CEO Duality and Firm Performance Under Endogeneity

This paper intends to lay out a framework to identify self-selection bias and how it can be used practically to analyze CEO duality in the field of corporate governance. Prior literature about agency theory and stewardship theory have argued extensively in this topic yet most of them treat duality as an exogenous variable. The model identifies whether this leadership structure is self-selected to maximize firm's profit. Also, it gives a correct estimate of duality effect after the bias is corrected. Because of different measurement of firm performance, the author uses four measures, including Tobin's q, return on assets (ROA), return on equity (ROE) and earnings per share (EPS) to represent firms' ability to make profit. The results are mixed. For Tobin's q and earnings per share, the author finds strong evidence that self-selection bias exists, that is, there is evidence that firms choose the structure to perform better. After controlling for bias, both Tobin's q and EPS show negative effect of duality on performance which means non-duality provides better performance than duality. However, for ROA and ROE, there is no evidence that firms do so to improve performance. Specifically, for ROE, the model has low predictability and thus, is not likely to provide meaningful and reliable results. We also compare the model with traditional methods including dummy variable approach and two stage least square approach. Two stage least square method produces similar result for EPS but opposite result for Tobin's q, while dummy variable approach shows there is no association for any measurement. The author also suggests not to use ROE as an approximation for profitability due to its unpredictability. Such results contradict the agency theory of corporate governance. These findings should be examined further by considering those firms that changed governance structure in sample period and more deeply the reasons such changes were made