Despite the progress in the assessment of transport planning and facilities investment, uncertainties in regards to the results are ever mounting. Unreliable traffic volume forecasting, in particular, has become the biggest financial risk in private investment and the fiscal costs incurred by the minimum revenue guarantee (MRG) and project bankruptcy are the biggest obstacles in the implementation of Public-Private Partnership (PPP) projects. Indeed, numerous PPP projects including the Incheon International Airport Expressway and Nonsan-Cheonan Expressway revealed that the actual traffic volume was far below the level agreed upon in the concession agreement, which has driven up fiscal costs and even put some projects such as the Uijeongbu Light Rail Transit into bankruptcy.

This study aims to present an analysis framework and suggest a selection of robust policies for when uncertainty runs deep. Desirable policy directions and agreement conditions are presented to enhance the sustainability of PPP projects. Specifically, a reduction in passenger fares and in MRG conditions is proposed, which would minimize the risks of bankruptcy and fallout from the additional fiscal burden on the government.

The method used in this study is the vulnerability-and-response-option approach which is a bottom-up approach that first considers the uncertainties in traffic volume forecasting or the underlying vulnerabilities and negative outcomes in PPP projects before presenting the optimal responses. It should be noted that the prevailing method is the predict-then-act method which is a top-down approach through which concession agreements are reached and implemented based on the assumption that traffic forecasts are accurate.

Based on the results from the analyzed PPP projects, it was found that the government’s fiscal burden could be minimized and the net present value (NPV) of projects maximized when: passenger fares are reduced to 86% of the current level and the actual revenue exceeds 60% of the expected level (or 50% of the current level) and; when 70% of the expected revenue (or 80% of the current revenue) can be guaranteed for five years after the commencement date and 62% of the expected revenue (or 70% of the current revenue) for 25 years thereafter. Furthermore, these conditions are expected to reduce the MRG payment from an average 448 billion won to 5.5 billion won and the NPV from 412.6 billion won to 71 billion won, implying that they could minimize the risk of bankruptcy and additional costs while also reducing passenger fares.

The policy suggestions are as follows. Efforts are needed to manage the contingent liabilities that can occur from government support or guarantee for PPP projects. As shown through the PPP projects in this study, implementing projects simply based on the assumption that traffic volume forecasts are accurate without an examination into the risks can ramp up the MRG payment and fiscal costs resulting from bankruptcies.

The dissemination and sharing of information on the associated risks is important in the implementation of PPP projects. For instance, empirical data on the inherent uncertainties in traffic volume forecasting is needed in order to preemptively recognize the underlying risks in feasibility evaluations so that more rational decisions can be made. Details on traffic volume forecasting should also be made transparent as openness and discussion on the forecast results will enable a better understanding of the risks.
Besides the aforementioned method, other measures to minimize the uncertainties in the short-term should be considered. PPP projects that involve low-forecasting risks should be actively implemented including those to improve aging and neglected facilities.

The vulnerability-and-response-option is unlike the existing predict-then-act method. This new approach is currently but rarely applied in the field of water management in Korea as a part of its efforts against climate change; there are no other cases of robust decision approaches in PPP projects in other countries. Therefore, applying this approach entails the following issues and limitations.

Above all, traffic volume was used as a variable for deep uncertainty in this study and the Geometrical Brownian Motion (GBM) model was used to assume the probability distribution of the traffic volume variable. And, by using the parameters of the GBM model, i.e. increase rate and volatility in traffic volume, a simulation was conducted of future traffic volumes. However, the accuracy of the actual traffic volume depends on analysts’ assumptions on and probability distribution model for the probability distribution of the traffic volume variable, which could, in turn, change the Pareto optimal solution. It should be noted that caution is needed when establishing the range for the strategy variable. In this study, the range (lower and upper limits) was based on the assumed minimum and maximum of conditions that are applicable to stakeholders. But, this range may also affect the pursuit of Pareto optimality. In this regard, prior consultation between stakeholders is important to establish the range of strategy variables.

Lastly, this study used two objective functions―performance indicators―to explore strategy variables with ‘robust’ Pareto optimal solutions that could minimize the fiscal costs generated by the MRG and maximize the NPV of PPP projects when there are deep uncertainties in traffic volume forecasting. This study, however, does not include discussions and analyses on how much Pareto weight should be given to each performance (objective) indicator. Indeed, despite the fact that performance indicators serve as the basis for final strategy decisions in trade-offs between the interests of the government and private project operators, this study only deals with cases wherein the weighting of the performance indicators (objective functions) is difficult or impossible.

Preface
Executive Summary

Chapter 1 Introduction
　Section 1 Necessity of the Study
　Section 2 Objectives and Scope of the Study

Chapter 2 Risk, Uncertainty, and Deep Uncertainty
　Section 1 Risk and Deep Uncertainty
　Section 2 Decision-Making Methods Under Deep Uncertainty
　Section 3 Deep Uncertainty in Traffic Volume Forecasting and Public-Private Partnership (PPP) Projects

Chapter 3 Case Study: PPP Project on Line A
　Section 1 Project Overview
　Section 2 Key Agreement Terms

Chapter 4 Decision-Making Model Under Deep Uncertainty in Traffic Volume Forecasting
　Section 1 Problem Definition
　Section 2 Procedures for Robust Decision-Making
　Section 3 Deep Uncertainty Variable―Traffic Volume Model
　Section 4 Performance Indicators (Objective Functions)
　Section 5 Optimization of Multi-Objective Functions
　Section 6 Robustness Evaluation of Strategies Under Uncertainty
　Section 7 Comparison Between Existing and Robust Strategies
　Section 8 Vulnerability Analysis and Final Strategy Selection

Chapter 5 Managing Deep Uncertainty in Traffic Volume Forecasting
　Section 1 Recognizing the Risks in Traffic Volume Forecasting
　Section 2 Implementing Projects That Minimize Uncertainty in Traffic Forecasting
　Section 3 Managing Fiscal Liabilities From PPP Projects

Chapter 6 Summary and Conclusion

References
ABSTRACT