AVIATION IN WORLD REGIONS; NEW FACTS & NEW APPROACHES

AVIATION IN WORLD REGIONS; NEW FACTS & NEW APPROACHES Paul van de Coevering Hugo Gordijn Ruimtelijk Planbureau (Netherlands Institute for Spatial Research) PO box 30314, 2500 GH The Hague, The Netherlands coevering@rpb.nl gordijn@rpb.nl 1. INTRODUCTION During the 1980s and 1990s, the worldwide air passenger transport sector grew continuously. The number of passengers increased and, subsequently, the supply of air transport rose. Several events influenced this growth. Firstly, the liberalization of the air transport market in the United States and Europe had an important impact on these developments. Secondly, in the new millennium the terrorist attack on the United States and the SARS epidemic in Asia led to a (temporary?) decline in the number of aircraft passengers. The air carriers had to cope with these developments in a more or less competitive market. This significantly affected the air transport network as a whole. This paper aims to asses and describe changes in the air passenger network during the last decennium (1996-2004). Changes in air transport supply and changes in airline network structure are taken into account. Where possible, causes of these changes are identified. Previous studies also described development of air passenger networks in several ways. However, these were often regionally oriented and comprised a limited period. In this study, we try to asses developments on a world level, thereby analyzing regional differences. This aim breaks down into the following research questions: How did the worldwide supply of airline transport change during the last decennium? To what extent did the structure of the airline transport network change? Are these worldwide phenomena or are there regional differences, and if so, where did growth and decline of airline transport supply take place? To answer these questions, the Netherlands Institute for Spatial Research has obtained the OAG (Official Airline Guide) dataset comprising all worldwide flights for the consecutive years 1996-2004. The supply of air transport is described by the total frequency, number of destinations reached and total seat capacity during September of each year. In addition, we look at aircraft types in service. Did the carriers start using higher capacity planes during the last decennium or is there a trend towards smaller airplanes? The structure of the airline network is described with a spatial concentration measure, the Gini index. This measure is based on the absolute difference in seat capacity

between every possible airport pair in the airline network scaled to the number of airports in that network and the average seat capacity per airport. It is important to note that these datasets only contain scheduled passenger flights. Data on non-scheduled flights were not (sufficiently) available. Hence, this study does not take into account the rise of holiday charter flights. To a certain degree, this may imply an underestimation of air transport supply, especially on smaller, regional airports. 2. GLOBAL DEVELOPMENTS IN AIR TRANSPORT SECTOR Before the development of air service provision is illustrated, we describe factors influencing these developments. During the research period, various external factors influenced the configuration of the airline networks. These include: Changes in air transport policy; Regional and global economic prospectives; Terrorist attacks and health issues. Traditionally, air transport policy was fragmented with airlines being treated as public entities, serving primarily national interests. During the late 1970s and the 1980s air transport policy in the USA transformed considerably. Deregulation and liberalization with free market entry, and fare-setting freedom led to free market competition. Consequently, new carriers entered the country, discount fares emerged, and the level of service increased considerably. The European Union followed the trend towards deregulation and liberalization but it was not until 1987 that the 'First Package' of deregulation measures was launched. There were liberal agreements between some member states before, but this was the first comprehensive initiative to liberalize the air transport market throughout the whole union. The 'First Package' was followed by a second (1990) and third (1993) package which successively uplifted most barriers for free market competition. During the 1990's these measures affected the air transport in the EU in a similar manner as in the United States, the total effects were smaller though (Button, 2001). The trend towards more liberalization in the air transport continues as countries in Southeast Asia are discussing the topic and recently the UK and Canada presented plans for liberalization of the transatlantic air transport markets. Regional and global economic conditions also affect the air transport sector significantly. During economic recession or financial crisis, the demand for air travel can drop considerably. During the research period, the world suffered from a number of crises and recessions. In the early 1990's mainly the USA and Canada, Australia and the UK were recovering from a major economic recession. In 1991 economic crisis struck Japan and the crisis halted the economic growth. The most prominent crisis - during the research period - however is the East Asian financial crisis that started in July 1997 in Thailand and affected several other Asian countries. Indonesia, South Korea and Thailand were the countries most affected by the crisis. Mainland China, Taiwan, Singapore and Vietnam were relatively unaffected. Japan was not affected much by this crisis but was going through its own long-term economic

difficulties. Major effects of the crises were felt throughout the globe. The last major recession occurred in the early 2000's and was felt in Western countries affecting the European Union mostly during 2000 and 2001 and the USA mostly during 2002 and 2003. Meanwhile, Russia's economy began to recover and Japan's 1990s recession continued (see also Kilgour 1999, Pettis 2001 & Tiwari 2003). Furthermore, security concerns such as terrorist attacks on '9-11', the war in Iraq and health issues as SARS and the bird flu reduce the regional demand for travel, especially in the short term. The attacks on the USA in 2001 for instance caused a temporal decline in passengers of 20% in the following four months. For 2001 as a whole, 6,6% fewer passages boarded (BTS, 2002). Directly following the SARS outbreak from November 2002, passenger numbers fell more than 50% on certain airports in Asia. The effects on the longer term remain to be seen but several studies indicate that during 2003 and 2004 the aviation industry has been catching up. Beside these external influences on the air travel market, the sector itself is also on the move. The entrance of 'low-cost-airlines' in the 1990 led to a revolution. Instead of adopting the 'hub-and-spoke' models of the established carriers, some carriers developed point-to-point network structures, linking airports directly with each other instead of using a hub for transfer of passengers. These carriers (also known as a no-frills or discount carrier / airline) use smaller airplanes and generally offer low fares in exchange for eliminating many traditional passenger services. They have had a significant downward impact on airfares in the markets they have entered (Reynold- Feighan, 1994). 3. GLOBAL DEVELOPMENTS IN AVIATION Development in air service provision Let us now turn to the developments in the air service provision between 1996 and 2004. Figure 1 illustrates the development for the most important service indicators. The data relate to September of each year.

Figure 1: Development air service provision 1996-2004 (in September of each year) 120 115 110 105 100 95 90 85 1996 1997 1998 1999 2000 2001 2002 2003 2004 Frequency Average seat capacity No. nodes Seat km. No. flights During the 1990s, there was an evident rising trend in the worldwide service provision. The number of flights and the total frequency 1 increased, resulting in a growth of approximately 10% in total seat kilometers by 2000 (compared to 1996). By 2001, we start to see the effects of a global recession. Note that the effects of '9-11' are not incorporated in figures for 2001 because the data represent September of each year. Nevertheless, the frequency and number of scheduled flights already decreased slightly between 2000 and 2001. The total amount of seat kilometers still rises however in 2001, this mainly due to the rise in average seat capacity of the airplanes in service. By 2002, the effects of the terrorist attacks on '9-11' combined with the effects of the worldwide recession are apparent. For the first time in years, the growth in service provision is halted, and more severely, a decline sets in. Furthermore, in 2003 the SARS epidemic and the war in Iraq begin. This prevents the aircraft industry to recover sufficiently from the events in 2001 and the growth remains at a low pace. It is not before 2004 that a substantial recovery can be observed. An increase in the number of flights and frequency, together with the continuing rise in average seat capacity leads tot a 10% rise in total seat kilometers compared to 2003. For the first time in years, the level of service exceeds the pre-2001 levels. The data also indicate that the number of flights has risen faster than the total frequency. This means that the expansion of the service level is largely caused by the expansion of the number of flight routes rather than adding frequencies on existing routes (see also Swan, 2002). Meanwhile, the number of airports served declined steadily with a few percents a year during 1996-2004. By 2004, approximately 10% less airports are served compared to 1996. A closer examination of the airports without scheduled provision in 2004 reveals that these are relatively small and already had a very low level of provision in 1996. Moreover, the market for small airports seems to be highly volatile, with winners and losers of scheduled flights every year (table 1). So although, there is a significant decline in the number of airports served, the worldwide service level is not affected proportionally. This is mainly because the airports concerned only represent a small fraction of the total market in both frequency and seat supply. In

addition, the average number of seats supplied on the newly connected airports is considerably higher than on the airport from which service is withdrawn. This is not because greater frequencies though, these are even smaller. The higher service provision on the newly connected airports is caused by the larger average seat supply per aircraft. Table 1: Airport volatility Market share frequency 1996 Market share seat supply 1996 Average seat supply Average seat supply per aircraft No. Airports Average frequency Airports 1996 3.903 100,0% 100,0% 1.382 140.332 102 Service redrew 2004 760 2,6% 0,9% 177 5.979 34 New service 2004 395 1,0% 0,7% 144 9.785 68 Service decline 2004 365 1,5% 0,2% The higher average seat supply to these newly connected airports and the abandoning of the smaller airports could be an effect of the emerging regional jets in the airline industry (see also Wong et al, 2005). Typically, regional jets are designed to seat 150 or fewer passengers. These aircrafts are suitable for expanding the networks by servicing more distant cities, previously beyond the older turboprops 2 range or with too few passengers to be profitably served by mainline jets. This creates opportunities for distant cities previously not served by airliners. However, evidence from the USA shows that the rise of the regional jets might just a well be accompanied by a loss of air service in some smaller markets. This seems counterintuitive but this has to do with the smaller markets that are now served by turboprops. It remains to be seen whether those markets may generate enough passenger traffic and revenue to be financially viable to sustain regional jet operations once the turboprops are retired (Wong, D.K.Y. et al., 2005; Arnoult, 2002; De Lollis and Hansen, 2002). Airplane size In 1958 the first successful jet carrier, the Boeing 707 began service, initiating the jet era. The first decades of this era were characterized by rapidly growing airplane capacities. By 1985 the growth was over and the Boeing 747, capable of containing approximately 400 seats 3 held the size record for more than 35 years, although it recently has been surpassed by the Airbus A380 (due to enter service in early 2007). The data on the air provision recorded a growth in the average seat capacity during the research period. Figure 2 shows the market share by flight route 4 of airplanes by size category in the research period. This clearly shows that smaller aircrafts have a larger market share in the flight schedules than bigger aircrafts. This is no surprise because the larger aircrafts are mainly in service on the heavily used routes and these represent a minority in the total worldwide flight schedule. The trend in market share of individual aircraft sizes is more interesting. The market share of the routes with small airplanes up to 50 seats decreased during the research period. This supports the suggestion in the previous section that routes with the lowest densities, served by small airplanes (mainly

turboprops) are in decline. As said, this development could be worrisome for small nodes connected via these services. Moreover, the decline in market share of the category 'up to 50' is not, as expected, accompanied by a steady increase in the category of 51 up to 100 seats. Only in recent years, a small increase in market share can be detected for this category after a decline in the preceding period. The medium sized carriers with 101 up to 105 and carriers with 150 up to 200 seats show the largest increases in market share. Furthermore, the data reveal that relatively large aircrafts holding 301 up to 350 seats are also gaining, while aircrafts holding more than 350 seats are losing market share. Figure 2: Development market share per size by routes 1996-2004 50% 40% 30% 20% 10% 0% 1996 1997 1998 1999 2000 2001 2002 2003 2004 tot 50 51-100 101-150 151-200 201-250 251-300 301-350 351-400 400+ The market share by route does not consider the frequency because it is intended to give insight in the development of flight routes suitable for certain aircraft sizes. However, the market share per size category with this frequency taken into account is important. In figure 3, the market share by total seat supply is displayed. With a solid market share of approximately 45% since the 1990s, the aircrafts with capacity of 101 up to 150 seats are clearly leading in the market followed by the airplanes with a capacity of 151 up to 200 seats, with a market share that rose to approximately 18% in 2004. Similar results are found by Swan (2002). The market shares of the lager aircrafts shows an interesting development. The market share of the two largest aircraft categories (350+) is declining steadily while the share of aircrafts with a capacity of 301 up to 350 seats is gaining market share. So, the year on year average growth of the average number of seat per aircraft is not prompted by the use of aircrafts holding more passenger seats. The use of aircraft of the largest categories even declines during the research period. Instead, growth is induced because medium sized aircrafts gain market share, at the expense of smaller aircrafts.

Figure 3: Development market share by seat supply 1996-2004 60% 50% 40% 30% 20% 10% 0% 1996 1997 1998 1999 2000 2001 2002 2003 2004 tot 50 51-100 101-150 151-200 201-250 251-300 301-350 351-400 400+ Network structure The impact of the regulation and liberalization in the USA was not limited to an increase in the level of service in the air transport network alone. The structure of the air network also changed considerably. After the deregulation 'hub-and-spoke' networks evolved on a large scale because the carriers could achieve considerable efficiency gains by concentrating their business around a small number of hubs. Effects on the network structure in Europe were less apparent because the carriers already focused their operations around mayor hubs before liberalization. In the past, diverse measures were employed to describe these variations in concentration in the airline networks. The 'Gini Index' was found most appropriate for describing these changes in network concentration (Reynold-Feighan, 1994). The gini index of concentration is defined as: G 1 2n = Y 2 i j y i y j with 'y i ' and 'y j ' representing the air traffic at airport i or j, with 'n' representing the number of airports and the Y is the mean air traffic per airport. A gini index of '1' indicates theoretically a 100% concentration of services on one airport and a gini index of '0' means that services are equally distributed over all the airports. Table 2 shows the worldwide gini index for the distribution of seat supply during the research period, comprising all airports served by scheduled flights. The high index values show that the worldwide air transport network is dominated by a limited number of 'hub' airports with a high level of concentration of services. This is not surprising regarding the fact that in each

country most international services are concentrated on one or a few airports. Moreover, the effects of the liberalization in the EU (and before the USA) seem to be crystallized by 1996 and the index proves to be stable between 1996 and 2004 with only minor fluctuations. More importantly however, the stable index indicates that the problems in the air transport industry that prompted a decline in service provision did not affect the network structure, at least so far. Table 2: Gini Index worldwide Gini-index Year passenger seat provision 1996 0,8918 1997 0,8918 1998 0,8915 1999 0,8925 2000 0,8918 2001 0,8913 2002 0,8875 2003 0,8838 2004 0,8833 4. REGIONAL ANALYSES Changes in air transport provision In the previous chapters, developments in air service provision are described on a worldwide level. The worldwide analyses show a growth of the provision during the 1990's and a significant drop in the beginning of the new millennium. Where did this growth and decline take place? For a better understanding of the underlying processes, regional analyses are performed, pinpointing similarities and differences across regions. For this purpose, the worldwide air travel market is split into nine regions (figure 4).

Figure 4: World air travel market regions AF: Afrika AS1:Asia 1 AS2:Asia 2 AS3: Asia 3 EU: Europe LA: Latin America ME: Middle-East NA: North-America SW: SouthWest Pacific Figure 5 shows the development of total seat kilometers for each region. The provision has grown worldwide in all regions, although Southwest Pacific and Southeast Asia show significant drops in service provision - probably prompted by the recession in this region- in the late 1990s. All regions - except the Middle East - are affected by the downturn in service provision after 2001 with losses in air provision or at least a stagnation of growth. The USA is heavily affected as well as Latin America. The Asian countries show growth mainly in late 1990s, after 2001 the growth diminished but a strong recovery sets in during 2004. Figure 5: Development total seat km. per region 1996-2004 160 140 120 100 80 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 AF AS1 AS2 AS3 EU LA ME NA SW Figure 6 shows the development of the total frequency for each region. Except for Southeast Asia and the Southwest pacific - the total frequency in all world regions went up from 1996 to 2001. In 2002, there is an eminent decline in the total frequency of the USA, Latin America and the Southwest Pacific. The

frequencies in these regions seem to recover gradually over 2003 and 2004. Although Europe is less affected, the growth stagnated in 2002 and 2003. The Southwest Pacific is the only region where the total frequency declined during the research period. In contrast, the frequencies in all Asian air industries grew considerably in the new millennium, despite external troubles as SARS. A comparison of figure 5 & 6 shows that the frequency in the Southwest pacific declined faster than the total seat kilometers while in the Asian countries, the opposite holds. This is mainly caused by differences in average seat capacity, which declined in the Southwest pacific yet increased in Southeast Asia. Figure 6: Development total frequency per region 1996-2004 140 120 100 80 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 AF AS1 AS2 AS3 EU LA ME NA SW The decline in the number of airports served by scheduled flights, recorded in the worldwide analyses, occurred in the majority of regions (figure 7). The most significant drops are recorded in the Southwest pacific and Latin America. On the contrary, there is a significant increase in airports served in Southeast Asia and the Middle East. Figure 7: Development number of airports served per region 1996-2004 120 110 100 90 80 70 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 AF AS1 AS2 AS3 EU LA ME NA SW

In the worldwide analyses, the air provision on smaller airport seemed to decrease because the use of small aircrafts up to 50 seats went down. Is this the case for the Southwest Pacific and Latin America? In figure 8, the development of aircraft provision by size is displayed for Latin America. The market share of the smallest airplanes is clearly descending in this region while the share of medium sized carriers of 101-200 seats rises. This trend is also apparent in the Southwest Pacific region. This supports the conclusions drawn in the worldwide analysis. Figure 8: Development aircraft market share by size in Latin America 1996-2004 50% 40% 30% 20% 10% 0% 1996 1997 1998 1999 2000 2001 2002 2003 2004 tot 50 51-100 101-150 151-200 201-250 251-300 301-350 351-400 400+ Top 20 airports and developments. Table 3 shows the top 20 of airports by seat supply in 2004. The Northern American region is by far the most listed region with 10 airports. The airports listed of this region are all within the USA, indicating the primacy of this country in the worldwide air transport market. Not surprisingly, the other airports are in the EU and Asia. A trend analyses reveals that this situation is remarkably stable during the research period. In none of these periods, airports of other regions appear in the top 20. Furthermore the data reveal that the top 20 -which represents no more than 0,6 % of all airports in 2004 - is responsible for 21,3% of all seat passengers offered in the worldwide air market. This figure is also fairly stable, but slightly declining from 22,9% in 1996 to 21,3% in 2004.

Table 3: Top 20 of airports in 2004 by total seat supply of arriving and departing aircrafts Airport Airport name Region total seat Market cumulative code supply share market share ATL Atlanta International Airport NA 9478324 1,9% 1,9% ORD Chicago O'Hare International Airport NA 8560968 1,7% 3,5% LHR London Heathrow International Airport EU 7675449 1,5% 5,0% LAX Los Angeles International Airport NA 6580133 1,3% 6,3% HND Haneda (Tokyo) Airport AS 6450357 1,3% 7,6% DFW Dallas/Fort Worth International Airport NA 6406994 1,3% 8,8% FRA Frankfurt International Airport EU 6028166 1,2% 10,0% CDG Charles de Gaulle International Airport (Paris) EU 5961487 1,2% 11,2% PEK Beijing Capital International Airport AS 4675392 0,9% 12,1% DEN Denver International Airport NA 4550476 0,9% 13,0% MAD Madrid international Airport EU 4459340 0,9% 13,8% BKK Bangkok International Airport AS 4438556 0,9% 14,7% PHX Phoenix International Airport NA 4415057 0,9% 15,6% HKG Hong Kong International Airport AS 4411424 0,9% 16,4% LAS Las Vegas International Airport NA 4304037 0,8% 17,3% AMS Amsterdam Schiphol International Airport EU 4246583 0,8% 18,1% JFK John F. Kennedy International Airport NA 4188671 0,8% 18,9% DTW Detroit International Airport NA 4152266 0,8% 19,7% MSP Minneapolis-Saint Paul International Airport NA 4092249 0,8% 20,5% NRT Tokyo Narita International Airport AS 3800581 0,7% 21,3% 5. CONCLUSIONS The provision of seat kilometers in the worldwide air transport market has grown considerably during the last decennium. This growth is prompted by an increase in the number of flight routes and the average seat capacity. The total frequency on these routes only increased slightly. External events are significantly related to the worldwide service provision but especially on the air service provision in the region involved. This was particularly apparent in 2002 and 2003. Both the worldwide and the regional air markets seem to have recovered strongly in 2004. The worldwide network is very highly concentrated around a few mayor hubs situated in the USA, EU and Asia. The top 20 of worldwide airports are all in these regions indicating that the regions of olden days are still leading the industry. Although Northern America has by far the largest market share in service provision, followed by Europe and Asia, the latest has had a higher growth path in recent years. During the research period, the structure of the air network did not change significantly. A minor decline of concentration set in the last years and the top 20 of airports are losing some market share but these changes are too small and to varied across regions and time to draw definitive conclusions. Moreover, the external events were clearly no reason for the carriers to radically change the way operations were undertaken and the rise of the low cost aircrafts has not affect the network much during the research period.

Despite the growth of the air provision in almost all regions, which can be met with enthusiasm, there also seems to be reason for concern because a significant number of smaller airports have lost their scheduled services over the past decade. This is most likely due to the decline in the use of the small turboprop aircrafts with a seat capacity mostly under 50 seats and a rise in the use of the somewhat larger aircrafts ranging from 100-200 seats. This could be troublesome for regions with low passenger potential that are being served by the turboprop aircrafts.

NOTES 1. The number of flights represents the number of different flight routes performed by the carriers. The total frequency represents the accumulated frequencies performed on all these routes. 2. Usually 50 seats or less, 36 seats on average during research period 3. Depending on layout, the 747 could accommodate 416 or 524 passengers 4. Type of Aircraft in service per individual flight route (frequency on these routes is not taken into account).

BIBLIOGRAPHY Arnoult, S. (2002) Express jet stretches to grow, Air transport world, November, 51-55. Button, K.J. (2001) Deregulation and Liberalization of European Air Transport Markets, Innovation, 14 (3) 255-275. BTS (2002) Air Carrier Traffic Statistics, Bureau of Transportation Statistics, Office of Airline Information. Available on the World Wide Web: <http://www.bts.gov> De Lollis, B., Hansen, H. (2002) Airlines abandon small cities, US Today, 22 October. Kilgour, A. (1999) The changing economic situation in Vietnam: A product of the Asian crisis? Available on the World Wide Web: <http://www.geogr.unigoettingen.de/kus/apsa/pn/pn12/vietnam.html>. Pettis, M. (2001) The Volatility Machine: Emerging Economies and the Threat of Financial Collapse, Oxford University Press. Reynolds-Feighan, A.J. (1994) The impact of US airline deregulation on airport traffic patterns. Geographical Analyses 30 (3) 234-253. Swan, W.M. (2002) Airline route developments: a review of history, Journal of Air Transport management, 8 (5) 349-353. Tiwari, R. (2003) Post-crisis Exchange Rate Regimes in Southeast Asia, Seminar Paper, University of Hamburg. Wong, D.K.Y., D.E. Pitfield, I.M., Humphreys (2005) The impact of regional jets on air service at selected US airports and markets, Journal of transport geography, 13 (2) 151-163.