Analysis of BRT systems and its implications on physical activity using a hurdle model statistical approach

• BRT’s implementation has a positive impact on the physical activity of the population living in its catchment area. • The benefits of the implementation have effects on different sectors of the population, according to their geographical area. • Cragg hurdle regression solves the problem of respondents reporting zero minutes of walking per week without considering them as outliers. Past research has explored the relationship between physical activity and public transit availability. Evidence suggests an increase in physical activity linked to public transit; however, research connecting physical activity to changes in transportation is scarce. Many studies focus on BRT systems, which are popular among emerging economies. Using data from Rio de Janeiro’s TransCarioca and Mexico City’s Metrobus, we investigate the effects of a new BRT service on the physical activity of catchment area residents. One problem related to traffic project evaluation research is that interviewed subjects in pre- and post-periods are different; hence, some research groups have applied propensity score matching techniques to reduce bias and achieve systematic concordance between treatment and control groups. Another limitation of this methodology is that it does not consider zero-inflated responses, thus lacking accuracy while estimating the average treatment effect. We present a novel approach where we conduct a matching process using the population’s sociodemographic variables —e.g. gender, age, and marital status— to build the after-mentioned groups, and then we use a hurdle statistical model in which the two processes generating the zeros and the positives are not constrained to be the same. Using the responses to over 8000 IPAQ questionnaires applied in Rio de Janeiro in 2011 and 2015, and in Mexico City in 2011 and 2014, after the implementation of BRT systems, we analyzed the time residents in the catchment area spent walking for utilitarian and recreational purposes using a Cragg double-hurdle regression model. Preliminary results show a statistically significant effect of Mexico City’s Metrobus system implementation on physical activity. At the same time, Rio de Janeiro experienced an increase in the same area after the TransCarioca system implementation, although the transport-mode change is statistically insignificant.