Wissenschaftliche Vorträge

Estimating product costs applying DEA and regression trees – insights from an automotive supplier
Andreas Dellnitz
Andreas Kleine
Wissenschaftliche Vorträge
Data Envelopment Analysis
International Conference on Data Envelopment Analysis, September 4-6, 2023, University of Surrey Guildford.

Data envelopment analysis (DEA) is a well-established method for measuring the efficiency among a comparable group of decision-making units (DMUs). DMUs mostly comprise profit or non-profit companies which are characterized by their time-related activities -- i.e., their inputs and outputs. Still, the notion decision-making unit is not reserved only for companies, a DMU can also be a project or product. In this paper, we focus on the latter type, using DEA-based efficiency scores to estimate the price for newly developed products to be purchased. DEA-based estimations of such product-related purchasing costs, however, suffer from one fact: DEA only accounts for deterministic input and output relationships and, hence, cannot handle unobservable negotiation behavior. Therefore, we combine DEA with regression trees in order to remedy this deficiency and, ultimately, to improve the estimations of a product's purchase price. To substantiate the superiority of our approach, we study a real-world application stemming from the automotive supplier industry.