Cities in developing Asia are growing at breakneck speed, and transport infrastructure is struggling to keep up with demand. One of the main challenges arising from rapid urban growth in the region is severe traffic congestion, which has become a major concern for urban transport practitioners as well as city residents.
But before we take action on traffic, we should look into what causes congestion, and how costly it is to cities. Surprisingly, little scientific research has been conducted on these topics by urban or transport scholars, and even less by economists. Among other factors, a key challenge is lack of good data to systematically measure people’s day-to-day travel.
In a recent Asian Development Policies Lecture hosted by ADB, Prof. Gilles Duranton from the University of Pennsylvania’s Wharton School shared a series of recent innovative studies on urban transportation that use information on real-time simulated trips gathered from Google Maps.
For each trip, the data contains day, time of day, origin, destination, length, total duration, and time spent in absence of traffic. Though they are not actual trips, they offer a good semblance of actual activity if observed closely enough.
Mobility, accessibility and city characteristics
The research proposed that mobility (the speed at which one can travel from origin to destination) and accessibility (the ease of reaching a destination) should not be considered separately. If we measure accessibility as the total time needed to travel, it is in fact a sum of distance and mobility. Mobility can be additionally broken down into mobility without congestion (free mobility) and congestion.
After gathering data from 22 million simulated Google Map trips in 154 large Indian cities, Prof. Duranton and his collaborators examined how mobility and accessibility are determined, and how city characteristics—such as population, roads and vehicle ownership—affect them.
Their research found an enormous variation in mobility and accessibility across Indian cities. The fastest city in terms of measured mobility was twice as fast as the slowest. This gap is much larger than among US cities, and the differences in accessibility and distance are even more pronounced.
The amount of free mobility explains 70% of the variance in overall mobility. This suggests that a city’s overall mobility is largely related to its urban characteristics such as road capacity that determine the travel speed in absence of congestion. The amount of traffic during peak hours appears to be a less important factor.
Distance explains 81% of variance in accessibility, demonstrating that living close to convenience stores, supermarkets or train stations is crucial for urban dwellers.
The analysis further shows that large population size lowers free mobility and increases congestion, but improves distance accessibility and has little overall effect on time accessibility. The amount of primary roads improves free mobility and distance accessibility, but barely affects congestion.
The impact of vehicle ownership on accessibility is significant: vehicles are associated with better free mobility, worse congestion, and longer distance of travel. In other words, more vehicles reduce overall accessibility.
The social cost of congestion
Most of us feel it is painful to commute in a congested city during peak hours, and wish we could go as fast as before (when cities were less crowded) or at least as fast as during non-peak hours. However, economic theory and evidence tell us this is unrealistic, as a certain degree of congestion in cities is unavoidable. The reason is that people travel more, and more people travel when road capacity increases. In other words, congestion is to some extent our own doing.
The above notion is closely related to how we measure congestion cost. The cost of congestion is not about the amount of time wasted between the trip’s actual duration and the duration under free traffic flow, simply because free flowing traffic at peak hours is not possible. The cost of congestion should rather be calculated in terms of the social cost – the longer travel time imposed on other drivers by one’s own decision to drive. In economics, this is referred to as deadweight loss.
Supply and demand of travel
The diagram above illustrates how we can calculate the deadweight loss. The horizontal axis is the volume of traffic, while the vertical axis represents the time cost of travel per unit of distance. To estimate the deadweight loss, we compare the demand (propensity to travel given the time cost of travel per unit of distance, shown on the downward sloping demand curve) and the supply, or time cost of travel given the number of travelers on the road, which yields the upward sloping curves showing the average cost and marginal cost of traffic
The deadweight loss—resulting from the wedge between the average cost to travelers and the marginal cost to society—is equal to the shaded area in the diagram.
Small side roads help ease congestion
Using data from a detailed travel survey and counterfactual trip data from Google Maps, Prof. Duranton estimated the deadweight loss due to congestion at Bogota in Colombia, a large and highly congested city.
The results show that the demand for urban travel is quite elastic, or sensitive to change in travel cost (-1.2 to -1.8 being equivalent to a 12%-to-18% drop in number of travelers with 10% increase in the time cost). But the elasticity of time cost of travel per unit of distance with respect to the number of travelers is only 0.06 on average (a 6% increase in time cost in response to 100% traveler increase), and a maximum of about 0.2.
The latter can be explained by smaller side roads that help divert Bogota traffic and thus reduce congestion. According to these estimates, the maximum deadweight loss is equal to a small fraction of travel time, which can be valued at less than 1% of the traveler’s daily wage.
While the results may sound counter-intuitive to many of us and are subject to further validation and debate, the methodology is well grounded in economic theory, and thus can be applied consistently to different cities.
Also, the estimated small social cost of congestion does not mean that policies to curb traffic are not beneficial. They still can bring tremendous gains to cities in the form of increased mobility and amount of travel.
Clearly, there is great scope for further academic research on urban transportation using these innovative new techniques.