Abstract
Reference class forecasting (RCF) is increasingly being adopted to mitigate two main factors causing persistent cost overruns in infrastructure projects—optimism bias and strategic misrepresentation. In practice, organizations tend to tailor generic estimating approaches to their unique operating environments by developing hybrid approaches blending RCF with other approaches such as the conventional fixed contingency approach. Yet, little empirical evidence exists on the accuracy of RCF or its variations. This study presents the results of estimation accuracy of a sample of road projects conducted by an Australian State Road & Traffic Authority using a hybrid estimating approach blending primarily RCF with fixed contingency approach. The results are then compared with historical results from literature on infrastructure projects and samples of two other dominant estimation methods, namely, the conventional fixed contingency approach and risk-based estimating (RBE). This study finds that the average accuracy of this sample using the hybrid approach compares favorably to historical results, which typically use the fixed contingency approach.