Peter Christensen, University of Illinois

Transit subsidies are often justified as a second-best instrument for automobile externalities, though most of available empirical evidence exists in developed country cities and relies heavily upon structural assumptions and traffic simulation.  This study develops a method for estimating of the impact of transit subsidies and other congestion abatement policies in developing country cities by combining observed trips from a transit survey with Google’s real-time traffic data to construct counterfactual trips for consumers in a transport market.  We use these constructed choice sets in a discrete choice framework to study the welfare effects of transit fare price changes as well as effects on traffic congestion.  We compare these estimates with ex post transit demand elasticities estimated using a natural experiment that makes use of sudden hikes in transit fares and daily ridership payment data for every bus route in the city.  Our results indicate that the elasticity of demand for public transit in São Paulo is between -.33 and -.4 and that these two methods yield consistent estimates.  Ex post estimates indicate that a $.50R hike in the public transit fare results in a 7.2% decline in transit ridership.  We find that a hike in São Paulo’s public transit fare results in a $3.79M BRL loss of consumer surplus per day, disproportionately affects low income individuals, and shifts 15,000-20,000 passengers per hour to private vehicles during peak hours.