Macroeconomic analysis of public transport competitiveness in megacities

MACROECONOMIC ANALYSIS OF PUBLIC TRANSPORT COMPETITIVENESS IN MEGACITIES MARCELO ALTIERI Dissertação submetida para satisfação parcial dos requisitos do grau de mestrado MESTRE EM PLANEAMENTO E PROJETO URBANO Orientador: Professora Doutora Cecília do Carmo Ferreira da Silva JUNHO DE 2016 MESTRADO EM PLANEAMENTO E PROJETO URBANO 2015/2016 DEPARTAMENTO DE ENGENHARIA CIVIL Tel. +351-22-508 1901 Fax +351-22-508 1446  [email protected] Editado por FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Rua Dr. Roberto Frias 4200-465 PORTO Portugal Tel. +351-22-508 1400 Fax +351-22-508 1440  [email protected]  http://www.fe.up.pt Reproduções parciais deste documento serão autorizadas na condição que seja mencionado o Autor e feita referência a Mestrado Integrado em Engenharia Civil - 2014/2015 - Departamento de Engenharia Civil, Faculdade de Engenharia da Universidade do Porto, Porto, Portugal, 2015. As opiniões e informações incluídas neste documento representam unicamente o ponto de vista do respetivo Autor, não podendo o Editor aceitar qualquer responsabilidade legal ou outra em relação a erros ou omissões que possam existir. Este documento foi produzido a partir de versão eletrónica fornecida pelo respetivo Autor. To my parents and Miki, for their unconditional affective support. Plato is dear to me, but dearer still is truth Aristotle Macroeconomic analysis of public transport competitiveness in megacities i ACKNOWLEDGMENTS Many people deserve acknowledgement for assistance during the course of this dissertation. First, I would like to thank my supervisor, Dr Cecília Silva, for accepting to supervise my thesis and whose support, advice, and comments were indispensable to achieve the objectives. In addition, I would like to thank my supervisor for the encouragement and support on this dissertation writing and draft revision. Thanks to Dr Fernanda Sousa for the advice on statistical issues. To my colleges Ana Amante, Ana Martins, and Miguel Lopes for the support to improve my research and references. I would like to thank Manuel Guimarães for the technical assistance on GIS-related issues and Bruno Spellanzon for advices and valuable comments. Thanks also to Professor Ricardo Nascimento for the support and advice on São Paulo data collection. I am greatly indebted to Tami Itabashi and Miki Itabashi for the assistant on translating documents in Japanese, as well as contacting Tokyo institutions and public authorities. I would like to thank Rui Viana and João Viana for the draft revision and comments. Finally, thanks to my family, for the support along this way. Macroeconomic analysis of public transport competitiveness in megacities ii Macroeconomic analysis of public transport competitiveness in megacities iii ABSTRACT According to official UN estimations, the intense urbanization process recorded in the last fifty years tends to intensify during the next half century. Moreover, the projections suggest that megacities - agglomerations with more than 10 million inhabitants - has the highest growth rate among the urban agglomeration categories. In addition, the prediction suggests that the number megacities should reach the mark of forty-one in 2030, most of them in developing countries. Megacities are, by essence, complex and diverse environments. Consequently, mobility patterns are also complex, significantly different from the other urban agglomerations categories behaviour. Nevertheless, it requires a transport infrastructure with high-capacity services, competitive, and diverse, offering to the commuters’ alternatives and the opportunity choice. The recent motorization rates increase and the remarkable urbanization patterns change are a current trend for some researchers. According to some authors, as a result of this double fusion, cities become more auto-dependent and distant from sustainable goals. In order to avoid such urban pathology, some planners advocate for a new urban development approach and restrictive measures to decrease private transportation. In this context, this thesis proposes to study how the mobility patterns change in accordance to contemporary urban trends and the introduction of the automobile and high-capacity public transport. The analysis is structured based on the comparison between megacities which implemented high-capacity public transport on different urbanization phases, as well as the intrinsic consequences. Thus, the study selected nine megacities with high capacity public transport service and distinct sociocultural features and backgrounds. In order to achieve the urbanization differences and the absence of the automobile, the case studies were divided into three groups: (a) cases developed exclusively by public transport; (b) cases developed either by private transport and high-capacity public transport; (c) cases developed by private transport and low-capacity public transport. Due to the complexity of megacities and the involving mobility patterns literature, this research proposes a macroeconomic and conservative analysis. The theoretical support approach urban planning debate, the mutual influence between land-use and mobility patterns, the public transport infrastructure impacts on the urban form, and dynamics that influence the elasticity of demand. The study set of variables selection was guided by the theoretical framework and conditioned by case studies availability of data. The methodology is composed by three progressive parts: bivariate correlation, linear regression, and elasticity of demand. The first part seeks to distinguish variables with and without significant correlation. This part discussed only data without significant correlation to the 0.01 level. The second part consists of the bivariate regressions analysis from correlations with greater relevance. When necessary, the values for the elasticity of demand, standard deviation and averages are presented. The main results suggest to find similar ground to some of the current mobility patterns debate and the correlation with the urban environment. This study results highlight the following findings: (a) land-use, socio-economic, the urbanization process with or without automobile, and earlier or late high-capacity public transport implementation suggest no influence to change public and private transport mode share in megacities. (b) the high-capacity public-transport mode share increases in accordance to the increasing of the network; nonetheless, HCPT coverage area expansion suggests no influence to both public and private transport modes. KEYWORDS: megacities, urbanization, competitiveness, public transport, private transport. Macroeconomic analysis of public transport competitiveness in megacities iv Macroeconomic analysis of public transport competitiveness in megacities v RESUMO Segundo as estimativas oficias da ONU, o intenso processo de urbanização registrado nos últimos cinquenta anos tende a se acentuar durante a próxima metade de século. Além disso, as projeções indicam que dentre as categorias de aglomerados urbanos, o grupo das megacidades – aglomerados com população superior a 10 milhões de habitantes – apresenta a maior taxa de crescimento. De acordo com as previsões, deve-se atingir a marca de quarenta e uma em 2030, das quais a maioria estarão em países em desenvolvimento. Megacidades são, por essência, ambientes complexos e diversos. Consequentemente, os padrões de mobilidade são igualmente intrincados, com comportamento sensivelmente diferente das demais categorias de aglomerados urbanos. Não obstante, exige-se uma infraestrutura de transportes com serviços de alta capacidade, competitividade e diversidade, oferecendo além de alternativas, a oportunidade para a livre escolha. O recente aumento dos índices de motorização e a evidente mudança dos padrões de urbanização tem recebido a atenção de parte dos pesquisadores. De acordo com alguns autores, como resultado dessa dupla fusão de eventos, as cidades tornam-se dependentes do uso automóvel e distantes dos objetivos sustentáveis. A fim de evitar essa patologia urbana, defendem novas abordagens para o desenvolvimento urbano e medidas restritivas ao automóvel. Nesse âmbito, esta dissertação propõe investigar de que maneira a evolução dos padrões de mobilidade se condicionou às mudanças contemporâneas. A análise é estruturada a partir da comparação entre megacidades que implementaram os serviços de transporte público de alta capacidade em diferentes processos de urbanização e quais as suas consequências. Assim, foram selecionadas nove megacidades com serviço de transporte público de alta capacidade e que são distintas em suas características socioculturais. Para cumprir as diferenças de urbanização e a ausência do automóvel, os casos de estudo foram divididos em três grupos: (a) casos desenvolvidos exclusivamente pelo transporte público; (b) casos desenvolvidos tanto pelo transporte privado como por público de alta capacidade; (c) casos desenvolvidos por transporte privado e público de baixa capacidade. Devido à complexidade das megacidades e da literatura envolvendo os padrões de mobilidade, esta pesquisa propõe uma análise macroeconómica e conservadora. Os fundamentos teóricos abordam o debate sobre o planejamento urbano, a mútua influência entre o ambiente urbano e os padrões de mobilidade, os impactos da infraestrutura de transportes públicos na estrutura urbana e as dinâmicas que influenciam a elasticidade da procura. O conjunto de variáveis utilizadas para a análise foi orientada pelo enquadramento teórico, sendo condicionada pela disponibilidade dos dados dos respetivos casos de estudo. A metodologia adotada é composta por três partes progressivas: correlação bivariável, regressão linear e elasticidade da procura. A primeira parte pretende distinguir as variáveis com e sem correlação significativa. Nessa etapa são discutidos os dados sem correlação significativa ao nível 0.01. A segunda parte é constituída pela análise de regressões bivariáveis das correlações com maior relevância. Quando necessário, são calculados os valores para a elasticidade da procura, desvio padrão e médias. Os principais resultados apresentam conformidade com parte do atual debate sobre os padrões de mobilidade e sua correlação com o espaço urbano. Pode-se destacar as seguintes evidências: (a) as questões da forma urbana, socioeconómicas, os diferentes processos de urbanização com e sem automóvel e da implementação dos transportes de alta capacidade parecem não condicionar a utilização do transporte público e privado nas megacidades; (b) a utilização do serviço público de transporte de alta capacidade aumenta consoante a dimensão de sua rede; contudo, não apresentam influência na performance do transporte público e do privado. PALAVRAS-CHAVE: megacidades, urbanização, competição, transporte público, transporte privado. Macroeconomic analysis of public transport competitiveness in megacities 1 1 INTRODUCTION 1.1. OVERVIEW Since the second half of the twentieth century, urban planners and policy makers combine efforts to control cities undesirable suburbanization and decentralization. Critics blame modern cities shapeless, economic failures and excessive real estate speculation, social conflicts and unattractive aesthetic to the lack of public regulation over land-use and increasing use of private automobile. Moreover, defend that ‘needless’ land consumption threats non-urban landscape, e.g. forests, farmland and open space, underuses infrastructure and demand longer commute. As a result, air pollution and traffic congestion increases, driving current urban outcomes far from sustainable practices and environmental friendly trends. In order to avoid such development, programs and planning practices had proposed several restrictions on urban expansion, as well as automobile usage. The artificial building limit reduce housing choice and raise costs, affecting mostly low and middle-income population. In the other hand, overtaxing automobile usage, e.g. limiting parking space, fuel price regulations, and tolled highways, impose limit choice for urban mobility, affecting mostly middle and high-income population. While most affluent population find more easily housing units that please their taste, poor people become, if possible, more public dependent for affordable housing. Similarly, strong automobile restriction fuel individual behaviour changing and constrain a bigger population share to use public transportation and nonmotorized alternatives, demanding investments to avoid overcrowding and poor services. Government bodies and some non-governmental urban planning institutions, e.g. New Urbanism, and Smart Growth, defend a more restrictive urban development agenda, find more in the Smartcode booklet (Pinnell, 2009). In practice, New Urbanism and Smart planners propose an alternative point-of-view regarding the current living standard and mobility pattern, suggesting some behavioural changes to communities’. In other words, the authors define some planning strategies to approach undesirable urban targets, reducing population ability to choice and impose constrains to both living standards and mobility. Nonetheless, urban planning had struggle to minimize suburbanization and decentralization effects since the early years of urban development. Differently from the popularly spread fallacy, urban expansion is not a post-automobile USA cities pathology that others cultures should avoid. In fact, suburbanization causes and related restrictive laws and public policies are much older. The first Queen Elizabeth prohibited buildings on London surroundings in the sixteenth century. Centuries later, the Greater London Plan of 1944 imposed severe planning changes and controls over urban growth on outer areas, achieving no success (Sowell, 2011). Macroeconomic analysis of public transport competitiveness in megacities 2 Additionally, sustainability planners denounce automobile as a catalyst for suburbanization, just as Duke of Wellington did to British’s passenger railroad pioneers for promoting “the common people to move about needlessly” (Jackson, 1987). Nevertheless, many critics blame current Los Angeles autodependence but ignore the fact that, during the nineteenth century, the city expanded outwards supported mainly by the Pacific Electric Railways cable cars service. By that time, the real estate companies used to give prime values to proprieties located along the public transport corridors (Wachs, 1984). In accordance, New York, Paris, Tokyo and many other prosperous cities by that time also implemented railways, subways and buses, expanding urban borders throughout the countryside powered by the public transport engine. Recent studies suggest that both suburbanization and decentralization (Bruegmann, 2006; Clark and Kuijpers-Linde, 1994; Gordon and Cox, 2012; Guerra, 2014a; Susilo and Kitamura, 2008), as well as increasing levels of private motorization (Buehler, 2011; Dargay and Gately, 1999; Giuliano and Dargay, 2006; Newman and Kenworthy, 2011) are common patterns in many distinct countries. Societies seem to behave similarly under increasing wealth and other socio-economic issues. Differences may concern to the intensity that changes happen – strongly relation to socio-economic development – and specificcountries cultural features. The post-World War period has just speed up economic growth and whether urban and social changes. Concerning travel pattern, Newman and Kenworthy (1999) defended that overall post-war period establish a turning point defining the rise of the ‘auto city’ and the death of the ‘transit city’. Hence, instead of restricting people’s freedom to choose – more public regulation and controlling measurements, this thesis focus on understanding how public transportation services can become more attractive and compete with others players on urban transportation market, and some macro dynamics on megacities transportation. This dissertation grounds on existing travel pattern, public transport elasticity and impact on urban form, towards understanding how urban environment and others factors influence public transportation performance. In addition, as complex and high capacity public transport 1 (HCPT) are usually related to huge urban agglomerations (Guerra, 2014b), only megacities with at least 8 million inhabitants will be considered. Finally, due to the minor number of studies approaching megacities (Priester et al., 2013), this study proposes in first place an macro-economic analysis. By the end of the document, new questions should emerge and base further researches. 1.2. MEGACITIES The research scope is limited to only megacities due to economic, demographic, technical and future urban projections reasons. First, in many countries it is common the concentration of both population and economic production in one or few cities. For example, Seoul, Republic of Korea capital’s and probably the most notable case, generates near one fourth of country’s overall Gross Domestic Product (GDP) (OECD, 2012), as well as held almost half of national bank deposits (Yim, 2003). In accordance, only the city of Seoul accounts for one fifth of national population and the Metropolitan Area for near half. Cities tend to concentrate high Gross Value Added (GVA) economic production and attract more population and more high skilled labour force. As a result, land value is generally higher than on middle and small size agglomerations. In order to reduce housing and production costs and scarcity, megacities experience higher densities and more activity concentration than smaller urban agglomerations. On a technical perspective, great population and densities not only provides enough patronage for HCPT (Guerra, 2014b), e.g. expressways, subway, commuter rails and BRT, as it fuels high levels of traffic congestion. Further, congestion produces harmful collateral effects, e.g. increasing air pollution, 1 This study considers high-capacity public transport all urban services running over dedicated tracks and with a minimum of five minute of services gap during peak time, as well as commuter rail services. Macroeconomic analysis of public transport competitiveness in megacities 3 delaying medical emergencies and accessibility to jobs and business, affecting overall city productivity. To avoid congestion deteriorating, investments should keep pace with urban growth and mobility pattern. In fact, the only way to solve congestion is through infrastructure upgrading and expansion. Moreover, megacities environment demand both roadway, e.g. automobiles and buses, and railway transportation system. Urban transportation alternatives would avoid a complete collapse of the traffic system, as well as increases service quality through market competition. Finally, the global urbanization and economic ‘boom’ registered during the second half of twentieth century may speed up through the next decades. Latest United Nations (UN) World Urbanization Prospect (WUP) (2014) forecasts that urban population percentage should achieve three quarters by 2050. A more detailed evaluation gives a sharper image for the predicted demographic change. The world urban population should climb from actual 3.4 thousand of million (2008) to 6.75 thousand of million inhabitants, roughly doubling. If all expectations materialize, this substantial increase would be 1.4 times greater than registered levels from 1961 until 2008. In addition, UN WUP (2014) reported two others remarkable predictions. First, indicates a pattern for the evolution of urban population distribution. Back in 1975, there were only three worldwide urban agglomerations with more than 10 million inhabitants, accounting for 3.5% of global population. Furthermore, all three cases consisted of only developed and high-income 2 cities. This number raised to twenty-two in 2015 and now represents 9.4% (+5.9%) of total population, seventeen of them located on developing countries 3 . For the same interval, cities with less than 500 thousands decrease from 56.6% to 50.5% (-6.1%), 500 thousands to 1 million from 11.2% to 9.1% (-2.1%) and 5 to 10 million from 7.7% to 7.1% (-0.6%) (UN, 2007). By 2030, estimation suggests a number of 41 urban agglomerations with more than 10 million inhabitants, the majority from global south and developing countries. Fig. 1 indicates actual urban population distribution. Fig. 1 - Percentage of urban population residing in urban agglomerations by size of urban settlement, 1975, 2005 and 2015. Source: UN WUP 2005. 2 High-income, middle-income and low-income countries classification in accordance to World Bank criteria based on GNI per capita. 3 UN WUP 2007 (UN, 2007) evidences that former megacities grew much less than the average annual world’s rate (2.4 per cent) during 1975-2005. Following, early 20th century megacities grew a little less than average world’s rate while developing Asia experienced growth higher than 2.4 per cent. Finally, both developing and developed metropolis should experience a very low population growth, resulting in stagnation and urban stability. Macroeconomic analysis of public transport competitiveness in megacities 4 Fig. 2 - Urban and rural population as proportion of total population, by major areas, 1950–2050. Source: UN WUP 2007. Urban projections mostly testify the running process of rapid urbanization in Asia and Africa, as well as low and middle-income countries. Fig. 2 illustrate past and predicted urbanization from 1950 to 2030 on six major world areas. While Europe, Oceania, North America and Latin America and the Caribbean had already achieved almost four fifths of urban population, restraining urban growth to a minimal or stagnate level, Asia and Africa has enough ceiling to growth and urbanize. Data from the World Bank also confirms UN WUP projections. While world average annual rate of change in urban population has decreased from 3.11% (1950) to 2.05% (2015), the ratio regarding less developed or developing and more developed countries annual rate increased from 1.81 (1950) to 4.26 (2015) (World Bank, 2015). Consequently, global urbanization should witness the emerging of bigger agglomerations on regions with limited resources and intense socio-economic changing. Complementary, recent studies evidence that urbanization patterns should not differ largely from past decades. Some authors argue that jobs-housing dispersion, polycentrism and decentralization and increasing of private automobile ownership are common urban development patterns (Bruegmann, 2006; Clark and Kuijpers-Linde, 1994; Dargay and Gately, 1999; Gordon and Cox, 2012). Hence, most cities tend to experience those patterns regardless of developed or developing status, as well as agglomeration size, cultural and historic differences. In fact, individual and cultural characteristics, historical framework and both private and public actions determine development intensity and advance. Macroeconomic analysis of public transport competitiveness in megacities 5 1.2.1. PRE AND POST-AUTOMOBILE Transportation technology advance played an important role on reducing land transport costs and hence expanding cities boundaries (Bruegmann, 2006; Jackson, 1987; Sowell, 2011; Wolmar, 2009). The opportunity to live outside central area were a relevant factor for both urban and economic development. Most notable in nineteenth century cases, the growing working-class and poor population remained victims of central areas high densities problems (Bruegmann, 2006). Without modern infrastructure support, poor population had to get along with severe urban disabilities, e.g. human waste disgorged on public space; lack of sunlight and decent ventilation on small dwelling; water and air pollution 4 ; proliferation of several diseases and epidemics; raging fires 5 and natural disasters 6 . Fig. 3 – Central cities population evolution and HCPT system opening. Note: The dots symbol represents the decade which HCPT service was introduced. Source: Population data. Tokyo and Osaka: Tokyo Statistical Yearbook. London: Office for National Statistics. New York: United States of America Census Report. São Paulo and Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística. Mexico City: Instituto Nacional de Estadística y Geografía. Seoul: Seoul Statistics. Jakarta: Badan Pusat Stastistik. Delhi: Census of India. Bangkok: Thai National Population Census. HCPT opening data. Tokyo: East Japan Railway. Osaka: Osaka Municipal Transportation Bureau. London: Transport for London. New York: Metropolitan Transportation Authority. São Paulo: Companhia do Metropolitano de São Paulo. Rio de Janeiro: MetrôRio. Mexico City: Sistema de Transporte Colectivo. Seoul: Seoul Metro. Jakarta: Project for the Study on 4 Great Stink of London 1858 and Paris 1880. 5 The Great fire of London 1666. 6 Lisbon Earthquake 1755. Macroeconomic analysis of public transport competitiveness in megacities 6 Jabodetabek Public Transportation Policy Implimatation Strategy in the Republic of Indonesia, JICA, 2012. Delhi: Delhi Metro Rail Company. Bangkok: Bangkok Mass Transit System. Initially, private horse-drawn carriages and better road infrastructures allowed particularly high-income population to commute longer and expand settlements on outwards areas, breaking the old urban demand for high densities and travel on foot (Bruegmann, 2006; Sowell, 2011; Wolmar, 2009). By that time, private transportation was expensive. Owning a horse require looking after, feeding and grooming, something that a small population share could afford (Wolmar, 2009). Moreover, similar to current automobiles, horse-based transportation had also some pollution problems. As an example: “The best estimate is that by 1830s, English towns had to cope with something like three million tons of droppings every year”. 7 In addition, the streets of nineteenth century New York shared similar behaviour: “Much of the muck followed from the still-unavoidable reliance on horses – forty thousand of them, who each working day generated some four hundred tons of manure, twenty thousand gallons of urine, and almost two hundred carcasses...”. 8 Differently from today’s critics, some nineteenth century cities experience urban growth and land expansion through almost exclusive public transportation. In spite of some informal public transportation running on the streets of London, there were no competing individual transportation mode. Horse-drawn buses and later urban railways and motorized-buses offered faster and cheaper urban transportation, allowing the working class to live farther from the city centre. For the first time, poor people no longer needed to live and work on the same district. Not surprisingly, as shown by Fig. 3, London more pronounced increasing rates of growth matches the opening of the Metropolitan Railway on 1863, as well as New York with the subway services in 1904. In accordance, Paris in 1900, Tokyo 9 1885 (subway in 1927) and Buenos Aires in 1913 and other early twentieth century prosperous cities opened public transportation services to support suburbanization and longer commuting. In contrast, post-automobile megacities started running HCPT when agglomerations were much greater and presented a more intense growth rate. As evidenced by Fig. 3, among post-automobile megacities there is a slight delay pattern on population development and a more pronounced for HCPT starting. While some cities opened rapid-rail systems approximately in the middle of the development curve, others had start operating just near the population peak. Consequently, the greater the delay in implementing high-capacity services, the longer inefficient low capacity public transport 10 (LCPT) services structured working-class commuting and hence suburbanization. In severely cases, public authority inertia to provide satisfying transportation services encourage people to cope with mobility problems by their own. In other words, recently available individual transportation and informal services had a more prominent role on commuting people outwards than registered before. Further, pre-automobile mildest urban development and the limited transport technology available by that time, at least compared to current technology, helped heavy commuter services to keep pace with the urban growth (Plotch, 2015; Wachs, 1984; Wolmar, 2009). Hence, as stated above, the mutual development between public transportation services and suburbanization becomes clearer. On the other hand, while some modern megacities had to pursuit an extremely rapid urban growth (Rolnik and 7 Wolmar (2009). 8 Sowell (2011), pg. 20. 9 Date referring to Yamanote elevated line opening. However, Tokyo had a particular urbanization development and high densities on central core are more related to building laws and other restraining policies (Sorensen, 1999). 10 In contrast to previous HCPT concept, this study considers low-capacity public transport all urban services running without any dedicated tracks and with more than five minutes of frequency in peak-time. Macroeconomic analysis of public transport competitiveness in megacities 7 Klintowitz, 2011; Villaça, 2012), e.g. Seoul, São Paulo, Mexico City, other cases started HCPT much later and no longer have the opportunity to pursued. For the latter, in order to recover the initial lethargy and track future growth rate, considerable amounts of public funds were addressed to high-capacity systems (Barter, 1999; BMPC, 2014; BTS, 2016; DMRC, 2015). However, this alternative is too far from the majority of developing economies reality. In general, commuter services tend to expanded in a slower pace, in accordance to the public budget and congestion demand. Finally, the overall increasing of congestion levels had affected other issues, such as air pollution. In order to reduce transportation-related GHG emission, several cities adopted car circulation restriction policies (Koh, 2004; Mahendra, 2008). Moreover, some studies suggests that in short-term, automobile restriction can reduce emissions, but in medium or long-term is inefficient and GHG levels back to the levels presented before (Davis, 2008; Lin, 2011). Equally, collateral effects such as congestion and automobile mode share reduction seem to behave in similar ways. Hence, public policies for car restriction needs complementary actions and the development of transportation alternatives. 1.2.2. PRIVATE PUBLIC TRANSPORTATION ENTREPRENEURISM Since the nineteenth century, many private entrepreneurs started exploring a wide number of urban services, e.g. water distribution, public transportation, energy generation, etc. Initially, transportation services were largely privately operated, running through public concessions or partnerships and sometimes over government-regulated fares (Saes, 2009; Wolmar, 2009). Public transportation companies offered a wide range of services, since high capacity systems, e.g. subway and railway, to regular buses and minor feeder systems. Public transportation private entrepreneurism was quite common in most cities around the world, even in contrasting situations. London, New York and Tokyo shared the same singularities that wealthy growing cities like Los Angeles or inexpressive like São Paulo. As the two latter refers to medium and small scale cities – when public transportation system started running, operational services were restricted to low capacity system, e.g. streetcars, cable cars and regular buses (Tomasevicius Filho, 2002; Wachs, 1984). However, railway network was already available on almost every case, offering a fast means for mobility between cities and goods exchange. In most cases, this model had such success and was very profitable, allowing private companies to operate a wide number of urban public transportation services. However, during the first half of the twentieth century many private public transportation companies faced severe profit reduction, leading the majority to bankrupt and consequent nationalization. Some critics blame that intense jobs-housing decentralization and increase on private automobile use constitute the main factors for public transportation market share and total ridership reduction (Cervero, 1998). In general, critics defend that lowering densities weakens overall public transportation performance, as well as massive investments on roadways encourage automobile use. Hence, reducing ridership impact directly on profit margins and running feasibility. On the other hand, others authors credit operational deficit to both private and public authorities wrong decisions (Cohen, 1988; Gomez-Ibanez, 1996; Wachs, 1984). First, fares faced inappropriate adjustment (Jones, 2008; Wolman and Reigeluth, 1980) and populist political actions for stagnation of fares over decades (Cohen, 1988; Plotch, 2015; Silva, 2015; Wachs, 1984). Thus, operational profit was consumed by increasing running costs and the inability to raise fares appropriately. In order to avoid bankrupt, public transportation companies cut infrastructure maintenance and fleet update spending. Consequently, level of service worsened drastically turning public transportation service an unreliable urban transporttation alternative. Macroeconomic analysis of public transport competitiveness in megacities 8 Additionally, the lack of public regulation on public transportation allowed the raise of informal competitors. Advanced by unregulated operation, informal public transportation offers poor services for a little profit. In general, old and unreliable road vehicles with irregular schedule and routes operate informal services. Despite usually linked with poor countries and developing economies (Cervero and Golub, 2007; Guerra, 2014b), informal transportation contributed significantly to decrease subway ridership and drop ticket revenues in London during the Underground early times (Wolmar, 2009). While struggling to balance the public finance, cities still experienced the continuous suburbanization and emerging of automobile (Snell, 1974; Wachs, 1984). Consequently, the demand for better public transport and road infrastructure also increased, with remarkable significance to poor neighbourhoods and working class residential areas in megacities (Bruegmann, 2006; Cohen, 1988; Rolnik and Klintowitz, 2011; Rolnik, 1997; Tomasevicius Filho, 2002). In addition, the great population and jobs opportunities in megacities require high-capacity services, commuting daily hundreds of thousands of workers (Wolmar, 2009). Nonetheless, new infrastructure, lines construction, and upgrading facilities required huge amount of investments. By that time, splitting a limited budget to both operational maintenance and expanding network were unfeasible for private companies, leading to contradictions and wrong decision-making (Cohen, 1988; Plotch, 2015; Wolmar, 2009). London’s underground railway concept emerged due to poor infrastructure on the surface level (Wolmar, 2009). Below streets, railways could run with no physical hindrance and impact on urban landscape. Even more after railways and trains electrification, which allowed a drastic reduction on ventilation holes and deeper lines (Wolmar, 2009). However, underground constructing requires more resources than surface and elevated. In order to recoup money and drive profit, London Underground pioneers used to build railway system in the cheapest possible way (Wolmar, 2009). Nevertheless, public transportation companies failed on recouping investment and were nationalized years after. In opposition, New York HCPT companies only operate elevated or surface services, refusing to invest on underground lines (Plotch, 2015). Tokyo commuter services also operated only elevated railways, building the first underground line only in 1927. However, after public transportation nationalization, some elevated railways in Manhattan–New York were shut down due to public pressure regarding better environmental conditions, urban landscape concerns, and real state deteriorating (Plotch, 2015). To replace public transportation services, new underground lines were gradually constructed based on similar routes. Gradually, bankrupt public transportation companies were nationalized. Under public domain, public transportation services could operate without essential operational profit concerns. In US, fare prices and infrastructure improvements were largely explored by political promotion (Plotch, 2015). Also, public bureaucracy and opposition prevented public transportation authorities to increase fares appropriately and stop the running debt (Gomez-Ibanez, 1996). Since nationalization, European and American public transportation operators remained unprofitable. Wolman and Reigeluth (1980) shows that during the 70s US largest metropolitan areas had only 45 percent (50 percent all cities) of operational recovery cost. By the same time, Europe had a much more contrasting scenario. While London led recovery cost by 76 percent, Vienna and Stockholm had only 35 and 31 percent respectively. In the 90s, US recovery ratio decreased to only 35 percent, while London achieved 93 percent (Kenworthy and Laube, 1999). However, European operators’ recovery costs average were about only 54 percent. Also, Kenworthy and Laube (1999) shows that in average both developed and developing Asian public transportation operators are profitable, with recover operational costs by 119 and 113 percent respectively. However, Tokyo is the only sample from Asian cities almost exclusively operated by private companies. Macroeconomic analysis of public transport competitiveness in megacities 9 Similarly, São Paulo and Los Angeles public transportation companies experienced similar administration failures and consequent bankrupt. Generally, inappropriate fares, severe economic inflation and public concessions adjustment restriction, wartimes and economic depression compromised private financial health (Jones, 2008; Silva, 2015; Tomasevicius Filho, 2002). Thereby, unprofitable public transportation companies required subsidies for both operating services and infrastructure network expansion (Cohen, 1988). Spending public funds on both expansion and operation were more severe on inefficient operational services, such as New York and Boston (Cohen, 1988; Gomez-Ibanez, 1996). Cohen (1988) shows that New York Subway suffered much more from unbalanced investment than insufficient public capitalization levels. During the New York City Transit Authority period – 1953-1967), investments were addressed mainly to fleet renewal and infrastructure update and only 17.5 percent for new routes. The prior for service quality, instead of network expansion, resulted on stable ridership levels (Plotch, 2015). Later, the Metropolitan Transit Authority administration adopted a more intense expansionist investment policy. From 1968 until 1980, new routes capital expenditure raised to 41.5 percent, while overall available capital only 9.87 percent, even when facing strong economic crisis. Thus, main service outcomes levels faced severe reduction and in response ridership once again experienced new decrease (Cohen, 1988). Lastly, despite the implicit literal sense, all post-automobile megacities public transport companies are of a public helm. If not entirely public owned, private companies run under concessions and strong regulation. Hence, no profit is expected and generally, services run subsidised (Summit, 2014). 1.3. RESEARCH QUESTIONS A macroeconomic analysis approaching such complex cases, with distinct development patterns, and huge historical and current disparities, is to structure concisely the research objective and questions. Towards achieving this purpose, this research tries to understand the dynamics among pre and postautomobile HCPT through a single and common outcome. The opposed condition, i.e. starting from the intrinsic differences and special cases, should lead to a more complex analysis, demanding since the beginning micro-data and detailed information. Thus, others questions should arise from the main question, leading to a progressive enlargement of the studied subject. Main question: Does the HCPT implementation concerning different urbanization stages affected the megacities competitiveness for mode split share? Hypothesis 1: Megacities competitiveness for mode split share react more positively to public transport in accordance to the respective urbanization phase that HCPT started running. However, considering the very distinctive nature regarding pre and post-automobile megacities, i.e. megacities that expanded exclusively by public transportation and others with the automobile presence, this hypothetical assumption lead to two contrasting answers. 1.1: Pre-automobile urban environment led to more HCPT share. Despite of owning greater and older HCPT network, pre-automobile megacities should have more public transport share mainly due to land-use diversity, more compact development, and cultural-specific reasons (Dieleman and Wegener, 2004; Ewing and Cervero, 2010; Ewing, 1997; Jacobs, 1961; Nivola, 1999; Pinnell, 2009). As stated above, the ability to build a HCPT network in accordance to a mildest urban growth resulted on a denser and more diverse urban environment. Therefore, a notable population share should live close to HCPT service. In addition, the feeder system can efficiently operate on farther neighbourhood, making the private transportation use sparse and needless. Furthermore, the dense ‘transit cities’ urban environment led to more congestion, harming the automobile use (Schimek, 1996). Macroeconomic analysis of public transport competitiveness in megacities 16 The idea that just through strong planning urban areas will become less auto-dependent is opposed by some researches and economists (Bruegmann, 2006; Giuliano and Narayan, 2003; Gordon and Richardson, 1997, 1989; Sowell, 2011). For example, Sorensen (1999) observed that in Tokyo region during 1960 and 1980, new urban settlement develop much faster outside planned areas. As a result, urbanization remains scattered all over the territory. Additionally, as urban infrastructures and facilities were built only on planned areas, such amenities become underused and non-functional. Finally, author suggest that the government is unable to avoid ‘unplanned’ development and that top-down regulation shows some weakness. Hence, planners and urban economists that support a more bottom-up development argue that travel patterns and travel decision are much more complex than a mere changing of density or any other urban features. For instance, in a response to Newman and Kenworthy research over gasoline consumption, Gordon and Richardson (1989) succinctly argue that several other factors influence gasoline prices and demand. As well funded in economic basis, the demand for fuel may vary more significantly concerning different social-economic environments and product supply than concerning densities (Gordon and Richardson, 1989; Snarr, 2014). Additionally, Brindle (1994) criticize the statistical value from Newman and Kenworthy research, Brugemann (2006) argue that suburbanization tend to become an universal phenomena, and Sowell (2011) suggest that excessive planning are unable to reverse towards the ‘re-urbanization’ era. 2.1.1. DENSITY Several empirical studies try to understand how transportation outcomes change regarding urban densities. In general, researchers compare the typical low-density American style cities to high-density European urban form (Bruegmann, 2006; Clark and Kuijpers-Linde, 1994; Dieleman and Wegener, 2004; Ewing, 1997; Ewing et al., 2002; Nivola, 1999) or denser Asian cases (Bruegmann, 2006; Gordon and Cox, 2012; Kenworthy and Laube, 1999; Newman and Kenworthy, 1989; Nivola, 1999). Some findings predominantly suggest that low-density urbanization are more auto-dependent and outcome a higher level of VMT and fuel consumption. In contrast, higher densities produce a mirrored result; hence, denser cities are less auto-dependent and has shorter trips. On the other hand, urban density shows opposite results concerning commuting time and travel speed (Giuliano and Dargay, 2006; Gordon and Cox, 2012; Kenworthy and Laube, 1999). In spite of producing longer commuting, people travel faster and save more time in low-density and labelled auto-dependent cities (Gordon and Cox, 2012; Lee et al., 2009). Moreover, high-density urban environment produces more congestion and consequently reduce travel speed, affecting air pollution by increases GHG emission (Gaigné et al., 2012), and weaken economic productivity and health assistance (Sowell, 2011). However, many studies evidence a weak and limited correlation regarding only density and a more autodependent transportation outcome (Brindle, 1994; Ewing and Cervero, 2010; Ewing, 1997; Ewing et al., 2002; Giuliano and Small, 1993; Gordon and Richardson, 1989; Lee et al., 2009). While some authors argue that density should be complemented by diversity and others variables (Ewing and Cervero, 2010; Ewing, 1997), enhancing demographics analysis and overall urban homogeneity, others defend that accessibility to amenities, personal or cultural preferences, and economic dynamics are relevant issues as well (Brindle, 1994; Giuliano and Narayan, 2003; Giuliano and Small, 1993; Lee et al., 2009). In spite of any low and high density clash, or even automobile or public transportation dependence labelling, some authors agree that density in a more detailed local level affect both automobile congestion and public transportation frequency and competitiveness (Levinson and Kumar, 1997), more specifically HCPT. The congestion occurs in accordance to the physical inability to expand road network capacity or discrepancies among urban development and investments (Balaker and Staley, 2006). Macroeconomic analysis of public transport competitiveness in megacities 17 Without available land to bear increasing traffic flow and expensive limited parking, automobile use find natural hindrances in dense urban environment (Schimek, 1996). Several authors tried to determine densities thresholds for automobile dependence and public transportation-supportive. Concerning minimum density to avoid auto-dependency, Newman and Kenworthy (2006) suggest that 3.500 inhabitants per km2 is a limit for absolute automobile dominance. In accordance, Newman and Kenworthy (1989) determine similar values concerning fuel consumption. However, Brindle (1994) argue that this findings are poorly reliable. As defended by the author, the argument that there is a threshold level where planners can manipulate to declassify a city as autodependent lead other ‘experts’ and academics to wrong conclusions. For instance, one may assume that minor automobile usage in India occurred due to high densities. However, Indian megacities, as well as many other developing agglomerations, combine high urban density with high dwellings density and significantly low-income, i.e. precarious slums neighbourhood. There is a lack of researches regarding urban public transportation minimum ridership densities. Hayashi et al. (1992) determined that above ten thousand inhabitants per km2, rail services ridership variance are insignificant. On a comparative study regarding cost efficiency and public transportation-oriented density, Guerra and Cervero (2011) argue that public transportation supportive density threshold may vary in accordance to the overall investment and public zoning and parking restrictions. However, authors suggest that in the US context, a value of near 20 jobs and population per acre are a reasonable point to evaluate future public transportation proposals. However, Brindle (1994) suggest that defining some density threshold is complicated and unlikely to workout. The author defend the argument based on three contradictory samples: (i) Copenhagen suburbs has densities inferior to auto-dependency limit and perform high public transportation share. (ii) More than 3 million residents of the Los Angeles County live at near 3.500 inhabitants per km2. (iii) While many Australian suburbs have densities above de threshold level, the Northern Suburbs Railway in Perth thrives in a region with half of the claimed density level. In other words, based on Newman and Kenworthy findings comparing only densities, Delhi and Mumbai should present lower autodependency levels than London and Tokyo. Moreover, megacities density is in average higher than any threshold existing on the literature. Among the ten greater urban agglomeration, Tokyo has the ‘lower’ mean density with 8.668 inhabitants per km2 while Kolkata has the highest with 24.429 inhabitants per km2. Table 2 compare the total population share living on several distinctive cities based on two different densities thresholds: public transportation minimum level of 3.500 inhabitants per km2 (Newman and Kenworthy, 2006, 1989), and the nonvariance level of ten thousands inhabitants per km2 (Hayashi et al., 1992). As noted, concerning only density, megacities levels disqualified any auto-dependent labelling, as well as offering enough levels for HCPT feasibility and automobile restriction through congestion and scarce parking area. Macroeconomic analysis of public transport competitiveness in megacities 18 Table 2 – Percentage of population living according to Newman and Kenworthy (1989) and Hayashi et al. (1992) density threshold. Calculations by the author. The cities were selected based on district, borough or ward population data availability for more detailed calculation, and the diversity of urban agglomerations sizes. Note: For UK: 2011 UK census. For USA: American 2010 Census. For South Korea: Korean Statistical Information services, data for 2010. For Portugal: Instituto Nacional de Estatística, data for 2011. For Japan: Statistics Bureau, data for 2010. For Brazil: Instituto Brasileiro de Geografia e Estatística, data for 2010. For Mexico City: Instituto Nacional de Estadística y Geografía, data for 2010. For Madrid: Instituto Nacional de Estadística, data for 2009. For Berlim: Berlin-Brandenburg Statistical Office, data for 2010. For Shanghai: Shanghai 2010 Census. 2.1.2. DIVERSITY As advocated by Jacobs (1961), cities demand high densities in order to produce high diversity and become more creative. Diversity is the main complementary variable for density. Similarly, existing literature base findings on empirical studies comparing distinct urban scenarios (Dieleman and Wegener, 2004; Ewing and Cervero, 2010; Ewing, 1997; Ewing et al., 2002; Gordon and Cox, 2012; Loo and Chow, 2011; Nivola, 1999). Therefore, the correlation among diversity and transportation outcomes is coarsely similar to density: less diverse urban environment generates a more auto-dependent urban form > 3.500 population/km² (autodependent threshold) > 10.000 population/km² (public transportation no variance threshold) Fewer than 500.000 Coimbra, Portugal 0% 0% Liverpool, UK 61.3% 0% Oxford, UK 51.6% 0% Cardiff, UK 52.2% 0% 500.000 to 1 million Lisbon, Portugal 94.5% 18.4% Kitakyushu, Japan 24.9% 0% Sorocaba, Brazil 0% 0% 1 million to 5 million Berlim, Germany 56.5% 8.5% Madrid, Spain 86.3% 68.1% Ulsan, South Korea 81.0% 0% Incheon, South Korea 96.8% 40.0% Birmingham, UK 64.5% 0% Chicago, USA 85.4% 12.1% Nagoya, Japan 93.4% 0% Fukuoka, Japan 72.3% 12.2% 5 million to 10 million New York, USA London, UK 87.6% 21.9% Mexico City, Mexico 86.4% 65.2% Jakarta, Indonesia 100% 100% Seoul, South Korea 100% 91.7% 10 million or more Shanghai, China 100% 61.7% Tokyo, Japan 94.5% 76.2% São Paulo, Brazil 96.8% 63.7% Macroeconomic analysis of public transport competitiveness in megacities 19 with higher levels of VMT. Once again, more diversity produces a mirrored image with lesser autodependency and shorter trips. In the words of Ewing (1997), “Compact development requires some concentration of employment, some clustering of housing, and some mixing land uses (but neither high density nor monocentric development)”. In other words, a homogenous and more static urban environment. In accordance, some studies suggests that both jobs and housing balance contribute significantly to reduce suburbanization and mono functional land use effects (Ewing et al., 2002, 1996; Loo and Chow, 2011). Concerning greater urban agglomerations, Loo and Chow (2011) investigated how significant jobs and housing dispersion could reduce excessive commuting in Hong Kong. Based on a geographical approach, authors suggest that jobs dispersion has potential to save significantly commuting time. In order to achieve a compact goal, authors suggest that policy makers should concentrate efforts on avoiding further expansion and encouraging jobs dispersion over new growth areas (Loo and Chow, 2011). In other words, similar to density, they defend more public regulation over land use and private real estate activity, intervening to rectify unplanned market distortions and weaken automobile use (Anas and Pines, 2008; Brueckner and Kim, 2003; Brueckner, 2005). In contrast, others authors defend that jobs-housing balance are significant for non-work daily trips and minor relevant for working trips (Giuliano and Narayan, 2003; Giuliano and Small, 1993). Accordingly, Giuliano and Small (1993) analysed journeys to work on the Los Angeles County finding minor relation between commuting time and jobs and housing ratio. Another research from Gordon and Cox (2012) suggests that despite of the auto-dependency and usually related to undesirable unplanned development, commuting time are lower and average speeds higher in American cities than in contrasting compact European. The main reason for the poor correlation relies on the minor influence that jobs proximity plays on choosing housing area (Bruegmann, 2006; Giuliano and Small, 1993). In fact, housing choice are likewise influenced by a wide range of others amenities, e.g. open space, schools, hospitals, as well as personal preferences (Bae et al., 2003; Giuliano and Narayan, 2003; Giuliano and Small, 1993; Gordon and Cox, 2012). Homogenous urban distribution should work only if accomplish some requisites, like: (a) employers and employees remains territorially stable for decades. (b) Housing costs are low enough to support population mobility in accordance to new jobs opportunities. (c) Personal preferences are ignored and both amenities and services are equally distributed through the territory. (d) The awkward household causality of only one working person per household or in the case of two workers, both finding work in the same district. Hence, compact and balanced development depend on unachievable issues for public controlling and planning, as well as a perfect overall socio-economic state. The relation between spatial interaction and housing and jobs distribution, and it intrinsic travel pattern result have been discussed since the past half of century (Wegener and Fürst, 1999). Among many proposed models, many researches have referred the Brotchie triangle, shown on Fig. 4 (Wegener and Fürst, 1999). The model correlates dispersion and interaction levels, relating some travel behaviour to each structure. Back to the compact concept defended by Ewing (1997), the proposed jobs and population dispersion should behave in accordance to B or C points. Point C represents a scenario where the travel pattern has less travelling and shorter trips. In this scenario, households tend to keep the overall mobility throughout the vicinities and the interaction between distant urban areas or districts is minimum. Moreover, is expected the absence of any HCPT and a predominance of non-motorized modes. On the other hand, point B represents the reverse situation, where housing, jobs and mobility is scattered all over the territory. In this scenario, there is no hierarchy on travel flow and travel pattern is complex and hard to predict. Hence, due the high number of possibilities, public transportation faces severe hindrances on competing with private modes. Macroeconomic analysis of public transport competitiveness in megacities 20 Fig. 4 – Brotchie triangle. Adapted from Wegener and Fürst (1999). Macroeconomic analysis of public transport competitiveness in megacities 21 Fig. 5 – Tokyo jobs concentration from 2010. Source: e-statistic of japan statistical bureau. Fig. 6 – London jobs concentration from 2011. Source: Office of National Statistics. Macroeconomic analysis of public transport competitiveness in megacities 22 At last, point A represents a monocentric scenario with no dispersion, where jobs are mainly located in central area and flow are well defined through traffic corridors. On such scenarios, urban mobility demand high capacity traffic structures, a fact that significant advance HCPT on transportation competition. A more effective approach of this last situation is the proposal of urban development under TOD umbrella (Cervero and Dai, 2014; Cervero, 1990; Curtis et al., 2009). As stated by Curtis et al. (2009) TOD seeks to combine the high capacity of heavy or segregated public transportation with the flexibility of non-motorized modes, i.e. walking and cycling, to enhance the competition with automobile. Consequently, planners and policy makers defend that population should concentrate around HCPT corridors, taking advance of land use and planning restrictions to force such development. However, transportation infrastructure and enhancing accessibility affect land price and use, as will be debated on item 2.3.1 of this dissertation. However, megacities are generally monocentric structures that concentrate more than half of employment places on the Central Business District (CBD) or inner areas (Barter, 1999). This condition is more evident on developing economies, remarkably on Asian cases, than in high-income and more mature economies. Considering the prediction for a predominance of Asian developing megacities, in those cases travel pattern should maintain the monocentric model rather than develop a more dispersed model, as shown by Fig. 5. Largely, high land costs in central areas and surroundings pushes working population to residential suburb, creating a pronounced commuting pattern from outwards to central area (Ântico, 2005; Guerra, 2014a; Rolnik, 1997; Villaça, 2012). Hence, the concentration of jobs in a specific area could determine the homogeneity of ‘monocentrism’ through megacities and the effects it produces on HCPT share. Concerning a macroeconomic analysis, high-capacity public transportation network and coverage area may vary regarding the urban dispersion and spatial interaction. In accordance to Brotchie triangle, on the monocentric model – point A, the commuting pattern has a pronounced radial shape, moving commuters from predominantly residential or suburban outwards neighbourhoods to central areas with high employment concentration. Many megacities evidence this commuting pattern, such as Tokyo (Fig. 5), London (Fig. 6), and São Paulo. On the other hand, concerning only the point B as a probable alternative, jobs and housing balance require a more capillary system, amplified by the urbanized size of megacities. 2.1.3. DESIGN, DESTINATION ACCESSIBILITY AND DISTANCE TO TRANSIT Along with density and diversity, spatial design and accessibility performance provide a finer detail regarding the built environment and travel pattern (Crowley et al., 2009; Ewing and Cervero, 2010). In general, both design and accessibility have a bottom-to-top structure, starting from tiny data and further producing major indicators for a whole territory or a specific partial subarea. As an example, the results from the SNAMUTS 11 methodology range from single transportation network segment to an average value for the entire city. Concerning megacities, in spite of the remarkable Manhattan grid, and the spread of similar standard through other boroughs, most cases presents a great heterogeneity of roads design and urban footprints that were built over many decades, layer by layer. During decades of urbanization, cities developed since industrial neighbourhoods to city garden planned districts, witnessing the natural change, decline and resurgence that the urban dynamic promotes. Hence, it is hard to determine an overall pattern, and perform a finer analysis is out of this dissertation scope. While accessibility require a deeper investigation, computing high-capacity public transportation coverage area require just minor information, i.e. station geographic location, and can produce a reasonable gross distance to public transportation output. The literature suggest that public 11 Spatial Network Analysis for Multi-modal Urban Transport Systems (www.snamuts.com). Macroeconomic analysis of public transport competitiveness in megacities 23 transportation proximity to urban activities have a relevant role on attracting more ridership (Biba et al., 2010; Crowley et al., 2009; Yigitcanlar et al., 2007). The attractive power weakens as distance increases, finding an edge value between 0.8km to 1km for commuter rail or subway (Crowley et al., 2009; Yang et al., 2013; Yigitcanlar et al., 2007). Moreover, on a study about the walking accessibility on Toronto’s transit-oriented North York City Center, Crowley (2009) found that, besides attracting patronage, increasing proximity to subway station provoke decreasing levels of automobile ownership per household and vehicle use during morning peak time. In accordance, on a study concerning Mexico City subway expansion comparing 1994 and 2007 surveys data, Guerra (2014a) found that while automobile trips grown 21.3 percent in the whole metropolis, on areas located around 1km from subway service the growth was of only 7.5 percent. Additionally, subway share grown 26.3 percent in areas 1km from the service and only 3.3 percent in the metropolis. Moreover, on a research concerning the relation between household ownership and use of automobile to densities, Schimek (1996) found that despite of the general increasing per capita income and vehicles per household, households within three blocks from public transportation stops have 0.20 fewer vehicles per household than outside the catchment area. Similarly, central city households have 0.16 fewer vehicles than non-central city households do. Finally, on a study comparing forty-one world cities, De Grange et al. (2012) found that increasing HCPT network in ten percent generates a decrease of two percent in automobile use and increases three percent on public transportation mode share. 2.2. SOCIO-ECONOMICS AND THE TRAVEL PATTERN Socio-economic and demographics complement both travel pattern and urban built environment analysis. Through several social and economic variables, researches seeks to understand the change on the opportunities and means to travel of a specific society (Dargay and Gately, 1999; Farber et al., 2014; Giuliano and Dargay, 2006; Giuliano and Narayan, 2003; Giuliano, 2005; Gordon et al., 1988; Levinson and Kumar, 1995). The most common transportation outcomes related to the demographic approach are trips per capita, trips length and time, work and non-work trips share and automobile ownership. Differently from the built environment, some demographic changes can affect travel pattern faster and incisively. For instance, while changing urban densities, diversity or public transportation network size require at least a couple of years, household income can change abruptly in accordance to any macroeconomic wrong decision or a deep crisis, like in 2008. 2.2.1. INCOME Urbanization has a mutual relation with economic production and wealth generation (Bruegmann, 2006; Sowell, 2011). Since earlier days, population moved to the cities seeking for better living standards and higher incomes. Consequently, as per capita income rises, the poorest population leaves the inevitable limitation of choice, leading to more opportunities and capacity to make decision. In accordance, while low-income population has no choice unless public transportation (Giuliano, 2005; Guerra, 2014a, 2014b), high-income can choose the service that best fits their needs. Through broadening the decision-making capacity, a greater population share can choose a preferred housing standard and means to travel. In accordance, Dargay and Gately (1999) found a consistent pattern regarding automobile ownership and GDP per capita increasing on several developed and developing countries during 1960 and 1992. Additionally, by that time, authors projected levels for the year 2015, suggesting that in developing economies automobile ownership will growth twice as rapidly as per capita income. Concerning housing options, higher population growth rate in the suburbs and Macroeconomic analysis of public transport competitiveness in megacities 24 exurb areas evidenced a general preference for bigger dwelling in less diverse and low-density neighbourhood (Gordon and Cox, 2012). Consequently, the preferred living standard change the travel pattern. On a comparative study between USA and GB, Giuliano and Dargay (2006) suggest that differences on each country travel pattern results from per capita income and related automobile ownership disparities, as well as lower fuel prices and densities and some country-specific culture. The study found that Americans households produce more trips, travel longer, and have and use more automobile than British does. On the other hand, British lower per capita income results on fewer opportunities and means for travel. Additionally, as fuel price are much higher than US, household have to use alternatives to private automobile and daily retail and services use routine. In addition, Giuliano (2005) evidenced that in USA national level, low-income households travel less in frequency and length than middle and high-income. Concerning living standards and preferred housing, Clark and Kuijpers-Linde (1994) found that both California and the Dutch Raanstad experience lowering densities and suburban growth as well as higher automobile ownership, trips per capita and VMT. The latter find similarities with Gordon and Cox (2012) and Bruegmann (2006), suggesting that land use policies and public regulation play minor relevance on determining people preferences. In opposition, there are minor studies investigating the result of decreasing income or rising unemployment rates. On a recent study, Newman and Kenworthy (2011) states that developed countries reached a virtual automobile peak, and forecasts declining levels for the next decades. As argued by the authors, this shift results in accordance to changing living and working behaviour, less drive license among younger population, successful public reforms to combat automobile, and cultural desire towards more sustainable and friendly environment. Similarly, Newman et al. (2013) suggest that the automobile peak occurs in accordance to a new twenty-first century cultural agenda. On the other hand, in spite of any automobile saturation threshold, Headicar (2013) suggests that the England travelled mileage decrease may result from the significant international immigration registered since the end of the twentieth century. Moreover, some studies omit the fact that in 2008 most developed countries faced a severe economic depression, and most of them, specifically European countries, still struggle to balance the running loss. Nonetheless, some developed cities household travel survey clarified some relevant transportation outcomes in post 2008 global crisis. Accordingly, the London Area Transportation Survey from 2011 evidenced a significant reverse regarding travel pattern tendency registered since 1971. For the first time the number of household with no automobile increased and automobile mode split share decreased. The turning point happened between 2007/08 and 2008/09, where trips per capita dropped from 2.64 to 2.42 (-0.22). The automobile leads the dropping with -0.15 points, while all public transportation modes remained stable. Similarly, São Paulo’s traffic agency (CET) reported a decline in congestion levels over arterial and expressways during peak time. Municipal authorities credit results to several traffic reforms and a new speed reduction policy 12 , which initially sought to reduce traffic fatalities. In contrast, others municipalities in São Paulo’s Metropolitan Area also reported decreasing congestion levels without any 12 News available in Folha de São Paulo web version: ‘Após redução, marginal Tietê tem piora de manhã e melhora à tarde’ from 19/08/2015. http://www1.folha.uol.com.br/cotidiano/2015/08/1670542-apos-reducao-marginal-tiete-tem-piora-demanha-e-melhora-a-tarde.shtml. Macroeconomic analysis of public transport competitiveness in megacities 25 speed reduction. However, CET and other academics experts related congestion reduction to Brazilian recent economic crisis 13 , which resulted on increasing levels of unemployment 14 and fuel prices. 2.2.2. HOUSEHOLD Changes on household features affect mainly trip rate, trip length and mode share. In general, current trend points to smaller household and more female participation on labour force. As a result, more workers commute, increasing trip level and congestion on both public transportation as on roadways. Levinson and Kumar (1995) evidenced that, over twenty years (1968-88) in the Washington Montgomery County – Maryland, while population and employment increased, household size decreased and the transportation outcomes registered increasing rates of automobile ownership and annual VMT. In accordance, through national survey data from 1969 to 1983, Gordon et al. (1988) found some similarities regarding decreasing household size to increasing automobile ownership and trips per household. However, in spite of the Levinson and Kumar (1995) findings on increasing trips length for a specific locality, others national wide researches suggest that trip length decrease in accordance to household size and income level (Gordon et al., 1988; Strathman et al., 1994). One reasonable hypothesis regarding this behaviour is the increasing rate of chained trips – journey-to-work and non-work, and more workers per household. Moreover, Sowell (2011) argue that as income rise, people tend to travel more and use the faster means to travel to build a more complex trip chain. For example, some workers take advantage of commuting to by-pass to a commercial centre or leave kids on school, as well as pick-up after class or do any other extra activity. In contrast, Susilo and Kitamura (2008) found that public transportation commuters makes a higher number of chain trips than automobile users in Osaka. 2.3. PUBLIC TRANSPORT AND THE CITY 2.3.1. PUBLIC TRANSPORTATION IMPACTS Infrastructure implementation and expanding systems are well accepted as a major factor for both urban and economic development (Wang, 2002). Although, several researchers devote attention on the relation between urbanization development – privately promoted – and infrastructure investment – usually public promoted. There is no consensus on hierarchical influence. While some authors argue that growing private sector productivity demand public investment to satisfy new needs, others suggests the opposed, relation public investment as the engine of growth (Agénor, 2010; Esfahani and Ramırez, 2003). Differently from geographic issues and natural resources, these advances are essentially produced by human intervention. However, there are some intrinsic distinction between land-use and infrastructure. The former is law based and differentiate urbanized area in several spatial fragments, allowing public authority to regulate urban development and address potential to specific areas. This condition makes land-use regulation much more flexible and to change over short-term. On the other hand, the latter split into two different dimensions. Some infrastructure tends to be much more homogeneous and equally distributed through urban territory, e.g. water distribution, sewage, capillary road network, while others specialize neighbourhoods, e.g. ports, landmarks, high-capacity public transportation lines. 13 News available in Folha de São Paulo web version: ‘Trânsito em horário de pico teve redução de 16.6% na capital de SP’ from 13/04/2016. http://www1.folha.uol.com.br/cotidiano/2016/04/1760446-transito-em-horario-de-pico-teve-reducao-de166-na-capital-de-sp.shtml. 14 In accordance to IBGE labour survey from 2016, the labour occupancy rate dropped from 56.6% in 2013 to 52.3% in 2016 on metropolitan areas. During the same interval, the unemployment rate growth from 6.4% to 8.1%. Macroeconomic analysis of public transport competitiveness in megacities 32 Table 3 – World’s greatest urban agglomeration. Source: UN WUP2005. Notes bold – Selected mega-city as study case. Italic – No high-capacity public transportation service. * – Available commuter rail service. ° – No available data or the last reported household travel survey without high-capacity public transportation. ¹ – High-capacity public transportation service operated exclusively by BRT systems. † – Chinese cities have lack of information about household travel surveys and no available data in English. ‡ – Population data for only Seoul Special City. Seoul Capital Area incorporate the city of Incheon and others municipalities in the Gyeonggi province, with 24 million inhabitants in 2010. Rank Agglomeration and Country Population (millions) Agglomeration and Country Population (millions) 1New York, USA 12.338 Tokyo, Japan 35.494 2Tokyo, Japan 11.275 Bombay, India° 21.869 3London, UK 8.361 Mexico City, Mexico 21.568 4Shanghai, China 6.066 São Paulo, Brazil 20.535 5 Paris, France 5.424 New York, USA 19.876 6 Moscow, Russia 5.356 Delhi, India 18.604 7 Buenos Aires, Argentina 5.098 Shanghai, China† 17.225 8Chicago, USA* 4.999 Calcutta, India 16.980 9Calcutta, India 4.513 Dhaka, Bangladesh 16.842 10 Beijing, China 4.331 Jakarta, Indonesia¹ 16.822 11 Osaka, Japan 4.147 Lagos, Nigeria 16.141 12 Los Angeles, USA* 4.046 Karachi, Pakistan 15.155 13 Berlin, Germany 3.338 Buenos Aires, Argentina 13.396 14 Philadelphia, USA 3.128 Cairo, Egypt 13.138 15 Rio de Janeiro, Brazil 2.950 Los Angeles, USA 13.095 16 Saint Petersburg, Russia 2.903 Manila, Philippines° 12.917 17 Mexico City, Mexico 2.883 Beijing, China† 12.850 18 Bombay, India 2.857 Rio de Janeiro, Brazil 12.770 19 Detroit, USA 2.769 Osaka, Japan 11.309 20 Boston, USA* 2.551 Istanbul, Turkey 11.211 21 Cairo, Egypt 2.494 Moscow, Russia 11.022 22 Manchester, UK* 2.422 Guangzhou, China† 10.420 23 Tianjin, China 2.374 Paris, France 9.858 24 São Paulo, Brazil 2.334 Seoul, South Korea‡ 9.545 25 Birmingham, UK* 2.229 Chicago, USA* 9.469 26 Shenyang, China 2.091 Kinshasa, DRC* 9.304 27 Rome, Italy 1.884 Shenzhen, China†° 8.958 28 Milan, Italy 1.883 Bogota, Colombia¹ 8.932 29 San Francisco, USA 1.855 London, UK 8.618 30 Barcelona, Spain 1.809 Tehran, Iran 8.423 1950 2015 Macroeconomic analysis of public transport competitiveness in megacities 33 3.1.1. PRE-AUTOMOBILE CASES Regarding Table 3, among the thirty biggest cities in the World by 1950, only Tokyo and New York had mega-city status, and one third were running any kind of subway or rapid urban rail service. Disparities on urban size are notable; the fourth ranked, Shanghai, has half the population of the two firsts, cases considered megacities. In the European context, only London resembles this urban status. In spite of dropping down the rank from 1950 to 2015, London changed little in population. While Paris and Moscow doubled, London remained stable on 8 million mark, evidencing that the most urbanization occurred before automobile spread. In more detail, Fig. 9 evidence that during this sixty-five years gap London had re-arranged population from inner to outer areas, achieving similar numbers back to 1950. Table 4 – Summarized data for pre-automobile cases. Fig. 9 – London demographic evolution through nineteenth and twentieth century. Source: National Statistic Office. Further, concerning only the central municipalities, while Tokyo and New York have similar built-up area, and hence, densities, London exhibit a modest dimension. Despite the spatial differences, all three cases similarities extent to starting high-capacity public transportation in the nineteenth century, share comparable levels of industrial and service GVA, as well as high levels of income per capita and country urban population share. Table 4 summarized relevant similarities regarding the three cases. Moreover, each case lies on a different continent with specific-culture background and spatially distant. Hence, based on the attempt to diversify the cultural background and considering the inevitable pioneers on high-capacity public transportation, London completes the selected group of megacities from preautomobile era. Tokyo New York London Tokyo New York London 5 385 071 7 891 957 3 680 821 8 489 653 8 175 133 3 231 900 571.59 789.00 319.00 621.98 789.00 319.00 9 421.21 10 002.48 11 538.62 13 649.40 10 361.39 10 131.35 Industrial - - - 27.54 20.33 20.78 Service - - - 71.28 78.51 78.48 Agriculture - - - 1.18 1.16 0.74 63.27 70.00 78.44 90.52 80.77 81.30 1885 1904 1863 2010 Urbanization index (%) HCPT (year) 1950 GVA (%) Population Built-up urban area (%) Density (population/km²) Macroeconomic analysis of public transport competitiveness in megacities 34 3.1.2. POST-AUTOMOBILE CASES The sixty-five years gap evidence a great change on the World’s megacities rank. Regarding the top ten, only four cities remained ranked, the overall chart introduces thirteen new cities, and twenty-two cases now have mega-city status. Among the megacities, eight cases were out of the 1950 list and had urbanized in few decades. Expect for former Tokyo and New York, due to the lack of household travel survey, socio-economic data, and no high-capacity public transportation availability in the top ten megacities, only four cities have conditions to integrate this dissertation study case selection, i.e. São Paulo, Mexico City, Delhi and Jakarta. Furthermore, the four selected post-automobile megacities evidence disparities, demanding a finer subdivision and arrange. As summarized by Table 5 and Table 6, São Paulo and Mexico City started high-capacity public transportation services decades earlier, and both countries GVA share are more close to high-income/pre-automobile cases than properly emerging Delhi and Jakarta. Consequently, income per capita and urban population percentage are higher. The economics facts and public transportation infrastructure find similar grounds in São Paulo and Mexico City, as well as in Delhi and Jakarta, defining the two post-automobile subgroups: earlier high-income and later low-income. Table 5 – Summarized data for post-automobile cases with higher-income and early high-capacity public transportation. Table 6 – Summarized data for post-automobile cases with lower-income and late high-capacity public transportation. In order to balance the number of cases per group, post-automobile subgroups introduce Seoul in highincome and Bangkok in low-income. Despite the lower UN WUP rank for Seoul and the absence of Bangkok in the chart, some facts clarify the choosing decision. São Paulo Mexico City Seoul São Paulo Mexico City Seoul 2 198 096 3 050 442 1 437 670 11 253 503 8 851 080 9 631 482 - - 605.21 968.30 790.19 605.21 - - 2 375.49 11 621.92 11 201.20 15 914.28 Industrial 41.75 21.31 26.81 27.36 35.13 38.27 Service 38.76 59.45 39.33 67.79 61.53 59.26 Agriculture 19.5 13.74 39.36 4.85 3.34 2.47 46.14 50.75 27.71 84.34 77.83 81.94 1974 1969 1974 Urbanization index (%) HCPT (year) Population Built-up urban area (%) Density (population/km²) GVA (%) 1950 2010 Delhi Jakarta Bangkok Delhi Jakarta Bangkok 1 744 072 1 430 000 1 178 881 16 753 235 9 607 787 8 305 000 - - - 579.28 653.63 1 026.00 - - - 28 920.79 14 699.12 8 094.54 Industrial 19.93 15.05 19.16 27.16 42.78 40.03 Service 38.29 33.50 45.00 54.64 40.67 49.54 Agriculture 41.77 51.46 35.84 18.21 13.93 10.53 17.92 14.59 19.67 30.93 49.92 44.08 2002 2004 1999 Urbanization index (%) HCPT (year) Population Built-up urban area (%) Density (population/km²) GVA (%) 1950 2010 Macroeconomic analysis of public transport competitiveness in megacities 35 UN WUP data for Seoul only consider the Special City administration, an urban area limited to only 605 km2. The whole Seoul Capital Area accounts for more than twenty-four million inhabitants and covers near 11.700 km2. The low ranking occurs due to methodology mismatches regarding administrative boundaries. For example, Shanghai data contemplate an area of near 6.300 km2, and New York more than 34.000 km2. Hence, as shown by Table 5, Seoul evidences strong similarities with São Paulo and Mexico City, as well as complementing the desired cultural difference. Additionally, South Korea development counterpoint the urban fallacy of the relation between fast urbanization and poverty. From 1960 until 2010, South Korea population doubled and urban population rose from 27.71% to 81.94%. In accordance, the ratio between high-income countries and South Korea decrease from 12.30 in 1962 to only 1.64 in 2014. As a comparison, during the same interval China decrease ratio gap from 19.32 to 8.13 and urbanized significantly lesser than Korea did. Table 7 – Public transportation system select by megacities Source: Tokyo: Companies annual report. New York, London, São Paulo, Mexico City, and Seoul: Transit Summit Leadership 2014 report. Delhi: DCMR annual report. Bangkok: MRT and BTS annual report. Like Seoul, UN WUP ranking standards ignored Bangkok as one of the World’s most populous urban agglomeration. In spite of the absence, Bangkok not only gathers sufficient conditions for the global analysis and subgroup composition, as has household travel survey data available for comparison. However, all low-income post-automobile are located in Asia, more specifically south Asia. This City Systems Annual Ridership (millions) Network lenght (km) Tokyo JR East 3.311 - Keikyu 164 18,30 Keio 603 77,60 Keisei 135 23,90 Metro 2.366 185,10 Odakyu 331 26,50 Tokyo Waterfront 88 12,20 Seibu 523 93,40 Tobu 320 28,20 Toei 1.028 106,20 Tokyo Monorail 45 17,80 Tokyu 684 60,00 Tsukuba 56 13,20 Yurikamome 36 14,70 New York New York City Subway 1.785 438,00 London Underground 1.260 402,00 Dockland Light Railway 101 34,00 National Rail 126 86,00 São Paulo Metrô 1.107 66,20 Companhia Paulista de Trens Metropolitanos 642 257,50 Mexico City Sistema de Transporte Colectivo 1.609 226,00 Seoul Seoul Metropolitan Subway 2.553 331,50 Delhi Delhi Metro Rail Company 871 193,26 Jakarta Transjakarta - - Bangkok BTS Skytrain 219 36,90 Mass Rapid Transit 92 20,00 Macroeconomic analysis of public transport competitiveness in megacities 36 subgroup lack of diversity is in accordance to no existence of similar cases in Europe and America, no available data for African cities, and due to the intense urbanization of Asia itself. 3.1.3. HIGH-CAPACITY PUBLIC TRANSPORTATION In accordance to the previous high-capacity public transportation definition, each selected case study have at least one high-capacity public transportation system. Table 7 introduces the selected transportation services and provides some basic information. 3.2. DATA Based on existing literature, a set of variables were selected to perform the macro-economic analysis and support the discussion. According to the literature structure on chapter 2, data selection follows the similar order: land-use and demographics, complemented by transportation outcomes and economics. The chosen land-use variables were density, percentage of jobs in CBD, and HCPT coverage area. For demographics, the selected variables were GDP per capita, household size, vehicles per household, and households with no automobile. The transportation outcomes expressed by the mode split share divided into four categories: (a) private mode, (b) public transportation mode, (b1) HCPT mode, (b2) bus mode, trips per capita, and the fare recovery ratio. Density was selected due to the controversy relevance of the variable in the literature. In spite of any Newman and Kenworthy (1989) and Kenworthy and Laube (1999) methodological shortcomings, both studies incurred on comparing urban agglomerations with remarkable size disparities. As evidenced previously by Table 2, disregarding socio-economic and cultural features, small, middle or large size cities experience huge differences concerning population distribution. Comparing Hong Kong with Brussels, Portland, or Winnipeg have enough potential to drive any study to tricky situations. However, as this dissertation focus only to megacities, density will be tested in a more controlled environment, giving the chance for new insights concerning the variable relevance on travel pattern. Following, due to the size of megacities agglomerations and the macroeconomic scope, diversity indicator seeks to determine succinctly the spatial dispersion of each case study. In accordance to the Brotchie triangle (Wegener and Fürst, 1999), through the concentration of jobs in the CBD it is possible to coarsely determine the urban spatial interaction behaviour of each mega-city. As pointed out previously, megacities that tend to more dispersion and high mobility pattern should demand a more capillary network than monocentric models. Moreover, jobs percentage in CBD will be used as an indicator to identify a specific spatial interaction, rather than a variable used for analytical correlations and regressions. As stated previously, this dissertation is an introductory investigation based on macroeconomic analysis; hence, there is no place for finer details and concerns about micro data. Although several studies approaching specific road network and detailed accessibility concerns, as well as aging support, urban amenities and open space design have defended some influence on urban travel pattern, this study will consider only distance to public transportation. However, differently from De Grange et al. (2012), that used the length of subway tracks as a variable for measuring the size of the service, this study had computed the catchment area of each HCPT station. This procedure attempt to reduce the absence of information regarding road network design and avoid parallelism or redundant services. Due to the lack of information regarding roadways and expressways, this research was unable to provide a general macro indicator for private transportation accessibility similar to public transportation. The private mode accessibility gap was filled by the significant literature referencing the correlation between automobile ownership and GDP per capita (Dargay and Gately, 1999; Giuliano and Dargay, 2006), as Macroeconomic analysis of public transport competitiveness in megacities 37 well as motorization indicators, i.e. vehicles per household and household with no automobile, and automobile usage (Schimek, 1996). However, both motorization indicators have evident limitations and were analysed with caution. Regarding specific demographic features, household size and trips per capita were selected based on the relevance documented on travel pattern literature (Dunphy and Fisher, 1996; Gordon et al., 1988; Levinson and Kumar, 1995; Strathman et al., 1994). However, the absence of indicators regarding the increase of woman labour force should bias this investigation to disregard the number of working person per households, an impacting variable in current travel pattern literature (Gordon et al., 1988; Levinson and Kumar, 1995) In addition, GDP per capita were selected concerning two different objectives. First, the establishment of criteria for the study cases selection and group segregation. Further, GRDP per capita, more specifically the current stable values, is one of the recurring socio-economic indicator to measure prosperity levels. Although consumer prices index, interest rates and inflation levels have a decisive effect on the overall per capita income, richer society tend to have greater choice ability than poorer ones, as well as more opportunities to travel (Farber et al., 2014; Giuliano and Narayan, 2003; Kemp, 1973). Table 8 – Fare recovery ratio data to aggregate scatter plot regression. Finally, as some household travel survey do not compute travelled distance and time, as well as annual VMT, the only available transportation outcome is the mode split share. Although the massive presence of the former variables on travel pattern literature and the recurrence in the attempt to define each city travel pattern, the mode share seeks to reflects an overall image for urban mobility and respective autoCity Year Tokyo 1980 n171% 1990 n170% 2012 n130% New York 1995 l60% 2012 l39% London 1976 l76% 1980 l78% 1990 p84% 2012 p83% São Paulo 1978 l20% 1988 l29% 1998 l79% 2008 l78% Mexico City 2012 l*46% Seoul 2012 l74% Delhi 2008 p91% Jakarta - - Bangkok 2010 p86% Fare recovery ratio (%) * Fare operational recovery ratio Macroeconomic analysis of public transport competitiveness in megacities 38 dependent or sustainable level. Therefore, concerning the influence of the financial stability on HCPT services maintenance and expansion (Cohen, 1988; Plotch, 2015; Silva, 2015; Wachs, 1984), the fare recovery ratio complement the regression analysis. Similar to mode split share, data for fare recovery ratio results from all HCPT companies, disregarding each company individual results, as summarized on Table 8. The fare recovery ratio results are also divided into two groups: (a) Profitable, where results are higher than 100%; (b) Loss-making, where results are lower than 100%. In addition, in accordance to the fact that the majority of HCPT are in public domains, it is expected that those companies do not seek for profit making. Hence, loss-making services should be sub-divided into two more categories: (b1) Neutral loss, where results are between 100% and 80%; (b2) Strong loss, where results are lower than 80%. Table 9 – Primary data sources part 1. In accordance to the significant variety of variables and institutional organization of each study case, Table 9 and Table 10 summarize the selected primary data, corresponding unit, data source, and available sets. Nonetheless, while some primary data are ready to be used, i.e. population, household, registered vehicles, produced trips, built-up urban area, and households with no car, others require some adjusts and homogenization to equalize units and values. In accordance, the level of processing depends on the complexity and the lack of uniformity that each variable presents. Moreover, the combination and arrangements of primary data generates compounded indicators. Merging two or more primary data into one sub-product allows a more complex and risky bi-variable evaluation, correlating three or more variables at the same time. Population and household Registered vehicles Households with no vehicle Mode split share Trips Units Absolute value Absolute value Percentage Percentage Absolute value Tokyo New York 2000 and 2010 USA census New York State Department of motor vehicle 2000 and 2010 USA census London 1971, 1981, 1991, 2001 and 2011 UK census São Paulo Mexico City 1970, 1980, 1990 and 2010 Mexico census Inegi: Vehículos de motor registrados en circulación Origin-destination survey for 2007 data. Seoul Delhi 2011 India census 2001 and 2008 Rites report Jakarta Statistic Indonesia Wismadi et al. (2013) Senbil et al. (2007) JICA (2012) Bangkok 2005 and 2010 Thailand census Tulyasuwan (2013) Primary data Tokyo statistical yearbook Travel in London: Report 6. Transport for London 1978, 1988, 1998 and 2008 Origin-destination survey Delhi economic survey for 2014 1978, 1988, 1998 and 2008 Tokyo metropolitan region trip survey 1997/98 and 2010/11 Regional household travel survey Thailand: Making transport more energy efficient. World Bank Report Statistics of Seoul Statistics of Seoul El transporte en la región centro de México report for 1972, 1983 and 1994 data. 2007 Origindestination survey Macroeconomic analysis of public transport competitiveness in megacities 39 Table 10 – Primary data sources part 2. 3.2.1. PRIMARY DATA PROCESSING Regarding all primary data, public transportation coverage area has the most complex and laborious processing. First, it demands the definition of a methodology to calculate the public transportation network catchment area. The literature regarding walking distance to public transportation station approach three traditional methodologies: overlapping circles, Thiessen Polygons and non-overlapping circles with Thiessen polygons (Upchurch et al., 2004). In order to obtain a macro value, this dissertation uses the non-overlapping circles methodology, merging all catchment area into a single shape. As detailed road network and urban design are out of scope, fixed buffers from an estimated station centroid determine each station catchment area, progressing station by station to form the whole network. As the literature significantly vary for the maximum walking distance from 0.8 km to 1km (Crowley et al., 2009; Guerra, 2014b; Guerra et al., 2012; Upchurch et al., 2004; Yang et al., 2013), the buffer distance is set to a mean value of 0.9 km. In addition, each station-opening year determines the public transportation network size for each available household travel survey. Fare box recovery ratio Public transportation coverage area GDP per capita Built-up urban area Jobs Units Percentage Km² US Dollar Absolute value Km² Percentage Tokyo Shoji (2001) Open street map 2016 World Bank; Yusuf and Nabeshima (2006) New York Transit Leadership Summit 2012-2014 report Open street map 2016 OECD 2000 and 2010 USA census NYS Department of Labor London Transport for London Annual Report; Shoji (2001) Open street map 2016 World Bank; OECD São Paulo Metrô annual financial report Open street map 2016 World Bank; IBGE Embrapa Rodrigues and Silva (2009) Mexico City Transit Leadership Summit 2012-2014 report Open street map 2016 World Bank, INEGI Duhua and Giglia (2007) Seoul Transit Leadership Summit 2012-2014 report Open street map 2016 Yusuf and Nabeshima (2006); Glanville and Glanville (2011) Statistics of Seoul Barter (1999) Delhi DMRC Annual report 2014 Open street map 2016 Slack and Chattopadhyay (2009) Bijender (2014) Jakarta Open street map 2016 World Bank, World Bank (2009) Statistic Indonesia Barter (1999) Bangkok BMPT Annual report 2014 Open street map 2016 World Bank, Choiejit and Teungfung (2005) Angel et al. (2011) Barter (1999) Primary data Tokyo statistical yearbook Office for National statistics Macroeconomic analysis of public transport competitiveness in megacities 40 Based on the previous concept, the calculation of the HCPT coverage area uses QGIS software and Open Street Map data, as well as each company information regarding stations opening year and operational status. Nevertheless, services outside administrative boundaries are ignored, taking effect only if station catchment area extends significantly through the territory to justify the computing. Concerning mode split share, as some household travel survey do not report non-motorized trips and make no distinction between automobile and motorcycle, mode split share variable only concerns to motorized trips and the transportation modes are divided into three groups: high-capacity, i.e. rail or BRT, low-capacity, i.e. buses, auto-rickshaw, and informal services, and finally private. Hence, the total modal share results from the sum of private and public transportation modes. However, as the majority of cases report non-motorized numbers, this circumstantial omission does not invalidate non-motorized variable, applying analysis only for cases where the variable exists. About fare recovery ratio, while some values are extracted directly from the Transit Leadership Summit (2014) in the final form, others are obtained through each company financial annual report. In these cases, fare box recovery is the ratio between operational income and expenditures only, disregarding any other non-transportation related source of income. In particular, due the lack of public financial report from Mexico’s STC, fare recovery ratio referees only for 2004 data set. The data regarding jobs concentration were obtained through two different sources and under different standards. While for Seoul, Jakarta and Bangkok data derive from Barter (1999) and are formatted in absolute percentages values in accordance to CBD and inner areas. On the other hand, for New York, London, Tokyo, and São Paulo, data from each respective labour survey were used to analyse the job distribution. Due to the lack of official CBD or inner areas boundaries, each study case spatial dispersion pattern was obtained based on the differences of jobs concentration. For example, in São Paulo near 43 percent of jobs are concentrated in only three districts, which account for 8.7 percent of the municipal area. Finally, in order to homogenize the GRDP, an aggregated data set estimates values for all cases and time-series. This concept seeks to avoid currency disparities, local inflation corrosion, and changes on GRDP calculation methodology. The following formula estimates the uniformed GRDP per capita: CITY GRDP = City share (%) * National GDP (1) Where city share is obtained through institutional reports and national GDP from World Bank database values for GDP at market prices (current US$). 3.2.2. COMPOUND VARIABLES PROCESSING In addition to mode split share, fare recovery ratio, and the traditional composite indicators of density, household size, vehicles per household, and GRDP per capita, this study introduces an another composite indicator. Based on the ratio regarding HCPT coverage area and the built-up urban area, the indicator ‘β’ produces a macro value for overall city HCPT accessibility, disregarding the road network structure and detailed station features in this introductory stage. Further, in accordance to price elasticity of demand theory, the ratio between ‘β’ and HCPT mode share lead to ‘φ’, a three composite indicator expressed by formula 2.1. This indicator seeks to measure the impact that the network size has on attracting more patronage and increasing share on transportation market. In other words, how much each percent of network increase contributes to HCPT mode share. The results behaviour in the same way of price elasticity of demand, where ‘φ’ = 0 are interpreted as perfectly inelastic, 0 < ‘φ’ < 1 as inelastic, ‘φ’ = 1 as unit elastic, and ‘φ’ < 1 as elastic. Macroeconomic analysis of public transport competitiveness in megacities 41 φ = β HCPT mode share (2) The same concept can evaluate if the ‘β’ value produce the similar or reverse results over private mode, as well as applying the same criteria for a private mode performance indicator. However, due the lack of data regarding road infrastructure, with highlight to developing countries, and the pronounced literature concerning increasing levels of automobile ownership, this introductory research approaches the relation between vehicles per household and households with no automobile. Thus, in accordance to HCPT coverage area indicator, future investigation should approach private transportation through a similar concept, i.e. the road network density. Table 11 – Land-use and socio-economics selected data Year Density (pop/km²) Household size (persons) GRDP per capita (US dollars) Jobs concentration in CBD (%) Vehicles per household (1) Household with no vehicle (%) Trip per capita Tokyo T1 1978 8 259.45 2.69 11 917$ 0.514 0.463 - - T2 1988 8 201.99 2.47 29 863$ 0.518 0.688 - - T3 1998 8 319.67 2.31 44 314$ 0.505 0.745 - - T4 2008 8 668.26 2.05 49 835$ 0.516 0.707 - 3.88 London L1 1971 4 587.29 2.72 3 624$ - 0.622 0.538 2.10 L2 1981 4 129.79 2.58 17 403$ 0.577 0.765 0.447 2.24 L3 1991 4 185.27 2.36 33 674$ 0.531 0.801 0.407 2.06 L4 2001 4 561.96 2.38 43 357$ 0.627 0.822 0.376 2.14 L5 2011 5 198.93 2.50 58 812$ 0.682 0.715 0.416 2.14 New York N1 1998 10 149.91 2.65 64 295$ 0.627 0.548 0.535 3.20 N2 2008 10 361.39 2.63 71 437$ 0.616 0.570 0.546 3.60 São Paulo P1 1978 7 766.54 4.14 3 844$ - 0.608 0.557 2.83 P2 1988 9 426.23 3.83 7 556$ - 0.604 0.569 2.20 P3 1998 10 179.54 3.60 8 201$ - 0.701 0.476 1.99 P4 2008 11 253.37 3.34 24 940$ 0.621 0.647 0.493 1.95 Seoul S1 1990 18 017.76 3.27 11 000$ 0.300 0.387 - - S2 1996 17 299.54 3.03 12 649$ - 0.570 - 2.66 S3 2002 16 986.70 2.84 14 297$ - 0.635 - 2.89 S4 2010 17 474.01 2.50 27 310$ 0.237 0.665 - 2.95 M1 1972 9 820.24 5.13 1 282$ - - - 1.61 M2 1983 11 175.89 5.23 5 458$ - 1.028 - 2.54 M3 1994 10 422.49 4.04 7 378$ - 0.978 - 2.82 M4 2007 11 173.10 3.81 26 809$ - 1.423 0.522 2.49 Bangkok B1 2003 6 623.78 3.75 7 845$ 0.580 1.436 0.250 2.85 B2 2015 8 094.75 2.88 12 680$ - 2.191 - 2.82 Delhi D1 2008 23 927.40 4.89 3 161$ - 1.545 0.401 - Jakarta J1 2010 14 699.12 3.80 13 282$ 0.590 4.390 0.280 - Mexico City Macroeconomic analysis of public transport competitiveness in megacities 48 Fig. 12 – Diagram with the expected four elasticity of demand curves. Macroeconomic analysis of public transport competitiveness in megacities 49 Macroeconomic analysis of public transport competitiveness in megacities 50 4 RESULTS AND DISCUSSION This chapter presents the most relevant results, debating findings and related evidences, as well as some hypothetical explanations based on the intrinsic literature. Moreover, the main question and the three initial hypothesis are revisited, validating or discarding in accordance to results and findings. Furthermore, the analysis is structured based on a progressive narrative composed by three parts, starting from the macro Pearson’s bivariate correlation; finer bivariate regression graphical analysis and concluding with the conceptual application of the elasticity variance. As the macroeconomic analysis seeks for a comprehensive and concise understating, correlations and observed facts are debated in accordance to each significance and the required level of detail. In other words, findings that suggest weaken correlation in the macro level would not be analysed on further stages, and results with strong correlation are analysed only in the second and third stages. 4.1. VARIABLE CORRELATION 4.1.1. PRIMARY AND BASIC COMPOUNDED VARIABLES At first glance, except for the flagged correlation regarding private and public transportation mode share, correlations presented by Table 14 initially evidence the absence of spurious occurrences, even concerning compounded variables that share common indicators. In those cases, household size, vehicles per household and households with no vehicles evidenced insignificant correlation among household-based variables, as well as density, HCPT coverage area, and built-up area ratio regarding land area-based variables. In addition, the high correlation regarding private and public transportation mode should necessarily be expected. This behaviour occurs in accordance to the fact that mode split share is composed by only motorized modes. The absence of this evident correlation should be treated as a spurious fact and demand the revision of all transportation outputs variables. A more detailed observation of the bivariate Pearson’s correlation results suggests five clear evidences. First, in a macro-economic analysis of megacities, density, vehicles per household, households with no vehicles, private transportation share, and public transportation share have insignificant correlation regarding any other of the selected variables. Second, in accordance to the classical socio-economic literature, there is a notorious pattern regarding the decrease of household size and the general increase of GRDP per capita, i.e. negative correlation. This expected behaviour partially validate the adjusted calculation to estimates GRDP per capita. Third, calculation suggests a significant correlation regarding both GRDP per capita and household size to LCPT and HCPT share, as well as HCPT coverage area and built-environment share. In addition, trips per capita and HCPT mode share also evidenced correlation with significance at level 0.01. Fourth, while the five public transportation related variables evidence a significant mutual influence, private automobile share suggests ignoring any other variable Macroeconomic analysis of public transport competitiveness in megacities 51 impact and behaving in accordance to density and household motorization lack of influencing. Fifth, trips per capita evidence significant correlation to only HCPT mode share. Following, only the results with no significant correlation, GRDP per capita, and trips per capita will be debated in this first level. In spite of presenting some bivariate correlation coefficient significant in level 0.01, GRDP per capita and trips per capita fail to provide relevance in the regression analysis and downgrade to this first analytical stage. 4.1.1.1. Results with no significant correlation Concerning density, the lack of correlation is remarkable and contributes to the intense literature debate about how relevant increasing or decreasing urban densities is to perform changes on urban travel pattern. In spite of the significant disparities regarding sample’s density values, 4.129,79 (L2) to 23.927,40 (D1), the variable was unable to perform any significant correlation to both motorization levels, any transportation outputs, and even concerning income and household size variables. The fact that, in a macro analysis, densities do not alter neither private nor public transportation mode counterpoint most of Newman, Kenworthy and Laube findings and suggestions as the main instrument for decreasing autodependency in cities. Fig. 13 – Schematic diagram of samples distribution according to density threshold. At least three apparently plausible hypothesis explain the resulting behaviour of megacities density irrelevance on a macro level. The first, hypothetically defended by researches in advocacy of denser urban areas, relies on the fact that megacities densities accomplish the auto-dependent threshold level, 3.500 inhabitants per km2 (Guerra and Cervero, 2011; Newman and Kenworthy, 2006, 1989). Hence, the urban structure does not constrain travel pattern to auto-dependency and variances may occur regarding other factors. Fig. 13 schematically presents the hypothesis concept. In contrast, if the auto-dependence threshold so sharply influences travel pattern, none sample should evidence a continuous predominance of private transportation mode. However, one third of the samples have the dominance of private transportation mode, and specifically London have never experienced the reversed scenario for over four decades. This hypothesis is based on Brindle (1994) arguments concerning the improbable density threshold efficiency to determine accurately where start or end any travel pattern dependency. Moreover, regarding the public transportation no variance density threshold (Hayashi et al., 1992), available samples which are above that mark suggests a weaken stability force of public transportation share. Concerning only cities with more than two samples, in order to perform longer observation, while São Paulo public transportation share have a standard deviation of .015, and Seoul .013, Mexico City have .053, as shown on Table 13. The latter result is closer to that presented by London with .065 and Tokyo with .067. However, Tokyo (8.300), Mexico City (10.600), and São Paulo (10.300) densities disparities are subtle enough to invalidate density as the catalyser for such discrepancy. Macroeconomic analysis of public transport competitiveness in megacities 52 Table 13 – Case study density and PT share average and standard deviation. Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Sample 1 8 259.45 0.649 10 149.91 0.465 4 587.29 0.424 4 587.29 0.424 Sample 2 8 201.99 0.705 10 361.39 0.467 4 129.79 0.348 4 129.79 0.348 Sample 3 8 319.67 0.721 4 185.27 0.325 4 185.27 0.325 Sample 4 8 668.26 0.810 4 561.96 0.333 4 561.96 0.333 Sample 5 5 198.93 0.474 Average 8 362.34 0.721 10 255.65 0.466 4 532.65 0.381 4 366.08 0.358 Standard Deviation 209.53 0.067 149.54 0.001 427.42 0.065 242.09 0.045 Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Density (pop./km²) PT share (%) Sample 1 7 766.54 0.663 9 820.24 0.893 18 017.76 0.740 6 623.78 0.465 Sample 2 9 426.23 0.559 11 175.89 0.808 17 299.54 0.754 8 094.75 0.535 Sample 3 10 179.54 0.533 10 422.49 0.823 16 986.70 0.731 Sample 4 11 253.37 0.558 11 173.10 0.788 17 474.01 0.759 Average 9 656.42 0.578 10 647.93 0.828 17 444.50 0.746 7 359.27 0.500 Standard Deviation 1466.12 0.058 655.86 0.046 432.08 0.013 1040.13 0.049 Seoul Bangkok Note: * Calculations without the last sample. Tokyo New York London London* São Paulo Mexico City Macroeconomic analysis of public transport competitiveness in megacities 53 Table 14 – Bivariate Pearson’s correlation for primary and basic compounded variables. Dens. HH size GRDP per capita VEH per HH HH with no VEH Trips rate Private share PT share HCPT share LCPT share HCPT CA BUAR Coef. 1 ,363 -,259 ,177 ,072 ,299 -,392 ,392 ,001 ,404 -,331 ,215 Sig.(bilat.) ,063 ,192 ,378 ,799 ,188 ,043 ,043 ,996 ,041 ,092 ,280 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,363 1 -,645 ,175 ,079 -,328 -,323 ,323 -,527 ,836 -,800 -,687 Sig.(bilat.) ,063 ,000 ,383 ,781 ,147 ,100 ,100 ,005 ,000 ,000 ,000 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. -,259 -,645 1 -,157 -,149 ,443 ,234 -,234 ,376 -,598 ,568 ,516 Sig.(bilat.) ,192 ,000 ,433 ,596 ,045 ,241 ,241 ,053 ,001 ,002 ,006 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,177 ,175 -,157 1 ,635 ,178 ,394 -,394 -,130 ,126 -,239 -,160 Sig.(bilat.) ,378 ,383 ,433 ,011 ,441 ,042 ,042 ,517 ,539 ,230 ,426 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,072 ,079 -,149 ,635 1 -,169 ,523 -,523 -,138 -,126 -,034 -,138 Sig.(bilat.) ,799 ,781 ,596 ,011 ,581 ,045 ,045 ,624 ,669 ,904 ,624 N15 15 15 15 15 13 15 15 15 14 15 15 Coef. ,299 -,328 ,443 ,178 -,169 1 -,215 ,215 ,653 -,318 ,268 ,490 Sig.(bilat.) ,188 ,147 ,045 ,441 ,581 ,350 ,350 ,001 ,160 ,240 ,024 N21 21 21 21 13 21 21 21 21 21 21 21 Coef. -,392 -,323 ,234 ,394 ,523 -,215 1 -1,000 -,341 -,448 ,069 -,137 Sig.(bilat.) ,043 ,100 ,241 ,042 ,045 ,350 0,000 ,082 ,022 ,733 ,495 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,392 ,323 -,234 -,394 -,523 ,215 -1,000 1 ,341 ,448 -,069 ,137 Sig.(bilat.) ,043 ,100 ,241 ,042 ,045 ,350 0,000 ,082 ,022 ,733 ,495 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,001 -,527 ,376 -,130 -,138 ,653 -,341 ,341 1 -,664 ,786 ,716 Sig.(bilat.) ,996 ,005 ,053 ,517 ,624 ,001 ,082 ,082 ,000 ,000 ,000 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,404 ,836 -,598 ,126 -,126 -,318 -,448 ,448 -,664 1-,855 -,564 Sig.(bilat.) ,041 ,000 ,001 ,539 ,669 ,160 ,022 ,022 ,000 ,000 ,003 N26 26 26 26 14 21 26 26 26 26 26 26 Coef. -,331 -,800 ,568 -,239 -,034 ,268 ,069 -,069 ,786 -,855 1,742 Sig.(bilat.) ,092 ,000 ,002 ,230 ,904 ,240 ,733 ,733 ,000 ,000 ,000 N27 27 27 27 15 21 27 27 27 26 27 27 Coef. ,215 -,687 ,516 -,160 -,138 ,490 -,137 ,137 ,716 -,564 ,742 1 Sig.(bilat.) ,280 ,000 ,006 ,426 ,624 ,024 ,495 ,495 ,000 ,003 ,000 N27 27 27 27 15 21 27 27 27 26 27 27 Note: HH - household; VEH - vehicle; PT - public transport; CA - coverage area; BUAR - Built-up area ratio Bold. The correlation is significant on level 0.01 (bilateral). HCPT share LCPT share Coverage area BUAR Italic. The correlation is significant on level 0.05 (bilateral). VEH per household HH with no VEH Trips rate Private share Public transport share Density Household size GRDP per capita Macroeconomic analysis of public transport competitiveness in megacities 54 Finally, in accordance to the second hypothesis, as density vary considerably throughout the large megacity urban area, a mean value ignores nuances and particularities inside the urban structure and bias the analysis to a false result. This final argument is based on structural accessibility concerns, defending that only microscope land use and transportation investigation should determine the real relevance of density or any other variable. Consequently, density may find more convincing relevance under microeconomic and specific analysis. Table 15 – Tokyo and Seoul Housing-Jobs dispersion comparison. Source: Tokyo Statistical Yearbook and Seoul Statistics. Districts Population density (population/km²) Jobs density (jobs/km²) Ratio (%) Ratio (%) Jobs density (jobs/km²) Population density (population/km²) Districts Chiyoda-ku 3 589.2 62 917.5 0.057 2.459 3 010.7 7 404.0 Jongno-gu Chuo-ku 9 665.9 53 979.4 0.179 2.323 5 968.5 13 865.0 Jung-gu Minato-ku 9 137.7 36 917.4 0.248 2.324 4 936.5 11 471.0 Yongsan-gu Shinjuku-ku 16 769.9 27 618.8 0.607 2.225 8 432.8 18 762.0 Seongdong-gu Bunkyo-ku 16 766.8 16 782.5 0.999 2.141 10 575.3 22 640.0 Gwangjin-gu Taito-ku 16 387.5 20 038.7 0.818 2.350 11 209.3 26 340.0 Dongdaemun-gu Sumida-ku 16 812.6 11 491.6 1.463 2.202 10 588.8 23 318.0 Jungnang-gu Koto-ku 10 536.9 8 186.5 1.287 2.270 8 685.6 19 717.0 Seongbuk-gu Shinagawa-ku 15 244.6 14 698.5 1.037 2.254 6 439.4 14 513.0 Gangbuk-gu Meguro-ku 17 963.5 9 321.1 1.927 2.308 7 765.8 17 925.0 Dobong-gu Ota-ku 11 195.3 5 724.0 1.956 2.425 7 153.5 17 344.0 Nowon-gu Setagaya-ku 14 482.9 5 261.2 2.753 2.255 7 080.7 15 969.0 Eunpyeong-gu Shibuya-ku 13 456.9 24 866.3 0.541 2.348 8 142.4 19 116.0 Seodaemun-gu Nakano-ku 19 924.8 7 509.9 2.653 2.197 7 477.1 16 425.0 Mapo-gu Suginami-ku 15 537.5 4 782.2 3.249 2.397 12 101.1 29 004.0 Yangcheon-gu Toshima-ku 19 261.0 17 481.2 1.102 2.231 6 257.0 13 960.0 Gangseo-gu Kita-ku 16 047.2 6 476.0 2.478 2.254 9 904.8 22 326.0 Guro-gu Arakawa-ku 18 745.8 8 580.4 2.185 2.281 8 845.8 20 180.0 Geumcheon-gu Itabashi-ku 16 260.0 6 180.9 2.631 2.349 7 640.0 17 947.0 Yeongdeungpo-gu Nerima-ku 14 375.8 4 448.1 3.232 2.207 11 304.6 24 954.0 Dongjak-gu Adachi-ku 11 744.5 4 554.9 2.578 2.119 8 726.7 18 493.0 Gwanak-gu Katsushika-ku 12 195.1 4 285.0 2.846 2.353 3 914.2 9 210.0 Seocho-gu Edogawa-ku 13 115.6 4 245.3 3.089 2.251 6 415.4 14 441.0 Gangnam-gu 2.249 9 055.3 20 365.0 Songpa-gu 2.214 9 022.2 19 973.0 Gangdong-gu Standard deviation 1.056 0.084 Standard deviation Tokyo Seoul Macroeconomic analysis of public transport competitiveness in megacities 55 Table 16 – Basic pattern behaviour evidenced by time-series data. Another relevant contribution to the density debate is the demystification of the famous Jane Jacobs (1961) assumption relating high density to more diversity. Expect for Delhi and Seoul, the former due to lack of data and the latter due to the remarkable deviation, most cases presented a well-defined monocentric urban spatial structure and diversity pattern in accordance to Brotchie triangle features (Wegener and Fürst, 1999). The attempt to defend Seoul exception based on the very high density (18.000), blaming the other cities ‘monocentrism’ to inferior density values is fallacious and uncertain. In spite of the lack of data for Delhi, the denser study case, Tokyo provides some insights to counterpoint this Jacobian argument. Tokyo central area, commonly called ‘Special 23 Wards’, and Seoul have similar built-up area and population. In spite of the strong employment magnetism that central Tokyo exerts over the entire metropolitan area, Table 15 suggests that even regarding similar land use and demographic, as well as cultural backgrounds, density seems to perform no influence on Tokyo diversity. Furthermore, Gordon and Ikeda (2011) illustrated that a prison have considerable high densities with no diversity and a sport stadium have a similar behaviour for only a brief period. The latter density hypothesis concerning the inability of macro analysis to produce convincing results, could also be applied to explain the lack of correlation regarding both households’ motorization variables. In accordance, the existing literature usually compare motorization levels inside and outside public transportation catchment area (Crowley et al., 2009; Guerra, 2014b; Schimek, 1996). Additionally, vehicles per household variable have plenty of shortcomings and households with no vehicle suffer from the lack of data. Concerning the former indicator, variances on a macro perspective not necessarily means an increase or decrease of household with access to vehicles. On a hypothetical Variable Pattern Cases Slight increasing Tokyo, New York, Londonª, and Seoul. Stable São Paulo, Mexico City, and Bangkok. Decreasing gradually Tokyo, Londonª, São Paulo, Mexico City, Seoul, and Bangkok. Stable New York. GRDP per capita Increasing All cases. Monocentric Tokyo, New York, London, São Paulo, Jakarta, and Bangkok. Dispersed Seoul. Vehicles per household Increasing All cases*. Households with no vehicle Decreasing New York, London*, and São Paulo*. Increasing London*, São Paulo*, Mexico City, and Bangkok. Decreasing Tokyo. Stable New York, and Seoul. Increasing Tokyo, and Bangkok. Decreasing London*, São Paulo*, and Mexico City. Stable New York, and Seoul. Increasing Tokyo, New York, São Paulo, Mexico City, Seoul, and Bangkok. Stable London*. Decreasing New York, London, São Paulo, Mexico City, Seoul, and Bangkok. Stable Tokyo. HCPT coverage area Increasing All cases. HCPT mode share Density Household size Jobs in CBD Private transportation mode share Public transportation mode share LCPT mode share Note: ª London experience a significant increasing on population between L4 and L5. Hence, density and household size evidence a slight increase. * The last sample results breaks the previous tendency. For London, the breakdown may be influenced by the 2008 subprime crisis. For São Paulo, there is no evidence of a specific cause for the registered breakdown. However, the intermediate survey from 2012 suggest that household motorization and private transportation mode share have back to increasing tendency track. Macroeconomic analysis of public transport competitiveness in megacities 56 scenario where the number of household maintain stable, the indicator may increase if household with one vehicle buy another one, as well as decrease if household with three sells one. However, in both situations the number of households with no vehicles remained stable. Even considering the number of working person per household, the variable seems unable to perform macro-economic analysis accurately. Finally, the most surprising lack of significant correlation concerns the private and public transportation model share. This fact becomes even more intriguing based on the observed convincing influence regarding HCPT service coverage area to the respective mode share, as well as LCPT negative correlation. These macro findings suggest that the selected set of variables have insufficient force to influence the global urban transportation share and drive travel pattern away from any ‘auto-dependent’ transportation style. Moreover, the set of variables with insignificant correlation suggests that the socalled ‘auto-dependent’ on megacities do not concern to the pre or post-automobile development, density, diversity, denying the initial hypothesis 1.1 and 1.2. Although the hypothesis 2 sounds plausible, the lack of correlation regarding private and public transportation share to HCPT network size evidences that distance to public transportation require other service quality variables to justify any influence on mode choice. In general, HCPT coverage area tend primarily to reduce LCPT share and later increase overall public transportation share. Hence, the dynamics that involve the urban transportation market suggests requiring other variables to perform macro analysis, or, the intrinsic inability to explain travel pattern through simple macro indicators. A finer observation of case studies time-series provides some insights for such phenomena, as summarized by table 16. Such exercise may only be applied for long-standing or with at least two sample cases, i.e. Tokyo, New York, London, São Paulo, Mexico City, Seoul and Bangkok. It is quite acceptable that during the observed period several changes occurred concerning an unpredictable range of events, institutions changes and hierarchy, new policies and land use planning, and many other. For example, during the nineties, São Paulo and Seoul re-structured the public company responsible for running the bus service. In addition, while São Paulo and Mexico City adopted some vehicle restriction policy, London implemented a congestion-charging zone over the central area. Except for the breakdown regarding the last sample from London, note that the set of variables on every study case generally presents a sharp and stable tendency. While some cases evidence minor variance and tenuous changing, others are more radical and experience a remarkable increasing or decreasing pattern. However, the intrinsic progressive tendency is always kept, and changes tend to occur in a reasonable predictable pace. Back to the L5 breakdown, the Travel in London report #6 (Transport for London, 2013) states between 2001 and 2011, London transportation trends reversed for the first time, with public transportation increasing substantially as population also increased whereas automobile share fell. Despite the previous interval, 1991 and 2001, evidencing a slight increase on public transportation, the L5 survey register an abrupt increase from 33.3 percent to 47.4 percent (42 percent). Additionally, the motorization levels also dropped significantly by 13 percent for vehicles per household and 11 percent for households with no vehicle. Thus, this exceptional occurrence should be considered an anomaly caused by external factors, such as the 2008 global subprime crisis. Otherwise, based on the ‘success’ achieved by London’s last sample, planners may suggest a periodic economic turbulence to adjust travel pattern in accordance to sustainable goals. 4.1.1.2. Socio-economic correlation and trips rate As initially expected, the first results coarsely confirm that socio-economic and travel pattern correlation behave in accordance to the classical literature. In other words, under urbanization and economic growth, demographic indicators usually evidence a decrease in household size and increase in GRDP per capita. This classical and well-accepted consequence is the most plausible explanation regarding the correlation Macroeconomic analysis of public transport competitiveness in megacities 57 of socio-economics and travel pattern. However, despite of sharing common backgrounds, household size showed more expressive results than GRDP per capita. Considering that both socio-economic variables found a demure inter correlation, two hypotheses can explain this notorious disparity in a macro level. First, the used GRDP adjustment calculation shows some inaccurate for finer analysis and bias results to relatively less significant correlation coefficient. On the other hand, even considering some inaccuracy, GRDP per capita has effectively minor influence than household size. Tokyo New York Seoul Delhi Jakarta Bangkok London São Paulo Mexico City Fig. 14 – Motorization and GRDP per capita time-series evolution In spite of the lack of significant correlation between GRDP per capita and household motorization, a time-series analysis of the development of both variables coincide to Dargay and Gately (1999) findings and estimations. As summarized by Fig. 14, with the exception of the last samples from both London and São Paulo, along the observed period, the number of vehicles per thousand population evidenced a clear increasing pattern, as well as GRDP per capita. Furthermore, as predicted by Dargay and Gately (1999), emerging economies and fast developing countries led the motorization growth rate. However, the disaggregation of the motorization data in automobiles and motorcycle suggests that early Asian cities find and alternative means to travel. Although no other result or data support any reasonable explanation, cultural background and socio-economic features may fuel the radical number of motorcycle on early Asian cases. Regarding the absence of the government to provide sufficient public transportation services, the high number of low-income households and the major household size, the scarcity of parking places in both public space and next or inside dwelling, the motorcycle emerges as a feasible alternative to both automobile and public transportation. This behaviour is clearly evidenced Macroeconomic analysis of public transport competitiveness in megacities 64 Fig. 19 – Regression 5: LCPT share x coverage area linear regression Furthermore, another relevant question arises: ‘Does the LCPT exponential and the HCPT linear trend line vary regarding the size of the urban agglomeration?’ This hypothesis is based on the favourable competing environment that megacities offer to massive transportation modes. Consequently, if cities with 5 to 10 million and 1 to 5 million inhabitants benefit LCPT mode, the result should reverse the mega-city output, deforming LCPT from exponential to liner and the opposed regarding HCPT. On the other hand, if both public transportation modes weaken competition force, LCPT tend to maintain or intensify its deformation and HCPT change from linear to exponential. As agglomerations with less than 1 million inhabitants in general hardly support HCPT services with similar megacities standards, this methodological public transport modes split should be inadequate. However, megacities or any other urban area has a scarce source of land area. A careless interpretation can lead to the idea that a continuous increasing of HCPT coverage area will, in a certain point, reach the top threshold level and successfully define urban transportation completely auto-independent. Although the size of the HCPT service found significant correlations to both public transportation modes, it obviously should never exceed the urbanized area. Macroeconomic analysis of public transport competitiveness in megacities 65 Fig. 20 – Regression 6: HCPT share x built-up area ratio linear regression with outliers. Further, based on the strong correlation regarding HCPT coverage area and built-up area ratio, the HCPT share elasticity will be analysed, as well as the predicted curve pattern. Considering the complexity of the following analysis, as well as previous regression results, this regression is presented by two regressions, one containing all samples and a reduced version excluding some outliers. For the reduced version, all samples from Tokyo, due to an evident plot displacement, and Mexico City second sample (M2), due to suspicious HCPT mode share value, were excluded. Initially, the regression regarding all samples evidence an abrupt R-squared reduction to .513, as shown on Fig. 20. Although Tokyo results are significantly far from the trend line and London results appear closer to the general plot cloud, most of the remaining samples maintain a pattern similar to the previous regression. In accordance, London samples seems randomly scattered with intense variance, Bangkok and Tokyo a similar high slope pattern, and São Paulo, Mexico City, and Seoul a slope lower than the trend line. If Tokyo samples are removed, the plot scatter graphic and the correlation become much stronger (.808) and seems to describe a completely autonomous correlation between HCPT share and built-up area ratio, as shown by Fig. 21. However, a finer observation suggests that this finding could led to shortcomings. Macroeconomic analysis of public transport competitiveness in megacities 66 Fig. 21 – Regression 7: HCPT share x built-up area ratio linear regression without outliers. First, concerning the strong positive linear correlation, HCPT share should respond to infrastructure supply based on a linear progression, where both market share and supply increasing in a constant ratio regardless graphic position. In addition, as the dominance of public transportation mode is reasonably impossible, the HCPT share would never cross the top axis. In contrast, although expensive and not necessarily plausible, the HCPT service could physically cover the entire urbanized area and achieve the totality. Hence, the trend line slope need be lower than 45º degrees; rather one might find HCPT dominance without covering the urban area. Besides the R-squared disparities, regression six and seven also differs regarding the trend line projections. The regression six, biased by Tokyo samples, suggest that the full extent of HCPT services would lead to HCPT mode share to almost 90 percent. On the other hand, the robust regression seven trend line suggests a lower value near 60 percent of HCPT mode share. This results suggests that others factors rise Tokyo HCPT ability to compete, producing more mode share with lower coverage area. The elasticity of demand presents relatively inelastic and very elastic results for long-term elasticity, as shown on Table 18. Hence, in accordance to the high variety of results and the previous lack of landuse and socio-economics correlation, the first hypothesis can be partially excluded. Moreover, some other results also move the second hypothesis under the shadow of doubts. Although Bangkok and Tokyo are placed in contrasting positions, elasticity values are similar and both experience high elastic behaviour. In addition, as Tokyo figure near the middle of the virtual curve, the most acceptable result should tend to unit elasticity, behaving as a turning point from increasing to decreasing or vice-versa. Macroeconomic analysis of public transport competitiveness in megacities 67 Instead, with the third worst built-up area ratio growth and the best HCPT share growth, Tokyo has the most elastic result regarding all cases. In other words, Tokyo produces more share with minor network expansion and hence, minor investments and capital. Therefore, as the HCPT share accounts for more than three fourths of the motorized trips and the majority of railways companies in Tokyo are independent from government authority rules, this pattern should be reaching a threshold level soon. Furthermore, after Mexico City, Seoul presents the most inelastic results. Differently from any other case, Seoul invested massively on HCPT network expansion, almost doubling the ratio of the network in two decades. Even concerning absolute values, it is the most significant coverage area increase computing 170.2 km2 of service network expansion. However, this brutal infrastructure increment was unable to attract transportation share in a similar pace. As will be analysed later, Seoul is the typical case where HCPT investments have resulted mainly on enhancing competition regarding the public transportation counterpart, rather than the automobile mode. Table 18 – HCPT share elasticity of demand based on built-up area ratio long-term results. In accordance, both negative and positive curves require an orderly progressive increase or decrease of elasticity values. As both extremes experience the same elastic results, the hypothesis 1.1 and 1.2 are denied. Hence, each case study has a particular HCPT share elasticity and are significantly influence by local factors. The general HCPT share elasticity results tend to a more chaotic and random aspect. The results from HCPT share elasticity of demand based on infrastructure supply suggests that some external factors influence the HCPT ability to compete for better results. Without motorization levels, land use and socio-economic supportive correlation it is impossible to determine which factors influence to increase elasticity. However, the high variety of results and values suggests that influences may vary regarding each specific case. For example, while for London and São Paulo the increasing number of households with access to vehicles suggest influencing public transportation mode share, the same cannot be applied to Bangkok. In accordance, even on a remarkable high-density environment, Seoul massive investment on HCPT was unable to attract private mode users in the same intensity that Tokyo did. It is remarkable that differently from all other cases, Tokyo HCPT service thrive is a sine qua non condition, regardless any of the traditional land-use and socio-economic variables. As services continue to be operated by private companies, with the exception of the Toei Tokyo Metro Co., profit making and strong competing conditions are necessary to keep services feasible and consequently sustain Tokyo public transportation domain. Note that under such urban transportation free-market, public authority spending on public transportation is arbitrary and policies seeking to induce urban development around stations or constrain private transportation will necessarily support private concerns. However, Case Long-term elasticity Tokyo 3,860 London 1,808; -0,667* New York 1,900 São Paulo 1,246; 0,495* Mexico City 0,198 Seoul 0,230 Bangkok 2,574 Note: * Values disregarding the last sample value. Macroeconomic analysis of public transport competitiveness in megacities 68 differently from Hong Kong, another westernized Asian city, Tokyo government is highly liberal and conservative, defending minor regulation over private activities. In addition, as in Tokyo both HCPT and LCPT is generally operated by the same company, the operational strategy is planned to maximise profit and efficiency concerns. In contrast, cases such as São Paulo and Mexico City, experience a huge competition regarding public transportation modes. While in São Paulo the clash is led by two different governmental levels, i.e. State and Municipal, Mexico City experience formal and informal competition. 4.2.2.2. Public transportation related regressions Concerning the HCPT and LCPT regression, in spite of been the only regression with a logarithm trend line, the elasticity of demand suggests a reasonable R-squared value (.682) and a laudable issue for further investigation. First, the regression shown on Fig. 22 matches the ‘law of demand’ negative hyperbolic curve. In addition, the logarithm trend line also evidences a significant magnitude, suggesting that, based on an unknown reason, e.g. low fuel price or huge availability of road infrastructure, the private transportation is a hard competitor, which demand a robust HCPT share in public transportation composition. Consequently, it is expected that during the opening of the HCPT service, the shift share occurred primarily from LCPT, changing to private mode and later shifting mostly to HCPT near the public transportation equilibrium line. At a first glance, the regression evidence that just Tokyo and New York experience HCPT dominance over LCPT. Moreover, Tokyo is the only case where the former fact occurs in accordance to a remarkable public transportation dominance. In addition, while the last sample from Seoul suggests that the transportation market tend to match a triple force equilibrium, all samples from London suggests a public transportation equilibrium under private mode dominance. Finally, the others cases present a remarkable initial LCPT dominance, losing gradually the share to both private and HCPT mode. Table 19 – HCPT share elasticity of demand based on LCPT share long-term results. It is important to inform that for Jakarta, due to the lack of split transportation share concerning HCPT and LCPT, the public transportation number is always related to the sum of both modes. Hence, when comparing HCPT and LCPT, the resulting value will always rely in the 45º degree equilibrium line. The regression chart presents three different behaviour concerning long-term elasticity. In accordance to the results shown on Table 19, except for London, all cases presented negative result. Moreover, São Paulo and Mexico City compute relatively inelastic, New York present almost the unit elastic, Seoul and Bangkok minor elastic results, and Tokyo and London tend to perfectly elastic. Once again, Tokyo samples are far away from the plot cloud, and London presents a fuzzy plot dispersion. In addition, London is the only positive case, evidencing that both public transportation mode Case Long-term elasticity Tokyo -33,200 London 15,667; 0,182* New York -1,118 São Paulo -0,335 Mexico City -0,255 Seoul -1,413 Bangkok -2,014 Note: * Values disregarding the last sample value. Macroeconomic analysis of public transport competitiveness in megacities 69 share increased mutually. Similar to previous regressions, disregarding the last sample from London the elasticity results change radically from strongly elastic to almost perfectly inelastic. This London conflicting behaviour seems to repeat on every variable, suggesting the collateral effects and magnitude of the 2008 economic crisis. The differences regarding the elasticity of demand based on the disaggregated public transportation provides a concise image of the changing response of commuters. Cases evidencing high elasticity reflects that by some reason, which is invisible regarding this level of investigation, HCPT increase occurs in accordance to private mode share decrease, instead of LCPT. Concerning the two highly elastic results, although presenting similar results, while Tokyo suggests that LCPT is near the minimum share, London is far away LCPT suppression, closer to market equilibrium. This significant difference defines a remarkable evidence. As Tokyo LCPT was already demure and insignificant, in the attempt to increase market share HCPT had to compete exclusively with the private transportation. Hence, the high elastic behaviour results from a strong shift from private mode for HCPT, which become the main transportation means. However, the macro-economic level is unable to determine which factors influence this progressive pattern. Does Tokyo high elastic results occur in response to commuter’s preferences, or due to strong land-use and private vehicle restriction? Or else, due to both causes? Note that the increasing pattern is continuous and the individual trend line is well defined. Whatever the cause, this tendency seems robust and long-term effective, evidencing the lack of external influences. Fig. 22 – Regression 8: HCPT share x LCPT share linear regression Macroeconomic analysis of public transport competitiveness in megacities 70 In contrast, London high elastic results suggest occurring in accordance to absolute different causes. First, as London plots are apparently scattered in random order, it is hard to visualize the pattern tendency. A finer observation of the fuzzy plots distribution suggests the existence of a progressive trend before L5, broken by a strong external event that reversed the previous trend. Moreover, differently from Tokyo, London samples are close to the market equilibrium, and both HCPT and LCPT are relatively free to shift up or downward. The positive high elastic result of L1-L5 suggests that London commuters have a strong tendency to shift from private mode to primarily HCPT and minor to LCPT. This fact makes London an exceptional case where public transportation take advance from any land-use, socio-economic or transport related feature and convert into market share. Based on the previous results, it is possible to predict a long-term scenario where private share tends to minimum and public transportation domain urban mobility under the equilibrium of modes. In other words, if this clearly deformed trend maintain, the public transportation equilibrium trend line would represent the London tendency for the next decades. However, if the last sample is ignored the results is contrasting, and once again, unique. Regarding the L1-L4 samples, London ratify the isolated positive elasticity case, presenting this time a positive inelastic result. Consequently, this result denies completely the previous hypothetical scenario, evidence exactly the opposed. The very low positive inelastic suggest a clear shift from LCPT to private mode. Hence, HCPT remained quite stable, evidencing that the network expansion was unable to boost the competition concerning neither private mode nor LCPT. This fact is unique concerning all study cases available to perform time-series analysis. The discrepancy regarding both results suggests that the final London transportation outcomes is far from the stability and next surveys should track travel pattern back to the regular trend. Instead, if this new trend is definitive, next surveys should follow the previous prediction and fit the public transportation equilibrium line. Furthermore, the elastic result experienced by Bangkok, Seoul, and New York evidence that each case HCPT system was able to shift both private and LCPT mode share. This result is more remarkable for Bangkok, where the percentage of households with no vehicle is the lowest regarding all study cases and there is a high pressure from the rapidly increasing fleet of motorcycles as alternatives transportation means. In addition, although B2 value still far from the ‘no private mode line’ and inside private dominance zone, if the successful Bangkok’s HCPT increasing share keep the previous pace, in shortterm public transportation should start competing directly with the private mode. This assumption bases on the B2 proximity to public transportation equilibrium, and the remarkable elasticity trend line. Hence, Bangkok transportation authority should address more investment on HCPT, expect decreasing on LCPT share, and hence reduce gradually the amount of investments for LCPT expansion. Seoul presents the most equilibrate transportation market in accordance to three remarkable reasons. First, the inevitable proximity to the triple mode equilibrium point, computing near one third of share to each transportation mode. Secondly, the private transportation dominance was already exceeded and HCPT starts heading to decrease LCPT share shift and compete more directly with the private mode. Finally, the Seoul progression seems quite stable and uniform through the observed time. This consistent behaviour leads to believe that the logarithmical trend line would not be followed, meaning that Seoul could possibly looks like how Tokyo looks today, but decreasing primarily LCPT. Similarly, the previous analysis suggests the Bangkok should achieve Tokyo standard experiencing both LCPT and private transportation decrease. In accordance to current values and future estimations, regardless of Tokyo and Mexico City, Seoul could be the only case to never experience private transportation mode market domination. Although the macro level and the selected set of variables do not provide any reasonable answer to explain such behaviour, the spatial homogeneity and the astonishing increase on HCPT service may suggest some Macroeconomic analysis of public transport competitiveness in megacities 71 insights to support this remarkable fact. Accordingly, as Mexico City never crossed the private mode zone, it is possible to achieve similar results. However, based on the progressive trend line, the public authority should manage public transportation and urban policies to intensify the change from inelastic to elastic sooner. Although New York shows a result that fits perfectly its position in the chart, the limited number of samples and the historical literature suggest that may be not consistent. As evidenced by Cohen (Cohen, 1988), since the end of the fifties New York Metropolitan Transportation Authority struggle to stop the continuous ridership decrease and loss making. One reasonable hypothesis is that similar to London, New York single time-series reflect just the turning point scenario. Finally, São Paulo and Mexico City relatively inelasticity results from the lack of alternatives for urban mobility. In both cases, the first samples register low motorization levels and primitive HCPT services. As a result, the only available transportation mode that working population have access was LCPT or informal services. The HCPT service inability to expand in accordance to urbanization pace, income increase levels, and the changing commuter travel demand, influenced a higher population share afforded their own transportation means. The inevitable consequence of this phenomena is the gradual LCPT share decreasing, primarily private mode and following HCPT. Hence, initially both cases need to struggle to exceed LCPT, reinforce overall public transportation competiveness, improve HCPT services to later progressively compete under more equally conditions with the private mode. Table 20 – LCPT share elasticity of demand based on public transportation share long-term results. The last regression, ratify the deformed behaviour presented by LCPT share. At first glance, as shown by Fig. 23, although the correlation is positive and the highest R-squared value suggests a linear trend line, the respective elasticity of demand values, presented on Table 20, suggests a remarkable pattern that contradicts this result. This conclusive assumption is mainly supported by the greatest range of possible elasticity outcomes, ranging since positive perfectly inelastic to negative perfectly inelastic. Moreover, note that if the regression ignores all Tokyo samples and Mexico City second sample due to previously mentioned reasons, the R-squared value reach .713 and the trend line describe a linear correlation. The combination of the regression and the elasticity of demand allows concluding that post-automobile megacities tend to loss both public transportation and LCPT share during the first years of HCPT. However, at a given moment, this linear tendency shifts from positive to negative. This shift could be a consequence of the HCPT introduction, forcing elasticity to slowly become inelastic and converge to a virtual turning point. If HCPT became able enough to compete with private transportation, based on free-market or imposed restriction to other modes, the public transportation reverses the tendency and Case Long-term elasticity Tokyo -32,200 London 16,667; 1,182* New York -0,118 São Paulo 0,665 Mexico City 0,745 Seoul -0,413 Bangkok -1,014 Note: * Values disregarding the last sample value. Macroeconomic analysis of public transport competitiveness in megacities 72 perform a sharp increasing with LCPT decreasing, changing positive to negative inelastic. The more combative public transportation is, the more elastic the demand curve result. Hence, the elasticity results suggest the existence of a disguised parabola with horizontal axis of symmetry, completely different from the linear trend line that disregard Tokyo sample. The lack of non-motorized share restrains the advance for a final macro-economic analysis. Considering that only four study cases presents such data, it is hard to produce any acceptable interpretation and sustain the results with a sufficient support. The full extent use of transportation modes in the HCPT share and LCPT share regression, complemented by distance travelled, should provide a robust instrument able to compare the results with the three Brotchie triangle models. Consequently, regarding the latter data, it is possible to distinguish between ‘A’, ‘B’ or ‘C’ models. For example, model ‘A’ should be expressed by lower LCPT share, longer trips, and medium scale HCPT network. Concerning model ‘B’, the huge variety of destinations and dispersion should demand a very large HCPT coverage area, with more balance between LCPT and HCPT, and neither longer nor shorter travel. At last, the ‘C’ model require shorter trips, dominance of non-motorized transportation modes and ignore HCPT network size. Fig. 23 – Regression 9: LCPT share x public transportation mode share linear regression with outliers. 4.3. FARE BOX RECOVERY RATIO INFLUENCE Although there is no clear evidence or any statistical result suggesting the influence that the fare box recovery ratio have on public transportation share, the previous analysis provides insights. Based on DEAD ZONE Macroeconomic analysis of public transport competitiveness in megacities 73 each case last sample, there are four cases with high loss-making – São Paulo, Mexico City, Seoul, and New York, three with minor loss-making – London, Bangkok, and Delhi, and just one profit-making, Tokyo. In addition, only Tokyo, Seoul, and Mexico City experience public transportation dominance over private share. However, the private entrepreneurism heritage from pre-automobile cities and the comparison between operational and recovery ratio provide some information able to briefly understand the some influences and constrains that loss-making have on HCPT competiveness. The sample plots distribution on Fig. 22 evidence that in spite of the last samples from São Paulo, Bangkok and Jakarta, only New York and London, loss-making pre-automobile cases, relies remarkably inside the private transportation dominance era. However, while São Paulo and Bangkok, which were always managed under public interest, strives to follow the trend line and tangle the private mode dominance zone, New York and London are suggest a longer permanence inside the zone with a sheepish recovery force. Actually, just the fact of relying inside this zone do not necessarily means loss making and the absence of competiveness. As some studies suggests, it becomes a problem when is chronic and systematic (Cohen, 1988; Gomez-Ibanez, 1996; Plotch, 2015; Silva, 2015; Tomasevicius Filho, 2002; Wachs, 1984), as experienced by New York and London. First, both cases are the only regarding all selected cities where private entrepreneurism failed on running public transportation services. Since then, transportation companies never recover the financial health, and year after year, delays on network expansion or maintenance savings undermined the capacity to compete with private mode (Cohen, 1988; Gomez-Ibanez, 1996; Wolmar, 2009). However, the differences regarding operational and recovery ratio set split the future perspective for both cases. Based on Table 21, while London generates 90 percent of operational costs and loses only 7 percent for property depreciation and loan interest, New York generates similar 82 percent of operational costs, losing impressive 43 percent for depreciation and administrative spending. In other words, public subsidize are more able to manage London loss making and allow future expansion. In contrast, as presented by Cohen (1988) and Plotch (2015), since the seventies that the Federal and State governments increased significantly the New York Metropolitan Transit Authority subsidy. However, the assistance were unable to stop the financial bleeding, and New York MTA still struggle to balance public counts. It is worthy to observe that London and New York HCPT companies had limited moves due to the very nature of public held institutions (Cohen, 1988; Gomez-Ibanez, 1996; Plotch, 2015). Generically, both cases can reduce expending, which without subsidies may affect service quality and potentially patronage, increase fares, which in short-term should decrease patronage, and constrain or overcharge private transportation, which led to limited individual choices. Moreover, it is relevant to remember that due to the lack of infrastructure maintenance and controversial massive investment in network expansion during the end of the sixties, New York MTA faced a vertiginous patronage decrease and financial crisis. In order to balance the books, others sources of income, such as publicising, real estate opportunities, capital diversification of the capital, and maximising profit should be addressed for saving companies healthiness. Nonetheless, those measures are intrinsic to private sector (Mises, 2010). However, note that based on the previous results and analysis from 4.2, London transportation future is completely uncertain. In addition, although available New York dataset sounds more promising, the lack of older travel surveys casts doubts about predicting future moves. Such scenario creates a hostile environment for planning strategies and recovery actions. On the other hand, in spite of presenting loss-making results, the public held HCPT services from postautomobile megacities presents positive operational costs. Hence, numbers become negative primarily due to debts interest for building and expanding the service network, and minor due to property depreciation, as explicit on Table 21. In such scenario, public companies take advance over private ones, Macroeconomic analysis of public transport competitiveness in megacities 80 6 FUTURE WORK This research results and debate lead to three future research sheds. First, the methodology should be improved and the missing data gathered, in the attempt to fill some open gaps and confirm the validation of the process. Consequently, the research for more data, and new cases, should continue in the future, monitoring the robustness of the development pattern and sights of change. An imaginative hypothetical scenario supports this first research field: considering the existence of the previous optimal variables, what would be the changing travel pattern from 1850 to 1900? Moreover, what to expect regarding the later 1900-1950, and future 2000-2050? Second, the same kind of analysis should be applied to other urban agglomerations sizes, regarding always a minimum of comparative standards and diversity. Would agglomerations with 5 to 10 million or 1 to 5 million behave similarly or significantly different? Are HCPT able to perform reasonable services on different urban realities? Would London finally found a similar pattern when compared to non-mega-city cases? Finally, instead of remaining in the macro level, further investigation should go deeper and research the microeconomics of megacities urban transportation. Consequently, the selected variables would change significantly, as well as the methodological approach. Moreover, while for macroeconomic analysis more comparable cases a determinant to robust the results, a more detailed micro approach would probably lead to a minor number of case study, refining the selection process to avoid critical situations. However, although it seems a quite different approach and universe, any further microeconomic analyses should necessarily regard much of the findings defined in the macro level. The changing of scope also change the questions concerns and objectives. Instead of researching for global patterns and distorted bias, the microeconomic focus on understanding the engine behind a certain pattern. (Almec Corporation Oriental Consultants Co. Ltd., 2012; Angel et al., 2011; Bijender and Joginder, 2014; Choiejit and Teungfung, 2005; Duhau and Giglia, 2007; Glanville and Glanville, 2011; Rodrigues and Silva, 2009; Shoji, 2001; Slack and Chattopadhyay, 2009; Tulyasuwan, 2013; Wismadi et al., 2013; Yusuf and Nabeshima, 2006) Macroeconomic analysis of public transport competitiveness in megacities 81 REFERÊNCIAS BIBLIOGRÁFICAS Agénor, P.-R., 2010. A Theory of Infrastructure-led Development. J. Econ. Dyn. Control 34, 932–950. doi:10.1016/j.jedc.2010.01.009 Almec Corporation Oriental Consultants Co. Ltd., 2012. Project for the study on Jabodetabek public transportation policy implementation strategy in the Republic of Indonesia (JAPTRAPIS). Jakarta, Indonesia. Anas, A., Pines, D., 2008. Anti-sprawl policies in a system of congested cities. Reg. Sci. Urban Econ. 38, 408–423. doi:10.1016/j.regsciurbeco.2008.05.001 Angel, S., Parent, J., Civco, D.L., Blei, A., Potere, D., 2011. The dimensions of global urban expansion: Estimates and projections for all countries, 2000-2050. Prog. Plann. 75, 53–107. doi:10.1016/j.progress.2011.04.001 Ântico, C., 2005. Deslocamentos pendulares na região metropolitana de são paulo. São Paulo em Perspect. 19, 110–120. doi:10.1590/S0102-88392005000400007 Bae, C.-H.C., Jun, M.-J., Park, H., 2003. The impact of Seoul’s subway Line 5 on residential property values. Transp. Policy 10, 85–94. doi:10.1016/S0967-070X(02)00048-3 Balaker, T., Staley, S., 2006. The road more traveled: why the congestion crisis matters more than you think, and what we can do about it, First. ed. Rowman & Littlefield Publishers, Inc., Oxford. Barter, P., 1999. An International Comparative Perspective on Urban Transport and Urban Form in Pacific Asia: The Challenge of Rapid Motorisation in Dense Cities. -. Murdoch University. Ben-Akiva, M., Morikawa, T., 2002. Comparing ridership attraction of rail and bus. Transp. Policy 9, 107–116. doi:10.1016/S0967-070X(02)00009-4 Biba, S., Curtin, K.M., Manca, G., 2010. A new method for determining the population with walking access to transit. Int. J. Geogr. Inf. Sci. 24, 347–364. doi:10.1080/13658810802646679 Bijender, S., Joginder, S., 2014. Land Use / Land Cover Change of Delhi : A Study using Remote Sensing and GIS Techniques. Int. Res. J. Earth Sci. 2, 15–20. BMPC, 2014. Annual Report 2014. Bangkok. Brindle, R., 1994. Lies, Damned Lies and “Automobile Dependence” - Some hyperbolic reflections. Australas. Transp. Res. Forum 19, 117–131. Brueckner, J., Kim, H., 2003. Urban Sprawl and the Property Tax. Int. Tax Public Financ. 10, 5–23. Brueckner, J.K., 2005. Transport subsidies, system choice, and urban sprawl. Reg. Sci. Urban Econ. 35, 715–733. doi:10.1016/j.regsciurbeco.2005.01.001 Bruegmann, R., 2006. Sprawl: A Compact History. Robert Bruegmann. Ann. Assoc. Am. Geogr. doi:10.1111/j.1467-8306.2006.00712.x BTS, 2016. Bangkok Mass Transit System Public Company Ltd. Buehler, R., 2011. Determinants of transport mode choice: a comparison of Germany and the USA. J. Transp. Geogr. 19, 644–657. doi:10.1016/j.jtrangeo.2010.07.005 Buehler, R., 2010. Transport Policies, Automobile Use, and Sustainable Transport: A Comparison of Germany and the United States. J. Plan. Educ. Res. 30, 76–93. doi:10.1177/0739456X10366302 Calvo, F., de Oña, J., Arán, F., 2013. Impact of the Madrid subway on population settlement and land use. Land use policy 31, 627–639. doi:10.1016/j.landusepol.2012.09.008 Macroeconomic analysis of public transport competitiveness in megacities 82 Cervero, R., 1998. The Transit Metropolis: A Global Inquiry. Island Press. Cervero, R., 1990. Transit pricing research - A review and synthesis. Transportation (Amst). 17, 117– 139. doi:10.1007/BF02125332 Cervero, R., Dai, D., 2014. BRT TOD: Leveraging transit oriented development with bus rapid transit investments. Transp. Policy 36, 127–138. doi:10.1016/j.tranpol.2014.08.001 Cervero, R., Golub, A., 2007. Informal transport: A global perspective. Transp. Policy 14, 445–457. doi:10.1016/j.tranpol.2007.04.011 Cervero, R., Kang, C.D., 2011. Bus rapid transit impacts on land uses and land values in Seoul, Korea. Transp. Policy 18, 102–116. doi:10.1016/j.tranpol.2010.06.005 Cervero, R., Kockelman, K., 1997. Travel demand and the 3Ds: Density, diversity, and design. Transp. Res. Part D Transp. Environ. 2, 199–219. doi:10.1016/S1361-9209(97)00009-6 Chakrabarti, S., Giuliano, G., 2015. Does service reliability determine transit patronage? Insights from the Los Angeles Metro bus system. Transp. Policy 42, 12–20. doi:10.1016/j.tranpol.2015.04.006 Choiejit, R., Teungfung, R., 2005. Urban Growth and Commuting Patterns of the Poor in Bangkok. Soc. Sci. 1–19. Circular, B., Giuliano, G., 1997. Land use policy and transportation. Why we won’t get there from here. Policies Foster. Sustain. Transp. Technol. 27. Clark, W.A. V., Kuijpers-Linde, M., 1994. Commuting in Restructuring Urban Regions. Urban Stud. 31, 465–483. doi:10.1080/00420989420080431 Clifton, G.T., Mulley, C., Hensher, D. a., 2014. Bus Rapid Transit versus Heavy Rail in suburban Sydney – Comparing successive iterations of a proposed heavy rail line project to the preexisting BRT network. Res. Transp. Econ. 48, 126–141. doi:10.1016/j.retrec.2014.09.010 Cohen, J., 1988. Capital investment and the decline of mass transit in New York, 1945-1981. Urban Aff. Q. 23, 369–388. Crowley, D., Shalaby, A., Zarei, H., 2009. Access Walking Distance, Transit Use, and TransitOriented Development in North York City Center, Toronto, Canada. Transp. Res. Rec. J. Transp. Res. Board 2110, 96–105. doi:10.3141/2110-12 Cullinane, S., Cullinane, K., 2003. Car dependence in a public transport dominated city: Evidence from Hong Kong. Transp. Res. Part D Transp. Environ. 8, 129–138. doi:10.1016/S13619209(02)00037-8 Curtis, C., Renne, J.L., Bertolini, L., 2009. Transit Oriented Development: Making it Happen. Ashgate, Surrey. Dargay, J.M., Gately, D., 1999. Income’s effect on car and vehicle ownership, worldwide: 1960–2015. Transp. Res. Part A Policy Pract. 33, 101–138. doi:10.1016/S0965-8564(98)00026-3 Davis, L.W., 2008. The Effect of Driving Restrictions on Air Quality in Mexico City. J. Polit. Econ. 116, 38–81. doi:10.1086/529398 De Grange, L., Troncoso, R., González, F., 2012. An empirical evaluation of the impact of three urban transportation policies on transit use. Transp. Policy 22, 11–19. doi:10.1016/j.tranpol.2012.04.003 Macroeconomic analysis of public transport competitiveness in megacities 83 Deng, T., Ma, M., Wang, J., 2013. Evaluation of Bus Rapid Transit Implementation in China: Current Performance and Progress. J. Urban Plan. Dev. 139, 226–234. doi:10.1061/(ASCE)UP.19435444.0000150 Dieleman, F., Wegener, M., 2004. Compact City and Urban Sprawl. Built Environ. 30, 308–323. doi:10.2307/24026084 DMRC, 2015. Annual Report 2014-2015. Delhi. Duarte, F., Ultramari, C., 2012. Making Public Transport and Housing Match: Accomplishments and Failures of Curitba’s BRT. J. Urban Plan. Dev. 138, 183–194. doi:10.1061/(ASCE)UP.19435444.0000107 Duhau, E., Giglia, A., 2007. Nuevas centralidades y prácticas de consumo en la Ciudad de México: Del microcomercio al hipermercado. Eure 33, 77–95. doi:10.4067/S0250-71612007000100005 Dunphy, R., Fisher, K., 1996. Transportation, Congestion, and Density: New Insights. Transp. Res. Rec. J. Transp. Res. Board 1552, 89–96. doi:10.3141/1552-12 Eboli, L., Mazzulla, G., 2011. A methodology for evaluating transit service quality based on subjective and objective measures from the passenger’s point of view. Transp. Policy 18, 172– 181. doi:10.1016/j.tranpol.2010.07.007 Esfahani, H.S., Ramırez, M.T., 2003. Institutions, infrastructure, and economic growth. J. Dev. Econ. 70, 443–477. doi:10.1016/S0304-3878(02)00105-0 Ewing, R., 1997. Is Los Angeles-Style Sprawl Desirable? J. Am. Plan. Assoc. 63, 107–126. doi:10.1080/01944369708975728 Ewing, R., Cervero, R., 2010. Travel and the Built Environment. J. Am. Plan. Assoc. 76, 265–294. doi:10.1080/01944361003766766 Ewing, R., Deanna, M., Li, S.-C., 1996. Land Use Impacts on Trip Generation Rates. Transp. Res. Rec. 1518, 1–6. doi:10.3141/1518-01 Ewing, R., Pendall, R., Chen, D., 2002. Measuring Sprawl and Its Impacts. Smart Growth America 175–183. Farber, S., Bartholomew, K., Li, X., Páez, A., Nurul Habib, K.M., 2014. Assessing social equity in distance based transit fares using a model of travel behavior. Transp. Res. Part A Policy Pract. 67, 291–303. doi:10.1016/j.tra.2014.07.013 Gaigné, C., Riou, S., Thisse, J.-F., 2012. Are compact cities environmentally friendly? J. Urban Econ. 72, 123–136. doi:10.1016/j.jue.2012.04.001 Giuliano, G., 2005. Low Income, Public Transit, and Mobility. Transp. Res. Rec. 1927, 63–70. doi:10.3141/1927-08 Giuliano, G., Dargay, J., 2006. Car ownership, travel and land use: a comparison of the US and Great Britain. Transp. Res. Part A Policy Pract. 40, 106–124. doi:10.1016/j.tra.2005.03.002 Giuliano, G., Narayan, D., 2003. Another look at travel patterns and urban form: The US and Great Britain. Urban Stud. 40, 2295–2312. doi:10.1080/0042098032000123303 Giuliano, G., Small, K., 1993. Is the Journey to Work Explained by Urban Structure? Urban Stud. 30, 1485–1500. doi:10.1080/00420989320081461 Glanville, A., Glanville, J., 2011. Economics from a global perspective, Third edit. ed. Glanville Books Ltd., Dokon, United Kingdom. Macroeconomic analysis of public transport competitiveness in megacities 84 Gomez-Ibanez, J.A., 1996. Big-City Transit Ridership, Deficits, and Politics: Avoiding Reality in Boston. J. Am. Plan. Assoc. 62, 30–50. doi:10.1080/01944369608975669 Gordon, P., Cox, W., 2012. Cities in Western Europe and the United States: do policy differences matter? Ann. Reg. Sci. 48, 565–594. doi:10.1007/s00168-011-0495-8 Gordon, P., Ikeda, S., 2011. Does density matter?, in: Andersson, D.E., Andersson, A.E., Mellander, C. (Eds.), Handbook of Creative Cities. Edward Elgar Publishing, Cheltenham, p. 576. Gordon, P., Kumar, A., Richardson, H.W., 1988. Beyond the journey to work. Transp. Res. Part A Gen. 22, 419–426. doi:10.1016/0191-2607(88)90045-3 Gordon, P., Richardson, H.W., 1997. Are Compact Cities a Desirable Planning Goal? J. Am. Plan. Assoc. 63, 95–106. doi:10.1080/01944369708975727 Gordon, P., Richardson, H.W., 1989. Gasoline Consumption and Cities: A Reply. J. Am. Plan. Assoc. 55, 342–346. doi:10.1080/01944368908975421 Guerra, E., 2014a. The Built Environment and Car Use in Mexico City: Is the Relationship Changing over Time? J. Plan. Educ. Res. 34, 394–408. doi:10.1177/0739456X14545170 Guerra, E., 2014b. Mexico City’s suburban land use and transit connection: The effects of the Line B Metro expansion. Transp. Policy 32, 105–114. doi:10.1016/j.tranpol.2013.12.011 Guerra, E., Cervero, R., 2011. Cost of a Ride: The Effects of Densitites o Fixed-Guideway Transit Ridership and Costs. J. Am. Plan. Assoc. 77, 267–290. doi:10.1080/01944363.2011.589767 Guerra, E., Cervero, R., Tischler, D., 2012. Half-Mile Circle. Transp. Res. Rec. J. Transp. Res. Board 2276, 101–109. doi:10.3141/2276-12 Hayashi, H., Tomita, Y., Doi, K., Rithica, S., Kato, H., 1992. A comparative study on urban transport energy consumption and its influences on the environment. Infrastruct. Plan. 15, 939–944. Headicar, P., 2013. The Changing Spatial Distribution of the Population in England: Its Nature and Significance for “Peak Car.” Transp. Rev. 33, 310–324. doi:10.1080/01441647.2013.802751 Hensher, D.A., King, J., 1998. Establishing Fare Elasticity Regimes for Urban Passenger Transport: Time-Based Fares for Concession and Non-concession Markets Segmented by Trip Length. J. Transp. Stat. 1, 43–61. Hensher, D.A., Li, Z., Mulley, C., 2014. Drivers of bus rapid transit systems – Influences on patronage and service frequency. Res. Transp. Econ. 48, 159–165. doi:10.1016/j.retrec.2014.09.038 Hensher, D.A., Stopher, P., Bullock, P., 2003. Service quality––developing a service quality index in the provision of commercial bus contracts. Transp. Res. Part A Policy Pract. 37, 499–517. doi:10.1016/S0965-8564(02)00075-7 Hidalgo, D., Pereira, L., Estupiñán, N., Jiménez, P.L., 2013. TransMilenio BRT system in Bogota, high performance and positive impact - Main results of an ex-post evaluation. Res. Transp. Econ. 39, 133–138. doi:10.1016/j.retrec.2012.06.005 Hirshleifer, J., Glazer, A., Hirshleifer, D., 2005. Price theory and applications: decisions, markets, and information, Finance. Jackson, K.T., 1987. Crabgrass Frontier: The Suburbanization of the United States. Oxford University Press. Jacobs, J., 1961. The Death and Life of Great American Cities. Macroeconomic analysis of public transport competitiveness in megacities 85 Jones, D.W., 2008. Mass Motorization + Mass Transit: An American History and Policy Analysis. Indiana University Press. Jun, M.J., 2012. Redistributive effects of bus rapid transit (BRT) on development patterns and property values in Seoul, Korea. Transp. Policy 19, 85–92. doi:10.1016/j.tranpol.2011.09.003 Kemp, M. a., 1973. Some evidence of transit demand elasticities. Transportation (Amst). 2, 25–52. doi:10.1007/BF00172614 Kenworthy, J.R., Laube, F.B., 1999. Patterns of automobile dependence in cities: an international overview of key physical and economic dimensions with some implications for urban policy. Transp. Res. Part A Policy Pract. 33, 691–723. doi:10.1016/S0965-8564(99)00006-3 Kindleberger, C.P., 1986. The World in Depression, 1929-1939. University of California Press. Koh, W.T.H., 2004. Congestion Control and Vehicle Ownership Restriction The Choice of an Optimal Quota Policy 38, 371–402. Lee, B.A., Gordon, P., Richardson, H.W., Moore, J.E.I., 2009. Commuting Trends in U.S. Cities in the 1990s. J. Plan. Educ. Res. 29, 78–89. doi:10.1177/0739456x09331549 Levinson, D., Kumar, a, 1997. Density and the journey to work. Growth Change 28, 147–72. doi:10.1111/j.1468-2257.1997.tb00768.x Levinson, D., Kumar, A., 1995. Activity, travel, and the allocation of time. J. Am. Plan. Assoc. 61, 458. doi:10.1080/01944369508975657 Lin, C., 2011. The Effects of Driving Restrictions on Air Quality: São Paulo, Bogotá, Beijing, and Tianjin. Agric. Appl. …. Litman, T., 2004. Transit Price Elasticities. J. Public Transp. 37–58. Litman, T., Laube, F., 2002. Automobile Dependency and Economic Development. Transportation (Amst). 1–19. Loo, B.P.Y., Chow, A.S.Y., 2011. Jobs-housing balance in an era of population decentralization: An analytical framework and a case study. J. Transp. Geogr. 19, 552–562. doi:10.1016/j.jtrangeo.2010.06.004 Mahendra, A., 2008. Vehicle Restrictions in Four Latin American Cities: Is Congestion Pricing Possible? Transp. Rev. 28, 105–133. doi:10.1080/01441640701458265 McDonald, J.F., Osuji, C.I., 1995. The effect of anticipated transportation improvement on residential land values. Reg. Sci. Urban Econ. 25, 261–278. doi:10.1016/0166-0462(94)02085-U McIntosh, J., Trubka, R., Kenworthy, J., Newman, P., 2014. The role of urban form and transit in city car dependence: Analysis of 26 global cities from 1960 to 2000. Transp. Res. Part D Transp. Environ. 33, 95–110. doi:10.1016/j.trd.2014.08.013 Mises, L. Von, 2010. Liberalismo: segundo a tradição clássica, 2a ed. Instituto Ludwig Von Mises Brasil, São Paulo. Newman, P., Kenworthy, J., 2011. “Peak Car Use”: Understanding the Demise of Automobile Dependence. J. World Transp. Policy Pract. 17, 31–42. Newman, P., Kenworthy, J., 2006. Urban Design to Reduce Automobile Dependence. Op. An Int. J. Suburb. Metrop. Stud. 2, 35–52. doi:Cited By (since 1996) 16\rExport Date 27 September 2011 Newman, P., Kenworthy, J., 1999. Sustainability and Cities: Overcoming Automobile Dependence. Island Press. Macroeconomic analysis of public transport competitiveness in megacities 86 Newman, P., Kenworthy, J., Glazebrook, G., 2013. Peak Car Use and the Rise of Global Rail: Why This Is Happening and What It Means for Large and Small Cities. J. Transp. Technol. 03, 272– 287. doi:10.4236/jtts.2013.34029 Newman, P.W.G., Kenworthy, J.R., 1989. Gasoline Consumption and Cities. J. Am. Plan. Assoc. 55, 24–37. doi:10.1080/01944368908975398 Nivola, P.S., 1999. Laws of the Landscape: How Policies Shape Cities in Europe and America. Brookings Institution Press. OECD, 2012. Development Centre Studies Industrial Policy and Territorial Development Lessons from Korea: Lessons from Korea. OECD Publishing. Paulley, N., Balcombe, R., Mackett, R., Titheridge, H., Preston, J., Wardman, M., Shires, J., White, P., 2006. The demand for public transport: The effects of fares, quality of service, income and car ownership. Transp. Policy 13, 295–306. doi:10.1016/j.tranpol.2005.12.004 Pinnell, P., 2009. Smart Code 9.2, The Town Paper Publisher. Plotch, P.M., 2015. Waiting More Than 100 Years for the Second Avenue Subway to Arrive. J. Plan. Hist. 14, 309–328. doi:10.1177/1538513215590496 Priester, R., Kenworthy, J.R., Wulfhorst, G., 2013. The diversity of megacities worldwide: Challenges for the future of mobility, in: Research, I. for M. (Ed.), Megacity Mobility Culture: How Cities Move on in a Diverse World, Lecture Notes in Mobility. Springer Berlin Heidelberg, Berlin, Heidelberg. doi:10.1007/978-3-642-34735-1 Rodrigues, E., Silva, H., 2009. Moradia é central: inclusão, acesso e direito à cidade, First edit. ed. Intituto Pólis, São Paulo, Brazil. Rodriguez, D.A., Vergel-tovar, E., Camargo, W.F., 2015. Land development impacts of BRT in a sample of stops in Quito and Bogotá. Transp. Policy 1–11. doi:10.1016/j.tranpol.2015.10.002 Rolnik, R., 1997. A cidade e a lei: legislação, política urbana e territórios na cidade de São Paulo. Studio Nobel. Rolnik, R., Klintowitz, D., 2011. (I)Mobilidade na cidade de São Paulo. Estud. Avançados 25, 89– 108. doi:10.1590/S0103-40142011000100007 Saes, A.M., 2009. Luz, leis e livre-concorrência: conflitos em torno das concessões de energia elétrica na cidade de São Paulo no início do século XX. História (São Paulo) 28, 173–234. doi:10.1590/S0101-90742009000200008 Schimek, P., 1996. Household Motor Vehicle Ownership and Use: How Much Does Residential Density Matter? Transp. Res. Rec. 1552, 120–125. doi:10.3141/1552-17 Shoji, K., 2001. Lessons from Japanese Experiences of Roles of Public and Private Sectors in Urban Transport. Japab Railw. Transp. Rev. 29, 12–18. Silva, A.C. e, 2015. Tudo é passageiro: Expansão urbana, transporte público e o extermínio dos bondes em São Paulo, First. ed. Annablume. Silva, C., 2013. Structural accessibility for mobility management. Prog. Plann. 81, 1–49. doi:10.1016/j.progress.2012.07.001 Silva, C., Reis, J.P., Pinho, P., 2014. How urban structure constrains sustainable mobility choices: Comparison of Copenhagen and Oporto. Environ. Plan. B Plan. Des. 41, 211–228. doi:10.1068/b37138 Macroeconomic analysis of public transport competitiveness in megacities 87 Slack, E., Chattopadhyay, R., 2009. Finance and Governance of Capital Cities in Federal Systems. McGill-Queen’s Press - MQUP. Snarr, H.W., 2014. Learning Basic Macroeconomics. A Policy Perspective from Different Schools of Thought. Snell, B., 1974. American Ground Transport: A Proposal for Restructuring the Automobile, Truck, Bus and Rail Industries. Washington, USA. Sorensen, A., 1999. Land Readjustment, Urban Planning and Urban Sprawl in the Tokyo Metropolitan Area 36, 2333–2360. doi:10.1080/0042098992458 Sowell, T., 2011. Economic Facts and Fallacies. Basic Books. Strathman, J.G., Dueker, K.J., Davis, J.S., 1994. Effects of household structure and selected travel characteristics on trip chaining. Transportation (Amst). 21, 23–45. doi:10.1007/BF01119633 Summit, T.L., 2014. Transit Leadership Summit 2012-2014. Susilo, Y.O., Kitamura, R., 2008. Structural changes in commuters’ daily travel: The case of auto and transit commuters in the Osaka metropolitan area of Japan, 1980-2000. Transp. Res. Part A Policy Pract. 42, 95–115. doi:10.1016/j.tra.2007.06.009 Te Brömmelstroet, M. Te, Bertolini, L., 2011. The Role of Transport-Related Models in Urban Planning Practice. Transp. Rev. 31, 139–143. doi:10.1080/01441647.2010.541295 Tomasevicius Filho, E., 2002. A regulação dos transportes coletivos por ônibus no Município de São Paulo. Rev. da Fac. Direito, Univ. São Paulo 97, 287. doi:10.11606/issn.2318-8235.v97i0p287303 Transport for London, 2013. Travel in London: Report 6. London. Tulyasuwan, N., 2013. Private Motorized Transport in Bangkok Thailand. Bangkok, Thailand. Turnovsky, S.J., 1977. Macroeconomic Analysis and Stabilization Policy. CUP Archive. UN, 2014. World Urbanization Prospects: The 2014 Revision, Highlights (ST/ESA/SER.A/352), New York, United. doi:10.4054/DemRes.2005.12.9 UN, 2007. World Urbanization Prospects The 2007 Revision Highlights. New York ESA/P/WP/2, 883. doi:10.2307/2808041 UN, 2006. World Urbanization Prospects: the 2005 Revision: The 2005 Revision. Executive Summary, Fact Sheets, Data Tables. Upchurch, C., Kuby, M., Zoldak, M., Barranda, A., 2004. Using GIS to generate mutually exclusive service areas linking travel on and off a network. J. Transp. Geogr. 12, 23–33. doi:10.1016/j.jtrangeo.2003.10.001 Villaça, F., 2012. Reflexões sobre as cidades brasileiras, 1st ed. Studio Nobel, São Paulo. Wachs, M., 1984. Autos, transit, and the sprawl of Los Angeles: the 1920s. J. - Am. Plan. Assoc. 50, 297–310. doi:10.1080/01944368408976597 Wang, E.C., 2002. Public infrastructure and economic growth: a new approach applied to East Asian economies. J. Policy Model. 24, 411–435. doi:10.1016/S0161-8938(02)00123-0 Wegener, M., Fürst, F., 1999. Land-Use Transport Interaction: State of the Art. Transport 119. doi:10.2139/ssrn.1434678 Macroeconomic analysis of public transport competitiveness in megacities 88 Wismadi, A., Soemardjito, J., Sutomo, H., 2013. Transport Situation in Jakarta. Study energy Improv. Transp. Sect. trough Transp. Improv. smart community Dev. urban area 29–58. doi:http://www.eria.org/RPR_FY2012_No.29_Chapter_4.pdf Wolman, H., Reigeluth, G.A., 1980. Financing Urban Public Transportation: The U.S. and Europe. Transaction Publishers. Wolmar, C., 2009. The Subterranean Railway: How the London Underground was Built and How it Changed the City Forever. Atlantic Books, Limited. World Bank, 2015. World Development Indicators 2015, World Bank. doi:10.1596/978-0-8213-73866 Yang, R., Yan, H., Xiong, W., Liu, T., 2013. The Study of Pedestrian Accessibility to Rail Transit Stations based on KLP Model. Procedia - Soc. Behav. Sci. 96, 714–722. doi:10.1016/j.sbspro.2013.08.082 Yigitcanlar, T., Sipe, N., Evans, R., Pitot, M., 2007. A GIS-based land use and public transport accessibility indexing model. Aust. Plan. 44, 30–37. doi:10.1080/07293682.2007.9982586 Yim, S.H., 2003. Geographical Features of Social Polarization in Seoul, South Korea, in: Mizuuzhi, T. (Ed.), Representing Local Places and Raising Voices from below. Osaka City University, Osaka. Yusuf, S., Nabeshima, K., 2006. Postindustrial East Asian Cities: Innovation for Growth. World Bank Publications. Macroeconomic analysis of public transport competitiveness in megacities 89 Macroeconomic analysis of public transport competitiveness in megacities 96 I.1.5. LONDON Coverage area (Km²) Built-up area ratio (%) 439.06 27.93% LONDON - 1971 Coverage area (Km²) Built-up area ratio (%) 443.50 28.21% LONDON - 1981 Coverage area (Km²) Built-up area ratio (%) 452.77 28.80% LONDON - 1991 Macroeconomic analysis of public transport competitiveness in megacities 97 I.1.6. DELHI Coverage area (Km²) Built-up area ratio (%) 471.81 30.01% LONDON - 2001 Coverage area (Km²) Built-up area ratio (%) 478.78 30.45% LONDON - 2011 Coverage area (Km²) Built-up area ratio (%) 106.90 15.23% DELHI - 2008 Macroeconomic analysis of public transport competitiveness in megacities 98 I.1.7. SÃO PAULO Coverage area (Km²) Built-up area ratio (%) 67.30 6.95% SÃO PAULO - 1978 Coverage area (Km²) Built-up area ratio (%) 101.64 10.50% SÃO PAULO - 1988 Coverage area (Km²) Built-up area ratio (%) 122.95 12.70% SÃO PAULO - 1998 Coverage area (Km²) Built-up area ratio (%) 171.70 17.73% SÃO PAULO - 2008 Macroeconomic analysis of public transport competitiveness in megacities 99 I.1.8. MEXICO CITY Coverage area (Km²) Built-up area ratio (%) 57.15 7.23% MEXICO CITY - 1972 Coverage area (Km²) Built-up area ratio (%) 108.99 13.80% MEXICO CITY - 1983 Coverage area (Km²) Built-up area ratio (%) 174.47 22.08% MEXICO CITY - 1994 Coverage area (Km²) Built-up area ratio (%) 200.52 25.38% MEXICO CITY - 2007 Macroeconomic analysis of public transport competitiveness in megacities 100 I.1.9. SEOUL Coverage area (Km²) Built-up area ratio (%) 206.60 34.15% SEOUL - 1990 Coverage area (Km²) Built-up area ratio (%) 247.70 40.94% SEOUL - 1996 Coverage area (Km²) Built-up area ratio (%) 354.38 58.58% SEOUL - 2002 Coverage area (Km²) Built-up area ratio (%) 376.80 62.28% SEOUL - 2010