Evaluation Transit Aid

The Experience with Transit Subsidies

Three arguments are usually offered in support of government assistance to mass transit. First, mass transit aid is said to reduce the level of automobile use and thereby to alleviate traffic congestion, the demand for costly urban highways, and automotive air and noise pollution and energy consumption. Second, subsidies to mass transit are thought to promote more concentrated, downtown-oriented forms of metropolitan development, with possible attendant infrastructure or energy savings. Finally, transit assistance is expected to benefit low-income house-holds, the elderly, and children—those too poor, too old, or too young to drive or own an automobile.

Reducing Auto Use

The failure of transit aid programs in the 1970s to increase transit ridership significantly was discouraging, especially given the rapid increase in government aid during the decade. Where transit ridership did increase, moreover, only a fraction of the new riders were former auto users; thus, the cost per auto trip avoided was often extremely high.

The new rail systems provided the most dramatic examples of the difficulties of reducing auto use at reasonable cost. Both the San Francisco Bay Area Rapid Transit (BART), opened in 1972, and the Washington D.C. Metrorail, opened in 1976, had anticipated that fares would cover at least operating costs and, in the case of BART, make some small contribution to construction costs as well. Construction and operating costs on both systems proved to be higher than anticipated, however, even adjusting for inflation, while ridership was overprojected by nearly a factor of two.^[24] In 1976, the average fare on BART was 72 cents, while the operating cost per rider was about $1.95 and the total public subsidies—operating and capital—amounted to $3.76 per rider. Since only 35

percent of all BART riders were former auto users (the remainder being former bus patrons or persons making new trips), the subsidy per auto trip avoided amounted to at least $10.74.^[25]

Metrorail has fared slightly better, perhaps in part because it is still under construction and only some of the most productive segments have been opened to date. When the commitment to build Metrorail was made in 1969, the cost of the planned 98-mile system was set at $2.5 billion. The 31-mile Phase III system, which was in operation by late 1979, alone cost over $3 billion to build (in 1980 dollars), and the cost to complete the entire system (since expanded to 101 miles) is now climbing toward $12 billion.^[26] With the Phase III system, total costs amounted to $2.73 per rail trip in 1980, two-thirds of which was for capital.^[27] Metrorail Phase III did reduce rush-hour auto trips to the downtown core by 4 percent.^[28] Since only 36 percent of all Metrorail riders were former auto or taxi users, however, the Metrorail capital cost per auto trip removed from the road amounted to $5.21, while the total cost (operating plus capital) might have been as high as $7.58.^[29] As the outer extensions of the system are opened, the cost per rider and per auto trip diverted are likely to increase.^[30]

Operating aid also brought disappointing ridership gains, despite the fact that the average operating subsidy per rider increased from 5 cents in 1970 to 76 cents in 1982. The small ridership increase was caused in part by the absorption of much of the aid in reduced productivity and higher unit costs rather than in fare reductions or service improvements. Between 1970 and 1982 the average wage for transit drivers grew 40 percent more than inflation, for example, while labor productivity, as measured in annual vehicle hours of service per full-time employee, fell by 20 percent.^[31] The rapid cost increases meant that the average real fare declined by only 19 percent from 1970 to 1982, while the number of vehicle miles of service offered by the transit industry increased by only 13 percent.^[32]

Even where operating aid led to significant fare reductions or service improvements, the ridership increases were often relatively small. In the Boston metropolitan area, for example, government operating subsidies for mass transportation increased from approximately $80 million in 1971 to $233 million in 1980, while passenger revenues remained relatively constant at $50–60 million per year. The operating aid was used to maintain existing service and keep the fare at 25 cents, despite rapid inflation in operating costs. Inasmuch as the consumer price index more than doubled bwteen 1971 and 1980, retaining the 25-cent fare amounted to reducing the fare by half (in real dollars) over the course of

the decade. Despite this substantial decline in price, ridership remained stable or declined slightly throughout the 1970s. In late 1980 and 1981, when financial realities finally forced Boston officials to raise fares, the response in ridership and auto usage was also relatively small.^[33] When bus fares increased by 100 percent and rail fares increased 50 percent in August 1981, for example, ridership fell by only 10 percent and car usage in the metropolitan area increased by at most 0.5 percent.^[34] Boston temporarily retained some mass transit riders by avoiding a fare increase during the 1970s, but the cost of doing so amounted to around $2 to $3 in added subsidy per passenger trip saved.^[35]

About a dozen metropolitan areas significantly increased their transit ridership during the 1970s, sometimes at a relatively modest subsidy cost of 50 cents or so per added rider. In most cases, however, fare reductions and service improvements were not the only factors contributing to the ridership gains. A few cities allowed their transit service to deteriorate greatly before public subsidies began; thus the increase was significant only compared to the very small base. In several other cities, policies to discourage auto use—such as parking taxes or controls, priority for buses in traffic, or exclusive bus lanes—appear to have contributed greatly to increasing patronage.^[36]

In short, the experience of the 1970s showed that it is difficult to expand transit patronage—and reduce auto use—much beyond transit's traditional markets. Transis has always been best in serving commuters to the downtowns of large and congested metropolitan areas, such as New York, Chicago, San Francisco, and Boston. The combination of high downtown parking fees and slow traffic speeds makes transit competitive with the automobile for these trips, particularly if transit can be protected from the congestion with either an exclusive rail or bus right-of-way or bus priority in traffic. The irony is that where transit contributes most to the alleviation of auto problems, it is probably least in need of government assistance. Where traffic congestion is severe and parking charges high, transit can offer relatively competitive service without significant government aid.

Concentrating Urban Development

The supporters of the new rail transit systems hoped they would lead to more concentrated forms of urban development as well as reduced auto use. A review of the experience of new rail systems in San Francisco, Washington, Montreal, and Toronto and major rail extensions in other cities suggests, however, that rail service is not sufficient to insure intensi-

fied development around suburban or outlying stations. A variety of other complementary factors are needed, including rapid economic and population growth in the metropolitan area and support from the community and local land use policies. In Toronto, for example, clusters of high-rise apartment buildings and retailing appeared around outlying stations, in part because rapid metropolitan growth encouraged new construction and local officials offered powerful zoning incentives for development around stations. With BART and Montreal's Metro, by contrast, there was little development around suburban stations, due in part to slower growth, community opposition, and difficulties of land assembly.^[37]

The principal land-use effect of these new rail systems appears to be in facilitating downtown high-rise office development. San Francisco, Toronto, and Montreal have experienced rapid downtown development since their new systems were completed; Washington, D.C., and Atlanta, where partial systems are now in operation, show signs of similar trends. Other non-rail factors probably contributed, however, since the downtown office booms often pre-dated the subway openings and other cities without new rail systems experienced similar developments.

If new rail systems promote downtown offices but leave suburban development dispersed, their land-use effects are similar to those of the radial expressways that transit advocates often oppose. By increasing the accessibility of the downtown to the suburbs, the rail system promotes more concentrated employment in the downtown and the displacement and dispersal of residences into the suburbs. The overall effect is probably to increase rather than reduce the average length of the commuting trip and to turn the downtown and its immediate neighborhoods into areas that are active during the day and have few residents and associated nighttime activities.^[38]

Helping the Poor

The experience of the 1970s also suggests that transit aid is not always an efficient or effective way to help poor people.^[39] Transit accounts for only a modest fraction of the trips poor people make. Although poor households are more dependent on mass transit than rich households, in 1977–1978 mass transit accounted for only 6.9 percent of the trips made by metropolitan residents with household incomes below $6000, a figure not much different from transit's 3.4 percent share of trips by metropolitan residents of all incomes.^[40] While 85 percent of all metropolitan households owned at least one car in 1977–1978, a surprising 54 percent of

households with incomes below $5000 and 83 percent of households with incomes between $5000 and $10,000 owned autos.^[41]

Poor people make up a small fraction of all transit riders, moreover. Travelers with 1977–1978 household incomes below $6000 accounted for only 25 percent of all transit trips in metropolitan areas, for example.^[42] The share of subsidy that goes to poor transit riders is probably even smaller, since low-income ridership tends to be concentrated on services where the subsidy per rider is relatively low. Subsidies per rider are typically three to four times higher on commuter railroad services than they are on buses or subways, for example, yet only 9 percent of all commuter rail trips were made by persons with 1977–1978 household incomes below $6000, while 38 percent were made by travelers with incomes above $25,000.^[43] Within rail and bus systems, transit subsidies tend to be higher for long-distance and peak-hour travelers, because the costs are higher while fares typically are not. Poor people ride much shorter distances than average because they live closer to the city center; the poor also travel disproportionately in the off-peak hours since many are unemployed.^[44]

Transit subsidies could better help the poor, of course, if they were targetted directly at poor travelers or the services they use. Virtually every transit system offers discount fares for handicapped and elderly persons, but these discounts help only slightly: 75 percent of the metropolitan poor are neither handicapped nor elderly.^[45] A more promising possibility is to subsidize poor users directly; poor persons might be issued discount fare identification cards or, to avoid the possible stigma of an identification card, could be sold monthly transit passes or a supply of transit tokens at a discount price. Targetting the subsidies to services such as bus, off-peak, and short-distance that poor people tend to use is less efficient but would be an improvement over the existing system. The federal government has sponsored several successful demonstrations of the possibilities of direct subsidies to poor people, but so far the practice has been adopted only in a handful of cities.^[46]

The Problems of Federal Aid

The Federal Rationale

The rationale for federal involvement in urban mass transit shares many of the weaknesses of the rationale for federal aid to urban highways. The argument most often cited in the early 1960s debates over the

initial federal capital grant program was the need to counterbalance federal highway aid. The federal and state highway trust funds, all financed with dedicated gasoline taxes, were thought to have induced state and local governments to channel too much capital spending into highways and too little into mass transit. Transit had declined because of undercapitalization, the argument continued, and federal transit aid was needed to correct the imbalance.^[47]

The failure of the transit investments of the 1970s to increase ridership significantly suggests that undercapitalization was probably not a major cause of the decline of mass transit patronage. Rising real household incomes, suburbanization of jobs and residences, and other demographic trends probably played more important roles in the postwar patronage losses. Even if local governments had seriously overinvested in highways and underinvested in transit, a massive new transit aid program may not have been the correct answer. By subsidizing both the highway and transit modes the federal government might reduce the balance between transit and highways only at the risk of overcapitalizing transportation in general. Reducing or eliminating the federal highway aid program might have encouraged more balanced spending on all forms of transportation.

Discretionary Capital Grants and the Rail Starts Problem

Whether mass transit was ever undercapitalized, the federal discretionary transit capital grant program now appears to encourage overcapitalization. The transit capital grant program has many of the same features as the Interstate Highway program. The federal government will pay up to 80 percent of the cost of a transit capital project (after 1982, 75 percent). Since grants are allocated for specific projects at the discretion of the Secretary of Transportation, moreover, there is no clear limit to the assistance a city may receive; the use of aid for one project does not preclude aid for others.

State and local governments have an obvious incentive to use the discretionary capital grant program to substitute federally paid capital costs for locally paid operating expenses.^[48] For example, one early study examined the possibility that transit authorities might save on bus maintenance costs by using federal capital grants to retire buses earlier than they would otherwise. That study never investigated whether transit authorities actually engaged in this practice but estimated that if they did, the waste would amount to more than 20 percent of the value of the federal capital grants.^[49]

Since most federal capital grants have been spent on rail transit, a most troublesome possibility is that local authorities may substitute federal capital grants for local operating expenses by building rail instead of bus systems. Among the advantages often cited for new rail systems is the potential for reducing labor costs by replacing several bus drivers with a single train operator. There are some offsets, of course, such as the added costs of maintaining track, signals, power distribution systems, stations, and other facilities not normally found on a bus system. In some situations, however, rail may offer savings in operating costs even though total costs, including capital, are higher.

The availability of federal aid has been a strong factor in the recent revival of rail transit. Rail systems are often classified as either heavy or light: on heavy rail systems, such as a conventional subway or metro, passengers board from platforms and power is distributed by a third rail; on light rail systems, such as streetcars or trolleys, passengers may board from street level and power is distributed by overhead catenary. From the 1930s through the late 1950s, rail transit mileage declined steadily in the United States as many cities replaced their streetcar lines with buses.^[50] When the federal capital grant program began in 1964, five United States metropolitan areas had rail systems that pre-dated World War II, Cleveland had opened a new heavy rail system in 1958, and San Francisco had already begun the construction of BART. Between 1964 and 1984, almost every older rail system extended its lines, four cities opened new heavy rail systems, one opened a new light rail system, and another five cities began building new light rail systems. As of 1984 at least 13 more cities were in various stages of designing or planning new rail systems, while virtually every existing rail system was planning extensions to existing lines (Table 7.8). While most of the rail construction in the 1960s and 1970s involved heavy rail systems, almost all the cities building or planning new systems in the 1980s are considering light rail.

Several studies have compared the operating and capital costs of using buses and heavy rail to serve metropolitan commuting trips. To insure that the cost comparisons are meaningful, these studies attempt to make the quality of service on the two modes comparable. While it is difficult to make the services exactly equivalent, the cost studies usually insure that the ratio of seated to standing passengers is the same on rail and bus and that bus travel times are comparable to heavy rail by, for example, operating the buses on an exclusive busway or an expressway whose access ramps are controlled to prevent congestion. The cost studies generally show that the bus is less expensive than heavy rail in radial corridors with

weekday peak-hour travel volumes below 15,000 transit passengers. When the peak travel volumes exceed 30,000 transit passengers, heavy rail is less expensive, and between 15,000 and 30,000 passengers the choice is usually determined by local conditions, such as the availability of a relatively free-flowing expressway that buses can use.^[51] Radial corridor volumes above 15,000 persons per peak hour probably are found only in metropolitan area with populations of at least several million, with high population densities and a strong downtown.^[52] Of the metropolitan areas that opened or began construction of heavy rail systems in the 1970s, probably only San Francisco and Washington, D.C., have corridor volumes that exceed the minimum heavy rail threshold. Even in those metropolitan areas, moreover, a careful bus-rail cost comparison would probably suggest much smaller heavy rail systems than the ones opened or planned.

The high cost of the heavy rail systems built in the 1970s stimulated interest in light rail systems in the 1980s. Unfortunately, there has been very little analysis of the comparative costs of light rail and bus.^[53] Light-rail proponents argue that it can be much cheapter than heavy rail because it can operate safely on city streets where building an exclusive right-of-way or subway would be expensive. Because the right-of-way need not be grade-separated, the argument continues, a light rail system does not need nearly as high corridor volumes to bring its per-passenger costs down to reasonable levels. If the light rail system is not grade-separated, however, the comparison bus service probably can also avoid expenditures for exclusive busways or similar facilities. Whether the savings in rail costs outweight the savings in bus costs may depend greatly on local conditions. Many of the metropolitan areas now building or planning light rail systems have such modest populations and densities as to make construction of even light rail systems very risky.

The high construction costs of some of the new rail systems built to date lend credence to the view that the total costs of many of these systems are greater than those of comparable bus systems. Even more striking is the possibility that at least some of the new rail systems may have increased rather than reduced locally paid operating costs. San Diego's new light rail system, for example, is widely touted as a success because farebox receipts cover 80 percent of its operating expenses, compared to only 40 percent on the San Diego bus system. The rail system consists of a single sixteen-mile line connecting the downtown with the southern suburbs and the United States–Mexican border, and it offers faster terminal-to-terminal travel times than the two bus routes it replaced.^[54] The

original bus routes were among the most heavily patronized in the San Diego system, however, and their farebox receipts exceeded operating expenses. Had the bus routes enjoyed the partially grade-separated right-of-way of the rail system, their travel times probably would have been comparable to rail and their operating profits even greater.^[55] Comparisons of bus costs for other larger rail systems are more difficult, since it is harder to identify the specific bus services they replaced. The steady increases in operating deficits in many of the metropolitan areas that recently opened rail systems suggests that the operating savings over buses may have been small.

Federal regulations requiring analysis of low-cost alternatives and restricting new rail starts to one line at a time have not been effective in limiting new rail construction.^[56] The 1976 requirement for alternatives analysis is credited with eliminating Denver's heavy rail proposal and reducing the size of the Buffalo light rail plan (when bus systems were found to offer comparable service at a fraction of the cost). Although the analysis requirement probably has discouraged or reduced the scale of other proposals, since the Denver and Buffalo decisions federal officials have not turned down any rail proposals after alternatives analysis, and the number of cities engaged in rail planning has increased.

As in the case of highways, the complexity of urban transportation planning helps to limit the effectiveness of federal rail transit regulations. Forecasting rail and bus patronage is a difficult art, for example, since ridership projections are sensitive to the detailed spatial distribution of future jobs and residences, which is itself uncertain. The non-rail alternatives can also appear relatively costly and unattractive if they are poorly or unimaginatively designed. Although experience has shown that many of the rail analyses overestimated patronage and understated costs, it is often difficult to demonstrate conclusively before the fact that a forecast is unreliable. Given the subtle judgment involved in accurately forecasting ridership and costs or designing alternatives, it is difficult for UMTA's small staff to detect or correct the potential errors in the alternatives analysis.

These problems are often further compounded by strong political support for rail from local officials and their congressional delegations. Local officials displeased with UMTA decisions are often quick to bring congressional pressure to bear on the agency, particularly when the UMTA position seems petty or unreasonable. The FY1982 and 1983 congressional earmarking of rail funds for specific cities is perhaps only the most extreme example of this problem.

In May 1984 the UMTA administrator announced that future rail-start decisions would be based on the added cost per new transit rider attracted, using a low-cost bus alternative as the baseline for measuring added rail costs and ridership. The hope is that the figures will embarrass Congress and discourage future earmarking for particularly cost-ineffective rail projects. The cost-per-rider estimates will be based on data from the alternatives analyses, however, which will increase the need for UMTA to monitor the quality of these studies.^[57]

Incentives of Operating Aid

The principal debate over federal operating assistance focuses on the extent to which it either substitutes for state or local aid or distorts state and local decisions in undesirable ways. Federal assistance was originally restricted to capital expenses in 1964, for example, partly out of fear that operating grants would reduce incentives for transit managers to control labor costs and improve productivity. Capital grants do not eliminate these incentives altogether, of course, since they free local resources that otherwise might have been absorbed by capital needs for operating expenses. With operating aid, however, the incentives might be more direct and obvious.

The relationship between operating aid and efficiency has received increasing attention because of the rapid inflation in transit operating costs during the 1970s. A careful study of the real (net of general inflation) increase in United States bus transit operating deficits between 1970 and 1980 found, for example, that increasing real labor compensation rates or declining labor productivity accounted for 48 percent of the deficit growth, while real fuel-cost increases accounted for only 11 percent of the deficit growth, service expansions only 19 percent, and fare reductions only 22 percent.^[58] Several cross-section and time series studies of transit agencies have also found that higher levels of government operating subsidies are strongly associated with higher wage levels and lower labor productivity. Correlation does not prove causation, however, and none of these studies has been able to demonstrate conclusively that higher subsidies caused operating cost increases rather than vice versa.^[59]

A more recent concern is the possibility that federal operating grants may encourage expansion of transit service into markets where transit has little chance of capturing significant ridership. Many smaller metropolitan areas greatly expanded their transit service during the 1970s even though low levels of auto congestion and low population densities make

it difficult to attract ridership. Some larger metropolitan areas greatly increased suburban and crosstown services, where the possibilities for attracting additional patronage at reasonable subsidy costs is probably also limited.

The cost inflation and the expansion of underutilized services encouraged by federal operating aid may well be centered on small, low-density metropolitan areas. State and local government aid accounts for 70 percent of all operating assistance for the United States transit industry as a whole, far less than the 50 percent minimum match required for federal grants. The formulas for distributing federal operating aid also provide less assistance per rider to many of the largest and most transit-oriented metropolitan areas; in the big Eastern and North Central cities such as New York, Chicago, Boston, and Philadelphia, federal aid accounts for less than 20 percent of all operating assistance. In many smaller and low-density metropolitan areas, federal aid is at the 50 percent limit and its potential effect on service and efficiency decisions is much greater.^[60]

Federal officials have rejected suggestions that they monitor the services or efficiency of operating grant recipients, in part for lack of simple and reasonable standards. The best remedy would probably be to increase the local matching requirement or to change the formula to favor the larger, transit-oriented cities more. The operating aid formulas were changed in 1978 and 1982 to increase the large cities' share of funding, but the need for broadly based congressional support has insured that many small, low-density metropolitan areas still receive a disproportionate share of aid.^[61]

Financial Effects on Urban Governments

Federal transit aid is targeted far more heavily at urban areas than federal highway aid, and urban taxpayers almost certainly receive more in federal transit grants than they pay in federal taxes to support the program. In 1980, for example, 34 states—all predominantly rural—received less than $10 per capita in federal transit aid, while only 8 states—all with cities that were building or expanding rail systems—received over $20 per capita.^[62] Since all federal transit aid was financed out of general revenues in that year, the federal tax burden for the program is probably far more equally distributed and urban areas are the net gainers.

As in the case of highways, however, the benefits to urban residents from a federal transit grant may be smaller than the dollar value of the

grant. To the extent that federal grants encourage too many rail transit extensions, expansion of underutilized services, or reduced productivity, the value of the grant to the recipient may be significantly lower than its dollar value. Capital grants may cause additional financial strains for urban governments if the capital projects increase rather than reduce local operating costs. The experience of the San Diego light rail system suggests, for example, that the construction of some rail systems may burden local governments with increased transit operating deficits.

The financial effects of federal transit aid also probably should not be viewed separately from those of federal highway aid. If urban representatives in Congress trade their support for highway aid in return for rural support for transit aid, then it would be impossible to eliminate or reduce the federal highway program without making comparable reductions in the federal transit programs. One recent study of the state-by-state allocations of federal transportation grants and taxes suggests that the urban bias of transit aid is not enough to offset the rural bias of highway aid. Of ten highly urbanized states, only two received more in highway and transit grants than they paid in federal highway and other taxes to support the programs. Almost all the states that received more in federal highway and transit grants than they paid in federal taxes were rural.^[63]