Análise do impacto de faixas de circulação exclusiva ao transporte coletivo em Fortaleza usando big data

Abstract

Public Transport is an essential mode of transport for some parts of the society, since other modes are not accessible to them. For it to become more efficient, public policies that prioritize it are necessary, promoting sustainable development and increasing its efficiency and effectiveness. One of the measures promoted by the National Urban Mobility Policy is the implementation of exclusive bus lanes, providing an increase in average speed, a decrease in travel time, a reduction in fuel consumption and a reduction in operational costs and pollution. However, these gains announced by transport agencies in operational indicators generally does not follow a rigorous method of evaluating public policies, the so-called methods of Causal Inference. Several factors can interfere in the evolution of each indicator, causing bias in these ex-ante and ex-post measurements. As an opportunity, a rich historical series of GPS, electronic ticketing and electronic surveillance data is available which can be used in a difference in differences causal inference approach, as it generally require a large amount of panel data. Thus, this work proposes to use this data, called the Big Data of Public Transportation with the aim of analyzing the real impact of the implementation of exclusive bus lanes in Fortaleza on operational performance indicators. To elaborate this method, we made a literature review on the Prioritization of Public Transport along with methods of Causal Inference and the Big Data of Public Transport. We divide the proposed method into five steps. In the first, we propose an integration and consolidation of the Fortaleza’s ticketing and GPS databases in order to make them suitable for the calculation of indicators. In the second step, we propose a method of defining a treatment and counterfactual group for the case of exclusive bands in Fortaleza. In the third step, we define the performance indicators we used and how to calculate them using the GPS data. In the fourth step, we propose a Difference in Differences approach and finally, we propose a DiD model with control for the volume and we discuss limitations of the proposed models. As a result, we found that the impact of the exclusive bus lanes on the indicators was well below that announced by the City Hall of Fortaleza, although statistically significant. In addition, we discuss the limitations encountered during this work and we give recommendations for future works.