Liberalization of the European Railway industry ...



Liberalization of the European Railway industry: Does vertical separation work?Erasmus University RotterdamErasmus School of EconomicsMSc in Economics & BusinessSpecialization: Urban, Port & Transport EconomicsSupervisor: Peran van ReevenAuthor: Ioannis Kougioumtzidise-mail:gianniskougioumtzidis@Student No.: 358557Rotterdam, August 2014“A developed country is not a place where the poor have cars. It’s where the rich use public transportation”Gustavo Petro, Columbian Politician & EconomistAcknowledgementsI would like to thank my supervisor, professor Peran van Reeven for his patience and the inspiring conversations we had. Without his guidance and valuable advice this paper would never be complete. I would also like to thank professor Martijn van der Horst for helping me structure the content of this paper, and transform my idea into measurable output. All errors are mine.Contents TOC \o "1-3" \h \z \u 1.Introduction PAGEREF _Toc396911363 \h 61.1 Background PAGEREF _Toc396911364 \h 61.2 Problem Statement PAGEREF _Toc396911365 \h 81.3 Research Questions PAGEREF _Toc396911366 \h 91.4 Thesis Outline PAGEREF _Toc396911367 \h 102. Literature Review PAGEREF _Toc396911368 \h 102.1 European Union directives on railway reforms PAGEREF _Toc396911369 \h 102.2 Existing railway structures PAGEREF _Toc396911370 \h 142.2.1. Sweden PAGEREF _Toc396911371 \h 152.2.2. Great Britain PAGEREF _Toc396911372 \h 162.2.3. Netherlands PAGEREF _Toc396911373 \h 172.2.4. Germany PAGEREF _Toc396911374 \h 172.2.5. Switzerland PAGEREF _Toc396911375 \h 182.2.6. France PAGEREF _Toc396911376 \h 182.3 Lessons from the world, the ideal case of Japan PAGEREF _Toc396911377 \h 192.4 Pros & Cons of vertical separation PAGEREF _Toc396911378 \h 202.5 Results of vertical and horizontal separation in literature PAGEREF _Toc396911379 \h 213. Data & Methodology PAGEREF _Toc396911380 \h 223.1 Sample PAGEREF _Toc396911381 \h 223.2. Similar Research and Variable selection PAGEREF _Toc396911382 \h 243.3 Data description PAGEREF _Toc396911383 \h 263.3.1 Dependent variables PAGEREF _Toc396911384 \h 263.3.2 Independent variable PAGEREF _Toc396911385 \h 273.3.3. Controlling variables PAGEREF _Toc396911386 \h 283.3.4 Dummy variables PAGEREF _Toc396911387 \h 293.3.5 Regressions PAGEREF _Toc396911388 \h 294. Results PAGEREF _Toc396911389 \h 314.1 Passenger Market PAGEREF _Toc396911390 \h 314.2 Freight Transport PAGEREF _Toc396911391 \h 334.3 Limitations & Discussion PAGEREF _Toc396911392 \h 345. Conclusion PAGEREF _Toc396911393 \h 36References PAGEREF _Toc396911394 \h 37Appendix PAGEREF _Toc396911395 \h 40 AbstractPurpose of this assignment is to map the trends occurring in the railways in the European territory and particularly to identify the impact of the vertical separation of national railway companies on their performance and competitiveness. Railways’ competitiveness has kept on being worsened during the last three decades, a fact that dictated the adoption of reforming measures by the side of the EU in order for the industry to compete with other modes more successfully. The most drastic and widely met reform has been the vertical separation of the integrated national railway companies. By analysing the structure of 28 European railways over the period 2003-2011 we will try to conclude whether the decision of vertically separating a national railway system did manage to increase the modal share and ridership of railways. We find that in the case of passengers vertical separation has been quite an obstacle in their attempt to increase their modal share and ridership, once it has brought the opposite results from the ones expected. Regarding the freight market our sample has not been able to prove any causal relationship. Introduction 1.1 BackgroundDuring the second half of the 20th century Europe has been dominated by publicly owned railway regimes (Cantos et al., 2012). This can be related to the ‘natural monopoly’ characteristics that the railway industry was believed to present (Jupe, Crompton, 2006, Laabsch, Sanner, 2012), a fact that discouraged the enhancement of competition in it. However, the development as well as the competitiveness of the national railway networks was found to be moderate during that period, compared to the other modes. Eventually, the public governance regimes was accused of assisting the establishment of publicly owned monopolies, that in most of the cases have been providing quite moderate services, followed by poor profitability. One of the most prevailing explanations on that has been that an insufficient growth in revenues led to sub optimum outputs (Cantos, Maudos, 2001). During the rising of European Union on the 50's, the member states partly agreed on 'developing a common transport policy in the Treaty of Rome'. Traditionally, the railway industry in Europe has been nationally oriented, though, mainly because of the special features and differences occurring between the states-markets (Marti-Henneberg, 2013). However no significant initiatives, towards that direction, were taken before 1985. The already disadvantageous position of the railways in comparison with other transportation modes was to be worsened after the deregulation that followed in road and air transportation (Barea, Dizy, Ruiz, 2007). The previously existing legislation regarding those two modes of transport was reformed in order to facilitate the cross-border commuting of goods and passengers. On the contrary the cross-border transportation by rail was lagging behind, due to the different national railway regulations. This led to a further reduction of the already small share of the railways regarding the intermodal competition (Stehmann, Zellhover, 2004). Eventually, during the past two decades the European Union has tried to give incentives that will lead to the liberalization of the national railway networks of its member states, through a number of reforms, the essential legislative steps being taken after 1985 (Stehmann, Zellhofer, 2004). From then on, the national railway networks in Europe started considering new governance regimes, in order to render the industry more financially efficient, as well as in an attempt to face the shrinkage of the market share of railways in the international traffic (Nash, 2008).Core of those incentives have been the four railway packages introduced by the Union since the eve of the new century, referring mainly to the freight market, and later also to the one for passengers (3rd and 4th package). The simple previous national integrated structures, of the one company responsible for the provision of the infrastructure as well as for the railway operations, have been replaced by more complicated ones (Thompson, 2003). Policies such as the vertical separation of the infrastructure provision and maintenance with the railway operations have been widely adopted, as well as particular reforms in order to facilitate competition and profitability for railway operators. The rapidly changing environment in the European railways has attracted a lot of attention during these two decades. A lot of scholars have tried to map the trends and investigate whether there is a relation between the reforms and the relative efficiency and productivity of the networks that they are applied into. However, we focus on the performance and attractiveness of the railway industry rather than productivity and efficiency. Additionally, the continuous evolution of the sector as well as the inclusion of even more countries to those with reformed organisational structures, give a reason of existence to our research. In this paper we will examine the impact of vertical separation on the performance of the railways in 28 European countries over the period 2003-2011. The sample contains all the European member states (except Malta and Cyprus once they have no railway network established) plus Switzerland, Norway and Croatia.1.2 Problem StatementDespite the numerous reforms even in big European countries (Germany, Spain, Italy, France), the ownership of the railway operating companies, as well as the responsibility of planning and regulating the network is still on the state. Beria et al. (2010), argues that the conflicting interests that are being generated out of this regime, 'tend to slow down the liberalization process', and harm the productivity of the businesses involved. This opinion is found to be widely supported on the broad literature existing on the topic (Friebel, Gonzalez, 2005). Most of the scholars vote for the liberalization of the market including the privatization of the main railway operators. Nevertheless, in countries that have already done so, like Great Britain, the academic community has been, in some cases, met to raise doubts. Jupe and Crompton in their paper (2006) argue that when it comes to a highly capital-intensive industry, which is dependent on governmental subsidy, privatization is inappropriate. However, this is a subject that will not be covered in our analysis due to the lack of available data.What we will do examine here is the effect of dividing the responsibilities of the incumbent railway companies in Europe, into at least one responsible for the provision and maintenance of the network and another, independent, company responsible for the railway undertakings, on the railways’ performance. This method for many countries constitutes the essential reform adopted in an effort to comply with the European directives, and it aims in an independent network provider, who will be willing to facilitate that use of the network by new railway operators. The independent nature of the network provider would, thus, ensure the minimising of discrimination against the candidate Railway Operating Companies (ROCs) that want to use the network, and in favour of the incumbent company. On the contrary, it would facilitate competition, increased quality of services, as well as international commuting of cargo and passengers by rail.1.3 Research QuestionsAs referred previously the vertical separation of the integrated national railway companies has been a core concern of the EU, and one of the first steps that the member states have adopted on their way of rethinking the structure of railways. Additionally the data available regarding all the other widely met reforms like the introduction of competition, through a range of measures, has not been enough for them to be included in a consistent econometric model. Thus, we will try to measure the impact of the European directives on the development of the railway industry in Europe, by measuring the impact of the most widely met reform proposed, that being the vertical separation. And as two of the main goals of the union in this procedure has been increasing the throughputs of the railways in terms of cargo and passengers in one hand, and improving the railways’ competitive position regarding the intermodal competition, on the other, we will try to give an answer to the following questions:What was the impact of vertical separation proposed by the EU, on the ridership of railways over the period 2003 – 2011?What was its impact on the modal share of railways compared to other modes, over the same period? In order to answer these questions we will apply a panel data analysis regarding all the European Member States that have a railway network established, plus candidate members (Croatia), as well as countries that have agreed to comply with the unions directives regarding the railway sector (Switzerland, Norway). We then use vertical separation as our main independent variable, and examine its impact on the two segments of railway industry, that is passenger and freight transport. For the first research question we regard ridership in terms of passengers-km and tonnes-km respectively, which we set as dependent variables. Accordingly, in our attempt to answer the second research question we replace the dependent variables with the modal share of railways for passenger and freight market. 1.4 Thesis OutlineThe next chapter is dedicated to the existing literature on the structure of the European railways at the moment, the legislation proposed by the EU and the results that it has brought by now according to the academic community. A detailed presentation of some of the most important national organisational structures is also presented there. In the third chapter we present and analyse the data to be used, and in the forth the methodological approach that will be followed in the paper. The fifth chapter presents our results, the limitations of our research and a discussion on our findings and tips for future research. Last, our conclusions are presented in chapter six.2. Literature Review2.1 European Union directives on railway reformsThe intention of the European Commission to revitalize the Union's railway network was put in track with three directives in the first half of the 90's, aiming specifically on the development of the community's railways, on the licensing of the Railway undertakings, on the allocation of railway infrastructure capacity and the charging of infrastructure fees (Holvad, 2006). Key elements in this respect have been:The establishment of independent management for railway undertakingsThe separation of accounts for infrastructure management and transport operationsThe dept restructuring for incumbent companiesAccess provision to railway infrastructure for new companiesAmendments on those three directives, would later (2001) form the base where the first 1st railway package was built on. The package was presented in 1998 and adopted in December 2000, and was mainly focused on the freight railway transportation. Its initial purpose was to revitalize the international goods transportation by rail in the Community, and to lead in a more integrated European railway network where the cross-border services would be facilitated. It was meant to be transposed in the Member States until 15 March 2003 and its particular goals were:the vertical separation of the basic functions of the railway industry, namely the infrastructure provision and the actual railway operationsthe establishment of an independent railway regulator in each Member State and the reduction of the governmental influence in the sectora reconsideration of the technical barriers between the different railway networks so that the international freight movements would be easeda gradual opening of the European rail freight network to all operators that have been licensed according to the Community's regulations. Within seven years (as of March 2008) from the adoption of the Council Directive, the aforementioned operators should have access to the whole international European railway freight networkHowever, despite the efforts put by the Union, it appeared that this gradual opening that would allow operators from all member states to use the ‘integrated European’ network was not as easy as was initially believed. At the moment there are countries that still lag behind and fail to follow the Union's directives regarding the international freight. (Holvad, 2006) Finally, an additional measure came to complete the 1st package in March 2001.The 2nd railway package was proposed by the union on the 30st of April 2004, and contained legislative proposals that, among others, pointed on:Proposing a common strategy to guarantee a greater degree of rail safety and establishing a European Railway Agency in charge of these issuesEnhancing the interoperability principles proposed with previous directives (safety and interoperability)Accelerating the opening of the rail freight marketJoining the Intergovernmental Organization for International Carriage by Rail (OTIF)After the 2nd package, the new date agreed for the opening of the rail freight markets would be the 1st of January 2007.Two more railway packages were adopted by the parliament during the years to follow, this time mainly focused on the passenger market. The 3rd railway package (2007) aimed in the opening of the passenger market to international services. In more detail, it proposed that starting in 2010 all the passenger service operators that have a licence and meet the safety preconditions should be able to provide international services in the Community. Two more particular legislative measures are included in this 3rd package, one on the protection of the passengers’ rights when using international services, and the other on the certification methods of the railway drivers.Finally, the 4th railway package was adopted by the European parliament on 26th of February 2014, and aimed to ensure the competitiveness of the railway sector and improve the quality of rail services by:enhancing competitioneliminating remaining market distortions and suboptimal structuresalleviating long and costly proceduresremoving access barriers for new entrants, andharmonising market access rules in the Member States.This last package is still to be confirmed by the European Council. 2.2 Existing railway structuresSumming up the directives sited above we can identify three basic axes of restructuring where the interest of European Union was focused on. That is the vertical, examining the relationship between the infrastructure provision and maintenance in one hand, and the railway undertakings on the other, the horizontal, examining the relationships between railway operating companies (ROCs) functioning on the same network; and a third one, regarding the existence or not of an independent regulatory authority, which would ensure the minimising of the ‘free rider’ behaviour from the side of the national holding integrated railway companies. Thus, it is obvious that vertical separation has played a very important role in the restructuring Europe’s railways, and is considered as a necessary condition for the industry to achieve a more competitive performance and minimize the negative effects in terms of governance and ownership. However, despite the fact that those directives are meant to be applied in the same way, significant differences can be identified in the way that MS’s have implemented them (European Commission, 2006). Hence, the implementation of the revised legislation on the railways in Europe has evolved to give birth to a number of different, and sometimes unique, structures, with varying levels of separation as well as induced competition, and a number of different policies adopted regarding the governance and independence of the particular national railway regulatory authorities. Table 1 (Appendix) summarizes the current situation in the countries of our sample, regarding the three aspects of reforms mentioned above. As we see there, 13 countries in the sample have either retained the traditional integrated model, or have adopted partially separated structures where two companies function under the same holding, but with partial division of key powers (Nash, 2008). 13 countries have adopted the fully separated model, while two countries have applied a hybrid model (France and Czech Republic), which will be further analyzed later in this chapter. There are cases where there is no vertical separation, but there is established within-rail competition (e.g. Germany, Austria, Italy), by licensing new operators to function in the same network with the incumbent company (Campos, Cantos, 2000). However, the enhancement of competition hasn’t experienced the progress that EU might have expected, once only seven countries (Netherlands, Germany, Austria, Italy, United Kingdom, Denmark and Sweden) have managed to establish ‘high’ levels of competition according to IBM (2011). . In Italy 49 have been licensed, though only 16 of them are operating (2009). For France and Spain the situation is pretty monopolistic, regarding the passenger market, once almost the absolute number of passengers is still being served by the incumbent company. The freight market is more promising, once in Spain the new entrants managed to serve 5% of the market by 2008, and in Italy 13% (Beria et al., 2010) Finally, growing independence is recorded for the railway regulatory authorities, once they are still directly controlled by governmental ministries only in six countries (Spain, Ireland, Romania, Estonia, Lithuania and Slovenia). After mapping the structures that can be met in a contemporary national railway network, we can now proceed in presenting a more detailed description of some of the most important national networks in Europe.2.2.1. SwedenSweden has been the first country to separate its holding railway company in Europe in 1988 (Nash, 2008), that is even before any initiative was taken by the side of the EU. Thus, the Swedish State Railways (SJ) was separated so that two companies were created, SJ, basic rail operator and Banverket, responsible for the network maintenance. Since then a raw of sequential reforms have developed the Swedish railway in what it is today. The vertical separation was followed by the opening regional railway undertaking services one year later (1989). The opening of interregional services was to follow in 1993 and the deregulation of the freight services in 1996 (Alexandersson, Rigas, 2013). Particular policies aiming mainly on inducing a greater degree of competition kept on being introduced by 2012.2.2.2. Great BritainThe second country in Europe to introduce vertical separation was Great Britain in 1994. The restructuring of the network at that point included also the opening of the market, expressed in terms of competition for the market, through competitive tendering (franchises). The incumbent British Rail, was thus fragmented to lead to the creation of several operators that through franchising would provide regional and interregional railway undertaking services initially through agreements that lasted for 7 to 10 years (the franchising periods were later expanded), and a private infrastructure manager, Railtrack. Seven years later though (2001), and after the failure of Railtrack to provide satisfactory results in the management and maintenance of the network without receiving governmental subsidy, the government decided its abolition. The new infrastructure manager would be Network Rail, a non-for-profit organisation, consisted by several public as well as private bodies, whose role was through governmental subsidies to fill in the gap created by their predecessors. The independent regulatory authority, Office of Rail Regulation (ORR) is supervising Network rail in a 5-year basis, to control its performance and decide on the new infrastructure charges. The results and opinions on this groundbreaking reform have been contradicting. The advocates of the reform underline that the ridership and safety was substantially increased after the reform, while there was a growth in rail services and significant modernization on the rolling stock (van de Velde, Rontgen, 2009). The other side though quotes that investment in rail services has been moderate, while there was no improvement in their quality, and that freedom of operators has led franchises functioning not in the interest of passengers (excessive pricing), (Jupe, Crompton, 2006).2.2.3. NetherlandsVan der Velde and Rontgen (2009) identify in the Netherlands the country that more than any other in the European territory achieved to be in total conformity with the First railway package that EC suggested in order to steer the European railway companies closer to financial success, by changing the legislation according to which the railway companies in Europe should function. Following a two step reform, Dutch railways are today vertically separated into two major companies. An 100% state-owned infrastructure provider (ProRail), and an 100% state-owned railway operator (Nederland Spoorwegen, NS), though organized according to private law (Van der Velde, 2009). Competition in-the-market has been successfully introduced in both passenger and freight markets. 2.2.4. GermanySince 1994 the railway operations in Germany are mostly being handled by Deutche Bahn (DB). DB is a 100% state-owned non-subsidized company, organized according to private law. DB operates under several brands (e.g. DB Fernverkehr provides long distance transport), while other private companies also provide long distance passenger transport services, though in a limited degree up to date. The infrastructure is owned and managed by DB Netz, a company functioning under the same holding, though with separated key powers. In comparison to the Dutch reality there are no contractual relations between the infrastructure provider and the operating companies and DB Netz is obliged to allow entrance to all competitors that are able to pay the infrastructure charge to be foreseen by law. Thus, Germany had already licensed 350 rail operators in 2008, when 330 of them are active. Though one of the most successful networks in Europe, a wide discussion has taken place during the previous years, considering the partial privatization of DB. However the procedure has been postponed for the time being.2.2.5. SwitzerlandA special feature of the Swiss railways is the fact that despite the numerous reforms that have taken place in the railway market of the country, the major railway operator and the infrastructure manager are still integrated, although they have to maintain separate accounting departments. Hence, the main rail operator, Schweizerische Bundesbahnen (SBB) is also the infrastructure manager, in order to avoid the need for contractual relationships between the two aforementioned bodies and increase efficiency. The department of SBB responsible for infrastructure is obliged to provide specific services through contracts that they sign with the governmental organizations in charge for the facility, safety and capacity provision (Bundesamt für Verkehr, BAV). SBB reports to BAV twice a year regarding the level of realization of the goals that were set in advance, according to 30 performance indicators that have been previously agreed. SBB is granted by BAV a 10-year concession for long distance passenger transport and operates in regional level as well. SBB, so as other small companies that operate in regional level, have to behave as private companies, though most of them (including SBB) are 100% owned by the Federal Government. What worth to be highlighted here is the introduction of competition in the market without privatization. Much attention seems to be given to the emulation of different companies' services, while further reforms are not to be expected (Van de Velde, 2009) once the network is functioning satisfactorily.2.2.6. FranceAs mentioned earlier in this chapter France has adopted a very special structure regarding the relationship between the different bodies in the railway industry. The infrastructure management, maintenance and development is being held by The French Railway Network (Reseau Ferre de France, RFF) which is a public company functioning with commercial orientation. The railway undertakings are being held by a separate company, SNCF (Societe Nationale des Chemins de fer Francais), while there are smaller organisations with specified regulatory responsibilities, like general competition regulation (Conseil de la concurrence), safety regulation (Etablissement public de securite ferroviaire), or network access issues (Mission de controle des activites ferroviaires). However, despite the fact that there is vertical separation introduced, RFF is expected to delegate some of its infrastructure management tasks back to SNCF, rendering this way SNCF as a delegated infrastructure manager. This feature underlines the uniqueness of the French model (something similar can only be met in Czech Republic). SNCF has also a commercial orientation and is asked by the State to generate profitability by the high-speed and the long distance services. But what is important here is the legislation existing regarding the social fare obligations of SNCF. According to them, the company is being subsidized from the particular regions in order to provide non profitable regional services that are necessary in social terms. Those amounts paid by the regions are derived by the French State. Even though France is considered to be lagging behind regarding the conformation with the European directives, according to the IBM’s Liberalization index (2011), it seems to secure the social character of railway transportation more than any other country. Also, despite the fact that the competition in the French market is still in very low level, due to the opposition of the public opinion and trade unions (van de Velde, Rontgen, 2009), there is evidence that French railways have in recent years increased performance and ridership.2.3 Lessons from the world, the ideal case of JapanThe process of restructuring railways was not initiated in Europe. A long time before such a need emerged here, same problems regarding moderate performance and reduced attractiveness for the railways have occurred in many sides of the world (e.g. Japan, U.S., Latin America and Africa), resulting in in-depth reforming procedures aiming in an improved performance as well as efficiency of the particular railway networks. However in some cases the path to success has not been the one that the EU is following. A quite representative example is Japan.The Japanese National Railways (JNR) was in 1986 divided into six vertically integrated companies orientated to provide regional services in six different regions of Japan. These companies are obliged to create profitability and maintain the operations as well as the infrastructure provision on their own cost, as it is prohibited for them to be subsidized by the government. The only case where separate infrastructure companies can be seen is when the network necessitated significant investments that was impossible to be held without governmental support. The creation of those infrastructure companies helps the government to indirectly capitalize on the operators and the needs they have on the network.After 28 years functioning under this regime the six operators (three private and three public) have been performing under absolute success and there are no thoughts for further reforms. Their concessions are not of limited time, and there is no competitive tendering for the creation of on the tracks competition (van de Velde, Rontgen, 2009), while there never were thoughts of implementing vertical separation.2.4 Pros & Cons of vertical separationThe increasingly wider adoption of vertical separation as well as enhancement of competition in railways has been followed by a growing debate on the suitability and efficiency of those reforms applied. We present here the basic points of the supporters of traditional vertically integrated railway companies, opposed by those supporting the vertical separation. A basic advantage of vertical separation as mentioned in the literature is that it installs the necessary preconditions for the enhancement of competition-in-the-market (Kurosaki, 2010), by establishing an independent infrastructure manager, who would ideally provide equal opportunities to new ROCs. An incumbent integrated company is more likely to discriminate against the new operators, basically by claiming unfair access charges in order for them to be licensed for functioning on the network. Another opportunity for discrimination from the side of the incumbent can possibly be pretentious delays in the process of licensing. However, even in the case that actual discrimination does not exist, the possibility of its existence alone might discourage new operators from entering the market (Laabsch, Sanner, 2012). Following the logic of vertical separation supporters, competition when enhanced will lead to lower prices and higher quality of services, thus revitalising the attractiveness of railways. Furthermore, the externalisation of transactions will increase transparency regarding the usage of governmental subsidies on infrastructure. However, problems regarding the access rights and charges might still remain (Holvad, 2006). Another advantage mentioned in the literature is that it promotes specialization, once when fragmenting the market the bodies created have more space to develop on their own field. Finally, it is argued that separation would bring the industry one step closer to another basic goal of the EU, which is the creation of a single European transport area, once separated infrastructure managers would co-ordinate with each other more efficiently.On the other end, the case of the integrated model creates undoubtedly a cost advantage by exploiting synergies and lower transaction costs. Williamson (1981) discusses the creation of cost advantages in the case of a holding company, while he underlines that co-ordination between the different actors is also enhanced, when parts of the same company. Another argument of Williamson is that in the case of separation the contracts to be created between operators and infrastructure managers are pretty likely to be incomplete. The bounded rationality which characterizes individuals makes it possible that there will be areas to remain unlighted, a fact that can possibly create conflicts in a case of disruption later. Additionally, vertical alliances could facilitate the achievement of high levels of efficiency, in order for the railways to compete with, for example, road transport more effectively. That would eventually render the railways more attractive and increase their market share. As both sides claim an increase in the modal share of railways in the case of applying their preferred model, it is interesting to examine in the following chapters the direct impact of vertical separation in Europe on the modal share of railways itself.2.5 Results of vertical and horizontal separation in literatureThe literature on the impact of vertical and horizontal separation on the performance of the industry has been rather limited, while most of the studies examine their particular impact on the efficiency of the railways. Its results have been contradicting, depending on the period that is examined in each study, the number of countries included, and the concept of efficiency considered (Cantos, Maudos, 2001). Cantos et al. (2012) conclude that horizontal separation, in the sense of introduction of new freight or passenger operators, increases efficiency, while their results about the impact of vertical separation are not clear. Furthermore, their dataset supports the idea that countries that have adopted both horizontal and vertical separation tools present the most efficient railway regimes, while Friebel, Ivaldi and Vibes (2010) in their paper for evaluating the effects of railway reforms in Europe for the period 1980 - 2000 conclude that in general these reforms have had a positive impact in both passenger and freight market. In the same paper attention is also being drawn on the way that reforms take place, suggesting that sequential reforms are more likely to have positive results, than all-in-once reform packages.On the contrary, Laabsch and Sanner (2012) conclude that vertical separation hasn’t had any positive impact on the passenger segment, once countries that have retained the integrated model perform much better in their sample. Regarding freight transport the find no robust results, in an analysis that examines the performance of railways of nine European countries for the period 1994-2009. 3. Data & Methodology3.1 SampleFor analysing our research questions in the paper a dynamic panel data analysis will be used. A panel of 252 observations from 28 countries will help us analyse the impact of vertical separation on the performance and competitiveness of the European railways, as analysed above. Our sample consists of all the European member states (excluding Malta & Cyprus once they do not have an established railway network), plus Switzerland and Norway, who have agreed to follow the European Union directives in an effort to be part of the ‘single European transport area’ that the Union aims to establish. These countries are analysed over the period 2003-2011, a period that represents the years of adoption and implementation of the first three railway packages proposed by the EU, of which vertical separation constitutes a significant component.The initial idea was to analyze both reforming dimensions, those being vertical as well as horizontal as presented in Chapter 2, in order to provide a more spherical insight on the impact of the reforming process on railway performance. However, the only consistent dataset that would help us proceed with analysing the changes in horizontal terms (level of liberalisation and induced competition), is the Rail Liberalisation Index conducted by IBM, which is available only for the years 2004, 2007 & 2011. Using IBM’s index would mean that we would have to reduce the number of our observations to 84 (now 252), thus reducing significantly the validity of our results. Consequently, we preferred the complete exclusion of the horizontal dimension from our analysis, in order to provide more solid and consistent evidence on the impact of the reconsideration of vertical relationships in the industry.We will try to draw conclusions on the freight as well as passenger market, once they were both affected by the vertical separation. By building two equations for each market, we will examine vertical separation’s impact on: a) performance and, b) competitiveness of the railways. The first by examining the variance of passenger-km and tonnes-km for passenger and freight market respectively, according to the decisions taken in terms of vertical separation, and the second by examining its impact on the modal split of railways, regarding the competition with air and road transport (as well as inland waterways were applicable). The data to be used in the analysis were taken from Eurostat, as well as World Bank and the IBM’s Rail Liberalisation Index 2011. More details about the data sources can be found in the Appendix (Table 2).3.2. Similar Research and Variable selectionIn this chapter we present the empirical research that has taken place on the very field of railway structures and reforms by now, and we explain the approach we choose to answer our research questions and the reasons why we do so.The majority of the work done in the field has tried to examine the impact of reforms on the railways’ productivity and efficiency. Friebel, Ivaldi and Vibes (2010) tried to measure the impact of reforms in European railways on the technical efficiency of the railways. In order to achieve that they use an input-output analysis, by applying a Cobb-Douglas function which implicitly assumes separability between inputs and outputs. As inputs they use the length of lines in the network and the number of employees, and as outputs the passengers-km and tonnes-km for passenger and freight market respectively. Their sample regards 11 European countries for the period 1980-2003. In the physical data referred earlier, they attach the three kinds f reforms met in Europe, namely the separation, the third party entry (competition) and the existence of an independent regulatory authority. Their results suggest that railway reforms have increased railway efficiency, and that reforms are more successful when applied sequentially, instead of all-in-one. Cantos, Monsalves and Martinez (2008) applied a Data Envelopment analysis to count the impact of reforms on both productivity and efficiency, using a sample of 16 European countries during the period 1985-2004. They find that both efficiency and productivity in railways have been positively influenced by vertical separation, even though the effect seems to be stronger when separation is accompanied be enhancement of competition. The variable used here as inputs are length of lines and employment as before, as well as rolling stock. As outputs we see again the ridership of railways in terms of passengers-km and tonnes-km respectively. Once again the same input and output variables are used by Driessen, Lijesen and Mulder (2006). In this case there is a more detailed inclusion of the reforms, once separation is divided to total or partial (separation of accounts), and competition to competitive tendering (competition for the market) and free entry (competition in the market). A dummy regarding the management independence from the government is also included in the model, while among others they control for GDP per capita and population density. Via the use of Data Envelopment Analysis, they conclude that competitive tendering has a positive impact on the productive efficiency of the railway. As for the rest, results regarding the vertical separation are not clear, while the introduction of in-the-market competition and the managerial independence seem to have the opposite results.While all the papers referred above are measuring the impact of reforms on either productivity or efficiency, Laabsch and Sanner examine their impact on merely the ‘success of rail’, that is their performance and attractiveness. Following the work of Drew and Nash (2011), they use as dependent variable the modal share of railways. According to them modal share measurement can take into account several reasons that can render the trains more attractive, except the lower prices arising due to cost reduction in a network with recently increased efficiency. Thus, the impact of higher quality of service can also be taken into account when setting modal share as a dependent variable. In addition, both supporters as well as adversaries of vertical separation argue that their preferable organisational structures are going to increase the modal share of railways if applied. This is a fair reason to examine the impact of the reforms directly on this measure. Their model also controls for other factors that can possibly influence the success of railways, such as GDP per capita, public budget contributions on railways, as well s share of rail in total transport budgets. The latter can give a good insight about the tendency of the government to assist railways on their effort to compete successfully with other modes. Interestingly, they conclude that separation has a negative effect on the modal share of rail in the case of passengers, while in the case of freight transportation their results are not reliable.This paper has two goals. In the first place we intend to, in line with Laabsch and Sanner, examine the impact of vertical separation on the attempt of railways to compete more successfully with other modes. For this purpose we also consider modal share of railways for passenger and freight transport as our dependent variables. Additionally, we want to have an insight on the impact of vertical organisations on the trains’ ridership itself, and that is why we run the same models, this time considering as dependent variables the passengers-km and tonnes-km for the two market segments of railways. Admittedly, the inclusion of public budget on railways and the share of rail in total transport budgets, as can be met in the model of Laabsch and Sanner, would render our model much more comprehensive. However, the lack of data did not allow us to do so. Consequently we expect the explanatory power of our model to be limited. Instead, we include controlling variables regarding the general economic coincidence, such as the GDP per capita. Finally, we control for population growth, once it is beyond doubt that variances in population can cause, though in a limited degree, similar variances in the ridership of rail.3.3 Data description3.3.1 Dependent variablesPassengers_Km / Tonnes_KmRepresents the millions of kilometres travelled multiplied by the number of passengers/tonnes (1 unit being: 1 kilometre travelled by 1 passenger/ tonne of product). This will be our dependent variable in the model examining the performance of passenger/freight railway market, once it manages to give a good insight on the variance in the traffic of passenger/freight market over the given period. Its relevance comes to the fact that increasing the usage of trains by passenger (and cargo) transporters has been of core concern for the EU. Cantos, Monsalves and Martinez (2008) are using passengers-km/tonnes-km, though in a different context, in an effort to measure the productivity of the railways. We consider that an increase in the throughput of railways in these terms is indicative of the successful (or not) implementation of organisational reforms, while it has also been an explicit goal of the legislation proposed by the European Union. As this variable appears in very big values, in our regressions we will use their logarithm in an effort to shrinkage the values in order to ease the procedure. Thus, passengers-km and tonnes-km in our models will be met as log_pas_km and log_tonnes_km respectively.Modal share of railwaysSo the supporters as the opponents of vertical separation argue that their preferred organisational models will increase the modal share of the railways. In that sense it would be interesting to examine how this measure fluctuates in countries that have adopted vertical separation compared to those that have not done so. Furthermore, increasing the modal share constitutes one more explicit goal of the EU. Laabsch and Sanner (2012) are using the modal split of railways as a dependent variable in a similar research examining the effect of vertical separation on the success of railways, regarding nine European countries examined over the period 1995 – 2009. 3.3.2 Independent variableVertical SeparationVertical separation is included in our analysis in the form of a dummy variable, which takes value 0 when not applied and switches to 1 when adopted by the given country. This classification is proposed by the European Commission (Laabsch, Sanner, 2012), although in our sample the countries adopted the ‘hybrid model’ (France, Czech Republic), are taken into account as separated structures, once an elementary form of separation has occurred in their organisational form. In part of our sample it can be found as time-invariant variable, for the countries that have either adopted the separation before the period we examine, or for countries that did not apply separation yet, at all. It is, though, time-variant for the countries that adopted the separation during the period 2003 – 2011. Cantos et al. (2008) have included vertical separation as an explanatory variable, in their paper examining the effect of the two-dimension reforming process on productivity of the railways.3.3.3. Controlling variablesGDP per Capita The lack of available data that could adequately control for factors which directly affect the performance and competitiveness of the railways, forces us to control our results through the general economic coincidence of the countries and period under examination. Laabsch and Sanner (2012) in their paper on the impact of vertical separation on the success of railways, control for GDP per capita, based on the assumption that ‘unemployment risk differs between users and non-users of public transport, or because the production of categories of goods that are more or less rail-affine is affected differently by a crisis’. Other reasons for using GDP per capita as a controlling variable could be the fact that an upward trend in the economy could justify the reduction of attractiveness of the rail, due to a possible increase in the car usage, or even contrarily that more developed economies can possibly present a better capacity in implementing such reforms. In the same logic with passenger-km and tonnes-km we will use the log of GDP per capita, from now on met as log_gdpcapita.Population growthAnother controlling variable that we include in our models will be the population growth in yearly basis. The reason is to control for the small changes in railway demand that might be caused by changes in population. We decide to include population in the form of population growth and not in absolute values, after running a test for the existence of a unit root in our variable. In order to achieve that we will use the Fisher- type unit root test, once it performs well in unbalanced data, which is the case in our sample. Our null hypothesis is that population has a unit root, which would be an adequate justification for non-stationarity. Our results show that we cannot reject the null hypothesis so we accept that population has indeed a unit root. Because of the fact that population is a non stationary variable, it would be here inappropriate to use population in absolute values. 3.3.4 Dummy variablesFinally we add a dummy variable (_Iyear_t) in order to account for events that possibly cause a simultaneous variation in all countries across our sample. Such an event can be a simultaneous downturn in the demand for cargo transportation due to the financial crisis of 2008 – 2010. We then have n-1 (8) new variables created, one for every year, having the _Iyear_2003, which is the first year analyzed in our sample automatically dropped by stata.3.3.5 RegressionsRegarding the fact that we use population in terms of population growth and not in absolute values, it is impossible to take into account the individual heterogeneity in terms of the population levels, as well as the other individual country-specific characteristics in our sample. Thus, this can be achieved only by using the fixed effects estimators, which help us account for individual heterogeneity.We then have for equations created in order to answer to our research questions, which all read under the form:Yit = β0 + β1 vertical_separationit + β2 log_gdpcapitait + β3 d_population it + γn Σ_Iyear_t + uitWhereYit represents our four dependent variables, one for each regression that we will run and those are:-mod_sh_pas is the modal share of railways in the passenger market, where i=country and t=year-log_pas_km is the logarithm of passenger-km -mod_sh_freight is the modal share of railways in the freight market, and-log_tonnes_km is the logarithm of tonnes-kmβ0 is the constantβ1- β3 are the coefficients of the independent and controlling variablesγn are the coefficients of the dummies created, anduit is the error termThe descriptive statistics of our dataset are presented in the Appenidx A (Table 4).4. Results4.1 Passenger MarketIn this chapter we present the tables with the results of our analysis for both markets and their interpretation in our view. In Table 5 we can see the results regarding the passenger market. What we see there is that somewhat surprisingly the impact of vertical separation on the performance and competitiveness of the passenger railways is not positive as expected, but negative. Table 5.Results - Passenger market Variablelog_pas_kmmod_sh_pasvertical_separation (P values)-0.070(0.018)**-0.008(0.000)***log_gdpcapita-0.221(0.001)***-0.027(0.000)***d_population 4.19(0.017)**0.100(0.417)_Iyear_2004-0.171(0.000)***-0.0005(0.781)_Iyear_2005-0.122(0.000)***0.001(0.398)_Iyear_2006-0.06(0.019)**0.006(0.003)**_Iyear_2007-0.011(0.624)0.009(0.000)***_Iyear_20080.021(0.349)0.116(0.000)***_Iyear_2009-0.043(0.082)*0.008(0.000)***_Iyear_2010-0.031(0.179)0.010(0.000)***_Iyear_2011(omitted)0.010(0.000)***Constant 10.75(0.000)***0.330(0.000)***Prob > F0.0000.000R2(observations)0.25 (207)0.24 (252)*/**/*** significant in the 10/5/1 per cent levels, respectively.The results show that vertical separations’ impact on both passengers-km and modal share of passenger railways is statistically significant in the 5% and 1% level respectively, with a marginal negative effect. Even though the small absolute values of the coefficients show us that the magnitude of the causality explained is not big, however the results verify that the effort put by the union to give incentives for fragmentation of the integrated railway authorities into smaller entities with more specialized responsibilities, has not brought the results expected for the moment, at least in the sense of increasing the attractiveness of railways. This could possibly be explained by the fact that the passenger market is rather quality driven. In a separated structure the competition will eventually lead to lower prices. Subsequently the quality of services is more likely to deteriorate than in a case of vertical integration. Following that logic, if we take into account that passengers are quality driven, they would possibly switch to another transport mode, e.g. private car usage, in case they do not receive the desired service quality. It can be also caused by the fact that communication and policy implementation can be facilitated by integrated responsibilities under the same holding, when such tasks can be interfered by possible lack of communication in a case of separated infrastructure management with railway undertakings. Accordingly, GDP per capita has also a negative and significant (1% level) impact on both passenger-km and modal share in our sample. This makes sense if we think that an increase in the population’s wealth could justify a switch from train to car usage or even air transportation for longer-distance travelling.On the other hand, population growth variations seem to be statistically significant only in the case of passenger-km. As we expected population is positively correlated with the creation of extra passenger-km, and statistically significant in the 5% level. It is also logic that no significant causality is observed in the relationship of population growth and modal share of rail, once there is nothing to indicate that newly settled population would make a different choice than the one existing before, in terms of their preferable transport modes. 4.2 Freight TransportIn the case of freight transport, vertical separation does not present any significant causal relationship with tonnes-km generated or variance in modal share of rail freight transportation. In both cases the coefficients are positive but not statistically significant. That could mean that either the vertical separation has been unsuccessful, or that it is still pretty early in the procedure for the results to be seen, in the sense that only 15 out of 28 countries of our sample have implemented the measure for the moment.Table 6. Results - Freight Market Variablelog_tonnes_Kmmod_Sh_freightvertical_separation (P values)0.024(0.641)0.001(0.885)log_gdpcapita0.089(0.447)-0.181(0.000)***d_population 0.603(0.852)0.242(0.651)_Iyear_20040.122(0.011)**-0.013(0.087)*_Iyear_20050.097(0.044)**-0.009(0.231)_Iyear_20060.120(0.016)**0.004(0.639)_Iyear_20070.126(0.019)**0.014(0.115)_Iyear_20080.055(0.319)0.014(0.132)_Iyear_2009-0.150(0.004)***-0.008(0.386)_Iyear_2010-0.053(0.322)0.005(0.551)_Iyear_2011-0.024(0.671)0.020(0.042)**Constant 7.98(0.000)***2.004(0.000)***Prob > F0.0000.000R2(observations)0.26 (238)0.40 (251)*/**/*** significant in the 10/5/1 per cent levels, respectively.This fact, does not let us have a clear view of what the results would be in a case of complete implementation of the measures proposed by the EU from all countries. However, disregarding that vertical separation is reported as insignificant, its coefficients are positive. This is in line with the fact that freight is more price-intensive, which means that any reduction in prices, as the one caused by the enhancement of competition, can have a positive impact on the ridership of freight trains, or their modal share. The population growth is also not significant in both regressions, while GDP per capita presents a significant, on the 1% level, negative effect on the modal share of freight railways. That could be explained in the sense that ‘richer’ countries would be more willing to switch to alternative transportation methods, in order to achieve the higher quality possible and avoid operational difficulties that have been surpassed in other modes like road and air transportation, but not in the case of railways.It is worth noting here that, regarding our dummy variable, we can see in the year 2009 a significant (1% level) and negative effect on the tones-km of products transported across Europe. That is quite reasonable once 2009 follows the outbreak of the global financial crisis in 2008, which caused a great reduction in the cargo transported globally.4.3 Limitations & DiscussionEven though modal share of railways represents a very good measure regarding what we examine here, its variation can be rather small, especially regarding a short period like the one in our sample. This might be caused by the fact that the habitual behaviour of users is enough inelastic in the short term, so that does not represent the actual change that would follow a structural change in railways. On the other hand, in the right side of the equations, which remains unchanged in all four, except the vertical separation we control population growth and the GDP per capita. Even though these two measures represent two strong demographic features that can have good explanatory power, there are no controlling variables directly related with the railway performance. This can reduce the validity and explanatory power of our analysis. More controlling variables that definitely influence the performance of railways and their path to success could be taken into account, like the public subsidies given provided by the government, but the inclusion of such measures has been impossible due to the lack of available data. Furthermore, an increase in ridership as examined here by using passenger-km and tonnes-km respectively as dependent variables, does not represent an improved performance when accompanied by a larger increase in public investment. Public investments constitute a leading factor in improved performance of railways, and can lead to catastrophic results when missing, as was the case of Railtrack in Britain during 1994 – 2001 (Jupe, Crompton, 2006). That is why their inclusion in a model like this is considered by the author very critical.Corrupted behaviour should also be taken into account in this point. The attitude of the incumbent national companies can play a very significant role in the way that the proposed reforms will be applied in a given country, and the effort that will be put in such a reforming initiative (Quinet, 2006, Cantos, 2006, Beria et al., 2010). The level of corruption and the ‘free riding’ behaviour, have played a very important role in maintaining structures that serve personal interests of the incumbents in the railways, while opposing to any possible reforming initiatives that point on minimizing this kind of events.The level to which the proposed legislation has been adopted by the union as a whole, is very important for us to understand the concept of the European case. As we have already mentioned in the second chapter, one of the long term goals of the European Union is to establish a united rail area with similar legislation, were operators would be facilitated in commuting passengers and cargo from one country to another. Taking into account the special features of different modes of transport, one could assume, though not unambiguously, that the value to be added in the attractiveness of railways, if international commuting of cargo is facilitated, can be much larger than the one in the case of passengers. And that is, in one hand because the cost effectiveness that can be generated in the case of cargo can successfully compete the one of air transportation, due to the inability of air transportation to commute large quantities, and on the other because in the case of passengers it is very difficult for the railways to compete with air transportation in long-distance travelling, in terms of pricing and travel duration. In that sense the results of vertical separation could be better evaluated in a later stage, when the directives of the Union will be adopted by more countries; at least in the case of freight transport where international transportation can be more significant. For the moment the reforms may occur in some countries, but this can only have a small impact on increase of cargo commuted internationally, until the legislation is implemented in the majority of the Member States.5. ConclusionThis paper has evaluated the impact of vertical separation on the performance and competitiveness of railways in the European context, by examining the extent to which there is correlation between vertical separation and variations on passenger-km and tonnes-km generated on one hand, and modal share of railways for passenger and freight market respectively, on the other. In the case of passenger market we find that vertical separation might interfere the generation of passenger-km as well as an increase in the modal share of railways. This might be caused by the fact that reduced quality in services due to lower prices encourages passengers to switch to other modes of transport. Another reason could be miscommunication between the interested bodies that is less likely to exist within the same holding. In the case of freight we do not see any significant causality from the side of vertical separation on neither generation of tonnes-km or variation on the modal share. The effect of vertical separation might be more significant though, in the case of a wider adoption of the policies proposed, which would lead somewhere closer to the integrated railway area that the Union is dreaming. For the moment only 15 countries out of the 28 of our sample have adopted vertical separation, while particular legislative differences still exist in the national level, a fact that renders it difficult for international freight to develop in the way that EU wants. Conclusively, if we want to take a look on the canvas of our findings combined with the findings of previous research on the vertical relationships in railways, we would say that the procedure of reforming the European railways has still a long way to go until it achieves its goals. Plus, the paths that are being followed in that direction are not necessarily the correct ones. The vertical separation which has been fervently supported by the EU, does not seem to be as socially desirable as expected, at least in regards with the passenger market. And that is in line with the fact that European Union managed to include in their planning the social responsibility of railway transportation only as late as the 4th railway package was proposed (2014).ReferencesAlexandersson, G., Rigas, K., (2013), ‘Rail Liberalisation in Sweden. Policy development in a European context’, Research in Transportation, Business & Management 6, 88-98Barea, J., Dizy, D., Ruiz, O., (2007), ‘The New Model Of The Railway Industry In Spain Within The European Framework’, Annals of Public and Cooperative Economics 78:3, 353-380Beria, P., Quinet, E., de Rus, G.,Schulz, C., 2010, 'A Comparison of Rail Liberalization Levels Across Four European Countries', 12th WCTR, July 11-15, Lisbon, PortugalCantos, P., Pastor, J.M., Serrano, L., 2012, Transport Policy 24, 67-72Cantos Sanchez, P., 2001, "Vertical Relationships for the European Railway Industry", Transport Policy 8, 77-83Cantos, P., Maudos, J., 2001, "Regulation and Efficiency: The Case of European Railways", Transportation Research Part A 35, 459-472Driessen, G., Lijesen, M., Mulder, M., 2006, ‘The Impact of Competition on productive Efficiency in European Railways’, CPB Discussion paper No. 71, Friebel, G., Ivaldi, M., Vibes, C., 2010,’Railway (De)Regulation: A European Efficiency Comparison’, Economica 77, 77-91Holvad, T., (2006), ‘Railway Reforms in a European Context’, European Railway Agency, Valenciennes, FranceJupe, Robert, Crompton, Gerald, "A Deficient Performance: The Regulation of the Train Operating Companies in Britain's Privatised Railway System, critical perspectives on accounting 17 (2006) 1035-1065, University of Kent, UKKirchner, C. (2004) Rail Liberalisation Index 2004. Comparison of the Market Opening in the Rail Markets of the Member States of the European Union, Switzerland and Norway. IBM Business Consulting ServicesKirchner, C. (2011) Rail Liberalisation Index 2011. Market opening: Rail Markets of the Member States of the European Union, Switzerland and Norway. IBM Business Consulting ServicesKurosaki, F., (2010), ‘An Analysis of Vertical Separation of Railways’, 1-24Laabsch, C., Sanner, H., (2012), ‘The Impact of Vertical Separation on the Success of Railways’, European Railway Policy, 120-128Marti-Henneberg, J., 2013, "European integration and national models for railway networks (1840-1910)", Journal of Transport Geography 26, 126-138Nash, C., 2008, "Passenger Railway Reform in the Last 20 Years - European Experience Reconsidered", Research in Transportation Economics 22, 61-70Rothengatter, W., 1991, "Deregulating the European Railway Industry: Theoretical Background and Practical Consequences", Transpn. Res. -A, Vol.25A, No.4, 181-191Thompson, L., (2003), ‘Changing Railway Structure and Ownership: Is Anything Working?’, Transport Reviews, Vol. 23, No.3, 311-355Van de Velde, D., Rontgen, E., (2009), Internationalization of Infrastructures: Proceedingsof the 12th Annual International Conference on the Economics of Infrastructures’, 211-229 Appendix Level of separationLevel of induced competition Governance of Regulatory authorityIntegration modelHybrid modelSeparation modelLowMediumhighIn the ministryIn the railway authoritySeparate regulatory authorityAustriaBelgiumBulgariaSwitzerlandCzech Rep.Germany DenmarkEstoniaGreeceSpainFinland FranceCroatiaHungaryIrelandItaly LithuaniaLuxemburgLatvia NetherlandsNorwayPolandPortugalRomaniaSwedenSloveniaSlovakiaUnited KingdomTable 1. Structures and reforms in EuropeSource: IBM – Rail Liberalisation Index 2011.VariablesSourcesPass-kmEurostat [table (rail_pa_total)]Tonnes-kmEurostat [table (rail_go_typeall)]Modal split freightEurostat [table (tran_hv_frmod)]Modal split passengerEurostat [table (tran_hv_psmod)]Vertical separation IBM (2011) ‘Rail Liberalisation Index’, World Bank (2011) ‘Railway Reform In South-East Europe and Turkey On The Right Track?’GDP per capitaEurostat [table (nama_aux_gph)]Population Eurostat[table (demo_gind)]Table 2. Data SourcesHo: All panels contain Unit RootsHa: At least one panel is stationaryNumber of panels = 28Number of periods = 9AR Parameter: Panel SpecificPanel means: IncludedTime trend: Not IncludedDrift term: Not includedAsymptotics: T -> InfinityADF regressions: 2 lags Statisticp-valueInverse chi-squared(56) PInverse normal Z Inverse logit t(129) L* Modified inv. chi-squared Pm 63.63505.45145.15910.72140.22561.00001.00000.2353Table 3. Stationarity test on populationTable 4. Descriptive StatisticsVariableObsMeanStd. Dev.MinMaxcountry_id25214.58.094128year25220072.58720032011mod_sh_pas2520.06340.0340.0070.179mod_sh_freight2510.21990.1640.0060.725pass_km20714783.222770.52019393918tonnes_km23815738.5320834.81079115652vertical_separation2520.5158730.50101gdpcapita2522.45E+0416667.850260080300population2521.82E+07223000004483008.25E+07d_population2520.00257330.008-0.028450.031359log_gdpcapita2529.8446450.77393977.86326711.29352log_pass_km2078.4977411.6178495.2626911.45018log_tonnes_km2388.9093991.4455684.36944811.65834_Iyear_20042520.11111110.314895101_Iyear_20052520.11111110.314895101_Iyear_20062520.11111110.314895101_Iyear_20072520.11111110.314895101_Iyear_20082520.11111110.314895101_Iyear_20092520.11111110.314895101_Iyear_20102520.11111110.314895101_Iyear_20112520.11111110.314895101 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download