Skip to main content

Competitive Analysis in Oligopolistic Aviation

Research reportView only

Create alertChat

February 28, 2025

What elements are needed to analyze the competitive position of an airline in the oligopolistic aviation market?

Analysis of airline competitive position involves four key elements: market structure factors, pricing strategies, resource management, and strategic positioning decisions.

Abstract

Airline competitiveness in oligopolistic markets depends on several interconnected elements. Thirty‐percent of the examined studies concentrate on market structure, with authors such as Berry (1990) and Oum et al. (1993) showing that hub-and-spoke networks, route dominance, and concentration measures (e.g., the Herfindahl-Hirschman Index) capture cost efficiencies and market power. Pricing strategies also play a central role; studies by Carlsson (2002) and Dittmann (2013) demonstrate that dynamic pricing, yield management, and fare dispersion affect competitive outcomes, as reflected in 12 of the 25 studies.

Resource-based factors further influence competitive position. Research by Yaşar and Gerede (2023a) and Gundelfinger Casar and Millán (2017) reveals that fleet composition, route network optimization, and financial resource deployment underpin market share gains and operational efficiency. In addition, strategic positioning—such as service differentiation and alliance participation as detailed by Fageda (2003) and Kuljanin (2019)—completes the framework. Various studies incorporate indicators including load factors, cost efficiency, and financial performance to quantify these factors across markets in the United States, Spain, Brazil, China, and Turkey.

Methods

We analyzed 25 sources from an initial pool of 500, using 6 screening criteria. Each paper was reviewed for 5 key aspects that mattered most to the research question. More on methods

Papers identified with Elicit search

n = 500

Papers screened using: Study Focus, Competitive Analysis, Comparative Scope, Study Type, Market Context, Study Scope

n = 500

Papers screened out

n = 475

Papers included for extraction

n = 25

Press enter or space to select a node.You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.

Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Paper search

Using your research question “What elements are needed to analyze the competitive position of an airline in the oligopolistic aviation market?”, we searched across over 126 million academic papers from the Semantic Scholar corpus. We retrieved the 500 papers most relevant to the query.

Screening

We screened in sources based on their abstracts that met these criteria:

• Study Focus: Does the study focus on commercial passenger airlines operating in oligopolistic aviation markets?
• Competitive Analysis: Does the research examine competitive positioning factors (such as market share, cost structure, or route network assessment)?
• Comparative Scope: Does the study analyze multiple airlines or provide comparative analysis between airlines?
• Study Type: Is the study an empirical study, case study, theoretical framework, systematic review, or meta-analysis focused on airline competition?
• Market Context: Does the study include explicit competitive context and market positioning implications?
• Study Scope: Does the study go beyond exclusively focusing on cargo airlines, technical aviation operations, airport operations, or regulatory frameworks without competitive analysis?

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

• Market Structure and Competitive Environment:

Describe the specific market context of the airline study:

• Identify the geographic market (e.g., Turkish international airline market, US domestic market)
• Specify the number of airlines competing in the market
• Characterize the market structure (e.g., oligopoly, number of competitors)
• Note any specific market conditions (e.g., deregulation, hub-and-spoke networks)

Look for this information in the introduction or methodology sections. If multiple market characteristics are mentioned, list all of them. If information is incomplete, note “insufficient information” and provide any partial details available.

• Competitive Analysis Approach:

Extract the specific methodological approach used to analyze competitive dynamics:

• Research methodology (e.g., panel regression, content analysis)
• Data collection period (start and end dates)
• Sample size of airlines studied
• Key analytical techniques or frameworks used

Locate this information primarily in the methodology section. If multiple approaches are used, list all. Ensure to capture the specific analytical techniques that were used to examine competitive positioning.

• Factors Influencing Competitive Performance:

Identify and list all factors found to impact airline competitive performance:

• Positive influencing factors (e.g., firm maturity, fleet homogeneity)
• Negative influencing factors
• Specific performance metrics affected (e.g., market share, revenue passenger kilometers)

Extract this information from the findings/results section. Ensure to note both statistically significant positive and negative factors. If quantitative impact is provided (e.g., correlation strength), include that information.

• Competitive Strategies and Actions:

Catalog the types of competitive actions identified in the study:

• Types of competitive moves (e.g., schedule-based, price-related)
• Strategies for market positioning
• Differences in strategies between airline types (e.g., traditional vs. low-cost carriers)

Search the results and discussion sections for explicit mentions of competitive strategies. If specific examples are provided, include them. Note the frequency or prevalence of different strategy types if such information is available.

• Performance Measurement Approach:

Detail the specific performance metrics used to assess competitive positioning:

• Operational performance indicators
• Financial performance indicators
• Comparative performance measurement methods

Locate this information in the methodology and results sections. Ensure to capture both the metrics used and how they were calculated or compared across airlines. If multiple performance dimensions are analyzed, list all of them.

Results

Characteristics of Included Studies

Study Focus

Methodology

Market Context

Key Variables Examined

Full text retrieved

Airline competition and product differentiation

Theoretical model

US domestic market (implied)

Airport presence, hub-and-spoke systems, market power

No

Price and frequency choice

Address model, econometric analysis

Eight European countries, 172 city-pair markets

Market structure, prices, flight frequency

No

Market structure and pricing strategies

Concentration indicators, econometric modeling

Brazilian domestic airline market

Market concentration, pricing, operational costs

Yes

Competitive behavior of airlines

Choice Theory, Pricing Theory

Chinese domestic airline market

Pricing strategies, market structure

No

Effect of market structure on prices

Panel data analysis

International air transport industry

Market structure, prices, government policy

No

Competition and liberalization effects

Empirical model, instrumental variables

Spanish domestic market

Market structure, product differentiation, scale advantages

No

Airline competition and conduct

Oligopoly models

Spanish air transport market

Market conduct, density economies

No

Airline competition and conduct

Demand and pricing equation system

Spanish airline market

Market conduct, density economies

No

Regulation, market structure, and performance

Regression analysis, factor analysis

27 OECD countries, 102 air routes

Regulation, market structure, performance

Yes

Oligopolistic competition

Instrumental variables, econometric modeling

Spanish domestic airline market

Demand, market structure, pricing

Yes

toof

Pageof

View 15 more rows

Analysis of the systematic review table suggests the following key points:

• Study Focus: Competition was the most common focus, appearing in 15 out of 25 studies. Other frequent focuses included pricing/price (4 studies), market structure (3 studies), market conduct (3 studies), performance (3 studies), and market power (3 studies).

• Methodology: Econometric analysis or modeling was the most common methodology, used in 6 studies. Empirical models or studies were used in 3 studies, and instrumental variables in 2 studies. We didn’t find methodology information in the abstracts of 2 studies where full texts were not retrieved. Various other methodologies were each used in one study.

• Market Context: The most studied markets were Spain and the US (4 studies each), followed by Europe, Brazil, and Turkey (3 studies each). China was the focus of 2 studies. Various other markets were studied in individual studies, and we didn’t find market context information in the abstract of 1 study where the full text was not retrieved.

• Key Variables: Pricing or price was the most commonly examined variable, appearing in 12 studies. Market structure was the second most common, examined in 8 studies. Market power, market conduct, and performance were each examined in 3 studies. A wide variety of other variables were examined in individual studies, reflecting the diverse approaches to studying airline competition.

This analysis suggests a strong focus on competition and pricing in airline markets, with a preference for econometric methodologies. The studies cover a range of geographic contexts, with a slight emphasis on European and US markets.

Thematic Analysis

Market Structure and Competitive Dynamics

Key findings related to market structure and competitive dynamics include:

• Hub-and-Spoke Network Effects:
  • Berry (1990) argues that hub-and-spoke networks allow airlines to reduce costs and increase service frequency, enhancing competitive position.
  • Oum et al. (1993) note that intensifying hub-and-spoke networks is a key strategy for major US airlines to solidify market power.
  • Fageda (2004, 2005) highlight that these networks enable airlines to benefit from economies of density and scale in the Spanish market.
  • Oliveira and Oliveira (2016) discuss how these networks contribute to market concentration and potential market power abuse in the Brazilian airline industry.
• Market Concentration Impacts:
  • Gonenc and Nicoletti (2000) find that lower market concentration is associated with increased productive efficiency and lower fares across OECD countries.
  • Piga and Bachis (2007) suggest that while dominance within a route is associated with higher fares, a larger share within a city-pair does not necessarily facilitate market power.
  • Carlsson (2002) finds that increased market concentration leads to higher Business class ticket prices in European markets.
  • Cavalcante de Souza et al. (2017) highlight how high market concentration in the Brazilian market allows leading companies to control pricing.
• Entry Barriers and Competitive Responses:
  • Fageda (2003) notes that scale advantages of major airlines in the Spanish market create barriers for smaller airlines.
  • Cavalcante de Souza et al. (2017) identify high costs for acquiring aircraft and slots as significant entry barriers in the Brazilian market.
  • Kuljanin (2019) examines how the entry of low-cost carriers (LCCs) into the long-haul sector has altered the competitive landscape in the transatlantic market.
  • Najda (2003) argues that the presence of LCCs is a more important determinant of route competitiveness than traditional measures of market concentration.
  • Kuljanin (2019) describes how full-service carriers use strategies such as “capacity extension” to mitigate the impact of LCCs.
  • Yaşar and Gerede (2023b) provide a detailed analysis of competitive moves, noting differences in strategies between traditional and low-cost carriers.

Resource-Based Competitive Factors

Key findings related to resource-based competitive factors include:

• Fleet Composition and Utilization:
  • Yaşar and Gerede (2023a) identify fleet homogeneity as a positive influencing factor for market share gain and operational performance.
  • Gundelfinger Casar and Millán (2017) find that average aircraft size negatively influences price, indicating that larger aircraft lead to lower prices, possibly due to economies of scale.
• Route Network Optimization:
  • Berry (1990) highlights how hub-and-spoke systems allow airlines to reduce the number of round-trips necessary while increasing passenger miles flown, potentially leading to cost advantages.
  • Oum et al. (1993) note that intensifying hub-and-spoke networks is a key strategy for major airlines to enhance their competitive position.
  • Oliveira and Oliveira (2016) discuss how traffic density and route-airport dominance influence market concentration.
• Financial Resource Deployment:
  • Yaşar and Gerede (2023a) identify financial resources as a positive influencing factor for market share gain.
  • Zhang (2020) highlights the importance of cost management in the competitive landscape of China’s civil aviation industry, noting that high costs, particularly from aviation fuel and airport fees, significantly impact profitability and competitive performance.
  • Cavalcante de Souza et al. (2017) note that efficient companies with lower operational costs can compete more effectively on price, thereby increasing their market share.

Strategic Positioning Elements

Key findings related to strategic positioning elements include:

• Pricing Power and Strategy:
  • Carlsson (2002) finds that market structure significantly affects Business class ticket prices in European markets, with increased concentration leading to higher prices.
  • Piga and Bachis (2007) note that while dominance within a route is associated with higher fares, a larger share within a city-pair does not necessarily facilitate market power.
  • Oum et al. (1993) mention the use of “skilful dynamic pricing and seat allocation techniques” as a competitive tool employed by major US airlines.
  • Dittmann (2013) discusses the application of yield management in the Chinese domestic market as a key pricing strategy.
• Service Differentiation:
  • Fageda (2003) highlights product differentiation as an essential feature of airline competition, allowing airlines to target different market segments.
  • Cavalcante de Souza et al. (2017) provide specific examples of service differentiation strategies, including shorter connection times, loyalty programs, onboard meals, and flexible baggage policies.
  • Kuljanin (2019) discusses how full-service carriers and low-cost carriers employ different strategies, with full-service carriers focusing on quality and network advantages while low-cost carriers compete primarily on price.
• Alliance and Partnership Effects:
  • Oum et al. (1993) mention the use of commission overrides to travel agents as a strategy to enhance market position.
  • Gonenc and Nicoletti (2000) note that airline alliances can lead to higher economy and business fares, suggesting that these partnerships can enhance market power and pricing ability.
  • Yaşar and Gerede (2023a) find that competing in the same alliance negatively affects Revenue Passenger Kilometers (RPK), a key operational performance metric.
  • Kuljanin (2019) discusses how major airlines benefit from alliances, using them as a tool to compete against the emerging threat of low-cost carriers in long-haul markets.

Competitive Position Indicators

Indicator Type

Measurement Approach

Competitive Impact

Industry Application

Market Concentration

Herfindahl-Hirschman Index (HHI)

Higher concentration may lead to increased market power and higher prices

Widely used across studies to assess market structure

Route Dominance

Percentage of flights or passengers on a route

Positive impact on pricing power and market share

Used to analyze competitive dynamics on specific routes

Network Efficiency

Hub-and-spoke metrics, connectivity indices

Enhances cost efficiency and service frequency

Applied in studies of network carriers’ competitive strategies

Pricing Power

Price elasticity, fare dispersion

Indicates ability to maintain higher prices and profitability

Analyzed in various market contexts to assess competitive strength

Operational Efficiency

Load factors, aircraft utilization

Improves cost position and profitability

Used to compare airline performance and competitive positioning

Financial Performance

Revenue Passenger Kilometers (RPK), Net Profit/Loss

Reflects overall competitive success and sustainability

Applied in comprehensive analyses of airline performance

Service Quality

Customer satisfaction scores, on-time performance

Impacts customer loyalty and ability to command price premiums

Considered in studies of service differentiation strategies

Market Share

Percentage of passengers or revenue in a market

Indicates competitive strength and potential for economies of scale

Widely used as a key indicator of competitive position

Cost Efficiency

Unit costs, cost per available seat kilometer

Enhances pricing flexibility and profitability

Analyzed to understand cost-based competitive advantages

Fleet Composition

Fleet age, aircraft type diversity

Impacts operational flexibility and cost structure

Considered in studies of airline efficiency and strategy

toof

Pageof

View 5 more rows

Analysis of the competitive position indicators table reveals:

• Indicator Types:
  • The studies collectively mentioned 15 different indicator types used to assess airline competitiveness.
  • The most common were:
    • Operational indicators (4 studies)
    • Market Structure indicators (3 studies)
    • Financial indicators (3 studies)
• Measurement Approaches:
  • 16 different approaches were identified:
    • Percentage-based measures were the most common (3 studies)
    • Indices and index-based measures were used in 2 studies
    • Other approaches were each found in only one study
• Competitive Impacts:
  • 23 different competitive impacts were identified across the studies:
    • Profitability was the most frequently mentioned impact (3 studies)
    • Higher prices and cost efficiency were each mentioned in 2 studies
    • All other competitive impacts were found in only one study each
• Industry Application:
  • The diversity of indicators, measurement approaches, and competitive impacts suggests a complex and multifaceted approach to assessing airline competitiveness in the literature.
  • The studies did not present a single dominant framework for measuring airline competitiveness, instead using a variety of indicators and approaches to capture different aspects of competitive performance.

References

Xavier Fageda\ (2005).Measuring Conduct and Cost Parameters in the Spanish Airline Market

Steven T. Berry\ (1990).Airport Presence as Product Differentiation

T. Oum, A. Zhang, Yiming Zhang\ (1993).INTER-FIRM RIVALRY AND FIRM-SPECIFIC PRICE ELASTICITIES IN DEREGULATED AIRLINE MARKETS.

Yannis Kerkemezos\ (2019).On the Dynamics of (Anti)competitive Behaviour in the Airline Industry

Mehmet Yaşar, Ender Gerede\ (2023).Results of competitive dynamics: Examination of the factors determining the operational and financial performance of airlines: The case of the Turkish international airline market. Journal of Entrepreneurship, Management and Innovation

Mehmet Yaşar, Ender Gerede\ (2023).HAVAYOLU PAZARINDAKİ REKABETÇİ EYLEMLERİN RAKİP ÇİFTLERİ BAĞLAMINDA DEĞERLENDİRİLMESİ: TÜRKİYE DIŞ HATLAR ÖRNEĞİ. Akademik yaklaşımlar dergisi

Lingyan Zhang\ (2020).Analysis of Market Conduct and Performance of China's Civil Aviation Industry

Xavier Fageda\ (2004).Measuring Conduct and Cost Parameters in the Spanish Air Transport Market

Javier Gundelfinger, P. Coto-Millán\ (2020).Price and demand behaviors in the Scandinavian airline industry following market liberalization

R. Gonenc, G. Nicoletti\ (2000).Regulation, Market Structure and Performance in Air Passenger Transportation

Thiago Cavalcante de Souza, Sarah Farias Andrade, Aniram Lins Cavalcante\ (2017).Estrutura de concorrência e estratégia de preço no mercado brasileiro de aviação civil. RACE - Revista de Administração, Contabilidade e Economia

Charles Najda\ (2003).Low-Cost Carriers and Low Fares : Competition and Concentration in the U

Marcus Oliveira, Alessandro V. M. Oliveira\ (2016).What Drives Effective Competition in the Airline Industry? An Empirical Model of City-Pair Market Concentration

Paul S. Dittmann\ (2013).MA Chinese-European Economics and Business Studies The Development of China's Civil Aviation Industry and the Competitive Behavior of its Airlines in the Domestic Market

Xiaoqian Sun, Changhong Zheng, S. Wandelt, Anming Zhang\ (2024).Airline competition: A comprehensive review of recent research. Journal of the Air Transport Research Society

C. Piga, Enrico Bachis\ (2007).Hub Premium, Airport Dominance and Market Power in the European Airline Industry

Xavier Fageda\ (2003).Hay competencia en el mercado europeo de transporte aereo?: el caso espanol

Alessandro V. M. Oliveira\ (2010).A Alocação de Slots em Aeroportos Congestionados e suas Consequências no Poder de Mercado das Companhias Aéreas

Junbiao Su, Hanjun Wu, K. Tsui, Xiaowen Fu, Zheng Lei\ (2023).Aviation resilience during the COVID-19 pandemic: A case study of the European aviation market. Transportation Research Part A: Policy and Practice

I. Kilinç, M. A. Öncü, Y. Taşgi̇t\ (2012).A study on the competition strategies of the airline companies in Turkey.

F. Carlsson\ (2002).Price and Frequency Choice under Monopoly and Competition in Aviation Markets

M. Dresner, M. Tretheway\ (2002).MODELLING AND TESTING THE EFFECT OF MARKET STRUCTURE ON PRICE: THE CASE OF INTERNATIONAL AIR TRANSPORT. IN: AIR TRANSPORT

Yang Xu\ (2011).Competitive Network and Competitive Behavior: A Study of the U.S. Airline Industry

Javier Gundelfinger Casar, P. Millán\ (2017).Measuring Conduct in the Spanish Air Transport Industry

J. Kuljanin\ (2019).A model for air transport market competition assessment: airline and alliance perspective

Download BIBDownload RISDownload TXT

Report

Status

Gather sources

500 sources found

Details

Screen sources

25 sources included

Details

Extract data

125 data points extracted

Details

Generate report

Generate PDF

BIBLaTeX, ZoteroRISZotero, MendeleyTXTAPA BibliographyGenerate PDFPDFGenerate DOCXMicrosoft Word

Chat

Got some follow-up questions?

Sign up or sign in to chat with this report.

Back

Regulation, Market Structure and Performance in Air Passenger Transportation

R. Gonenc, G. Nicoletti

2000·

117 citations

SourceDOI

Plain textPDFFigures

Searching for PDF

Unable to find PDF from source

Back

Market Structure and Competitive Environment

- Geographic Market: 27 OECD countries, covering both national and international routes. - Number of Airlines: Generally a few carriers dominate, often through incumbent flag-carriers or alliances. - Market Structure: Oligopoly, with significant market concentration and control over slots by single airlines. - Specific Market Conditions: Deregulation, liberalization, hub-and-spoke networks, and airline alliances.

Competitive Analysis Approach

- Research methodology: Ordinary Least Squares (OLS) regression, factor analysis - Data collection period: 1996-97 - Sample size: 27 OECD countries, 102 air routes - Key analytical techniques: Regression analysis, factor analysis

Factors Influencing Competitive Performance

Positive Influencing Factors: - Competitive regulatory and market environment - Low market concentration - Presence of challenger airlines - Liberal charter rights

Negative Influencing Factors: - Airport congestion and dominance - Airline alliances - Government control (mixed effects)

Specific Performance Metrics Affected: - Productive efficiency - Fare levels (business, economy, discount) - Occupancy rates

Competitive Strategies and Actions

- Types of competitive moves: Price-related (fare regulations), schedule-based (slot dominance) - Strategies for market positioning: Liberalization of entry and charter rights - Differences in strategies between airline types: Traditional carriers often form alliances that can increase fares; challenger airlines use liberal charter rights to reduce discount fares

Performance Measurement Approach

The "Performance Measurement Approach" in R. Gonenc, G. Nicoletti (2000) involves assessing operational performance through load factors (occupancy rates) and financial performance through airfares. These metrics are analyzed using regression models to evaluate the impact of regulatory and market environments on efficiency and pricing. Comparative performance is measured across countries and routes using detailed indicators from factor analysis, focusing on productive efficiency and fare levels.

The paper uses a data base on regulation, market structure and performance in the air passenger transportation industry, to analyse the links among liberalisation, private ownership, competition, efficiency and airfares at national and route levels. Covering the 1996-97 travel season, 21 aggregate indicators have been developed for 27 OECD countries, and 23 micro indicators for 102 air routes connecting 14 major international airports. These data, summarised by means of factor analysis show that i) regulations affecting the air industry vary heavily across countries and routes; ii) in most markets, air services are still provided by a few carriers, generally dominated by an incumbent flag-carrier or by an airline alliance between incumbents; iii) only in a few cases new entrant airlines play a significant role; iv) in a large number of airports, a single airline controls more than half of the available slots; v) as a result, few international routes are truly open to ... Cette etude utilise une base de donnees concernant la regulation, la structure des marches et la performance dans le transport aerien de passagers, au niveau national ainsi qu’au niveau des lignes, pour analyser les liens entre d’une part la liberalisation, la privatisation et la concurrence et, d’autre part, l’efficacite de l’industrie et les tarifs des voyages aeriens. Pour la saison 1996-97, il a ete possible de developper 21 indicateurs agreges pour 27 pays de l’OCDE et 23 indicateurs detailles pour 102 lignes aeriennes reliant 14 grands aeroports internationaux. Ces donnees, synthetisees a l’aide de l’analyse factorielle, montrent que i) la regulation de l’industrie du transport aerien varie remarquablement selon les pays et les lignes ; ii) dans la majorite des marches, les services de transport aerien sont encore fournis par un nombre restreint de vecteurs, qui sont generalement domines par une compagnie nationale presente depuis longtemps sur le marche ou par une alliance ...

\\\\ Cette étude utilise une base de données concernant la régulation, la structure des marchés et la performance dans le transport aérien de passagers, au niveau national ainsi qu'au niveau des lignes, pour analyser les liens entre d'une part la libéralisation, la privatisation et la concurrence et, d'autre part, l'efficacité de l'industrie et les tarifs des voyages aériens. Pour la saison 1996-97, il a été possible de développer 21 indicateurs agrégés pour 27 pays de l'OCDE et 23 indicateurs détaillés pour 102 lignes aériennes reliant 14 grands aéroports internationaux. Ces données, synthétisées à l'aide de l'analyse factorielle, montrent que i) la régulation de l'industrie du transport aérien varie remarquablement selon les pays et les lignes ; ii) dans la majorité des marchés, les services de transport aérien sont encore fournis par un nombre restreint de vecteurs, qui sont généralement dominés par une compagnie nationale présente depuis longtemps sur le marché ou par une alliance entre plusieurs compagnies ; iii) les cas où les nouveaux entrants jouent un rôle majeur sont largement minoritaires ; iv) dans un grand nombre d'aéroports une seule compagnie contrôle plus de la moitié des créneaux horaires disponibles ; v) par conséquent, très peu de lignes internationales sont vraiment ouvertes à la concurrence ; vi) la création d'espaces aériens régionaux et la signature d'accords "ciel ouvert" constituent des pas importants mais limités dans la direction de l'ouverture des marchés. En contrôlant, pour la dimension des marchés et des réseaux, ainsi que pour d'autres différences de nature technologique et économique, et en tenant compte des interactions entre les caractéristiques des lignes et de l'industrie nationale, les résultats des régressions transversales aux pays et aux routes suggèrent que i) l'efficacité augmente et les tarifs diminuent quand la régulation et la structure de marché deviennent plus favorables à la concurrence ; ii) l'efficacité dans l'utilisation des avions est affectée de façon positive par les pressions concurrentielles, mesurées par un indice de concentration bas ; iii) les tarifs aériens réagissent à la libéralisation indépendamment de la structure du marché mais, quand l'environnement réglementaire est libéral, leur diminution est renforcée par l'existence de concurrents sur la ligne ; iv) les tarifs affichés des classes affaires et économique ont tendance à baisser quand les prix sont libéralisés et la concentration du marché diminue, mais ont tendance à augmenter quand les marchés sont dominés par des alliances entre compagnies aériennes ; v) les tarifs soldés sont affectés par le degré de concurrence globalement présent dans les pays reliés par la ligne ainsi que par les régulations concernant les charters et la présence effective de compagnies challengers sur la ligne ; vi) la congestion et la domination des aéroports par une compagnie aérienne tendent à augmenter les tarifs dans les segments du marché qui sont plus sensibles aux temps de départ et d'arrivée.

Classification JEL : K23, L51, L93, L32, L1 Mots-clés : Industries réglementées, économie de la régulation, transport aérien, structure de marché, libéralisation, concurrence, entreprises publiques, stratégie d'entreprise.

REGULATION, MARKET STRUCTURE AND PERFORMANCE IN AIR PASSENGER TRANSPORTATION

Rauf Gonenc and Giuseppe Nicoletti 1

1.Summary and conclusions

The airline industry transports passengers and goods within and across national borders on a scheduled and non-scheduled basis. In 1999, the world scheduled air traffic amounted to a total of 1.5 billion passenger journeys and a volume of 26 million tonnes of freight and is growing at a higher rate than world-wide GDP growth. Scheduled transportation accounts for most of passenger traffic, with charter flights representing 15 per cent of total travel output. Even though its share in GDP is limited (between 0.5 and 1 per cent in OECD countries), air transportation provides a critically important infrastructure to the global economy.

Until recently, scheduled passenger transportation has been heavily regulated in most countries at both the domestic and international levels. Domestically, entry into the air passenger business, initiating service on specific routes, operating aircraft above given sizes, reducing or discontinuing services, investing in airlines, establishing and applying various categories of passenger fares have been subject to a detailed set of rules and regulations. International regulations compound these domestic rules. They govern the conditions of entry and ownership, the selection of operable destinations, and the freedoms to set capacity and fares on international routes, which represent more than 60-70 per cent of all passenger-kilometres performed in most countries, with the exception of continental size countries.

The policies and regulations which have governed the air transport industry for several decades have various motivations (including safety, national prestige, national defence, regional and urban development, environmental sustainability, public service and other non-commercial objectives) specific to each country. However, there is growing consensus that unnecessarily restrictive regulations may have led to significant losses of economic efficiency, and thereby failure to secure low-cost air transportation to the largest possible proportion of the population -the ultimate objective of air transport policies. Recognising these shortcomings, several OECD governments have initiated reforms in the past two decades. Their aim was to improve efficiency and reduce airfares by increasing competition, encouraging the rationalisation of air networks, and enhancing airline governance.

A large literature on airline economics has supported reform policies in the past two decades, but cross-country examinations of the relationship between regulatory frameworks, market structures and 1.

OECD Economics Department. The authors wish to thank Mike Feiner, Jorgen Elmeskov for their help in improving earlier versions of this paper. Useful comments were also provided by Ignazio Visco, Sally van Siclen and Nick Vanston. Special thanks also go to Wolfgang Hübner, Michel Nicolas and Patrice Dubus for their support as well as to Charles Dudley and Attilio Costaguta of ICAO, and Rebecca Rowland and Jon Manning of IATA for their advice. Martine Levasseur and Anne-Claire Saudrais provided statistical and technical assistance. The opinions expressed in the paper are those of the authors and do not reflect necessarily the position of the OECD or its Member countries.

performance have been relatively few. Most of the empirical research has concerned the United States, at first inspiring the 1978 domestic airline reforms and later looking at the economic implications of liberalisation and enhanced competition (Caves et al., 1984; Bailey et al., 1985; Liu and Lynk, 1999; Reed, 1999). 2 U.S. research also demonstrated that certain feared outcomes of liberalisation --such as declines in safety levels, or deprivation of small communities of air service --did not materialise. 3 Only a few quantitative analyses have documented the implications of airline competition for efficiency and air fares in other parts of the OECD, such as the trans-Pacific routes (Gillen et al., 1998; Kissling, 1998) and the European internal market (Marìn, 1995; Morrell, 1998), while a monographic analysis looked at the productivity performances of large-sized international airlines (Oum and Yu, 1998).

This body of research has unveiled the peculiar competitive features of air travel markets and documented the efficiency gains and fare changes that followed liberalisation. 4 Results have been less clear-cut outside of the U.S. perhaps due to more limited liberalisation and the presence of exogenous factors damping competition -such as more intense airport congestion and social and political constraints on airline restructuring (Marìn, 1998; Lapautre, 2000). Research also helped identify those structural factors such as airport domination and market concentration which may temper the benefits of reforms. 5 verall, these analyses focused on the effect of market concentration on performance, side-stepping the role of regulation per se and failing to distinguish the influence of various kinds of regulatory and market arrangements. 6 Moreover, no study has looked at the effects on performance of the presence of "flag carrier airlines", often controlled by governments. Finally, few studies have accounted, within a unified framework, for influences on performance originating from regulation, market structure and infrastructure access conditions.

This paper attempts to fill some of these gaps by analysing the impact that different types of regulatory and market arrangements have on the efficiency of supply of scheduled air passenger travel services in the OECD area, controlling for "framework conditions" such as airport dominance and the role of government-controlled flag carriers. 7 The analysis is based on a detailed set of indicators of regulation 2. Derthick and Quirk (1985) provide a detailed account of how economic studies have inspired and contributed to the U.S. air transport reforms.

Certain negative effects were handled by pro-competitive regulations. Research showed that air services were discontinued in certain very small communities, inspiring a budget-funded "essential air services" programme. Surveys also helped detect the unsatisfactory safety performances of financially-strained small size airlines, justifying additional safety policies directed to this category of carriers.

Several studies were devoted to testing versions of the "contestable markets" hypothesis (for instance, Graham et al, 1983; Hurdle et al., 1989). Other studies showed that price differentials between business and discount fares widen under competition, raising in certain cases business fares and previously crosssubsidised short-distance fares to above pre-deregulation levels, but provoking a substantial overall decrease in total travel costs for all passengers (see, among others, Borenstein, 1992; Evans and Kessides, 1993; Morrison and Winston, 1999).

The effects of hub dominance on airfares are highlighted by Kahn (1993) and documented empirically by Abramowitz and Brown (1993), Kim and Singal (1993) and US Department of Transportation (1999).

Among the studies focusing on specific kinds of regulatory and market arrangements, Dresner and Tretheway (1992) et al. (1989) highlighted the role of "likely potential entrants" (carriers whose entry in a route is not deterred by economies of scale and scope) in moderating air fares over U.S. routes; and Kim and Singal (1993) suggested that airline mergers increased air fares in U.S. routes in the late eighties.

Regulations concerning charter flights affect competition and carrier performance in scheduled operations, as charters frequently substitute for scheduled services in "non-time sensitive" demand segments. Therefore, regulations which govern charter flights are also considered in this study as part of the regulatory framework of scheduled services.

and market structure. It follows a top-down approach looking first at the impact of the overall market and regulatory environment (represented by means of summary indicators) and then turning to the impact of specific regulation and market features, such as rules on prices, capacity and entry on individual routes or the role of challengers, third party carriers and airline alliances. The linkage between regulation, market structure and performance is studied both at the level of national industries supplying domestic and international travel services, and at the level of individual routes, focusing on a large set of international non-stop connections between the airports of the major OECD countries. Interactions between industry and route features, such as pressures for overall network optimisation originating at both the domestic and route level, are also accounted for. Two important dimensions of air travel performance are considered: productive efficiency (in its two main dimensions of capital use efficiency and efficiency of use of all production factors) and air fares. By distinguishing among different classes of fares (business, standard economy and discount), the disaggregated impacts of route-specific regulations, market structures and performance on business and leisure travel between individual city-pairs are highlighted.

The rest of the paper has three main sections. Section 2 overviews competition issues and regulation trends in OECD air transport, providing the necessary background for the empirical analysis. Section 3 describes patterns of regulation and market structure in OECD countries and in a large sample of international routes in (or around) 1996, the reference year for subsequent regressions. In this section, detailed information about regulations and market structures is summarised by means of factor analysis techniques. Finally, Section 4 uses the resulting indicators to estimate the impact of several regulation and market features on industry-wide and route-level performance.

The cross-country and cross-route analysis of regulatory and market arrangements highlighted some stylised features of the environment in which air travel services are supplied in the OECD area:

-Despite the wide-ranging reforms implemented by OECD governments in the past two decades, regulations affecting the air travel industry domestically and on individual routes still vary a lot across countries and routes.

-Relatively few international (long-haul) routes are truly open to competition. This inhibits the creation of a global aviation market and hinders network optimisation by air carriers. Because they prevent competition from third party carriers, regional arrangements and open sky policies constitute only a partial step in this direction.

-In many OECD countries and on most of the international routes included in the sample market concentration is significant and a few carriers supply air travel services. This generally occurs in the presence of an incumbent flag-carrier (often controlled by the government) and in only a few cases new entrants (such as third party carriers or challenger carriers) play a significant role. In addition, airline alliances (often between incumbents) have been formed in a large number of routes.

-In a large number of international airports congestion phenomena are reported to exist and a single airline controls more than half of the available slots.

The analysis of the linkages between regulatory and market environments, framework conditions and performances at the industry and route level suggests the following main conclusions:

-At both the national and route level there is clear evidence that productive efficiency and fares are affected by regulatory and market arrangements. Overall efficiency and the rate of occupancy of aircraft seats tend to increase and all categories of fares tend to decline as the regulatory and market environment becomes friendlier to competition.

-Industry and route (productive) efficiency is particularly sensitive to actual competitive pressures (as proxied by low market concentration on individual routes and at the national level), while fares react to changes in regulation independently from market structure, suggesting that potential entry has a disciplining role on prices. However, the effect of liberalisation on certain categories of fares is amplified by the existence of competitive market conditions. In addition, both route efficiency and certain categories of fares are also affected by overall market conditions prevailing in industries at route ends.

-Different categories of fares are sensitive to different types of regulations and market arrangements. Business and economy fares are particularly sensitive to pricing regulations and (when price competition is possible) market concentration, while discount fares are affected mainly by charter regulations and (when charter rights are extensive) the presence of challenger airlines, possibly reflecting the use of these rights as an entry device for competitors wishing to establish themselves on a route.

-There is some evidence that economy and (to a lesser extent) business fares are higher in routes dominated by airline alliances.

-There is some evidence that airport congestion and dominance tend to raise fares in the time-sensitive segments of the city-pair markets.

-The effects of government control are ambiguous at both the industry and route level. Business fares and the rate of occupancy of aircraft seats tend to increase with the role of government-controlled carriers, while discount fares tend to decrease.

On the whole, these results confirm that air transport reforms aimed at liberalising entry (e.g. by eliminating bilateral designation rules or extending charter rights) and prices involve significant benefits for all categories of travellers. The empirical relevance of scale effects and the finding that both route and industry environments matter for performance point to the potential gains to be obtained from the simultaneous liberalisation of domestic/regional markets and international (long-haul) routes, which encourages network optimisation and cost-efficiency while reducing price-cost margins. For these policies to fully bear their fruits, however, constraints on airport access must be relaxed and strategic behaviour by incumbents (e.g. through alliances and slot dominance) must be kept in check by appropriate competition policies. 8

Competition issues and regulation trends in air passenger transportation

The technology of the air travel industry involves economies of scale, but increasing returns are exhausted at relatively low levels of output. Economies originate principally from the so-called economies of density, which reflect the diminishing costs of the additional seats, passengers and flights on individual routes. Research on the economics of air transportation has shown that these economies do not hinder competition within routes (White, 1979; Caves et al., 1984; Liu and Lynk, 1999). Furthermore, when airlines have free access to airports, they can exert potential competition on each other's routes because reallocating existing equipment (aircraft) and personnel (flying crew) to new destinations is relatively easy.

Competition between air carriers takes place in a multiplicity of separate markets. Passenger services between individual cities, and at different times, form a web of markets with little cross-substitutability of demand between them. Connections between cities can be further segmented into 8.

To be effective, these policies may involve co-operation and co-ordination at the international level.

time-sensitive (i.e. business travel) versus non-time-sensitive (i.e. tourist travel) services, and operations between individual end-points (i.e. airport-to-airport routes). 9 In servicing this web of markets, airlines exploit a wide range of scope economies.

The development of "hub-and-spoke" networks over the past two decades, which resulted from the exploitation of both scale and scope economies, has added new dimensions to competition in air transport. In a hub-and-spoke network, traffic is collected from feeder points and consolidated at a hub before being redistributed by further flights to other destinations (short "spoke" or long "trunk" destinations). One airport, or no more than a small number of airports become the focus of the operations of air carriers, permitting the use of smaller numbers of larger aircraft and more frequent flights (Figure 1 ). Competition is affected in several ways:

-Higher total traffic on individual spoke and trunk routes raises output levels on the constant-cost segments of supply curves, facilitating entry competition.

-If interconnection at hubs between airlines is effected at low cost and at little consumer inconvenience, competition is further increased.

-Hub-and-spoke networks shift competition from the route level to the network level, with different hubs competing to attract passengers in transit to the same origin and/or destination cities. 10 -However, if transaction costs between airlines are significant, hub-and-spoke generates firm-specific scope economies between routes and may damp competition.

To date, the net effect of these factors has been to increase the degree of competition in the airline industry (US Department of Transportation, 1999).

[Figure 1 . Hub-and-spoke infrastructures and economies of density]

The inherently competitive character of the airline industry may be threatened by obstacles to airport access and anticompetitive behaviour of incumbent carriers (Pera, 1989). The terms of access of competing airlines to airport and air traffic infrastructures (runways, terminals, air traffic control) may have an important bearing on the degree of competition in downstream airline services (Kahn, 1993; OECD, 1998; Federal Aviation Administration, 1999). 11 The quality of infrastructure access is particularly important for efficiency and competition in hub-and-spoke networks where operations are organised in closely timed "flight banks", and delays in individual landing or take-off operations perturb and may 9.

Transportation of business travellers from New York JFK to London Heathrow and transportation of tourists from New York-New Jersey (Newark) to London Stansted via Amsterdam do not take place in the same market.

10. Kleit and Maynes (1992) underscore this new dimension of airline competition and explore its implications for the definition of relevant antitrust markets and competition policy.

Fixed ground infrastructures often present natural monopoly characteristics and other forms of market failures (e.g. environmental externalities) and generally confer a significant degree of market power to infrastructure operators. However, not all airports are natural monopolies. Costs flatten when traffic attains a threshold of 1.5 to 3 million yearly passengers (Doganis, 1992). Urban areas with intense air traffic, such as New York, Washington, Chicago, Tokyo, London, Paris, and Milan, can bear the efficient operation of more than one airport and may sustain airport competition. Generally, even with limited competition between airports, the pricing of infrastructures, the allocation of access rights under congestion, and the needed capacity extensions need to be appropriately regulated.

paralyse the entire network. Access problems are compounded when airport runways, terminals and traffic control infrastructures come in short supply and suffer congestion (under existing access rules and prices). 12 Anticompetitive business practices often involve the use of "slot dominance" in an airport to foreclose competitors or raise rival costs on certain routes (e.g. by strategically increasing flight frequencies). They also arise under privileged vertical relations between airlines and ancillary upstream and downstream services, such as computer reservation systems, travel agents, ground handling services etc. (Morrison and Winston, 1999; Esperou and Subremon, 1997). 13 15.

Despite its essentially competitive nature, air transport has been typically subject in OECD countries to strict domestic and international regulations concerning entry and exit, pricing and business operation. 14 The regulation of international exchanges of air transport services is often described by a typology of "freedoms" granted to carriers (Box 1). These international regulations are established and enforced either multilaterally, bilaterally between individual countries, or regionally among groups of countries (Box 2).

The common international practice is to give incumbent carriers "grandfather" rights (i.e. the right to maintain control over slots that were controlled by them in previous years). Remaining rights are sometimes reserved to new entrants and the trade of existing rights may or may not be authorised. For instance, the United States and EU Members reserve a quota of spare slots for new entrants; the United States authorises slot trading in some airports, and the EC only slot barters. In all instances incumbent airlines usually continue to control the wide majority of the airport utilisation rights.

Other marketing practices may also have potentially anti-competitive effects, such as instantaneous price discounts in response to new entries and certain types of consumer loyalty programmes. Instances where specific price discounts and frequent flyer programmes may hinder competition without being themselves a form of competition are difficult to establish in practice, but clearer competition rules in these areas are important for preserving market contestability (Van Miert, 1998; U.S. Department of Transportation, 1999; OECD, 2000).

14. This paper is not concerned with regulations addressing the external effects of air transportation, such as exhaustion gases, noise impacts, traffic congestion and air accidents. It should be noticed, however, that these are often handled in OECD countries in ways that may curb competition. For instance, quantitative ceilings on airport movements (aimed at containing pollution, congestion and safety risks) often favour incumbents and hinder new entry. The use of economic instruments, such as slot pricing and taxes and tradable permits for gas and noise emissions, is still rare even though they are likely to be more competitively neutral policies (see Gonenc et al., 2000). Certain international rules for civil aviation, especially those concerning the professional licensing of air crews and the safety certification of aircraft, and their gas and noise emission effects, are settled multilaterally, in the International Civil Aviation Organisation (ICAO). 15 Several OECD governments stress today the need for a more dependable enforcement of multilateral standards, their adaptation to growing air traffic and multiplication of market participants, and to new aviation technologies -such as computer and software-based navigation which necessitate new maintenance procedures (Olster et al., 1992; Savage, 1999; ICAO, 1999).

The rules which most directly affect the organisation of the industry, such as the designation of authorised carriers, their entry on specific routes, their freedom to establish capacity and prices, and the authorisation of charter flights are embedded in "bilateral air service agreements" (ASAs) between governments. There are at present more than 3000 of them in application. They spell-out the traffic rights between the two underwriting countries and describe in detail the air routes operable, the names of carriers allowed to enter, the aircraft types and flight frequencies authorised, the types of fares applicable, and the mechanisms available for fare settlement.

ASAs traditionally granted only the first four traffic freedoms described in Box 1.

A new generation of ASAs called "Open Skies" have been proposed by the US government in its bilateral agreements as from early 1990s, and there are at present 34 of them in application. They imply: i) no limits on the number of airlines that may be designated by either country; ii) unrestricted capacity and frequencies on all routes; iii) full fifth-freedom and sixth-freedom rights and unlimited "change of gauge" (change of aircraft type) on all routes; and iv) full pricing freedom unless fares are contested simultaneously by both governments (double disapproval regime). However, these agreements do not trigger full international competition because carriers continue to be designated by their respective governments, cross-country equity investments are not liberalised, and seventh (right of one country's airlines to carry independent traffic between two other countries) and eighth (consecutive service into domestic market) traffic freedoms, which are essential for network optimisation, are generally not available.

"Regional aviation markets" are a recent innovation in international civil aviation. They are aviation equivalents of free-trade areas and transform sub-sets of bilateral air routes between participating countries into domestic routes within the common aviation area. Two prominent regional aviation markets are the Australia-New Zealand 16. Domestic and international regulations on entry, pricing and service are often supplemented by restrictions concerning the ownership structures of airlines as well as corporate strategies (such as participation in domestic and international alliances). These restrictions, prescribed by domestic laws, bilateral ASAs or regional market provisions, require that carriers which service air routes within a country, between two countries or in a regional market must be "owned and effectively controlled" by the locals of concerned countries. They have been instrumental in creating the "flag carrier" concept where government or nationally-owned airlines gain a de facto monopoly on the domestic and international routes they service. International mergers, acquisitions and joint-ventures are therefore difficult, and the growth of firms across national boundaries takes indirect forms not involving capital exchanges, such as commercial alliances and code-sharing agreements.

In the past certain international fare structures, fare levels and service norms were established by the trade association of international airlines (IATA), to which governments delegated authority and granted anti-trust immunities. This role of IATA has declined in the recent period in most of the OECD area.

The regulatory straightjacket imposed on domestic and international air travel has affected business strategies, industry organisation and market structures. Government-owned or regulated airlines were sheltered from competition and developed strategies characterised by high operation costs, high quality of services, and relatively restricted output growth. 16 The lack of competitive pressures often resulted in a low utilisation rate of aircraft capacity, wages with a significant rent element, and relatively generous working conditions. In the air routes where prices were regulated but more than one incumbent competed, such as the busiest U.S. routes before deregulation and the busiest international routes, "quality of service competition" based on the frequency and comfort of flights was the main outcome. Entry, pricing and service regulations also have made the optimisation of networks difficult. Individual airlines could not freely choose their hub locations, route structures and flight frequencies according to economic considerations, but generally had to apply operation patterns prescribed by regulatory arrangements (domestic and bilateral route schedules centred on national hubs). Dynamically, they could not adjust their operations to structural and seasonal moves in travel demand, and to the shift of the technical frontier as new vintages of aircraft implied the adaptation of their network, 17 or may have been able to do so only with delay. Furthermore, government ownership and the "national ownership" rules did not permit adjustments in the governance of airlines, and ASA bans on mergers countered the emergence of optimised company boundaries and network structures. 18 .

Reforms introduced in the past two decades aimed at exploiting the potential for free competition in air travel markets. Prominent reforms included the pioneering U.S. domestic deregulation of 1978, the Australia-New Zealand Common Aviation area of 1992, the European Single Aviation Market of 1997 and a limited number of Open Sky air agreements signed in the course of 1990s (OECD, 1998; Gaudry and Mayes, 1999). Under these reforms, previously protected national, regional and bilateral markets became potentially open to challenge by new entrants, which in principle may compete on fares, frequency of flights, degrees of comfort, connection paths etc., to the extent permitted by airport availability (US General Accounting Office, 1996b). In addition, price discrimination became possible, whereby flight fixed costs can be funded according to the price-elasticities of different groups of customers. However, liberalisations remained nationally or regionally fragmented relative to the inherently global aviation marketplace: the bulk of international routes is still governed by highly restrictive bilateral ASAs. Reforms therefore failed to fully exploit the potential for global competition and trade specialisation (Box 3).

As a by-product, OECD-area airlines have continuously focused on the safety and reliability of their services, making high safety standards a positive legacy of the regulated era.

For instance, the introduction of new generations of larger jet aircraft increases the efficiency of hub-andspoke technologies (as a topical example, the recently announced 600-800 seat mega carriers are destined to operate on 20 or so international trunk routes), this may conflict with nationally centred route networks.

In the absence of stock market listings and takeover threats, incumbent management tends to become non-contestable and restructuring opportunities often go unnoticed.

Box 3. Potential for global competition and trade

Global competition and trade in air transportation are restricted by domestic and international regulations. Domestic routes are reserved to national carriers, except in regional aviation agreements which give open access to foreign airlines registered in partner countries. The vast majority of international routes are reserved to the national airlines of route-end countries, except in a small number of markets where third-country airlines may operate consecutive flights to-and-from their home countries. (Boxes 1 and 2) .

The limited scope for global competition contrasts with the technological characteristics of the sector, which are protrade. Aircraft and aircrew are perfectly mobile and can access any domestic or international market in a matter of hours, which make air transportation services genuinely tradable. Still, limited forms of trade are practised, via international subcontracting arrangements such as code-share and wet-lease agreements. In these arrangements "principal" airlines hire the aircraft and operating personnel of lower cost subcontractors. Code-shares and wet-leases permit the provision of lower-cost air services via the established brands and traffic rights of incumbents. They are more and more accompanied by cross-frontier equity investments which are nevertheless constrained by the prevailing "national ownership and control" rules.

International differences in labour costs hint at an underlying potential for global competition and trade. As major airlines use similar aircraft and technologies, no sustainable sources of long-term productivity differences are likely to compensate for large wage differentials. Labour costs represent up to 30 percent of airlines' operating costs, and the table below documents the magnitude of the wage differentials among countries (Table ).

Liberalisations also aimed at encouraging network rationalisation by facilitating more extensive recourse to hub-and-spoke. For instance, U.S. airline deregulation unleashed a previously unsuspected network restructuring potential. Network rationalisation has been comparatively more limited outside the United States, either because reforms have covered limited regional areas (such was the case of the Australia-New Zealand aviation integration) or because ASAs continue to constrain network remodelling. Open Sky air agreements certainly facilitate network optimisation when fifth freedom traffic rights in a succession of countries are available, but perfect matching between independent agreements is rare. On the other hand, Open Sky agreements do not offer seventh and eighth traffic freedoms which are important for network optimisation (See above Boxes 1 and 2). In addition, the persistence of political support for national hubs and their domination by national airlines may have slowed down restructurings in Europe.

Finally, reforms have aimed at opening up the capital structures of airlines. Mergers between companies have been facilitated and airline privatisations, often part of liberalisation packages, introduced new flexibility in capital and financing structures. In the United States, the wave of new entries, mergers and acquisitions which followed the 1978 deregulation brought about shake-ups in the ownership and governance of US airlines (including a number of bankruptcies), and led to sharp improvements in performance (Kole and Lehn, 1999). In Europe, privatisations and the establishment of the "community carrier" status waived national ownership constraints and opened the way to trans-European mergers. The European Commission required that equity injections to government-owned airlines be subject to the "private market investor principle" whereby investments are not cleared when they are not justified from a business perspective -therefore preventing hidden subsidies. The Australia-New Zealand agreement liberalised ownership adjustments between two countries' airlines. However, in all these reforms, "local ownership and control" rules remained and ruled out mergers, acquisitions and equity financings across large regional zones: In the US, foreign investors are not authorised to acquire more than 25 per cent of the voting capital of airlines and carrier boards must be controlled by US citizens. European Union Member countries, and Australia-New Zealand, as well as all other OECD countries have similar foreign investment restrictions, generally at higher thresholds such as 49 per cent.

The locally (regionally) important but globally limited impact of liberalisation reforms have been born out by a number of studies. In the US, early studies demonstrated that reform in "deregulated islands" such as Texas and California enhanced air service productivity and decreased local fares without affecting national markets (Levine, 1965). Studies after the 1978 reform revealed that efficiency and fare gains were concentrated on routes where true competition and network optimisation unfolded, free from operational obstacles such as infrastructure congestion, hub domination by incumbent carriers, and mergers foreclosing low-to-medium density ends of the networks to competition (Morrison and Watson, 1989; Hurdle et al., 1989; Borenstein, 1989; U.S. General Accounting Office, 1996; Kim and Singal, 1993; Liu and Lynk, 1999; U.S. Department of Transportation, 1999). Research in Europe shows that the benefits of the European single market fell short of expectations, because of hindrances to the free entry and growth of new competitors (European Commission, 1999), to the reorganisation of incumbent airlines (Marin, 1998; Neven et al., 1998), and to cross-country network optimisation (Lapautre, 2000). Analyses of the liberal air agreements indicate that benefits are maximised when true competitors to incumbents can actually enter the markets (Caves and Higgins, 1993; Gillen et al., 1998). Liberal agreements are in all cases confined to bilateral markets and effects are generally limited to long-haul point-to-point destinations (although consumer welfare gains on these large trunk routes are important), giving limited incentives to cross-country network redesign (for some aspects see Scott, 2000).

Comparing regulations and market structures across countries and routes

Air transportation reforms have been implemented at different times and unevenly across OECD countries and routes. Therefore, the variability of recent regulatory arrangements and market structures is large. To describe this variety of country and route-level situations and investigate their impact on efficiency in the provision of air services, it is useful to focus on a set of regulatory and market structure indicators for which comparative cross-country or cross-route information is available. Using a variety of sources (including the replies of OECD countries to an ad hoc questionnaire) a total of 21 indicators have been developed at the aggregate level for 27 OECD countries, and a total of 23 additional indicators at the micro level for a set of 102 air routes connecting 14 major international airports. 19 Underlying data for different indicators and countries generally refers to the 1996-1997 air travel season. Detailed information on sources and methodologies is provided in the Annex. 20

Three main areas were covered by the indicators: regulation (including government control), market structure and infrastructure access. Regulatory indicators focus on entry conditions (including for charters), pricing rules and government control (focusing on public ownership, subsidies and governance rights). Market structure indicators cover market concentration at the route and country levels, the presence of challenger and/or third party carriers, and the role of alliances. Indicators of infrastructure access conditions take into account both slot dominance and congestion. To make the analysis of differences in regulatory and market environments manageable, the large set of cross-country and cross-route indicators was summarised by means of statistical techniques (see Box 4). Box 4. Using multivariate data analysis to assess regulations and market structures Patterns of regulation and market structure in the air transportation industry are summarised by means of factor analysis. Factor analysis is a statistical technique aimed at finding the minimum number of "latent" variables which explain the maximum amount of the overall variance of the observed variables. The factors, which are linear combinations of the observed variables, can be interpreted in economic terms (e.g. entrenchment of a flag carrier or openness of international regulations). Each factor is characterised by a set of coefficients (factor loadings) expressing its correlation with the observed variables and the variables are assigned to the factor in which they are most "loaded". As a result, the regulatory and market structure indicators are split into disjoint sets, each of which is associated with one factor. The estimated factor loadings applied to the country-specific or route-specific regulation and market structure indicators make it possible to "score" countries and routes according to each of the factors, so that rankings of countries and routes can be obtained in terms of factor-specific scores. It is standard practice to retain a number of factors that cumulatively explain a substantial part of the overall variance. For more details on the use of multivariate data analysis techniques for studying patterns in regulation and market structure see Nicoletti et al. (1999).

The regulatory and market environment of scheduled air passenger transportation was measured along multiple dimensions (Table 1 ). Regulations included both domestic and international provisions (such as regional and/or open sky agreements) and government involvement in carriers' operations (such as government ownership, subsidies and public service obligations). The length of application of liberalisation policies, a proxy for the degree of maturation of their economic effects, was also taken into account. Regulatory information at the country level did not include pricing provisions, as these vary across routes. Market structure indicators included both the domestic and international dimensions, attempting to catch the overall competitive pressures impinging on national carriers. All indicators have been expressed on a decreasing (1-0) scale where competition-friendly regulations and market structures 19.

The data set includes the routes between the main national hub of 12 OECD countries to the main hub of the United States, Japan, Germany, France and United Kingdom, with route additions for the United States and Italy which have more than one national hub.

General information on the OECD International Regulation Database, which includes the data used in this paper, and on statistical methodology for aggregating basic data on individual regulatory provisions into summary indicators can be found in G. Nicoletti et al (1999). The database and its documentation are accessible on the OECD Website at http://www.oecd.org/subject/regdatabase/.

reflect in smaller indices. Details on how these regulatory and market dimensions were mapped into quantitative indicators are provided in the Annex.

[Table 1 . Country-level indicators of regulation and market structure]

Based on these regulatory and market structure indicators factor analysis identified three main discriminating factors, which were interpreted as: (i) the overall market environment, (ii) the entrenchment of a flag carrier, and (iii) the openness of international regulations (Table 2 ). Flag carrier entrenchment covers both the openness of domestic markets (to domestic carriers) and government control over a large-size incumbent carrier. A summary indicator of the country-level environment of air transport industries was computed as a weighted average of the score of each country on individual factors (weighted, according to the contribution of each factor to the overall variance of country-level characteristics). The summary indicator shows that the United States and, to a lesser extent, the United Kingdom and the Netherlands are the OECD countries offering the most liberal environment, and some Southern European and new Member countries the most protected ones (Figure 2 ). Country scores on individual factors are broadly consistent with the overall rankings, with the exception of smaller-size countries which may present a concentrated industry structure even when their regulations are relatively liberal (due to minimum efficient size effects) and the United States whose international regulations appear imperfectly open as this country does not participate to any integrated regional aviation market. Apart from the United States, where several equally-sized incumbents coexist, in virtually all countries a flag carrier is strongly entrenched in domestic and international markets.

Figures 3 and 4 present country clusters along "pairs of factorial axes". Figure 3 suggests that flag-carrier entrenchment is not necessarily synonymous with lack of competition. There is an important distinction between countries which have nurtured a dominant carrier within competitive market conditions (where one large-sized airline may thrive on economic grounds, such as in the United Kingdom or Japan) and countries in which national airlines operate in relatively closed markets (where the national carrier seem to dominate as a result of policy design). Figure 4 indicates that for a majority of countries, a liberal regulatory environment on international routes is associated with a relatively competitive overall market structure (and vice versa). However, there are sets of countries in which the international regulatory stance bears little relationship with market structure. A dose of competition may be sustained within protected industries (Japan, Mexico, Canada and Korea seem to present such a pattern of "managed competition"), while in some small countries international openness has not resulted in low market concentration (perhaps due to minimum size effects).

3.2

Regulation and market structure at the route-level

Differences in regulation and market structure at the route level are particularly strong in international air transport due to the overlap of domestic, bilateral and multilateral provisions. The main data source for regulation are bilateral air agreements applicable on the 102 routes included in the sample in (or around) 1996. These provided detailed information about regulations concerning: the designation of carriers authorised to service the routes, route capacity limitations, the setting of air fares and the authorisation of charter flights on the route. 21 The regulatory information has been supplemented with data on the combined market share of publicly-controlled carriers on each route (defined as those in which the government owns more than a third of the shares), as an indicator of the direct role of government. Market structure on each route has been analysed from the viewpoint of both the concentration of supply and its distribution between different categories of carriers. The taxonomy stresses different types and degrees of competitive pressures exerted by different types of airlines and market arrangements: "challengers" (or limited-size incumbents) are those which were not traditional incumbents on a route, entered relatively recently, and succeeded in carving out a minimum market share; "third party" carriers are those that do not belong to any of the countries signatories of the ASA on a particular route; airline alliances are captured by the combined route market share of the airlines participating in a strategic alliance. Finally, to measure infrastructure access conditions, data were included on the degree of congestion and slot concentration by incumbent carriers at route-end airports. 22 .

Table 3 summarises these route-level indicators. Following the same approach adopted for ranking countries, indicators are expressed on a decreasing (1-0) scale where more liberal regulatory and market arrangements generate smaller indices (see Annex for details).

[Table 3 . Route-level indicators of regulation, market structure and infrastructure access]

Route-level regulations and market structures were summarised by factor analysis. Government control patterns and airport characteristics, which are not truly route-specific but reflect national characteristics of the industry at route ends were maintained as stand-alone indicators. Four main factors explained most of the cross-route variance in the data (Table 4 ): i) route regulations; ii) route market structure; iii) the role of third-party carriers; and iv) the role of challenger airlines.

[Table 4 . Route-level regulation and market structure: the discriminating factors] 30.

Combining the "scores" of each route on the four factorial axes (weighted by the contribution of each factor to the variance of route characteristics) generates an indicator of "openness to competition", which summarises the regulatory and market environment on the route. City-pairs can then be characterised by "high competition", "limited competition" and "low competition" (Figure 5 ). 23 For illustrative purposes, the figure reports a selection of route names (their key can be found in the Annex). The scores of routes along the four individual axes are generally consistent with their overall ranking, with important exceptions: certain potentially competitive routes grant little room for third-party competition (such as several intra-European routes where access by carriers of a third country is still rare), and there are routes open to only moderate competition which nevertheless host significant third-party airlines (such as certain Asian routes where fifth traffic freedoms are available).

[Figure 5. Route characteristics]

Designation rules may provide for either single or multiple carriers, with or without route restrictions. Route capacity limitations may predetermine total capacity on the route, or stipulate free capacity with ex post monitoring by signatory governments (as originally in Bermuda I agreements between the United Kingdom and the United States). Price setting rules may provide for free pricing or define approval procedures by the signatory governments (double approval, country of origin approval, double disapproval).

It is particularly difficult to measure congestion empirically. For the purposes of this paper an airport was tagged as "congested" when it was reported as such to IATA by national authorities.

The classification was determined by looking for "breaks" between route clusters in the sample distribution of the indicators. The identification of breaks remains, however, somewhat arbitrary.

Evaluating the effects of regulatory reform on performance in air passenger transport

The effects of the regulatory and market environment on the efficiency of provision of air passenger services are analysed at both the country and route level. At the country level, the focus is on government control of domestic carriers and the openness and competitive stance of domestic and international markets. These features of the domestic industry are related to proxies for overall industry efficiency. Route-level analysis is richer and more pertinent because data availability makes it possible to concentrate on relevant markets (individual markets with little cross-substitutability of demand, where carriers are direct competitors), and the price dimension is also brought into the picture. Therefore, the effects of route market structure and route regulations on route efficiency and air fares in different market segments (such as leisure and business travel) are looked at in more detail: potentially important external influences, such as airport dominance at route ends, are taken into account, and the differential impact of several kinds of regulations (price-setting rules, charter rights) or market arrangements (airline alliances, presence of challenger carriers) is measured.

The econometric analysis followed a top-down approach. The effects of regulation and market structure on performance were first sought at the country-level and, subsequently, at the route-level. The analysis is based on ordinary least squares on a cross-section of 27 OECD countries or 100 major international routes in (or around) the 1996/1997 air travel season. 24 At both the country and route levels regulatory and market environments are summarised by the indicators described in the previous section. The summary indicators constructed by means of factor analysis are particularly appropriate for econometric analysis because they make it possible to specify parsimonious regression models, with synthetic explanatory variables that approximate well the cross-country variance originally present in the detailed regulatory and market structure data. Summary indicators are subsequently unbundled into their main components to check the differential impact of various regulatory and market arrangements on route performance.

Interactions between country and route-level regulatory and market influences are accounted for in the analysis. On the one hand country-level factors such as public ownership of carriers, propensity to travel by air or fleet structure can have an impact on airline performance on individual routes; on the other hand route-specific factors, such as competitive pressures faced by domestic carriers on individual international routes, can contribute to shape the organisation of the domestic industry (e.g. overall network features, technology choices, etc.). Empirically, the following interactions are accounted for: the combined effects of regulations concerning international routes on domestic industry efficiency; the combined effects of government control over route carriers on route efficiency and prices; the combined effects of the market environment faced by air transport industries in countries at route ends on route efficiency and prices; the combined effects of economic and fleet structure in industries at route ends on route efficiency and prices.

To facilitate the understanding of the empirical results, the discussion focuses on the distance of the country-level and route-level efficiency proxies from best practice. In this way, the performance and the regulatory and market environment variables are both cast on a decreasing scale. Higher values reflect low efficiency and restrictive environments, lower values reflect high efficiency and liberal environments. The estimated relationship between the indicators of performance and regulatory and market structure is therefore expected to be positive if competition is to have beneficial effects on the air travel industry.

In cross-route regressions, two Asian routes were dropped due to missing data.

At the country-level, efficiency of the air travel industry is measured in two ways. A simple (partial) measure is the average aggregate load factor 25 of the major domestic carriers (airlines carrying more than 400 000 passengers per year) on international routes. This can be taken to measure the capital productivity of the air travel industry on international markets served by domestic carriers. A more adequate proxy is the efficiency in the use of all factors of production. This is measured by means of Data Envelope Analysis, a non-parametric procedure that estimates a global measure of static efficiency in production accounting for multiple inputs and outputs. The present analysis has total passengers transported and total passenger-kilometres as outputs, and total personnel, capacity, fleet, fuel and average stage length as inputs (Box 5). The distance of country-level efficiency from best practice is measured by the reciprocals of the two efficiency proxies: the average aggregate inoccupancy rate (defined as the complement at unity of the average aggregate load factor) and the distance of the domestic industry from the efficiency frontier (defined as the percentage gap of the DEA-indicator of each country from that of best-practice countries).

Box 5. Measuring air transport efficiency by Data Envelope Analysis

Data Envelope Analysis (DEA) is a non-parametric procedure that estimates the distance of the input-output choices of a decision unit from the production efficiency frontier (so-called X-efficiency). 26 The estimation is based on observations concerning multiple inputs and outputs of different decision units. Using linear programming techniques efficiency is estimated by comparing the output(s) of the decision unit to its inputs. Each output and each input is assigned a weight and the ratio of weighted outputs to weighted inputs is maximised under a set of linear constraints, the maximand being the weights. The procedure can accommodate variable returns to scale and restrictions on outputs and inputs in order to avoid corner solutions (in which some outputs or inputs are not produced or used in the optimal program). 27 Comparisons between efficiencies of different units are made by rescaling their inputs (or outputs) and checking how much inputs (or outputs) of less efficient units should decrease (or increase) to make the unit efficient. The DEA methodology is particularly attractive for measuring efficiency in service sectors where production technologies and organisational forms vary widely, and inputs and outputs can be better approximated by a collection of physical indicators than by value added statistics.

Here, the method has been applied to analyse the relative efficiency of the air travel industries of OECD countries. Meaningful DEA efficiency comparisons require that the same technology is available to all decision units, which seems a reasonable assumption in cross-country comparisons of air travel industries. The production possibilities frontier is determined by the best performing countries, which dominate the other countries operating within the frontier. The data concerned around 100 major OECD carriers in 1996 and the DEA results suggest that the US, the UK and Japan were best practice countries, followed closely by the Netherlands and New Zealand, while efficiency was particularly low in the Eastern European countries, Austria, Switzerland and Belgium.

A useful starting point for the empirical analysis is to relate the summary indicator of the overall regulatory and market environment (see Figure 2 above) to industry efficiency across countries. Figure 6 shows that a strong positive correlation exists between this indicator and both the average aggregate inoccupancy factor and the distance from the production efficiency frontier. Thus, the air travel industries 25.

For each carrier, aggregate load factors (as opposed to route load factors) are defined as the percentage share of seats occupied per year in total aircraft seat capacity on international routes served by the carrier. The average aggregate load factor of the domestic industry is the weighted average of carriers' aggregate load factors.

For an extensive discussion of the concept and empirical relevance of x-efficiency, see Frantz (1997).

For a survey of developments in DEA methodology, see Charnes, A., et al. (1978), and the papers in The Journal of Productivity Analysis (1996).

of countries in which the regulatory and market environment is relatively friendly to competition appear to be more efficient than industries facing a more protected environment.

[Figure 6 . Industry efficiency and the regulatory and market environment] 37.

To explore further this relationship, a reduced form multivariate model was estimated on the cross-section of countries, relating the two measures of industry efficiency to their potential determinants (Table 5 ). The inoccupancy rate and the distance from the efficiency frontier were related to the regulatory and market structure indicators as well as to a set of control variables expressing economic structure and industry structure. Economic and industry structures were proxied by the propensity of the population to travel by air, the average age of the fleet and the average size of planes in the fleet. Air travel propensities and fleet characteristics can be assumed to be exogenous to industry efficiency at any point in time, but they may be related to the regulatory and market structure indicators due to the likely time-series correlation of regulatory and market arrangements. 28 To avoid excessive multicollinearity and given the few degrees of freedom available, a parsimonious specification was chosen. Policy and market influences were therefore proxied by the overall indicator of the regulatory and market environment (regression A) or by separate indicators for regulation and market structure (regression B).

The regression results suggest the presence of strong economies of scale and density related to market and aircraft size, respectively, and a significant impact of the market and regulatory environment on industry efficiency (Table 6 ): the more competitive this environment, the higher is industry efficiencyhowever measured. The market environment alone has a significant impact in both regressions, with more competition (at home and on international routes) being associated with higher efficiency. Regulations on domestic and international routes per se play a lesser role and their effect is significant only in regressions based on the DEA measure of efficiency.

[Table 6 . Performance of the airline industry at the country-level and the regulatory and market environment] Results of cross-country OLS regressions

At the route level, standard measures of performance are load factors and fares. Load factors express the efficiency in the use of aircrafts on each route. Following the same approach as for the country-level analysis, the distance of aircraft efficiency from best practice on the route is proxied by the complement at unity of the load factor (inoccupancy rate). Three types of fares referring to the 1998/99 air travel season were considered: business, standard economy and discount fares. Figure 7 suggests some positive correlation between the overall route-specific regulatory and market environment (i.e. the "openness to competition" indicator of Figure 5 ) and air fares in a cross-section of 100 routes connecting the main hubs of the 12 largest OECD countries. Competitive routes tend to be associated with lower fares. By contrast, no bi-variate correlation can be observed between the openness to competition indicator and inoccupancy rates. However, inoccupancy rates and fares are affected by a number of different factors and these need to be captured by multivariate analysis.

Standard empirical tests (based on variance inflation factors) did not provide evidence of this possible source of multicollinearity, which however may have biased downwards the significance of the coefficient estimates of the regulatory and market structure indicators.

[Figure 7 . Performance on routes and the regulatory and market environment] 40.

To put some structure on the specification of the multivariate model, air fares can be assumed to depend on marginal costs and mark-ups (for a similar specification, see Graham et al., 1983). Marginal costs depend on the costs of inputs and various kinds of route-specific and industry-level economies (or diseconomies) proper to air transport, while mark-ups mainly depend on route-specific factors --such as regulatory restrictions, competitive pressures, and airport conditions at route ends. A number of empirical proxies were chosen for these variables (Table 7 ). The costs of inputs were proxied by inoccupancy rates (productivity of capital), stage length (fuel) and an overall measure of the purchasing power of the currencies at route ends, which controls for exchange rate effects on input costs. 29 No reliable data were available for labour costs at the route level. Economies (or diseconomies) of scale were assumed to act through both capital productivity (see below) and other (unmeasured) cost components. Therefore, proxies for economies of scale (the propensity to travel by plane at route ends), economies of density (the average size of aircrafts in fleets at route ends) and diseconomies (difficulties of airport access, as reflected in the degree of slot concentration by incumbents and congestion) were included in the fare equation. Route regulations and market structures were summarised by the indicators described in the previous section. To account for the influence on mark-ups of the potential limitations to competition implied by government ownership of incumbents, a summary measure of the share of the city-pair market jointly held by government-controlled carriers was also included in the model.

Inoccupancy rates were assumed to depend on the structure of the fleet (average size and age of planes), on economies of scale, density and stage length, and on policy and market influences shaping the x-efficiency of carriers. Influences on x-efficiency operate mainly by affecting competitive pressures and the governance of firms (Frantz, 1997). In a network industry such as air transport, the efficient use of capital depends not only on competitive pressures at the route level but also, more generally, on pressures exerted at the industry level, which may enhance network design and the allocation of capital over the network. For this reason, inoccupancy rates are assumed to depend on both route-specific regulations and market structures and the overall market environment faced by carriers in home industries at route ends. The influence of governance on x-efficiency is proxied by the share of the city-pair market jointly held by publicly-controlled carriers.

These assumptions led to the following two equations for inoccupancy rates (IR) and fares (P):

where STAGE is the average stage length, u and v are stochastic disturbances and all other variables are defined as in Table 7 . Using equation [1] to substitute for IR in equation [2] leads to the following reduced-form equation for air fares:

Fares are expressed in US$ at current exchange rates and therefore are affected by deviations of national currencies from their PPP values.

Equations [1] and [3] were estimated by OLS based on the sample of international routes (see the Annex for details on data, sources and methodologies). To avoid endogeneity problems (e.g. between fares and propensity to travel by air) the data on fares refers to the 1998-99 air travel season, while the other data (load factors, economic and industry structure, regulation and market structure) refers to the 1996-97 air travel season. 30 Given current regulatory arrangements at the international level, the likelihood that a route is governed by restrictive bilateral air service agreements increases with stage length. Since this leads to strong collinearity between stage length and the regulatory and market indicators used in the analysis, it was decided to adjust both fares and load factors by stage length prior to estimation. Therefore, these variables were redefined in terms of deviations from values predicted by stage length to improve the quality of the regressions. The estimation strategy was to explore the impact of route-specific regulations and market structures (the REG i and MKT j variables) at increasing levels of disaggregation, looking first at the overall indicator of regulatory and market environment and next at its various components.

Table 8 presents the results of regressions for load factors and the three types of air fares when the regulatory and market environment on the route is summarised by means of the openness to competition indicator shown in Figure 5 above. In estimating model [1], three observations were dropped because they were identified as outliers by standard statistical procedures. In estimating model [3], the variables expressing average aircraft age and size at route ends have been omitted because they were highly collinear with other explanatory variables and their contribution to the model fit was insignificant. The included variables are jointly strongly significant, as implied by the F-tests. However, the fit of the models is relatively poor (adjusted R 2 s range from 0.2 to 0.4), reflecting the very high volatility of the price data and suggesting that some important route-specific effects are not captured by the explanatory variables.

[Table 8. Efficiency, fares and the regulatory and market environment: overall route-specific effects]

Results of cross-route OLS regressions 45.

In general, regression results suggest that (a) scale economies are a significant phenomenon in air travel; and (b) the effects of regulation and market structure (at both the route and industry levels) are at least as significant. The efficiency in the use of capital increases with average aircraft size and the size of the market, and as the route-specific and country-specific regulatory and market environment faced by route carriers becomes friendlier to competition. By contrast, there is no evidence of adverse effects on productive efficiency of airport conditions at route ends or public control of route carriers.

A route-specific environment friendly to competition also tends to reduce all categories of fares, with the strongest effects observed in the business segment. However, the various types of fares react somewhat differently to the other economic and policy factors. As the size of the market expands, airlines use scale economies for decreasing economy and discount fares, but at the same time business fares rise, increasing price discrimination on the route. Similarly, a competitive national market environment at route ends pushes up business fares, possibly because (mostly domestic) competitive pressures force airlines to shift the bulk of price discrimination onto international business travel, where competition is weaker. By contrast, competitive pressures on industries at route ends are the single most significant influence that reduces discount fares, possibly because these pressures force airlines to adopt better yield management 30.

If these variables are autocorrelated over time, the introduction of a time lag only partially avoids this possible source of endogeneity bias.

strategies in a segment of the market where demand is highly elastic and competition by charters may be vibrant. Fares also seem to react differently to airport conditions at route ends. Airport dominance and congestion appear to push up prices in time-sensitive business and standard economy travel (although at 10 per cent significance levels), while no effect can be detected on discount travel, which is not timesensitive. Finally, government control over route carriers tends to push up business fares, perhaps because (mostly public-owned) flag carriers are often more prone to compete for business travellers by increasing service quality, while it tends to reduce discount fares (also at 10 per cent significance levels).

To explore further the separate contributions of the market structure and regulation variables to explaining differences in performance across routes, Table 9 shows the estimates of models [1] and [3] when the regulation component (i.e. the first factorial axis) of the overall route-specific summary indicator is isolated. The market structure components (market structure, role of third party carriers and role of challenger carriers) are summarised by a single indicator obtained by weighting the corresponding route scores by the contribution of each factorial axis to the overall variance of the data.

[Table 9 . Efficiency, fares and the regulatory and market environment: separating the effects of route-specific regulation and market structure] Results of cross-route OLS regressions 48.

Overall the effects of the variables that are not route-specific are generally consistent with the previous regression, with the exception of the influence of the national market environment at route ends on inoccupancy rates, which is now captured by the route-specific market structure. The regression results suggest that a competitive route market structure is of the outmost importance for improving efficiency in the use of aircrafts, but there is an inverse significant relationship between the route-specific regulatory indicator and inoccupancy rates. Thus, relaxing route regulations appears to impact negatively on the efficiency of aircraft use. Perhaps this reflects the effect of strategic behaviour of incumbents, which react to liberalisation by increasing flight frequencies to pre-empt new entry on the route. At the same time, the inverse relationship between government control over route carriers and inoccupancy rates is more difficult to explain.

Fares appear to be affected only by the route regulatory environment and the national market environment at route ends. The summary indicator of route-specific market structure is insignificant in all fare regressions. This is consistent with the so-called "potential entry" hypothesis, whereby route liberalisation per se submits incumbent carriers to the competitive pressure of other carriers that could easily enter the route. Another possible interpretation is that the presence of several airlines on a route is effective in reducing fares only if route regulations allow competitive pressures to unfold.

The conjectures about the differential effects of regulation and market structure on productive efficiency and different types of fares can be partially checked by looking at the effects on performance of individual components of the regulatory and market environment. The specific role of different regulatory provisions and market characteristics is highlighted in Table 10 , which reports a selection of the results obtained in regressions that include the detailed indicators of regulation and market structure among the explanatory variables. Due to high collinearity between the detailed indicators, the strategy was to focus on the most significant regulatory and market influences on each of the performance measures. For brevity, the table omits the results concerning the variables expressing economic structure, which remained consistent with previous regressions. 31 To explore the potentially different impact of various combinations of regulation and market structure characteristics, two kinds of fare regressions were performed: (i) with 31.

Full regression results are available from the authors upon request.

separate regulatory and market structure indicators (regression A); and (ii) with an interaction term that allows for different effects of market structure in strict and lax regulatory environments (regression B). 32

[Table 10 . Efficiency, fares and the regulatory and market environment: exploring the effects of different regulatory and market conditions] Summary of results of cross-route OLS regressions 51.

While the efficiency of aircraft use is improved by both low concentration of capacity on the route and the presence of challenger airlines, the only significant regulatory influence is the extent of access rights for charters, with more extensive rights leading to lower capital efficiency. This result tends to corroborate the hypothesis that the negative relationship between liberalisation and efficiency in aircraft use is driven by the strategic reaction of incumbents to potential entry (such as charters). On routes in which the impact of regulatory reforms on market structure has matured and carrier competition is effective, raising rival costs and foreclosure (e.g. through the increase in flight frequencies) turn out to be more difficult strategies and incumbents must compete on costs, including through a better use of aircraft capacity.

Both business and economy fares are affected by fare regulations. By hindering price competition, pricing rules requiring the double approval of the governments involved in bilateral air service agreements or the approval of the country to which the carrier belongs are conducive to higher fares. At the same time, the route-specific market environment has an impact through airline alliances, which also tend to increase fares (especially standard ones) as the market share they cover on the route becomes larger. Apart from this effect, market structure appears to have no independent effect on fares. When the interaction between market structure and fare regulations is accounted for, the regressions for business fares provide only very weak evidence that a low capacity concentration on the route may have a sobering effect on fares when it is associated with a relatively liberal price setting environment. Thus, the conjecture that market structure can only affect business and economy fares when genuine price competition is possible and the coexistence between several airlines on a route is not "managed" by the signatories of bilateral or multilateral service agreements is not clearly supported by the data.

In the regressions allowing for specific regulatory and market structure effects, overall competitive pressures in countries at route ends continue to exert significant market influence on discount fares. However, route-specific regulations concerning access rights for charters and (to a lesser extent) the presence of challengers on the route also play a role (Regression A). Charters are the main competitors of incumbents in the market for leisure travel and, therefore, it is not surprising that liberal charter rights exert a downward pressure on discount fares. In addition, there is a close relationship between charter activity and the role of challengers on a route: challenger carriers may operate in the charter market in order to compete with incumbents and, conversely, charter companies sometimes establish themselves as challenger carriers over time. The existence of this relationship is confirmed by the results of the regression for discount fares that accounts for the interaction between the role of challengers and the extent of access rights for charters (Regression B). Challengers bring about a significant reduction in discount fares only when regulations concerning charters are relatively liberal.

On the whole, the route-level empirical results can be summarised as follows. Actual competition at route ends and on the route is essential for improving efficiency and lowering discount-fares. Potential competition is helpful in disciplining business and economy fares, but its effects are dampened as the role of airline alliances on the route increases. The effects of actual competition on discount and (to a lower 32.

For each kind of regulation and in each route, interaction terms were constructed by subdividing countries into a "liberal" group and a "protectionist" group. The effect of market structure in the two regulatory environments were isolated by using dummies identifying the two groups of countries.

extent) business fares are more significant when the regulatory environment encourages entry and price competition. 33 There is some evidence that difficulties in accessing infrastructures related to airport dominance and congestion, as well as government control over route carriers increases fares in time-sensitive market segments. However, government control tends to improve efficiency in aircraft use and moderate prices in market segments that are not time-sensitive.

Thus the results are consistent with those of Hurdle et al. (1989), who find that the presence of competitors reduces fares over and above the mere effect of potential entry in a study concentrating on liberalised U.S. routes. FEDERAL AVIATION ADMINISTRATION (1999), "Airport business practices and their impact on airline competition", Task Force Study, October.

International regulation:

airlines Participation in a regional single aviation market Number of major airlines (carrying more than 400 000 Maturation of the regional aviation market passengers per year)

Establishment of an 'Open Sky' air service agreement Market share of the largest carrier in the domestic market with the United States Market share of the largest carrier in the international market Maturation of the 'Open Sky' agreement Carrier concentration on domestic market (Herfindahl index)

Government control:

Carrier concentration on international market (Herfindahl index) Share of government in the equity capital of the largest Proportion of the 100 busiest international routes serviced national airline by more than 2 carriers Presence of a special government voting right (i.e. golden share) in a major national airline Government loss make-ups in airlines in the past five years Public service obligations of large national airlines Hub-and-spoke network 2 1. 9 cities connected by bilateral routes = 36 routes Total traffic = T, density of passengers per route = T/36 2. 9 cities connected by hub-and-spoke network = 8 routes Total traffic = T, density of passengers per route = T/4 (assuming that all passengers connect through a hub) Open (competitive) international regulatory environment and non-competitive market environment.

Restrictive (non-competitive) international regulations and non-competitive market environment.

Relatively open international regulations and relatively competitive market environment.

Restrictive international regulations and competitive market environment (managed competition).

Low competition

Role of third-party carriers

Role of challenger airlines Rrp (price regulation on the route): double approval = 3; country of origin approval = 2; double disapproval = 1; free = 0; Rrch (authorisation of charter flights on the route): no formal provision and traffic rights for charter services = 1; explicit provisions and traffic rights for charter services = 0

O(comp):

A composite "continuous" indicator which cumulates (from 0 to 4): = 4 if Ogs>50% (effective government control of the largest airline), = Ogs+Ggs+Ogd+Opso otherwise Ogs: Share of government in the equity capital of the largest airline (<1) Ggs: Presence of a special voting right (i.e. golden share) for government in a major airline (no = 0, yes = 1) Ogd: Government loss make-ups in major airlines in the past five years (no = 0, yes = 1) Opso: Formal public service obligations of the largest airline (no = 0, yes = 1) , Traffic: Commercial Air Carriers, 1993-1997, Series T57; and ICAO, On-Flight Origin and Destination Statistics, 1996, Series OFOD80. Capacity and traffic data apply to 1996 unless indicated otherwise. When applicable, market structure indicators have been calculated by consolidating the capacity and traffic of large size "flag carriers" with their subsidiary companies. Capacity and traffic on certain routes had to be estimated: the capacity of Air New Zealand on its routes was estimated as being equal to the capacity of its main competitor on each route. The capacity of continental on Milan-New York-Milan and the capacity of Air New Zealand on Sydney-Los Angeles-Sydney could not be estimated and were not taken into account. Domestic market concentration indexes of countries with no sizeable domestic markets (Belgium, Netherlands) were considered equal to their international market concentration indexes.

INTERNATIONAL REGULATORY FRAMEWORK

R2i: Openness of international market to competition: ((R2os+Tos)+(R2reg+Treg))/2: R2os: Existence of an 'Open Skies' air service agreement with the US (yes = 0, no = 1). R2reg: Existence of a Regional Single Aviation Market (yes = 0, no = 1, cabotage exception = 0.5) 35 Tos and Treg: maturation mark-ups (= 0 if R2os and R2reg established before 1993, = 1 if established after 1993, = 2 if no liberalisation yet) 36

ROUTE-LEVEL REGULATORY FRAMEWORK 37

Rrd (designation of carriers on the route): single designation = 3; multiple designation with route limitation = 2; multiple designation = 1; free entry (no designation) = 0; Rrc (capacity regulation on the route): predetermination = 3; hybrid = 2; Bermuda I = 1; free/no clause = 0;

34. EU countries' domestic markets were not considered fully deregulated until 1997 (entry into force of cabotage freedoms within the European single aviation market). For econometric tests (not in the database) in a case (Australia) where domestic deregulation did not lead to a change in market structure R2d was established at 1.

Norway is considered entirely integrated to the EU single aviation market, via its flag carrier SAS. Switzerland, Iceland, Czech Republic, Hungary and Poland have not been considered as full participants to the single European aviation market. R2reg for Portugal and Greece are established at 0.5 because of cabotage exceptions to single aviation market in 1997 (Açores and Agean Islands).

When a long-term commitment to Open Sky policy is demonstrated, Tos is established at 0 for economic tests. This is the case of New Zealand which signed several Open Sky agreements in the 1990s (with Singapore, Malaysia, Brunei and United Arab Emirates) and "Open Sky plus" (a unique Open Sky agreement including 7 th right freedoms) with the United States in 1997.

Route-level regulation indicators have been established on the basis of International Civil Aviation Organisation (ICAO): Digest of Bilateral Air Service Agreements, 1998 edition and 1995 update. When information on post-1995 changes to bilateral agreements were available, this recent data has been used. For bilateral aviation relations of European countries, the provisions of the Third Aviation Package have been used (1997).

OPENNESS OF INTERNATIONAL MARKET TO COMPETITION

Mitr: 1-share of the 100 busiest international routes serviced by more than 3 carriers (<1) Mhei: Carrier concentration on international market (International Herfindahl)

INFLUENCE OF FLAG CARRIERS

Mdsm: Market share of largest (national) carrier in domestic market (<1) Mism: Market share of largest (national) carrier in international market (<1)

Route level 39

OPENNESS OF ROUTE MARKET TO COMPETITION

Mrn: 1/number of main carriers on the route (with >5% market share) (<1) Mrsm: capacity share of the largest carrier on the route (<1) Mrhf: Herfindahl index of capacity concentration on the route (<1)

Airline alliances are a new form of market organisation. However, there is no formal definition of alliances and no formal lists. Certain carriers are involved in more than one alliance, it is then difficult to locate them in a given strategic family. For the purposes of this study, four main alliances have been distinguished: i)

The "Star" group: United Airlines, Lufthansa, SAS, Air Canada, Thai, Varig, Air New Zealand, Singapore Airlines, All Nippon Airways; ii) The "OneWorld" group: American Airlines, British Airways, Qantas, Iberia, Finnair, US Airways, Japan Airlines; iii) The "Delta" constellation (formerly "Qualiflyer"): Delta Airlines, Swissair, Sabena, Austrian Airlines, Turkish Airlines, Air France (the arrival of Air France may subsequently change the structure of this group); iv) The "KLM/Northwest" group: Northwestern Airlines, KLM, Alitalia, Braathens, Continental Airlines, Japan Air System (this group also manifests signs of recomposition in 2000).

AIRPORT CONGESTION ON THE ROUTE 41

Ardc: reported congestion in departure airport (no = 0, yes = 1) Arac: reported congestion in arrival airport (no = 0, yes = 1);

AIRPORT SLOT CONCENTRATION ON THE ROUTE 42

Ards: largest carrier's share of slots in the departure airport (<1) Aras: largest carrier's share of slots in arrival airport (<1)

National level

NFScomp: Overall regulatory and market structure score of countries according to factor analysis NFS1: Score of countries on axis 1 (Market structure) NFS2: Score of countries on axis 2 (Entrenchment of flag carriers) NFS3: Score of countries on axis 3 (International regulations)

Route level

NRFScomp: Overall regulatory and market structure score of air routes according to factor analysis NRFS1: Score of air routes on axis 1 (Route regulations) NRFS2: Score of air routes on axis 2 (Route market structures) NRFS3: Score of air routes on axis 3 (Third party presence) NRFS4: Score of air routes on axis 4 (Presence of challenger airlines)

PRODUCTIVE EFFICIENCY

P(DEA-s2) = The percentage score-gap of each country from the most DEA-efficient country(ies) (adjusted for average stage length, defined as simple average stage length of domestic carriers).

P(il):

International inoccupancy rate: 100-average % international load factor.

Airport congestion has been identified on the basis of a list of airports reported as "congested" to the International air Transport Association. There may be a reporting bias in these reports as there is no formal definition and formal test congestion.

Data on slot concentration have been extracted from multiple country-specific and international sources. They refer to the share of the largest carrier in the total number of departure flights from an airport. There are differences in definition and in some periods -in certain airports only the concentration of international flights are reported. Slot share may be underestimated in certain airports when there is double-counting due to code-share flights.

PRICE PERFORMANCE 43

PPrb: rate of deviation of announced business fare (-or +) from benchmark (>-

Failed to load PDF:

StripeM-Inner

Paper sources

Abstract screening pilot

Abstract screening results

Extraction pilot

Extraction results

Research report

Modify setup