The Effect of ERM implementation on
CEO compensation, risk information
disclosure and shareholder
contribution
Nicole A. Marlin
Studentnumber: 1688537
University of Groningen
(n.a.marlin@student.rug.nl)
Master Thesis Accountancy
University Supervisor
Dr. B.Qin
July 2011
Abstract
There have been many practical studies but not theoretical studies on the effect of enterprise risk management (ERM) implementation. The objective of this study is to compare the after effects for companies that have implemented an ERM system with regard to CEO compensation, risk information disclosure and shareholder contribution. Using a sample of 112 ERM firms that have implemented ERM in the year 2005, this study provides evidence that ERM implementation does affect CEO compensation and the amount of risk information disclosure in annual reports. However, with regard to ERM contributing to shareholder value no significant evidence was found.
Keywords: Enterprise risk management, CEO compensation, risk information disclosure
1 1. Introduction
In the last years Enterprise Risk Management (ERM) has become very popular especially since the economic crisis. A research carried out by Ernst & Young (2009) over the economic crisis shows that in various organizations the problem was caused by the failure of the risk management system. ERM is the process of managing risk on a corporate level and is defined by COSO (2004) as “a process, effected by an entity’s board of directors, management and other personnel, applied in strategy setting and across the enterprise, designed to identify potential events that may affect the entity, and manage risk to be within its risk appetite, to provide reasonable assurance regarding the achievement of entity objectives.”
Risk management decisions are made by the manager of the organization (Rogers, 2002). This leads us to the other major topic that has been criticized namely, executive compensation. For this reason it is of grand importance to integrate compensation committees in the risk management process to protect the stakeholder from risks related to executive compensation (Lynn, 2011).
The “agency theory” (Jensen and Meckling, 1976) defines the problems caused by the separation between the principal (stakeholders/owners) and the agent (manager). The risk-averse differences between the agent and the principal may lead to actions by the agent that is not always in the best interest of the principal. A good example here is the infamous Enron scandal with evidence of excessive risk taking activity by executives. Researchers (Frydman and Jenter, 2010) provide evidence that the manner in which CEOs are compensated affects the performance of the organization this is the target of effective contracting. Even though incentive contracting is meant to align interest of both parties research has proven that they do not always succeed in their goal. Therefore, the economic crisis and scandals such as Enron and Worldcom have led to controversy on executive compensation. These scandals also led to investors losing trust in the data disclosed by organizations. The CEO of an organization has an immense amount of influence on the information disclosed publicly.
The composition of CEO compensation is used as an incentive and affects the manner in which the CEO operates a firm. This composition is of grand importance for investors. The optimal contract can be seen as one that aligns the incentives of the CEO with those of the investors. One that diminishes the agency problem (Jensen
2
and Meckling, 1976) caused by the separation between the principal and the agent. However, inefficient compensation contracts can lead to additional organisation risks, by causing CEO to commit actions contrary to the purpose of incentive compensation contracts which is interest alignment. Adequate CEO compensation contracts is therefore one of the risks to be controlled by an organization. The above mentioned scandals and crisis has led to greater emphasis on risk management with regard to organisational- and compensation risks.
Turning back to ERM, the main goal of ERM is to add value. According to Pagach and Warr (2007) another goal of ERM is to control the risks associated with CEO compensation. Knowing how important the CEO is to ERM this study would like to focus on the relationship between the two.
There have been various studies (Droogsma, 2009; Beasley et al., 2005) on the importance of ERM and these studies are mainly related to the implementation of ERM. Prior studies (Beasley et al., 2005; Desender, 2007; Droogsma, 2009) have provided evidence that the CEO support is of importance to ERM. The CEO is actually the one ultimately responsible for enterprise risk management (COSO 2004). Studies on ERM has become of grand necessity due to the fact that it is a new growing issue. As stated above the CEO plays a crucial role in ERM. However, researches such as Droogsma (2009) found evidence that in practice the CEOs are not as involved as they should be in ERM implementation. Some organizations have implemented ERM while others have not, that has been the main focus of previous researches. The need for research on ERM is essential.
In this paper I would for this reason like to research the association between the two variables CEO compensation and ERM. To my knowledge a study specifically focusing on the relation between the two in a variety of ways does not exist. However, evidence was found by Pagach and Warr (2011) of a positive correlation between the likely hood of ERM implementation and the risk taking incentives of CEO. The explanation given is that firms might be adopting ERM to offset the risk taking incentives that they give to CEOs. Pagach and Warr (2011) focused on how risk taking incentives of CEO can lead to ERM implementation my study will extend to their study by focuses on how after a firm has implemented ERM how this affects the CEO compensation. The main focus of this paper is to study the effect of ERM on CEO compensation.
3
Moreover, I also focus on two other important aspects that are influences by ERM implementation. These are; risk information disclosure in annual reports and the main goal of ERM which is value creation for shareholders. Organisations cannot just implement an ERM system and expect investors to be assured that risks are well managed; this must also be done by effective communication (Beretta and Bozzolan, 2004). The manner through which organisations communicate with investors is through the annual report. In this paper instead of focusing on disclosure as a whole the focus will be on risk information disclosure in annual reports of firms that have implemented ERM due to the fact that ERM is a risk management system. Lastly this paper focuses on the main goal of ERM which is namely to add value to a firm. Two studies focused on this aspect and found contradicting results. The study that provided results used the Tobin’s Q as measure. In this research the following value measures will be used; return on assets (ROA), return on equity (ROE), economic value added (EVA) and stock return (SR).
This research will add to the growing necessary research on ERM and will provide information on the effects of ERM implementation with regard to three aspects. Therefore, the main research question is as follows;
- Does implementing an ERM system affect the level of CEO compensation, the
risk information disclosed in annual reports or add value to shareholders contribution?
In order to answer the main research question I have formulated three sub-questions. These sub-questions focus on various aspects of an organization that could be affected after ERM implementation.
The sub- questions are as follows;
- Does the implementation of ERM in organizations affect the level of CEO
compensation with regard to fixed- and incentive compensation?
- Does ERM implementation influence the amount of risk information disclosed in
annual reports?
4
This study contributes to research as follows; there have been various researches (Beasley et al., 2005; Desender, 2007; Droogsma, 2009) on factors influencing risk management of an organization whereby CEO has been seen as an important factor that has influence on the risk management. There have also been many studies on the influence of CEO compensation on the actions and decisions made by CEOs (Hanlon et al., 2004; Rogers, 2002). However, the relationship between the CEO compensation and ERM has not yet been researched to my knowledge. With regard to the last two aspects, effective communication and value creation are essential for a firms’ continuity.
My study mainly focuses on the effects of ERM implementation not the factors leading to ERM implementation. Therefore my study compare the variables prior to and after ERM implementation. Using a sample of 112 ERM firms from the article of Gordon et al. (2009) that have implemented ERM in the year 2005, this study provides evidence that ERM implementation does influence the level of CEO compensation and the amount of risk information disclosure in annual reports. However, with regard to ERM adding value to a firm no significant evidence was found.
The rest of this paper will be organized as follows. The next section will contain a literature review of prior researches relating to the above mentioned aspects; CEO compensation, risk information disclosure and contribution to shareholders value. In that section the hypotheses will also be formulated. In the third section the research method will be presented. The formulated hypotheses will then be tested and the results will be presented in the fourth section. In the last section the discussion of the results, conclusions, limitation of the study and suggestions for future research will be given.
2 Literature review and Hypothesis development
In this section I will be presenting prior research and formulating hypotheses in association with the effects of ERM implementation with regard to CEO compensation, risk information disclosure in annual reports and contribution to shareholders value. The prior research and findings presented will be used to formulate the hypotheses.
5
2.1 CEO and Enterprise Risk Management (ERM)
Many Researches (Beasley et al., 2005; Desender, 2007) mainly focused on the determinants of factors influencing why certain firms embrace ERM and why others do not. Leadership is an important factor, not only in the implementation of ERM but also in the quality of ERM (Droogsma, 2009). Desender (2007) researched the relation between board composition and the degree of ERM and found that the highest level of ERM is found in organizations with an independent board and where the CEO is separated from the chairman. Beasley et al. (2005) who also studied factors associated with the degree of ERM in organizations also found that a more independent board and mainly the tone at the top coming from both CEO and CFO is critical to ERM deployment other factors also associated to extent of ERM implementation found are size, auditor type, industry and country of domicile. Not only does literature provide evidence of the importance of the CEO with regard to ERM but studies of ERM in practice also show the importance of CEO with regard to ERM. A research report provided by The Conference Board (2006) studied the role of U.S. corporate boards in ERM in which evidence was provided that directors are unanimous in the belief that the overall responsibility for the ERM should be with the CEO. Even though in theory and practice it is expected that the CEO is mainly involved with ERM in the study of Droogsma (2009) results are presented from an interview held with one hundred respondents working in companies with a turnover of 100 million to five billion and shows that only in 4% of the cases the CEO is the head sponsor of the ERM. Droogsma (2009) is astonished of the results and suggests further research for the reason of the lack of involvement of the CEO in ERM.
The hiring of a CRO is used by quite a lot research (Hoyt and Liebenberg, 2009; Pagach and Warr, 2010) as a proxy for ERM adoption. Pagach and Warr (2011) provide evidence that when the CEO of firms have incentives mainly based on options compensation these firms are also more likely to hire a CRO. They found a positive significant Vega coefficient, therefore as CEO compensation become more sensitive to stock volatility it is more likely for that firm to implement ERM. According to them it is the board that makes the decision to implement ERM and does this to not only encourage risk taking from the CEO but to also implement a program that can manage coordinate and understand the risks. Another explanation given is the fact that ERM is implemented to protect cash flows that come from risky growth options.
6
As a result not only does the CEO affect the risk management system of an organization but CEO compensation self is a risk that needs to be managed by the organization. Pagach and Warr (2007) also found a positive correlation between the likelihood of adoption of ERM and an increase in risk taking incentives. In order to diminish the risks associated with CEO compensation with regard to the alignment of the interest between the principal and the agent, organizations make compensation contracts consisting mainly of the following components; base salary, pension, short term incentives (bonus) and long term incentives (options). It is of critical importance to know what influence the composition of executive compensation has and the best manner in which the above mentioned risk related to incentive compensation can be mitigated. Efficient CEOs contracts are made by composing the contracts out of the following; a fixed salary, short term incentive and long term incentive. Short term incentives are mainly bonuses and long term incentives are either stock options or the holding of firm shares. The compensation policy of a firm is an important factor in the success of an organization (Jensen and Murphy, 1990). Jensen and Murphy (1990) provide evidence that the biggest problem is the composition of the compensation contract. They found that CEO incentives generated by stock ownership had declined over the past years. Another problem mentioned by Jensen and Murphy (1990) is the fact that CEOs are awarded through stock options with the target of aligning the interest of the CEO with the shareholders however CEOs tend to sell their options which causes the target to be undone.
The goal of incentive compensation is a positive one in which the shareholders and managers interest are aligned however there are risks that come with incentives compensation such as “pump and dump”, caused when contracts are not optimal and can lead to actions by the managers not in alignment with the interests of the shareholders (Hall, 2003). CEO contracts should for this reason include manners in which the risks of mitigating the incentives can be reduced. Hall (2003) gives the following six challenges when it comes to optimal contracting; mismatched time horizon, gaming, the value-cost wedge, leverage fragility trade off, aligning risk taking incentives and avoiding excessive compensation. In this paper the one of importance is aligning risk taking incentives.
Prior studies (Rajgopal and Shevlin, 2002; Rogers, 2002) provide evidence that risk taking incentives is related to risk management behaviour of organizations. Rajgopal and Shevlin (2002) found that Employee Stock Options (ESO) provide
7
managers with incentives to mitigate risk-related incentive problems. In contrast, Hanlon et al. (2004) found evidence that CEOs stock options are statistically associated with greater risk-taking behaviour, e.g. investing in riskier projects.
A manner in which compensation risk can be controlled is the use of restricted stock that restricts the CEO of disposing of their stock prior to the due date. Even though restricted stock option can be seen as a better motivator then unrestricted stock option the latter is more popular, the possible reason for this fact is that issuing restricted stock has always required expensing and that unrestricted stock since 2004 (Scott, 2009).
Researchers (Beasley et al., 2005; Desender, 2007; Droogsma, 2009) have provided evidence that the CEO support is of importance to ERM. The CEO is the one ultimately responsible for ERM (COSO 2004). Therefore, composition of CEO compensation and the manner in which organizations control these risks are indirectly of importance for an organizations risk management system. According to Nocco and Stulz (2006) and Blanchard and Dionne (2003) the main goal of risk management is to maximize the firms share values, however according to Blanchard and Dionne (2003) another objective can be maximization of the well-being of executives. When executives are compensated with stock options the two objectives can be in conflict with each other (Blanchard and Dionne, 2003).
It is expected that if boards implement ERM to control the risk incentives from CEOs the composition of CEOs compensation will differ after ERM implementation. For this reason I will specifically like to study the relation between ERM implementation and CEO compensation. Pagach and Warr (2007) found evidence that firms are more likely to adopt ERM when the CEOs option and stock portfolio is increasing in stock volatility. Therefore a reason for ERM implementation is seen as a control method to the decisions made by CEO due to their risk incentives. For this reason I expect that ERM implementation will influence the level of CEO compensation with regard to the fixed- and incentive compensation. According to Pagach and Warr (2007) firms with CEOs that are compensated with high levels of option based will benefit from ERM implementation if it leads to a reduction in managers incentives (e.g. compensation) to adopt very risky projects. Another manner in which CEOs will benefit from ERM implementation is due to the fact that ERM should lead to an increase in firm value. If the value of the firm increases so will
8
the compensation of the CEO because of the fact that CEO compensation (stock options) is linked to firm value. Therefore, it is expected that ERM implementation will lead to a difference in the level of CEO compensation.
The first hypothesis to be tested is as follows;
HYPOTHESIS 1: Implementation of ERM leads to significant changes in the level of CEO compensation with regard to fixed- and incentive compensation.
2.2 Risk information disclosure
The manner in which companies inform investors of their risks and how they manage these risks is primarily through annual reports. Many studies (Beretta and Bozzolan, 2004; Linsley and Shrives, 2006) focused on risk information disclosure in annual reports. The studies on disclosure levels includes relating disclosure level to size (Linsley and Shrives, 2006; Zarzeski, 1996), legal system (Archambault and Archambault, 2003; Hope, 2003), culture (Zarzeski, 1996) and many other aspects. According to economic theory increasing disclosure should lower the cost of capital by lowering the information asymmetry component of it (Diamond and Verrecchia, 1991). Therefore, risk information disclosure is of grand importance.
The importance of an appropriate risk management is related to risk information disclosure. It may not be in an organizations best interest to fully inform investors of how they manage their critical risks. However, risk management disclosure in annual reports is an important device in informing investors of the risks that organizations face and the manner in which they are managing these risks. Investors need to be assured that risks are managed well, not only by implementing a risk management system but also by effective communication (Beretta and Bozzolan, 2004). A study by PWC (2006) with regard to the benefits of ERM shows that companies with ERM in place are paying more attention to external risk reporting and that a percentage of 80% of the respondents include risk reporting in their annual report.
The CEO of an organization has a major influence on the information disclosed and plays an important role in risk reporting (Abraham and Cox, 2007). There are several markets involved in private incentives for managers to produce information; managerial labour market, capital markets and takeover market (Scott, 2009). For
9
example, CEOs influence investors’ expectations by making voluntary disclosures in which positive information is delayed and negative information is publicised around the time of stock option awards in order to maximize their stock option compensation (Aboody and Kasznik, 2000; Brockman et al., 2010). Hermalin and Weisbach (2011) who studied the relation between information disclosure and corporate governance found evidence that greater disclosure tends to raise executive compensation this because of the fact that the preferences between investors and management with regard to disclosure are opposing therefore more mandatory disclosure with regard to the quantity or quality of disclosure should lead to higher management compensation. Additionally according to them disclosure is seen as a two-edged sword and can lead to both less or additional agency problems. This lies in the fact that the principal (investor) will always like to be better informed and the information may benefit them while harming the agent (manager). Therefore the agent has to be compensated for this extra risk. This risk includes the possibility that predictions made by the managers do not come through and this causes damage to their reputation. By providing more information managers can be easily monitored thus doing the contrary limits the monitoring ability of the capital and labour market (Shleifer and Vishny, 1989). A positive association was also found by Nagar et al. (2003) between firm disclosure and CEO compensation related to stock options and the value of shares held by CEO. According to them stock compensation provide managers with an incentive to improve price informativeness through disclosure. When prices are very informative about managerial actions managers then make sure that the information given of the prices are very clear this then improves the view that capital and labour market have with regard to the managers and can lead to a compensation increase. Dobler (2008) describes risk reporting as being dependent on disclosure incentives. He found that risk reporting is not as common as research suggest and finds that it depends on the information that the company possess and on decisions made by the manager to disclose or not disclose the information, which are in turn influenced by the incentives of the manager.
This study extends to prior ERM researches by namely focusing on the relation between ERM implementation and the amount of risk information disclosure in annual reports. ERM implementation by an organization is a manner in which they can better manage their risks. I expect that this will also lead to more risk
10
information in their annual reports. Investors might see the implementation of ERM by an organization as a manner in which the organization is portraying that it is effectively managing their risks. However, as mentioned by Beretta and Bozzolan (2004) organizations can not only implement a risk management system but must effectively communicate to the investor. I expect that organizations that implement ERM as a manner to manage risks will also provide more risk information in their annual report and will not only rely on the fact that they have implemented ERM. This hypothesis therefore focuses on ERM implementation and risk information disclosure in annual reports and predicts that ERM implementation will increase risk information disclosure levels.
The second hypothesis to be tested is formulated as follows;
HYPOTHESIS 2: Firms that have implemented ERM tend to disclose more risk information in their annual report.
2.3 ERM and shareholder contribution
ERM as defined by Emanuels et al. (2006) is “a system that enables management to identify, prioritise, analyse and control the relevant risks which may be threatening the organisation’s ability to meet its objectives”. There is an immense rise in research about ERM. According to theory (COSO, 2004) the main goal of ERM is to create value. An overview is given in the article of Nocco and Stulz (2006) of the manner in which ERM creates value. According to them ERM creates value through first enabling firms to both rationalize and quantify the risk they face while getting access to the necessary resources and taking the risks that create value. Secondly, all risks of material value and their risk-return trade off in association with individual risks are internalized.
Many firms are now implementing ERM and it is becoming of grand importance. Firms are implementing ERM because it is expected to bring benefits. The main benefit of ERM according to theory is to enhance the capacity of firms to build value (COSO, 2004). The manner in which ERM does this is by dealing with the risks and opportunities faced by firms, by identifying the potential events that may affect the entity (COSO, 2004). A study carried out by PWC (2006) with regard to firms that have implemented ERM questioned the benefits of ERM. The result shows
11
that the top two scoring categories according to firms with regard to the benefits of ERM implementation are; better control of uncertainty and loss avoidance. Other measurable variables that have proven to be beneficial due to ERM implementation are; lower volatility of earnings and contribution to improved shareholder value.
Even though there has not been many research on the effects of ERM firm value and variables relating to firm performance two research provided by Hoyt and Liebenberg (2003) and Pagach and Warr (2010) show contradicting results with regard to the effects of ERM on firm value. Hoyt and Liebenberg (2009) found evidence that ERM implementation enhances firm value. They compared firms that have implemented ERM to firms that have not implemented ERM and found that firms with ERM implemented are larger, less leveraged have lower return volatility and have less financial slack. Their study focused specifically on the insurance industry and they found evidence that insurers that have ERM implemented are valued roughly 20% higher than insurance firms that have not implemented ERM. Their research was the first to document the value relevance of ERM. Unlike this research the study of Pagach and Warr (2010) who also studied the effect of ERM implementation on value creation found little evidence to support the proposition that ERM is value creating. However, there are various differences between the two studies; e.g. Pagach and Warr (2010) did not focus on a specific industry as did Hoyt and Liebenberg (2009). Another difference is the fact that the results presented by Pagach and Warr (2010) as mentioned by them can be due to the lower power test used in their study. One similarity with regard to results between the two studies is that they both found some evidence that firms that have implemented ERM tend to have a reduction in earnings volatility.
In my paper I will also like to focus on the association with ERM and firm value. As mentioned above practical studies show that firms relate ERM to improved shareholder benefits. I would therefore like to test this empirically to see if there is a relation between the two variables. Even though there has not been much research on the value of ERM the studies mentioned above that studied this gave contradicting results. Value creation as mentioned is the primary goal of implementing an ERM system. According to Lam (2003) ERM can help an organization maximize shareholder value. I expect therefore that ERM implementation should lead to value creation.
12
The third and final hypothesis to be tested is formulated as follows;
HYPOTHESIS 3: Implementation of ERM leads to an increase in the value of shareholder contribution.
3. Method Section
3.1 Research Method
In my research I use a sample of 112 firms, and 448 firm year observations for the years 2003 through 2007 excluding the year 2005. The year 2005 is excluded due to the fact that this is the year in which the firms implement an ERM system. Most of the data used in this research are collected by using various databases. All variables used in my research will be further discussed below. I analyse the data using the SPSS program. The goal of this study is to research the affect implementing an ERM system has on the following dependent variables: CEO compensation, risk information disclosure in annual reports and contribution to shareholder value. This will be tested by using the paired samples t-test and Wilcoxon Signed Ranks (WSR)-test to determine the influence on the various variables after the implementation of an ERM system in an organization.
In order to use the t-test the variables must meet two criteria’s they must be continuous and they must be normally distributed. All the variables to be tested with the t-test are continuous. However, with regard to normally distribution I first test all the variable using the “test of normality” in SPSS, variables that are already normally distributed will be analysed using t-test without transformation. The variables that are not normally distributed will be transformed to their natural logarithm (LN) and then tested again for normal distribution prior to using the t-test. My study focuses on prior to and after ERM implementation therefore the dependent t-test is used, which assumes equal variances.
When using the WSR-test the variables do not have to meet any criteria’s in this test I will test the variables at their natural state. The results of both tests will be compared. Just as Pagach and Warr (2010) the averages of the variables two years prior to ERM implementation (2003-2004) are taken and are then compared to the averages of the two years after ERM implementation (2006-2007). Because the t-test and WSR-test alone lack power I also run regression analysis using various control variables. The models to be tested in this research are as follows;
13
(1) CEOcomp= β Xi + δ ERM + e (2) RIDI= β Xii + δ ERM+ e
(3) CSV = β Xiii + δ ERM + e
The dependent variables in the above models are; CEO compensation (CEOcomp), risk information disclosure index which represents the total risk information disclosure in annual reports (RIDI) and contribution to shareholders value (CSV), respectively.
In the various models ERM is a dichotomous variable and indicates the status of ERM implementation in an organization (0=prior ERM implementation, 1=after ERM implementation). The X’s in the various models represents the group of control variables that according to prior research is known to influence the dependent variables and are explained below. All models estimate the effect of ERM implementation on the various dependent variables.
3.2 Data Collection
As mentioned above the primary objective of this study is to examine the effect of firms implementing ERM. In this research data collected from the Execucomp database will be used, this database contains information on executive compensation. Other databases used include the “Thomson One Banker” and “Wharton Research Data Services (WRSD)” which contains information on companies and characteristics of companies in various countries. Information not available in the various databases will be hand-collected through the analysis of annual reports. For the analysis of the amount of risk information disclosure I manually analyse firm’s annual reports.
ERM firms
The sample that will be used consists of the 112 firms used in the article of Gordon et al. (2009). All these firms have implemented ERM in the year 2005. The manner in which Gordon et al. (2009) selected these firms are as follows; they used Chief Risk Officer (CRO) or risk management committee or other words such as “enterprise risk management and corporate risk management” as a proxy for organizations that have implemented ERM this manner was also used in other research (Hoyt and Liebenberg, 2009; Pagach and Warr, 2010). Gordon et al. (2009)
14
first identified 273 US firms as having implemented ERM in 2005 among which some firms were eliminated due to missing data. The final sample consists of 112 firms identified as having implemented ERM in 2005 and this sample will be used in this paper. An overview of the list of firms is presented in Appendix A. All these firms are vested in the United States.
CEO Compensation
Information with regard to CEO stock compensation is taken from Execucomp. With regard to CEO compensation I collect the following data; The total compensation (TC), the total current compensation (TCC) which includes the salary & bonus, the value of the restricted stock holdings (RSH), the value of options granted as reported measured by Black-Scholes (OPTGRCBLCK), the value of options granted as reported by the company (OPTGRC), and lastly the salary and bonus are collected separately. Therefore the RSH gives the total amount of restricted stock held by the CEO for the given year. An overview of the description of the variables is presented in Appendix A. The reason I take both measures with regard to options granted is because firms are permitted to calculate options following either the Black-Scholes value or the simpler rule of thumb where the options future values is reported under the 5% or 10% assumption of annual stock price appreciation (Yermack, 1998). As described in the data definition of Execucomp database the Black-Scholes value from the options granted indicates the total value of all options received in that year this figure also includes the volatility of the individual company. Companies usually calculate both values and report the lowest to the SEC which is most likely the Black-Scholes value (Yermack, 1998). The difference between the two are as follows; even though companies might use Black-Scholes to measure the options granted there are many discount and other ways in which they can influence the Black-Scholes formula. Thus, the Black-Scholes value is the formula used by Execucomp database therefore it is not biased by the companies individually: and the value as reported by the company is the specific measures the company uses to calculate the value, which measures are unknown.
Risk information disclosure
For the measure of risk information disclosure in annual reports content analysis is used. This method has been also used in various articles such as Abraham
15
and Cox (2007) and Linsley and Shrives (2006). In performing content analysis, number of words, page proportions and sentences can be used. In this article I measure amount of risk information disclosure by using count of number of words in 304 annual reports of the ERM firms for the years 2003-2007 excluding 2005. Data for all four years was collected for 76 out of the 112 firms. The words used have been taken from the article of Abraham and Cox (2007) and is presented in Appendix B. The last five words in the list presented under “risk management” were added by me. I added these words by analysing annual reports and there information on risk management these were words that stood out under risk management information. The words added are; hedge, mitigate, impact, manages and risk management. Appendix B also gives an overview of the total count of words in all the annual reports of ERM firms for the years included in this study.
Shareholder contribution value measures
In the article of Hoyt and Liebenberg (2009) the value measure of Tobin’s Q was used, I would further like to add to this research by now focusing on other value measures with regard to ERM. According to theory (Stern, 1990) EVA is described as the performance measure that is the closest to the creation of shareholders wealth. The manner in which EVA analyses shareholder creation is by focusing on the following five key factors; net operating profit after tax and before financing cost, the weighted average cost of capital, investment in the business, the rate of return on investment and the competitive advantage period (Putnam, 1997). Brummer and Hall (1999) who studied the relationship between market value and internal performance of a company found evidence that EVA has the highest positive correlation with market value added (MVA). They also found slightly lower correlations with other variables such as ROA, ROE, EPS and DPS. In this study four value measures will be used; EVA, ROA, ROE and Stock Return (SR).
Control variables
All control variables used in this paper was derived from prior research. All control variables used in the three equations will be discussed. As mentioned above with regard to CEO compensation various CEO compensation components will be tested. I begin with the discussion of all the control variables used in the equations with regard to CEO compensation. The exact control variables used to test the CEO
16
compensation component will be presented in the table with the results of the regression. The control variables used with regard to CEO compensation are; size, return on assets (ROA), leverage, return on equity (ROE), net income, total compensation (TC), stock return (SR), dividend pay-out and total current compensation (TCC). With regard to the size, studies (Chourou et al., 2008) provide evidence that CEO in larger firms is compensated higher then CEO in smaller firms. This is due to the fact that these firms have more growth opportunities and are more difficult to monitor therefore the managers need more incentives. The measure for size will be the natural logarithm of sales. Second, financial performance measures are known as an important factor in determining CEO pay especially the incentive compensation components. ROA is one of the most commonly used profitability measure (Bhattachereejee et al., 2006) and for this reason will be used as a control variable in this equation. ROA is positively related to CEO pay. Other financial performance variables included as control variables are ROE and net income. With regard to leverage, evidence was provided by prior studies that the financial leverage (Chourou et al., 2008) tends to have a negative effect on stock compensation. According to Chourou et al. (2008) highly leveraged firms make less use of stock options to mitigate the stock- bond holder conflict. Last, research (Cadman et al., 2009) also provide evidence that stock return influences CEO compensation and will also be included as a control variable.
Now I will discuss the control variables used in the second equation. In the second equation with regard to risk information disclosure I use two control variables; firm size and leverage. (Linsley and Shrives, 2006) provide evidence that larger firms tend disclose more risk information in their annual report. With regard to the relation between leverage and risk information two studies (Zarzeski, 1996; Gray et al., 1995) provide evidence and found evidence that disclosure decreases with leverage.
Last I turn to the control variables used in the third and final equation with regard to contribution to shareholder value. The control variables used in the last equation are; size, sales growth, net income, dividend pay-out, ROA and ROE. Studies (Lang and Stulz, 1994) provide evidence that size and firm value are significantly negatively related. Brummer and Hall (1999) found evidence that ROA and ROE are also internal measures that are correlated to market value. Sales growth and dividend pay-out was also added due to the fact they was used as control
17
variables in the article of Hoyt and Liebenberg (2003) who also studied the influence of ERM on firm value but then with the variable Tobin’s Q.
An overview of all variables used in this paper and the source from which they were collected is presented in Appendix C. Also descriptive statistics of all the variables used in my research is presented in Appendix D. An overview is given of “N” the total valid cases in the sample for the specific variable. Further information is presented of the minimum, maximum, mean and standard deviation and the values at the various percentiles. There have not been many prior research on the effect of ERM. However, I compare my sample with the study of Liebenberg and Hoyt (2009) who studied the effect ERM has on firm value. Their means are presented in brackets size 8.020 (8.039) and sales growth 14.076 (13.791). They also used the performance measure ROA in their sample however, the mean for their ROA is much lower than the ROA for my sample 4.279 (0.015). This can be due to the fact that in my sample ROA is measured in percentage while prior studies used original value. With regard to RIDI just as in the article of Abraham and Cox (2007) I used content analysis. Even though the sample used in my article is larger, comparing the frequency of the words with the 2003 annual reports due to the fact that the sample in the article of Abraham and Cox (2007) was for the year 2002 I find that the count of words are within a 30% range. With regard to CEO compensation some of the variables are compared with the study of Griffith and Najand (2006) who studies the effect of CEO compensation on firm performance; salary 7.70 (6.05) Bonus 9.48 (6.19) TCC 11.75 (6.8) OPTGRCBLCK 14.31 (6.10). With regard to the CEO compensation variables I use natural logarithm of the variables to control for a too big range and outliers.
A correlation matrix of the variables is presented in Appendix E. The correlations between the dependent variable and the explanatory variable ERM are as follows; the highest correlation between ERM and CEO pay variables are between the variable salary and ERM (0.266) significant at the p-value 0.05. The correlation between ERM and RIDI is 0.196 also significant at the p-value 0.05. Last with regard to ERM and the shareholder contribution value variables the highest correlation is the negative correlation between ERM and stock return (-0.150) significant at the p-value 0.05. With regard to the other independent variables used in the models the moderately high correlations in the matrix are between the following independent and dependent variable; size and Black-Scholes value of options granted (0.638)
18
significant at p-value 0.05. With regard to the independent variables mutually there are no high correlations. None of the correlations exceed the value of 0.85 therefore I conclude that multicollinearity should not be a problem in my regression analysis (Schroeder, 1990).
4. Results
4.1 Hypotheses testing
In my research I use the paired sampled t-test, Wilcoxon signed rank (WSR) test and regression analysis to examine the effects of ERM implementation on the following variables of interests; CEO compensation, risk information disclosure in annual reports and contribution to shareholder wealth. Prior to testing I transform the variables relating to CEO compensation and risk information disclosure to their natural logarithm (LN) in order to control for a too big range and outliers.
The hypotheses that I will be testing are as follows;
H1: Implementation of ERM leads to significant changes in the level of CEO
compensation with regard to fixed- and incentive compensation.
H2: Firms that have implemented ERM tend to disclose more risk information
in their annual report.
H3: Implementation of ERM leads to an increase in the value of shareholder contribution
The results of the various hypotheses with regard to the various variables are presented separately in the following paragraphs. First the results with regard to the
t-test and the WSR-test are presented, and then I present the results with regard to
the regression analysis. At the end of each paragraph I give my conclusion with regard to the formulated hypothesis. In the final section of this paper the results will be discussed and I will give my concluding remarks.
4.1.1 ERM implementation and CEO compensation
The results with regard to the t-test are presented in Appendix F1. With regard to CEO compensation variables tested in the t-test, four variables present significant results. Three of which are presented as being significantly positively related to ERM implementation and these variables are: restricted stock holding (RSH) (T=1.754,
19
p<.01), total compensation (TC) (T=4.153, p<.1) and salary (T=1.860, p<.1). The total
current compensation (TCC) gave significantly negative results in relation to ERM implementation. The WRS-test gave similar results with regard to the variables, three of the four variables present significant results in the same direction as the t-test; RSH (Z=-4.373, p<.01), TC (Z=-3.609, p<.01) and TCC (Z=-3.813, p<.01). The difference between the two tests lies in the following two variables; salary and bonus. The WRS-test found no significant results with regard to salary however; with regard to bonus the WRS-test provide significantly negative results (Z=5.274, p<.01) with regard to ERM implementation. See the notes of Appendix F2 for interpretations with regard to the results of WRS-test.
We then turn to the regression results with regard to Equation (1) which are presented in Table 1. The coefficients of determination (adjusted R-squared) which gives the proportion of variables in the data that can be explained by the independent variables in the models with regard to CEO compensation vary from 0.014 to 0.302. The significant F-values vary from 2.829 to 10.991. With regard to CEO compensation the results vary among the various CEO compensation variables used. The CEO compensation variables that resulted in significant results with regard to ERM implementation are as follows; total current compensation (TCC), the options granted as reported by the company (OPTGRC) and the variable SALARY. Their coefficient of determination is presented in brackets behind the variable TCC (0.051), OPTGRC (0.361) and SALARY (0.279). Results provide evidence that ERM implementation leads to a significant increase in the salary the result for the interaction was positively significant (β=1.653, p<.05). With regard to the other two significant CEO compensations components; OPTGRC is negatively significant (β=-5.226, p<.05) and TCC is also negatively significant at (β=-.883, p<.10) the results show that ERM implementation leads to a decrease in these components. I must mention that the F-value for the model with regard to TCC is not significant therefore the results with regard to this model cannot be interpreted with confidence. With regard to the other models for OPTGRC and SALARY the F-values are significant these models can be interpreted with confidence and the results will be further discussed in the discussion section. The dependent variables are log transformed therefore the interpretation is that the dependent variables changes by 100% for a unit increase in the independent variable. The coefficient estimates that the salary after ERM implementation increases with 165%. With regard to OPTGRC the
20
coefficient estimates a decrease of 525%. When observing the control variables two provide significant positive results consistent with prior research (Bhattachereejee et al. 2006; Chourou et al. 2008); ROA and size.
The results of all three tests will now be compared. In two of the tests evidence was provided that ERM implementation leads to a decrease in the total current compensation (TCC) of a CEO. With regard to the SALARY and RSH of the CEO two of the three tests provide evidence that ERM implementation leads to an increase in the CEOs salary and RSH. However, as mentioned above TCC consists of salary and bonus of a CEO. The only test presenting results with regard to the bonus was the WRS-test and this test provided significant results that ERM implementation is significantly negatively related to CEO bonus. Other variables that were only significant in one of the tests were; OPTGRC.
With regard to fixed- and incentive compensation my results show that ERM implementation leads to an increase in the fixed compensation of CEOs (SALARY). However, when focusing on incentive compensation the RSH which is an incentive compensation component increases after ERM is implemented and OPTGRC the other incentive component decreases. Last the TCC which consists of a fixed and an incentive component decreases after ERM is implemented. These results will be further discussed in the following section.
Even though significant results was not found in relation to all the CEO compensation variables, I accept my first hypothesis which stated that implementation of ERM leads to significant changes in the level of CEO compensation with regard to fixed- and incentive compensation.
21 Table 1
OLS estimation of the relationship between ERM implementation and CEO compensation
Independent Variables
TC TCC RSH OPTGRBLKSCH OPTGRREPCO Salary Bonus
ERM 0.233 [0.368] -.883* [0.451] 0.476 [0.494] -.376 [1.095] -5.226** [2.042] 1.653** [0.630] -0.90 [0.632] Size 0.140 [0.177] 0.079 [0.222] 0.792*** [0.177] 0.485*** [0.160] 0.130 [0.525] 0.855*** [0.267] 0.485* [0.249] Leverage -0.220 [0.189] 0.170 [0.246] -0.043 [0.048] -0.005 [0.086] 0.138 [0.225] 0.106** [0.048] ROA 0.027 [0.032] 0.033 [0.041] 0.189*** [0.042] 0.090* [0.050] 0.383** [0.038] ROE 0.008 [0.006] 0.005 [0.008] -0.015* [0.009] -0.007 [o.o17] -0.79* [0.045] LnTC .352*** [0.090] LnTCC -0.085* [0.047] 0.086 [0.139] SR -1.539E-5 [0.000] -1.133E-6 [0.000] 5.554E-5 [0.000] NetInc 6.678E-5 [0.000] 8.497E-5 [0.000] 0.000 [0.000] -9.1711E-5 [0.000] 0.000 [0.000] Salesgrowth -0.003 [0.013] R-squared 0.055 0.051 0.267 0.402 0.361 0.279 0.054 Adjusted R-squared 0.014 0.022 0.243 0.302 0.234 0.248 0.023 F-value 1.335 1.786 10.991*** 4.033*** 2.829** 8.896*** 1.70
22
#observations 193 205 187 56 48 96 127
***, **, * Coefficient is significant at 1%, 5% en 10%
Notes: Of all the dependent variables the natural logarithm are taken. ERM is dummy variable equal to 0 prior to ERM implementation and 1 after ERM has been implemented. Size is equal to the natural logarithm of the total assets. Leverage is measured as debt/total assets. TC is the natural logarithm of the total compensation. TCC is the total current compensation. OPTGRBLKSCH the total options granted measured in the Black Scholes value. OPTGRREPCO is the natural logarithm of the total options granted as reported by the company.SR is stock return. NetInc is the net income for the year. Salesgrowth is the one year sales growth. ROA is the return on assets measured in percentage. ROE is the return on equity. Standard errors appear in brackets.
4.1.2 ERM implementation and risk information disclosure
The results with regard to the second hypothesis which is formulated as follows: firms that have implemented ERM tend to report more risk information in their annual reports is now discussed. I first turn to the results of the t-test and the WRS-test presented in Appendix F. In both tests (T =4.981, Z=-4.917) the RIDI which is the measure for risk information disclosure in annual reports and which is also log-transformed show a significantly positive relationship between the amount of risk information disclosure in annual reports prior to and after ERM has been implemented with a p-value<0.01. The results of the regression analysis are presented in Table 2 in the first column. The adjusted R-squared for this model is 0.195 and the F-value is 10.281 and is significant at the p-value<0.01. The results of the regression analysis also suggests that ERM is significantly positively associated with the amount of risk information disclosed in annual reports (β=0.213, p<0.01). Most importantly the coefficient is positive at 0.213 which shows an increase of 21.3% in risk information disclosure. With regard to the control variables used in this equation only the size variable shows a significant positive association consistent with the notion that larger firms disclose more.
All tests give positive significant results with regard to the amount of risk information disclosure in annual reports and ERM implementation. Therefore I accept my second hypothesis which stated that: ERM implementation leads to more risk information disclosure in annual reports.
23 Table 2
OLS estimation of relationship between ERM implementation and risk information disclosure and value of shareholder contribution
Independent Variables
RIDI EVA ROA ROE Stock
Return ERM 0.213*** [0.082] -251.392 [274.084] 0.029 [0.792] -2.420 [2.847] -20.386** [9.347] Size 0.138*** [0.030] -160.973 [110.494] 1.315*** [0.234] 5.270*** [0.886] 8.733** [3.511] Leverage -0.005 [0.008] -0.236* [0.149] -0.284 [0.539] -1.218 [1.570] ROE -1.634 [10.039] ROA 0.862*** [0.212] 0.199 [0.933] Divpay 10.026* [5.250] 0.000 [0.015] -0.064 [0.053] -0.123 [0.174] Net Income 0.057 [0.061] Sales growth -8.254 [5.952] -0.030* [0.016] -0.284 [0.056] 0.929*** [0.220] R-squared 0.394 0.031 0.129 0.218 0.128 Adjusted R-squared 0.195 0.009 0.114 0.202 0.105 F-value 10.281*** 1.422 5.537*** 13.348*** 5.576*** #Observations 171 271 256 293 233 ***, **, * Coefficient is significant at 1%, 5% en 10%
Notes: Dependent variables; RIDI is the index used to measure the total risk information disclosure in annual reports. EVA, ROA, ROE and SR are all measures used to measure value of shareholder contribution. Independent variables; ERM is dummy variable equal to 0 prior to ERM implementation and 1 after ERM has been implemented. Size is equal to the natural logarithm of the total assets. Leverage is measured as debt/total assets. ROE is the return on equity. ROA is the return on assets measure in percentage. DivPay is
24
the total dividend pay-out. NetInc is the net income for the year. Salesgrowth is the one year sales growth. Standard errors appear in brackets.
4.1.3 ERM implementation and shareholder contribution
My final hypothesis focuses on the effect of ERM implementation on value of shareholder contribution. The results of the t-test are insignificant for all the variables except stock return. The t-value for stock return is -2.785 and is significant at p-value 0.01; this shows that after ERM is implemented there is a decrease in stock return. With regard to the WSR-test one variable gives significant results the ROA which provides significant results (Z=-2.096, p<.05) with a negative sign which shows that ERM leads to an increase in shareholder contribution measured by ROA. With regard to ROE the sign is also negative but not significant. However, with regard to EVA the sign of the WRS-test is positive which states that ERM implementation leads to a decrease in the value of contribution to shareholders, measured by EVA but the results are also not significant therefore no statements can be made.
The results of the regression analysis with regard to this final hypothesis also provide no significant results for the first three measure variables of shareholder contribution; EVA, ROA and ROE. With regard to the variable EVA the F-value (1.422) is insignificant. The other F-values for the variables ROA (5.537), ROE (13.348) and Stock return (5.576) are all significant at p-value 0.01. However, as mentioned above the results with regard to the effect of ERM on shareholder contribution measured by ROA and ROE are both insignificant. The only variable that shows significant results with regard to ERM is stock return (β=-20.39, p<.05). The coefficient is negative and estimates a decline of 20 in stock return after ERM implementation.
With regard to the effect of ERM on shareholder contribution the results of the three tests are mostly insignificant. The only variable that produced significant results in two tests is stock return and these results shows a decrease in stock return after ERM is implemented. Therefore on the basis of stock return as a measure for shareholder contribution, I reject my final hypothesis which stated that ERM leads to an increase in shareholder contribution.
25 5 Conclusion & Discussion
In this section I will interpret and further discuss the results presented in the prior section and give my concluding remarks. Due to the fact that I am dealing with log-transformed data plus a dichotomous independent variable (ERM) the results are interpreted differently with regard to equation one and two.
The first equation pertained to the effect of ERM implementation on CEO compensation. With regard to CEO compensation I focused on seven variables, four of which gave significant results; TCC, SALARY, BONUS, RSH and OPTGRC. The results suggest that ERM implementation does have a significant effect on some CEO compensation with regard to fixed -and incentive compensation.
First the results with regard to the TCC will be discussed. The TCC consists of both a fixed (salary) and an incentive compensation component (bonus). The results of all test show that ERM implementation leads to a decrease in the TCC. Now turning to the two components separately that the TCC is made up of salary & bonus the results (two tests) provide evidence that ERM leads to an increase with regard to salary and a decrease (one test) with regard to bonus. Here you can conclude that ERM leads to an increase in a CEOs fixed compensation and a decrease with regard to incentive compensation.
Moreover, there are two other incentive compensation variables that are also effected by ERM; the restricted stock held (RSH) by the CEO and the options granted as reported by the company (OPGRC). A comparison between RSH and OPGRC is not optimal due to the fact that one is “flow” compensation (OPGRC) while the other is a “stock” compensation (RSH). Therefore the effect on both will be explained separately. Even though only one test (regression analysis) provide evidence that ERM leads to a decrease in OPGRC this result holds due to the fact that this test have power. The other two tests did not give significant with regard to this variable because in these test the averages prior and after ERM implementation were compared, but with regard to this variables there is a significant lack of options granted for the years 2006-2007 which are the years used in my research to test the effect “after ERM”. This is in line with Frydman and Jenter (2010) that states that CEO pay declined from 2000 to 2008 and restricted stock grants has replaced option grants. According to Hall and Murphy (2003) the most pronounced change in CEO compensation in the last decade has been the escalation and decline of executive stock options. The big decrease (525%) in options grants in this study leads me to
26
conclude that another external force may be the cause of this grand decrease in options.
One possible cause is the revisited standard by the FASB in December 2004 that mandates that stock options be recognized in the income statement as an expense. Prior to this most companies disclosed the stock options expense in the footnotes rather than the income statement. This may be the primary cause of such a decline in options grants. (Frederickson et al., 2006)
With regard to RSH the results (two tests) provided evidence that after ERM implementation there is more RSH held. Paying executives with restricted stock tend to increase their risk aversion (Walker, 2009).
In conclusion with regard to fixed- and incentive compensation we get the following. After ERM is implemented the fixed pay increases and with regard to incentive pay we get an increase in the RSH and a decrease in options granted and bonus. According to theory (Core and Guay, 1999, 2002; Coles et al., 2006) stock option compensation relative to stock will induce managers to more likely to choose risky projects. Namely, stock options can increase executive risk-taking appetites (Walker, 2009). This is in line with the statement of Pagach and Warr (2011) that firms may be implementing risks to offset the risk-incentives from CEOs. And that firms in which CEOs compensation are mainly based on stock options are more likely to hire a CRO, which is known as a proxy for ERM implementation.
Therefore, the results with regard to ERM implementation and CEO compensation suggest that the board may indeed be implementing ERM not only to manage risks but then as a way to control the risks with regard to CEO compensation, by increasing the amount of non-incentive related compensation component SALARY and diminishing the more incentive related compensation component BONUS and OPGRC and increases the incentive component RSH. The two incentive components that are diminished after ERM is implemented are both component that can be easily influenced by the CEO and most probably increases the executive risk-taking appetites. However, one incentive compensation component increases after ERM is implemented that is the RSH that cannot be easily influenced by the CEO self and is known to diminish the risk-appetite of executives.
Second I turn to the results with regard to ERM implementation and risk information disclosure in annual reports. As mentioned by Beretta and Bozzolan
27
(2004) firms cannot just implement a risk management system and expect that this is sufficient for investors. My results provide evidence that after firms have implemented ERM they significantly report more risk information in their annual report. This may be due to the fact that the system helps them to better manage their risks and therefore they can indeed provide investors with more information with regard to the risks and the manner in which they will be managing these risks in their annual report. This result is also cohesive with the practical study by PWC (2006) that showed that firms that have implemented ERM are paying more attention to external reporting.
Last, with regard to the effect that ERM implementation has on shareholder contribution one measure provides significant results, stock return. These results are not in line with my final hypothesis which states that ERM leads to an increase with regard to contribution to shareholder value. However, these results are somewhat in line with the findings of Pagach and Warr (2010) who also found little evidence that ERM leads to an increase in firm value. A possible explanation is that ERM takes time to add value. My study focused on the two years prior to and after ERM implementation this period might be too short to really study the value creation due to ERM implementation. With regard to the other measures only ROA provide significant results in one test (WRS-test). However, this test alone does not have sufficient power in order to make interpretations based on these results alone.
In summary, the results indicate that ERM implementation does have an effect on the level of CEO compensation and risk information disclosure in annual reports.
5.1 Limitations
Limitations of my research include the fact that the sample size used in the t-test and WRS-t-test vary among the variables due to the fact that for some variables there were more data available than for others in the various databases. Another limitation is the scope of the study which is the two years prior to and after ERM implementation. Some effects with regard to ERM implementation may take longer to influence the company then the first two years after an ERM system is implemented. Effects of ERM implementation may also vary across various industries. In this research due to the fact that the largest industry among the
28
companies consisted of a sample of fifteen no tests were done with regard to industry specific differences.
5.2 Future research
Future research with regard to ERM is of grand importance. ERM implementation is not only costly but it is said to lead to many benefits to firms and it is necessity to firms. More research on ERM can lead to better understanding of the best manner in which firms can implement ERM by comparing the ERM benefits. This can be done by comparing firms in various industries or by just comparing prior and after effect of ERM implementation as was done in this study. With regard to risk information in annual reports evidence was found that firms that have implemented ERM disclose more risk information in their annual reports. However, more risk information does not necessarily mean better quality risk information. Future research can also focus on the effect that ERM implementation has on the quality of the risk information produced in annual reports.
Finally, the results presented in my research should be generalized with caution. With regard to risk information disclosure the word count of content analysis was used which is a good measure but it is not the best. Also with regard to analyzing the data used in this research three tests were used and the results were then compared not all the results were significant in all three tests. However, all the tests provided cohesive directions with regard to a positive or negative affect.
This research has broadened my interest with regard to the effects that implementing an enterprise risk management (ERM) system may have on an organization. It has also enhanced my knowledge with regard to CEO compensation and risk information disclosure in firms that have implemented an ERM system. In closing, I hope that my research paper can be of assistance to future researchers interested in researching subjects with regard to ERM implementation.
