Why Do Some Toll Roads Fail?
By Douglas M. Fried and Jeremy S. Rosenshine
The toll road landscape includes a number of failed projects. Sponsors of new projects — and the banks that lend to them — can learn valuable lessons from what went wrong in these earlier projects. This article looks at four problem areas: poor traffic forecasting, inadequate legal foundations, politically-motivated sovereign actions, and unanticipated events.
Poor Traffic Forecasting
One of the most important elements in evaluating the viability of a toll road project is determining whether drivers will use the road and to what extent they will be willing to pay a toll. These questions are extremely difficult because the answer is based on a prediction of human behavior. Sponsors and lenders will typically evaluate a broad range of statistical and other data to predict the use of a toll road. Traffic consultants will review traffic trends in the country and particular region where the road is to be located, congestion levels, land-use trends, current and projected levels of community development, socio-economic data and other information.
One area of vulnerability associated with faulty traffic forecasts is reliance on traffic patterns of existing roads that are irrelevant to predicting traffic patterns of the planned toll road. For example, the lower than anticipated traffic at the Garcon Point bridge in Santa Rosa County, Florida, which opened to traffic in May 1999, is attributed by some critics to the reliance on traffic patterns on a busy bridge in the next county that led to a popular resort area. In contrast, the Garcon Point bridge led to no similar attraction. Traffic patterns associated with developed areas, in which there is a pent-up demand to release congestion, are not appropriate to predict traffic in underdeveloped areas. On occasion, consultants have made use of regional travel demand models that are intended for non-traffic planning purposes but that may be inappropriate for use to predict vehicular traffic. Such a model was used to model traffic flow on the San Joaquin Hills toll road in California, which opened in various stages in the mid-1990s.
Another pitfall is the use of “steady-state” forecasts (forecasts that assume a steady, unchanging pattern) that fail to incorporate the likelihood of traffic fluctuation during economic cycles. For example, the recession that affected Orange County, California in the early 1990s contributed to the failure of the Foothill Eastern toll road that also opened in various stages in the mid-1990s. Furthermore, the lack of success experienced by both the Pocahontas Parkway, a toll road in Virginia that was completed in late 2002, and the San Joaquin Hills toll road demonstrates that weekend and truck traffic patterns can vary significantly from the underlying models used to forecast traffic patterns. The underlying models made use of over-simplified assumptions about weekend traffic by equating weekend traffic levels to those experienced during a weekday.
A critical component in every traffic model is the difficult task of determining the monetary value to attach to the time being saved by toll road users and their willingness to pay for such time savings. Travel demand models should not assume that people will always choose, and pay for, the fastest route. Often they will not. The assumption that truckers will be willing to pay to use a road to save time has proven to be inaccurate. The lower than expected proportion of trucks on the San Joaquin Hills road contributed to its failure. A detailed truck analysis is necessary if truck traffic is expected to be a significant part of future revenues.
Another difficult task is accurately to determine the “ramp-up period,” or how quickly drivers will accept a faster, less congested road. Historically, the “ramp-up period” has received insufficient attention, but is now considered an important factor that needs to be carefully considered. With respect to toll roads that have been financed with debt, structuring debt service payments using ascending debt service repayment schedules and capitalized interest beyond the projected opening of the toll road may be used to mitigate the risk of payment defaults as a result of revenue shortfalls during the early years of operation. In addition, longer “ramp-up periods” should be considered where traffic growth is dependent on future development along the road.
Sponsors and lenders should take a conservative approach to reliance on data and assumptions to protect against forecasting failures. Travel demand models that reflect the perspective of metropolitan planning organizations should be assessed and revised, if necessary, to better reflect the characteristics of toll road projects. A good example is the Central Texas turnpike, due to be operational in December 2007. The land-use and socio-economic assumptions in the forecasting model were modified downward compared to both historical data and future expectations to reflect more conservative traffic scenarios.
Future development plans should also be examined in view of the likelihood that they will actually come to fruition. Traffic consultants are developing increasingly sensitive models to reflect more accurately the inherent nuances attendant upon traffic patterns and demand in order to capture peak, off-peak, midday, night and weekend traffic.
Forecasts should incorporate the likelihood of multiple scenarios. Forecast sensitivities should incorporate the compounded effects of different assumptions and changed conditions such as changing economic environment, the possibility of sudden acceleration of improvements on competing roads, delay in construction of a complimentary artery or the delay in implementation of a toll increase. Sensitivities should also reflect a slower acceptance rate by users, longer “ramp-up periods,” and other road-specific factors.
When the road has an electronic toll collection system, forecasting has unique complications. Since modern toll roads depend on electronic means instead of toll booths for toll collection, drivers may believe that it will be easier to avoid paying tolls. As a result, greater attention needs to be paid to attempted toll evasions. Revenue forecasts should take into account the measures that will be put in place to enforce toll collection. Such measures may include policing the road, fines and penalties, revocation of drivers’ licenses and vehicle registration, and impoundment of vehicles, among others. The cost and probable success of such measures should be factored into the road’s revenue forecast.
Toll road developers and lenders should, at the outset, analyze as many different traffic and revenue sensitivities and scenarios that are possible to help account for possible revenue shortfalls. From a lender’s perspective, while certain mechanisms such as cash traps and mandatory prepayments may help protect against certain traffic and revenue shortfalls, they will not protect against a major forecasting error. One of the best ways for the toll road developer and lender to mitigate the risk of inaccurate traffic and revenue forecasts is to have a creditworthy government or government entity provide a traffic or revenue guaranty for an agreed percentage of the expected traffic or revenues. For the government or government entity to get comfortable with providing such a guaranty, it will need to undertake an analysis of the reliability of the forecasts.
Inadequate Legal Foundations
Since some governments lack the financial resources or willingness to finance the development of road networks, governments may choose to grant concessions to the private sector to build, operate and maintain toll roads.
For a private or quasi-private toll road to succeed, lenders and sponsors need to assure that adequate legislation and governmental authority is in place to permit not only the road’s development but also the uninterrupted ability to collect tolls, at a minimum, for as long as the road needs to service its project debt. Legislation also needs to be in place allowing for, among other things, the relevant land to be legally and timely delivered to the developer and also authorizing not only the collection of tolls but also the enforcement against delinquent users, when necessary. Legislative authority also needs to be in place that authorizes, within reasonable parameters, or does not prevent, the concessionaire’s right effectively to raise toll rates to service the road’s debt. Sponsors must confirm that such legislation and authority are in place at the outset and are not subject to any conditions or further governmental action.
A government’s decision to build a toll road can raise a broad range of objections. Some societies view access to any road as a fundamental right and consider collection of tolls by private entities to be unconstitutional and fundamentally discriminatory. They may argue that toll roads create a two-tier system in which people of ordinary means drive on dilapidated roads while the affluent pay to drive on new or improved highways. Sometimes, the public simply is not ready for a toll system.
In Hungary, the public was not prepared to pay for the higher level of service provided by a toll road. The Hungarian M1 motorway, which opened in 1996, captured annually less than 50% of the initially-estimated traffic. Due to the traffic shortfall, only half of the toll revenue forecast was achieved. While the concession contract for the M1 permitted the concessionaire to raise toll rates to cover the revenue shortfall, the general political environment prevented the concessionaire from effectively doing so. Soon after the toll road opened, litigation ensued on the grounds that the high toll rates were socially unjustifiable. These proceedings effectively led to the capping of the toll rates by courts that found the right to raise toll rates under the concession contract was not supported by appropriate legislation. The Hungarian experience demonstrates how the legal framework failed to accommodate and enforce toll increases vital to the road’s success even though toll increases were allowed under the concession contract. Extensive public relations about the benefits of the road are required well in advance to educate the public and “sell” the concept to prospective users. It will be difficult for a toll road to be successful if toll levels, in the long run, exceed socially-acceptable levels.
In Poland, the government awarded a concession to build the A1 motorway between Gdansk and Torun. Following the award, the scope of the government’s authority to support the project under the Motorways Act of 1994 was called into question. The Motorways Act assumed that traffic and GDP growth would enable major road projects to become self financing irrespective of state funds and with only limited government guaranties. Under the Motorways Act, government funds could only be used for preliminary activities related to the development of the road such as acquisition of the land. The authorized level of government financial support proved to be insufficient to secure the financing package for the road. Consequently, the project needed to be restructured to provide for government grants, “availability payments” and a “shadow toll” mechanism.
The “availability payments” allow the concessionaire to be paid irrespective of whether actual demand exists for the toll road so long as the concessionaire meets certain performance criteria set by the government. Such criteria may include ensuring the availability of a specified number of lanes over a specified period of time. When a “shadow toll” mechanism is in place, the government (and not the actual users) pays a “shadow toll” per user to the sponsor based on either an actual or estimated level of traffic. The restructuring required an amendment of the Motorways Act to permit the government to make the availability payments, make grants and provide interest-free loans to the concessionaire to support investment in the road. However, despite the amendment to the Motorways Act, the A1 motorway has not yet been constructed, even though the Polish government claims to support the project.
Failure by the government to condemn land and secure rights-of-way can result in extended construction periods or even prevent completion of the project altogether. Resistance by landowners to surrender their land and protracted negotiations over damages can be detrimental to the concessionaire’s ability to build the road. Construction of portions of the cross-Israel highway toll road, which opened in early 2004, were delayed when local communities resisted turning over land for political reasons. Appropriate legislation supporting the road enabled expedited condemnation, landowner compensation and an appeal process so that local resistance did not prevent timely completion of the road. If the rights-of-way for the road are not secured prior to construction commencement and the necessary and proper power of eminent domain is not vested in the correct entities, delays may occur. Sponsors need to have a good understanding of the applicable condemnation or other relevant process, legislation, and risks.
Sovereign Action and Unanticipated Events
Some toll roads have failed because of actions by sovereign governments and other unanticipated events. These events have included expropriation of the toll road by the government or repudiation of a government obligation to support the project, declaration of a moratorium or other restrictions against payments on foreign debt, detrimental regulatory actions, civil and political unrest, labor strikes and change in law generally.
A good example of such an event occurred in Venezuela. In 1997, the Venezuelan Ministry of Infrastructure awarded Autopista Concesionada de Venezuela — called “Aucovan,” a Mexican-Venezuelan joint venture — a concession to construct, operate and maintain a new highway between Caracas and the northern coast of Venezuela. Under the concession contract, the Venezuelan government guaranteed Aucovan an internal rate of return of over 15% on its investment. In the early stages of the project, the road encountered massive unexpected protests from the trucking industry over toll increases that the concessionaire had a right to charge under the concession contract. Notwithstanding this contract right, the Venezuelan government succumbed to heavy public pressure and prohibited the concessionaire from charging any tolls at all. As a result, the concessionaire was unable to fund construction and gave notice of termination of the concession contract. The Venezuelan government characterized the trucker protests as a force majuere event and argued that the trucker demonstrations were unexpected. However, it failed to convince the arbitration panel, an essential element required in order to prove the force majuere. In this case, the Venezuelan government’s force majuere argument did not cause the road to fail. But it could have.
While the Venezuelan government failed on its force majeure argument, it was successful in avoiding its obligation to guarantee the concessionaire its 15% rate of return. In the arbitration, the government convinced the arbitrators that the concession contract contained no clear formula for determining lost profits. The concession contract also did not specify clearly how to calculate the net present value of the alleged lost profits. Even though a financial model was incorporated into the contract, the arbitrators determined that the project was unlikely to generate profits even if the concession contract had not been terminated. This case demonstrates the risks involved when a foreign government succumbs to domestic pressure and decides to avoid its obligation to support the project.
While risks associated with the politics of a country are difficult to protect against, it may be helpful to evaluate the government’s history of honoring government contracts as a guide to whether it will honor future contracts.
The occurrence of unanticipated events, unrelated to actions taken by a host country government, have also contributed to toll road failure. Each of these events or risks has the effect of hindering or preventing the road development or collection of revenues. The types of unanticipated events are diverse and far too numerous to enumerate, but a few examples highlight the risks. In Denver, for example, revenues on Highway E-470, which was opened in various segments over the course of the 1990s, fell after Denver International Airport suffered a falloff in passengers in the wake of the September 11 terrorist attacks. In California, traffic on the San Joaquin Hills toll road suffered after competing roads (the I-5 and I-405) were widened after the toll road was completed. Similarly, traffic on the Dulles Greenway was affected by the later widening of Route 7. Occurrences such as San Joaquin and Dulles can be mitigated by a thorough review of regional transportation plans. In other cases, even a careful review will not reveal future competing roads that are planned after the fact. Sponsors should negotiate for a government undertaking to restrict construction of any future roads that could compete with the toll road. Should the government refuse to surrender that right, the concession contract could provide for compensation if the toll road is adversely affected by a competing road. While some risks are truly unforeseeable, others are foreseeable. Concession contracts should include provisions that broaden a sponsor’s protection against the many future events that may occur outside the sponsor’s control.
Conclusion
The lessons learned from prior toll road failures can help contribute to the success of new toll road projects. While toll road sponsors and lenders can use different methods to address risks on which prior toll roads faltered, an analysis of the unique facts of a new project will be crucial to the project’s success and to securing a robust stream of revenues.