Does public transit reduce car travel externalities? Quasi-natural experiments' evidence from transit strikes

The provision of public transit is thought to reduce travel time losses that are due to car congestion. For this reason, it is economically justified to subsidise public transit from a welfare perspective as it creates a congestion-relief benefit. The main goal of this paper is to quantify the congestion-relief benefit of public transit for Rotterdam by analysing car speed during public transit strikes. Arguably, strikes can be interpreted as exogenous transit supply shocks and therefore as a quasi-natural experiment. We are aware of two other papers that use the same methodology. Lo and Hall (2006) and, more recently, Anderson (2014) both analyse the effect of (the same) single transit strike lasting 35 days on highway speed for Los Angeles. Anderson (2014) finds a substantial congestion relief benefit with a decrease in time delays experienced by car drivers of 0.12 minutes per kilometer traveled. It is unknown to what extent this result can be generalised to cities where the share of public transit use is much higher or to cities where bicycle use might be a viable alternative. Our analysis differs from Anderson (2014) and Lo and Hall (2006) in a number of ways. First, we focus on a city, Rotterdam in the Netherlands, which, as we will document, is only mildly congested. Second, we analyse the effect of multiple strikes of various public transit modes (e.g. bus, light rail) that are citywide. Third, we examine the strike effect on car speed (and flow) both for the highway ring road as well as within the inner-city of Rotterdam. Finally, by examining heterogeneity in the effects of strikes, we are able to improve our understanding when the public transit relief benefit is particularly pronounced. For example, as one may expect, we find a particularly strong effect of strikes on car speed during weekday rush hours (but no clear effect during weekends and outside rush hours). In addition, our results suggest that the speed effects of strikes that last a few hours are similar to full-day strikes indicating that a continuous supply of public transit during the day is essential for travelers. We show that the congestion relief impact in Rotterdam is by a factor ten larger for inner city roads than for highway ring roads. For the latter we found a several times smaller effect than Anderson (2014). It turns out that the congestion relief benefit of public transit for Rotterdam is substantial, and about 50% of the current subsidy level. This result is particularly noteworthy as we focus on what can be considered an uncongested city. This suggests that subsidies to public transit are welfare improving, even for cities that exhibit low congestion levels.