1
The relationship between CEO duality &
Firm Performance
Kino Alexander
10794786
Bsc Economie en Bedrijfskunde Specialisation: Finance & Organization Supervisor: Stephan D. Jagau
January 2018 Words: 5125
2 Statement of own work
I, Kino Alexander, hereby declare that I have written this thesis myself and that I take full responsibility for its contents. I confirm that the text and work presented in this thesis is original and that I have not used any sources other than those mentioned in the text and in the references. The Faculty of Economics and Business is only responsible for the submission of the thesis, not for the content.
3
Abstract
CEO duality is the practice of an individual practicing both titles of CEO and chairman of the board. It is a topic widely discussed in corporate governance, but lacks clear theoretical and empirical evidence. In this paper I will be studying the relationship between CEO duality and firm performance using an exogenous shock. The shock is used is the terrorist attack on September 11 2001 in the U.S. The results reported a significant positive relationship between CEO duality and firm performance, namely Tobin’s Q and return on assets.
4
Table of content
1.0 Introduction 5
2.0 Theoretical framework 7
2.1 Arguments in favor or against CEO duality
2.2 Former research on the duality-performance relation
3.0 Methodology 9
3.1 the model
3.2 sample & data source
3.3 the exogenous shock
3.4 empirical method
4.0 Results 13
4.0 Hausman test
4.1 impact of dual leadership on return on equity 4.2 Impact of dual leadership on return on assets
5.0 robustness check 14 6.0 conclusion 15 5.1 summary 5.2 discussion Bibliography 17 Appendix 19
5
Figure 1.
This figure is attained from Yang & Zhao (2014). It shows the percentage of U.S firms with dual leadership over time.
1.0 Introduction
This paper studies the effect of CEO duality on firm performance using an exogenous shock. CEO duality is the practice of one person serving as both titles of CEO and chairman of the board (COB). CEO duality is a topic widely discussed in the corporate governance world. This is also noticed in past events concerning leadership structures. For example, The Sarbanes-Oxley act in 2002 which stated stricter rules regarding board governance. Secondly, The Troubled Asset Relief Program (TARP) obligated firms, who made use of it, to change their leadership structure into separate CEO and chairman titles. Furthermore, multiple proposals were submitted by the congress of the United States to abolish the dual leadership structure. Lastly, the Dodd-Frank act in 2010 that forced the Securities and Exchange Commission (SEC) to set obligatory rules which state that firms have to report the reasoning behind their leadership structure.
These events, among other things, have presumably led to a change in board leadership structure over the world. Looking at the S&P 500 firms, the transformation from CEO duality to CEO non-duality has increased from 20% to 40% (Krause, Semadeni & Cannella, 2014). Spencer Stuart (2010) showed in his paper that there was a growth in truly independent management, from 9% to 20%. Furthermore, in the United States the number of firms with a dual leadership structure fell over the years. Between the
1970’s and the 1990’s 80% of the largest firms in the U.S had a dual CEO. This number has decreased to 54% in 2010 (Yang & Zhao, 2014). See figure one.
An important question to be asked is: where does this dislike towards CEO duality come from, considering there is not much clear theoretical support and mixed empirical results. An answer to this question could be for example the financial crisis in 2008. This event resulted in regulators and stakeholders questioning the role and responsibilities of executives and board members. While simultaneously questioning the actions taken by the executives and board members. They encountered false practices resulting from abuse of power (Carty & Weiss, 2012). Considering regulators and stakeholders wanted more transparency and independency (Palanissamy, 2015), which aren’t characteristics associated with CEO duality, the idea of dual leadership deteriorated. This is only one of multiple events not supporting dual leadership
6 Literature on CEO duality use two main theories to debate about the subject. The first being the agency theory, which argues against dual leadership and the second being the Stewardship theory, which argues in favor of dual leadership.
The agency theory claims that combining the titles of CEO and chairman of the board compromises the independency of both practices, thereby weakening the separation of ownership and control. This could evolve into conflict of interest between the CEO and the shareholders. Which could lead to the CEO acting in his own interest at the expense of the shareholders. Adding to this conflict of interest, dual CEO’s have significantly higher salaries and their turnover is less sensitive to firm performance (Core et al., 1999; Goyal & Park, 2002). Furthermore, Kesner and Johnson (1990) argue that boards with less independent members are more frequently convicted of wrongdoing. Their results only held if the CEO was also the COB. Lastly, the great power a dual CEO holds could also lead to managerial entrenchment
(Finkelstein & D’aveni, 1994). Managerial entrenchment refers to the situation where a manager acquires so much power that they could use the firm to explore their own interest instead of the interest of shareholders. The significantly higher salaries of dual CEO’s and their less sensitive turnover to firm performance could be an outcome of managerial entrenchment (Core et al., 1999; Goyal & Park, 2002).
The stewardship theory claims that the organizational structure determines the performance of a firm (Donaldson & Davis, 1991). A good structure will be facilitative to the CEO when formulating and implementing strategies. The theory argues that having a dual CEO contributes to this, because the firm will benefit from unified leadership. Furthermore, dual leadership makes the information flow within a firm more efficient. This will ultimately lead to a reduction in transferring and processing of information costs (Jensen & Meckling, 1995). Lastly, the dual title is also part of the incentive scheme. Not rewarding a CEO with the chairman title could lead to a lack of motivation (Dalton, 1998).
Besides literature, there have also been researchers studying the effect of dual leadership. One study in particular concerning this paper is that of Yang & Shao (2014). In their paper they studied the effect of CEO duality on firm performance. They used an exogenous shock in there research to mitigate the endogenous problem, which will be explained later in this paper. They concluded that CEO duality is beneficial when competition intensifies.
This paper will study 45 listed firms in the S&P 500 list from 1998 till 2006 on the relationship between CEO duality and firm performance, using return on equity, return on assets and Tobin’s Q as dependent firm performance measurements. I used an exogenous shock to mitigate problems arising from the endogeneity aspect of board leadership structure. The shock used was the terrorist attack on September 11 2001. Furthermore, I also applied a
7 robustness check to see whether the results hold when another firm performance measurement is used.
My results concluded a significant positive relationship between CEO duality and firm performance for the models with return on assets and Tobin’s Q as dependent variables. Furthermore, for all three models reported significantly lower firm performance post-shock.
This paper is organized as follows: section 2 describes the existing literature on the topic, section 3 describes the model, sample, data, exogenous shock and the empirical method. Section 4 states the results coming from the regressions and section 5 describes the robustness check and section 6 consists of the conclusion, under which a summary and discussion.
2.0 Theoretical framework
2.1 Arguments in favor or against CEO duality
Arguments in support of CEO duality are mostly based on the Stewardship Theory (Donaldson & Davis, 1991). This theory argues that the variation in firm performance arise from the organizational structure of a firm. The right structure should empower, authorize and make clear what the expectations are of the firm, specifically regarding the CEO. When power and command are combined, thus with CEO duality, there is no question who is authorized or responsible for a particular situation.The company will benefit from this unified leadership, but also the CEO. Namely, a dual leadership structure is facilitative to a CEO when formulating and implementing strategies. The COB title is also part of the incentive scheme. Dalton (1998) argues that when a CEO isn’t rewarded with the chairman title, it could lead to a lack of motivation or the CEO potentially leaving the firm. Another argument in favor of dual leadership comes from Jensen & Meckling (1995). In their paper they argued that the CEO is usually the person who has the best knowledge of the opportunities and obstacles that a firm faces. Therefore, if the CEO is also in charge of the board, information flow will become more efficient, making it easier and faster to implement strategic plans to act on these opportunities and obstacles. When transferring and processing of information becomes more efficient, the cost related to this will also decrease (Jensen & Meckling 1995). Additionally, Brickley, Coles, and Jarrell (1997) emphasize the critical issue of the potentially questionable incentives of the independent chairman. They argue that an independent chairman could reduce the agency costs associated with the CEO, but on the other hand also introduce its own agency costs, as a result of the great power he himself holds. Furthermore, within corporations there is often a process where a retiring CEO holds on to his chairman title for a while. This is practiced so that the retiring CEO can pass on valuable information to the new CEO. Meanwhile the board can
8 examine the new CEO to see if he or she is qualified for the job. If this is true, the retiring CEO passes on the chairman title to the new CEO. This process has its advantages. From an incentive perspective, the probability of also attaining the chairman title could trigger extra motivation, what wouldn’t be the case if the two titles are practiced separately (Brickley, Coles, & Jarrell, 1997). From a psychological perspective, this method could ease out the succession process, making sure that the retiring CEO doesn’t stick to his title for too long (vancil, 1987). This advantage may mitigate the problem argued by Core et al (1999), Goyal and Park (2002) further in this paper.
Arguments against CEO duality are mostly based on the agency theory (Eisenhardt, 1989; Fama & Jensen, 1983a). Within a corporation you have an agent and a principal. In this case, the CEO being the agent and the shareholders are the principals. The agent can make decisions on behalf of the principal and should act to maximize their value (Jensen and Meckling, 1976). The idea behind the theory is that when the goals of the both parties aren’t aligned, the agent could have the incentive to act in his own interest at the expense of the principal (Fama & Jensen, 2008). To mitigate this problem, there is a separation between ownership and control within modern corporations. The executives practices the decision management and the board the decision control. Decision management is the right to initiate and execute plans for resource allocation, while Decision control means the right to validate and monitor these decisions (Brickley, Coles, & Jarrell, 1997). The separation should ensure that conflicts of interest doesn’t arise and thereby protecting the interest of the shareholders. With a dual CEO, the independence of the two titles will be compromised, considering that the CEO practices both titles. Adding to this conflict of interest, Core et al (1999), Goyal & Park (2002) claim that dual CEO’s have significantly higher salaries and that turnover, meaning the replacement of a CEO, is less correlated with firm performance. Thus dual CEO’s are less likely to be fired when firm performance deteriorates. This suggests that dual CEO’s use their greater power for their own benefit. Furthermore, it also suggests that dual CEO’s are less accountable for their actions. Another argument against dual leadership is that an independent chairman could also be of more value. The CEO could focus on running the company, while the COB can concentrate on running the board of directors and thereby complementing each other (Dalton et al ,1998). Furthermore, Kesner and Johnson (1990) claim that boards with less independent directors were more often convicted of wrongdoing. In this situation, wrongdoing being the amount of suits against the board done by the shareholders. Their results held only if the CEO was also the COB.
Furthermore, there have been studies claiming that CEO duality is beneficial in certain circumstances. They found that having a dual CEO would be linked to type of industry condition the firm is in. Boyd (1995) concluded that when there is resource scarcity and environmental
9 complexity, the firm could benefit from CEO-duality, because of the unified leadership and swift decision making. Whereas in low complexity environments, the likelihood of opportunism of the CEO increases, making non-duality preferable.
2.2 Empirical research on the duality-performance relation
There have been studies in the past researching the relationship between CEO-duality and firm performance. Here, I summarized some studies similar to this one, which also use financial variables as their firm performance measurement. Rechner and Dalton (1991) studied 141 firms from the fortune 500 between 1978 and 1983 and found a negative effect of CEO duality on return on equity, profit margin and return on investment. Donaldson and Davis (1991) found a positive effect on return on equity when studied 337 firms in the United States in 1987. Furthermore daily and Dalton (1992, 1993) performed two studies concluding no effect on return on assets, return on equity and P/E ratio. The first study consisted of 100 fastest-growing small listed firms in 1990 and the second one out of 186 small listed firms in the United States in 1990. A study especially relating to this paper is that of Yang & Zhao (2014). In their research they studied the effect of CEO duality on firm performance. Their main firm performance measurement was the Tobin’s Q, but they also tested the effect of CEO duality on return on equity and return on assets for completeness of their study. Evidence on the relation between CEO duality and firm performance was mixed due to endogenous problems caused by the board leadership structure aspect within their research, therefore they applied a new framework that mitigated these problems. Namely, by using an exogenous shock they could measure the relation between post-shock performance and pre-shock board structure. They concluded that dual leadership firms exceed non-dual leadership firms by 3-4% in the event of a change in the competitive environment. This effect gets stronger the better the corporate governance and the greater the information costs.
3.0 Methodology
This section explains the model, sample & data source, the exogenous shock and empirical method applied in this research.
3.1 the model
With this paper I will be testing the relation between CEO duality and firm performance. The tested models are:
10 Return on equityit = β0 + β1Dual*Post-shockit + β2Post-shockit + β3DebtRatioit + β4FirmSizeit + β5Volatilityit + εit
Return on assetsit = β0 + β1Dual*Post-shockit + β2Post-shockit + β3DebtRatioit + β4FirmSizeit + β5Volatilityit + εit
Tobin’s Qit= β0 + β1Dual*Post-shockit + β2Post-shockit + β3DebtRatioit + β4FirmSizeit + β5Volatilityit + εit
The hypotheses tested for all models are:
H0: There is no difference in firm performance between dual leadership firms and non-dual leadership firms.
H1: There is a difference in firm performance between dual leadership firms and non-dual leadership firms.
H0: There is no difference in performance pre- and post-shock H1: There is a difference in performance pre and post-shock
It is unlikely that one indicator could essentially capture the entire performance of a firm, considering performance could be driven by different indicators (Begley and Boyd ,1986). For example growth, efficiency or quality. I decided to use three ratio’s as firm performance measurement.These variables consist of the return on equity, return on assets and the Tobin’s Q. Return on equity (ROE) is calculated by dividing the earnings before interest, taxes and depreciation by the common equity (Yang & Zhao, 2014). The ROE is a good proxy for firm performance, because it essentially shows how much profit is earned with the money invested by shareholders (Berk & DeMarzo, 2007). This could be an indication of how effective a firm finances its projects with equity.
The Return on Assets (ROA) is calculated by dividing the earnings before interest, taxes and depreciation by the total book assets (Yang & Zhao, 2014). The ROA is often used to determine the profitability of a firm in relation to its total assets. In comparison to the return on equity, it is less sensitive to leverage (Berk & DeMarzo, 2007)
The Tobin’s Q is calculated as dividing the market value of a firm by total book value of assets. Yang & Zhao (2014) argue that the Tobin’s Q is a good way to measure firm performance, considering it is frequently used in corporate finance literature, including studies on the relation between firm performance and board structure.
The independent variables consist of dual*post-shock, post-shock, debt ratio, firm size and volatility. The last three are control variables.
11 Dual is a dummy variable that equals 1 if the CEO is also the Chairman of the board and 0 if the CEO and Chairman are two independent people. Considering that there is an exogenous shock, we would like to test the firm performance pre- and post-shock. Therefore the dummy variable post-shock is added to the model. It equals 1 for the years 2001 up until 2006 and 0 for the years 1998 up until 2000.
The interaction variable Dual*Post-shock is added to see if there is a significant difference in firm performance between CEO duality firms and CEO non-duality firms.
The debt ratio is calculated by dividing long-term debt by total book assets. In essence stating how much of the assets is financed
by debt (Yang & Zhao, 2014). The variable firm size is calculated by taking the natural logarithm of the total assets (Dang et al., 2018). Lastly, volatility is calculated by taking the standard deviation of monthly stock return and
multiplying it with the square root of twelve (Yang & Zhao, 2014) See table 1 for the sample description.
3.1 sample & data source
The sample consist of 45 firms listed on the S&P 500. In order to get data for the variables I used Warton Research Data Service. Within this database there are multiple databases with each their own specification. For the ratio’s, ROE & ROA, the Financial ratio suite by WRDS was used. It provided the monthly time series per firm. For the duality dummy Institutional Shareholder Services (ISS) was applied. The yearly data gave the names of the chairman and CEO of each firm. Thereafter I manually wrote down the leadership structure, dual or separate. For the debt ratio, firm size and Tobin’s Q Compustat North America was used for the gathering of total assets, long-term debt and total market values. Lastly, database CRSP gave the monthly stock prices for the calculation of the volatility variable.
3.2 exogenous shock
In previous researches there were problems finding clear results, about the relation between CEO duality and firm performance, caused by endogeneity problems. CEO duality is related to the board leadership structure and therefore most likely endogenous. The reverse causality
table 1
Sample description
Variables N Mean Std. Dev.
Roe 405 29.37% 18.76% Roa Tobin’s Q 405 405 8.09% 1.29 7.74% 2.05 Total assets (in millions
$)
405 93184.41 221344.1
volatility 405 28.66% 13.25%
Debt ratio 405 11.85% 11.02%
12 behind it causes the problem. Seeing that firm performance could be an outcome of CEO duality,
but also the reason why CEO duality is adopted within a firm (Hermalin and Weisbach, 2001). Yang & Zhao (2014) used an exogenous shock in there research to mitigate this problem,
thereby disrupting the relation between the dependent variable (firm performance) and the endogenous independent variable (CEO duality).
This study faces the same problem and for this reason I am also using an exogenous shock to weaken the problem. The shock applied to this research is terrorist attack 9/11. On September 11 2001 four planes got hijacked by terrorist organization Al Qaida. Two of the planes crashed into the Twin Towers in New York, the third one crashed into the Pentagon in Washington D.C and the fourth crashed down in Shanksville. Almost 3000 people were killed during the attack (History.com, 2010). Because the attack was unexpected and improbable it counts as exogenous shock (brander, 1991; Thompson, 1993; Guadalupe and Wulf 2010)
3.3 empirical methodology
Considering this paper concerns a combination of the cross-sectional and time-series data, I will be using the panel data analysis to do my regressions. Panel data analysis, also known as
longitudinal analysis, observes a multitude of variables over time on a multitude of cross-sectional entities, which in this case S&P 500 firms. This analysis has its advantages, under which control of heterogeneity and reducing the collinearity between predictor variables. Thus resulting in a more exact inference of the model framework.
There are two main techniques when using the panel data analysis. There is the fixed effects model and the random effects model. By means of the Hausman test I will be deciding whether to use the fixed or random effects model. The Hausman test determines if there is a correlation between the error terms (ui) and the regressors. H0 states that both models are applicable, however the random effects model makes tighter assumptions, making it preferable. H1 states that only the fixed effects model is applicable. If the Prob>chi2 is smaller than 0.05 (significance level) H0 should be rejected.
4.0 Results
4.1 Hausman test
Column 1 in table 3 reported a p<0.05, therefore the null hypothesis is rejected. The fixed effect model will be used for the regression of model 1 with return on equity as dependent variable. Furthermore, column 2 and 3 from table 2 report a p>0.05. The null hypotheses will not be
13 rejected, therefore both fixed and random effects model can be used. However, the random effects model has tighter assumptions, so this one will be applied.
4.2 Impact of dual leadership on return on equity
Column 1 in table 3 show the regression results for the model with return on equity as dependent variable. The results show a positive relation between the coefficient dual*post-shock, but this relationship isn’t significant. The coefficient post-shock is significantly negative. This result suggest that the return on equity of firms were lower (-0.0623) post-shock, ceteris paribus. Furthermore, the F value is significant at 1%. Indicating that the model jointly explains the dependent variable significantly.
4.3 Impact of dual leadership on return on assets
Column 2 in table 3 report the regression results from the model with return on assets as dependent variable. The Coefficient dual*post-shock is significantly positive. This suggests that firms with CEO duality had higher return on assets (+0.0135) than firms with CEO non-duality, ceteris paribus. The coefficient post-shock is significantly negative. This indicates that the return on assets were lower (-0.0161) post-shock, ceteris paribus. Lastly, the Wald-Chi statistic is significant at 1%. Therefore, the model jointly explains the dependent variable significantly.
4.4 Impact of dual leadership on Tobin’s q
Column 3 in table 3 show the regression results from the model with Tobin’s Q as dependent variable. The coefficient dual*post-shock is significantly positive. This result suggests that firms with CEO duality had a higher Tobin’s q (+0.587) in comparison to firms with CEO non-duality, ceteris paribus. Furthermore, the coefficient post-shock is significantly negative. This suggest that the Tobin’s q was lower (-0.961) post-shock for all firms, ceteris paribus. Lastly, the Wald-Chi statistic is significant at 1%. Thereby saying that the model jointly explains the dependent variable significantly.
Table 2 Hausman test
(1) (2) (3)
roe roa Tobin’s
q
Chi2 35.06 2.10 1.13
Prob>chi2 0.0000* 0.8358 0.9514 This table reports the results from the Hausman test. (p<0.05*)
14 Table 3
Impact of dual leadership
(1) (2) (3)
Dependent variable= roe roa tobin’s q
dualpostshock 0.0401 0.0135** 0.587*** (0.0247) (0.00553) (0.201) postshock -0.0623** -0.0161*** -0.961*** (0.0253) (0.00546) (0.194) volatility -0.0224 -0.00973 -1.524*** (0.0594) (0.0133) (0.486) debtratio 0.973*** -0.0774** -3.746*** (0.178) (0.0375) (1.265) lnassets -0.0830*** -0.0164*** -0.467*** (0.0240) (0.00441) (0.133) Constant 1.067*** 0.267*** 7.371*** (0.240) (0.0456) (1.375) Observations 405 405 405 R-squared 0.172 Number of company F-value Wald-chi2 Fixed effects 45 14.70*** Yes 45 42.70*** No 45 82.44*** No This table reports the regression results for the models with return on equity, Return on assets and Tobin’s Q as dependent variable. The standard errors are in parentheses.
(*** p<0.01, ** p<0.05, * p<0.1)
5.0 Robustness check
In this section an alternative firm performance measurement will be used to check for
robustness of the model. The firm performance measurement applied is the natural logarithm of total sales. The regression consists of a new dependent variable, however the independent variables remain unchanged. The model is LnSalesit= β0 + β1Dual*Post shockit + β2Post shockit + β3DebtRatioit + β4FirmSizeit + β5Volatilityit + εit
The Hausman test concluded a p>0.05, H0 is therefore not rejected. Both fixed and random effects model can be used, but because the random effects model makes tighter assumptions, it is therefore preferred. The panel data analysis reported the following results in table 5. The coefficient dual*post-shock is significantly negative. This result suggest that CEO duality firms have 7.57% lower sales in comparison to CEO non-duality firms. Furthermore, the coefficient post-shock is positive, but not significant. Lastly, the Wald-chi statistic is significant at 1%. Therefore indicating that the model jointly explains the dependent variable significantly.
15 The results from the robustness check contradict the results from the first three models. It suggests that CEO duality is negatively related to firm performance.
Table 4
Impact of dual leadership- robustness check Dependent variable= Ln(sales)
dualpostshock -0.0757** (0.0305) postshock 0.0319 (0.0310) volatility 0.0346 (0.0733) debtratio 0.0830 (0.217) lnassets 0.729*** (0.0284) Constant 1.612*** (0.314) Observations 405 Number of company Wald-Chi 1013.11*** 45
This table report the regression results with the natural logarithm of sales as dependent variable. The standard errors are in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
6.0 Conclusion
5.1 summary
With this paper I tried to answer the question: Does CEO duality have an effect on financial firm performance? To answer this question I performed a literature review and multiple regressions. In the literature revolving CEO duality there were multiple arguments in favor or against dual leadership. The stewardship theory argued the benefits of CEO duality. Stating that a firm with an organizational structure that applies CEO duality will benefits from the unified leadership. The benefits include swift decision making and lower information costs. The agency theory argues the disadvantages of CEO duality. Stating that combining the two titles compromises the independency of both practices, thereby weakening the separation between ownership and control. This creates a circumstance that could lead to conflict of interest at the expense of the firm and shareholders.
16
5.2 discussion
The purpose of this study is to examine the effect of CEO duality on firm performance using an exogenous shock to the market. The shock is used to mitigate the endogenous problems arising from the board leadership structure aspect within this study. To answer the research question I tested three models using panel data analysis. The firm performance measurements were return on equity, return on assets and the Tobin’s q. The following results were reported. For all three models a significant negative relationship between firm performance and the post-shock variable was concluded. This suggests that firm performance was significantly lower post shock. This could be linked to the impact of the terrorist attacks on capital markets, but is unlikely. Chen & Siems (2004) concluded a significant negative impact of the terrorist attack on September 11 2001 on the capital market worldwide, but this effect is only significant up until the 11th day event window. The post shock dummy equaled 1 for the years 2001 up until 2006, therefore a much longer period. An implication for the lower firm performance could be the global financial crisis. Maybe the early effects of this crisis were influencing the firm
performances of the sample firms.
Furthermore, the coefficient Dual*Post-shock is positive for all three models. However, the relationship is only significant for the models with return on assets and Tobin’s Q as dependent variables. These results suggest that for these models the performance was higher for duality firms than for non-duality firms. Firstly, the Tobin’s Q performance measurement applied in this paper can be compared to the study of Yang & Zhao (2014). They also reported a significant positive relationship between the Tobin’s Q and CEO duality and concluded that CEO duality being beneficial whenever competition intensifies. This could also be true for this paper, considering the shock or perchance the financial crisis had an impact on competition intensity. You could argue that firms with CEO-duality have the ability to implement strategic plans more swiftly (Jensen & Meckling, 1995). Thereby reacting faster to market opportunities or obstacles. This may result in higher firm performance in comparison tot non-dual leadership firms. The return on assets and return on equity were used as a robustness checks in the paper of Yang & Zhao (2014), but not in this paper. They concluded no significant impact of dual leadership on return on assets. This is contrary to my results, considering I found a significant positive effect. The difference could be due to various factors. For example, sample differences, timeframe differences or model differences.
One limitation of this study is the relatively small sample. Due to time restriction and lack of data I had to shrink my sample size. If the sample size was of a larger amount, there would be a lot more data, leading to perhaps more significant results. Furthermore, because of the board leadership structure aspect within this study, I had to deal with endogeneity
17 this is that I don’t know how far the problem has been weakened. Furthermore, it was hard finding firms with a stable board leadership structure throughout the timeframe of this study. This could have consequences on the effect of the exogenous shock to mitigate the endogeneity problem. Looking at the data, the first limitation occurred when the majority of the data was only available yearly and not monthly. Monthly data would be preferable, for more precise results. Secondly, accounting variables, like return on assets and return on equity, are sensitive to manipulation by management (Yang & Zhao, 2014). Taking this into account, the reliability of the data provided by the firms could be questioned.
Bibliography
Berk, J. B., & DeMarzo, P. M. (2007). Corporate finance. Pearson Education.
Boyd, B. K. (1995). CEO duality and firm performance: A contingency model. Strategic
Management Journal, 16(4), 301-312.
Brander, J. A. (1989). Election polls, free trade, and the stock market: evidence from the canadian
general election (No. w3073). National Bureau of Economic Research.
Brickley, J. A., Coles, J. L., & Jarrell, G. (1997). Leadership structure: Separating the CEO and chairman of the board. Journal of corporate Finance, 3(3), 189-220.
Carty, R., & Weiss, G. (2012). Does CEO duality affect corporate performance? Evidence from the US banking crisis. Journal of Financial Regulation and Compliance, 20(1), 26-40.
Chen, A. H., & Siems, T. F. (2004). The effects of terrorism on global capital markets. European
journal of political economy, 20(2), 349-366.
Core, J. E., Holthausen, R. W., & Larcker, D. F. (1999). Corporate governance, chief executive officer compensation, and firm performance. Journal of financial economics, 51(3), 371-406. Daily, C. M., & Dalton, D. R. (1992). The relationship between governance structure and corporate performance in entrepreneurial firms. Journal of Business Venturing, 7(5), 375-386. Daily, C. M., & Dalton, D. R. (1993). Board of directors leadership and structure: Control and performance implications. Entrepreneurship: Theory and Practice, 17(3), 65-82.
Dalton, D. R., Daily, C. M., Ellstrand, A. E., & Johnson, J. L. (1998). Meta‐analytic reviews of board composition, leadership structure, and financial performance. Strategic management
journal, 19(3), 269-290.
Dang, C., Zhichuan, F.L., & Yang, C. (2018) Measuring firm size in empirical corporate finance.
Journal of Banking and Finance, 159-176
Donaldson, L., & Davis, J. H. (1991). Stewardship theory or agency theory: CEO governance and shareholder returns. Australian Journal of management, 16(1), 49-64.
18 Eisenhardt, K. M. (1989). Agency Theory: An Assessment and Review Academy of Management Review 14 (1): 57–74. CrossRef Google Scholar.
Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. The journal of law and
Economics, 26(2), 301-325.
Fama, E. F., & Jensen, M. C. (2008). Agency Problems and Residual Claims', Journal of Law and Economics, XXVI (2), June, 327-49. INTERNATIONAL LIBRARY OF CRITICAL WRITINGS IN
ECONOMICS, 214(3), 365.
Finkelstein, S., & D'aveni, R. A. (1994). CEO duality as a double-edged sword: How boards of directors balance entrenchment avoidance and unity of command. Academy of Management journal, 37(5), 1079-1108.
Goyal, V. K., & Park, C. W. (2002). Board leadership structure and CEO turnover. Journal of
Corporate Finance, 8(1), 49-66.
Guadalupe, M., Wulf, J., 2010. The flattening firm and product market competition: the effect of trade liberalization on corporate hierarchies. American Economic Journal: Applied Economics 2, 105–127.
Hermalin, B. E., & Weisbach, M. S. (2001). Boards of directors as an endogenously determined
institution: A survey of the economic literature (No. w8161). National Bureau of Economic
Research.
History.com staff (2010). 9/11 attacks. Retrieved from
http://www.history.com/topics/9-11-attacks
Hsiao, C. (2014). Analysis of panel data (No. 54). Cambridge university press.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.
Jensen, M. C., & Meckling, W. H. (1992). Specific and general knowledge and organizational structure.
Kesner, I. F., & Johnson, R. B. (1990). An investigation of the relationship between board composition and stockholder suits. Strategic Management Journal, 11(4), 327-336.
Krause, R., Semadeni, M., & Cannella Jr, A. A. (2014). CEO duality: A review and research agenda. Journal of Management, 40(1), 256-286.
Palanissamy, A. (2015). CEO DUALITY–AN EXPLORATIVE STUDY. European Scientific Journal,
ESJ, 11(10).
Rechner, P. L., & Dalton, D. R. (1991). CEO duality and organizational performance: A longitudinal analysis. Strategic Management Journal, 12(2), 155-160.
Seetaram, N., & Petit, S. (2012). Panel data analysis. Handbook of research methods in tourism:
Quantitative and qualitative approaches, 127-144.
Thompson, A. J. (1993). The anticipated sectoral adjustment to the Canada-United States Free Trade Agreement: An event study analysis. Canadian Journal of Economics, 253-271.
19 Yang, T., & Zhao, S. (2014). CEO duality and firm performance: Evidence from an exogenous shock to the competitive environment. Journal of Banking & Finance, 49, 534-552.
Appendix A
Table 2 Hausman test
(1) (2) (3)
roe roa Tobin’s
q
Chi2 35.06 2.10 1.13
Prob>chi2 0.0000* 0.8358 0.9514 This table reports the results from the Hausman test. (p<0.05*)
Table 3
Impact of dual leadership
(1) (2) (3)
Dependent variable= roe roa tobin’s q
dualpostshock 0.0401 0.0135** 0.587*** (0.0247) (0.00553) (0.201) postshock -0.0623** -0.0161*** -0.961*** (0.0253) (0.00546) (0.194) table 1 Sample description
Variables N Mean Std. Dev.
roe 405 29.37% 18.76% roa tobin’s q 405 405 8.09% 1.29 7.74% 2.05 totalassets (in millions
$)
405 93184.41 221344.1
volatility 405 28.66% 13.25%
debtratio 405 11.85% 11.02%
20 volatility -0.0224 -0.00973 -1.524*** (0.0594) (0.0133) (0.486) debtratio 0.973*** -0.0774** -3.746*** (0.178) (0.0375) (1.265) lnassets -0.0830*** -0.0164*** -0.467*** (0.0240) (0.00441) (0.133) Constant 1.067*** 0.267*** 7.371*** (0.240) (0.0456) (1.375) Observations 405 405 405 R-squared 0.172 Number of company F-value Wald-chi2 Fixed effects 45 14.70*** Yes 45 42.70*** No 45 82.44*** No This table reports the regression results for the models with return on equity, Return on assets and Tobin’s Q as dependent variable. The standard errors are in parentheses.
(*** p<0.01, ** p<0.05, * p<0.1)
Table 4
Impact of dual leadership- robustness check Dependent variable= Ln(sales)
dualpostshock -0.0757** (0.0305) postshock 0.0319 (0.0310) volatility 0.0346 (0.0733) debtratio 0.0830 (0.217) lnassets 0.729*** (0.0284) Constant 1.612*** (0.314) Observations 405 Number of company Wald-Chi 1013.11*** 45
This table report the regression results with the natural logarithm of sales as dependent variable. The standard errors are in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
