Sunday, November 2, 2014

Induced demand is a real problem and it makes congestion relief nearly impossible

For some reason, contrarian articles on "induced demand" seem to be the topic du jour. While they are right to point out that induced demand is a more subtle and nuanced concept than the way it is sometimes popularly handled, both articles seem to err too far in the other direction. Induced demand is a real problem, even if some people sometimes misuse it when making arguments.

Reliability, frequency, and speed


I want to take a quick detour to talk about goals in transportation investment. People tend to make decisions about transportation based on many factors, which I will boil down into three: reliability, frequency, and speed. Reliability encompasses factors such as predictability of trip time, probability of vehicle breakdown, and in general, the ability to make a confident statement about your location at a future time. Frequency is a metric of opportunity: the number of chances that will you have to undertake your trip in the window of time that you have available. And speed is the cost of the trip in time. To be sure, there is some crossover: frequency affects average speed and predictability, for instance. But let's roll with this for now.

There are two major categories: scheduled transportation and unscheduled transportation. Scheduled transportation is largely mass transit, whether public or private. Unscheduled transportation includes modes such as walking, biking and most driving. Frequency is obviously a huge deal for transit; there is also a frequency limitation on unscheduled transportation caused by things like traffic signals, but it is usually small enough to go without notice.

For many people, reliability is the most important characteristic. Walking and biking tend to have the most reliability because they are the least complex modes. Transit with a dedicated right-of-way is also supposed to be highly reliable: a properly run railroad should operate like clockwork (the MBTA on the other hand...). Modes such as driving are very susceptible to being caught in unpredictable and chaotic traffic congestion because they are highly inefficient in terms of space. Cars are also complex pieces of machinery that can easily break down. And mixed-traffic transit gets the worst of it because it has to contend with traffic congestion as well as passenger loading.

Speed should be broken down into average and top speed. Walking is obviously the slowest in average and top speed. Biking does not have high top speed (for most people) but can often maintain decent average speed comparable to a local city bus or a slow train. Transit can vary depending upon quality of right-of-way and maintenance. Driving is usually the mode with highest top speed but in typical traffic conditions can have significantly slower average speeds; possibly even slower than transit or biking.

The speed and reliability of walking is almost always independent of the number of people walking, except in the extreme cases (important as they are). Biking is also mostly able to enjoy that same feature. Transit in a dedicated right-of-way has the capability of operating reliably independent of the number of people attempting to ride it -- up to a point. Once the number of people reaches a certain "crush" capacity then dwell times can be impacted if the operator takes no further action. Finally, driving and mixed-traffic transit are most clearly the modes that are bothered the most by other vehicles on the road. As a driver, or as a mixed-traffic transit rider, the reliability and speed of your trip is highly dependent upon the decisions made by thousands of other people.

Perception and reality don't always match. For example, many people perceive to be in "more control" when driving. But in fact, they're highly dependent on the behavior of other people. Another: many transit riders don't perceive slowness because they are absorbed in some other activity such as reading or working; whereas drivers traveling at the same average speed might be seething and pounding the steering wheel in frustration. And of course: "There is a very strong element of psychology behind traffic patterns."

So, when considering a transportation investment, there are multiple ways to increase capacity. But they do not all have the same effect, even when viewed in such an abstract fashion.

Induced demand


Halloween just passed, so let's talk about candy. Prior to Halloween, there is a high, and one might say inelastic, demand for candy. Many people want to buy candy for Halloween, almost no matter what. This usually shows up in stores as an increase in both supply and prices. Luckily for kids (and the young at heart), supply of candy is relatively easy to increase.

Post-Halloween, the demand drops off. People remember why they don't binge on candy, or let their kids do so, all year. Excess supply is sold off as prices drop in response to the lower, much more elastic, demand. Then the market returns to normal, at least, until the next spike in demand.

Normally, I don't eat much candy. I try to avoid too much sugar for health and personal reasons. And I don't want to spend money on something like that. I could easily go for weeks and months without having a piece of candy. But if you put a bowl of chocolates in front of me, I'll probably take one sooner or later. The low price and ready supply has literally induced my demand. But not every type of candy will have this effect. I'm mostly not a fan of hard candy and will probably just ignore it. And perhaps I've generated economic activity by consuming a piece of chocolate, but I may also be ruining my diet and harming my health.

You probably guessed where I'm going with this. One major difference between candy and transportation is that the supply of the latter is much, much harder to increase. With transportation, very large, very expensive fixed investments are usually expected to accommodate all fluctuations in demand over time. Furthermore, in the United States, people generally do not expect the nominal monetary cost of everyday transportation to vary based on supply and demand, even though prices regularly change for commodities such as candy, food and other items. Instead, the cost fluctuation of everyday transportation is usually expressed as traffic congestion or overcrowding.

When a transportation investment is made, it can both increase the supply and possibly lower the overall cost of travel (either perceived or real) in that corridor. And the way that travel patterns change is hard to predict, despite many efforts.

Former Bostonian, and now Los Angeles transportation blogger VamonosLA writes:
In a large city, there’s almost always going to be trips that people want to make but don’t because of large travel times. This is especially true in large US cities, where we underprice road capacity to the point that new lanes are almost always quickly filled. We misinterpret the construction of the new lane as having caused the demand, but it was there all along.
This is true but is only part of the story. To be sure, some trips generated do represent increased economic activity. Other trips might represent less efficient usage of the now-cheaper resource. For example, you might make multiple trips where one would have sufficed in the past. Or, it might be due to drivers who were enticed into giving up existing car-pools, or who ditched transit. They might even have switched from walking or biking in cases where the distances were feasible (perhaps not in LA).

For a rather stark example, suppose you live in a neighborhood with sidewalks and you are able to walk for many trips. Suddenly, the DOT decides to "improve transportation" by taking away your sidewalks and replacing them with additional car travel lanes. Most people would then choose to stop walking and start driving everywhere. Nowadays it seems unimaginable that such a thing would happen, but it has in the past, and continues to a lesser extreme: oftentimes it is not that sidewalks are removed entirely, but rather, walking is made more unpleasant for the benefit of motor vehicle traffic.

An unknown contributor to the Transportationist named JW writes:
I hate the term “Induced Demand”. I hate the idea that induced demand is something bad; something to be avoided. 
[...] From the individual’s point of view, access provides opportunities to more jobs, more entertainment and social options, and more alternatives for consumption of goods and services. From a business’ point of view, access provides a larger pool of labor and more raw materials. From a retailer’s point of view, access provides a larger pool of consumers. From a municipal government’s point of view, access allows more efficient provision of police protection, fire protection and ambulance service by reducing the number of facilities necessary for a given response time. 
[...] Transportation improvements that provide greater access per unit of time lower transaction costs. Lower transaction cost lead to great efficiency in the economy and a higher standard of living.
Although JW is careful to couch it in terms of access, which is a good thing, this otherwise sounds like an argument that could be lifted straight out of a 1950s highway builder's manual: new highways lead to greater efficiency and higher standard of living! According to this logic, the congestion on the new highway facilities justifies the construction of even more highways. Clearly something is broken in this reasoning.

The mistake that JW is making is that not all transportation improvements are equal, and that there are more considerations than just raw "access per unit of time", which itself is a difficult thing to measure in practice. For one thing, the increase in access by one mode (say, private car) can lead to the degradation in access by another mode (such as walking, or emergency vehicles). For another, some modes lead to far more pollution and other bad side effects like poor urban planning decisions (e.g. minimum parking requirements, urban renewal, pro-sprawl zoning laws, etc).

Congestion relief


But I think that by far the biggest confusion stemming from the debate around induced demand comes because people are often talking about different things. When highway builders are trying to sell expansion, and when the general public is getting excited about it, they often refer to "congestion relief" as the major feature of the proposal. But what is "congestion relief?" Is congestion relief equivalent to the fact that more people can travel more miles by car? Is congestion relief the possibly increased economic activity and higher standard of living that JW talks about? Is congestion relief the fulfillment of "latent demand" as described by VamonosLA?

I would argue that congestion relief is none of those things in the popular understanding of the term. Instead, congestion relief is generally understood to be an increase in reliability of travel and a subjective decrease in the frustration that each individual has when using the transportation facility. Highway builders sell highway expansion to a public that is hungry for predictable travel times and less stress behind the wheel. However, that sort of congestion relief is largely a false hope, precisely because of the phenomenon that has come to be known as induced demand. It doesn't matter whether that comes from latent demand, or even that it might represent increased economic activity: if the typical driver still experiences large variance in travel times, and high levels of stress, then the congestion relief was a mirage as far as he or she is concerned.

Even the level of spending associated with projects like the Big Dig did not make it immune from this effect. From the Globe:
A Globe analysis of state highway data documents what many motorists have come to realize since the new Central Artery tunnels were completed: While the Big Dig achieved its goal of freeing up highway traffic downtown, the bottlenecks were only pushed outward, as more drivers jockey for the limited space on the major commuting routes.

Ultimately, many motorists going to and from the suburbs at peak rush hours are spending more time stuck in traffic, not less. The phenomenon is a result of a surge in drivers crowding onto highways - an ironic byproduct of the Big Dig's success in clearing away downtown traffic jams.

[...] The findings also call into question the promises made when ground was broken in 1992. At the time, state officials said in a promotional mass mailing to the public that, when it was all done, "people will find the commute to their jobs faster and easier than ever."
The new tunnels may have been able to handle the flow coming into them, but those outer links themselves were not able to live up to the induced demand.

Conclusion


So what can be done about congestion, then? It has rightly been pointed out that transit improvement and biking facilities do not "reduce congestion" in this sense. Actually that point should be clarified further: those improvements do not "reduce congestion" when the travel demand is virtually inexhaustible. However, it is possible that in some cases demand is actually exhaustible: for example travel along Comm Ave in Allston, in which case those kinds of improvements can be very effective at real congestion relief. But for destinations such as downtown Boston, there will almost always be another driver willing to replace a trip that is mode-switched away from private vehicle.

All that does not make transit, walking and biking improvements worthless. Jarrett has discussed several reasons why transit is still very worthwhile even without "congestion reduction" and I won't reiterate, but I will add that dedicated-lane transit, walking and biking facilities all increase the supply of "reliable transportation" that boosts accessibility without the harmful side effects imposed by increased car travel. Walking and biking also have great implications for the health of urban neighborhoods and are worthwhile goals on their own.

The idea behind congestion relief is to try and make personal motor vehicles as reliable as walking, biking, or riding a well-run railroad. There are only two known ways to make that actually happen: (a) rigorous scheduling, or (b) decongestion pricing. Railroads avoid congestion and conflict by imposing very strict discipline and heavily analyzed schedules on the movement of trains. To try and apply that same discipline to private cars on roads is politically and practically unfeasible. Therefore that leaves decongestion pricing as the only known way to make private motor vehicle travel reliable, and it works by allowing the price of travel to vary based upon the supply and demand for travel. This is, of course, standard procedure for allocating just about every other resource in our society, but for some reason is completely ignored when it comes to roads. In any case, this article is about induced demand and I have already gone on too long, so I will leave discussion of pricing to others.