Duranton and Turner on the fundamental law of traffic congestion in AER

This paper in AER is getting its kicking around the web from the transit fanboys and those outside the transport field who don’t get why managing congestion is treated as a goal for public transit. The commentariat is in umbrage: surely transit riders benefit from transit, yada, yada, and this result means nothing. Andrew Gelman gets a buy on his comments because he’s brilliant, I love his Bayes book, and I learn more from his blog than I learn from most books. Everybody else needs to chill.

The Duranton and Turner paper is significant for multiple reasons. First of all, transit fanboys have nobody to blame but themselves for the widely held perception that transit investment decreases congestion. It’s part of every “More rail, more rail, more rail” chant I’ve ever read in about 20+ years of professional life in transport planning. Why? Because if you didn’t promise those who don’t ride transit a benefit from the billions we spend on transit, they’d never hand over the billions to you. Outside of the few major transit markets in the US, transit riders themselves have never been a big enough constituency to hold their own in budget battles, which is one reason why they are at such a disadvantage in Federal budget talks.

Promising nonuser benefits has been the major marketing strategy of transit agencies for at least 40 years. Transit saves the air! It makes us skinny! It decreases congestion! And so on and so forth. Promises of this type, however, have the tendency to prompt empirically minded researchers like Matt Turner to get out their datasets and their instrumental variables and get all hypothesis testy on you.

Again: Gelman gets a buy because he readily admits he hasn’t been at the party for 20+ years, but it’s some serious gaslighting at this stage of the game, after transit advocates have spent decades schilling the investment based on nonuser benefits, to respond with “how silly those economists are! Transit provides mobility! Of course, that’s the benefit of transit!” Especially when ridership figures on many systems are so disappointing. That’s a pretty politically dangerous response for everybody who, unlike Gelman, doesn’t live in NYC because if we do cost-benefits on transit investment based on benefits to riders alone, we’d see a lot less investment. I assume that’s not what the fanboys want.

Anyway, so what’s interesting in the manuscript itself? Here’s the actual citation:

Duranton, Gilles, and Matthew A. Turner. 2011. “The Fundamental Law of Road Congestion: Evidence from US Cities.” American Economic Review, 101(6): 2616–52.

Here’s the abstract:

We investigate the effect of lane kilometers of roads on vehicle-kilometers traveled (VKT) in US cities. VKT increases proportionately to roadway lane kilometers for interstate highways and probably slightly less rapidly for other types of roads. The sources for this extra VKT are increases in driving by current residents, increases in commercial traffic, and migration. Increasing lane kilometers for one type of road diverts little traffic from other types of road. We find no evidence that the provision of public transportation affects VKT. We conclude that increased provision of roads or public transit is unlikely to relieve congestion. (JEL R41, R48)

So what? Anthony Downs (and other smart people) pointed out the theory of triple convergence quite some ago–that additional capacity on an unpriced system will erode until a congested re-occurs. In the absence of money prices, the only thing that disciplines demand on a facility are the time costs, and the time costs rise with…congestion. So one of the most misguided commenters asks: Where do all the extra drivers come from? The answer is easy:

a) population growth or
b) nowhere, since you don’t need additional bodies. You just need additional trips.

So if we provide a whole bunch of new supply, transit or otherwise, on a high-demand corridor, that supply will get used as the time costs are lower, and the out-of-pocket money costs of car ownership at that point are sunk and unrelated to trip time of day–until congestion starts in again. So if the congestion on the 405 clears up suddenly because we’ve provided commuter rail (I’ll just hold my breath until that happens), other drivers may opt on to the facility, or some of the drivers left may decide to sneak in a few more trips during the day.

There is a point when supply can become saturated: if you put a 50 lane road down in Des Moines, I doubt you’ll get gridlock. But that’s a flummery example. Nobody proposes such things.

Transit fanboys are reacting so strongly to the Duranton and Turner paper because for a very long time, people have argued Down’s triple convergence only in terms of highway supply. It was a rational for all those who said “You can’t build your way out of congestion” at the same time they argued for building more rail. The problem appears to be–and most people who understand economics have known this for awhile—that triple convergence holds regardless of whether the additional supply is highway or transit.

The problem that Duranton and Turner highlight concerns the highly counterfactual nature of most purported environmental benefits in public investment, not just transit. The promise that transit “clears the air” or “reduces congestion” or “reduces auto use!” contains an implicit caveat that few people acknowledge: transit is a cleaner mode than if we were to meet the additional travel demand with highway supply rather than transit supply. But it’s much snappier to say “Transit clears the air” than it is to say “Transit clears the air relative to what it would be had transit users driven cars.” These are benefits that occur from shifting future user behavior.

That’s why the California HSR advocates argue that their new $98 billion HSR investment is a bargain compared to the $127 billion of airport and highway expansion that nobody has actually proposed yet.

The point from Duranton and Turner: if your metro area has a problem with cars now—either related to congestion or to air quality—you are going to keep your problem, even if you build transit.

HOWEVER. And this is for the fanboys:

a) If you don’t have a problem yet with auto-related externalities, new transit supply may forestall those problems. Probably not forever, but you may buy yourself some time, and

b) Restated: if you already have a problem with auto-related externalities, new transit supply may help change the slope of how bad those problems get over time.