Moderative influence of compensation policy on the relationship between the personal characteristics of the CEO and the company's risk

Размещено на http://www.allbest.ru/

1

ПЕРМСКИИ? ФИЛИАЛ ФЕДЕРАЛЬНОГО ГОСУДАРСТВЕННОГО АВТОНОМНОГО ОБРАЗОВАТЕЛЬНОГО УЧРЕЖДЕНИЯ ВЫСШЕГО ОБРАЗОВАНИЯ «НАЦИОНАЛЬНЫИ? ИССЛЕДОВАТЕЛЬСКИИ? УНИВЕРСИТЕТ «ВЫСШАЯ ШКОЛА ЭКОНОМИКИ»

Факультет экономики, менеджмента и бизнес-информатики

Выпускная квалификационная работа

Модеративное влияние компенсационной политики на связь между личными характеристиками генерального директора и риском компании

Ячменева Екатерина Владимировна

Пермь, 2020

Аннотация

В данном исследовании мы объединили теорию контрактов и теорию высших эшелонов для того, чтобы выявить модеративное влияние компенсационной политики на связь между личными характеристиками генерального директора и риском компании. Исследуемая выборка состоит из 170 торгуемых компаний из Великобритании, рассматриваемых в течение 2008-2015 гг. В ходе тестирования моделей с фиксированными эффектами, было обнаружено, что при увеличении доли опционов генеральные директора в возрасте от 55 до 64 лет имеют тенденцию к снижению корпоративного риска, в то время как генеральные директора в возрасте от 45 до 54 лет увеличивают риск компании. Увеличение доли бонусов в структуре вознаграждения руководителей имеет противоположное влияние на связь между данными возрастными группами генеральных директоров и риском. Существует положительное модеративное влияние бонусов на связь между самоуверенностью и риском, опционы оказывают отрицательный эффект на данную связь.

In this study we combined contract theory and upper-echelon theory to reveal the moderation effect of CEOs' compensation on the link between CEOs' personal traits and corporate risk. The sample used in the research comprises 170 traded companies from Great Britain between 2008 and 2015. By testing fixed effects models, we find that a higher share of options in compensation package of CEOs aged 55-64 makes them decrease the level of corporate risk while CEOs aged 45-54 increase corporate risk. An increase in the share of bonuses in the remuneration structure of managers has the opposite effect on the relationship between these age groups of CEOs and risk. The link between overconfidence and corporate risk is positively moderated by bonuses share and negatively by options share.

Table of contents

Introduction

1. Theoretical background

1.1 Personal traits of a CEO and corporate risk

2. Compensation of CEO and corporate risk

2.1 Optimal contracting approach vs. managerial power approach

2.2 The influence of various types of compensation on corporate risk

2.3 Measurement of variables

3. Results of models' estimation

3.1 The issue of reverse causality between CEOs' compensation and corporate risk

Conclusion

References

Appendix

Introduction

Almost every day newspapers, forums and blogs covering financial events report on strategic decisions made by companies around the world. One of the most controversial issues in the field of company management is the level of risk taken by the organization. Corporate risk taking has become one of the core topics in the research on corporate finance (Cain and McKeon, 2016; Sayari and Marcum, 2018). Previous literature has shown that corporate risk taking decisions do also depend on personal characteristics of firm decision makers (Li, Griffin, Yue and Zhao, 2013; Serfling, 2014). It can be explained by upper echelon theory which postulates that CEO characteristics impact financial strategic actions such as investment policy, financial policy, diversification, etc. (Bertrand and Schoar, 2003; Wang, Holmes, Oh and Zhu, 2016). The effect of personal biases and preferences is likely to be even more influential in risk taking than in other corporate decisions.

In addition, the issue of corporate risk is reflected in agency theory (Ross, 1973; Jensen and Meckling, 1976). According to this theory, one party, called principal (this role is devoted to shareholders in our case), gives to another party (so-called agent (manager)) rights to rule the company. However, an agent may deviate from an optimal risk level which principal supposes him/her to maintain. In the company manager may have his/her individual picture of the world due to manager's personal beliefs, thoughts, prejudices, former experience and character traits which form particular risk-profile of the company. As a rule, CEOs do not have enough intrinsic motivation to maintain the desired level of risk for the company that is why principals (shareholders) have to set external conditions which are aimed to change CEO's risk-appetite (Humphery-Jenner et al., 2016).

According to contract theory, which is a part of agency theory, principal designs remuneration package that helps to align agent's interests with company's ones. Hou et al. (2017) in their paper note that nowadays major institutional shareholders even have special guidelines which depict preferences of compensations they would like the CEO to be awarded.

As far as currently compensations in companies have complicated nature, the consequences of their nomination may be two-fold. From the one side, managers can increase the level of the risk in order to raise their welfare that is tied with corporate results. From the other side, managers choose lower level of risk because they are afraid of losing their welfare (Martin et al., 2009; Benischke et al., 2019). Furthermore, according to classic contract theory, compensations are awarded to eliminate purposeful breach of contracts. In this case, agency problem occurs due to the reason that the CEO rationally and purposefully deviate from the behavior which maximizes value of the company. As a consequence, for the classic contract theory the influence of compensation can be surely expected. On the contrary, it is hard to undoubtedly define the influence of compensations on those deviations which are not under CEO's control and happen naturally due to CEO's behavioral and psychological peculiarities. That is the reason why this field of research is of particular interest.

We joined both of the mentioned areas of the research (contract theory and upper-echelon theory) and study how the remuneration packages may moderate the impact of CEO personal characteristics on the risk taking of the firms they run.

We analyze a sample of 170 traded companies from Great Britain between 2008 and 2015. The obtained results depict that the increase in options share positively influence the link between CEO's age from 45 to 54 and corporate risk whereas there is a negative effect of options on the link between older CEOs group (55-64 years old) and risk. Furthermore, bonuses share, on the contrary, negatively moderates the link between 45-54 years old CEOs and share price volatility and positively influence on this link for CEOs aged 55-64 years. Moreover, we could find that overconfident CEOs when being awarded more options tend to decrease corporate risk and vice versa increase it with the growth of bonuses share.

This paper contributes to the literature mostly in two ways. First, we analyze how the influence of CEO's personal traits on corporate risk might be managed by the structure of remuneration package. Previous literature usually tests the connection between personal traits and corporate risk or between remuneration and corporate risk, but do not regard them together. Secondly, we extend previous research that has primarily focused on US firms to European market.

The research takes 55 pages and consists of five chapters, references and appendix.

1. Theoretical background

1.1 Personal traits of a CEO and corporate risk

The upper echelons approach is a tool that has provided theoretical explanations for a number of corporate strategic issues. The main idea is that corporate strategy reflects the experiences, personalities and values of their top managers which have the power of ruling the company (Hambrick, 2005). A consequence of this insight is that focusing on the traits of the managers will yield strong explanations about organizational outcomes, in particular, risk. The studies analyzing relationship between different personal characteristics of CEO and company's level of risk have become widespread over the past thirty years.

The influence of the characteristics of CEOs on the performance of the company has been studied in many papers, the issue has been considered both for companies in developed markets and for organizations placed in emerging markets. Furthermore, scientists revealed the influence of personal traits on various types of companies' performance indicators: risk, profitability, growth, etc. We believe that the influence of personal traits on corporate risk is the most interesting issue as far as the interests of managers and shareholders in this matter are often different (Humphery-Jenner et al., 2016). Moreover, the decisions concerning the level of risk are inevitably tied with personal attitude to the risk which is formed by former experience, personal beliefs, characteristics.

The process of determination and measurement of personal traits which can have an effect on company's risk can be challenging since most characteristics of CEOs are latent and difficult to measure. As a rule, academics tend to use certain proxies that imply specific way of thinking and making decisions.

In this paper the attention will be devoted to different personal traits of CEO which are observed in empirical studies as predictors of company's risk level.

A vast literature has studied the role of a person age in making decisions, the results are somewhat conflicting and call for further research. It is supposed that strategic actions and decisions of the CEO are strongly tied up with the planning horizon (Rayan & Wiggins, 2001).

Vroom and Pahl (1971) were ones of the first that empirically proved negative relationship between the age and corporate risk-taking. Subsequent papers of Barker and Mueller (2002), Bertrand and Schoar (2003), Graham, Harvey and Puri (2013), Ting et al (2015), Farag and Mallin (2018) and Peltomдki, Swidler and Vдhдmaa (2018) confirm this finding. Barker and Mueller (2002) suppose that older CEOs tend to choose more conservative business strategies. The closer the retirement for the CEOs the less risky he/she is because of already formed reputation and no need to prove his/her capabilities. Ting et al. (2015) believe that older CEOs are less stress resistant and flexible, for them it can be harder to frequently analyze information and make non-conservative decisions. Chowdhury and Fink (2017) go step ahead and show that it is specifically through the channel of R&D investment that CEO age is associated with the reduction in firm equity risk. Serfling (2014) in his research also address the issue of investment choices which distinguish between older and younger CEOs noting that older CEOs are less interested in risky investments in R&D.

On the contrary, Xie (2015), Li & Tang (2010) find positive relationship between managers age and risk-taking. The researchers explain this influence by arguing that older CEOs are usually more confident about decisions they make and for them it's easier to take actions which may increase the risk. Rayan & Wiggins (2001) in their research found out that younger and older CEOs are the least risk averse groups of managers compared to those of middle age. This happens due to the reason that younger CEOs eager to build their reputation as soon as possible whereas older CEOs choose those projects which pay off before their retirement. Holmstrom (1999) & Zweibel (1995) assume that younger CEOs are cautious in their decisions as far as at the beginning of their career they do not want to break future perspectives by risky (probably harmful) investments.

CEO age is not the only one trait which is observed in empirical papers. Many researches claim that CEO tenure also influences company risk level. Nonetheless, the nature and direction of this influence is still being questioned.

In their paper Rayan & Wiggins (2001) argue that more tenured CEOs have already built their reputation, reached important goals which allowed them to stay in this position and do not have any induces to increase the level of risk in the company because he/she has rooted in the company. The same idea is supported by Barker and Mueller (2002) and Musteen, Barker and Baeten (2006) who explain negative influence between tenure and risk by arguing that tenured CEOs are incline to choose stable strategy. Although, the scientists suppose that previous nature of career (marketing, engineering) may also have an impact on the relationship. This is consistent with Martino, Rigolini and D'Onza (2018) and Kabir, Li and Veld-Merkoulova (2018), who report that CEOs do not curtail some high risk activities such as R&D investments as their career horizons become shorter.

Nonetheless, Rayan & Wiggins (2001) put forward notion that the long tenure of CEO can be connected with lasting growth of the company which implies taking decisions that increase company risk and shareholders' welfare. So that in order to continue his/her career as a manager and maintain desired price level CEO may increase the risk. Chen and Zheng (2014) and Xie (2015) also find a positive relation between CEO's tenure and corporate risk taking. Antia, Pantzalis and Park (2010) show that longer tenure is associated with higher levels of information risk, which supports the career concerns hypothesis. This is not the only paper where contradictory influence of CEO tenure was noticed. Lin et al (2011) and Chen (2013) measuring risk as R&D investments found U-shape relationship between CEO tenure and risk: authors suppose that young CEOs are very cautious when making decisions about R&D keeping in mind the idea of stable career and reputation.

Even though, managerial positions are frequently taken by males, with the spread of equality rights and the ideas of corporate social responsibility, nowadays there are more women in CEO positions than there were some decades ago (Faccio et al, 2016; Palvia et al, 2015). This tendency contributed to the development of researches aimed to evaluate peculiarities of males' and females' decisions as CEOs. Behavioral differences connected with gender are usually tied up with information perception by males and females, their levels of conservativeness, responsibility and risk tolerance (Palvia et al, 2015). It is common to believe that women are more risk averse than men (Brachinger et al, 1999; Martin et al, 2009; Faccio et al, 2016, Ting et al, 2015). To be precise, Bernasek and Shwiff (2001) in their paper argue that women are less risky even if it counts for their personal financial portfolio. Martin et al. (2009) noticed that market level of risk in the company decreases when it is ruled by a woman. Even more, according to authors, some companies appoint women as CEOs in order to decrease risk level. Faccio et al. (2016) came to comparative similar results: researchers reveled negative relationship between woman as a CEO in a company and price volatility and financial leverage. Palvia et al. (2015) put their attention to bank sector and noticed that those banks which are ruled by women are less probable to go bankrupt despite crisis. Nonetheless, there are studies such as Shao and Liu (2014) where no statistically significant difference between risk in companies managed by males and females wasn't found.

In addition, education is also treated as one of those characteristics which influences risk in the company. It is supposed that better education indicates higher cognitive capabilities of the CEO, his ability to make right decisions in hard times and higher awareness of company's perspectives.

Lin et al (2011), Kitchell (1997), Thomas et al (1991), Farag and Mallin (2018) note that more educated CEOs are usually more open-minded that is the reason why they invest more in R&D. It should be notices that R&D is not the only measure of risk which was observed regarding this personal trait. For example, Ting et al (2015) and Rakhmayil and Yuce (2011) defined that higher education of CEO is associated with larger financial leverage. MacCrimmon and Wehrung (1990) and Barker and Mueller (2002) also support the idea about positive relationship but Barker and Mueller (2002) haven't found direct connection between years spent on education and risk.

One more trait is overconfidence. Overconfidence can be viewed as an overestimation of one's own abilities and outcomes relating to one's own personal situation (Brown and Sarma, 2007, Mertins and Hoffeld, 2015). The impact of overconfidence on economic decisions is a controversial issue. On the one hand, as a behavioral bias, overconfidence can lead to wrong decisions. Overconfident agents rely too much on their own perceptions and can stop acquiring relevant information about corporate prospects and the environment (Bйnabou, 2012). On the other hand, managerial overconfidence can induce some risk appetite and make up for managerial risk aversion, which leads to sub-optimal investment decisions (Chava and Purnanandam, 2010, Kim and Lu, 2011, Liu et al., 2013).

Recently the growing popularity was obtained by papers which analyze the influence of marital status of CEO on his/her strategic decisions. For instance, Ruossanov & Savor (2014) found out that, as a rule, single CEOs tend to increase risk of the company compared to married CEOs. Authors suppose that single CEOs tend to make riskier decisions because they are searching for a partner and compete on marriage market.

Furthermore, some researchers follow the idea that CEO duality also can affect risk of the company (Olson et al., 2018; Benischke et al., 2019). Nonetheless, this influence is contradictory.

To sum up, the connection between personal characteristics of CEO and company risk has been observed in many papers. It should be noted that in most papers in order to explain the influence researches take biographical or demographical data to measure the trait instead of psychological measures. This way of measurement makes the results biased and “noisy” (Hambrick and Mason, 1984) the problem still not solved.

Furthermore, the process of making decisions is not that obvious as it seems to be. First of all, this is due to the fact that managers are often chosen because they have certain experience, knowledge and characteristics, that is, this process is not random (Hambrick and Mason, 1984).

Corporate risk

Risk management is treated as one of the most important objectives for the company (Froot et al, 1993). The nature of risk is very controversial: higher risk cannot be unconditionally good or bad for the company because it is somehow alighted with company's strategy, it's life cycle and industry. Nonetheless, Dittmann et al. (2017) in their paper argue that company can become riskier even not changing its core strategy but just due to one-time risky actions taken by company's management. Furthermore, as it was noted by MacCrimmon & Wehrung (1990), risk can hardly be captured by the only one type of measure because risk-propensity of the company is reflected in various areas, risk itself can be addressed to different types of companies' financial and operational outcomes.

Furthermore, according to the majority of economic papers, corporate risk is assumed to be the mean of enterprise growth. Nonetheless, risk do not always bring benefits to firm's shareholders as it may lead the firm to “big losses” and destructive problems. Sanders & Hambrick (2007) in their paper mentions Bear Stearns and Lehman Brothers as an example of this failure.

Risk is usually measured by either market or book indicators in empirical papers (Bryan et al., 2000). Book indicators of risk are assumed to show certain area of risk-based decisions which are aimed to attract cheaper financing (increasing financial leverage of the company), increase sales and, as a consequence, market share through higher R&D investments and follow other clear goals. For comparison, market-based indicators of risk do not reflect particular area of corporate decisions, instead they show overall risk inclination of the company.

Common risk measures comprise volatility, financial leverage, liquidity, sharp ratio, credit rating, etc. In this research we stick to efficient market hypothesis and believe that market measures of risk include all information about the company which is available at the time so that all events and indicators which may signal the growth of the decrease of company risk are reflected in share price (Benischke et al., 2019).

It is supposed that risk taken by CEO is immediately reflected in company's prices and volatility. Company value growth is proclaimed to be the main goal of large enterprises in the modern economy. The growth can be reached by company development, investments in R&D, making new decisions and choices - large set of actions which imply adoption of higher level of risk.

In this paper we believe that this overall risk in various forms is captured by share price volatility.

2. Compensation of CEO and corporate risk

2.1 Optimal contracting approach vs. managerial power approach

It is thought that CEO is the person who has the most profound and precise information about investment and organizational prospective of the company. Moreover, keeping this information in mind CEO makes strategic decisions based in his/her believes and preferences, company risk level and opportunities (Rogers, 2002).

As a rule, managers suppose that high level of risk which they can take through participating in risky projects and investments in innovations might be unreasonable for them because it bears possibility of reputation and welfare loss (Rayan & Wiggins, 2001). It should be noted that usually investment portfolio of CEO is not diversified because large part of his/her wealth and future savings is connected with company success and stability (Sheikh, 2012). According to agent theory the ideas about means for reaching prosperity (on individual and company level) differ among shareholders and managers (Jensen & Meckling, 1976). Furthermore, as far as the management rights and ownership are divided in the company the problem of information asymmetry occurs. The question is: can this problem be solved by compensation awarded to the CEO?

Fundamentally, there are two main ideological streams in this field: optimal contracting approach and managerial power approach (Bebchuk & Fried, 2003). The followers of the first one support the idea that shareholders have possibility to decrease the level of asymmetry through compensation assignment. The compensation package can be structured in such a way to align manager's interests and shareholders' ones (Harvey & Shrieves, 2001; Dittmann et al., 2017; Zavertiaeva & Smirnova, 2017). In a company the right of representing shareholders' interests is devoted to a board of directors. The board's members are interested in resolving the possible conflict between manager and shareholder. That is the reason why the board appoint compensation committee which set compensation size and structure which, as a consequence, shifts CEO's behavior to the desired direction (Matsumura and Shin, 2005).

The supporters of the second stream argue that compensation itself cannot solve agency problem, moreover, this problem is exacerbated by compensation issue. The researchers suppose that this situation occurs as the board of directors, which play important role in renumeration of CEO, are affiliated: they are not fully independent from the CEO. This dependence is a result of the fact that CEOs play significant role for the decision of further directors' re-appointment as well as their renumeration so that they have to favor CEO in order to keep their positions (Bebchuk & Fried, 2003). Ryan& Wiggins (2004) in their paper could reveal that if the board of directors is not independent CEO can influence not only the amount of compensation but also it's structure. The authors note that the problem occurs in the companies with large board of directors, with the CEO belonging to the family which founded the company and with tenured CEOs which have more power over board of directors. Nonetheless, the followers of managerial power approach claim that CEO cannot fully influence the amount and structure of his/her compensation as far as the company, as a rule, operates under the gaze of the media. The researchers suppose that this peculiarity may lead to the camouflage of income. (Bebchuk & Fried, 2003). Van Essen et al. (2012) in their meta-analytic paper note that in the companies, where the board of directors is not independent, CEO, as a rule, can manage the amount of compensation he/she receives and rise the amount of cash rather than manage the whole structure of compensation. The authors also emphasize that the higher managerial power is devoted to CEO with duality (Van Essen et al., 2012). In 2002 the Sarbanes-Oxley Act (SOX) was issued, being the reaction to the dot-com crisis happened in the early 00's. This act describes the rules relating to corporate governance and CEO compensation in companies. Interestingly, Nourayi et al. (2012) found out that such factors as board size, CEO duality and the percentage of external directors had significant influence on CEO pay before SOX but after the time act was presented to general public these variables lost their significance.

In this research we support the first theory and believe that in the countries with developed economies there is no or weak affiliation of boards of directors and they fully represent the interests of shareholders so that CEO has minimal or no power over his/her compensation. This idea is being supported due to the reason that more than ten years ago SEC rules related to compensation and Sarbanes-Oxley Act were issued. These rules proclaim board of directors' and compensation committee's independence so that nowadays the opposite is being monitored and punished. Undoubtedly the problem can remain in the companies where the CEO owns high share of the enterprise but most developed traded companies hire relatively independent CEOs.

2.2 The influence of various types of compensation on corporate risk

The most common types of compensation cover salary, bonuses, options and benefits which are awarded based on the results of previous financial year. It is believed that compensation tied with balance indicators is not the best one because it induces manager to choose traditional means of balance indicators' improvement through expenditures and quality decrease. This situation happens as the CEO is focused on the short-run perspective. Furthermore, such type of compensation as salary may even have no influence on managers decisions because, as a rule, it is stable and does not change much from year to year (Miller et al., 2002; Coles et al., 2006).

Stable salary which usually depends on certain KPIs make managers reluctant to taking on more risk through cheaper but riskier loans, investments in R&D and other risky corporate decisions. Nonetheless, it seems to be obvious that without such actions the growth of the company will be slower and it may lose a good business opportunity (Coles et al., 2006). Such situations may lead to share price decrease and deterioration of future prospects of the company. For example, Miller et al. (2002) in their research proved that there is a negative relationship between share of compensation tied with balance results in total compensation package and company risk. Compensation committee awards CEO with market-based types of compensation (shares and options) to eliminate such behavior of CEO and achieve permanent growth and development of the company (Coles et al., 2006).

Although, it should be noted that to the date scientists haven't found the answer to the question whether market-based or other components of compensation have strong positive or negative influence on company risk (Wowak & Hambrick, 2010).

The outcome (decisions being made by CEO) of the award is strongly tied up with behavioral economics. Economists involved in this scientific field support one out of two fundamental economic theories: classical agent theory and prospective theory (Benischke et al., 2019). The paper of Kahneman & Tversky (1979) is treated as basic research of prospect theory which postulates that uncertainty makes people overestimate bad outcomes and possible losses rather than believe that they can benefit from the situation. This psychological peculiarity influences CEO in the following way: the bigger the share of options in the CEO compensation package the more cautions and conservative he/she becomes as far as it means that future welfare of the CEO is uncertain and he/she is afraid to lose it (Martin et al., 2009). Accordingly, the followers of prospect theory suppose that the more options (regarding other parts of compensation package) the CEO has, the lower the risk of the company will be.

On the contrary, the supporters of agent theory believe that options represent an incentive for CEO to make prospective and progressive decisions tied with risk. The option itself gives the manager the right to buy shares in some time at predetermined price. For the manager in this context it is beneficial to make such decisions which can increase share price of the company. If share price increases it gives manager possibility to gain more on the sale of securities and receive income from the difference between the purchase price (predetermined) and the sale price (market) (Armstrong et al., 2012; Sandrez & Hambrick, 2007).

Agent theory followers assume that under circumstances of uncertainty CEO value benefits of higher possible income above losses he/she may suffer from due to higher level of risk for the company.

To sum up, CEO endowment with options forms additional incentives for him/her to take risky actions which therefore may lead to company growth (Coles, 2006; Jin, 2002; Armstrong et al., 2012; Sandrez & Hambrick, 2007).

Nowadays it is hard to imagine compensation package of the CEO in a big traded company which doesn't include options. Even in the 90-th of the last century one-third of CEOs compensations in such companies in the USA was formed by options (Jin, 2002). Nowadays this proportion is much higher as shareholder want CEO to think about future of the company.

One of the risk measures is price volatility. According to economic theory options price increases as volatility of basic asset (shares, as a rule) also grows. In this case it is beneficial for the CEO to increase corporate risk if he/she owns company's options. Nonetheless the direct positive influence is not supported by whole science community. For example, Gormley et al. (2013) argue that initially CEOs are risk averse and options may induce him/her to take more risk until the risk is too high: the CEO utility function from risk growth is U-shaped. Since the risk is too high CEO becomes more concerned about negative outcomes of high risk rather than about opportunities. To sum up, authors believe that starting from a certain level of risk stimulation CEO through options being awarded does not further enhance risk of the company.

Nonetheless, it should be noticed that some researches connect the wide-spread tendency of options being awarded with dot-com crisis and financial crisis which happened in 2008: they assume that CEOs aiming to increase their option-based wealth make decisions which push up share prices in the short-run and worsen the company's positions in the long-run (Kim et al., 2011). Interestingly, the researchers failed to prove this notion empirically which could happen due to the reason that options are awarded for a long time (as a rule, exercise period for options comes in several years since they were awarded) so that CEOs strive to maintain qualitative stable growth for the company.

There is one more instrument used in the companies to induce CEO to make certain decisions which are restrictive shares. Restrictive shares give CEO the right to vote and can be sold only when certain effectiveness indicators are reached in some time (Guay, 1999; Gopalan et al., 2014). Such type of compensation stimulates CEO to participate in the long-run investment projects which pay off in five or more years. For instance, risky investments in R&D may lead to significant company growth but it takes time. Although, researchers found out that restrictive shares are less effective than options as stimulating instruments for the risk (Gopalan et al., 2014).

A very common type of compensation awarded in almost every big traded company are bonuses. Bonuses are awarded to CEO for achieving KPIs. As a rule, KPIs assumed to be reached in two- maximum three years so that bonuses program is oriented towards short-run perspective rather than long-run growth of the company (Gopalan et al., 2014). The influence of bonuses of company risk is not unequivocal. On the one side, higher proportion of bonuses in the CEO compensation package does not imply higher risk because focusing on short-run performance indicators CEO may increase his/her welfare with expenditures decrease. For example, these can be secondary spending the reduction of which wouldn't lead to bad outcomes in the short run: marketing, advertisement, development of employees, maintenance. On the other hand, short-run perspective implies making decisions which pay off in the short run and CEO can make risky decisions to achieve these results (Gopalan et al., 2014).

Furthermore, compensation committees of many companies award CEOs with so-called benefits. Benefits represent company spending on CEO's personal driver, healthcare, help with legal issues and other additional spending. This type is included in compensation package of the CEO but it is very rarely analyzed regarding company level of risk.

The relationship between compensation incentives and risk was observed in the vast number of empirical papers some of which will be mentioned further.

One of the papers where the relationship was analyzed is written by Jin (2002): the author demonstrates the sensitivity of CEO compensation to the change of market share price. Welfare of the CEO is positively connected with idiosyncratic risk and has no connection with systematic risk of the company.

In alignment with Jin (2002) results Coles et al. (2006) based on the sample of S&P 500 companies came to alike findings: they noticed that higher sensitivity of managers compensation to shares volatility (vega) forms incentives for CEOs to invest in riskier assets through increase in investments in R&D and decrease of investments in PPE, to follow more aggressive debt policy and stick to the strategy of lower diversification. Nonetheless, researcher haven't found positive relationship between risk of company and sensitivity of CEO compensation to change in company financial results (delta). On the contrary, they noticed that with the growth of delta there is a decrease in lower-risk assets.

Armstrong et al. (2012) studied the influence of CEOs' options compensation on systematic and idiosyncratic risk with vega and delta indicators. The authors concluded that an increase in vega indicator motivates CEOs to increase the overall risk of the company through an increase in systematic risk rather than idiosyncratic risk, since at a given level of sensitivity of vega, an increase in systematic risk always leads to a larger increase in the value of the CEO options portfolio, rather than when this occurs with an increase in the unique (idiosyncratic) risk; delta was connected with all risk indicators of companies

Rogers (2002) also found positive connection of these two indictors of compensation sensitivity and risk of a company measured as interest rate. Bryan et al (2000) studied all compensation components awarded to CEOs in detail. This approach allowed researchers to find that higher number of options in CEO's compensation package prevents the decrease of company risk level and supports further development of company by providing higher growth potential. On the contrary, scientists came to the result that monetary par of compensation expressed in salary and bonuses awarded based on financial results does not have any influence on the risk taken by the company.

Sheikh (2012) in their research took the measure of quotient of compensation change to one-percent change in share price as indicator of compensation stimulus of the CEO. It was defined that this indicator is positively connected with investments in R&D and number of patents and patent applications. Moreover, Sheikh (2012) divided compensation package of CEO on components which allowed them to make conclusion that options have more impact on innovation activity of the company than shares.

Xue (2007) observed how CEO compensations influence decision whether to buy R&D from external companies or to develop new products internally. Internal R&D are treated as riskier ones as the consequences of internal R&D may be uncertain whereas external R&D have more obvious results. The author revealed that with the growth of options in compensation package the probability of internal R&D increases.

Some empirical studies have shown that the relationship between CEO compensation and company risk can be quite complex. For example, in their work, Bryan et al (2000) showed that with an increase in the number of options in the portfolio of the CEO, they cease to have the same stimulating effect on the acceptance of a higher level of risk by the CEO.

To sum up, the relationship between the structure and volume of the compensation package of the CEO and the risk of the company is still an interesting question to study.

In this study we focus on two particular types of compensation: options and bonuses because other compensation components such as salary or benefits do not vary much during career of the CEO and company level of risk (Bryan et al., 2000).

Research question

In empirical economic papers it is very common to analyze different phenomena's simultaneously as they can condition each other. According to theoretical background, the connection between CEO personal traits and company risk-profile is controversial: there was found positive, negative or no influence at all for different characteristics which may indicate that this direct connection is being conditioned by external factors.

Primarily such factors can be reflected in external circumstances under which manager works. Compensations may be one of those conditions which bias the influence of CEO characteristic on company risk as compensation is awarded to the particular person considering his/her traits (Humphery-Jenner et al., 2016). Sanders & Hambrick (2007) in their paper argued that risk, compensation and personal traits should be analyzed all together. Furthermore, Humphery-Jenner et al (2016) claim that concerning the relationship between personal traits and corporate risk there should be applied so-called behavioral agency theory. They assume that the vital role in the connection is devoted to CEO extrinsic (monetary) motivation. Even though the issue is being debated, it is hard to find any profound empirical research where scientists questioned how compensation policy of company condition the link between CEO characteristics and risk.

As a rule, according to behavioral economics, decisions made in the company by CEO are grounded on his/her believes and personal traits which form certain risk level for the company. This company risk level may differ from the one desired by shareholders. Compensation committee representing the interests of shareholders sets particular structure of compensation which considers CEO personal traits. remuneration corporate risk compensation

Jensen & Murphy (1990) in the research noted that except the amount of compensation it is important to put attention to compensation structure. According to their analysis of American companies the impact of change in compensation structure exceeds the effect of compensation value on company risk.

In this paper it is assumed that CEOs decisions are driven by both intrinsic motivation (personal traits) and extrinsic influence of compensation which allows to shift the impact of intrinsic driving force. The idea of this separation is somehow presented in the paper of Humphery-Jenner et al. (2016). The researchers revealed that overconfident CEOs are more likely to be awarded with market-based compensations. Nonetheless, the effect of these compensations on the link between overconfidence and firm value was not proved. Compared to the research of Humphery-Jenner et al. (2016), more CEO's traits are included in the present paper and more resent data is analyzed.

Research question of this study can be formulated in the following way:

How does particular compensation structure awarded to CEO moderates the connection between CEO personal characteristics and risk?

The research question is represented on Fig. 1.

Fig. 1. Research question

Methodology

Database

The analysis in this study is based on the data about 170 traded companies from Great Britain observed during 2008-2015 years. The data has panel structure and the full sample consists of 750 company-year observations. The panel is unbalanced due to the lack of information about some metrics at certain time periods. The data includes financial information, data about companies' risk, personal traits of CEOs and their compensations.

Great Britain's companies were analyzed due to the reason that they tend to report information about CEO compensations in their annual reports in a more detailed way compared to other European countries which makes this data more suitable for the research. Furthermore, inclusion of one country implies similar corporate governance of companies located there. In addition, we believe that companies in such a developed economy as Great Britain's one, as a rule, have independent boards of directors and CEOs so that the optimal contracting approach can be applied.

The database used in the research was provided by IDLab HSE Perm, a part of the database: companies' financial and market data and information about CEO compensations for 2014-2015 years, was collected on our own. The data was collected through the analysis of companies' annual reports and information portals Factiva (for overconfidence), Investing.com and Amadeus. In the collected dataset there are companies related to 7 industries: construction, manufactory, energy, services, trade, finance and professional services.

2.3 Measurement of variables

To construct the models firstly we had to choose variables that could be used in analysis based on the research question which was formulated earlier. As a rule, many variables are measured by proxies in empirical papers as it is hard to choose one and only right measure for the phenomena which cannot be measured directly. In this section detailed description of variables is presented.

Risk

In the study we measure risk as annual volatility of company share prices. This market measure of risk previously was observed in the papers of Guay (1999), Rajgopal & Shevlin (2002), Cadenillas et al. (2004), Bargeron et al. (2010) in the context of compensation influence on the risk and corporate decisions relating company risk-taking. Annual volatility of company share price is measured as the mean of daily share price standard deviation during the year. It is believed that volatility shows company risk due to the efficient market hypothesis which postulates that all the news about company, its strategy, decisions about R&D are immediately reflected in company share price. So, following the efficient market hypothesis in this research we believe that this measure of risk allows to capture the whole firm-specific risk.

CEO characteristics

In the research we pay attention to three characteristics of CEO: CEO age, tenure and overconfidence. These characteristics were chosen due to the reason that tight relationship between these traits and corporate risk was emphasized in previous research papers.

Even though calculation of CEO age is clear, based on the previous papers, in this research it was decided to make dummy variables for the age which could reflect certain groups of CEOs of the same age. This division was made in accordance with previous results where the influence of CEO's age in the risk was non-linear. We believe that this approach gives as possibility to analyse each age group in detail. To be precise, we divided CEOs into four age groups: 35-44, 45-54, 55-64 and more than 65 years old. Such groups were formed due to the idea that 35-44 years old CEOs are at the beginning of their career and their incentives and wishes differ from other CEOs' ones: according to previous research, younger CEOs are more likely to take more risk and participate in challenging projects. 45-55 years old are in the middle of their career which also have an impact on their behavior: they are more tenured compared to younger CEOs but still have wide horizons and much time before retirement to reach the goals they set for themselves. 55-64 years old CEOs are closer to their retirement and seem to start changing their behavior to more conservative one which, undoubtedly, have an impact on company risk. Those CEOs who are 65 or older seem to be the most conservative because this is the age of retirement in The Great Britain. Furthermore, this age division has already been applied in the paper of Corci & Prencipe (2019) who studied the factors of companies' innovations.

In this paper we initially measure tenure as number of years during which CEO works as CEO (no matter what company CEO worked for previously). Nonetheless, we for the tenure we apply the same logic which was used for age: the tenure was divided into four groups: 0-4 years of tenure, 5-9, 10-14 and more than 15 years tenure as CEO. CEOs tenured 0-4 years seem to be both more cautious and ambitions. Those who take this job position for 5-9 years are less cautious, they already know a lot about the management, internal processes and still have high ambitions. CEOs who work for 10-14 years are the experts though they have already rooted in this position and may have less incentives to rise company's level of risk. We believe that those tenured 15 years and more have the least motivation to prove that they deserve their job position through making extreme decisions which can lead to company growth. Nonetheless, such CEOs can make risky decisions without concerns about probable failure due to the fact that they know a lot about business.

The division of tenure was used in the paper written by Cook & Burres (2013) but scientist divided CEO only into two groups: more that 10 years and less that 10 years. We believe that in order to capture non-linear influence of CEO, more groups are required.

Even though there is no doubt that continuous variables reflect more information and seem to be preferable in the research, the usage of binary variables gives opportunity to reveal the influence of different age and tenure groups on the risk (which is required when there is no certainty in the form of the connection between continuous variables and risk) and to describe moderation effect of compensation on each group. Compared to continuous variables for which it is hard to explain moderation effect (due to the reason that the multiplication of two continuous variables does not give the understanding of which variable exactly is growing and which does not).

Overconfidence is even more controversial metric. In the research it was measured based on the press-based metric proposed by Malmendier & Tate (2008). “Cautious” and “confident” words connected with particular CEO in the press were counted to construct the variable. These words should be related to the company the CEO works for. Afterwards, average values of “confident” and “cautious” articles during the period CEO took his/her position (during 2008-2015) are calculated. Ultimately, these average vales are compared and in the case of excess of “confident” average for the CEO, he/she is being treated as overconfident (Zavertiaeva et al., 2018).

As a result, in the research we analyze binary measures depicting CEO's traits.

CEO compensation

In order to reveal how CEO compensation influences company's risk, we have taken the data about CEO options and bonuses. Initially compensation package of CEO consists of larger number of components: salary, benefits, but these components are not the point of interest as far as salary is usually constant and doesn't influence risk of the company, benefits often cover additional perks for CEO: medical care, car service, insurance etc. and we hardly can expect this component to somehow influence the risk and performance of the company.

Furthermore, it was decided to put attention to the share which particular type of compensation takes in CEO's compensation package due to the reason that in previous papers it was found that the monetary amount of compensation is closely related to company size which does not reflect main functions of each type of compensation (Humphery-Jenner et al., 2016).

Options value was measured based on Black-Scholes model in accordance with Rogers (2002) and Gormley et al. (2013). The share of each type of compensation was measured as quotient between particular compensation amount and total amount of compensation which includes salary, bonuses and options. Benefits weren't included due to the reason that, as a rule, they take a very low part in total compensation and rarely presented in annual reports. Both applied measures of compensation are quantitative.

Controls. We use company size, share of CEO interests in the company, return on assets, innovativeness and company growth as control variables.