Accessibility without congestion: Towards a less expensive lunch

Big cities have lots of congestion. Big cities have lots of accessibility. Therefore congestion causes accessibility. . Therefore accessibility causes congestion.

I don’t think the last sentence is necessarily correct either. It might be more accurate to say that congestion is an indicator of economic activity (and poor resource management). However increasing the number of people per unit space (density) without increasing the transportation capacity to move them, or changing their travel behavior, will increase congestion. Similarly, increasing the number of people per unit space (density) without changing their travel costs will increase accessibility, and may increase accessibility even if you increase their travel costs.

As a reminder, accessibility is a measure of the ease of reaching valued destinations, which we often operationalize as the number of things (jobs, shops, schools, etc.) you can reach in a given time (10, 20, 30 minutes) by a particular mode (car, bus, walk, bike) at a particular time of day (7 am, noon, 5 pm, midnight). This operationalized measure is called cumulative opportunities, and is easily explained, even to politicians. Big cities have more things, and higher density, and thus usually have more accessibility, especially at longer time distances. The opposite of accessibility is isolation (or access to nothing, how much of nothing I can reach in 10, 20 or 30 minutes). We know people care about accessibility, since the cost of land in the most accessible places is much higher than the least accessible places.

Congestion occurs when a facility is overcrowded. So if a bridge can carry 3600 people an hour (its capacity), but 4000 people want to use it (its demand), we say it is congested. At the end of the hour, there will be 400 people left waiting to use the bridge (who will presumably be first in line to use it during the next hour). Congestion occurs potentially on all modes, since every mode has a capacity, and on any facility, which again have some capacity. Many modes however are operating very far from capacity, have excess, unused, or spare capacity, and thus don’t see congestion with additional users. Congestion matters to users because of the delay that results, the travel time above and beyond what the user would have experienced in its absence.

So can we increase accessibility, which people value, without increasing congestion, which people dislike?

There are several possible strategies:

  1. Increase capacity
    1. Expand capacity on the congested mode or facility.
    2. Expand capacity on an alternative mode or facility.
    3. Build new facility.
    4. Better manage existing capacity
  2. Reduce per capita demand
    1. Price
    2. Ration
    3. Exhortation

There are many different ways these can be implemented.

To start, we could simply increase capacity without trying to address demand. Unfortunately, there are rising costs (and diminishing returns) to capacity investments. All the high benefit, low cost projects have already been done (the economist’s proverbial $10 bill is not lying on the street is it?). So we are left with really expensive projects to get more capacity, which often involve tunneling or elevated structures (since all the good rights-of-way are already used). Perhaps there will be technological breakthroughs which reduce the costs of tunneling, or other construction. It hasn’t happened yet.
We might better manage capacity. This is certainly an option. If we can get the driver out of the loop, so cars can drive themselves, we could narrow lanes (and thus get more of them per unit of pavement) and cars could follow more closely. Both of these things will increase throughput. There are simpler technologies (ramp metering, rapid incident clearance, and so on) that have already been widely deployed in many cities.

Pricing comes in numerous flavors. It might be very precise to time and location, or might be in selected areas, or on selected facilities, or at selected times of day, or everywhere. HOT Lanes that charge by time of day on selected facilities may guarantee that a particular facility is uncongested (by limiting demand on that facility). This is as much pricing for reliability as pricing to reduce congestion. The political advantage of HOT lanes is their voluntary nature, and the untold alternative. We could certainly extend this to truck-only toll lanes (TOTs) and perhaps to some other domains. The London Congestion Charge affects most traffic entering Central London, regardless of time of day or miles traveled. In Germany heavy goods vehicles pay to travel on the autobahn. Some people have even come up with the silly concept of paying people not to drive. By pricing more in the peak (and less in the off-peak) flexible travelers will switch travel times from when capacity is fully utilized to when there is spare. Of course, induced demand operates in all cases, but that can be considered when setting prices.

Rationing seems fair, and is used in many megacities throughout the world, either rationing number of vehicles, or the days they can be driven. During World War II, the US rationed rubber tires and fuel. Rationing however often devolves into a black market (and thus pricing), as people pay for rations. For instance there is a large market in license plates in cities that have day of week license plate rationing.
We often mock exhortation (well at least in my social circles), but on occasion exhortation has been used fairly effectively, often coupled with monetary incentives: smoking has gone down, and recycling up in the US in part due to exhortation. (My 5 year old daughter tells me how recycling is better than throwing in the trash, so clearly the education campaigns start young). How much of the change is due to changing mores and social preferences, and how much to cigarette taxes or trash collection discounts for recycling is hard to say. On the other hand, modern campaigns to encourage transit use and carpooling have been notable failures.

In short, to paraphrase John Lennon and Yoko Ono, Congestion is over! If you want it. We can have accessibility without (or with less) congestion. I don’t think we want it badly enough. The choice is really congestion OR pricing, and the political cost of pricing has to date outweighed the political cost of congestion. We don’t value the time savings of accessibility enough for politicians to do the things necessary to save time. On the other hand, voluntary tolls (HOT lanes, TOT lanes, and so on) are more politically acceptable, giving people travel time reliability for an uncertain price, which is better than nothing, but certainly not the best possible.

A large part of the problem is thus its political nature. We would not in the US tolerate periodic blackouts from an electric utility because they could not manage supply and demand. In fact this was the what in large part led to the downfall of Governor Gray Davis of California and the rise of Arnold Schwarzenegger. Why do we tolerate the transportation equivalent,  queueing? Because roads are owned and operated by governments. If they were separate (and regulated) organizations, not directly responsible to the state legislature, we might have different outcomes. The notions of “free” and “already paid for” and “double-taxation” that are used to politically defeat tolling proposals would be replaced with a “fee for service” concept common in public utilities. My mental model of governance is that “institutions loosely coupled“, each with specific missions, management, and revenue, would outperform a giant monolithic government that tries to do everything for everyone.

Getting from our current world to a slightly better one can be achieved calmly and rationally via white papers and deliberation, or through a real or politically generated “crisis” (the preferred mode of governance in the US). I have the feeling that so many people have played the “crisis” card that there is “crisis-fatigue”, and as a rationalist would certainly rather we went at this systematically. However, politics does not seem to want to make transitions without some stress. Perhaps the rise of electrification and the collapse of gas tax revenue will be the crisis required to move to a new and different organizational regime in surface transportation. But this is a slowly building crisis that could take the rest of my career, and I am impatient.

Directions for Research in Transport and Land Use

Below I posit some directions for research in transport and land use. Comments welcome.
1. We need more panels and time series and fewer cross-sectional analyses. If we want to establish causation, we need to look across time, otherwise, we are stuck simply with correlations. [And as we know, correlation is not causation]. We need data that examines the evolution and dynamics of transport and land use systems. I have not quite come to the conclusion that all analyses must be temporal (that is rejecting any atemporal analysis), but I am really tempted to do so as a reviewer.
2. We need to improve the scientific rigor of our research. The discipline is ripe for continuing meta-analysis to establish the magnitude of effects, and to reduce the range of estimates (and explain the range that exists through different underlying causal factors).
3. We need to more systematically consider network structure when looking at explanations of travel behavior. This includes measures of topology, morphology, and hierarchy. The measures that have historically been used have been relatively easy to estimate, but don’t get at the gestalt of the network as an integrated system.
4. We need to systematically look at the difference between travelers perceptions of how systems operate and how long are travel times, and what we analysts measure. The differences can be systematically explained, at least in part, and people of course make decisions based on how they think the world works, not on how we think it does. We could then examine why perceptions differ from measurements, how much is simply differences in linguistic interpretation (when a trip begins and ends is somewhat ambiguous, e.g.), and how much is differences in time perception, and how much is “rounding” error, and how much is strategic to either impress with the length of the commute (which brings to mind the Four Yorkshiremen sketch) or to exaggerate in order to get sympathy or a policy response.
5. We need to increase the inter-disciplinarity in the study of transport and land use research, with planners, geographers, engineers, economists, and others working together looking at these problems.
6. We need more international and historical cases in the field to build towards a general truth. Reasoning is both inductive and deductive, but so much of what we are doing is complex, one often cannot simply derive from theory whether a change will lead to more or less travel, it depends on parameters, for instance. the fixed costs of engaging in a trip vs. the variable (and non-linear) costs of travel.