Decision Support in Car Leasing: A Forecasting Model for Residual Value Estimation

摘要

The paper proposes a methodology to support pricing decisions in the car leasing industry. In particular, the price is given by the monthly fee to be paid by the lessee as compensation for using a car over some contract horizon. After contract expiration, lessors are obliged to take back the vehicle, which will then be sold in the used car market. Therefore, lessors require an accurate estimate of cars' residual values to manage the risk inherent to their business and determine profitable prices. We explore the organizational and technical requirements associated with this forecasting task and develop a prediction model that complies with identified application constraints. The model is rigorously tested within an empirical study and compared to established benchmarks. The results obtained in several experiments provide strong evidence for the proposed model being effective in generating accurate predictions of cars' residual values and efficient in requiring little user intervention.