THE BCG MATRIX REVISITED: A COMPUTATIONAL APPROACH
Although the Boston Consulting Group’s growth share matrix has been the subject of many critiques, there has been surprisingly little empirical research that directly examines the effectiveness of the model. The current study uses a computational model based on Nelson and Winter’s (1982) evolutionary model of economic change totest whether firms using BCG’s investment rules outperform firms using Nelson and Winter’s investment rules. We found that the original BCG rules were not capable of outperforming the Nelson and Winter rules even under the most favorable conditions. However, we were able to use this data to construct a set of modified BCG rules that outperformed the Nelson and Winter on almost every occasion. Theimplications of these results are discussed. Key Words: portfolio planning; simulation; agents; evolutionary economics
BPSPAP12482 The Boston Consulting Group’s (BCG) growth share matrix is one of the best known and persistent tools in strategic management. At the height of its success between 1972 and 1982, the BCG matrix was used by around 45% of the Fortune 500 (Bettis & Hall, 1981;Haspeslagh, 1982). The JSTOR database reports that no fewer than six major journal articles were authored on the BCG matrix in 1982. However, in the first decade of the 21st century, the BCG matrix is certainly in the decline phase of its product life cycle; perhaps qualifying for ‘dog’ status in its own terminology. References to the BCG matrix have disappeared from graduate textbooks and academicjournals, and are slowly being phased out of undergraduate and marketing texts except, perhaps, as historical footnotes. There are several sound reasons for this decline, including: the model’s use of only two dimensions (growth and share) to assess competitive position, the focus on balancing cash flows rather than other interdependencies, the emphasis on cost leadership rather than differentiationas a source of competitive advantage, and the poor correlation between market share and profitability (Morrison & Wensley, 1991). Despite the numerous theoretical critiques of the BCG model, empirical studies that directly examine whether the BCG matrix delivers superior profitability as a portfolio management system are surprisingly scarce (Hambrick, MacMillan, & Day, 1982; MacMillan, Hambrick,& Day, 1982; Armstrong & Brodie, 1994). In fact, Armstrong and Brodie (1994) report they could find only one empirical study prior to their own that directly tested whether firms adopting the BCG matrix outperformed those that did not. Furthermore, while 66% of students familiar with the BCG matrix thought it would produce better decisions under certain circumstances, the same students were unableto describe any such circumstances (Armstrong et al., 1994). The primary purpose of this paper is to take up this challenge and discover the circumstances under which the BCG matrix works as a comprehensive resource allocation system. A computational model is used in the current study to analyze the efficacy of the BCG matrix under various contingencies. Specifically, we test differentformulations of the matrix’s investment rules against a benchmark investment strategy using Nelson and Winter’s evolutionary economic model (Nelson & Winter, 1982). Using a technique pioneered at the RAND corporation (Bankes, 1993), the relative performance of the various BCG strategies was measured across several environmental contingencies to discover regions where the BCG rules were dominant. A secondarypurpose of this study was to make a contribution to methodology by demonstrating that any rule-based system can be tested using a computational approach. Despite calls for a model-based organization science (McKelvey, 1997), computational methods have not been widely adopted in strategic management research. We hope that our approach will demonstrate that such studies can produce valid insights...