Studies from around the world have validated the existence of induced demand:  each improvement to freeway capacity in urban areas generates more traffic.

The best available science worldwide—in Europe, Japan and North America—shows a “unit-elasticity” of travel with respect to capacity:  A 1 percent expansion of capacity tends to generate 1 percent more vehicle miles traveled.

The fundamental law of road congestion requires us to fix broken traffic models and stop widening highways in a futile effort to reduce congestion.

Call it what you will:  Jevons Paradox, Braess Paradox, Marchetti’s Constant or Downs’ Triple Convergence, the science confirms them all.

Induced demand:  More road capacity produces more traffic

At City Observatory, we’ve related the classic example of North America’s widest freeway, the 23-lane Katy Freeway in Houston.  It’s been successively widened many times, most recently at a cost of $3 billion, and within three years of its expansion, commute times were even longer than before.

But there’s much more than anecdotes like the Katy Freeway to buttress the observation of induced demand.  Sophisticated, in-depth studies of transportation infrastructure and traffic levels, that look at entire nations and measure traffic changes over decades find what is now being called “the fundamental law of road congestion.”  An increase in road capacity directly generates a proportional increase in traffic, with the effect that congestion and travel times quickly return to (or worsen from) pre-expansion levels.  Simply put, expanding road capacity is a futile, and self-defeating effort.  Urban highway expansion is the labor of Sisyphus.

Two recent and definitive studies are Duranton and Turner’s “Fundamental Law of Road Congestion,” and more recently Kent Hymel’s “If you build it they will drive” both of these studies use data for the US and find a unit elasticity of traffic with respect to roadway expansion. Hsu and Zhang found a nearly identical result for roadway expansion projects in Japan.

Europe:  Still more evidence Induced Demand and the fundamental law of road congestion

The latest evidence of the universality of the fundamental law comes from Europe.  Three researchers from the University of Barcelona use two decades of data for hundreds of European cities to replicate the methodology used by Duranton and Turner and Hymel in the U.S.  They find very similar results, confirming the fundamental law of road congestion.  The best estimate is that the elasticity of travel with respect to capacity is essentially unitary:  a one percent increase in highway capacity generates a one percent increase in vehicle travel.

We use data for the 545 largest European cities to estimate the elasticity of a measure of congestion with respect to highway expansion. The results indicate that this elasticity is in the range close to 1. This suggests that expansion of the highway network induced the demand for car travel, and so, on average, the level of congestion remained roughly unchanged in the period 1985–2005. In other words, we show that investments in highways did not effectively relieve traffic congestion.

Congestion in Edinburgh (The Herald)

Tolling is the only way to avoid the induced demand trap

Garcia-Lopez, Pasidis and Viladecans-Marsal examine how the prevalence of tolled roadways affects the induced demand effect.  Cities that toll a higher proportion of their highway system have a much smaller induced demand effect.  Their analysis concludes that traffic is highly elastic in response to capacity expansions in cities with no tolls (each 1 percent in capacity results in a nearly 2 percent incrase in travel; which traffic is highly inelastic in cities with 100 percent tolled highways (a 1 percent increase in capacity results in a 0.3 percent increase int raffic:

(1) highway improvements increase congestion, (2) the effect is smaller in cities with tolls, and (3) the fundamental law is mainly related to cities without tolls or with a low percentage of tolled highways. In particular, and focusing on our preferred specification in column 3 (using a continuous interaction), a 1% increase in lane kilometers increases congestion by 1.9% in cities without tolls and by only 0.3% (=1.9-1.6) in cities with tolls in all their highways (100% share of tolled highways). Some simple computations show that the fundamental law applies to cities with a share of tolled highways below 56%. These results can be regarded as novel evidence in line with recent literature suggesting that the solution to traffic congestion is the adoption of ’congestion’ pricing policies.

Policy implications:  Fix broken traffic models; Stop widening highways

The fundamental law of road congestion is a demonstrated scientific fact, with unambiguous implications for public policy.  First and foremost, the fundamental law signals that the folk wisdom (repeated by highway boosters) that we can somehow “build our way out” of traffic congestion is utterly false.  More roads simply generate more traffic and more sprawl.  Cities and states should stop spending money on road widening projects to reduce congestion.  There’s a second, more subtle and technocratic point as well.  The fiction that more capacity will somehow reduce congestion is actually hard-wired into many of the “four-step” traffic models highway departments use to plan (and justify) highway widening.  The models are calibrated in a way that either ignores or denies the existence of induced demand, usually by simply assuming that the level of traffic demand is fixed, and unaffected by either journey times, delays or congestion.  At best models may crudely re-route traffic in response to congestion, but they fail to alter aggregate trip demand, especially in the long run.  As a result, as Jamey Volker, Susan Handy and Amy Lee show, most existing travel models create the false illusion that a wider road will lead to faster traffic.  Transportation planners–and funding entities, like the Federal Highway Administration—should insist that transportation models be updated to reflect scientific reality. The induced travel calculator shows how this can be done, now.

A truth with many names and discoverers

While this new study from Barcelona, and the similar papers by Duranton & Turner, Hymel, and Hu and Zhang all elaborate in great statistical detail on this finding, the basic concept is well understood, and has been for decades (or longer).  The scientific validation of the phenomenon of induced demand buttresses a series of related explanations:  the Jevons Paradox, the Braess Paradox, and Marchetti’s constant and the Triple Convergence.

Jevons Paradox holds that an increase in efficiency in resource use will generate an increase in resource consumption rather than a decrease.  English economist William Stanley Jevons predicted greater efficiency in using coal would increase its use in 1863.  The efficiency gain seems paradoxical, if one assumes that demand is unaffected by the lower price of a more efficient process, but by making something more efficient, we generate additional demand.

Braess’s Paradox is the application of this general idea specifically to traffic.  German engineer Dietrich Braess postulated exactly this in 1968.

Marchetti’s constant is a corollary to these paradoxes:  It observes that the amount of time human’s devote to daily travel remains constant regardless of improvements in transportation technology. Whether walking, riding horses or streetcars, or driving cars, we devote about an average of an hour a day to travel.  Marchetti’s constant means that we use improvements in transportation to travel further, rather than saving time, a result fully consistent with the fundamental law of road congestion. (The observation has been made independently by many observers including Bertrand Russell as early as 1934.

Down’s Triple Convergence, in his 1992 book Stuck in Traffic, economist Anthony Downs described the existence of a “triple convergence” in which changes in road infrastructure prompted changes in the mode (i.e. transit to car), time, or destination of trips in ways that would lead to congestion reappearing even after an expansion of road capacity.

Miquel-Àngel Garcia-López & Ilias Pasidis & Elisabet Viladecans-Marsal, 2020. “Congestion in highways when tolls and railroads matter: Evidence from European cities,” Working Papers wpdea2011, Department of Applied Economics at Universitat Autonoma of Barcelona.