Carpooling is a flop: NEW RESEARCH ON HOT LANES

By Robert Poole

August 14, 2002

This spring I came across a brilliant PhD dissertation from UCLA on
HOT lanes. Along with some other research on rush-hour capacity, it
may revolutionize our thinking about how to best manage the lanes on
congested urban freeways. I wrote a recent Public Works Financing
column about these implications, and I'm pleased to share it with you
here.

Time to Rethink Carpool Lanes

The recently released journey-to-work figures from the 2000 census
reveal what many of us have long suspected: carpooling is a flop.
Despite the expenditure of billions of dollars adding carpool lanes
to congested freeways, carpooling declined from 13.4% of work trips
in 1990 to 11.2% in 2000. Carpooling's mode share declined in 36 of
the largest 40 metro areas-including highly congested Los Angeles and
San Francisco. So what do we do now?

Some transit advocates have resented the use of these lanes by
carpools from the outset, reminding us that high-occupancy vehicle
(HOV) lanes were originally conceived as busways and should be
converted, accordingly. Some highway advocates argue vociferously for
converting most carpool lanes to general-purpose (GP) lanes. And the
Federal Highway Administration's Value Pricing office tries valiantly
to persuade metro areas to convert under-performing HOV lanes to HOT
lanes.

What's been missing from this discussion has been serious
quantitative analysis of the tradeoffs involved. That gap has
recently been filled by a little-noticed UCLA Ph.D. dissertation by
Eugene Kim, "HOT Lanes: A Comparative Evaluation of Costs, Benefits,
and Performance." Kim used a logit travel-demand model to estimate
the comparative travel times that would come about by converting an
existing HOV lane on a congested freeway to (a) a GP lane, (b) a HOT
lane, or (c) a Toll lane. He also estimated long-term (20-year) costs
and benefits of each alternative, as well as environmental impacts.

The results are striking. In almost all cases, HOT or Toll lanes
provide a greater degree of fiscal, consumer welfare, and
environmental benefits than any other expressway investments. Echoing
previous research by UC Berkeley's Joy Dahlgren, Kim shows that there
is a very limited set of conditions under which HOV lanes can be the
best option. But in most cases, society would be better off if the
lanes were converted. Converting to GP lanes is most defensible when
HOV use is less than 7% of all corridor trips, and there are under
700 vehicles/hour in the HOV lane. But in almost all cases,
converting to a Toll lane produces greater benefits, primarily
because it can preserve free-flow conditions as traffic continues to
grow and freeway congestion worsens. And because Toll lanes generate
substantial revenues for the highway system.

Whether to convert only to HOT (carpools still go for free) or go all
the way to Toll receives detailed attention in Kim's work.
Intuitively, one might expect that conversion to Toll would produce
less delay-reduction than conversion to HOT, because fewer people
will continue to carpool if those vehicles have to pay. But the
modeling shows that conversion to Toll produces large delay-reduction
benefits "regardless of whether the conversion . . . results in a
significant increase or decrease in the initial proportion of HOVs."

These results clearly support the idea that many of today's HOV lanes
are candidates for conversion to Toll. As Kim points out, tolling
indirectly preserves economic incentives to ride-share, by (1)
spreading the toll over more than one person, and (2) by providing
insurance against travel-time uncertainty in the event that a carpool
participant unexpectedly cancels-an effect already observed on the
I-15 HOT lanes. But there's a way of guaranteeing that these Toll
lanes would continue to serve large numbers of people in
high-occupancy vehicles: let express buses use the Toll lanes at no
charge. One 60-passenger express bus takes up no more room than three
cars, but carries as many people as 20 three-person carpools.

What about the operating characteristics of Toll lanes? Both existing
California HOT Lanes (91 Express and I-15) use variable pricing to
maintain 65-mph speeds during peak hours. Critics have argued that
this is excessive, because (1) maximum throughput is obtained in the
35-50 mph range and (2) because emissions are significantly worse at
65 mph than in the lower speed range. Recent research challenges both
beliefs.

Pravin Varaiya at UC Berkeley analyzed speed-flow data on congested
LA-area freeways, using the new Caltrans Performance Measurement
System. Measuring actual speed and throughput during entire rush-hour
cycles, Varaiya validated the traditional bullet-nose-shaped
speed/flow curve-i.e., as volume increases, speed gradually decreases
until reaching a point of instability at around 2100
vehicles/lane/hour, at which point both speed and flow decrease
significantly (with speeds dropping to 15 mph and flows cut to as low
as 1300/hour). He concludes that 60-mph is the most efficient
rush-hour speed and that lower speeds, such as 45 mph, are not
sustainable (because flow becomes unstable).

In his dissertation, Kim compared the environmental impacts of his
four alternatives using the EMFAC 2000 model. He concludes that the
HOV case "produces a greater output of ROG, NOx, and CO than
converting to either GP or Toll lanes." And "a Toll lane produces the
largest emissions reductions because it eliminates some vehicle trips
(like an HOV lane) while reducing congested conditions more
effectively than a GP lane." This is especially the case when the
analysis encompasses a 20-year period, as Kim's does.

Kim's results are reinforced by a recent study of Houston's
now-rejected 55-mph freeway speed limit, imposed briefly as an
air-quality remedy. A new study by Environ International Corp., using
the EPA's recent MOBILE6 model, found no measurable effects of the
55-mph limit on ozone and only modest impact on NOx (with most of
that impact coming from heavy vehicles, not cars).

Thus, a growing body of research supports the case for converting
most of America's HOV lanes into value-priced Toll/Bus lanes,
operating at a throughput-maximizing 60-65 mph design speed. How
about it, FHWA and FTA?

Bob Poole of the Reason Foundation publishes a periodic email newsletter called Surface Transportation Innovations.  If you would like to subscribe to his newsletter, send an email to bobp@reason.org.