Commute mode choice in the city of São Paulo: an empirical analysis

• Ano de defesa: 2017
• Autor(a) principal: Anchante, Jayme Tolpolar
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Transporte - Trânsito de passageiros; Transportes - Planejamento; Transportes - Aspectos sociais; Economia
• Link de acesso: http://www.locus.ufv.br/handle/123456789/13090

The present works deals with the urban mobility in the city of São Paulo, Brazil, specifically the commute mode choice process. Our aim is to analyze how travel time and travel cost, as well as the commuter characteristics, are associated with the probability of choice of a certain mode of transport. Our main hypothesis is that both time and cost are negatively associated with the choice of any mode. Our method is the Alternative-Specific Conditional Logit, a discrete choice econometric model that enables the inclusion of alternative-specific and individual characteristic variables, which has as main assumption the Independence of the Irrelevant Alternative (IIA): the probability ratio of two alternatives depends solely on the characteristics of the these alternatives. We also employed the Nested logit model, which does not relay on the IIA assumption, because it nests similar alternatives. Our main data sources are the microdata from the 2007 and 2012 Origin Destination Surveys conducted by the Companhia do Metropolitana de São Paulo; besides that, we also used monthly gasoline prices from the Agência Nacional do Petróleo, Gás Natural e Biocombustíveis and vehicle mileage from the Instituto Nacional de Metrologia, Qualidade e Tecnologia. The Origin Destination Surveys have information about the household and the individual demographic and economic characteristics; the position of the household, workplace and school of the interviewees; and information about the previous workday travels: weekday, origin/destination zone, purpose at origin/destination, mode of transport, hour and minute of departure/arrival. The dependent variable, the choice of the commute mode of transport, has three alternatives: transit, private motorized transport and non- motorized modes. The independent variables are travel time and cost (alternative- specific characteristics), age, sex, if study, employment relationship and degree of education (characteristics of the individual). The results corroborate our initial hypothesis that travel time and cost are negatively correlated with mode choice, besides that, the Conditional logit revealed a Value of Travel Time of 1.78 and 1.09 (nominal R$) for 2007 and 2012, respectively (the Nested logit revealed a Value of Time between R$ 0.04 and R$ 2.47): this is the value the average citizen of São Paulo is willing to pay to save one hour of commute travel time. The average marginal effect shows both the own-elasticity and the cross-elasticity of a mode’s cost and time, with these information it is possible evaluate a series of transport policies that raise or reduce the mode’s cost or time in order to discourage or encourage it. We perform several tests for the IIA assumption and we get mixed results: some reject it and some do not, depending on the model chosen and how we nest the alternatives. We conclude that it is possible to change the behavioral pattern of choosing the commute mode of transport by changing the travel cost and time, grounding possible transport policies that tax or improve the speed flow of certain modes relative to others.