A risk analysis framework for toll road revenue forecasts

In the United States, the shortage for traditional public sector funding sources for a safer and more effective highway infrastructure has made it more increasingly difficult for transportation agencies to keep up with the increasing demand. As a consequence, transportation agencies are considering a number of alternatives to raise funds such as charging tolls or cooperating with the private sector. One of the most important steps when analyzing the feasibility of a toll road is the traffic forecast process. However, the uncertainty of the traffic forecast has become apparent across the U.S. as evident in the common overestimation of traffic demand (usually between 25% to 30%) as well as toll revenue. Usually, the traffic forecasting model is a function of several key variables, which include Value of Time (VoT) and toll rate. The incorrect values or assumptions of these variables applied to the model can be critical when determining the traffic volume of a potential toll road. A risk analysis framework was developed to quantify the risks (or variations) of toll road revenue imposed by uncertainty in model inputs. The methodology incorporates Monte Carlo Simulation to uncertain model inputs such as the VoT. The proposed framework was then applied to a case study in El Paso, Texas. The results show that the Beta General distribution provides the flexibility (due to its shape parameters) to fit the forecasted toll revenue data. The framework should provide the analyst with a basic approach on how to develop probability distribution functions of revenue with respect to uncertainty in model variables