Search for: All records

Abstract Consumer choice models used in optimal product design typically ignore potential context effects by assuming the utility of each product is independent of the attributes of other products in the choice set. We characterize implications of context effects for profit-maximizing designs by deriving the first-order conditions of the design problem under alternative utility formulations, and we propose a utility function that incorporates context effects and has well-defined optimal design solutions for all products in the choice set. We then conduct a discrete choice survey experiment of automobile options and find statistically significant context-effect parameters and superior out-of-sample prediction when context-effect parameters are used in both logit and mixed logit models. These results suggest that context effects can be important in engineering design contexts and have the potential to affect optimal design differentiation. 

Abstract Global product platforms can reduce production costs through economies of scale and learning but may decrease revenues by restricting the ability to customize for each market. We model the global platforming problem as a Nash equilibrium among oligopolistic competing firms, each maximizing its profit across markets with respect to its pricing, design, and platforming decisions. We develop and compare two methods to identify Nash equilibria: (1) a sequential iterative optimization (SIO) algorithm, in which each firm solves a mixed-integer nonlinear programming problem globally, with firms iterating until convergence; and (2) a mathematical program with equilibrium constraints (MPEC) that solves the Karush Kuhn Tucker conditions for all firms simultaneously. The algorithms’ performance and results are compared in a case study of plug-in hybrid electric vehicles where firms choose optimal battery capacity and whether to platform or differentiate battery capacity across the US and Chinese markets. We examine a variety of scenarios for (1) learning rate and (2) consumer willingness to pay (WTP) for range in each market. For the case of two firms, both approaches find the Nash equilibrium in all scenarios. On average, the SIO approach solves 200 times faster than the MPEC approach, and the MPEC approach is more sensitive to the starting point. Results show that the optimum for each firm is to platform when learning rates are high or the difference between consumer willingness to pay for range in each market is relatively small. Otherwise, the PHEVs are differentiated with low-range for China and high-range for the US.