Vanishing Automobile update #57

Does Light Rail Pay for Itself?

8 December 2005

Among the many arguments rail advocates use to promote their cause is the claim that rail transit costs less to operate than buses. The savings, they claim, will soon pay for the cost of rail construction.

Rail supporters point to St. Louis, where the Bi-State transit agency says it spent 29 cents per passenger mile operating light rail but 88 cents per passenger mile on buses in 2003. Since the agency reports that it carried 125 million light-rail passenger miles, this savings amounts to $72 million per year. That would cover the $800 million cost of building light rail in a little more than 11 years. Of course, this does not take into account interest or the difference in inflation between 1992 (which much of St. Louis' light-rail line was built) and 2003.

As pointed out on page 353 of The Vanishing Automobile, a major flaw in this reasoning is that light-rail operating costs are not comparable to bus operating costs. New light-rail lines "skim the cream" of transit riders because they tend to be built in the busiest transit corridors. Since costs per passenger mile depend heavily on ridership, the cost of any transit running in a busy corridor is likely to be less than the cost of a bus roaming through low-density suburbs.

But let's take the operating cost averages for granted for a moment and look at the nation's new light-rail lines to see whether operational savings can ever cover rail's high construction costs. Table one, which is based on the Federal Transit Administration's 2003 National Transit Database, shows how much money light rail saves in operating costs if we believe the average operating costs for buses represent the operating costs of buses in light-rail corridors.

Table One
Potential Operating Savings from Light Rail
Light Rail       LR PM   LR $/PM  Bus $/PM  Savings
Baltimore        48,554    0.71     0.72       $386
Boston          173,900    0.60     0.75     25,924
Buffalo          14,444    1.18     1.01     -2,396
Dallas          120,674    0.48     0.97     59,564
Denver           45,495    0.44     0.73     13,145
Hudson-Bergen    12,347    1.29     0.59     -8,687
Los Angeles     225,712    0.38     0.51     28,668
New Orleans      13,475    0.70     0.70        -44
Philadelphia     61,018    0.72     0.81      5,377
Pittsburgh       31,988    1.00     0.77     -7,248
Portland        169,572    0.33     0.72     67,163
Sacramento       47,365    0.64     0.91     12,626
Salt Lake City   91,173    0.36     0.92     30,827
San Diego       159,356    0.24     0.55     48,366
San Francisco   109,941    0.98     0.86    -13,578
San Jose         26,815    1.90     1.39    -13,692
St. Louis       124,973    0.29     0.88     72,798

The table shows that bus and light-rail costs per passenger mile are nearly identical in Baltimore and New Orleans, and buses actually cost less than light rail in Buffalo, New Jersey (Hudson-Bergen), Pittsburgh, San Francisco, and San Jose.

Still, light rail achieves significant annual savings in ten cities. Are these savings sufficient to justify the cost of light-rail construction? It is difficult to answer this question in Boston, New Orleans, Philadelphia, and San Francisco since light-rail lines in these cities are upgrades of older rail transit lines, so we really don't know how much they cost to construct.

The other cities are "new" light-rail cities, meaning they had bus-only transit systems before 1980 and built new rail transit lines since then. Approximate construction costs of light rail in each of these cities were published in a recent paper by Nathan Baum-Snow and Matthew Kahn. The dollar costs in table 1 of Baum-Snow/Kahn's paper are in "current" dollars, meaning dollars in the year they were spent. Baum-Snow & Kahn did not have data for Hudson-Bergen and Salt Lake City, so I used numbers reported by the lightrailnow.org web site for Salt Lake City and newspaper reports for Hudson-Bergen.

I used standard GNP price deflators to convert these costs to 2003 dollars so they are comparable to the 2003 savings shown in table one. Since rail lines take several years to construct, I used the year prior to the opening date of each rail line as the average date that the dollars were spent.

Capital costs are one-time only while operating costs are annual. To convert construction costs to an annual figure, I amortized them over 30 years at 7 percent interest. Thirty years represents the average lifespan of rail hardware: roadbed, tracks, electrical transmission systems, cars, stations, and so forth. I used 7 percent both because it is prescribed by the Federal Transit Administration for amortizing capital costs and because it is likely that a local government with a few hundred million dollars on its hands could find an investment somewhere that would return 7 percent. For comparison, I also used 4 percent.

Table Two
Annualized Capital Cost v. Operational Savings
Light Rail    Cap Cost  Amort 7%  Amort 4%  Op Savings
Baltimore        $757     $60.4     $43.4       $0.4
Buffalo         1,007      80.4      57.7       -2.4
Dallas          1,540     122.9      88.2       59.6
Denver            413      33.0      23.7       13.1
Hudson-Bergen   1,128      73.6      52.8       -8.7
Los Angeles     2,028     161.9     116.2       28.7
Portland        1,538     122.8      88.2       67.2
Sacramento        302      24.1      17.3       12.6
Salt Lake         451      36.0      25.8       30.8
San Diego         997      79.6      57.1       48.4
San Jose        1,024      81.8      58.7      -13.7
St. Louis         918      73.3      52.6       72.8

Table two shows that operational savings on none of the light-rail systems are greater than amortized capital costs at 7 percent, though St. Louis is close. At 4 percent, both St. Louis and Salt Lake cover their capital costs with operational savings.

What makes St. Louis and Salt Lake City special? According to the FTA data, one important factor is the huge difference between the average number of people riding light-rail cars and the number riding buses in those cities. St. Louis and Salt Lake light-rail cars both carry an average of 24 people over the course of a day while St. Louis buses have only 7 and Salt Lake buses average only 5.

Does anyone really think that express or limited buses running in the corridors now served by light rail would average only 5 to 7 riders? In the case of Salt Lake, if buses carried just 20 percent more riders, light-rail's operational savings would be less than annualized construction costs even at 4 percent.

If transit agencies truly want to save money on operating costs, the data point to a much better way than spending hundreds of millions or billions on rail transit. Several transit agencies that have light rail also contract out some of their bus services to private operators. This results in a substantial savings: as shown in table three, private operators are able to operate and maintain buses for about half the costs of government agencies.

Table Three
Cost Per Vehicle Revenue Mile
               Directly  Contracted  Percent
City           Operated     Out      Savings
Los Angeles     $8.69      $4.45       49%
Philadelphia     9.67       2.41       75%
Dallas           7.89       3.59       54%
New Jersey       7.82       4.36       44%
Baltimore        9.74       5.33       45%
Denver           7.13       3.83       46%
Boston          10.16       2.61       74%

The advantage of contracting out is that transit agencies can use the operational savings to expand service into other areas. This is much more difficult if agencies must spend a large portion of their tax base on paying back the debt incurred to build rail transit.