Proceedings of the ASME 2009 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference IDETC/CIE 2009 August 30-September 2, 2009, San Diego, California, USA DETC2009-87165 METHODS FOR EVALUATING SUITABILITY OF ECONOMETRIC DEMAND MODELS IN DESIGN FOR MARKET SYSTEMS Bart Frischknecht ∗ Optimal Design Laboratory Mechanical Engineering University of Michigan Ann Arbor, Michigan 48109 Email: bdfrisch@umich.edu Katie Whitefoot Optimal Design Laboratory Design Science University of Michigan Ann Arbor, Michigan 48109 Email: kwhitefoot@umich.edu Panos Papalambros Optimal Design Laboratory Mechanical Engineering University of Michigan Ann Arbor, Michigan 48109 Email: pyp@umich.edu ABSTRACT This paper articulates some of the challenges for what has been an implicit goal of design for market systems research: To predict demand for differentiated products so that counterfactual experiments can be performed based on changes to the product design (i.e., attributes). We present a set of methods for examin- ing econometric models of consumer demand for their suitability in product design studies. We use these methods to test the hy- pothesis that automotive demand models that allow for nonlinear horizontal differentiation perform better than the conventional functional forms, which emphasize vertical differentiation. We estimate these two forms of consumer demand in the new vehi- cle automotive market, and find that using an ideal-point model of size preference rather than a monotonic model has model fit but different attribute substitution patterns. The generality of the evaluation methods and the range of demand model issues to be explored in future research are highlighted. 1 INTRODUCTION A goal in design for market systems is to integrate models of demand, cost, and product performance in order to implement a game-theoretic formulation of producer behavior where pro- ducers choose the products they will produce, the attributes of the products, and the prices they will charge in order to maxi- mize an objective such as profit. Economic theory of produc- ∗ Address all correspondence to this author. ers, differentiated products, and market equilibrium are well- established and ongoing fields. Economic theory recognizes that in differentiated-product markets firms compete on product at- tributes as well as price [1]. Engineers develop explicit repre- sentations of firms’ technology capabilities, proving opportune to conduct forward-looking product planning or design scenar- ios in a market system context given suitable consumer decision models. This paper suggests that conventional consumer deci- sion models from economics should be evaluated as to their suit- ability for use in a producer decision model and is a first attempt at presenting methods to accomplish this task. Section 2 provides background on econometric models with specific highlights to applications in the automotive market and developments in the engineering design literature. Section 3 identifies challenges stemming from current demand modeling practice. Section 4 presents methods for evaluating a given de- mand model for use in a design optimization context. Section 5 presents a mixed-logit choice model from the new vehicle au- tomotive market estimated from disaggregate consumer data and discusses the performance of this model with respect to the evalu- ations described in Section 4. Section 6 summarizes conclusions. 2 BACKGROUND We focus on econometric models of product demand derived from observed consumer behavior. We assume consumers seek to maximize their expected utility, and the demand (or choice) 1 Copyright c  2009 by ASME