A New Look at Corporate Effects – by Samuel Phineas Upham
In 1991, Richard Rumelt wrote an essay on corporate effects, stating that the corporate effect on the variance of the performance of a company is very insignificant. Many years later authors Thomas H. Brush and Philip Bromiley decided that Rumelt’s essay wasn’t as accurate as he thought. In their essay, “What Does a Small Corporate Effect Mean? A Variance Components Simulation of Corporate and Business Effects,” the authors study, reinterpret, and retest Rumelt’s findings. In their essay, Brush and Bromiley design a model where there was vital corporate effects and then used Rumelt’s metrics to see if they would get similar results and to analyze the findings. This approach reveals just how important it is for researchers to clearly explain their methods, not only for evaluations but also for critical analysis.
Rumelt’s findings were controversial. According to Rumelt, 46 percent of variance has to do with business unit effects, 8 percent with industry effects, and 1 percent is explained by corporate effects. He believes that companies are not able to transfer success between product lines. He claims that companies don’t have enough resources that can be used internally to help them succeed.
However, authors Brush and Bromiley disagree with Rumelt’s findings. The first reason they find his essay faulty is because they believe his findings don’t mean what he thinks they mean. According to the authors, Rumelt should have made statements of important based on an estimated parameter not explained variance.
Through a unique and clever method, Brush and Bromiley create a simulation of data in order to rest Rumelt’s findings. The test attempts to answer two questions, and the authors use different methods to answer both questions. The authors discover that variance component seem to be the square of importance and do not seem to mirror importance in a linear way. They also found that the industry is 40 percent as important as the business unit. Finally, they found that findings vary depending on the simulation. Brush and Bromiley’s study can mean that corporate effect is not overwhelming and that it is much smaller than business unit effect, but not inconsequential.
Brush and Bromiley have spent a great deal of time and energy on disproving Rumelt’s findings, a scholar whom they respect nonetheless. While I don’t know enough about variance component methodology, I do question the use of “scale” as measures by the difference between top and bottom quartiles. Whether the authors are right or wrong, it would be interesting to get Rumelt’s reaction to their findings.
————————————————————————–
This article was written by Samuel Phineas Upham. Samuel Phineas Upham is an investor in New York City and San Francisco where he has worked in macro and special situations. He currently works at a family office / hedge fund. To find out more visit Samuel Phineas Upham Website and Samueal Phineas Upham Twitter page.