Monday, July 26, 2010

More on rail transit in America

Cato's Randal O'Toole has assembled data for a cross-section of U.S. rail transit lines and cannot find any good news. Here is the abstract of his paper.

Over the past four decades, American cities
have spent close to $100 billion constructing rail
transit systems, and many billions more operating
those systems. The agencies that spend taxpayer
dollars building these lines almost invariably call
them successful even when they go an average of
40 percent over budget and, in many cases, carry an
insignificant number of riders. The people who
rarely or never ride these lines but still have to pay
for them should ask, “How do you define success?”
This Policy Analysis uses the latest government
data on scores of rail transit systems to
evaluate the systems’ value and usefulness to the
public using six different tests:

• Profitability: Do rail fares cover operating
• Ridership: Do new rail lines significantly
increase transit ridership?
• Cost-Effectiveness: Are new rail lines less
expensive to operate than buses providing
service at similar frequencies and speeds?
• The “Cable Car” Test: Do rail lines perform
as well as or better than cable cars, the oldest
and most expensive form of mechanized
land-based transportation?
•The Economic Development Test: Do new rail
lines truly stimulate economic development?
• The Transportation Network Test: Do rail
lines add to or place stresses upon existing
transportation networks?

No system passes all of these tests, and in fact few
of them pass any of the tests at all.

Paige Elise Kolesar helped me augment Randal's data. In particular, we added auto external cost estimates from a recent American Economic Review paper by Ian Parry and Ken Small. According to these authors, such costs are in the range of 6 cents/mile (off-peak, Washington, DC) to 31 cents/mile (peak, Los Angeles). We also used Tom Rubin's estimate of how many rail riders are new-to-transit (median near 25%) and asked: What if all of them had substituted away from their cars?

The annualized values allowed us to compute costs (losses) per boarding after non-rider benefits as well as fares had been counted on the plus side of the ledger. The results for eight commuter rail lines were $19.50 per boarding (weighted average; almost $40 per round-trip); for six heavy rail projects, the comparable weighted averages were $8 and $16. For 20 light rail projects, $9 and $18.

These numbers speak for themselves and simply update suspicions that anyone who bothers to look at the data has long held. Of course, none of this makes a whit of difference to the many who know that we "need" to build many more such projects.