Investment risk based on traffic forecasting accuracy: Case studies of U.S. highway public-private partnerships

The scarcity of public funding, coupled with crumbling infrastructure, and demands for infrastructure improvements has led to increased usage of public-private partnerships (PPP) as an alternative delivery method for surface transportation infrastructure projects in the United States. PPPs however have a somewhat mixed track record; an unusually high number of PPPs have defaulted on debt due to lower than expected traffic and revenue. These observations led to this study to investigate the potential losses to debt and equity due to the inaccuracies in traffic and revenue forecasting. Building on the traffic forecast research of Dr. Bent Flyvbjerg and Dr. Robert Bain, a top down approach was taken to investigate the problem. Five U.S. highway PPP projects were selected for investigation based on project type, location and availability of financial data. Financial data from Private Activity Bond transcripts was used to construct Simulated Cash Flow Models that emulated the concessionaires' cash flow models and accounted for funds during the project term. Next, a Traffic Distribution Instrument was created to vary the lender's base case traffic forecasts to a user-specified confidence interval on a normal probability distribution function. These new traffic levels were entered into the Simulated Cash Flow Models and their impacts were observed on equity IRR and debt default. The effects of optimism bias in traffic forecasting were investigated by creating two different traffic forecasting accuracy distributions, one with no optimism bias (u=1.00, sd=0.34) and one with optimism bias (u=0.77, sd=0.26). It was found that the risk of default on debt in PPP projects is three times greater than would be anticipated by a projects BBB- credit rating. The equity IRR and equity risk premium for the base case scenarios is consistent with other investments but returns are at serious risk when accounting for the volatility of traffic forecasting. It was also shown that concessionaires do not accurately price the cost of traffic revenue risk. From these results it was concluded that there is a general underestimation of risk in direct toll PPP projects by both debt investors and equity participants. It was not possible to conclude that optimism bias is present or the cause of the unusually high default rate in PPP projects; however, it was possible to demonstrate that if present, optimism bias would be detrimental to returns on investments in the selected cases..