 shows the frequency
30
Number of carriers
The data set used consists of OAG/ABC data for the years 1990–1999. The OAG/ABC data set contains variables based on published information on scheduled flights. Variables include departure airport, destination airport, flight frequency, aircraft type and seat capacity. The data are based on a representative week in July of each year. We refer to Burghouwt and Hakfoort (2001) for a description of the data. In order to analyse the changing network configurations of different types of European carriers as discussed in Section 1, the European airlines were categorised into four groups
20
10 Std. Dev = 0.16 Mean = 0.61 N = 122.00
0 0.00
0.13 0.25 0.38 0.50 0.63 0.75 0.88 0.06 0.19 0.31 0.44 0.56 0.69 0.81 0.94
Network concentration for intra-EU traffic
Source: OAG/ABC data Fig. 2. Frequency distribution of NC values in 1990 (above) and 1999 (below).
distribution of NC both in 1990 and 1999. It can be concluded that between 1990 and 1999 more and more airlines adopted radial networks with high NC levels. Both mean and median moved towards a higher NC rate. Linear networks with low NC values were less frequent in 1999 compared to 1990. Fig. 3 shows the evolution of NC between 1990 and 1999 according to carrier type. The national carrier network configurations have overall higher NCindices than other airline types. This is a plausible outcome. It reflects the national carriers orientation towards their national hubs. Because of the asymmetric deregulation of the European aviation market, carriers are still bounded to their country of registration for their intercontinental services. These intercontinental ‘chains’ have resulted in a stable NC between 1990 and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323 low-cost
regional
national
extra-EU
type of carrier regional
extra-EU
0.9
national
low-cost
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Source: OAG/ABC data Fig. 3. Evolution of NC levels for different airline types, 1990–1999.
Network concentration for intra-EU traffic
1.0
0.8 Network concentration
313
0.8
0.6
0.4
0.2
0.0 0
Extra-EU airlines have the lowest NC scores. This is largely caused by the nature of air services of these carriers within Europe: when they operate intra-EU services, their network is linear, based on fifth freedom rights (Weber and Dinwoodie, 2000).
6. National airlines National airline networks can be characterized as medium or large radial networks (Fig. 4). Differences between the national carriers in terms of concentration indices for intra-EU traffic are small. They varied between 0.63 (Austrian) and 0.8 (Aer Lingus) in 1999. The NC levels of national European carriers have been quite stable between 1990 and 1999 (Table 1). Ten out of sixteen nationals faced an increase in the number of destinations. From a dynamic point of view, four types of national carrier network strategies can be distinguished (Figs. 5–8): 1. Concentrated network builders: Carriers such as British Airways, Finnair and SAS have not only enlarged the scope of their network but also focused this growth on their major hubs and/or a limited number of intra-EU destinations. In the case of SAS, the carrier focused its network growth both in terms of intra-EU capacity and number of routes on its primary airports Copenhagen, Stockholm and Oslo. The three primary airports in the SAS network experienced above average growth levels. The result was a higher NC-index. 2. Deconcentrated network builders: These carriers distributed seat capacity more equally over a larger
20
40
60
80
100
120
number of intra-EU airports
Source: OAG/ABC data Fig. 4. Number of airports and NC levels for different airline types in 1999.
Table 1 Number of airlines, average network concentration level per airline, average Gini scores per airline, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, national airlines, 1990–1999
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Network concentration
Gini index
No. Airports
Capacity %
16 16 16 16 16 16 16 16 16 16
0.744 0.739 0.730 0.730 0.729 0.728 0.728 0.730 0.726 0.729
0.705 0.702 0.694 0.695 0.693 0.693 0.694 0.695 0.691 0.695
44.8 46.4 46.4 46.6 45.9 47.1 48.8 49.1 51.6 53.4
3.1 3.2 3.4 3.5 3.5 3.5 4.2 4.4 5.1 5.5
Source: OAG/ABC; own calculations.
number of airports. The KLM network demonstrates a clear and growing radial network structure focused on Amsterdam Schiphol. Nevertheless, the carrier faced a small decrease in NC values. This is largely caused by the more equal capacity distribution of flow sizes over the routes originating in Amsterdam in 1999 compared to 1990. 3. Concentrated network rationalisers: Some flag carriers limited the number of intra-EU destinations
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
TAP Air Potugal
Swissair
Sabena
SAS
Austrian
Olympic Airways
Lufthansa
KLM
Iberia
Icelandair
Aerlingus
British Airways
Alitalia
Finnair
Luxair
1999
1990
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Air France
Network concentration for intra-EU traffic
314
Fig. 5. NC indices for national airlines, 1990 and 1999. Source: OAG/ABC data
1999
120 100 80 60 40 20 TAP Air Potugal
Swissair
Sabena
SAS
Austrian
Olympic Airways
Lufthansa
Luxair
KLM
Iberia
Icelandair
Aerlingus
British Airways
Alitalia
Finnair
0 Air France
Number of airports served
1990
Fig. 6. Number of EU airports served, national airlines, 1990 and 1999. Source: OAG/ABC data
between 1990 and 1999 but faced an increase in the NC-index. TAP Air Portugal, for example, concentrated the growth of seat capacity on Lisbon and Oporto during the period of analysis. At the same time, the airline rationalised its network by cutting the number of direct destinations from Lisbon, Oporto and Faro. 4. Deconcentrated network rationalisers: A small group of nationals not only faced a decline in network size but also deconcentrated the capacity of its network. Swissair, for example, outsourced a large number of its intra-EU Geneva destinations to its regional daughter Crossair but maintained most of its services
out of Zurich. . The network moved from a dual to a single hub-structure resulting in a decrease in the NC-index (Appendix B).
7. Regional carriers Regional carriers have much smaller networks and lower concentration rates (NC) than the national carriers (Fig. 4, Table 2). In 1999 they accounted for about 26 percent of total intra-EU capacity. The group of regional carriers is diverse: the group ranges from single city-pair carriers to regional carriers operating at
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
a pan-European level. The Swiss airline Crossair for example served more intra-EU destinations in 1999 than most of the flag carriers (Figs. 4 and 8). In 1999, most of the regional carriers had less concentrated networks than the nationals. But also with respect to NC, a large variation in seat capacity distributions can be observed (Figs. 4 and 9): the regional network configurations range from typical linear networks (Wideroe’s, Suckling Airways), radial networks combined with a large amount of point-topoint routes (Skyways, Regional Airlines) to concentrated radial networks (Crossair, Air Liberte! , Braathens). Yet, the radial network dominated the scene in 1999. Only Wideroe’s, Suckling Airways and Oltostfriesische Lufttransport had considerable linear networks in 1999.
315
Regional carrier network configurations have been less stable than those of the national carriers between 1990 and 1999, both with respect to the network size as to the level of concentration (Table 2). While changes of NC of national carriers did not exceed 0.1 points between 1990 and 1999, NC levels of a number of regional carriers have changed dramatically (Fig. 10). Table 4 and Fig. 10 show the different network strategies of a number of regional carriers. Most of the regional airlines concentrated their network in some extent around one or two central hub airports. This can be an independent, mini-hub expansion strategy in the case of the French carrier Regional Airlines at Clermont-Ferrand (Appendix B). However, the carrier had a relatively low NC score because of the large amount of point-to-point routes it still operated in
Change in network concentration, 1990-1999
0.08 0.06 AY
TP
BA
0.04 SK 0.02
FI
0.00
IB
OA
EI
LG AZ
AF
-0.02
SN
-0.04 -0.06
LH
OS
KL
-0.08 SR -0.10 -20
Table 2 Number of airlines, average network concentration level per airline, average Gini scores per airline, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, regional airlines, 1990–1999
-10
0
10
20
30
40
Change in number of airports, 1990-1999
Network concentration
Gini index
No. Airports
Capacity %
124 110 120 116 121 120 116 112 116 124
0.560 0.576 0.581 0.578 0.582 0.569 0.558 0.589 0.563 0.614
0.299 0.284 0.296 0.298 0.306 0.305 0.322 0.322 0.307 0.321
8.3 8.4 8.5 8.5 8.4 8.7 9.4 9.7 9.3 9.3
0.2 0.2 0.2 0.3 0.3 0.3 0.3 0.3 0.3 0.3
Source: OAG/ABC; own calculations.
1990
1999
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 IT IS H BR M AA ID LA TH N EN D S S. A. AI F. R E LI TT O R AL AI R LI BE R TE C R O SS AI EU R R O W IN G S R KL EG M IO U N K AL TY R A IR O W LE LI ID N AN ER ES O AI E' R S W FL AY YV S ES EL SK AP
0
BR
Network concentration for intra-EU traffic
Fig. 7. Change in network size and NC for national carriers, 1990–1999 (see Appendix A for airline codes). Source: OAG/ABC data
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Fig. 8. NC indices for selected regional carriers, 1990 and 1999. Source: OAG/ABC data
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
1990
70 60 50 40 30 20 10 0
1999
BR IT IS H BR M AA ID LA TH N EN D S S. A. AI F. E R LI TT O R AL AI R LI BE R TE C R O SS AI R EU R O W IN G S R KL EG M IO U N K AL TY AI R O R W LE LI ID N AN ER ES AI O E' R W S AY FL YV S ES EL SK AP
Number of airports
316
Fig. 9. Number of airports for selected regional carriers, 1990 and 1999. Source: OAG/ABC data
Table 3 Network strategies of regional airlines
Network growth
Network decline
Concentration
Deconcentration
Concentrated network builders: British Midland, Braathens S.A.F.E., Deutsche BA, Aurigny Air Services, Hahn Air, Air Libert!e, Augsburg Airways, Jersey European Airways, Skyways, Air Botnia, Crossair, Eurowings, Olt Ostfriesische Lufttransport, KLM UK, Regional Airlines, Wideroe’s, Aero Lloyd, Monarch Airlines, Muk Air Concentrated network rationalizers: Air Littoral, Manx Airlines
Deconcentrated network builders: Suckling Airways, Maersk Air, Sata Air Acores, Tyrolean Airways
Deconcentrated network rationalizers: Air Sicilia
Source: OAG/ABC; own calculations.
0.4 YQ
change in NC, 1990-1999
0.3
0.2
UK IJ JY KF
DI FU
-0.0
JZ
IQ
0.1
BU
ZB
BD
LX YP
WF JE ZR OL
BM
VM
EW
VO
DM
SP IG
-0.1
-0.2
CB
-0.3 -20
-10
0
10
20
30
40
change in number of airports, 1990-1999 Fig. 10. Change in network size and NC for regional carriers having scheduled service both in 1990 and 1999, 1990–1999 (see Appendix A for airline codes). Source: OAG/ABC data
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
1999. It can also be the consequence of alliances between national and regional carriers, in which the regional carriers play a hub-feeding role (Graham, 1997). For example, Eurowings and Air UK centered their network on KLM’s hub Amsterdam Schiphol between 1990 and 1999. Even the Norwegian carrier Wideroe’s-probably the best example of (PSO subsidized) linear carrier in Europe-focused part of its network around Oslo during the period of analysis (Appendix B). Regional linear strategies among regional carriers are rare in Europe. It underlines the conclusion of various authors that spatial concentration of traffic at a limited number of airports has cost advantages, even without the transfer of passengers from one flight to another. These cost advantages include the spread of fixed costs over more passengers as well as easy crew rotation and aircraft maintenance.
Table 4 Number of airlines, average network concentration level per airline, average Gini scores, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, low-cost airlines, 1990–1999 Network concentration
Gini index
No. Airports
Capacity %
2 2 3 3 5 6 8 9 9 9
0.654 0.742 0.763 0.657 0.660 0.611 0.667 0.630 0.647 0.617
0.280 0.389 0.566 0.498 0.524 0.484 0.518 0.543 0.561 0.551
8.0 6.5 8.5 9.3 15.2 13.7 13.5 19.9 21.3 24.1
0.2 0.2 0.3 0.2 0.3 0.6 0.6 1.0 1.1 1.3
8. Low-cost carriers The entrance of low-cost carriers to the European aviation market was one of the most profound effects of deregulation. In the US, a number of low-cost carriers, such as Southwest, operate linear networks with low levels of NC. In Europe, the low-cost carrier networks seem to be as concentrated as the networks of national carriers (Table 4, Figs. 3, 11 and 13). All the low-cost carriers operated out of central airports. Ryanair, for example, used Dublin and Stansted as its central airports in 1999. The only exception is Virgin Express (former Eurobelgian). Besides its Brussels-network, Virgin Express also operated services between Madrid–Barcelona–Rome and Shannon–Stansted in 1999. The difference in the network configurations between low-cost and national carriers can mainly be found in network size (Fig. 12). This is because low-cost carriers have been operating air services only for a short period. Therefore, it makes little sense to categorise the carriers according to their network strategies. However, the described networks of the low-cost carriers are about to change. The low-cost carriers experienced fast growth after 1999. Low-cost carriers did not suffer as much from the crisis in the air transport industry after September 11th because of the low fare levels and the fact that low-cost carriers transport relatively few passengers from the American continent
base year
AIR EUROPA
EASYJET
SPANAIR
HAPAG LLOYD FLUGGESSELSCHAF
RYANAIR
CONDOR FLUGDIENST
1999
VIRGIN EXPRESS
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
DEBONAIR AIRWAYS
Network concentration level
Source: OAG/ABC; own calculations.
From Fig. 10 it can be concluded that the only carrier with a significant, regional, linear strategy is Suckling Airways. In 1990 the carrier operated a Cambridge– Manchester and Cambridge–Amsterdam route. In 1999 the airline had services to 11 destinations but did not operate out of a central airport in the UK.
TRANSAVIA AIRLINES
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
317
Source: OAG/ABC data. Base year for low-cost carriers: Debonair: 1996; Virgin: 1995; Condor: 1993; Ryanair: 1990; Hapag Lloyd: 1997; Transavia: 1991; Spanair: 1994; easyJet: 1996; Air Europa: 1994 Fig. 11. NC indices for low-cost carriers, base year and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323 50 45 40 35 30 25 20 15 10 5 0
base year
AIR EUROPA
EASYJET
SPANAIR
TRANSAVIA AIRLINES
HAPAG LLOYD FLUGGESSELSCHAF
RYANAIR
CONDOR FLUGDIENST
VIRGIN EXPRESS
1999
DEBONAIR AIRWAYS
Number of intra-EU airports
318
Source: OAG/ABC data. Base year for low-cost carriers: Debonair: 1996; Virgin: 1995; Condor: 1993; Ryanair: 1990; Hapag Lloyd: 1997; Transavia: 1991; Spanair: 1994; easyJet: 1996; Air Europa: 1994 Fig. 12. Number of intra-EU airports served by low-cost carriers in base year and 1999.
Network concentration level
0.9 0.8 0.7 0.6 0.5 0.4 0.3 1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
JK
U2
UX
Year 2G
BQ/TV
DE
FR
HF
HV
Source: OAG/ABC data Fig. 13. Change NC for low-cost carriers 1990–1999 (see Appendix A for airline codes).
compared to the national airlines. New low-cost subsidiaries have entered the market (Buzz, Go). British low-cost carriers have expanded their services on the European continent (Ryanair, easyJet). Ryanair, for example, opened new base airports at Frankfurt Hahn and Brussels Charleroi. In March 2003, Ryanair served about 65 destinations in contrast to 33 in 1999. Based on the flight schedule available on the website of Ryanair, we have computed the NC-index for Ryanair in 2003. Ryanair had an NC value of 0.7 in March 2003 (against 0.68 in 1999). Hence, the network of Ryanair still centred on multiple central airports (London Stansted, Dublin, Brussels Charleroi, Frankfurt Hahn, Glasgow Prestwick, Stockholm Skavsta and Shannon). Ryanair does not have a criss-cross or linear network such as the network of low-cost carrier South-
west in the US with a NC value of 0.47 in 1999. However, further expansion of Ryanair at the European continent and the increase in the number of home bases may eventually result in a lower NC value for the network.
9. Extra-EU airline networks For a long time, the use of 5th freedom rights by international airlines was the only possibility to create world-embracing network. The 5th freedom is ‘the right of an airline of one country to carry traffic between countries outside of its own country of registry as long as the flights originates or terminates in its own country of registry’ (Button and Taylor, 2000). This traffic right
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
was defined at the Chicago Conference of 1944 and could be agreed upon in air service agreements between states. It enables international airlines to operate routes that would not be economically viable as non-stop destination. Overall, the 5th freedom networks of extra-EU airlines have a linear character with very low NC values compared to the national and regional airline networks. However, a significant decline of NC values together with a decline in capacity share/number of airports served took place during the period of analysis. This development is in line with the study of Weber and Dinwoodie (2000) on 5th freedom traffic operations in the Single European Aviation market. They state that, in the early 1990s, 5th freedom traffic was the only opportunity for extra-EU airlines to develop a European network with destinations that would not be
Table 5 Number of airlines, average network concentration level per airline, average Gini scores, average number of airports per airline and average share (%) in total intra-EU seat capacity per airline, extra-EU airlines, 1990–1999
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Number of airlines
Network concentration
Gini index
No. Airports
Capacity %
72 66 67 68 37 30 41 34 31 34
0.362 0.349 0.301 0.337 0.403 0.423 0.353 0.435 0.373 0.302
0.159 0.136 0.123 0.131 0.108 0.118 0.079 0.078 0.064 0.034
4.8 4.5 4.6 4.5 3.6 3.9 2.9 2.7 2.5 2.2
0.08 0.06 0.06 0.06 0.05 0.10 0.03 0.02 0.02 0.02
viable as non-stop routes. Some airlines, such as Delta (the airline acquired the extensive 5th freedom network of Pan Am), TWA and United Airlines set up mini-hub operations within Europe based on 5th freedom rights. These kind of operations resulted in relatively high NC values in 1990 compared to 1999. Weber and Dinwoodie (2000) identify a number of factors that limited the 5th freedom traffic in Europe during the period of analysis: *
*
*
*
Source: OAG/ABC; own calculations.
1990
0.9
1995
Global airline alliances: Alliances form a substitute or even a refinement of 5th freedom traffic: European partner airlines can offer better connectivity to more destinations. In 1991, for example, the US carrier United Airlines, used London Heathrow as its European hub, connecting its intercontinental services with smaller Boeing 727s to Brussels, Amsterdam and Paris Charles de Gaulle. Gradually, the carrier shifted its European gateway to Frankfurt, anticipating to the alliance-agreement with Lufthansa in 1993. Now, the European partners of the Staralliance (e.g. Lufthansa, SAS) provide the intra-EU services. Technological developments: Introduction of the Airbus A340 and Boeing 777 has enabled airlines to operate economically thin long-haul routes incapable of supporting Boeing 747–400 s. Product life cycle of international routes: A number of 5th freedom routes has reached maturity and have sufficient demand to justify non-stop service. Competition: Competition of European carriers with higher corporate strength in the markets served makes it less attractive to operate 5th freedom routes. The implementation of the ‘Third Package’ in 1992 resulted in a significant decrease of 5th freedom operations of extra-EU carriers (Table 5).
1999
Network concentration level
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 AC
CX
DL
GA
KU
319
PA
QF
SQ
TG
TW
UA
Source OAG/ABC data Fig. 14. NC values for selected extra-EU airlines (see Appendix A for carrier codes).
ARTICLE IN PRESS 320
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
These developments resulted in a decline in market share of extra-EU airlines and lower NC values. A number of US carriers and Asian carriers gave up most of their fifth freedom operations (Fig. 14).
10. Conclusion Both the spatial concentration of seat capacity on a limited number of hubs and the temporal concentration of flights around a number of synchronized waves of flights per day, are essential features of a hub-and-spoke system. This paper has presented an overview of the changes in the spatial concentration of airline networks in Europe during the period 1990–1999. Spatially concentrated networks possess a radial route structure whereas deconcentrated networks possess a linear or criss-cross route structure. The NC-index, derived from the Gini-index, has been used to measure the spatial distribution of traffic flows in airline networks. Combined with the number of airports served, the NC-index makes it easy to detect changes in the structure of airline networks. We have applied the NC to answer the question to what extent airlines in Europe operate radial or linear networks and how these networks have developed over time. Taken together, the networks of European airlines are increasingly concentrated. Radial networks have been developed between 1990 and 1999. Yet, large differences exist between the various airline types. The national carrier networks can be characterised as large radial networks. The ‘flag’ carriers are still bounded to their national airports because of bilateral regulation of intercontinental services. This implies the operation of radial network structures. Moreover, large networks are complex networks and have many dimensions to absorb new developments (Swan, 2002). Therefore, changes in the NC level of national carriers have been quite small. Radical geographical restructuring of networks of major carriers, as in the US after deregulation, cannot be observed in Europe. Nevertheless, at a smaller scale, radial strategies can be observed among regional carriers. Taking advantage of the liberalised market, some regional airlines restructured their networks from linear into radial networks. Others started from a small traffic base at a central airport and, due to alliances with national carriers, benefited from traffic feed of large intercontinental carriers. Some of the regional carriers, such as Crossair, had even larger European networks than most of the national airlines. Large linear networks were an exceptional phenomenon in Europe at the end of the nineties. Regional airlines such as Suckling Airways, the PSO subsidized Norwegian carrier Wideroe’s and the German Oltfrie-
sische Lufttransport operated some kind of linear network in 1999. Only Suckling Airways intensified its linear network structure. In contrast to the US, European low-cost carriers focused their service on a limited number of airports until 1999. Yet, signs of change are already there: very fast growth, inauguration of new routes, entrance of British low-cost carriers at the continent and use of secondary airports might be the first signs of linear/crisscross networks of low-cost carriers. The evolution of the intra-European aviation network can be characterised by both a spatial deconcentration trend at the airport level and a spatial concentration trend at the airline network level. Intra-EU traffic is spreading over more and more airports (Burghouwt and Hakfoort, 2001). Although we cannot draw any firm conclusions about the factors decisive for the network developments observed based on our data, deregulation seems to have given regional airlines and airports the opportunity to play an important role in the European aviation scene. Economic growth and lower prices generated sufficient demand for new services, not only from primary but also from smaller airports. At the same time, a number of regional carriers concentrated their networks around one or two central airports during the period of analysis. National carrier networks were already focused on key nodes at the beginning of deregulation. In the planning process of these central airports, airport authorities and governments will have to cope with periods of very fast traffic growth at hub airports. This may result in increasing noise pollution in airport regions, peaking problems and new terminal layout requirements. Airports forming part of the network of hubbing carriers, will be more and more dependent of network decisions of these airlines and their alliances for their future network position.
Appendix A. Carrier codes National carriers AF AIR FRANCE AY FINNAIR AZ ALITALIA BA BRITISH AIRWAYS EI AER LINGUS FI ICELANDAIR IB IBERIA KL KLM-ROYAL DUTCH AIRLINES LG LUXAIR LH LUFTHANSA GERMAN AIRLINES OA OLYMPIC AIRWAYS OS AUSTRIAN AIRLINES SK SAS SCANDINAVIAN AIRLINES
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
SN SR TP
SABENA SWISSAIR TAP AIR PORTUGAL
Low cost carriers 2G DEBONAIR AIRWAYS TV VIRGIN EXPRESS DE CONDOR FLUGDIENST FR RYANAIR HF HAPAG LLOYD FLUGGESSELSCHAF HV TRANSAVIA AIRLINES JK SPANAIR U2 EASYJET UX AIR EUROPA Regional carriers BD BRITISH MIDLAND BM AIR SICILIA BU BRAATHENS S.A.F.E CB SUCKLING AIRWAYS DM MAERSK AIR EW EUROWINGS FU AIR LITTORAL IG MERIDIANA IJ AIR LIBERTE! IQ AUGSBURG AIRWAYS JE MANX AIRLINES JY JERSEY EUROPEAN AIRWAYS JZ SKYWAYS
KF SP UK VM VO WF YP YQ ZB ZR Extra-EU AC CX DL GA KU PA QF SQ TG TW UA
321
AIR BOTNIA SATA AIR ACORES KLM UK REGIONAL AIRLINES TYROLEAN AIRWAYS WIDEROE’S FLYVESELSKAP AERO LLOYD HELIKOPTERSERVICE EURO AIR MONARCH AIRLINES MUK AIR carriers AIR CANADA CATHAY PACIFIC AIRWAYS DELTA AIR LINES GARUDA INDONESIA KUWAIT AIRWAYS PAN AMERICAN WORLD AIRWAYS QANTAS AIRWAYS SINGAPORE AIRLINES THAI AIRWAYS INTERNATIONAL TRANS WORLD AIRLINES UNITED AIRLINES
Appendix B Mini-hub expansion strategy in the case of the French carrier Regional Airlines at Clermont-Ferrand:
ARTICLE IN PRESS 322
G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
Development of the network of Swissair between 1990 and 1999.
ARTICLE IN PRESS G. Burghouwt et al. / Journal of Air Transport Management 9 (2003) 309–323
References Allison, P.D., 1978. Measures of inequality. American Sociological Review 43, 865–880. Bootsma, P.D., 1997. Airline Flight Schedule Development; Analysis and Design Tools for European Hinterland Hubs. University of Twente, Utrecht. Brueckner, J.K., Spiller, P.T., 1994. Economies of traffic density in the deregulated airline industry. Journal of Law and Economics 37, 379–415. Burghouwt, G., Hakfoort, J.R., 2001. The European aviation network, 1990–1998. Journal of Air Transport Management 7, 311–318. Button, K.J., Taylor, S.Y., 2000. International air transportation and economic development. Journal of Air Transport Management 6, 209–222. Button, K.J., Haynes, K., Stough, R., 1998. Flying into the Future. Air Transport Policy in the European Union. Edward Elgar, Cheltenham. Caves, R.E., Gosling, G.D., 1999. Strategic Airport Planning. Pergamon, Amsterdam. Dempsey, P.S., Gesell, L.E., 1997. Airline Management. Strategies for the 21st century. Coast Aire Publications, Chandler. Dennis, N.P., 2001. Developments of hubbing at European airports. Air and Space Europe 3, 51–55. de Wit, J., Uittenbogaart, P., Wei-Yun, T., 1999. Hubbing and hubbypassing. Network developments in a deregulated European airport system. ATRG Conference, Hong Kong. Graham, B., 1995. Geography and Air Transport. Wiley, Chichester. Graham, B., 1997. Regional airline services in the liberalized European Union single aviation market. Journal of Air Transport Management 3, 227–238.
323
Hakfoort, J.R., 1999. The deregulation of European air transport: a dream come true? TESG 90, 226–233. O’Kelly, M.E., 1998. A geographer’s analysis of hub-and-spoke networks. Journal of Transport Geography 6, 171–186. Pels, E., 2001. A note on airline alliances. Journal of Air Transport Management 7, 3–7. Reynolds-Feighan, A., 1995. European and American approaches to air transport liberalisation: some implications for small communties. Transportation Research-A 29, 467–483. Reynolds-Feighan, A., 1998. The impact of US airline deregulation on airport traffic patterns. Geographical Analysis 30, 234–253. Reynolds-Feighan, A., 2000. The US airport hierarchy and implications for small communities. Urban Studies 37, 557–577. Reynolds-Feighan, A., 2001. Traffic distribution in low-cost and fullservice carrier networks in the US air transport market. Journal of Air Transport Management 7, 265–275. Sen, A., 1976. Poverty: an ordinal approach to measurement. Econometrica 44, 219–231. Swan, W.M., 2002. Airline route developments: a review of history. Journal of Air Transport Management 8, 349–353. Veldhuis, J., Kroes, E., 2002. Dynamics in Relative Network Performance of the Main European Hub Airports. European Transport Conference, Cambridge. Viscusi, W.K., Vernon, J.M., Harrington, J.E., 1998. Economics of Regulation and Antithrust. MIT Press, Cambridge. Weber, M., Dinwoodie, J., 2000. Fifth freedoms and airline alliances. The role of fifth freedom traffic in an understanding of airline alliances. Journal of Air Transport Management 6, 51–60. Williams, G., 2001. Will Europe’s charter carriers be replaced by ‘‘nofrills’’ scheduled airlines? Journal of Air Transport Management 7, 277–286.
Related Documents
The Spatial Configuration Of Airline Networks In Europe
November 2019 1
38 Huber Spatial Concentration In Europe Versus The Us
December 2019 3
The Globe Of Europe
October 2019 62
The Tragedy Of Europe
June 2020 42
The Wealth Of Networks
August 2019 45