Financial reporting comparability and accounting based relative performance evaluation in the design of CEO cash compensation contracts

THE ACCOUNTING REVIEW Vol 95, No May 2020 pp 343–370 American Accounting Association DOI: 10.2308/accr-52588 Financial Reporting Comparability and Accounting-Based Relative Performance Evaluation in the Design of CEO Cash Compensation Contracts Jonathan Nam The Hong Kong Polytechnic University ABSTRACT: This study examines how cross-firm differences in financial reporting practices affect how peer-firm accounting information is used to evaluate CEO performance I propose that efficient relative evaluation using accounting performance requires peer firms to have comparable financial reporting systems, allowing boards to reduce the information processing costs associated with differences in firms’ financial reporting practices Supporting this view, when peer selection takes financial reporting comparability into account, I find evidence that the earnings of peer firms with high financial reporting comparability serve as a performance benchmark for determining CEOs’ cash compensation My paper empirically corroborates the substantial anecdotal evidence of the use of peer firms’ accounting performance as a significant element in boards’ evaluation of CEO performance JEL Classifications: D8; G3; J33; M41 Keywords: financial reporting comparability; relative performance evaluation; CEO compensation I INTRODUCTION T his paper examines how cross-firm differences in financial reporting practices affect the use of peer-firm accounting information in the evaluation of CEO performance Theory predicts that an optimal incentive scheme would include peer-firm accounting performance as a means of distinguishing between managers’ contribution to accounting performance and the effect of exogenous shocks, that is, accounting-based relative performance evaluation (RPE) (e.g., Lazear and Rosen 1981; Holmstrom 1982; Nalebuff and Stiglitz 1983; Holmstrom and Milgrom 1987).1 Despite its theoretical appeal and substantial anecdotal evidence, there is surprisingly little large-sample evidence in the U.S that supports the use of peerfirm earnings, as opposed to peer-firm stock returns, as a performance benchmark in CEO compensation contracts (e.g., Gibbons and Murphy 1990; Albuquerque 2009) Because a firm’s own accounting earnings are widely used as a performance measure in evaluating its CEO performance (e.g., Bushman and Smith 2001), it is puzzling that CEO compensation contracts appear to rely exclusively on peer firms’ stock returns while disregarding peer firms’ earnings To shed light on this puzzle, I examine two related issues First, I investigate whether, when designing RPE contracts, boards take into account financial reporting comparability (i.e., the degree to which a firm’s financial reporting is comparable to I am very grateful for the valuable feedback from Rodrigo S Verdi (editor) and two anonymous reviewers I am deeply indebted to my advisor, Pervin Shroff, for her invaluable guidance and support, and my dissertation committee members, Ivy Zhang, Aiyesha Dey, and Gerard McCullough I also appreciate the helpful comments from Gus De Franco, Zhaoyang Gu, Agnes Cheng, Matthew Pinnuck, Jeffrey Ng, Nancy Su, Allen Huang, Inder Khurana, Kam-Ming Wan, Chao Tang, Eunhee Kim, and other seminar participants at the University of Minnesota, The Hong Kong Polytechnic University, The Chinese University of Hong Kong, IESE Business School, IE Business School, HEC Paris, and the 2016 American Accounting Association Annual Meeting This paper is based on my dissertation, completed at the University of Minnesota Previously, this paper was titled ‘‘Financial Reporting Comparability and Relative Performance Evaluation.’’ Editor’s note: Accepted by Rodrigo S Verdi, under the Senior Editorship of Mark L DeFond Submitted: August 2016 Accepted: July 2019 Published Online: October 2019 For example, Pacific Gas and Electric (PG&E), an energy provider, determines its CEO’s bonus based on the firm’s performance relative to that of preselected competitors like Consolidated Edison and Xcel Energy In other words, the more PG&E outperforms its peer firms, the higher the bonus its CEO receives 343 344 Nam that of peer firms) Second, I study the ways in which empiricists could employ financial reporting comparability to improve their ability of documenting the use of accounting-based RPE in CEO compensation contracts The first part of the paper is based on the notion that financial numbers from a peer firm with a comparable accounting system are useful in evaluating CEO performance Two firms have comparable accounting systems (i.e., high financial reporting comparability) if, for a given set of economic events, they produce similar financial statements Therefore, comparable accounting systems facilitate the financial statement users’ ability to identify similarities and differences in firms’ economic performance (De Franco, Kothari, and Verdi 2011) That is, the time and effort needed to assess a certain firm’s earnings relative to those of a peer decrease in the degree to which the two firms exhibit comparable financial reporting systems This assertion suggests that boards of directors must address substantial cross-sectional variation in the information processing costs inherent to intensive comparison of accounting numbers across firms I, thus, propose that boards select peers with higher financial statement comparability when designing accounting-based RPE, which, in turn, facilitates the assessment of own-firm, relative to peer-firm, earnings To investigate whether boards’ RPE peer selection considers financial reporting comparability, I identify the overlap between a firm’s peer group selected on the basis of firm characteristics and the firm’s actual peer group as disclosed in its proxy statement.2 Using the method proposed by De Franco et al (2011) (hereafter, DKV), I measure the extent to which, given similar economic events, two firms’ accounting systems produce similar accounting earnings The compensation contracting practices of U.S firms show that financial reporting comparability (FRC) is a significant determinant of boards’ peer selection When I consider FRC in peer selection, along with industry and size, the overlap increases to 35.04 percent relative to when peer firms are randomly selected from the same industry (10.15 percent) or when they are similar-sized firms from the same industry (28.34 percent) Furthermore, after controlling for various dimensions of firm-peer economic similarities (e.g., industry, size, market-to-book [MTB], firm complexity), I find that financial reporting comparability is a significant determinant of boards’ peer selection This result is consistent with my prediction that peer firms with high financial reporting comparability are more likely to be included as performance benchmarks in CEO compensation contracts The second part of the paper focuses on the empirical implication of boards’ peer selection practices Specifically, I examine how researchers could enhance their ability to detect the use of accounting-based RPE in CEO compensation contracts The empirical evidence on RPE usage varies significantly depending on a researcher’s choice of peer selection (e.g., Albuquerque 2009; Dikolli, Hofmann, and Pfeiffer 2013) Prior compensation studies commonly use industry membership and/ or firm size as peer selection criteria in tests of RPE (e.g., Joh 1999; Albuquerque 2009) While industry membership and/or firm size help to identify peer firms that are subject to common shocks, high financial reporting comparability is an essential additional screen to identify peer firms that account for common shocks in a similar way, making peer-firm accounting earnings effective metrics for relative performance assessment Given the significance of financial reporting comparability in the design of RPE, empirical tests that disregard this factor create a discrepancy in peer selection between boards and empiricists Furthermore, disregarding financial reporting comparability introduces a bias against detecting the use of RPE by regressionbased tests (Dikolli et al 2013) I, thus, conjecture that the paucity of evidence in support of accounting-based RPE can be traced to researchers’ reliance on conventional peer definitions that overlook boards’ likely consideration of the information processing costs associated with differences in firms’ financial reporting practices To investigate the above conjecture, I introduce an innovation in the identification of RPE peer firms in conducting regression-based tests Specifically, to test for the use of accounting-based RPE, I refine peer selection by considering financial reporting comparability, in addition to conventional firm characteristics For each treatment firm, I select the ten same-industry peer firms that are closest in size (market capitalization) and financial reporting comparability Using these selected peers, I find empirical evidence of accounting-based RPE in CEO compensation for a large sample of U.S firms.3 In contrast, when peer firms are selected solely based on the conventional matching on industry and/or size, I find no support for the use of accounting-based RPE, consistent with prior studies.4 I also show that the specific form of RPE (i.e., accounting- or pricebased) is affected by how CEO compensation is structured My results suggest that accounting-based RPE is used in cash compensation, whereas it has an insignificant effect on equity compensation, where price-based RPE plays a significant role This asymmetric use of accounting-based RPE across compensation components also mirrors prior findings that accounting performance measures are mainly used to determine cash compensation (Bushman and Smith 2001) Consistent with Gong, Li, and Shin (2011), I find that after the 2006 Securities and Exchange Commission (SEC) rule change, about 14 percent of the firm-year observations in the Incentive Lab database report self-selected RPE peers My results are robust to using an alternative measure of financial reporting comparability based on the Dechow and Dichev (2002) accruals-generating process, rather than the earnings-returns relationship that DKV employ Albuquerque (2009, Table 1) summarizes the empirical tests for RPE in prior studies The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 345 Prior studies test whether boards of directors identify poor-performing CEOs through RPE While there is empirical evidence of price-based RPE in CEO termination decisions in the U.S (e.g., J Barro and R Barro 1990; Gibbons and Murphy 1990), the results for accounting-based RPE are inconclusive.5 I find that accounting-based RPE is significantly related to CEO turnover when peer firms with high financial reporting comparability are considered as RPE benchmarks However, when using industry-size peer firms, I find no evidence of accounting-based RPE in CEO turnover decisions This finding further highlights the role of financial reporting comparability in peer selection for accounting-based RPE My study contributes to the literature on how boards choose peers for CEO performance evaluation by introducing financial reporting comparability as an important determinant While boards of directors make a wide range of choices in implementing RPE, we know relatively little about their peer selection process, which is mostly unobservable By examining firms’ proxy statement disclosures, I provide evidence that U.S boards take into account financial reporting comparability, in addition to economic fundamentals (e.g., size, industry), when selecting RPE peers To the extent that comparable accounting systems increase the quantity and quality of information available (DKV), RPE peers with high financial reporting comparability likely help mitigate the information processing costs associated with integrating multi-firm accounting numbers, thereby facilitating the use of accounting-based RPE in contracting decisions Furthermore, my results offer a methodology by which empirical researchers can improve their ability to document the use of accounting-based RPE Given the significant anecdotal evidence of accounting-based RPE (Gong et al 2011), the lack of prior empirical support for accounting-based RPE in large-sample U.S studies (e.g., Albuquerque 2009; Antle and Smith 1986; Vrettos 2013; Ozkan, Singer, and You 2012) is surprising I conjecture that it can be attributed to empiricists’ choice of RPE peer groups that are not matched on financial reporting comparability Including financial reporting comparability as an additional peer selection criterion in regression-based RPE tests, I am able to present evidence suggesting the significant use of accounting-based RPE in the evaluation of CEO performance Last, my study contributes to the literature on CEO compensation structure Many prior studies of RPE in the U.S focus on the CEO’s total compensation and not examine the effect of accounting- versus price-based RPE on different components of CEO compensation My results show that accounting-based RPE is used in cash compensation, but not in equity compensation, consistent with the notion that a cash bonus rewards realized performance captured by accounting earnings The remainder of the study is organized as follows Section II discusses the literature and develops the hypothesis Section III examines boards’ peer selection and financial reporting comparability Sections IV and V empirically test for the use of accounting-based RPE Section VI concludes II LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT Accounting Information in CEO Performance Evaluation Contracting Value of Own-Firm Accounting Information Anecdotal evidence and compensation surveys strongly support the role of accounting data in firm performance evaluation and managerial compensation Accounting performance measures can provide better managerial incentives because (1) they are based on audited numbers and are, therefore, reliable, and (2) relative to market-based or nonfinancial measures, accounting performance measures give managers a better picture of the impact created by their specific actions (Indjejikian 1999) However, prior research raises a concern that CEO evaluation based on accounting performance measures could encourage managerial actions that sacrifice long-term profitability for short-term gains (e.g., Dechow and Sloan 1991) Nevertheless, accounting performance measures continue to play a role in compensation contracts (Baber, Kang, and Kumar 1998) Both theory and empirical evidence suggest that accounting performance is useful in CEO evaluation In particular, the ‘‘informativeness’’ principle of Holmstrom (1979) suggests that a certain performance measure is useful in contracting if observing it provides incremental information about the agents’ actions and decisions Consistent with this view, accounting performance measures are useful in contracting because they provide information that helps identify the manager’s actions and decisions Prior empirical studies (e.g., Lambert and Larcker 1987; Jensen and Murphy 1990; Murphy 1999; Leone, Wu, and Zimmerman 2006; Gong et al 2011) report a significant presence of accounting metrics in CEO compensation contracts, and show that the contracting value of own-firm earnings is incremental to that of own-firm stock returns, particularly in determining the CEO’s salary and cash bonus.6 DeFond and Park (1999) show that a firm’s industry-adjusted accounting performance is inversely related to CEO turnover Bushman and Smith (2001) question whether this result can be viewed as evidence for RPE, given that the effect of peer performance is not separately tested Sloan (1993) proposes that own-firm accounting measures are used in CEO compensation contracts mainly to filter out the noise in stock returns Bushman and Smith (2001) observe that the documented positive relation between earnings and CEO compensation is inconsistent with Sloan’s (1993) argument The Accounting Review Volume 95, Number 3, 2020 Nam 346 Contracting Value of Peer-Firm Accounting Information Holmstrom’s (1979) informativeness principle implies that a performance measure is useful if it results in managers being compensated for the outcomes of their own actions and decisions That is, compensation contracts should be designed to filter out common shocks to firm performance so that managers are not rewarded or penalized for factors outside their control Consistent with agency theory, the managerial accounting literature emphasizes the principle of ‘‘controllability’’—that managers should be able to control or influence the performance measure used for their evaluation (e.g., Antle and Demski 1988; Lambert 2001) Relative performance evaluation using peer firms’ performance meets this standard, as it offers a means of filtering out common shocks from the firm’s performance measures Because a firm’s accounting numbers could be more informative when they are integrated with those of its peers rather than when considered in isolation (e.g., De Franco et al 2011; Chen, Collins, Kravet, and Mergenthaler 2018; Cheng, Jo, Ng, and Wu 2017), compensation committees may find it useful to examine peer firms’ accounting performance as a means of isolating the CEO’s contribution to own-firm accounting performance Consistent with agency theory (e.g., Holmstrom 1982; Holmstrom and Milgrom 1987), evaluating own-firm accounting performance relative to peer-firm accounting performance can incentivize CEOs while insulating them from common uncertainty, thus resulting in more efficient compensation contracts Peer-firm accounting performance is likely to be a more efficient benchmark for filtering out exogenous shocks from own-firm accounting performance, as the benchmark measures the same attribute, i.e., accounting earnings In contrast, the use of peer firms’ price performance is less likely to successfully parse out the effect of factors beyond the CEO’s control from the firm’s accounting earnings, given that stock returns are affected by many factors, in addition to earnings Nevertheless, in contrast to the extensive empirical support for price-based RPE, there is limited evidence for the use of accounting-based RPE in the U.S (e.g., Gibbons and Murphy 1990; Janakiraman, Lambert, and Larcker 1992; Albuquerque 2009) The limited evidence in the U.S also contrasts with the significant amount of non-U.S evidence that accounting-based RPE is used to evaluate CEO performance (e.g., Ozkan et al 2012; Wu and Zhang 2010; Chen, Liang, and Zhu 2012).7 Hypothesis Development: Financial Reporting Comparability and RPE In reviewing the financial statements of multiple firms, more time and effort are needed to assess relative performance when the firms’ financial reporting is not easily comparable Supporting this view, the Financial Accounting Standards Board’s (FASB 2010) conceptual framework states that ‘‘Users’ decisions involve choosing between alternatives, for example, selling or holding an investment, or investing in one reporting entity or another Consequently, information about a reporting entity is more useful if it can be compared with similar information about other entities.’’ Even for firms under the same accounting standards regime, the degree to which one firm’s financial reporting is comparable to its peers will differ (DKV) Therefore, the information processing costs associated with integrating multi-firm accounting information are inversely related to financial reporting comparability Several studies document the benefits of highly comparable financial reporting in evaluating external entities, such as analyst activities (DKV), mergers and acquisitions (M&As) (Chen et al 2018), equity investors (Kim, Li, Lu, and Yu 2016), and hedge funds (Cheng et al 2017) Because accounting-based RPE requires a periodic review of multi-firm accounting numbers, I conjecture that the benefits of financial reporting comparability also apply to the use of accounting information for CEO compensation contracts That is, when accounting systems are not easily comparable, it is costly for boards to distinguish between managers’ contribution to firm performance and exogenous effects, which limits the contracting advantage of RPE In line with this view, the enhanced comparability that results from the mandatory adoption of International Financial Reporting Standards (IFRS) has given rise to the use of accounting-based RPE in continental European countries (Ozkan et al 2012; Wu and Zhang 2010).8 I, thus, expect that U.S boards of directors are more likely to consider accounting-based RPE that uses peer firms with high financial reporting comparability as a low-cost and efficient means of filtering out exogenous shocks from the firm’s accounting performance Hence, the hypothesis, stated in the alternative form, is: Ha: Boards use peer firms with high financial reporting comparability in implementing accounting-based RPE in CEO compensation contracts Note that I focus only on CEO compensation contracts There is prior evidence that division-level accounting-based RPE is used for the internal evaluation of managers (e.g., Blackwell, Brickley, and Weisback 1994; Matsumura and Shin 2006; Cichello, Fee, Hadlock, and Sonti 2009) Although these results could be seen as a consequence of enhanced comparability, the adoption of IFRS has simultaneously led to changes in other aspects of the adopting firms’ financial reporting Thus, it is difficult to pinpoint the incremental effect of financial reporting comparability on the contracting value of accounting numbers Moreover, as Cascino and Gassen (2015) show that the effect of mandatory IFRS adoption on comparability is marginal, the post-adoption increase in RPE usage might be due to other factors The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 347 The above hypothesis is a joint hypothesis that has two key elements First, when designing RPE contracts, boards select peers with high financial reporting comparability (Section III) Second, empiricists are able to document the use of accounting-based RPE when their peer selection process takes into account financial reporting comparability (Sections IV and V) In the subsequent sections, I present evidence for both elements III BOARDS’ PEER SELECTION AND FINANCIAL REPORTING COMPARABILITY Measuring Financial Reporting Comparability To empirically measure financial reporting comparability, I follow DKV’s approach This measure captures how closely the same economic events (as reflected by stock returns) of two firms are mapped into their accounting earnings.9 For each firm-year observation from Execucomp, I first estimate the following equation using data from the previous 16 quarters (12 quarters minimum): Earningsit ẳ ỵ bi Returnit ỵ eit ; 1ị where Earnings equals quarterly income before extraordinary items scaled by lagged total assets; and Return equals the contemporaneous quarterly stock returns The coefficient estimates from Equation (1), a^i and b^i , are firm-specific accounting system parameters that map economic events of firm i into reported earnings numbers of firm i Similarly, I estimate a^j and b^j as the parameters of the accounting system of peer-firm j.10 I calculate the predicted earnings of firm i by applying the parameter estimates of the accounting systems of firms i and j, respectively, to firm is stock returns as follows: EEarningsịiit ẳ a^i ỵ b^i Returnit ; 2ị EEarningsịijt ẳ a^j ỵ b^j à Returnit : ð3Þ Equation (2) estimates the predicted earnings of firm i based on the accounting system parameters of firm i, whereas Equation (3) estimates the predicted earnings of firm i based on the accounting system parameters of firm j Financial reporting comparability between firms i and j, FRCij, is defined as the absolute difference between the expected earnings estimated from Equations (2) and (3), averaged over the preceding 16 quarters (and multiplied by À1): Xt jEðEarningsÞiit À EðEarningsÞijt j: 4ị FRCijt ẳ 1ị 1=16ị t15ị Therefore, a greater value of FRC indicates higher accounting comparability between firm i and firm j Proxy Disclosure of Peer Selection and Financial Reporting Comparability My main hypothesis jointly tests whether a CEO compensation contract is written on peer-firm accounting performance and whether firms use financial reporting comparability when selecting RPE peers In this section, I first investigate the extent to which financial reporting comparability (FRC) explains the actual choice of RPE peers observed in the Compensation Discussion and Analysis (CD&A) section of proxy statements Before testing for accounting-based RPE, it is important to verify whether using financial reporting comparability (as measured by DKV) as a selection criterion enhances a researcher’s ability to select a peer firm that the firm of interest is actually likely to choose The SEC’s executive compensation disclosure rules, effective December 15, 2006, require firms to disclose detailed information that includes the performance targets (e.g., RPE peers) used in setting executive pay.11 Incentive Lab provides detailed data from proxy statements (Form DEF 14A) on various aspects of executive compensation contracts for the largest 750 firms in a given year, resulting in 11,920 firm-years for the fiscal years 1998 to 2013 If firms implement RPE solely using indexes (e.g., Standard & Poor’s [S&P] 500, industry) or if they mention the use of RPE without providing peer firms’ details, 10 11 The DKV measure is an output-based measure of financial reporting comparability In contrast to some prior studies that use measures comparing a broad set of specific accounting choices (e.g., Bradshaw, Miller, and Serafeim 2009), the DKV method does not require data on accounting choices and does not rely on assumptions to develop a composite empirical proxy of comparability Furthermore, as Daske, Hail, Leuz, and Verdi (2013) note, an additional weakness of input-based measures of comparability is that similar accounting inputs may not necessarily result in comparable outputs in the absence of effective incentive alignment and/or enforcement mechanisms Consistent with De Franco et al (2011), I classify firms into three reporting-interval groups based on the month when the fiscal year ends: group (March, June, September, or December); group (April, July, October, or January); and group (May, August, November, or February) To match the timing of estimating the accounting system parameters, both firms i and j are required to belong to the same reporting-interval group See SEC Final Rule 33-8732a (available at: https://www.sec.gov/rules/final/2006/33-8732a.pdf ) The Accounting Review Volume 95, Number 3, 2020 Nam 348 then I classify them as non-RPE disclosers After retaining observations for which financial reporting comparability (FRC) is measurable, my sample includes 1,436 firm-year observations (12 percent) of explicit RPE users with self-selected peer groups Explicit versus Implicit Peers For each RPE peer group retrieved from Incentive Lab, I select peer groups based on alternative criteria in which each group consists of the same number of peer firms as disclosed by the firm Table presents the extent to which peer selection overlaps with each of the alternative peer groups (i.e., implicit RPE peers) and the corresponding actual peer group disclosed in the firm’s proxy statement (i.e., explicit RPE peers) The mean (median) number of peers in the disclosed peer groups is 13.55 (11).12 When peers are randomly selected from Compustat, the overlap is as low as 0.34 percent, on average The selection efficiency increases when economic characteristics are included as peer selection criteria The overlap increases to 10.15 percent when peers are randomly selected from the same industry, and it further rises to 28.34 percent when similar-sized peers are selected from the same industry Matching same-industry peer firms on financial reporting comparability (FRC) also enhances a researcher’s peer selection ability The peer group consisting of high FRC peers from the same industry exhibits an overlap of 16.07 percent When firm-peer size closeness and financial reporting comparability (FRC) are jointly considered in peer selection, the overlap increases to 35.40 percent The percentage of firms identifying at least one peer that appears in both the implicit and explicit RPE peer groups exhibits a similar pattern By selecting same-industry peers randomly, 53.01 percent of firm-year observations match at least one peer firm appearing in the firms’ self-selected peer groups When I additionally consider size (and also financial reporting comparability) in peer selection, the matching rate goes up to 81.81 percent (87.24 percent), a substantial proportion of the firm-year observations (untabulated) Selected versus Unselected Peers To further examine the role of FRC in actual peer selection, I estimate the following logistic regression:   XK À Á  à Controls Prob Selectedijt ¼ ¼ U q0 ỵ q1 FRCijt ỵ ijkt ; kẳ1 k ð5Þ where Selectedijt is a dummy variable that equals if firm j is selected as a member of the RPE peer group by explicit discloser i in year t, and otherwise Because the sample of RPE peers selected by firms is highly unbalanced compared to the Compustat universe, the estimated coefficients from logistic regressions may be biased and inefficient To alleviate this concern, I follow a procedure similar to Gong et al (2011) to form a control group of potential peer candidates that are not chosen by a certain RPE discloser Out of the population of Compustat firms that are not chosen by the explicit RPE discloser, I randomly choose the same number of control firms as the number of explicit peers Panel A of Table reports the univariate statistics of various characteristics of the selected RPE peers relative to the firm of interest (Columns (1) and (2)) and the characteristics of the control (i.e., unselected) firms relative to the firm of interest (Columns (3) and (4)) Relative to the control firms, the selected peer firms are more likely to be in the S&P 500 index, in the same industry, and similar in size and profitability to the firm of interest Important to my prediction, I also find that relative to the control firms, the selected RPE peers have financial reporting systems that are more comparable with that of the firm of interest One may question the effectiveness of the DKV measure in capturing how boards view similarities and differences in multi-firm accounting systems For example, rather than relying on the mapping between earnings and returns (Equations (1)– (4)), a board may examine intuitively whether two firms choose the same accounting methods To address such concerns, I investigate the homogeneity in accounting methods, including depreciation, inventory valuation, and leases, between an RPEdisclosing firm and its peer firm In Table 2, Panel B, the percentage of peers choosing the same accounting input method chosen by their own RPE discloser is greater for the selected peer group than it is for the unselected peer group Moreover, FRC is also greater for the selected peers for all three accounting methods That is, a researcher can use DKV’s output-based comparability measure as an aggregate proxy for the differences and similarities in accounting method choices, overcoming empirical challenges arising from an input-based comparability measure, such as which accounting choices to use in boards’ peer selection, how to weight them, how to account for variation in their implementation, etc Focusing on accounting outputs (i.e., earnings), the DKV methodology abstracts from those challenges and avoids the difficulty of collecting data on a broad set of accounting choices for a large sample of firms (De Franco et al 2011) Finally, Panel C of Table reports the results of the logistic regression investigating the determinants of actual RPE peer selection (Equation 5) using 37,820 firm-year-peer observations (18,910 disclosed explicit peers and 18,910 matched control firms) I find that FRC is a significant determinant of actual RPE peer selection after controlling for own- and peer-firm 12 With no data restriction for measuring FRC as in Equation (4), prior studies find an average of roughly 15 peers from proxy disclosures (e.g., Gong et al 2011; Black, Dikolli, and Hofmann 2015) The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 349 TABLE Explicit versus Implicit RPE Peers Mean # of reported peers in Incentive Lab 13.55 # of matched peers between Incentive Lab and sampling based on alternative criteria (% of matched peers) Random selection from Compustat 0.08 (0.34%) Random selection from the same industry 1.50 (10.15%) Peers of similar size from the same industry 3.71 (28.34%) Peers with high financial reporting comparability from the same industry 2.30 (16.07%) Peers of similar size and with high financial reporting comparability from the 4.79 same industry (35.40%) Q1 (0.00%) (0.00%) (9.09%) (0.00%) (13.79%) Median 11 (0.00%) (4.65%) (25.00%) (11.76%) (33.33%) Q3 17 (0.00%) (16.67%) (43.50%) (25.00%) (53.33%) This table shows how similar a group of peers selected by a researcher (the implicit peer group) is to the peers that are disclosed in firms’ proxy statements (the explicit peer group) by examining the overlap between the two groups Incentive Lab collects the relative performance evaluation (RPE) peers that are reported in proxy statements For a given firm-year, I select the same number of peer firms as reported in Incentive Lab under five alternative sampling processes: (1) pure random selection from the Compustat Universe; (2) random selection from the same two-digit SIC group; (3) the most similar-sized peers from the same two-digit SIC group; (4) the most comparable peers with respect to financial reporting, measured as in De Franco et al (2011), from the same two-digit SIC group; and (5) the peers that are the most similar-sized and the most comparable with respect to financial reporting from the same two-digit SIC group For FY 1998–2013, there are 1,160 Incentive Lab firm-year (447 unique firms) observations for which all five groups of implicit peers are available fundamentals To alleviate concerns that FRC proxies for other firm characteristics unrelated to accounting systems, I further control for the number of business segments and the number of geographical segments (Albuquerque, De Franco, and Verdi 2013).13 In the presence of input-based comparability measures (Selected), FRC’s explanatory power for disclosed RPE peer selection is not attenuated, but including FRC does attenuate the explanatory power of the input-based comparability measures (Column (6) versus Columns (3)–(5)) Collectively, the results in Tables and reveal that FRC is a significant determinant of boards’ RPE peer choice and validate the use of FRC as a peer selection criterion for regression-based RPE tests.14 IV EMPIRICAL DESIGN: THE USE OF ACCOUNTING-BASED RPE The Effect of the Discrepancy in Peer Selection between Boards and Empiricists RPE in CEO compensation contracts can be empirically observed only if two factors are met: (1) compensation committees indeed use peer firms as benchmarks, and (2) the researcher’s definition of RPE peer firms matches the compensation committee’s peer-firm selection criteria (Gibbons and Murphy 1990; Baker 2002) Because a board’s actual peer selection process is mostly unobservable, in tests of empirical support for RPE usage, empiricists need to detail their own assumptions regarding peer firms’ characteristics Dikolli et al (2013) show that the resulting discrepancy in peer selection by an empirical researcher translates into a bias against detecting RPE practice in regressions (i.e., Type II inferential errors) If compensation 13 14 Moreover, after controlling for the historical correlation between two firms with respect to their earnings, operating cash flows, and returns that DKV use as alternative measures of accounting comparability, the coefficient estimate on financial reporting comparability (FRC) remains positive at the percent significance level (untabulated) Only 14 percent (2 percent) of firm-year observations from Incentive Lab disclose their selection of RPE peers after (before) the 2006 SEC rule change The low number of RPE-peer disclosers, in contrast to the empirical evidence of the widespread use of RPE (e.g., Albuquerque 2009), does not necessarily mean that RPE practice is, in fact, rare because non-disclosers may apply RPE implicitly (e.g., at the board’s discretion or via subjective evaluation) rather than pre-committing to a formulaic explicit RPE (e.g., Gong et al 2011; Ferri 2009; Matsumura and Shin 2013) Moreover, the SEC acknowledges firms’ concern about potential competitive harm from disclosing performance targets (for example, see the SEC comment letter to Blackbaud Inc on October 28, 2008 at: https://www.sec.gov/Archives/edgar/data/1280058/000000000008053799/filename1.pdf ) Consistent with this conjecture, Black et al (2015) show that the use of price-based RPE is significant in both the non-disclosers and the disclosers samples Interestingly, I find that the proportion of RPE peer disclosers in Incentive Lab increases to 20 percent in 2013 from 11 percent in 2007, an increase that coincides with the growing attention to CEO compensation, which may have resulted in more firms adopting formulaic RPE Further, given the evidence of a lack of self-serving behavior in peer selection (e.g., Albuquerque et al 2013; Cadman and Carter 2014), it is rather unlikely that the extent of non-disclosure of RPE peers in Incentive Lab is mainly due to strategic managerial decisions The Accounting Review Volume 95, Number 3, 2020 Nam 350 TABLE Proxy Statement Disclosure of RPE Peer Selection Panel A: Selected versus Unselected Peers Selected (n ¼ 18,910 Firm-Year-Peers) Same_SIC2 Same_SIC4 S&P500 FRC SALES_Gap MVE_Gap RET_Gap ROA_Gap MTB_Gap Multi_Bus_Seg Single_Bus_Seg Multi_Geo_Seg Single_Geo_Seg Unselected (n ¼ 18,910 Firm-Year-Peers) Mean (1) Median (2) Mean (3) Median (4) 0.708*** 0.377*** 0.560*** À0.011*** 1.384*** 1.637 0.242 0.048*** 0.496*** 0.414*** 0.247 0.434*** 0.376*** 1.000*** 0.000 1.000*** À0.006*** 0.580*** 0.643*** 0.168*** 0.023*** 0.242*** 0.000*** 0.000 0.000*** 0.000*** 0.057 0.011 0.133 À0.030 1.748 1.728 0.420 0.132 0.956 0.240 0.241 0.252 0.243 0.000 0.000 0.000 À0.015 0.959 0.957 0.295 0.058 0.509 0.000 0.000 0.000 0.000 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively, for the two-sample mean (median) test of the difference between the samples of selected and unselected peers Incentive Lab provides 18,910 firm-year-peer observations for which financial reporting comparability (FRC) is available Selected firms are actual RPE peer firms observed in the proxy statements of RPE disclosers Using under-sampling (e.g., Gong et al 2011), the same number of firm-year unselected peers are randomly chosen from the unchosen domestic Compustat firms This panel provides the characteristics of selected and unselected peers as compared to their matched RPE discloser firm Variable Definitions: Same_SIC2 (SIC4) ¼ a dummy that equals if an RPE discloser and a peer are in the same two-digit (four-digit) SIC group; S&P500 ¼ a dummy that equals if a peer belongs to the S&P500; FRC ¼ the financial reporting comparability between an RPE discloser and a peer firm, using the method proposed by De Franco et al (2011); SALES_Gap (MVE_Gap) ¼ the absolute difference in sales (market capital) between an RPE discloser and a peer firm, divided by the sales (market capital) of the RPE discloser; ROA_Gap (RET_Gap) ¼ the absolute difference of ROA (annual stock returns with no scaling) between an RPE discloser and a peer firm; MTB_Gap ¼ the absolute difference in the market-to-book ratio between an RPE discloser and a peer firm; Multi_Bus_Seg (Single_Bus_Seg) ¼ a dummy that equals if both an RPE discloser and a peer have multiple (single) business segments; and Multi_Geo_Seg (Single_Geo_Seg) ¼ a dummy that equals if both an RPE discloser and a peer have multiple (single) geographic segments Panel B: Accounting Method Choice and Observed RPE Peers Accounting Method Selected Unselected Depreciation n (firm-year-peer) Using the same method FRC 12,085 78.2% À0.014 10,771 65.3% À0.035 Inventory Valuation n (firm-year-peer) Using the same method FRC 10,391 41.8% À0.015 7,044 27.3% À0.030 Lease n (firm-year-peer) Using the same method FRC 13,773 57.6% À0.013 13,059 43.50% À0.033 This panel shows the accounting method choices of selected and unselected peers as compared to their matched RPE discloser firm For the same sample as in Panel A, I examine whether an RPE discloser and its peer use the same accounting method in depreciation, inventory valuation, and lease Compustat provides a firm’s choice of depreciation method (DPACT_FN) and inventory valuation method (INVVAL) Two firms are regarded as the users of the same accounting choice in lease if both have (1) non-zero capital lease leverage and operating lease leverage, (2) non-zero capital lease leverage exclusively, or (3) non-zero operating lease leverage exclusively For each subsample (i.e., disclosed RPE peer choice, along with accounting method choice), the mean of financial reporting comparability (FRC) is reported together with the percentage of firm-year-peer observations where the same accounting method is used (continued on next page) The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 351 TABLE (continued) Panel C: Logistic Analysis of Observed RPE Peers (Dep Var ¼ Selected) (1) FRC 26.686*** (17.67) (2) (3) (4) 7.121*** (3.19) Same_Method: Depreciation 0.419*** (6.01) Same_Method: Inventory 0.392*** (6.50) Same_Method: Lease Same_SIC2 Same_SIC4 S&P500 3.102*** (36.69) 1.661*** (17.13) 2.191*** (27.37) SALES_Gap MVE_Gap RET_Gap ROA_Gap MTB_Gap Multi_Bus_Seg Single_Bus_Seg Multi_Geo_Seg Single_Geo_Seg SALES_Peer MVE_Peer RET_Peer ROA_Peer MTB_Peer Volatility_RET_Peer Volatility_ROA_Peer # of obs Pseudo R2 (5) 37,820 0.4991 3.108*** (30.66) 1.982*** (16.95) 1.092*** (13.77) À0.896*** (À18.89) À0.388*** (À9.46) À0.384*** (À4.61) 1.311*** (À3.18) À0.410*** (À8.48) 0.406*** (7.21) À0.173*** (À2.33) 1.392*** (19.25) 0.283*** (3.38) 0.000*** (6.41) 0.000* (1.65) 0.089 (1.14) 0.472* (1.43) 0.183*** (4.15) À1.502*** (À2.37) À4.542** (À2.19) 37,593 0.5935 2.690*** (23.77) 2.551*** (14.68) 1.152*** (11.85) À0.846*** (À17.60) À0.350*** (À7.64) À0.402*** (À4.75) 0.438 (0.93) À0.423*** (À8.89) 0.398*** (6.02) À0.023 (À0.28) 0.903*** (12.28) 0.214** (1.72) 0.000*** (5.33) 0.000 (1.24) 0.180** (2.27) 0.251 (0.64) 0.126*** (2.84) À2.139*** (À2.77) À1.886 (À0.71) 22,712 0.5621 2.298*** (18.15) 2.763*** (13.26) 1.187*** (11.42) À0.886*** (À15.71) À0.369*** (À8.11) À0.375*** (À4.10) À0.004 (À0.01) À0.330*** (À6.33) 0.411*** (5.95) À0.074 (À0.82) 0.729*** (9.70) 0.474*** (2.90) 0.000*** (4.87) 0.000 (0.65) 0.171** (2.15) À0.253 (À0.61) 0.135*** (2.64) À2.251*** (À2.80) 3.152* (1.63) 17,421 0.5466 0.235*** (4.33) 2.743*** (25.14) 2.585*** (16.01) 1.028*** (11.47) À0.864*** (À16.84) À0.345*** (À8.05) À0.393*** (À4.63) 0.629* (1.33) À0.444*** (À9.10) 0.272*** (4.52) À0.005 (À0.06) 1.122*** (15.77) 0.372*** (3.69) 0.000*** (6.74) 0.000 (1.06) 0.150** (1.89) 0.289 (0.75) 0.163*** (3.71) À1.853*** (À2.68) À3.242* (À1.32) 26,654 0.5607 (6) 10.563*** (4.41) 0.367*** (4.70) 0.383*** (6.46) À0.099** (À1.76) 2.302*** (17.87) 2.766*** (12.70) 1.224*** (11.40) À0.857*** (À16.43) À0.349*** (À7.38) À0.339*** (À3.53) 0.616* (1.30) À0.320*** (À5.96) 0.421*** (5.63) À0.098 (À1.08) 0.686*** (8.34) 0.481*** (2.97) 0.000*** (5.27) 0.000 (0.46) 0.143** (1.77) À0.146 (À0.35) 0.121** (2.27) À1.794** (À2.13) 0.392 (0.18) 16,166 0.5538 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (two-tailed) This panel presents the result of estimating Equation (5) Year FE are included Wald Chi-square statistics (in parentheses) are based on robust standard errors clustered at the firm level (continued on next page) The Accounting Review Volume 95, Number 3, 2020 Nam 352 TABLE (continued) Variable Definitions: Selected ¼ the dependent variable is a dummy that equals if a potential peer firm is chosen as an RPE peer firm in the proxy statement disclosure of RPE practice, and otherwise; Same_SIC2 (SIC4) ¼ a dummy that equals if an RPE discloser and a peer are in the same two-digit (four-digit) SIC group; S&P500 ¼ a dummy that equals if a peer belongs to the S&P500 as of December 31, 2013; FRC ¼ the financial reporting comparability between an RPE discloser and a peer firm, using the method proposed by De Franco et al (2011); Same_Method ¼ a dummy that equals if an RPE discloser and a peer use the same accounting method in depreciation, inventory valuation, or lease; SALES_Gap (MVE_Gap) ¼ the absolute difference in sales (market capital) between an RPE discloser and a peer firm, divided by the sales (market capital) of the RPE discloser; ROA_Gap (RET_Gap) ¼ the absolute difference in ROA (annual stock returns with no scaling) between an RPE discloser and a peer firm; MTB_Gap ¼ the absolute difference in the market-to-book ratio between an RPE discloser and a peer firm; Multi_Bus_Seg (Single_Bus_Seg) ¼ a dummy that equals if both an RPE discloser and a peer have multiple (single) business segments; Multi_Geo_Seg (Single_Geo_Seg) ¼ a dummy that equals if both an RPE discloser and a peer have multiple (single) geographic segments; Volatility_RET_Peer ¼ the standard deviation of peer firm returns over the previous 48 months; and Volatility_ROA_Peer ¼ the standard deviation of peer firm ROA over the previous 16 quarters committees are cognizant that the effectiveness of accounting-based RPE increases with financial reporting comparability, as illustrated in Section III, then a researcher who disregards this criterion in peer selection could have difficulty detecting the use of accounting-based RPE in CEO compensation contracts To address the potential for a bias that results from a divergence between a board and a researcher in terms of their respective peer selections, I include financial reporting comparability as an additional peer selection criterion Using peers with high financial reporting comparability, I examine the use of accountingbased RPE for CEO performance evaluation in the following sections Sample Construction I collect the annual CEO compensation data of S&P 1500 firms from Execucomp for the 1992–2013 period The initial dataset contains 37,053 CEO-year observations I exclude firm-year observations when a firm has more than one CEO in a given year or when the CEO has held the position for less than 12 months I also exclude observations with a non-positive book value of common equity and missing values for compensation, sales, and market value RPE Peer Selection Strategy For each firm-year observation of CEO compensation, I also collect data for all firms in the same industry (two-digit SIC) from the merged CRSP-Compustat database This ensures that potential peer candidates are included even if they are not in Execucomp I require peer firms to have total assets of at least $10 million (Albuquerque 2009) To ensure that peer candidates are not too few in number, I require that a firm-year observation has at least 20 same-industry peer firms that satisfy the data requirements (Jayaraman, Milbourn, Peters, and Seo 2018) The final sample consists of 20,025 firm-year observations for 2,626 distinct firms (see Table 3).15 Albuquerque (2009) selects peers matched on industry and firm size based on the premise that different-sized firms are exposed to different exogenous shocks and face different constraints in responding to those shocks.16 In addition to industry and size, I introduce financial reporting comparability (FRC) as an additional criterion for identifying peer firms Comparable accounting systems enable the integration of multi-firm accounting information and, thus, facilitate RPE Specifically, for a firm-year (treatment) observation on Execucomp, I estimate DKV’s financial reporting comparability (FRC) for firm i and each of its peers within the same two-digit SIC group I then select peer candidates belonging to the top quartile of FRC From this selected pool, I choose the top ten (minimum five) peer firms that are closest in size relative to the treatment firm based on beginning-of-year market capitalization This peer selection procedure (SIC2-FRC-SIZE) results in an RPE peer group consisting of same-industry peer firms that are similar in size and have similar financial reporting systems as the treatment firm.17 Peer-firm performance is measured as the average performance of the selected peer group 15 16 17 Because the measure of financial reporting comparability (FRC) is based on the earnings-returns mapping over the preceding 16 quarters, this data requirement leads to an additional percent cut, approximately, in the sample size compared to prior RPE studies that only use industry membership and firm size as matching attributes Albuquerque (2009) argues that an ideal peer group is composed of firms that are not only in the same industry, but that also have characteristics similar to the treatment firm She contends that selecting industry-size peers is a parsimonious way of capturing various dimensions of firm characteristics My empirical results are robust to (1) selecting peers from the top quintile (instead of the top quartile) of financial reporting comparability, and (2) excluding firm-year observations with fewer than ten SIC2-FRC-SIZE peers The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 357 opportunities (measured by R&D intensity) CEOs who chair their respective boards and those with low equity ownership are also more highly compensated.22 Panel B of Table tests for RPE in CEO equity compensation I find that the use of accounting-based RPE is insignificant in determining CEO equity compensation, reaffirming that the contracting value of accounting information is prevalent in determining cash compensation (e.g., Jayaraman and Milbourn 2012) In contrast, the use of price-based RPE appears significant with industry-comparability-size (SIC2-FRC-SIZE) or industry-size (SIC2-SIZE) peers (Columns (1) and (3)) When selecting peers based on industry and financial reporting comparability only, the coefficient estimate on RET_Peer is insignificant (Column (2)), consistent with firm size playing a crucial role in detecting empirical support for price-based RPE (Albuquerque 2009).23 CEO Turnover and Peer Performance The role of RPE may be more prominent in CEO turnover, as firing a CEO is an extreme event resulting from poor performance Prior studies examine the use of either price- or accounting-based RPE in CEO dismissal decisions Barro and Barro (1990) find evidence that supports a complete filtering of peer performance in a sample of CEO turnover from 1982 to 1987 Consistent with RPE usage, Gibbons and Murphy (1990) document that CEO retention decisions filter out both market and industry shocks DeFond and Park (1999) show that RPE is used more in CEO termination decisions in high-competition industries than it is in low-competition industries, which might lead to the lack of empirical support for RPE found in those studies that disregard industry competition Examining the Execucomp database for the time period from 1993 to 2009, Jenter and Kanaan (2015) find that boards partially adjust for peer-firm price performance when making CEO turnover decisions, but the adjustment is too small to remove all the peer performance effect Accounting performance significantly explains CEO turnover (e.g., Bushman and Smith 2001; Engel, Hayes, and Wang 2003) Hermalin and Weisbach (1998) posit that the stock price embeds the market’s expectations about hiring a new CEO, so that earnings are a cleaner signal of the current CEO’s talent than is the stock price Given that CEO turnover is associated with the firm’s accounting performance measures, one would expect peer accounting performance measures to help screen out poorperforming CEOs Examining the benefits of cross-country comparable accounting systems in the European Union (EU), Wu and Zhang (2010) document that accounting-based RPE is used more in CEO turnover decisions following mandatory IFRS adoption In this section, I examine the relationship between peer group performance and CEO turnover using peer selection based on cross-firm financial reporting comparability (FRC) In Table 6, I find that CEO turnover is inversely related to both own-firm accounting and stock price performance.24 When the peer group is selected based on financial reporting comparability (Columns (1) and (2)), the role of accounting-based RPE is supported, as evidenced by the significantly positive coefficient on peer-firm accounting performance, whereas the role of price-based RPE is insignificant In contrast, using conventional industry-size (SIC2-SIZE) peers, I find no relation between CEO turnover and peer-firm price, as well as peer-firm accounting performance (Column (3)) Thus, as with the results for CEO compensation, the results for CEO turnover reaffirm that the use of peer-firm accounting information as an evaluation benchmark is empirically evident only when the peer group includes firms with high financial reporting comparability Strong-Form RPE Holmstrom and Milgrom (1987) state that an optimal contract compensates the CEO only for unsystematic firm performance where common risk factors are completely removed (i.e., strong-form RPE) The main focus of this paper is to test weak-form RPE (Equation (6)) Specifically, I assume that peer-firm performance at least partially filters the effect of common risk from evaluation Weak-form RPE is a necessary, but not a sufficient, condition for the existence of strong-form RPE Studies often document support for weak-form RPE, but not for strong-form RPE (e.g., Janakiraman et al 1992; Rajgopal, Shevlin, and Zamora 2006) To estimate strong-form RPE, I modify Equation (6) by replacing own-firm and peer-firm performance with measures of unsystematic and systematic performance (Antle and Smith 1986): 22 23 24 Results are unchanged when I estimate year-wise cross-sectional regressions following Fama and MacBeth (1973) In untabulated tests using CEO Total Compensation (the logarithm of plus TDC1) as the dependent variable, I find that the use of accounting-based (price-based) RPE increases (decreases) in the fraction of CEO Cash Compensation to Total Compensation CEO Turnover is a dummy variable that equals if the firm’s CEO does not continue as CEO or does not hold any other director or executive position in the same firm in the following year (as reported by Execucomp), and otherwise This definition rules out cases such as the former CEO becoming chairman the year after stepping down from the CEO position The average CEO turnover rate is percent in my sample Jenter and Kanaan (2015) search the Factiva news database to classify CEO turnover according to whether the turnover was forced or voluntary They report that the forced (voluntary) CEO turnover ratio is 2.77 percent (7.85 percent) The Accounting Review Volume 95, Number 3, 2020 Nam 358 TABLE CEO Turnover and RPE Peer Selection SIC2-FRC-SIZE (1) RET_Firm RET_Peer ỵ ROA_Firm ROA_Peer þ AGE AGE_6365 CEO Tenure CEO Chair CEO Ownership SALES MTB # of obs Pseudo R2 À0.693*** (À3.89) 0.202 (1.01) À4.043*** (À6.56) 1.513** (1.73) 0.083*** (9.97) 0.804*** (6.76) 0.113** (1.65) À0.529*** (À4.97) 0.995*** (8.26) À0.017 (À0.50) 0.002 (1.04) 16,218 0.0336 SIC2-FRC (2) À0.647*** (À3.55) À0.057 À0.23 À4.088*** (À6.46) 1.456* (1.64) 0.083*** (9.90) 0.803*** (6.76) 0.118** (1.72) À0.547*** (À5.18) 0.974*** (8.18) 0.002 (0.05) 0.002 (1.03) 16,218 0.0335 SIC2-SIZE (3) À0.743*** (À3.91) 0.209 (0.89) À3.332*** (À6.93) 0.748 (0.85) 0.082*** (9.80) 0.801*** (6.73) 0.124** (1.81) À0.548*** (À5.19) 0.968*** (8.12) À0.008 (À0.19) À0.045 (À0.80) 16,218 0.0334 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (p jzj) The table provides the results of a logistic regression of CEO turnover on performance measures and control variables The dummy dependent variable, CEO Turnover, is equal to in the year in which the current CEO is replaced, and otherwise Industry FE, Year FE, and an Intercept are included in the estimations, but untabulated for brevity Z-statistics (in parentheses) are based on firm-level clustered standard errors Continuous variables are winsorized at the percent and 99 percent levels All variables are defined in Appendix A Payit ¼ v0 þ v1 à SystematicRETit þ v2 à UnsystematicRETit þ v3 SystematicROAit ỵ v4 UnsystematicROAit XK ỵ w Controlsikt ỵ xit ; kẳ1ị k 7ị where systematic firm performance (unsystematic firm performance) is defined as the predicted (residual) value from regressing own-firm performance on a constant and peer-firm performance If the optimal amount of noise is filtered out in determining CEO compensation, then the CEO is not compensated based on the component of own-firm price (accounting) performance subject to exogenous factors, i.e., v1 ¼ (v3 ¼ 0) Table reports the results of the test H0: v1 ¼ (v3 ¼ 0) against HA: v1 6¼ (v3 6¼ 0) When financial reporting comparability is used as an additional peer selection criterion (SIC2-FRC-SIZE or SIC2FRC), the coefficient on systematic accounting performance is positive and significant, albeit substantially lower in magnitude relative to unsystematic accounting performance (Columns (1) and (2)) With conventional industry-size peers (SIC2-SIZE), I continue to find no significant evidence for strong-form accounting-based RPE.25 Therefore, despite the evidence of the use of 25 Consistent with Albuquerque (2009), when using industry-size (SIC2-SIZE) peers, I find significant evidence of strong-form price-based RPE at the total or the equity compensation levels (untabulated) The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 359 TABLE Strong-Form RPE Tests Peer Selection SIC2-FRC-SIZE Pay Type ¼ Cash (1) Systematic RET_Firm Unsystematic RET_Firm Systematic ROA_Firm Unsystematic ROA_Firm SALES MTB R&D Idiosyncratic Risk Regulation CEO Tenure Interlock Chair Ownership # of obs Adj R2 0.180*** (6.41) 0.202*** (16.33) 0.421*** (3.13) 1.136*** (10.96) 0.322*** (33.92) À0.007 (À0.76) 0.018** (2.33) À1.095*** (À3.16) 0.051 (0.67) 0.036*** (3.00) À0.109*** (À2.75) 0.129*** (6.73) 0.070*** (3.48) 20,025 0.499 SIC2-FRC Pay Type ¼ Cash (2) 0.242*** (8.49) 0.194*** (15.44) 0.399*** (3.02) 1.192*** (11.81) 0.322*** (34.19) À0.008 (À0.85) 0.018** (2.44) À1.128*** (À3.29) 0.051 (0.67) 0.036*** (2.97) À0.109*** (À2.75) 0.128*** (6.68) 0.069*** (3.43) 20,025 0.499 SIC2-SIZE Pay Type ¼ Cash (3) 0.143*** (6.36) 0.225*** (16.34) 2.368*** (7.82) 0.758*** (7.52) 0.309*** (31.20) À0.012 (À1.25) 0.021** (2.53) À0.755** (À2.24) 0.059 (0.76) 0.033*** (2.79) À0.105*** (À2.64) 0.128*** (6.69) 0.068*** (3.40) 20,025 0.499 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (two-tailed) This table presents the results from regressing CEO compensation on systematic and unsystematic firm performance (strong-form RPE) Systematic firm performance is the predicted value from regressing firm performance on the average peer group performance Unsystematic firm performance is the error term from the regression Industry FE, Year FE, and an Intercept are included in estimations, but omitted from the table for brevity t-statistics (in parentheses) are based on firm-level clustered standard errors Continuous variables are winsorized at the percent and 99 percent levels All variables are defined in Appendix A accounting-based RPE (in the weak form), common risks are not completely filtered out from the accounting-performance measures (in the strong form) when determining CEO compensation Alternatively, Dikolli et al (2013) posit that strong-form RPE tests are more subject to Type II inferential errors arising from an empiricist’s crude peer selection than are weak-form RPE tests.26 As such, there may be other ways to refine further an empiricist’s peer selection so as to detect strong-form accounting-based RPE 26 Dikolli et al (2013) presume an empiricist testing for weak-form RPE based on the coefficient estimates from a regression that originally tests for strong-form RPE, such as Equation (7) They state that the impact of the empiricist’s peer selection is less salient to the weak-form RPE tests than it is to strong-form tests However, rearranging the terms of their analysis demonstrates that the empiricist’s crude peer selection non-trivially biases upward the coefficient estimate on peer-firm performance from Equation (6) Refining the empiricist’s peer selection to reduce biases against detecting RPE usage, this study and Albuquerque (2009), thus, provide empirical support for Dikolli et al.’s (2013) intuition The Accounting Review Volume 95, Number 3, 2020 Nam 360 Additional Tests Alternative Measure of Financial Reporting Comparability Because the DKV measure is based on the earnings-returns relation in Equation (1), it might be affected by factors external to the accounting system In cases where the relation between earnings and returns is weak (e.g., high-tech firms), it is difficult to know to what extent the DKV measure can capture the similarities between two firms’ accounting systems Further, although earnings are arguably the most important summary measure of accounting performance, the DKV measure does not capture financial reporting decisions beyond income statements (De Franco et al 2011, 927) De Franco et al (2011) also show that firms with similar economic characteristics (e.g., industry, size, growth) exhibit higher financial reporting comparability This raises the question of whether peer selection based on DKV’s measure (FRC) enables a researcher to capture fundamental similarities between two firms rather than the firms’ accounting comparability Since firms with similar characteristics would face similar exogenous shocks (Albuquerque 2009), one may argue that the empirical evidence supporting the use of accounting-based RPE, as presented in this study, may not necessarily reflect the benefits of accounting comparability across firms To alleviate these concerns, I replace DKV’s base model, Equation (1), with the accruals-cash flows relation Because accruals and cash flows are direct outputs of a firm’s financial reporting decisions, mapping accruals onto cash flows can reflect the effect of accounting factors on cross-firm financial reporting comparability I first estimate the following firm-level regressions over the previous 16 (minimum 12) quarters.27 Accrualsit ẳ ỵ bi1 CFOit1 ỵ bi2 CFOit ỵ bi3 CFOitỵ1 ỵ eit ; 8ị 28 where Accruals is the change in working capital; and CFO is the cash flow from operations (OANCFY) The coefficient estimates from Equation (8), b^i1 , b^i2 , and b^i3 , are the firm-specific accounting system parameters that map the accruals of firm i onto firm i’s lagged, current, and lead cash flow from operations.29 Similarly, I estimate b^j1 , b^j2 , and b^j3 as the parameters of peer-firm j’s accounting system I calculate the predicted accruals of firm i by applying the parameter estimates of the accounting systems of firms i and j, respectively, to firm i’s cash flow from operations, as follows: EðAccrualsÞiit ẳ a^i ỵ b^i1 CFOit1 ỵ b^i2 CFOit ỵ b^i3 CFOitỵ1 ; 9ị EAccrualsịijt ẳ a^j ỵ b^j1 CFOit1 ỵ b^j2 CFOit ỵ b^j3 CFOitỵ1 : 10ị Equation (9) estimates the predicted accruals of firm i based on the accounting system parameters of firm i, whereas Equation (10) estimates the predicted accruals of firm i based on the accounting system parameters of firm j FRCDD is defined as the negative value of the absolute difference between the predicted accruals estimated from Equations (9) and (10) averaged over the preceding 16 quarters From the top quartile of financial reporting comparability measured by FRCDD in the same industry (SIC two-digit), I choose the top ten (minimum five) peer firms that are closest in size relative to the treatment firm based on beginning-of-year market capitalization Using these selected peer firms (SIC2-FRCDD-SIZE), the coefficient on (own-firm) peer-firm accounting performance is (0.860) À0.366, which is significant at the percent level (Column (1) of Table 8) This outcome continues to support the use of accounting-based RPE When peers are selected based on industry and financial reporting comparability (SIC2-FRCDD), I continue to find evidence of accounting-based RPE (Column (2)).30 This finding is difficult to reconcile with the view that DKV’s methodology merely reflects economic, rather than accounting, similarities between two firms 27 28 29 30 Cascino and Gassen (2015) similarly estimate financial reporting comparability based on the contemporaneous relation between total accruals and cash flow from continuing operations, but they not include lagged and lead cash flows Accruals is measured as the increase in accounts receivable (RECCHY) plus the increase in inventory (INVCHY) plus the decrease in accrued taxes (TXACHY) plus the decrease in accounts payable and accrued liabilities (APALCHY) plus the increase (decrease) in other assets (liabilities) (AOLOCHY), multiplied by (À1) The variables are scaled by lagged total assets I find that the characteristics of the accruals-cash flow relationship, as documented in prior studies, largely hold at the quarterly level (untabulated) Consistent with Dechow and Dichev (2002), I find that accruals are negatively related to current operating cash flow, and positively related to past and future operating cash flow In firm-specific regressions, the mean coefficient on current cash flow ðb^i2 Þ is À0.72 and the mean coefficients on lagged cash flow ðb^i1 Þ and lead cash flow ðb^i2 Þ are 0.09 and 0.13, respectively Because the data requirement for Equation (8) is more restrictive than the requirement for Equation (1), using FRCDD reduces the sample size in Table The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 361 TABLE Financial Reporting Comparability Based on Accruals-Cash Flow Relation (FRCDD) Peer Selection RET_Firm RET_Peer ROA_Firm ROA_Peer SALES MTB R&D Idiosyncratic Risk Regulation CEO Tenure Interlock Chair Ownership # of obs Adj R2 SIC2-FRCDD-SIZE Pay Type ¼ Cash (1) SIC2-FRCDD Pay Type ¼ Cash (2) SIC2-SIZE Pay Type ¼ Cash (3) 0.195*** (13.99) À0.003 (À0.15) 0.860*** (8.24) À0.366*** (À3.45) 0.320*** (28.15) À0.004 (À0.35) 0.016** (1.99) À1.020** (À2.54) 0.120 (1.34) À0.134** (À2.50) 0.022 (1.59) 0.141*** (6.16) 0.063*** (2.79) 15,523 0.507 0.196*** (14.24) À0.003 (À0.18) 0.838*** (8.20) À0.279*** (À4.13) 0.317*** (28.64) À0.005 (À0.49) 0.017** (2.17) À1.018** (À2.55) 0.123 (1.37) À0.133** (À2.49) 0.022 (1.55) 0.142*** (6.19) 0.062*** (2.71) 15,523 0.507 0.217*** (14.60) À0.079*** (À3.40) 0.674*** (6.37) 0.616*** (4.69) 0.305*** (26.25) À0.008 (À0.81) 0.019** (2.27) À0.804** (À2.01) 0.133 (1.48) 0.021 (1.48) À0.133** (À2.48) 0.140*** (6.15) 0.061*** (2.74) 15,523 0.508 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (two-tailed) This table provides the regression estimates of Equation (6), testing the use of RPE in CEO compensation For financial reporting comparability, I use an alternative measure, FRCDD, which is based on the accruals-cash flow relation (Dechow and Dichev 2002) Coefficients reported in bold are consistent with the existence of RPE Industry FE, Year FE, and an Intercept are included t-statistics (in parentheses) are based on firm-level clustered standard errors Continuous variables are winsorized at the percent and 99 percent levels All variables are defined in Appendix A Economic Comparability versus Financial Reporting Comparability Given that financial accounting measures a firm’s underlying economics, it is natural that economically comparable firms would have comparable financial numbers To the extent that boards know better how to compare the economics between firms rather than whether a firm accounts for economic events in the same way as its peers, using FRC as a selection criterion may enable a researcher to better identify economically comparable peers I study further this alternative explanation by selecting peer firms based on economic comparability (EC) To measure economic comparability, I follow DKV and run a regression of firm i’s earnings against firm j’s earnings over the preceding 16 quarters The adjusted R2 from this regression is defined as the firm i-firm j economic comparability (CompEAR-R2) I repeat this procedure for operating cash flow and stock return (CompCFO-R2 and CompRET-R2, respectively) Additionally, following Chen et al (2018), I estimate economic comparability by replacing the earnings variables in DKV’s methodology (Equations (1)–(4)) with operating cash flows If similarities in the underlying economics drive my result, then peer selection based on economic comparability should also detect the use of accounting-based RPE The Accounting Review Volume 95, Number 3, 2020 Nam 362 TABLE Economic Comparability (EC) in Peer Selection Panel A: R2-Based Measure (DKV) EC Measure CompEAR-R (DKV) RET_Firm RET_Peer ROA_Firm ROA_Peer # of obs Adj R2 CompCFO-R2 (DKV) CompRET-R2 (DKV) Peer Selection SIC2-EC-SIZE (1) Peer Selection SIC2-EC (2) Peer Selection SIC2-EC-SIZE (3) Peer Selection SIC2-EC (4) Peer Selection SIC2-EC-SIZE (5) Peer Selection SIC2-EC (6) 0.208*** (16.86) À0.002 (À0.09) 0.660*** (6.62) 0.507*** (5.66) 0.208*** (17.71) 0.014 (0.84) 0.760*** (7.96) 0.152** (2.35) 0.208*** (16.88) À0.031* (À1.82) 0.705*** (7.13) 0.438*** (4.81) 0.207*** (17.16) 0.000 (0.03) 0.793*** (8.38) 0.117 (1.41) 0.220*** (17.99) À0.051*** (À2.93) 0.680*** (6.91) 0.429*** (4.55) 0.205*** (17.26) 0.005 (0.27) 0.804*** (8.57) 0.042 (0.26) 18,415 0.500 20,028 0.498 17,125 0.501 19,029 0.499 18,422 0.499 19,026 0.498 Panel B: Modified-FRC Based on CFO (Chen et al 2018) EC Measure Modified-FRC based on CFO (Chen et al 2018) RET_Firm RET_Peer ROA_Firm ROA_Peer # of obs Adj R2 Peer Selection SIC2-EC-SIZE (7) Peer Selection SIC2-EC (8) 0.208*** (16.82) À0.044** (À2.29) 0.806*** (8.03) À0.07 (À0.759) 0.204*** (17.24) 0.02 (1.32) 0.902*** (9.31) À0.160*** (À2.82) 17,228 0.503 19,021 0.499 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (twoÀtailed) This table presents the results of estimating Equation (6) using economic comparability (EC) in peer selection EC is estimated by one of four measures Following DKV, I use three R2-based measures from firm-peer pairwise regressions of earnings, cash flow from operations, and returns, respectively Additionally, following Chen et al (2018), a modified DKV measure based on CFO is used to proxy for EC The dependent variable is Cash Compensation Industry FE, Year FE, an Intercept, and control variables are included t-statistics (in parentheses) are based on firm-level clustered standard errors Continuous variables are winsorized at the percent and 99 percent levels All variables are defined in Appendix A The results in Table not provide support for this prediction in most specifications The coefficient estimate on peerfirm accounting performance is insignificant in all columns except Column (8), which uses the modified FRC measure based on CFO, following Chen et al (2018) Table 9, thus, reaffirms that DKV’s methodology is not merely capturing economic comparability Nevertheless, I acknowledge that economic comparability is a broad constraint that is difficult to measure using any single variable or group of variables One may argue that DKV’s measure of comparability captures similarities in fundamentals rather than accounting A narrow interpretation of my findings would be that using a technique to identify more similar peer firms mitigates inferential errors that create a bias against finding support for RPE The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 363 TABLE 10 Alternative Specifications Panel A: Alternative Measures of Peer Performance Peer Selection SIC2-FRC-SIZE 15 Peers Pay Type ¼ Cash (1) RET_Firm RET_Peer ROA_Firm ROA_Peer # of obs Adj R2 SIC2-FRC 15 Peers Pay Type ¼ Cash (2) 0.196*** (15.75) 0.002 (0.13) 1.502*** (14.68) À0.511*** (À4.17) 20,025 0.494 SIC2-FRC-SIZE Median Pay Type ¼ Cash (3) 0.192*** (15.37) 0.019 (0.99) 1.537*** (15.07) À0.530*** (À4.67) 20,025 0.495 0.196*** (15.44) 0.019 (1.08) 1.159*** (11.53) À0.823*** (À5.83) 20,025 0.499 SIC2-FRC Median Pay Type ¼ Cash (4) 0.186*** (14.54) 0.067*** (3.04) 1.582*** (15.21) À0.678*** (À5.26) 20,025 0.495 SIC2-SIZE for RET_Peer; SIC2-FRC-Size for ROA_Peer Pay Type ¼ Cash (5) 0.199*** (14.91) À0.013 (À0.70) 1.486*** (14.51) À0.477*** (À4.22) 20,025 0.499 Panel B: CEO Fixed Effects and Change in Compensation Peer Selection SIC2-FRC-SIZE (CEO Fixed Effects) Pay Type ¼ Cash (6) RET_Firm RET_Peer ROA_Firm ROA_Peer # of obs Adj R2 Pay Type ¼ Bonus (7) 0.181*** (12.90) À0.011 (À0.83) 1.134*** (12.79) À0.131 (À1.27) 20,025 0.771 0.936*** (16.79) À0.243*** (À3.59) 5.873*** (13.17) À0.677* (À1.86) 20,025 0.492 SIC2-FRC (CEO Fixed Effects) Pay Type ¼ Cash (8) 0.184*** (22.59) À0.006 (À0.46) 1.258*** (21.79) À0.195* (À1.66) 20,025 0.772 Pay Type ¼ Bonus (9) 0.928*** (24.40) À0.121** (À2.05) 6.102*** (22.59) À0.865*** (À2.88) 20,025 0.492 SIC2-FRC-SIZE (Change Regression) Pay Type ¼ DCash (10) 0.206*** (7.66) À0.046** (À1.97) 0.553*** (6.01) 0.28 (1.04) 11,285 0.048 *, **, *** Denote significance at the 10 percent, percent, and percent levels, respectively (two-tailed) This table presents the results of estimating Equation (6) using various specifications Industry FE, Year FE, an Intercept, and control variables are included In Columns (6)–(9), CEO fixed effects are included In Column (10), the dependent variable is the logarithm of plus the percentage annual change in cash compensation for the same CEO, and the changes in ROA_Firm, ROA_Peer, and control variables are included t-statistics (in parentheses) are based on firm-level clustered standard errors Continuous variables are winsorized at the percent and 99 percent levels All variables are defined in Appendix A Alternative Regression Specifications Table 10 provides the test results for Equation (6) under alternative specifications First, I expand the peer group size by selecting up to 15 peer firms (with a minimum of five) Second, I choose the median of the peer group performance instead of the mean In Panel A, Columns (1)–(4), I continue to find empirical support for the use of accounting-based RPE For the third specification, I estimate Equation (6) using industry-size (SIC2-SIZE) peers for calculating peer stock price performance and industry-comparability-size (SIC2-FRC-SIZE) peers for calculating peer accounting performance Although The Accounting Review Volume 95, Number 3, 2020 Nam 364 this study focuses on accounting-based RPE, the effect of financial reporting comparability on price-based RPE is also of interest It is ex ante unclear to what extent comparable accounting systems contribute to comparable stock price performances across firms Considering the low explanatory power of earnings for stock returns (Lev 1989), financial reporting comparability may not be a relevant peer selection criterion in assessing the contracting value of peer-firm stock price performance in CEO compensation This is consistent with anecdotal evidence on the use of alternate peer groups for accounting- versus price-based RPE.31 When I employee different peer groups for accounting- and price-based RPE (Table 10, Panel A, Column (5)), I continue to find significant use of accounting-based RPE I also estimate Equation (6) with CEO fixed effects The empirical support for accounting-based RPE becomes weaker (Table 10, Panel B, Columns (6)–(9)) Moreover, the use of accounting-based RPE becomes insignificant in determining salary when CEO fixed effects are included (untabulated), perhaps due to the relation between CEO reputation and accounting-based RPE.32 The contracting value of peer-firm earnings may be lower when contracting with a reputable CEO, who often receives a higher reservation wage (Rajgopal et al 2006).33 When examining changes in cash compensation for the same CEO (Column (10)), I find that accounting-based RPE is insignificant This result suggests that peer-related compensation is more likely to be a function of the relative accounting profitability of a firm compared to its peers, rather than of relative changes in accounting profitability VI CONCLUSION Prior studies find that peer firms are a valuable source of information when evaluating a firm of interest (e.g., De Franco et al 2011; Kim et al 2016; Shroff, Verdi, and Yost 2017) I investigate whether peer-firm accounting numbers are useful in designing CEO compensation contracts I propose that efficient relative evaluation using peer-firm accounting performance requires peer firms to have a comparable financial reporting system To the extent that comparable accounting systems increase the quantity and quality of information available (DKV), RPE peers with high financial reporting comparability are likely to facilitate the use of accounting-based RPE in CEO performance evaluation Consistent with this assertion, I find that financial reporting comparability is a significant determinant of boards’ RPE peer selection as it is disclosed in proxy statements The results of this study suggest that disregarding financial reporting comparability when selecting RPE peers limits empirical tests’ ability to detect the use of accounting-based RPE Using financial reporting comparability as a peer selection criterion, along with industry and size, I document the use of accounting-based RPE in determining CEO cash compensation in the U.S In addition, using an alternative measure of financial reporting comparability that only includes financial reporting outputs in the base model (i.e., accruals and cash flows), I continue to find empirical evidence supporting the use of accountingbased RPE Overall, I find that highly comparable peer-firm accounting measures provide incremental information useful for CEO compensation contracts Moreover, the results highlight that financial reporting comparability is a key attribute that facilitates the use of peers’ accounting information in evaluating CEO performance REFERENCES Albuquerque, A M 2009 Peer firms in relative performance evaluation Journal of Accounting and Economics 48 (1): 69–89 https://doi org/10.1016/j.jacceco.2009.04.001 Albuquerque, A M 2014 Do growth-option firms use less relative performance evaluation? The Accounting Review 89 (1): 27–60 https://doi.org/10.2308/accr-50574 31 32 33 See Appendix B for one such example relating to International Paper Co Adding CEO fixed effects effectively means using the within-CEO time-series variation in peer ROA to explain the within-CEO time-series variation in compensation Given that innate (i.e., time-invariant) CEO characteristics (e.g., reputation, education, and employment experience at the time of appointment) may affect peer ROA, which, in turn, affects compensation, the removal of these characteristics via CEO fixed effects could control for relevant factors that impact the relation between compensation and peer ROA That is, we may be ‘‘throwing the baby out with the bathwater’’ when including CEO fixed effects This is of concern with regard to cash compensation (particularly salary), in which reservation wages are often paid From an RPE perspective, it does not matter what underlying factors drive peer ROA For example, peer ROA may be higher because the focal firm has a less reputable/capable CEO who is not good at handling competition Nevertheless, if peer firms perform better, especially within the same industry (which is the reason I use industry fixed effects in the main specification), then the CEO of the firm of interest should be paid less After controlling for CEO reputation, as proxied for by industry-adjusted ROA (Rajgopal et al 2006), the use of accounting-based RPE remains significant, although accounting-based RPE is used less for reputable CEOs (untabulated) That is, not only an empiricist’s crude peer selection, but also the substantial cross-sectional variation in RPE usage could make it difficult for prior studies to document on-average evidence for accountingbased RPE The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 365 Albuquerque, A M., G De Franco, and R S Verdi 2013 Peer choice in CEO compensation Journal of Financial Economics 108 (1): 160–181 https://doi.org/10.1016/j.jfineco.2012.10.002 Albuquerque, A M., B Chen, F Dong, and E J Riedl 2014 Do boards exercise discretion to reduce costly ex post settling up? Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id¼2479330 Antle, R., and J S Demski 1988 The controllability principle in responsibility accounting The Accounting Review 63 (4): 700–718 Antle, R., and A Smith 1986 An empirical investigation of the relative performance evaluation of corporate executives Journal of Accounting Research 24 (1): 1–39 https://doi.org/10.2307/2490802 Baber, W R., S-H Kang, and K R Kumar 1998 Accounting earnings and executive compensation Journal of Accounting and Economics 25 (2): 169–193 https://doi.org/10.1016/S0165-4101(98)00021-4 Baker, G 2002 Distortion and risk in optimal incentive contracts Journal of Human Resources 37 (4): 728–751 https://doi.org/10.2307/ 3069615 Banker, R D., M N Darrough, R Huang, and J M Plehn-Dujowich 2013 The relation between CEO compensation and past performance The Accounting Review 88 (1): 1–30 https://doi.org/10.2308/accr-50274 Barro, J R., and R J Barro 1990 Pay, performance, and turnover of bank CEOs Journal of Labor Economics (4): 448–481 https:// doi.org/10.1086/298230 Black, D E., S S Dikolli, and C Hofmann 2015 Peer group composition, peer performance aggregation, and detecting relative performance evaluation Working paper, University of Nebraska–Lincoln Blackwell, D W., J A Brickley, and M S Weisback 1994 Accounting information and internal performance evaluation: Evidence from Texas banks Journal of Accounting and Economics 17 (3): 331–358 https://doi.org/10.1016/0165-4101(94)90032-9 Bradshaw, M T., G S Miller, and G Serafeim 2009 Accounting method heterogeneity and analysts’ forecasts Working paper, The University of Chicago, University of Michigan, and Harvard University Bushman, R M., and A J Smith 2001 Financial accounting information and corporate governance Journal of Accounting and Economics 32 (1/3): 237–333 https://doi.org/10.1016/S0165-4101(01)00027-1 Cadman, B., and M E Carter 2014 Compensation peer groups and their relation with CEO pay Journal of Management Accounting Research 26 (1): 57–82 https://doi.org/10.2308/jmar-50598 Cascino, S., and J Gassen 2015 What drives the comparability effect of mandatory IFRS adoption? Review of Accounting Studies 20 (1): 242–282 https://doi.org/10.1007/s11142-014-9296-5 Chen, C-W., D W Collins, T D Kravet, and R Mergenthaler 2018 Financial statement comparability and the efficiency of acquisition decisions Contemporary Accounting Research 35 (1): 164–202 https://doi.org/10.1111/1911-3846.12380 Chen, D., S Liang, and P Zhu 2012 Relative performance evaluation and executive compensation: Evidence from Chinese listed companies China Journal of Accounting Research (2): 127–144 https://doi.org/10.1016/j.cjar.2012.02.002 Cheng, A., K Jo, J Ng, and H Wu 2017 Financial statement comparability and hedge fund activism Working paper, The Hong Kong Polytechnic University Cichello, M S., C E Fee, C J Hadlock, and R Sonti 2009 Promotions, turnover, and performance evaluation: Evidence from the careers of division managers The Accounting Review 84 (4): 1119–1143 https://doi.org/10.2308/accr.2009.84.4.1119 Core, J E., R W Holthausen, and D F Larcker 1999 Corporate governance, chief executive officer compensation, and firm performance Journal of Financial Economics 51 (3): 371–406 https://doi.org/10.1016/S0304-405X(98)00058-0 Daske, H., L Hail, C Leuz, and R Verdi 2013 Adopting a label: Heterogeneity in the economic consequences around IAS/IFRS adoptions Journal of Accounting Research 51 (3): 495–547 https://doi.org/10.1111/1475-679X.12005 Dechow, P M., and I D Dichev 2002 The quality of accruals and earnings: The role of accrual estimation errors The Accounting Review 77 (s-1): 35–59 https://doi.org/10.2308/accr.2002.77.s-1.35 Dechow, P M., and R G Sloan 1991 Executive incentives and the horizon problem: An empirical investigation Journal of Accounting and Economics 14 (1): 51–89 https://doi.org/10.1016/0167-7187(91)90058-S DeFond, M L., and C W Park 1999 The effect of competition on CEO turnover Journal of Accounting and Economics 27 (1): 35–56 https://doi.org/10.1016/S0165-4101(98)00044-5 De Franco, G., S P Kothari, and R S Verdi 2011 The benefits of financial statement comparability Journal of Accounting Research 49 (4): 895–931 https://doi.org/10.1111/j.1475-679X.2011.00415.x Dikolli, S., C Hofmann, and T Pfeiffer 2013 Relative performance evaluation and peer-performance summarization errors Review of Accounting Studies 18 (1): 34–65 https://doi.org/10.1007/s11142-012-9212-9 Engel, E., R M Hayes, and X Wang 2003 CEO turnover and properties of accounting information Journal of Accounting and Economics 36 (1/3): 197–226 https://doi.org/10.1016/j.jacceco.2003.08.001 Fama, E F., and J D MacBeth 1973 Risk, return, and equilibrium: Empirical tests Journal of Political Economy 81 (3): 607–636 https://doi.org/10.1086/260061 Ferri, F 2009 Discussion of explicit relative performance evaluation in performance-vested equity grants Review of Accounting Studies 14 (2/3): 307–313 https://doi.org/10.1007/s11142-009-9092-9 Financial Accounting Standards Board (FASB) 2010 Conceptual Framework for Financial Reporting Statement of Financial Accounting Concepts No Norwalk, CT: FASB Available at: http://www.fasb.org/resources/ccurl/515/412/ Concepts%20Statement%20No%208.pdf The Accounting Review Volume 95, Number 3, 2020 366 Nam Gaver, J J., and K M Gaver 1993 Additional evidence on the association between the investment opportunity set and corporate financing, dividend, and compensation policies Journal of Accounting and Economics 16 (1/3): 125–160 https://doi.org/10.1016/ 0165-4101(93)90007-3 Gibbons, R., and K J Murphy 1990 Relative performance evaluation for chief executive officers Industrial and Labor Relations Review 43 (3): 30S–51S https://doi.org/10.1177/001979399004300303 Gong, G., L Y Li, and J Y Shin 2011 Relative performance evaluation and related peer groups in executive compensation contracts The Accounting Review 86 (3): 1007–1043 https://doi.org/10.2308/accr.00000042 Hermalin, B E., and M S Weisbach 1998 Endogenously chosen boards of directors and their monitoring of the CEO American Economic Review 88 (1): 96–118 Holderness, C G., and D P Sheehan 1988 The role of majority shareholders in publicly held corporations: An exploratory analysis Journal of Financial Economics 20: 317–346 https://doi.org/10.1016/0304-405X(88)90049-9 Holmstrom, B 1979 Moral hazard and observability Bell Journal of Economics 10 (1): 74–91 https://doi.org/10.2307/3003320 Holmstrom, B 1982 Moral hazard in teams Bell Journal of Economics 13 (2): 324–340 https://doi.org/10.2307/3003457 Holmstrom, B., and P Milgrom 1987 Aggregation and linearity in the provision of intertemporal incentives Econometrica 55 (2): 303– 328 https://doi.org/10.2307/1913238 Indjejikian, R J 1999 Performance evaluation and compensation research: An agency perspective Accounting Horizons 13 (2): 147– 157 https://doi.org/10.2308/acch.1999.13.2.147 Janakiraman, S N., R A Lambert, and D F Larcker 1992 An empirical investigation of the relative performance evaluation hypothesis Journal of Accounting Research 30 (1): 53–69 https://doi.org/10.2307/2491091 Jayaraman, S., and T T Milbourn 2012 The role of stock liquidity in executive compensation The Accounting Review 87 (2): 537–563 https://doi.org/10.2308/accr-10204 Jayaraman, S., T Milbourn, F Peters, and H Seo 2018 Product market peers and relative performance evaluation Working paper, University of Rochester Jensen, M C., and K J Murphy 1990 Performance pay and top-management incentives Journal of Political Economy 98 (2): 225–264 https://doi.org/10.1086/261677 Jenter, D., and F Kanaan 2015 CEO turnover and relative performance evaluation Journal of Finance 70 (5): 2155–2184 https://doi org/10.1111/jofi.12282 Joh, S W 1999 Strategic managerial incentive compensation in Japan: Relative performance evaluation and product market collusion Review of Economics and Statistics 81 (2): 303–313 https://doi.org/10.1162/003465399558094 Kim, J-B., L Li, L Y Lu, and Y Yu 2016 Financial statement comparability and expected crash risk Journal of Accounting and Economics 61 (2/3): 294–312 https://doi.org/10.1016/j.jacceco.2015.12.003 Lambert, R A 2001 Contracting theory and accounting Journal of Accounting and Economics 32 (1/3): 3–87 https://doi.org/10.1016/ S0165-4101(01)00037-4 Lambert, R A., and D F Larcker 1987 An analysis of the use of accounting and market measures of performance in executive compensation contracts Journal of Accounting Research 25: 85–125 https://doi.org/10.2307/2491081 Lazear, E., and S Rosen 1981 Rank-order tournaments as optimum labor contracts Journal of Political Economy 89 (5): 841–864 https://doi.org/10.1086/261010 Leone, A J., J S Wu, and J L Zimmerman 2006 Asymmetric sensitivity of CEO cash compensation to stock returns Journal of Accounting and Economics 42 (1/2): 167–192 https://doi.org/10.1016/j.jacceco.2006.04.001 Lev, B 1989 On the usefulness of earnings and earnings research: Lessons and directions from two decades of empirical research Journal of Accounting Research 27: 153–192 https://doi.org/10.2307/2491070 Li, F., M Minnis, V Nagar, and M Rajan 2014 Knowledge, compensation, and firm value: An empirical analysis of firm communication Journal of Accounting and Economics 58 (1): 96–116 https://doi.org/10.1016/j.jacceco.2014.06.003 Matsumura, E M., and J Y Shin 2006 An empirical analysis of an incentive plan with relative performance measures: Evidence from a postal service The Accounting Review 81 (3): 533–566 https://doi.org/10.2308/accr.2006.81.3.533 Matsumura, E M., and J Y Shin 2013 Relative performance evaluation: A review of managerial accounting research Journal of Management Accounting, Japan (Supplement 2): 3–12 https://doi.org/10.24747/jma.Supplement2.0_3 Murphy, K J 1999 Executive compensation In Handbook of Labor Economics, Volume 3, Chapter 38, Part B, 2485–2563 Available at: https://doi.org/10.1016/S1573-4463(99)30024-9 Nalebuff, B J., and J E Stiglitz 1983 Prizes and incentives: Towards a general theory of compensation and competition Bell Journal of Economics 14 (1): 21–43 https://doi.org/10.2307/3003535 Ozkan, N., Z Singer, and H You 2012 Mandatory IFRS adoption and the contractual usefulness of accounting information in executive compensation Journal of Accounting Research 50 (4): 1077–1107 https://doi.org/10.1111/j.1475-679X.2012.00453.x Rajgopal, S., T Shevlin, and V Zamora 2006 CEOs’ outside employment opportunities and the lack of relative performance evaluation in compensation contracts Journal of Finance 61 (4): 1813–1844 https://doi.org/10.1111/j.1540-6261.2006.00890.x Shroff, N., R S Verdi, and B P Yost 2017 When does the peer information environment matter? Journal of Accounting and Economics 64 (2/3): 183–214 https://doi.org/10.1016/j.jacceco.2017.03.005 The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 367 Sloan, R G 1993 Accounting earnings and top executive compensation Journal of Accounting and Economics 16 (1/3): 55–100 https:// doi.org/10.1016/0165-4101(93)90005-Z Smith, C W., Jr., and R L Watts 1992 The investment opportunity set and corporate financing, dividend, and compensation policies Journal of Financial Economics 32 (3): 263–292 https://doi.org/10.1016/0304-405X(92)90029-W Vrettos, D 2013 Are relative performance measures in CEO incentive contracts used for risk reduction and/or for strategic interaction? The Accounting Review 88 (6): 2179–2212 https://doi.org/10.2308/accr-50548 Wu, J S., and I Zhang 2010 Accounting integration and comparability: Evidence from relative performance evaluation around IFRS adoption Working paper, University of Rochester The Accounting Review Volume 95, Number 3, 2020 Nam 368 APPENDIX A Variable Definitions Variable Cash Compensation Equity Compensation Total Compensation RET_Firm RET_Peer ROA_Firm ROA_Peer SIC2 (SIC4) SIZE FRC SIC2-SIZE SIC2-FRC-SIZE SIC2-FRC SALES MTB Idiosyncratic Risk Regulated Industry CEO Tenure CEO Chair CEO Ownership AGE AGE_6365 CEO Turnover Selected Description The sum of salary and bonus Log-transformed in the regression The sum of the value of the restricted stock granted and stock options granted (using the Black-Scholes formula) before December 15, 2006; the sum of the grant-date fair value of stock awards and the grant-date fair value of option awards after December 15, 2006 Log-transformed in the regression TDC1 from Execucomp, log-transformed in regression TDC1 comprises salary, bonus, other annual, the total value of the restricted stock granted, the total value of the stock options granted (using the Black-Scholes formula), long-term incentive payouts, and all other compensation for fiscal years ending before December 15, 2006 (OLD_DATAFMT_FLAG ¼ 0) For fiscal years ending after December 15, 2006 (OLD_ DATAFMT_FLAG ¼ 1), TDC1 is the sum of salary, bonus, non-equity incentive plan compensation, grantdate fair value of stock awards, grant-date fair value of option awards, deferred compensation, and other compensation Own-firm stock price performance; log(1 ỵ RETANN), where RETANN is the annual compounded stock returns using CRSP monthly returns (Albuquerque 2014) Peer-firm stock price performance The average stock returns of firms in an RPE peer group Log-transformed in the regression Own-firm accounting performance The annual income before extraordinary items divided by the lagged total assets (ROA) It is the natural logarithm of plus ROA in the regressions (Albuquerque 2014) Peer-firm accounting performance The average accounting performance of firms in a certain RPE peer group Log-transformed in the regression Two-digit (Four-digit) SIC industry group The beginning-of-year market capitalization Used in RPE peer selection Financial reporting comparability, as measured by the procedure in De Franco et al (2011) See Section III for details A peer group of the top ten firms closest in size (beginning-of-year market capitalization) and in the same two-digit SIC industry group as the treatment firm A peer group based on industry, financial reporting comparability, and firm size From the same-industry firms in the top quartile of financial reporting comparability, I select the top ten (minimum five) peer firms that are closest in size to the treatment firm A peer group of the top ten financial reporting-comparability peers from the same industry as the treatment firm The natural logarithm of sales The market-to-book ratio, measured by assets (AT) plus the market value of equity (CSHO à PRCC_F) minus the book value of common equity (CEQ) minus deferred taxes and investment tax credit (TXDITC), divided by the assets (AT) Idiosyncratic risk equals the standard deviation of residuals from a regression of firm-level stock returns on industry stock returns (two-digit SIC) estimated over the previous 36 months (similar to Gong et al 2011) A dummy variable equal to for firms that are gas and electric utilities with SIC codes 4900–4939, and otherwise (Albuquerque 2009) The number of months between the CEO assignment date (BECAMECEO) in Execucomp and the fiscal yearend date Log-transformed in the regression A dummy that equals if the CEO is the board chair, as identified in the TITLEANN filed in Execucomp, and otherwise A dummy variable equal to if the CEO ownership share, SHROWN_EXCL_OPTS/(SHROUT à 1000), is lower than the median for the year across CEOs in the sample, and otherwise CEO age (AGE from Execucomp) A dummy variable equal to if CEO age is between 63 and 65, and otherwise A dummy variable equal to if the current CEO is not included in the executive/director list of the same firm in the following year, and otherwise A dummy variable equal to if a firm is disclosed as an RPE peer following the expanded SEC disclosure rule, effective December 15, 2006, and otherwise Obtained from Incentive Lab (continued on next page) The Accounting Review Volume 95, Number 3, 2020 Financial Reporting Comparability and Accounting-Based RPE in CEO Compensation Contracts 369 APPENDIX A (continued) Variable Description Same Method A dummy variable equal to if a (Selected or Unselected) peer chooses the same accounting method in depreciation, inventory, or lease, and otherwise As reported in Compustat, I retrieve DPACT_FN (depreciation method) and INVVAL (inventory valuation method) For lease, I first estimate the present value of operating lease as MRC1/(1.1) ỵ MRC2/(1.12) ỵ MRC3/(1.13) ỵ MRC4/(1.14 ) þ MRC5/(1.15 ) and the capital lease value as DCLO, both scaled by total assets Two firms are identified as users of the same lease method if both firms use: (1) both operating and capital lease, (2) operating lease only, or (3) capital lease only EC Economic comparability between firms i and j, which is captured by one of four measures as follows CompEAR-R2 is the adjusted R2 from a regression of firm i’s ROA against firm j’s ROA over the preceding 16 quarters Similarly, CompCFO-R2 (CompRET-R2) is the adjusted R2 from regressing firm i’s operating cash flow (returns) on that of firm j over the preceding 16 quarters (De Franco et al 2011) Finally, following Chen et al (2018, 190), I repeat the DKV’s estimation of comparability based on the mapping between stock returns and operating cash flows Same_SIC2 (Same_SIC4) A dummy variable equal to if a selected peer firm belongs to the same SIC2 (SIC4) group as an RPE discloser, and otherwise SALES_Gap The absolute difference in sales between an RPE discloser and a selected peer firm, divided by the sales of the RPE discloser MVE_Gap The absolute difference in market capitalization between an RPE discloser and a selected peer firm, divided by the market capitalization of the RPE discloser RET_Gap The absolute difference in annual stock returns between an RPE discloser and a selected peer firm ROA_Gap The absolute difference in ROA between an RPE discloser and a selected peer firm Volatility_ROA_Peer Standard deviation of peer-firm ROA over the previous 16 quarters (or peer-firm RET over the previous 48 (Volatility_RET_Peer) months) S&P500 A dummy variable equal to if a peer firm belongs to the S&P 500 in that year, and otherwise RPE Discloser A dummy variable indicating if the firm discloses a list of specific firms that are used as RPE peers If a firm uses an index to represent peer performance, then the firm is not classified as an RPE discloser APPENDIX B Examples of RPE in CEO Compensation Contracts Below are excerpts from CD&A reports in the proxy statements of two companies for fiscal year 2013 Pacific Gas and Electric Corp uses price-based RPE in determining executive compensation And International Paper Co applies different peer groups for accounting-based versus price-based RPE Both firms disclose their RPE peer groups Pacific Gas and Electric Corp Each year, PG&E Corporation and the Utility also identify a Performance Comparator Group that is used only for evaluating the company’s relative TSR (total shareholder return) performance to determine payouts for LTIP performance share awards In determining the composition of the Performance Comparator Group for 2013, the Committee decided that the Performance Comparator Group will include companies (1) that are categorized consistently by the investment community as ‘‘regulated,’’ as opposed to ‘‘less regulated,’’ based on analysis of revenue sources (i.e., the companies have business models similar to PG&E Corporation and the Utility), and (2) that have a market capitalization of at least $4 billion The Committee first selected companies listed on the Philadelphia Utility Index that meet these criteria and then selected additional companies that also meet these criteria A total of 12 companies were included in the 2013 Performance Comparator Group American Electric Power CMS Energy Consolidated Edison DTE Energy Duke Energy NiSource, Inc Northeast Utilities The Accounting Review Volume 95, Number 3, 2020 Nam 370 Pinnacle West Capital SCANA Corp Southern Company Wisconsin Energy Corporation Xcel Energy, Inc International Paper Co Our incentive compensation plans are designed around achievement of preestablished performance objectives that will drive improved financial performance of the Company Each year, the Committee assesses the appropriateness of the performance metrics, and makes adjustments based on the financial objectives most critical to the Company’s success ROIC (Return on Invested Capital) Peer Group ROIC is calculated as operating earnings before interest (including both earnings from continuing and discontinued operations up through the date of sale), and before the impact of special items and non-operating pension expense, divided by average invested capital Invested capital is total equity (adjusted for pension) plus interest-bearing liabilities Boise Inc Domtar Inc Fibria Celulose S.A Klabin S.A Metsa Board (formerly M-real Corp.) MeadWestvaco Corp Mondi Group Packaging Corporation of America RockTenn Company Smurfit Kappa Group Stora Enso Corp UPM-Kymmene Corp TSR (Total Shareholder Return) Peer Group Alcoa Inc Domtar Inc Dow Chemical Company E.I DuPont de Nemours & Co Fibria Celulose S.A Klabin S.A MeadWestvaco Corp Mondi Group Packaging Corporation of America RockTenn Company S&P 100 Index S&P Basic Materials Index Sappi Limited Smurfit Kappa Group Stora Enso Corp United States Steel Corp UPM-Kymmene Corp The Accounting Review Volume 95, Number 3, 2020 Copyright of Accounting Review is the property of American Accounting Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission However, users may print, download, or email articles for individual use ... changes in other aspects of the adopting firms’ financial reporting Thus, it is difficult to pinpoint the incremental effect of financial reporting comparability on the contracting value of accounting... that the use of accounting-based RPE is insignificant in determining CEO equity compensation, reaffirming that the contracting value of accounting information is prevalent in determining cash compensation. .. Using financial reporting comparability as a peer selection criterion, along with industry and size, I document the use of accounting-based RPE in determining CEO cash compensation in the U.S In