*Title Page/Author Identifier Page/Abstract

The Role of Information and Financial Reporting In Corporate Governance and Contracting
Christopher S. Armstrong carms@wharton.upenn.edu Wayne R. Guay guay@wharton.upenn.edu The Wharton School University of Pennsylvania 1300 Steinberg-Dietrich Hall Philadelphia, PA 19104-6365 Joseph P. Weber jpweber@mit.edu MIT Sloan School of Management 50 Memorial Drive, Room E52-325f Cambridge, MA 02142

September 21, 2009

Abstract: We review recent literature on the role of financial reporting and information transparency in the areas of corporate governance and contracting. The paper explores the importance of financial reporting transparency in reducing governance-related agency conflicts between managers, directors and shareholders, as well as in reducing agency conflicts between shareholders and outside contracting parties, such as creditors. Some key themes emphasize the governance implications of information asymmetry between managers and outside directors, the endogenous nature of debt contracts and governance mechanisms, and the role of financial reporting in facilitating implicit contracts between managers, directors, shareholders, and creditors. Keywords: financial accounting; corporate governance; board structure; contracting; executive compensation; debt contracts; implicit contracts

This paper has benefitted from numerous discussions with Stan Baiman, Anne Beatty, John Core, Thomas Hemmer, Mirko Heinle, Rick Lambert and, Dave Larcker, Michael Roberts, and Michael Willenborg.

*Manuscript

The Role of Information and Financial Reporting In Corporate Governance and Contracting
Christopher S. Armstrong carms@wharton.upenn.edu Wayne R. Guay guay@wharton.upenn.edu The Wharton School University of Pennsylvania 1300 Steinberg-Dietrich Hall Philadelphia, PA 19104-6365 Joseph P. Weber jpweber@mit.edu MIT Sloan School of Management 50 Memorial Drive, Room E52-325f Cambridge, MA 02142

September 21, 2009

Abstract: We review recent literature on the role of financial reporting and information transparency in the areas of corporate governance and contracting. The paper explores the importance of financial reporting transparency in reducing governance-related agency conflicts between managers, directors and shareholders, as well as in reducing agency conflicts between shareholders and outside contracting parties, such as creditors. Some key themes emphasize the governance implications of information asymmetry between managers and outside directors, the endogenous nature of debt contracts and governance mechanisms, and the role of financial reporting in facilitating implicit contracts between managers, directors, shareholders, and creditors. Keywords: financial accounting; corporate governance; board structure; contracting; executive compensation; debt contracts; implicit contracts

This paper has benefitted from numerous discussions with Stan Baiman, Anne Beatty, John Core, Thomas Hemmer, Mirko Heinle, Rick Lambert and, Dave Larcker, Michael Roberts, and Michael Willenborg.

Table of Contents 1. Introduction ............................................................................................................................................... 4 2. The role of contracts ................................................................................................................................. 5 2.1. Agency conflicts between managers, boards, and shareholders: Governance ................................. 10 2.2. Agency conflicts between shareholders and external contracting parties ........................................ 15 2.3. Debt contracting ............................................................................................................................... 16 2.4. Auditing ........................................................................................................................................... 17 2.5. Properties of financial accounting reports and outputs from the accounting system ....................... 19 3. The role of information in corporate governance ................................................................................... 23 3.1. The role of information in structuring corporate boards .................................................................. 23 3.1.1. Severe forms of information asymmetry: Accounting fraud and restatements ......................... 35 3.2. Mechanisms to commit to a transparent information environment.................................................. 38 3.2.1. Timeliness of financial reports in facilitating information for governance............................... 40 3.2.2. The role of conservative financial reporting in facilitating information for governance .......... 42 3.2.3. The governance role of the auditing process in the information environment .......................... 46 3.2.4. Audit committee of the board of directors ................................................................................ 49 3.2.5. Outside directors as a bonding mechanism to mitigate agency conflicts with creditors and other contracting parties ...................................................................................................................... 53 3.2.6. The governance role of active investors in facilitating the information environment .............. 55 3.2.7. The governance role of financial expert directors in facilitating the information environment 58 3.2.8. The role of management incentives in facilitating the information environment ..................... 61 3.2.9. Discussion of mechanisms to commit to a transparent information environment .................... 63 3.3. The use of accounting information in assessing management ......................................................... 65 3.3.1. Overview of agency theoretic incentive contracting results ..................................................... 67 3.3.2. Types of performance measures in executive compensation contracts ..................................... 72 3.3.2.1. Multiperiod compensation contracting .................................................................................. 74 3.3.2.2 The stewardship and valuation roles of accounting ................................................................ 83 3.3.2.3 The implications of stock price as the primary executive performance measure ................... 85 3.3.3. Implicit contracting with executives ......................................................................................... 90 3.3.4. Senior executive turnover ......................................................................................................... 95 3.3.5. Contractual incentives for earnings management ..................................................................... 98 3.4. The role of information in multi-party contracting ........................................................................ 110 3.5. Good vs. bad governance ........................................................................................................ 112

4. Accounting information and ownership structure ................................................................................. 122 5. The role of financial accounting in debt contracting ............................................................................ 129 5.1. The role of financial accounting in determining whether and from whom the firm obtains debt financing ............................................................................................................................................... 136 5.2. How do elements of the firms accounting system affect the design of the debt contract? ........... 149 5.2.1. Association between financial reporting and interest rates ..................................................... 149 5.2.2. Association between financial reporting and the design of debt covenants ............................ 154 5.2.3. Association between elements of the accounting system and other contract features ............ 158 5.3. The effect of debt contracts on accounting choice ......................................................................... 160 5.3.1. Are covenant violations costly? .............................................................................................. 161 5.3.2. The effect of debt contracts on earnings management ............................................................ 166 5.3.3. The effect of debt contracts on the extent to which firms report conservatively .................... 168 5.4. Theoretical advances and incomplete contract theory ................................................................... 170 5.5. Synthesis and directions for future research .................................................................................. 177 6. The role of auditors ............................................................................................................................... 180 6.1. Research on auditors reducing agency costs .................................................................................. 181 6.2. Research on agency problems arising from the contract between the firm and its auditor ............ 185 7. Other parties in the nexus of contracts .................................................................................................. 188 8. Conclusion ............................................................................................................................................ 189 References ................................................................................................................................................. 194 Appendix A: A model of constrained shareholder value maximization ................................................... 212 Appendix B: Overview of Theoretical Contracting Results ..................................................................... 222

1. Introduction We review recent literature on the role of financial reporting in resolving agency conflicts that arise in the contracting relationships that comprise a firms nexus of contracts. We define contracting, as well as financial reporting, in a broad sense. Contracting encompasses the complex set of explicit and implicit agreements among shareholders, managers, boards, creditors, regulators, auditors, customers, suppliers, and employees, among others. Financial reporting entails the set of financial reports, public filings, and other public disclosures that corporations release for consumption by external parties. Although we touch upon a broad spectrum of contracting settings, our review emphasizes corporate governance and debt contracts as primary areas of focus (with some discussion of contracts with auditors and to an even lesser extent with regulators). This choice stems largely from the observation that much of the theoretical and empirical work on contracting in the accounting literature in recent years is within these areas. A key theme of our review is the notion that a firms contractual arrangements and its corporate information environment evolve together over time to resolve agency conflicts. That is, certain contractual arrangements work more efficiently with certain information and financial reporting choices. As a result, one does not necessarily expect to observe firms converging to a single dominant type of corporate governance structure, compensation contract, debt contract, or financial reporting system, but rather one expects to observe heterogeneity in these mechanisms that is a function of firms economic characteristics. Although this notion appears to be more widely accepted in the literature on debt and other contracts, the governance literature seems more burdened by the idea that some forms of governance are good and some are bad.

In Section 2, we briefly discuss the firm as a nexus of contracts, the general nature of contracts related to governance, debt, auditors, and regulators, and properties of the information environment and financial reports that are relevant to various contracting settings. Section 3 discusses the role of information in corporate governance, with an emphasis on corporate boards and executive compensation arrangements. In Section 4 we discuss the role of accounting information in ownership structure with an emphasis on minority shareholders. In Section 5, we discuss the role of financial reporting in the design of debt contracts. Sections 6 and 7 provide brief reviews of the contracting literatures related to auditors and regulators, respectively. Section 8 provides a synthesis of the main themes in our review and a discussion of what we consider to be fruitful areas for future research. Appendix A develops an illustrative model that provides a structure for tying together most of main points in this review.

2. The role of contracts In this survey, we take the perspective of the firm as a nexus of contracts among the various factors of production, where contracts serve to mitigate agency conflicts and can be both explicit and implicit (Coase, 1937; Alchian and Demsetz, 1972, Jensen and Meckling, 1976; Fama and Jensen, 1983). Although explicit contracts, such as debt and executive compensation contracts, might be the first thing that comes to mind when one considers the role of financial reporting in contracting, implicit contracts are vitally important to the functioning of the firm and in mitigating agency conflicts. In this survey, we devote considerable attention to the role of financial reporting in implicit contracting settings, largely because this type of contracting arrangement is the focus of many of the papers we discuss and, we believe, will be an important source of future progress in the contracting and governance literatures.

We note that an explicit contract is typically quite narrow in scope, covering an arrangement between contracting parties that specifies various duties, responsibilities, quantities, prices, and what to do in the event of certain foreseeable contingencies. Examples of explicit contracts include public or private offerings of equity or debt, employee compensation contracts, purchase or sales contracts, derivatives instruments, exchange listing requirements, SEC financial reporting and disclosure regulations, litigation settlements, tax agreements with governmental entities, and service contracts with auditors or law firms. In contrast, implicit contracts constitute the broad set of (usually) unwritten or informal arrangements that allow the firm to engage in activities that are otherwise non-contractible in the sense that it is either prohibitively costly or impractical to write an explicit contract. Although implicit contracts do not typically have a formal enforcement mechanism, concerns regarding loss of reputational capital, costly signaling, costly commitment or investment, or the fear of retaliation in a repeated game setting discourage contracting parties from reneging. One operative (economic) definition of an implicit contract is a non-contractual agreement that corresponds to a Nash equilibrium to the repeated bilateral trading game other than the sequence of Nash equilibria to the one-shot trading game. (Bull, 1997). Another, more legal, definition of an implicit contract is found in the United States Supreme Court Case Baltimore & Ohio R. Co. v. United States, 261 U.S. 592 (1923), which defines an agreement implied in fact as one which is founded upon a meeting of minds, which, although not embodied in an express contract, is inferred, as a fact, from conduct of the parties showing, in the light of the surrounding circumstances, their tacit understanding. This latter definition is more akin to an explicit contract in that it establishes criteria for an unwritten contract to be legally binding. In this survey, our discussion of implicit contracts goes far beyond those implicit contracts that are

legally binding (although the legal enforceability of an implicit contract may well be a consideration by contracting parties in some settings). To illustrate the breadth of a firms explicit and implicit contracting arrangements, consider a closely-held private firm that decides to go public, list its shares on an organized stock exchange, and sell equity to outside investors. This decision requires the firm to enter into an array of new contractual arrangements, many of which have received substantial attention from accounting researchers. For example, the firm enters into relationships with its investment bank and audit firm to help list and sell its shares. The newly public firm also enters into contractual agreements with an organized stock exchange and with the SEC to make certain disclosures and to periodically prepare and file audited financial accounting reports in accordance with GAAP. A variety of relationships also evolve over time as the firm develops a reputation for various strategic policies. Customers and suppliers build relationships with the firm and grow to expect certain quality, price and safety levels in products or services. Employees, most of whom are employed at will, establish relationships with the firm regarding compensation, benefits, working conditions, promotion opportunities, etc. Creditors establish lending relationships with the firm. Commitments to certain types of governance structures, financial reporting and disclosure quality, payout policies, risk management practices, and operating and investing practices evolve over time that shareholders, creditors and others grow to rely upon. Firms also establish reputations with regulators, taxing authorities, and politicians. Analysts, the financial press, institutional investors, and other information intermediaries begin to follow the firm and grow to expect certain financial reporting practices and other behaviors.

For purposes of this survey, implicit and explicit contracts related to governance and debt markets are the primary areas of focus. With respect to the firms system of corporate governance, both explicit and implicit agreements determine the number of insiders and outsiders on the board of directors, the size of the board, the required background of board members, the number of board meetings, the types of shareholder protections and rights that are allowed/required, etc. Specifically, although corporate charters, formal regulation and stock exchange rules shape some of the basic governance structure, many of the governance-related responsibilities and actions of board members and shareholders require significant judgment and discretion. Contracts with executives, a key focal point in this survey, also have both explicit and implicit components, although the latter seems much more interesting and important. For example, many CEOs do not have explicit written employment agreements with their employers (Gillan, Hartzell, and Parrino, 2009). In these cases, most of the CEOs responsibilities with respect to operating, investing and financing activities, as well as compensation and the length of employment, are governed through an implicit agreement between the CEO and the board. Further, within an implicit employment agreement, parties often operate under significant uncertainty about factors that may result in the termination of the agreement, and the rights and obligations of each party if the manager is terminated.1 Incentive structures for executives and board members are also set through both explicit and implicit agreements. For example, although boards sometimes impose minimum equity
1

For example, empirical evidence indicates that executives are terminated by boards in the wake of poor stock price and accounting performance (e.g., Engel, Hayes and Wang, 2003), but precisely how poor this performance must be before a board asks an executive to leave the firm is setting specific. At the same time, executives are more likely to terminate the agreement when outside labor market opportunities are stronger or when retention incentives are weaker. Further, the specifics of the termination agreement, in terms of severance payouts, option and restricted stock vesting, and other benefits, is frequently agreed upon outside of an explicit contract.

ownership guidelines for executives and directors, many firms do not, and even for those firms that do impose such guidelines, CEOs are frequently expected to hold substantially more equity than is explicitly required. We also note that for those CEOs that do have explicit employment contracts, only a small proportion of the CEOs responsibilities, compensation, etc. are specifically addressed in such agreements. When a firm chooses to raise debt capital, it enters into an array of explicit and implicit contractual arrangements with its creditors. Explicit contracts with debt holders include details on the amount that can be borrowed, the interest rates charged, covenant thresholds, and the maturity date of the loan. Implicit relationships, however, also influence these decisions by creditors because the decisions, in part, depend upon the reputation the firm has established with respect to financial transparency, corporate governance, risk management, and other strategic and operating policies. Further, decisions made by debt holders with respect to exercising decision-making rights over the life of the contract (e.g., at the time of a covenant default) will also depend on both the terms of the explicit contract and the implicit relationships that have been established. We could continue this example to discuss many other explicit and implicit contracting settings related to customers, suppliers, lower-level employees, tax authorities, financial intermediaries, etc. The key point here is simply that both explicit and implicit agreements are critical to the efficient functioning of the firm. We also note that we restrict our survey to the set of explicit and implicit contracts that serve to resolve agency conflicts, as opposed to contracts that serve other purposes, such as the valuation role of financial reporting.2 Similarly, we ignore

A large literature examines investors demands for financial reporting information that is used in the valuation of equity, debt, and other financial claims. This valuation role for financial information exists even in the absence of agency conflicts between managers and shareholders (i.e., even if shareholders and managers interests were perfectly aligned, shareholders would still require financial information about the firm to facilitate a liquid market

contractual arrangements between the firm and both its customers and suppliers that may serve purposes other than to resolve agency conflicts. For the most part, we also ignore the explicit and implicit contracts between regulators and the firm, as this topic is too broad to cover in this survey. 2.1. Agency conflicts between managers, boards, and shareholders: Governance We view corporate governance as the subset of contracts in the firms nexus of contracts that help align the actions and choices of managers with the interests of shareholders, and suggest that the firms accounting system plays an important role in this process. This definition is broadly consistent with the views of authors such as Jensen (1993), Mehran (1995), Shleifer and Vishny (1997), Core, Holthausen, and Larcker (1999), Holderness (2003), and Core, Guay and Larcker (2003). Some features of this definition of governance are worth discussing. First, we adopt a shareholder perspective or outsiders perspective with respect to governance. That is, one can think of a hierarchy or chain of command within the corporation that includes shareholders, the board of directors, and managers. Shareholders are the owners of the equity capital in the firm, and are the residual claimants on the firms cash flows. Typically, however, they are not involved in the day-to-day activities and decision-making in the organization (at least in widelyheld firms). Instead, they delegate these responsibilities to a board of directors. Most members of the board of directors also are not typically involved in the day-to-day activities and decision-making in the organization. Rather, they delegate these responsibilities to a team of managers. Corporate governance consists of the mechanisms by which the shareholders ensure

for buying and selling shares). For a review of this literature, see Kothari (2001), or Lee (2001). Also note that the contracting role of information may sometimes be in conflict with its valuation role (Gjesdal, 1981). This is a common theme in the executive contracting literature and we discuss this issue in greater detail below.

10

that the board of directors, in turn, ensures that managers interests are aligned with those of shareholders. Researchers on corporate governance issues often analyze one of two types of agency problems in the linkage between managers interests with shareholders interests. One type of agency problem arises when the interests of shareholders and the board of directors are aligned, but where managers interests are not aligned. Examples include papers that examine various compensation plans or other monitoring mechanisms, such as a commitment to timely accounting information, that are used by boards to mitigate managers incentives to act in ways that are detrimental to shareholders. In these papers, researchers typically assume that the board of directors is comprised of individuals that make governance decisions that are in the best interests of shareholders. The second type of agency problem arises when boards and managers interests are aligned with each other but are not completely aligned with the interests of shareholders. In these papers, management is typically believed to have substantial influence over the board of directors actions. Although there are a variety of mechanisms the firm can use to mitigate governancerelated agency conflicts, in this survey we focus on the research investigating how the firms financial accounting system can help reduce these agency costs. In particular, we emphasize information asymmetries that exist between managers and both outside directors and shareholders. For example, if left unchecked, managers might be expected to withhold information that is detrimental to their personal interests (e.g., information indicating poor performance, extraction of private benefits, etc.). Thus, one potential role for the financial reporting system is to provide outside directors and/or shareholders with relevant and reliable information that aids in the monitoring of management and/or directors.

11

Before moving on, we point out that although our survey does not delve deeply into the legal literature on governance, the law appears to view boards as having a primary fiduciary responsibility to shareholders rather than the firms other contracting parties. Huebner and McCullough (2008) highlight that in North American Catholic Educational Programming Foundation, Inc. v. Gheewalla, 930 A.2d 92 (Del. 2007), the Delaware Supreme court recently summarized the duties of directors as follows: It is well established that the directors owe their fiduciary obligations to the corporation and its shareholders. While shareholders rely on directors acting as fiduciaries to protect their interests, creditors are afforded protection through contractual agreements, fraud and fraudulent conveyance law, implied covenants of good faith and fair dealing, bankruptcy law, general commercial law and other sources of creditor rights. Delaware courts have traditionally been reluctant to expand existing fiduciary duties. Accordingly, the general rule is that directors do not owe creditors duties beyond the relevant contractual terms. Further, they go on to note that in Gheewalla, the Delaware courts uphold the boards responsibility to shareholders even when firms enter a zone of insolvency: [T]he need for providing directors with definitive guidance compels us to hold that no direct claim for breach of fiduciary duties may be asserted by the creditors of a solvent corporation that is operating in the zone of insolvency. When a solvent corporation is navigating in the zone of insolvency, the focus for Delaware directors does not change: directors must continue to discharge their fiduciary duties to the corporation and its shareholders by exercising their business judgment in the best interests of the corporation for the benefit of its shareholder owners. In general, it appears that a boards legal responsibility to creditors (or other stakeholders) shifts only once the creditors (or other stakeholders) take over ownership of the firm and effectively become the firms shareholders. Although we emphasize the maximization of shareholder value in our discussion of governance, it is useful to briefly consider how this approach relates to the interests of other contracting parties in the firms nexus. At first glance, it may seem that aligning managers

12

interests with shareholders implies that managers interests are not aligned with other contracting parties. While this will sometimes be true, in many (or most) cases, managers, boards, and shareholders have strong incentives to consider the interests of the firms other contracting parties. As an illustration, consider a firm where the interests of the managers, board and shareholders are all perfectly aligned. In other words, from a shareholder perspective, there are no residual agency conflicts between the key parties of interest. In such a firm, managers will attempt to maximize shareholder value when they enter into contracts with other parties on behalf of shareholders. Of course, resolving agency conflicts between managers and shareholders does not necessarily resolve agency conflicts between shareholders and other parties within the firms nexus of contracts, and it may sometimes exacerbate such conflicts. When outside parties contract with the firm, they expect agency conflicts to exist, and will take measures to protect themselves from potential actions by the firms managers that might harm them after entering into the contractual relationship. Protection could take the form of charging a higher price for goods, services or funds, writing a contract that specifies penalties for certain outcomes or that renegotiation occur in the event of various foreseen contingencies, or by simply refusing to contract with the firm. As a result, shareholders are not expected to benefit, ex ante, from the managers ability to take actions that are detrimental to other contracting parties, ex post. In fact, it is well recognized that shareholders bear residual losses when they cannot commit ex ante to not take such actions in the future (i.e., when there are residual agency conflicts; Jensen and Meckling, 1976). As a result, shareholders welcome mechanisms that allow managers ex ante to commit to maximize other contracting parties

13

interests ex post. That is, a commitment to maximize other contracting parties interests ex post increases firm value (and therefore shareholder value) ex ante. As an example, consider a board of directors that could consist of a majority of inside directors or a majority of outside directors. Assume that in a healthy business environment, all directors take actions that are in the best interest of shareholders. In times of financial distress, however, assume that inside directors are more likely than outside directors to take actions that expropriate wealth from other contracting parties, such as bondholders or employees, even if it is in shareholders interests to do so (e.g., because insiders have substantial stock and option holdings as well as human capital tied to the firm). In this setting, having outside directors on the board serves as a mechanism for shareholders to commit ex ante not to take actions, ex post, that are detrimental to other contracting parties (we return to this point below in the context of the relation between outside directors and the information environment). As a result, creditors and other parties will be willing to contract with the firm on better terms, monitoring costs will be lower, and firm value will be greater. These outcomes (reduced agency costs) are likely to be reinforced if outside directors, who frequently serve as senior executives at other corporations, fear reputational damage from taking actions that damage these other parties (see Fama and Jensen, 1983; Gerety and Lehn, 1997; Srinivasan, 2005). Of course, shareholders will generally not be able to fully commit to this course of action, and so residual agency conflicts will persist that induce the firm and other contracting parties to incur monitoring costs, as well as cause other contracting parties to protect themselves, thereby reducing firm value.3 However, for our purposes, the important point is that

An additional type of agency problem stems from conflicts of interest between shareholders and other parties that have a vested interest in the firm, but who do not directly contract with the firm, for example, where there are externalities related to societal interests. Shareholders are expected to internalize some, though certainly not all of these societal effects of their operations (e.g., pollution issues, community impact, etc.). These societal

14

it is in shareholders interests to commit ex ante to maximize the value of the firm, which corresponds to maximizing the value of the claims of all parties in the nexus of contracts. We dedicate a substantial portion of this survey to the agency conflicts between shareholders/boards/managers and other contracting parties, such as creditors. As a largely semantic issue, we view these conflicts as a separate class of agency problems. Specifically, we break our survey into two broad categories of agency conflicts: 1) the set of agency conflicts among shareholders, boards, and managers when these parties interests are not completely aligned (which we refer to as internal agency problems related to governance), and; 2) the set of agency conflicts with all other contracting parties to the firm when the interests of shareholders, boards, and managers are completely aligned (which we refer to as external agency problems). 2.2. Agency conflicts between shareholders and external contracting parties As noted above, a host of external parties, such as creditors, suppliers, distributors, customers, regulators, auditors, analysts, and taxing authorities are parties to explicit and implicit contracts with the firm. Some of these contractual relationships, such as debt contracts, have received relatively more attention in the accounting literature. Others have received less attention, either because the contractual relationship is less interesting to researchers, there are problems with data availability, or perhaps the accounting system does not play a prominent role in the contracting process. In this survey, we focus on implicit and explicit contracts with creditors, and to a lesser extent, with auditors. For reviews of contracting relationships with financial intermediaries and taxing authorities (IRS), see Beyer et al. (2009) and Hanlon and Heitzman (2009), respectively. Further, although contracting relationships with regulators are a

externalities and their implications are addressed in the literature that views corporate governance from a stakeholder perspective (e.g., Tirole, 2001). Further reconciling the differences between shareholder and stakeholder governance perspectives are beyond the scope of our survey.

15

very important research area, the vast scope of that literature makes it infeasible for us to discuss in any detail here. 2.3. Debt contracting Several important papers in the finance and economics literature develop models of agency problems that stem from interactions between debt holders and equity holders. To introduce the basic problem, we focus on Jensen and Mecklings (1976) analysis of the relationship between managers and outside providers of capital (equity holders or debt holders). In their framework, the owner-manager has incentives to engage in actions (e.g., perquisite consumption) that benefit the owner/manager to the detriment of the debt holder. The external capital provider rationally anticipates the managers actions and price-protects its claims from potential losses. Price protection and monitoring costs give rise to agency costs that are borne by the owner/manager. Their framework suggests that implicit contracts that reduce the managers ability to expropriate wealth, or explicit contracts that include provisions that restrict owner/managers from engaging in certain actions can potentially reduce agency costs and increase the efficiency of the contracting process. Building on Jensen and Mecklings insights, the accounting system can play at least two important roles in reducing the agency costs that arise in the debt contracting process. First, a commitment to provide high quality accounting reports can serve as an implicit contract between the debt holder and the firm, which facilitates debt contracting. Because borrowing is typically a repeated game, managers have incentives to develop their reputation and make commitments to lower the cost of debt contracting in order to increase shareholder value. Second, outputs from the accounting system can serve as inputs (or parameters) in the explicit contract between the firm and debt holders. More specifically, outputs from the accounting

16

system can be incorporated into the contractual mechanisms that restrict managers from incurring (increasing) agency costs. For example, debt agreements often contain covenants based on accounting numbers that restrict managerial actions which, in turn, reduces agency costs. Although the focus of our paper is on the role of accounting information in reducing agency costs, in the debt contracting setting, accounting information has the potential to play an enhanced role in the contracting process. Debt contracts often include clauses (covenants) that are based on accounting information. One of the reasons these clauses (covenants) are included in the contract is to reduce agency costs by providing creditors with the potential to increase their control rights when the firm performs poorly (Aghion and Bolton, 1992). However, these clauses may exist even if there were no agency problem, as they could also serve as a mechanism to allow for efficient contract renegotiation. That is, since it is too costly to write a contract that stipulates the rights and obligations of the contracting parties in all states of the world, firms often write incomplete contracts that allow for renegotiation under certain conditions. Renegotiation of debt contracts is often based on outputs of the accounting system. While the accounting literature has mostly focused on the agency problems associated with debt, we will discuss the role of the accounting system in renegotiation in an incomplete contracting setting. 2.4. Auditing Watts and Zimmerman (1986) note that the efficacy of the accounting system in reducing agency conflicts depends, at least in part, on whether accounting principles are applied in a manner consistent with Generally Accepted Accounting Principles (GAAP), or other standards outlined in the contracting setting. The auditor serves a central role in this process. That is, in

17

order for accounting reports to be effective in reducing agency costs, contracting parties must be provided with some assurance that the financial reports have been prepared in accordance with the provisions of the contract. This gives rise to a demand for auditing, as an audit certification provides some assurance that the financial statements have been prepared in accordance with a set of recognized principles. This assurance can come from either the insurance role of the audit (i.e., the auditor can be sued if there is an error or irregularity), or from the auditors reputation. When financial accounting numbers play a more prominent role in the contract, the demand for auditing is, ceteris paribus, expected to be greater. Regarding the nature of the audit relationship, there is an explicit contract between the audit firm and the client, where the client pays the auditor to attest to the accuracy of its financial statements. There is also an understanding between the auditor and the shareholders, as well as other contracting parties that rely on the financial statements, that the auditor will discover and correct errors and irregularities contained in the firms financial statements. Since auditors are typically paid by the firms they audit, there is a potential conflict of interest that can compromise their independence. That is, rather than ensuring that the financial reports are prepared in accordance with a prescribed set of accounting principles, auditors may shirk or simply ignore errors and irregularities. Further, the likelihood of shirking depends, in part, on the auditors reputation, as well as the overall nature of the relationship between the auditor and the firm. We also note that the regulatory process has created a demand for auditing, as all publicly traded companies (and some privately held companies) are required to prepare audited financial statements. However, given the regulatory requirement that all public firms must be

18

audited, if competition among auditors is weak, audit firms may have incentives to shirk, again leading to unresolved agency problems. 2.5. Properties of financial accounting reports and outputs from the accounting system In the contracting settings that are the focus of this paper, the corporate information environment plays an important role in reducing the agency costs that arise from explicit and implicit contracts. For example, in governance settings, board members require financial and non-financial information about firm performance, risk, and investment to evaluate executives. In debt contracts, stockholders must provide bondholders with credible information about performance, risk, assets, and net worth to achieve efficient debt contracts. In this survey, we interpret financial reporting rather broadly to include financial statements, required footnote disclosures, and other mandatory disclosures.4 The role of financial accounting information is likely to be somewhat different depending on whether the contracting setting is implicit as opposed to explicit. Explicit contracts are typically written over recognized financial statement numbers or other standardized financial numbers in the accounting reports. Implicit contracts, on the other hand, rely more generally on overall information content (as well as a commitment to providing information), as opposed to specific line items or figures in the financial reports. To highlight the importance of this distinction consider the relatively recent controversy over the expensing of employee stock options. Prior to the adoption of SFAS 123R, firms were not required to recognize the expense associated with employee stock option grants in the

To make our survey more manageable, we do not discuss in detail the literature related to voluntary disclosures outside of financial reports (e.g., press releases, management forecasts, conference calls, etc.) or the vast disclosure literature that is concerned with the disclosure of a generic information signal (which can often be interpreted as an accounting report).

19

income statement, but the underlying information was instead disclosed in the footnotes to the financial statements. The recognition versus disclosure debate is likely to have different implications for explicit and implicit contracts. Explicit contracts put in place prior to SFAS 123R are likely to be mechanically affected by this change in financial reporting (unless the contract anticipated the rule change, or otherwise adjusted for stock option expense). Implicit contracts, on the other hand, are likely to be affected less by whether the cost of employee stock options is, or is not, included in the reported income number. Rather, it would be sufficient that this cost (or its underlying inputs) is adequately and credibly disclosed to contracting parties so that their decisions can incorporate this information (except to the extent that the actual recognition of stock option expense has implications for the information content of these reports; e.g., in the event that managers distort option expense figures once they are required to be recognized). The Concept Statements issued by the FASB highlight relevance and reliability as the primary characteristics that make accounting information useful, and go on to discuss other attributes that different financial statement users might value, such as comparability, consistency, and materiality. In this survey, we focus on the role of accounting information as a mechanism to reduce agency costs that arise through the firms implicit and explicit contracts. As such, the attributes of relevance and reliability are clearly important, and it is difficult to imagine that a set of financial statements that are either unreliable or irrelevant will be effective in reducing agency costs. Inherent in any discussion of relevance and reliability, however, is the tradeoffs that firms must make in order to provide relevant or reliable information to financial statement users. We believe that the essence of this tradeoff might be most intuitively characterized as one between verifiability and timeliness. That is, in order for information to be

20

useful (or relevant) in most contracting settings, the information should be timely. However, as firms try to provide timely information to decision makers they often have to trade-off against the verifiability of the underlying information. For example, the general nature of accrual accounting implies that when measured over a long enough time period, cumulative accounting earnings and cash flows are the same. Therefore, eventually, the accounting system will capture and convey (through earnings) all information about cash flows. However, by introducing accruals, which are estimates of future cash flows, accounting reports introduce less reliable measures into the system. As managers attempt to provide more timely information to decision makers (more accruals) the verifiability of the information typically deteriorates. While we do not intend to diminish the importance of the concepts of relevance and reliability, we hope to streamline the discussion on the attributes of accounting that are likely to be important in a contracting setting by focusing on the verifiability and timeliness of information about current and future cash flows.5 Many contracting settings rely on timely signals about various aspects of firm performance. For example, Adams, Hermalin and Weisbach (2009) discuss a variety of theoretical models of governance. Each of these models discusses information signals that are obtained by the board about various aspects of the CEOs actions and performance. Further, in some models, the CEO is able to influence the information that is received by the board. Although most of these models do not specifically discuss how financial reporting fits into the

One nuanced perspective on timeliness is the view that a primary role of financial statements is to provide verifiable evidence on whether management has done what it said it would do (e.g., whether the firm, in fact, has generated the profits that it has conveyed to investors through other disclosures). One take on this perspective is that it views the financial statements as conveying timely information about the verifiable profits that the firm has generated. Alternatively, one could view this perspective as too narrow in that it does not appear to consider the role of the forward-looking information that the accounting process attempts to incorporate into financial statements.

21

model, they often provide sufficient structure to link the role of the information in the model to various general attributes of accounting information. Given the recent prominence of accounting research on conservatism in financial reporting, we discuss this literature as somewhat separate from the more general research on the role of timely financial reporting. Conservatism is a property of financial reports that is frequently argued to serve an important role in contracting in general, and corporate governance in particular. Typically, conservatism is defined as the differential verifiability required for the recognition of accounting gains versus losses that generates a persistent understatement of net assets (LaFond and Watts, 2008; Holthausen and Watts, 2001; Basu, 1997). In particular, conservative financial reports recognize information about difficult-to-verify losses in a timelier manner than difficult-to-verify information about gains. The distinction between difficult- and easy-to-verify information is important. There is generally symmetry in the recognition of easy-to-verify information about gains and losses. For example, both accounts receivable (easy-to-verify gain) and accrued wages expense (easyto-verify loss) are recognized in a timely manner. In contrast, while goodwill write-downs are required when this intangible asset is impaired (difficult-to-verify loss), writing up the value of goodwill when it has appreciated (difficult-to-verify gain) is not allowed. Difficult-toverify gains are only recognized in earnings over time when the gains materialize either as realized revenues or as other easy-to-verify gains. We note two subtle, but important, alternative interpretations of the potential contracting benefits of conservatism. The first interpretation views both the timely recognition of losses and the untimely recognition of gains as being important for contracting (gains are recognized in financial reports only when they become easy to verify). This view appears to mainly

22

emphasize explicit contracts, such as executive compensation and debt contracts, and argues that timely (untimely) recognition of losses (gains) produces accounting numbers with properties that are beneficial in reducing agency conflicts (see for example Guay, 2008 and Beatty, Weber and Yu, 2009). The second interpretation views the timely loss recognition aspect of conservatism as being much more important than the untimely recognition of gains. This view emphasizes implicit contracts, and argues that reporting rules that force managers to recognize losses in a timely manner enhances the overall information environment. That is, managers are expected to have economic incentives to voluntarily disclose gains (i.e., good news), and to suppress losses (i.e., bad news) (see for example Watts, 2003a and 2003b and Guay and Verrecchia, 2009). Reporting rules that commit managers to disclose bad news in a timely manner result in a more complete disclosure environment that facilitates a wide array of contracts. In the next section we begin our survey of the research on the role of accounting information in contracting and corporate governance by focusing on the relatively recent research on how the firms accounting system fits into their overall system of corporate governance. We focus on papers that have been published in accounting journals, as well as papers that are likely to be of interest to accountants but were published in finance and economics journals.

3. The role of information in corporate governance 3.1. The role of information in structuring corporate boards Information plays a critical role in corporate governance. As noted above, the key issue in corporate governance is constructing mechanisms that align executives objectives with the

23

interests of shareholders. Boards of directors are tasked with much of the work in advising and monitoring executives toward this end. Corporate boards are typically comprised of both outside directors and inside directors.6 The value of having outside directors on the board derives, in part, from their broad expertise in areas such as business strategy, finance, marketing, operations, and organizational structure. Further, outside directors bring an independence that carries with it an expectation of superior objectivity in monitoring the behavior of management. Their diligence in this respect may stem partially from monetary incentives attached to being a board member, but possibly even more importantly from their desire to protect the significant reputational capital that most outside directors have developed. Inside directors, who are typically executives of the firm, are expected to be much better informed about the firms activities than outside directors, but may withhold their information if it can be used to discipline the executives or to take away private benefits. Thus, a key advantage of inside directors is also the key disadvantage of outside directors: specifically the differential costs and difficulty in ensuring that the director has adequate information with which to make decisions. Such information transfer between insiders and outsiders is not trivial, and is the focus of much of the theoretical literature on corporate governance. Outside directors are typically high quality individuals that already have tremendous demands on their time. It is unrealistic to expect that an outside director can or will invest the time and effort necessary to become as informed as the executives of the firm they

Pursuant to Item 470(a) of Regulation S-K, firms must disclose whether each director is independent within the definition prescribed by the exchange on which the firms shares are traded. Directors are typically classified as insiders, outsiders, and affiliates (or gray directors). Insiders are those who are currently employed by the firm, such as the CEO, CFO, president and vice presidents. Outsiders have no affiliation with the firm beyond their membership on its board of directors. Affiliates are former employees, relatives of the CEO, or those who engage in significant transactions and business relationships with the firm as defined by Items 404(a) and (b) of Regulation SK. Directors on interlocking boards are also considered to be affiliated, where interlocking boards are defined by Item 402(j)(3)(ii) of Regulation S-K as those situations in which an inside director serves on a non-inside directors board.

24

are governing. Further compounding these informational problems is the fact that outside directors must to a large extent rely on the same executives that they are monitoring and advising to provide them with the information necessary to achieve effective corporate governance (although auditors, regulators, analysts, the media, and other information intermediaries also assist outside directors in this regard). These executives are unlikely to be entirely forthcoming with information that might have a detrimental effect on the way outside directors view the executives effort, performance, and ability (e.g., see Section 3 of Verrecchia, 2001). Holmstrom (2005, 711-712) provides a helpful characterization of the information flow issues between management and boards: Getting information requires a trusting relationship with management. If the board becomes overly inquisitive and starts questioning everything that the management does, it will quickly be shut out of the most critical information flow the tacit information that comes forward when management trusts that the board understands how to relate to this information and how to use it. Management will keep information to itself if it fears excessive board intervention. A smart board will let management have its freedom in exchange for the information that such trust engenders. Indeed, as long as management does not have to be concerned with excessive intervention, it wants to keep the board informed in case adverse events are encountered. Having an ill-informed board is also bad for management, since the risk of capricious intervention or dismissal increases.

The tradeoffs between retaining inside and outside directors, having a large board as opposed to a small board, as well as other features of board and committee structure, have been the subject of considerable research during the last several years. Specifically, a growing literature explores the ways in which firms select governance mechanisms to help align the actions of managers with shareholders, reduce contracting costs, and make better strategic decisions. Hermalin and Weisbach (2003) and Adams, Hermalin and Weisbach (2009) survey theoretical and empirical work on the role of corporate boards, and how heterogeneity in the

25

agency conflicts that firms face is expected to give rise to heterogeneity in the governance mechanisms that firms use to solve these agency conflicts. Since the information environment influences many of the choices over the structure of these governance mechanisms, we focus a substantial portion of our review on this literature. Board size and board independence are two of the most intensely studied governance characteristics. Clearly, there are tradeoffs between having a small vs. large board and having more vs. fewer insiders on the board. A board with only one or two members may not have the range of expertise to govern effectively, just as a board with fifty members may have coordination and free-riding problems that prevent it from governing effectively. Similarly, a board composed of entirely insiders may not be effective because of the potential for allowing managerial entrenchment, just as a board with no insiders may not be effective if the board has a limited understanding of the firm they are governing and cannot readily overcome this information asymmetry. Hypotheses regarding optimal board size and director independence stem largely from the body of theoretical and empirical literature that argues boards serve two key functions: 1) advising executive management, and; 2) monitoring executive management.7 Along the advising dimension, a common prediction is that larger and more complex firms require larger boards with a diverse group of outside directors. A large group of outside directors, it is argued, can provide management with collectively superior advice through their broad range of expertise (but recognizing that there are information acquisition and processing costs to convert their general expertise into firm-specific knowledge). A large board can also handle the relatively greater number of tasks and committees in large, complex firms. At some point,

For example, see Fama and Jensen (1983), Raheja (2005), Boone, Field, Karpoff, and Raheja (2007), Drymiotes (2007), Lehn, Patro, and Zhao (2008), Linck, Netter and Yang (2008), and Harris and Raviv (2008).

26

however, coordination costs and free-rider problems outweigh the benefits of additional board members, thereby placing limits on the efficient size of the board. Further, in firms with substantial investment opportunities and complex investments, considerable firm-specific knowledge may be necessary to effectively advise management. In these settings, the informational advantage of insiders vis--vis outsiders may impede the advisory role of outside directors and lead to a greater proportion of inside directors (e.g., see Coles, Daniel, and Naveen, 2008). With respect to the boards monitoring function, hypotheses frequently emphasize how the firms information and operating environment influence the monitoring costs and benefits of certain board structures. Specifically, it is argued that firms in noisy business environments, such as high growth firms with substantial R&D, intangible assets, and volatility, are more difficult and costly to monitor (e.g., Demsetz and Lehn, 1985; Gillan, Hartzell, and Starks, 2006; and, Coles, Daniel and Naveen, 2008). Smaller boards, with lower coordination costs and free-rider problems, are better able to cost-effectively monitor high-growth firms. Further, because it is costly for outside directors to acquire the information necessary to effectively monitor managers, firms characterized by greater information asymmetry between managers and outsiders are predicted to have a higher proportion of inside directors. Finally, larger and more profitable firms are expected to provide managers with greater scope to obtain private benefits, thereby requiring a greater proportion of outside directors, which are assumed to provide better monitoring.8 Inside directors, for their part, can facilitate effective decision-making because they are a valuable source of firm-specific information about constraints and opportunities (e.g., see
8

As noted by Hermalin (2005), an additional cost associated with greater monitoring by outside directors is that CEOs are expected to demand greater pay as compensation for greater job security risk and the greater effort that results from more intense monitoring.

27

Raheja, 2005; Harris and Raviv, 2008; and Adams, Hermalin, and Weisbach, 2009), and can also be particularly helpful in educating outside board members on the firms activities (Fama and Jensen, 1983). At the same time, however, inside directors are potentially conflicted in their incentives to monitor due to their lack of independence from the CEO, as well as a desire to protect their own private benefits.9 Further, even though better informed outside directors are likely to more effectively advise the CEO, if a well-informed outside board is more likely to interfere with the CEOs strategic decisions, insiders may be reluctant to share their information with outside directors (Adams and Ferreira, 2007). The above arguments lead to the general prediction that a key cost of having outsiders on the board decreases with the ease with which outside directors can obtain and assimilate the superior information known to inside directors. In other words, if outside directors could costlessly become equally informed as inside directors, one would expect shareholders to prefer the board to consist largely (or completely) of outside directors outsiders would bring independence and additional perspectives to the decision-making without being disadvantaged in the information upon which the decisions are made. Bushman, Chen, Engel and Smith (2004, 179) summarize this tradeoff clearly: An important question of board composition concerns the ideal combination of outside and inside members. Outsiders are more independent of a firms CEO, but are potentially less informed regarding firm projects than insiders. Insiders are better informed regarding firm projects, but have potentially distorted incentives deriving from their lack of independence from the firms CEO.

However, see Drymiotes (2007) for a setting where more inside directors might actually improve the efficiency of the monitoring role of the board. In his model, outside directors have an incentive to shirk in their monitoring duties, and to shortchange the CEO with respect to his performance ex post. Inside directors, who represent the CEOs interests, can commit to expending monitoring effort ex post, thereby increasing the CEOs incentives to exert productive effort.

28

Note also that the hypothesis described above relates to information asymmetry between managers and outside directors, as opposed to information asymmetry between managers and outside investors, information asymmetry between investors, or simply general uncertainty about a firms business environment. For example, outside directors will undoubtedly have some information that the general population of investors does not have (although most of this information is likely filtered by management when it is provided to outside directors). Further, managers and boards will both face general uncertainty about their firms future prospects, and such uncertainty does not necessarily translate into information asymmetry between managers and outside directors (or with investors). From a practical perspective, however, researchers cannot directly observe information asymmetry between managers and outside directors, and therefore must construct proxies for this asymmetry. Examples of such proxies include various measures of the degree of growth options in the investment opportunity set (e.g., R&D expenditures, capital investment, etc.), as well as measures of uncertainty in the business environment (e.g., stock price volatility). Although these proxies likely capture variation in the extent to which managers are in a position to obtain information that is not easily known to boards or investors, these measures generally do not (and quite possibly cannot) discriminate between manager-director information asymmetry. Finally, we emphasize that our discussion of an expected positive relation between information transparency and the proportion of outside directors is distinctly different from the frequently found view in the literature that a board with a high percentage of outside directors facilitates better governance than a board with a lower percentage of outside directors. Or equivalently, the view that a board with 100% outside directors will unconditionally govern asymmetry and manager-investor information

29

better than a board with 70% outside directors, which will unconditionally govern better than a board with 40% outside directors. When a firm is characterized by high information asymmetry, a board with a high proportion of outside directors will likely lack the information necessary to govern effectively. Descriptively, most boards consist of a large majority of outsiders (a simple majority of outside directors is actually required for listing on both the NYSE and NASDAQ). Linck et al. (2008) documents that in a broad sample of firms between 1990 and 2004, the typical board consisted of roughly 66% outside directors. This figure has been increasing steadily over time: in 1990, the typical percentage of outsiders was about 60%-65%, whereas by 2004, this percentage appears to have grown to about 70%. Further, as of 2004, inside directors dominate 10% of boards (roughly). Lehn et al. (2008) document a similar declining trend in the proportion of inside directors over the substantially longer time period from 1935-2000, with the sharpest decline occurring over last three decades of that sample period. In contrast to the proportion of outside directors, Linck et al. (2008) do not find much of a trend in board size over the 1990 to 2004 period. In particular, they find an average of roughly seven or eight directors per board. Over the longer 1935-2000 time period, however, Lehn et al. (2008) find that board size increased during 1935-1960 sub period, and then declined during the 1960-2000 sub period, with the ending board size in 2000 being somewhat smaller than in 1935. The growth in the proportion of outside directors over time is an interesting development for accounting researchers that warrants investigation. Specifically, why has this trend occurred, and how does this trend relate to changes in the corporate information environment over time? As noted above, outside directors require a transparent information environment to govern effectively. Therefore, it may be that the quality of the corporate

30

information environment has improved over time (due to some trend in firm characteristics, accounting standards, or other regulation that is exogenous to governance), thereby reducing information asymmetry and endogenously enabling firms to invite more outside directors to sit on the board. Alternatively, perhaps the increase in outside directors is due to regulatory pressure stemming from a belief (justified or not) that firms have too few outside directors to ensure socially optimal decision-making by boards (see McConnell, 2003). It is also possible that regulatory actions removed frictions that (for whatever reasons) prevented shareholders from instituting their desired proportion of outside directors. In either of these latter two cases, both firms and outside directors should recognize that a more transparent information environment is essential for outside directors to govern effectively, and one expects to observe a simultaneous movement by firms (and/or by regulators) toward improved financial reporting and lower information asymmetry.10 A third alternative is that the increase in outside directors stems endogenously from a widespread concern about a decline in the quality of the corporate information environment over time, and outside directors are added to the board to increase monitoring of the information environment and to actively reduce information asymmetries.11 The reasons for changes in board structure over time, and the expected simultaneous changes (or possible an unintended mismatch) in corporate information environments is an interesting area for future research.
10

In this case, of course, a reasonable question to ask is whether or not such a regulatory push toward a greater proportion of outside directors, in fact, improved the efficiency of corporate governance for either shareholders or society at large (see also McConnells, 2003 call for research on this issue). 11 Note that the three potential explanations given here for the trend toward a greater proportion of outside directors can likely be differentiated by testable predictions. For example, the first explanation suggests a positive relation between the quality of the information environment and the proportion of outside directors, but possibly with a lag. The regulatory explanation also suggests a positive relation between the quality of the information environment and the proportion of outside directors, but either contemporaneously, or a lag in the opposite direction. The regulatory explanation also suggests a positive relation between regulatory actions and the proportion of outside directors. The third explanation suggests a more complex relation between the quality of the information environment and the proportion of outside directors, with low or declining information quality followed by an increase in the proportion of outside directors, followed by an improvement in information quality.

31

In addition to temporal variation in board characteristics, there is also cross-sectional variation in board size and board independence. Linck et al. (2008) document that small firms typically have a smaller proportion of outside directors than large firms (an average of 58% vs. 73%, respectively during the 1990-2004 period), and also tend to have boards with fewer directors than large firms (about six directors vs. ten directors, respectively). The larger board size and proportion of outside directors at large firms is consistent with the prediction that large firms require a broader range of director expertise, and have a higher demand for monitoring due to greater scope for managers extracting private benefits. Interestingly, Lehn et al. (2008) find that the variation in board size across firms has declined from 1935 and 2000, which suggests that firms have converged somewhat in their preferences regarding the number of directors on the board. No similar convergence is found regarding the proportion of inside directors. A growing body of empirical literature examines the relation between firm-specific information asymmetry and board structure. In particular, several papers explore the hypothesis that when outside directors face greater information acquisition and processing costs, they are likely to be less effective advisors and monitors, and are less likely to be invited to sit on boards.12 Information acquisition and processing costs are generally predicted to increase with information asymmetry, and information asymmetry (and monitoring difficulty in general) is measured in these papers using various proxies, such as the market-to-book ratio or Tobins Q, R&D expenditures, stock-return volatility, firm size, number of analysts, and analyst forecast errors.

12

See, for example, Boone et al. (2007); Coles, Daniel and Naveen (2007); Linck et al. (2008); Lehn, Patro, and Zhoa (2008); and Cai, Liu, and Qian (2009).

32

Across a variety of research designs and samples, empirical evidence generally supports the idea that the proportion of outside directors is smaller at firms with greater information asymmetry between insiders and outsiders, and where firm-specific knowledge is likely to be most important (although we note that there are some mixed results in this literature, e.g., Boone et al., 2007). Linck et al. (2008) examine the hypothesis that outside directors face information acquisition and processing costs when transforming their general expertise into firm-specific expertise. Using a broad sample of 53,000 firm-year observations for the years 1990-2004, they document a negative cross-sectional relation between such informational costs and the proportion of outside directors (where information and processing costs are measured using market-to-book, R&D expenditures, and stock-return volatility). Cai et al. (2009) explore hypotheses similar to Linck et al. (2008) but with greater emphasis on the role of information asymmetry in shaping governance. Specifically, in addition to testing for a relation between information asymmetry and board structure, they also examine the relation between information asymmetry and both CEO equity incentives and anti-takeover provisions. Their results are generally consistent with the predictions that direct monitoring via a high proportion of outside directors is decreasing in the degree of information asymmetry between management and outside directors (as in Linck et al., 2008), but also that firms implementing lower direct monitoring choose greater indirect monitoring via greater equity incentives and fewer anti-takeover provisions. Although it is not surprising that various governance structures would complement or substitute for each other, there is relatively little research on these relations. We return to this point periodically throughout this review. Boone et al. (2007) and Lehn et al. (2008) use a time-series research design to examine the relation between information asymmetry and board structure. Lehn et al. (2008) examine a

33

sample of surviving manufacturing firms over the period 1935-2000 and document an inverse relation between various proxies for information asymmetry and the proportion of outside directors. Although such a sample obviously raises serious concerns about survivorship, it is an interesting setting in which to observe changes in governance structures over time. Boone et al. (2007) examine a sample of IPO firms, and follow the evolution of their governance structures over time. In contrast to the other papers discussed above, they do not find evidence of a relation between the proportion of outside directors and proxies for the costs such directors are expected to incur to monitor management. Ajinkya, Bhojraj, and Sengupta (2005) and Karamanou and Vafeas (2005) document that firms with a greater proportion of outside directors provide more frequent earnings forecasts. These forecasts are also more specific, accurate, and less optimistically biased when firms have a greater proportion of outside directors. These results suggest that the presence of outside directors plays a role in guiding the firms disclosure practices. What is less understood, however, is why outside directors are associated with more timely and accurate disclosures. One possibility is that such disclosures facilitate a more transparent information environment that allows outside directors to more effectively monitor and advise management. Another dimension of board structure that has received attention in the literature is the CEOs role on the board, as well as the CEOs bargaining power in influencing board characteristics. Regarding the CEOs role on the board, the most common feature examined in the literature is whether the CEO is also the Chairman of the Board (which occurs in about half of large US corporations). Brickley, Coles and Jarrell (1997) argue that the prospect of becoming the Chairman of the Board acts as an incentive mechanism for CEOs, suggesting that more successful and talented CEOs are likely to be awarded the Chairman role. A prediction

34

more closely related to this survey is that because CEOs typically have the most detailed firmspecific information, firms characterized by greater information asymmetry are more likely to allow the CEO greater control (Brickley, Coles and Linck, 1999). Linck et al. (2008), however, fail to find a significant relation between information asymmetry and the combined roles of CEO and Chairman. In addition to the CEOs role on the board, the quality of the CEO is also predicted to influence the evolution of board independence. CEOs of superior ability with a history of strong performance may develop significant bargaining power which can be used to surround themselves with loyal directors, thereby reducing the independence of the board (Hermalin and Weisbach, 1998). At the same time, shareholders may feel that more board independence is necessary to provide a check on a powerful CEO, particularly when information asymmetry has the potential to lead to agency conflicts (although whether implementation of such a board is feasible in this setting is an empirical question). The complexity and inter-relations between these hypotheses makes it difficult to construct a clear prediction of the relation between information transparency and the combined roles of CEO and Chairman. 3.1.1. Severe forms of information asymmetry: Accounting fraud and restatements A manifestation of extreme distortion and asymmetry in the information environment is the incidence of accounting fraud or accounting misstatements (which is reasonably distinct from more benign informational issues stemming from incomplete and/or untimely disclosure or financial reporting). In addition to distorting the information environment in which outside directors make decisions, accounting fraud and misstatements subject the firm to regulatory, litigation, and other costs (Karpoff, Lee and Martin, 2008). Given that inside directors are typically executives of the firm, and that executives are typically knowledgeable parties to

35

fraudulent/irregular accounting activities, inside directors are unlikely to be effective monitors of fraudulent accounting activities. Outside directors, on the other hand, seem more likely to have incentives to monitor and prevent such perverse activities, but are at an informational disadvantage vis--vis inside directors (i.e., management is unlikely to voluntarily inform outside directors that they are committing accounting fraud). However, causality may also run in the opposite direction, whereby outside directors are not invited to sit on the board (or choose not to sit on a board) unless the firm has committed to a high quality, transparent information environment where fraudulent activities are unlikely to occur. Thus, the appointment of a high quality outside director can act as a signal to external parties about the transparency of the firms information environment. We discuss this issue further in Section 3.2.7 in the context of the appointment of financial experts to the board. A natural question that arises in this setting is: What are the incentives of outside directors to monitor the disclosure and financial reporting activities of the firm? When considering the incentives of management (or inside directors), one can envision a probabilityweighted tradeoff between the expected benefits of distorting the information environment (e.g., extracting excess compensation or equity gains, hiding poor performance to maintain employment, etc.) and the associated costs (e.g., job loss, criminal or civil prosecution, destruction of firm value). With respect to outside directors, there are also expected benefits and costs from condoning (or failing to identify) distortions, although the benefits are likely quite small. A typical outside director receives a relatively modest annual fee ($50,000-$200,000), does not receive bonuses tied to accounting performance, and usually does not hold a large fraction of his or her wealth in firm stock. At the same time, most outside directors have substantial reputational capital to protect, and public detection of accounting

36

fraud/misstatements have been shown to result in significant reputational costs on outside directors.13 Therefore, with minimal benefits and substantial costs, it is difficult to see why an outside director would knowingly support management in manipulating financial information. Rather, it seems more likely that outside directors sometimes do not (because of information gathering costs) or cannot (because of managements active suppression of information) obtain sufficient information to uncover financial manipulation by management.14 The empirical evidence on the relation between board structure and the incidence of financial accounting fraud and misstatements is mixed. Beasley (1996) documents that the incidence of SEC accounting enforcement actions is lower in firms with a greater proportion of outside directors. Dechow, Sloan and Sweeney (1996), Gerety and Lehn (1997), and Farber (2005) conduct similar studies but use a matched-pair research design. Dechow et al. (1996) and Farber (2005) find a lower incidence of SEC accounting enforcement actions for firms with a greater proportion of outside directors (Dechow et al. finds no significant relation between the proportion of outside directors and board size). Gerety and Lehn (1997), however, find no significant relation between enforcement actions and the proportion of outside directors.

13

Gerety and Lehn (1997) examine the labor market for corporate directors and document that the number of other directorships held by the directors of firms charged with accounting fraud declines significantly compared to a control sample. In a similar study, but in the context of accounting restatements, Srinivasan (2005) documents that, following restatements, director turnover is greater and the number of other directorships declines (the result is even stronger for audit committee members). Finally, Fich and Shivdasani (2007) examine a sample of firms that face lawsuits alleging financial misrepresentation and find that outside directors at these firms bear costs in the form of lost directorships at other firms. These authors note, however, that there are several possible reasons for the loss of directorships. One possibility is that directors suffer reputational harm when the firm they oversee is sued for financial misrepresentation, and that other firms no longer want such tainted directors. Alternatively, directors may voluntarily leave boards of other firms to either allow more time to be spent governing the firm that faces the lawsuit, or to reduce their exposure to litigation generally by reducing the number of boards on which they sit. Regardless of which explanation (or explanations) is correct, directors appear to incur costs when they sit on boards that are sued for financial misrepresentation. 14 One implication from this line of thought is that for accounting fraud or irregularities to go undetected by outside board members requires substantial information asymmetry between insiders and outsiders. An interesting area for future research would be to examine the relation between information asymmetry and fraud or accounting irregularities, or to use the existence of fraud and irregularities to develop a measure of such information asymmetry.

37

Further, Farber (2005) finds that following an SEC enforcement action, firms appear to increase the proportion of outside directors, suggesting that the incidence of an enforcement action causes firms to improve their information transparency thereby facilitating the increase in outside directors (or possibly that firms use the appointment of outside directors as a commitment mechanism to induce improvements in information transparency). Finally, Agrawal and Chadra (2005) find no significant relation between earnings restatements and the proportion of outside directors. Beasley (1996), Dechow et al. (1996), and Agrawal and Chadra (2005) also examine the relation between accounting irregularities and combining the CEO and Chairman roles. Ex ante, the expected relation is unclear. As noted above, in settings characterized by greater information asymmetry where the CEO has substantial private information, it may be efficient to combine the CEO and Chairman roles. The combination of greater information asymmetry and greater CEO power, however, may lead to a greater propensity for accounting irregularities. Alternatively, the appointment of CEO as Chairman may be more likely when the CEO has established a reputation (or adopted commitment mechanisms) for being forthcoming with high quality information. The empirical evidence bears out these conflicting predictions, with Dechow et al. finding that SEC accounting enforcement actions are more frequent for firms where CEO is also Chairman, but Beasley and Agrawal and Chadra finding no significant relation. 3.2. Mechanisms to commit to a transparent information environment If we accept that outside directors require high quality information to perform their monitoring and advisory roles, the next question that arises is how managers commit (or are enticed by outside directors) to more fully convey their private information about the firms

38

activities and financial health. That is, because outside directors are unlikely to know precisely the extent of their information disadvantage, they require commitment mechanisms to gain comfort that the information environment is transparent. The accounting literature has identified several such commitment mechanisms, including: a commitment to report timely financial accounting information in general (e.g., earnings timeliness), as well as a commitment to report negative information in a timely manner in particular (e.g., conservative accounting), hiring a high quality auditor, inviting financially sophisticated outside directors to sit on the board, maintaining or encouraging more institutional investors or creditors as monitors, and providing executive incentive structures to resolve information-related agency problems. We discuss each of these mechanisms in further detail below. Of course, one might question whether the resolution of asymmetric information problems between inside and outside directors could instead be resolved through mechanisms that keep information private (i.e. within the firm). That is, outside directors do not necessarily need to obtain their information through public channels such as financial reports. In some respects, the private transfer of information between boards and managers would be preferable to public disclosure of information as public information can be used by competitors, regulators, and other parties to the detriment of the firm. And, certainly, a considerable volume of information is conveyed from managers to outside directors through internal reports, presentations, memos, etc. It seems unlikely, however, that outside directors rely solely on information supplied by (and filtered through) managers to obtain all information necessary to perform their duties in monitoring and advising management. As discussed above (and as will be discussed below), agency conflicts exist between managers and shareholders where managers have incentives to

39

take actions that are not in the best interests of shareholders. And, although managers are likely to be forthcoming in sharing a considerable amount of information with outside directors, they are unlikely to share information with outside board members that is detrimental to their own interests (e.g., information about bad projects, poor performance, perquisite consumption, and accounting irregularities). In other words, managers are likely to be completely forthcoming only with information that is relatively unhelpful in assisting outside directors with their monitoring activities. In light of this concern about managers selective filtering of information, it seems plausible that financial accounting systems and public disclosures do play an important role in board governance. As noted by Bushman et al. (2004), public disclosures likely have greater credibility than private communications, in part because such disclosures are subject to SEC rules and enforcement and, in the case of financial reports, to the oversight of an auditor. Further, public disclosures can be scrutinized by other external parties, such as analysts and the business press to assimilate and to uncover distortions in various pieces of information. Finally, other parties beyond boards play a role in corporate governance (e.g., blockholders, institutional investors, the market for corporate control, and executive labor markets), and these parties require public disclosures to effectively monitor the firm (these parties may sometimes interact directly with outside board members in their monitoring role). These other outside parties also require timely and accurate public information for efficient contracting and valuation purposes beyond corporate governance. We conclude from the various arguments above that the nature and extent of the role of financial reports in governance is ultimately an empirical issue. 3.2.1. Timeliness of financial reports in facilitating information for governance

40

Bushman et al. (2004) note that outside directors will demand timely information to assist them in carrying out their monitoring and advising responsibilities, and that timely financial reporting in general, and earnings in particular, has the potential to facilitate these informational needs.15 With respect to predicting the relation between earnings timeliness and board structure, however, the authors suggest that constructing a straightforward hypothesis might be more difficult than it at first appears. On one hand, the theoretical arguments above suggest that outside directors are unlikely to be effective unless they are governing a firm that has made a commitment to low information asymmetry between insiders and outsiders. Thus, one might expect to find a positive relation between the proportion of outside directors and high financial reporting quality (i.e., implying low information asymmetry). Alternatively, the authors argue that low transparency can increase the scope for agency conflicts between shareholders and managers, thereby requiring a greater proportion of outside board members to monitor management in settings characterized by low earnings timeliness. Within this second hypothesis, it is important to consider how outside board members can be effective monitors when there is low transparency. One possibility might be that low transparency is correctable and that outside directors will work to improve transparency so that they can more effectively carry out their monitoring and advising roles. If this is true, however, the negative relation between earnings timeliness and outside directors should be temporary. Possibly as a result of these conflicting hypotheses, Bushman et al. (2004) fail to find a significant relation between earnings timeliness
15

The authors measure timeliness as the extent to which current earnings capture the information set underlying contemporaneous changes in stock price. The authors acknowledge that measuring timeliness using the relation between earnings and stock returns raises the question of whether directors (and other outside monitors) could not simply observe the stock return directly and use that as a source of information to carry out its activities. They also reasonably point out, however, that earnings and its components might help directors ascertain the detailed information set underlying stock returns. Further, earnings timeliness may be a reasonable proxy for the overall informational properties of financial reports the boards would find valuable.

41

and the proportion of outside directors (they also fail to find a relation between earnings timeliness and board size). Incidentally, when researchers measure various aspects of the information environment (e.g., earnings timeliness, accounting quality, etc.), it is expected that both firm- and industryspecific effects as well as manager-specific effects influence such measures. That is, the important element of the information environment in the preceding discussion is information asymmetry between managers and outside directors, not a general difficulty in conveying relevant and reliable information about the firm. For example, even when managers are doing their best to convey their private information, a high growth firm in an uncertain business environment may still have low earnings timeliness. In other words, just because a firm has low earnings timeliness does not imply that there is substantial information asymmetry between managers and outside directors. Thus, financial accounting properties like earnings timeliness may or may not be a good proxy for the underlying asymmetry between managers and outside directors. 3.2.2. The role of conservative financial reporting in facilitating information for governance Similar to Bushman et al. (2004), Ahmed and Duellman (2007) recognize that outside directors require high-quality timely information to effectively monitor and advise managers, but that managers have better information than outside directors and also may have incentives to distort and conceal their private information. In contrast to Bushman et al.s focus on the general timeliness of earnings, Ahmed and Duellman emphasize the timeliness of bad news. It seems reasonable to view bad news as central to the informational conflict between management and outside directors. Management is typically expected to be quite forthcoming with information about good news, since such information will paint their performance in a

42

favorable light. In contrast, when there is bad news, managers are expected to be less forthcoming, both with their disclosures to outside board members and to outside parties generally (see, for example, discussions by Watts, 2003a; Ball and Shivakumar, 2005; and Kothari, Shu and Wysocki, 2009). Within the accounting literature, the term conservatism is ascribed to the property of accounting reports that requires the recognition of bad news in earnings in a more timely manner than the recognition of good news in earnings. The more timely recognition of bad news in accounting reports is operationalized through a variety of reporting rules and choices that commit managers to expend greater effort in recognizing and disclosing difficult-to-verify information about losses than gains. For example, negative changes in the value of inventory, goodwill, and other long-lived assets are recognized in a timely manner, but positive changes in difficult-to-verify values are recognized only when they become easy-to-verify (e.g., when they are realized in cash). Thus, it seems reasonable to interpret conservatism as a set of financial accounting rules and conventions that facilitate more complete and timely corporate disclosure by committing managers to report bad news sooner than it might otherwise be provided. Ahmed and Duellman (2007) measure conservatism using three variables: 1) the market-to-book ratio; 2) a three-year measure of persistent negative accounting accruals; and 3) the Basu (1997) stock-returns-based measure of the asymmetric timeliness of earnings. None of these measures, however, seems to capture the essence of the argument above that outside directors require timely information about bad news to complete their understanding of the firms operations and performance. The market-to-book ratio and three-year negative accounting accruals appear to be measures of a cumulative downward bias in earnings as opposed to information-based measures of timely financial reporting about bad news. The Basu

43

(1997) measure begins with a regression of (typically annual) earnings on contemporaneous stock returns, where stock returns are used as the proxy for news about the firm in a given period. In addition to the main effect coefficient of returns on earnings, the regression also allows the coefficient on returns to vary according to whether returns are negative. The incremental coefficient on negative returns is typically viewed as a measure of conservatism, or the asymmetric timeliness with which earnings captures bad news vis--vis good news. However, if managers have strong incentives to disclose good news about the firm, then it is unlikely that the disclosure of good news is particularly problematic to the efficacy of the monitoring and advising role of outside directors. Rather, the timely disclosure of bad news is the key informational problem. This suggests that the overall timeliness of bad news in the Basu (1997) regression is the relevant measure to study (i.e., the sum of the coefficient on returns and the coefficient on negative returns), rather than simply the incremental timeliness of bad news as compared to good news.16 As an illustration, consider two firms, one with very untimely earnings, but where bad news is reflected in earnings somewhat more quickly than good news, and another firm with very timely earnings, where both good and bad news are reflected very quickly in earnings. The first firm will have more conservative earnings according to the asymmetric timeliness measure, but it is the second firms earnings that are likely to be more helpful to outside directors in

16

More formally, the basic Basu (1997) regression is: NI = 0 + 1NEG + 2RET + 3NEG*RET + where NI is (annual) net income, RET is contemporaneous stock returns which is the proxy for news, and NEG is an indicator variable that equals one for negative returns and zero otherwise. In the most common interpretation of this regression, 2 indicates how news, in general, is reflected in NI, where a larger coefficient indicates that earnings exhibits a greater response for a given amount of news in returns. 3 indicates whether this mapping is incrementally different for bad news (i.e., the differential timeliness of bad news). When 3 > 0, bad news is reflected in net income more timely than good news, and accounting is considered to be conservative. Note that although 2 is sometimes interpreted as good news sensitivity, this labeling is misleading because other things equal, when 2 is smaller, both good news and bad news are less timely. Therefore, the overall (or absolute) timeliness of bad news in earnings (presumably what outside directors are concerned about knowing) is captured by the sum of 2 and 3.

44

fulfilling their monitoring and advising roles (see Guay and Verrecchia, 2006 for a discussion of this point). It may well be, however, that the incremental timeliness of bad news in the Basu (1997) regression is fairly highly correlated with the total timeliness of bad news recognition, particularly if timeliness of good news recognition does not vary substantially across firms. Regardless, however, we encourage researchers to more carefully consider whether their hypotheses speak to a firms commitment to asymmetric recognition of good and bad news, or rather a commitment to timely recognition of bad news. The latter seems more consistent with a wide array of economic hypotheses in the disclosure, contracting and governance literatures. Measurement issues aside, Ahmed and Duellman (2007) find that the degree of conservatism in accounting earnings is greater for firms with a higher proportion of outside directors, suggesting that timely recognition of bad news assists these directors in carrying out their monitoring and advisory roles. The authors also find that conservatism is greater when outside directors hold greater equity incentives in the firm. This latter result appears consistent with the notion that outside directors will demand even more timely information about bad news when they have stronger incentives to expend greater effort in their monitoring and advisory roles. The authors find no consistent relation between conservatism and either board size, whether the CEO is also Chairman, or whether outside directors hold a greater number of additional directorships. In a related study, LaFond and Watts (2008) note the importance of financial statements for a variety of implicit contracts between managers and shareholders, and argue that shareholders demand more timely reporting when they are concerned about managers withholding information. They document that firms with greater information asymmetry between inside and outside investors commit to greater conservatism in financial reporting.

45

They also provide evidence that changes in information asymmetry lead to a greater commitment to conservative accounting. These results appear consistent with the hypothesis that when there is greater information asymmetry between insiders and outsiders, managers and/or boards commit to reporting rules that convey timely disclosure about bad news.17 If the informational problems between insiders and outsiders stem mainly from insiders reluctance to convey timely information about bad news, a commitment to more conservative accounting naturally leads to lower information asymmetry between insiders and outsiders. An interesting and important implication of this perspective is that when one views conservative accounting simply as a commitment to more complete disclosure (as opposed to an asymmetric accounting practice for recognizing good slowly and bad news quickly), the economic benefits of conservative reporting can be extended to a broad array of contracting and governance settings as well as to a valuation setting. Further, we note that the potential for conservative accounting to assist investors in valuation is an interesting area for future research given the common perception (or possibly misperception) that although conservative accounting may assist firms in contracting and governance settings, such benefits come at the expense of accountings role in valuation. 3.2.3. The governance role of the auditing process in the information environment Broadly viewed, the auditing process is a set of mechanisms that facilitates the monitoring of the financial reporting system. As such, managers, shareholders, directors and other contracting parties all have an interest in structuring an auditing process that provides

17

As with many papers in the conservatism literature, LaFond and Watts (2008) emphasize the asymmetry between good news and bad new recognition. As noted above, however, the timely recognition of gains in accounting reports is less likely to be necessary to achieve full disclosure since managers are expected to be more forthcoming with good news irrespective of accounting rules and conventions. Thus, the asymmetry, per se, in the timeliness of gain and loss recognition may be much less important than simply the commitment to timely loss recognition.

46

external parties with confidence that the financial reporting system distributes timely, relevant and credible information about the firms current and future cash flows and their riskiness. In this subsection, we begin by briefly discussing literature on how the firms choice of auditor influences the financial reporting process. We then focus on research addressing how board structure, including structure of the audit committee, influences the financial reporting process. Auditors play a prominent role in the production and distribution of accounting information by certifying that the financial statements have been prepared in accordance with GAAP. Jensen and Meckling (1976) suggest that an audit of financial reports by an independent party is one mechanism by which shareholders can reduce agency costs that arise from information asymmetries between managers, shareholders, and other contracting parties. Specifically, because outside contracting parties recognize and penalize firms for the existence of agency conflicts and the risk of expropriation by insiders, the shareholders that bear these costs have an incentive to hire an auditor that monitors their actions and reports this information to the outside investors. Watts and Zimmerman (1983) expand on this argument by suggesting that quality/independence of the audit is a function of the probability that the auditor will uncover accounting problems, conditional on the existence of an accounting problem. A large literature documents that the choice of a high quality auditor can influence the quality of the financial reporting process (where a high quality auditor is typically considered to be a large auditor that puts substantial reputational and financial capital at risk when conducting an audit). For example, Francis, Maydew and Sparks (1999) argue that managers have incentives to choose high quality auditors as a bonding mechanism to credibly commit to constrain their ability to opportunistically manipulate the financial reporting process. They argue that managers of high accrual firms are likely to have greater opportunity to manipulate

47

financial reports, and predict that such firms are more likely to select a high quality (Big 6) auditor. Consistent with this prediction, the authors find that Big 6 auditors are more common at firms with longer operating cycles and greater capital intensity, both of which are proxies for the propensity to use accruals. They also document that Big 6 auditors are associated with lower discretionary accruals. This result is corroborated by findings in Becker, DeFond, Jiambalvo, and Subramanyam (1998). These results are intuitive and consistent with the idea that larger auditors provide higher quality audits and these audits influence the outputs of the accounting process. However, it is not clear what role self-selection may play in this research. Since the choice of auditor is clearly an endogenous decision (by the board), the research design should consider what types of firms are more likely to select high quality auditors, and what types of financial reports these firms would have had absent a high quality auditor (i.e., the counterfactual outcome). Given the importance of the endogenous choice of auditors in this setting, it seems important to carefully consider and account for the effect of self-selection. Other papers in this literature document that a high quality auditor is associated with higher earnings response coefficients (Teoh and Wong, 1993), a lower incidence of litigation (Palmrose, 1988), a lower incidence of accounting errors and irregularities (DeFond and Jiambalvo, 1991), and more informative audit reports (Weber and Willenborg, 2003). However, Dechow et al. (1996), Gerety and Lehn (1997), and Agrawal and Chadra (2005) find no relation between Big 5 (or 6 or 8) auditors and the incidence of SEC accounting enforcement actions and earnings restatements (but, Farber, 2005 finds a negative relation between Big 4 auditors and SEC enforcement actions). A more comprehensive review of this literature is beyond the scope of this survey.

48

3.2.4. Audit committee of the board of directors The audit committee is a subcommittee of the board of directors that has oversight responsibility for the financial reporting process. Outside directors on the audit committee are likely to bring more independence to the monitoring of managements financial reporting activities. Further, outside directors on the audit committee, like outside directors generally, are expected to require more transparency in the information environment to carry out their responsibilities. At the same time, regardless of their efforts, outside directors on the audit committee are unlikely to understand the firms financial reporting process as well as inside directors. Early research on the role of audit committees examined whether the existence of an audit committee played a role in monitoring fraudulent accounting activities. For example, Beasley (1996) and Gerety and Lehn (1997) do not find a relation between the existence of an audit committee of accounting fraud. Dechow, Sloan, and Sweeney (1996) find that the incidence of SEC accounting enforcement actions is lower for firms with an audit committee. In more recent research, Klein (2002a and 2002b) examine hypotheses similar to those in Bushman et al. (2004), but in the context of outside directors on the audit committee as opposed to outside directors on the overall board. With respect to the determinants of outside directors on the audit committee, Klein (2002a) predicts and finds that more complex firms and firms with greater uncertainty and growth opportunities are less likely to have outside directors on the audit committee. This result is consistent with outside directors being asked to serve on the board in settings where there is sufficient information transparency to allow the outsiders to effectively fulfill their advising and monitoring roles.

49

Conditional on having a larger proportion of outside directors on the audit committee (or an increase in the proportion of outside directors), Klein (2002b) documents lower earnings management behavior (measured using discretionary accruals). She also documents that earnings management is negatively related to the proportion of outside directors on the overall board. Krishnan (2005) documents that the proportion of outside directors on the audit committee (but not the size of the audit committee) is negatively related to the incidence of internal control problems (as publicly disclosed on the Form 8-K when a change of auditor occurred). These results are consistent with outside directors having incentives and the ability to monitor the financial reporting process and to effectively reduce earnings management activity that is not in shareholders interests.18 Further, assuming that less earnings management reduces the costs associated with outside directors becoming well-informed, these results are consistent with firms recognizing that outside directors demand more transparent financial reporting to effectively perform their job.19 However, the findings in this literature are somewhat inconsistent. For example, Farber (2005) fails to find a relation between the proportion of outside directors on the audit committee and SEC enforcement actions. Also, Karamanou and Vafeas (2005) find little evidence that the frequency or accuracy of managements earnings forecasts are associated with either (i) the proportion of outside directors on the audit committee, (ii) the proportion of audit committee members with financial expertise, or (iii) the size of the audit committee.

18

Some earnings management activities are in the interests of shareholders (e.g., avoiding the violation of debt covenants, avoiding regulatory scrutiny, etc.). Such earnings management may be acceptable to both inside and outside directors, and so would not necessarily be related to the proportion of outside directors. 19 Klein (2002b) also hypothesizes that because a large blockholder frequently retains the ability to intervene in the firms strategic decisions, such a large shareholder has incentives to ensure an information environment that is sufficiently transparent so as to remain well informed about the firms activities. She examines that relation between blockholders and earnings management, but does not find evidence of an association.

50

As an aside, Klein (2002b) indicates that her results suggest that boards structured to be more independent of the CEO are more effective in monitoring the corporate financial accounting process. This interpretation emphasizes a direction of causality running from the actions of outside directors to the accounting process; that is; outside directors actively constrain earnings management. Although this interpretation is certainly plausible, an alternative interpretation that is also consistent with the collective evidence is that management and shareholders recognize and ensure that their corporate financial reporting process is transparent when they invite more outside directors to sit on the board. This alternative interpretation emphasizes the incentives that shareholders (and potentially management) have to proactively mitigate agency conflicts that arise when financial reporting is not transparent. These two interpretations give rise to a host of interesting questions. For example, how do outside directors ensure that the financial reporting process is transparent and that information is accurate and timely? How do management and/or shareholders commit to provide outside directors with accurate and timely information (recognizing that directors may not know whether any specific piece of information is accurate or timely)? Does the relation between outside directors and financial reporting depend on whether the outside directors are actively sought out by management and shareholders (endogenously), or whether they are in some sense, imposed on the firm by some (largely) exogenous influence, such as regulatory requirements? Carcello and Neal (2000) and (2003) explore the role of the audit committee during times of financial distress. Carcello and Neal (2000) find that the probability of a financially distressed firm receiving a going concern audit report is higher when there are more outside directors on the audit committee. The authors interpret this result as evidence that inside

51

directors on the audit committee of financially distressed firms actively seek to prevent the auditor from issuing a going concern report. It is inferred, therefore, that affiliated (inside) directors on the audit committee impede the independence of the auditor and reduce the quality of accounting reports. This is an interesting inference because it suggests that outside directors take actions to protect the integrity of the financial reporting system even when it might be against the shareholders interests to do so. An alternative interpretation of this finding, however, is that insiders on the audit committee are in a better position to provide the auditor with information that reduces uncertainty about the firm's ability to return to financial health, thereby reducing the frequency of going-concern reports. In a follow-up paper, Carcello and Neal (2003) examine whether audit committee characteristics influence the likelihood that an auditor is dismissed following the issuance of a going concern report. The underlying presumption in this study is that management will attempt to dismiss an auditor who issues a going concern report, either in an attempt to find a more pliable auditor, to punish the auditor (presumably to establish a reputation that will influence the next auditor), or due to irreparable damage to the relationship between management and the auditor stemming from the conflict. The authors argue that the audit committee can be structured to commit ex ante not to dismiss the auditor, thereby encouraging the auditor to remain independent and improving the quality of the auditing process. The authors hypothesize that the following features of the audit committee reduce the probability that the auditor will be dismissed following a going concern report: 1) a greater proportion of outside directors; 2) directors that sit on a greater number of other boards (as a proxy for reputational capital); 3) directors with greater financial expertise (as a proxy for the ability of directors to understand the reasons for the going concern report), and; 4) directors holding less company stock (as a proxy

52

for the directors incentives to be concerned about the market effects of the going concern report).20 The authors empirical work is taken to support all of the predictions except the relation with financial expertise of the directors. 3.2.5. Outside directors as a bonding mechanism to mitigate agency conflicts with creditors and other contracting parties An interesting aspect of the Carcello and Neal (2000) and (2003) papers, is the notion that outside directors can be considered an ex ante commitment device to bond against taking certain actions, ex post. As noted above, shareholders can potentially increase firm value ex ante by committing to mechanisms that will maximize other stakeholders interests when decisions are made by the firm ex post. The arguments and findings in the Carcello and Neal papers are consistent with arguments by Fama and Jensen (1983) that outside directors have reputational capital that improves monitoring and constrains their willingness to take ex post opportunistic actions with respect to financial reporting that may benefit shareholders, but are potentially detrimental to other stakeholders, such as bondholders, employees, and suppliers. At the same time, inside directors, most of whom are executives of the firm, may have great difficulty convincing stakeholders that management will not distort financial reports in certain states of the world. Further, outside directors, shareholders, and other external parties are largely aligned in their demand for high quality information. For example, firms that use transparent financial reporting to credibly convey timely and reliable information to outside

20

The predictions and findings regarding the number of additional directorships and stockholdings are interesting, because of their contrast with other predictions in the literature. Specifically, additional directorships, which are also used in the literature as a proxy for busy board members, are frequently considered to impair the monitoring ability of the directors, whereas the results in Carcello and Neal (2003) are consistent with additional directorships measuring reputational capital (a seemingly reasonable alternative interpretation). Another interesting prediction regards stockholdings by directors, which are frequently argued to align directors interests with shareholders. In this paper, however, stockholdings are thought to impair the monitoring role of board members by aligning their interests with shareholders and compromising their commitment to protect the auditor from being dismissed, ex post, in the event of a going concern report.

53

directors, simultaneously convey this information to other outside investors and contracting parties. A growing literature explores the notion that certain governance structures not only control agency costs between managers and shareholders, but also control agency conflicts between managers/shareholders and other stakeholders. Several of the papers in this area explore agency conflicts with creditors, and how certain governance structures may lower debt costs by facilitating the information environment and/or constraining actions that are detrimental to bondholders. These papers generally rely on the notion that timely and credible corporate disclosure practices are expected to reduce information asymmetry between managers and creditors, and to allow creditors to accurately evaluate default risk and agency conflicts. Sengupta (1998) provides evidence consistent with this hypothesis by documenting that firms with higher disclosure quality ratings are able to borrow at lower rates. Further, Ashbaugh, Collins and LaFond (2006) document that firm credit ratings are positively related to the timeliness of earnings and accrual quality. Bhojraj and Sengupta (2003) also provide evidence on the relation between governance structures and the cost of debt. In particular, they document that firms with higher credit ratings are able to borrow at lower rates when they have a higher proportion of outside directors on the board. Anderson, Mansi and Reeb (2004) also document this relation between the cost of debt and overall board independence, as well as a negative relation between the independence of the audit committee and the cost of debt. These results are consistent with two non-mutually exclusive explanations. First, because outside directors require timely information to perform their monitoring and advising roles, the information environment is expected to be more transparent at firms where a greater proportion of outside directors are asked to sit on the board.

54

And, if a more transparent information environment is known to facilitate less costly contracting with creditors, one expects to find a negative relation between the proportion of outside directors and the cost of debt (but, importantly, this result is not necessarily causal). Second, the personal reputational concerns of outside directors provide incentives for outside directors to monitor and constrain managers ability to engage in self-interested actions. If these self-interested actions are detrimental to either firm value, or to creditors specifically, the proportion of outside directors is expected to be negatively related to the cost of debt. Ashbaugh, Collins and LaFond (2006) extend the work of Sengupta (1998) and Bhojraj and Sengupta (2003) to a broader set of financial reporting and governance variables. They document a similar positive relation between credit ratings and the proportion of outside directors on the board. They also find that credit ratings are positively related to the proportion of directors that hold company stock and the proportion of directors that hold directorships at other firms (which may proxy either for expertise, or more likely, reputational capital). However, they find no evidence that the proportion of outside directors on the audit committee is related to the cost of debt. In a related study, Anderson, Mansi and Reeb (2004) document that the cost of debt is negatively related to both board size and the size of the audit committee. However, these authors find no relation between the cost of debt and director equity ownership. 3.2.6. The governance role of active investors in facilitating the information environment Jensen (1993, p.867) argues the merits of active investors as a governance mechanism: Active investors are individuals or institutions that simultaneously hold large debt and/or equity positions in a company and actively participate in its strategic direction. Active investors are important to a well-functioning governance system because they have the financial interest and independence to view firm management and policies in an

55

unbiased way. They have the incentives to buck the system to correct problems early rather than late when the problems are obvious but difficult to correct.

To make efficient investing decisions, active investors require timely and reliable information about the firm so they can monitor managements actions and participate in the firms strategic direction. Further, as noted by Jensen (1993), active investors have the financial incentives and clout to influence managements decisions regarding the timeliness and reliability of the information that is conveyed to outsiders. Thus, one hypothesis is that information transparency and the presence of active investors are complements and the two should be positively correlated. An alternative hypothesis, articulated by Demsetz and Lehn (1985) and Bushman et al. (2004), is that effective monitors, such as active investors, are most important in settings characterized by low information transparency. This perspective views information transparency and active investors as substitutes, and leads to the hypothesis that transparency and the presence of active investors are negatively correlated. Shleifer and Vishny (1997) offer a competing view of active investors, suggesting that concentrated ownership can allow blockholders to influence management and secure private benefits that are detrimental to fragmented shareholders and creditors. And, if timely and reliable disclosures constrain the ability of blockholders to secure such private benefits, one expects to find a negative relation between blockholders and information transparency. Perhaps as a result of these conflicting effects, the empirical literature is mixed on the relation between various types of active investors and the degree of information transparency. Empirical evidence on this issue includes the following results:

56

i. Bushman et al. (2004) document a negative relation between earnings timeliness and concentrated ownership, where concentrated ownership is a composite measure incorporating institutional ownership, blockholders, and average holdings per shareholder. This result is consistent with earnings timeliness and concentrated ownership being substitute-monitoring mechanisms. ii. Ajinkya, Bhojraj, and Sengupta (2005) also find a negative relation between concentrated institutional ownership and information transparency, as measured by the frequency of voluntary earnings forecasts. Their argument, however, is not that active investors provide a substitute monitoring function, but rather that concentrated institutional owners do not require transparent disclosure because they have access to private information. iii. Dechow, Sloan and Sweeney (1996) and Farber (2005) find that SEC accounting enforcement actions are less frequent for firms with outside blockholders. However, Beasley (1996) and Agrawal and Chadra (2005) find no significant relation between blockholders and SEC enforcement actions and earnings restatements, respectively. iv. Ajinkya, Bhojraj, and Sengupta (2005) and Karamanou and Vafeas (2005) document that greater institutional holdings are associated with more frequent voluntary earnings forecasts and more accurate forecasts. However, Farber (2005) finds no relation between institutional holdings and SEC enforcement actions. v. Bhojraj and Sengupta (2003) find that institutional investor ownership is associated with lower interest rates and higher credit ratings, a result interpreted as institutions monitoring both the information environment and managers self-interested actions.

57

However, Ashbaugh, Collins and LaFond (2006) find no relation between credit ratings and institutional ownership. vi. Bhojraj and Sengupta (2003) find that concentrated institutional ownership (as opposed to disperse institutional ownership) is associated with higher borrowing rates. vii. Ashbaugh, Collins and LaFond (2006) document that blockholders are negatively related to credit ratings. The wide array of predictions and findings in this literature do not lead to clear inference on the relation between active investors and the information environment. We suggest further research be conducted in this area to flesh out the economic forces at play. We also use this set of findings to highlight the importance of considering the full set of results on a particular economic hypothesis. Generally speaking, it is often the case that (likely for the sake of brevity), authors highlight their significant findings, but may not spend much, if any discussion on their insignificant results. To draw proper inferences within any field of research requires the consideration of both significant and insignificant results. For example, if ten authors test a given hypothesis, and only one or two find results consistent with the hypothesis (and the others find no results), what should one conclude from this line of research? We do not attempt to answer this question here, but suggest that it is important that thought be given to how nonresults should be factored into inferences when accumulating results within a literature. 3.2.7. The governance role of financial expert directors in facilitating the information environment As discussed above, the board of directors requires timely and reliable information about the firms activities to effectively carry out its monitoring and advising roles. Directors, however, are not homogeneous with respect to their expertise, and therefore are unlikely to be

58

homogeneous with respect to their informational needs. In the wake of several high-profile accounting scandals in the early 2000s and the passage of disclosure rules in the 2002 SarbanesOxley Act (SOX), the role of financial experts on the board of directors became a topical issue. Financial expert directors, it is thought, should have different capabilities vis--vis nonfinancial expert directors with respect to monitoring and advising on financial reporting and disclosure issues. We are aware of no well-accepted definition of a financial expert director (at least within the academic community).21 It seems intuitive that a director with a background in public accounting, auditing, or financial operations (e.g., CFO, controller, etc.) would be considered a financial expert. However, SOX uses a broader definition of how a director can obtain financial expertise that includes, for example, experience in managing individuals that carry out financial reporting and financial operations. As a result, the SOX definition of financial expert includes individuals such as CEOs and company presidents, who may not have detailed skills in analyzing financial reports or accounting practices. Absent regulatory requirements, a firm will presumably invite a financial expert to sit on the board for one of two reasons: 1) when management either requires advice with respect to corporate finance or financial reporting strategy, or; 2) when management wants to commit to more intense monitoring of corporate finance or financial reporting strategies that a financial expert director will carry out in his/her role on the board (or when shareholders actively require management to bring an expert onto the board due to concerns about insufficient monitoring). In the former case, the outside financial expert director will only be able to perform this

21

Regulation S-K Item 401 defines an audit committee financial expert as a person with various attributes, including an understanding of GAAP and financial statements, experience in preparing, auditing, or analyzing financial statements, an understanding of internal control over financial reporting, and an understanding of audit committee functions.

59

advisory role if the firms financial reporting and information environment are transparent. Thus, one might expect a positive relation between information transparency and financial expert directors on the board. In the latter case, the outside financial expert director may be asked to sit on the board when the firm believes that its financial reporting and information environment are not transparent, but that additional monitoring and advice from the financial expert will make it more transparent. In this case, one might expect to initially observe a negative relation between information transparency and financial expert directors on the board. Over time, however, information transparency will improve as the financial expert carries out his/her duties such that in a cross-sectional test one could find a negative relation, a positive relation, or no relation between information transparency and financial experts. Empirical research on these hypotheses is mixed, but generally supports a positive relation between information transparency and financial experts on the board. Xie, Davidson, and DaDalt (2003) find that board and audit committee members with corporate or financial expertise are associated with lower earnings management (as measured by discretionary current accruals). Agrawal and Chadha (2005) find that the frequency of an earnings restatement is lower in companies with an outside financial expert director on either the board or the audit committee. Further, Farber (2005) finds that firms subject to an SEC enforcement action have fewer financial experts on their audit committees than a group of control firms. In related work, Krishnan (2005) documents that the proportion of audit committee members with financial expertise is negatively related to incidence of internal control problems. There are, however, several studies that find no results for this relation. In particular, Karamanou and Vafeas (2005) do not find a consistent relation between the proportions of financial experts on the audit committee and the frequency, accuracy or bias in management earnings forecasts. Ashbaugh,

60

Collins and LaFond (2006) find no evidence that a financial expert on the audit committee is associated with credit ratings, and Anderson, Mansi, and Reeb (2004) find no relation between the cost of debt and financial experts serving on the audit committee. Finally, Carcello et al. (2003) find no significant relation between the proportion of audit committee members with financial expertise and the frequency of auditor dismissals following going-concern reports. The invitation to have a financial expert sit on the board can also act as a signal that the firm believes its corporate finance and financial reporting practices are of high quality (i.e., a financial expert director will presumably investigate the firms financial reporting practices prior to joining the board, and will only accept the invitation to sit on the board if such practices are of acceptable quality). In addition (or simultaneously), the invitation to have a financial expert sit on the board can act as a signal that the firm is serious about improving corporate finance or financial reporting practices, and is actively seeking advice and monitoring to achieve such improvements (see DeFond, Hann, and Hu, 2005, and Engel, 2005 for discussions of these points). DeFond, Hann, and Hu (2005) examine this signaling hypothesis and find that the stock market reacts positively when a director with accounting expertise is appointed to the audit committee (this result does not hold for the non-accounting experts that meet the broader SOX definition of a financial expert). 3.2.8. The role of management incentives in facilitating the information environment Another tool that boards have at their disposal to ensure a transparent information environment is managerial incentive structures. That is, they can provide managers with incentives to provide reliable and timely information about their activities. One approach is for boards to commit to remove executives from office in the event that they are shown to have distorted the information environment. And, on this point, a number of

61

researchers have examined senior executive turnover as a penalty for engaging in accounting irregularities. The motivation for these studies is similar to that in the studies discussed above that examine whether director turnover and related labor market penalties follow accounting irregularities. Early research by Beneish (1999) and Agrawal, Jaffe and Karpoff (1999) generally failed to find an increase in turnover frequency following accounting irregularities, specifically GAAP violations and the revelation of corporate fraud, respectively. In a more recent study, Desai, Hogan and Wilkins (2006) examine the frequency of managerial turnover following accounting restatements. They find that 60% of the firms in their restatement sample experienced turnover of at least one top manager within the 24 months following the restatement compared to 35% among their year-size-industry-matched counterparts (they also find that such managers bear poorer subsequent employment prospects). They conclude that corporate boards and the labor market impose significant discipline on financial reporting. An interesting area for future research in this literature would be to construct sharper hypotheses of types of earnings management that influence turnover decisions. For example, perhaps it is the case that there is an increased frequency of CEO turnover following earnings management that was detrimental to the firms shareholders, but no change in frequency for earnings management that was beneficial to the firms shareholders (e.g., earnings management to avoid violating covenants in an attempt to buy more time from creditors, or earnings management to meet analysts forecasts). Another incentive-based approach to promoting a transparent information environment is to tie executive compensation to performance measures that are sensitive to the quality of the information environment. For example, Nagar, Nanda, and Wysocki (2003) argue that the stock price is increasing in the quality of the information environment, and therefore providing

62

executives with equity-based incentives should promote a commitment to high quality disclosure. They find a positive relation between equity incentives and the frequency of management earnings forecasts and AIMR analysts rankings of firms disclosure practices. As we discuss in much more detail below, however, there is considerable debate among researchers about whether equity incentives encourage executives to improve or instead distort the quality of accounting information. 3.2.9. Discussion of mechanisms to commit to a transparent information environment The discussion above supports the notion that outside directors require a transparent information environment to effectively carry out their monitoring and advising roles. There is also ample support that firms, boards, and managers employ a variety of mechanisms to commit to information transparency. We know relatively little, however, about how firms select among the various mechanisms available to them, as well as how the various mechanisms interact and serve as complements and/or substitutes for each other. For example, consider a firm that wants to increase the proportion of outside directors on the board. Recognizing that information transparency is critical to such directors performing their roles, the firm decides to implement certain mechanisms to commit to lower information asymmetry between managers and outsiders. Which mechanism or combinations of mechanisms should they choose? Do some mechanisms work better (or worse) when implemented together (i.e., are there interactive effects)? We recommend more research into the relations among these mechanisms, and to illustrate the types of directions we envision such research might take we now discuss an example of links that have been identified in the literature related to information asymmetry, conservative financial reporting, and executive equity incentives.

63

We begin by noting the positive relation documented between information asymmetry and conservative financial reporting (LaFond and Watts, 2008), a result consistent with firms committing to timelier reporting of bad news when they suffer from low information transparency. At the same time, LaFond and Roychowdhury (2008) document that conservative reporting and CEO equity incentives are substitute monitoring-mechanisms. The idea here is presumably that when a firm suffers from low information transparency and cannot employ (or chooses not to employ) more conservative reporting that would reduce information asymmetry and thereby allow more direct monitoring by the board, then the firm resorts to indirect monitoring via equity incentives. This result is similar conceptually and empirically to Bushman et al.s (2004) finding that earnings timeliness is negatively related to the equity incentives of both executives and outside directors. In fact, if variation in earnings timeliness is driven largely by timely recognition of bad news, then the result in LaFond and Roychowdhury may simply be a re-characterization of the Bushman et al. (2004) result. The results in LaFond and Watts (2008) and LaFond and Roychowdhury (2008) seem closely related to Cai et al.s (2009) finding that information asymmetry is positively associated with executive equity incentives. This positive relation is consistent with firms choosing equity incentives as an indirect monitoring mechanism when information asymmetry makes direct monitoring more difficult. Collectively, these results suggest that while both conservatism and equity incentives may be employed in settings characterized by greater information asymmetry, firms with a given level of information asymmetry tradeoff the costs and benefits of using conservative reporting and equity incentives to resolve informational problems. However, an explicit test of how and when conservatism resolves information asymmetry would be useful for advancing this literature.

64

This discussion only scratches the surface of the links between various mechanisms that firms use to commit to a transparent information environment, as well as links with mechanisms firms use to control agency conflicts when they are not able to commit to such information transparency. For example, reflecting on the Bushman et al. (2004) and LaFond and Roychowdury (2008) papers, one might ask how firms choose between a commitment to overall timely earnings (i.e., timely good and bad news recognition) versus a commitment to just timely bad recognition. It is also reasonable to ask whether this is even a choice. For example, do some firms operate in business environments that simply do not allow for timely recognition of gains or losses? We also note that some other links between various commitment mechanisms have begun to be identified. For example, Hui, Matsunaga, and Morse (2009) document that a commitment to more frequent and timely earnings forecasts is a substitute for conservative financial reporting. Bushman et al. (2004) find that the presence of active investors is a substitute for the timely reporting of earnings. A more complete understanding of the tradeoffs among these various mechanisms, as well as among alternative monitoring mechanisms when information transparency is not achievable (e.g., equity incentives, monitoring by creditors, regulatory monitoring, etc.) could significantly advance the literature. 3.3. The use of accounting information in assessing management The previous section discussed the relation between board structure and corporate financial reporting and disclosure, and emphasized that outside directors will be more effective carrying out their monitoring and advising roles when there is a transparent information environment. In this section, we focus on how boards and other contracting parties use their available information to guide, assess, and reward the actions and performance of senior

65

management. Board activities in this regard include designing incentive compensation contracts to attract, retain, and motivate managers, deciding when and what type of senior executives to hire and fire, and guiding the strategic operating, investing and financing activities of the firm. A large body of research examines the role of financial (and to a lesser extent nonfinancial) information and various accounting measures in the design of executive compensation contracts, incentives structures, and turnover. This contract design literature has its intellectual roots in the information economics and mechanism design literatures, and has produced a rich set of theoretical models that have guided empirical researchers. Lambert (2001) and Bushman and Smith (2001) collectively provide a comprehensive review of the literature related to the theoretical and empirical role of accounting information in the executive compensation. We therefore focus our discussion of this area of the literature on advancements since the time of those reviews, as well as opportunities for future research, and confine our introductory discussion to those earlier theoretical results that are essential for understanding recent theoretical and empirical developments. As in the previous section on board structure, we also spend considerable time in this section discussing the role of implicit contracting in the relationship between boards and executives. Among other motivations for this discussion, we note that a minority of CEOs have formal employment agreements with their employers, and even those that do have explicit agreements, only a small proportion of the issues related to the CEOs duties, responsibilities compensation, and dismissal are articulated in such agreements (Gillan, Hartzell, and Parrino, 2009). And, as with explicit contracts, accounting information potentially plays a role in resolving agency conflicts within the broad set of implicit contractual relations among executives, boards, and shareholders.

66

We also highlight a rich set of predictions and issues related to the positive accounting theory championed by Watts and Zimmerman (1978, 1986). In particular, much of the existing positive accounting research on executive compensation focuses on executive incentives to influence accounting reports in the presence of explicit contracts where the payoff to the manager is a mechanical function of the reported accounting numbers (e.g., accounting-based bonus plans). Our discussion of implicit contracts emphasizes that, in many cases, accounting reports do not have a mechanical or formulaic influence on compensation, but rather that boards use their subjective assessment of accounting information (as opposed to simply accounting numbers) in many contracting agreements with executives. As a topical example, a significant body of research argues that by lowering accounting earnings, conservative financial reporting serves to improve contracting efficiency by constraining managers ability to reap excessive accounting-based bonuses (e.g., Watts, 2003a, b). Irrespective of the merits of this argument, we conjecture that conservative financial reporting can improve the contracting efficiency with executives by providing boards (as well as other monitors, such as creditors, blockholders and the capital markets) with more timely information about projects that generate losses. That is by providing boards with timely information about losses that managers might not otherwise voluntarily provide, financial reports give boards a more complete information set with which to infer and evaluate the executives performance, and to make adjustments to investing and operating activities as well as decisions regarding changes to management. 3.3.1. Overview of agency theoretic incentive contracting results Both Lambert (2001) and Bushman and Smith (2001) adopt an agency perspective and view incentive compensation contracts as a key corporate governance mechanism used to align the interests of managers with those of shareholders. Lambert (2001) identifies the four most

67

common sources of divergence between the objective functions of the contracting parties: (i) effort aversion by the agent, (ii) diversion of resources by the agent for private consumption, (iii) differential risk aversion between principal and agent, and (iv) differential time horizons of the principal and agent. It is typically infeasible, however, for the principal to fully resolve these agency conflicts (i.e., the first-best solution). This stems, in large part, from information asymmetry between the principal and agent with respect to the agents actions and/or type. That is, if the principal had full knowledge of the agents actions and type, many of these agency conflicts could be resolved by the principals careful use of available contract designs to hire good managers, and to require/induce managers to take value-maximizing actions. As discussed in detail in the prior sections, however, boards and shareholders do not have complete information about managers actions and types. As a result, the board uses available compensation contract designs, performance measures, and other information to attract appropriate managers, and to provide the managers with incentives to take valuemaximizing actions. However, because there are typically no perfect contract designs or performance measures to fully resolve all of the informational problems among shareholders, boards, and managers, the divergence between the principals and agents objectives is not completely resolved. The literature refers to this contracting outcome as the second-best solution, with the recognition that contracts result in lower welfare than the nirvana of a first-best contract. The second-best outcome is therefore the typical benchmark against which the efficiency achieved by alternative contracts (or mechanisms) is compared. The notion of the second-best outcome is also useful because it illustrates that the expected utilities of the contracting parties can be increased when additional performance measures or other contract

68

design features bring new information that the principal can use to attract more appropriate managers or to better align incentives. Most theoretical agency research in accounting adopts a moral hazard perspective where the principal (i.e., the board acting on behalf of the firms shareholders) designs an incentive contract to induce the agent (e.g., the CEO) to take action(s) that affect firm value. Bushman and Smith (2001) identify three roles for accounting information in these models: (i) creating incentives for the agent to take actions, (ii) filtering noise from other performance measures (e.g., stock price) and (iii) balancing managerial effort across multiple activities. One distinguishing feature of a pure moral hazard model is that the contracting parties have homogeneous information at the inception of the contract, so the agents actions can be perfectly anticipated by the principal. The notion of homogeneous information between executives (agents) and directors (principal) in these moral hazard models is interesting in light of our discussion in the prior section where information asymmetry between executives and outside directors was a key problem in reconciling governance-related agency conflicts. In particular, an assumption of homogenous information between principals and agents appears to abstract away from the issue of whether directors have effective mechanisms at their disposal (e.g., high quality financial reporting, high quality auditors and audit committee, active investors, etc.) to elicit more complete information from executives and, in particular, information that executives would otherwise not want to reveal to the board. Adverse selection models relax the homogeneous information assumption and allow one or both of the contracting parties to possess private information at some stage of the contracting process. In these models, the compensation contract becomes a mechanism with which the principal can induce the agent to reveal (at least in part) his private information through his

69

contract choice.22 Thus, to the extent that financial reporting can reduce information asymmetry between boards and managers at various points in the contracting process, adverse selection problems may be reduced. To help frame our discussion of the important theoretical incentive contracting results, we provide a stylized depiction of the main features and timing of the contracting process in Figure 1. We note that in order to maintain tractability, any given model contains only a subset of the features depicted in the figure. We also note, consistent with most models in this literature, that Figure 1 depicts a single-period game between the principal and agent. Figure 1 begins with the possibility that there is adverse selection when the agent and principal meet. That is, the manager has private information (denoted ) about some contractrelevant information such as his risk tolerance, talent at identifying profitable projects, or time horizon. Next, the principal and agent negotiate a contract (which is typically explicit in singleperiod settings; below we discuss implicit contracts in the context of multiple-period settings). In addition to adverse selection, the principal also recognizes that the agents action(s) during the contract will affect firm value, but cannot be directly observed, and that the agents interests are not aligned with those of the principal (i.e., that there is moral hazard). To mitigate the adverse selection problem, the principal typically designs a menu of contracts from which the agent chooses, with the objective being to elicit the agents private information through his contract selection. To mitigate the moral hazard problem, the principal first identifies various performance measures, such as accounting earnings and stock price that are sensitive to the agents productive actions. The general idea is for the principal to select performance measures that are sensitive to the agents actions and are also closely related to the outcome that the
22

Since the principals problem is to induce the agent to take an unobservable action in a moral hazard model and to induce the agent to reveal his private information in an adverse selection model, these models are sometimes referred to as models of hidden action and hidden information, respectively.

70

principal is trying to achieve, such as maximizing firm value (referred to as congruence, and denoted as in Figure 1). The principals problem in selecting performance measures to resolve agency conflicts is more difficult than it might at first seem. This is because even well selected performance measures reflect the agents actions with noise, and also because these performance measures capture the effect of the agents productive action on firm value with noise. The principal, however, may have other tools at her disposal to mitigate some of these problems. Following the agents actions, the principal may be able to observe a signal 1 that is not affected by the agents actions, but is affected by a noise term 1, which also affects the contractual performance measures. This common source of noise in the signal 1 will have contracting value if it can be used to filter noise from the performance measure used in the compensation contract. The principal may also be able to observe a signal 2 that is again not related to the agents actions, but does affect firm value (e.g., the effects of a natural disaster, war, or other uncontrollable crises). As with 1, 2 can be used to remove noise from the compensation contract. Figure 1 also illustrates how the market values the firm using the realized performance measures, the outcome from the compensation contract, plus some noisy signal about firm value, labeled Other Information. In the case where some elements of the principals compensation contract with the agent are unknown, the market can also learn from the observed contractual outcome. In addition, in the case of adverse selection where the agent chooses from a menu of contracts, the market (in addition to the principal) can learn about the agents private

71

information from the agents choice of compensation contracts.23 Although the market in models of executive incentive compensation is typically the equity market, as we discuss below, it includes other markets as well, such as the market for the firms debt and the product markets in which the firm competes. Further, note that after the market determines firm value, firm value itself becomes a performance measure that the principal and agency could (and nearly always do) use in the contract. To settle up on the compensation contract, the principal (and agent) observe the performance measures (e.g., stock price, accounting earnings, cash flows, revenues, etc.), and feed them into the compensation contract according to their agreed upon pay-for-performance sensitivity. The realized Contractual Outcome is therefore computed according to the aggregation and weighting of performance measures that is specified in the contract. The range of contractual outcomes is broad and includes not only the agents periodic remuneration, but also promotion and termination decisions. In summary, Figure 1 provides a brief overview of the compensation process, which we discuss in more detail in Appendix B for the interested reader. 3.3.2. Types of performance measures in executive compensation contracts Although shareholders and boards can observe some of the actions by executives, many actions are unobservable (or at least are prohibitively costly to observe). This is true even after the board takes many of the actions discussed above related to facilitating a transparent information environment. Perhaps even more importantly, however, even if shareholders and board could observe all of the actions of executives, it is unlikely that they would be able to determine whether the executives choose the best actions from among the set of possible
23

As an example, consider a manager who chooses a relatively high powered contract that consists primarily of out-of-the-money options over a relatively low powered contract that has a large fixed salary component. In this case, the market can reasonably infer that the executive is relatively risk tolerant.

72

actions (where best means the actions that maximize value to shareholders). Given this dilemma, boards use various performance measures to infer the executives actions, as well as the appropriateness of the executives actions. By conditioning the executives compensation (or more broadly, the executives wealth) on these performance signals, the board attempts to align the executives incentives and actions with the interests of shareholders. Thus, one obvious starting point for compensation research is to understand which performance measures firms typically use in executive compensation contracts, the weights assigned to such measures, and how the measures and weights vary across firms and economic settings. A key difficulty with this approach, however, is that because the details of actual compensation contracts are often not observable (e.g., most executives do not even have written employment contracts), the researcher typically lacks knowledge of both the performance measures that are used in explicit and implicit compensation arrangements, and the weights assigned to these measures. As a result, researchers are generally constrained to examine performance measures that are readily observable and measureable to the researchers (e.g., income, sales, cash flow, stock returns, etc.). A typical empirical study in this literature might estimate pooled cross-sectional regressions of contractual outcomes (such as the level of annual pay or CEO termination) on realized values of performance measures that are thought to be included in the contract, with theory-driven predictions on the relative importance of certain performance measures in various settings (e.g., more or less weight will be placed on a given performance measure as a function of its statistical or economic properties, or as a function of firm characteristics or contracting demands). Early empirical work on performance measures in executive compensation contracts documented that both earnings and stock returns are useful performance measures in explaining

73

the observed level of executives cash pay, and that the weight assigned to each of these measures is increasing in its precision (Lambert and Larcker, 1987). Over time, researchers have advanced this literature by documenting that executive annual cash pay and annual total pay (i.e., cash pay plus grants of equity and other long-term pay) co-vary with many additional financial and non-financial performance measures, and that the use of particular performance measures depends upon the economic characteristics of the firms, executives, and other features of the contracting environment. Lambert (2001) and Bushman and Smith (2001) review the theoretical and empirical work, respectively, in this literature in great detail. We, therefore, focus our attention on some of the theory and empirical work in this area since the time of their reviews. 3.3.2.1. Multiperiod compensation contracting Most empirical studies of the determinants of executive pay are inspired by predictions from single period agency models (as in Figure 1). However, as noted by Demski and Sappington (1999), executive compensation contracts are multiperiod arrangements where the executive is responsible for taking multiple actions at different points in time, and where these actions manifest in output (and the performance measures) at various points in time. Extending the theoretical work in this literature into a (more realistic) multiperiod setting raises a host of interesting issues and questions. In particular, multiperiod contracting settings introduce at least two issues in the context of explicit contracts that cannot be addressed in a single period setting.24 First, as noted by Lambert (2001, pp. 9 and 77), multiperiod models are necessary for any discussion of the role of accrual accounting in contracting because cash flows are equivalent to net income in a single period model. This observation also encompasses the role

24

In this section, we introduce the multiperiod contracts in the more familiar territory of explicit contracts. In Section 3.3.3 below we discuss the more interesting setting (at least in our view) of implicit multiperiod contracts.

74

of conservative accounting numbers in the contracting process because these are the product of accrual accounting. Second, as noted by Dutta and Reichelstein (2003), certain single-period contracting results such as Holmstroms (1979) informativeness principle (i.e., that any costless signal that is informative about, or sensitive to an agents actions is useful for contracting) are not necessarily immediate in a multiperiod setting. In a multiperiod setting, leading indicators of managerial performance such as stock price and accrual accounting numbers convey information about managerial actions and decisions (e.g., investment policy) at an early stage, but eventually become supplanted by the actual results at a later stage (e.g., realized cash flows from an investment), which can also be used as a performance measure. Both types of performance measures are likely to have advantages: leading performance measures reflect timely expectations of the market or accounting system as to how the managers actions will affect cash flows and firm value, whereas trailing performance measures reflect the actual result of the managers actions. Although both types of performance measures capture the managers actions with error, the types of error are likely to be different. Further, differences in the time horizons and risk preferences of principals and agents, as well as differences in the costs of contracting over leading vs. trailing information, may also effect the tradeoffs to contracting over a leading performance measure as compared to a trailing performance measure. There is an emerging theoretical literature on the role of leading performance measures in general, and accrual accounting information in particular, in managerial performance evaluation. For reasons discussed above, the papers in this literature necessarily involve multiperiod contracting settings and are generally concerned with how compensation contracts (optimally) use accrual accounting information (among other performance measures) to provide

75

the agent with incentives to undertake an investment in the current period when the payoff is realized in a future period. Accrual accounting can generate investment incentives for the agent because it can match the future revenues generated by an investment with the cash expenditure on the investment in the current period and can help offset the agents (typically) lower discount rate. In addition, this work has been conducted almost exclusively within the so called LEN framework which derives its name from (i) the constraint that requires the compensation contract to be linear, (ii) the assumption of negative exponential utility for the agent (which exhibits constant absolute risk aversion (CARA) which, in turn, implies that the agents wealth does not affect his risk aversion and therefore does not affect his incentives), and (iii) the assumption of normally distributed performance measures (e.g., stock price).25 Although the LEN framework considerably improves the tractability of most contracting problems, this does not come without costs which we discuss in greater detail below. One of the early studies in this literature is Dikolli (2001) who examines a quasi twoperiod agency model in which the principal must motivate the agent to take both a short- and farsighted productive action that affect firm value in the first and second periods, respectively.26 The agent receives a payout at two different points in time and discounts the later payment more heavily than the principal (which is interpreted as the agent having a shorter time horizon than the principal). The principals problem is therefore to motivate the agent to exert personally costly effort on the farsighted action using the three performance measures that are available for contracting: (i) a short-term trailing measure of the first period outcome that is
25

Lambert (2001) provides an excellent discussion of the underlying features of the LEN framework in Section 3.1 of his survey. 26 Dikollis (2001) model is quasi two-period in the sense that it posses elements of both a multi-period and a single-period model. In particular, the agent receives compensation at two different times and discounts the compensation received later in time which is a characteristic of multi-period models. However, the setting can be viewed as a single period since there is only one compensation contract and both of the agents actions are made simultaneously during the first period (but their productive effects are realized at different times).

76

measured at the end of the first period, (ii) a short-term forward-looking measure of the second period outcome that is measured at the end of the first period, and (iii) a long-term trailing measure of the second period outcome that is measured at the end of the second period. The two trailing measures are interpreted as accounting profitability and the forward-looking measure (which can be interpreted as non-accounting information, such as customer satisfaction surveys, or accounting information such as accruals), which noisily captures future value creation. Consistent with the informativeness principle of Holmstrom (1979), Dikolli finds that the principal can benefit from using the current but noisy indicator of future value (i.e., the shortterm farsighted measure) in addition to the long-term trailing measure to reward farsighted effort exerted by the agent. In addition, the optimal contracting weights depend on the agents employment horizon so that the contract will place more weight on the forward-looking performance measures the shorter the managers horizon. Dutta and Rechelstein (2003) extend Dikollis (2001) analysis to a true two period setting. In their model the principal faces a double-moral hazard problem of motivating the agent (who is assumed to have a different horizon than the principal) to take a productive action in each period as well as to undertake a soft investment in the first period for which the payoff is not realized until the end of the second period. The soft investment is not directly contractible, but its value is captured with noise by a verifiable and therefore contractible leading performance measure, such as accruals, that match the first periods investment expenditure with the second periods cash flows. The performance measures available for contracting are the realized cash flow in each period of the two periods and the leading performance indicator. The leading performance indicator can be viewed as an accrual that

77

matches the cash flows that are realized from the investment in the second period with the investment expenditures that are incurred in the first period. Dutta and Reichelstein (2003) find that the optimal weight on the leading indicator relative to the weight on cash flows depends on whether the principal offers a long-term (i.e., two period) contract or a sequence of short-term (i.e., one period) contracts. In the case where long-term commitment is possible, the leading indicator is valuable if and only if the underlying moral hazard problem differs across both periods (i.e., it is not stationary). However, absent long-term commitment, the leading indicator becomes the only instrument for motivating the manager to undertake any investment since it acts as an instrument that allows the principal separate the investment problem from the periodic moral hazard problem. Thus, in the shortterm contracting setting, the leading indicator is crucial because it is the only way for the principal to generate investment incentives. Dutta and Reichelstein (2005) extend their earlier analysis by incorporating a stock market (which provides a perfect assessment of investment returns at each point in time) in addition to an accrual accounting system. In this setting, the principal now has three performance measures available to provide the manager with incentives to undertake an investment: (i) realized cash flows in each period, (ii) accrual accounting earnings (which potentially shields the manager from current investment expenditures by capitalizing them) and (iii) a forward looking stock price (which is available at the end of the first period). They show that optimal performance measure will rely on all three sources of information.27 In particular, the weight on current market price relative to the weight on current income (i.e., the sum of

27

The optimal contract will not rely on stock price alone to generate investment incentives since the market price reflects all value relevant components of investment returns; even those that are beyond the managers control. This is similar to the tension found in models of the relation between the stewardship and valuation roles of accounting earnings, except for stock price.

78

cash flows plus accruals) reflects the tradeoff in balancing the measurement error inherent in market price (from investors inability to perfectly identify investment expenditures) and the variability of future cash flows reflected in the current market price. Their results imply that the optimal weight on price relative to accounting earnings should be decreasing in the volatility of the investment return and increasing in the noise in the accounting system. The findings of Dutta and Reichelstein (2005) are also broadly consistent with the empirical results of Bushman et al. (2004) who find that firms place greater value on equity-based incentives as accrual accounting information becomes less reliable. One concern with their setup, however, is that the contract is constrained to be a (linear) function of aggregate accounting earnings, which is what makes the capitalization of expenditures valuable from a contracting perspective. It is unclear whether an equivalent benefit of accrual accounting would obtain if the principal and agent were able to contract over a more general contract space that included the disaggregated components of earnings. Dutta and Zhang (2002) examine a multiperiod setting to study the role of revenue recognition from a stewardship perspective by examining how various revenue recognition principles (e.g., historical cost, lower of cost or market, and mark-to-market accounting) affect the incentive properties of performance measures based on aggregate accounting data (e.g., net income). They show that when revenues are recognized when realized (i.e. the realization principle), residual income provides optimal effort and productive incentives (within the class of linear incentive schemes), where optimality is defined relative to compensation schemes based on disaggregated accounting information. They also show that mark-to-market accounting generally fails to provide efficient aggregation of raw information for stewardship purposes for two reasons. First, since the manager possesses private information, market-to-

79

market accounting, which is based on only public information, fails to align the managers incentives since it is based on the anticipated performance of the manager and does not induce the manager to take the expected action.28 Second, although mark-to-market accounting aggregates future expected cash flows using the firms true cost of capital, this is generally not the optimal discount rate from an incentive alignment perspective. As noted above, most of the recent theoretical accounting research on incentive compensation has adopted the LEN framework which has enabled researchers to explore a number of interesting, but otherwise difficult to analyze contracting problems, such as the use of multiple performance measures and multiperiod contracting settings. The enhanced tractability afforded by the LEN framework however is not without costs. One of these costs is that it precludes the analysis of nonlinear contracts, such as stock options, bonus contracts with performance thresholds, and termination and promotion decisions, as well as managerial wealth effects. Further, its assumption of normally distributed outcomes, where the agents action represents a mean shift in outcome (but with exogenous variance), makes it unsuitable for analyzing managerial risk-taking incentives which are often thought to be a central feature of incentive contracting. In addition, Hemmer (2004, 2007) raises a more fundamental concern that by restricting the contract space to consist of only linear contracts, the LEN framework precludes the researcher from identifying the second-best contract from the outset. Instead, the researcher must be content to settle for the third-best contract, which might entail a substantial welfare loss relative to the second-best outcome, and is therefore not expected to be

28

This notion of anticipated versus delivered performance is similar to the effect noted by Barclay, Gode, and Kothari (2005) who argue that the forward-looking property of stock price can be detrimental for incentive purposes because it reflects the markets expectations about actions that are expected to be (but have not yet been) taken by the manager. Mark-to-market accounting numbers inherit this feature since they are based on market values.

80

an equilibrium outcome.29 Lambert (2001, p.29) echoes this concern since agency theorys intellectual roots lie in information economics, where information systems are based on the optimal use of information generated by the system. Restricting the contract to be linear is a very significant philosophical departure because in single-period models linear contracts are rarely optimal. In order to more fully appreciate these concerns, it is important to remember the generality of the standard principal agent model. In particular, since Holmstrom (1979), it is well known that the optimal second-best contract in a general moral hazard setup is characterized by the following equation:

where x is the output, s(x) is the compensation contract as a function of output, G() and U() are the principals and agents utility functions, respectively (and G() and U() are the principals and agents marginal utilities, respectively), and are the Lagrange multipliers

associated with the agents (IR) and (IC) constraints, respectively, f(x|e) is the probability density of the outcome conditional on the agents effort, e, and fe(x|e) is the derivative of f(x|e) with respect to effort. As noted by Lambert (2001), Hemmer (2007), and others, this solution is quite general and very little can be said about the optimal contract absent additional structure. In particular, Lambert (2001, p. 18) notes that the shape of the optimal second-best contract depends on the functional form of the term as well as the shape of the

29

Of course it is possible for a linear contract to emerge as the optimal second-best contract when the contract space is unrestricted. However, this seems to occur in only highly stylized settings (e.g., Holmstrom and Milgrom, 1987; Hellwig and Schmidt, 2002). This point is underscored by Hemmer (2007) who notes that for every (interesting) problem formulation, there is only one second-best (in allocations), in information economics analysis there are infinitely many third bests.

81

principals and agents utility functions and that it is easy to construct examples where the optimal contract is linear, convex, or concave.30 Given the generality of this solution, Hemmer (2007) describes how the purpose of the original Holmstrom and Milgrom (1987, 1991) analysis was to facilitate second-best analysis of interesting institutional and organizational phenomena by developing a very tractable model that could be easily analyzed and solved. It was not to legitimize the study of contracts restricted to be linear. (emphasis original). In other words, the original purpose of the linear approximation of Holmstrom and Milgrom (1987) was to provide researchers with a justification for examining linear contracts (which afford the researcher an additional degree of freedom) within the context of a broader research question. Hemmer (2004) therefore argues that invoking the LEN framework should be done on the basis of theoretical justification rather than a simple appeal to its tractability and that the appropriateness of the framework should be judged in light of the research question that is being addressed.31 For example, for research questions where the shape of the contract is likely to be of first-order importance (e.g., risktaking), imposing linearity on the contract space is likely to significantly affect the conclusions and should likely be avoided. Although a further discussion of the debate about the utility of the LEN framework is beyond the scope of our survey, it is important for empirical researchers to be aware of the major concerns and limitations of this line of research.

30

In addition, Hemmer (2007) notes that there is nothing in this expression that would preclude a positive relation between risk and incentives, which is a popular comparative static that obtains from the LEN framework. In a more general setup where risk is endogenous (which is thought to be a popular reason for the use of stock options to encourage risk taking), this result will not necessarily obtain. 31 Dye (2001, p. 197) makes a similar critique of the appropriateness of the assumption of linear prices in the rational expectations literature. In particular, he advocates that theorists should build models based on how they believe the world works, and not be confined by what constitutes current measurement technology.

82

3.3.2.2 The stewardship and valuation roles of accounting A relatively large body of theoretical and, more recently, empirical literature explores the distinction between the stewardship and the valuation roles of accounting information. A key issue in this literature is whether financial reports that are best suited to aid investors in valuing the firm are also best suited to aid shareholders and directors in contracting with executives to mitigate agency conflicts. At first glance, it is easy to see the overlap between the valuation and contracting roles of accounting reports. Market price aggregates a variety of valuation relevant sources of accounting and other information to provide what is likely to be the best available estimate of shareholder value. And, since maximization of shareholder value is the key objective of shareholders, stock price is an obvious choice for a performance measure that can serve to align the interests of managers with those of shareholders. Beginning with Gjesdal (1981), however, the literature has noted that that the relative ranking of an accounting system for contracting purposes need not coincide with its ranking for valuation purposes. Lambert (1993) makes the related point that valuing the firm is not the same as evaluating the managers marginal contribution to firm value. Similarly, Lambert (2001) notes that it is the sensitivity (and therefore congruence) of a performance measure to the managers actions that is relevant for contracting purposes, but it is the real shock to cash flows (i.e., 2 in Figure 1) that is relevant for valuation purposes.32 This is because under the information structure of most moral hazard models, such as Paul (1992), the principal, agent and investors all have common knowledge so the agents actions are perfectly anticipated by the market (as well as the principal) in equilibrium. Therefore the realization of a performance measure is composed of the agents (perfectly anticipated action) plus noise and, consequently, the market does not learn anything about the agents actions from the realization of the
32

This is in contrast to 1 which represents noise in the performance measure that does not affect firm value.

83

performance measures. Thus, the valuation weight placed on the performance measures (e.g., earnings) does not depend on the sensitivities of the performance measures to the agents actions. In contrast, a real shock to the firms cash flows (i.e., 1 if Figure 1) is relevant for valuation because the market does not anticipate it. Bushman, Engel and Smith (2006) empirically examine Pauls (1992) prediction that the weight placed on earnings by investors when valuing the firm, or the valuation earnings coefficient (VEC), and the weight assigned to earnings in CEO annual cash compensation, or the compensation earnings coefficient (CEC), should be unrelated. Probably not surprisingly, they reject the null hypothesis of no relation between the VEC and the CEC and instead find that the two coefficients exhibit a strong positive relation. They develop two alternative models that predict a positive correlation between the VEC and the CEC. Their first model assumes that the managers actions have multiperiod effects that are not fully captured in the current periods earnings. This implies that the VEC is included in the incentive coefficient to motivate the manager to internalize the discounted value of his current period actions. The second model assumes that the managers marginal productivity and the sensitivity of earnings to the managers actions are correlated random variables (with parameters that are common knowledge). In this case, both valuation and the incentive contract exploit the correlation between earnings and the marginal product of the managers effort because valuation seeks to infer the managers true marginal productivity, and incentive contracting seeks to motivate managerial actions that are consistent with the true marginal product. Banker, Huang and Natarajan (2009) extend the analysis of Bushman et al. (2006) and consider the relation between the pay-for-performance sensitivity and the value relevance of both earnings and cash flows. They develop a model similar to Feltham and Xie (1994) in

84

which a CEO takes two productive actions that affect the value of the firm. The two actions are captured with noise by two separate performance measures, which also share a common noise component. Their model predicts that both the valuation and compensation weights decrease when the total variance of the performance measure increases due to an increase in its idiosyncratic noise. The performance measure, however, becomes more informative for valuation purposes and less informative for contracting purposes when the increase in variance is attributable to the common source of noise. Examining annual CEO cash and total compensation, they provide evidence confirming the results of Bushman et al. (2006) that the pay-for-performance sensitivity of earnings is increasing in its value relevance. 33 3.3.2.3 The implications of stock price as the primary executive performance measure As noted above, boards have many performance measures available to them for use in providing executives with incentives. Empirically, however, stock price performance appears to account for the vast majority of financial incentives for most top corporate executives. Although the typical U.S. CEOs annual pay exhibits some covariation with identifiable performance measures such as corporate earnings and stock price, the covariation of the value of their stock and option portfolio with stock price is typically much larger in magnitude, thereby suggesting that CEOs monetary incentives tied to stock price are much larger than from any other performance measure (Hall and Liebman, 1998; Core, Guay and Verrecchia, 2003).34 In fact, in the majority of cases, CEOs stock price incentives appear so large relative

33

Heinle and Hofmann (2009) move beyond the stewardship versus valuation roles of accounting-based performance measure (i.e., earnings) to examine the stewardship and valuation roles of soft (i.e., non-verifiable and non-contractible) information disclosures. They find that although the disclosure of soft information can be beneficial from a valuation perspective, it can detrimental from a stewardship perspective. This occurs when the disclosure of soft information result in a noisier and therefore less congruent stock price, which can diminish managerial incentives and therefore firm value. 34 Or, viewed another way, if stock price does not provide the vast majority of CEOs incentives (say, because one views accounting earnings as much easier for the CEO to understand and influence than stock price), it seems extremely inefficient for boards to impose such a large amount of unnecessary risk on the CEOs.

85

to the monetary incentives provided by all other possible performance measures (combined), that one could question the merits of continuing to studying non-price incentives in executive compensation at all, at least for U.S. CEOs. Although a large body of theoretical research clearly articulates that stock price need not be the dominant performance measure for U.S. CEOs, this theory also does not preclude this possibility. Indeed, stock price has many desirable properties as a performance measure. It is very highly correlated with the shareholders objective function, is likely to be quite sensitive to a CEOs actions, and is expected to be more difficult to manipulate than many other performance measures, such as accounting earnings. Further, the relatively minor role played by accounting measures vis--vis stock price in CEO monetary incentives also suggests that boards have mechanisms available to them to alleviate concerns that performance measures based on anticipated performance (e.g., stock price) do not provide adequate incentives for managers to actually deliver performance. For example, boards may use explicit vesting restrictions on equity grants, or may use implicit contracts to pressure executives to hold equity after vesting restrictions lapse (both of which are empirically descriptive). Regardless, the main point here is not to debate whether stock price should be the dominant performance measure for U.S. CEOs, but instead to discuss the implications of the empirical result that it is the dominant performance measure and to suggest some areas for future research. To begin, we question the merits of continuing to pursue new theories that articulate reasons why stock price might not dominate accounting-based performance measures in CEO compensation contracts. Because stock price and financial statement numbers largely relate to aggregate firm performance, they are likely to be most useful as performance measures for the very top executives in the organization. And, empirically, most top executives compensation

86

arrangements are built around stock price performance. This is not to say, of course, that executives are ambivalent about financial accounting performance measures, such as earnings. Earnings and other financial statement numbers are well known to be important components of the information set that investors use in setting stock prices. To this point, below we discuss in detail the large and growing literature that addresses how executives stock and option holdings influence their incentives with regard to reported accounting earnings. We also note that as one moves down to lower levels in the organization, both stock price and financial statement numbers are likely to be less relevant for providing incentives. Thus, theoretical and empirical research on the relative importance of various performance measures is undoubtedly both interesting and important once one moves down in the organization to less senior executives. Alternatively, if one is determined to test such theories on CEOs, the subsample of CEOs that do not have large stock and option portfolios would be a more reasonable group of CEOs to examine. Related to the points above, we also stress that most agency theory predictions on the relative use of price and non-price performance measures cannot be tested by examining annual CEO cash pay, or even annual CEO total pay. This is because agency theory speaks to the weights in total compensation (e.g., including cash pay, other annual pay, and changes in the CEO's equity portfolio value), not to the weights in a portion of the CEOs compensation.35 Because annual cash pay regressions omit the incentive weight on stock price provided by other annual pay as well as the CEO's equity portfolio, such regressions, in general, cannot be

35

Ideally, a researcher would also include incentive weights in the boards ongoing assessment of the quality of the CEO, as this assessment will influence turnover risk, as well as changes in the present value of the CEOs expected future stream of compensation.

87

interpreted as evidence of agency predictions.36 Further, Core, Guay, and Verrecchia (2003) show explicitly that results from testing agency predictions on annual cash pay do not necessarily hold when using total annual pay or when including incentives embedded in CEOs stock and option portfolios. And, if a prediction does not hold for total annual pay, a test of whether it holds for cash pay is uninformative (or at the very least it cannot be interpreted as evidence supporting the theory). A similar argument holds for annual total pay regressions, since the weight on stock price provided by the CEOs equity portfolio is omitted. Our point here is not to argue against conducting studies that explore the role of financial reporting in the context of evaluating or compensating executives. Rather, we simply make the point that such research must acknowledge the reality that the compensation of most top U.S. executives is tied explicitly (and implicitly) to stock price performance. Toward this end, we offer two potentially interesting areas for future research. First, we note that nearly all U.S. corporations establish accounting-based bonus plans (annual and/or long-term) for their executives, including the CEO. As argued above, such plans are unlikely to provide the CEO and some of the very top executives with significant monetary incentives in light of their stock and option holdings. Thus, a natural question to ask is: Why do we observe so many accounting-based bonus plans if they do not provide top executives with incentives? As suggested by Core, Guay, and Verrecchia (2003), one explanation might be that bonus plans are actually designed to provide incentives for lower-level executives, as opposed to the CEO. Murphy (2000) documents that the median executive bonus plan includes 122 executives beyond the CEO. In many of these bonus plans, the pool of funds to be allocated for bonuses is a function of aggregate accounting performance (e.g., 3% of annual earnings can be

36

For example, Bushman and Smith (2001, p. 264) argue that the estimated slope coefficients in cash pay regressions do not have the theoretical interpretations derived from the model.

88

paid to eligible executives). The actual payouts to individual executives (often excluding the CEO) are frequently based on a combination of both objective and subjective assessments of each executives performance. Consistent with this observation, Rajan and Reichelstein (2009) discuss the role of bonus pools based on financial accounting metrics, as well as the distinction between objective and subjective performance measures in compensating executives out of this pool. It would seem, however, that further research on the team aspect of accounting-based bonus pools might significantly advance our understanding of why such plans are observed. We also note that even though the CEOs monetary incentives may not be greatly influenced by accounting-based bonuses, contracting frictions in real institutional settings may make it efficient to have the CEO's bonus plan parallel that of lower-level executives. Further, we note that because tax-deductible salaries are capped at $1 million, firms are effectively forced to tie the majority of CEOs compensation to some explicit performance measure (even if such measures do not provide significant incentives). Bonus plans that are explicitly tied to a complex function of accounting-based performance measures may also serve as a signal to shareholders and other investors that the board is exerting effort to both obtain information about the firms activities and to monitor executives. As a second area for future research, we note that financial reports provide a decomposition of the performance and actions of executives that is not provided by the stock price. That is, although the stock price performance (and comparisons of stock price performance with competing firms) may provide boards with an excellent overall signal of whether the CEO has done a good or bad job over a period of time, it is unlikely to provide much information about what specific actions the CEO might have taken to achieve this performance. Knowledge about executives specific strengths and weaknesses can aid boards in

89

determining both the quality of the firms top executives and whether the executives are a good fit for the organization, as well as what kinds of personnel changes might improve the overall executive team. Such information can also help the board identify and correct certain agency conflicts, such as perquisite consumption, poor investments, and misleading disclosures. Thus, one could view financial reports as a tool that boards use to disaggregate stock price performance. This role for accounting information is along the lines of that in Berger and Hann (2003) in the context of segment data. Those authors argue that greater disaggregation of segment information under SFAS 131 served to reveal previously hidden information about firms diversification strategies, and that such information resulted in improved monitoring of executives.37 The use of accounting information as described in this paragraph is more likely to be consistent with implicit contracting arrangements with executives than with explicit contracting arrangements. We turn now to a discussion of implicit contracts between boards and executives. 3.3.3. Implicit contracting with executives Although formal, explicit contracts are one way to induce cooperation among selfinterested parties with divergent objectives, when one views actual contracting arrangements between executives and firms, it is striking how little of the relationship takes the form of a standard written contract. Gillan, Hartzell, and Parrino (2009) note:

37

A related set of recent studies disaggregate net income to explore whether its components better explain the observed level of compensation. Adut, Cready and Lopez (2003) and Comprix and Muller (2006) examine how boards weight restructuring changes and pension expense, respectively, when determining annual CEO cash compensation (i.e., salary and bonus). Adut et al. (2003) find that the extent to which boards shield CEOs from restructuring charges varies as a function of characteristics of the contracting environment (e.g., managerial tenure and whether restructuring charges are a repeated issue with the manager). Comprix and Muller (2006) find that CEO cash compensation is more sensitive to pension income than to pension expense. Although these findings are severely limiting in their focus on cash compensation, they are suggestive that boards do not simply take earnings at face value when monitoring the performance and actions of executives, but instead consider the separate, disaggregated components of earnings.

90

When a firm hires a Chief Executive Officer (CEO), it enters into a complex relationship that has significant long-term implications for its stockholders. Establishing the terms of this relationship requires determining the CEOs responsibilities, compensation, perquisites, and term of employment, the conditions under which either party can sever the relationship, and restrictions on the CEOs outside activities, among other considerations. Despite the complexity of these arrangements, many public companies, including some of the largest, choose not to put such terms in writing. In 2000, less than half of the firms in the S&P 500 had a comprehensive written (or explicit) employment agreement (EA) with their CEOs. The other firms had either no written agreement at all, or agreements that covered only limited aspects of their relationship with the CEO, such as change of control, non-disclosure, non-compete, or nonsolicitation agreements. These latter firms and their CEOs relied on implicit EAs through which the CEO was employed at will (authors emphasis).

Further, even when an explicit contract is used, it typically speaks to only a fraction of the issues that likely comprise the full understanding between an executive the board. For example, Gillan, Hartzell, and Parrino (2009) document that the most common features in a CEO employment contract relate to the CEOs initial salary, initial target bonus, initial equity grants, perquisites, and provisions for dismissal with cause (or good reason), change in control, and non-compete. These contracts typically cover a period of about three years, and are thirteen pages in length. Noticeably absent from these contracts is any explicit discussion about the CEOs specific duties, how performance will be assessed over time, what the level and form of future compensation will be, how long the CEO will be employed, for what reasons (other than cause) the CEO might be terminated, and how future contract negotiations will be carried out. If one accepts that these unwritten features of explicit contracts are extremely important (and in light of the fact that many CEOs do not even have written employment agreements), it seems relatively straightforward that implicit arrangements between firms and CEOs constitute much of their contracting relationship. An implicit contract represents the equilibrium behavior of a multiperiod repeated game. Hermalin (2001) notes that, in many situations, inducing cooperation through repeated play is 91

cheaper than inducing it contractually and that, in certain situations, desirable outcomes can be supported in repeated games that cannot be supported contractually. An example would be a setting where the variables on which an ideal contract would be based are not verifiable and therefore not contractible. Alternatively, it may be that the performance measures are verifiable and contractible (e.g., a financial accounting variable), but that the board wishes to use these measures subjectively in assessing the executives performance.38 Multiperiod implicit contracting introduces issues beyond those discussed above in the context of multiperiod explicit contracting. First, the role of commitment on the part of the principal and the agent takes on a somewhat different meaning in an implicit multiperiod setting. Second, the principal, agent and other parties can learn about contract relevant information in a multiperiod setting. For example, the board, shareholders, creditors and other contracting parties can update their beliefs about a managers productivity, risk tolerance, or employment horizon in a multiperiod contracting setting. Although an explicit multiperiod contract could incorporate provisions that are conditional on outcomes from earlier periods of the contract, this is a different concept from revising beliefs over time.39 Given that explicit contracts appear to constitute relatively little of the contracting relationship between the board and executives, the scope for research on implicit contracting with executives seems vast. Some of these implicit relationships have been fleshed out through
38

For example, consider a compensation contract that has a bonus weighted 50% on EPS and 50% on the boards subjective view of the executives performance. When the board forms its subjective opinion about the managers performance for the year, it is not hard to imagine that they might consider a host of financial accounting measures in this assessment. The use of objective performance measures in a subjective way might be an interesting area for research. 39 The level of commitment between the principal and agent in a multiperiod contract is related to the concept of contract renegotiation. Hermalin and Katz (1991) argue that renegotiation is a mechanism that allows the contracting parties to contract on otherwise unverifiable information. A frequent modeling convention is to assume that the principal and agent can credibly commit to not renegotiate their initial contract. However, in reality, the renegotiation of long-term contracts is likely to be common especially if both parties agree to an ex post change in the terms of the contract. The impact of contract renegotiation on the demand for information has been studied in both the economics (e.g., Fudenberg and Tirole, 1990; Hermalin and Katz, 1991; Hart and Tirole, 1988) and accounting literatures (e.g., Demski and Frimor, 1999; Gigler and Hemmer, 2004; Christensen et al., 2005).

92

a rich body of empirical and theoretical research. For example, issues related to executives holdings of equity incentives (through stock and option portfolios) have been extensively researched. Most CEOs hold substantially greater amounts of stock and options than is required by any explicit arrangement with directors. And, these implicit agreements regarding equity incentives have been shown to influence executives actions with respect to various operating, investing, and financing activities, as well as accounting choices. A complete discussion of the wide array of implicit contracting arrangements between shareholders, boards, and executives is beyond the scope of our review. We therefore confine ourselves to briefly noting a few areas of research on implicit contracts that might be interesting to accounting researchers. We begin by considering the multiperiod career concern of executives with respect to how they signal their quality to the labor market, and how boards dynamically contract with executives over time as information about an executives quality is revealed. For example, how does a board determine whether their CEO is worth $2 million per year or $10 million per year (note that this is a question of the CEOs level of pay, as opposed to variation in pay due to incentive risk that is imposed on the CEO)? Presumably, this depends upon the CEOs quality. But, how does one determine the CEOs quality? It seems hard to imagine that such a determination can be made based on a single year or two of performance data. All CEOs (like academic researchers) will make some good decisions and some bad decisions in each period. Quality, therefore, is likely to be revealed only over time as one aggregates data on the (value-weighted) proportion of good decisions as compared to bad decisions. It is interesting to conjecture about the types of signals that reveal an executives quality. Among the various possibilities are stock price performance, aggregated or disaggregated

93

accounting measures, and a host of other objective and subjective performance signals, such as acquisitions, product development, employee development, investor relations, etc. We are aware of very little research examining how or whether accounting measures provide information about executive quality. Boschen, Duru, Gordon and Smith (2003) examine the dynamic effects of alternative performance measures (earnings and stock returns) in CEO compensation contracts. They find that unexpectedly good accounting performance initially increases compensation in the short-run, but then this compensation reverses in later years such that the total effect of the accounting performance has little effect on cumulative pay. This is consistent with ratcheting with respect to accounting earnings.40 In contrast, stock price performance increases current-period pay, but also increases future pay so that there is a substantial relation between compensation and stock price performance that persists in the long run.41 Indjejikian and Nanda (2002) also examine the dynamic effects of accounting performance on executive compensation. The authors examine the extent to which firms adjust for prior performance when setting annual target bonus thresholds (i.e., whether there is ratcheting in performance benchmarks). They predict that achieving a bonus target in a given year should be independent of whether a target was achieved in the prior year if compensation committees fully adjust the target to incorporate performance in the prior period. The authors find evidence of positive serial correlation in executives annual bonuses, which is consistent with firms not fully adjusting performance standards for executives past performance. These
40

Milgrom and Roberts (1992, p. 232) define the ratchet effect as the tendency for performance standards to increase after a period of good performance in a in a dynamic contracting environment. When agents anticipate the ratchet effect in their contract, they trade off the better contractual outcomes (e.g., higher compensation and promotion) from better performance in the current period against higher performance thresholds in the future. Indeed, the ratchet effect is an example of where the inability to commit to multiperiod contract is inefficient. 41 Based on the findings from their vector autoregressive model, Boschen et al. (2003) suggest that empirical models of compensation should allow for at least three lags of compensation and firm performance in order to fully capture the dynamics of executive contracting arrangements.

94

results suggest that the setting of target bonuses is a complex process that likely considers multiperiod implicit contracting issues, such as the intertemporal revelation of an executives quality, how bonuses interact with other components of the compensation plan, and costs associated with making changes to compensation arrangements over time. A severe limitation of this study, however, is that the analysis is restricted to annual cash pay, which constitutes only a small portion of executives pay. 3.3.4. Senior executive turnover In addition to selecting and compensating senior executives, the board is also tasked with periodically reassessing whether the CEO and other high level executives should continue their employment with the firm. A number of authors, such as Gibson (2002), contend that one of the primary purposes of corporate governance mechanisms is to ensure that poorly performing managers are removed. This view is consistent with Jensen and Ruback (1983) who argue that poorly performing managers who resist removal might be the costliest manifestation of agency problems in the firm. Thus, the decision to terminate an unwanted executive can have a large impact on shareholder value and is a very important decision of the board. Turnover decisions can be motivated by both incentive and matching objectives. Regarding the former, the threat of termination can provide managers with an incentive to take the appropriate actions (or suffer from the disutility that typically accompanies a termination). Regarding the latter, turnover can be used to facilitate a better match between firms and CEOs on the basis of some characteristic such as CEO leadership qualities, risk preferences, or expertise with the firms production technology. Early research on executive turnover documented that both accounting and stock price performance are used in the turnover decision. For example, Hermalin and Weisbach (1998)

95

show that the historical nature of accounting earnings makes it useful in the turnover decision because stock price reflects both the markets expectations of the CEOs continued employment and the anticipated effect of his replacement, with only the former being useful in motivating the CEO. More recently, Bond, Goldstein and Prescott (2009) sharpen this result in a more general setting by noting that learning from price is often difficult because two or more fundamentals may be associated with the same equilibrium price (or technically that price is not necessarily monotonic with respect to fundamentals). For example, in the case of the board deciding whether to replace the CEO, a moderate price could either indicate that (i) the market expects the board to replace the CEO and impounds the replacement CEOs actions or (ii) that the CEO is not performing poorly enough to justify replacement. Bond et al. (2009) show that whenever a decision maker such as the board takes a corrective action (e.g., replacing the CEO) on the basis of information inferred from the market price of the firms securities, there is a complementarity between market information and the decision makers other information.42 If the decision makers other information is not sufficiently precise, then market price is not fully revealing which impairs the corrective action. Their analysis therefore suggests a role for accounting information to facilitate the boards learning from price in order to make a more informed turnover decision.43 In another recent study, Engel, Hayes and Wang (2003) examine the relation between the weight placed on accounting and market return information in the CEO turnover decision and the relative sensitivity and precision of these measures. Their tests are motivated by moral
42

It is important to note that in their model, the market price of a security endogenously affects its own real value through the information it provides the decision maker. In other words, price both reflects and affects the decision makers actions. 43 Note that the intuition in Bond et al. (2009) applies to any actions that are taken on the basis of information gleaned from price. Thus, there is an indirect role for accounting information (in addition to any direct effect) in facilitating learning from price for a wide array of contracting parties when making a variety of governance decisions.

96

hazard models with multiple signals (e.g., Holmstrom, 1979; Banker and Datar, 1989) which stress that the optimal aggregation of signals is in relation to their sensitivity to the agents action relative to the precision with which the signal captures that action (or, the signal-to-noise ratio). Engel et al. find that the weight placed on earnings in the CEO turnover decision is increasing in the timeliness of earnings (measured as the R2 from a reverse regression of earnings on returns) and decreasing in the variance of earnings (which proxies for its noise). A key assumption underlying this analysis, however, is that the ability of earnings to explain returns is a good proxy for the sensitivity of earnings to the CEOs actions. Although, in many respects, this seems like a reasonable assumption, it would be useful and interesting to see some confirmatory evidence that cross-sectional variation in the returns-earnings relation reflects cross-sectional variation in CEO actions. Finally, recent theoretical work by Dutta and Reichelstein (2003) models a multiperiod double-moral hazard problem where the principal must induce the agent to both take a productive action in each period and choose the level of long-term investment in the first period. They find that in certain settings where the agent has incentives to overinvest in the long-term project, the principal is better off foregoing a long-term contract with the agent in favor of a series of single-period contracts. They interpret this result as managerial rotation, or the principal firing the agent after each period. A related idea is that an executive who has accumulated a large amount of wealth might have a relatively low marginal utility and is therefore more difficult to motivate with monetary incentives. Kadan and Swinkels (2008) show that a poorly designed compensation package can make a manager too rich too soon and therefore unresponsive to additional monetary incentives as a result of the diminished marginal utility of income that results from the wealth effects inherent in their utility function. Thus,

97

ceteris paribus it would be more efficient to replace the manager with a less wealthy manager who is more responsive to monetary incentives. Our knowledge of the turnover decision is still incomplete and our understanding could undoubtedly benefit from additional theoretical and empirical work. 3.3.5. Contractual incentives for earnings management A large literature, primarily inspired by the positive accounting theory of Watts and Zimmerman (1986), entertains the possibility that both explicit and implicit compensation contracts provide incentives for executives to take actions that might not be intended by the board (and/or shareholders) when designing the compensation contract.44 Earlier studies focused primarily on the incentives provided by explicit contracts (e.g., accounting-based bonus plans) while a more recent literature has emerged that focuses on implicit contracts (e.g., evidence that boards reduce bonus payouts when earnings miss analysts forecasts). Thus, this literature straddles both the explicit and implicit contracting paradigms. The theoretical motivation for these studies is based on the idea that a manager has the choice between (i) a productive action that both affects firm value and increases the value of a contractual performance measure such as earnings or stock price, and (ii) an unproductive action that does not necessarily increase firm value (and often reduces it) but nevertheless increases the performance measure used in assessing the manager. Examples of unproductive actions include cutting value-enhancing R&D expenditures in order to increase net income in the current period and outright earnings manipulation (fraud) that has no real effect on firm value. When the net cost of productive actions is high relative to the net cost of unproductive
44

Examples of such actions include cutting dividends (Lewellen et al., 1987; Lambert, Lanen and Larcker, 1989), cutting R&D and other discretionary expenditures when nearing retirement (Dechow and Sloan, 1991), forgoing risky projects (Smith and Stulz, 1985), empire building through negative NPV investments (Jensen, 1986), option backdating (Heron and Lie, 2007 and 2009), and managing earnings (Healy, 1985; Holthausen, Larcker and Sloan, 1995; Cheng and Warfield, 2005).

98

actions (both of which include the personal cost of taking such actions and any potential legal and reputational penalties if detected), the agent has an incentive to substitute the unproductive actions for the productive actions. Of course, there is also a large literature on how accounting discretion can be used by managers to increase firm value (e.g., through lowering taxes, avoiding covenant violations, or improving earnings as a reliable and timely measure of firm performance). A number of early studies focused on the manipulation of reporting earnings because earnings-based measures are common explicit contractual performance measures that are typically thought to be directly controllable by the agent (e.g., Healy, 1985). Although this early literature was motivated by appealing to the explicit use of financial performance measures in executive compensation contracts (e.g., earnings-based bonus plans), a more recent series of papers examines their implicit use in executive compensation contracts. For example, Matsunaga and Park (2001) examine the effects of missing three distinct quarterly earnings benchmarks (consensus analyst forecasts, earnings of the same quarter of the prior year, and zero earnings) on CEOs annual bonuses and restricted stock and stock option grants. They find a negative relation between the CEOs annual bonus and missing either the first or second benchmark two, three or four times during the year, but not for missing it just once. They also find no effect of missing the zero earnings benchmark on the annual bonus regardless of the number of times it is missed. Mergrnthaler, Rajgopal, and Srinivasan (2009) document that the CEO bonus results in Matsunaga and Park (2001) extend to equity compensation and forced turnover, as well as to CFOs. Specifically, CEOs and CFOs both receive lower equity compensation and face a greater likelihood of forced turnover after missing quarterly earnings benchmarks. Since certain benchmarks such as the consensus analyst forecast might be difficult

99

to incorporate into explicit contracts, the effect of missing an earnings benchmark on a CEOs compensation is likely to result from the compensation committee exercising its discretion in reducing CEO pay. These studies raise a number of interesting questions, such as why boards use missing multiple earnings forecasts as a measure of performance. The authors suggest that the use of certain earnings benchmarks as a performance measure might encourage managers to provide the market with better information so that analysts are able to more accurately forecast earnings. Therefore, missing a benchmark multiple times could be a signal of the managers disclosure efforts or that the manager does not have adequate control of the firms operating activities. This is similar to the argument by Nagar, Nanda and Wysocki (2003) that equity-based compensation provides managers with incentives to be more forthcoming with disclosure. These papers also raise the interesting question of whether CEOs are rewarded for managing earnings to achieve a target. In Section 3.1.1, we discussed a number of papers that find that managers and directors face relatively severe penalties for earnings manipulation (e.g., restatements, AAERs, etc.). The findings in Matsunaga and Park (2001) and Mergrnthaler et al. (2009) suggest that boards might, in certain instances, encourage and reward earnings management. It would be useful for future research to reconcile these findings by examining in what settings earnings management is encouraged or deterred from managing earnings. For example, if a manager meets earnings forecasts by manipulating earnings, does he still get rewarded? Or is it only when he meets analysts forecasts by either giving better guidance or by achieving earnings without using manipulation? Evidence in support of the former would seem to contradict the findings presented in section 3.2. In addition, it would be useful for future studies to identify situations in which earnings management is actually beneficial to the

100

contracting parties (e.g., in order to facilitate renegotiation of the contract). Finally, in models of private information it is essential that a violation of one of the three necessary conditions of the revelation principle occurs in order to generate earnings management. If the conditions for the revelation principle are satisfied, then any equilibrium in which earnings management occurs is weakly dominated by an alternative equilibrium that entails truthful reporting by the agent (Dye, 1988). We move now to the large recent literature that examines the relation between executive equity incentives and earnings management, because manipulated earnings are thought to have a causal effect on stock prices. The recent focus on equity incentives stems largely from the realization that most top executives incentives are tied to stock price performance through the executives holdings of stock and options, as well as that executives have some control over the timing of sales of these securities. Thus, the literature has evolved from examining the direct effect of earnings manipulation on accounting-based bonuses to examining the indirect effect of earnings on stock price, which in turn has a large direct effect on a managers wealth. Two wrinkles with this more recent line of research, however, are: (i) stock price is believed to be a more difficult performance measure to manipulate than earnings since investors are sophisticated in their processing of information and anticipate or adjust for many forms of earnings management, and; (ii) executives must not only manipulate the stock price, but must also convert such manipulation into liquid wealth. This typically requires exercising options and selling stock prior to investors realizing that earnings are not as strong as previously thought, as well as ensuring that the manipulation is undetected both ex ante and ex post (ex ante to ensure that investors are fooled and ex post to ensure that the manager does not bear any repercussions from manipulating earnings, such as being fired or sued).

101

Like earlier studies, most recent work in this area generally treats compensation contracts as exogenous and looks for a positive relation between equity incentives of the CEO, the CFO, or the senior executive team and the degree of earnings management (Larcker, Richardson and Tuna, 2007; Cheng and Warfield, 2005), the frequency of accounting restatements (Harris and Bromiley, 2007; Burns and Kedia, 2006; Efendi, Srivastava and Swanson, 2007; Armstrong, Jagolinzer and Larcker, 2009), and SEC Accounting and Auditing Enforcement Reselases (Erickson, Hanlon and Maydew, 2006; Johnson, Ryan and Tian, 2009). Most studies adopt a rent extraction perspective and interpret a positive relation between equity incentives and accounting manipulation as a symptom of bad governance and misaligned managerial incentives. An alternative perspective, however, is that boards anticipate both the productive and unproductive actions that a given set of incentives will induce and so expect a certain amount of accounting manipulation. And, for any given incentive structure that is implemented, the board decides that the costs associated with the expected manipulation are outweighed by the expected productive benefits of the chosen incentive structure. Most earnings management studies focus on the incentives of the CEO, both because earnings are generally thought to be under the control of this individual and because the CEO typically has the highest equity incentives in the organization (and therefore the greatest potential benefit). Another group of studies, however, examines the incentives of the CFO, since this executive is directly in control of the financial reporting and typically has a sizable level of equity incentives. Finally, a few studies examine the incentives of the firms five highest paid executives arguing that earnings management typically requires the complicity of other senior executives (e.g., Erickson, Hanlon and Maydew, 2006; Johnson, Ryan and Tian, 2009).

102

As discussed by Armstrong, Jagolinzer and Larcker (2009), and summarized in their Table 1, studies in this literature tend to adopt a variety of different research designs, measures of incentives, sample periods and proxies for earnings management and offer mixed results. For example, Harris and Bromiley (2007) find a positive relation between the CEOs option mix (i.e., the ratio of the value of annual stock option compensation to the value of total annual compensation) and accounting restatements, but no such relation for bonus mix. Burns and Kedia (2006) and Efendi, Srivastava and Swanson (2007) both find a similar result that the sensitivity of the CEOs option portfolio to stock price is positively association with the frequency of accounting restatements.45 Larcker, Richardson and Tuna (2007) find a positive relation between the compensation mix of the CEO and abnormal accruals. However, neither Larcker, Richardson and Tuna (2007) nor Barber, Kang and Liang (2007) find a relation between the compensation mix of the CEO and accounting restatements. In contrast, Erickson, Hanlon and Maydew (2006) find a positive relation between the compensation mix of the CEO and the incidence of accounting fraud, but no such relation when portfolio delta is used as the measure of incentives.46 A recent study by OConnor, Priem, Coombs and Gilley (2006) suggests that the relation might not be as direct and conjecture that certain governance characteristics might instead mediate the relationship. In particular these authors examine the relation between the risk-neutral (Black-Scholes) intrinsic value of the CEOs option portfolio and the incidence of accounting restatements. The authors find a positive relation if either (i) the CEO is also the
45

Burns and Kedia (2006) also find a positive relation when portfolio delta is used as the measure of incentives. However, they find no relation between the other sources of CEO incentives (e.g., the stock and restricted stock portfolio, long-term incentive plans, salary and bonus) exhibit no such relation. 46 An important measurement issue is whether compensation mix (i.e., the proportion of annual pay that comes from a single component of pay, such as stock or options) is a meaningful proxy for the incentives of the CEO. Specifically, as previously discussed, since the vast majority of a CEOs equity incentives stem from stock and option holdings, it is difficult to see how any measure constructed from annual pay alone is a valid proxy for CEO incentives.

103

chairman of the board and the firms other directors do not receive options or (ii) the CEO is not also the chairman and the other directors receive options, but no relation otherwise. Thus, the evidence is mixed and any evidence that supports a link between CEO equity incentives and accounting irregularities appears to be somewhat fragile and sensitive to measurement issues. Moreover, the results of OConnor, Priem, Coombs and Gilley (2006) suggest that other characteristics of the governance structure might also play a role in either mediating or moderating any relation between executives equity incentives and various accounting outcomes. Finally, a number of recent studies acknowledge that the Chief Financial Officer (CFO) often has as much, if not more influence over the firms financial reporting than the CEO. As noted by Feng, Ge, Luo and Shevlin (2009), CFOs are in a unique position to carry out accounting manipulations, from transaction structuring, to choosing an improper accounting method, to making false journal entries.47 Similarly, Jiang, Petroni and Wang (2009) argue because CFOs primary responsibility is financial reporting, we argue that CFO equity incentives should play a stronger role than those of the CEO in earnings management. Indeed, the important role played by the CFO in the firms financial reporting is highlighted by the requirement in the Sarbanes-Oxley Act of 2002 that both the CEO and CFO personally certify the material accuracy and completeness of the financial information and disclosures of the

47

This is related to the idea that is explored in a series of recent papers that CFOs have their own unique financial reporting and disclosure style, which can be captured by following CFOs as they change firms (Yang, 2009; Ge, Matsumoto and Zhang, 2009; Dyreng, Hanlon and Maydew, 2009). In particular, these studies adopt the research design in Bertrand and Schoar (2003) which estimates a series of CEO fixed effects in a sample of CEOs that were employed by at least two different firms during the sample period. This research design thus attempts to separately identify the effect of the CEO from firm-specific effects. This argument is consistent with Geiger and North (2006) who find that the magnitude of discretionary accruals decreases significantly following the appointment of a new CFO (compared to a matched sample of firms that did not hire a new CFO) which they interpret as evidence of the CFO exerting an influence over the firms financial reporting.

104

company. In addition, as one of the more senior executives in the firm, the CFO also typically receives a large component of their compensation in the form of equity-based compensation. As in the case of the CEO, however, the evidence appears to be sensitive to the research design and inconsistent across outcome variables. For example, Feng, Ge, Luo and Shevlin (2009) find that the CFO equity incentives (measured as portfolio delta) within their sample of AAER firms are not significantly different from that of either (i) other non-CEO executives within the same firm or (ii) those of CFOs is a sample of non-AAER control firms. They do, however, find that both the equity incentives of the CEO and the ratio of the CEOs to CFOs equity incentives are higher in their sample of AAER firms than in the control firms. Feng et al. (2009) also find that the penalties faced by CFOs can be quite severe. In that regard, they find that in roughly 20% of the 496 AAER firms during 1982-2005, the CFO was charged with fraud while the CEO was not. This is consistent with the evidence in Hennes, Leone and Miller (2008) who find that CFO turnover is more frequent than CEO turnover following accounting restatements. Thus, they conclude that CEOs have strong financial incentives to manipulate accounting reports while CFOs do not. However, Jiang, Petroni and Wang (2009) who find a positive relation between the magnitude of CFO equity incentives and both absolute total and discretionary accruals and the probability of beating analyst forecasts, which they interpret as consistent with CFOs engaging in earnings management.48 One problematic feature of this line of research, which might explain the contradictory findings, is that the compensation contracts are typically treated as exogenous. Thus, the research design of most studies precludes the endogenous matching of executives with

48

It is interesting to note that Jiang, Petroni and Wang (2009) find a positive relation between CFO equity incentives and the absolute value of both total and discretionary accruals in the pre-SOX period, but not in the post-SOX period. However, they find a positive relation between the CFO equity incentives and the probability of beating analyst forecasts both prior to and following SOX.

105

compensation contracts on some attribute that is potentially correlated with both the choice of the contract and the executives propensity to manipulate accounting reports. Armstrong, Jagolinzer and Larcker (2009) relax the assumption of exogenous contracts and instead allow for the endogenous matching of executives and contracts. In particular, they adopt a propensity score matched pair research design where they first model the probability that a CEO will have a certain level of equity incentives. They then form matched pairs of CEOs who have similar propensity score (i.e., a similar probability of having a certain level of incentives conditional on observable attributes of their contracting environment) but dissimilar levels of equity incentives.49 Their results suggest that the results in this literature are sensitive to the omission of relevant variables such as the CEOs degree of risk aversion or talent that could affect both the choice of the compensation contract as well as the executives willingness to manipulate earnings. Another problem with this literature is that the incentives for earnings management are often taken for granted. In particular, the most common research design is to examine the relation between portfolio delta, or some other ex ante measure of the managers incentives to increase stock price, and the executives propensity to manipulate accounting reports. Although ex ante measures of earnings management, such as portfolio equity incentives, are readily available, one problem is that they do not necessarily indicate whether the manager benefits from managing earnings. Armstrong, Foster and Taylor (2009) make this point in the context of whether there are incentives to manage earnings around initial public offerings. Moreover, it is consistent with Schippers (1989) definition of earnings management as a purposeful

49

Another feature of the research design in Armstrong, Jagolinzer and Larcker (2009) is that they perform a sensitivity analysis of their results to hidden bias that could result from the omission of a relevant variable from the propensity score model. Their sensitivity analysis entails quantifying the strength of a correlated omitted variable that would be necessary to alter the statistical significance of their results.

106

intervention in the external financial reporting process with the intent of obtaining some private gain. A more comprehensive research design would examine whether executives, in fact, benefit (or gain) by exercising options and selling shares of stock when the manipulated earnings artificially inflate the stock price. Another related study is by Cornett, Marcus and Tehranian (2008). They note that if the degree of earnings management is a function of the governance structure (including executive compensation contracts), then tests that examine the relation between governance and firm performance should take into account that some of the performance is cosmetic rather than real. In other words, there is a potential distortion in earnings that is correlated with the firms governance structure. They find that adjusting for the effect of earnings management increases the relation between measures of governance and firm performance and decreases the relation between incentives and firm performance. They suggest that empirical studies should be careful when correlating measures of governance with firm performance to distinguish actual and perceived performance (i.e., window dressing or earnings management). There has also been recent theoretical work on modeling the principals contracting problem when managers can engage in earnings management. For example, Demski, Frimor and Sappington (2004) examine the optimal behavior of an organization (i.e., principal) when an agent can manipulate the organizations accounting system to his private advantage. The authors first note that agents can influence the performance measures produced by the accounting system in one of two ways: (i) by delivering productive efforts which are captured by the accounting system (with noise) and (ii) by manipulating the accounting system. The principal can limit the losses from manipulation in at least three ways: (i) the principal can reduce the direct payoff that agents anticipate from manipulation, (ii) the principal can increase

107

the agents cost of manipulating the accounting system, or (iii) the principal can facilitate manipulation. The paper explores the third mechanism and shows that encouraging manipulation in early periods can reduce the agents opportunity cost of productive effort in later periods, which mitigates the principals incentive problem in those later periods. The analysis in Gigler and Hemmer (2004) also has implications for earnings management. In particular, their model implies that the contracting (or stewardship) value of information should determine whether it is mandatorily or voluntarily disclosed with more contract relevant information being subject to mandatory disclosure. This implies that managers should be afforded more discretion over the firms financial reporting when earnings (and other accounting numbers) serve less of a contracting role. Their results also relate to the result that a commitment to not produce information can be valuable whenever contracting parties cannot commit to not renegotiate a long-term contract (e.g., Demski and Frimor, 1999; Christensen, Demski and Frimor, 2002).50 Similar to the benefits of a conservative accounting system identified by Christensen, Demski and Frimor (2002), it is the ability to manage earnings that generates a benefit rather than the act of managing earnings per se. Thus, earnings management generates strict contracting benefits even though everyone sees through it in equilibrium. A related idea is that when managers are in possession of private information, the discretion afforded by accounting rules can provide a channel through which managers can reveal this information. Finally, Chen, Hemmer and Zhang (2007) also develop a model in which earnings are used for both stewardship and valuation purposes and show that the (current) owners of the firm
50

In particular, as Gigler and Hemmer (2004) note, in cases where the contracting parties cannot commit to not renegotiate a long-term contract at an interim stage, although the use of information in the renegotiation is ex post efficient, it results in contracts that are ex ante less efficient than full commitment contracts. Accordingly, these models generally find that information destruction, through aggregation [or] earnings management improves the ex ante efficiency of renegotiation-proof contracts.

108

have incentives to engage in earnings management which, in turn, reduces the stewardship value of earnings and leads to inferior risk sharing. They find that conservative reporting standards dampen insiders incentives to manage earnings since, in their model, the current owner has an incentive to bias earnings upward even though investors rationally anticipate earnings management and price protect against it. Also, the upward bias induced by earnings management reduces the informativeness of earnings for stewardship purposes which, in turn, leads to inefficiencies in the incentive contract. They show how conservative accounting can improve contracting efficiency (i.e., the stewardship value of earnings) because the reduction in earnings management more than offsets the loss in value of the earnings signal. Their model predicts that conservative accounting is more likely to arise when accounting numbers play both roles and when those with control over the financial reporting process have significant equity stakes in the firm (and therefore have high incentives to misreport). Their model also predicts that the degree of earning management is lower when accounting standards are conservative than when they are unbiased. Although their results are intriguing, they rely critically on the assumption that the current owner of the firm cannot credibly commit to not manage earnings and is therefore forced to manage earnings to fulfill potential investors rational conjecture. This suggests that the benefit of conservative financial reporting in mitigating the incentives for earnings management are greatest when the firm is unable to credibly commit to not manage earnings. The accounting literature has long recognized that in order for earnings management (or non-truthful reporting in general) to be an equilibrium, one of the three necessary conditions for the revelation principle must be violated (Dye, 1988). These three conditions have been referred to as the three Cs and are (i) unrestricted communication between the principal and agent, (ii)

109

full commitment between the principal and agent on how the information will be used and (iii) optimal contracts. Accordingly, the revelation principle can be rendered invalid through (i) restrictions on communication between the contracting parties, (ii) limited communication between the contracting parties and (iii) suboptimal contracts. 3.4. The role of information in multi-party contracting The previous section on the relation between information (e.g., financial accounting numbers) and executive incentive contracts generally focused on agency problems between management and the board (acting on behalf of the firms shareholders). However, other contracting parties in the firms nexus will also generally observe the compensation contracts that are in place and, in turn, will anticipate managerial actions that are induced by these contracts. Some of these actions might be beneficial to certain contracting parties and detrimental to others. Contracting parties will rationally anticipate the costs and benefits of the anticipated managerial actions and will thus price the effects into their contracts. Therefore, to maximize shareholder value, the board might consider the broader set of contracts into which the firm has entered or plans to enter when designed an optimal compensation contract. In other words, even though the other contracting parties in the firms nexus do not necessarily have an explicit role in designing compensation contracts, they do play an implicit role to the extent the board and shareholders anticipate the effects of how these parties will correspondingly adjust their own contracts to compensation contracts are put in place. Therefore when designing compensation (or any other contracts in the firms nexus), it necessary to explicitly take the interests of other constituencies into account in the contract design problem. The firms debt contracts are the most frequently cited contracts that are also affected by the compensation contracts put in place. Brander and Poitevin (1992) develop a model in which

110

the firms shareholders design a compensation contract to induce the manager to take an action. Since the shareholders do not have sufficient funds to finance the managers project selection, they must rely in part on debt funding. Since the firms creditors rationally anticipate the managers action under the compensation contract, they price the expected effect into the debt contract. Accordingly, they show that the shareholders problem is equivalent to maximizing firm value (i.e., the value of the firms equity plus its debt) rather than just maximizing equity value. Thus, the compensation contract is an important (and credible) mechanism for mitigating conflicts of interests between a firms shareholders and bondholders. In general, they argue that managerial compensation contracts are an important determinant of financial structure. Our discussion until now has generally considered the role of equity markets in the incentive contracting process. However, there is nothing that prevents other markets from playing a role in the compensation process. If we think of the market in Figure 1 as the market for the firms debt then a number of interesting points can be discussed. In this case, the market can learn about the managers private information from his choice of the contract.51 The basic insight of Fershtman and Judd (1987, p. 934) is that profit-maximizing owners may not want to give profit-maximizing incentives to their managers because an owner can influence the outcome of the competition between the managers in his favor by distorting his managers incentives. internal relationships and incentives can be distorted and manipulated for inter-firm strategic reasons, giving a new and fundamentally different role of internal contracts. In other words, competing firms owners will often distort their managers

51

Note that the manager choosing his compensation contract can be construed broadly enough to include not only the choice of the explicit contract, but also choices that are made during the life of the contract that affect the managers contractual payoff. An example includes when to exercise stock options and sell shares of stock. This is, to a large extent, an example of where the manager has a choice about the compensation contract. In other words, allowing the manager choices in his compensation contract (such as when to exercise options and sell shares of stock) may be a mechanism by which the principal can induce the manager to reveal his private information.

111

objectives away from strict profit maximization for strategic reasons. Substituting accounting performance measures for stock price might be one way to accomplish this. 3.5. Good vs. bad governance Our discussion of corporate governance has emphasized the role of financial reporting in resolving agency problems between managers and directors, and to some extent between controlling and minority shareholders. Underlying this discussion is the broad notion that contracting parties strive to mitigate these agency problems, but that there are contracting costs and frictions that limit the extent to which these agency problems can be resolved. The cost of transferring governance-relevant financial and non-financial information to outside board members and shareholders is one such friction that limits a firms ability to reduce agency conflicts. Because the costs and benefits of transferring information among managers, directors and shareholders differs across firms, industries and countries, one expects to observe firmspecific, industry-specific, and country-specific variation in governance structures, such as board structure and incentive compensation plans. Further, observed variation in governance mechanisms can allow researchers to better understand how agency conflicts, as well as the costs and benefits of different governance mechanisms vary across economic settings. Our intention, however, is not to suggest that information asymmetry is the sole or even the most important friction that researchers emphasize when studying governance mechanisms. Indeed it is quite common for researchers to examine cross-sectional variation in board structure or compensation plans with little or no emphasis on the informational problems that these mechanisms are designed to mitigate. One such line of research emphasizes frictions between shareholders and directors. Specifically, it is frequently argued that contracting costs sometimes prevent shareholders from selecting directors who are independent of management

112

or, more generally, directors whose interests are aligned with their own. For example, various authors have argued that directors sometimes take actions that are not in the best interests of shareholders, but are nevertheless easily observed by the firms shareholders (e.g., adopting anti-takeover provisions, staggered elections for board members, or granting excessive compensation to senior executives). Under the assumption that these governance features are, in fact, undesirable from the perspective of shareholders (a debate is too detailed to discuss here), the frictions that allow the agency conflict to persist appear to stem from shareholders inability to select a mix of (inside and outside) directors to effectively monitor and advise management.52 Consistent with this line of reasoning, regulators have recently devoted considerable attention to the issue of whether and how shareholders might be allowed to exert greater influence in selecting and removing corporate directors. Since information asymmetry between contracting parties does not appear to be a key driver of these other important governance issues (i.e., when shareholders know there is an undesirable governance mechanism in place, but cannot take corrective action), we do not discuss these issues in detail in our survey. Instead, we start from the assumption that shareholders can influence the governance structure, and then consider how information asymmetry and the accompanying information processing and transfer costs endogenously influence shareholders governance choices (and vice versa). Or, said another way, if the frictions that prevent shareholders from selecting their preferred governance structure were eliminated, how would the information environment influence shareholders choices with respect to governance?
52

One commonly cited friction is that individual shareholders must each bear significant costs to both become sufficiently informed about agency conflicts, and to take actions to correct the agency conflicts. The benefit that accrues to any particular shareholder, however, is proportional to his ownership interest in the firm, and for most shareholders in widely held firms, may be small relative to the costs. The first of these costs, information costs, is the primary focus of our survey. The second type of cost, the cost of taking action, is beyond the scope of our survey.

113

At this point in our discussion, we emphasize the importance of carefully identifying (and even quantifying) the economic frictions in any discussion of the efficiency of governance structures. In making this point, we highlight the large literature that points to certain governance structures (e.g., characteristics of the board and its committees, and compensation plans) as being unconditionally good (strong) or bad (weak). Based on our understanding of this literature, we infer that bad (weak) governance is broadly intended to mean the following: a firm suffers from a serious agency conflict between shareholders and managers, and that some (often unarticulated) contracting cost or friction prevents the firm from moving toward a good or at least better governance mechanisms that would mitigate the agency conflict. In light of the discussion in the prior sections, an obvious first note of caution when asserting that some governance structures are good or bad is to ensure that, in fact, what is being called a bad structure is not instead a good structure for certain firms. For example, many papers identify firms with a relatively higher proportion of outside directors as having a bad governance structure, and even further that firms with the highest proportion of outside (inside) directors have the very best (worst) governance (and this cross-sectional variation in good/bad governance is asserted even if all/most firms in the sample have a majority of outside directors). In many cases, this view ignores the extensive economic arguments and empirical evidence related to why some firms that are labeled bad governance firms might, in fact, have appropriately selected a board with relatively few outside directors. An alternative way of characterizing this critique is to suggest that the notions of good and bad governance are conditional, rather than unconditional concepts. That is, only after conditioning on all of the relevant (observable and unobservable) economic characteristics of the firm and its

114

operating and information environment can statements be made about whether certain governance structures are good or bad. It is also the case that after conditioning on economic characteristics, one must consider that firms with too much or too little of certain structures, such as outside directors, should likely be characterized as bad. However, we will proceed with the assumption that certain governance structures are, in fact, (unconditionally) less desirable than other governance structures from the perspective of shareholders (at least for particular firms at particular points in time). As an example of this literature, we point to the large body of research that examines the relation between various types of governance structures (e.g., board structure, executive incentive structures, and shareholder protection indices) and either firm performance or some other outcome variable that is assumed to measure undesirable behavior (e.g., earnings management or excess executive compensation). An implicit (or explicit) assumption in any study that hypothesizes a relation between governance structures and firm performance or perverse behavior is either that: 1) shareholders behave rationally, but economic frictions prevent shareholders at some firms from instituting the good governance structure (or at least that frictions slow the process by which shareholders can institute the good structure), or; 2) shareholders behave heuristically or irrationally and do not attempt to maximize shareholder value. We believe that many researchers working in this literature would endorse the first of these two assumptions, which is equivalent to the notion that governance structures are expected to be efficient within contracting costs.53 What is less developed, however, is precisely what these frictions are, and how these frictions vary cross-sectionally in a way that generates
53

For researchers that endorse the second assumption, and view heuristic or irrational behavior as the more plausible explanation for observed governance structures, research on frictions in the market for corporate control would seem a fruitful area of research. That is, if groups of irrational shareholders persist in controlling firms with suboptimal governance structures, an obvious question is what are the frictions that prevent a well-functioning market for corporate control from acting as a correction mechanism.

115

variation in the efficiency of governance structures.54 In order for a researcher to successfully argue that a firm or a group of firms has an undesirable governance structure, it is incumbent on the researcher to describe how that unwanted structure came to be, why it persists, and how long it is expected to remain in effect (or better yet, to provide empirical evidence on these issues). Further, to convincingly argue that some firms have less appropriate governance structures than other firms, it is important to identify the evolutionary process and/or crosssectional variation in contracting costs that allows such differences in governance to persist across firms and over time. In other words, if it is easy for shareholders to identify undesirable governance structures (which seems to be the case if one accepts the academic literature arguing that good and bad governance structures can be documented with seeming ease), and if it was not costly to correct such inefficiencies, shareholders would relatively quickly adjust to the more desirable governance structure. Explaining how and why some firms may have less desirable governance structures than other firms presents a host of interesting challenges and opportunities for future research. Toward this end, we provide several thoughts and suggestions here. To begin, we suggest that firm-specific evolution is likely to be important in explaining observed governance practices. Early on, most firms are closely held, with equity concentrated among entrepreneurs, venture capitalists, private equity firms, or other institutional or sophisticated investors. At this stage in the firms life, the selection of governance structures seems less hampered by the frictions that are expected to exist in widely held firms. As a result, one might expect that observed
54

As an analogy to emphasize the importance of identifying frictions, we note the large literature that explores efficiency in financial markets. When researchers identify apparent anomalous behavior in capital markets, such as profitable trading strategies, it is common to subject the trading strategies to a litany of tests to explore whether investors behave heuristically, or instead behave rationally, but are hampered by trading costs and frictions that would reduce or eliminate the profitability of the trading strategy. Granted, the speed with which capital markets can adjust prices to eliminate profitable opportunities is likely to be greater than the speed with which firms can adjust governance structures to eliminate profitable opportunities. The economic forces at play, however, are conceptually similar.

116

governance structures would be reasonably optimal at this stage of development (certainly, the owners of such closely-held firms have powerful incentives to structure governance optimally if they are to maximize the price at which the firm is sold to outside investors). Over time, firm change. Closely held firms become widely held, which creates a variety of frictions, informational demands, and free-rider problems with respect to making adjustments to governance structures. Growing firms become mature firms. Firms that originally had difficulty conveying information related to their operating strategy and potential for creating value find that financial reporting systems and other disclosure mechanisms are better able to reduce informational asymmetries between managers and outside investors. Thus, the appropriate governance structure from the shareholders perspective is an evolutionary process. There is no standard set of best governance practices that applies uniformly to every firm in the economy at any point in time. Accordingly, governance is developed through trial and error, being guided by the expertise of directors and shareholders, possibly with assistance from management and advice from consultants regarding governance practices at competing firms. This evolutionary governance process also highlights interesting and important measurement issues. Because all of the firms governance structures evolve simultaneously, one expects to find that certain combinations of governance mechanisms will be observed together (e.g., the proportion of outside directors and financial reporting quality). Larcker, Richardson, and Tuna (2007) and Dey (2008) are good examples of papers that attempt to identify how governance mechanisms are grouped together, with an eye toward developing our understanding of how the various dimensions of governance might be measured. These insights are likely to be helpful for future research in at least two ways. First, a better understanding of

117

the correlations between governance mechanisms should aid researchers in developing sharper hypotheses regarding substitution effects and complementarities among these mechanisms (an issue we currently know relatively little about). Second, better measurement of the various dimensions of governance might allow more powerful tests to identify firms where frictions are likely to prevent shareholders and directors from moving to a better governance structure. As evidenced by Larcker et al. (2007) and Dey (2008), it is difficult to document robust relations between governance measures and outcome variables, such as firm performance and accounting quality. Larcker et al. (2007) attribute this result to the difficulty in generating reliable and valid measures for the complex construct that is termed corporate governance (authors emphasis). We agree with this inference, but we would modify the statement slightly to state that it appears difficult to generate reliable and valid measures of good corporate governance. In other words, the various dimensions of the construct labeled corporate governance are likely to be much easier to identify and measure than whether a firms position along that dimensionality is the best place for it to be at any given point in time. However, we also acknowledge that better measures of the various dimensions of governance would also allow more refined predictions and more powerful tests regarding when shareholders and boards would be better off adjusting the firms position along the continuum of any given governance dimension.55 As a starting point, one might consider examining a firms position along the continuum of a certain governance dimension (e.g., board composition) relative to other firms with similar

55

There is also a preoccupation with attempting to collapse the multidimensional governance construct into a single scalar value as evidenced by the proliferation of univariate measures such as the G-Score (Gompers, et al. 2003). Although this exercise produces a single-dimensional variable which might be more amenable to empirical analysis, it undoubtedly discards information about how the various governance mechanisms interact. Instead, these measures implicitly assume some structure about how the various governance components interact (e.g., linearly) rather than use information about its underlying structure.

118

economic characteristics. For example, if most similarly-situated firms have a greater (or lesser) percentage of outside directors on their board, one might conjecture (i) that the shareholders would prefer to have the firm move toward the more commonly observed governance structure (given the firms type), (ii) that there is some firm-specific reason why a particular firm chooses to deviate from the typical structure at similar firms, or (iii) that the governance mechanism of interest deviates from that at similar firms because an alternative governance mechanism is being used instead to mitigate agency problems. As an example of this approach, Core and Guay (1999) provide evidence that when CEO equity incentives deviate from the typical level observed for similarly-situated firms, a dynamic process ensues where boards grant more or less new equity incentives to move the CEO toward the predicted target level. It is also important to note that one cannot interpret the mere existence of agency conflicts or actions stemming from agency conflicts as evidence that shareholders should prefer a different governance structure. At the most obvious level, Jensen and Meckling (1976) point out that no governance structure is likely to eliminate all agency conflicts. Thus, researchers should expect to observe evidence of agency conflicts in the actions taken by executives at even the most well governed firms. More subtlety, however, even when a researcher observes more severe agency conflicts, caution must be taken in interpreting such evidence as bad governance. Governance structures that are ex ante efficient may give rise to either anticipated or unanticipated agency conflicts ex post as the firm evolves over time. Further, even when shocks to the evolutionary process cause shareholders to view the current governance structure as less desirable than an alternative structure, switching costs may be sufficiently large so as prevent or impede progress toward the more desirable structure.56

56

Another analogy from the capital markets literature relates to the optimal rule for rebalancing a portfolio. In particular, a portfolio that is initially (or ex ante) optimal in terms of risk and expected return is likely to be sub-

119

As an example, consider that over time, firms hire and fire CEOs, with unsuccessful CEOs being removed, and successful CEOs becoming more powerful as an increasing function of their success and tenure.57 It is tempting to view powerful CEOs and evidence of agency conflicts, such as perquisite consumption, empire building and accounting distortions, as indicative of a breakdown of the governance system. However, as noted by Hermalin and Weisbach (1998), a by-product of a successful CEO is that this individual will gain bargaining power that can be used to extract rents. For example, Baker and Gompers (2003) find evidence consistent with successful CEOs being able to bargain for less independent boards. 58 Preventing the CEO from influencing the board structure is likely to be quite difficult, and firing the CEO may also be quite difficult and costly. Instead, shareholders might attempt to monitor the CEO through other mechanisms, such as using the influence of blockholders, the market for corporate control, or greater scrutiny of financial reports.

optimal at a later point in time since the prices of the component securities will change and will thus affect the weights of the various positions. Constantinides (1986) shows that, in the presence of transaction costs, the optimal rebalancing rule is to wait until the portfolio is far enough away from the optimum in risk and expected return space before rebalancing. In particular, an envelope can be drawn around the position of the optimal portfolio in risk and expected return space where the gains from rebalancing (in terms of an improved risk and expected return combination) outweigh transaction costs outside of the envelope. However, within the envelope, any gains from rebalancing are outweighed by transaction costs. Constantinides (1986) shows that the optimal rebalancing rule is not to rebalance back to the optimal portfolio, but instead to rebalance only to the boundary of the envelope. If we think of the firm as a nexus, or portfolio of contracts (which includes the firms various governance mechanisms), then this analogy suggests that when a firms governance environment deviates from the optimum as a result of changes in the underlying economic environment, we should not expect to see the firm alter its governance mechanisms back the optimum (in the absence of transaction costs). Instead, the optimal strategy is for the firm to adjust its governance mechanisms to get sufficiently close to the optimum to the point where the benefits from further improvement equal the costs of further adjustment. Since transaction costs and frictions are likely to be far more costly (in terms of both time and monetary outlay) in the context of corporate governance than in a capital markets setting, this suggests that it might in fact be efficient for firms to tolerate relatively large deviation from what appears to be optimal governance. In addition, unlike capital markets where shares are generally modeled as perfectly divisible, many governance mechanisms are discrete (e.g., terminate the CEO, adopt a poison pill, number of directors) which represents an additional source of friction. 57 Of course, firms can avoid being run by powerful CEOs by hiring and retaining less talented individuals, but this hardly seems like an attractive option. 58 One mechanism though which this occurs is the CEOs influence on the appointment of both inside and, in particular, outside directors to the board. A longer serving CEO will have had the ability to influence the appointment of a larger proportion of the firms directors.

120

Guay (2008) makes a related point with respect to decision-making discretion provided by boards to CEOs. In widely-held corporations, it is well understood that substantial decision rights are delegated by shareholders to the board of directors, due in part to the considerable information costs and coordination costs that would be necessary for shareholders to make many key decisions themselves. In turn, and for many of the same reasons, the board of directors delegates substantial decision rights to executive management. Executive managers involved in the day-to-day activities of the firm are likely to have much better decision-relevant knowledge, and can take actions in a timelier manner than either the board or shareholders. At the same time, shareholders and boards are rightfully concerned about agency conflicts with executive managers and the possibility that such managers will take opportunistic actions that benefit themselves to the detriment of shareholders. For example, it is not hard to imagine that executives who are granted greater decision-making discretion, in general, will have greater discretion across a variety of dimensions, including project selection, financing activities, mergers and acquisitions, executive compensation, perquisite consumption, and accounting policy. We also suggest that more research be conducted to identify and quantify the costs and frictions that prevent or impede firms from adjusting their governance structures, as well as how these frictions vary cross-sectionally. Cross-sectional variation in frictions is likely to be a function of economic characteristics of firms, such as organizational structure, ownership structure, information asymmetry between managers and shareholders, and the firms other governance structures that are in place. In addition to frictions that vary with firm characteristics, cross-sectional variation in regulatory regimes (e.g., across-state or acrossstock-exchange variation in regulatory environments) might lead to testable hypotheses

121

regarding variation in firms governance structures (e.g., relatively greater or lesser ability of shareholders to select directors or vote on proposals).59 Further, even when a group of sample firms operate within the same regulatory environment, regulations can also produce variation in frictions or contracting costs across firms. In addition, one might question what frictions prevent firms from finding a non-regulatory solution to undesirable governance structures, at least over time. To date, much of the literature on the relation between governance structures and regulatory environments has been conducted at the cross-country level. Yet, many of the studies on inefficient governance structures are conducted using single country samples of U.S. firms.

4. Accounting information and ownership structure In this section, we discuss agency conflicts between controlling shareholders (i.e., shareholders with sufficient ownership interest to influence or control company policy) and minority interest shareholders. In firms with no controlling shareholder, or where the objective of the controlling shareholder is aligned with the objective of minority shareholders (i.e., to maximize the present value of the firms expected future cash flows), the interesting agency conflicts are between shareholders (as a group) and other contracting parties, such as directors, managers, and creditors. However, when there is a controlling shareholder that has a somewhat different objective function than minority shareholders, such as extraction of private benefits of control, the minority shareholder becomes more like other suppliers of capital, such as creditors, in terms of agency conflicts and contracting issues.
59

As an example of across-state variation in regulatory environments that gives rise to cross-sectional variation in governance structures, a number of recent studies beginning with Bertrand and Mullainathan (2003), examine the effect of changes in the state business combination laws that occurred in the mid- to late-1980s and early 1990s. These laws generally made it more difficult to acquire firms and thus are thought to have reduced the efficacy of the market for corporate control.

122

Controlling shareholders are typically assumed to have fewer agency conflicts with managers and boards of directors (vis--vis widely-held firms) because there is little separation between ownership and control, board members and managers can be hand selected, and the direct monitoring of managers is more intense. As a result, the demand for high quality public disclosures and financial reporting for the purpose of monitoring management seems less important in firms with controlling shareholders compared to in firms with dispersed ownership that rely more heavily on outside directors to monitor management (see LaFond and Watts, 2008 for a discussion of this point). However, although manager-shareholder conflicts are smaller in firms with controlling shareholders, there are important agency conflicts between minority shareholders and controlling shareholders, the latter having the ability and incentives to extract private benefits from control, such as perquisite consumption, self-serving investments, etc. (e.g., see Schleifer and Vishny, 1997). Agency conflicts between controlling and minority shareholders are frequently categorized in the literature as a governance issue (which we defined above as conflicts in aligning the actions of management with the interests of shareholders). We view the conflicts between controlling and minority shareholders, however, as being more akin to agency problems with creditors and other outside (non-shareholder) contracting parties. At some point in the evolution of a firms ownership structure, controlling shareholders decide whether or not to raise capital from minority shareholders or creditors, or both. Analogous to creditors expectations regarding borrowers, minority shareholders recognize the potential for agency conflicts with controlling shareholders. And, as with creditors, minority shareholders are expected to price-protect their claims against these agency conflicts (i.e., they will pay less for shares when agency conflicts are greater).

123

Controlling shareholders, for their part, can put bonding mechanisms in place ex ante to commit to not take advantage of minority shareholders and thus mitigate the degree of price protection. That is, controlling shareholders are expected to internalize the costs associated with maintaining liberal private benefits of control, in much the same way that borrowers internalize the retention of operational and investment flexibility in contracting with creditors. But, since monitoring and bonding is costly, and residual agency costs are expected to persist (e.g., inefficient investment policy in certain states), one expects these agency conflicts to lower firm value (again similar to the effects of residual stockholder-bondholder conflicts). It is important to note, however, that the welfare loss to the controlling shareholder stemming from agency conflicts is less than the decline in firm value. This is because the controlling shareholder obtains private benefits that at least partially offset the lower proceeds received from minority shareholders (Jensen and Meckling, 1976; DeAngelo and DeAngelo, 1985). Further, because controlling shareholders have paid for their private benefits of control through a lower share price, controlling shareholders are not expected to willingly relinquish such benefits. Indeed, ex post actions (or regulations) that limit the controlling shareholders ability to obtain their private benefits of control would serve as a windfall gain to minority shareholders (who paid a price for their shares that considered such benefits). This setting is again similar to contractual agreements with creditors, whereby lenders charge a higher rate when borrowers have greater ability to make wealth transfers in future states. And, ex post actions (or regulations) that prevent borrowers from making such wealth transfers would result in a windfall gain to creditors.60

60

Interestingly, it would seem that, ex ante, both controlling shareholders and borrowers might welcome regulations or other commitment mechanisms that restrict their ability to obtain private benefits and wealth transfers, respectively. That is, although such restrictions would result in a windfall loss to these parties when implemented after the contracts are in place, such restrictions would benefit controlling shareholders and

124

The above discussion highlights the important security design issues that controlling shareholders face with minority shareholders, and the similarity to shareholder-creditor agency conflicts. As we discuss in detail below, accounting information plays an important role in mitigating agency conflicts between classes of shareholders that have different objective functions, voting rights or cash flow rights. In particular, a commitment to high quality, transparent financial reporting may lower the cost of monitoring by minority shareholders, thereby reducing concerns about controlling shareholders extracting private benefits of control. In this section, we briefly review some of the accounting literature on these issues. DeAngelo and DeAngelo (1985) examine the firms decision to issue dual class stock. Specifically, they focus on a small sample of firms (45) that have two classes of stock with equal cash flow rights but different voting rights. They find that managers are more likely to have the class of stock with superior voting rights. They suggest (but do not test) that managers may be more likely to retain control when it is costly for them to provide external shareholders with information that allows outside investors to make decisions regarding selection of investment opportunities and/or evaluation of managerial performance. This argument is akin to the notion that decision rights are more efficiently allocated to the party that has the best information about the decisions that need to be made (Jensen and Meckling, 1995), and appears to suggest a rich set of predictions on the relation between the quality of accounting information and the evolution of ownership structure. Francis, Schipper and Vincent (2005) examine the financial reporting informativeness of firms with dual class stock, and find that dual class stock firms have lower earnings informativeness (as measured by the return-earnings relation) compared to a sample of firms

borrowers if they are implemented prior to the contracting arrangement and would aid controlling shareholders in bonding against taking certain actions that are detrimental to other parties.

125

that have a single class of stock holders. In a related paper with similar findings, Fan and Wong (2002) examine a large sample of East Asian companies that have cross-sectional variation in ownership structure and weaker minority rights, and provide evidence that ownership control is negatively related to earnings informativeness. These results appear consistent with the DeAngelo and DeAngelo conjecture that dual class stock is a response to costs that some firms face in providing external investors with adequate information to ensure efficient decisionmaking if control is relinquished to outsiders. An alternative interpretation for the Francis et al. result, and one advanced by the authors (and by Fan and Wong, 2002), is that controlling shareholders in dual class firms use their voting rights to obscure information to outside parties, thereby enabling them to obtain greater private benefits of control. These are interesting competing explanations that highlight the importance of establishing causality between the choice of a controlling shareholder system and the financial reporting environment. Specifically, the DeAngelo and DeAngelo (1985) conjecture is that the lack of transparency in the information environment drives the decision to have a dual class system (or a controlling shareholder), whereas the Fan and Wong (2002) and Francis et al. (2005) conjecture is that the choice to have controlling shareholders that drives the decision to produce an information environment that lacks transparency (and presumably the controlling shareholders also choose to bear the costs imposed on them by minority shareholders at the time they choose to maintain control). Like Fan and Wong (2002) and Francis et al. (2005), Fan and Wong (2005) also recognize the informational advantage of controlling shareholders vis--vis minority investors, and that agency conflicts between controlling and minority investors can be exacerbated by this informational advantage. Fan and Wong (2005), however, also emphasize that concentrated

126

ownership can serve to improve contracting efficiency in certain environments. In particular, in emerging markets with weak property rights, controlling shareholders may be better able to negotiate and enforce contracts with outside parties. Therefore, because the purpose of controlling shareholders in this setting is not primarily to extract private benefits of control, controlling shareholders have incentives to introduce monitoring and bonding mechanisms to limit their ability to extract private benefits. One such mechanism is a commitment to timely and credible financial reporting that allows minority investors to more clearly see and potentially take action against certain behaviors by controlling shareholders. Consistent with this hypothesis, the authors examine a broad sample of East Asian firms, and find that concentrated ownership is positively related to the likelihood that a high quality auditor (i.e., Big 5) is selected to review and approve financial statements. Wang (2006) and Ali, Chen and Radhakrishnan (2007) make similar arguments in the context of family-owned firms (i.e., firms where a founder (or descendents) is a large shareholder or holds a top management position). As with controlling shareholders, family firms recognize the potential agency conflicts with outside/minority investors, and are expected to respond with a commitment to high quality financial reporting to mitigate such conflicts. Ali, Chen and Radhakrishnan (2007) find that family-owned firms are more timely in reporting bad news, have higher quality earnings, greater analyst following, and smaller bid-ask spreads. Wang (2006) finds similar results, reporting that founding family ownership is associated with lower abnormal accruals, greater earnings informativeness, and less persistence of transitory loss components in earnings. An interesting and seemingly contradictory finding is provided by Chen, Chen and Cheng (2008) who find that family ownership is negatively related to voluntary disclosures via earnings forecasts and conference calls. They interpret this result as consistent

127

with family-owned firms being more actively involved with management and not requiring public disclosures to effectively monitor management. A seemingly interesting (unanswered) question with respect to ownership structures with controlling shareholders is whether such firms use financial reporting as a commitment mechanism that restricts their ability to extract private benefits of control, or instead actively distort financial reporting to facilitate greater extraction of private benefits. The eventual answer to this question likely rests on whether controlling shareholder ownership structures serve to facilitate more efficient investing or operating decisions (e.g., as suggested by DeAngelo and DeAngelo, 1985), or instead are simply chosen to allow entrepreneurs or founding families to extract private benefits of control.61 If the later, one must take into consideration that minority shareholders will require entrepreneurs to bear the costs associated with ownership structures that facilitate private benefits, and as a result, entrepreneurs must derive sufficient utility from the private benefits to outweigh the residual agency costs. Further complicating the expected relation between financial reporting quality and ownership structure are economic arguments predicting that the demand for accounting information by outside contracting parties varies cross-sectionally with ownership structure. For example, Fan and Wong (2002) and Francis et al. (2005) raise the hypothesis that because controlling shareholders can monitor management without public disclosures, there may be less governance-related demand for high quality financial reporting, thereby allowing controlling
61

We note that in addition to (or instead of) using financial reporting as a commitment mechanism, firms may instead elect to use other corporate governance mechanisms to protect minority shareholders. For example, Reese and Weisbach (2002) suggest that firms domiciled in countries with weak protection of minority shareholders have incentives to cross-list in the U.S. when raising capital, as the U.S. has stronger shareholder rights and better financial reporting, both of which have been shown to reduce cost-of-capital. Consistent with this idea, Lang, Raedy, and Yetman (2003) find that firms that cross list in the U.S. have less earnings management, more timely loss recognition, and more conservative earnings reports than a matched sample of firms that does not have crosslisted shares. These differences are attributable to both innate differences in the firms, as well as the likely perceived benefits of cross-listing. See Coffee (2002) for a more thorough discussion of the literature examining the effects of cross-listing.

128

shareholders to protect proprietary information through less transparent reporting. Ajinkya, Bhojraj, and Sengupta (2005) make a similar argument, suggesting that institutions with concentrated (blockholder) ownership have access to privately obtained information and are less likely to demand high quality, timely disclosures. Finally, Ball and Shivakumar (2005) investigate differences in the quality of accounting reports (using timely loss recognition as a metric) across public and privately held firms in the U.K. They argue that there is greater demand for high quality financial reporting in public firms vis--vis private firms because the former engages in substantially more arms length transactions with outside contracting parties. Outside contracting parties face greater information asymmetries, which can be reduced to some extent through timely financial reporting. Ball and Shivakumar provide evidence consistent with this hypothesis, finding that publicly held firms have more timely loss recognition than privately held firms.

5. The role of financial accounting in debt contracting One of the more important and widely researched aspects of the firms nexus of contracts is the contractual agreement between the firm and various suppliers of capital. In the corporate governance section of this survey, we focused on the set of contracts that govern the relationship among managers, boards of directors, and the shareholders that supply equity capital. In the previous section, we focused on the scenario where the firm has a controlling shareholder, and there are agency conflicts between this shareholder and minority shareholders. In this section, we emphasize that firms often raise additional capital beyond equity, such as debt and preferred stock, which can give rise to agency conflicts between these suppliers of capital and the firms shareholders. In these contractual settings there is an allocation of voting

129

rights (or control rights), cash flow rights, and information rights that defines the contractual relationship. We begin our discussion by focusing on the importance of accounting information in the contractual relations between the debt holders and the firm, and then briefly discuss the role of accounting information in agency conflicts between controlling and minority shareholders. Debt is the primary source of capital raised by United States (U.S.) corporations. During the calendar year ending December 2006, corporations domiciled in the U.S raised more than $2.6 trillion of capital.62 Of the $2.6 trillion, more than 95%, or about $2.5 trillion, was some form of debt financing (i.e., bonds, syndicated debt, or other types of loans). The remaining 5%, or about $0.1 trillion, was in the form of common and preferred equity. Similar statistics hold for 2005 and 2004. Despite the relatively larger size of the debt markets as compared to the equity markets, accounting researchers have focused much more on the role of accounting information in equity markets. For example, in the survey papers prepared for the 2000 Journal of Accounting and Economics Conference, Kothari (2001) and Fields et al. (2001) cite approximately 700 papers related to capital markets and accounting choice research. By our count, fewer than 10 percent of these papers (roughly 50 papers) focus on accounting issues related to debt markets. 63 There are several reasons for this emphasis on equity research. First, equity markets are typically much more liquid and active (in terms of trading volume) than debt markets, are subject to greater volatility in prices, and have more individual investors, potentially making them more interesting to researchers. Second, data on equity returns, investors, and equity analysts has

62

The data supporting these statistics was obtained from the following website, http://www.federalreserve.gov/econresdata/releases/corpsecure/current.htm. 63 Note these are rough approximations. Kothari (2001) cites roughly 550 papers and Fields et al (2001) cite roughly 150 papers; we did not eliminate overlapping papers when calculating these statistics.

130

been readily available while data on debt contracts, debt analysts, and debt market prices has historically been much more limited. Over the last decade, however, there has been a noticeable increase in the number of published papers focusing on accounting issues in the debt contracting process. We find this trend to be encouraging. As new data become available, and the research intensifies, we expect to have a greater understanding of the role of accounting information in the debt contracting process. Much of the debt research has developed with an agency theoretic view of the firm. Jensen and Meckling (1976) provide a framework in which agency costs arise from the differing objectives of debt and equity holders. They argue that owner/managers have incentives to engage in actions to further their own self-interests to the detriment of external capital providers. Rationally anticipating that external capital providers will price-protect their claims in anticipation of this behavior, owner/managers have incentives to encourage monitoring activities and to bond themselves against engaging in such behavior until the costs of monitoring and bonding exceed the benefits. Complete resolution of agency conflicts (i.e., the first-best outcome), however, is likely to be either undesirable or prohibitively costly (relative to the second-best outcome), resulting in residual agency costs that are borne by the owner/managers (in addition to bearing the monitoring and bonding costs). Fama and Miller (1972), and Myers (1977) provide similar motivations for how agency problems can arise between debt and equity holders. In particular, these papers highlight the differential payoff of debt and equity holders, and that debt holders typically only have state-contingent voting rights or control rights.64 These differences give rise to incentives for equity holders to engage in

64

The literature has long recognized that creditors can assert their control rights in bankruptcy proceedings (e.g., Gilson, 1990). Recently there has been recognition within the literature that creditors can exert control over investing and financing activities in settings where the borrower has not missed a payment, such as when a covenant is violated (e.g., see Roberts and Sufi, 2008; Chava and Roberts, 2008; and Nini, Smith and Sufi, 2009).

131

actions ex post (i.e., after the contract) that will result in a transfer of wealth from debt holders. Smith and Warner (1979), Watts and Zimmerman (1978, 1986, 1990), and others suggest that accounting information can play an important role in reducing the agency costs that arise in the debt contracting process. Smith and Warner (1979) postulate that there are four categories that encompass typical agency conflicts that arise between debt and equity holders. First, there is a conflict of interest between these two stakeholders over dividends. When bonds or other forms of debt are initially priced, they are priced based on assumptions about expected future dividend policies. If, after the bond is issued, the firm were to increase its dividend payments more than was expected ex ante (i.e., more than the lenders price protected themselves against), it would reduce the resources available to pay off debt holders claims and would result in a wealth transfer from debt holders to equity holders. Creditors and equity holders recognize this incentive and frequently agree to include contractual features that limit the firms ability to pay dividends. Second, there is a conflict between a lender and equity holders over future increases in debt levels that reduce the probability of the lender being repaid. When a firm issues more debt than was expected by earlier creditors, a wealth transfer can occur from the earlier creditor to the equity holders, especially if the new debt is of equal or higher priority, or of shorter maturity than the earlier debt. Equity holders recognize that this concern will cause creditors to charge a higher price for debt capital, and so frequently agree to contractual restrictions against issuing certain types of debt. The third and fourth sources of conflict between debt and equity holders relate to asset substitution and under investment, respectively. Following a debt issuance, firms can have incentives to shift their asset mix toward riskier investments, which results in a wealth transfer

132

from debt to equity holders.65 In addition, when firms become riskier and approach default they may choose to forgo positive net present value (NPV) projects because the benefits would accrue to the firms creditors (rather than its equity holders). Although it is relatively easy to prevent firms from shifting tangible assets by using security agreements to collateralize the firms assets and/or restricting its future investment, it is much more difficult for firms to commit to invest in all positive NPV projects or to maintain a given risk profile when exercising future growth options. Typically banks and equity holders can only devise contractual mechanisms that indirectly mitigate these agency problems. Smith and Warner (1979) provide examples of actual corporate debt agreements, and a discussion of typical bond provisions included in the Commentaries on Corporate Indentures, to illustrate how debt contracts are designed to reduce these stockholder-bondholder agency problems. They highlight that many of the provisions included in debt contracts to reduce agency costs (e.g., covenants, sweeps, borrowing base requirements) rely on outputs from the borrowers accounting system. Thus, attributes of the borrowers accounting system are likely to be important to the lender both at the time the firm enters into the debt contract and over the life of the contract (Watts and Zimmerman, 1986, 210-213). For example, if the lender is concerned with the dividend payout problem, then the borrower might agree to include a clause in the contract to limit the firms ability to pay dividends. Such a clause may allow the firm to pay dividends up to a set dollar amount (e.g., $1,000,000), or alternatively, may allow dividends to be paid as a function of certain outputs of
65

The risk-seeking incentives of equity holders are best illustrated by viewing the equity holders claims as a call option on the value of the firms assets with an exercise price equal to the face value of the firms debt (Merton, 1974). Option valuation theory indicates that the value of a call option (to a risk-neutral holder) is strictly increasing in the volatility of the underlying asset (which, in this case, is the firms assets). The magnitude of such risk-shifting incentives, however, is expected to be relatively small for most financially healthy firms (when the option is substantially in-the-money), but of greater significance for financially distressed firms (when the option is at- or out-of-the-money). See Guay (1999) and Parrino and Weisbach (1999) for empirical evidence on the magnitude of stockholder-bondholder conflicts arising from risk-shifting incentives.

133

the accounting system (e.g., dividends are limited to 50% of positive net income). The choice between these two alternative constraints is likely to depend on a variety of factors, such as the extent to which the firm has made conservative accounting choices, the discretion the accounting system provides to the manager to make income-increasing or income-decreasing accounting choices (e.g., changes in accounting methods), and the managers ability to delay or accelerate revenue recognition. Consider the case where the borrower and lender agree to limit dividends to be 50% of positive net income. When negotiating the contract, the lender and the borrower must make an assessment of the costs and benefits of including this clause in the contract. By making the dividend restriction a function of net income (as opposed to a fixed amount), the borrower has more flexibility to make dividend payments if they have good outcomes, and less flexibility to make dividend payments if they have bad outcomes. Thus the lender is likely to assess the quality of the firms accounting reports when deciding to include this clause in the contract. In addition, by making the threshold a function of net income, the borrower may be provided with the flexibility to make accounting choices that increase (or decrease) their ability to make dividend payments. Thus the decision to include this clause in the contract will depend on the costs and benefits of increases (or decreases) in the flexibility to make dividend payments. One of the factors affecting these costs and benefits is the quality of the firms financial reporting system. If the firms financial reporting system provides borrowers with significant discretion to make income increasing accounting choices, then the borrower will have more flexibility in paying dividends, potentially affecting the probability (and costs) of default. Alternatively, by providing the borrower with discretion to make income increasing accounting choices, the lender also affords the borrower with the flexibility to return excess cash to

134

shareholders. By reducing excess free cash flows, the manager would be less likely to invest in negative NPV projects or consume perquisites. Conditional on including this clause in the contract, the debt contract is likely to influence the managers subsequent accounting choices. For example, a firm may consider switching from first-in first-out (FIFO) to last-in first-out (LIFO) to account for its inventories, but this choice may cause income to drop below zero, making the dividend constraint binding. Thus this accounting choice will affect whether the firm violates their covenant, whether the firm changes their dividend policies, or whether the firm has to forgo this (otherwise beneficial) accounting change in order to continue paying dividends, thereby affecting managers accounting choices over the life of the contract (Watts and Zimmerman, 1986, 213-217). Alternatively, the firm might seek to renegotiate the covenant, increasing its ability to pay dividends. If the borrower and lender do renegotiate, the lender is likely to reassess the borrowers performance since the inception of the contract. Thus the accounting choices the firm has made since the inception of the contract are likely to affect the outcome of this renegotiation and ultimately whether the bank provides the borrower with the flexibility to increase dividends as well as how much flexibility is provided to the borrower. To summarize, there are a variety of different attributes of a firms accounting reports that are likely to affect whether the firm is able to obtain debt financing, the design of the contract, whether covenants are violated and, if so, how the violation is handled, and other features of the contract. As a result, the debt contract provides firms with incentives (both ex ante and ex post) to make accounting choices to minimize expected contracting costs. In the following summary and synthesis of the debt contracting literature, we focus on the importance of attributes of the accounting system at various stages in the contracting process, and how debt

135

contracts provide incentives that influence the firms accounting choices. First we focus on how elements of a firms accounting system affect whether the firm is able to enter the debt market (and if so which segment(s) of the debt market are accessible). Next we discuss how attributes of the accounting system affect the design of debt contracts, highlighting features of the debt contract affected by outputs of the accounting system. We then discuss how, conditional upon obtaining debt financing, the debt contract is likely to affect outputs of the accounting system. Towards the end of this section, we provide some thoughts on how future research might focus on how attributes of the accounting system may affect the outcomes of the renegotiation process. We then discuss a relatively recent extension of agency theory to incomplete contracts, which suggests a new and potentially important role for accounting information in the debt contracting process. We conclude this section by offering more general avenues of future research. 5.1. The role of financial accounting in determining whether and from whom the firm obtains debt financing At the most fundamental level, characteristics of the firms accounting system are likely to affect whether the firm is able to obtain debt financing. Typically, before providing borrowers access to capital, lenders require firms to provide audited financial statements, budgets, forecasts, and other financial information to assist the lender in assessing the borrowers ability to repay the debt. In particular, when deciding whether to provide funds to the borrower, the lender is likely to evaluate the firms collateral, as well as the timing and riskiness of the firms expected future cash flows from existing projects and anticipated investments. The firms financial reporting system plays a vital role in this analysis. If the firms accounting reports provide unreliable asset values, opaque performance

136

measures, or performance measures that make it difficult to forecast either future cash flows or their riskiness, then the lender is likely to have difficulty assessing the firms credit quality and the firm is less likely to be able to cost-effectively access the debt market. There has been relatively little research on the role of financial reporting in the lending decision, with researchers instead tending to focus on whether, conditional on obtaining debt, certain attributes of the accounting system affect the borrowers choice of debt markets. We conjecture that this is because researchers cannot directly observe firms that would like to borrow but are unable to find a willing lender and therefore do not participate in the debt markets. It seems however, that one could construct an empirical model to predict the demand for debt financing as a function of various firm characteristics, and then use the predictions derived from such a model to study firms that are expected to have debt financing but do not. We also encourage researchers to also explore the more fundamental decision of how financial reporting influences the firms decision regarding the type of financing to pursue (e.g., whether to borrow from the debt markets, issue equity, or choose some alternative source of financing). Although a detailed discussion of capital structure is beyond the scope of this review, we note the importance of information asymmetry between owners/managers and capital providers in some of key theories of capital structure (e.g., Myers and Majluf, 1984). Although there is relatively little research on the role of financial reporting in capital structure decisions, Liang and Zhang (2006) is a relatively recent paper on the role of financial accounting in the decision to issue debt vs. equity. Conditional on deciding to access the debt markets, firms recognize that there are institutional differences between various suppliers of debt capital that give rise to differing demands for certain attributes of accounting information. For example, the simplest partition of

137

the debt market is whether the debt is provided by banks (i.e., private debt) or by public investors. Private debt holders, like banks, typically have access to private information, while public debt holders typically do not. More specifically, since the enactment of Regulation FD, both private debt holders and ratings agencies are exempt from the disclosure restrictions promulgated under this regulation, and thus borrowers are allowed to provide these parties with access to management and to non-public information. Private debt holders base their lending decisions on a variety of proprietary, non-public information, including financial forecasts, detailed sales data (by region, store and/or segment), aging of inventory, payables and receivables, capital expenditure budgets, and a variety of other information that the public can not access. Private debt agreements often stipulate the types of public and proprietary information that must be provided to the private debt holder prior to the loan initiation, and throughout the life of the loan. If the ratings agency is solicited by the firm (and thus paid for by the firm) to make their ratings forecast, the firm is allowed to share nonpublic information with the ratings agency as well. Presumably this information is considered when the ratings agency develops its debt rating. There are other institutional differences across these two markets. Public debt is often diffusely held and typically requires the approval of two thirds of the bondholders for a renegotiation (Smith and Warner, 1979). As a result, public debt has relatively larger renegotiation costs than private debt. Public debt is often underwritten by an investment bank, subject to SEC regulations and filing requirements, and is bought and sold on an exchange by pension funds, insurance companies and other institutional investors. Private debt is not underwritten, and generally is not subject to SEC regulations and filing requirements. Some private debt (typically syndicated debt) is, however, traded and held by institutional investors,

138

but the primary capital providers are banks (see Wittenberg-Moerman, 2008 for a discussion of the institutional features of the syndicated debt market). The differential access to private information as well as other institutional differences that exist across these two markets are likely to affect the contractual features that are included in the debt agreement, and thus the incentives that the debt contract is likely to provide the borrower over the life of the contract. The differential access to private information is also likely to result in differences in the types of information that the borrower requires at inception of the loan and over the life of the contract. For example, Smith and Warner (1979) suggest that ceteris paribus, if public debt contains covenants then they will be looser than private debt covenants because of the higher renegotiation costs. Consistent with this conjecture, Begley and Freedman (2004) focus on the use of covenants and provide evidence that public debt seldom has financial covenants. They suggest that the lack of financial covenants in public debt is likely because coordination costs across public debt holders would make renegotiation very costly. Research on the role of the accounting system in the choice to obtain public versus private debt has also focused on issues related to the costs of obtaining information and renegotiating contracts. For example, Bharath et al. (2008) predict that banks have better access to private information and superior information processing abilities, which allows them to reduce adverse selection costs and lend to firms with lower accounting quality. Consistent with this prediction, they examine the choice of public debt versus private bank debt and find that firms with lower accounting quality are more likely to borrow from banks. Further, Dhaliwal et al. (2005) focus on the role of disclosure costs in accessing public versus private debt markets and find that when the cost of making public disclosures is high, firms are more likely to access

139

private debt markets in order to avoid having to make costly disclosures. Beatty, Liao, and Weber (2008) investigate the determinants of the decision to lease as opposed to use equity or other forms of debt financing. They find that firms with lower quality accounting are more likely to enter into leases, arguing that the legal benefits afforded to lessors in bankruptcy proceedings allow them to provide financing to firms with lower quality accounting. They note, if the lessee goes bankrupt, the lessee can either accept or breach all lease obligations. If the lessee accepts the obligations, the lessor is entitled to continue receiving payments based on the original lease agreement. The rental expenses, along with lessors other claims, are classified as administrative expenses in the bankruptcy code, which must be paid immediately or within a reasonable period. Thus, the bankruptcy filing does not affect the lessors claims. Alternatively, if the lessee breaches the lease obligations, then the lessor can immediately repossess the property. This provides them with relatively more security as other creditors have no assurance that their claims will be recovered. To summarize, the research investigating the role of the accounting system in the firms decision to access public debt, private debt, or leasing suggests that the institutional features of the different debt markets allows them to cater to different types of borrowers that have accounting systems of differing quality. Public debt holders who do not have access to private information or management appear to require the highest quality accounting reports in order to provide financing. Public debt is also the least likely to include accounting based covenants, as coordination among the lenders if there is a technical default is difficult. Borrowers with lower quality accounting can still obtain financing from private debt holders, who can use private information obtained from management to perform additional due diligence, or from lessors, who both have access to private information, and are afforded greater protection if the firm

140

faces financial distress. There are also institutional differences across lenders who provide capital within the private debt markets, as well as differences across the features of debt contracts offered by these lenders. For example, private debt can be supplied by insurance companies (or other institutional investors) or through banks.66 As El-Gazzar and Pastena (1990, 1991) discuss, banks and insurance companies typically have different investment horizons. Bank liabilities (i.e., deposits) typically have a shorter maturity than the liabilities of insurance companies and thus they typically look to provide shorter term lending to better match the duration of their assets and liabilities. Further, private bank loans typically involve multiple lenders (i.e., syndication), while insurance company loans typically have a single lender. These institutional differences lead to differences in monitoring and renegotiation costs, which in turn lead to differences in covenant packages and differences in the role of financial accounting numbers. El-Gazzar and Pastena (1990) find that 100% of the private placements (i.e., insurance funded debt) in their sample made tailoring adjustments to GAAP in their financial covenants, while only 21% of the bank debt contracts had similar adjustments. Further, they find that insurance companies typically impose more restrictions (covenants) than banks, but El-Gazzar and Pastena (1991) show that insurance company restrictions have more slack at initiation. Because tailoring of GAAP and greater restrictions require more monitoring by lenders, these findings are consistent with insurance company loans being characterized by greater monitoring and lower renegotiation costs (due to the presence of only a single lender). At the same time, the greater covenant slack at initiation of the longer-term insurance company loans is consistent with borrowers requiring somewhat greater flexibility to efficiently run the

66

Private placements are a form of debt financing which is typically provided by insurance companies or other non-banking institutional investors

141

operations of the firm over a longer time horizon. More recent research examines multiple lender arrangements (i.e., syndicated loans) in greater detail. In recent years, the syndicated loan market has been one of the fastest growing sectors of the U.S. capital markets growing at a 27% annual rate between 1991 and 2003, with over $1 trillion in new loans each year, and comprising more than 50% of annual issuances in U.S. equity and debt markets (see Wittenberg-Moerman, 2008). Sufi (2007) and Ball, Robin and Sadka (2008) examine how information asymmetry and accounting quality affect the role of lead arrangers in the loan syndicate. Specifically, they argue that when there is greater information asymmetry between syndicate lenders about the financial characteristics of the borrower, the lead lender is induced to more actively monitor the borrower, which in turn induces the lead lender to hold a larger proportion of the loan and form a more concentrated syndicate. Sufi (2007) provides evidence consistent with these predictions. Ball et al. (2008) extend the evidence in Sufi (2007) to examine the role of financial statements in reducing the information asymmetry among lenders. They find that when financial reports provide more informative signals about future credit quality, lead lenders are able to hold a smaller proportion of the overall syndicate and the syndicate can become less concentrated. Within the syndicated loan markets, some loans are traded publically, and others are not. Wittenberg-Moerman (2008) investigates the relation between the quality of financial reporting and information asymmetry between investors in the syndicated debt market. It is well understood that firms have incentives to produce high quality accounting reports to ensure low information asymmetry between investors in the equity market (i.e., lower cost of capital through lower bid-ask spreads). For the same reason, firms also have incentives to produce high quality accounting reports to ensure low information asymmetry among investors in the debt

142

market. Wittenberg-Moerman uses the timeliness of accounting earnings as a proxy for high quality accounting reports, and uses the bid-ask spread on debt securities as a measure of information asymmetry between investors in the syndicated debt market. She finds that timely loss recognition, but not timely gain recognition, is negatively associated with the size of bidask spreads. This result is consistent with the discussion in Section 3 that because managers are expected to be forthcoming with good news but reluctant to disclose bad news, a commitment to timely loss recognition improves the transparency of the information environment. This argument is even more compelling with respect to debt markets; because debt is a fixed claim on a firms cash flows (i.e., it has a limited upside), the value of a lenders investment is more sensitive to bad news vis--vis good news. Biddle and Hilary (2006) take an alternative approach to provide evidence on the relation between accounting quality and access to debt markets. They note that when investors do not have adequate information about a firms activities and performance, they may withhold capital, thereby financially constraining the firms and restricting its ability to invest. They argue that high quality accounting reports can mitigate information asymmetry and reduce firms reliance on internal capital for investment. Using both country- and firm-level tests, they find that when firms have higher quality accounting there is less reliance on internally generated cash flow when making investments. They then hypothesize that accounting quality is more important in economies that rely on arms-length financing through public stock markets than in economies that rely more on private creditor financing, since in the latter case creditors can obtain additional information about firms through private channels. The authors use the U.S. as their setting where public stock markets are most important, and Japan as their setting where private credit financing is most important. They find that the relation between accounting

143

quality and investment-cash flow sensitivity (which is a measure of the extent to which firms face financing frictions) holds in the U.S., but not in Japan. A concern with this approach, however, is that the differential use of debt financing between firms domiciled in the U.S. and in Japan is not clear. As noted above, the vast majority of capital raised by U.S. corporations is debt, with privately-structured syndicated debt arrangements accounting for a substantial and growing proportion of this financing. Beatty, Liao, and Weber (2008) expand on this paper, providing evidence that when firms borrow from private lenders in the U.S., accounting quality appears to reduce investment-cash flow sensitivity. They also provide evidence that certain contractual features such as investment restrictions and lender monitoring may substitute for accounting quality and reduce its importance in financing decisions in debt markets. In addition to the decision on whether to borrow from public versus private lenders, or and deciding among the various different sources of private debt financing, firms must also decide whether to enter into a relationship lending arrangement, where they borrow from the same lender over time. There is an extensive theoretical finance literature on this topic.67 Relationship lending involves both an explicit debt contract, and also has an implicit contract, where the lender and borrower can build reputational capital over time. Thus relationship lending has many aspects of a repeated game, and many of the concepts that we discussed in Section 3 in the context of multi-period executive compensation contracting are also applicable here. Firms that decide to pursue a relationship with a bank may potentially benefit as a good payment record will help to establish the firms reputation, will reduce the need for monitoring,

67

As we discuss in Section 3, the ideas of contract memory and reputation become important in repeated game settings.

144

and will lead to improvements in future loan contract terms (for some examples of the literature supporting this point see Diamond, 1991; Rajan, 1992; and Boot and Thakor, 1994). However, relationship lending does not come without a cost. Firms that enter into a relationship with a bank may potentially suffer from a hold-up problem, where the bank seeks to extract rents before providing additional financing, or renewing existing financing (Rajan, 1992). One might expect that the firms accounting system would play a vital role in relationship lending. For example Ongena and Smith (2001) suggest that firms facing large information asymmetries with outside investors stand to benefit most from long-term bank relationships but are also particularly susceptible to holdup problems and high switching costs. This argument implies that firms whose accounting systems are least effective in reducing information asymmetry may benefit the most from relationship lending, but also may incur the most in holdup costs once they are in a relationship with a bank. One could extend this argument along several dimensions on how attributes of the firms accounting system affect whether the firm establishes a relationship with a bank, and the outcome of this lending relationship. For example, firms with the lowest accounting quality should reap the most benefits (lower interest rates, looser contract terms) when they enter into a relationship with a bank. Similarly, the arguments of Ongena and Smith (2001) would also suggest that contract renegotiation costs (in terms of waiver fees, higher interest rates, etc) will be the highest when the firm has a relationship with its bank, and has an accounting system that is relatively ineffective in reducing information asymmetries. Alternatively, one might conjecture that relationship lending may instead reduce the importance of accounting information in the lending process. Firms that have established a relationship with a bank would presumably have reputational capital and a proven track record,

145

which would reduce the reliance on the firms accounting system to reduce information asymmetries. This argument would suggest that the role of accounting information in lending decisions would, in part, depend on whether the firm has a relationship with their lender. We are unaware of any papers testing this conjecture. In summary, the evidence from research examining the role of the accounting system in the choice of lenders suggests that elements of the accounting system are associated with the choice of lender, the lending market entered, the amount of the loan held by the lead lender, and the size of the bid-ask spread offered in the syndicated loan market. Although we are unaware of any empirical evidence on the topic, we conjecture that attributes of the firms accounting system are also likely to affect whether or not the firm decides to enter into (and maintain) a relationship with their bank. Consistent with the observations in Beatty (2008), we note that it is unclear exactly which elements of the accounting system are important in these decisions, and each of the papers discussed above seems to focus on a different attribute of the accounting system in isolation. It would be useful to know the relative importance of the extent to which firms provide conservative reports, the quality (and quantity) of voluntary disclosure, the debt contracting value of the accounting information, and the quality of the firms accruals. While it is reasonable to expect elements of the accounting system to play an important role in these decisions, establishing which elements of the accounting system are most important would be particularly useful. We also suggest that future research expand on the analyses in Bharath et al. (2008) and Dhaliwahl et al. (2005) by further exploring why accounting quality seems to affect the choice of which debt market the firm enters. Do firms with higher quality accounting, as measured in

146

these studies, have less earnings management, more informative disclosures, or reveal more proprietary information in their financial reports? At a minimum, researchers should investigate whether the accounting quality metrics commonly used in debt research (e.g., absolute value of accruals and the Dechow and Dichev, 2002 measure of earnings quality) are reasonable predictors of credit quality, or changes in credit quality. Future research might also consider a more direct test of the relationship among the various measures of accounting quality and the extent to which firms have agency problems. For example, perhaps firms that have high accounting quality are more likely to have boards and other monitors that can uncover or prevent agency problems. Alternatively, poor accounting quality may exacerbate agency problems in the firm. Or maybe instead, when boards recognize that managers have substantial discretion and potential for agency conflicts, they demand that the managers commit to high quality accounting. Ultimately, more research on how the various measure of accounting quality employed in the empirical debt contracting research affect agency costs, and the causality of this relationship will help deepen our understanding of the role of accounting information in lending decisions. It has also been suggested that accounting quality is important in the choice of lending market because accounting quality makes a firms operations more transparent and thus facilitates better monitoring. Firms that have better accounting quality can borrow in markets (or from capital providers) where monitoring is weaker (see Bharath et al. 2008; Ball et al. 2008; Sufi, 2007). It is not clear whether the measures of accounting quality used in these studies are consistent with more transparent operations, or exactly how they facilitate better monitoring. For example, the extant research assumes that firms that have larger accruals as a

147

percentage of assets have more opaque operations, larger moral hazard problems, and thus a greater need for monitoring. Additional theoretical and empirical evidence on the relationship between this measure of accounting quality and the opacity of firm operations, the extent of moral hazard problems, and the need for monitoring would help us better understand the role of accounting information in the contracting process. For example, how do moral hazard costs manifest themselves in accruals? consumption? Is there evidence that accruals capture perquisite

Perhaps survey or experimental evidence on how banks use accounting

information in lending decisions, in the selection of monitoring mechanisms, or decisions regarding the extent of monitoring would allow us to obtain a better understanding of how attributes of the accounting system affect lending decisions. For example, one could envision an experiment where the proportion of income that is accruals versus cash is manipulated, and lenders are asked to make monitoring choices. One could observe the monitoring choices made by lenders and provide evidence on which monitoring mechanisms are put in place when a larger proportion of the firms income consists of accruals. Alternatively, one could argue that accounting quality allows for the design of more efficient contractual mechanisms, which reduces agency costs (and potentially renegotiation costs). Additional empirical evidence on how accounting quality affects the choice of monitoring mechanisms used by lenders and how these mechanisms influence agency costs would be useful (as we discuss above, Bharath et al., 2008 is one step in this direction). For example, one could argue that firms with better accounting quality are able to obtain debt with a longer maturity, less restrictive covenants, and lower interest rates. Which attributes of accounting quality are best suited for these three alternative choices? Does increased transparency result in longer maturity debt or looser covenants? Does increased conservatism

148

lead to lower rates or longer maturity? Perhaps more importantly, additional theoretical insight as to why different accounting attributes might be better suited to influence different contractual mechanisms would also be helpful. Ultimately, providing evidence on why accounting attributes are important in the lending decision (or in the choice of lending market) would help rule out the possibility that accounting quality is correlated with some other underlying firm attribute that is important in lending decisions. 5.2. How do elements of the firms accounting system affect the design of the debt contract? The literature on how attributes of the accounting system affect the design of debt contracts is both relatively new and growing. In Table 1 we provide a comprehensive list and definition of the common features of debt contracts that may be affected by different aspects of firms accounting systems. We also provide a brief description of the motivation for why the accounting system might affect each element of the debt contract. We then discuss the literature supporting (or rebutting) these conjectures, and conclude by offering our thoughts on additional research that would advance this literature. 5.2.1. Association between financial reporting and interest rates The central premise underlying Jensen and Mecklings (1976) framework is that creditors are expected to demand higher nominal returns (i.e., higher interest rates) as compensation for the agency costs associated with the managers incentives to engage in actions that benefit the shareholder at their expense. We use the phrase nominal returns here because a creditors expected return is the nominal return less the expected costs associated with anticipated agency problems (e.g., expected wealth transfers, monitoring costs, and the cost of writing covenants that are borne by the creditor). This cost wedge between the interest

149

rate charged by creditors and the creditors expected return represents an agency cost borne by shareholders.68 It has been argued that elements of the firms accounting reports can exacerbate or mitigate these agency costs, and in turn that these elements are likely to be associated with the interest rate charged on the loan. The early literature indirectly tested this hypothesis, where the cost of debt was measured by either interest expense or credit ratings. For example, Ahmed et al. (2002) investigate the association between the extent to which firms report conservatively and the credit rating the firm reports on its public debt. They measure conservatism using adjusted measures of book-to-market and accruals and find that firms with better credit ratings report more conservatively. Francis, LaFond, Ohlsson, and Schipper (2005) investigate whether measures of accounting quality are associated with the interest expense reported on a firms income statement. They find that the ratio of interest expense to interest bearing debt is lower for firms that have better accounting quality. However, one potential problem with this test is that interest rates are likely determined at the time the contract is initiated. Correlating current accounting quality and interest rates at the time the contract is initiated assumes that either accounting quality or credit quality (or both) is relatively persistent or sticky. Ashbaugh et al. (2006) extend these tests using a similar measure of accounting quality but using credit ratings as a measure of the cost-of-capital. More specifically, they investigate the relation between credit ratings at debt issuance and accounting quality prior to debt issuance and find that measures of financial transparency and accounting quality are associated with firms bond ratings (controlling for other features of firms operating environment and corporate governance).
68

As noted by Jensen and Meckling (1976), the value loss to shareholders is somewhat less than the full amount of these agency costs, however, because the shareholders reap any expected wealth transfers that stem from the stockholder-bondholder agency conflicts.

150

As machine-readable data on actual interest spreads (over LIBOR) charged in loan contracts has become available, there have been more direct tests of the relation between financial reporting and cost of debt. Bharath et al. (2008) investigate the relation between interest spreads charged on the loan contract and measures of accounting quality prior to entering into the loan agreement. They divide their sample according to whether the firm has public or private debt and find that accounting quality is associated with lower interest spreads in both markets (after controlling for firm characteristics, bond characteristics and default risk). They also find that accounting quality has a relatively larger affect on interest rates in public markets compared to private debt markets, a result they argue is consistent with either lenders in private markets having access to additional information beyond public reports, or alternatively private lenders are able to more cost-effectively monitor the borrower and place restrictions on the borrowers actions through covenants (which can act as substitutes for high quality accounting information). Zhang (2008) investigates the relation between the extent to which firms report conservatively and the interest rate spread charged on the loan and finds that firms that report more conservatively are able to negotiate lower interest rates. Beatty, Ramesh, and Weber (2002) investigate the relation between interest rates and the flexibility to make accounting changes and finds that firms that are afforded the flexibility to make accounting changes are charged a premium in their cost of debt. These findings are consistent with lenders requiring firms to bear the costs of agency conflicts, as well as lenders expecting lower agency problems when firms commit to timely disclosure of information and when they bond themselves against making opportunistic use of the flexibility in financial accounting rules. In addition to the papers that examine the direct effect of accounting quality on the cost

151

of debt, there is also a stream of research examining factors that are expected to indirectly influence (or at least be associated with) the quality of firms financial statements and, in turn, affect the cost of debt. For example, Pittman and Fortin (2004), and Mansi, Maxwell and Miller (2004) provide evidence that when firms engage higher quality (larger) auditors they have a lower cost of debt. Anderson, Mansi, and Reeb (2004) find that firms that have more independent boards, more independent audit committees, and smaller boards have a lower cost of debt. Ashbaugh et al. (2006) take a broader approach and find that numerous governance factors that are likely to affect the quality of the firms accounting system are associated with the cost of debt. Asquith, Beatty and Weber (2006) highlight that in addition to considering whether to adjust the interest rate charged on the loan to reflect aspects of the firms accounting system, the lender also must decide whether to include a performance pricing provision in the contract. Performance pricing contracts provide for a pricing grid, which maps measures of firm performance in period t to the interest rate charged on the loan during period t+1. For example, a typical performance-pricing contract would have several different interest rates that would correspond to different values of the firms debt-to-EBITDA ratio. As the debt-to-EBITDA ratio varies, interest rates charged by the lender reset according to the terms of the pricing grid. By ex ante committing to a repricing schedule, performance pricing may affect the cost of writing the contract, the design of the other features included in the contract (e.g., covenants and maturity), and the probability and costs of renegotiation. Thus through performance pricing features, accounting information is likely to have an important affect on contracting costs. Asquith, Beatty and Weber (2006) find that debt contracts are more likely to include performance-pricing provisions when renegotiation, adverse selection, and moral hazard costs

152

are high. Manso et al. (2009) develop a screening model, where the lender provides the option to include a performance pricing provision in the contract, and the firms choice to include (or not include) performance-pricing features allows the lender to sort firms by type and identify firms with more promising growth options. Ball et al. (2008) investigate the choice of performance metrics that are included in performance pricing contracts. They find that performance-pricing provisions are more likely to be based on accounting information when the debt contracting value of the accounting information is high. Summarizing this research, the quality of a firms financial reporting is related to the cost of debt. We believe these results are interesting because they highlight that accounting information reduces contracting costs, and plays an important role in determining the cost a firm pays for capital. Similar to the discussion above, what is less clear however, is precisely why debt holders value high quality accounting information. It is unclear whether elements of the accounting system are priced because high quality accounting reduces agency costs, reduces the costs of writing the contract, reduces renegotiation costs, or reduces information asymmetry. Evidence on why accounting quality is associated with interest rates would provide insight into lenders information demands and borrowers incentives to provide this information. There are a host of other interesting questions in this area that are unexplored. For example, since performance pricing directly links outputs of the accounting system to the cost of capital, it likely provides incentives for firms to manage earnings. The decision to include this feature in the contract, and the design of various elements of the grid are therefore likely depend on elements of the accounting system. In addition, performance pricing appears to be a mechanism to make an incomplete contract more complete, reducing the costs of writing the contract. Performance pricing also changes the payoffs of a debt contract from fixed to variable,

153

allowing the lender access to returns above the initial rate negotiated in the loan. Thus, in performance pricing contracts, outputs from the accounting system change the payoff structure of the contract. These features of performance pricing suggest that such provisions have the potential to significantly affect a firms accounting choices. Researchers might want to develop a measure of the incentives (to increase net income) provided by performance pricing contracts, such as how a dollar of additional income affects the cost of debt (similar to the portfolio delta measure used as a measure of equity incentives in the compensation literature). If we had such a measure, we might have a better sense of how (or whether) performance-pricing provisions affect accounting choices. Such a measure would also allow us to consider how debt contracts affect firms investments or risk decisions. As discussed above, Ball et al. (2008) examine whether performance pricing is based on an accounting metric (e.g., net worth or EBITDA) or debt ratings. There are a number of unanswered questions related to how the accounting system affects the design of the performance-pricing grid. For example, how do elements of the accounting system affect the convexity or concavity of the grid? Similarly, how does the accounting system affect whether the repricing metric is based on an income metric or a measure of net worth? Finally, what elements of the accounting system determine the size of the spread covered by the grid? Do elements of the accounting system affect whether the starting point is the top of the grid (with interest rates only adjusted downward without renegotiation), in the middle, or at the bottom of the grid (where interest rates can only be increased without renegotiation)? 5.2.2. Association between financial reporting and the design of debt covenants The financial covenants in debt contracts are designed to reduce a variety of the agency

154

costs discussed in Smith and Warner (1979). Because many of these agency conflicts are the result of information asymmetry between borrowers and lenders, or at least can be more effectively mitigating or monitored within a transparent information environment, it is natural that elements of the firms financial reporting system are likely to be associated with attributes of the covenants.69 At the most basic level, one would expect covenants to be designed around attributes of the firms financial reports (both current attributes and anticipated future attributes) at the time the firm enters into the contract. Beatty, Weber and Yu (2008) and Nicholaev (2008) investigate the relation between the extent to which firms provide conservative financial reports and attributes of the firms financial covenants. Beatty et al. (2008) examine the use of income escalators and tangible net worth covenants in debt contracts, arguing that these modifications potentially make contract calculations more conservative. They find that firms that provide more conservative reports are more likely to have conservative modifications (e.g., income escalators) in their debt covenants. They suggest that covenant modifications are unlikely to fulfill the lenders demands for conservative financial statements. Nikolaev (2008) examines the relationship between conservative financial reporting and the extent of covenant use in public debt. He finds that firms that provide more conservative reports are more likely to have (and have more) covenants in their debt agreements. He also suggests that including covenants and the degree of timely loss recognition act as complementary mechanisms for reducing agency costs. Guay (2008) suggests that the results in Beatty et al. may stem from covenant modifications and conservative financial reporting serving different (and complementary) roles

69

Although they do not directly examine the effect of attributes of the accounting system on the design of covenants, Black, Carnes, Moseback, and Moyer (2004) investigate whether the use of covenants is associated with the extent to which regulators monitor the borrower. They find that when regulatory monitoring increases, lending agreements from banks are less likely to include covenants.

155

in resolving agency conflicts between borrowers and lenders. Specifically, as previously noted, conservative reporting likely serves to commit managers to timely recognition of information about losses, whereas covenant modifications provide no new information, but may serve to mitigate concerns about upward biases in reported financial numbers. Frankel, Seethamraju, and Zach (2008) examine how the firms existing accounting methods and mandatory changes in those methods shape covenants in the context of goodwill accounting. They find that changes in the accounting rules for goodwill (i.e., SFAS 141 and 142) have led to modifications of covenants in debt contracts. More specifically, net worth covenants are more likely to exclude goodwill from covenant calculations after the promulgation of these standards. They also find that firms that have goodwill on their balance sheets are less likely to have covenant modifications that exclude goodwill from covenant calculations, which is consistent with creditors recognizing that intangible assets are an important source of information about the firms financial health and its ability to pay future principal and interest on the debt. Beatty, Ramesh, and Weber (2002) examine the determinants of the decision to exclude mandatory and voluntary accounting changes from the calculation of covenant compliance. Consistent with Jensen and Meckling (1976), they conjecture that lenders will price protect themselves when they provide firms with discretion to make accounting changes to influence contract calculations. Their results suggest that lenders do charge borrowers more (i.e., a larger spread over LIBOR) when they provide borrowers with the discretion to make voluntary accounting changes. They also find that contracts are more likely to exclude voluntary accounting changes when accounting discretion is likely to result in larger moral hazard costs, and that mandatory accounting changes are more likely to be excluded from loan agreements

156

when the lender faces higher costs of investigating covenant violations. The results from this paper suggest that lenders understand that accounting discretion may increase contracting costs, and they consider these costs when they design (enter into) debt contracts. There are a number of unanswered research questions regarding the role and design of financial covenants. In particular, we encourage more research on how attributes of financial reports affect features of the covenant package. For example, how does financial reporting quality affect the tightness of covenants, or the number of financial covenants included in the contract? In addition, what factors affect the choice of financial ratios over which covenants are written? Similarly, there is also the opportunity to investigate the financial reporting attributes that debt holders value by examining the modifications to GAAP that are made in the calculation of compliance with covenants (i.e., setting of covenant thresholds). While Beatty et al. (2008) and Frankel et al. (2008) have taken some preliminary steps investigating why firms exclude intangible assets from net worth calculations (i.e., include goodwill in covenant calculations), there are a number of other frequent modifications that are made to covenant calculations that have not been investigated. In addition, there is a need for additional evidence on why some covenants include goodwill in covenant calculations while other covenants exclude goodwill from calculations. Holthaussen and Watts (2001) suggest that debt holders should exclude goodwill from covenant calculations, as goodwill is unlikely to have any value if there is a default. Yet Frankel et al. (2008) find that firms with goodwill already on their balance sheet are less likely to have it excluded from covenant calculations. Beatty et al. (2008) fail to find a relation (either positive or negative) between the exclusion of goodwill from covenant calculations and the extent to

157

which firms report conservatively. They conjecture that the exclusion of goodwill from covenant calculations may not actually make covenant calculations more conservative.70 Additional research on this topic would potentially help to resolve some of the conflicting results in the literature. Ultimately, we believe that there is a need for additional evidence on the types of modifications to GAAP that are made to calculate compliance with covenants, and uncovering the factors that cause firms to make these modifications can help us understand the attributes of the accounting system that debt holders value. 5.2.3. Association between elements of the accounting system and other contract features In addition to examining the relation between financial reporting and the source of debt financing, the interest rates charged on the loan, and the design of the financial covenants, there are also a few recent studies that examine other features of debt contracts. For example, Bharath et al. (2008) investigate the relation between accounting quality and the maturity and security requirements included in private and public debt contracts. They find that in private debt contracts, firms with higher quality accounting are able to negotiate a longer maturity and less restrictive collateral requirements when they enter into the contract. 71 They do not find similar results in public debt markets. As an explanation for these different results, they suggest that in private debt markets, monitoring and renegotiation are relatively less costly, allowing lenders to more cost-effectively tailor the debt contract to the individual borrower by adjusting price and

70

If goodwill is included in covenant calculations, and the firm has goodwill on its balance sheet, then the amortization or write off of goodwill will cause covenants to become tighter. If goodwill is excluded from covenant calculations, then the write off of goodwill has no effect on covenant calculations. Thus the exclusion of goodwill from covenant calculations does not unambiguously make covenants more conservative. 71 Bharath et al (2008) suggest that since there is no consensus on how to measure accounting quality it is appropriate to use a factor analysis of common measures of accounting quality in order to develop a relatively comprehensive measure. The three measures they use to develop one factor are (i) a measure based on Dechow and Dichev (2002), (ii) a measure based on Dechow, Sloan and Sweeney (1995), and (iii) a measure based on Teoh, Welch and Wong (1998).

158

non-price terms to reflect differences in accounting quality. Firms with better accounting quality may be easier to monitor, leading to debt contracts with a longer maturity and less security. However, the authors conjecture that in public debt markets it is more difficult to monitor and renegotiate, so poor accounting quality impacts interest rates (and not maturity or security). The premise that there are differences in renegotiation costs across private and public debt markets is plausible, and likely affects the role of accounting information across the two markets. It is less clear that higher renegotiation costs in public debt markets makes it more efficient to only adjust interest rates and not other features of the contract. Perhaps public debt is more likely to use boiler plate language in the contract, which reduces the scope for accounting quality to affect other contractual features. Alternatively, when monitoring and renegotiation costs are large, it could be that the debt contract will have fewer features that might require monitoring and renegotiation, and since many of these features rely on accounting performance measures, high quality accounting is not very useful in public debt agreements. Beatty, Liao, and Weber (2009) investigate the determinants of the decision to include cross-acceleration provisions in public debt agreements, where a cross-acceleration provision is a contractual feature that allows the debt holder the right to accelerate debt payments if a another class of debt holder (typically a more senior debt holder) makes a similar election. They find that cross-acceleration provisions are more likely to be included when firms have more accounting discretion. They suggest that firms delegate monitoring to specialists when the accounting system provides the debt holders with more flexibility. Marquardt and Wiedman (2005) examine the effect of the accounting system on the decision to include the option to make debt convertible into equity (conversion feature). They

159

find that the decision to include an equity conversion feature in a debt contract depends on the impact of the equity conversion feature on fully diluted EPS. The literature has generally ignored the influence of accounting quality on other attributes of the debt contract. For example, many debt contracts contain cash flow sweeps, where borrowers are required to pay down a portion of their debt if they realize earnings or cash flows beyond certain thresholds. We conjecture that firms may use cash flow sweeps to resolve free cash flow problems of the type discussed in Jensen (1986) (e.g., excessive perquisite consumption, over investment, etc.). Although debt, in general, has been hypothesized to help resolve free cash flow problems, cash flow sweeps might be a mechanism to further reduce such agency problems. One could hypothesize that the decision to include a cash flow sweep, and the measure of performance on which this contractual feature is based, are likely to depend on elements of the firms accounting system. Similarly, the decision to use a cash flow sweep versus financial covenants or a direct restriction on capital investments is likely to depend on elements of the firms accounting system. Finally, many contracts contain features like borrowing base restrictions, which limit the amount a firm can borrow to be based on the firms accounts receivable and inventory balances. It is also likely that the inclusion and design of these features depend on the quality of the firms accounting system, and the accounting choices the firm has made (e.g., FIFO and LIFO).72 5.3. The effect of debt contracts on accounting choice The literature on how debt contracts affect accounting choices started in the late 1970s and early 1980s and has been evolving over the last 30 years (see for example Deakin, 1979). The central premise underlying this line of research is that, conditional on having debt

72

The effect of characteristics of the firms accounting system on other elements of the contract, like capital expenditure restrictions and asset sale restrictions, are less clear and have not been explored in the literature.

160

financing, the features of the firms debt contract influence the managers accounting choices. A relatively recent survey of by Fields et al. (2001) suggests that there are two streams of research in this literature.73 The first stream of research investigates whether managers change accounting methods to avoid covenant violations. The second stream of research investigates whether stock market participants reaction to mandated accounting changes is associated with the probability of a covenant violation. Fields et al. (2001) suggest the second stream of literature (i.e., the stock price reaction to mandated accounting changes) has mostly died out in the late 1980s and early 1990s.74 We extend Fields et al. (2001) by first focusing on recent research on one of the central premises in this literature; namely that covenant violations are costly. We then delve into the literature suggesting that debt contracts provide incentives for firms to provide conservative accounting reports. We conclude this section by focusing on the literature that examines the effects of debt contracts on other accounting choices. 5.3.1. Are covenant violations costly? One of the central hypotheses in the literature on debt contracts and accounting choice is that managers change accounting policies to circumvent financial covenants found in debt contracts. As Sweeney (1994) notes, there are two central assumptions underlying this argument: (i) violating the covenant and entering technical default is costly, and (ii) managers have sufficient accounting discretion to avoid violating the covenant. There has been substantial
73

There have been several surveys of this literature, including Lev and Ohlson, (1982), Dopuch (1989), Bernard (1989), Watts and Zimmerman (1990), and Fields Lys and Vincent (2001). In summarizing the previous literature, we focus on the Fields et al (2001) discussion, because it is both the most recent literature review, and because it subsumes these prior reviews in their framework. 74 Consistent with Fields et al (2001), our review of the recent literature suggests that there are relatively few studies examining the association between the stock markets reaction to mandated accounting changes and proxies for tightness of debt covenants. The one exception we noted was Espahbodi, Espahbodi, Rezaee, and Tehranian (2002).

161

debate as to whether technical default is costly. Smith (1993) provides some guidelines for the assessment of whether covenant violations are costly. He suggests that the lenders reaction to a default fall along a continuum (from most to least costly): (i) grant a permanent waiver without renegotiation, (ii) grant a temporary waiver without renegotiation, (iii) no waiver, no renegotiation, (iv) renegotiation and provide the borrower with a waiver, (v) renegotiation fails, no waiver is granted, and the firm seeks alternative financing or enters bankruptcy. There are no renegotiation costs associated with the first three categories. The fourth and fifth category result in default costs, and they are potentially significant. Thus, based on Smith (1993), there appears to be a continuum of costs associated with a covenant violation. The early research suggests that, on average, violating covenants is costly. For example, Beneish and Press (1993) indicate that over 50% of the firms that violate covenants incur costs, and the average cost of technical default ranges between 1.2% and 2.0% of the firms market value of equity. Similarly, Sweeney (1994) finds that approximately two-thirds of the firms in her sample that disclosed a technical default experienced some form of default cost (e.g., increased collateral, restricted borrowing, or increases in interest rates). The other third incurred no direct default costs.75 Thus the early research suggests that in a majority of the cases, violating covenants is costly. Early research relied on technical violations disclosed in the firms financial statements to identify firms that violated covenants. Dichev and Skinner (2002) use a slightly different approach, using a machine readable database of financial covenants to identify covenant thresholds, and then matching this data to actual performance in COMPUSTAT to identify
75

By direct default costs we mean that the terms of the contract were not adversely affected by the covenant violation. This does not mean that there were no reputational or other costs (e.g., an adverse stock market reaction) to the disclosure of these covenant violations.

162

firms that violate covenants. They find that roughly 30% of all loans have some form of covenant violation. They suggest that private lenders set debt covenants tightly and use them as trip wires for borrowers, that technical violations occur relatively often, and that violations are not necessarily associated with financial distress. One potential explanation for the differing interpretations of these two studies is that although many contracts explicitly provide for a an increase in interest rates when there is a covenant violation, technical defaults also typically give rise to indirect costs to the firm, as a result of the implicit contract between the firm and the lender and measuring these indirect costs is difficult. There is also a burgeoning finance literature on this topic (see Roberts and Sufi (2009a) for a discussion of this literature). For example, Chava and Roberts (2008) examine the effect of covenant violations on the firms investment behavior. They provide evidence that the lenders threat of accelerating the debt payments affects the firms investment decisions, resulting in a net reduction in investment, Sufi, and Smith (2009) examine changes in a firms corporate governance following covenant violations. They find that covenant violations are often followed by an increase in CEO turnover, the hiring of turnaround specialists, and a reduction in shareholder payouts. Sufi (2009) examines whether covenant violations affect firms ability to access credit markets and finds that covenant violations lead to a reduction in the borrowers access to lines of credit. Similarly, Roberts and Sufi (2009b) find that covenant violations lead to an increase in interest rates and restrictions on the firms ability to access credit markets. A second recurring problem with research on whether covenant violation is costly is that firms are not required to disclose covenant violations when they fall into Smiths (1993) first three categories. Thus the extent to which we observe these less costly covenant violations is

163

limited to those firms that voluntarily disclose these violations. Graham, Harvey, and Rajgopal (2005) survey managers to ascertain the role of debt contracts in managers accounting choices. Their survey evidence suggests that managers do take actions to avoid violating debt covenants, but the extent to which debt covenants motivate managerial actions is likely to vary cross-sectionally. Compared to public firms that are financially healthy, they find that firms that are privately held or are close to violating covenants are more likely to make accounting choices to avoid violating covenants. The importance of covenant violations (and the accompanying costs of violating covenants) remains a controversial topic for several reasons. First, as we allude to above, the disclosure requirements for covenant violations are such that firms can and do violate covenants, but they do not have to disclose those violations in their financial statements. Firms can also renegotiate contracts if they anticipate violating a covenant, and they do not have to disclose the renegotiation. Thus, the covenant violations that are disclosed in the financial statements are likely to be the most costly violations, as the cost of curing these violations may be so prohibitive that the firm is unable to cure the violations before the financial statements are released. Thus it is difficult to identify firms that violate (or would have violated) covenants and the ones that are easily identified are likely to have systematically different costs associated with the violation than the ones that cannot be easily identified. Second, covenant calculations are often based on modified GAAP. These modifications are typically not included in any of the electronic databases, making it difficult to infer when a covenant violation occurred. Thus, without a disclosure of a covenant violation, it is difficult to identify firms that violate or would have violated covenants. Given the problems related to the lack of disclosure about firm and lender actions as the

164

firm approaches and potentially violates a covenant, and the problems in actually measuring the threshold at which violation would occur, it is unlikely that we will soon obtain a clean answer to the question of whether covenant violations are costly. Perhaps more importantly, given the variety of other contractual features that are affected by outputs of the accounting system, it is not clear how important it is to have a definitive answer to this question. If, for example, we assume that most covenant violations are not costly, does this imply that the firm does not have an incentive to take accounting actions to reduce the costs that arise from the firms debt contract? If there are performance pricing provisions, borrowing base restrictions, security requirements, cash flow sweeps, or other contractual features that are based on the outputs of the firms accounting system, then the firm will have incentives to make accounting choices to reduce the contracting costs that arise from these features of the contract. Thus, even if violating a financial covenant were costless, the firms debt contract is still likely to influence the managers accounting choices. In fact, other contracting features may have an even larger impact on the managers accounting choices. In other words, although some covenant violations are likely to be costly, the cost of violating a covenant relative to a change in interest rates in the performance pricing grid, a reduction in amount available to borrow, or a restriction on investment are unknown, and consequently the relative importance of these contractual features is unknown. Instead of focusing on the costs of violating a covenant, researchers might want to consider how all of the attributes of the contract jointly increase the significance of the accounting system. That is, a loss of $1 of income can reduce the amount of funds available to the borrower, increase their interest rate, and restrict their ability to invest. Attempting to measure these costs may be easier and more useful than trying to measure the cost of violating a covenant.

165

5.3.2. The effect of debt contracts on earnings management Historically, researchers often used the leverage ratio as a proxy for the costs associated with violating debt covenants, and investigated the association between this measure and firms accounting choices (see Watts and Zimmerman, 1986 for a summary of the earlier papers and Fields et al., 2001 for a summary of the more recent papers that use this measure). However, Fields et al. (2001) criticize the use of the leverage ratio, arguing that leverage is determined endogenously, is not necessarily correlated with the tightness of covenants, and is likely to be a poor proxy for default risk. They suggest that researchers should use actual covenants to measure the probability of default, and as a result the literature using leverage as a measure of default costs has dwindled. Perhaps in response to the suggestions made by Fields et al. (2001), researchers have become more creative when examining the effects of debt contracts on accounting choices. First, we note that in addition to accounting choices, firms have a variety of alternative mechanisms that they can use to reduce contracting costs, such as renegotiating the contract, or taking real actions such as cutting R&D or dividends. The costs and benefits of these alternatives are likely to vary cross-sectionally and inter-temporally. In addition, covenants and other factors such as performance pricing, investment restrictions, and borrowing base requirements may provide incentives for managers to make accounting choices to reduce debt contracting costs. For example, Healy and Palepu (1990) examine the accounting and the dividend payment choices of a sample of firms that approach the dividend constraints included in their debt contracts. They find that as firms approach their dividend constraints, they are more likely to cut dividends as opposed to make accounting changes to avoid covenant violations. Daniel,

166

Denis, and Naveen (2008) extend this research to look at the accounting choices of the dividend paying firms in the S&P 1500. They find that firms whose earnings are below expected dividend levels tend to manage earnings upwards, but that this effect only exists for firms that have debt in their capital structure. The conflicting results of these two studies highlight some of the issues that arise when studying the effects of debt contracts on accounting choices. Healy and Palepu (1990) identify firms that are paying dividends, have debt covenants based on dividend restrictions, and are close to violating those covenants. For these firms, it would appear that cutting dividends is the least costly alternative (in terms of net benefits). In the research setting examined in Daniel et al. (2008), it would appear that some feature of debt provides incentives for firms to make income-increasing accruals when current earnings are below historical dividend levels. However, it is unclear which features of debt provide these incentives, whether accruals management is less costly than changing accounting methods, whether accruals management provides more flexibility than changing accounting methods, and whether the costs of cutting dividends differs dramatically across these two settings. Other papers have adopted an approach similar to Daniel et al. (2008), focusing on whether debt in the capital structure impacts the accounting choices that firms make. For example, Roychowdhury (2006) examines firms incentives to engage in real activities management to meet income thresholds. He finds that firms with debt in their capital structure are more likely to engage in real activities management to meet thresholds than firms that do not have debt. Altamuro, Beatty, and Weber (2005) investigate the effect of having bank debt on firms revenue recognition decisions. They find that firms with bank debt are more likely to accelerate revenue recognition than firms that do not have bank debt in their capital structure. Other papers have taken a slightly different approach, investigating the effect of specific

167

contract features on accounting choices. For example, Beatty and Weber (2003) examine the effect of the provision that allows accounting changes to affect contract calculations on the decision to make voluntary accounting changes. They find that when accounting changes affect contract calculations (i.e., increases in income associated with the accounting change relaxes the covenant), managers are more likely to make income increasing accounting changes. Beatty and Weber (2006) examine the interaction between covenant tightness and the flexibility to include accounting changes in contract calculations on the adoption decisions associated with SFAS 142. Firms with tight covenants where accounting changes affected covenant compliance calculations were less likely to take write-offs when adopting SFAS 142. 5.3.3. The effect of debt contracts on the extent to which firms report conservatively As we discussed above, several papers suggest that the extent to which firms report conservatively affects the design of debt contracts. In particular, researchers have suggested that firms that report more conservatively are charged a lower interest rate, have conservative covenant modifications, and are more likely to have smaller bid-ask spreads in the syndicated debt markets (i.e., a lower cost of capital). There are two elements of the debt contract that suggest that the results from this research are plausible. First, debt contracts have an asymmetric payoff, where the lender is subject to a limited upside but has a large downside (which is capped at the entire principal borrowed).76 This feature of the debt contract provides incentives for lenders to obtain more timely information about losses. Second, debt contracts often contain provisions limiting managerial discretion, preventing managers from changing accounting policies (e.g., recognizing certain types of accruals) in order to increase income. If at the contract initiation
76

While a performance pricing contract may allow lenders to increase the interest rates they charge on the loan, usury laws will generally limit the interest that lenders can expect to earn on the loan which will preserve the asymmetric payoff structure.

168

date lenders value timely information about losses, and contracts can be designed to prevent managers from changing their accounting principles or limit the effects of income increasing earnings management, then it is reasonable to expect lenders to reward firms with more favorable contract terms. The literature on the effects of the debt contract on the choices firms make to report conservatively is more contentious. One stream of research suggests that firms that report conservatively are more likely to violate covenants after experiencing negative news (Zhang, 2008). A second stream of research suggest that if a firm provides conservative reports after the contract has been negotiated, there will be a suboptimal number of covenant violations resulting in an increase in contracting costs (Gigler et al., 2009 and Li, 2009). They suggest that the increase in contracting costs associated with covenant violations that are not indicative of default are greater than the benefits of more timely signals of the reduction in credit quality, and thus, from the lenders perspective, conservative reporting is suboptimal. Reconciling these two streams of research, it is ultimately an empirical issue whether covenant violations that arise from conservative accounting choices are beneficial or detrimental. One could argue that a covenant violation that allows the lender to call the debt and preserve their principal provides the lender with a benefit that outweighs the cost of a Type I error (i.e., a false positive). However, if there are a sufficient number of false positives, then it is less clear whether there is a net benefit to reporting conservatively. Country-level studies also provide mixed guidance on whether lenders value conservative financial reports. Bushman and Piotroski (2006) find that economies with strong legal systems and a prevalence of debt financing are more likely to have timely loss recognition which suggests that debt holder demands drive conservative accounting choices. Similarly, Ball

169

et al., (2008) establish that the extent to which an economy has asymmetrically timely earnings is positively correlated with the size of the economys debt market. Ball and Shivakumar (2005) investigate the conservatism choices of U.K. private and public firms and find that timely loss recognition is more prominent in publically traded firms than in privately held firms. They suggest that external shareholders drive conservative accounting choices (as opposed to debt markets). 5.4. Theoretical advances and incomplete contract theory Most of our discussion of debt contracting has focused on the recent empirical advances in this literature. There have also recently been some notable theoretical advances in this literature. Since many of the theoretical advances have occurred in the finance and economics literature, we refer the interested reader to Roberts and Sufi (2009a) for a more thorough and complete discussion of this literature. Instead, we focus on the theoretical advances that we think will be particularly useful to accounting research. We begin by discussing the incomplete contracting line of research, then focus on renegotiation in complete contracting models, and conclude this section by discussing other theoretical advances in the literature that may be useful to accounting researchers. Hart (1995) and Tirole (1999) provide an introduction to a burgeoning stream of research in economics and finance loosely described as incomplete contracting. They highlight that one of the shortcomings of the standard principal-agent view of the firm is that it ignores the costs of writing contracts. That is, when two parties negotiate a contract, it can be very costly (and often impossible) to gather the information necessary to write a complete contract that specifies all contingencies for all states of the world that may arise after the contract is written. Incomplete contracting suggests that if the costs of writing a complete

170

contract are high, then the contract will include clauses or terms that allow for renegotiation in certain states. Thus, an incomplete contract will stipulate outcomes under the expected states of nature and provide for renegotiation if an unexpected state of nature arises.77 While the economics and finance literatures have taken the ideas central to incomplete contracting theory in a variety of directions, in this section we focus on its application to debt contracting. In the traditional agency theoretic view, accounting information is useful in the debt contracting process because shareholders can use accounting information to commit to not expropriate wealth from debt holders. As we will discuss below, accounting information is useful in the incomplete contracting view of debt agreements because it can both reduce agency costs, and the costs associated with writing a more complete contract. That is, when it is difficult for parties to specify future states of the world, the parties cannot write a complete contract. They will instead write an incomplete contract that allows for renegotiation in certain states of the world, which are frequently described in debt contracts by referencing accounting information. Thus the incomplete contracting literature may be of particular importance to accountants because it provides a framework to address issues related to renegotiation. A recent paper by Garleanu and Zwiebel (2008) provides a rich set of predictions about the implications of information asymmetry between entrepreneurs and lenders on the design and renegotiation of debt contracts. The authors begin by recognizing that a lender lacks complete information about the future actions that the owner might take to expropriate wealth from the lender (i.e., lack of transparency about the types and severity of agency problems). As a result, the authors note it is typically not feasible to delineate contractually all future positive NPV projects ex ante, or for a court to ex post enforce such a vague contractual provision.

77

When referring to unexpected states of nature, we mean states of nature that are statistically very unlikely, as well as states of nature that are unknowable at the time the contract is entered into.

171

Consequently, covenants are instead conditioned on more easily observable and verifiable accounting variables, such as financial ratios, that are likely to be imperfectly correlated with the availability of good future projects, and are then renegotiated when more information becomes available. Thus, accounting-based covenants arise endogenously in the debt contract as a mechanism for owners to commit to restrict certain actions, and as a trigger for renegotiating the contract as new information about the firms investment opportunity set becomes available. The incomplete contracting approach adopted in Garleanu and Zwiebel (2008) has the potential for new and interesting avenues of accounting research. For example, their theory predicts that the greater the information asymmetry between the owner and the lender about the form and magnitude of potential wealth transfers, the more restrictive (i.e., tighter) the covenants are expected to be. Further, they predict that such information asymmetry is increasing in the complexity and lack of transparency within the firm. These results are intuitive in that owners will grant lenders stronger decision rights (that allow for prohibitions on certain investments) when lenders are less informed about the form and magnitude of future wealth transfers. The authors also argue that covenant tightness should be positively associated with the magnitude and predictability of cash flows and cash holdings since fungible assets enable managers to make payments for wealth transfers. Finally, they suggest that, in most cases, when financial covenants are violated, borrowers and lenders should agree to looser covenants in the future (as opposed to tighter covenants). In addition to Garleanu and Zweibel (2008), there have been other theoretical papers in the accounting literature that have adopted an incomplete contracting approach to model debt contracting problems. A relatively early paper in the stream of literature by Sridhar and Magee

172

(1996) assumes that a firm is able to generate two signals regarding future outcomes that are observable to both the debt holder and the equity holder, but the debt holder cannot contract on the second signal (i.e., it is unverifiable). Thus, from an incomplete contracting perspective, it is too costly to contract on the second signal. Given these assumptions, Sridhar and Magee develop a model that highlights attributes of the variable that will be included in the debt contract. In particular, their model suggests that, from the borrowers perspective, uncertainty in reporting latitude for the contractible variable is not desirable. They suggest that borrowers will tailor debt agreements to reduce the discretion in the contractible performance measure, and this tailoring should result in either a lower interest rate or looser covenants because such changes facilitate more effective monitoring by debt holders. Their analysis also predicts that if the parties expect future standards to diminish (increase) the uncertainty in the firm's reporting latitude, the debt contracts will have lower (higher) interest rates and/or looser (tighter) covenants. Li (2009) also uses an incomplete contracting framework to investigate the role of conservative accounting choices in the debt contracting setting. She suggests that the extent to which lenders will demand conservative accounting depends on the expected renegotiation costs, the firms investment opportunities and expected liquidation values.78 Within the incomplete contracting literature, there are numerous other papers investigating issues besides the design of covenants and renegotiation that are potentially interesting to accountants. For example, Diamond (1991) investigates the decision to issue short-term debt versus long-term debt. He models the maturity decision to be a function of a trade-off between liquidity risk, which relates to the lenders unwillingness to refinance the loan
78

Gigler et al. (2008) reaches a slightly stronger conclusion, that conservatism reduces the efficiency of debt contracts. However, they do not use an incomplete contracting approach in their research design, which potentially explains the difference in results.

173

when bad news arrives, and the borrowers private information regarding future credit ratings. His model yields the predictions that both high quality (i.e., good credit ratings) and low quality firms (i.e., poor credit ratings) borrow short term, which is consistent with the stylized fact that higher quality borrowers tend to issue commercial paper, and lower quality borrowers tend to borrow through short term bank loans. From an accounting perspective, the disclosure of bad news to the lender is an important consideration in the debt maturity decision, as is the borrowers private information regarding their credit quality. One could envision empirical tests of whether the timeliness of bad news is associated with the maturity structure of corporate debt. Similarly, the extent to which the lender has access to information to evaluate the firms credit quality during the life of the loan may also be related to the maturity of the debt. There are also a number of papers that investigate the liquidation value of assets and how the liquidation value of the firms assets affects the availability of credit, the proportion of debt and equity in the firms capital structure, and the maturity of the firms debt (e.g., Aghion and Bolton, 1992; Shleifer and Vishny, 1992; Bolton and Scharfstein, 1996; and Hart and Moore, 1994 and 1998). These papers are potentially of interest to accountants for a variety of reasons. For example, Hart and Moore (1994), suggest that asset specificity and durability will affect the maturity of the firms debt. They suggest that firms borrow short term to finance working capital like inventory and accounts receivable and long term to finance property plant and equipment. The firms accounting system and more specifically the accruals recorded in this accounting system are likely to be a central source of information regarding the firms working capital. Extending these theoretical models to consider how the quality of information regarding accruals affects the maturity of the debt will help guide future empirical research on the elements of the debt contract that are likely to be affected by the firms accounting system.

174

Similarly, one can envision extensions of these models that lead to additional insights into the properties of accruals that are likely to be valued by creditors, or how other elements of the accounting system (beyond accruals) are important information in lending decisions. For example, a central hypothesis from this literature is that the liquidation values of assets are of central importance in a variety of lending decisions. This suggests that lenders have demands for information regarding liquidation values. Liquidation values are unlikely to equate to market values or to fair value for a firm that is a going concern. Theoretical and empirical research extending these models and deriving insights on the properties of accounting information that are useful to lenders may be helpful to standard setters as they determine the role of market values, fair values, and historical costs in accounting standards.79 Despite these and other theoretical advances, there has been relatively little empirical research inspired by the incomplete contracting literature (especially in the accounting literature). There have been some advances in the finance literature. For example, Benmelech and Bergmen (2008) investigate the role of liquidation values and cash flows in airline renegotiations of their lease contracts. Consistent with the hypotheses developed in the incomplete contracting literature, both the firms financial position and the liquidation value of its asset (i.e., airplanes) influence the outcome of the renegotiation. A related paper by Benmelech, Garmaise, and Moskowitz (2005) focuses on a different aspect of liquidation value (the redeployability of property) focusing on how the liquidation value of commercial property affects the size, maturity and interest rate charged on the loan. They find that all of these factors are affected by the extent to which the property is redeployable which is consistent with the hypothesis that liquidation values affect the design of the debt contract.
79

As noted by Roberts and Sufi (2009a), the concept of liquidation value has been extended by both Williamson (1988) and Schleifer and Vishny (1992) to encompass the idea of the redeployability of the asset, and the financial condition of potential buyers.

175

Outside of the incomplete contracting models, there are other theoretical models that investigate a variety of different attributes of debt such as the tradeoff between short-term debt and long-term debt and renegotiation, both of which are of potential interest to accounting researchers. For example, the costly state verification avenue of theoretical research pioneered by Townsend (1979) constructs a model in which there is information asymmetry between borrowers and lenders, and the lender has a contractually agreed upon option to incur a cost to verify some piece of asymmetric information (e.g., managerial effort and asset values). He goes on to develop a model where the audit costs incurred by the lender serve to force the borrower to report truthfully. There have been numerous extensions of this line of research over the last 30 years that accounting researchers may find useful. For example, Gale and Hellwig (1979) extend the ideas in Townsends (1979) model to suggest that the act of observation or verification in a debt contract could be interpreted as bankruptcy. When the borrower fails to make a debt payment the lender exercises their option to verify the firms assets and forces the firm into bankruptcy. Once in bankruptcy, the lender is able to verify the borrowers true assets, and receive a payment on their claim (which is generally consistent with Townsends (1979) model). Although Gale and Hellwigs (1979) model does not have any direct implications for the importance of accounting information, it seems as if a slightly different extension would be to reformulate the model where the observation or verification occurs when the firm violates a financial covenant, which is triggered by the receipt of an accounting report. Here a covenant violation leads to renegotiation, where there is a potential for random verification as the accounting report is generated from a stochastic process, and verification can lead to the borrower being worse off (forced to pay down the debt), or better off, and the act of

176

renegotiating the covenant may lead to better terms. In this extension, the accounting report serves as the trigger for the costly state verification, and thus attributes of the firms accounting system will endogenously become important in the lending process. This model could lead to hypotheses regarding what types of accounting information the debt contract bases covenants on, how tight covenants are made, and on how covenants are reset after a covenant violation. In addition to the costly state verification and incomplete contracting frameworks, there have been a variety of additional theoretical advances in the debt contracting literature. Rather than attempt to summarize these theoretical developments, we direct interested readers to surveys by Roberts and Sufi (2009a), Hart (2001), Allen and Winton (1995), Harris and Raviv (1991). 5.5. Synthesis and directions for future research Over the last decade there has been an increased interest in investigating how attributes of the accounting system affect the design of debt contracts. This change in research focus is likely attributable to the greater availability of machine-readable data on contract terms and improved techniques for extracting debt related data from SEC filings. We find this increase in research to be promising. As we discussed in the introduction, debt is a significant source of capital for publicly traded corporations that has been relatively neglected in the accounting literature, in large part because of difficulty in accessing data. Researchers have established that the importance of accounting quality in the debt markets is a function of the particular debt market from which the firm has obtained its debt financing. Public debt and private debt have different institutional features, and thus lenders in these markets are likely to have different preferences for various accounting attributes. This also implies that the incentives provided by the debt contract are also likely to depend on the

177

source of financing. For example, researchers have generally asserted that covenants provide incentives for managers to use accounting discretion. As Begley and Freeman (2004) discuss, public debt has higher renegotiation costs than private debt and thus seldom has financial covenants. Thus the importance of covenants on accounting choices likely depends on whether the debt is public or private. Researchers have also established that accounting quality likely affects numerous features of the debt contract (e.g., interest rates, covenants, maturity and collateral requirements). Ten years ago it was difficult to obtain data on the features contained in a debt contract. Researchers can now examine what information lenders contract on, how they modify GAAP when measuring performance, and how accounting information directly affects the amount firms can borrow, the interest rates they pay, and other features of the debt contract. Consistent with the suggestions made in Fields et al. (2001), there has also been a corresponding increase in the use of actual covenant data and other contractual features in the debt contract in the accounting choice literature. Thus, rather than using indirect proxies for debt contracting motivations to make accounting choices (e.g., leverage), researchers now attempt to measure the actual degree of covenant tightness and other contractual features that might affect accounting choices. By developing better measures of how accounting information is used in the contract, researchers have also been able to develop sharper (and more convincing) tests of how debt contracts affect accounting choices. While we have provided numerous suggestions on how the existing lines of research in the debt contracting area can be extended, we believe that there a few potential new lines of research that appear promising. In particular, although we know much about how attributes of the accounting system may affect the design of the debt contract and how debt contracts may

178

influence outputs of the accounting system, we know comparatively little about which attributes of the accounting system are most valuable to lenders. For example, it has been argued that capital providers prefer transparent financial statements. Lenders have access to private information, and can require borrowers to provide additional detail, making less transparent financial reports more transparent. Thus it is not clear how important transparency is in the lending markets. Additional research investigating the accounting attributes lenders value would likely help standard setters as they are often forced to consider various tradeoffs as they design accounting standards. Many of these tradeoffs are based on equity holders informational demands for valuation purposes. It would seem that if we knew more about the informational demands of debt holders, standard setters would be better able to incorporate their needs when setting standards. Second, there has been relatively little research on the role of accounting reports in the renegotiation process. When firms violate covenants or renegotiate or amend the debt contract, the renegotiation is likely to be affected by the outputs of the accounting process. The accounting system provides the lender with information about future credit quality and cash flows, and about past investment choices and perquisite consumption. Thus although the accounting system likely plays a very important role in the renegotiation process, we know relatively little about its role in this process. Finally, we believe that innovations in theoretical research on debt contracts have the potential to make a significant contribution to this literature. Jensen and Meckling (1976) provide for a framework for understanding how agency costs may manifest in debt contracts. However, Jensen and Meckling (1976) describe a static one period game. Accounting is

179

important in the debt contracting process because it not only affects the design of the contract at initiation, but it also plays a role in the allocation of decision rights and control rights throughout the life of the contract. Innovations in the incomplete contracting paradigm are potentially useful for researchers to model the debt contracting process, and the role of accounting information in this process.

6. The role of auditors Compared to the other contracting parties we have discussed in this survey, the auditors role in the firms nexus of contracts is somewhat more complex (at least with respect to financial reporting issues). Similar to other contracting parties, the auditor and (public) firm typically enter into explicit and implicit contractual arrangements, in part due to regulatory requirements that all publicly traded firms prepare audited financial statements in accordance with Generally Accepted Accounting Principles (GAAP), but more fundamentally because independent auditors serve to help firms lower agency costs by certifying the integrity of financial reports. However, the added complexity in the firm-auditor contracting setting is that the role of the auditor is to directly influence the quality of the financial reports, which in turn affects contracts related to governance as well as between the firm and other outside parties. By certifying that the financial statements have been prepared in accordance with GAAP, the auditor serves as a monitor, and potentially reduces the agency costs that arise in these other contractual settings.80
80

There exists a large body of research on auditing issues, largely due to the prominent role that auditors play in the production and dissemination of accounting information. Papers in this area examine the procedures used by auditors (e.g., sampling procedures and the use of bright line standards), outputs of the auditing process (e.g., going concern opinions and earnings management in the financial reports), how various institutional factors affect the auditor and the outputs of the auditing process (e.g., litigation risk and fees paid by the client), and regulations affecting auditors (e.g., SOX and the Private Securities Litigation Reform Act (PSLRA)). Most of this research, however, is beyond the scope of this survey.

180

In this section, we focus on the research that investigates how auditors reduce agency conflicts, as well as the research related to agency conflicts that arise between the auditor and the firm. As discussed above, Jensen and Meckling (1976) emphasize how owners and managers have incentives, ex ante, to commit to monitoring mechanisms that reduce costs arising from agency problems. In this regard, they suggest that an audit by an external party is a mechanism that can reduce agency costs. Watts and Zimmerman (1983) expand on this argument. They hypothesize that for an auditor to be effective in reducing agency costs, there must be a non-zero probability that the auditor will detect and report breaches in the contract (i.e., accounting irregularities). They suggest that one approach to defining auditor independence is the probability that the audit firm will report a contract breach (accounting irregularity) conditional on its existence. Since audit firms are typically hired and paid by (contract with) the firm, there is the potential for an agency problem, as the auditors incentives to report irregularities may be affected by the fees and relationships they receive from the firm to conduct their audits. 6.1. Research on auditors reducing agency costs The auditor can potentially play an important role in reducing agency costs associated with all of the previously discussed contracting settings (e.g., between managers and directors, between boards and external shareholders, between the firm and creditors, and between the firm and regulators). We begin with the body of research suggesting that auditors can reduce agency problems between debt holders and equity holders. In particular, the process of auditing financial statements increases the probability that accounting irregularities and fraud are detected. Recognizing that auditors may discover fraud, managers are less likely to engage in

181

these types of activities, thereby increasing the reliability of financial reports and increasing the efficiency with which creditors can contract with the firm. Consistent with this idea, Blackwell, Noland, and Winters (1998) identify a set of small private companies that recently obtained bank financing, some of which had obtained an audit prior to obtaining a loan, some of which did not. They examine the effect of obtaining an audit on interest rates and find that firms that have been audited are able to negotiate lower interest rates. Pittman and Fortin (2004) and Mansi, Maxwell, and Miller (2004) provide evidence that firms receive lower interest rates when larger (better quality) auditors audit them. Both papers suggest that larger auditors reduce agency problems and borrowers that hire larger auditors are rewarded with lower interest rates. However, using an alternative measure of audit quality (the ratio of audit fees paid to the auditor to the total fees earned by the auditor), Ashbaugh et al. (2006) fail to find a relation between audit quality and interest rates. In addition to providing mixed evidence, these papers provide at best indirect tests of whether auditors actually reduce the agency costs between debt holders and equity holders. For example, one explanation for the results in these papers is that larger auditors actually have deeper pockets, and thus act as insurers to the lender. While this research has tried to rule out such an alternative explanation, it is very difficult to separate the insurance effect of a large auditor from the effects of improved audit quality. In addition, Ashbaugh et al. (2006) highlight that there are other governance mechanisms (several of which are described in the governance section above) that may substitute for high quality auditors. While the Pittman and Fortin (2004) and Mansi et al. (2004) papers both provide some controls for other measures of corporate governance, neither paper uses measures that are as extensive as those in Ashbaugh et al. (2006).

182

The above critique is not intended to suggest that auditors do not play an important role in the debt contracting process. Many private debt agreements contain specific covenants that require the borrower to be audited by a Big 5 auditor, or by a specific (typically larger) auditor. This suggests that lenders consider auditor choice to be very important when contracting with a borrower. It is less clear whether this auditor provision is included in the contract because certain auditors are believed to be able to reduce agency problems, or because these auditors have deep pockets in the event there is an error or irregularity in the accounting reports. Another stream of research investigates whether auditors reduce agency problems between equity holders and boards (or between boards and managers). This literature has taken a variety of approaches to address this question. One line of research investigates the relationship between the cost of capital and audit firm size. For example, a theoretical model developed by Balvers, McDonald and Miller (1988) suggests that there will be less IPO underpricing when firms employ higher quality auditors. The basic argument is that investment bankers with a better reputation are more likely to select better auditors, and both investment banker reputation and auditor reputation reduce IPO under-pricing. The empirical research testing this hypothesis finds that firms with larger auditors experience less under-pricing at the time of their IPO (for examples see Beatty, 1989; Hogan, 1997; and Willenborg, 1999). Khurana and Raman (2004) extend these tests by investigating the relationship between a more general measure of cost-of-capital and auditor size. They find that audit quality is associated with the cost of capital in the U.S., a relatively more litigious environment, but not in the U.K., Australia, and Canada, where litigation is less prevalent.

183

Fan and Wong (2005) examine the auditors role in resolving agency conflicts between minority shareholders and controlling shareholders in emerging markets (China). They suggest that other corporate governance mechanisms are less effective in emerging markets, and thus the auditor may play a more prominent role in reducing agency problems. Consistent with this idea, they find that firms have smaller price discounts when they hire Big 5 auditors, and that Big 5 auditors charge a larger fee when they audit firms that are more likely to have agency problems. Wang, Wong and Xia (2008) extend this hypothesis by investigating the role of the auditor in companies in which the state or government owns a significant stake. They find that state-owned firms in China are more likely to hire small local accounting firms to conduct their audits. They argue that governments are able to exert more control over these auditors, suggesting that state ownership dilutes the effectiveness of an external audit. A third body of research investigates the stock market reaction to changes in auditor quality. These papers generally find a negative reaction to events that signal a reduction in audit quality, suggesting that equity market participants value higher quality auditors, again either for their ability to facilitate high quality financial reports or for their insurance role (for examples, see Baber Kumar and Verghese, 1995; Menon and Williams, 1994; Chaney and Philipich, 2002; and Weber, Willenborg, and Zhang, 2008).81 Blouin et al. (2007) extend this analysis, investigating how firms select new auditors when they are forced to switch auditors (they focus on audit switching decisions following the downfall of Arthur Andersen.) They find that firms with greater agency problems are less likely to follow their former Andersen audit team as they

81

A recent paper by Nelson, Price, and Roundtree (2008) questions the findings that equity market participants reacted to the changes in audit quality that were signaled as the events surrounding the demise of Arthur Andersen unfolded. Although the paper is unable to refute the hypothesis that equity market participants value audit quality, the paper does highlight problems in using market studies to identify auditor reputation.

184

select a new auditor. Their paper highlights the importance of agency costs in the auditor selection decision. A final stream of research on the role of auditors in reducing agency conflicts examines whether financially distressed firms that hire high quality auditors have financial reports with less accounting discretion. For example, Gaver and Paterson (2001) investigate the relation between auditor quality and the reserves for claim losses in distressed property and casualty insurance companies. They find that distressed property and casualty insurance companies that have high quality auditors (actuaries) have less under-reserving behavior, suggesting that high quality auditors reduce agency problems. Gaver and Paterson (2007) extend this study, investigating whether the contractual relationship between auditor and client influences outputs of the accounting process. Using the same sample of distressed property casualty companies, they examine whether distressed property and casualty companies that represent a relatively larger proportion of an audit firms fees are afforded more accounting discretion. They find that larger clients are actually afforded less accounting discretion, which suggests that reputational effects dominate auditors incentives to provide additional discretion to their larger clients. 6.2. Research on agency problems arising from the contract between the firm and its auditor The potential for agency problems between shareholders and auditors, or between external contracting parties and auditors has been discussed in the accounting literature for many years. In publically traded U.S. corporations, auditors are typically paid by the firm, with the choice of auditor being largely decided by managers and directors. As Watts and Zimmerman (1983) discuss, this creates the potential for an independence problem. Although the value of an audit stems from the expertise and independence of the auditor, there is the potential for the fees an auditor earns from a particular client to compromise the auditors

185

integrity in reporting accounting irregularities. This may occur if managers have influence over the auditor selection and do not want an accounting irregularity revealed to directors and shareholders (for fear of being fired or losing compensation). Alternatively, there may be accounting irregularities that managers/directors/shareholders do not want to be revealed to other contracting parties, such as creditors, taxing authorities or regulators. This hypothesis regarding audit selection and fees has been further refined, breaking down the fees an auditor receives into the portion related to the audit and the portion related to other non-audit services. Researchers have hypothesized that the larger the fee associated with non-audit services, the more likely the auditor is to provide the firm with accounting discretion. Early evidence on this hypothesis was provided by Davis, Risshiute, and Trompeter (1983), Simunic (1984), and Beck et al. (1988). All of these papers suggest that firms purchasing more non-audit services are more likely to have agency problems. These studies used relatively small samples of firms that responded to surveys (Simunic, 2004), from public accounting firms that provided the data to the researchers (Davis et al., 1983), or from a brief disclosure period mandated under Accounting Series Release (ASR) # 250 (Beck et al., 1988). Beginning in 2001, all publicly traded firms in the U.S. were required to disclose data on audit fees and non-audit fees in their proxy statements. A number of papers use these new data to revisit the question of whether the size of audit and non-audit fees creates an agency problem with the firms auditor. One of the first studies in the post-2000 period was Frankel, Johnson, and Nelson (2002), who find that firms paying relatively larger non-audit fees have relatively larger discretionary accruals, suggesting that independence problems increase with the size of non-audit fees. As noted by Defond and Francis (2005), however, several papers reexamine the question addressed by Frankel et al. (2002), and find that the Frankel et al. results

186

are sensitive to sample selection, research design, and model specification. Papers highlighting these issues include Chung and Kallapur (2003), Ashbaugh et al. (2003), Larker and Richardson (2004) and Reynolds et al. (2004). Thus, the preponderance of the evidence suggests that, on average, large non-audit fees do not compromise auditor independence. The discussion of these issues in Larcker and Richardson (2004) is the most salient to our survey. They examine the association between measures of abnormal audit fees (non audit fees) and discretionary accruals across different subsamples of the data. They find that when the auditor is likely to play the most significant role in the overall system of corporate governance (small firms, high insider holdings, low institutional holdings, less independent boards) the relation between discretionary accruals and abnormal fees is negative. This implies that audit fees do not compromise independence in settings where the auditor is expected to be most critical to corporate governance. The litigation risk faced by the auditor, or the potential loss of clients due to a damaged reputation appears to provide incentives for auditors to reduce managerial discretion in the settings where their opinions are likely to be the most valuable. In many studies, researchers focus on one contracting party (e.g., boards, debt holders, or auditors) and investigate the incentives that their contract provides on outputs of the accounting process. Larcker and Richardson (2004) innovate along at least two dimensions. First, they map the relationship between incentives and outputs, and investigate how other stakeholders affect this relationship. Second, they use a latent class mixture analysis, which allows the authors to analyze several measures of audit fees and accruals, and more than one regression model to investigate the relationship between auditor independence and audit fees. Both of these innovations extend our understanding of the relationship between abnormal accruals and audit fees.

187

Kinney, Palmrose, and Scholz (2004) also provide evidence on the relation between non-audit fees and restatements. They find that for most categories of non-audit fees there is no significant relation between restatements and non-audit fees. These results are consistent with most of the results we discuss above, suggesting that the provision of non-audit services does not impair auditor independence. A second stream of literature investigates whether the firms decision to hire former partners of their audit firm as officers or directors influences the outcomes of the accounting process. The fundamental hypothesis in this literature is that auditors may allow former partners more accounting discretion. Menon and Williams (2004) and Lennox (2005) both investigate this hypothesis. Menon and Williams (2004) find that signed and unsigned accruals are relatively larger for firms that employ former audit partners, while Lennox (2005) finds that firms that hire former audit partners are more likely to receive clean audit opinions. Thus, both papers suggest that the efficacy of the auditor as a mechanism to reduce agency problems is impeded when the firm hires former audit partners.

7. Other parties in the nexus of contracts To this point, we have discussed in some detail the interactions of shareholders, directors, managers, creditors and auditors in the nexus contracts, and how financial reporting plays a role in mitigating agency conflicts and reducing contracting costs between these parties. There are, however, several other important contracting parties, such as regulators, taxing authorities, suppliers, and customers that are noticeably absent from our review. The first of these parties, regulators, are omitted because the breadth and depth of the literature on regulation is extensive, and giving it the emphasis it deserves would double the

188

length of our paper.82 We suggest that readers interested in this topic may find a relatively recent survey by Leuz and Wysocki (2009) as a useful starting point that summarizes the research in this area. With respect to taxing authorities, again, giving it appropriate emphasis would require a significant number of pages (and even more importantly, the authors of this review acknowledge our lack of expertise is the area of tax). Rather, we point the interested reader to the review on tax research by Hanlon and Heitzman (2009) in the same issue of this journal. Finally, on the topics of suppliers and customers, we are not aware of a sufficient body of accounting research to develop a detailed literature review of how financial reporting influences firms contracts with these outside parties. We conjecture that this lack of research may stem from a lack of data, but suggest that this may be an interesting area for future research.

8. Conclusion In this survey we review the corporate governance and contracting literatures with a particular emphasis on the developments over roughly the last decade, as well as areas that are likely to be of future interest to accounting researchers. The paper focuses on the firm as a nexus of contracts among the various factors of production and highlights the role of the accounting system in reducing the agency costs that arise among various stakeholders within the firms nexus of contracts. We emphasize that the research conducted in this area has made significant progress, and we believe the professions understanding of the role of the accounting system in the nexus of firm contracts has been significantly expanded over the last decade.

82

For example a recent survey by Coates (2007) of the literature examining the effects of the Sarbanes Oxley (SOX) Act cites over 70 papers on the topic, most written by accounting academics. One might expect that there at least as many papers examining issues associated with the adoption of IAS as well as a large number of other important regulatory topics.

189

We have deliberately adopted a rather broad view of contracting by considering not only explicit contracts, which have historically been the focus of a large proportion of accounting studies, but also implicit contracts. Throughout this survey we have interspersed discussions of directions for future research. These suggestions, for the most part, relate to extensions of existing lines of literature. Rather than repeat those suggestions here, we instead discuss what we believe are the main themes that emerge from this survey, and potential new lines of research that we believe would be useful for enhancing our understanding of the role of accounting information in corporate governance and contracting. One of the main themes in each of the contracting settings is the importance of implicit contracts. Although implicit contracts are not governed by formal written documents and in most cases are not subject to the same degree of legal enforcement, they represent an important mechanism for facilitating a variety of economic transactions that might otherwise be noncontractible, or contractible with lower efficiency. Understanding implicit contracts, however, poses a number challenges from both a theoretical and empirical perspective. Theoretically, implicit contracts represent the equilibrium behavior of the players in a repeated game, and therefore necessarily entail a multiperiod model. This poses a number of technical challenges and also introduces additional issues that may be tangential to the question of interest. Empirically, implicit contracts require time-series (or panel) data to estimate the equilibrium behavior of the contracting parties. In addition to the standard econometric concerns that are inherent in analyzing panel data, implicit contracts also raise issues about the power of tests to examine rare outcomes that are infrequent in equilibrium. This point is made by Demski and Sappington (1999, p. 30) in the context of senior executive turnover where they note that key

190

dimensions of penalty and reward structures may not be observed by the empirical researcher because these dimensions are not exercised in equilibrium. Another theme emerging from our survey is the interrelation among various governance characteristics. We know very little about how these mechanisms interact with each other and whether these mechanisms serve as substitutes or complements for one another. This is potentially a significant problem for accounting research, as, from a theoretical perspective, it is unclear whether the accounting system enhances existing governance mechanisms, substitutes for those mechanisms, and/or is the mechanism that managers use to disguise the extent of the firms agency problems. Similarly, a variety of contractual features are often included in contracts between external providers of capital and the firm. Additional research on how these mechanisms interact with each other, and how accounting substitutes or complements these various mechanisms would help expand our understanding of the role of accounting information in the contracting process. We also have very little evidence on which governance structures are most efficient for mitigating the various types of agency problems that arise within a firm. Studies such as Larcker, Richardson and Tuna (2007) and Dey (2008) provide a useful first step towards informing researchers about how governance mechanisms fit together. These studies are, however, admittedly exploratory. They are mostly silent on the theory regarding why we might expect to observe certain governance structures together in some settings, while other structures are more effective in other settings. The governance literature would therefore greatly benefit from additional theoretical work modeling the relation among alternative governance mechanisms, focusing on the role of the accounting system in the reduction of agency costs.

191

Our review also points out that the accounting field has accumulated a large body of descriptive evidence over the last decade. It would be useful for future research seeking to extend these results to focus on establishing a causal relationship between governance characteristics, attributes of firms accounting systems, attributes of various contracts and outcomes such as firm performance, risk-taking, or other outcomes of interest to accounting researchers. This is a difficult task because of the endogenous nature of the firms accounting system, governance characteristics and outcomes of interest. Indeed the heterogeneity in observed governance structures and debt contract features that we highlight in this survey attest to the endogenous nature of these mechanisms. Researchers have a number of tools available to potentially address these concerns. For example, careful selection of research settings and research designs can help to mitigate the endogenous nature of these relationships. Alternatively, there have been advances in the statistics and econometrics literatures in developing instrumental variable techniques to allow researchers to infer causal effects. Finally, throughout the preceding discussion we have suggested that researchers should give greater attention to the accounting attributes examined in the research setting of interest. For example, a number of accounting attributes appear to be important in debt contracting settings (including conservatism, earnings timeliness and discretionary accruals, among others). Other than some relatively recent research on conservatism, we have very little theoretical understanding about why these accounting measures are likely to be important in a debt contracting setting, and very little empirical evidence on why lenders value these accounting attributes. Similarly, there are a variety of instances where a relationship between an accounting attribute and an outcome in one particular setting is not necessarily applicable in another setting. For example, from a theoretical perspective, a greater degree of earnings timeliness is

192

likely to reduce information asymmetries between insiders and outsiders. However, even when managers are doing their best to convey their private information, a high growth firm in an uncertain business environment may still have low earnings timeliness. In a firm experiencing high growth in an uncertain business setting, one might expect that other mechanisms will arise to reduce information asymmetry, and thus low earnings timeliness will not necessarily imply that a high degree of information asymmetry exists between inside and outsiders, controlling shareholders and minority shareholders, or debt holders and equity holders. A more careful application of our existing theories in the research settings we examine will hone our understanding of the role of the accounting system in reducing agency costs.

193

References Adams, R.B., Ferreira, D., 2007. A theory of friendly boards. The Journal of Finance 62, 217250. Adams, R.B., Hermalin, B.E., Weisbach, M.S., 2009. The role of boards of directors in corporate governance: A conceptual framework and survey. Working Paper. Adut, D., Cready, W.H., Lopez, T.J., 2003. Restructuring charges and CEO cash compensation: a reexamination. The Accounting Review 78, 169-192. Aghion, P., Bolton, P., 1992. An incomplete contracts approach to financial contracting. Review of Economic Studies 59, 473-494. Agrawal, A., Chadha, S., 2005. Corporate governance and accounting scandals. The Journal of Law and Economics 48, 371-406. Agrawal, A., Jaffe, J.F., Karpoff, J.M., 1999. Management turnover and governance changes following the revelation of fraud. Journal of Law and Economics 42, 309-342. Ahmed, A.S., Billings, B.K., Morton, R.M., Stanford-Harris, M., 2002. The role of accounting conservatism in mitigating bondholder-shareholder conflicts over dividend policy and in reducing debt costs. The Accounting Review 77, 867-890. Ahmed, A.S., Duellman, S., 2007. Accounting conservatism and board of director characteristics: an empirical analysis. Journal of Accounting and Economics 43, 411-437. Ajinkya, B., Bhojraj, S., Sengupta, P., 2005. The association between outside directors, institutional investors, and the properties of management. Journal of Accounting Research 43, 343-376. Alchian, A., and Demsetz, H., 1972. Production, information costs, and economic organization. American Economic Review 62, 777-795. Ali, A., Chen, T., Radhakrishnan, S., 2007. Corporate disclosures by family firms. Journal of Accounting and Economics 44, 238-286. Allen, F., Winton, A., 1995. Corporate finance structure, incentives, and optimal contracting. In R. Jarrow, V. Maksimovic, and W. Ziemba (eds.), Finance. North-Holland, Amsterdam, 693717. Altamuro, J., Beatty, A., Weber, J., 2005. The effects of accelerated revenue recognition on earnings management and earnings informativeness: evidence form SEC staff accounting bulletin no. 101. The Accounting Review 80, 373-401. Anderson, R., Mansi, S., Reeb, D., 2004. Board characteristics, accounting report integrity, and the cost of debt. Journal of Accounting and Economics 37, 315-342. Antle, R., Smith, A., 1986. An Empirical Investigation of the Relative Performance Evaluation of Corporate Executives. Journal of Accounting Research 24, 1-39. Armstrong, C.S., Foster, G., Taylor, D.J., 2009. Earnings management around initial public offerings: a re-examination. Working Paper. Armstrong, C.S., Jagolinzer, A.D., Larcker, D.F., 2009. Chief executive officer equity incentives and accounting irregularities. Journal of Accounting Research. Forthcoming. 194

Armstrong, C.S., Larcker, D.F., Su, C.L., 2009. Stock options and chief executive officer compensation. Working Paper. Ashbaugh, H., LaFond, R., Mayhew, B.W., 2003. Do nonaudit services compromise audit independence? Further evidence. The Accounting Review 78, 611-639. Ashbaugh, H., Collins, D., LaFond, R., 2006. The effects of corporate governance on firms credit ratings. Journal of Accounting and Economics 42, 203-243. Asquith, P., Beatty, A., Weber, J., 2005. Performance pricing in debt contracts. The Journal of Accounting and Economics 40, 101-128. Baber, W., Kang, S., Liang, L., Shareholder rights, corporate governance, and accounting restatement. Working paper, Georgetown University. Baber, W., Kumar, K., Verghese, T., 1995. Client Security Price Reactions to the Laventhol and Horwath Bankruptcy. Journal of Accounting Research 33, 385-395. Baker, M., Gompers, P.A., 2003. The determinants of board structure at the initial public offering. Journal of Law and Economics 46, 569-598. Ball, R., Robin, A., Sadka, G., 2008. Is financial reporting shaped by equity markets or by debt markets? An international study of timeliness and conservatism. Review of Accounting Studies 13, 168-205. Ball, R., Shivakumar, L., 2005. Earnings quality in UK private firms: comparative loss recognition timeliness. Journal of Accounting and Economics 39, 83-128. Baltimore & Ohio R. Co. v. United States, 261 U.S. 592 (1923) Balvers, R., McDonald, B., Miller, R., 1988. Underpricing of New Issues and the Choice of Auditor as a Signal of Investment Banker Reputation. The Accounting Review 63, 605-622. Banker, R.D., Datar, S.M., 1989. Sensitivity, precision, and linear aggregation of signals for performance evaluation. Journal of Accounting Research 27, 21-39. Banker, R.D., Huang, R., Natarajan, R., 2009. Incentive contracting and value relevance of earnings and cash flows. Journal of Accounting Research 47, 647-678. Barclay, M.J., Gode, D., Kothari, S.P., 2005. Matching delivered performance. Journal of Contemporary Accounting and Economics 1, 1-25. Basu, S., 1997. The conservatism principle and the asymmetric timeliness of earnings. Journal of Accounting and Economics 24, 3-37. Beasley, M., 1996. An empirical analysis of the relation between the board of director composition and financial statement fraud. The Accounting Review 71, 443-465. Beatty, R.P., 1989. Auditor reputation and the pricing of initial public offerings. The Accounting Review 64, 693-709. Beatty, A., 2008. Discussion of the debt-contracting value of accounting information and loan syndicate structure. Journal of Accounting Research 46, 289-296. Beatty, A., Liao, S., Weber, J., 2008. Evidence on the determinants and economic consequences of delegated monitoring. Working Paper, Massachusetts Institute of Technology and Ohio State University. 195

Beatty, A., Liao, S., Weber, J., 2008. Financial reporting quality, private information, monitoring and the lease-versus-buy decision. Working Paper, Massachusetts Institute of Technology and Ohio State University Beatty, A., Liao, S., Weber, J., 2009. The effect of private information and monitoring on the role of accounting quality in investment decisions. Contemporary Accounting Research, Forthcoming. Beatty, A., Ramesh, K., Weber, J., 2002. The importance of accounting changes in debt contracts: the cost of flexibility in covenant calculations. The Journal of Accounting and Economics 33, 205-227. Beatty, A., Weber, J., 2003. The effects of debt contracting on voluntary accounting method changes. The Accounting Review 78, 119-142. Beatty, A., Weber, J., 2006. Accounting discretion in fair value estimates: An Examination of SFAS 142 Goodwill Impairments. Journal of Accounting Research 44, 257-288. Beatty, A., Weber, J., Yu, J.J., 2008. Conservatism and debt. Journal of Accounting and Economics 45, 154-174. Beck, P., Frecka, T., Solomon, I., 1988. An empirical analysis of the relationship between MAS involvement and auditor tenure: Implications for auditor independence. Journal of Accounting Literature 7, 65-84. Becker, C., Defond, M., Jiambalvo, J., Subramanyam, K., 1998. The effect of audit quality on earnings management. Contemporary Accounting Research 15. Begley, J., Freeman, R., 2004. The changing role of accounting numbers in public lending agreements. Accounting Horizons 18, 81-96. Beneish, M.D., 1999. Incentives and penalties related to earnings overstatements that violate GAAP. The Accounting Review 74, 425-457. Beneish, M., Press, E., 1993. Costs of technical violation of accounting-based debt covenants. The Accounting Review 68, 233-257. Benmelech, E., Bergman, N.K., 2008. Liquidation values and the credibility of financial contract renegotiation: evidence from U.S. airlines. Quarterly Journal of Economics 123, 16351677. Benmelech, E., Garmaise, M.J., Moskowitz, T.J., 2005. Do liquidation values affect financial contracts? Evidence from commercial loan contracts and zoning regulation. Quarterly Journal of Economics 120, 1121-1154. Bernard, V., 1989. Capital market research in accounting during the 1980s: a critical review from the state of accounting research as we enter the 1990s. Board of Trustees of the University of Illinois, Champaign. Bertrand, M., Mullainathan, S., 2003.Enjoying the quiet life? Corporate governance and managerial preferences. Journal of Political Economy 111, 1043-1075. Bertrand, M., Schoar, A., 2003. Managing with style: The effect of managers on firm policies. Quarterly Journal of Economics 118, 1169-1208.

196

Beyer, A., Cohen, D., Lys, T, Walther, B., 2009. Corporate information environment: management reports and forecasts, analysts reports, other disclosures and auditing. Working Paper. Bharath, S.T., Sunder, J., Sunder, S., 2008. Accounting quality and debt contracting. The Accounting Review 83, 1-28. Bhojraj, S., Sengupta, P., 2003. Effects of corporate governance on bond ratings and yields. The Journal of Business 76, 455-475. Biddle, G., Hilary, G., 2006. Accounting quality and firm-level capital investment. The Accounting Review 81, 963-982. Black, E.L., Carnes, T.A., Moseback, M., Moyer, S.E., 2004. Regulatory monitoring as a substitute for covenants. Journal of Accounting and Economics 37, 367-391. Blackwell, D., Noland, T., Winters, D., 1998. The Value of Auditor Assurance: Evidence from Loan Pricing. Journal of Accounting Research 36, 57-70. Blouin, J., Grein, B.M., Roundtree, B.R., 2007. An analysis of forced auditor change: the case of former Arthur Andersen clients. The Accounting Review 82, 621-650. Bolton, P., Scharfstein, D.S., 1996. Optimal debt structure and the number of creditors. Journal of Political Economy 104, 1-25. Bond, P., Goldstein, I., Prescott, E.S., 2009. Market-based corrective actions. Review of Financial Studies, Forthcoming. Boone, A., Field, L., Karpoff, J., Raheja, C., 2007. The determinants of corporate board size and independence. Journal of Financial Economics 85, 66-101. Boot, A.W.A., Thakor, A.V., 1994. Moral hazard and secured lending in an infinitely repeated credit market game. International Economic Review 35, 899-920. Boschen, J.F., Duru, A., Gordon, L.A., Smith, K.J., 2003. Accounting and stock price performance in dynamic CEO compensation arrangements 78, 143-168. Brander, J.A., Poitevin, M., 1992. Managerial compensation and the agency costs of debt finance. Managerial and Decision Economics 13, 55-64. Brickley, J., Coles, J., Jarrell, G., 1997. Leadership structure: Separating the CEO and chairman of the board. Journal of Corporate Finance 3, 189-220. Brickley, J.A., Linck, J.S., Coles, J.L., 1999. What happens to CEOs after they retire? New evidence on career concerns, horizon problems, and CEO incentives. Journal of Financial Economics 52, 341-377. Bull, C., 1997. The existence of self-enforcing implicit contracts. Quarterly Journal of Economics 102, 147-159. Burns, N., Kedia, S., 2006. The impact of performance-based compensation on misreporting. Journal of Financial Economics 79, 35-67. Bushman, R., Chen, Q., Engel, E., Smith, A., 2004. Financial accounting information, organizational complexity and corporate governance systems. Journal of Accounting and Economics 37, 167-201. 197

Bushman, R., Engel, E., Smith, A., 2006. An analysis of the relation between the stewardship and valuation roles of earnings. Journal of Accounting Research 44, 53-83. Bushman, R., Indjejikian, R., 1993a. Accounting income, stock price, and managerial compensation. Journal of Accounting and Economics 16, 3-23. Bushman, R., Indjejikian, R., 1993b. Stewardship value of distorted accounting disclosures. The Accounting Review 68, 765-782. Bushman, R., Indjejikian, R., Smith, A., 1996. CEO compensation: The role of individual performance evaluation. Journal of Accounting and Economics 21, 161-193. Bushman, R.M., Piotroski, J.D., 2006. Financial reporting incentives for conservative accounting: the influence of legal and political institutions. Journal of Accounting and Economics 42, 107-148. Bushman, R.M., Smith, A.J., 2001. Financial accounting information and corporate governance. Journal of Accounting and Economics 32, 237-333. Cai, J., Liu, Y., Qian, Y., 2009. Information asymmetry and corporate governance. Working Paper. Carcello, J., Neal, T., 2000. Audit committee composition and auditor reporting. The Accounting Review 75, 453-467. Carcello, J., Neal, T., 2003. Audit committee characteristics and auditor dismissals following new going-concern reports. The Accounting Review 78, 95-117. Chaney, P., Philipich, K., 2002. Shredded Reputation: The Cost of Audit Failure. Journal of Accounting Research 40, 1221-1245. Chava, S., Roberts, M.R., 2008. How does financing impact investment? The role of debt covenants. Journal of Finance 63, 2085-2121. Chen, S., Chen, X., Cheng, Q., 2008. Do family firms provide more or less voluntary disclosure? Journal of Accounting Research 46, 499-536. Chen, Q., Hemmer, T., Zhang, Y., 2007. On the relation between conservatism in accounting standards and incentives for earnings management. Journal of Accounting Research 45, 541565. Cheng, Q., Warfield, T.D., 2005. Equity incentives and earnings management. The Accounting Review 80, 441-476. Christensen, P.O., Demski, J.S., Frimor, H., 2002. Accounting policies in agencies with moral hazard and renegotiation. Journal of Accounting Research 40, 1071-1090. Christensen, P.O., Feltham, G.A., Sabac, F., 2005. A contracting perspective on earnings quality. Journal of Accounting and Economics 39, 265-294. Chung, H., Kallapur, S., 2003. Client Importance, Nonaudit Services, and Abnormal Accruals, The Accounting Review 78, 931-955. Coase, R., 1937. The Nature of the Firm. Economica 4, 386-405. Coates, J.C., 2007. The goals and promise of the Sarbanes-Oxley Act. Journal of Economic Perspectives 21, 91-116. 198

Coffee, J.C., 2002. Racing towards the top? The impact of cross-listings and stock market competition on international corporate governance. Columbia Law Review 102, 1757-1831. Coles, J., Daniel, N., Naveen, L., 2007. Co-opted boards: costs, benefits, causes and consequences. Working Paper, Temple University. Comprix, J., Muller, K.A., 2006. Asymmetric treatment of reported pension expense and income amounts in CEO cash compensation calculations. Journal of Accounting and Economics 42, 385-416. Constantinides, G.M., 1986. Capital market equilibrium with transaction costs. Journal of Political Economy 94, 842-862. Core, J.E., Guay, W.R., 1999. The use of equity grants to manage optimal equity incentive levels. Journal of Accounting and Economics 28, 151-184. Core, J.E., Guay, W.R., Larcker, D.F., 2003. Executive equity compensation and incentives: a survey. Federal Reserve Economic Policy Review, 27-50. Core, J.E., Guay, W.R., Verrecchia, R.E., 2003. Price versus non-price performance measures in optimal CEO compensation contracts. The Accounting Review 78, 957-981. Core, J., Holthausen, R.W., Larcker, D.F., 1999. Corporate Governance, Chief Executive Officer Compensation and Firm Performance. Journal of Financial Economics. 51, 371-406. Cornett, M.M., Marcus, A.J., Tehranian, H., 2008. Corporate governance and pay-forperformance: the impact of earnings management. Journal of Financial Economics 87, 357-373. Daniel, N.D., Denis, D.J., Naveen, L., 2008. Do firms manage earnings to meet dividend thresholds? Journal of Accounting and Economics 45, 2-26. Datar, S., Kulp, S.C., Lambert, R.A., 2001. Balancing performance measures. Journal of Accounting Research 39, 75-92. Davis, L., Ricchiute, D., Trompeter, G., 1993. Audit effort, audit fees, and the provision of nonaudit services to audit clients. The Accounting Review 68, 135, 150. Deakin, E.B., 1979. An analysis of differences between non-major oil firms using successful efforts and full cost methods. The Accounting Review 54, 722-734. DeAngelo, H., DeAngelo, L., 1985. Managerial ownership of voting rights: a study of public corporations with dual classes of common stock. Journal of Financial Economics 14, 33-69. Dechow, P., Dichev, I., 2002. The quality of accruals and earnings, The Accounting Review 77, 35-59. Dechow, P., Sloan, R., 1991. Executive incentives and the horizon problem: an empirical investigation. Journal of Accounting and Economics 14, 51-89. Dechow, P., Sloan, R., Sweeney, A., 1996. Causes and consequences of earning manipulation. Contemporary Accounting Research 13, 1-36. DeFond, M., Francis, J. 2005. Audit research after Sarbanes-Oxley? Auditing: A Journal of Practice and Theory 24, 5-30. DeFond, M., Hann, R., Hu, X., 2005. Does the market value financial expertise on audit committees of board of directors? Journal of Accounting Research 43, 153-193. 199

DeFond, M., Jiambalvo, J., 1991. Incidence and circumstances of accounting errors. The Accounting Review 66, 643-655. Demski, J.S., Frimor, H., 1999. Performance measure garbling under renegotiation in multiperiod agencies. Journal of Accounting Research 37, 187-214. Demski, J.S., Frimor, H., Sappington, D.E.M., 2004. Efficient manipulation in a repeated setting. Journal of Accounting Research 42, 31-49. Demski, J.S., Sappington, D.E.M., 1999. Summarization with errors: a perspective on empirical investigations of agency relationships. Management Accounting Research 10, 21-37. Demsetz, H., Lehn, K., 1985. The structure of corporate ownership. The Journal of Political Economy 93, 1155-1177. Desai, H., Hogan, C. E., Wilkins, M. S., 2006. The reputational penalty for aggressive accounting: earnings restatements and management turnover. The Accounting Review 81, 83112. Dey, A., 2008. Corporate governance and agency conflicts. Journal of Accounting Research 46, 1143-1181. Dhaliwal, D., I.K. Kumar, and R. Pereira., 2005. Costly public disclosure and the choice between private and public debt. Working Paper, University of Arizona. Diamond, D., 1991. Monitoring and reputation: the choice between bank loans and directly placed debt. Journal of Political Economy 99, 689-721. Dichev, I., Skinner, D., 2002. Large-sample evidence on the debt covenant hypothesis. Journal of Accounting Research 40, 1091-1123. Dikolli, S.S., 2001. Agent employment horizons and contracting demand for forward-looking performance measures. Journal of Accounting Research 39, 481-494. Dopuch, N., 1989. The impact of regulations on financial accounting research. Contemporary Accounting Research 5, 494-500. Drymiotes, G., 2007. The monitoring role of insiders. Journal of Accounting and Economics 44, 359-377. Dutta S., Reichelstein, S., 2003. Leading indicator variables, performance measurement, and long-term versus short-term contracts. Journal of Accounting Research 41, 837-866. Dutta, S., Reichelstein, S., 2005. Stock price, earnings, and book value in managerial performance measures. The Accounting Review 80, 1069-1100. Dutta S., Zhang, X., 2002. Revenue recognition in a mutliperiod agency setting. Journal of Accounting Research 40, 67-83. Dye, R. A., Earnings management in an overlapping generations model. Journal of Accounting Research 26, 195-235. Efendi, J., Srivastava, A., Swanson, E.P., 2007. Why do corporate managers misstate financial statements? The role of option compensation and other factors. Journal of Financial Economics 85, 667-708.

200

El-Gazzar, S., Pastena, V., 1990. Negotiated accounting rules in private financial contracts. Journal of Accounting and Economics 12, 381-396. El-Gazzar, S., Pastena, V., 1991. Factors affecting the scope and initial tightness of covenant restrictions in private lending agreements. Contemporary Accounting Research 8, 133-151. Engel, E., 2005. Discussion of does the market value financial expertise on audit committees of board of directors? Journal of Accounting Research 43, 195-204. Engel, E., Hayes, R.M., Wang, X., 2003. CEO turnover and properties of accounting information. Journal of Accounting and Economics 36, 197-226. Engel, E., Hayes, R.M., Wang, X., 2007. The Sarbanes-Oxley Act and firms going-private decisions. Journal of Accounting and Economics 44, 116-145. Erickson, M., Hanlon, M., Maydew, E.L., 2006. Is there a link between executive equity incentives and accounting fraud? Journal of Accounting Research 44, 113-143. Espahbodi, H., Espahbodi, P., Rezaee, Z., Tehranian, H., 2002. Stock price reaction and value relevance of recognition versus disclosure: the case of stock-based compensation. Journal of Accounting and Economics 33, 343-373. Fama, E., Jensen, M., 1983. Separation of ownership and control. The Journal of Law and Economics 26, 301-325. Fama, E. F., Miller, M., 1972, The Theory of Finance. Holt, Rinehart and Winston, New York. Fan, J.P.H., Wong, T.J., 2002. Corporate ownership structure and the informativeness of accounting earnings in East Asia. Journal of Accounting and Economics 33, 401-425. Fan, J.P.H., Wong, T.J., 2005. Do external auditors perform a corporate governance role in emerging markets? Evidence from East Asia. Journal of Accounting Research 43, 35-72. Farber (2005) Restoring trust after fraud: does corporate governance matter? The Accounting Review 80, 539-561. Feltham, G., Indjejikian, R., Nanda, D., 2006. Dynamic incentives and dual-purpose accounting. Journal of Accounting and Economics 42, 417-437. Feltham, G.A., Xie, J. 1994. Performance measure congruity and diversity in multi-task principal/agent relations. The Accounting Review 69, 429-453. Fershtman, C., Judd, K.L., 1987. Equilibrium incentives in oligopoly. American Economic Review 77, 927-940. Fich, E.M., Shivdasani, A., 2007. Financial fraud, director reputation, and shareholder wealth. Journal of Financial Economics 86, 306-336. Fields, T., Lys, T., Vincent, L., 2001. Empirical research on accounting choice. Journal of Accounting and Economics 31, 255-307. Fisher, R.A., 1922. On the mathematical foundations of theoretical statistics. Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character 222, 309-368. Francis, J., LaFond, R., Ohlsson, P., Schipper, K., 2005. The market pricing of accruals quality. Journal of Accounting and Economics 39, 295-327. 201

Francis, J., Maydew, E., Sparks, H., 1999. The role of big 6 auditors in the credible reporting of accruals. A Journal of Practice and Theory 18, 17-34. Francis, J., Schipper, K., Vincent, L., 2005. Earnings and dividend informativeness when cash flow rights are separated from voting rights. Journal of Accounting and Economics 39, 329360. Frankel, R., Johnson, M, Nelson, K., 2002. The Relation between Auditors' Fees for Nonaudit Services and Earnings Management. The Accounting Review 77, 71-105. Frankel, R., Seethamraju, C., Zach, T., 2008. GAAP goodwill and debt contracting efficiency: evidence from net worth covenants. Review of Accounting Studies 13, 87118. Fudenberg, D., Tirole, J., 1990. Moral hazard and renegotiation in agency contracts. Econometrica 58, 1279-1320. Gale, D., Hellwig, M., 1985. Incentive-compatible debt contracts: The one-period problem. Review of Economic Studies 52, 647-663. Garleanu, N., Zwiebel, J., 2008. Design and renegotiation of debt covenants. Review of Financial Studies, Forthcoming. Gaver, J.J., Paterson, J.S., 2001. The association between external monitoring and earnings management in the property-casualty insurance industry. Journal of Accounting Research 39, 269-282. Gaver, J., Paterson, J., 2007. The influence of large clients on office-level auditor oversight: evidence from the property-casualty insurance industry. Journal of Accounting and Economics 43, 299-320. Gerety, M., Lehn, K., 1997. The causes and consequences of accounting fraud. Managerial and Decision Economics 18, 587-599. Gibson, M.S., 2002. Is corporate governance effective in emerging markets? Working Paper, Federal Reserve Board. Gigler F.B., Hemmer, T., 2004. On the value of transparency in agencies with renegotiation. Journal of Accounting Research 42, 871-893. Gigler, F., Kanodia, C., Sapra, H., Venugopalan, R., 2009. Accounting conservatism and the efficiency of debt contracts. Journal of Accounting Research 47, 767-797. Gillan, S.L., Hartzell, J.C., Parrino, R., 2009. Explicit vs. implicit contracts: evidence from CEO employment agreements. Journal of Finance, Forthcoming. Gillan, S., Hartzell, J., Starks, L., 2006. Evidence on corporate governance. Working Paper. University of Texas at Austin. Gilson, S., 1990. Bankruptcy, boards, banks, and blockholders. Journal of Financial Economics 27, 355-387. Gjesdal, F., 1981. Accounting for Stewardship. Journal of Accounting Research 19. 208-231. Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of Economics 118, 107-155.

202

Graham, J., Harvey, C., Rajgopal, S., 2005. The economic implications of corporate financial reporting. Journal of Accounting and Economics 40, 3-73. Grossman, S., Stiglitz, J., 1980. On the Impossibility of Informationally Efficient Markets. The American Economics Review 70, 393-408. Guay, W. R., 1999. The sensitivity of CEO wealth to equity risk: an analysis of the magnitude and determinants. Journal of Financial Economics 53, 43-71. Guay, W., 2008. Conservative financial reporting, debt covenants, and the agency costs of debt. Journal of Accounting and Economics 45, 175-180. Guay, W., Verrecchia, R., 2006. Discussion of an economic framework for conservative accounting and Bushman and Piotroski (2006). Journal of Accounting and Economics 42, 149165. Hagerty, K.M., Segal, D.R., 1988. On the observational equivalence of managerial contracts under moral hazard and self-selection. Quarterly Journal of Economics 103, 425-428. Hall, B.J., Liebman, J.B., 1998. Are CEOs really paid like bureaucrats? Quarterly Journal of Economics 113, 653-691. Hanlon, M., Heitzman, S., 2009. A review of tax research. Working Paper. Harris, J., Bromiley, P., Incentives to cheat: the influence of executive compensation and firm performance on financial misrepresentation. Organizational Science 18, 350-367. Harris, M., Raviv, A., 2008. A theory of board control and size. Review of Financial Studies 21, 1797-1832. Harris, M., Raviv, A., 1991. The theory of capital structure. Journal of Finance 46, 297-355. Hart, O., 1995. Firms, Contracts, and Financial Structure. Oxford University Press. Hart, O., 2001. Financial contracting. Journal of Economic Literature 39, 1079-1100. Hart, O., Moore, J., 1998. Default and renegotiation: A dynamic debt model. Quarterly Journal of Economics 113, 1-41. Hart, O., Moore, J., 1994. A theory of debt based on the inalienability of human capital. Quarterly Journal of Economics 109, 841-879. Hart, O.D., Tirole, J., 1988. Contract renegotiation and coasian dynamics. Review of Economic Studies 55, 509-540. Hayes, R.M., Schaefer, S., 2000. Implicit contracts and the explanatory power of top executive compensation for future performance. RAND Journal of Economics 31, 273-293. Healy, P.M., 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics 7, 85-107. Healy, P., Palepu, K., 1990. Information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature. Journal of Accounting and Economics 31, 405-440. Heinle, M.S., Hofmann, C. Soft information and the stewardship value of accounting disclosure. Working Paper. 203

Hemmer, T., 2004. Lessons Lost in Linearity: A Critical Assessment of the General Usefulness of LEN Models in Compensation Research. Journal of Management Accounting Research 16, 149-162. Hemmer, T., 2006. On the subtleties of the principal-agent model. In Essays on Accounting Theory in Honour of Joel S. Demski, eds. R. Antle, F. Gjesdal, and P. Liang. Hemmer, T., 2007. Discussion of Limited commitment in multi-agent contracting. Contemporary Accounting Research 24, 377-385. Hermalin, B., 2001. Economics and corporate culture. The Handbook of Organizational Culture and Climate, John Wiley and Sons. Hermalin, B., 2005. Trends in corporate governance. The Journal of Finance 60, 2351-2384. Hermalin, B.E., Katz, M.L., 1991. Moral hazard and verifiability: the effects of renegotiation in agency. Econometrica 59, 1735-1753. Hermalin, B.E., Weisbach, M.S., 1998. Endogenously chosen boards of directors and their monitoring of the CEO. American Economic Review 88, 96-118. Hermalin, B., Weisbach, M., 2003. Boards of directors as an endogenously determined institution: a survey of the economic literature. Economic Policy Review 9, 7-26. Hermalin, B. E., Weisbach, M.S., 2008. Information disclosure and corporate governance. Working Paper. Heron, R.A., Lie, E., 2009. What fraction of stock option grants to top executives have been backdated or manipulated? Management Science 55, 513-525. Heron, R.A., Lie, E., 2007. Does backdating explain the stock price pattern around executive stock option grants? Journal of Financial Economics 83, 271-295. Hogan, C.E., 1997. Costs and benefits of audit quality in the IPO market: a self-selection analysis. The Accounting Review 72, 67-86. Holderness, C.G., 2003. A survey of blockholders and corporate control. Federal Reserve Economic Policy Review, 51-64. Holmstrom, B., 1979. Moral hazard and observability. Bell Journal of Economics 10, 74-91. Holmstrom, B., 2005. Pay without performance and the managerial power hypothesis. Journal of Corporation Law, 711-712. Holmstrom, B., Milgrom, P., 1987. Aggregation and linearity in the provision of intertemporal incentives. Econometrica 55, 303-328. Holmstrom, B., Milgrom, P., 1991. Multitask principal-agent analysis: incentive contracts, asset ownership, and job design. Journal of Law, Economics, and Organization 7, 524-552. Holthausen, R.W., Larcker, D.F., Sloan, R.G., 1995. Annual bonus schemes and the manipulation of earnings. Journal of Accounting and Economics 19, 29-74. Holthausen, R.W., Watts, R.L., 2001. The relevance of value-relevance literature for financial accounting standard setting. Journal of Accounting and Economics 31, 3-75.

204

Huebner, M., McCullough, H., 2008. The fiduciary duties of directors of troubled U.S. companies: emerging clarity. The International Comparative Legal Guide to: Corporate Recovery and Insolvency 2008. Hui, K., Matsunaga, S., Morse, D., 2009. The impact of conservatism on management earnings forecasts. Journal of Accounting and Economics 47, 192-207. Ijiri, Y., 1967. The Foundations of Accounting Measurement. Prentice Hall. Indjejikian R.J., Nanda, D., 2002. Executive target bonuses and what they imply about performance standards. Accounting Review 77, 793-819. Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76, 323-329. Jensen, M., 1993. The modern industrial revolution, exit, and the failure of internal control systems. The Journal of Finance 48, 831-880. Jensen, M. C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305-360. Jensen, M.C., Meckling, W.H., 1995. Specific and general knowledge, and organizational structure. Journal of Applied Corporate Finance 8, 4-18. Jensen, M., Murphy, K., 1990. CEO incentives: it's not how much, but how. Harvard Business Review. Jensen, M., Ruback, R., 1983. The market for corporate control, Journal of Financial Economics 11, 5-50. John, T.A., John, K., 1993. Top-management compensation and capital structure. The Journal of Finance 48, 949-974. Johnson, S.A., Ryan, H.E., Tian, Y.S., 2009. Managerial incentives and corporate fraud: the source of incentives matters. Review of Finance 13, 115-145. Kadan, O., Swinkels, J.M., 2008. Stock or options? Moral hazard, firm viability, and the design of compensation contracts. Review of Financial Studies 21, 451-482. Karamanou, I., Vafeas, N., 2005. The association between boards, audit committees, and management earnings forecast: an empirical analysis. Journal of Accounting Research 43, 453486. Karpoff, J., Lee, D., Martin, G., 2008. The consequences to managers for financial misrepresentation. Journal of Financial Economics 88, 193-215. Khurana, I., Raman, K., 2004. Litigation Risk and the Financial Reporting Credibility of Big 4 versus Non-Big 4 Audits: Evidence from Anglo-American Countries. The Accounting Review 79, 473-495. Kinney, W., Palmrose, Z., Scholz, S., 2004. Auditor Independence, Non-Audit Services, and Restatements: Was the U.S. Government Right? Journal of Accounting Research 42, 561-588. Klein, A., 2002a. Economic determinants of audit committee independence. The Accounting Review 77, 435-452.

205

Klein, A., 2002b. Audit committee, board of director characteristics, and earnings management. Journal of Accounting and Economics 33, 375-400. Kothari, S.P., 2001. Capital markets research in accounting. Journal of Accounting and Economics 31, 105-231. Kothari, S.P., Shu, S., Wysocki, P.D., 2009. Do managers withhold bad news? Working Paper. Krishnan, R., 2005. The effect of changes in regulation and competition on firms demand for accounting information. The Accounting Review 80, 269-287. Kyle, A. S. 1985. Continuous auctions and insider trading. Econometrica 53, 1315-1335. LaFond, R., Roychowdhury, S., 2008. Managerial ownership and accounting conservatism. Journal of Accounting Research 46, 101-135. LaFond, R., Watts, R., 2008. The information role of conservatism. The Accounting Review 83, 447-478. Lambert, R. A., 1983. Long-term contracts and moral hazard. Bell Journal of Economics 14, 441-452. Lambert, R. A., 1993. The use of accounting and security price measures of performance in managerial compensation contracts: a discussion. Journal of Accounting and Economics 16, 101-123. Lambert. R.A., 2001. Contracting theory and accounting. Journal of Accounting and Economics 32, 3-87. Lambert, R.A., Lanen, W.N., Larcker, D.F., 1989. Executive stock option plans and corporate dividend policy. Journal of Financial and Quantitative Analysis 24, 409-425. Lambert, R.A., Larcker, D., 1987. An analysis of the use of accounting and market measures of performance in executive compensation contracts. Journal of Accounting Research 25, 42-54. Lang, M., Raedy, J., Yetman, M., 2003. How Representative Are Firms That Are Cross-Listed in the United States? An Analysis of Accounting Quality. Journal of Accounting Research 41, 363-386. Larcker, D.F., Richardson, S.A., Tuna, I., 2007. Corporate governance, accounting outcomes, and organizational performance. The Accounting Review 82, 963-1008. Larcker, D.F., Richardson, S.A., 2004. Fees paid to audit firms, accrual choices, and corporate governance. Journal of Accounting Research 42, 625-658. Lee, C.M.C., 2001. Market efficiency and accounting research. Journal of Accounting and Economics 31, 233-253. Lehn, K., Patro, S., Zhao, M., 2008. Determinants of the size and composition of US corporate boards. Working Paper. University of Pittsburgh. Lennox, C. 2005. Audit quality and executive officers affiliations with CPA firms. Journal of Accounting and Economics 39, 201-231. Leuz, C., Wysocki, P.D., 2009. Economic consequences of financial reporting and disclosure regulation: a review and suggestions for future research. Working Paper. 206

Lev, B., Ohlson, J., 1982. Market-based empirical research in accounting. Journal of Accounting Research 20, 249-322. Lewellen, W.C., Loderer, C., Martin, K., 1987. Executive compensation contracts and executive incentive problems: an empirical analysis. Journal of Accounting and Economics 9, 287-310. Li, J. 2009. Accounting Conservatism and Debt Contracts: Efficient Liquidation and Covenant Renegotiation. Working Paper, Columbia University. Liang, P.J., Zhang, X.J., 2006. Accounting treatment of inherent versus incentive uncertainties and the capital structure of the firm. Journal of Accounting Research 44, 145-176. Linck, J., Netter, J., Yang T., 2008. The determinants of board structure. Journal of Financial Economics 87, 308-328. Mansi, S.A., Maxwell, W.F., Miller, S.P., 2004. Does auditor quality and tenure matter to investors? Evidence from the bond market. Journal of Accounting Research 42, 755-793. Manso, G. Strulovici, B., Tchistyi, A., 2009. Performance sensitive debt. Working Paper, Massachusetts Institute of Technology. Marquardt, C., Wiedman, C., 2004. Earnings management through transaction structuring: contingent convertible debt and diluted earnings per share. Journal of Accounting Research 43, 205- 243. Matsunaga, S.R., Park, C.W., 2001. The effect of missing a quarterly earnings benchmark on the CEOs annual bonus. The Accounting Review 76, 313-332. McConnel, J.J., 2003. Outside directors. Financial Review 38, 25-31. Mehran, H., 1995. Executive compensation structure, ownership, and firm performance. Journal of Financial Economics 38, 163-184. Menon, K., Williams, D., 1994. The use of audit committees for monitoring. Journal of Accounting and Public Policy 13, 121-139. Menon, K., Williams, D., 2004. Former audit partners and abnormal accruals. The Accounting Review 79, 1095-1118. Mergenthaler, R., Rajgopal, S., Srinivasan, S., 2009. CEO and CFO career penalties to missing quarterly analysts forecasts, Working Paper. Merton, R.C., 1974. On the pricing of corporate debt: the structure of interest rates. Journal of Finance 29, 449-470. Milgrom. P., 1987. Adverse selection without hidden information. Working Paper. Milgrom, P., Roberts, J., 1992. Economics, Organization, and Management. Englewood Cliffs NJ: Prentice Hall. Modigliani, F., Miller, M., 1958. The cost of capital, corporation finance and the theory of investment. American Economic Review 48, 261-297. Murphy, K.J., 1998. Executive Compensation in Handbook of Labor Economics, Volume 3, 2485-2557. 207

Murphy, K., Oyer, P., 2003.Discretion in executive incentive contracts. Working Paper, University of Southern California and Stanford University. Myers, S.C., 1977. Determinants of corporate borrowing, Journal of Financial Economics 5, 147-175. Myers, S.C., Majluf, N.S. 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, 187-221. Nagar, V., Nanda, D., Wysocki, P., 2003. Discretionary disclosure and stock-based incentives. Journal of Accounting and Economics 34, 283-309. Nelson, K.K., Price, R.A., Rountree, B.R., 2008. The market reaction to Arthur Andersens role in the Enron scandal: Loss of reputation or confounding effects? Journal of Accounting and Economics 46, 279-293. Nicholaev, V., 2008. Debt covenants and accounting conservatism. Working Paper, University of Chicago. Nini, G., Smith, D.C., Sufi, A., 2009. Creditor control rights, corporate governance, and firm value. Working Paper, University of Pennsylvania. OConnor, J.P., Priem, R.L., Coombs, J.E., Gilley, K.M., 2006. Do CEO stock options prevent or promote fraudulent financial reporting? Academy of Management Journal 49, 483-500. Ongena, S., Smith, D.C., 2001. The duration of bank relationships. Journal of Financial Economics 61, 449-475. Palmrose, Z., 1988. An analysis of auditor litigation and audit service quality. The Accounting Review 63, 55-73. Parrino. R., Weisbach, M.S., 1999. Measuring investment distortions arising from stockholderbondholder conflicts. Journal of Financial Economics 53, 3-42. Paul, J., 1992. On the efficiency of stock-based compensation. Review of Financial Studies 5, 471-502. Pittman, J.A., Fortin, S., 2004. Auditor choice and the cost of debt capital for newly public firms. The Journal of Accounting and Economics 37, 113-136. Prendergast, C., 2002. The tenuous trade-off between risk and incentives. Journal of Political Economy 110, 1071-1102. Raheja, C., 2005. Determinants of board size and composition: A theory of corporate boards. Journal of Financial and Quantitative Analysis 40, 283-306. Rajan, R.G., 1992. Insiders and outsiders: the choice between informed and arms-length debt. Journal of Finance 47, 1367-1400. Rajan, M.V., Reichelstein, S., 2009. Objective versus subjective indicators of managerial performance. The Accounting Review 84, 209-237. Reese, W.A., Weisbach, M.S., 2002. Protection of minority shareholder interests, cross-listings in the United States, and subsequent equity offerings. Journal of Financial Economics 66, 65104.

208

Reynolds, J.K., Read, W.J., Whisenant, J.S., 2004. Professional service prices and auditor objectivity. Auditing: A Journal of Practice and Theory 23, 29-52. Richardson, S., Tuna, I., Wysocki, P., 2009. Accounting anomalies, fundamental analysis, and returns to investment strategies. Working Paper. Roberts, M., Sufi, A., 2008. Financial contracting: a survey of empirical research and future directions. Annual Review of Financial Economics. Roberts, M.R., Sufi, A., 2009. Financial contracting: a survey of empirical research and future directions. Annual Review of Financial Economics 1, 1-20. Rogerson. W., 1985. Repeated moral hazard. Econometrica 53, 69-76. Roychowdhury, S., 2006. Earnings management through real activities manipulation. Journal of Accounting and Economics 42, 335-370. Schmidt, K.M., Schnitzer, M. 1995. The interaction of explicit and implicit contracts. Economic Letters 48, 193-199. Schipper, K., 1989. Commentary on earnings management. Accounting Horizons 3, 91-102. Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance, 52, 737783. Sengupta, P., 1998. Corporate disclosure quality and the cost of debit. The Accounting Review 73, 459-474. Schleifer, A., Vishny, R.W., 1992. Liquidation values and debt capacity: a market equilibrium approach. Journal of Finance 47, 1343-1366. Shleifer, A., Vishny, R., 1997. A survey of corporate governance. The Journal of Finance52, 737-783. Simunic, D., 1984. Auditing, Consulting, and Auditor Independence. Journal of Accounting Research 22, 679-702. Smith, C.W., 1993. A perspective on accounting-based debt violations. The Accounting Review 68, 289-303. Smith, C.W., Stulz, R.M., 1985. The determinants of firms hedging policies. Journal of Financial and Quantitative Analysis 20, 391-405. Smith, C.W., Warner, J., 1979. On financial contractingan analysis of bond covenants. Journal of Financial Economics 7, 117-161. Sridhar, S., Magee, R., 1996. Financial contracts, opportunism, and disclosure management. Review of Financial Studies 1, 225-258. Srinivasan, S., 2005. Consequences of financial reporting failure for outside directors: Evidence from accounting restatements and audit committee members. Journal of Accounting Research 43, 291-334. Sufi, A. 2009. Bank lines of credit in corporate finance: an empirical analysis. Review of Financial Studies 22, 1057-1088.

209

Sufi, A. 2007. Information asymmetry and financing arrangements: evidence from syndicated loans. The Journal of Finance 62, 629- 668. Sweeney, A., 1994. Debt-covenant violations and managers accounting responses. Journal of Accounting and Economics 17, 281-308. Teoh, S., Welch, I., Wong, T.J., 1998. Earnings management and the underperformance of seasoned equity offerings. Journal of Financial Economics 50, 63-99. Teoh, S., Wong, T., 1993. Perceived auditor quality and the earnings response coefficient. The Accounting Review 68, 346-366. Tirole, J., 1999. Incomplete contracts: where do we stand? Econometrica 67, 741-781. Tirole, J., 2001. Corporate Governance. Econometrica 69, 1-35. Tirole, J., 2006. The Theory of Corporate Finance. Princeton University Press. Townsend, R.M., 1979. Optimal contracts and competitive markets with costly state verification. Journal of Economic Theory 21, 265-293. Verrecchia, R.E., 2001. Essays on disclosure. Journal of Accounting and Economics 32, 97180. Wang, D., 2006. Founding family ownership and earnings quality. Journal of Accounting Research 44, 619-656. Wang, G., Wong, T., Xia, L., 2008. State ownership, the institutional environment, and auditor choice: Evidence from China. Journal of Accounting and Economics 46, 112-134. Watts, R., 2003a. Conservatism in accounting, part I: explanations and implications. Accounting Horizons 17, 207-221. Watts, R. 2003b. Conservatism in accounting, part II: evidence and research opportunities. Accounting Horizons 17, 287-301. Watts, R.L., Zimmerman, J.L., 1978. Towards a positive theory of the determination of accounting standards. Accounting Review 53, 112134. Watts, R.L., Zimmerman, J.L., 1983. Agency problems, auditing, and the theory of the firm. Journal of Law and Economics 26, 613. Watts, R.L., Zimmerman, J.L., 1986. Positive Accounting Theory. Prentice-Hall, Englewood Cliffs, NJ. Watts, R.L., Zimmerman, J.L., 1990. Positive accounting theory: a ten-year perspective. The Accounting Review 65, 131156. Weber, J., Willenborg, M., 2003. Do Expert Informational Intermediaries Add Value? Evidence from Auditors in Microcap IPOs. Journal of Accounting Research 41, 681-720. Weber, J., Willenborg, M., Zhang, J. 2008. Does Auditor Reputation Matter? The Case of KPMG Germany and ComROAD AG. Journal of Accounting Research 46, 941-972. Willenborg, M., Empirical analysis of the economic demand for auditing in the initial public offerings market. Journal of Accounting Research 37, 225-238.

210

Williamson, O.E., 1988. Corporate finance and corporate governance. The Journal of Finance 43, 567-591. Wittenberg-Moerman, R., 2008. The role of information asymmetry and financial reporting quality in debt trading: evidence from the secondary loan market. Journal of Accounting and Economics 46, 240-260. Xie, B., Davidson, W., DaDalt, P.J., 2003. Earnings management and corporate governance: the role of the board and the audit committee. Journal of Corporate Finance 9, 295-316. Zhang, J. 2008. The contracting benefits of accounting conservatism to lenders and borrowers. The Journal of Accounting and Economics 46, 27-54.

211

Appendix A: A model of constrained shareholder value maximization This appendix develops a model of the shareholders problem of maximizing firm value subject to various constraints that comprise the nexus of the firms contracts. In our model, we capture the self-interested behavior of the various contracting parties through the objective function of the shareholders (which is to maximize net firm value) and the incentive compatibility (IC) and individual rationality (IR) constraints of the other contracting parties. Our model is deliberately general and a majority of the papers that we discuss in this survey can be classified in this framework.83 In addition, we also confine our model to consist of a single period. Although this has the benefit of making the model relatively transparent, it comes at the cost of omitting a number of key features that are empirically descriptive of observed contracting environments (e.g., accrual accounting, renegotiation, reputation and commitment). Our primary concerns are to capture how the behaviors of the various parties to the firms contracts interrelate to one another and to supply signs on the comparative statics that we believe to be intuitive. We formulate the problem from the perspective of the firms shareholders with the objective of maximizing net firm value (i.e., cash flows less any costs such as interest expense and compensation payments to the manager). We adopt the standard convention of viewing the problem from the perspective of the shareholders but acknowledge that certain decisions that we model as the choice of the shareholders are often delegated to other parties (e.g., the board of directors).84 This delegation can potentially add another layer of agency problems that are not

83

We also refrain from specifying the functional for of the various contracts and variables in the model. This allows us to discuss a number of realistic features of observed contracting environments (e.g., non-linear compensation contracts and non-normal cash flow distributions) without worrying about the tractability of the resulting model. 84 This modeling convention is discussed by Tirole (2006) who notes that in principle, the shareholders, perhaps through the board of directors, are in control. In practice, asymmetric information between insiders and outside

212

captured in our model.85 The primary goal of our model is to illustrate how the various objective functions that are bound together in the firms nexus of contracts interrelate and how there are often competing demands for on the firms accounting attributes. The shareholders constrained maximization problem is as follows.

subject to: (IC1) (IR1) (IC2) (IR2) (IC3) (IR3) (Accounting Properties) (Interest Rate) (Cash Flow) (Net Income) (Valuation Signal) (Rational Price)

A.1. Shareholders Objective We model the shareholders as risk-neutral and concerned with maximizing the expected value of the firm which is defined as (gross) cash flows, , less the cost of the managers

shareholders introduces an important distinction between formal authority, held by shareholders, and real or effective authority, often enjoyed by managers. 85 As an example, consider the case where a captured board is responsible for contracting with the CEO. In this case, there is likely to be an additional layer of agency costs that result in a further reduction in the shareholders objective relative to the first-best solution that obtains in the absence of agency costs.

213

compensation contract,

, and interest payments to creditors,

. The shareholders choice

variables are (i) the form of the compensation contract, including the relative weights to place on price, , and accounting earnings, , (ii) characteristics of the board of directors, b (e.g., which directors to elect, the composition of inside and outside directors and which committees to have), (iii) a social choice function, , that aggregates the preferences, , of the shareholders,

management, directors and creditors (indexed by i = 1-4, respectively) to decide on the features of the financial accounting (e.g., the degree of timeliness, conservatism and the degree of discretion accorded to management when preparing the accounting reports) and (iv) their preferences for financial accounting attributes and policies of the firm, A.2. Management The managers are represented by their incentive compatibility constraint and their participation or individual rationality constraint, denoted (IC1) and (IR1), respectively. Without loss of generality, the manager takes two actions, and , that have different marginal effects .86

on the firms cash flows. As is typical in the multitasking literature, we assume that one of the actions, say , results in a greater marginal increase in cash flows so that , but that the .87 Our

second action provides a greater non-pecuniary benefit to the agent so that

formulation allows for the possibility that certain properties of the accounting system can reduce the benefits of the undesirable action so that .88 We also allow for the possibility that

86

As we discuss further below, it is not always the case that the attributes of the firms financial reporting are the choice of the firms shareholders. Empirical researchers often assume that the financial accounting system is the choice variable of either the board or management. We discuss how both cases can be accommodated by altering the incentive compatibility constraints of either of these parties. The point is that it is important to consider who controls the firms financial reporting, which depends on the research design, and to consider how that affect both that partys incentives as well as those of other parties in the firms nexus of contracts. 87 Alternatively, rather than providing non-pecuniary benefits to the manager, the second action might be increase net income without providing a corresponding increase in cash flows. In this case, this action would be consistent with window dressing or earnings management. 88 An example that is frequently cited in the literature is that conservative accounting reports reduce the ability of managers to continue pursuing negative net present value projects from which they derive private benefits.

214

certain attributes of the board related to its monitoring function (which we discuss in more detail below) can also reduce the benefits of these actions, or . The managers compensation

contract is a function of both the realized stock price the mangers degree of risk aversion, , and is represented by f(P(),). The managers problem is to choose the actions that maximize total (pecuniary and non-pecuniary) expected utility conditional on the compensation contract, the properties of the accounting system (i.e., c) and the boards monitoring attributes (i.e., m). The manager maximizes his expected utility from the contract, , that is a function of , from

his degree of risk aversion, . The manager also derives some non-pecuniary benefit,

taking each action which is also a function of the financial accounting attributes and the amount of monitoring by both the board and the firms creditors, denoted manager incurs disutility associated with each action, . Finally, the

that is also a function of the strategic

advice supplied by the board, denoted s. It is reasonable to assume that a board with more insiders is in a better position to offer valuable strategic advice such that it reduces the disutility associated with taking each action, .

Discuss how there can be selection on, say, risk-aversion or talent (which might be modeled as reducing the marginal cost of effort). Most empirical contracting research suffers from an identification problem that stems from the endogenous matching of executives with firms (which includes both firm characteristics and the accompanying compensation contracts). A.3. Board of Directors The board of directors is represented by the constraints (IC2) and (IR2). The boards objective consists of maximizing their expected utility, , from firm value less their cost of . The boards choice , the amount of

supplying strategic advice to and monitoring the firms management,

variables are the amount of strategic advice, , the amount of monitoring,

215

debt, and the its preferences for the firms financial accounting policies. Our representation of the boards utility function accommodates a risk-averse board which can add an additional layer of agency costs since their objective differs from that of the shareholders. The most plausible comparative statics for the boards cost function is that it is increasing in both the amount of strategic advice and the amount of monitoring, or and

(e.g., Harris and Raviv, 2008). The more interesting comparative statics, however, are likely to come from the cross-partial derivatives. For example, it is often assumed that the presence of creditors acts as a substitute monitoring mechanism that facilitates the monitoring actions of the board, or . Another common assumption is that outsiders are better able , where b represents the fraction of the

to monitor, which would be represented as

directors that are unaffiliated with the firms management (i.e., outsiders). A related comparative static that, to our knowledge, has not been examined empirically in the accounting literature is that outsiders raise the cost of supplying strategic advice to management due to their lack of firm-specific knowledge, or . A related extension would be that higher quality

accounting information ameliorates this cost by providing outsiders with more relevant information for their strategic decisions, or , where ci captures some dimension of

quality of the accounting reports. A similar argument is often made that certain properties of financial accounting information (e.g., conservatism) monitoring, or . Note that this

argument relates to the second partial derivative of the cost function and is therefore a statement about the average benefit irrespective of any characteristics of the board. The previous discussion treated the financial reporting preferences of the board as independent of the boards characteristics (i.e., ). However it is more likely that boards

preferences will be related to its characteristics. For example, it is often argued that more 216

independent boards demand higher quality financial reports in order to facilitate monitoring, which translates into the prediction that the board and A.4. Creditors The creditors are captured by the constraints (IC3) and (IR3). They are modeled as riskneutral and therefore concerned with maximizing the expected value of their payoff from the debt contract (defined as the minimum of (i) the firms cash flows and (ii) the sum of interest and principal) less the cost of monitoring, monitoring effort, . The creditors choice variables are the amount of . represents accounting quality. , where b represents the independence of

and their preferences for accounting characteristics,

We model the creditors as offering a menu of interest rates that are a function of the amount borrowed, the anticipated actions of the manager (denoted anticipated monitoring effort (denoted and ), the creditors

) and the accounting policies of the firm. One natural .

comparative static is that the cost of borrowing is increasing in the amount borrowed, or

It is also natural to assume that the interest rate is increasing in the anticipated amount of monitoring in order to compensate the lender for the cost of this effort, actions of the agent (e.g., . If certain

) affect the second and higher moments of the distribution of cash

flows the creditor will price protect itself in the form of a higher interest rate to the extent this risk-substitution is anticipated, or . We also model the creditors participation constraint

(IR3) is written as an equality to reflect a competitive lending market that incorporates the creditors anticipated cost of monitoring. A multiperiod extension of the model might also include the most recent periods cash flow and/or net income as determinants of the interest rate to capture the relatively recent security design innovation of performance pricing, or 217 such that

implies that the interest rate is decreasing in accounting income from the previous period. Since the managers payoff is a function of price (which is decreasing in the interest rate), performance pricing might provide the manager with incentives to inflate earnings. A.5. Accounting Properties As discussed above, we model the firms accounting properties as resulting from a social choice function, , (that is chosen by the shareholders) that maps the preferences of the

shareholders, managers, board and creditors into the accounting properties of the firm. This reflects the fact that the accounting properties of the firm are potentially shaped by a variety of parties that have different preferences and incentives. An example of one such rule might be that creditors receive zero weight unless certain states occur (e.g., the violation of a debt covenant) in which case the creditors preferences determine the accounting policy (i.e., receive 100% weight). A multiperiod extension of this model would have a realization of cash flow and net income in both the first and second periods (i.e., and , respectively) and

one such decision rule would be that the creditors receive zero weight in the first period and their weight in the second period would be specified as a function of the first period realizations of cash flow and/or net income. Although we take the social choice function, , as a given in our model, it likely to be

a complex mechanism design problem where the shareholders have to design a mechanism that provides incentives for the various parties to act on the basis of their private information. The allocation of control right for purposes of determining the financial reporting properties of the firm is an area that has received comparatively little research and is likely to be a fruitful avenue for future theoretical and empirical research. A.6. Cash Flow and Accounting Income

218

The function

maps the managers efforts, the amount of debt and the interest rate into

the firms cash flows. It is common to assume that expected cash flows are concave in the amount of debt raised by the firm such that and . This captures the fact that cash

flows are at first increasing in the amount of debt (as the firm has growth opportunities that can absorb the additional capital) but then decreasing as the firm exhausts its profitable investment opportunities. It might also be reasonable to assume that higher moments of the cash flow distribution are affected by the amount of debt in the firms capital structure if the firm invests in risky positive net present value projects with the proceeds. This, of course, is inconsistent with the Modigliani and Millers (1958) theorem which assumes that the probability distribution of cash flows is independent of the firms capital structure but is consistent with a large body of agency literature. Although many moral hazard models (e.g., those in the LEN tradition) model the managers action(s) as shifting the mean of the outcome distribution, it is probably more fruitful to assume that the managers efforts can also affect higher moments of the distribution (e.g., both the mean and the variance of the cash flow distribution are parameterized by the managers action(s)). For example, it could be that the variance rather than (or in addition to) the mean of the outcome distribution is parameterized by such that the variance is increasing in effort. This

would capture the notion that risk-taking incentives provided by convex compensation contracts can increase the expected value of the firm to the benefit of the shareholders but to the detriment of the firms creditors. The financial accounting system is modeled as a function that maps the firms cash

flows into net income according to the financial accounting properties of the firm. A.7. Rational Price

219

The final piece of the model is a price function,

, that maps the valuation signal (that

is a noisy function of the firms cash flows) and accounting income into price. In order to prevent price from being fully revealing, we follow the convention in the rational expectations literature and introduce a random aggregate supply of the firms shares, . The price is expressed net of both the payoff to the manager and interest payments to the firms creditors. There are a number of alternative choices for the price function . One possible choice

is to assume perfect competition where managers have homogeneous information, and a pricing function such as that in the CAPM will obtain. Another plausible choice is one of imperfect competition with heterogeneous information as in Kyle (1985). A multiperiod extension of this model might incorporate the role of price as a disciplining mechanism on management through the market for corporate control. In this case, a low realization of price after the first period might prompt a takeover attempt by an institutional shareholder. A multiperiod extension would also likely make use of price as mechanism for learning information that is not available from other sources in order to make a better intermediate decision. For example, the board could learn about firm performance from price which might help it to make a better intermediate turnover decision (i.e., fire the manager or allow him to continue). The board could condition its choice of monitoring effort in the second period on the price observed after the first period. Similarly, in the event of a covenant violation after the first period, the firms creditors could condition their decision on whether to demand repayment of forgive the violation on price. Another role for price related to the firms creditors is as a signaling mechanism that conveys information about the firms choice of capital structure. The degree to which price is useful as a mechanism for conveying information (as opposed to

220

simply clearing the market for the firms shares) depends crucially on what is assumed for the price function, A.8. Synthesis Although the above model contains only a few of the many parties that contract with the firm, it should be sufficient to illustrate the complexities that are inherent in analyzing the various features of a firms governance and contracting environment. It also illustrates how the firm can be viewed as a portfolio or nexus of contracts, which we model through a series of constraints. The contracts are not likely to be independent and, in many cases, can serve as either substitutes or complements for one another. From an empirical perspective, our model should illustrate that observed data such as board characteristics (i.e., m), compensation contracts (i.e., f(P())), interest rates (i.e., r), and attributes of the financial accounting system (i.e., c) are unlikely to be exogenous, but are instead the result of a complex multi-level and multi-dimensional optimization problem involving multiple parties that contract with the firm. As we discussed earlier, the degree to which this should be accounted for in the empirical design is a function of both the research question and the strength of conclusions that are sought by the researcher (i.e., association vs. causality). .

221

Appendix B: Overview of Theoretical Contracting Results This appendix provides a more thorough discussion of the theoretical contracting results that are previewed in Section 3.3. The discussion is based on Figure 1 and generally progresses chronologically. Moral Hazard The standard moral hazard model (e.g., Holmstrom, 1979) begins with the principal designing a contract to induce the agent to take a single personally costly and unobservable (and therefore non-contractible) action that affects the terminal value of the firm according to the productivity of the action. Although the contract can be either explicit or implicit, most agency models involve only a single period so the resulting contract must be explicit (i.e., there is no repeated game that is necessary to support an equilibrium in which the contracting parties have an incentive to cooperate). A natural extension of a single action model is one in which the agent is responsible for taking multiple actions. In this case, the principals problem is more complicated because the contract must also be designed to induce the agent to balance his effort across the various activities according to their marginal productivity. Regardless of the number of actions, in moral hazard models, it is the unobservability of the agents actions that creates an agency problem in that if the principal desires for the agent to take an action other than the least costly, he must provide the agent with incentives to do so. Fortunately for the principal, the agents actions often affect various performance measures that are sensitive to the agents actions and are also related to the outcome that the principal is trying to achieve (e.g., maximizing firm value). Such measures include accounting earnings and stock price (among many others), and allow the principal access to some information regarding the agents effort. As noted above, the difference between the sensitivity

222

of a performance measure to the agents action and the sensitivity of firm value to the action is referred to as the congruence of the performance measure and is denoted due to its geometric interpretation.89 For example, if the principals objective were to maximize terminal value of the firm, and terminal value is one of the performance measures available for contracting then this would be a perfectly congruent performance measure (i.e., = 0).90 However, as noted by Feltham and Xie (1994), the first-best outcome can be achieved if and only if a performance measure is both perfectly congruent and noiseless. That is, even in the case where there is a perfectly congruent performance measure, if the measure reflects the agents actions with noise, it will not receive all of the weight in the compensation contract. This is because noise exposes the agent to risk from shocks to the terminal value that are beyond the agents control (such as 2 in Figure 1). Instead, the principal can typically achieve better risk sharing with the agent by also putting weight on other available performance measures (Feltham and Xie, 1994; Datar, Kulp and Lambert, 2001). As an example, consider the firms stock price, which is typically thought to be a highly congruent performance measure. Despite its high level of congruence, this measure typically does not receive all of the weight in a CEOs compensation contract. This may be because there is a role for other, less congruent performance measures to achieve better risk sharing or to better allocate the agents effort across multiple activities. In addition, as we discuss in Section 3.3., when the principal (i.e., the board acting on behalf of the firms shareholders) designs the contract with the objectives of all of the firms contracting parties in mind (e.g., the firms creditors), it can be
89

Feltham and Xie (1996) and Datar, Kulp and Lambert (2001) provide related definitions of the congruence of a performance measure as the angle between how it maps into the performance measure and firm value. 90 Accordingly single-period models such as Bushman and Indjejikian (1993a, b) typically do not permit the parties to contract over the terminal cash flow. In contrast to single period models, multiperiod models can allow the parties to contract over terminal value since there is still an agency problem that results from differential discount rates in a multiperiod model that is absent in a single period model. However, as we discuss further below, even when terminal cash flow is one of the performance measures available for contracting in a multiperiod model, it does not necessarily receive all of the weight in the optimal contract.

223

optimal to commit to deliberately sacrifice congruity with respect to their own objective in order to maximize the value of all of the firms expected contractual payoffs.91 Controllability and Relative Performance Evaluation Agency theory generally delivers clear predictions on the desired characteristics of variables that should be used as performance measures in executive compensation contracts. Specifically, the informativeness principle of Holmstrom (1979) implies that when there are multiple performance measures available for contracting, any (costlessly observed) variable that is correlated with the agents actions should be included in the contract as long as one of the other included variables is not a sufficient statistic for the candidate variable.92 In other words, any variable that is controllable by the agent and provides the principal with additional information about the agents actions is valuable for contracting. Lambert (2001, p.24) defines a variable as controllable if the agents actions influence the probability distribution of that variable. Although this principle describes the features of performance measures that should appear in the contract, it does not say how those variables should be aggregated and weighted, or how much value the variables have from a contracting efficiency perspective. We return to this issue in the next subsection. In addition to contracting over performance measures that are sensitive to the agents actions, it is often useful to include signals that are insensitive to the agents actions (i.e., uncontrollable by the agent). These signals (denoted 1 in Figure 1) are valuable from a
91

Datar, Kulp and Lambert (2001) also show that unlike the single action case, an increase in the sensitivity of a performance measure to an agents action does not necessarily increase the weight placed on the performance measure when the agent is responsible for taking multiple actions. In particular, they find that in some cases, the combination of an increased compensation weight and increased performance measure sensitivity will motivate the agent to allocate too much effort toward relatively unprofitable actions. 92 A sufficient statistic relates to the idea of a noisy transformation of a variable. More specifically, a statistic is said to be sufficient with respect to a parameter when no other statistic which can be calculated from the same sample provides any additional information as to the value of the parameter (Fisher, 1922). From an inferential perspective, a variable for which the principal already has a sufficient statistic provides no additional information about the action taken by the agent and therefore has no contracting value.

224

contracting perspective because they share a common source of noise (denoted 1 in Figure 1) with the contractual performance measures. These uncontrollable signals can therefore be used in the compensation contract to filter this common source of noise from the contractual performance measures. Reducing the noise in a performance measure delivers a more precise signal about the agents actions and thus exposes the agent to less risk which improves overall contracting efficiency. Compensation and Incentives As noted above, the principals basic problem is to choose the weights assigned to various performance measures and the functional form of the contract in order to induce the agent to take the desired actions. As noted by Core, Guay and Verrecchia (2003), the incentives provided by a performance measure are increasing in both the sensitivity of the performance measure to the agents action and the weight assigned to the performance measure in the contract (i.e., the product of the sensitivity and the pay-for-performance sensitivity which is the composite mapping from the agents actions to the contractual outcome). The agent observes the contract and then responds to the incentives in the contract by taking the action(s) that maximize his expected utility. After the agent takes his action(s) and the performance measures are observed, the compensation contract aggregates and weights the performance measures (which can be either objective or subjective) and maps the resulting values into various contractual outcomes such as the agents monetary remuneration, promotion or termination (i.e., the pay-for-performance sensitivity).93

93

Although the optimal form of aggregation and weighting is highly dependent on the assumed information structure of the problem, in certain special cases it can be shown that when signals are aggregated linearly, it is generally optimal for the relative weights assigned to the performance measures to be proportional to their signalto-noise ratio (e.g., Banker and Datar, 1989; Lambert, 2001; Datar, Kulp and Lambert, 2001).

225

It is important to note the distinction between the compensation and the incentives provided by the contract since the two concepts are closely related. In an agency theoretic setting, the notion of compensation relates to the net utility derived by an agent (including the cost to the agent, or disutility, associated with taking the equilibrium actions) during the contract period. Thus, compensation relates to the agents participation (or IR) constraint, which considers the amount of utility the agent must expect to receive to accept the principals contract.94 The notion of incentives relates to the mapping from the agents actions to various contractual outcomes (Core, Guay and Verrecchia, 2003) and, as the name suggests, relates more to the agents incentive compatibility (or IC) constraints. That is, incentives are less concerned with the magnitude of compensation, but rather how that compensation covaries with actions the agent might consider taking. For example, a bonus plan that pays $3 million for a good outcome and $2 million for a bad outcome provides more compensation, but the same incentives, as a bonus plan that pay $2 million for a good outcome and $1 million for a bad outcome (under certain assumptions about the agents wealth and risk aversion). A number of authors such as Antle and Smith (1986), Jensen and Murphy (1990) and Core, Guay and Larcker (2003) have noted that in practice, executives are provided incentives through three primary mechanisms: (i) flow compensation which consists of the executives annual salary and bonus payments, the expected value (conditional on the agents equilibrium actions and type) of new restricted stock and stock option grants and the value of other annual compensation such as the change in value of supplemental executive retirement plans (SERP) and long-term incentive plans (LTIP), (ii) changes in the value of the executives equity portfolio
94

It is important to note that the relevant measure of the agents utility is net of any cost associated with taking the desired actions (or, in the case of private information models, the costs associated with the agent revealing his private information). Indeed it is the wedge between the benefit to the agent (net of his personal cost) and the benefit to the principal (who does not incur the personal cost of effort) that gives rise to an agency problem in a moral hazard setting.

226

and (iii) changes in the value of the executives human capital (which includes changes in the probability of termination, promotion, and additional directorships). Valuation Since realizations of performance measures, such as financial accounting measures, provide information about the agents productive actions, these variables can also be used by financial markets to value both the firms equity securities and various claims against the firm (e.g., debt securities). Indeed, one of the primary functions of financial markets is the aggregation of information from publicly available performance measures (Grossman and Stiglitz, 1980). In addition, the value of the firm is affected by forces other than the agents actions (e.g., macroeconomic shocks, changes in the firms investment opportunity set, etc.) which are denoted 2. This is depicted in Figure 1 as the market receiving a noisy measure of terminal value.95 Finally, to the extent the compensation contract is observable, the market can, at least in part, anticipate the agents actions that will be induced by the contract in equilibrium and can use this conjecture for valuation purposes (e.g., if the contract consists of primarily outof-the-money options, the market can anticipate risky project selection). If, however, the parameters of the compensation contract are unobservable or the contract is based on private (and therefore unobservable) performance measures, the market can use realized contractual outcomes (e.g., realized bonus payments) to infer both the form of the contract and the realization of the unobservable performance measures. Stewardship versus Valuation Financial accounting information is typically used both as a performance measure for incentive contracting (or stewardship) purposes and by markets for valuation purposes. As
95

The noise term 2 is often given the theoretical interpretation as the random supply of the firms shares in a noisy rational expectation model. This modeling convention has the effect of precluding the market price from fully revealing investors information

227

noted by Gjesdal (1981), these two roles are often in conflict and the ranking of an information system for one purpose (e.g., contracting) need not coincide with its ranking for another purpose (e.g., valuation). As noted by Paul (1992), Lambert (2001) and others, when the contracting parties have symmetric information, it is the sensitivity/congruence of a performance measure that is relevant for contracting purposes, but it is the real shock to cash flows (i.e., 2) that is relevant for valuation purposes. This is because under the information structure of these models, all of the parties (including investors) have the same information set and the agents equilibrium actions can be perfectly anticipated. Therefore the realization of the contractual performance measures is useful only to the extent it is also sensitive to valuation relevant shocks since the market does not learn anything about the agents actions from the realizations of the performance. In contrast, 2 represents a real shock to the firms cash flows and is therefore valuation relevant. Adverse Selection In addition to the moral hazard setting, where contracting parties have symmetric information and the principal designs an incentive contract to induce the agent to take appropriate unobservable actions, it can also be the case that one or both of the contracting parties possesses private information before contracting or acquires private information after contracting. The most common case is where the agent has private information about his attributes (i.e., his type such as his degree of risk tolerance) prior to contracting. The solution is for the principal to design a menu of contracts from which the agent selects. The compensation contract therefore becomes a (screening) mechanism with which the agent reveals (at least in part) his private information to the principal (and possibly other parties such as debt and equity investors) through his choice of contract. The extent to which the agents

228

private information is revealed generally depends on the type of equilibrium achieved. A fully separating equilibrium implies the complete revelation of information, a partially separating equilibrium implies only partial revelation of information, and a pooling equilibrium implies no information revelation. The type of equilibrium, in turn, is function of the shape of the contract space and the similarity of the different types of agents. Specifically, contracts are designed by the principal to exploit differences among the agents in their participation constraints so that certain types of agents will find it profitable to select certain contracts but not others.96,97 In addition, a number of models consider the case of both adverse selection followed by moral hazard in which the agent acquires private information either before or after contracting with the principal but before taking productive actions. The difference between moral hazard and adverse selection models is not only of theoretical interest but also has empirical implications. In particular, if managers match with firms (and compensation contracts) on the basis of their private information (e.g., their level of risk-aversion or their talent in identifying profitable investment opportunities), this can result in a correlated omitted variable and/or endogeneity problem which may confound attempts to ascribe the causality of an outcome (e.g., firm performance) to the characteristics of the contract rather than the managers type. For example, if a positive relation is observed between the use of stock options (which provide risktaking incentives) and firm risk, it could be either that the convexity of the contract causes the
96

Note that the typical adverse selection setup involves the principal selecting from a pool of agents and the principal must contract with every type of agent. An alternative formulation is where the contract is designed to attract only a certain type of agent and dissuade (or exclude) all other types (i.e., the expected utility from the contract will not meet the other agents reservation utility given any action). Armstrong, Larcker and Su (2009) formulate and solve both modeling assumptions in a principal agent model with both moral hazard and adverse selection and show how the latter problem is more difficult for the principal (because of the additional constraints). 97 Milgrom (1987) and Hagerty and Segal (1988) demonstrate that in some settings, there is an equivalence between moral hazard and adverse selection. In particular, when the agent does not have limited liability in a model of adverse selection, the principal can design a menu of contracts such that there is only a single contract for each type of agent that will meet that agents reservation utility (captured by the agents IR constraint). In that case, the various IR constraints become equivalent to each agents IC constraint if the model were instead one of moral hazard. However when agents have limited liability (as is typically the case), the equivalence between the two models is not straightforward.

229

manager to undertake risky investments, or it could be that more risk-tolerant managers are attracted to riskier compensation contract (because they demand less of a risk-premium) and these managers are more inherently predisposed to undertake risky investments. Multiperiod Contracting Finally, although the preceding discussion focused on a single-period explicit contracting arrangement, the discussion and accompanying figure could be extended to encompass multiple periods in order to accommodate dynamic contracting issues. As we discuss in Section 3.3, this extension is crucial for understanding implicit contracts that arise in an incentive compensation contracting setting since these contracts represent the equilibrium of a repeated game between contracting parties. Examples of such contracting arrangements include dealings with executives that do not have formal employment contracts (which is, in fact, the majority of executives), implicit pressure on executives to hold significance positions of vested and saleable firm equity, and negotiations over changes in bonus and equity pay plans over time. Further, some important conceptual issues introduced in this setting include the degree and method of commitment on the part of the principal and agent, how previous realizations of performance measures are used by the contract in the current period (i.e., the degree of contract memory) and the use of accrual (and conservative) accounting numbers in contracts (since the reversing nature of accrual accounting is by construction a multi-period concept). As will be apparent in our discussion below, we believe that theoretical and empirical development of implicit contracting arrangements related to executive compensation and incentives has considerable potential for future research.

230

Figure 1

(Adverse) Selection / Screening

Uncontrollable signal 1

Filtering Aggregation and Weighting

Noise 1

Objective

Performance Measures

Subjective
Pay-for-performance Sensitivity

Compensation Contract(s)

Contractual Outcome

Sensitivity Valuation

Valuation

Managers Private Information

Managers Unobservable Actions(s)

Productivity

Market
= Congruence Valuation

Other Information

Noise 2

(Terminal) Firm Value

Uncontrollable Signal 2

Managers Acquire Private Information

Manager takes unobservable action(s)

Performance Measures Realized

Compensation Contract Aggregates and Weights Performance Measures

Contractual Outcome Determined

Firm and Manager contract (Manager possibly selects contract on the basis of his private information)

Noise and Uncontrollable Signals realized

Market Price Forms

231

Table 1 Characteristics of the Debt Contract that may be affected by Attributes of the Accounting System Attribute Interest Rate Maturity Security/collateral Performance pricing provision Performance pricing metric Concavity/Convexity of pricing grid Number of steps in the pricing grid Size of the loan Borrowing base restrictions Investment restrictions The rate charged by the lender. For bank loans this is typically stated as a spread over Libor, while for public debt the rate is typically fixed The period the loan is made available to the borrower A lien on the assets of the company A contracting feature that provides for changes in interest rates over the life of the debt contract based on measures of performance (typically accounting measures or debt ratings) The measure that the performance pricing provision is based on (e.g. Debt/EBITDA, Debt/Net Worth, S&P credit rating) The rate of change in the pricing metric to the rate of change in the interest rate. This ratio is not necessarily constant over the pricing grid A typical pricing grid can have anywhere from two to ten pricing levels The amount the lender borrows Restrictions on the amount that the firm can borrow. This is similar to a security/collateral requirement. Typically a borrowing base restriction limits the amount to be borrowed to be a percentage of receivables or inventory Restriction preventing firms from making capital expenditures Provision requiring the borrower to pay down debt if cash flows (EBITDA) are greater than pre-specified level, or if a sale exceeds a pre specified dollar amount, or if long term debt or equity is issued. Covenants that are based on accounting ratios Covenants restricting firms ability to pay dividends Net worth covenants where the covenant thresholds are increased by a percentage of positive net income Provisions included in the contract that limit the borrowers ability to make voluntary (or mandatory) accounting changes Modifications to GAAP included in the contract, like eliminating goodwill or intangible assets from the calculation of covenants.

Cash flow, asset sale, and debt/equity sweeps Financial covenants Dividend restrictions Income escalators

Fixed GAAP Tailored GAAP

232