The Case for Bus Rapid Transit

Much has been written about America’s love affair with the automobile. That’s why Nieman Reports is devoting the bulk of this edition to highways. What has been overlooked has been America’s love affair with electric trains — what I have called the “Lionel Complex.” If the object of this passion were just model trains that speed around Christmas trees we could understand the pleasure that adults, as well as children, enjoy in running them. Unfortunately, the passion extends to full-size trains that urban planners have foisted on city after city in an alleged effort to solve commuter problems. The result has been failure and the cost has been billions of dollars.

Los Angeles is the prime example of choosing the wrong technology for improving local transportation needs. Despite proof that fare reductions would increase bus ridership dramatically, officials pushed ahead with construction of a hugely expensive heavy rail line and an extensive light rail system. The choice of rail to serve Los Angeles’s dispersed employment and resident pattern was clearly a mistake, and I know of no competent transport planner who thinks the choice made any sense. To make matters worse, Los Angeles’s error has been, and continues to be, repeated in city after city in the United States and in other parts of the world. More than $200 billion of government subsidies has been spent in a largely unsuccessful effort to increase transit ridership. A large fraction of this money has been wasted. Portland, Ore., San Diego, Houston, Atlanta and several other cities duplicated Los Angeles’s success in increasing transit ridership by spending taxpayer dollars to reduce fares and to expand bus service. With the exception of Houston, all of these cities abandoned these successful policies in favor of the construction of costly and ineffectual rail systems. Small ridership increases or declines were the result.

Nothing can be done about these past mistakes, although a good case can be made for discontinuing rail service in a number of cities. The argument for not extending these systems or building new ones is even stronger. In Los Angeles and elsewhere, transit operators should stop throwing good money after bad in an effort to make ill-conceived and inappropriate rail systems “work.” They should instead use available transit subsidies to expand and improve their bus systems, which remain the backbone of their transit systems.

Local and national media must accept a significant portion of the blame for the nation’s costly love affair with electric trains. In city after city, metropolitan dailies and TV stations have been uncritical boosters of ill-conceived rail proposals and have grossly mislead their readers and viewers about the benefits and costs of these investments. While there is nothing they can do about past sins, it is not too late to make amends. See, for example, editorials in The Houston Chronicle from February through August this year supporting Dallas’s light rail system.

Here are facts for editors and reporters to keep in mind as their cities consider transit improvements:

In 1947 transit ridership in the United States reached an all-time high of 22.5 million annual hoardings (unlinked trips). Thereafter, it declined, first precipitously and then more gradually, until 1972 when it was less than a third of its 1947 level.

Between 1972 and 1980, aided by more than $200 billion in government subsidies and by a temporary boost from higher gas prices and shortages, hoardings increased by 26 percent. After 1980, ridership declined and then recovered until a slower growth in real subsidies and continued suburbanization led to small, but persistent, declines. For the entire period from 1972 to 1995 annual hoardings grew by 15 percent. This period, however, was characterized by large increases in rail operations, which have higher transfer rates. When the hoardings data are corrected for this spurious source of ridership growth, annual transit trips grew by less than 5 percent.

Long-term trends in land use patterns and urban tripmaking, which were interrupted first by the depression and then World War II, are largely responsible for postwar declines in transit ridership. For the most part, these trends were efficient responses to changes in technology and rising real per capita incomes. At the same time mistakes in public policy and errors by transit operators in their spending of subsidy dollars caused the declines in ridership and the associated increases of public subsidies for transit to be larger than they needed to be. In the case of the transit industry, the principal errors were the replacement of private by public ownership and the use of disproportionate amounts of available subsidy dollars for the construction and operation of costly and ineffective rail transit systems. Alternatives were available. Subsidy dollars could have been used to pay private transit operators to provide unprofitable, but desirable, services, and the huge sums spent to build and operate rail systems after World War II could have been used to buy more vehicle miles of bus service and lower fares.

The most serious error of public policy, however, was arguably the failure of governments, which are the principal providers of highways and public transport services, to charge urban tripmakers prices that reflect the long run social costs of providing urban transportation facilities and services.

With one exception, scholarly studies broadly agree with the following conclusions about the costs and performance of alternative public transport technologies:

In “Urban Rail in America,” Boris Pushkarev and Jeffrey Zupan of the New York Regional Plan Association concluded that new rail transit systems would have lower costs than bus transit in a wide range of circumstances in United States urban areas. Their analysis, however, has been discredited by Don Pickrell, Chief Economist, John A. Volpe National Transportation Systems Center, Cambridge, Mass., who found they had: (a) seriously underestimated both the capital and operating costs of new rail systems relative to bus systems, and (b) used optimistic ridership projections when they applied their cost estimates to particular metropolitan areas. When Pickrell’s corrections are taken into account, the differences between the Pushkarev-Zupan results and those of other scholarly studies disappear. In addition, Pickrell failed to mention, or overlooked, the fact that the Pushkarev-Zupan analysis assumed commuter buses operate on congested streets and roads at an average speed of only 12 miles an hour. Their cost comparisons are much less favorable to rail transit when express buses speeds are used.

During the past quarter century new light and heavy rail systems have been built in nearly two dozen U.S. cities. An Urban Mass Transit Administration study of 10 federally funded rail systems found that in every case the planners of these systems seriously overestimated future ridership and seriously underestimated their operating and capital costs. Forecasts of total costs per trip ranged between $1.53 to $3.04 per rail passenger, while actual total costs per rail passenger for the same eight systems varied between$ 5. 06 and $16.77. Actual total cost per rider as a percentage of projected total cost per rider similarly ranged from a low of288 percent for Washington to a high of 969 percent for Buffalo (all figures are in 1988 dollars).

The findings do not in themselves directly prove anything about the relative cost-effectiveness of exclusive busways and rail systems. They are relevant, however, for at least three reasons. First, they provide strong confirmation for the widely held view that rail system planners typically use optimistic assumptions when they develop rail system costs. Second, there is substantial evidence that rail system planners frequently “gold-plate” exclusive busways and employ a variety of other assumptions that reduce the effectiveness of bus alternatives. Finally, over-prediction of future ridership biases system choices toward rail. Overly optimistic forecasts of future transit use favor rail alternatives because of their higher capital costs and because advocates of these systems often rely on a variety of ad hoc arguments to justify their construction. One of the most common of these arguments is to assert that central area streets would be unable to accommodate a huge projected numbers of buses. The large number of buses projected to use these streets in some future year is, of course, a result of overly optimistic projections of transit ridership and central area employment.

Proponents of rail rapid transit systems have used extravagant projections of future ridership to justify the large capital costs these systems entail. In fact, because new rail rapid transit systems are typically built in well-developed transit corridors, where the new rail lines replace the most heavily traveled bus line or lines, they attract relatively few new riders. There may be benefits from building new rail transit lines in such situations, but a large increase in the number of new transit passengers is not among them. Any growth in ridership that occurs would have to be due to reductions in door-to-door travel time, which in most instances are modest or nonexistent.

The El Monte Busway in Los Angeles, the Shirley Highway in the Washington area, and Houston’s North Contra-flow lane had much larger impacts on transit ridership than new rail systems have had. There are four principal reasons why these busways induced far more growth in transit ridership than new rail systems. First, they provided large savings in travel time relative to the situation that existed before they were implemented. The impact of new rail systems on door-to-door travel times has been far more problematic. Second, each was implemented in a rapidly growing and heavily congested corridor. Third, in contrast to most new rail systems, which have been constructed in existing well-developed transit corridors, the three busways referred to above and most bus-carpool facilities have been designed to serve new transit markets. Finally, the implementation of the busway was typically accompanied by a major expansion in transit service levels and by the provision of transit service to areas that were not served previously.

None of the recently constructed rail systems uses more than a fraction of its peak hour capacity to accommodate peak hour demand. In contrast to light and heavy rail lines, where expensive excess capacity must go unused, carpools, vanpools and other high occupancy vehicles can use the extra capacity of bus-carpool lanes. If transit ridership increases, a larger fraction of the capacity of these facilities can be allocated to buses.

The advantage bus-carpool facilities have in sharing capital costs is even more evident in the case of I-66, the HOV Parkway between Northern Virginia and the District of Columbia. Both peak direction lanes of this four-lane parkway are used as a bus and 3+ carpool facility during the morning and evening rush hours. The use of I-66 by buses, moreover, is quite limited. Only 19 buses per hour, carrying under 700 passengers, used I-66 during the morning peak hour in the year shown. If transit had to bear the entire cost, the capital cost per daily trip would be almost $52,478. But when total daily trips are used in the denominator, construction costs per trip plummets to $3,523.

It is tempting to conclude that because buses using High Occupancy Vehicle (HOV) lanes are able to share excess capacity with carpools, it is always better to build shared bus-carpool facilities than exclusive busways. The Ottawa and Pittsburgh experiences, however, suggest such a conclusion would be premature. First, their costs per total trip are very similar to those of the El Monte, Shirley Highway and Houston bus-carpool facilities. In addition, they differ in important respects and serve rather different markets than bus-carpool facilities.

Both the Ottawa and Pittsburgh busways have on-line stations and carry significant amounts of walk-on traffic. In this respect, they closely resemble many light and heavy rail systems. Both busways, however, are also used by large numbers of express bus routes that collect their passengers at park-and-ride lots or in suburban residential areas and use the busway for a fast nonstop trip to the central area. While it is possible to have on-line stations on a shared bus-carpool facility (the El Monte Busway has one, for example), the provision of on-line stops (stations) on shared bus-carpool facilities creates numerous engineering, safety and operational problems. Overcoming these problems may significantly increase capital costs. Where transit demand is sufficient to use a large fraction of an exclusive busways capacity as in Ottawa, or in the case of the South Busway in Pittsburgh, it may simply not be worth it to allow carpools and van pools to use them.

Successful exclusive busways are likely to be short and be located in built-up and fairly dense areas with relatively high levels of transit ridership. In such situations, they may provide a superior service for those boarding at busway stops and provide a reliable and high-speed right-of-way for express services from suburban residential areas. It should be understood, moreover, that the choice need not be all or nothing. A well-designed, high-performance bus system might include both exclusive busways and shared bus-carpool facilities. Where an extensive network of regional expressways already exists, it almost certainly makes sense to include a number of miles of shared, bus-carpool facility in the system design. Within existing built-up areas, however, where densities are higher, and where there is a potential for walk-on ridership, exclusive busways may be the preferred solution. This is particularly true where there exist, as in Pittsburgh, portions of underutilized or abandoned rights-of-way that are too narrow for a major highway but have sufficient width for an exclusive busway.

When construction costs per trip for both light and heavy rail systems are compared to the same statistics for exclusive busways and shared, bus-carpool facilities, the situation is markedly different. Mean capital costs per transit trip for four light rail systems are five times as large as the capital cost per transit trip for the Pittsburgh exclusive busways and nearly 10 times as large as for the Ottawa exclusive busways. Finally, the mean capital cost per total trip of the same four light rail systems is nearly eight times as great as those of the El Monte Busway and Houston Transitways and almost ten times as great as those of the Shirley Highway HOV lanes.

Advocates of light and heavy rail systems frequently acknowledge the higher capital costs of these systems, but contend that these higher capital costs will be more than offset by lower operating costs, arguing that rail costs less to operate because fewer drivers are required. When the cost of maintenance labor and station personnel are added, this supposed operating cost of rail becomes much smaller and in many situations disappears. Rail advocates also frequently compare the per trip costs of a particular rail line with average bus system costs. Such comparisons, however, are highly misleading as new rail lines typically replace the most productive bus routes (that is why they are built there) and rail often requires low productivity feeder bus routes. When the costs of providing comparable service are compared and when the impact of rail on bus system costs are properly accounted for, bus operating costs are frequently lower than rail.

Several authors have questioned the large sums of public money that are used to build new rail systems or to extend existing ones and have suggested these subsidy dollars might have been better spent on improvements and expansion of existing bus services. Los Angeles and Atlanta are but two examples.

Thomas A. Rubin, Assistant General Manager of the Alemeda-Contra Costa Transit District, and James E. Moore II, an Associate Professor at the University of Southern California, document one of the most grievous examples, where the decision to build rail transit came at the expense of largely low-income bus users and resulted in much smaller increases in transit ridership than if the same amount of money had been spent for lower fares or more bus service. They provide the following analysis of what they refer to as “the most successful transit ridership experiment in United States history.” Specifically, they describe the impact on ridership of a three-year fare reduction from 85¢ to 50¢, a subsequent increase to its former level in 1986 and then to $1.10 in 1990.

Over the three years of the 50¢ fare program, District transit ridership rose over 40 percent, and was still increasing in the last month of the experiment. Very little about the bus system was changed except the fare. Revenue service miles increased only 1.5 percent, including special service added for the 1984 Los Angeles Olympics. Rubin and Moore tell what happened:

"Beginning in fiscal year 1986, the Proposition A funds that had been used to subsidize the 50¢ bus fare were reallocated to rail construction. Fares were increased to 85¢ in fiscal year 1986 and then to $1.10 in fiscal year 1989. By fiscal year 1990, ridership had decreased by over 96 million passenger hoardings per year, or 19.3 percent.

“The decline in bus ridership coincided exactly with the beginning of the Los Angeles County rail transit construction program. The funds transferred away from the fare subsidy program paid for about 35 to 40 percent of the reported construction costs of the Blue Line. In fiscal year 1995, Blue Line ridership hit 12 million passenger boardings, each at a public sector operating subsidy of more than three times higher than that of the average bus passenger. Thus the money that had been used to move people on buses was used to build rail projects that will never be able to move more than a small fraction of the number of passengers already lost from the bus systems.”

Atlanta’s experience was strikingly similar. The Metropolitan Atlanta Regional Transit Authority, which came into being with the public purchase of a private company in 1972, achieved a 38 percent increase in linked trips during its eight years of all-bus operations. During the subsequent 13 years, after it began operating its heavy rail system, linked trips grew by three percent.

The large difference in ridership growth between MARTA’s periods of all-bus and rail-bus operations is easily explained. There is overwhelming econometric evidence that decreases in real fares and increases in vehicle miles of service produce increases in transit ridership. In contrast, there is essentially no empirical support for the view that rail transit per se increases transit ridership. During its period of all-bus operations, MARTA reduced real fares by 61 percent and increased vehicle miles of service by 57 percent. During the subsequent period as a rail-bus system, it increased real fares by 61 percent and vehicle miles of service (rail plus bus) by 1.5 percent.

Using a transit ridership model, I projected what MARTA’s cumulative ridership during 1980-93 would have been if it had spent its subsidy dollars on further bus system improvements rather than building its rail system. One of the more interesting of 10 alternative all-bus scenarios indicated that if MARTA had simply kept real fares at their 1978 level and operated an all-bus system with the same net operating deficit as its bus-rail system, total ridership (linked trips) would have been 9 percent greater. The cumulative total cost (operating cost plus annualized capital cost) for this all-bus alternative, moreover, would have been only 31 percent as large as MARTA’s expenditures for the same period. A second all-bus scenario, which assumed 1978 real fares and cumulative total expenditures that were slightly less than MARTA’s actual ones, yielded a total 1980-93 ridership that was more than twice MARTA’s actual ridership. These and similar analyses provide compelling support for the view that if MARTA had continued to pursue the combination of low fares and service expansion it implemented during its first nine years of existence, instead of choosing rail, it could either have bought much larger increases in transit ridership with the same amount of money or the same ridership with less than one-third its actual expenditure.

In spite of the steadily accumulating evidence that the planners and other proponents of proposed light and heavy rail systems systematically underestimate system costs and ridership, proponents of these systems continue to argue that building them will reduce congestion and improve air quality since electric powered rail systems do not emit pollutants. These claims ignore the pollutants produced by electric power plants and fail to consider significant reductions in bus emissions that can be obtained at modest cost from “clean” burning buses. The latter improvements are particularly important because new rail systems have very little impact on bus miles. More importantly, reductions in emissions and in congestion from transit system improvements depend primarily on reductions in the number of cold starts and vehicle miles of travel by passenger cars. Moreover, as I have discussed, spending subsidy dollars on bus system improvements, particularly by reducing fares and increasing vehicle miles of service, provides substantially larger reductions in cold starts and vehicle miles of passenger car use. As a result, serious analyses of the reductions in congestion and air pollution that result from alternative transport system improvements strongly favor bus system improvements over spending for costly and ineffective rail systems.

The lesson for journalists in metropolitan areas considering changes in rapid transit is obvious: study the facts. Don’t let transit operators throw good money after bad in an effort to make ill-conceived and inappropriate rail systems “work.” It is clear that they should use subsidies to expand and improve their bus systems. We are not dealing with Lionel train sets. We are dealing with real economic facts of life.