Earnings Smoothing, Executive Compensation, and Corporate Governance: Evidence from the Property-Liability Insurance Industry

Beginning of article

ABSTRACT

Unlike studies that estimate managerial bias, we utilize a direct measure of managerial bias in the U.S. insurance industry to investigate the effects of executive compensation and corporate governance on firms' earnings management behaviors. We find managers receiving larger bonuses and stock awards tend to make reserving decisions that serve to decrease firm earnings. Moreover, we examine the monitoring effect of corporate board structures in mitigating managers' reserve manipulation practices. We find managers are more likely to manipulate reserves in the presence of particular board structures. Similar results are not found when we employ traditional estimated measures of managerial bias.

INTRODUCTION

Earnings management has long been a topic of interest to academic researchers, stakeholders, industry practitioners, and regulators. Despite the ample evidence of earnings manipulation and the importance of corporate governance in curtailing such manipulation, little research has examined the collective impact of executive compensation and board structure on earnings management. In this study, we extend previous literature by investigating the combined effect of executive compensation and board structure on firms' earnings management behavior. We jointly consider that (1) managers use discretionary accounting components to affect earnings for compensation enhancement and (2) firms structure boards to exert more control over managers, thereby potentially mitigating some of these effects.

Consistent with the prior literature utilizing estimated abnormal accruals, our results show a direct link between the incentive component of executive compensation and earnings management. Specifically, we observe that managers who derive larger proportions of their compensation from bonus payments and restricted stock are more likely to engage in earnings management. We further see that corporate governance through board monitoring plays an important role in curbing managers' manipulation of earnings, with different corporate governance structures being associated with varying degrees of manipulation by managers. We find these results using observed outcomes of managerial decision making (i.e., insurer loss reserve errors) instead of estimated measures of earnings management typically used in the accounting and finance literature. Moreover, our results are not found when we employ estimated measures of managerial bias. Our research contributes to the literature in several other important ways. First, this study augments the earnings management literature in general. It is well known that insurance companies are subject to much heavier regulation than almost all other industries. Hence, our study serves as a stronger test for earnings management than prior studies examining other industrial firms not subject to the same level of regulatory scrutiny as insurance companies.

Second, this research expands the current limited understanding of corporate governance in the insurance industry specifically. Despite the abundant evidence of loss reserve manipulation among property-liability insurers (Petroni, 1992; Beaver, McNichols, and Nelson, 2003; Gaver and Paterson, 2000, 2004) and the link between executive compensation and earnings management (Healy, 1985; Holthausen, Larcker, and Sloan, 1995; Eckles and Halek, 2010), no prior research in the insurance literature has investigated how insurers' loss reserve practices are affected by the joint impact of executive compensation incentives and board structure. Further, this is the first article that uses reserve errors to observe the initial impact of the Sarbanes-Oxley Act (SOX) on corporate governance in the insurance industry.

Third, our article significantly adds to the narrow, yet growing literature regarding the interaction among executive compensation, board structure, and earnings management. Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) are the only two studies we know of that examine the joint effect of corporate governance and executive compensation on earnings management. Yet, both Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) proxy earnings management by abnormal accruals, which are estimated indirectly using a regression model. Inferences drawn from their empirical evidence are thus limited, "due to the difficulty of measuring the level of accruals absent managerial bias, since accruals are the expected future cash receipts and payments resulting from all current and past transactions, and researchers cannot directly measure managerial expectations" (Petroni, 1992, pp. 485-486). We significantly advance Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) by utilizing a more accurate proxy of managerial bias, insurer loss reserve errors, which are disclosed in accounting results filed with state regulatory authorities and available through the

National Association of Insurance Commissioners' Database (the NAIC Database). Specifically, statutory reporting requires insurers to estimate losses and to ultimately report the observed, realized values of the estimated losses. Hence, managerial manipulation is directly captured by the differences between the estimated losses and the actual realized losses reported by management, both of which are disclosed in annually filed regulatory reports. (1)

Moreover, neither Cornett, Marcus, and Tehranian (2008) nor Cornett, McNutt, and Tehranian (2009) include insurance holding companies. Cornett, Marcus, and Tehranian (2008) study a sample of industrial firms in the Standard & Poor's (S&P) index from 1994 to 2003, whereas Cornett, McNutt, and Tehranian (2009) focus on the 100 largest bank holding companies in the United States from 1994 to 2002. According to Gillan, Hartzell, and Starks (2003), industry factors account for most of the explainable variation in overall governance structure and appear to dominate time effects and firm factors. Therefore, the results from Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) may not apply to the insurance industry. We contribute to understanding the relation between executive compensation, board structure, and earnings management by pioneering an investigation of the U.S. property-liability insurance industry.

Finally, by focusing solely on the U.S. property-liability insurance industry, we better control for the differential effects of regulation and political pressure, which allows us to assess more directly the influence of executive compensation and board structure on earnings management. The corporate governance literature is known for being fraught with difficult to overcome endogeneity problems. The likelihood that our results are due to the spurious correlation caused by unobserved heterogeneity is significantly reduced because our sample firms come from a single industry, and are thus more homogeneous (Blackwell, Brickley, and Weisbach, 1994; He and Sommer, 2010). Compared to Cornett, Marcus, and Tehranian (2008), our article is less prone to endogeneity, particularly the problem of omitted variables.

The remainder of the article is organized as follows. The "Literature Review" section examines the extant literature. The "Hypotheses Development and Model Framework" section develops the research hypotheses and the economic models. The "Data and Descriptive Statistics" section provides details on the sample data and analysis methodology. The "Regression Results" section presents the empirical results, and the final section concludes.

LITERATURE REVIEW

Prior research has documented the manipulation of insurance accounting results for various reasons, including avoiding regulatory scrutiny (Grace, 1990; Petroni, 1992), smoothing tax liabilities (Grace, 1990; Petroni, 1992; Petroni and Shackelford, 1999), and increasing the compensation of executives (Healy, 1985; Holthausen, Larcker, and Sloan, 1995; Eckles and Halek, 2010). (2) There also exists a literature investigating the oversight capacity that corporate governance mechanisms have in mitigating executives' manipulation of earnings (see Klein, 2002; Xie, Davidson, and DaDalt, 2003; Peasnell, Pope, and Young, 2005). Despite evidence of earnings manipulation and the impact of corporate governance in curtailing such manipulation, little research has jointly examined the influence of executive compensation and board structure on earnings management. To our knowledge, Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) are the only papers to investigate earnings manipulation with respect to both corporate governance oversight and managerial compensation incentives.

Our article examines the joint hypothesis that managers use discretionary accounting components to affect firm performance and that a firm's board structure can potentially mitigate some earnings management incentives. We provide a brief review of these two research areas.

The first strand of research focuses on managers' motives to manipulate earnings. Some incentives are operational decisions that benefit the business entity, whereas others directly benefit managers at the expense of business owners. Dechow and Skinner (2000) categorize capital market based incentives for earnings management into two streams: (1) incentives provided by stock market participants (e.g., analysts and money managers) for managers to meet relatively simple earnings benchmarks (see Burgstahler, 1997; Burgstahler and Dichev, 1997; Degeorge, Patel, and Zeckhauser, 1999; Myers, Myers, and Skinner, 2007) and (2) incentives for managers to improve the terms of equity offerings by engaging in earnings management at the time of seasoned equity offers (see Rangan, 1998; Teoh, Welch, and Wong, 1998a). While these incentives are more likely operational decisions, managers' incentives to manipulate earnings can also be self-serving. In particular, the growth of stock-based compensation has significantly affected managerial motives for earnings management, as "managers have become increasingly sensitive to the level of their firms' stock price and their relation to key accounting numbers such as earnings" (Dechow and Skinner, 2000, p. 237).

Despite widespread belief among practitioners and regulators that earnings management is pervasive and problematic, academic research has not convincingly proved this to be the case, as indicated by Dechow and Skinner (2000). McNichols (2000) suggests that "much of the controversy over interpretation of the literature's findings is due to the extensive use of aggregate accruals models to characterize discretionary behavior" (p. 314). She further indicates that "future contributions to the earnings management literature will come from papers that model the behavior of specific accruals with and without manipulation" (p. 338).

Studies in both the accounting and insurance literatures have considered the effects of managerial compensation packages on earnings manipulation. In the accounting literature, Healy (1985), Gaver, Gaver, and Austin (1995), and Holthausen, Larcker, and Sloan (1995) link the bonus component of executive compensation with earnings management through accruals. (3) In the insurance literature, Eckles and Halek (2010) find evidence that incentive-based components of executive compensation are positively associated with earnings manipulations. Similarly, Browne, Ma, and Wang (2009) show significant correlation between insurer reserve errors and the options granted to executives. However, neither study considers the role of board structure in mitigating the impact of the executive compensation packages on earnings management.

Studies in the corporate governance literature show that board structure affects the probability and/or magnitude of earnings management. For example, Xie, Davidson, and DaDalt (2003) and Peasnell, Pope, and Young (2005) show that the management of earnings through abnormal accruals is somewhat mitigated by more independent boards. But none of these studies consider the impact of executive compensation on earnings management, and more importantly whether such an impact is mitigated by the firms' board structure.

Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) are the only papers we know of that consider both managerial compensation and board structure within the context of accounting manipulation. Our article builds upon Cornett, Marcus, and Tehranian (2008) and Cornett, McNutt, and Tehranian (2009) in two important ways. First, our direct measure of managerial bias allows for a more accurate proxy for managerial discretion than estimated aggregate accruals models. (4) Second, we consider multiple detailed, incentive-based executive compensation components rather than one proxy for compensation structure. (5)

HYPOTHESES DEVELOPMENT AND MODEL FRAMEWORK

Healy (1985), Gaver, Gaver, and Austin (1995), and Holthausen, Larcker, and Sloan (1995) all indicate some level of manipulation of reported earnings by managers through the use of discretionary accounting practices. Loss reserve estimation by insurance executives is one such discretionary practice. Ceteris paribus, we postulate that insurance executives who derive a larger portion of their total compensation from incentive-based components will have larger reserve estimation errors in the direction that optimizes their compensation. Further, recent corporate governance literature documents that managers of firms with particular strong (weak) governance mechanisms are less (more) likely to manipulate earnings. (6)

In this section, we examine how various executive compensation components and several corporate governance characteristics relate to executives' discretionary accounting practices in insurance firms. The compensation components examined include bonuses, restricted stock held, stock options exercised, and stock options awarded. Prior research indicates that each component does not necessarily induce the manager to act in a consistent manner. We therefore investigate each compensation element and discuss whether it should induce earnings-decreasing behavior (e.g., restricted stock awards), earnings-increasing behavior (e.g., exercising options), and either earnings-decreasing or earnings-increasing behavior (e.g., structured bonus plans).

Executive Compensation and Reserve Errors

Several observable compensation items can be characterized as long-term incentive schemes. Awarding stock options and restricted stock is intended to align the long-term incentives of the executive and company. The "value" of these securities is not obtained during the year in which they are awarded but rather are ultimately realized in the future, contingent on the overall value of the firm. Stock options are commonly granted with an exercise price set equal to the price of the stock on the award date (Aboody and Kasznik, 2000). Additionally, restricted stock grants are made at the current stock price. Therefore, as Aboody and Kasznik (2000) argue, executives prefer lower current stock prices to reduce the strike price of the option and to potentially increase the number of shares awarded. (7) Hence, these compensation variables should provide motivation for managers to make earnings-decreasing decisions (i.e., relatively over-reserve for losses) in the current year. Over-reserving creates the temporary perception that the insurer has incurred larger losses, therefore reducing current stock prices. (8) This leads to our first testable hypothesis:

H1: Firms whose managers have larger proportions of stock options awarded and restricted stock awarded as compensation components relative to their total compensation in period t are more likely to make earnings-decreasing decisions in period t.

In addition to long-term compensation components, we observe incentive-based compensation elements that are realized or exercised during the current year, such as bonuses and exercised stock options. The earnings management incentives created by these compensation components are not as straightforward as some other components. Specifically, Healy (1985), Gaver, Gaver, and Austin (1995), and Holthausen, Larcker, and Sloan (1995) indicate managers may have incentives to make both earnings-increasing and earnings-decreasing decisions based on the structure of a bonus plan. The payoff schemes of the bonuses investigated by Healy (1985), Gaver, Gaver, and Austin (1995), and Holthausen, Larcker, and Sloan (1995) resemble a generic "call spread," with upper and lower thresholds setting bounds on the potential bonus. (9) The manager's incentives then were determined by where the firm s earnings were relative to the thresholds. (10)

We examine bonus plans similar to the plans studied by Holthausen,

Larcker, and Sloan (1995) and Healy (1985). (11) Unfortunately, our data do not provide sufficient detail …