Demand Uncertainty, Dynamic Learning and Exit in Competitive Markets

q shen,h liu
2014-01-01
Abstract:Incidences of firm exit are common in a market full of uncertainty. We consider two sources of uncertainty that may affect a firm’s exit decision. One is the uncertainty about consumers’ preference toward its own product. The other is the uncertainty toward the demand for the competitors’ products. These uncertainties are gradually resolved as a firm continues to gather information by staying in the market. Exit occurs when a firm learns that it is of low competitiveness. We model firms’ optimal exit decisions under demand uncertainty in competitive markets using a dynamic game. Specifically, firms use the sales in each period to update their beliefs on the true types of themselves as well as their competitors in a Bayesian fashion. The optimal decision on whether and when to exit a market depends on the expected value of additional learning, the fixed cost of operation, and the expected actions of the competitors. We estimate the model using data from China’s evolving microwave oven industry. We then perform counterfactual simulations to evaluate how different factors affect individual firms’ optimal exit decisions and the industry evolution path. ∗Qiaowei Shen is Assistant Professor of Marketing, The Wharton School, University of Pennsylvania; Email: qshen@wharton.upenn.edu. Hongju Liu is Assistant Professor of Marketing, School of Business, University of Connecticut; Email: hliu@business.uconn.edu. We would like to thank Andrew Ching, Jean-Pierre Dubé, Brett Gordon, Günter Hitsch, K. Sudhir and J. Miguel Villas-Boas for their valuable comments. We would also like to thank participants at the Columbia-NYU-Wharton-Yale 4-School conference, Marketing Science conference, INFORMS International conference and CKGSB Marketing Research Forum as well as seminar participants at Indiana University, National University of Singapore, Peking University, Shanghai Jiaotong University, Singapore Management University and University of British Columbia for their helpful feedback. Both authors contributed equally to this work. Financial support from the Connecticut Center for Entrepreneurship and Innovation is gratefully acknowledged.
What problem does this paper attempt to address?