Urban congestion, more commonly known as traffic, is probably the most painful daily experience for people living in modern cities. It costs our time, ruins our mood, and harms our health. From an economic perspective, traffic increases costs of our travel within a city and thus restricts the areas we can reach for jobs and amenities. Therefore, it counteracts the higher productivity and quality of life that cities are anticipated to offer. However, there seems no easy solution to it and situations have not improved much, and have become worse, in many places over time.

To tackle the problem, we first need a better understanding of it. How to quantify urban congestion, what are the city attributes that give rise to it, and what are options to address it?

Congestion is typically considered as traffic volume exceeding the road capacity. However, it is largely a location-specific measurement and no good data is available so far to allow for a cross-city comparison. Alternatively, it can be measured as additional time to travel during peak hours as opposed to travel time in non-peak hours. With access to big, real-time travel data, in platforms like Google Maps, economists have developed a novel approach to construct congestion indicators across many cities in India. In the recently released report Fostering Growth and Inclusion in Asia’s Cities, the theme chapter of  Asian Development Outlook 2019 Update, we adopted a similar approach to measure traffic levels and correlate them with several city attributes across 278 Asian cities, each with an estimated population greater than a half million in 2016.

The data show that Metro Manila, Kuala Lumpur and Yangon City rank as the most congested cities in the 25 developing countries in Asia that were part of our research. The ratio of driving durations in rush hours to off-peak hours, averaged across 400 randomly selected routes, ranges from 1.7 to 1.9 in these cities. It means that drivers need to spend almost 100% more time during peak hours in the most congested city, Metro Manila.

For some travelers, these figures may still underestimate their daily experience. Although the ratio for some routes in Manila is as high as 3.1, there could be different explanations for this gap. First, our estimates are averages based on a large number of routes while individuals usually only experience a very small fraction of them. Second, the off-peak travel duration could be long in some routes, which also suggests bad traffic but could lower the congestion level. Third, our sample is not fully representative as trips on 400 routes on a single workday may not be a large sample for a metropolitan area like Metro Manila.

In any case, the results do suggest severe traffic in a number of Asian cities. We then relate it to some city attributes using regression analysis. This shows us that more densely populated cities are more congested. This may not be surprising. If population indicates demand for urban travel while area indicate supply, congestion reflects exactly the tension between demand and supply for urban travel. Second, we use average luminosity level to proxy for economic activity of a city. It turns out not to be related to congestion. To the extent that luminosity represents the richness of a city reasonably well, congestion does not seem to get alleviated if a city has more resources to use. In short, prosperous cities suffer traffic much like poor cities.

An unexpected result is that the amount of roads, regardless of measured as absolute length, length per capita or density, is positively correlated with congestion in a city. This seems counter intuitive although our intuition is not always correct.

  Neither rich or nor poor cities are immune from traffic and urban congestion.

Does a city with more roads have less traffic? Not necessarily. In economics, there is a so-called fundamental law of road congestion which states that with the increase in road capacity due to the construction of new roads or public transit infrastructure, people will switch to or increase driving such that the congestion level is pushed back to where it was. Some studies validate the law in reality, while others show it holds under certain conditions. The fundamental law of road congestion implies that road capacity and traffic is likely unrelated, but it does not explain why they would be positively related. A plausible explanation is that cities with severe congestion have invested more than their peers in roads, but just not enough yet. In this case, we see traffic levels and road availability move the same way across cities. 

A couple of caveats are warranted regarding the above analysis. First, the model leaves out many important factors due to limitations in data availability. For instance, we know car ownership has been rising significantly in the region, which must contribute to congestion. Unfortunately, we do not have information on car ownership at the city level so that it can be included in the model. Public transit systems are another factor too important to miss. We show in the report that efficiency, coverage and affordability, remain a challenge for many cities in Asia. Second, the regressions do not tell the causal relationship between the city attributes and traffic levels. While causal relationship is more relevant for decision making, it is much harder to get. In our case, the correlations shown do carry some novel and useful information while more work is needed to draw causality of factors influencing congestion.

To be sure, we cannot wait until we understand every cause before we address the region’s traffic problems. We need to try whatever looks right and draw lessons from these attempts. There is a long list of things a city could do to mitigate congestion. On top of the list are increasing road capacity, expanding public transit services and developing a multimodal transport system that encompasses motor-based and non-motor-based travel as well as formal and less formal transport services. Some believe addressing fragmented urban governance including land use planning, urban infrastructure development, and traffic management is crucial, especially when the metropolitan area involves several local governments. Cities in developing economies also need aid to strengthen their capacity in urban governance.

Supply-side measures, i.e. those focused on increasing the supply of the transport system or improving its efficiency, are not enough. Essentially, traffic increases because drivers do not take into account the cost they bring to other drivers, which economists call negative externality.  As the fundamental law of road congestion suggests, the efficacy of supply-side measures will eventually be offset by induced demand for automobile travel. To substantively reduce congestion in the long run, demand-management measures are imperative.

Governments should control the growth of car ownership through a quota system. Some people’s privilege to buy a car must be delayed. To be fair, the existing owners should be charged an additional annual fee for owning a vehicle. Coding schemes that allow driving on four out of five weekdays per week seem inadequate. Though politically unpopular, pricing measures such as stiffer parking tariffs, gasoline taxes, and congestion fees may be indispensable. When weighing the benefits and costs of these interventions, note they will mitigate air pollution at the same time that they reduce traffic. Other options include promote commuting arrangements such as work-from-home and flexible working hours. However, they may only improve the situation marginally and are better used as complements to the quota system and pricing measures.  

Finally, congestion is a textbook case of market failure due to externalities. It is imperative that governments play a pivotal role in fighting congestion. For many emerging cities, it is an uphill battle that demands resolution and capacity of the government.

This article is based on the findings of the ADB report  Asian Development Outlook 2019 Update: Fostering Growth and Inclusion in Asia’s Cities Update.