Forecasting the costs and performance of major infrastructure projects with a reasonable degree of approximation is very difficult. Despite relying upon the finest technology, the most adept computer models and the sharpest minds, projections for many major infrastructure projects have been exceedingly inaccurate.

• The new Denver International Airport was to have cost $1.7 billion when it was approved for construction. Eight months into construction, costs had increased 60 percent. After opening 16 months late, the cost had escalated to $4.8 billion (each month of delay cost nearly $20 million), a construction related cost overrun of $3.1 billion. The higher levels of bonded debt would require approximately $2 billion in additional interest payments, raising the cost overrun alone to $5.1 billion --- a more than 300 percent increase over the cost estimate on which the decision to proceed was made (all figures in 1996$).⁽¹⁾

• The Channel Tunnel between England and France was to have been built for $7.8 billion. Costs escalated to $18.6 billion --- an increase of nearly 140 percent (not including the higher cost of interest due to larger borrowing requirements than projected).⁽²⁾ After opening a year late, its first year of operation produced a loss of $1.5 billion. The competitive response of cross-channel ferry operators reduced tunnel traffic to below expectations. After failing to pay interest on its debt for more than a year, a financial bail-out was negotiated with creditors converting half of their loans to equity. This project was privately financed as both the British and French government were unwilling to provide either public subsidies or debt guarantees.

• The cost of Boston's Central Artery/Tunnel expressway project has nearly doubled from a projected $5.5 billion to $10.4 billion (1996$). The project is scheduled to open six years late in 2004.⁽³⁾

• Amtrak, which was created to salvage the U.S. passenger rail system, was intended to achieve profitability shortly after its establishment in 1971. Yet Amtrak continues to post significant losses and taxpayers subsidies have exceeded $15 billion. Amtrak claims that fares and other commercial revenues will eventually exceed its operating, but not capital costs. The United States Government Accounting Office has found that Amtrak's financial condition is deteriorating and that it is unlikely to earn commercial revenues that exceed its operating costs, much less its capital costs.⁽⁴⁾ Amtrak is now seeking a new federal tax.

• Large urban rail projects have consistently cost more to build, cost more to operate, attracted fewer passengers and generated less passenger revenue than projected. During the 1980s, federally financed urban rail projects cost 46 percent more to build, and 78 percent more to operate than projected. Ridership averaged 59 percent below projections. So few new passengers were attracted that the annual cost per new passenger exceeded the cost of leasing a car in virtually all new systems.⁽⁵⁾ In response to ridership shortfalls, transit agencies have begun to issue radically reduced ridership estimates shortly before system openings.

• The more recently completed Los Angeles-Long Beach light rail project was to have cost $210 million when the Los Angeles County Transportation Commission⁽⁶⁾ decided to proceed with the project (1981). Costs rose to $500 million by the time final plans had been formalized and nearly $900 million when completed, a cost escalation of more than 300 percent. Annual operating costs were 150 percent above projection.⁽⁷⁾

• Miami's Metrorail cost 33 percent more to build and 42 percent more to operate than projected. Daily ridership was to have been 239,900 by 1988 but by 1995 was only 47,800, 80 percent below projection.⁽⁸⁾ As a result, the cost per rail passenger was nearly nine times the projection.⁽⁹⁾ Miami's Metro mover (people mover) cost 106 percent more to build and 84 percent more to operate than projected. Daily ridership was to have been 41,800 by 1988, but was under 13,300 in 1995, 68 percent below projection despite a more than doubling of the route's length.⁽¹⁰⁾ As a result, the cost per rail passenger was nearly seven times the projection.⁽¹¹⁾ These two infrastructure projects were to have substantially increased transit ridership in Miami. By 1995 ridership was 65 percent lower than the level predicted for 1988.⁽¹²⁾

Inaccuracy in highway usage forecasts illustrates the difficultly inherent in projections even where there is a wealth of experience. A recently released National Research Council study of the international experience notes that rail projections are "more problematic" than road projections. The study finds:

... the main lessons are that cost overruns of 50 to 100 percent are common for large transportation infrastructure projects: overruns above 100 percent are not unusual. Traffic forecasts that are off by 20 to 60 percent are also common ... The result is that decisions based on misleading forecasts --- often presented to parliaments, to other decision makers, and to the general public --- may lead to a misallocation of funds and to underperforming projects during construction and operation.⁽¹⁴⁾

... forecasts that underscore a priority which is out of political favor are likely to be ignored, whereas forecasts that support politically favorable positions are likely to be embraced.⁽¹⁵⁾

... most of the forecasts used in the planning of America's rail transit systems are statements of advocacy, rather than unbiased estimates.⁽¹⁶⁾

Major infrastructure projects can take on a "life of their own." The experience demonstrates that, once authorized, even cost escalation that double or triples the cost of a project will not result in its cancellation.

There will always be detailed explanations for cost escalation and failure to attract projected ridership and revenue --- some are more valid than others. But in publicly financed projects the "bottom line" is the same --- the cost of unreliable forecasts is paid by users. Or, if public subsidy is involved, the excess cost is paid by the taxpayers.