What does a small corporate effect mean? A variance components simulation of corporate and business effects

In a widely cited paper, Rumelt (1991) presents estimates of the relative influence of corporate, business unit, and other influences on business unit profitability and finds the corporation explains almost none of the variability in business unit profitability. Using a Monte Carlo simulation, we examine the relation of variance component magnitudes to other indicators of the importance of a particular effect. Our results demonstrate that variance components can be an extremely nonlinear indicator of importance. We also question whether Rumelt's corporate effect represents the possible contributions of corporate strategy to business unit performance. This addresses a puzzle raised by Rumelt (1991) concerning the small effect of corporations in explaining performance, and suggests that Rumelt's findings should not be seen as demonstrating the insignificance of corporate strategy. © 1997 John Wiley & Sons, Ltd.