The CEO effect and performance variation over time

Abstract

While CEO effect scholars agree that variation in firm performance tends to persist over time and that CEOs’ performance contribution should be gauged against a changing context, recent CEO effect studies addressing these issues have made extreme but opposite claims concerning the magnitude of the CEO effect. We show why recent findings that indicate a much larger CEO effect are spurious. We also argue and empirically demonstrate that a multilevel model that includes autocorrelation can properly gauge CEOs’ performance contribution against a changing context while simultaneously avoiding confounding. Our empirical result shows that the opposite claim positing that the CEO effect is nearly indistinguishable from chance is likewise unwarranted. These findings have implications for CEO effect studies and other studies that use longitudinal firm performance data.