Forecasting Performance Evaluation: Does a Complicated Model Outweigh a Simple One?

A wide variety of alternatives are available to forecast future outcomes; methods range from intuitive to systematic processes. Forecasting construction volume should be an interest for all stakeholders in the construction marketplace. Contractors, especially, will benefit because they will be making management decisions and planning for future market trends on a short-term basis. Thus in this study, the authors compare the forecasting performance of two methods, simple and complicated models, and provide a decision support tool for short-term construction volume forecasting. Various forecast evaluation methods explored in this study include: Mincer and Zarnowitz's method; Theil's U and U2 statistic; Mean Square Forecasting Errors (MSFE); and turning point errors. Forecasting realization diagrams are also utilized to compare the degree of error deviations. Furthermore, advantages of a combined forecasting method are discussed. In conclusion, the authors recommend using a simple method in forecasting future construction volumes because it is comparable to the complicated model in forecasting performance.