They may be resistant to change within the business, or hold on to their position to the detriment of the company. They may simply have a poor work ethic. Previous studies on the effect of CEO power on company performance show mixed results, with a predominantly negative relationship. Our study argues that the relationship between CEO power and company performance can be better understood by analysing the moderating role of two contingency factors – the oversight of the company board and the effect of the Sarbanes–Oxley Act (SOX) of 2002.

Our results

Our research results indicate that the negative effect of powerful CEOs on firm performance is mitigated by the oversight of a board. In addition, they underline that the passing of SOX – a United States federal law that established new auditing and financial regulations for public companies – has reduced the negative effect of CEO power and amplified the positive effect of board oversight.

As such, our study highlights to investors and company directors that corporate boards should be designed to take into account the power of individual CEOs. In addition, our results suggest to policymakers that tight regulations can be an effective deterrent to bad behaviour by CEOs and be closely aligned with the interests of the company.

In the past, many scholars and practitioners have seen the optimal design of corporate boards as ‘box-ticking’. We argue that board design, in fact, plays a key role in the success of the company, and this may be more pronounced in the post-SOX period. Having an appropriately sized board and including directors who are also shareholders may support the board’s effective monitoring of powerful CEOs and the company’s overall performance.

Our research also examined the effect of having external director-CEOs on the board. While there are some potentially negative effects (e.g. closed friendship circles or cliques forming), positive effects include strengthening the board with highly competent and experienced individuals. It can also foster a sense of shared responsibility for the company’s success. These positive effects are stronger in the post-SOX period and are still highly relevant to businesses today.

Interestingly, our results also underline that corporate governance practices, and board oversight in particular, do not always have positive implications for company performance. This suggests that corporate boards should be designed in relation to the power of CEOs. In other words, when CEOs are powerful, high levels of board oversight can balance their power and its potentially negative consequences on company performance. However, when CEOs are less powerful, high levels of board oversight can undermine company performance. This is particularly true when there is an emphasis on controls, as in the post-SOX setting. Anecdotal evidence seems to support this conclusion. Many CEOs question what they see as the zero-trust and zero-tolerance environment after SOX, and underline the risks of a shift of power from company CEOs and executive directors to non-executive directors. In this way, our study also provides guidance for policymakers, suggesting that tight regulations can be an effective deterrent to bad behaviour by CEOs and be closely aligned with the interests of the company’s owners. In summary, our findings provide support for a contingency perspective on corporate governance, under which governance mechanisms should be designed, taking into account both the firm and institutional context.

Our research methods

We hand-collected data on both CEOs and boards of directors for S&P 500 US firms in the period 2000-2006 (three years before and four years after SOX). We measured company performance with return on assets (ROA). Following Finkelstein’s model of CEO power (1992), we measured CEO power through a composite index. We measured board oversight using three board attributes: size, the presence of director-shareholders, and the presence of director-CEOs from other companies. We introduced controls for company size, age, past performance, length of CEO tenure and the percentage of directors from outside the company. To test our hypotheses, we employed the fixed-effects model for panel data methodology.

Limitations of our study

Our study has some limitations. First, our analysis is biased towards large US firms. This limits the applicability of our results to other companies (e.g. those with concentrated ownership) and to other countries (e.g. where corporate law and regulations may be very different). In addition, we used well-established proxy variables (i.e. a proxy used in place of a variable that is hard to observe or measure) to explore complex constructs, such as the interaction among directors and CEOs. Furthermore, while our study shows the impact of SOX in the short-term, our results cannot be easily extended to capture its long-term effects.