When armchair urbanists talk about why expanding highways will never work to solve transit, which is true, they claim the reason is due to “Induced demand” and that the highway will simply fill up with more cars, making the point moot.
But they then claim that the solution to housing prices is to build more housing, which is true.
The problem is that in both of these cases, induced demand is at play, and at the same level.
Either induced demand is the reason why highway construction does not work and densifying cities is pointless in terms of affordability, or there is something else going on.
Now it is true that at the scale of transportation demand and the necessity to commute by car in most of the United States, the amount of people who want to travel between points A and B are larger than the total capacity of basically any highway, even the Katy freeway in Houston, Texas which still gets congested despite its ridiculous size.
Leaving armchair urbanism and entering the realm of peer-reviewed research, it becomes clear that a main reason for traffic is not “induced demand” but in fact lane changes. https://www.sciencedirect.com/science/article/abs/pii/S0378437123006155#:~:text=Such%20decisions%20can%20reduce%20the,to%20traffic%20congestion%20and%20accidents
This presents the real reason why two lane highways are often faster than wide interstate highways. As lane changes increase, people slow down to merge, making the people behind them slow down, leading to more and more congestion until the highway is turned to gridlock. It’s a combination of people driving below the speed limit and changing lanes which leads to congestion in actual scientific research. More lanes means more merging. More merging means slower speeds.
Not induced demand.
It is true that if every person in every car kept going at the same speed, there would be no traffic. But in reality, that is impossible.
Now, there are ways to reduce traffic. Fewer cars on the roads means fewer opportunities for people to change lanes. Fewer lane changes means less traffic.
So how do we end up with fewer cars on the road? There are several options, divided into supply side and demand side solutions.
Demand side solutions:
- Congestion charge. This will lead to fewer people choosing to drive. However, its impact on traffic is proportional to the availability of substitutes.
- Car tax. Fewer people driving means fewer cars on the road. However, like a congestion charge, its impact on traffic is proportional to the availability of substitutes.
- Substitution effect. Offer people substitutes to reduce the demand for the road which can include:
- Living close to the office. As long as everyone in the house works in the same neighborhood and never changes jobs this can work great!
- Bike trails. As long as everyone lives close enough to their offices that everyone can bike to work every day of the year this can work great!
- Mass transit. It actually works.
Supply side solutions:
- Narrow the road. Fewer lanes, fewer merges, fewer problems.
Substitutes will be most functional as long as the substitute for driving is faster than driving. Once they reach equity they will stop making any significant impact on congestion. This means that if you invest in a substitute you should make sure it is better than the available alternatives. The better the substitute, the more demand it will attract.
The stronger argument
There is a much stronger argument against widening highways than misapplied basic economic theory.
Mixed traffic, highways or surface streets, carry far fewer people than any other mode of transportation.
Induced demand is a weak argument. The better argument is a simple cost/benefit ratio.
According to Transformative Mobility and other sources, highways carry 10% of the capacity of light rail, and 3% of the capacity of heavy rail.
To estimate the cost of building a highway, let’s use a real life example. The Washington State Department of Transportation is adding lanes to I-405 south of Bellevue. This is a 40 mile long project and it is going to cost $705 million, or over $17 million per mile. This is on par with the costs found on this website, which estimates $20 million per mile, or $32 million per kilometer, for a new highway.
The West Seattle Link which is a 4 mile long project is estimated to cost around $1.5 billion or $275 million per mile. Transit Costs Project puts it even higher at $649.50 million per kilometer. New York is even worse with the subway East Side Access expected to cost almost $4 billion per kilometer.
The San Francisco Central Subway is expected to cost $708 million per kilometer.
This is insane.
Eric Goldwyn is the expert on this topic. Explore the Transit Costs Project for details.
Now when we keep capacity in perspective, the appropriate costs for the following modes would be the following, if we make the inane assumption that $20 million per mile for a new highway is reasonable.
At $20 million per mile for a new highway the cost for a new highway is $10,000 per mile per passenger per hour. If we keep that cost constant we get the following table for each of these modes.
| Mode | Passengers per hour | Cost per kilometer | Cost per mile |
| Highway | 2000 | $32,000,000.00 | $20,000,000.00 |
| Regular bus | 5000 | $80,000,000.00 | $50,000,000.00 |
| BRT single lane | 9000 | $144,000,000.00 | $90,000,000.00 |
| Cyclists | 12000 | $192,000,000.00 | $120,000,000.00 |
| Pedestrians | 15000 | $240,000,000.00 | $150,000,000.00 |
| Light rail | 20000 | $320,000,000.00 | $200,000,000.00 |
| BRT double lane | 43000 | $688,000,000.00 | $430,000,000.00 |
| Heavy rail | 60000 | $960,000,000.00 | $600,000,000.00 |
| Suburban rail | 90000 | $1,440,000,000.00 | $900,000,000.00 |
These are the costs per mile which keep the cost on par compared to the already inflated cost of building a highway. A simple rule of thumb (whether it is good or not) is if the cost is higher than this, its a bad deal, but if it is lower than this, it is a good deal.
But no matter how you look at it, American costs for some transit projects are too high to justify the costs from a homo economicus perspective.
This is the insane thing.
I don’t care about the money as an end to itself. But if we were able to cut the cost of building transit in half, we could then theoretically double the amount of transit we are building. It also makes the case of building transit much weaker as opposed to highway expansion. At prices like these, highway expansion is sometimes the more economical option for increasing transit modal share!
It is absolutely insane that even though light rail is capable of carrying ten times the number of people as a highway, it can cost 20 times the cost of building a new highway in our current regulatory environment.
Heavy rail like the subway carries 30x the capacity of a highway yet costs 125x the cost of building a new highway.
Transportation costs are too damn high and just shoveling more money into the system will not fix it.
We need to bring costs down by making construction more efficient.
The argument I wanted to make was to say that expanding highways is not economical because they are low capacity and that it would be more cost-effective to just build rail rather than a new highway.
This is true in many cases, even in the United States. There are good projects in the United States which fit in the benchmarks in the table above. Basically every heavy rail project outside of New York city fits under this benchmark. These include:
- The San Francisco Central Subway: $708 million per km
- Los Angeles Purple Phase 3: $857 million
- Los Angeles Regional Connector: $611 million
- Los Angeles Purple Phase 1: $479 million
- Vancouver Broadway: $450 million
- Boston Green Line extension: $315.3 million
- Washington Silver Line Phase 2: $151
- Miami Metrorail extension to MIA: $129 million
But there are also projects which do not meet the benchmark:
- Seattle projects
- New York City subway projects
- Honolulu HART
It’s really just these three cities which are blowing their budgets above the baseline for matching the cost of building a highway in terms of capacity, at least from projects on the Transit Costs Project Database.
One issue I keep thinking is if we changed how bidding worked. Open bidding to foreign contractors. Each contractor sets a bid and proposal which is then accepted by the transit agency and/or city based on its costs and benefits. The contractor then must deliver the project as proposed at the costs laid out. If they underestimate the costs, they cover the difference. If they overestimate the costs, they are unlikely to get the bid. This bidding process in a free market will drive prices down, encourage innovation, and force contractors to stick to their word. If they don’t survey properly, don’t get things done on time, or other issues, they need full liability for that. Private contractors need to take the liability or the construction needs to be done in house.
I think that will be very effective in driving costs down.
But the funny thing is that even with outrageous spending costs, the cost of building transit is still cheaper than widening a highway outside of Seattle, Honolulu, and New York.
This is a factual argument based on real-life data which should be able to convince basically any politician regardless of their political affiliation.
It’s very simple. We should stop building highways because they are just not worth the money. It is usually cheaper to build rail, even in the United States.