Stronger risk management, lower risk : evidence from the top 100 S&P 500 non-financial companies

(1)

Stronger Risk Management, Lower Risk: Evidence

From the Top 100 S&P 500 Non-Financial

Companies

Date

Author Robin Oussoren Student number 5744679

First supervisor Name: J.E. Ligterink

Grade:

---

Signature: --- Second supervisor Name:

J. van de Ven

Signature:

--- Master Business Economics

Specializations Finance Organization Economics

Pages 59

(2)

Abstract

Risk management can improve firm value by facilitating shareholders’ risk management (Demarzo and Duffie, 1991). Risk management can add value by decreasing the costs and probability of financial distress (Smith and Stulz, 1985). It can increase value by decreasing the risks faced by managers holding large fractions of equity (Meulbroek, 2001). It can increase firm value by increasing debt and thereby reducing taxes (Modigliani and Miller, 1963). Finally, risk management can improve firm value by providing internal funding for investment projects (Froot, Scharstein and Stein, 1994).

This research estimates multiple regressions to examine the relationship between risk management and risk and performance for non-financial companies. The results of this research are in line with the results of Ellul and Yerramilli (2013) and finds evidence that risk management reduces tail risk. However, like the research of Jin and Jorion (2006) and Guay and Kothari (2003), the research finds no evidence for the relationship between risk management and Tobin’s Q.

(3)

1 Introduction

August 9th_{2007 was pinpointed as the day the housing bubble blew up and the economic}

crisis had officially started (Larry Elliot, The Guardian). This was the day that the French bank BNP Paribas announced that it would cease their investment in three major hedge funds that were specialized in US mortgage debt. This was also the day that investors and bankers found out that many derivatives were worth much less than assumed. No-one knew the exact magnitude of the losses and the exposure to individual banks. This caused a lack of trust between banks and they stopped lending money to each other. Still at this time trust had not yet evaporated and banks still had the status “too big to fall”. Within a year the US investment bank Bear and Stearns was acquired by JP Morgan and the British bank Northern Rock had to be nationalized. But on September 15th 2008 the US investment bank Lehman Brothers went bankrupt and the notion “too big to fall” no longer held true. Banks were considered risky and stock prices plummeted.1

In their paper Fahlenbrach, Prilmeier and Stulz (2012) compared the performance of financial companies during the financial crisis in 2008 with their performance during the crisis in 1998. Their ‘learning hypothesis’ stated that companies who had performed poorly in the first crisis in 1998 would learn from it and would perform better during the crisis in 2008. However, they found that these companies had more risky funding, grew faster in the three years prior to crisis but still performed worse during the financial crisis of 2008. They did not appear to have changed their risk culture or business model.

According to Ellul and Yerramilli (2013) financial companies with a strong risk management outperformed those with a weaker risk management during a crisis. As a consequence the risk management within financial companies increased enormously after a crisis. Their research showed that in 1994 only 40% of their bank holding companies sample had appointed a Chief Risk Officer (CRO) and this increased to 100% in 2008. As a matter of fact they proved that companies with a stronger risk management index, before the crisis started, performed equally before but outperformed those with a weaker risk management index during the crisis. Companies with a strong risk management had lower tail risk and a higher profitability during the crisis years.

The research of Ellul and Yerramilli (2013) is limited to bank holding companies. Fahlenbrach, Prilmeier and Stulz (2012) extended their research with a non-financial control group to find out whether the bad performance of the financial companies was due to what they call the financial crisis factor. Meaning that the crisis itself is causing bad performance. Interestingly, they indeed found a positive coefficient for the non-financial control group, however it was

1_{Elliot (2001), available at:}

(6)

6

about half the coefficient they found for banks. So the performance of their non-financial company sample was less affected by the crisis.

Non-financial companies differ in many ways from banks. They are not subject to bank runs and shareholders react less fiercely during the financial crisis (see Fahlenbrach, Prilmeier and Stulz, 2012). Ellul and Yerramilli (2013) exposed the importance of a strong risk management, especially during crisis years for bank holding companies. Non-financial companies’ shareholders react less fiercely on a crisis and the benefit of a strong risk management is that it influences performance of financial companies especially during crisis years. So what are the benefits for non-financial companies for having a strong risk management and do those benefits differ during crisis years?

This research examines the following research question: Do non-financial companies with a

strong risk management index have lower tail risk and do they perform better compared to non-financial companies with a weak risk management index?

This thesis answers this question through several hypotheses that test the relationship between risk management and tail risk, Altman Z-score, Tobin’s Q and annual return. To find out what effect risk management has on the risk exposure and performance of a company. The literature shows a lot of contradicting empirical results on the relationship between risk management and firm performance. Allayannis and Weston (2001), Graham and Rogers (1999), Mayers and Smith (1982) and Smith and Stulz (1985) all found a positive significant relationship between risk management and firm performance, while Jin and Jorion (2006) and Guay and Kothari (2003) did not find significant results. This paper contributes to the existing literature by researching the relationship of risk management in a different way. The most researches looked at the implementation of ERM (Liebenberg and Hoyt (2003), Beasley, Pagach and Warr (2008), etc.) This research focusses more on risk management in the form of governance.

More specific, the variables in the risk management index are based on, activeness of the board of directors, board efficiency, activeness of the audit committee and the level of risk-averse behavior of the CEO. According to Yermack (1996) large boards are inefficient and have a negative impact on performance. Vafeas (1999) find a positive effect of the frequency of board meetings on firm performance. In a study conducted by Tufano (1996), evidence showed that managers who own more stock are less likely to take on risk than managers who own more options. While Haushalter (2000) showed that there is a negative correlation between risk management and the compensation of officers and directors. He found evidence that the number of exercisable options held per officer was negatively associated with the company’s hedging activities.

(7)

7

This thesis consists of four parts. The first part elaborates the literature research. It discusses the hypotheses and examines and compares previous literature to explain why risk management adds value to a company, to specify and to discuss the relevance of the dependent and independent variables. The second part elaborates the methodology, discusses the gathered data and describes the used sources. The third part shows and discusses the results of the research. Finally, the last part gives some concluding remarks and advice about the follow up and improvements that can be done.

(8)

8

2 Literature Review

This literature review gives an insight in risk management. All the literature contributes to answer the research question, to substantiate the hypotheses, the performance measures and the control variables. Chapter 2.1 discusses how risk management can add value. However, according to Modigliani and Miller (1958) does risk management not increase shareholder value in perfect capital markets. Nevertheless markets are not perfect, chapter 2.2 discusses the capital market imperfections and in particular the agency theory which has a huge impact on the risk exposure of a company. Chapter 2.3 discusses corporate governance, how companies can control risks and reduce the principal-agent problem. Chapter 2.4 discusses the empirical findings of the relationship between risk management, the performance and risk measures.

2.1 Risk Management, Value Enhancing

In perfect capital markets, as described by the capital asset pricing model of Markowitz (1952) and Sharpe (1964), managers who want to maximize profits do not invest in risk management. However, Smith and Stulz (1985) showed that because perfect market assumptions violation do occur, reducing risk adds value to shareholders. Implementing a strong risk management system can enhance firm value in several ways. This chapter discusses why it is important for companies to have a strong risk management system and how risk management can increase firm value.

Managing risk can have multiple motives that differ from firm to firm. Some want to hedge their downside risk while retaining the upside risk. Some firms want to hedge their full risk exposure while others only hedge partially. Meulbroek (2002) argued that the fundamental goal of risk management is to maximize shareholder value. According to Meulbroek (2002) companies can improve firm and maximize shareholder value through risk management in the following ways:

• Facilitating shareholder’s risk management and lower external monitoring costs • Decreasing costs of financial distress

• Decreasing the risks faced by managers that hold large fractions of equity • Reducing taxes by increasing debt and the tax shield

(9)

9

2.1.1 Facilitating shareholder’s risk management

Shareholders can protect themselves from systematic risk by using futures, swap contracts or by changing their safe cash holdings and risky assets. It can be costly for shareholders to constantly adjust their portfolio at the expense of transaction costs (Hu, 1996). They can only do so at a cost of lowering their own expected returns and they need to know the firm’s risk exposure. However, for shareholders it is far more difficult to estimate the firm’s risk exposure than for a manager who knows everything about the firm’s current and future activities that can change the whole risk exposure. Demarzo and Duffie (1991) stated that companies have exclusive information. This information can be kept private, because of the strategic value or due to the high costs of constantly informing shareholders about every update in a company’s production plan. It is impossible for shareholders to hedge based on information they do not have, a manager can do this on their behalf. A manager can stabilize the risk exposure by managing the risk at a targeted level. This improves the certainty about the risk exposure to systematic risk for holding that stock. Therefore shareholders will accept a smaller risk premium that lowers the costs of capital. Meulbroek (2002) argued that shareholders can adjust their own risk exposure more easily when the level of risk is constant. Moreover, enterprise risk management can create value in ways that cannot be duplicated by shareholders. Shareholder wealth can be increased by reducing monitoring costs and transaction costs inflicted by information asymmetry and a volatile risk exposure.

2.1.2. Increasing firm value by decreasing costs of financial distress

Firms can increase their value with a proper risk management system. According to Meulbroek (2002) financial distress is less likely to occur when a firm reduces its total risk. Robicheck and Myers (1966) argued that not only the costs of financial distress, but also the expected costs once in distress can destroy substantial firm value. A company incurs costs of financial distress once it is threatened with bankruptcy, even when bankruptcy can be avoided. According to Shapiro and Titman (1986) the first costs a company encounters after being under distress is the negative influence on contracts with creditors, suppliers and customers. Stulz (2000b) confirmed that the negative effect of payment problems affects the customers’ willingness to buy a product. Financial distress can also lead to loss of reputation and human capital (Shapiro and Titman, 1986). Employees want to be compensated for the risk they bear of losing their job or they can leave the company to find job security elsewhere. Smith and Stulz (1985) developed the direct bankruptcy cost argument, companies’ direct distress costs arise from legal expenses and lawyer fees. However, illiquidity can still induce the highest costs for companies in financial distress, companies have to pay higher financing costs due to lower credit ratings.

(10)

10

The direct and indirect costs of financial distress are not immediately affected by a company’s risk management. However a strong risk management can strongly reduce the probability of financial distress by reducing the volatility of cash flows. Whereas a reduction in financial distress increases the value of a firm, it also increases shareholder value by improving the firms potential to carry debt (Ross, 1996).

2.1.3 Increasing firm value by decreasing the risks faced by managers that

hold large fractions of equity

In her paper Meulbroek (2001) researched the deadweight cost of management’s stock and option-based compensation. Since undiversified managers are exposed to the total risk of a firm, but are only compensated for the systematic portion of that risk, they value the compensation under its market value. Due to stock-based pay an increasing number of managers possess large portions of company stock. Of course this can be beneficial to the firm to align the manager’s incentives with the shareholders. However, managers cannot diversify the company’s non-systematic risk that normally would be diversified by holding an efficient passive portfolio. Risk-averse managers therefore want to be compensated for bearing these risks. The firm has to pay a higher value of shares or has to offer the shares at a discount and this would cost more than a compensation paid in cash. The higher the risk exposure of the firm the bigger the gap between the managers private valuation of shares compared to the market valuation. Jin (2001) underlined that as firm specific risk increases, pay becomes less sensitive to performance. So companies with managers holding large fractions of stock can increase company value by reducing the risk the company and these managers face.

2.1.4 Increasing firm value by reducing taxes

Modigliani and Miller (1963) were the first to argue that tax benefits of debt increase firm value. Interest expenses are tax deductible and therefore a company’s value increases with its debt-to-equity ratio. Companies can increase their value by increasing their debt capacity to the optimal debt level and therefore obtain a higher tax shield associated with that debt. Additional debt can also lead to extra risks and costs. According to DeAngelo and Masulis (1980) financial distress costs of debt offset some of the tax benefits. However Talmor, Haugen and Barnea (1985) showed that the increasing amount of debt can increase the tax benefits faster than it increases the probability of bankruptcy and Parrino and Weisbach (1999) showed that the agency costs of debt are too small to offset the tax benefits of debt. According to Graham and Smith (1996) the effect of risk management is stronger when the volatility of a company’s corporate income is higher. It is important for companies to keep

(11)

11

their cash flows constant to raise new debt from debt-holders. Volatility determines whether a debt-holder gets paid or not and whether the debt-holder is willing to lend money. Firms with strong risk management systems are able to reduce risks and support a greater debt-to-equity ratio. Smith and Stulz (1985) were the first ones to analytically describe the effect of risk management on the corporate tax bill. When the income is subject to a convex tax code, the corporate tax bill can be lowered by reducing the volatility of the income through risk management.

2.1.5 Provide internal funding for investment projects

Froot, Scharstein and Stein (1994) argued that through risk management, corporate investment and financing policies can be coordinated more efficiently and hereby increase shareholder value. The value of a firm can be increased by undertaking positive net present value projects, however funding is needed for these projects. Corporate cash flows can be volatile, so internal funds are not guaranteed at every point of time. Risk management can smooth out cash flow volatility to ensure a stable cash flow enabling companies to fund investment opportunities internally. Raising external debt as well as equity, in imperfect markets, is not optimal since agency and transaction costs lead to additional costs.

The issuance of equity is, according to Asquith and Mullins (1986), also not always optimal. Managers have inside information that external investors do not have. Companies do not want to fully disclose all their private information because their competitors can use that information too. External investors believe that companies act in the best interest of the current shareholders, companies that issue new equity are most likely to do that when the stock price is overvalued. By issuing new equity the company subconsciously signals the market that the stock price is overvalued this leads to a reduction in stock price.

Financing projects with external debt has less asymmetry problems and is therefore preferred over equity issuance (Stiglitz and Weiss, 1981). However agency costs that arise from the relationship between shareholders and debt-holders increase the costs of external debt funding. In addition, external financing is also subject to transaction costs like bank fees and syndication fees (brokerage fees, registration fees, legal fees, etcetera).

Due to these costs companies generally choose internal funding over external funding (Myers, 1993). Froot, Scharfstein and Stein (1993), argued that risk management can smooth out cash flow volatility and thereby create stable internal funding to guarantee the realization of NPV projects and at the same time avoid higher cost of capital. This eventually leads to an increase in shareholder value.

(12)

12

2.2 Capital Market Imperfections

The previous chapter shows that risk management can add value to the company and its shareholders. According to the Modigliani and Miller (1958) proposition does risk management not increase shareholder value in perfect markets (without transaction costs, information asymmetry or taxes). Shareholders can always replicate the actions of the company with transactions in the capital market. So the only way to increase firm value is by realizing positive NPV projects and it is irrelevant whether they are financed with debt or equity. However, Fite and Pfeiderer (1995) show that the Modigliani and Miller propositions do not hold in real world. Capital market imperfections exist, therefore risk management has a positive impact on firm and shareholder value.

The most common form of information asymmetry is studied in the context of the principal-agent problem.2_{This chapter gives an insight in how capital market imperfections arise}

through information asymmetry and how companies can align managers’ interest with the interest of shareholders by giving the right incentives through compensation.

2.2.1 Agency theory

One of the most cited articles about the agency theory was published in 1976 by Jensen and Meckling. An agency relationship is a contract under which one or more principals appoint an agent, who is given some decision power from the principal to perform some services on their behalf. The problem that arises, under the assumption that all parties try to maximize their own utility, is that the agent not always acts in the best interest of the principal. The principal can introduce incentives to reward the agent when he act in the best interest of the principal. He therefore incurs monitoring costs, but ensures that the agent is motivated or punished if he takes actions in his own, instead of the principal’s, best interest. As the managers’ fraction of equity falls, his fraction of the total outcome also falls and may encourage him appropriating larger amounts of the corporate resources. So when the equity of the agent falls it increases the desire for the principal to monitor the agent’s behavior. In the past decades a lot of research has been done based on the theory provided by Jensen and Meckling (1976). For instance it has been applied on dividend policies3_{, issues on why}

entrepreneurs concentrate large fractions of their wealth in firm equity4_{, and by Ang et al.}

(2000) about ownership structures. Ang et al. (2000) found results on ownership structures that are in line with the outcomes of Jensen and Meckling (1976). The more equity a manager owns the lower the agency costs due to better incentive alignment.

2_{See Akerlof (1970), Spence (1973) and Stiglitz (1981).} 3_{See Lopez de Silanes, Vishny and Shleifer (2000)} 4_{See Bitler, Moskowitz and Vissing-Jørgensen (2005)}

(13)

13

2.2.2 Incentives and compensation

Baker, Jensen and Murphy (1988) discussed a lot of different compensation methods in their paper. There are a lot of ways to compensate employees and managers. In addition to the performance based compensation, also promotion-based incentive systems are widely used. Other compensations are holiday bonuses, profit sharing, tenure and up-or-out promotion systems. However, at the higher level, bonus based incentives are more important since the CEO is not promotable and financial incentives therefore must come from bonuses. There is a lot of criticism on monetary pay for performance systems.5_{However, according to Baker,}

Jensen and Murphy (1988) the problem of the performance system is not that they are ineffective but that they can be too effective. It motivates people exactly to do what they are told to do, it is however sometimes difficult to express exactly what employees should do. An advantage of bonus-based compared to promotion based incentives is that they can be applied on everyone and are not limited to the few individuals who actually have promotion opportunities.

In a study conducted by Tufano (1996), evidence showed that managers who owned more stock were less likely to take on risk than managers who owned more options. This is supported by Knopf, Nam and Thornton (2002), who tested the relationship between hedging activities and the risk preference of managers. They found results that showed that companies hedge more when the sensitivity of the total portfolio to stock price increases, and the opposite happens when the stock option portfolio to stock return volatility increases. They also compared the contradictory results of Gay and Nam (1998) who found a positive relation between options and hedging and Smith and Stulz (1985) who found a negative relation between options and hedging. Knopf, Nam and Thornton (2002) disentangled the effects of sensitivity to stock price and sensitivity to stock return volatility and found evidence that supports the theory of Smith and Stulz (1985).

The research of DeFusco, Johnson and Zorn (1990) also supports the theory that executives undertake more risky investments when having a stock option incentive. Besides the effect of compensation on managers they also examined the market reaction to the announcement of a stock option plan. Remarkably, their results showed that the wealth of shareholders increases after an option compensation announcement and the bond market showed a significant negative reaction surrounding the announcement. This is consistent with rational investors anticipating on an increase of the managerial risk taking. However, the downside of option compensation is also an advantage. When a manager does not have options at all the possibility exists that this manager is more concerned about job security and may reject positive NPV projects. This would not benefit the shareholders. With option

5_{See Baker, Gibbons and Murphy (1994), ‘Subjective Performance Measures in Optimal Incentive Contracts’,}

The Quarterly Journal of Economics, Vol. 109, pp. 1125-1156. Or Hamner (1975), ’How to Ruin Motivation with Pay’, Compensation Review, Vol. 7, pp. 17-27

(14)

14

incentives this effect can be avoided. A finding of Ofek and Yermack (2000) about stock and option rewarding is remarkable. They examined the incentive effects of equity-based compensation and compared the results of higher equity ownership with lower equity ownership. When a manager had higher-ownership and he received options they sold the stock that they already owned to reduce their risk exposure. Lower-ownership managers kept the shares they already possessed after receiving options. So the findings of their research suggest that equity compensation only lifts the incentives of lower-ownership managers.

2.2.3 Characteristics Risk Taking Managers

Also characteristics can increase the risk taking behavior of managers. Mac Crimmon and Wehrung (1990) showed that a higher maturity of a manager leads to less risk taking. The most risk-averse managers were the older managers who work longer at the same firm. Another remarkable result from their research is that executives who worked for large banks were more risk-averse than their colleagues who worked in different industries. Tufano (1996) found no proof that the age of a manager influenced his risk-taking behavior. The results of job tenure of the CFO, on the other hand, were negatively associated with risk management. Likewise, new CFOs managed more of their firm’s risk and were more likely to engage in greater risk management activities compared to CFOs with a higher tenure. Tufano (1996) also studied firm’s characteristics in relation with risk management but he found no significant relationships.

(15)

15

2.3 Corporate Governance

The Financial Reporting Counsil (combined code, June 2008) set out the purpose of good corporate governance as helping a board discharge its duties in the best interest of shareholders contributing to better company performance. Good governance should facilitate effective and efficient management that can create shareholder value over the long term. The OECD Principles of Corporate Governance underpin this view. An effective corporate governance system lowers the cost of capital and encourages firms to use resources more efficiently and thereby sustaining growth.6

According to Shleifer and Vishny (1997) corporate governance deals with the way investors assure themselves of getting a return on their investment. Their perspective on corporate governance is the perspective of ownership and control. So on how investors can motivate management to return their investment. Lemmon and Lins (2003) stated that the firm’s ownership structure is a key determinant of the extend of agency problems between controlling inside and outside investors. Core, Holthausen and Larcker (1999) found evidence that companies with weaker governance had greater agency problems, performed worse and CEOs received higher compensations.

The previous chapters are about how risk management can create additional firm and shareholder value and how capital market imperfections (in particular the principal-agent problem) makes risk management a necessity in creating this value. The following chapter of the literature review focusses on corporate governance, so on how risk management controls risks and reduces excessive risk taking by CEOs. This chapter discusses internal corporate governance in particular, because an important part of the methodology is based on the effectiveness of ERM systems, board of directors and the audit committee.

2.3.1 Enterprise Risk Management

Enterprise risk management is a phenomenon that grew rapidly after 1990, firms increased their efforts to organize uncertainty rapidly. Managing uncertainty became more important to prevent catastrophes and large losses. Decision making, accountability, risk management, objective-setting and business strategy, the ERM approach seeks to link them all together (Arena, Arnaboldi and Azzone, 2010). The corporate governance rules set up by the New York Stock Exchange already require listed companies to discuss the risk management policies in their audit committees. Standard and Poor assesses ERM processes as part of corporate credit ratings analysis.

6_{OECD (2004), ‘OECD Principles of Corporate Governance’, Available at:}

(16)

16

2.3.1.1 The Benefits of Enterprise Risk Management Systems

According to Bowling and Rieger (2005) is a good ERM system important for a company as well for the stakeholders. There is little tolerance for negative returns and ERM is essential for managing stakeholder’s expectations and for good governance. Investors do not want companies to eliminate all risk, since taking and managing risk can create value. Investors want companies to manage the risks effectively to ensure financial stability and fewer excessive negative returns. The perfect form of corporate governance takes all the stakeholders’ interest into account. From shareholders, customers, employees, etcetera, since they all share the same interest which is a successful continuation of the company. Enterprise risk management is one of the most important mechanisms of the framework of corporate governance and management should have broad understanding and experience on how to manage risks and dealing with all stakeholders to make the company prosper (Bowling and Rieger, 2005).

Internal auditors are an important element in making a ERM system successful. While the ultimate responsibility must lie with the parent board, the management at lower levels should manage the wide variety of risks (Fraser and Henry, 2007). This idea is supported by Tufano (1996) who argued that not only the CEO is associated with the risk management choices but the entire managerial team.

2.3.1.2 Chief Risk Officer

Supporters of ERM argue that firms which adopt an ERM strategy should make a person or a group responsible for implementation, coordination and communicating the goals and results to the board. Arena et al. (2010) argued that during the rise of the ERM philosophy the new role of Chief Risk Officer (CRO) formed within organizations. CROs hold advantages compared to traditional risk managers. They possess the communication skills to promote ERM to the board and other internal and external stakeholders and often have a high level of technical expertise. They are appointed by the board and they report directly to the CEO and/or the CFO. Liebenberg and Hoyt (2003) argued that a company appoints a CRO if they want to implement an ERM program and need someone to manage the whole process. This does not mean that companies which have not appointed a CRO do not have an ERM program in place since the responsibilities can also lie with a risk champion or in the function of the CEO or CFO.

This is somehow confirmed by a paper published by Forbes Insights in association with Deloitte. In their survey, under 192 U.S. executives from companies in the health care, life sciences, consumer and industrial products, and technology/media/telecommunications industries, they found that most of the respondents (26%) think that the main responsibility

(17)

17

for overall risk management belongs to the CEO, second came the CFO (23%) and third came the CRO (19%).7_{Suder (2006) observed that in non-financial organizations the function of}

CRO relatively new is and that most non-financial companies do not even have a CRO function. In that case are the functions of a CRO fulfilled by the CFO and when they do have a CRO than he often has to report directly to the CFO. Ellul and Yerramilli (2013) showed in their research that in 1994 only 40% of their Bank Holding Companies sample had appointed a CRO and this increased to 100% in 2008. The crisis was an important factor to appoint a CRO and for a good reason. They prove that companies with a stronger risk management index, before the onset of the crisis, performed equally before but outperformed those with a weaker risk management index during the financial crisis.

2.3.1.3 Characteristics of Firms hiring a CRO

The study of Liebenberg and Hoyt (2003) provided an initial attempt at identifying the determinants of ERM adoption. They constructed a sample of firms that have signaled their use of ERM by appointing a CRO who is charged with the responsibility of implementing and managing the ERM program. They found that companies that were highly leveraged were more likely to appoint a CRO. Also the wave of corporate governance scandals increased the willingness to appoint CROs to signal to external shareholders the company’s commitment to risk management.

Pagach and Warr (2011) extended the research of Liebenberg and Hoyt (2003) by adding more firms to their sample, they analyzed a larger number of determinants of CRO hiring and they used a hazard model instead of a logit-model. In addition to the results of Liebenberg and Hoyt (2003) they found that firms were more likely to hire CROs and adopt ERM systems if they were larger, had a greater institutional ownership, had a greater risk of financial distress and had more volatile operating cash flows. Also when the CEO had more incentives to take risk increased the probability that a firm hired a CRO to provide control mechanism against the CEO’s incentive to take risk. Companies are hiring CROs for the benefits of ERM, it is of course also a way to show extern stakeholders the willingness to focus their attention on risk management.

7_{Deloitte and Forbes Insight (2012), ‘Aftershock: Adjusting to the new world of risk management’, p. 4.}

Available at:

http://www.deloitte.com/view/en_AU/au/industries/tmt/c56e46fd75ef8310VgnVCM3000001c56f00aRCRD.ht m# (accessed June 7, 2014)

(18)

18

2.3.2 Supervisory Board

A board consists of three types of players, the CEO, inside and outside directors. The CEO can propose either good or bad projects to the board based on his own incentives. Inside directors are the firm’s senior managers, they are more informed about the firm’s projects compared to outside directors. Outside directors can motivate inside directors to reveal their private information by evaluating them for their CEO succession votes. So competitive inside directors can distinguish themselves from other inside directors by sharing their private information. Inside directors can also reveal their information due to a lack of independence from the CEO or due to possible private benefits to firm’s managers by accepting ‘bad’ projects.

According to Raheja (2005) the optimal board composition is a trade-off, a higher percentage of inside directors makes is difficult for outside directors to win a majority vote to reject negative projects, because more insiders have to defect from the CEO. Nevertheless this can still be favorable if it is difficult for independent outside directors to evaluate complex projects and when the benefits of becoming the CEO for inside directors increases. Therefore smaller sized boards are preferred when the incentives of internal directors are aligned with those of the shareholders. When it is easy for outside directors to evaluate projects (and not too costly) than they are preferred in boards. However, when projects are too complicated for outsiders, like in high tech firms sometimes can be the case, a higher proportion of inside directors is favorable.

On November 4 2003 the SEC approved the final corporate governance rules of the New York Stock Exchange. Companies listed on the NYSE must meet these corporate governance rules. Rule 7 c iii D is explicitly about risk assessment and risk management. This rule states that the audit committee must discuss policies with respect to risk management and assessment. While it is still the task of the management to assess and manage risk, the audit committee must discuss the major financial exposure of the firm and the way the management monitors and controls this exposure.8

2.3.2.1 Independent Directors

Shivdasani and Yermack (1999) researched the effects of the involvement of the CEO in the selection process of new board members. They found that when the CEO was involved in the directors appointment, the director was less likely to monitor the CEO aggressively and stock price reactions were significantly negative.

8_{New York Stock Exchange (NYSE), 2003, Final NYSE: Corporate Governance Rules, Available at:}

(19)

19

Yermack (2004) found that when a director started at a company, in most cases he had no or little stock. As their tenure increased, their fractional ownership in the firm increased with almost a factor 12 after the fifth year compared to their first year fractional ownership. He also found that more than half of a directors’ incentives comes from stock and stock options, they represented about 55% of the total incentives. Obtaining new directorships accounts for about 40% of the motivation of the directors and only 5% for losing a directorship and an increase in annual turnover.

Klein (2002) argued that using a 51% majority of outside directors is desirable. Since it is favorable but costly to get a board that only consists of independent directors. The findings of Xie, Davidson and DaDalt (2003) support the recommendation for audit committee members to be independent board members with financial expertise. With more independent outside directors earnings management (mislead stakeholders by changing financial reports to obtain personal gain) was less likely to occur. Also meeting frequently by the board and audit committee reduced earnings management on lower management levels. So active boards and a large portion of outside directors are associated with better monitoring. Also smaller boards are more effective than larger boards. According to Ferris, Jagannathan and Pritchard (2003) when a director serves larger firms and larger boards he is more likely to attract new directorship. Further does their evidence not support that a higher number of directorships held by an individual director affects his work negatively.

2.3.2.2 Board Size & Meeting Frequency

Cheng (2008) studied the effects of the board size on the variability of corporate performance. He found that larger boards were associated with less negative occurrences like variable accruals, restructuring activities and analyst forecast inaccuracy. But also with less positive occurrences like frequent acquisition and R&D spending. According to Cheng this is most likely due to the fact that larger boards have more difficulty to reach consensus and have to compromise more. This led to less risk taking but does not have to be optimal for all stakeholders. Yermack (1996) criticized large boards due to poor communication and decision making. He found prove that as board sizes grew, profitability and operating efficiency appeared to decline. The threat of dismissal and performance incentives through compensation operated less strongly. Shareholders reacted negatively on board expansion and positively on significant board size reductions. Lipton and Lorsch (1992) recommend in their paper to limit the total membership of the board to ten people. They stated that the optimal total board size consists of eight or nine persons. Vafeas (1999) studied the frequency of board meetings and the effects on firm performance. His results indicated that boards are reactive, rather than proactive. Boards met more frequently responding to poor performance. However, he did find that the operating performance of a firm who’s board met abnormally frequently improved.

(20)

20

2.4 Risk Management and Performance,

Empirical Evidence

The previous chapters discusses the value of risk management as well as the associated corporate governance. This chapter adds some additional empirical evidence of the relationship between risk management and firm value, tail risk and CEO compensation.

2.4.1 Relationship with Firm Value

Jin and Jorion (2006) investigated the effects of hedging activities in the American oil and gas industry. They tested the difference in firm value between firms that hedged and firms that did not hedge their gas an oil price risk. They used Tobin’s Q to compare the different firm values but they did not find a significant different result for companies that hedged their price risks and companies that did not hedge. Guay and Kothari (2003) argued that the derivative positions of non-financial companies are too small to affect firm value. Contrary results were found by Allayannis and Weston (2001) who found positive significant evidence of hedging on Tobin’s Q. They examined the use of foreign currency derivatives in a sample that consisted of 720 non-financial companies. Companies that faced currency risk and used derivatives to hedge these risk had on average a 4.87% higher Tobin’s Q than companies that did not use derivatives. Graham and Rogers (1999) also found evidence that companies use derivatives to hedge in a way that is in line with optimal risk management. Their results indicate that companies respond on the high costs of financial distress and underinvestment by hedging. Mayers and Smith (1982) and Smith and Stulz (1985) found the same results. Graham and Rogers (1999) also found that hedging led to a higher debt capacity and extra interest deductions, which resulted in an increase between 2.2% and 3.5% to firm value.9 They also found a positive relation between hedging and R&D expenses, which is in line with the theory of Froot, Scharfstein and Stein (1993) that firms hedge to minimize underinvestment problems. Bartram, Brown and Conrad (2011), Carter, Rogers and Simkins (2006) and Nelson, Moffitt and Affleck-Graves (2005) also found a positive relationship between risk management and firm value.

In 2003 Gompers, Ishii and Metrick studied the effects of shareholder rights on the performance of companies. To do so they built a corporate governance index for approximately 1500 companies that consisted of 24 governance rules. They divided their sample in subgroups based on the level of the governance index. The highest G-index group consisted of companies with the worst shareholder rights and the lowest G-index group consisted of companies with the best shareholder rights. They cross checked the index with

(21)

21

firm characteristics and found that firms with a high G-index were large S&P firms with relatively high share prices, trading volume and institutional ownership. They also looked at the performance and found that firms with high G-index performed worse compared to those with a low governance index. The high governance index companies had poor stock-market performance and relatively poor sales growth. When you invested $1 in the relatively good corporate governance companies’ (low G-index) portfolio in 1990 you would have earned $7,07 in 1999 compared to $3,39 for the portfolio of high G-index companies. That is an annual return difference of more than nine percent. Gompen et al. (2003) also found significant evidence that companies with high shareholder rights outperformed companies with low shareholder rights.

2.4.2 Relationship with Enterprise Risk Management

Liebenberg and Hoyt (2003) constructed a sample of companies that signaled their implementation of ERM by appointing a CRO. Their findings were in line with the hypothesis that companies hire CROs to reduce information asymmetry. They found that companies that appointed a CRO had a greater financial leverage. Pagach and Warr (2011) also researched the characteristics of companies that hired a CRO. They found that companies that were larger, had higher earnings volatility, had greater institutional ownership, had poorer stock returns and had a CEO whose compensation increased with stock volatility were more likely to appoint a CRO to implement ERM. Beasly, Clune and Hermanson (2005) found that the stage of ERM implantation was positively related to the appointment of a CRO and the independency of the board of directors.

The value of risk management was researched by Hoyt and Liebenberg in 2011. Their research focused on the American insurance industry and they used Tobin’s Q as a standard proxy for firm value. They found a positive significant relation of 20 percent from the use of ERM on firm value. Beasley, Pagach and Warr (2008) conducted a study of the market reaction to the appointment of a senior executive to oversee the ERM process. They found a positive market reaction for non-financial companies but not for the financial companies. The research of Gordon, Loeb and Tseng (2009) was based on a sample on 112 American companies that disclosed their ERM activities in the 10K and/or 10Q statement. They used one-year excess returns as a measurement for performance. They found that the relationship between ERM and firm performance is conditional on the match between environmental uncertainty, firm size, the monitoring of the board of directors and industry competition. A greater deviation from this match is associated with a lower firm performance. McShane, Nair and Rustambekov (2011) did not find a significant positive relationship between firm value and ERM implementation. They found that Traditional Risk Management increases firm value but that it did not further increase when firms achieved ERM.

(22)

22

2.4.3 Relationship with Tail Risk

Ellul and Yerramilli (2013) researched the effects of strong risk management functions and the risk oversight of the directors at Bank Holding Companies (BHC). They found proof that BHC with a strong and independent risk management function had lower tail risk. Ellul and Yerramilli (2013) found that especially during the financial crisis years the better performing BHCs were those with a higher risk management index. They had lower tail risk, higher annual returns and higher return on assets. Taken the results over the whole time period (1995-2010) they still found better operating returns and lower tail risk but no higher stock returns. Ellul and Yerramilli (2013) also looked at the determinants of companies with a high risk management index. In contrast to Gompen et al. (2003), Ellul and Yerramilli (2013) found that larger BHC had stronger risk management systems. The RMI was higher for firms with a lower tier 1 capital ratio, with a larger fraction of income in non-banking activities and with larger derivative trading operations. The RMI was also higher for the BHC that motivated CEOs to take on risk through option compensation and had more independent boards. The logic behind this is that boards give managers the incentive to take risk but at the same time control these risks more intensively. Ellul and Yerramilli also proved that BHCs with a high RMI were also more likely to be funded with riskier short term debt. This is in line with the theory stated by Meulbroek (2002) that companies with stronger risk management systems can obtain debt more easily. Haushalter (2000) also supports this theory by showing that companies with a larger financial leverage managed their price risk more intensively. Pagach and Warr (2011) argued that companies with more volatile cash flows would benefit more from ERM by smoothing cash flows and hereby reducing the probability of a lower tail cash flow outcome. This is supported by Kashyap, Rajan and Stein (2008) who stated that financial companies with an independent and strong risk management should experience lower enterprise wide tail risk and that a strong risk management is necessary to identify and correct excessive risk-taking. This theory is also supported by Nocco and Stulz (2006) who found that managing risk on the long run, created value by reducing costly lower tail outcomes. A strong and effective risk management protects assets from lower-left-tail outcomes preventing destruction of shareholder value.

2.4.4 Relationship with CEO Compensation

Core, Holthausen and Larker (1999) looked at the effects of board and ownership structure on the CEOs compensation. They found that companies with weaker governance had greater agency problems, performed worse and CEOs received higher compensations. The CEO’s compensation was higher when the board of directors was larger, had a higher percentage of directors appointed by the CEO and when the CEO was also the chairman of the board. On

(23)

23

the other hand, the compensation was smaller when the CEO had a larger stake in the company. A remarkable outcome of their study, conflicting with the SEC’s guidelines for improving corporate governance, was that they found no evidence that more outside directors led to a more effective board. Nor did they find results that governance systems were improved by giving outside directors more equity ownership. Haushalter (2000) showed a negative correlation between hedging and the compensation of officers and directors. He found evidence that the number of exercisable options held per officer was negatively associated with the fraction of hedged production. These results are consistent with the results of Schrand and Unal (1998), Tufano (2006) and Smith and Stulz (1985) who also showed that managers awarded with options have less incentives to reduce risk and hedge. However the researches of Graham and Rogers (1999) and Géczy, Minton and Schrand (1997) found no significant proof that CEOs stock and options are related to hedging.

(24)

24

3 Research Question & Hypotheses

This chapter formulates the research question and the hypotheses in this paper.

3.1 Research Question

Ellul and Yerramilli (2013) built a risk management index based on governance and risk oversight. They proved the positive effects of a strong risk management index on risk and performance for BHCs. The importance of a strong risk management function for non-financial companies has not been tested in a similar way yet and the literature shows a lot of contradictions between the relationship of risk management, risk and performance. Fahlenbrach, Prilmeier and Stulz (2012) showed that the effects of the crisis were less for non-financial companies. Non-financial companies are less sensitive to the risk that financial companies have to account for, they are not subject to bank runs and shareholders react less fiercely during crisis years. Also the results of Mac Crimmon and Wehrung (1990) that executives who worked for large banks were more risk-averse than their colleagues who were working in different industries would suspect that results could be different for non-financial companies.

This research contributes to the existing literature by answering the following research question:

Do non-financial companies with a strong risk management index have a lower tail risk and do they perform better compared to non-financial companies with a weak risk management index?

3.2 Hypotheses

Risk management can improve firm value by facilitating shareholders’ risk management (Demarzo and Duffie, 1991). Risk management can add value by decreasing the costs and probability of financial distress (Smith and Stulz, 1985). It can increase value by decreasing the risks faced by managers holding large fractions of equity (Meulbroek, 2001). It can increase firm value by increasing debt and thereby reducing taxes (Modigliani and Miller, 1963). Finally, risk management can improve firm value by providing internal funding for investment projects (Froot, Scharstein and Stein, 1994).

The literature shows a lot of contradicting empirical results on the relationship between risk management and firm performance. Allayannis and Weston (2001), Graham and Rogers

(25)

25

(1999), Mayers and Smith (1982) and Smith and Stulz (1985) all found a positive significant relationship between risk management and firm performance, while Jin and Jorion (2006) and Guay and Kothari (2003) did not find a significant result. This paper is contributing to the existing literature by researching the relationship of risk management in a different way. Where most researches looked at the implementation of ERM (Liebenberg and Hoyt (2003), Beasley, Pagach and Warr (2008), etc.), is the focus of this research more on risk management in the form of governance. Most of these papers were written before the onset of the crisis, however, Ellul an Yerramilli (2013) showed that risk management is even more important during crisis years. The first hypothesis is therefore divided in two:

Hypothesis 1a:

Companies with a stronger risk management index perform better than companies with a weaker risk management index during non-crisis years.

Hypothesis 1b:

Companies with a stronger risk management index perform better than companies with a weaker risk management index during crisis years.

Smith and Stulz (1985) and Meulbroek (2002) argued that risk management can also reduce the probability of financial distress. Empirical evidence showed that risk management can reduce tail risk and therefore reduce the probability of financial distress. Ellul and Yerramilli (2013) and Nocco and Stulz (2006) found a positive relationship between risk management and tail risk. The second hypothesis examines this relationship to find out whether risk management can reduce tail risk and lower the chance of default.

Hypothesis 2a:

Companies with a stronger risk management index experience less risk than companies with a weaker risk management index during non-crisis years.

Hypothesis 2b:

Companies with a stronger risk management index experience less risk than companies with a weaker risk management index during crisis years.

Risk management can add value by reducing the risk of a company and therefore the risk faced by managers that hold large fractions of equity (Meulbroek, 2001). Tufano (1996), showed that managers who owned more stock were less likely to take on risk than mangers who owned options. Haushalter (2000) also found that the number of options was negatively related to hedging activities. However a risk-averse manager can also reduce firm value. DeFusco, Johnson and Zorn (1990) argued that when a manager does not have options at all the possibility exists that this manager is more concerned about job security and may reject positive NPV projects. This would not benefit the shareholders and with

(26)

26

option incentives this can possibly be avoided. Reducing risk can add firm value while reducing the risk-taking behavior of CEOs also can destroy value. These two theories are tested in the following hypothesis.

Hypothesis 3:

Companies with a stronger risk management index and a risk-taking CEO perform better than companies with a stronger risk management index and a risk-averse CEO.

(27)

27

4 Data & methodology

The purpose of this chapter is to define the variables used to build the risk indexes, how and wherefrom the data is gathered and to explain which methodology is used.

4.1 Descriptive statistics

This research examines the effect of a strong risk management index on companies’ risk and performance. Ellul and Yerramilli (2013) set up a risk management system to estimate the effects of a strong index on the risk and performance of BHCs. The most important variable in the index was the existence and relative power of the CRO. Over the years an increasing number of banks hired a CRO and the value of the index grew almost equally. The other variables of the index were governance related like the independency of the members of the board of directors, financially educated or experienced directors and the amount of meetings of the risk committee in the previous year.

This research build a risk management index that suits a non-financial company sample better. The independency and financial experience of board members is not added to the index because of the corporate governance rules set up by the New York Stock Exchange. According to those rules every audit member must be financially literate or become financially literate after his appointment within a reasonable time. Additionally, a board should have a majority of independent directors.10_{Also the CRO measures do not suit the}

non-financial sample, in non-financial companies the function of CRO is not a common one. According to a research of Forbes and Deloitte (2012) the primary responsibility for overall risk management lies with the CEO in the first place. The level of risk-aversion of the CEO is therefore added to the index to replace the CRO variables.

4.2 Risk management index

The estimated RMI consists of a number of governance, risk oversight variables and a CEO risk-averse dummy variable. According to Yermack (1996) large boards are inefficient and have a negative impact on performance. Vafeas (1999) found a positive effect of the frequency of board meetings on firm performance. Therefore a dummy variable is created for board efficiency that is based on the size of the board of directors. The variable gets the value one when the number of board members in a company is lower than the mean of the

10_{New York Stock Exchange (NYSE), 2003, Final NYSE: Corporate Governance Rules, New York Stock Exchange,}

New York, World Wide Web: http://www.ecgi.org/codes/documents/finalcorpgovrules.pdf (accessed June 24, 2014)

(28)

28

total number of directors each year. Also two dummy variables are created for the total amount of board meetings. The board of directors consists of a number of directors who meet several times during a year with the whole board. Besides these meetings the board of directors also have several committees which also meet to discuss matters like compensation, governance and risks. The committee for the risk oversight is the audit committee or risk committee. Every company has an audit that consists of independent and financial experienced directors. When the audit committee met more during a year than the medium of the sample than the dummy variable gets the value one. The same applies for the dummy variable for the total board meetings (which consists of board and committee meetings excluding audit committee meetings).

The last dummy variable is created to include the risk-taking behavior of the CEO. The CEO’s Delta and Vega are compared to estimate the incentives of the CEO to take risks.11_According

to Knopf et. al (2002) the Delta is an estimate of the CEO’s sensitivity to changes in the stock price and the Vega is an estimate of the CEO’s sensitivity to stock return volatility. A high Vega encourages managers to take risks while the Delta discourages risk-taking. The Delta and Vega are estimated following the method described by Core and Guay (2002) and Coles, Daniel and Naveen (2006). This is actually the Black-Scholes (1973) option valuation model modified by Merton (1973) to account for dividends. The values of the Delta and Vega are calculated for every year across the whole sample and the dummy variable gets the value one when the particular CEO has less incentives to take risks (stock price sensitivity – stock volatility sensitivity) compared to the medium of a year’s sample.12

The RMI is the sum of four dummy variables and therefore can have a value between zero and four were zero means that a company has weak risk management and four that it has a strong risk management in place. To test hypothesis 3, a risk-taking CEO improves the performance of a company with a strong risk management, additional RMI sets are created. RMI-1 is an index including a dummy variable that gets the value one when the CEO is risk-averse while in RMI-2 the dummy gets the value one when the CEO is risk-taking. Also a third RMI is created, RMI-3 is without a CEO risk-averse or risk-taking dummy and is only looking at the activeness of the board of directors and audit committee.

The RMIs are created to research the effects of risk management indexes that are based on governance, risk oversight and incentives to take risks by managers:

RMI-1 Value between 0-4

Efficient board: Board members < medium sample = 1 Active board : Board meetings > medium sample = 1

11_{See appendix for comprehensive calculations Delta and Vega.}

(29)

29

Active audit committee: Audit committee meetings > medium sample = 1 CEO risk averse: CEO risk-averse > medium sample = 1

Efficient board: Board members > medium sample = 1 Active board : Board meetings > medium sample = 1

Active audit committee: Audit committee meetings > medium sample = 1 CEO risk averse: CEO risk-averse < medium sample = 1

Active board : Board meetings > medium sample = 1

Active audit committee: Audit committee meetings > medium sample = 1

The first RMI treats meetings as favorable for future performance. A lot of meetings combined with an efficient board and a risk-averse CEO should lead to less risk and a better performance. RMI-2 treats the CEO risk-averse variable as not favorable, therefore the criteria for risk-averse CEO is reversed. A high RMI-2 means that the board meets often and has a CEO that has more incentives to take risk due to his option compensation. RMI-3 supports the first RMI, but is only looking at the board and committee meetings variables. The difference can be assigned to the absence of the dummy variables, efficient board and risk-averse CEO.

4.3 DATA

The companies that are included in the sample are non-financial listed companies in the top 100 of the S&P 500 based on the book value of total assets in 2013. Large publicly listed companies are used, because they are required to file an annual 10-K statement with the SEC and the compensation information of large companies are publicly known. The collected data is from 2002 up and until 2008, to exclude the effects of the Internet Bubble and include the effects of the recent financial crisis. There are 85 non-financial companies in the S&P 100, four companies have been excluded because they were not publicly traded for the whole sample period, eight companies because of insufficient governance information provided in their proxy statements and one due to a lack of compensation information. The remaining 72 companies are operating in the following industries:

(30)

30 Table I

The data regarding the variables of the size and meetings of the board of directors and its committees are hand collected from the annual 10-K statements and proxy statements filed by the companies with the Securities and Exchange Commission (SEC). For all of the remaining 72 companies the data is gathered for the years 2002 up and until 2008, so over 500 fillings are gathered from the SEC EDGAR database.

The information to calculate the tail risk and annual returns are obtained from the CRSP Database (a database for historical security prices and return information). Other financial data is collected from balance sheets and profit and loss statements available at the Compustat Database. The information about the CEO’s compensation is obtained from the Compustat Execucomp Database. This information is used to calculate the CEO’s Delta and Vega, however due to a change in reporting format in 2005 additional information is gathered from the CRSP and Compustat.13

4.3 Methodology

There are three RMI’s created to prove or reject the hypotheses. To show value of the different methods a summary of statistics has been created over the period 2002 up and until 2006. Instead of calculating the RMI’s year to year the sum of all index variables (2002-2006) are used to calculate one RMI over the whole period. The logic behind this is that companies which board met most over the whole period have a stronger risk management culture than those who only met when the performance was bad. The RMI’s are divided over a high versus low index score so an univariate analysis could be carried out.

13_{See the appendix for comprehensive calculation and explanation of Delta and Vega.}

Information technology 10 Health Care 11 Consumer Discretionary 10 Energy 11 Industrials 14 Consumer Staples 8 Materials 4 Telecommunications services 2 Utilities 2 Total 72

(31)

31

To get a better understanding of RMI a panel regression is estimated to find the determinants of RMI. One of the most important control variable is debt-to-equity ratio, because according to Graham and Smith (1996) debt increases with a better risk management and this increases firm value which is tested in hypothesis one.

RMIj,t = α + β ∗ Xj,t−1 +Year FE (1)

In the equation above, subscript j denotes the companies and t denotes the year. In this regression, important company’s financial characteristics (Xj,t−1) that may affect the

company’s RMI are controlled for.14_{To test hypothesis 1 and 2 multiple regressions are}

estimated to find the relationship between RMI and risks and the relationship between RMI and performance:

Tobin’s Qj,t= α + β ∗ RMIj,t-1 + γ ∗ Xj,t-1 +Year FE (2)

Annual Returnj,t= α + β ∗ RMIj,t-1 + γ ∗ Xj,t-1 +Year FE (3)

Tail Riskj,t= α + β ∗ RMIj,t-1 + γ ∗ Xj,t-1 +Year FE (4)

Altman Z-scorej,t= α + β ∗ RMIj,t-1 + γ ∗ Xj,t-1 +Year FE (5)

Tobin’s Q and Annual Return are used as dependent variables to test the relationship between RMI and performance (hypothesis 1). Tail Risk and Altman Z-score are used as dependent variables to test the relationship between RMI and risk (hypothesis 2). Ellul and Yerramilli (2013), argued that risk management affects future performance. So if a board acts today it is prepared for future risks and therefore performs better. Therefore a lagged RMI is used in the regressions.

Additional to the equations above, a dummy variable is used to test the differences between hypothesis 1a and 1b and the difference between hypothesis 2a and 2b. So separating the RMIs of equations (2) to (5) in a crisis RMI and a pre-crisis RMI to identify whether the risk and performance results are different before and during the crisis. The equation (2) to (5) are estimated as follows:

Tobin’s Qj,t = α + β1 ∗ RMIj,t-1 * Crisis Year+ β2 ∗ RMIj,t-1 * (1-Crisis Year) (6)

+ γ ∗ Xj,t-1 +Year FE

Annual Returnj,t = α + β1 ∗ RMIj,t-1 * Crisis Year+ β2 ∗ RMIj,t-1 * (1-Crisis Year) (7)

Tail Riskj,t = α + β1 ∗ RMIj,t-1 * Crisis Year+ β2 ∗ RMIj,t-1 * (1-Crisis Year) (8)

Altman Z-scorej,t = α + β1 ∗ RMIj,t-1 * Crisis Year+ β2 ∗ RMIj,t-1 * (1-Crisis Year) (9)

(32)

32

To test whether a company performs better with a risk-taking CEO and a strong risk management index an additional RMI-2 is created. To test hypothesis 3 all of the equations above are also tested with this additional RMI-2. The differences between the RMI sets are explained in chapter 4.1.2.

(33)

33

5 Results

This chapter discusses all the estimated regressions. Also the hypotheses are either proved or rejected.

5.1 Summary statistics

Over a period of 7 years data is collected to create a risk management index to evaluate the results of effective risk management. Table II shows a summary of the variables that are used to estimate the different RMI sets. Over the years boards changed their composition a lot. The lowest number of board of directors was six and the maximum was twenty-one members. However over the years the average (and medium) was stable around twelve board members. The number of board meetings increased slightly before the onset of the Financial Crisis but did not change significantly. The most remarkable observation is the number of times the audit committee met. While boards do not seem more active, the meetings of the audit committee increased from 500 in 2002 to 700 in 2006 and 2007. The CEO’s Delta shrunk in 2008 due to the fall in stock prices while the exercise prices stayed the same and the CEO’s Vega did not change much over the years.

Table II statistics variables RMI

The RMI variables do not indicate much variance across the years. As described before those variables are used to create three different RMI-sets. Separating board meetings from risk-taking incentives of the CEO. Table III shows the results of a univariate analysis between the companies with high RMI score to those with a low RMI score.

Year # Board Members # Board Meetings # Audit Board Meetings Committee Meetings Delta CEO Vega CEO 2002 845 589 543 961 0.7178 19.0049 2003 855 595 634 1157 0.6995 23.4649 2004 844 589 653 1123 0.7173 22.7928 2005 850 602 693 1158 0.7176 23.9194 2006 859 649 723 1188 0.7221 21.7616 2007 864 588 727 1230 0.6819 20.0402 2008 845 610 678 1196 0.4863 19.5398

Stronger risk management, lower risk : evidence from the top 100 S&P 500 non-financial companies