Model for Efficient Risk Allocation in Privately Financed Public Infrastructure Projects Using Neuro-Fuzzy Techniques

Abstract

Risk allocation plays a critical role in privately financed public infrastructure projects. Project performance is contingent on whether the adopted risk-allocation strategy can lead to efficient risk management. Founded primarily on the transaction cost economics, a theoretical framework was recently developed to model the risk allocation decision-making process in privately financed public infrastructure projects. In this paper, a neuro-fuzzy model adapted from an adaptive neuro-fuzzy inference system was further designed based on the framework by combining fuzzy logic and artificial neural network techniques. Real project data were used to train and validate the neuro-fuzzy models. To evaluate the neuro-fuzzy models, multiple linear regression models and fuzzy inference systems established in previous studies were used for a systematic comparison. The neuro-fuzzy models can serve the purpose of forecasting efficient risk-allocation strategies for privately financed public infrastructure projects at a highly accurate level that multiple linear regression models and fuzzy inference systems could not achieve. This paper presents a significant contribution to the body of knowledge because the established neuro-fuzzy model for efficient risk allocation represents an innovative and successful application of neuro-fuzzy techniques. It is thus possible to accurately predict efficient risk-allocation strategies in an ever-changing business environment, which had not been achieved in previous studies.