Why Expanding Roads Fails to Reduce Traffic Congestion

Increasing lanes sounds like it should reduce traffic congestion – research shows it doesn't.

Nominal News is an economics newsletter written by a PhD Economist that translates the latest economic research into clear, policy‑relevant insights on current issues. Join 4,000 readers to stay-up-to-date with Nominal News directly in your inbox:

---

Recently, The Georgia Department of Transport posted on their social media account the following statement with the below picture:

The widening of I-95 near the Savannah River will improve travel for thousands of drivers each day while supporting the continued growth of the Port of Savannah –one of the fastest-growing ports in the nation. A smarter, safer corridor is on the way. https://bit.ly/4ux3vjW

The picture suggests that widening the road by 1 lane will significantly reduce congestion and improve commute times. Economics research tells us that this extra lane will have no impact on congestion whatsoever.

The Fundamental Law of Road Congestion

Economist Anthony Downs, in 1962, posited the Law of Peak-Hour Traffic Congestion:

This Law states that on urban commuter expressways, peak-hour traffic congestion rises to meet maximum capacity.

Over time this law was rephrased as the Fundamental Law of Road Congestion. Economists have often studied this law to see if it actually holds, and research has shown this to be the case.

Evidence from US Cities

Duranton and Turner (2011) investigated whether the Fundamental Law of Road Congestion holds in the context of US cities. The main question Duranton and Turner were interested in answering is what impact would changing the number of lane kilometers (i.e. increasing the number of lanes on road) have on vehicle-kilometers traveled (VKT). VKT is a measure of the total number of kilometers traveled by vehicles on a given highway system. If one vehicle travels 1 kilometer, that would be considered to be 1 VKT.

Suppose a one lane highway is 100 kilometers and 10 cars travel it in its entirety. This would be a road with 100 lane kilometers (1 lane times 100 kilometers) and VKT would be 1000 (10 cars times 100 kilometers). If we add an extra lane to this highway, the overall distance does not change, so 10 cars would still travel 100 kilometers, resulting in 1000 VKT. Lane kilometers would double since we have two lanes (2 times 100 kilometers). If VKT remains at 1000, while lanes doubled, then we would consider this road to be less congested. On the other hand, if the extra lane results in 20 cars on the road (2000 VKT), then the additional lane will have no impact on congestion.

Data and Results

To estimate the impact of adding lanes, Duranton and Turner looked at 228 metropolitan areas in the US over a 30 year time period, between 1983 to 2003. The number of vehicles on roads in the US is collected quite accurately – for example, states must provide accurate daily interstate highway usage to the federal government.

Overall, VKT in the US doubled from 1983 to 2003, from 7,700 VKT to 15,900 VKT for interstate highways. For major urban roads, a similar pattern occurred – VKT went up from 15,000 to 30,000. In the same time span, the number of lane kilometers of interstate highways remained basically constant, while on major urban roads, it went up from 3,800km to 6,500km.

It is important to note that in order to determine the impact on VKT of adding additional roads (causality), we have to ensure we are not capturing other factors. For example, overall population growth will naturally result in higher VKT and it may seem like adding roads has no impact on congestion. Duranton and Turner do control for many factors and even use an innovative approach called “Instrumental Variables regression”.

So what is the impact of increasing the number of lane kilometers on VKT? Duranton and Turner estimate that increasing lane kilometers by 1% will increase VKT by 1.03%. Economists call this an elasticity¹ of 1.03. Similarly, a 10% change in lane kilometers will increase VKT by, approximately, 1.03 times 10%, so 10.3%.

The Duranton and Turner result tells us that any feasible increase in roadways will have no impact on congestion – any increase in roads will be met by more vehicles on the road, more than offsetting the increase in roads.

Where Do The Extra Vehicles Come From

Duranton and Turner were also interested in figuring out what is the source of extra vehicles. They looked at four potential sources:

• Additional commercial traffic;

• People and economic activity migrating to places with additional roads;

• Diversion of traffic from smaller, local roads;

• Changes in individual behavior with people choosing to drive more.

Duranton and Turner studied how a hypothetical 10% increase in lane kilometers in interstate highways, that increases VKT by 10.3%, would be apportioned between the above 4 causes.

Duranton and Turner found that a 10% increase in interstate highways results in a 10 to 20% increase in truck road usage. As trucks account for approximately 13% of all vehicle traffic, increases in commercial truck traffic explains about 20% to 30% (thus, 2%-3% of the 10.3% increase) increase in VKT due to increased interstate highways.

Individual behavior is also a major contributor to increasing VKT. A 10% increase in interstate highways results in about a 1% increase in road usage by each individual driver annually. Although this may not sound like much, the largest source of VKT on the roads are individual drivers (87% of all VKT). Thus, a minor increase in car usage by each person, results in a significant increase of cars on the road. Individual behavior changes account for approximately 9% to 39% (1%-4% of the total VKT increase) of the increase in VKT due to additional interstate highways.

The other two reasons appear to have much less impact on congestion. Population growth from migration is minimal – a 10% increase in interstate lane kilometers increases population in the regions around the interstate by 2% over 20 years and can explain only around 5% to 20% of the increase in VKT. Diversion from other smaller local roads is almost non-existent – only 0% and 10% of the increase in VKT.

These results suggest that construction of additional roads have limited positive impacts – it does not reduce congestion on any roads nor does it have any meaningful increase on population growth, which would translate to economic development.

The increase in commercial vehicles could be a positive economic sign (more goods being transported) but it depends whether this increased truck traffic is driven by increased economic activity or a substitution from other forms of goods transportation like trains.

The main cause for increasing congestion when new lanes are built is that individual drivers choose to drive more. This fact has been termed often as “induced” or “latent” demand. Any time a new road is built or an old one is expanded, it quickly ends up getting congested, without any relief of congestion on other roads. However, the people undertaking these ‘new trips’ must have been doing something before the road expansion. They either used an alternative mode of transport or did not make this trip. This ‘pent-up’ demand keeps congestion constant, as any road expansion will immediately be filled up by cars.

Interestingly, this also means that some solutions that might work – won’t. Duranton and Turner looked at whether increasing public transit can reduce congestion. By studying the provision of buses, Duranton and Turner found that public transit has no impact on VKT. This is consistent with induced demand. Even if people shift some of their car trips to public transit, there are ‘new trips’ ready to take their place.

Duranton and Turner also found that, over-time, regions slowly converge to having the same average daily number of vehicles per lane. This suggests that there is a ‘constant’ level of congestion all roads tend to. People appear to migrate out of regions that have higher than average congestion levels, to areas that are lower, resulting in the convergence of congestion.

Solving Traffic

Research shows that expanding roads will not alleviate congestion. So what will? The only solution demonstrated to have had impacts on congestion is congestion pricing. Congestion pricing internalizes the externality² caused by using a vehicle. The externalities caused by car usage include increased travel times, increased air and noise pollution, as well as increased probability of accidents. These costs are borne by both drivers and non-drivers. By putting a monetary cost on the travel, we internalize this cost, as the driver now pays for the costs that car usage entails. Research shows congestion pricing does reduce car usage – a holistic overview by Lehe (2019) has shown that congestion pricing does reduce the number of cars (depending on the implementation) and that people are very responsive to the additional congestion cost.

In the US, New York City implemented congestion pricing last year, and, as expected, the benefits have been significant, including a fall in commute times and pollution.

If you would like to support us in reaching our subscriber goal of 7,000 subscribers, please consider sharing this article and pressing the like❤️ button at top or bottom of this article!

---

Interesting Reads from the Week

1. Even Before the Iran War, There Was a Growing Inflation Problem – Mike Konczal goes over the inflation picture in the US, and why inflation had been rising even before the Iran war.

2. In view of the above post, March inflation, as being forecast by the Cleveland Federal Reserve, may be very high – at 0.62 percent (normally it should be at 0.15%)

   

3. The Dirty Secret of Election Models: Carl Allen explains why judging election models based on hindsight (i.e. after the result) isn’t helpful for picking good models. What matters more is the transparency and validity of assumptions:

All models have assumptions. There’s no such thing as a “purely data-driven” model – only models whose assumptions are obvious…

The difference between a good model and a bad one isn’t whether assumptions exist. It’s whether those assumptions are valid, explicit, and aligned with the real-world process being modeled.

1. Elasticity is a measure that tells how a percentage change in one variable impacts another variable.

2. Externalities are indirect costs or benefits that impact uninvolved parties caused by activities of another party (pollution impacting people’s health caused by factories and traffic congestion from people driving cars are examples of negative externalities; vaccinating oneself has positive externalities as it reduces the likelihood of others getting sick). Since these costs and benefits are not taken into account by private individuals when making decisions (i.e. they don’t ‘internalize’ them), some activities happen too much (pollution), while others not enough from the perspective of society.