Title Bypassing the hubs - The potential of secondary European airports in the long haul sector

Paper prepared for the "7th Conference on Applied Infrastructure Research" , October 10-11, 2008,

Sven Maertens Remscheider Str. 107 51103 Cologne / Current occupation: Researcher at the DLR-German Aerospace Establishment Cologne/Germany Tel.: +49 (0)2203 - 6012596 [email protected]

The study this paper refers to was mainly conducted at the Institute for Transport Economics (IVM), University of Muenster, where Sven Maertens worked until April 2008.

Abstract Decentralised, i.e. non-hub long haul air traffic is unevenly distributed between secondary airports in Europe. There are a few airports, such as London-Gatwick, Manchester or Düsseldorf, that offer various long haul flights every day. Most secondary airports, however, hardly ever welcome those services, even though many of them provide infrastructure dedicated to widebody aircraft. In these cases, the provision of such “long haul - specific” infrastructure is not economically efficient. This paper presents the main results of a research project in which we discuss and test factors that might influence secondary airport choice of long haul carriers. Our sample consists of timetable and catchment data for all 224 secondary airports in Europe. OLS regression results indicate that GDP is the main driver of the supply of long haul flight supply at secondary airports, while the proximity of a larger airport has a negative impact. Airports in the UK tend to easier attract long haul services, possi- bly caused by the high number of immigrants from overseas living in the UK and by strong ethnic and economic relations with the USA. A sufficient length is a necessary, but not a sufficient condi- tion for long haul services.

Keywords: airports, airlines, networks, hubs, long haul JEL codes: L93, R41, R42 2

1 INTRODUCTION

Long haul air traffic from and to Europe is concentrated on a couple of large European hubs with a high rate of transfer traffic. In contrast, most secondary airports struggle to attract a noteworthy number of long haul air services, although many of them provide adequate infrastructure for widebody aircraft. In these cases, the provision of infrastructure dedicated for long haul flights is not economically efficient. Hereinafter, potential factors influencing airport choice for long haul flights from secondary airports in Europe are discussed and empirically tested.

In a first step, an overview of the current distribution of weekly long haul flight supply between secondary airports in Europe is given. It shows that only a dozen out of 224 secondary airports represent about 80% of all decentralised long haul air traffic from and to Europe. Factors that could potentially influence the supply of long haul flights at secondary airports are derived from the theories of airline network design and airline business models. Some factors have a largely or at least partly endogenous character, such as airport infrastructure and airport marketing. Other factors, however, are exogenous, like regional GDP and industry structure, an airport’s proximity to a larger airport and the political importance of the airport region. Multiple OLS regression analysis and multiple logistic regression analysis are applied to test the relative importance of these factors. The results indicate that

GDP is the main driver of the supply of long haul flights at secondary airports, while the proximity of a larger airport has a negative impact. Airports in the UK tend to easier attract long haul services, possibly caused by relatively high numbers of citizens from overseas and by strong ethnic and economic relations with the USA. A sufficient runway length seems to be a necessary, but not a sufficient condition for long haul services. Long haul all-cargo services mainly depend on runway length, airport location and GDP in the 90 minute catchment area.

While there has been much research on airport choice by hub carriers, as well as on general factors affecting air travel demand, long haul air traffic bypassing the hubs has been widely neglected. For this reason, the results of this paper might be of relevance for airport planning and management and could help reduce future resource misallocation related to airport infrastructure. The topic of this paper

2 3 might also be of relevance in the ongoing discussion about the impact of new medium sized long haul aircraft, such as B787 and A350, on future air network structure.

2 DEFINITIONS

As there is no unique definition, in this paper, the terms “long haul” and “intercontinental” flights refer to flights from Europe to all overseas destinations except for countries around the Mediterranean.

Table 1: Longhaul destinations from Europe Region Countries all countries except Northern Africa (Algeria, Egypt, Libya, Marocco, Africa Tunisia) Asia / all countries except those located around the Mediterranean (Asian part of Turkey, Australia Israel, Lebanon, Syria) and Jordan, but incl. East Uralian Russia America all countries Source: Own production.

There also is no unique definition for secondary airports.1 To exclude thousands of small airfields from the sample, all European airports that are equipped with ILS facilities and with one or more runways of more than 2.000m length are referred to as secondary airports, as long as they are open for commercial flights. Exceptions are Amsterdam, Paris CDG, Frankfurt, London LHR, Madrid, ,

Rome FCO, Vienna and Zurich because these airports fulfil at least 3 out of 4 hub criteria shown in table 2 and thus can be clearly allocated to the group of major hubs.2 Hubs that focus completely on intra-European traffic, however, such as Lyon or Palma de Mallorca, are assigned to the group of secondary airports. As a result, our sample includes 224 and thus virtually all secondary airports in

Europe (as shown in appendix 1).

Table 2: Hub criteria Transfer rate > 20% Wave structure of the hub&spoke operation Passenger numbers > 20 Mio. Main airport of the national carrier Source: Own production.

1 Many different approaches of airport classifications have been reviewed: Most institutions such as Airports Council International (ACI), ICAO or the European Union simply classify airports according to their output, that means passenger and cargo figures, or to technical characteristics and thus ignore the variety of flights and the passenger mix. (ACI, 2005, Airports Authority of India, 2005, EU Commission, 2005). 2 For Burghouwt/de Wit (2005), a hub is characterized by a high number of connections and by flights operated in waves. Roth (2002) refers to all those airports as hubs that are dominated by one carrier.

3 4

3 STATUS QUO: LONG HAUL FLIGHT DISTRIBUTION BETWEEN EUROPEAN AIRPORTS, TYPOLOGIES OF LONG HAUL AIRLINES AND INFRASTRUCTURE MISALLOCATION Nowadays, more than 90% of all passengers leaving Germany on a direct long haul flight depart from

Lufthansa’s hubs Frankfurt and Munich.3 In the rest of Europe, the situation is similar: All major

European carriers concentrate their long haul operations on their respective main hub and hardly operate any long haul services from secondary airports.4 Figure 1 illustrates the distribution of those long haul services that do take off from secondary airports and thus bypass the European hubs. Again, there is no equal distribution between airports, but a massive concentration on a few secondary airports led by London Gatwick, Manchester, Milan, Paris Orly and Brussels. Other airports that han- dle a few regular long haul services are either secondary and third class airports in large European countries – such as Cologne/Bonn, Berlin and Hamburg in Germany, Nice and Marseille in France or

Edinburgh and London Stansted in the UK – or primary airports in smaller Eastern European coun- tries, such as Prague or Budapest.

London LGW Others 18% 22%

Birmingham 2% Manchester 10% Glasgow 2%

Stockholm 3% Athens 3% Mailand MXP Dusseldorf 9% 3% Helsinki 4% Paris ORY Copenhagen Dublin 6% Brussels 4% 4% Lissabon 5% 5%

Figure 1: Long haul services from secondary airports in Europe (2007) Source: Own calculation based on total weekly MTOW supplied on long haul flights in October 2007. Most regular charter flights have been included manually.

3 Source: Own calculations based on data provided by the German Federal Statistical Office. 4 The main exception is British Airways which operates a decreasing but still considerable number of long haul services from Gatwick, assumingly caused by capacity constraints at Heathrow. Air France has shifted most long haul flights to Charles de Gaulle airport, with the exception of a handful of leisure services to the French Overseas Territories. Lufthansa, finally, has recently (in May 2008) based three Airbus 340 aircraft in Düsseldorf, operating to North American destinations.

4 5

Carriers operating these decentralised long haul flights can be divided into the following groups. Some

European flag / Full Service Network carriers (FSNC) operate from secondary airports to globally important destinations such as New York or Chicago (LOT from Warsaw, Malev from Budapest, CSA from Prague, Aer Lingus from Dublin, Lufthansa from Düsseldorf…). Hub carriers from outside of

Europe connect their respective hubs in overseas with second- and third-tier cities in Europe. The main carriers to be cited in this context are Continental Airlines, Delta Airlines and Emirates which operate from their hubs in the US and Dubai respectively to the likes of Manchester, Barcelona or

Düsseldorf, and to even smaller airports like Hamburg or Glasgow (table 3). As these flights “feed” the respective hub in overseas, they are hereafter referred to as “long haul feeder flights”. Leisure carriers fly from Europe to typical overseas holiday destinations. While in Germany, long haul holiday flights concentrate on the largest secondary airports like Düsseldorf, virtually every UK airport handles at least a few of such services every week. Typical airlines are (former) charter carriers, such as Condor, LTU (both from Germany), Corsair (France), Excel Airways, Monarch and

Thomsonfly (all British). From abroad, Canadian carriers Air Transat and Zoom serve various

European destinations, mainly located in France, Germany and the UK (Zoom, however, ceased operations in summer 2008). Ethnic carriers offer so-called ethnic flights from airports in major conurbations with high numbers of immigrants to the respective home countries, focusing on VFR

(Visiting Friends and Relatives) traffic. Prominent examples are PIA from Manchester and

Birmingham to various cities in Pakistan and Mahan Air and Iran Air from Western Europe to Tehran.

Last but not least, all-business class airlines like Privatair (Lufthansa), Silverjet, EOS, Maxjet,

Openskies and L’Avion offer(ed) non-stop, all-premium class services from secondary airports in

Europe to destinations in the USA.5

5 Since October 2007, EOS, Maxjet and Silverjet have all ceased operations, and Lufthansa has changed the equipment on the DUS-EWR and DUS-ORD routes from Boeing BBJ to 3-class-widebody aircraft. However, L’Avion and Openskies are new airlines operating all-business class Boeing 757 aircraft between Orly and New York.

5 6

Table 3: Western European hub and non-hub destinations served by Continental Airlines and Emirates (2007) Continental Airlines Emirates Amsterdam Geneva Birmingham Nice Athens Glasgow Dusseldorf Paris CDG Barcelona Hamburg Frankfurt Rome Belfast Lisbon Glasgow Venice Berlin London LGW Hamburg Vienna Birmingham Madrid Larnaca Zurich Bristol Manchester London LGW Brussels Milan London LHR Cologne/Bonn* Paris CDG Malta Copenhagen Rome Manchester Dublin Shannon Milan Edinburgh Stockholm Munich Frankfurt Zurich Newcastle Non-hub destinations are in bold type. *) until 5 September 2008 Source: Official timetable information from December 2007.

Based on this, long haul flights supply can be divided in the following groups: • Scheduled flights by national (i.e. European) legacy carriers [LEGACY] • Long Haul Feeder flights operated by network carriers from overseas [LHFEEDER] • Leisure and Ethnic flights [L&E] • All-Business-Class flights [ALLBIZ].

Table 4 shows the importance of these market segments for different size classes of secondary airports in Europe, measured by the number of long haul services per week. Large secondary airports with several daily long haul services, such as Manchester, Dusseldorf, Brussels, Milan, London-Gatwick or

Copenhagen, show a more uniform distribution of the respective groups than airports handling only a few, mostly leisure, long haul services per week.

Table 4: Importance of different market segments for different size groups of secondary airports in Europe Segment Classes of secondary airports according to the total number of weekly long haul flights < 7 7-50 > 50 n = 53 n = 27 n = 13 LEGACY 9% 18% 30% L&E 80% 27% 35% LHFEEDER 11% 51% 34% ALLBIZ 0% 4% 1% Source: Own calculation based on flights operated in October 2007, adjusted to consider seasonal differences.

6 7

This long haul flight distribution leads to resource misallocation of airport infrastructure dedicated for long haul flights. While 2.000m can be regarded as a minimum for any long haul operation, a runway length of more than 2.700m is usually not necessary for any short and medium haul flight. Thus, runway sections exceeding a total length of 2.700m can be defined as “long haul - specific infrastructure”. Currently, 115 secondary European airports are equipped with such infrastructure.

Even assuming a very moderate construction price of 400 €/m², an asset depreciation range of 30 years, an interest rate of 5% and unrealistically high long haul specific per seat revenues of 20 €, only

41 of these 115 airports would generate enough revenues from long haul services to independently finance the infrastructure provided exclusively for these services. This – admittedly – very simple calculation gives a rough idea of the dimension of long haul runway misallocation.

4 DETERMINANTS OF LONG HAUL FLIGHT SUPPLY AT SECONDARY AIRPORTS 4.1 Consistency of hub bypass long haul flights with airline business models As most airlines flying from and into Europe are private, profit maximising firms, we assume profit maximising route networks to be chosen. Thus, we try to explain the observed choice of secondary airports in deducting possible factors of airport choice from the literature on airline network strategies and airline business models. Airport choice has been widely discussed in the literature (for a recent article and further references see Blackstone et al., 2006), but there has hardly ever been any particular focus on decentralised long haul flights. In the following, we will briefly discuss in how far long haul operations to secondary airports are consistent with the business models of the typical groups of airlines. Based on this, potential factors on the supply of long haul flights at secondary airports will be compiled, discussed and empirically tested. A distinction between internal, semi-external and external factors is made to account for the degree in how far airport management can independently control the respective factors.

As shown, apart from the small group of all business carriers, the main types of passenger long haul airlines are “legacy carriers” and “leisure or ethnic carriers”. Legacy carriers rely on the hub and spoke model because it offers advantages both on the cost (economies of scale, scope and density, see

7 8

Caves et al., 1984, Hansen/Kanafani, 1989, Kanafani/Ghobrial, 1982) and demand side. From the consumer perspective, the latter aspect is especially important: In offering connections at the hub, the hub airline supplies a large variety of O&D connections to choose from, including niche markets which could never be served directly.6 Every additional flight out of a carrier’s hub yields in a multiple of new O&Ds and additional economies of density. Consequently, it usually makes not much sense for a hub carrier to start decentralised flights from secondary airports. Legacy carriers from other continents, however, are more likely to serve secondary airports in Europe, as they have their hub at the other end of the route. An example is Delta Airlines connecting Düsseldorf with its Atlanta hub, where onward connections to all over North and South America are offered.

Leisure (and ethnic) carriers have always been operating long haul flights, selling the bulk of their capacities to tour operators. They achieve low unit costs in focusing on direct point to point flights using aircraft with high-density seating. As holidaymakers usually stay one week or longer at their destination, a temporal concentration of demand to each destination on few flights per week is usually accepted. Furthermore, most leisure travellers are prepared to cover longer distances to their departure airport, allowing the airlines to spatially concentrate passenger flows on flights from few departure airports. The same applies to ethnic carriers (Brons et al., 2002, Graham, 1995, Windle/Dresner,

1995). However, the range of economically viable long haul destinations suitable for charter and ethnic carriers is rather limited.

Low cost carriers (LCC), finally, achieve very low costs per seat mile in concentrating on simple point to point flights and in eliminating frills. They mainly offer short haul routes between relatively small and uncongested airports allowing for immediate turnarounds. Thereby, LCC manage to extend fleet utilization, leading to lower capital costs per seat mile. While some experts claim that the low cost carrier business model could well be transferred to the long haul market, we question this because some key elements of the strategy of LCC either cannot be implemented on long haul routes (higher

6 For a detailed analysis of the function of hub-and-spoke networks, see Bailey et al. (1986), Hansen/Kanafani

8 9 daily fleet utilisation, lower personnel costs in avoiding hotel accommodation), or they are of relatively less importance as the cost structure in the long haul market is different. Discounted airport fees and the abandonment of amenities such as lounges, for instance, have a much smaller impact on total cost of long haul flights compared to fuel, personnel and capital costs (Ionides, 2006).

4.2 Factors influencing long haul flight supply 4.2.1 Internal factor: Airport marketing In times of increasing deregulation and competition in the aviation market, most airports have intensified their marketing activities. However, airports usually cannot independently control all 4 P’s of their marketing mix, namely product, price, promotion and place. Instead, crucial decisions on product characteristics such as runway length, terminal size or curfew are usually subject to governmental approval. The size and wide of the catchment area - dimension “place” in the marketing mix – mainly depend on rail and road access and on the proximity of competing airports and thus are also exogenous. While promotion is supposed to have no or only marginal impact on the attraction of long haul services, pricing is the only dimension in an airport’s marketing mix which could have a major impact on route acquisition. The success of smaller airports in attracting low cost carriers by offering reduced airport fees shows how price sensitive some airlines are.7 On long haul routes, however, airport charges represent a relatively small part of total operating costs, making it unlikely that low airport fees, volume based discounts and marketing support alone would help attract new long haul services. 8

4.2.2 Semi-external factors Long haul aircraft require a better airport infrastructure than short- and medium-haul aircraft. The minimum runway length an aircraft requires depends on various factors such as take off weight, humidity, altitude above sea level, weather and pavement surface. A runway length of at least 3.600 m is generally sufficient for all long haul aircraft and MTOW’s. On shorter runways, pilots might be

(1989) and Hanlon (1996). 7 Barrett (2004), p. 36, BCG (2004), S. 22, Gillen/Morrison (2005), pp. 164-165, Morrison/Mason (2006), pp. 6-10 and Gillen/Lall (2004), pp. 47-48 show the importance of low airport-related costs for low-cost-carriers.

9 10 forced to depart with payload restrictions, reducing profitability.9 Since extensions of airport capacities are usually subject to governmental approval, airport infrastructure can be regarded as a semi-external factor.10 For the same reason, operating restrictions represent another semi-external factor. Bans of night flights, for example, negatively damage an airport’s attractiveness especially in the eyes of cargo airlines, integrators and forwarders.11 Limits on movements, such as in Düsseldorf, can cause or worsen slot shortage and hence reduce an airport’s attractiveness if slots for long haul flights and feeder services are no longer available at attractive times.12 Aircraft size restrictions, finally, as practiced in Düsseldorf-, can effectively inhibit any long haul services.13 Since these kinds of operating restrictions are usually imposed by governmental institutions or court decisions, airport management can only indirectly influence them through lobbying efforts. A good connectivity to other modes of transport (i.e. motorways and long distance trains), finally, enlarges an airport’s catchment area and reduces passenger access costs. Thus, improved airport land access might enhance an airport’s attractiveness for long haul carriers.14

8 See Knibb (1993). Graham (2003) provides a good overview of the structure of aeronautical charges and explains how discounts are used to attract new services. 9 Apart from a runway’s length, its strength is important because widebody aircraft generally require a more concrete runway than smaller aircraft. A runway’s stability is usually denoted by its pavement classification number (PCN). A PCN of 60 can be regarded as an absolute minimum for long haul aircraft (Malina 2005). 10 In Germany, it can take more than 20 years to plan and construct a new runway (Bickenbach et al., 2005). 11 Facing a night ban at following the construction of a fourth runway, Lufthansa Cargo (2006) stresses the importance of night flights for the cargo industry. 12 At Düsseldorf, a maximum number of aircraft movements allowed between May and October each year and a maximum hourly runway usage prevent the airport from growing in line with the market (Düsseldorf Airport, 2007). 13 At Weeze Airport, despite a 2440m runway, only aircraft up to ICAO Code C and B757 are permitted, as shown on http://www.airports-worldwide.com/germany/niederrhein_germany.htm [as of 04 August 2008]. 14 There are many articles focusing on the importance of airport accessability. In an early paper focusing on the New York – New Jersey area, Augustinus (1974) showed that airport choice depends on ground access time. Later studies conducted by Weisbrod at al. (1993) or Windle/Dresner (1995) confirm the importance of access time.

10 11

4.2.3 External factors While in hub-and-spoke networks15, demand from various origins is concentrated at the transfer hub, sufficient local demand is supposed to be a necessary condition for airlines to start direct long haul services from a secondary, non-hub airports. As discussed in the literature, the following factors describe the attractiveness of an airport’s catchment area and could thus influence local demand for long haul flights: Number of inhabitants and immigrants from overseas, economic power (GDP), industry structure, political importance and the region’s importance for tourism.16 Of special importance is the demand generated by business travellers as this group is usually by far less price elastic than leisure travellers and thus willing to pay for unrestricted, full-fare economy, business and first class tickets (Oum et al., 1986, Dresner, 2006). Additionally, local demand could be boosted in case of severe capacity constraints at the national hub. This does mainly apply to Frankfurt or

London.17 Especially new and independent long haul carriers that do not possess enough slots at the major hubs could be forced to make use of nearby secondary airports (Ewald, 1990).18 The presence of a larger, better-known airport nearby, however, could also reduce a secondary airport’s chance of attracting long haul carriers, as carriers and passengers might prefer to use the larger airport. In bilateral air service agreements (ASA’s), finally, governments rule which airlines are allowed to offer how many flights between which airports in the contracting states. While restrictive ASA’s usually allow the (former) national carriers only to operate scheduled services, “open sky” agreements permit all airlines of the countries involved to fly as often as they want and from whatever airport they like

(Gillen et al., 2001, pp. 31-32). Today, only the aviation markets within Europe and between Europe and the US are highly liberalised, while flights to most other long haul destinations are still relatively

15 For a detailed analysis of the function of hub-and-spoke-networks, see Bailey et al. (1986), Hansen/Kanafani (1989) or Hanlon (1996). 16 See for example Brons at al. (2002), Hanlon (1996), Janic (2006) and Pagliari (2005). 17 In the end of the 1990s, the demand for slots at Frankfurt airport had already reached a level of more than 110 movements per hour on working days. Between 7:30 and 21:30, it was constantly higher than the airport’s capacity of 80 movements per hour (Bundesregierung, 2000, p. 36). Up to now, supply has slightly increased to 82 movements per hour, but demand is supposed to have risen even faster causing severe peak load problems.

11 12

restricted. Direct flights in these regions from secondary airports are often prohibited. Thus, a further

liberalization of ASA’s would make it easier for secondary airports to attract new long haul services

(Gillen et al., 2001, pp.185-187, Haworth, 1996, p.68). Table 5 summarizes and sorts the factors

identified and discussed above.

Table 5: Potential factors on long haul flights (airport perspective)

Degree of Airport-related factors (Marketing-Mix) Airport-unrelated external factors Price Promotion Product Place influence Airport charges Public relations Incentives (discounts, Market research Low rebates) Airline marketing Infrastructure Operational Access Medium restrictions Catchment area Competition by hubs high Bilaterals Source: Own production.

4.3 Empirical test of the discussed factors 4.3.1 Determination and quantification of input and output variables Multiple OLS regression analysis is used to test the dependence of one output variable from one or

more independent input variables. Here, the output variable has to describe the supply of long haul

services at secondary airports. For the passenger segment, we use total weekly MTOW on long haul

services as output variable as this indicator comprises not only seats but also belly cargo capacity. The

analysis is conducted for all long haul flights from all 224 secondary European airports in a week in

October 2007, adjusted by seasonal variations. In contrast to timetable analyses based on OAG data

only, our data set has been manually collected and even includes long haul charter flights, for instance

from Scandinavia and the UK to Thailand or Florida, which are not listed in OAG and other booking

engines. For the all-cargo segment, timetable data are virtually non existent. Thus, we will employ a

binary output variable which takes the value of “1” if the respective airport handles at least irregular

all-cargo flights, and use logistic regression analysis. Table 6 shows the quantification of the

independent variables.

18 Slots are allocated in a system based on grandfather rights. This means an airline can use a slot as long as it uses it regularly which makes it nearly impossible for “new” airlines to access a congested hub (Graham,

12 13

Table 6: Determinants of the supply of long haul flights Factor Quantification Internal Marketing Mix Measures Market research / Marketing Market research and marketing activities are barely quantifiable. Pricing Rebates and discounts are usually not published. However, it is likely that airports focusing on short haul low cost flights would also give discounts for long haul carriers. Thus, we employ the dummy variable [LCAIRPORT] for airports on which low cost carriers have a market share of about 50% or more. Semi- Airport Infrastructure External Runway Length [RWY] Physical runway length (in metres) (additional estimations considering runway strength as well have not led to substantially different Operational restrictions results) Night ban Dummy [NIGHT] (0 = no 24 hour operations; 1= 24 hour operations) Limit on aircraft movements Limitations on aircraft are barely quantifiable. External Catchment area Economic power in the [GDP04]: GDP of 2004 in all NUTS 2 regions whose largest city is located catchment area [GDP] less than 60 minutes from the airport (more recent data not yet available) Industry structure [GBI] GBI-Index developed by the Globalization and World Cities (GaWC) Study Group mirroring the number of global service companies and their branch offices located in the 60-minute catchment area Importance of inbound [BEDS]: Number of hotel beds in the catchment area in 2006 tourism for the region Ethnic minorities These and other national characteristics are considered in using country dummys for Ireland/U.K. [UK], Spain [SPAIN] and eastern Europe [EAST] Political importance of the Dummy [CAPITAL] (0 = no national capital located in the catchment area; catchment area 1 = national capital located in the catchment area] Proximity of and Dummy [COMP] is used if a larger airport is located less than 60 minutes competition by a larger away from the airport in question Degreeairport of congestion at the Dummy [HUBCAP] (0=idle capacity available; 1=congestion) (only national hub Frankfurt and London Heathrow are regarded as congested) Bilaterals The degree in how far bilaterals are liberalized is barely quantifiable. Others Total passengers in 2006 [PAX06] Total passenger numbers at the airport in question in 2006 Source: Own production.

4.3.2 Discussion of the results At first, we regard factors on long haul flight supply in the passenger segment. With regard to the

classical assumptions of linear regression, it shows that double log regressions models lead to better

results than non-log estimates. Table 7 shows the regression results for models A (which includes total

passenger numbers PAX06) and B (which does not include PAX06 and GBI). In both cases, the

variables COMP, UK and LNRWY seem to have a significant impact on long haul flight supply. If

PAX06 is included in the estimation, it is also highly significant, along with SPAIN and GBI. If

PAX06 is not included, LNGDP and CAPITAL turn out to be significant. Not surprisingly, GBI and

LNGDP are highly correlated.

2003).

13 14

Table 7: Impact of the discussed variables on direct long haul flight supply in the passenger segment (in total MTOW/week) at secondary European airports (stepwise double log OLS regression analysis) Model A B N 224 (all secondary airports) 224 (all secondary airports) R² / R² adj. 0,636 / 0,626 0,551 / 0,541 F-Value 63,103 53,612 Coefficients B Beta B Beta (constant) -53,458*** -53,798*** LNPAX06 0,801*** 0,390 n.i. SPAIN -1,954*** -0,176 n.s. COMP -0,952*** -0,113 -1,901*** -0,225 UK 3,299*** 0,281 4,194*** 0,357 LNRWY 5,596*** 0,263 6,034*** 0,284 LNGBI 0,403*** 0,219 n.i. LNGDP n.s. 0,815*** 0,342 CAPITAL n.s. 2,849*** 0,274 Dependent variable: LNMTOWWK (weekly MTOW on passenger long haul flights) n.s.: not significant at the 5% level, n.i.: not included in the estimation *)/**)/***) significant at the 10% / 5% / 1% level Source: Own production.

The results indicate that long haul flight supply at secondary airports is highly dependent of the economic power and industry structure in the catchment area. If PAX06 is included in the estimation, it does (not very surprisingly) turn out to be very significant, as airports with high passenger numbers are usually located in economically viable areas. In this case, a negative coefficient for SPAIN is also significant, because many airports in holiday regions in Spain handle high passenger numbers, but virtually no long haul flights, since all holidaymakers arrive from within Europe. Table 8 displays long haul flight supply and total passengers numbers at Spanish secondary airports in 2006/2007.

Table 8: Long haul flight supply and total passenger numbers at Spanish secondary airports Airport Passenger number (2006) Total seats offered on long haul flights (2007) Barcelona 30.008.152 548.028 Palma 22.408.302 0 Malaga 13.076.252 16.614 Gran Canaria 10.286.635 0 Alicante 8.893.749 9.412 Teneriffa Sur 8.845.668 0 Lanzarote 5.626.337 0 Valencia 4.969.113 0 Ibiza 4.460.141 0 Fuerteventura 4.424.880 0 Teneriffa Norte 4.025.601 13.260 Bilbao 3.876.062 0 Sevilla 3.870.600 0 Girona 3.614.223 0 Menorca 2.690.992 0 Santiago 1.994.519 0

14 15

Murcia 1.645.886 0 Reus 1.385.157 0 Jerez 1.381.560 0 Vigo 1.187.730 0 Granada 1.086.221 0 Almeria 1.055.545 0 Santander 649.447 0 Vallalodid 457.618 0 Vitoria 173.607 0 Source: Own production.

Airports in the UK and Ireland [UK] seem to easier attract long haul services. This can be explained by the high number of immigrants from overseas and by strong ethnic connections to North

America. In contrast to this, immigrants in Germany, for instance, mainly come from Southern Europe and thus do not generate any demand for ethnic long haul services. Runway length has a positive impact because longer runways enhance an airport’s technical capability to handle larger long haul aircraft. However, figure 2 indicates that runway length is only a sine qua non, because a sufficient runway length does not automatically yield in a good number of long haul flights.

Figure 2: Correlation between runway length at secondary airports and long haul flight supply (in total weekly MTOW) in 2007

Looking at cargo traffic, it is difficult to employ a metric output variable as cargo timetables are virtually not publicly available and lots of ad-hoc charters are not listed anywhere. Thus, we use logistic regression analysis to find out which variables have an impact on the binary output variable

15 16

LOGCARGO which takes the value of “1” if the respective airport handles all-cargo flights at least sometimes. Planespotter databases and websites helped us find out which 70 airports belong to this group.

Table 9: Factors on the likelihood that a secondary airport handles long haul cargo flights Model LOGCARGO Dependent Variable CARGO (MTOW/week) N 224 McFadden-R² 0,40 Chi-Quadrat 111,304*** Hosmer-Lemeshow statistic 0,846 Percentage better predicted by logistic regression 68,6% / 85,7% Estimation results B Exp(B) (constant) -11,707*** 0,000 RWYLENGTH 0,003*** 1,003 CAPITAL 1,518*** 4,565 GDP04BIOEUR 0,006*** 1,006 GDP04BIOEUR90MIN = GDP in Billion € in the 90 minutes catchment area *)/**)/***) significant at the 10% / 5% / 1% level Source: Own production.

The results indicate that runway length, capital and GDP within the 90 minutes catchment area seem to have a significant impact on long haul cargo flight supply at secondary airports. This comes not surprisingly as cargo shipments usually arrive in regions with a certain demand, and not in rural areas.

In most European countries, except for Germany and Italy, these regions are supposed to be located not too far away from the respective capitals. Runway length is more crucial for cargo than for passenger services as long haul cargo aircraft are usually large widebody aircraft.

5 CONCLUSIONS

In this paper, factors that influence long haul flight supply at European airports have been discussed and empirically tested based on a data sample featuring 224 and thus nearly all secondary airports in

Europe. It shows the economic power in the catchment area of an airport does heavily and positively influence long haul flight supply, while proximity to a larger airport reduces its chances of gaining long haul services. Long runways alone are not enough to boost long haul flights. As a consequence, there are already dozens of airports in Europe which do not gain enough revenues (if any) from long haul services to finance their long haul specific infrastructures. To avoid further misallocation,

16 17 runways should only be extended at airports that are – among other things – located in economically viable regions, and not too close to larger airports.

References

ACI (2005) Airport Traffic Report December 2004. Airport Council International. Brussels and

Geneva.

Airports Authority of India (2005) Aerodrome Operations Manual. Available online at

http://www.airportsindia.org.in/aai/aai-emp-rt.htm [28.11.2005].

Augustinus, J.G. (1974) An air passenger airport distribution model for the New York – New Jersey

area. Airport Economic Planning, 193-209.

Bailey, E. E., Graham, D. R., Kaplan, D. P. (1986). Deregulating the airlines. Cambridge and London.

Barrett, S.D. (2004). How do the demands for airport services differ between full-service carriers and

low-cost carriers? Journal of Air Transport Management 10, 33-39.

BCG – The Boston Consuling Group (2004) Airports – Dawn of a new era. Munich.

Bickenbach, F., Kumkar, L., Sichelschmidt, H. et al. (2005) Ausbau der Flughafeninfrastruktur – Kon-

flikte und institutionelle Lösungsansätze. Kieler Studien 335, Berlin.

Blackstone, Buck, Hakim (2006) Determinants of Airport Choice in a Multi-Airport Region. Atlantic

Economic Journal 34, issue 3, 313-326.

Brons, M., Pels, E., Njikamp, P. et al. (2002) Price elasticities of demand for passenger air travel: a

meta analysis. Journal of Air Transport Management 8, 165-175.

Bundesregierung (2000) Flughafenkonzept der Bundesregierung. Bundesministerium für Verkehr,

Bau- und Wohnungswesen. Berlin.

Burghowut, G., de Wit, J. (2005) Temporal configuration of European airline networks. Journal of Air

Transport Management 11, 185-198.

Caves, D.W., Christensen, L.R., Tretheway, M.W. (1984) Economies of density versus economies of

scale: why trunk and local service airline costs differ. Rand Journal of economics 15, 471-489.

Dresner, M. (2006) Leisure versus business passengers: Similarities, differences, and implications.

Journal of Air Transport Management 12, 28-32.

17 18

Düsseldorf Airport (2007) Oberverwaltungsgericht bestätigt aktuelle Betriebsgenehmigung des Düs-

seldorfer Flughafens. Press release from 16 May 2007.

EU Commission (2005) Community guidelines on financing of airports and start-up aid to airlines

departing from regional airports. Brussels.

Ewald, R. (1990) Mittelgroße Flughäfen rücken ins Rampenlicht. FVW International, Nr. 4/1990, 61-

62.

Gillen, D.W. et al. (2001) The impact of liberalizing aviation bilaterals. Aldershot.

Gillen, D., Lall, A. (2004) Competitive advantage of low-cost carriers: some implications for airports.

Journal of Air Transport Management 10, 41-50.

Gillen, D., Morrison, W.G. (2005) Regulation, competition and network evolution in aviation. Journal

of Air Transport Management 11, 161-174.

Graham, B. (1995) Geography and air transport. Chichester.

Graham, A. (2003) Managing airports. Oxford.

Hanlon, P. (1996) Global airlines. Oxford.

Hansen, M., Kanafani, A. (1989) Hubbing and airline costs. Journal of Transportation Engineering

115, 581-590.

Haworth, M. (1996) Good relations. Airline Business, June 1996, 68-71.

Janic, M. (2006) A macroscopic analysis and forecasting of the airport passenger demand, ATRS 2006

conference paper. Delft.

Kanafani, A., Ghobrial, A.A. (1982) Aircraft evaluation in air network planning. Transportation

Engineering Journal 108, Nr. TE3, 282-300.

Knibb, D. (1993) The Gateway Game. Airline Business, November 1993, 44-47.

Lufthansa Cargo (2006) Very good results in difficult operating conditions. Press release from March

30, 2006, available online at http://www.lufthansa-

cargo.com/content.jsp;jsessionid=73C8CDCBF03413AB77C99F335763BBFE?path=0,1,19141

,19156,77793&bhcp=1 [14.09.2006]

Malina, R. (2005) Market power and the need for regulation in the German airport market. Discussion

paper 10, Institut für Verkehrswissenschaft. Münster.

18 19

Morrison, W.G., Mason, K. (2006) So what is a low cost carrier?, Working paper presented at the 10th

Air Transport Research Society World Conference in Nagoya/Japan. Waterloo/Cranfield.

Oum, T.H./Gillen, D.W./Noble, S.E. (1986) Demands for fareclasses and pricing in airline markets.

The logistics and Transportation Review 22, 195-222.

Pagliari, R. (2005) Developments in the supply of direct international air services from airports in

Scotland. Journal of Air Transport Management 11, 249-257.

Porger, V. (1978) Europäischer Flugtourismus in der BRD, Teil 1. Zeitschrift für

Verkehrswissenschaften.

Roth, P. (2002) Targeting air travel markets in a fluctuating context: a methodology for non-hub

airports. Conference paper for STRC 2002, Session Travel. Lausanne.

Windle, R./Dresner, M. (1995) Airport choice in multiple-airport regions. Journal of Transportation

Engineering, 332-337.

Appendix 1: List of secondary airports in Europe Country Airport Country Airport Austria Graz Italy Lamezia Terme Austria Innsbruck Italy Mailand Linate Austria Klagenfurt Italy Mailand Malpensa Austria Linz Italy Neapel Austria Salzburg Italy Olbia Belgium Bruxelles Italy Palermo Belgium Charleroi Italy Pescara Belgium Liege Italy Pisa Belgium Ostend Italy Rimini Bulgaria Bourgas Italy Rom - Ciampino Bulgaria Plovdiv Italy Treviso Bulgaria Sofia Italy Trieste Bulgaria Varna Italy Turin Czech Republic Brno Italy Venedig Marco Polo Czech Republic Karlovy Vary Italy Verona Czech Republic Ostrava Latvia Riga Czech Republic Pardubice Lithuania Kaunas Czech Republic Praha-Ruzyne Lithuania Vilnius Cyprus Larnaca Luxembourg Luxemburg Denmark Aalborg Malta Malta Denmark Aarhus Nederlands Eindhoven Denmark Billund Nederlands Maastricht Denmark Copenhagen Norway Bergen Denmark Esbjerg Norway Bodo Denmark Karup Norway Evenes Estonia Parnu Norway Haugesund

19 20

Estonia Tallinn Norway Kristiansand Finland Helsinki Norway Lakselv-Banak Finland Jyväskylä Norway Oslo Finland Kittilä Norway Stavanger Finland Kuopio Norway Torp Finland Oulu Norway Tromsö Finland Rovaniemi Norway Trondheim Finland Tampere Norway Alesund Finland Turku Poland Bydgoszcz Finland Vaasa Poland Gdansk France Ajaccio Poland Katowice France Bastia Poland Krakau France Beauvais Poland Poznan-Lawika France Biarritz-Anglet Poland Rzeszow France Bordeaux Poland Stettin France Brest Poland Warsaw-Okecie France Calvi Poland Wroclaw France Chateauroux Portugal Faro France Cherbourg Portugal Lissabon France Clermont-Ferrand Portugal Madeira France Deauville Portugal Porto France Epinal Mirecourt Romania Bukarest-Baneasa France Figari Romania Bukarest-Otopeni France Grenoble Romania Constanta France La Rochelle Romania Timisoara France Lille Slovakia Bratislava France Limoges-Bellegarde Slovakia Kosice France Lourdes Slovenia Ljubljana France Lyon Slovenia Maribor France Marseille Spain Alicante France Metz Spain Almeria France Montpellier Spain Barcelona France Nantes Spain Bilbao France Nizza Spain Fuerteventura France Paris ORY Spain Girona France Pau Spain Gran Canaria France Perpignan Spain Granada France Strasbourg Spain Ibiza France Toulouse Spain Jerez France Vatry Spain Lanzarote Germany Berlin-Schönefeld Spain Malaga Germany Berlin-Tegel Spain Menorca Germany Bremen Spain Murcia Germany Dortmund Spain Palma Germany Dresden Spain Reus Germany Düsseldorf Spain Santander Germany Erfurt Spain Santiago Germany Frankfurt-Hahn Spain Sevilla Germany Friedrichshafen Spain Teneriffa Norte Germany Hamburg Spain Teneriffa Sur Germany Hannover Spain Valencia Germany Karlsruhe/Baden-Baden Spain Vallalodid Germany Köln/Bonn Spain Vigo Germany Leipzig-Halle Spain Vitoria

20 21

Germany Lübeck Sweden Goteborg-Landvetter Germany Münster/Osnabrück Sweden Karlstad Germany Nürnberg Sweden Kiruna Germany Paderborn/Lippstadt Sweden Lulea-Kallax Germany Saarbrücken Sweden Malmö-Sturup Germany Stuttgart Sweden Orebro Germany Weeze Sweden Ostersund-Froson Greece Athen Sweden Skavsta Greece Iraklion Sweden Stockholm Greece Corfu Sweden Vasteras Greece Kos Sweden Vaxjö Greece Mytilene Switzerland Basel-Mulhouse Greece Zakynthos Switzerland Genf Greece Rhodos UK Belfast Int'l Greece Thessaloniki UK Birmingham Hungary Balaton UK Bournemouth Hungary Budapest UK Bristol Hungary Debrecen UK Cardiff Ireland Dublin UK Doncaster Robin Hood Ireland Knock UK Edinburgh Ireland Shannon UK Exeter Italy Alghero UK Glasgow Italy Ancona UK Leeds-Bradford Italy Bari UK Liverpool Italy Bergamo UK London-Gatwick Italy Bologna UK London-Luton Italy Brescia UK London-Stansted Italy Brindisi UK Manchester Italy Cagliari UK Manston Italy Catania Fontanarossa UK Newcastle Italy Forli UK Nottingham-East Midlands Italy Genua UK Prestwick Source: Own production.

21