1 Introduction

Recent research has shown that the behavioral characteristics of chief executive officers (CEOs) may affect a company’s payout policy. The risk preferences of CEOs are among such behavioral characteristics. Given the significance of risk tolerance and a specific CEO’s appetite for risk, the board of directors set up a framework to determine the level of risk that the CEO should take. Within such a framework, the incentives component of executive pay packages could play an important role. Research shows that a CEO will pay out less to investors if his or her compensation plan is risk-oriented (Sundaram and Yermack 2007; Burns et al. 2015; Geiler and Renneboog 2016), and a CEO will pay out more if the compensation is less risk-oriented (Minnick and Rosenthal 2014). Risk-averse CEOs also tend to pay higher dividends despite market trends and investor preferences (Sundaram and Yermack 2007; Caliskan and Doukas 2015).

In addition to compensation policies, a board of directors may use its monitoring power to induce CEOs to pay out more (Bhabra and Luu 2015; Yarram and Dollery 2015; Detthamrong et al. 2017; Green and Homroy 2018). It is assumed that if the board of directors is not too small and not too big, and/or if the number of independent directors and women on the board is optimal, the board will be efficient in setting corporate policies and will have sufficient monitoring power. While a CEO’s risk preferences may be influenced by different board policies, there is still no clear evidence as to whether corporate governance reduces the possible detrimental effects of CEO behavior in payout decisions.

Although the literature shows that risk preferences may affect corporate decision-making, there are some limitations. First, the results for total payout are mixed: some authors find positive relationships (Geiler and Renneboog 2016) and others find negative relationships (Cuny et al. 2009). Second, the impact of a CEO’s risk preferences on the decision to start paying out has not yet been adequately examined. Third, the influence of a CEO’s risk preferences on the decisions to switch between dividends and share repurchases is also under-studied. Fourth, there are no significant results on the ability of corporate governance to overcome the negative effects of a CEO’s risk preferences on various aspects of payout policy.

Thus, in filling these informational gaps we aim to improve the understanding of the role of boards of directors in eliminating the negative effects of CEO risk preferences on payout decisions. This paper provides new empirical evidence on the ability of the strategic oversight of boards to offset the possible negative impact of a CEO’s risk preferences on payout policy and the choice between dividends or share repurchases.

To investigate the remuneration policy as a tool for defining a CEO’s risk preferences, we not only examine cash compensation, but also compensation by way of restricted stocks and the relative proportion of total stocks that are owned by the CEO. Cash compensation and restricted stocks have not been assumed to encourage a CEO to take higher risk, because the former component is accounted for by salary, which in almost all cases does not depend on the company’s value, whereas the latter component is used to compensate for the achievement of long-term goals. Thus, such CEOs may be reluctant to invest in high-risk projects and may distribute money among the shareholders instead.

In contrast, a higher share of stocks in the CEO’s remuneration may stimulate CEOs to take additional risks to increase the expected return of the companies’ stocks in the short-term and to make some speculative profits. Dividends decrease the price of the shares and the value of the CEO’s portfolio. As a result, he or she may end up paying out less to the shareholders and not acting in their interests.

Our findings add to the literature in the following ways. First, we fill in the gaps in understanding how CEO risk preferences impact the decision to initiate paying dividends. We then add new empirical results on the role of compensation schemes set up by the boards for a CEO (to prevent him or her from taking more risk) and document how this stimulates higher levels of payout. Given these results, we demonstrate how the boards influence CEO risk preferences through remuneration policies. Our empirical evidence shows that risk preferences affect the probability of starting the payment of cash dividends: less risky CEOs are more likely to initiate cash dividends than riskier ones. The decision to initiate share repurchases is not affected by the risk preferences of the CEOs.

Second, our findings show that the quality of the board’s work matters for overcoming the possible negative impact of CEO risk preferences on payout policies. By introducing our corporate governance quality index, we differentiate between companies which have higher and lower rankings and show that higher-quality governance may reduce or even eliminate the negative effects of CEO risk preferences on the payout policy. We also show that the ability of corporate governance to eliminate the negative impact of CEO behavior on the payout policy decreases for companies with the highest levels of payout. Risk preferences still significantly influence the level of payout and the choice of payout channel in these companies. We assume that shareholders are satisfied with such levels of payout and do not ask for protection against the negative effects of CEO risk preferences.

Third, we provide evidence for the impact of CEO risk preferences on the choice of payout channels. We show that CEOs who are encouraged to take more risks pay out more through repurchases than through dividends to shareholders: the level of total payout is made up mostly of repurchases. We divide the sample into quartiles by the size of total payouts and show that companies with the highest levels of payout are more likely to follow a policy of repurchases than companies with the lowest levels of payouts. We show that companies from top quartiles are more affected by the CEOs risk preferences than the companies from the lowest quartiles. From this, we may conclude that the level of repurchases is more sensitive to the CEO’s risk preferences in the companies from top quartiles. Therefore the fundamental financial variables, and not the behavioral ones, determine payout decisions in the companies from the lowest quartiles. It may ultimately be postulated that when a company starts to generate more cash flows, and hence more cash is made available for distribution among shareholders, the CEO’s decisions on payout policy become more acutely affected by his or her risk preferences. CEOs may therefore be more inclined to seek more investment opportunities with high risks instead of those with less risky but lower payouts.

We organize the paper as follows. In Sect. 2, we review the payout policy literature with respect to the risk preferences of CEOs and the literature on the ability of corporate governance to influence the strategic policies of companies, including the payout policy. Section 3 outlines the hypotheses, and discusses models and data. Section 4 presents the results. Section 5 concludes the analysis and introduces possible future research agendas.

2 Literature review

Although there are several conceptual explanations of dividend policies and their empirical tests (based on the assumption of rational behavior), they cannot fully explain the drivers of payout decisions. These approaches are not sufficient to fully account for agent behavior, especially when it is subject to biases. The existence of a number of such biases (for example, overconfidence, over-optimism, hindsight, anchoring) and their ability to affect the decisions of top management are already recognized conceptually and have been confirmed in empirical papers (Kahneman and Tversky 1979). In addition to these anomalies, the patterns of behavior of top management, based on a variety of managerial traits beyond individualism and opportunism may be significant drivers of payments to shareholders. According to behavioral studies in finance, CEOs act within bounded rationality; their decisions are significantly influenced by cultural values, emotions and cognitive biases (Anilov 2017). Given the leading role of CEOs in the decision-making process, there are a number of existing studies on their personal traits and behavioral biases. Such studies attempt to approximate, for example, the behavioral patterns of top management and their consequences for key financial decisions on strategic deals (Graham et al. 2013) and capital structure (Chava and Purnanandam 2010). However, direct empirical evidence outlining the behavioral foundations of payout policies is still missing (Baker and Wurgler 2013; Breuer et al. 2014).

In this section, we discuss the literature with respect to (1) the determination of CEO risk preference in relation to compensation policy; (2) the effects of risk preferences on the different aspects of payout policy; (3) the quality of corporate governance.

2.1 Remuneration policy and CEO risk preferences

Managers are assumed to be affected by several biases, which are related to various levels of risk preferences. Given these biases, CEOs may promote a payout policy that may not be the one most favorable for shareholders. It has been shown that the pay-for-performance mix can motivate an agent to change his or her appetite towards risk and therefore it has an impact on corporate policies. Risk-taking CEOs are paid with a higher proportion of performance-based compensation packages and less with cash-based packages.

Boards introduce equity-based payments, particularly in the form of stock options, to induce optimal risk-taking behavior (Financial behavior 2017). Stock options allow a CEO to make a profit in two ways. First, a CEO may want to increase the company’s equity value to exercise options to get more profit. In this case, a CEO must combine profitable projects with positive NPV while also concentrating on risk control. This is to keep the spread between return on capital and the required rate of return at a positive level for the overall portfolio of projects. Second, he or she may want to increase the volatility of underlying shares to increase the value of the related options. To do so, the CEO needs to invest more heavily in projects with higher risk. As the CEO’s investment set increases, so does the risk (Caliskan and Doukas 2015). Higher risk results in higher volatility of the company’s stocks, which leads to an increase in the value of executive stock options. This is why he or she would rather pursue investing in projects with a higher-than-average risk factor (from the company’s perspective), hoping that it will boost the company’s capitalization, volatility and the CEO’s pay (Douglas 2007; Sundaram and Yermack 2007; Burns et al. 2015; Geiler and Renneboog 2016). The evidence also shows that lower risk levels prevail in those firms with low levels of stock options pay in their CEO incentive plans, compared to companies with risky CEOs (Low 2009).

To motivate CEOs to adopt less risk, compensation may be tied to the market value of the company’s debt, which has a negative correlation with risk (Sundaram and Yermack 2007). If the board of directors implement this compensation policy, the CEO becomes a creditor of the company and does not benefit from an increase of the share price or its volatility. Instead, he or she will allocate as many resources within the company as possible to decrease the probability of default. This makes the CEO less risky. If the compensation of the CEO is based only on salary and bonuses, he or she is not encouraged to increase the company’s value. In these cases, the investment set may be limited to projects with low risk that guarantee an acceptable level of cash flow (Berger et al. 1997).

Therefore, the compensation policy may define the risk preferences of the CEO through the available investment set. Now, we move to a discussion of how risk preferences may affect the payout policy.

2.2 CEO risk preferences and payout policy choice

We  continue with a discussion of how risk preferences may affect the level of shareholders payout. If the CEO’s compensation is equity-based, especially in the form of stock options, the CEO does not have any strong risk burdens. Such CEOs will allocate more money to investment projects pursuing high returns and will choose the projects that are riskier. As a result, they will be left with a lower cash flow and will pay out fewer dividends (Douglas 2007; Burns et al. 2015; Geiler and Renneboog 2016). On the other hand, such CEOs may consider the shares of their company to be undervalued and will distribute more through share repurchases. However, the increase in repurchases is usually not enough to cover the dividend reduction, so the total payout will be less if the CEO is a risk-taker (Cuny et al. 2009).

The companies where compensation is tied to the market value of the company’s debt pay out more on average, as CEOs avoid risky projects and have free cash flows that can be distributed among shareholders (Caliskan and Doukas 2015). In addition, the risk appetite of the CEO may also be reduced, if the compensation scheme is built on restricted stock units (RSU—those which cannot be sold before a specified point in the future but that bear dividends). Such a remuneration policy encourages CEOs to pursue long-term goals. If a CEO takes additional risks, he or she may not achieve these goals in the future and will not get compensation. The CEO will choose investment projects more carefully, with lower risk and will have more cash to be distributed among shareholders (Minnick and Rosenthal 2014). If such compensation plans are not used in the company, the shareholders will be left with lower levels of dividends.

The risk preferences of the CEO may affect not only the level of dividends or repurchases, but also the choice of payout channel. The literature provides some insights into the relationship between stock-option-based compensation and the choice of share repurchase to pay shareholders (Kahle 2002). The use of executive stock options and restricted stock by boards is associated with a reduction in cash dividends and a shift to share repurchases (Aboody and Kasznik 2008). Geiler and Renneboog (2016) come to the same conclusion and show that the use of stock options and RSU as remuneration is positively related to the choice of share repurchase.

CEO risk preferences do not always serve to increase shareholder wealth. Research shows that these adverse effects can be mitigated through corporate governance practices. Given both the monitoring and the conformance roles of boards, the directors aim to reduce agency conflicts and to provide strategic oversight of a company. The empirical evidence shows that the more efficient the mechanism of corporate governance, the more the company pays out to investors (Jiraporn et al. 2011; Sharma 2011; Ambardnishvili et al. 2017). CEOs may be forced to pay out more due to better protection of shareholders’ rights in such companies.

In the next subsection, we discuss the existing approaches to measuring the quality of corporate governance.

2.3 The quality of corporate governance

Several approaches have been developed to define the quality of corporate governance. The first approach is to use an index that is based on several measures chosen by the authors. The elements of the index may include gender and age diversity (Bernile et al. 2018; Cosma et al. 2018); the size of the board of directors and its committees (Chan et al. 2014; Nguyen et al. 2016; Ararat et al. 2017); the level of the company’s transparency (Braga-Alves and Shastri 2011; Hwang et al. 2013); the presence of independent directors on the board and in committees (Mande et al. 2012). These researchers conclude that high-quality corporate governance increases the company’s value, shareholder payouts, and reduces the agency problem. The second approach is to use commercial indexes, which are provided by professional agencies, for example, RiskMetrics (Zagorchev and Gao 2015), G-Index (Chang et al. 2014), ISS (Jiraporn et al. 2011; Zhu 2014), and Globe&Mail (Adjaoud and Ben-Amar 2010). These authors conclude that high-quality corporate governance increases operational efficiency, increases shareholder payouts, and reduces the cost of capital.

In this study, to assess the quality of corporate governance, we have developed an index. As a huge number of elements with equal weights may increase measurement errors (Bozec and Bozec 2012), we limit our index to 5 components. We also focus only on the quality of the board of directors as the main corporate governance body to capture its effects on the relationship between payout policy and the CEO risk preferences. We discuss the index more thoroughly in the next section.

The evidence on the ability of the board of directors to offset the negative effects of the CEO’s risk preferences on payout policy is rather limited. To deeper understand the role of the strategic direction of the board in payout decisions, it is very important to fill the gaps in studies on the impact of CEO risk preferences. The literature still demonstrates contradictory results on the effects of CEO risk preferences on both the level of payouts and the choice of payout channel. There is limited evidence for the effect of CEO risk preferences on the decision to initiate payments to shareholders (repurchases or cash dividends).

To address these issues, this paper provides new empirical evidence on the role of the board of directors in protecting shareholder interests against the adverse effects of the CEO’s risk preferences.

3 Hypotheses development, model and data

Research findings suggest that the most conservative policy is to pay dividends. Risky CEOs are more likely to stick to a policy of increasing investments in projects that are associated with higher than average risks (relative to his or her specific company). This approach implies higher levels of volatility in expected future cash flows. This increase in volatility may result in a decrease in the amount of cash available for distribution among shareholders, and consequently lower payouts. To investigate the relationship between CEO risk preferences and payout policy, we study the variation in the levels of payouts, the decisions to start paying out, and the changes in the repurchase-dividend mix.

We use several compensation-based proxies to measure CEO risk preferences. Based on previous results, we assume that the compensation scheme aligns the CEO’s risk preferences with those of the board of directors and shareholders. Following the literature, we apply the fraction of the total cash amount of the CEO’s compensation, the fraction of company shares owned by the CEO (Burns et al. 2015) and the fraction of compensation in the form of restricted stocks (Minnick and Rosenthal 2014). We also control for executive option-based compensation schemes by the ratio of exercisable options to the total executive options. This measure may reflect the level of CEO overconfidence—another characteristic of CEO behavior (Fenn and Liang 2001; Deshmukh et al. 2013). Equity-based compensation, due to the capital gains of the CEO, may involve him or her gambling on the high marginal cost of investing in projects which forego cash dividends, thus yielding different utility (Kahneman and Tversky 1979).

We also assume that risky CEOs may prefer repurchases rather than cash dividends. Such managers may consider the company’s stocks undervalued and be willing to repurchase them at what they think is a low price (Sundaram and Yermack 2007; Geiler and Renneboog 2016).

Finally, the age of the CEO may be a proxy for a type of CEO risk preferences. There is evidence that younger CEOs pursue risky investment policies, seeking riskier (and more lucrative) components in their compensation plans (Kempf et al. 2009; Serfling 2014).

CEOs who are risk-takers will search for funds to initiate additional risky investment projects. Additional capital expenditure will lead to a decrease in the level of payout (net income being constant) (Minnick and Rosenthal 2014) and to postpone or even avoid initiating payouts altogether (Burns et al. 2015).

Therefore, we test the following hypotheses:

Hypothesis 1

The higher the risk preferences of the CEO, the lower the level of both cash dividends and repurchases.

Hypothesis 2

The higher the risk preferences of the CEO, the lower the probability of initiating both cash dividends and repurchases.

Hypothesis 3

The higher the risk preferences of the CEO, the more the company switches to repurchases.

To test these hypotheses, we used the following models:

$$\begin{aligned} Payout_{i,t} & = \alpha + \beta_{1} \cdot Payout_{i,t - 1} + \beta_{2} \cdot RiskPref_{i,t} + \beta_{3} \cdot Age_{i,t} \hfill \\ & \quad + \beta_{4} \cdot Ex\_Opt_{i,t} + \mathop \sum \limits_{k = 5}^{12} \beta_{k} \cdot Control_{i,t,k} + \theta_{i} + \delta_{t} + \varepsilon_{i,t} \hfill \\ \end{aligned}$$
(1)
$$\begin{aligned} pr(DTP_{i,t} = 1) &= \varphi \Bigg\{ \mu + \gamma_{1} \cdot DTP_{i,t - 1} + \gamma_{2} \cdot RiskPref_{i,t} + \gamma_{3} \cdot Age_{i,t} + \gamma_{4} \cdot Ex\_Opt_{i,t} \\ &\quad+ \mathop \sum \limits_{k = 5}^{12} \gamma_{k} \cdot Control_{i,t,k} + \theta_{i} + \delta_{t} \Bigg\} , \end{aligned}$$
(2)

where \(Payout_{i,t}\) is one of the three “Payout” variables; \(pr(DTP_{i,t} = 1)\) is the probability that \(DTP_{i,t}\) = 1; \(DTP_{i,t}\) is one of the two “Decision to pay” variables; \(\varphi \left\{ x \right\}\) is the standard normal cumulative distribution function; \(RiskPref_{i,t}\) is the set of “Risk preferences” variables; \(Age_{i,t}\) is the age of the CEO; \(Ex\_Opt_{i,t}\) is the ratio of the value of exercisable options to the value of all executive options; \(Control_{i,t,k}\) is the set of control variables; \(\alpha ,\beta_{k} ,\mu ,\gamma_{k}\) are regression coefficients; \(\varepsilon_{i,t} ,\) are normally distributed error terms; \(\theta_{i}\) are industry effects; \(\delta_{t}\) are the year’s effects; i is the company index; t is the year index.

The definitions of the variables for the Models (1) and (2) are presented in Table 1.

Table 1 The variables
Full size table

Table 1 summarizes four specifications to measure the risk preferences of the CEO. We assume that the use of cash compensation and RSU both tend to lower the risk preferences of CEOs. As the CEO has no incentives to boost capitalization in the short-run or volatility, he or she might choose less risky projects with more certain outcomes and therefore a CEO is more likely to distribute cash. In contrast, stock compensation may encourage CEOs to bear additional risks to increase stock return in the short-run. This is why we think that CEO ownership should have a negative effect on payouts.

As shown, it is assumed that older people are more cautious and less willing to take certain risks. Given these previous results, we use the CEO’s age to capture the attitude towards risk.

We assume that the value of exercisable options which have not yet been exercised may reflect CEO overconfidence, as he or she is confident of a stock price increase and postpones the decision to exercise the options. Exercisable options are those for which the vesting period has already expired and that can be exercised at any time from now until the expiration date and are already “in the money”. Such CEOs may be more willing to repurchase stocks if they consider them undervalued. However these CEOs may be reluctant to pay dividends as he or she does not want dividend payouts to negatively affect the value of the stock options. As a result, the impact on the total payout may be mixed. We summarize our predictions in Table 2.

Table 2 Predicted signs of the impact of risk preferences, age and overconfidence on the payout ratios
Full size table

Given the monitoring and conformance roles of corporate governance mechanisms, we study not only how the board aligns the risk preferences of the CEO with the required risk levels of the corporate strategy by the induced compensation scheme, but also its capability to overcome the negative effects of the CEO’s risk preferences. We construct the corporate governance quality index (CGQI) based on the following board characteristics which were studied in prior research:

  1. 1.

    The gender diversity of the board (Green and Homroy 2018).

  2. 2.

    The percentage of independent directors (Black et al. 2012; Zagorchev and Gao 2015).

  3. 3.

    CEO duality (Yarram and Dollery 2015).

  4. 4.

    The frequency of board meetings (Black et al. 2012).

  5. 5.

    The size of the board (Mande et al. 2012).

We then apply the principal components analysis with the use of a correlation matrix to derive the value of the index. To construct the index, we use two components with the highest values. Once the index is standardized, we use the dummy variable to distinguish between governance of good quality (dummy = 1, provided the value of the index is greater than the sample’s average) and governance of poor quality (dummy = 0, provided the value of the index is less than the sample’s average). In the next section, we discuss the variability of the index in our sample.

Finally, we test the impact of governance with the following hypothesis:

Hypothesis 4

The high quality of the board of directors reduces the negative effects of the CEO’s behavior on payout policy.

To test Hypothesis 4, we add a dummy variable for high-quality corporate governance. We then extend Models (1) and (2) and assess Models (3) and (4):

$$\begin{aligned} Payout_{i,t} & = \alpha + \beta_{1} \cdot Payout_{i,t - 1} + \beta_{2} \cdot RiskPref_{i,t} \hfill \\ & \quad + \beta_{13} \cdot RiskPref_{i,t} \cdot D_{i,t} + \beta_{3} \cdot Age_{i,t} + \beta_{14} \cdot Age_{i,t} \cdot D_{i,t} \hfill \\ & \quad + \beta_{4} \cdot Ex\_Opt_{i,t} + \beta_{15} \cdot Ex\_Opt_{i,t} \cdot D_{i,t} \hfill \\ & \quad + \mathop \sum \limits_{k = 5}^{12} \beta_{k} \cdot Control_{i,t,k} + \beta_{16} \cdot D_{i,t} + \theta_{i} + \delta_{t} + \varepsilon_{i,t} \hfill \\ \end{aligned}$$
(3)
$$\begin{aligned} pr(DTP_{i,t} = 1)& = \varphi \Bigg\{ \mu + \gamma_{1} \cdot DTP_{i,t - 1} + \gamma_{2} \cdot RiskPref_{i,t} + \gamma_{13} \cdot RiskPref_{i,t} \cdot D_{i,t} \\ &\quad+ \gamma_{3} \cdot Age_{i,t} + \gamma_{14} \cdot Age_{i,t} \cdot D_{i,t} + \gamma_{4} \cdot Ex\_Opt_{i,t} + \gamma_{15} \cdot Ex\_Opt_{i,t} \cdot D_{i,t} \\ &\quad+ \mathop \sum \limits_{k = 5}^{12} \gamma_{k} \cdot Control_{i,t,k} + \gamma_{16} \cdot D_{i,t} + \theta_{i} + \delta_{t}\Bigg \} , \end{aligned}$$
(4)

where \(D_{i,t}\) is the dummy variable for high-quality corporate governance; \(\beta_{13}\), \(\beta_{14}\), \(\beta_{15}\) and \(\gamma_{13}\), \(\gamma_{14}\), \(\gamma_{15}\) are the coefficients for companies with high-quality governance.

If corporate governance eliminates completely the impact of CEO behavioral characteristics on their decisions, then the following equations should hold:

$$\beta_{2} = - \beta_{13} \;\;{\text{and}}\;\;\gamma_{2} = - \gamma_{13} ;$$
$$\beta_{3} = - \beta_{14} \;\;{\text{and}}\;\;\gamma_{3} = - \gamma_{14} ;$$
$$\beta_{4} = - \beta_{15} \;\;{\text{and}}\;\;\gamma_{4} = - \gamma_{15} .$$

We use Wald statistics to check whether these equations hold.

In addition to these variables and based on previous research (see Table 1), we use a set of control variables (Cash holdings, Tobin’s Q, Debt-to-Equity ratio, Long-term Debt ratio, Capital and R&D expenditures, ROA and Size) representing the financial position of the company. To capture possible effects, we also included industry dummies and year dummies.

To sum up, unlike previous studies, we include in the analysis the relationship between behavioral characteristics and the repurchases-dividends mix, the impact of risk preferences on the decision to initiate payouts, and the influential power of corporate governance. We also check the results for different quartiles of the levels of payout.

We collect a sample of non-financial and non-utility companies from the US for 2007 to 2016 from the S&P 1500 Index, which represents the largest and most stable companies in the US. We further restrict the sample to companies that had a positive payout at least once during the period of observation. After adjusting for missing data and outliers, we come up with a final sample of 671 companies. The data was obtained from the S&P Capital IQ and Bloomberg databases.

To assess Models (1) and (3), we use the dynamic panel data method, namely the Arellano-Bond estimator. We do so because lags are included in our specifications causing endogeneity problems. We also report Arellano-Bond tests for autocorrelation, and the Hansen test for specification. To address the lagged dependent variable and the initial conditions problem, for Models (2) and (4) a panel probit model regression has been applied (Wooldridge 2005). For all models the robust standard errors at firm level and standardized variables have been used.

To check our predictions for companies with different levels of payout, we implement quantile regressions for the 25th, 50th and 75th percentiles of the sample. Given the panel structure of data and endogeneity, we use Powell’s estimator (Powell 2016).

4 Empirical results

Table 3 presents the descriptive statistics for the sample (for the purpose of this table we use unstandardized variables).

Table 3 The descriptive statistics
Full size table

Table 3 shows that the companies in our sample differ in various respects: from companies with a high concentration of CEO ownership, to companies where no stocks are owned by the CEO; from companies with high payout ratios and to those with no payouts; companies with very high levels of debt, and companies with no debt. There are also companies with different board quality levels, but we can see that most companies in our sample have a high CGQI value. Within the sample, repurchases are, on average, more common than cash dividends (the average repurchase ratio for our sample is 0.035 and the average dividend ratio is 0.014). These are in line with previous findings (Fama and French 2001). Given the changes in the fractional amount of repurchases relative to the total payout for the period from Fig. 1, the data suggest that repurchases have been becoming increasingly popular.

Fig. 1
figure 1

The dynamics of the fractional amount of repurchases relative to the total payout

Full size image

We can see from Fig. 1 that from 2009 to 2011 and from 2012 to 2015 both the mean and median fraction of repurchases in the total payout increased. The major shocks that happened in 2009 (the “Great Recession”) and 2012 (tax reform and the tightening of monetary policy) dramatically reduced the overall fraction of repurchases, but the subsequent trends were upward and in 2014 repurchases reached pre-recession levels. To consider the effects of 2009 and 2012 we use dummy variables for both years in our Models.

Table 4 presents the means for variables in 4 quartiles divided by the total payout. The fraction of repurchases increased for the companies with the highest levels of payout. This means that the companies that pay out more prefer repurchases rather than cash dividends. The companies that pay out less prefer cash dividends instead. This shows that companies tend to distribute some base level of funds among shareholders through cash dividends and distribute extra funds through repurchases. We can also see that companies from the upper quartile use less cash and shares, but more stock options as part of their compensation policies.

Table 4 Mean values for the 1-st, 2-nd, 3-rd and 4-th quartiles
Full size table

Table 5 provides the correlation matrix for the chosen variables. All the correlations are below 50%, which means that there will be no multicollinearity in the Models.

Table 5 Correlation matrix
Full size table

Table 6 summarizes the results of the tests for Models (1) and (2). Tables 7, 8 and 9 summarize the results for the 25th, 50th and 75th percentiles, respectively.

Table 6 Results of testing Hypotheses 1, 2 and 3
Full size table
Table 7 Determinants of the repurchase ratio per quartiles
Full size table
Table 8 Determinants of the dividend ratio per quartiles
Full size table
Table 9 Determinants of the fraction of repurchases per quartiles
Full size table

Hereafter the results are reported for the five dependent variables that are described in Table 1: repurchase ratio, cash dividend ratio, fraction of repurchase, decision to initiate repurchases, and decision to initiate cash dividends.

In line with our predictions from Table 2 and previous findings (Caliskan and Doukas 2015) less risky CEOs tend to distribute more funds among the shareholders. Table 6 shows that restricted stock compensation stimulates CEOs to increase the level of repurchases. We can see that a 1 standard deviation increase in restricted stock compensation leads to a 0.129 standard deviation increase in the repurchase ratio. CEO ownership and exercisable executive options also lead to an increase in the repurchase ratio. We think that CEOs with large stock holdings tend to repurchase stocks in order to signal to markets that a company’s stocks are undervalued. This should result in more demand for this stock and a price increase. The value of the CEO’s portfolio should increase as well. As we pointed out, more options that are exercisable may be evidence of a CEO’s overconfidence in the undervaluation of his or her company’s stocks. Again, such a CEO prefers repurchases: a 1 standard deviation increase in exercisable option compensation increases the repurchase ratio by 0.031 standard deviations. The impact of the CEO’s age on payout policy was not significant.

When we implement the quantile regressions (Table 7), we can see that exercisable options increase the level of repurchases in companies from all quartiles. The impact of exercisable options increases with quartiles. The impact of restricted stocks and CEO age also increases with quartiles. It is conceivable that when the level of payout through repurchases in these situations starts to increase, CEOs start to be affected by their risk preferences. When the level of repurchases is low, the payout policy is mostly determined by financial variables, because CEOs seek opportunities to increase payout ratios with limited resources. When the level of repurchases is high, the impact of risk preferences increases. This may happen if the CEO, having satisfied all the demands of the shareholders, starts to look for investment opportunities in accordance with his or her risk preferences.

The risk-preferences and overconfidence of CEOs have no impact on the cash dividend ratio according to the results from Table 6. Only the control variables represented by financial measures are important for the level of cash dividends: cash holdings (a 1 standard deviation increase in cash holdings decreases the dividend ratio by 0.069 standard deviations), Tobin’s Q (a 1 standard deviation increase in Tobin’s Q increases the dividend ratio by 0.135 standard deviations), and debt-to-equity ratio (a 1 standard deviation increase in debt-to-equity ratio decreases the dividend ratio by 0.097 standard deviations).

However, when we analyze the per quartile results from Table 8, the risk preferences and the overconfidence of the CEO start to influence the dividend ratio in companies with median and high levels of dividends. When accounting for the impact of the increase in RSU and stock option compensation and the age of the CEO, the level of dividends also increases. It may happen because in the companies from low quartiles the dividend policy should be defined by the limited financial resources. When the available funds increase, the CEO’s decisions may be influenced to a higher extent by CEO’s risk preferences. As shown in Table 8, age affects dividends positively and repurchases negatively. The magnitude of this negative effect on repurchases increases as one looks at the quartiles from lowest to highest. The older CEOs, being less risky, prefer dividends rather than repurchases. This may be due to the fact that dividends have traditionally been more popular than repurchases and have been displaced by repurchases only recently.

What are the determining factors which influence the switching of an approach based on repurchases to one based on cash dividends? The level of exercisable options has a positive effect on the fraction of repurchases in the total payout. Again, this is a result of the CEO’s belief that the company’s stocks are undervalued. Another explanation is that dividends have a negative effect on the value of stock options. In order to eliminate these negative effects, companies may use repurchases that do not decrease options’ value. The level of cash compensation also has a significant impact on the fraction of repurchases: a 1 standard deviation increase in the level of cash compensation decreases the repurchase fraction by 0.118 standard deviations. This means that less risky CEOs prefer cash dividends rather than repurchases. This could be explained by the absence of sound policies to induce CEOs to create share value because their compensation is not based on the equity value. As a result, shareholders are left with a base level of dividends and do not receive additional cash distributions in the form of repurchases.

Table 9 shows that the impact of restricted stocks and exercisable stock options increases with the quartiles. The fraction of repurchases in total payout is stronger affected by the CEO’s behavior in companies with the highest levels of payouts, and is less affected in the companies with the lowest levels of payout.

This argument is also supported by the significant positive impact of the level of cash compensation on the probability of initiating cash dividends. From Table 10 we can see that a 1 standard deviation increase in the level of cash compensation leads to an increase in the probability of initiating paying dividends by 0.033 standard deviations. The level of exercisable options also has a positive impact on the probability of both initiating repurchases and initiating the payment of cash dividends: 1 standard deviation increase in the level of exercisable options increases the probability of repurchases and cash dividends by 0.018 and 0.029 standard deviations respectively.

Table 10 Marginal effects for the model (2), on average
Full size table

Our findings show that the previous levels of both repurchases and cash dividends have a significant positive impact on the current levels of payouts. This means that dividends are “sticky”, and CEOs are reluctant to change their payout policies. For all specifications (except of the model with the dividend ratio), ROA affects payout policy positively and the dummy variables for 2009 affect payout levels negatively, which can also be observed in Fig. 1. R&D expenses also have positive effects on the levels and the probability of repurchases. Therefore, profitable companies do not consider investments and payouts as substitutes for each other but having a strong cash flow tend to increase both investments and payouts. We also found that the origin of the industry matters only in terms of the level of repurchases and the decision to initiate repurchases. For other specifications of the payout policy, the impact of industry is not robust.

To verify Hypothesis 4, on the mitigating role of the boards, we assess models (3) and (4) and make linear tests on coefficient equality. We also check whether both coefficients are statistically significant. Cells in Tables 11 and 12 are highlighted with italics if both coefficients are significant and the test shows that the equations \(\beta_{n} = - \beta_{m}\) and \(\gamma_{n} = - \gamma_{m}\) hold. If the equations do not hold, we highlight them with bold font and italics. If the equations do not hold but the coefficients have different signs, we use bold font, which means that corporate governance reduces but does not completely eliminate the effects of the CEO’s risk preferences, age, and overconfidence. We do not use any highlight if at least one of the coefficients is insignificant in the first place. These results are summarized in Tables 11 and 12.

Table 11 Results of testing the ability of corporate governance quality to reduce the negative effects of CEO’s risk preferences
Full size table
Table 12 Results of testing the ability of corporate governance quality to reduce the negative effects of CEO’s risk preferences per quartiles
Full size table

According to the results in Tables 6 and 11, we can conclude that efficient corporate governance may eliminate the negative effects of the CEO’s behavioral traits, namely overconfidence, on the fraction of repurchases and decision to repurchase. The boards with better governance also reduce the influence of the personal risk preferences on the level of repurchases. The quality of corporate governance does not reduce the negative effects of the CEO’s ownership, exercisable options, and age on the level of repurchases and the effect of CEO’s risk preferences on the fraction of repurchases and decisions to pay dividends. Thus, the results for the whole sample document that corporate governance has a limited ability to overcome the negative effects of CEO’s behavioral characteristics on payout decisions. However, the overall picture of governance impact changes when we study these interrelations for each quartile of the sample. The results are presented in Table 12.

We can now also differentiate the impact of corporate governance quality on the effects of risk preferences, age, and overconfidence between the companies with low and high levels of dividends and repurchases. Table 12 shows that corporate governance has limited power to completely eliminate the negative effects of a CEO’s risk preference. However, it has an ability to significantly reduce these effects in companies with the lowest levels of payouts, but not in the companies with the highest levels of payout. We can see that with the increase in quartile, with the exception of cash compensation level, the number of cells highlighted with bold font or italics decreases while the number of cells highlighted with both bold font and italics increases. This is especially clear for the repurchase ratio and the fraction of repurchases. The shareholders in these companies might still be satisfied with the level of payout, even though they do not get the maximum payout, i.e. what the level of payout would have been if the CEO’s risk preferences had not influenced payout decisions. However, corporate governance still has enough power to decrease the negative effects of CEO behavior.

Comparing these results with Tables 7, 9 and 11 we can see that the impact of age becomes significant for the repurchase ratio (Q1) and the level of cash compensation becomes significant for the repurchase ratio (Q1), dividend ratio (Q2), and the fraction of repurchases (Q1, Q2 and Q3). These effects, though, are eliminated by efficient corporate governance mechanisms. We assume that in such companies there might be some additional tools to cope with CEO behavior that are not considered in our CGQI.

5 Discussion and conclusions

In this study we explore whether the boards of directors are able to overcome negative influences of CEO behavior in terms of his or her preferences to risk. Our first set of findings fills the gap in the research on the relationship between CEO risk preferences and various aspects of payout policies. We show that CEOs that bear more risk tend to set lower levels of payout than their less risky colleagues do, which means that Hypothesis 1 cannot be rejected. It is also shown that repurchases are the more preferred method of payout in companies with higher levels of executive stock options. This is due to CEOs’ overconfidence in these types of companies, and their awareness regarding the undervaluation of stocks. Moreover, they may avoid dividends due to their negative impact on the options’ value. On the contrary, less risky CEOs tend to maintain higher levels of payout: the compensation policy that stimulates a CEO to bear less risk is associated with higher levels of payout.

As for the decisions to start paying to shareholders, we found that less risky CEOs were more likely to initiate dividend payments. More risky CEOs, on the contrary, have a lower probability of initiating either repurchases or cash dividends, which means that Hypothesis 2 cannot be rejected.

We also document the effects of CEO risk preferences on the choice of the payout channel itself. We show that more risky CEOs choose to distribute profits among shareholders through repurchases rather than through dividends, which means that Hypothesis 3 cannot be rejected. Risky CEOs consider the company stocks to be undervalued and tend to repurchase them at what they think is a good price. Less risky CEOs prefer to distribute cash through the dividends instead.

Our second set of findings is related to the role of boards of directors in offsetting the negative effects of CEO behavior on payout policies. Our study provides new empirical evidence on the role of compensation schemes set up by the boards to align CEO risk preferences with the strategic vision of the boards. We document that efficient boards are able to eliminate or to reduce the negative influence of CEO behavioral characteristics. We show that the impact of a CEO’s risk preferences is lower in the companies with higher quality of corporate governance. This means that Hypothesis 4 cannot be rejected.

However, we found that this ability decreased with the increase in payout levels. High quality corporate governance has an ability to significantly reduce negative effects in companies with the lowest levels of payouts, but not in the companies with the highest levels of payout. In the latter case the CEO’s risk preferences still affect both the level of repurchase ratio and the choice of payout channel. One of the reasons for this is that in companies with the highest payout levels, shareholders may be satisfied with these high levels in spite of opportunities to get increase payouts if the negative effects of the CEO’s behavior are overcome.

Based on the results of this study we strongly believe that remuneration policy and the pay for performance mix should be considered as a tool for influencing CEO behavior within the company. Moreover, major shareholders should force the development of highly efficient governance processes, especially in those companies with low levels of payout, to protect themselves against the negative effects of the CEO’s behavior. The appropriate CG efficiency should be set in accordance with the shareholders’ interests and the peculiarities of the CEO’s behavior.

The aspects of CEO risk aversion studied here are only a part of the behavioral traits that predetermine different styles in developing corporate policies. We believe that further research should focus on a deeper understanding of the influence of the overall set of behavioral characteristics of CEOs, which could be assigned to the bounded rationality of decision-making by top executives. It seems important to understand better how CEO overconfidence, which is based on an underestimation of future risks, and CEO optimism, which is an overvaluation of future outcomes and of favorable trends, interact in corporate payout policies. Future research on the interaction of CEO behavioral biases, along with the biases of members of the board, might be a productive angle for understanding the future of corporate payout policies.

In addition to the above, it would be beneficial to gain a deeper understanding of the board’s ability to eliminate the negative effects of other behavioral biases. It may be the case that to treat overconfidence and optimism appropriately, or to gain the most benefit from hindsight, boards of directors need to develop approaches that differ from those used to deal with CEO risk preferences. Such a research agenda may indeed help shareholders to protect their interests more effectively from the adverse effects of CEO behavior.