investigate the relationship between interstate highways and highway
vehicle kilometers traveled (vkt) in us cities. We find that vkt
increases proportionately to highways and identify three important
sources for this extra vkt: an increase in driving by current
residents; an increase in transportation intensive production activity;
and an inflow of new residents.
The provision of public
transportation has no impact on vkt. We also estimate the aggregate
city level demand for vkt and find it to be very elastic. We
conclude that an increased provision of roads or public transit is
unlikely to relieve congestion and that the current provision of roads
exceeds the optimum given the absence of congestion pricing.
first inclination of most urbanists, when confronted with something
like this, is probably to bristle and conclude that the authors are
nuts. Transit can’t relieve congestion? What would happen to New York
or Washington if transit systems were shut down for a day? There would
There would indeed be chaos. But that doesn’t mean that the authors are wrong, and it’s important to understand why.
a transit system is built, two things happen to area roadways. Some
subset of drivers will find that commuting by transit is faster, or
cheaper, or more convenient (or all three) than driving and will switch
to transit. If that were where things ended, then transit construction
would indeed reduce congestion.
But there is a knock-on
effect. When drivers switch to transit, roads become less congested,
and driving therefore becomes more attractive. Drivers who were
previously commuting at off-peak times to avoid congestion will switch
to peak times, and drivers who were otherwise adapting to congestion by
working at home some days or taking longer, alternative routes will
adjust their behavior as well.
The end result will be … a
return to road congestion. This may not happen immediately. When new
capacity of any sort — roads or rails — is built, there may be a time
period during which traffic flows more freely, but ultimately
settlement and transportation patterns will adjust until roads are
This happens because roads are under-priced
(and often free). If drivers don’t have to pay to use scarce road
space, and don’t have to pay to cover the cost of congestion their
driving imposes on others, then drivers will use the road until it is
congested. Because the government isn’t using price to ration demand
(as is done with most consumer goods), demand will rise until the cost
of lost time rations demand, and pushes drivers to take other routes or
I don’t think there’s any point in denying this. Instead, there are key points that urbanists should take to heart.
is that roads are overbuilt. Years of trying to fight congestion,
ineffectually, by building new capacity have generated a road network
that is far too large.
Another is that road congestion can
only be addressed through imposed rationing, and the most efficient way
of doing this rationing is through variable tolling. This, of course,
should raise a lot of new and desperately needed revenue.
third point is that just because transit can’t permanently reduce
congestion doesn’t mean that there aren’t good reasons to build
transit. Indeed, the authors of the above study conclude that it will
often make sense to do so.
What kinds of reasons? Well, just
because the road network is overbuilt doesn’t mean that all
transportation demand is adequately met. In many areas — particularly
in center cities and job concentrations — new capacity is needed, and
transit offers an effective way to move a lot of people through dense
Another reason is that in the busiest areas,
congestion tolls are likely to be fairly high, and transit can offer an
affordable alternative to driving to low- and middle-income travelers.
finally, transit can help address one of the key contributors to
transportation demand — poor land use. Transit networks offer a
framework around which denser, mixed-use neighborhoods can be built,
allowing more people to walk or bike to work or reduce their travel
needs. Infrastructure is expensive to build and maintain, and ideally
we’d like to grow in a more intensive way that reduces that needed
infrastructure for a given amount of land.
Urbanists ought to
trumpet these results. Drivers are more likely to accept congestion
tolls if it is made clear to them that such tolls are the only
way to reduce road congestion. And tolling will create both the demand
and the revenue for new transit capacity. The relationship between
capacity and congestion is one we’d all do well to understand.