Statistical Considerations for Predicting Residual Value of Heavy Equipment

Abstract

Residual value needs to be considered in owning cost calculations for used heavy construction equipment. Its dependency on factors such as manufacturer and model, equipment age, and condition rating can best be examined by analyzing real market data from equipment auctions. Macroeconomic indicators can also be included to examine any potential influence of the overall economy on auction prices. This paper discusses statistical considerations for performing such a residual value analysis. Considerations include the study type, data properties, identifying outlier observations, regression assumptions, and formulating and selecting an appropriate regression model using the adjusted coefficient of determination. A second-order polynomial of equipment age with additive factors appears promising as the final regression model. Adjusted confidence and prediction intervals are created to correctly display residual value. Cross-validation using randomly split halves of the dataset is performed. Actual data for medium track dozers are used to illustrate the validity of the methodology.