Vanishing Automobile Update #40

Federal Transit Data Show That Rail Doesn't Work

July 23, 2003

In June, 2003, MaryPIRG published a naive and simplistic report claiming that Rail Transit Works: Light Rail Success Stories from Across the Country. The first problem with the report, as Wendell Cox notes, is that two of the "light rail success stories" involve heavy rail (subways), not light rail.

The report also relies on transit agency propaganda about ridership projections and claims. For example, MaryPIRG says that "planners expected 100,000 passengers daily on a new section of the Red Line and had to add cars when ridership rapidly reached 120,000."

Thomas Rubin points out that the original ridership projections for the Red Line were far higher -- 376,000 per day -- than the projections made just before it opened. Moreover, he adds, the Los Angeles transit agency has recently admitted that it greatly overestimated ridership on the Red Line, and now admits that daily ridership has never in any month averaged 100,000.

In the end, the numbers game of projections-vs.-actual is the wrong way to look at transit. Transit agencies once greatly overestimated ridership thinking that they needed to do so to convince local governments to support rail projects. Now they know that many local officials support rail projects no matter what the numbers say, so they often project low numbers just so they can be on target when the rail line actually opens. Many politicians are so innumerate that they are impressed when transit advocates say "It will take 9,000 cars off the road each day!" even though 9,000 auto trips is less than a drop in the bucket in a major metropolitan area.

The real questions that need to be asked are: How much does rail cost? What do you get for your money? And what would you get if you spent that money on something else such as bus or highway improvements? The MaryPIRG report, of course, ignores these questions.

To help people put these numbers into perspective, I have posted a summary of 2001 National Transit Data Base (NTBS) numbers for all U.S. commuter-rail, light-rail, and heavy-rail lines. (Trolley lines such as the New Orleans streetcar are not included.) You can download this Excel spreadsheet from the American Dream Coalition web site.

Here is what you will find on the spreadsheet:

All of the above are taken straight from the National Transit Data Base, meaning they are data reported to the Federal Transit Administration by the transit agencies themselves. "Directional route miles" are miles in each direction. So if you have a 15-mile light rail line, it has 30 directional route miles since the rail cars (probably) go in both directions on all 15 miles.

Note that capital costs are NOT included. You can get some estimate of capital cost for light-rail projects from You will have to get heavy-rail and commuter-rail costs from the agencies themselves. The average light-rail project cost about $40 million a route mile, but some cost as little as $10 million a route mile. Heavy rail is about twice as expensive and commuter rail is highly variable but usually less expensive than light rail. When calculating costs per directional route mile, remember that there are twice as many directional route miles as route miles.

The next few columns are calculated based on the previous data.

Column M is automobile passenger miles for each urban area. It is taken from Highway Statistics 2001, table HM-72, which lists daily vehicle miles traveled. This number is multiplied by 365 to get the annual number and multiplied again by the average occupancy of 1.6 to get passenger miles. NOTE: These numbers aren't strictly additive because several urban areas are listed more than once. I've corrected for this in the totals in rows 55-58.

Column N is transit's share of motorized passenger travel. It is based on total transit passenger miles in each urban area (not otherwise shown in the table) divided by transit passenger miles plus auto passenger miles.

Column O is the share of motorized travel held by each rail system. It is the passenger miles (column E) divided by auto plus transit passenger miles.

Column P is passenger miles divided by trips, resulting in average trip length. This is around 23 miles for commuter rail, 5 miles for heavy rail, and 4 miles for light rail.

Column Q shows the number of daily passenger miles per freeway lane mile in each urban area. Like column M, this is taken from table HM-72 of Highway Statistics 2001. See note in column M above.

Column R is rail passenger miles/directional route mile as a percent of freeway passenger miles per lane mile (J/Q).

I used columns T, U, and V to calculate the correct totals for columns M and Q. However, they only come out correctly when the main list is sorted by column A. You can ignore them.

Rows 55-58 give the totals or averages for light rail, commuter rail, and heavy rail, plus a grand total for all three modes.

Here are some points you can make based on these data.

Buses Can Carry More People Than Most Light-Rail Vehicles Carry

The average light-rail occupancy is only 27 people. Many express and commuter bus lines carry more people per bus than this. Particularly pathetic, at average occupancies of 10 and 15 people respectively, are the light-rail lines in New York (actually the Hudson-Bergen line in New Jersey) and San Jose.

Although occupancies will obviously be higher than average during rush hour, and many light-rail lines are run with two or more cars hooked together, you can also run more than one bus when you need to. Any light-rail line with an average occupancy of less than 40 people -- meaning all but one system shown here -- could more effectively be handled by buses. Heavy-rail and commuter-rail lines are more often run in lengthy trains so the occupancy figures are less significant.

Information about buses can also be found on the American Dream Coalition web site, an Excel file in an almost identical format as the rail table but for all modes of transit and all agencies that report to the Federal Transit Administration. This Excel file is about 370 kb in size and contains data for 1,145 agencies and modes, compared with just 52 on the rail spreadsheet.

Buses Cost Less Per Trip Than Most Light- or Heavy-Rail Lines

Light- and heavy-rail lines can be evaluated by comparing cost per trip with similar costs for buses. Nationwide, the average cost per bus trip is $2.40, but rail lines usually aren't built in average corridors: they are built in the most heavily used corridors in a region. We can get an idea of the cost per bus trip in heavily used corridors by looking at bus agencies that tend to focus on such corridors. These operators, such as the Trans-Hudson Express in New York, Laidlaw Transportation in California, and San Francisco Muni, have costs of about $1.50 to $1.60 a trip.

All but two light-rail systems (Boston and Los Angeles) and all but three heavy-rail systems (Boston, New York, and Philadelphia) have costs that are greater than this. This suggests that buses could cost-effectively replace rails in all other cities.

Buses Cost Less Per Passenger Mile Than Most Commuter-Rail Lines

Because commuter-rail trips are longer than other rail trips, cost per passenger mile is a better measure of their cost than cost per trip. These can be compared with the costs of bus operators that focus on long-distance commuter trips, such as Golden Gate in San Francisco, Loudoun County in Virginia, and Hudson Transit in New York-New Jersey. These operators spend an average of 15 to 20 cents per passenger mile on operating costs.

Only New York has commuter-rail costs in this range. Boston is 23 cents a passenger mile, Chicago is 26 cents, and most of the rest are 28 to 36 cents. Dallas (in sharp contrast to Ft. Worth) is $1.00. Most of these commuter rail lines could be handled by buses at a lower cost.

One Freeway Lane Carries Far More People Than Any Rail Line Outside of New York Subways

Rail advocates often like to say that a rail line can carry as many people as an eight- (or sometimes twelve-) lane freeway. But column R in the rail table shows that no rail line outside of New York City subways carries as many people as a single freeway lane.

Note that the number of people carried on a freeway lane is highly variable. Los Angeles freeway lanes, for example, carry 50 percent more people than those in New York, partly because they are more modern and thus have greater capacities. And remember that, unlike most rail transit lines, freeways also carry freight.

Rail Transit Only Works When You Have Millions of Jobs in One Place

Wendell Cox likes to say that transit is really about downtown. In my presentations, I point out that transit is really only an important mode of travel in urban areas with a huge number -- roughly a half-million or more -- centrally located jobs. That pretty much limits it to six urban areas: New York, Boston, Chicago, San Francisco, Philadelphia, and Washington.

The numbers in the table show that rail transit does pretty well in New York and makes a marginal showing in the other five cities. Elsewhere, it is insignificant when measured by market share of travel, and it has particularly failed in cities that have mainly grown in the last fifty years, such as Dallas, San Diego, and San Jose.

Light-Rail Transit Doesn't Even Work in the New York Urban Area

Notice also that the transit that does work in New York and the other five cities is mainly heavy rail and commuter rail. Light rail doesn't even work in the New York urban area, as shown by the Hudson-Bergen failure (4.2 percent of a freeway lane!). The other New York-area light-rail line in the table is a Newark rail line, sometimes called a subway, originally built in 1935. It also has pretty low ridership and certainly didn't deserve the $200 million that New Jersey Transit recently put into it.

In sum, light rail doesn't work, and neither, outside of the New York urban area, do heavy rail or commuter rail.

Thoreau Institute | Vanishing Automobile | Vanishing Automobile Updates