An Empirical Study Of The Determinants Of Executive Compensation

The thesis uses a holistic model that considers the impact of the overall mix of governance and incentive variables simultaneously on executive compensation of China's listed companies. We try to show some evidence that other firm-level and individual level variables will also influence executive compensation such as individual characteristics of executive, risk, the managerial complexity, industry and corporate governance mechanisms besides firm performance, size and state-owned shares, thus provide a clearer picture of how executive compensation is determined.In the study, we lay emphasis on disputed variables while controlling for firm performance, size and state-owned shares because the three variables have been found to be important in previous research on the determination of executive compensation. We employ a two-stage procedure, Enter method provided by SPSS to estimate model coefficients with least square regression. In order to show the explanatory powers of the variables better, we also choose Stepwise method to get the optimal result. We compare the two results in our research.The results of this study support some findings and arguments of other researchers, while also suggesting some new evidence. For all the variables we consider, we find that firm size, state-owned shares, industry variable and performance are significant. Our research partially supports the argument that internal shareholdings are additional managerial perquisites and the establishment of independent director system is just to follow the regulation. The results also prove that firm diversification has a positive impact and firm risk has negative impact on executive compensation, and there is an inverted U-form relationship between the age of executive and his or her compensation. We also find that the phenomenon of chairman with zero compensation and significant multicollinearity seriously affects the efficacy of the estimated coefficient of CEO duality.Besides, we find that the meaning of the measure for industry we used is different in nature from its traditional meaning in the west literature. We argue that the measure of state-owned shares we designed contribute at least partly to its superior statistical result as measured by coefficient significance. At last, we have to say that executive compensation practice of our listed companies still has a long way to go, and the information disclosure system still need improving.