Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effect of governance structures on airport efficiency performance – the North American case Qi, Zhao 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_spring_zhao_qi.pdf [ 1.27MB ]
Metadata
JSON: 24-1.0071624.json
JSON-LD: 24-1.0071624-ld.json
RDF/XML (Pretty): 24-1.0071624-rdf.xml
RDF/JSON: 24-1.0071624-rdf.json
Turtle: 24-1.0071624-turtle.txt
N-Triples: 24-1.0071624-rdf-ntriples.txt
Original Record: 24-1.0071624-source.json
Full Text
24-1.0071624-fulltext.txt
Citation
24-1.0071624.ris

Full Text

The Effect of Governance Structures on Airport Efficiency Performance – The North American Case  by  QI ZHAO  B.A. in International Economics and Trade, Zhejiang University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE IN BUSINESS ADMINISTRATION in  THE FACULTY OF GRADUATE STUDIES  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) March 2011  © Qi Zhao 2011  Abstract Over the last two decades, there have been widespread moves to corporatize, privatize, and deregulate airports around the world. These changes have created a great diversity of airport ownership and governance structures. Against this backdrop, this paper applies a stochastic cost frontier model to examine how the two dominant governance forms of publically owned airports in North America, namely operation and governance by a government branch and by an airport authority, affect airport efficiency performance. The data for this study is taken over the 20022008 period from 54 airports in Canada and the US and provided for this thesis in confidence by the ATRS Global Airport Performance Benchmarking Project. This study sets out to prove that these two types of governance structures can have significant effects on the efficiency performance of airports in North America, with the results showing that (1) the airports operated by an airport authority achieve higher cost efficiency than those operated by a government branch; and (2) the airports operated by a government branch tend to have lower labour share than those operated by an airport authority. Moreover, by separating Canadian and US airport authorities, our study also attempts to determine whether Canadian and US airport authorities differ in their impact on airport (cost) efficiency performance and hence should be considered as different types of airport governance. However, our regression models have not discerned there is any statistically significant difference as to the efficiency performance between airports operated by US and Canadian airport authorities. It seems therefore that US and Canadian airport authorities are similar in nature and should not be considered as different types of airport governance.  ii  Table of Contents Abstract...................................................................................................................... ii Table of Contents................................................................................................ iii List of Tables............................................................................................................ iv List of Figures ......................................................................................................... v Acknowledgements ........................................................................................... vi 1  Introduction ............................................................................................................................ 1 1.1 Literature Review ............................................................................................................. 3 1.1.1  Debate over Ownership Efficiency Problem in a Broad Economic Context ........... 3  1.1.2  Efficiency and Productivity of Airports: Different Methodologies .......................... 4  1.1.3 Determinants of Airport Productivity and Efficiency: Ownership and Governance Structures ............................................................................................................................... 6 1.2 Scope and Objective ......................................................................................................... 9 2  Background Information on Airport Governance Structures ............................................ 11 2.1 Overview of Airport Governance Structures .................................................................. 11 2.2 Airport Governance in Canada ....................................................................................... 12 2.3 Airport Governance in the United States ........................................................................ 14 2.4 The Effects of Governance Structure on Airport Performance ...................................... 17 2.5 Conclusion ...................................................................................................................... 25  3  The Econometric Model ....................................................................................................... 26 3.1 Model of Airport Performance ....................................................................................... 26 3.2 The Framework of the Stochastic Cost Frontier Model ................................................. 28 3.3 The Specifications of the Stochastic Cost Frontier Model ............................................. 29 3.3.1  Specification of the Cost Frontier ........................................................................... 29  3.3.2  Effects of Governance Structures on Airport Cost Efficiency................................ 32  3.4 Conclusion ...................................................................................................................... 37 4  Sample and Variables ........................................................................................................... 39 4.1 Variable Construction ..................................................................................................... 39 4.1.1  Outputs and Inputs .................................................................................................. 39  4.2 Characteristics of Sample Airports ................................................................................. 41 4.2.1  Summary Statistics of the Airports Operated by the Government Branch ............. 42 iii  4.2.2  Summary Statistics of the Airports Operated by the Airport Authority ................. 43  4.3 Conclusion ...................................................................................................................... 45 5  Empirical Results and Discussion ....................................................................................... 47 5.1 Model Issues and Alternative Considerations ................................................................ 47 5.2 Empirical Results and Discussion .................................................................................. 51 5.2.1  The Effects of Airport Characteristics on Cost Frontier ......................................... 51  5.2.2  The Effects of Airport Governance Structures ....................................................... 59  5.3 Hypotheses Tests for Effects of Airport Governance Structures .................................... 63 5.4 Conclusion ...................................................................................................................... 67 6  Conclusion ............................................................................................................................ 68 6.1 Summary of Key Findings .............................................................................................. 68 6.1.1  Effects of Other Airport Characteristics ................................................................. 68  6.1.2  Effects of Airport Governance Structures............................................................... 69  6.2 Suggestions for Further Research ................................................................................... 69 6.3 Conclusion ...................................................................................................................... 70 Bibliography ................................................................................................................................. 71 Appendices.................................................................................................................................... 77 Appendix A: Classification of Airport Governance Structures ............................................. 77 Appendix B: The Sample Airports ......................................................................................... 78  iv  List of Tables Table 1 Effects of Governance Structure on Airport Operating Total Cost ................................. 21 Table 2 Effects of Governance Structure on Airport Non-aeronautical Revenue ........................ 22 Table 3 Effects of Governance Structure on Shares of Non-aeronautical Revenues ................... 22 Table 4 Effects of Governance Structure on Airport Labour Share ............................................. 23 Table 5 Effects of Governance Structures on Airport Labour Price ............................................. 24 Table 6 Summary of Average Statistics -- Airports Operated by Government Branch ............... 43 Table 7 Summary of AverageStatistics -- Airports Operated by Airport Authority..................... 45 Table 8 Explanatory Variables and Regression Model for Stochastic Frontier Analysis ............. 50 Table 9 Stochastic Cost Frontier Model A Results ....................................................................... 52 Table 10 Stochastic Cost Frontier Model B Results ..................................................................... 53 Table 11 Stochastic Cost Frontier Model C Results ..................................................................... 54 Table 12 Stochastic Cost Frontier Model D Results ..................................................................... 55 Table 13 Hypothesis Tests for Stochastic Frontier Model A ........................................................ 65 Table 14 Hypothesis Tests for Stochastic Frontier Model B ........................................................ 65 Table 15 Hypothesis Tests for Stochastic Frontier Model C ........................................................ 65 Table 16 Hypothesis Tests for Stochastic Frontier Model D ........................................................ 65  v  List of Figures Figure 1 Non-aeronautical Revenue for Sample Airports in 2008 ............................................... 56 Figure 2 Share of Non-aeronautical Revenue for Sample Airports in 2008 ................................. 56 Figure 3 Labour Cost Share for Airports Operated by Government Branch in 2008 ................... 60 Figure 4 Labour Cost Share for Airports Operated by Airport Authority in 2008 ....................... 61  vi  Acknowledgements I would like to express my sincerest gratitude to my supervisor, Professor Tae H. Oum, for all of his support and encouragement during my master studies and throughout all stages of this thesis. This thesis would not be completed without his guidance and support. I would like to extend my appreciation to Professor Anming Zhang for serving on my committee and for his continuous encouragement and help during my years at UBC. I would like to thank Professor Chunyan Yu for serving on my committee and for her invaluable advice along the way. Special thanks to Professor David Gillen for serving as my university examiner and his great experience and support for my thesis. I would also like to thank Prof. Keth Head, Prof. Vadim Marmer and Prof. Shinichi Sakata for their help and advice during my study. In addition, I will be always grateful to Zian Zhao and Natalie Anderson for their attention and help during my study. Special thanks to my friend, Nancy Roberts, for helping me go through the darkest time in my life.  vii  1  Introduction  As the most important infrastructure for air transport systems, airports play a critical role in both the modern economy and society. From the earliest years of airport transport system development, it was natural for governments to own and operate airports around the world. Therefore, public ownership and governance of airports were not questioned throughout most of the history of aviation. Even though economists started expressing concerns about the overall inefficiency of air transport systems in the 1970s, the initial efforts focused more on the deregulation and privatization of airlines than on airports or air traffic control systems. Their reasons for focusing on airlines over airports underline the reality that more efficiency gains could be made from the former than the latter. However, as deregulation and privatization of the airline industry and its markets have become the norm, attention has slowly been shifting to questioning the efficiency of airports and air traffic control systems that have been operating in a near monopoly market for a long time. Since the British government privatized their airports in 1987, a great diversity has emerged in the ways which countries tackle the airport ownership and governance issues. In Europe and Australia, lots of airports have been fully privatized and governed under different regulatory structures. However, in Canada and the United States (henceforth referred to as ―North America‖ for convenience), the majority of commercial airports are still publicly owned , and operated either by a government branch (mostly city or state governments) or an airport authority that is also viewed as a quasi-governmental organization. Despite the uniqueness of the North American airport governance structures, few studies have investigated on their effects on airport efficiency performance. Many insightful studies, 1  such as that of Oum, Zhang and Zhang (2004), are devoted to examining the effects of ownership forms and regulatory policies on airport efficiency. Nonetheless, by far the most extensively studied influence on airport efficiency performance has been that of ownership forms rather than governance structures. Indeed, there has been little discussion about the way in which different governance structures affect airport efficiency or productivity performance. Furthermore, very few studies have focused solely on measuring and comparing the effects of airport governance forms on the efficiency performance of North American airports. To fill theses gaps, this study attempts to examine how the two dominant types of governance forms (i.e., a government branch vs. an airport authority) affect the cost efficiency performance of airports in North America. We apply a stochastic cost frontier model to investigate the effects of these two types of airport governance structures, not only on technical inefficiency but also on allocative inefficiency, in particular, variable input usage. This study sets out to prove that these two types of governance structures can have significant effects on the efficiency performance of airports in North America, with the results showing that (1) the airports operated by an airport authority achieve higher cost efficiency than those operated by a government branch; and (2) the airports operated by a government branch tend to have lower labour share than those operated by an airport authority. Moreover, by separating Canadian and US airport authorities, our study also attempts to determine whether Canadian and US airport authorities differ in their impact on airport (cost) efficiency performance and hence should be considered as different types of airport governance. However, our regression models have not discerned there is any statistically significant difference as to the efficiency performance between airports operated by US and Canadian airport authorities. It seems therefore that US and Canadian airport authorities are similar in nature and should not be considered as different types of airport governance. 2  To set the stage, this chapter starts off an overview of the academic discussion about the relationship between ownership reform and productivity or efficiency performance. Based on recent economic literature, we summarize the impact of ownership and governance structure on productivity and efficiency in the airport sector. In section 2, we outline the scope and objective of our study. 1.1  Literature Review  It seems that various countries have tackled the airport ownership and governance issue in a wide variety of ways. How do such different packages adopted by countries affect airport efficiency? At first glance, it seems a relatively straightforward question, but the answer is neither simple nor straightforward, and any attempt to answer it almost immediately reveals a history of prolonged controversy. 1.1.1 Debate over Ownership -- Efficiency Problem in a Broad Economic Context  The 1970s saw the resurgence of the notion that transferring assets from the public to the private sector would raise both allocative and technical efficiency, leading to greater economic wellbeing for all. Some early economic studies enthusiastically reported that the evidence stood in favour of private-sector performance, as found, for instance, by De Alessi (1980) and Bennert and Johnson (1980). Later studies, however, have been more cautious. In their widely quoted 1983 survey, Millward and Park (1983) arrive at the conclusion that there is no systematic difference in performance under public or private ownership. Further, by assuming that both public and private firms implement the optimal contract, De Frajia (1993) demonstrates that public ownership yields a higher level of productive efficiency than private ownership. In addition, many scholars also point out that complementarities exist between the optimal choice 3  for the form of privatization and the design of the legal framework within which privatized firms will operate. Indeed, the evidence provided by Vickers and Yarrow (1988) casts doubt on whether regulating a private firm by strictly defined rules leads to better performance efficiency than the internal regulation that characterizes public ownership.  Meanwhile, the presence of mixed private and public ownership only further complicates the debate over the relationship between ownership and efficiency. Although several interesting properties of mixed enterprises have been revealed, the answer to the impact of mixed ownership is not obvious and has not fully been settled. For instance, in their 1986 survey, Boardman, Eckel and Vining (1986) argue that mixed enterprises tend to be more efficient in production than public enterprises, but less profitable than those totally owned by the private sector. However, by solely focusing on firm’s profitability, Boardman and Vining (1989) arrive at somewhat different conclusion and suggest that the mixed enterprises perform no better and, in fact, often worse than the private or public enterprises. 1.1.2 Efficiency and Productivity of Airports: Different Methodologies  Prior to 1990, there was relatively little attention paid to questions of airport performance, in terms of productive efficiency. In the latter half of the 1990s, however, there has been a small boom in modeling airport efficiency. Different methodologies have been proposed to provide an adequate measure of airport performance. These methods can be broadly classified into nonparametric and parametric. Non-parametric methods include indexes of partial and total factors productivity (TFP), and data envelopment analysis (DEA). Parametric methods involve the estimation of neoclassical and stochastic cost or production function. At least in part, these two methods can be reconciled, and it is important to do so. 4  Thus far, much of the work which has been done on airport efficiency and productivity, such as that of Doganis (1992), has taken the form of partial productivity measures. Partial productivity measures are useful to compare performance across airports operating in similar operating environments or over time within an airport when the operating environment and input prices remain relatively stable. However, partial productivity measures related an airport’s output to a single input factor. The productivity of any one input depends on the level of other inputs being used; high productivity performance in one input may come at the expense of low productivity of other inputs. Therefore, it is desirable to have a more comprehensive measure of productivity in order to make more confident assessments of airport performance.  As an alternative to construct indexes of partial productivity, a Total Factor Productivity (TFP) index has been widely used in measuring airport productivity performance. It is noticeable, however, that a TFP index is a ratio of a total (aggregate) output quantity index to a total (aggregate) input quantity index. The TFP index, itself, only yields a ―gross‖ measure of productivity changes. It does not distinguish among sources of productivity growth. Thus, in order to make inference about airport efficiency it is necessary to separate out the influences, such as operating environments and scale of outputs, on the ―gross‖ measure of TFP. Hooper and Hensher (1997), Nyshadham and Rao (2002), and Yoshida (2004) have used regression analysis to decompose a TFP index and further investigate the productivity efficiency of airports from different regions. Nonetheless, since the TFP method requires detailed specifications on both an airport’s outputs and inputs, the lack of accurate and consistent capital input measures1 inevitably limits  1  Since different countries often have different accounting convention and construct official data in different ways, it is nearly impossible to compile accurate and consistent capital input measures. 5  the application of this method, as the results of its analyses of airport performance over time and across countries are unreliable. To overcome this problem, the Air Transport Research Society (ATRS) proposes a Variable Factor Productivity (VFP) method, which only requires the data on variable input factors. This method has been applied to measuring the level of productivity in ATRS annual airport benchmarking reports. Further, many researchers, such as Cooper and Gillen (1994), Gillen and Lall (1997, 2001), Sarkis (2000) and Barros and Dieke (2007), have also applied a Data Envelopment Analysis (DEA) method to evaluate the efficiency performance of airports under different circumstances.  More recently, a Stochastic Frontier Analysis (SFA) has been incorporated to analyze airport efficiency performance. As an econometric technique, SFA specifies the underlying production/cost function when conducting efficiency analysis. In contrast to the conventional econometric approaches, SFA introduces an additional one-side error term, alongside the traditional symmetric noise term, to capture unexplained inefficiency in the production/cost frontier. Barros (2008) and Martin et al. (2009) have used this method (SFA) to investigate the cost efficiency performance of Spanish airports. 1.1.3 Determinants of Airport Productivity and Efficiency: Ownership and Governance Structures  From the 1990s on, the momentum to privatize airports has been gaining strength throughout most of the world. These changes in airport ownership are usually accompanied by explicit price regulations. This subject at hand has attracted the attention of a wide range of researchers, who focus on the effects of ownership and associated regulations on airport efficiency performance. There are many fine studies already on this topic: for the UK, Beesley (1999) and Starkie (2001);  6  for Australia, Forsyth (1997, 2002) and Hooper et al (2000). Nevertheless, these studies have chosen a geographical structure, which highlights diversity rather than drawing out unifying themes. A highly influential study, by Oum, A. Zhang and Y. Zhang (2004), investigates 60 airports worldwide under different ownership forms and provides both theoretical and empirical evidence on the impact of different economic regulations on airport efficiency performance. Their results support the argument that dual-till regulation would be more economically efficient than the single-till regulation, especially for large, busy airports. Moreover, their empirical results indicate that privately owned airports do not necessarily achieve higher capital input productivity or total factor productivity than publicly owned airports do. However, it is not only ownership and its associated regulations that determine performance but also airport governance structures themselves can have significant effects on performance. In his insightful work, Gillen (2010) not only adopts a descriptive approach to examine the evolution of airport governance but also proposes two-sided platforms2 to consider airport governance structure issues. Apart from this work, a limited number of studies have explicitly analyzed the impact of governance structures on airport efficiency performance. Moreover, it remains inconclusive as to how different governance structures affect the efficiency and productivity of North American airports. Although most empirical evidence suggests that airports operated by a port authority are less efficient than those run by either a government branch or an airport authority, it has not been fully determined how the airports operated by the latter two governance structures differ in their efficiency and productivity performances.  2  This is a new study, with its initial stream of articles emerging in 2003. Some economists would argue that twosided platforms (airports and airlines) internalize usage externalities that agents cannot internalize efficiently Rochet and Tirole (2005) define two sided markets as a situation in which the volume of transactions between end-users depends on the structure and not only on the overall level of the fees charged by the platform (Gillen 2010). 7  On the one hand, some studies argue that the difference in efficiency between airports operated by these two governance structures (i.e. a government branch vs. an airport authority) is negligible in North America. For instance, by using the VFP method to measure the productivity, the study by Oum, Adler and Yu (2006) examines the impact of six different governance structures3 on the productivity performance of 116 airports worldwide. One of their finding suggests that, in terms of productivity of North American airports, there is no significant difference between the airports operated by an airport authority and airports operated by a government branch. On the other hand, other studies, such as Oum, Yan and Yu (2008) and Craig, Airola and Tipu (2005) detect a better performance from the airports operated by an airport authority. Oum, Yan and Yu (2008) compare the cost efficiency among 109 airports worldwide with respect to seven different airport governance structures4. By estimating a stochastic cost frontier via a Bayesian approach, they argue that the airports run by an airport authority perform far more efficiently than those operated by a government branch in North America. Based Solely on US airport data, Craig, Airola and Tipu (2005) find consistent results as Oum et al (2008) that airports operated by airport authorities outperform those operated solely by the government branches in term of technical efficiency. Nevertheless, the omission of airport non-aeronautical service may bias the results from Craig et al (2005), since the majority of US airports generate a high proportion of their total revenues from non-aeronautical services.  3 4  See appendix A See appendix A 8  1.2  Scope and Objective  While it is clear that there has been a worldwide move towards privatizing airports over the last two decades, not all countries have joined this push towards privatization; the United States and Canada have done little in the way of privatization and keep their airport sectors mainly under government ownership. Given the recent growth in the privatization of airports, it is perhaps more interesting to ask how airports are governed and operated in countries in which privatization has not been undertaken. Also, the literature on airport governance structures provides few concrete, quantitative analyses of how different governance structures affect airport efficiency or productivity. Especially in North America, there has been no consensus, so far, as to which airport governance structure is better to foster airport efficient performance. Therefore, this paper seeks to present new empirical evidence on the way in which two different governance structures—a government branch and an airport authority—affect the cost efficiency performances in North American airports. To achieve this objective, we apply a stochastic cost frontier model to an unbalanced panel of 54 airports over 2002-2008. In particular, our model will not only capture unobserved airport inefficiency, but also examine whether different airport governance structures can explain them. Moreover, we argued that governance structures will exert influence on airport operation or production in a non-neutrally input-augmenting fashion and thus employed a parametric method to measure the impact of governance structures on airport variable input usage. To achieve these objectives, the structure of this paper is as follows: In Chapter 2, we begin with a short section on reviewing the evolution of airport ownership and governance around the world. Then, we turn on exploring fundamental characteristics of airport governance structures adopted in North America. Chapter 3 starts by comparing and contrasting different 9  models used for measuring airport efficiency performance. After explaining motivations for using stochastic frontier model, this chapter further discusses the specification of our stochastic cost frontier model. Chapter 3 contains a summary of the data, where the sample airports are categorized into two groups: those operated by a government branch and those run by an airport authority. We present the empirical results and a discussion of our findings in Chapter 4. Finally, Chapter 5 summarizes our study and considers some potential follow-up studies.  10  2  Background Information on Airport Governance Structures  It is not an easy question to determine whether the form of governance makes a significant difference in achieving airport operational efficiency in North America. This question is complicated because: (1) the definition of airport governance varies across countries; (2) it remains a challenge to provide an adequate measure of airport efficiency; (3) airport performance is the result of multiple factors, governance structure being one of the factors. Since the following two chapters will provide detailed answers to the latter two questions, we here attempt to develop a basis of understanding the evolution of airport governance and the ways in which different airport governance types may influence efficiency performance of North American airports. This kind of scrutiny provides a factual profile of airport governance forms in North America and fosters the identification and discussion of the most crucial matters: how do governance structures affect airport efficiency performance in North America. 2.1  Overview of Airport Governance Structures  In the 1980s, long after a wave of privatization swept through utility industries around the world, policy-makers began to turn their attention to reforming airport governance. At the time, most airports round the world were owned and operated by the public sector. One potential catalyst for airport reform in the 1970s and 1980s was the dramatic growth in air travel brought about by deregulation in the airline industry in North America and elsewhere. Rising passenger demand led to airport congestion and the need to invest in additional capacity, and, more importantly, to increase the productivity of airports. To respond to these needs, there was an extensive attempt to alter the governance framework in which airports operate. Different regimes adopted by countries (Oum, Adler and 11  Yu, 2006, and Oum, Yan and Yu, 2008) can incorporate different degrees of private-sector involvement, ranging from a complete sell-off of public airports to private investors (privatization) to simply demanding more self-sufficiency with respect to public services. The move towards full privatization has been strongest in the UK and, later, Australia and New Zealand. In continental Europe, there has been a preference for partial privatization, with the public sectors remaining with majority ownership. Surprisingly, with all of the changes brought about by consolidation and restructuring of the airline industry, Canada and the United States (henceforth referred to as ―North America‖ for convenience) are closer to the opposite end of the spectrum. Even though there is a long tradition of privately owned utilities and transport industries, virtually all major airports in North America are publicly owned. What distinguishes North American airports is, therefore, not their form of ownership but their unique governance structures. 2.2  Airport Governance in Canada  From the 1960s through the 1980s, airports in Canada were governed by the Canadian Air Transportation Administration (CATA), a division of Transport Canada. In the 1970s, growth in air transport together with technological changes placed substantial stress on the airport system. Under the existing airport governance framework, public-sector managers had little incentive to increase revenues or improve cost efficiency. To promote commercialization and efficiency, the initiative to reform Canadian airports was first taken in 1987. The federal policy, ―A Future Framework for Airports in Canada‖, allowed provincial, regional or local authorities to assume financial responsibility for airports, and directly manage and operate airports by virtue of a longterm ground lease drafted by Transport Canada. As a result, ―local airport authorities‖ (LAAs) were established in Montreal, Calgary, Vancouver and Edmonton in 1992. 12  The Canadian government further advanced airport governance reform with the introduction of the National Airports Policy in 1994. Under this policy, 64 regional airports and 30 small airports have sold to their communities, usually for a nominal amount. 8 arctic airports have been transferred to provincial or territorial governments, whereas 13 remote airports are remained under federal government operation. Most importantly, the National Airports Policy guarantees the transfer of responsibility for the operation, and management of 17 large national airports to ―Canadian Airport Authorities‖ (CAAs) on long-term leases. The leased LAA and CAA airports comprise 22 airports that link Canada from coast to coast and internationally. Currently, these 22 airports serve 90 per cent of all scheduled passenger and cargo traffic in Canada and are the points of origin and destination for almost all interprovincial and international air service in Canada. Given their importance to the country and society, our following discussion centers on the governance regime under which both the LAAs and CAAs operate. In Canada, the concept of airport authority governance is that of a private sector corporation which operates an airport. What differentiates an airport authority from the private corporation is that an airport authority is not-for-profit and thus has no shareholders. Between the LAAs and CAAs, there is a slight difference in terms of public accountability provisions 5. The CAAs must publish 60-days’ advance notice and justification for price increases in the local media, while Transport Canada may audit the LAAs’ records and procedures at any time and review the LAAs’ performance every five years. Despite of this, both LAAs and CAAs are private, self-financing, not-for-profit, non-share-capital corporate entities that do not pay income  5  Canadian federal government has examined the principles under which the local airport authorities (LAAs) were created and has revised them in the areas relating to the CAA's accountability to the communities they serve. 13  tax. Their leases on Canadian federal infrastructure are for 60 years with an option to renew for an additional 20 years. Although some business practices are controlled through the lease document, the LAAs and CAAs are not subject to economic regulation through legislation. Furthermore, the LAAs and CAAs enjoy the freedom to set up the prices for various airport activities (e.g., parking, rent, landing aircraft, terminal use, etc.) and determine service levels within the safety regulatory framework. In addition, the LAAs and CAAs operate airports with virtually no federal assistance or subsidy. To the contrary, with required ground lease payments, the Canadian airport authorities have become a source of significant general treasury revenues for the federal government. 2.3  Airport Governance in the United States  Governance and ownership of airports is actually quite complicated in the US, as it can differ for each state. Therefore, we do not attempt to draw sweeping conclusions, but seek to depict general pattern that has emerged in the types of airport governance in the US. Although there is no single path by which US airports came to their present governance form, state governments have exerted a strong influence on restructuring and forming airport ownership and governance in the United States. In the first part of the 20th century, state governments in the US began enacting legal statues that explicitly authorized local governments to establish and operate airports. These statutes promoted a wave of lawsuit from nearby communities, from different government subdivisions competing for control of the new airports, and from taxpayers challenging the use of tax revenues to build and operate airports. By and  14  large, the statues survived these challenges. Nowadays, with few exceptions6, majority of commercial airports in the US are owned by local governments. As state governments experimented with different forms of airport governance, state legislatures enabled state or local governments to establish their own aviation departments. Moreover, these legislatures either created individual airport authorities, such as State of Michigan and Minnesota, or authorized local governments to create individual airport authorities. As a result, these two governance types have become the most common forms of airport governance structure in the United States. The most recent ACI-NA survey, conducted in 2003 primarily among larger airports, has revealed that more than 60% of the airports responding to the survey were operated either by a local/regional government branch or by an airport authority. The rest of our discussion in this section therefore will focus on these two dominant types of airport governance structures (i.e., a government branch vs. an airport authority). It has long been the tradition that airports are operated by local or regional government branches (i.e., a division or department of aviation). Such an aviation department is usually separated from other departments, but often uses some functions of local government, for example accounting services, or purchasing decisions. Within an aviation department, the board directors are appointed by the chief executive officer of the local government and are ultimately responsible to the councils. Generally speaking, these board directors in an aviation department cannot enter into contracts without the approval of the (city) councils which literally own the airport. Moreover, the annual budget, bond sales and other similar measures of an aviation  6  Two airports, Dulles and Reagan National Airports are owned by the federal government, while the following state own and operate their own airports: Alaska, Arizona, Connecticut, Hawaii, Maryland, Minnesota, New Hampshire, and Rhode Island. 15  department also need to be approved by the councils. In some states, such as Alaska and Maryland, the aviation department can call upon general-fund revenues to subsidize airport operations and capital development, whilst, in other areas, such as Illinois and Nevada, the aviation department self-supports and operates without financial assistance from local/regional government. In most areas, the aviation department not only administers all aspects of airport operation and development, but also is responsible for setting, modifying and implementing rules and regulations that will affect the airport. Elsewhere the aviation department operates the airports directly but with the help of an advisory board as happens at Atlanta. It should be noted, however, that the advisory board is only involved in setting and modifying rules and regulations rather than operation or management of airports. As an alternative to direct control by local/regional government, airport authorities were first established to assume control over public airports during the 1950s and 1960s. Unlike those in Canada, US airport authorities are considered as public agencies since they are created by the local/regional governments that own airports. Although few airport authorities, such as the Great Orlando Aviation Authority and the Metropolitan Washington Airports Authority, lease the airport from the government, majority of local/regional governments directly transferred and delegated all airport managerial responsibilities to an airport authority at no cost or nominal cost. At large, an airport authority resembles an autonomous corporation with its own functional departments, such as finance and procurement departments. While airport authorities are structured as independent and self-supporting institutions, the board members of an airport authority are always elected by the local/regional government. The boarder members are authorized to appoint the chief executive officer of an airport authority and veto authority decision. Therefore, the local government, to a greater or lesser degree, can exercise varying 16  levels of oversight and control via the makeup and structure of the board. In some states, such as state of Florida, the elected public officials are allowed to serve and always server as the board members, whistle, in other states, such as state of Michigan, state legislatures have ruled out the elected public officials as the board member of airport authorities. For instance, Mayor of the City Orlando and other public officials are serving as the board member of the Great Orlando Aviation Authority, whereas the board of the Cincinnati/Northern Kentucky International Airport only consists of both civic and business leaders. 2.4  The Effects of Governance Structure on Airport Performance  Of critical importance to our study is to differentiate airport governance structures adopted in North America. As discussed previously, the model that airports are operated by local or regional government branches is clearly defined and unique to the US. However, the terminology of Airport Authority has some ambiguity to it. It has been used in Canada as a private sector corporation, whilst the terms airport authority, in the US, is considered as a quasi-governmental operation model. In previous research it is conventional to consider both Canadian and US airport authorities as one type of airport governance and simply refer to them as ―airport authority‖. This argument rests on the fact that these airport authorities are not-for-profit/nonshareholder entities that re-invest retained earnings into future airport development programs and are by-and-large financially self-sustaining. However, differences occur between US airport authorities and their Canadian counterparts. Regardless of governance structures, US airports have developed particular relationship with their customers, airlines for example, and financial structures which distinguish them from airports in Canada. For instance, US airports enter into legally binding contracts known as airport-use agreements which detail the conditions for the use of both airfield and terminal 17  facilities. These contracts are negotiated between the airport and its airline customers. The contracts will specify the fees and rental rates which an airline has to pay and the method by which these fees are to be calculated. In respect to sources of capital investment, many US airports are financed partly or largely from the private sector through the bond market. Moreover, US airports are eligible to be funded by the federal government via the Airport Improvement Program (AIP), which is administered by the Federal Aviation Administration (FAA). In addition, while Canadian airports are not directly regulated, US airports are subject to some general pricing rules. For instance, US airports are required to set aeronautical fees so as to collect revenues that reflect the costs of providing the services. Despite of their particular relationship with airlines and financial sources available, US airport authorities are different from their Canadian counterparts as the selection of Board members. As for US airport authorities, their board members have to be appointed by the state and local government which own the airport, while the board members of the airport authority in Canada are generally appointed by local community organizations. However, it remains a question how such different selection processes affect airport efficiency performance. There is no doubt that political motivated appointment of the Board members leaves US airport authorities vulnerable to changes in administration and to the exertion of political decisions of a business nature. It is noticeable, nonetheless, that the board members of US airport authorities either serve on a voluntary basis or are paid by a small stipend for each official meeting or activity. It is therefore possible that the Board members of US airport authorities are more likely to represent the communities that airports serve and thus have a strong interest in the performance of the airport. On the other hand, the Board members of Canadian airport authorities receive compensation for their service on the Board and thus would be distant from 18  the communities that airports serve. It is then likely that Canadian airport authorities enjoy greater autonomy from the electorate and from a single accountable body than US airport authorities. Such greater autonomy may trigger greater rent seeking activities of bureaucrats in the authority and potentially induce inefficiencies in airport operation. While the definition of an airport authority varies, this model rivals direct control by local/regional governments, i.e., an aviation department runs the airports owned by corresponding local/regional governments. Compared to airport authorities, directors of aviation departments may be more sensitive to voter demands but also consider airports as stimulating development. Hence they would like to pursue more cost effective strategies and articulate clear incentives for efficient performance. The benefits of direct control often are balanced against the belief that greater autonomy can lead to improved performance and greater efficiency. It is often noted that airport authorities are less liable to political interference. To a large extent, airport authorities avoid many of the civil services, contract approval, and other constraints of aviation departments. Moreover, managers of airport authorities may have greater knowledge and expertise regarding the specialized aviation industry. In addition, it has been long argued that there is some potential inefficiency in procurement practices of aviation departments. As discussed in Section 2, airports run by a government branch rely on the local government staff to make purchasing decisions which often make the process longer and less efficient. In most cases, the political influence form local government ultimately prevents aviation departments from procuring services from the most cost-effective source.  19  Here, we adopt some simple assessments to see whether there are any clear patterns of performance between different airport governance structures. For the assessment, we identify a sample of 54 airports in North America (see appendix B). The data covers the time frame from 2002-2008 and all the monetary values have been converted into US dollars. The sample airports have been split with 26 airports operated by a government branch and 28 airports operated by an airport authority. We applied the one-way analysis of variation (ANOVA) to compare the performance of the sample airports as regards the following statistics. As a simple summary means of evaluating the overall (cost) efficiency performance, we first compare the total operating cost of the sample airports by using a one-way ANOVA analysis. Over 7 years, the average total operating cost of airports operated by a government branch was US $ 131 million, as opposed to US $73 million for those operated by an airport authority. The result from Table 1-A shows that airports operated by an airport authority have significantly lower total operating cost than those operated by a government branch. To bring more insights, we further break down airports operated by an airport authority for airports run by US airport authorities and Canadian airport authorities. As shown in Table 1-B, there is statistically significant difference between airports operated under these three governance types with regard to their total operating costs. Airports operated by a government branch tend to have higher total operating costs than those by either US or Canadian airport authorities.  20  Table 1 Effects of Governance Structure on Airport Total Operating Cost  Table 1-A Governance Structure vs. Total Operating Costs Count Average(  Croups Airport Authority Government Branch Government Structures Residuals Groups  192 175 DF  7.32 13.2 Sum SQ(  1  3.15  )  Variance( 3.80 12.4 F Value 39.76  365 28.9 Table1-B Governance Structure vs. Total Operating Costs (Alternative Classification) Count Average( Variance(  US Airport Authority  143  8.38  44.1  Canadian Airport Authority  49  4.22  7.65  Government Branch  175  13.2  124  DF  Sum SQ(  2 364  3.78 28.3  Governance Structures Residuals  )  )  F Value 24.33  Since many airports aim to increase revenues from commercial services and other nonaeronautical activities, we also examine the effects of governance forms on airport nonaeronautical revenue. Surprisingly, Table 2 –A and B indicate that airports operated by a government branch tend to have significantly higher non-aeronautical revenue than those under other governance structures. However, the picture looks different as to the percentage of nonaeronautical revenue. It is clear from Table 3 A and B that airports operated by a government branch tend to derive a lower percentage of their revenue from non-aeronautical activities than their counterparts under other two governance forms.  21  Table 2 Effects of Governance Structure on Airport Non-aeronautical Revenue  Table 2-A Governance Structure vs. Non-Aeronautical Revenue Croups  Count  Average(  Variance (  Airport Authority Government Branch  192 175 DF  6.85 9.27 Sum SQ(  3.23 4.85 F Value  Governance Structures  1  5.35  13.40  Residuals  365 146 Table 2-B Governance Structure vs. Non-Aeronautical Revenue(Alternative Classification)  Groups  Count  Average (  Variance (  US Airport Authority  143  7.30  3.60  Canadian Airport Authority Government Branch  49 175 DF  5.55 9.27 Sum SQ (  1.96 4.84 F Value  Governance Structures  2  6.46  8.12  Residuals  364  144  Table 3 Effects of Governance Structure on Shares of Non-aeronautical Revenues Table 3-A Governance Structure vs. Shares of Non-Aeronautical Revenue Croups  Count  Average  Variance  Airport Authority  192  0.54  0.015  Government Branch  175 DF  0.40 Sum SQ  0.013 F Value  Governance Structures  1  0.182  12.95  Residuals 365 5.146 Table 3-A Governance Structure vs. Shares of Non-Aeronautical Revenue (Alternative Classification) Groups  Count  Average  Variance  US Airport Authority  143  0.55  0.018  Canadian Airport Authority  49  0.52  0.004  Government Branch  175 DF  0.50 Sum SQ  0.013 F Value  Governance Structures  2  0.206  7.32  Residuals  364  5.123 22  Finally, different governance structures may induce different levels of provisions over airport hiring and procurement activities. Therefore, we also look at some statistics with regard to airport personnel expenses. Although there is no statistically significant difference between airports under different governance structures as regards their labour price (shown in Table 5), airports operated by a government branch tend to have a lower labour share those operated under other governance structures (shown in Table 4). Table 4 Effects of Governance Structure on Airport Labour Share  Table 4-A  Governance Structure vs. Labour Share  Croups  Count  Average  Variance  Airport Authority  192  0.43  0.010  Government Branch  175 DF  0.39 Sum SQ  0.011 F Value  Governance Structures  1  0.14  13.46  Residuals  365 3.89 Table 4- B Governance Structure vs. Labour Share (Alternative Classification)  Groups  Count  Average  Variance  US Airport Authority  143  0.44  0.010  Canadian Airport Authority  49  0.40  0.007  Government Branch  175 DF  0.39 Sum SQ  0.011 F Value  Governance Structures  2  0.18  8.64  Residuals  364  3.85  23  Table 5 Effects of Governance Structures on Airport Labour Price  Table 5-A  Governance Structure vs. Labour Price  Croups  Count  Average (  Variance (  Airport Authority Government Branch  192 175 DF  6.88 6.92 Sum SQ (  2.90 5.11 F Value  Governance Structures  1  1.55  0.04  Residuals Table5-B  365 14400 Governance Structure vs. Labour Price (Alternative Classification)  Groups  Count  Average (  Variance (  US Airport Authority Canadian Airport Authority Government Branch  143 49 175 DF  6.83 7.02 6.92 Sum SQ (  2.78 3.29 5.11 F Value  Governance Structures  2  1.55  Residuals  364  1440  )  0.20  Although we have detected some difference in the performance of airports under different governance forms, these average and simple comparisons should not be viewed as conclusive. Many other factors relating airport performance are not within the control. Therefore, more rigorous statistical method should be used to establish a relationship between governance structure and airport performance.  24  2.5  Conclusion  In this chapter, we briefly surveyed the current state of airport governance around the world and shed more light on discussing the types of airport governance in North America. In addition, by applying a simple ANOVA analysis, we have discerned some evidence from a cursory review of different types of financial data that airports operated by a government branch perform differently than those run by an airport authority. However, given the preliminary nature of these assessments, it is too early to make any definite statement about the impact of governance forms on airport efficiency performance. The inherent complexity of the airport industry calls for a more rigorous analysis that can better control for the factors other than governance structures to correlate airport governance and its efficiency performance.  25  3  The Econometric Model  In Chapter 2, we provided some preliminary analysis on the effects of governance structures on the efficiency performance of North American airports. In this chapter, we will discuss how to analyze such effects in a more sophisticated econometric manner. We start off by comparing and contrasting different methodologies or models used to measure airport efficiency performance. In Section 2, we review the basic framework of our stochastic frontier model. Based on this framework, we then proceed, in Section 3, to discuss each constituent element of our model. Particularly, we provide specific details on the formulations which permit us to examine the effects of governance structures on both technical inefficiency and variable input usage.  3.1 Models of Airport Performance As indicated in Chapter 1, a lot of efforts and resources have been expanded in exploring measures of performance in air transport industry over the last two decades. The 1990s, nonetheless, was a distinctly late time for there to be an awakening of interest in airport performance. One reason why models of airport performance have been only relatively recently developed lies with the difficulties associated with the task. It can be argued that it is more difficult to develop satisfactory models than it is for other sectors in transport, airlines for example. To provide a confident measure of airport performance, a model must solve the aggregation problem, and ensure that the diverse outputs and inputs of the airport being aggregated in a meaningful way. Furthermore, since there are many factors that affect the relationship of inputs to outputs, it is necessary to be able to relate these factors to measured productivity. Among the standard approaches introduced in Chapter 1, non-parametric approaches, such as TFP and DEA, readily handle a large number of input and output categories, 26  more easily than they can be accommodated in econometric estimation methods. But econometric estimation methods, if data are available, build on established economic theory relationships and separate out the influences on cost/productivity. In this study, we argue that it is not sufficient to simply describe airport performance but also to be able to assess and understand how different governance structures can affect it. Hence econometric methods are preferred since it is desirable to isolate governance structures from other sources which also affect measured airport performance. In regard to econometric methods, traditional econometric methods for estimating cost or production functions implicitly assume that all firms are successful in reaching the efficient frontier (and only deviating randomly). If, however, firms are not always on the frontier, then the conventional estimation method would not reflect the (efficient) production or cost frontier against which to measure efficiency. For this reason, we would like to estimate either frontier production or cost functions recognizing that some firms may not be on the efficient frontier. The use of frontier cost functions requires the specification of assumptions on firm behaviour and the properties of functional forms. The assumptions on production technology depend on the specification of the cost function, but in all cases the firm pursues the strategy of cost minimization. As an alternative to cost functions, one could use production functions, for which the assumption of cost minimization is not necessary. However, it is difficult to estimate a production function when firms produce more than one output. Granted that multiplicity of output is a key feature of airports, cost-function approach enables us to bypass the aggregation problem but also incorporate the diverse outputs and inputs of the airport in our analysis. Taking everything into consideration, our study will rest on a stochastic cost frontier model and  27  econometric techniques presented in the following sections will further explore the ways in which we analyze the effects of governance structures on airport efficiency performance.  3.2 The Framework of the Stochastic Cost Frontier Model A large body of literature commencing with work by Knight (1933), Debreu (1951), and Farrell (1957) has argued that the conventional production/cost function is merely an efficient frontier and that the existence of inefficiency necessitates the incorporation of a nonzero disturbance in the function. In agreement with this argument, Aigner, Lovell and Schmidt proposed a so-called Stochastic Frontier Analysis (SFA) in their pioneering empirical study in 1977. The basic empirical framework for SFA is a regression specification involving a logarithmic transformation that adds a positive random error term, along with the traditional symmetric noise term, to capture unexplained inefficiency. Following this path, we outline the basic framework of our stochastic cost frontier model as follow. In the short run, if an airport tries to minimize its production cost (C) given the outputs (Q), variable input prices (W), and capital inputs (K), then the cost minimization frontier in a logarithmic form can be expressed as  ,where i represents airport and t  represents time. In reality, airports may deviate from their cost minimization objective for various reasons and such deviations indicate the existence of inefficiency. To reflect this reality, for airport , we denote a positive random term  as technical inefficiency, which indicates the  deviation of airport actual cost from its cost frontier. Further, since our interest centers on determining whether and/or how governance structures can affect airport efficiency performance, we assume that the technical inefficiency term variable  , with the dependence expressed as  depends on airport governance structure . Moreover, it is possible that 28  governance structures can assert influence not only on airport technical inefficiency but also on allocative inefficiency. Although we cannot fully analyze the impact of governance structures on allocative inefficiency due to estimation problems, we model between airport governance structure variable  as the interaction  and variable input price variables (  ). By  applying Shephard’s lemma, this specification will allow us to investigate how different governance structures alter airport variable input usage. Taken together, the observed actual production cost7 of airport i at time t can be expressed as  (3.1) where  represents the traditional symmetric noise term. In particular, our model includes  three outputs in vector non-aeronautical output price of the soft cost input number of gates  (number of passengers  ; number of aircraft movements  ), two variable input prices in vector  (labour price  ), and three capital inputs8 in vector  ; and terminal size  ; and ; and  (number of runways  ;  ). We shall discuss these variables in Chapter 4.  3.3 The Specifications of the Stochastic Cost Frontier Model 3.3.1  Specification of the Cost Frontier  Stochastic cost frontier analysis begins with the selection of a functional form. In previous literature, there are three ―basic‖ functional forms that have been applied with the stochastic cost frontiers model: the Cobb-Douglas, Leontief and constant elasticity of substitution (CES)  7  From this point on, we will no longer make an explicit notational distinction between a variable and its logarithm. To be consistent with most empirical specifications, we will assume all variables are measured in logs. 8 As pointed out in Chapter 2, there are no accurate and consistent measures of airport capital inputs available. We have no choice but to replace the capital input measures with three physical measures of capital stocks. 29  specifications. Among all of these functional forms, the Cobb-Douglas is one of the most commonly used. But, some scholars, such as Hasenkamp (1976), noted that a function having the Cobb-Douglas form cannot accommodate multiple outputs without violating the requisite curvature properties in output space. In addition, the simple functional form of Cobb-Douglas can hardly capture the true complexity of the production technology, therefore leaving the unmodeled complexity in the error term and biasing estimates of cost inefficiency. In this regard, translog cost function has been incorporated into stochastic frontier analysis. There are some difficulties attached to this function, however. For example, the translog multiproduct cost function cannot deal with zero output (Caves et al., 1980). Moreover, some restriction may not be met: a common finding in the applied economic literature is that an estimated translog functional form is not globally concave (Diewert and Wales, 1987). Despite these difficulties, the translog cost frontier is usually preferred in the literature over more simple specifications such as the Cobb-Douglas function. Our study also adopts translog specification which can be expressed as follows:  (3.2) Given that out study rests on a translog cost-function specification, among all challenges that have to be further addressed is how to measure and specify capital inputs in an airport. Due 30  to different accounting or reporting system across airports worldwide, transportation researchers have long struggled for finding accurate and consistent measures of airport capital inputs. To overcome the difficulty in data collection, we proxy capital inputs by some physical measures, e.g., measuring airport capital input by the number of runways or total terminal size. Particularly, we treat physical measure as a fixed input, i.e., the physical measure enters the cost frontier rather than the price of capital. However, in practice airports may not be able to adjust their capacity quantities as output changes. In order to account for the short-run disequilibrium adjustment in capacity, we estimate the following restricted translog cost frontier, which is loglinear in the physical measure of capital inputs.  (3.3) Moreover, all airports are, to an extent, unique; no airport is simply a larger or smaller version of another. In addition to governance structures, many other factors, such as location, are important with airports but beyond airport managerial control. Therefore, we augment our cost frontier in (3.3) to account for observed airport heterogeneity by adding to two sets of variables: percentage of international passengers in total passenger traffic (  ), country dummy variable (  ). In addition, worldwide economic climate is the most powerful driver of changes in air transport industry at large. At its simplest, we include a set of time dummies (  ) in the cost  frontier to reflect the impact of general economic condition on airport operating cost.  31  Adjusted to these specifications, the stochastic cost frontier adopted in our study can be written as follows:  3.3.2 3.3.2.1  Effects of Governance Structures on Airport Cost Efficiency Effects of Airport Governance Structures on Variable Input Usage  Although the types of airport governance are generally beyond the control of airport managers, airport governance structure indeed capture features of the environment in which airports operate. It may influence the airport production process itself, as is the case with network characteristics in transportation studies. If so, it would be entirely appropriate to include governance structure along with other explanatory variables in a stochastic cost frontier, which we write as  (3.5) where  represents airports,  stochastic cost frontier,  is the deterministic kernel of the captures the effect of technical inefficiency ,  32  captures the effect of symmetric random noise term , and the parameters vector  to be estimated  now includes both technology parameters and governance structure parameters. In this formulation, we assume that airport governance structure influence airport operating cost directly, by affecting the structure of the cost frontier relative to which the efficiency of airports is estimated. However, as noted in Chapter 2 (see Section 2.2), it is more likely that governance structures will exert indirect influence on airport operation or production, through its effect on the use of various airport inputs. Therefore, we seek to model the effect of airport governance structure in a non-neutrally input-augmenting form rather than in a mere shift of cost frontier. To analyze such an effect, we initially tried to break down the overall cost inefficiency into technical and allocative inefficiency. We soon discovered, however, that there is an inherent econometric challenge in estimating both technical and allocative inefficiency with panel data: it either requires a restrictive functional form, or estimates a highly nonlinear specification involving the allocative errors. To avoid these complexities, we narrow the scope of our investigation to the effect of governance structure on airport variable input usage. Further, to reduce the number of parameters to estimate, we model this effect ( an interaction between the dummy variable of airport governance structures ( variable input prices (  ) as  ) and the  ). This interaction term is expressed as  (3.6)  33  where  stands for the price of variable input  for the airport at the time ,  variable indicating the governance structure of the airport , and  is a dummy  is the parameter to be  estimated. Our stochastic frontier model, therefore, can be rewritten as follows:  (3.7) where  represents airports,  cost frontier,  is the deterministic kernel of the stochastic  captures the effect of governance structures on airport variable input usage,  captures the effect of technical inefficiency , symmetric random noise term , and the vector  and  captures the effect of  are the parameters to be estimated.  3.3.2.2 Effects of Governance Structures on Technical Inefficiency  What is accomplished by the formulation (3.7) is a more accurate characterization of airport production possibilities than would be provided by a formulation similar to (3.4) that excluded the effect of governance structures from the model, and consequently more accurate estimates of airport efficiencies. However variation in efficiency is left unexplained by the formulation (3.7). As much of this paper addresses on whether airports operated under different governance types differ in their efficiency levels, it is necessary for us to associate variation in estimated efficiency with variation in airport governance forms. To do so, many empirical analyses have proceeded in two steps. Had this two-step approach applied to our case, in the first step, one would ignore the effect of airport governance structures and estimate the stochastic frontier model and firm’s efficiency levels. In the second step, one would try to see how the estimated  34  efficiency levels varied with types of airport governance structure, perhaps by regressing a measure of efficiency on variables representing airport governance structures. Unfortunately such a two-step procedure is not appropriate for our analysis. First of all, any attempt to calibrate the two-step procedure to our analysis has so far relied on the assumption that other exogenous variables in the cost frontier should be uncorrelated to airport governance structures. Contrary to this assumption, the formulation (3.6) has explicitly assumed the input price variables are correlated to airport governance structures and thus prevents us from applying the two-step formulation. Secondly, even if it is a judgment call in regard to our attitude toward the effect of governance structures on airport input usage, there is a more serious econometric problem with this two-step procedure. This method implicitly assumes that in the first stage that the inefficiencies are identically distributed. But this assumption is contradicted in the second-stage regression in which predicted efficiencies are assumed to have a functional relationship with airport governance structures. In these circumstances it is not clear whether a two-step formulation successfully contributes anything to our understanding of airport governance structures on efficiency variation. To overcome the drawbacks of the two-step approach, we adopt the formulation proposed by Battese and Coelli (1995) in which the distribution of technical inefficiency depends on firm characteristics, i.e. governance structures in our case. Specifically, we assume that technical inefficiency  may change linearly with respect to airport governance structures.  is then  written as follows:  35  9  (3.8) where  is a dummy variable indicating what type of the governance structures the airport has  at the time ,  and  is the parameter to be estimated, and  is a random variable defined by  the truncation of the normal distribution with zero mean and variance truncation is –  , i.e.,  , such that the point of  .  Based on these specifications, the stochastic cost frontier model adopted in our study can be expanded as follows:  (3.9)  9  Not including an intercept parameter, , in the mean, may result in the estimator for the -parameter associated with the governance structure variable being biased and the shape of the distribution of technical inefficiency , being unnecessarily restricted.  36  where  As the cost function must be linearly homogeneous in input prices, the following restrictions are required:  ,  ,  ,  ,  ,  ,  ,  Moreover, we also impose the symmetric restrictions which are given as  ,  ,  ,  3.4 Conclusion In this chapter, we gave a detailed description of the stochastic frontier model adopted in our study. Our model not only captured unobserved airport inefficiency levels, but also examined 37  whether different airport governance structures can explain them. Moreover, we argued that governance structures will exert influence on airport operation or production in a non-neutrally input-augmenting fashion and thus employed a parametric method to measure the impact of governance structures on airport variable input usage. In addition, our model included three sets of variables that represent the heterogeneity of cost frontier across individual airports.  38  4  Sample and Variables  In previous two chapters, we have clarified the definition of airport governance structures and present our econometric model. We will now focus on the data used in estimation. Our sample contains an unbalanced panel of 54 North American airports over the period of 2002-2008. These airports are located across two countries (Canada and the United States) and governed / operated under two different governance structures (a government branch and an airport authority). The present chapter presents a brief overview of variable constructions and some characteristics of the sample airports10.  4.1 Variable Construction 4.1.1 Outputs and Inputs In the field of transportation research nothing is more valuable yet simultaneously more limiting to the validation of theory and models than are data. In dependence of the precise methodology used, modeling airport performance requires the definition of inputs (or input prices in our case of determining cost frontier) and outputs. Such definitions are not straightforward and give rise to some controversy. On the output side, it is noticeable that airports do not provide final services; they provide intermediate services to the airline and other related industries. This makes it difficult to define the outputs of airport precisely. The output categories commonly employed in economic analysis include the number of passengers, the volume of air cargo, and the number of aircraft movements (ATM). In classical transport economic there have been little questions raised as to consider the number of passengers as a separate output. Here we follow the 10  For a more detailed description of the data, readers should refer to a series of the ATRS Global Airport Benchmarking Reports (2003-2009). 39  conventional approach and treat the number of passengers as one set of airport outputs. As for the latter two output categories, airport cargo services are generally handled by airlines, thirdparty cargo handling companies, and others that lease space and facilities from airports. In this study, we do not consider airport cargo services as a separate output since airports derive a very small percentage of their income indirectly from air cargo services. Are aircraft movements to be regard as outputs of an airport? A contrary view might be that the airport is providing air-land interchange services for passengers and freight, and that aircraft movements are not separate outputs, but rather the means by which these interchange services are affected. However, granted that a significant portion of air port activities are related to the movement of aircraft, we consider the number of aircraft movements at an airport as an important indicator of airport activity and thus a separate output in our analysis. In addition to the three output categories mentioned above, airports also derive revenues from concessions, car parking and other numerous services. These services are not directly related to aeronautical activities in a traditional sense, but they are becoming increasingly more important for airports around the world. For some airports, such as Richmond and Nashville International airport, the revenues from non-aeronautical services account for as high as 70% of their total revenue in 2008. Moreover, the airport inputs are not usually separable between the aeronautical and the non-aeronautical activities. Any productivity measure which excludes the non-aeronautical services as the output would, therefore, bias empirical results seriously against the airports generating a large portion of their revenues from commercial activities. For this reason, we construct the third output to take into account the revenues from all non-aeronautical services. Further, since non-aeronautical services include various items and activities, it is difficult to construct an ―exact‖ price index consistent across airports in different regions and 40  over time. At this point, we construct the non-aeronautical output index by deflating the total non-aeronautical revenues by the cost of living index (COLI) for all sample airports. On the input side, we initially considered four general categories: (1) labor, which is measured by the number of employees who work directly for an airport operator; (2) purchased goods and materials; (3) purchased services, including those contracted out to external parties; and (4) capital, which consists of various facilities and infrastructure. In practice, however, few airports provide separate expense accounts for input categories (2) and (3). We therefore combine them to a single input category, which is referred as ―the soft cost input‖. This soft cost input consists of all operating expenses not directly related to personnel and capital expenditures. As such operating expenses are measured in different currencies and airports operate in regions with very different price levels, we use the COLI as proxies for the soft cost prices of the sample airports. Moreover, as indicated in Chapter 3, we use some physical measures as the proxies for capital inputs. In particular, we consider three physical measures: the number of runways, the number of gates and the terminal size.  4.2 Characteristics of Sample Airports In this section we illustrate general data and statistics for the sample airports in terms of governance structures, in both absolute and relative figures 11. Summary statistics are shown for each governance structure, so as to allow an easy comparison between all variables by readers themselves. Note that these summary statistics presented are based on raw data from 2002-2008 and provided in confidence by the ATRS Global Airport Performance Benchmarking project. In  11  Without additional notations, all monetary values are measured in US dollars.  41  the ensuring estimation, we normalize total operating costs, outputs, proxy capital measures and variable input prices at their sample means. 4.2.1 Summary Statistics of the Airports Operated by the Government Branch  As is evident from Table 6, there are significant differences among the airports operated by a government branch in terms of operation scale and orientation. For example, while the average passenger traffic was 28 million in 2008, 30.8% of the airports had annual passenger traffic below 10 million and 42.3% had annual passenger traffic above 30 million. The annual passenger traffic ranged from 90.2 million for Hartsfield-Jackson Atlanta International Airport to 6.5 million for Albuquerque International Airport in 2008. Furthermore, with an average of 38% in 2008, labour expense accounts for 56.8% of San Francisco International Airport’s total operating costs in 2008, whereas comprising only 18% of the total operating costs at Louis Armstrong New Orleans International Airport. Nonetheless, as for all the sample airports in this category, a substantial portion of airport revenue comes from non-aeronautical activities. In 2008, the share of annual non-aeronautical revenue ranged from 31% for Miami International Airport to 65% for Phoenix Sky Harbor International.  42  Table 6 Summary of Average Statistics -- Airports Operated by Government Branch 2002  2003  2004  2005  2006  2007  2008  Number of Passengers (million)  23 (19)  24 (20)  26 (21)  27 (22)  25 (19)  28 (23)  28 (23)  Number of Aircraft Movements(000's) Non-Aeronautical Revenue (million COLI deflated $)  328 (221)  335 (225)  343 (240)  347 (251)  334 (218)  361 (246)  349(245)  74(51)  76 (53)  76 (55)  79 (56)  79 (55)  87(61)  91 (69)  Number of Runways  3.4(1.2)  3.5(1.3)  3.5 (1.3)  3.5 (1.3)  3.5 (1.3)  3.6(1.3)  3.6 (1.2)  Number of Gates  73 (46)  77 (48)  76 (47)  79(47)  73 (41)  80(49)  79(48)  Terminal Size (000's Squared Meter)  200 (183)  220 (187)  244 (224)  224 (195)  217 (184)  230 (194)  212(159)  Wage (000's US$)  58 (16)  63(19)  64(18)  69(21)  68(22)  79(24)  86(27)  Soft Cost Input Price ( COLI)  1.04(0.2)  1.07(0.2)  1.12(0.3)  1.14(0.2)  1.19(0.2)  1.22(0.3)  1.26(0.3)  39(10)  39(10)  40(9)  40(11)  38(11)  39(11)  38(11)  Percentage of International Passengers (%)  8(11)  8(12)  8(11)  8(11)  8(11)  8(11)  7(12)  Share of Non-Aeronautical Revenue (%)  49(11)  48(12)  49(12)  49(12)  50(12)  51(12)  50(11)  Number of Observation  26  25  26  25  25  25  24  Airports Operated by Government Branch  Output Measures  Proxy Capital Measures  Variable Inputs' Prices  Variable Inputs' Share Labour Cost Share (%) Other Characteristics  Note: The numbers in parentheses are the standard errors.  4.2.2 Summary Statistics of the Airports Operated by the Airport Authority In this category, the sample airports are located across two countries – Canada and the United States – and the majority of these airports are in the US. For instance, in 2008, the US airports comprised 75% of the sample airports, while 25% of the airports were Canadian. Table 7 presents some summary statistics of the airports in this category. Overall, the traffic volume of the airports in this category is relatively small. With an average 15 million, the annual number of passengers ranged from 2.8 million for Albany International Airport (US) to 57.1 million passengers for Dallas Fort Worth International Airport (US) in 2008. 42.9% of the airports had the passenger volume below 10 million, while 10.7% of the airports accommodated more than 30 million passengers during the same period. Similar to the airports operated by a 43  government branch, the majority of the airports in this category also generate a large portion of their revenue from commercial activities and facilities, such as concessions and car parking. For instance, in 2008, the overall average share of non-aeronautical revenue reached 56%, ranging from 28.6% for Memphis International Airport (US) to 85% for Richmond International Airport (US). Moreover, within this category, labour expense accounts for almost 50% of the total operating costs for most airports in this category over 2002-2008. For example, on average, labour expense comprised 43% of the total operating expense for the airports in 2008. In the majority of the airports (79%), the labour share in total operating costs varied from 47% to 66% in 2008. In comparison between Canadian and the US airports, Canadian airports accommodate more international traffic. Whereas, on average, international traffic accounts for 30% of total passenger traffic for Canadian airports in 2008, the average percentage of international traffic for the US airports are significantly lower (4%).  44  Table 7 Summary of Average Statistics -- Airports Operated by Airport Authority 2002  2003  2004  2005  2006  2007  2008  Number of Passengers (million)  14(12)  13(12)  15(13)  15(14)  15(13)  15(13)  15(12)  Number of Aircraft Movements(000's) Non-Aeronautical Revenue (million COLI deflated $)  245(162)  236(159)  252(171)  242(160)  232(143)  228(140)  222(134)  52(39)  53(38)  59(47)  61(46)  67(53)  73(69)  75(75)  Number of Runways  3.2(1.1)  3.1(1.2)  3.3(1.3)  3.3(1.2)  3.2(1.2)  3.2(1.2)  3.3(1.2)  Number of Gates  59(42)  60(43)  61(42)  58(39)  63(41)  65(44)  64(41)  Terminal Size (000's Squared Meter)  121(102)  119(101)  123(104)  126(100)  127(94)  134(10)  133(10)  Wage (000's US$)  55(12)  60(12)  64(12)  67(13)  72(17)  77(15)  84(19)  Soft Cost Input Price ( COLI)  0.99(0.1)  1.03(0.1)  1.06(0.2)  1.08(0.2)  1.13(0.2)  1.16(0.2)  1.18(0.2)  46(10)  43(12)  43(10)  42(10)  42(9)  42(9)  43(9)  Percentage of International Passengers (%)  10(14)  10(15)  10(13)  10(15)  10(15)  11(15)  11(15)  Share of Non-Aeronautical Revenue (%)  52(13)  52(11)  53(12)  54(12)  55(12)  56(12)  56(13)  Canadian Airports (%)  26  27  26  26  25  25  25  US Airports (%)  74  73  74  74  75  75  75  Number of Observation  27  26  27  27  28  28  28  Airports Operated by Airport Authority  Output Measures  Proxy Capital Measures  Variable Inputs' Prices  Variable Inputs' Share Labour Cost Share (%) Other Characteristics  Geographic Distribution of Airports in Percentage  Note: The numbers in parentheses are the standard errors.  4.3 Conclusion This chapter discussed our construction of the variables and presented a short summary of the data used in our study. It is evident that the airports differ in the degree to which they engage in aeronautical and non- aeronautical activities. Except that of governance structures, the importance of other variables on airport performance has been acknowledged and illustrated in the previous literature, such as Oum et al (2008). Therefore, it is important to control for the effects of these variables when testing hypothesis concerning the effects of governance structures  45  on airport efficiency performance. Moreover, it would be also interesting to see how these variables affect the observed cost performance of the airports.  46  5  Empirical Results and Discussion  In the preceding chapters, we have discussed the model and data issues put to our study about the effects of governance structures on airport efficiency performance. In this chapter, we trace our initial classification of airport governance structures and present our modified categories. We then interpret our empirical results and critique the impact of governance structures on airport cost efficiency performance in Section 2.  5.1 Model Issues and Alternative Considerations The theme of this paper centers on examining the effects of airport governance structures, and asking whether airports operated under different governance forms differ in their efficiency performance. Thus far the spread of mathematical modeling to applied fields of transport economics and the development of econometrics greatly facilitate our research. However, a fundamental challenge to this study is how to define and classify governance of airports across countries. As most of the changes in ownership and governance of airports (with the exception of the UK) have taken place only within the last decade, there has been little scope for economic researchers to systematize and synchronize the different models that countries have adopted. In literally every circumstance different classification of airport governance forms has been defended by its proponents based on various political or economic allegations. For instance, it has been often argued that US and Canadian airport authorities are similar in nature as they are not-for-profit and financially self-sustaining. But as we saw in our discussion in Chapter 2, US airport authorities indeed differ from their Canadian counterparts in terms of their relationship with airlines, financial sources available and selection of the Board members. The question could, therefore, arise as so whether Canadian and US airport authorities differ in their impact on  47  airport (cost) efficiency performance and hence should be considered as different types of airport governance. For the sake of conceptual clarity, we experiment two alternative classifications of airport governance structures in our analysis. In Model A, we postulate the differences between Canadian and US airport authorities can be dismissed as without importance. Consequently all sample airports are classified into two categories: (1) airports operated by a government branch and (2) airports operated by an airport authority. In Model B, we separate Canadian and US airport authorities in our classification, which is concerned with three types of airport governance structures: (1) a government branch; (2) the US airport authority; and (3) the Canadian airport authority. Despite of unsettling classification of airport governance forms, another issue also needs to be taken into account. In Chapter 3, our approach of modeling airport heterogeneities lies within a simple framework: we established three sets of variables – country dummy variable, time dummy variables and percentage of international traffic – to control observed airport heterogeneities. Merely adding country dummy variable and the variable capturing the percentage of international traffic in airport total traffic implies that traffic pattern is similar among airports located in different countries. The statistic summary in Chapter 4, however, has revealed that, on average, Canadian airports appear to have much higher percentage of international passengers in their total traffic than US airports. Therefore, at least, the estimated coefficient of percentage of international traffic would be biased as without allowing the percentage of international traffic to differ across countries. To response to this finding, we create Model C and Model D by adding an addition interaction term to Model A and Model B, respectively. This new interaction term is the product of location (country dummy variable) and  48  the percentage of international traffic in airport total passenger traffic. By incorporating this new interaction term, we assume that the percentage of international traffic in airport total passenger traffic will differ between two countries. The designs of the four stochastic frontier models enable us to examine a pooled regression model for the two countries, in which the United States is the benchmark for the country dummy variable. In Model A and C, we set up the airport authority as the benchmark for the dummy variable of the governance structure, whereas the US airport authority is considered as the benchmark for the dummy variable of the governance structure in Model B and D. We use the Gauss maximum-likelihood computer program to obtain the empirical results. Table 8 shows the four stochastic cost frontier models that are analyzed in this paper – Models A, B, C and D. The econometric results for these four models are discussed in detail in the following two sections.  49  Table 8 Explanatory Variables and Regression Model for Stochastic Frontier Analysis  Table 8- A Explanatory Variables for Stochastic Cost Frontier Observed Airport Heterogeneities Model A Model B × × Year 2003) × × Year 2004) × × Year 2005) × × Year 2006) × × Year 2007) × × Year 2008) × × (%International ) × × (Canadian dummy) Outputs × × non-aeronautical output) × × passengers ) × × aircraft movements) Proxy Capital Measures × × runway) × × number of gates) × × terminal size) variable input's prices × × wage ) Interactions between Governance Structure and Variable Input Price × × × Interactions among Outputs non-aeronautical * non-aeronautical) (passenger * passenger) aircraft movements * aircraft movements) non-aeronautical * passenger ) non-aeronautical * aircraft movements ) passenger *aircraft movements) Interaction between Inputs' Price wage *wage ) Interaction between Outputs and Variable Inputs' Prices  Model C × × × × × × × × ×  Model D × × × × × × × × ×  × × ×  × × ×  × × ×  × × ×  ×  ×  × -  × ×  × × × × × ×  × × × × × ×  × × × × × ×  × × × × × ×  ×  ×  ×  ×  × × × × × × passenger * wage) × × × aircraft movement * wage) Table 8-B Explanatory Variables for Airport Technical Inefficiency Governance Structures Model A Model B Model C × × × (Government -branch dummy) × ) ** X denotes the variables used for that particular regression model  × × × Model D × ×  50  5.2 Empirical Results and Discussion This section presents our empirical results and their economic implications, based on the four stochastic frontier models described in Section 1. As presented in Table 9-12, there is no significant statistical difference among the results obtained from the four different model specifications. Therefore, we will not discuss the empirical results obtained from each model separately but interpret these results within a unifying framework which is compatible with the four proposed scenarios. The first part of this section then discusses the effects of airport characteristics on the stochastic cost frontier. The second part, which is of particular interest in this paper, reveals the effects of governance structures on airport cost efficiency and evaluates whether such effects vary under different model specifications.  5.2.1 The Effects of Airport Characteristics on Cost Frontier As noted previously, translog cost frontier includes first-order and quadratic terms. Since total operating costs and all regressors are in natural logarithms, and the regression has been normalized at the mean data point, the first order coefficients of the translog cost frontier can be interpreted as cost elasticities evaluated at the sample means. As it is difficult to interpret the quadratic terms in the translog specification, we here draw more attention on discussing the firstorder coefficients of the cost frontier and the coefficients of explanatory variables of particular interest in Table 9-12.  51  Table 9 Stochastic Cost Frontier Model A Results  Table 9-A Estimation Results for the Airport Characteristics Parameters  Coefficient  constant) Year 2003) Year 2004) Year 2005) Year 2006) Year 2007) Year 2008) (%International ) (Canadian dummy) Coefficients of Outputs non-aeronautical output) passengers ) aircraft movements) Coefficients of Proxy Capital Measures runway) number of gates) terminal size) Coefficients of variable input's prices wage ) Impact on Variable Input Usage Coefficients of Interactions among Outputs non-aeronautical * non-aeronautical) (passenger * passenger) aircraft movements * aircraft movements) non-aeronautical * passenger ) non-aeronautical * aircraft movements ) passenger *aircraft movements) Coefficients of Interaction between Inputs' Price wage *wage ) Coefficients of Interaction between Outputs and Variable Inputs' Prices  -0.934 0.043 0.022 0.048 0.066 0.085 0.117 0.684 -0.241  -3.79** 3.29** 0.94 1.89* 2.82** 3.01** 2.61** 2.34** -1.13  0.325 0.290 0.081  3.36** 3.21** 1.17  0.088 0.108 0.017  0.50 1.05 0.26  0.394  2.99**  -0. 069  -1.57  0.361 0.276 -0.241 -0.247  2.33** 0.61 -0.88 -0.97  -0.075 0.044  -0.29 0.24  0.172  -0.056 passenger * wage) -0.393 aircraft movement * wage) 0.511 Table 9-B Estimation Results for the Impact of Airport Governance Structures Parameters Coefficient  variance parameter)  -0.798 0.141 0.058 0.744  ** Significant at  t-statistics  0.76 -0.28 -1.48 2.30** t-statistics -3.1** 1.75* -  ;* Significant at  52  Table 10 Stochastic Cost Frontier Model B Results  Table 10-A Estimation Results for the Airport Characteristics Parameters Coefficient constant) -0.936 Year 2003) 0.040 Year 2004) 0.021 Year 2005) 0.044 Year 2006) 0.059 Year 2007) 0.082 Year 2008) 0.115 0.652 (%International ) (Canadian dummy) -0.236 Coefficients of Outputs non-aeronautical output) 0.338 passengers ) 0.269 2 aircraft movements) 0.077 Coefficients of Proxy Capital Measures runway) 0.095 number of gates) 0.111 2 terminal size) 0.013 Coefficients of variable input's prices wage ) 0.416 Coefficient for Interactions between Governance Structure and Variable Input Price Government branch wage -0.058 Canadian Airport Authority wage 0.014 2 Coefficients of Interactions among Outputs non-aeronautical * non-aeronautical) 0.356 (passenger * passenger) 0.243 22 aircraft movements * aircraft movements) -0.270 non-aeronautical * passenger ) -0.232 2 non-aeronautical * aircraft movements ) -0.026 passenger *aircraft movements) 0.083 2 Coefficients of Interaction between Inputs' Price wage *wage ) 0.185 2 Coefficients of Interaction between Outputs and Variable Inputs' Prices -0.055 non aeronautical wage -0.305 passenger * wage) 2 aircraft movement * wage) 0.505 Table 10-B Estimation Results for the Impact of Airport Governance Structures Parameters Coefficient Constant -0.766 0 (Government -branch dummy) 0.145 Canadian Airport Authority dummy) 0.028 2 2 2 + (Variance Parameter) 0.061 z 2  v 2  Ratio of the Variances)  ** Significant at  0.755  t-statistics -3.47** 3.21** 0.87 1.72* 2.46** 2.78** 2.50** 2.32** -1.12 3.60** 3.18** 1.21 0.61 1.06 0.21 2.91** -1.29 0.31  2.43** 0.54 -1.02 -0.93 -0.17 0.31 0.97 -0.29 -1.11 2.18** t-statistics -3.04* 1.77** 0.22 -  ; * Significant at 53  Table 11 Stochastic Cost Frontier Model C Results  Table 11-A Estimation Results for the Airport Characteristics Parameters constant) Year 2003) Year 2004) Year 2005) Year 2006) Year 2007) Year 2008) (%International ) (Canadian dummy) Coefficients of Outputs non-aeronautical output) passengers ) aircraft movements) Coefficients of Proxy Capital Measures runway) number of gates) terminal size) Coefficients of variable input's prices wage ) Impact on Variable Input Usage Coefficients of Interactions among Outputs non-aeronautical * non-aeronautical) (passenger * passenger) aircraft movements * aircraft movements) non-aeronautical * passenger ) non-aeronautical * aircraft movements ) passenger *aircraft movements) Coefficients of Interaction between Inputs' Price wage *wage ) Coefficients of Interaction between Outputs and Variable Inputs' Prices  Coefficient -0.802 0.044 0.021 0.049 0.065 0.086 0.118 0.022 0.013 -0.195  -3.26** 3.20** 0.84 1.67* 2.53** 2.81** 2.47** 1.26 2.02 -0.49  0.330 0.283 0.085  3.25** 3.31** 1.16  0.061 0.105 0.020  0.44 1.21 0.45  0.395  3.01**  -0. 071  -1.60  0.363 0.281 -0.222 -0.261  2.30** 0.59 -0.67 -0.96  0.015 0.021  0.17 0.10  0.165  0.71  -0.061 passenger * wage) -0.385 0.508 aircraft movement * wage) Table 11-B Estimation Results for the Impact of Airport Governance Structures Parameters Coefficient  variance parameter)  ** Significant at  t-statistics  -0.28 -1.45 2.26** t-statistics  -0.772 0.139 0.055  -3.11** 1.71* -  0.738  -  ; * Significant at 54  Table 12 Stochastic Cost Frontier Model D Results  Table 12-A Estimation Results for the Airport Characteristics Parameters constant) Year 2003) Year 2004) Year 2005) Year 2006) Year 2007) Year 2008) (%International )  Coefficient -0.916 0.040 0.020 0.045 0.059 0.083 0.116 0.017 0.007 -0.181  (Canadian dummy) Coefficients of Outputs non-aeronautical output) 0.332 passengers ) 0.273 aircraft movements) 0.080 Coefficients of Proxy Capital Measures 0.095 runway) 0.106 number of gates) 0.016 terminal size) Coefficients of variable input's prices 0.411 wage ) Coefficient for Interactions between Governance Structure and Variable Input Price -0.059 0.014 Coefficients of Interactions among Outputs non-aeronautical * non-aeronautical) 0.357 (passenger * passenger) 0.247 aircraft movements * aircraft movements) -0.246 non-aeronautical * passenger ) -0.247 non-aeronautical * aircraft movements ) -0.003 passenger *aircraft movements) 0.056 Coefficients of Interaction between Inputs' Price 0.179 wage *wage ) Coefficients of Interaction between Outputs and Variable Inputs' Prices -0.061 -0.289 passenger * wage) 0.495 aircraft movement * wage) Table 12-B Estimation Results for the Impact of Airport Governance Structures Parameters Coefficient -0.758 (Government -branch dummy) 0.142 ) 0.049 0.058 + (Variance Parameter) Ratio of the Variances) ** Significant at  0.731  t-statistics -3.53** 3.19** 0.84 1.73* 2.44** 2.77** 2.49** 1.64 2.11 -0.55 3.46** 3.21** 1.22 0.61 0.99 0.24 2.89** -1.33 0.32 2.39** 0.54 -0.91 -0.94 -0.02 0.21 0.93 -0.33 -1.05 2.11** t-statistics -3.08* 1.84** 0.55 -  ; * Significant at 55  Generally speaking, the four stochastic frontier models convey remarkably similar information as regards the effects of airport characteristics on the cost frontier. All first order coefficients, except those for the three proxy capital measures, have the expected coefficient signs in the cost frontier. The following overview highlights the effects of these variables:   Non-aeronautical output is one of the most statistically significant variables in all four regressions and has a positive coefficient. In all four scenarios, the coefficient for nonaeronautical output is very close to the value of 0.3, which indicates that a 1% increase in non-aeronautical output, holding other variables constant, causes an increase in the total operating cost of about 0.3%. Moreover, these results further demonstrate that nonaeronautical activities have become one of the primary economic forces driving mordent airport operation and have a substantial impact on the observed performance of an airport (as shown in Figure 1 and 2). Thus, the omission of this important variable can lead to misleading inferences concerning the effects of governance structures on airport efficiency performance. Figure 1 Non-aeronautical Revenue for Sample Airports in 2008  56  Figure 2 Share of Non-aeronautical Revenue for Sample Airports in 2008    Proxy capital measures include the number of runways, number of gates and terminal size. Although not statistically significant, the coefficients of these three variables are all positive and thus indicate upward shifts of the cost frontier. These counter-intuitive results are in part due to the indivisibilities of airport investments in major infrastructure facilities, and in part, perhaps, due to a lack of accurate and consistent measurement of capital inputs12.    Input Price (Wage) has a positive and significant coefficient. This first-order coefficient of labour input price indicates that, at the sample mean data, labour input accounts for almost 40% of the total operating cost, which leaves the soft cost input to account for 60% of the total operating cost.    Time Dummies capture the effect of time-specific variables, omitted in the model, which vary over time but are constant across airports. In our regressions, all coefficients for time  12  As pointed out in Chapter 3, transportation researchers have long struggled to find accurate and consistent measures of airport capital inputs. Given the data available, we have no choice but to replace the capital input measures with three capital stock variables.  57  dummies are positive. Except the coefficient for year 2004, all coefficients for other time dummies are statistically significant. These positive coefficients indicate upward shifts of the cost frontier in the post-2001 period and further reflect the fact that airports in North America have to bear the cost of the recently raised security threats and impacts of recovery from the recent recession.   Percentage of International Traffic has a positive coefficient in all four regressions. In the first two regressions—Model A and B— the coefficient for percentage of international traffic reaches 0.6 and statistically significant. This result enthusiastically suggests that accommodating international traffic imposes significant upward pressure on airport operating cost. However, we know from our previous data examination that this result may be attributed to the omission of the interaction between airport location (country) and the percentage of international traffic in airport total traffic volume. As is evident from Model C and D, the effect of percentage of international traffic has been reduced after adjusting the percentage of international traffic to airport location (country). In Model C and D, the coefficient of percentage of international traffic is still positive but not statistically significant. On contrary, the cross term with Canadian dummy is statistically significant with a positive coefficient. These results provide some evidence that in Canada, airports with a heavy reliance on international traffic face a higher cost frontier since international traffic requires more airport services and resources than domestic traffic.    Country dummy has a positive coefficient but is not statistically significant in all four models. This result provides weak evidence that Canadian airports face a lower cost frontier than US airports. It is noticeable that, after incorporating the new interaction  58  terms between airport location (country) and the percentage of international traffic, the coefficient for country dummy has become less significant (see Table 11 and Table 12). Thus, it is probably much more difficult to conclude that airports located in these two countries face different cost frontiers.  5.2.2 The Effects of Airport Governance Structures In this paper, we seek to develop a linkage between airport (efficiency) performance and governance structures, arguing that airport governance structures do not affect airport technical inefficiency but also allocative inefficiency, in particular, airport variable input usage. By comparing the results obtained from the four different model specifications, we found that adding the new interaction term—the product of location and the percentage of international traffic—does not highly influence the estimated effects of governance structures on either airport technical inefficiency or variable input usage: Model A and Model C show the similar estimated effects of governance structures on airport efficiency, as Model B and D do. It is therefore interesting to see whether our empirical results can provide a convincing rationale for alternative classification of airport governance structures in North America.  5.2.2.1 The Effects of Airport Governance Structures on Airport Input Usage Recall from Chapter 3 that the impact of the airport governance structures on variable input usage is identified via the coefficient of the input price variable interacted with the governance structure dummy. Thus, by applying Shephard’s lemma, we are able to use this coefficient to analyze the effect of the different governance structures on the level of airport input usage (labor and soft cost input). In Model A and Model C, the coefficient for government-branch with wage is negative  59  but not statistically significant. By applying Shephard’s lemma to the estimated cost functions, the negative coefficient for this cross term suggests that the airports operated by a government branch appear to have a lower labour cost share, or conversely a higher soft cost share, than those operated by an airport authority (Figure 3 and Figure 4). As discussed in Chapter 2, this result is partly because airports operated a government branch do not have some functional departments (e.g accounting and security) and use these services from other local government departments. Partly as a result of the procurement provisions of the local government, airports run by a government branch may not purchase services from the most cost-effective source, and thus tend to have a higher soft cost share than those run by an airport authority. Figure 3 Labour Cost Share for Airports Operated by Government Branch in 2008  Labour Cost Share for Airports Operated by Government Branch in 2008 0.60 0.50 0.40 0.30 0.20 0.10 MSY MDW SNA BWI CLT HNL PHX ORD SJC PHL MCI CLE SMF MKE LAS IAH MIA AUS ATL SLC ABQ ONT LAX SFO  0.00  Labour Cost Share  Average Labour Cost Share  60  Figure 4 Labour Cost Share for Airports Operated by Airport Authority in 2008  Labour Cost Share for Airports Operated by Airport Authority in 2008 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1  YYC FLL SAN MCO YVR DTW PIT YUL IAD DFW YOW MEM JAX STL DCA SDF YHZ ALB TPA MSP IND BNA RDU RIC CVG YWG YEG RNO  0  Labour Cost Share  Average Labour Cost Share  The results obtained from Model B and D also provide some supporting evidence to the argument as to potential inefficiency in procurement of airports run by government branches. As presented in Table 10 and Table 12, the negative coefficient for government-branch with wage, although not statistically significant, also indicates that the airports operated by a government branch appear to have a lower labour cost share (and thus a higher soft cost share) than those operated either by US or Canadian airport authorities. A interesting point brought out by Model B and D is that , compared to airports operated by US airport authorities, airports operated by Canadian airport authorities tend to have a high labour cost share. Provided that airport authorities, in general, are less impaired by government-wide hiring and procurement rules, there is a possibility that greater independence from the electorate and from a single accountable body makes Canadian airport authorities more prone to higher pay for their employees. Nonetheless, it requires more caution over this argument as the t-statistics of the cross term with Canadian 61  airport authority is highly insignificant in both regressions.  5.2.2.2 The Effects of Airport Governance Structures on Technical Inefficiency The part B of Table 9-12 reveals the effect of governance structures on airport technical inefficiency. In Model A and C, the coefficient for airports operated by a government branch is positive and significant at the 10% level. This result suggests that airports run by a government branch perform less efficiently than those run by an airport authority. Consistent with the results obtained by Craig et al (2005) and Oum et al (2008), our finding confirms that, independent institutions, such as the airport authorities, achieve a higher efficiency performance since they enjoy sufficient freedom to operate the airports in a commercially-oriented manner. While separated from other government departments, the aviation branch operates under the general requirements of the local government bureaucracy and thus is influenced by other political activities. Such factors may hinder efficient airport operation. Based on the alternative classification of airport governance forms, Model B and D suggest that airports operated by either the US or Canadian airport authority indeed outperform those operated by a government branch. These results further confirm that greater autonomy over airport decision-making can lead to improved performance and greater efficiency. Moreover, the results obtained from Model B and Model D raise doubts as to whether there is any efficiency difference between airports operated by US airport authorities and those run by Canadian airport authorities. As noted in Chapter 2, conventional wisdom has often in favour of Canadian airport authorities since political motivated appointment of the Board members leaves US airport authorities vulnerable to changes in administration and to the exertion of political decisions of a business nature. Although not statistically significant, the empirical results obtained from Model B and Model D do not support this argument but indicate that airports operated by US airport 62  authorities are more efficient than those run by Canadian airport authorities. However, t-statistics from both regressions are too low to discern there are any clear patterns of performance between airports run by US airport authorizes and those by Canadian airport authorities. This section analyzes the effects of governance structures on the cost efficiency performance of North American airports under four different model specifications. Moreover, the views taken from Model B and Model D results include determine whether Canadian airport authorities differ from US airport authorities in terms of their impact on airport cost efficiency performance and hence should be considered as a separated category of airport governance. Our estimation results13, as shown in Table 10 and 12, did not provide a convincing empirical rationale for this classification: there was no significant difference in either technical inefficiency or variable input usage between the airports operated by US and Canadian airport authorities. However, it is too early to arrive at the conclusion the airport authorities from both countries are similar in nature. Although we have established country dummy variable to delineate the country-specific effects on the cost frontier, the lack of data capturing the relationship between US airports and airlines may inherently limit our analysis on the efficiency difference between US and Canadian airport authorities.  5.3 Hypotheses Tests for Effects of Airport Governance Structures Thus far we have not only examined how different governance structures affect unobserved airport technical inefficiency, but also argued that the effect of governance structure is more likely to be embodied in a non-neutrally input-augmenting fashion rather than in a mere shift of cost frontier. However, it is not obvious whether the governance structure should be a  13  In our original regression, the United States is the benchmark for the country dummy variable and the US airport authority is the benchmark for the dummy variable of the governance structure. 63  characteristic of airport production technology or a determinant of airport productive efficiency. It is frequently a judgment call as regards how to model the effects of airport governance structures. Therefore, after discussing and making inference from our empirical results, we shall use a series of hypothesis tests to further examine our specifications on the effects of airport governance forms. Roughly speaking, the most commonly used the test procedures for maximum likelihood estimation include: the likelihood ratio, Wald, and Lagrange multiplier tests. Based on ease of computation, we here use the likelihood-ratio test. Specifically, let to be estimated, and  be a vector of parameters  specify some sort of restrictions on these parameters. Let  maximum likelihood estimator of  obtained without regard to the constraints, and  constrained maximum likelihood estimator. If  and  be the be the  are the likelihood functions  evaluated at these two estimates, then the likelihood ratio test is based on  where the test statistic  has approximately chi-square distribution with degrees of freedom equal  to the number of restrictions imposed in the null hypothesis hypothesis is rejected if the value of  , provided  is true. The null  exceeds the appropriate critical value from the chi-squared  tables. Since the likelihood ratio test cannot be used to test a simple null hypothesis against a simple alternative, the hypothesis with regard to the effects of governance forms on airport variable input usage is only conducted based on Model B and Model D. Thus, Table 14 and 16 contain four sets of hypothesis and test statistics based on Model B and D, whereas Table 13 and 15, based on Model A and C, only present three sets of hypothesis and test statistics. 64  Table 13 Hypothesis Tests for Stochastic Frontier Model A Table 13 Hypothesis Tests for Stochastic Frontier Model A Null Hypothesis  -  ** Significant at  Log[Likelihood(  )]  264.39 264.39 264.39  261.19 258.88 259.71  Test Statistics  -Value  5.99 7.82 5.99  6.38** 11.01** 9.35**  ;* Significant at  Table 14 Hypothesis Tests for Stochastic Frontier Model B Table 14 Hypothesis Tests for Stochastic Frontier Model B Log[Likelihood(  Null Hypothesis  )]  263.69 262.10 258.88 259.71 ** Significant at  -Value  266.12 266.12 266.12 266.12  5.99 7.82 11.07 9.49  Test Statistics  4.86* 8.04** 14.48** 12.82**  ;* Significant at  Table 15 Hypothesis Tests for Stochastic Frontier Model C Table 15 Hypothesis Tests for Stochastic Frontier Model C Null Hypothesis  -  ** Significant at  Log[Likelihood(  )]  261.87 258.96 260.15  -Value  265.07 265.07 265.07  5.99 7.82 5.99  Test Statistics  6.40** 12.21** 9.84**  ;* Significant at  Table 16 Hypothesis Tests for Stochastic Frontier Model D Table 16 Hypothesis Tests for Stochastic Frontier Model D Null Hypothesis  Log[Likelihood(  263.97 262.10 258.96 260.15 ** Significant at  )]  -Value  266.78 266.78 266.78 266.78  5.99 7.82 11.07 9.49  Test Statistics  5.62* 8.05** 15.64** 13.26**  ;* Significant at 65  As shown in Row 2 of Table 14 and 16, the null hypothesis indicates that the effect of governance types on airport variable input usage are absent from Model B and Model D, respectively. The chi-square statistic with two degree of freedom is significant at 10% level for Model B and D. As shown in Table 9-12, the t-statistics have not discerned any significant deference between airports operated under different governance structures as to their variable input usage. These results, however, indicate that the joint effect of governance structures on airport variable input usage is significant and thus cannot be omitted after separating US and Canadian airport authorities. In Row 3 of the four tables above, the null hypothesis specifies that technical inefficiency does not change linearly with respect to the governance structures but is half-normal distributed. As for all the four models, this null hypothesis is rejected at the 5% level of significance. This implies that, by incorporating difference between airport governance types, the proposed stochastic frontier cost model is a significant improvement over the corresponding stochastic frontier model without incorporating the effect of governance structures on technical inefficiency.  Based on these results, we can conclude that governance structures exercise  substantial influence on airport technical inefficiency and should hence be considered as factors which affect airport productive efficiency. In Row 4 of Table 13-16, the null hypothesis specifies that the effects of governance structures on both airport technical inefficiency and variable input usage are absent from the model and that the model collapses to the conventional half normal stochastic frontier specified in Aigner, Lovell and Schmidt (1977). This null hypothesis is rejected at the 5% level of  66  significance for all the four regressions. In Row 5 of the four tables above, the null hypothesis specifies that effects of governance structures on both airport technical inefficiency and variable input usage are absent from the model but that the model collapses to the truncated normal stochastic frontier model as specified in Stevenson (1980). This null hypothesis is also rejected at the 5% level of significance. Although the impact of governance structures on airport variable input usage, especially based on the t-statistics, is not statistically significant, the results of these latter two hypotheses tests confirm that the effects of governance structures on airport cost efficiency, namely both technical inefficiency and variable input usage, are indeed statistically significant.  5.4 Conclusion Based on different model specifications, we have estimated the impact of airport governance structures on the cost efficiency performance of North American airports. Our findings indicated that there was no significant difference between airports operated by US airport authorities and Canadian airport authorities as to their impacts on airport efficiency performance. It seems therefore that US and Canadian airport authorities are similar in nature and should not be considered as different types of airport governance. Moreover, our findings revealed that the airports operated by an airport authority indeed outperformed those operated by a government branch. In addition, although not statistically significant, our findings implied that the airports run by a government branch tend to have a lower labour cost share than those operated by an airport authority. Based on a series of hypotheses tests, we further confirmed that governance structures exercise substantial influence on airport efficiency and should thus be considered as determinants of airport performance.  67  6  Conclusion  This paper assessed the ways in which governance structures could affect the airport cost efficiency performance. With a focus on North America, we investigated the impact of two dominant governance structures – a government branch and an airport authority – on not only airport technical inefficiency but also airport variable input usage. To achieve this objective, we applied a stochastic frontier model to the unbalanced panel data of 54 North American airports over the time period 2002-2008. In this concluding chapter, we open with a summary of our key empirical findings and their contributions to the air transportation field. Section 2 discusses some potential extensions of this study.  6.1 Summary of Key Findings 6.1.1  Effects of Other Airport Characteristics  In addition to governance structures, a number of other airport characteristics have potential effects on airport cost performance. For instance, we find that non-aeronautical output is significantly related to the observed airport cost performance. 1% increase in percentage of nonaeronautical output is expected to increase the airport’s total operating cost by 0. %. Therefore, it is important to control for the effects of these airport characteristics when testing hypotheses concerning the effects of governance structures on airport efficiency performance.  68  6.1.2  Effects of Airport Governance Structures  This study has confirmed that the two dominant forms of governance structures (i.e. a government branch vs. an airport authority) indeed exercise substantial influences on the cost efficiency performance of airports in North America. In particular, our findings suggest that the airports operated by an airport authority outperform those operated by a government branch in terms of technical efficiency. This result provides new supporting evidence for the argument that the governance structures in which management can exercise a greater degree of autonomy and face less political pressure are more likely to stimulate airport efficiency performance. Moreover, by modeling the interrelationship between governance structure and airport variable input usage, our study provides weak evidence that the airports run by a government branch tend to have lower labour cost share than those run by an airport authority. Since little attention has been paid to the influence of governance structures on airport inputs, this paper provides a fuller account of the impact of governance structures on efficiency performance and offers a new platform for the future study. 6.2  Suggestions for Further Research  The empirical constructs of the current study offer a useful starting point for more in-depth analysis of the impact of airport governance structures. Based on the framework of our study, incorporating the share equation could improve the efficiency in estimation. To further extend our study, more flexible formulations can be applied that account for the presence of observed and unobserved heterogeneity across individual airports. Since our hypothesized model structure was reduced due to estimation problems, the interrelationship between governance structures and airport input usage also have the potential to be further analyzed. In addition, the incorporation of 69  various regulatory and environmental factors, such as the degree of competition and contract between airport and their airline customers, is another area that warrants further research. The complex specification required for such work, however, may increase the computational difficulty, if the stochastic frontier model is estimated via maximum likelihood. 6.3  Conclusion  In summary, this paper contributes to the discussion on the relative merits of the airport authority over the government branch in simulating airport efficiency performance. Most previous studies, on the other hand, have focused barely on variable input considerations. This paper assess the broader extent to which governance structures affect airport operation efficiency by investigating the impact of governance structures on airport variable input usage. The future possibilities of constructing complex specifications and relating governance structure to various regulatory and environmental factors are two related areas to this study that are rich in their potential to provide additional insight into the issue of airport efficiency performance.  70  Bibliography  Abbott, M., & Wu, S. (2002). Total Factor Productivity and Efficiency of Australian Airports. The Australian Economic Review 35, 244-60. Aigner, D., Lovell, K., & Schmidt, P. (1977). Formulation and Estimation of Stochastic Frontier Production Models. Journal of Econometrics 6, 21-37. Backx, M., Carney, M., & Gedajlovic, E. (2002). Private and Mixed Ownership and the Performance of International Airlines. Journal of Air Transport Management 8, 213-220. Barros, C., & Dieke, P. (2007). Performance evaluation of Italian airports: A data envelopment analysis. Journal of Air Transport Management 17, 184-191 . Battese, G. E., & Coelli, T. J. (1995). A Model for Technical Inefficiency Effects in a Stochastic Frontier Production Function for Panel Data. Empirical Economics 20, 325-332. Battese, G. E., & Coelli, T. J. (1988). Prediction of Firm-Level Technical Efficiencies with a Generalized Frontier Production Function and Panel Data. Journal of Econometrics 38, 387-399. Beesley, M. E. (1999). Airport Regulation. In M. E. Beesley, Regulating Utilities: A New Era? (pp. 35-47). London: Institute of Economic Affairs. Bennett, J. T., & Johnson, M. H. (1980). Tax Reduction without Sacrifice: Private Sector Production of Public Services. Public Finance Quarterly 8, 363-396.  71  Boardman, A., Eckel, C., & Vining, A. (1986). The Advantages and Disadvantages of Mixed Enterprises. Research in International Business and International Relations, 221-44. Boardman, A. E., & Vining, A. R. (1989). Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private,Mixed, and State-Owned Enterprises. Journal of Law and Economics 32, 1-33. Boardman, A. E., & Vining, A. R. (1992). Ownership versus Competition: Efficiency in Public Enterprise. Public Choice 73, 205-239. Bos, D. (1991). Privatization : A Theoretical Treatment. Oxford: Clarendon Press. Caves, D. W., Christensen, L. R., & Swanson, J. A. (1980). Productivity in US Railroads, 19511974. Bell Journal of Econometrica, 65-83. Coelli, T. J., Rao, D. S., & Battese, G. E. (2005). An Introduction to Efficiency and Productivity Analysis. Springer. Cooper, D., & Gillen, D. (1994). Measuring Airport Efficiency and Effectiveness in the California Aviation System. Berkeley: Online Manuscript . Craig, S., Airola, J., & Tipu, M. (2005). The Effects of Institutional Form on Airport Governance Efficiency. On-line manuscript . De Alessi, L. (1980). The Economics of Property Rights: A Review of the Evidence. JAI Press, Greenwich,CT. Debreu, G. (1951). The Coefficient of Resource Utilization. Econometrica 19, 273-292. De Neufville, R. (1999). Airport privatization: issues for the United States. A draft paper, Massachusetts Institute of Technology, Cambridge, MA .  72  Diewert, W. E., & Wales, T. J. (1987). Flexible Functional Forms and Global Curvature Conditions. Econometrica, 43-68. Farrell, M. (1957). The Measurement of Productive Efficiency. Journal of the Royal Statistical Society, Series A, 253-291. Forsyth, P. (2002). Privatization and Regulation of Australia and New Zealand Airports. Journal of Air Transport Management , 19-28. Forsyth, P. et al (2004). The economic regulation of airports: recent deveopment in Australia, Asia, North America and Europe. Burlington : Ashgate Publishing . Forsyth, P. (2006). Airport Policy in Australia and New Zealand : Privatisation,Light Handed regulation and Performance., (p. Paper presented at the Conference"Comparative Political Economy and Infrastructure Performance: the Case of Airports‖). Madrid. Gillen, D. (2010). The Evolution of Airport Ownership and Governance. Journal of Air Transport Management, Forthcoming. Gillen, D.W., Oum, T.H. and Tretheway, M.W. (1985). Airline Cost and Performance: Implications for Public and Industry Policies. Centre for Transportation Studies. U.B.C. Gillen, D.W., Oum, T.H. and Tretheway, M.W. (1989). Privatization of Air Canada: Why it is Necessary in a Deregulated Environment. Canadian Public Policy 3,285-299. Gillen, D.W., Oum, T.H., and Tretheway, M.W. (1990). Airline Cost Structure and Policy Implications: A Multiproduct Approach for Canadian Airlines.  Journal of Transport  Economics and Policy.9-34.  73  Hasenkamp, G. (1976). A Study of Multiple-output Production Functions:Klein's Railroad Study Revisited. Journal of Econometrics, 253-262. Hooper, P. G., & Hensher, D. A. (1997). Measuring Total Factor Productivity of Airports -- An Index Number Approach. Transportation Research Part E 33, 249-259. Hooper, P. (2002). Privatization of Airports in Asia. Journal of Air Trabsport Management 8, 289-300. Hooper, P., Cain, R., & White, S. (2000). The Privatization of Australia's Airports . Transportation Research Part E 36, 181-204. Huang, C. J., & Liu, J. (1994). Estimation of a Non-Neutral Stochastic Frontier Production Function. The Journal of Productivity Analysis 5, 171-180. Knight, F. (1933). The Economic Organization. New York: Harper and Row. Kumbakar, S. C. (1997). Modeling Allocative Inefficiency in a Translog Cost Function and Cost Share Equations: An Exact Relationship. Journal of Econometrics 76, 351-356. Kumbhakar, S. C. (1990). Production Frontiers, Panel Data, and Time-Varying Technical Inefficiency. Journal of Econometrics 46 , 201-212. Kumbhakar, S. C., & Knox Lovell, C. A. (2000). Stochastic Frontier Analysis. Cambridge University Press. Martin, J. C., Roman, C., & Voltes-Dorta, A. (2009). A Stochastic Frontier Analysis to Estimate the relative Efficiency of Spanish Airports. Journal of Productivity Analysis 31, 163-176.  74  Millward, R., & Parker, D. M. (1983). Public and Private Enterprise: Comparative Behaviour and Relative Efficiency. London: Longman. Nyshadham, E.A., and Rao, V.K. (2000), ―Assessing efficiency of European airports: a total factor productivity approach,‖ Public Works Management and Policy, 5, 06-144. Oum, T. H., Adler, N., & Yu, C. (2006). Privatization, corporatization, ownership forms and their effects on the performance of the world's major airports. Journal of Air Transport Management , 109-121. Oum, T. H., Yan, J., & Yu, C. (2008). Ownerhsip Forms Matter for Airport Efficiency: A Stochastic Frontier Investigation of Worldwide Airports. Journal of Urban Economics 64, 422-435. Oum, T. H., Zhang, A., & Zhang, Y. (2004). Alternative Forms of Economic Regulation and Their efficiency Implications for Airports. Journal of Transport Economics and Policy 38, 217-246. Reifschneider, D., & Stevenson, R. (1991). Systematic Departures from the Frontier: A Framework for the Analysis of Firm Inefficiency. International Economic Review 32, 715723. Sarkis, J. (2000). A comparative analysis of DEA as a discrete alternative multiple criteria decision tool . European Journal of Operational Research, 543-557 . Schmidt, P., & Knox Lovel, C. A. (1979). Estimating Technical and Allocative Ineffiency Relative to Stochastic Production and Cost Frontier. Journal of Econometrics 9, 343-366.  75  Starkie, D. (2001). Reforming UK Airport Regulation . Journal of Transport Economics and Policy 35, 119-135. Stevenson, R. F. (1980). Likelihood Functions for Generalized Stochastic Frontier Estimation . Journal of Economietrics 13, 57-66. Thomsen, S., & Pedersen, T. (2000). Wonership Structure and Economic Performance in the Largest European Companies. Strategic Management Journal, 689-705. Vickers, J., & Yarrow, G. (1988). Privatisation: An Economic Analysis . Cambriage, Mass: MIT Press. Yang, X., Tok, S. K., & Su, F. (2008). The Privatization and Commercialization of China's Airports. Journal of Air Transport Management 14, 243-251. Yoshida, Yuichiro (2004), ―Endogenous-weight TFP measurement: methodology and its application to Japanese-airport benchmarking,‖ Transportation Research Part E, 40 (2004), 151-182. Zhang, A., Zhang, Y., & Zhao, R. (2001). Impact of Ownership and Competition on the Productivity of Chinese Enterprises. Journal of Comparative Economics 29, 327-346.  76  Appendices Appendix A: Classification of Airport Governance Structures If one differentiates between the degree and mode of the shift of airports out of government ownership, there are at least seven possible ownership/governance forms: a) Government Ownership: operated by a municipal or regional government department; (e.g. ATL) b) Government Ownership: operated by a local-based and single-purpose management authority via a long term lease; (e.g. YVR) c) Government Ownership: operated by a local-based and multipurpose management authority via a long term lease; (e.g. JFK)14 d) Government Ownership: operated by the airport operator owned by multiple governments; (e.g. AMS) e) Government Ownership: operated by corporatized public airport operators; (e.g. OLS) f) Majority Government Ownership (Mixed Public-Private Ownership): public sectors owning a majority share in the airport operator; (e.g.PEK) g) Private Ownership: private sectors owning 100% or a majority share in the airport operator;(e.g. SYD)  14  In Oum, Alder and Yu (2006), the authors combine category (b) and (c) as one category. 77  Appendix B: The Sample Airports North America - United States Code Airport Name Governance Structures 1 ABQ Albuquerque International Airport Government Branch 2 ALB Albany International Airport Airport Authority 3 ATL Hartsfield-Jackson Atlanta International Airport Government Branch 4 AUS Austin Bergstrom Airport Government Branch 5 BNA Nashville International Airport Airport Authority 6 BWI Baltimore Washington International Airport Government Branch 7 CLE Cleveland-Hopkins International Airport Government Branch 8 CLT Charlotte Douglas International Airport Government Branch 9 CVG Cincinnati/Northern Kentucky International Airport Airport Authority 10 DCA Ronald Reagan Washington National Airport Airport Authority 11 DEN Denver International Airport Government Branch 12 DFW Dallas/ Fort Worth International Airport Airport Authority 13 DTW Detroit Metropolitan Wayne County Airport Airport Authority 14 FLL Fort Lauderdale Hollywood International Airport Airport Authority 15 HNL Honolulu International Airport Government Branch 16 IAD Washington Dulles International Airport Airport Authority 17 IAH Houston-Bush International Airport Government Branch 18 IND Indianapolis International Airport Airport Authority 19 JAX Jacksonville International Airport Airport Authority 20 LAS Las Vegas McCarran International Airport Government Branch 21 LAX Los Angeles International Airport Government Branch 22 MCI Kansas City International Airport Government Branch 23 MCO Orlando International Airport Airport Authority 24 MDW Chicago Midway Airport Government Branch 25 MEM Memphis International Airport Airport Authority 26 MIA Miami International Airport Government Branch 27 MKE General Mitchell International Airport Government Branch 28 MSP Minneapolis /St. Paul International Airport Airport Authority 29 MSY Louis Armstrong New Orleans International Airport Government Branch 30 ONT Ontario International Airport Government Branch 31 ORD Chicago O'Hare International Airport Government Branch 32 PHL Philadelphia International Airport Government Branch 33 PHX Phoenix Sky Harbour International Airport Government Branch 78  North America - United States (cont.) Code Airport Name Governance Structures 34 PIT Pittsburgh International Airport Airport Authority 35 RDU Raleigh-Durham International Airport Airport Authority 36 RIC Richmond International Airport Airport Authority 37 RNO Reno/Tahoe International Airport Airport Authority 38 SAN San Diego International Airport Airport Authority 39 SAT San Antonio International Airport Government Branch 40 SDF Louisville International Airport Airport Authority 41 SFO San Francisco International Airport Government Branch 42 SJC Norman Y. Mineta San Jose International Airport Government Branch 43 SLC Salt Lake City International Airport Government Branch 44 SMF Sacramento International Airport Government Branch 45 SNA John Wayne Orange County Airport Government Branch 46 STL St. Louis-Lambert International Airport Airport Authority 47 TPA Tampa International Airport Airport Authority North America - Canada Code Airport Name Governance Structures 48 YEG Edmonton International Airport Airport Authority 49 YHZ Halifax International Airport Airport Authority 50 YOW Ottawa International Airport Airport Authority 51 YUL Montreal-Pierre Elliot Trudeau international Airport Airport Authority 52 YVR Vancouver International Airport Airport Authority 53 YWG Winnipeg International Airport Airport Authority 54 YYC Calgary International Airport Airport Authority  79  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071624/manifest

Comment

Related Items