Determinants and Consequences of Risk Management Committee Formation

Transcription

1 University of Arkansas, Fayetteville Theses and Dissertations Determinants and Consequences of Risk Management Committee Formation Chris Hines University of Arkansas, Fayetteville Follow this and additional works at: Part of the Accounting Commons, Business Administration, Management, and Operations Commons, and the Finance and Financial Management Commons Recommended Citation Hines, Chris, "Determinants and Consequences of Risk Management Committee Formation" (2012). Theses and Dissertations This Dissertation is brought to you for free and open access by It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of For more information, please contact

2 DETERMINANTS AND CONSEQUENCES OF RISK MANAGEMENT COMMITTEE FORMATION

3 DETERMINANTS AND CONSEQUENCES OF RISK MANAGEMENT COMMITTEE FORMATION A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Business Administration By Chris S. Hines Missouri State University Bachelor of Science in Accounting, 1994 Missouri State University Master of Business Administration, 2001 August 2012 University of Arkansas

4 ABSTRACT This paper examines the determinants and consequences of financial institutions voluntarily forming risk management committees (RMCs). Specifically, I determine whether RMCs are related to a change in risk outcomes, an increase in profitability, a change in hedging and trading derivative structures, and greater financial reporting quality during the post-committee formation period compared to a control group. I use a sample of financial institutions that form a RMC in any year from 1994 through 2008 and a control sample of financial institutions that do not form a RMC during the sample period. The results provide evidence to suggest that financial institutions with a higher level of risk, a Chief Risk Officer (CRO), lower financial reporting quality, a larger board, a board that has a greater proportion of independent members, international banking activity, merger and acquisition activity, a CEO as Chairman of the Board, and a Big N auditor, are more likely to form a RMC. There is some evidence to suggest that RMC formation is negatively associated with a change in risk only for financial institutions with a high-level of risk in the year prior to RMC formation. RMCs and CROs act as substitutes (not complements) when it comes to reducing non-performing assets and loan charge-offs in high risk firms. Additionally, the results provide some evidence to suggest that RMC formation is associated with an increase in the notional values of both hedging and trading derivatives. Interestingly, for firms with a high notional value of trading derivatives, a CRO serves as a mitigating force to reduce the notional value of trading derivatives. Moreover, for firms with a low level of profitability, RMC formation is not associated with a change in profitability. However, the presence of a CRO is related to an increase in profitability. The presence of both a RMC formation and a CRO is associated with a decrease in profitability. Therefore, RMCs and CROs do not act as substitutes r complements when it comes to increasing profitability in firms with a low level of profitability.

5 This dissertation is approved for recommendation to the Graduate Council. Dissertation Director: Dr. Gary F. Peters Dissertation Committee: Dr. James Nelson Myers Dr. Cory A. Cassell Dr. Timothy John Yeager

6 DISSERTATION DUPLICATION RELEASE I hereby authorize the University of Arkansas Libraries to duplicate this dissertation when needed for research and/or scholarship. Agreed Chris S. Hines Refused Chris S. Hines

7 ACKNOWLEDGMENTS I thank the members of my dissertation committee, Cory Cassell, James Myers, Gary Peters (Chair), and Timothy Yeager for helpful comments and suggestions. I also thank Linda Myers for helpful comments and support. In addition, I thank University of Arkansas seminar and workshop participants for helpful comments.

8 DEDICATION This dissertation is dedicated to my wife (Deanna), my son (Jared), and my daughter (Julia).

9 TABLE OF CONTENTS 1. Introduction 1 2. Background and hypothesis development Risk committee background Determinants of risk management committee formation Committee formation and governance structures Committee formation and risk management practices Risk management committees and the effect on risk outcomes Risk management committees and the effect on risk hedging activities Risk management committees and profitability Risk management committees and financial reporting quality Sample Empirical models Determinants of risk management committee formation Consequences of risk management committee formation Hypothesis 1 and 2 models Hypothesis 3 and 4 models Hypothesis 5 model Hypothesis 6 model Results Descriptive statistics and correlations Probit regression: determinants of risk management committee formation Multivariate analyses: risk management committee formation and changes in risk outcomes Multivariate analyses: risk management committee formation and changes in notional values of derivatives Multivariate analyses: risk management committee formation and changes in profitability Multivariate analyses: risk management committee formation and loan-loss provision validity Additional analyses Risk management committee formation and changes in risk within firm analysis of treatment firms Risk management committee formation and changes in notional values of derivatives within firm analysis of treatment firms Risk management committee formation and changes in profitability within firm analysis of treatment firms Risk management committee formation and changes in risk, changes in derivatives, and changes in profitability - within firm analysis of treatment firms Risk management committee structural characteristics Conclusion 47 References 49

10 1. Introduction Risk management committees (RMCs) are important corporate governance mechanisms (Moore and Brauneis 2008; Schlich and Prybylski 2009). An important unresolved question is why does a board of directors voluntarily form a RMC while other boards do not form a RMC? Prior literature does not address whether the board of directors should designate a stand-alone RMC to oversee the process of identifying and managing critical risks. The answer is dependent upon the effectiveness of stand-alone RMCs. To address these issues, I examine the determinants and consequences of financial institutions voluntarily forming a RMC (i.e., the first incidence of disclosure indicating the presence of a RMC). Specifically, I first address factors that determine whether financial institutions voluntarily form a RMC. I then determine whether RMCs are related to changes in risk outcomes, an increase in profitability, a change in hedging and trading derivative structures, and greater financial reporting quality during the post-committee formation period. Kaplan (2011) suggests that the governance practice of risk management, in general, is an important issue for accounting researchers to address. He indicates that risk management intersects several accounting research areas such as financial reporting, disclosure, auditing, and corporate governance. He charges accounting academics to develop research questions which determine when risk management processes are effective and when they are ineffective. Moreover, Carcello et al. (2011) bring to light the fact that very little governance research investigates the interactions between the board and its committees, or the interactions among board committees. They call for research on interactions among board committees and management and examining certain outcomes such as risk management practices. The research questions in this paper are an attempt to answer these calls. 1

11 In light of the recent financial crisis and current economic environment, there is ongoing debate among regulators and practitioners about what type of governance role related to risk the board of directors should have. In part, this debate includes whether the board of directors should designate a stand-alone RMC to oversee the process of identifying and managing critical risks. The current New York Stock Exchange (NYSE) listing standards explicitly require the audit committee, or a different board committee, to have risk oversight responsibility and to discuss policies related to governing risk assessment and risk management (Krus and Orowitz 2009). Since the financial crisis, governance reform has resulted in audit committees focusing more attention on risk management oversight that is beyond their traditional role of monitoring the risk of material financial misstatements. According to a 2010 survey of executives by the American Institute of Certified Public Accountants and North Carolina State University, over 60 percent of audit committees now have some risk oversight responsibility, and more than one-third of these audit committees focus on strategic risks (Johnson 2010a). However, this is troubling to many practitioners who believe that overall risk oversight should not reside solely with the audit committee (e.g., Johnson 2010b; Keizer 2010). In addition, some organizations and regulators recommend the use of stand-alone RMCs (Keizer 2010). Firms with a stand-alone RMC are not common, and those firms that have this committee tend to be in highly regulated industries (Bates and Leclerc 2009). The benefits of forming a RMC have not been addressed in previous research. Although, the ways in which firms address the governance practice of risk management is an increasingly important professional and regulatory issue. In response to the increased focus on corporate accountability, the Securities and Exchange Commission (SEC) has expanded proxy disclosure rules that affect risk oversight, corporate governance, and executive compensation matters. The recent SEC proxy disclosure rule No. 33-2

12 9089, effective February 28, 2010, gives firms flexibility in how the board oversees risk. For example, firms can disclose that they administer risk oversight through the entire board, through the audit committee, or through a separate committee such as a RMC. Financial institutions represent an important risk management group because they are exposed to certain types of risks (e.g., liquidity risk, credit risk, regulatory and compliance risk) that are more critical in their industry than in non-regulated industries. Financial firms, by their nature, take on specific risks as part of their primary business plan and they must effectively manage those risks in order to be profitable (Eggleston and Ware 2009). According to a 2010 survey conducted by Bank Director and Grant Thornton LLP, risk management is such a priority for financial institutions that approximately one-third of them have formed a RMC (Keenan 2010). I compare the changes in risk, the changes in hedging and trading derivatives, and the changes in profitability from the period prior to RMC formation to the period subsequent to RMC formation for financial institutions that voluntarily form a RMC to those that do not form a RMC. I also compare the level of financial reporting quality in the post RMC formation period for financial institutions that voluntarily form a RMC to those that do not form a RMC. To do this, I use a sample of financial institutions that form a RMC in any year from 1994 through 2008 and a control sample of financial institutions, matched on the year of formation, two-digit SIC code, and size (i.e., total assets), that did not form a RMC during the sample period. I find that riskier financial institutions and financial institutions that have a Chief Risk Officer (CRO) are more likely to form a RMC. Also, the results provide evidence to suggest that financial institutions with lower financial reporting quality, a larger board, a board that has a greater proportion of independent members, international banking activity, merger and acquisition activity, a CEO as Chairman of the Board, and a Big N auditor, are more likely to form a RMC. 3

13 Overall, with respect to risk outcomes, there is some evidence to suggest that RMC formation is positively associated with an absolute change in loan charge-offs around the year of risk management committee formation but is not associated with an absolute change in nonperforming assets. Additionally, the results suggest that risk management committee formation is negatively associated with a change in risk only for financial institutions with a high-level of risk in the year prior to RMC formation. RMCs and CROs act as substitutes (not complements) when it comes to reducing non-performing assets and loan charge-offs in high risk firms. The results provide some evidence to suggest that RMC formation is associated with an increase in the notional values of both hedging and trading derivatives. Interestingly, for firms with a high notional value of trading derivatives, a CRO serves as a mitigating force to reduce the notional value of trading derivatives. Moreover, there is evidence to suggest that for firms with a low level of profitability, RMC formation is not associated with a change in profitability. However, the presence of a CRO is related to an increase in profitability. The presence of both a RMC formation and a CRO is associated with a decrease in profitability. Therefore, RMCs and CROs are not substitutes and they are also not complementary when it comes to increasing profitability in firms with a low level of profitability. Additionally, I find that RMC formation is not associated with financial reporting quality during the post-committee formation period. My study contributes to research on corporate governance (e.g., effectiveness of governance, board structure, etc.). This study also has practical and regulatory implications in light of the SEC s recent regulatory agenda related to risk management practices. Prior accounting research has paid little attention to the determinants and consequences of specific governance structures related to risk management. Research that investigates determinants and value relevance 4

14 of broader aspects of Enterprise Risk Management (ERM) programs is beginning to emerge (e.g., Baxter et al. 2010), 1 but prior research has not examined the effectiveness of specific governance structures related to risk management such as RMCs. Furthermore, the results of my study could have broader implications for understanding the benefits of RMC formation for not only regulated industries, but for non-regulated industries as well. My findings inform both practitioners and regulators about certain characteristics of risk governance reform that may have implications related to reducing risk, changing notional values of derivatives, and improving profitability in financial firms. My focus on financial institutions is also relevant to the post financial crisis debate over governance regulation related to risk management. Recent regulatory proposals focus directly on financial institutions and their attempt to shape governance and risk management practices in the aftermath of the financial crisis. Because of this governance regulation, examining the determinants and consequences of RMC formation is important. The remainder of the paper is structured as follows. In section two, I summarize relevant background literature and develop the testable hypotheses. Section three addresses the sample selection. Section four discusses the empirical models including the procedures used for analyzing the data. The results are provided in section five and section six concludes. 1 Baxter et al. (2010) examine the determinants and value relevance of relative ERM program quality for a sample of financial firms and find benefits of higher ERM quality reflected by firm and market measures. 5

15 2. Background and hypothesis development 2.1 Risk committee background There is current debate among regulators and practitioners regarding whether the best governance solution related to risk management is for firms to form a stand-alone RMC, as opposed to keeping risk management practices under the sole authority of the audit committee. The NYSE s Listed Company Manual requires the audit committee to address risk assessment and risk management by discussing primary financial risks and the monitoring role that management has in place to control risk exposures (Bates and Leclerc 2009). However, a stand-alone committee can be charged with overseeing the risk management function as long as the audit committee reviews this oversight function. Similarly, as part of a review of corporate governance in UK financial institutions, Integrated Governance Solutions (IGS) 2 recommends that the board of a bank should establish a board risk committee separate from the audit committee with responsibility for oversight and advice to the board on the current risk exposures of the entity and future risk strategy (IGS 2009, p. 5). Consistent with the above regulatory guidance, since the financial crisis there has been a shift in the audit committee s focus to overall risk oversight responsibility. For example, Johnson (2010a) highlights recent changes in priorities for audit committees in response to events such as the sudden demise of Lehman Brothers and the strained commercial paper market. In general, this involves more discretion in audit committee agenda setting and more probing questions asked of Chief Financial Officers (CFOs). CFOs indicate that audit committees spend more time analyzing 2 Based on 25 years of corporate governance experience, the founder established IGS to transform how firms are governed. The organization s mission is to help create, restore, and sustain trust through next generation board oversight and integrated risk management ( IGS reviews its solutions with several key interested stakeholders such as The Council of Institutional Investors, an SEC commissioner, nationally recognized business schools, etc. 6

16 the types of cash investments firms are allowed to make, scrutinizing routine cash management transactions, and monitoring cash investment safety (Johnson 2010a). Moreover, Ernst & Young s 2008 Global Internal Audit Survey reveals that audit committees are now increasingly seeking internal audit s participation in discussions regarding process improvement recommendations, emerging risks and trends and risks in the business (including insights around risk appetite and tolerance levels) (Ernst & Young 2008, p. 5). A potential outcome of RMC formation may be an increased risk oversight emphasis, and a relief of some of the risk oversight responsibilities borne by the audit committee. For example, in response to this shift in audit committee focus, J. Michael Cook, a former chairman and CEO of Deloitte & Touche who has served on various audit committees, states there has been some confusion about the role of audit committees that is sorting itself out. The audit committee s reason for existing is to address one very significant enterprise risk: that you will issue inaccurate, or misleading, or fraudulent financial statements (Johnson 2010b, p. 1). Moreover, other governance experts believe that the best governance solution related to risk management may be to form a stand-alone RMC. Henry Keizer, the Global Head of Audit for KPMG International, offers suggestions to help boards put RMC formation into context and states a risk committee may help provide a more focused oversight of risk, and relieve the audit committee of some of its responsibilities for risk (Keizer 2010, p. 2). Keizer (2010) also describes the current trend in risk committee formation and suggests that a number of financial institutions have formed a RMC comprised of members with industry knowledge. Regulators are responding to the call for governance reform due to governance failures responsible for the financial meltdown. In August 2010, the Dodd Frank Wall Street Reform and Consumer Protection Act (the Dodd-Frank Act ) was enacted and requires financial institutions 7

17 with more than $10 billion in total assets and certain financial institutions that report to the Federal Reserve Bank to form a stand-alone RMC (Nixon Peabody Corporate Responsibility Alert, August 11, 2010). A primary role of RMCs is to review and evaluate risks that threaten firms existence. In general, a financial institution s RMC oversees the entire risk management framework which includes operational risk and credit risk. An example proxy statement disclosure of a financial institution s RMC role is provided in Appendix A. 2.2 Determinants of risk management committee formation Committee formation and governance structures Because prior research does not specifically address the determinants of RMC formation, I initially draw from early literature on the determinants of voluntary audit committee formation prior to any SEC mandate regarding audit committee formation. Prior accounting research investigates why some firms voluntarily form an audit committee when there is no regulatory mandate. Using a 1986 sample of NASDAQ firms, Pincus et al. (1989) examine incentives to form an audit committee and find that firms are more likely to voluntarily form an audit committee when managerial equity ownership is lower, leverage is higher, firm size is greater, the proportion of outside directors is greater, and they employ Big 8 auditors. Moreover, Eichenseher and Shields (1985) find that firms with newly hired auditors are more likely to voluntarily form an audit committee when the successor is a Big 8 auditor. In contrast, Bradbury (1990) uses a sample of firms listed on the New Zealand Stock Exchange and finds that managerial equity ownership, leverage, firm size, and the employment of Big 8 auditors are not associated with voluntary audit committee formation. However, he does find that a greater number of directors on the board and a 8

18 greater proportion of outside directors are positively associated with voluntary audit committee formation Committee formation and risk management practices Resource dependency theory relates to ways in which a board provides access to various resources (e.g., skills, information, constituents) in order to reduce uncertainty (Awotundun et al. 2011). According to resource dependency theory, the primary role of insiders on the board is to provide firm-specific information to the board. Hillman et al. (2000) refer to this as insider information provision. Hillman et al. (2000) also argue that given this key role of insiders on the board, insider decision-making will take on greater importance when firms are under regulation. Because the firms examined in this study are financial institutions (i.e., regulated firms), resource dependency theory provides a potential motivation for why these firms form a RMC. For example, one possibility is that boards form a separate RMC which consists of certain influential decisionmakers (insiders) who can provide necessary information regarding risk management practices. Related to the adoption of risk management mechanisms or practices, there is a stream of research that examines the determinants of ERM program adoption and ERM program quality although prior research does not specifically address the determinants of RMC formation. For example, Baxter et al. (2010) use Standard & Poor s ERM quality ratings of financial firms to examine potential determinants of ERM program quality. They find that financial firms that are larger, are more complex, are more liquid, are less mature, and have stronger internal controls over financial reporting have higher ERM program quality on average. In addition, Hoyt and Liebenberg (2009) study U.S. insurance firms and simultaneously model the determinants of ERM 9

19 adoption and the effect of ERM on firm value. They find that firms that are larger, with greater institutional ownership, and with lower leverage are more likely to adopt ERM. Beasley et al. (2005) provide a foundation for research in this area by examining factors associated with ranges of ERM adoption levels. Based on survey data from 123 firms, they find that the ERM adoption level is positively associated with the employment of a CRO, board independence, the presence of a Big 4 auditor, firm size, and firms in the banking, education, and insurance industries. Kleffner et al. (2003) examine a sample of Canadian firms and find that the determinants of ERM adoption include the influence of a risk manager, pressure from the board of directors, and compliance with Toronto Stock Exchange guidelines. Using a sample of firms that employ a CRO charged with managing the ERM program, Liebenberg and Hoyt (2003) find that firms with greater leverage are more likely to appoint a CRO, but also find that earnings and stock price volatility, the market-to-book ratio, and the level of institutional ownership are not associated with CRO appointments. More recently, Pagach and Warr (2010) also use the employment of a CRO as a proxy for ERM adoption and find that larger firms, firms with more operating cash flow volatility, and firms with greater institutional ownership are more likely to adopt ERM. Given that Pagach and Warr (2010) find that firms with more operating cash flow volatility (i.e., higher risk) are more likely to adopt ERM, I expect that riskier financial institutions will be more likely to form a RMC. In addition, given that Beasley et al. (2005) find that the ERM adoption level is positively associated with the employment of a CRO and Kleffner et al. (2003) find that the influence of a risk manager is a determinant of ERM adoption, I expect that financial institutions employing CROs are more likely to form a RMC. Since prior literature is limited in scope related to the determinants of RMC formation, I utilize an approach similar to Doyle et al. (2007). Their study is a first step in examining 10

20 determinants of internal control quality. Similarly, as a first step, I investigate firm characteristics that may be determinants of RMC formation. 2.3 Risk management committees and the effect on risk outcomes Despite the recent emphasis on stand-alone RMCs, several boards and investors question whether the audit committee is the optimal body to oversee a firm s risk management function (Bates and Leclerc 2009). Directly related to my financial institution setting, Moore and Brauneis (2008) suggest that risk management will play a significant role in restoring balance after the subprime mortgage crisis and that RMCs will exert more influence over business decision-making. They state that, risk committees of the board of directors at many financial institutions, alarmed by the reminder of the damage that can result from imprudent risk decisions, will increase scrutiny of risk activities (Moore and Brauneis 2008, p. 26). Schlich and Prybylski (2009) suggest that since the banking crisis in 2007, financial institutions are making substantial changes to their risk governance structures in order to more effectively manage critical risks. However, I argue that these changes to governance structures may be unnecessary, and even ineffective, if financial institutions already operate at an acceptable risk level prior to any risk governance changes. Because the benefits of a RMC have not been addressed in prior accounting research, determining whether a RMC affects firm risk remains an important empirical question. For example, practitioners stress the importance of considering RMC formation although they acknowledge it is difficult to determine the benefits a priori: Whatever the level of risk a company faces and however well it currently manages that risk, it would probably do well in this political, economic, and regulatory environment to at least ask the question whether a risk committee will benefit the company. The answer to that question, however, is often unknowable in advance (Eggleston and Ware 2009, p. 1). 11

21 In general, a financial institution s RMC oversees the entire risk management framework including operational risk and credit risk. Stand-alone RMCs are purported to improve risk management outcomes since they provide a focused coordination between finance and risk and make risk management a central priority throughout the entire organization. The impact of RMC formation can manifest itself in the form of potential benefits and costs. Nixon Peabody s Corporate Responsibility Alert suggests that a firm may form a RMC based on several considerations which include the level and complexity of operational risk as well as the firm s risk tolerance (Nixon Peabody Corporate Responsibility Alert, August 11, 2010). The benefits of RMC formation may include setting a risk management tone for the firm, obtaining expertise in managing and reducing operational risks, and facilitating communication at the board level regarding various critical risks. On the other hand, drawbacks to RMC formation may include an increase in demand for directors time, an increase in director compensation costs, and a potential duplication of duties between the RMC and audit committee (Nixon Peabody Corporate Responsibility Alert, August 11, 2010). In addition, some firms with low risk profiles may not have a need for a RMC if they already have sufficient risk oversight processes in place with the current board and audit committee structure. When RMCs scrutinize risk activities, their objective is to ensure that risk is within acceptable risk tolerance limits, not to eliminate all risk. Each decision that a RMC makes increases some risks while reducing others. I argue that an effective RMC will make decisions which place the firm closer to an acceptable range of risk tolerance. Therefore, if the likelihood of RMC formation increases when the level of risk is higher, then I predict that the level of risk will be lower in the period after committee formation than in the period prior to committee formation. The reason for this prediction is that if financial institutions with high risk levels tend to form 12

22 RMCs, then they will attempt to reduce risk in order to restore risk to an acceptable level. This argument also draws from prior literature which indicates that audit committees are more effective at preserving financial reporting quality, and that the market reacts favorably, when an individual with financial expertise is on the audit committee (e.g., Abbott et al. 2004; DeFond et al. 2005). The previous arguments lead to the following alternative hypotheses: H1: RMC formation is positively associated with absolute changes in risk from the period prior to formation to the period subsequent to formation. H2: RMC formation is related to a reduction in risk for high risk firms from the period prior to formation to the period subsequent to formation. 2.4 Risk management committees and the effect on risk hedging activities Prior literature recognizes that large U.S. firms have increasingly used different types of derivatives over the past two decades to reduce risk exposures related to changing interest rates, foreign exchange rates, and commodity prices. Tufano (1996) illustrates that one theory for why firms institute more extensive risk management activities is that managers use them to maximize shareholder wealth. Additionally, theory developed by Smith and Stulz (1985) suggests that risk management can increase the expected value of the firm by decreasing the likelihood of financial distress. Ahmed et al. (2006) indicate that financial institutions hold derivatives for trading on their own account (trading purposes) and to manage risk (hedging purposes). According to Ahmed et al. (2011), financial institutions hold hedging derivatives to decrease exposures to macroeconomic risk factors such as interest rate risk and foreign exchange rate risk. Therefore, the objective of holding hedging derivatives is to reduce systematic risk and default risk (Froot et al. 1993; Ahmed et al. 2011). Trading derivatives at financial institutions can be comprised of customer-related positions (i.e., positions that can increase or reduce risk), speculative positions (i.e., positions that 13

23 increase risk), and economic hedges (i.e., positions that decrease risk) (Ahmed et al. 2011). Therefore, the relation between RMC formation and changes in trading derivatives can be positive or negative depending on the net trading derivative positions. The previous arguments lead to the following alternative hypotheses: H3: RMC formation is positively associated with changes in the notional value of hedging derivatives from the period prior to formation to the period subsequent to formation. H4: RMC formation is associated with changes in the notional value of trading derivatives from the period prior to formation to the period subsequent to formation. 2.5 Risk management committees and profitability A substantial amount of research examines how corporate governance characteristics are related to firm performance. Gompers et al. (2003) examine the importance of the twenty-four variables in the Investor Responsibility Research Center (IRRC) database and find that firms with stronger corporate governance have higher firm value and higher profits during the 1990s. Bhagat and Bolton (2008) find that stronger corporate governance ratings, greater board member stock ownership, and a lack of CEO power are associated with higher return on assets (ROA). This finding is consistent with Core et al. (2006) who also find a positive relation between stronger corporate governance and ROA. In general, prior research documents a negative relation between board size and firm performance. Guest (2009) investigates the association between board size and firm performance for a sample of UK listed firms. He finds that board size is negatively associated with ROA. Adams and Mehran (2008) examine the relation between board size, board composition, and ROA for a sample of bank holding companies. They find that board size is not related to ROA, however the percentage of independent directors is negatively associated with ROA (although robustness 14

24 tests do not appear to hold). I extend this line of research by considering whether RMC formation is associated with a subsequent increase in profitability. Minton et al. (2009) find that the presence of a CRO is positively associated with firm stock performance during the financial crisis period ( ) for a sample of financial firms. Additionally, according to a recent Ernst & Young 2012 report, firms with more mature risk management practices financially outperform their competitors. Specifically, Ernst & Young (2012) find that these firms generate higher growth in revenues and earnings before interest, taxes, depreciation, and amortization (EBITDA). They suggest that firms accomplish this performance by strengthening risk governance and oversight for risk management at the board and executive levels, integrating risk and performance management, investing in risks that influence profitability, and improving key business process controls. In this study, a disclosed RMC formation signals a financial institution s board-level commitment to risk management. This leads to the next alternative hypothesis: H5: RMC formation is positively associated with changes in profitability from the period prior to formation to the period subsequent to formation. 2.6 Risk management committees and financial reporting quality Audit committee responsibilities are evolving with an increase in banking regulation. For example, audit committees are now reviewing management s calculation of the allowance for loan losses (Keenan 2010). This is in response to regulators emphasizing the importance of a financial institution s ability to estimate potential loan losses. Beasley et al. (2009) suggest that it is often difficult for audit committees to provide effective oversight, especially in large, complex firms. This ineffective oversight can be a problem because audit committees have a high level of responsibility over financial reporting quality and the audit process. Baxter et al. (2010) find that ERM quality is associated with better internal 15

25 controls over financial reporting. In addition, using a sample of financial institutions, Altamuro and Beatty (2010) find that the Federal Depository Insurance Corporation Improvement Act (FDICIA)-mandated internal control requirements lead to an increase in loan-loss provision validity along with a corresponding increase in the predictability of cash flows. Loan-loss provision validity increases as the association gets stronger between the loan-loss provision expense in the current period and the actual loan charge-offs in the subsequent period. Taken together, this recent accounting research forms the basis for my next prediction. The GAO (1991, 1994) identifies the loan-loss provision as a bank s largest accrual and the account most likely affected by internal control deficiencies. I argue that if RMC formation relieves the audit committee of some of its overall risk oversight responsibilities, then the audit committee will be better able to provide oversight over financial reporting quality. Moreover, if RMC formation reduces the amount of loan charge-offs (i.e., reduces risk), then the loan-loss provision should be more accurately estimated due to the reduction in credit risk uncertainty. Therefore, I expect there to be a stronger association between the loan-loss provision and subsequent period loan charge-offs for financial institutions that form a RMC. The previous arguments lead to the final alternative hypothesis: H6: RMC formation is positively associated with loan-loss provision validity. 16

26 3. Sample I identify financial institutions that voluntarily form a RMC for the first time from 1994 through 2008 by using a 10-K Wizard key word search. The two-digit SIC codes of financial institutions in my sample are 60 and 61. I use this sample period because the 10-K Wizard search tool does not contain filings prior to My initial sample is 210 financial institutions using risk committee and risk management committee as key word searches. This process is similar to (but more specific than) prior studies (e.g., Gordon et al. 2009; Hoyt and Liebenberg 2009) that identify risk management practices by using similar key search terms. I eliminate 74 observations that are not stand-alone RMCs and 45 observations that have unavailable data in Compustat and/or SEC filings. 3 The 74 observations I eliminate are comprised of 36 observations with the word committee within a three word proximity to the phrase risk management but do not represent stand-alone RMCs; 23 observations where the audit committee is named audit and risk management committee, for example, but are not stand-alone RMCs; 9 observations where a board director serves on another firm s risk management committee ; and 6 observations where there is no stand-alone RMC disclosed in the proxy statement or 10-K filing in year t+1. I also perform a search in the proxy and 10-K for all sample firms in year t-1 to verify that the year of RMC formation (year t) is the first year of formation. This process yields a final sample of 91 financial institutions that form a stand-alone RMC from 1994 through Eighty treatment firms have a two-digit SIC code of 60 and 11 treatment firms have a two-digit SIC code of 61. I provide a sample selection summary in Table 1. I select a control sample of financial institutions, matched on year of RMC formation, twodigit SIC code and size (total assets), which did not voluntarily form a RMC from 1994 through 3 I do not eliminate an observation from the final sample if there is available data for at least one of the three risk measures I use in the empirical models. 17

27 2008. To match the control sample of financial institutions on size, I select the financial institution that is closest to the sample financial institution s total assets. Since all of the control firms are in the financial industry, I accept control firms that have total assets within +/- 50% of the sample firm s total assets. In order to satisfy this constraint, two sample firms are matched at the one-digit SIC code level (e.g., a sample firm with a 61 SIC code is matched to a control firm with a 60 SIC code). In my final sample, 90.1% (82 of 91) of the control firms have total assets within +/- 30% of the sample firm s total assets. I also verify that all control firms do not form a RMC during the sample period. 4 I obtain all financial variable data from Compustat Bank Fundamentals Annual or from 10-K filings for any missing Compustat data, and obtain the governance variable data from the Corporate Library or from proxy statements (i.e., DEF 14 SEC filings) and 10-Ks for any missing data from the Corporate Library. 5 Finally, I collect hedging and trading derivative variable data from 10-K filings under the notes to the financial statements. 4 I verify that each control firm does not form a RMC by performing a search of the 10-K and proxy statement for year t (i.e., the year that the matched sample firm forms a RMC) and for Three control firms have unavailable data for the Board Size and Independence variables. I replace the missing data with the control sample Board Size and Independence variable means for these three control firms. 18

28 4. Empirical models 4.1 Determinants of risk management committee formation To investigate firm characteristics that may be determinants of RMC formation, I estimate the following probit regression model (time t represents the year that a financial institution forms a RMC in all regressions): RMC t = β 0 + β 1 NPA t-1 + β 2 Charge Offs t-1 + β 3 Capital t-1 + β 4 CRO t-1 + β 5 Hedging t-1 + β 6 Trading t-1 + β 7 Risk Adj. ROA t-1 + β 8 Foreign t-1 + β 9 CEO Chair t-1 + β 10 CEO Turnover t-1 + β 11 Leverage t-1 + β 12 Independence t-1 + β 13 Auditor Change t-1 + β 14 Board Size t-1 + β 15 BigN t-1 + β 16 Merger t-1 + β 17 Restatement t-1 + β 18 Size t-1 (1) Where the variables are described below and summarized in Table 2: Dependent Variable RMC is an indicator variable that captures whether a financial institution voluntarily forms a RMC during the sample period. RMC equals 1 if the financial institution voluntarily forms a risk management committee, 0 otherwise. Year t represents the year that a financial institution forms a RMC. Independent Variables I investigate firm characteristics and risk measures that may be determinants of RMC formation. The first two risk measures (non-performing assets and loan charge-offs) are standard for financial institutions in prior accounting and banking literature (e.g., Altamuro and Beatty 2010; Beatty et al. 2002; Ng et al. 2010; Rosen 2005). NPA equals the total non-performing assets, scaled by lagged total assets. Charge Offs equals the extent of loan charge-offs, net of recoveries, scaled by lagged total assets. Following prior research (e.g., Fang 2010; Hodder et al. 2002), I also use a risk-adjusted capital ratio (the ratio of Tier 1 and Tier 2 capital scaled by risk-adjusted assets) as a third risk measure. Capital equals the ratio of Tier 1 plus Tier 2 capital scaled by risk-adjusted assets. Higher levels of NPA and Charge Offs represent higher risk and higher levels of Capital 19

29 represent lower risk, therefore I multiply Capital by negative one in my analysis so that sign predictions for all three risk measures are consistent. Based upon key word searches, I also identify whether the firm has a Chief Risk Officer or equivalent position. I utilize CRO to capture whether the financial institution has an officer dedicated to risk management leadership in the year prior to RMC formation. CRO equals 1 if the financial institution has a Chief Risk Officer as of the end of the fiscal year, 0 otherwise. As mentioned earlier Beasley et al. (2005) and Kleffner et al. (2003) find that enterprise risk management practices are associated with the presence of a CRO. Therefore, I predict a positive association between the employment of a CRO and the likelihood of RMC formation. In Equation (1), I also examine other variables from prior accounting literature related to the determinants of the voluntary adoption of similar director committees, as well as the determinants of enterprise risk management (ERM) program adoption. To capture the firm s proclivity to engage in risk-taking transactions, I also measure the extent of the firm s use of derivative instruments. Ahmed et al. (2006) indicate that financial institutions hold derivative financial instruments for trading purposes (trading products for their own accounts, or speculative risk-taking trading) and hedging purposes (to offset risk exposures). Hedging equals the total notional amount of hedging derivatives scaled by total assets. Trading equals the total notional amount of trading derivatives scaled by total assets. I do not make a sign prediction for the coefficients since derivatives consist of positions that can increase or decrease risk and may reflect the firm s overall need or desire to employ substitutionary or complementary risk management mechanisms. Baxter et al. (2010) find that ERM quality is positively associated with firm performance. I predict that poorer performing financial institutions are more likely to form a RMC in order to 20

30 improve risk decisions and performance. Therefore, I include Risk Adj. ROA t-1 as an explanatory variable and expect a negative sign on the coefficient. Risk Adj. ROA equals the ratio of net income before extraordinary items scaled by lagged risk-adjusted assets. Foreign equals 1 if the financial institution conducts international banking (i.e., has a nonzero foreign currency translation [FCA]), 0 otherwise. Financial institutions that engage in foreign operations are exposed to greater foreign exchange rate risk. Baxter et al. (2010) find that firms engaging in foreign operations are more likely to have higher ERM quality. Therefore, I include Foreign as a potential determinant of RMC formation and I expect a positive sign on the coefficient for Foreign t-1. While Baxter et al. (2010) find only marginally significant positive associations between the CEO as Chairman of the Board and ERM quality, I predict that a CEO s power over the board will influence the decision to form a RMC. CEO Chair equals 1 if the Chief Executive Officer is the Chairman of the Board, 0 otherwise. I predict a positive sign on the coefficient for CEO Chair. I also include CEO Turnover as an explanatory variable, which captures the structural changes in the board s governance structure. CEO Turnover equals 1 if there is CEO turnover during the fiscal year, 0 otherwise. Liebenberg and Hoyt (2003) find that firms with greater leverage are more likely to appoint a CRO and Pincus et al. (1989) find that firms are more likely to voluntarily form an audit committee when leverage is higher. These results suggest that when leverage is higher, there is a greater need for internal monitoring. Therefore, I include Leverage as a potential determinant of RMC formation and expect a positive sign on the coefficient for Leverage t-1. Leverage equals total debt [DLTT + DLC] divided by total assets [AT]. Bradbury (1990) finds that firms with a greater proportion of outside directors are more likely to voluntarily form an audit committee. Again, I expect that determinants of voluntary audit 21

31 committee formation may also affect the likelihood of voluntary RMC formation. So, I predict a positive sign on the coefficient for Independence t-1. Independence equals the percentage of independent directors on the board. I include Auditor Change t-1 as a control variable, which captures the audit committee s decision to replace the current auditor (or an auditor s decision to resign). Auditor Change equals 1 if the financial institution changes auditors during the fiscal year, 0 otherwise. An auditor change represents a change in external monitoring which may be associated with changes in internal monitoring processes, such as changes in governance structure. Therefore, I expect a positive sign on the coefficient for Auditor Change t-1. Bradbury (1990) finds that firms with a greater number of directors on the board are more likely to voluntarily form an audit committee. Therefore, I include Board Size as a potential determinant of voluntary RMC formation and expect a positive sign on the coefficient for Board Size t-1. Board Size equals the total number of directors. Pincus et al. (1989) find that firms with a Big 8 auditor are more likely to voluntarily form an audit committee. Additionally, Beasley et al. (2005) find that firms with a Big 4 auditor have higher ERM adoption levels. I include BigN as a potential determinant of voluntary RMC formation and expect a positive sign on the coefficient for BigN t-1. BigN equals 1 if the financial institution employs a Big N auditor, 0 otherwise. Because frequent merger and acquisition activity is common in the financial institution industry, I include Merger t-1 as a control variable. Merger equals 1 if the financial institution has a non-zero value in the after-tax acquisition/merger account [AQA], or discloses merger & acquisition activity in the 10-K or proxy statement, 0 otherwise. Mergers and acquisitions 22

32 represent a major event in the life of a firm and can often result in management and board-level structural changes. Therefore, I expect a positive sign on the coefficient for Merger t-1. Restatement equals 1 if the financial institution restates financial statements, 0 otherwise. I expect a positive sign on the coefficient for Restatement (Baxter et al. 2010) because it is consistent with the notion that financial institutions with a higher risk of material financial misstatement may be more likely to form a RMC so that the audit committee can focus more attention on internal controls over financial reporting, fraud risk, etc. Finally, Pincus et al. (1989) find that larger firms are more likely to voluntarily form an audit committee, therefore I control for Size (i.e., the natural log of total assets) and expect that larger financial institutions are more likely to form a RMC. 4.2 Consequences of risk management committee formation Hypothesis 1 and 2 models To test H1 and H2, I estimate the following ordinary least-squares (OLS) regressions: ABS (Risk Measures) t-1, t+n = β 0 + β 1 RMC t + β 2 CRO t + β 3 RMC t *CRO t + β 4 Leverage t-1, t+n + β 5 Independence t-1, t+n + β 6 Board Size t-1, t+n + β 7 CEO Chair t + β 8 BigN t + β 9 Foreign t + β 10 Credit Institution t +β 11 Size t-1, t+n + β 12 Time Fixed Effects t + ε (2) (Risk Measures) t-1, t+n = β 0 + β 1 RMC t + β 2 CRO t + β 3 RMC t *CRO t + β 4 High Risk + β 5 RMC t *High Risk + β 6 CRO t *High Risk + β 7 RMC t *CRO t *High Risk + β 8 Leverage t-1, t+n + β 9 Independence t-1, t+n + β 10 Board Size t-1, t+n + β 11 CEO Chair t + β 12 BigN t + β 13 Foreign t + β 14 Credit Institution t +β 15 Size t-1, t+n + β 16 Time Fixed Effects t + ε (3) Dependent Variables The dependent variable in Equation (2), ABS (Risk Measures),captures the absolute change in a financial institution s risk from the year prior to RMC formation to the year subsequent to RMC formation whereas the dependent variable in Equation (3), (Risk Measures), 23