peasnell et al - 2005 - board monitoring and earnings management - do outside directors influence abnorm

Board Monitoring and Earnings Management: Do Outside Directors Influence Abnormal Accruals?K.V.. Results indicate that the likelihood of managers making income-increasing abnormal accrua

Board Monitoring and Earnings Management: Do Outside Directors Influence Abnormal Accruals?

K.V PEASNELL, P.F POPE AND S YOUNG*

Abstract: This paper examines whether the incidence of earnings ment by UK firms depends on board monitoring We focus on two aspects of board monitoring: the role of outside board members and the audit commit- tee Results indicate that the likelihood of managers making income-increasing abnormal accruals to avoid reporting losses and earnings reductions is negatively related to the proportion of outsiders on the board We also find that the chance of abnormal accruals being large enough to turn a loss into a profit or to ensure that profit does not decline is significantly lower for firms with a high proportion of outside board members In contrast, we find little evidence that outside directors influence income-decreasing abnormal accruals when pre-managed earnings are high We find no evidence that the presence of an audit committee directly affects the extent of income-increasing manipulations to meet or exceed these thresholds Neither do audit commit- tees appear to have a direct effect on the degree of downward manipulation, when pre-managed earnings exceed thresholds by a large margin Our find- ings suggest that boards contribute towards the integrity of financial state- ments, as predicted by agency theory.

manage-Keywords: corporate governance, boards of directors, abnormal accruals, earnings management

* The authors are from Lancaster University They gratefully acknowledge the helpful comments of the anonymous referee and those of Steve Lim, Gilad Livne, Scott Richardson, Lakshmanan Shivakumar, Judy Tsui, and seminar participants at Bristol, U.C Dublin, Dundee, Glasgow, Lancaster, London Business School, University of Science and Technology, Hong Kong, Feng Chia University, Taiwan, Stockholm, the Scottish Institute for Research in Investment and Finance, the 2000 Annual Meeting of the European Finance Association and the 2001 Annual Conference of the British Accounting Association Financial support was provided by the Research Board of the Institute of Chartered Accountants in England and Wales, The Leverhulme Trust, and the Economic and Social Research Council (contract No H53627500497) (Paper received June 2004, revised and accepted November 2004)

Address for correspondence: Ken Peasnell, Management School, Lancaster University, Lancaster LA1 4YX, UK.

e-mail: k.peasnell@lancaster.ac.uk

# Blackwell Publishing Ltd 2005, 9600 Garsington Road, Oxford OX4 2DQ, UK

1 INTRODUCTIONBoards of directors are widely believed to play an importantrole in corporate governance, particularly in monitoring topmanagement (Fama and Jensen, 1983) The governance debateemphasises the distinctive contribution that outside directorscan make in helping to ensure that managers act in the interests

of outside stockholders (Fama, 1980; and Fama and Jensen,1983) Prior US research indicates that outside directors influ-ence a wide range of board decisions, including CEO removal(Weisbach, 1988), negotiation of tender offer bids (Byrd andHickman, 1992) and resistance to greenmail payments (Kosnik,1987) In this paper, we test whether the influence of outsidedirectors extends to the financial reporting process In particu-lar, we examine whether the incidence of earnings managementdepends on board monitoring

Studies that examine the relation between board monitoringand financial reporting have mostly focused on blatant viola-tions of Generally Accepted Accounting Principles (GAAP) Forexample, Beasley (1996) reports that the incidence of financialstatement fraud in the US is lower for firms where the propor-tion of outside directors is relatively high Dechow et al (1996)report similar findings for firms subject to SEC accountingenforcement actions These studies provide evidence of a linkbetween gross violations of accounting standards and boardstructure However, whether this link extends to more subtleaccruals-based earnings management permissible within GAAPremains an unresolved issue that this paper seeks to address.Opportunities for earnings management arise because offlexibility permitted by GAAP The purpose of this paper is toinvestigate whether boards actively monitor and take actionsthat reduce the incidence of earnings management when theincentives for manipulations are high Our tests focus on situa-tions where performance is either poor (in which case theincentive will be to manage earnings upwards), or exceptionallygood (in which case the incentive will be to manage earningsdownwards)

We use abnormal working capital accruals to proxy for ings management Our focus is on two dimensions of boardmonitoring in the UK: the proportion of outside directors and

earn-the presence of an audit committee The UK provides an esting experimental setting for investigating these issues sincethere is greater variation in outside director representation onboards in the UK than in the US (Peasnell et al., 1999) and auditcommittees are not mandatory Our results show that when pre-managed earnings are negative or below last year’s reportedearnings, abnormal working capital accruals are less positive ifthe proportion of outsiders on the board is relatively high.Furthermore, we find that the chance of abnormal accrualsbeing sufficiently large as to result in reported earnings exceed-ing the thresholds is significantly lower for firms with a highproportion of outside board members Our findings are consis-tent with the prediction that outside directors contributetowards the integrity of financial statements However, boardsappear only to intervene in the case of income-increasing earn-ings management: we find little evidence that outside directorsare associated with income-decreasing accruals when the incen-tives to manipulate earnings downwards might be deemed to bestrong Finally, we find that the mere presence or absence of anaudit committee has no measurable impact on earnings man-agement While audit committees are voluntary in the UK, 84%

inter-of the companies in our sample had set up such a committee,which might account for this null result

The next section reviews the literature on the monitoring role

of boards of directors and develops our hypotheses Section 3presents the methods used to identify earnings managementand explains the research design The data and sample aredescribed in Section 4 Section 5 presents the empirical results.Our conclusions appear in Section 6

2 HYPOTHESIS DEVELOPMENT(i) Governance and Earnings Management

Schipper (1989) defines earnings management as purposefulintervention in the external financial reporting process, with aview to obtaining private gain for shareholders or managers.Shareholders will gain when earnings management is used tosignal managers’ private information (Healy and Palepu, 1995),

to avoid costly debt re-contracting or to reduce political costs

(Watts and Zimmerman, 1986) Managers can also use earningsmanagement to extract rents from shareholders Such gainscould take the form of increased compensation (Healy, 1985;and Holthausen et al., 1995) or reduced likelihood of dismissalwhen performance is low (Weisbach, 1988) We focus on therole of corporate governance in constraining earningsmanagement.

Two measures of board monitoring that have been widely used

in the literature are the proportion of board members classified asoutside directors (e.g., Vicknair et al., 1993) and whether the boardhas an audit committee (e.g., Pincus et al., 1989; Collier, 1993; andCollier and Gregory, 1999) The governance literature emphasisesthe role of outside directors in resolving agency problems betweenmanagers and shareholders through the creation of appropriateemployment contracts and the subsequent monitoring of manage-rial behaviour Fama (1980) and Fama and Jensen (1983) arguethat outside directors have incentives to be effective monitors inorder to maintain the value of their reputational capital Prior USresearch generally supports the prediction that board effectiveness

in protecting shareholders’ wealth is a positive function of theproportion of outsiders on the board (Weisbach, 1988;Rosenstein and Wyatt, 1990; Byrd and Hickman, 1992; Brickley

et al., 1994; and McWilliams and Sen, 1997).1

The effectiveness of the board’s monitoring activities mightalso depend on how the board is structured and organized.Boards often delegate work on important tasks to standingcommittees reporting directly to the main board (Klein, 1998).The board committee responsible for financial reporting is theaudit committee, if one exists The audit committee has specificresponsibility for the production of financial statements, andusually for communicating with the external auditor Auditcommittees are not mandatory in the UK, although listed firms

1 The board’s effectiveness at monitoring the financial reporting process will depend on the ability of outside directors to understand earnings management methods Although the level of accounting expertise, and hence monitoring effectiveness, will vary across boards, there are at least two reasons for having confidence in outside directors’ general ability in this regard First, outside directors often have a financial background For example, Peasnell et al (1999) report that over a quarter of all UK board members are professionally qualified accountants Second, outside directors frequently hold senior management positions in other large corporations As such, they are likely to be familiar with financial reporting from a senior management perspective.

are encouraged to form them (Collier, 1993; and Collier andGregory, 1999).2However, the creation of an audit committeedoes not absolve the full board from its financial reportingresponsibilities The full board will often have other reasons toconsider financial reporting issues, because accounting numbersfigure prominently in such matters as determining managementcompensation, reviews of operating results and making invest-ment decisions The incremental monitoring contribution of

an audit committee is an empirical issue on which this studyaims to shed light

The preceding discussion takes the perspective that earningsmanagement is undesirable because it is costly to shareholders.However, signalling-based motivations for earnings managementneed to be considered as well Earnings management will benefitshareholders if managers use accounting discretion to signalprivate information about future performance Subramanyam(1996) shows that, in the US, discretionary accruals predictfuture profitability and dividend changes, suggesting that man-agers do use their discretion to improve the informativeness ofearnings In such circumstances, shareholders will not expectboards to constrain earnings management We therefore startour empirical analysis by considering whether signalling canexplain discretionary accruals behaviour, before investigatingthe monitoring roles of the board and the audit committee.(ii) Earnings Management Relative to Thresholds

Recent evidence suggests that the incidence of earnings ment is particularly pronounced when earnings fall below certainthresholds Three thresholds have been considered in the litera-ture: avoiding reporting a loss; reporting a growth in profits; andmeeting the analysts’ consensus forecast Burgstahler and Dichev(1997) and Degeorge et al (1999) find that there is a higher-than-expected frequency of firms in the US with slightly positive

manage-2 Collier (1993) provides evidence on the factors affecting the formation of audit committees in the UK prior to the governance recommendations set out in the Cadbury Report (1992) Since the London Stock Exchange’s decision to adopt the Cadbury recommendations on board composition and audit committee formation, the number of firms with such committees has increased dramatically However, unlike in the US, firms still have the option not to form an audit committee As we show later in the paper, a minority of firms exercise their option not to form an audit committee.

reported earnings (and earnings changes) and a expected frequency of firms with slightly negative reported earn-ings (and earnings changes) The same pattern has been observed

lower-than-in the UK (Gore et al., 2002) Such discontlower-than-inuities lower-than-in the tions are consistent with managers trying to beat the benchmarks

distribu-in question.3 A key issue is how managers decide which mark to try to beat when the benchmarks conflict Degeorge et al.(1999) report that there appears to be a hierarchy to the bench-marks, with firms behaving as if reporting a profit is of mostimportance, followed by reporting growth in earnings, with meet-ing analysts’ forecasts mattering only if the other two thresholdshave been met We focus on avoiding losses and declines in profitbenchmarks since as well as being more important, we are able toinclude in our analysis the many firms for which consensus fore-casts are not available, and thereby to achieve more powerful tests.There are a number of possible reasons for such threshold-targeting earnings management behaviour One possibility isthat capital market participants’ implicit contracts with manage-ment are defined in terms of these simple thresholds Forexample, there is considerable anecdotal evidence of share-holders increasing their monitoring activities when a loss or adecline in earnings is reported, with significant knock-on costsfor management in the form of reduced compensation and anincreased probability of dismissal Consistent with this view, USresearch reveals that firms subject to shareholder activism tend

bench-to be characterised by poor earnings performance (Karpoff,1998) Another reason might be the fear that failing to meet athreshold will result in a large decline in stock price.4Whatever,the reason, Dechow et al (2000) report that attempts by USfirms to avoid reporting losses and earnings declines appear to

be driven by managers’ desires to opportunistically delayreporting poor performance Thus, if board monitoring isassociated with earnings management, we expect that our

3 Other factors might play a role as well For example, Beaver et al (2003) show that, in the US, the asymmetric treatment of income taxes and special items for profit and loss firms can account for about two-thirds of the discontinuity in the distribution of earnings.

4 For example, Skinner and Sloan (2002) have found that growth stocks in the US tend

to exhibit an asymmetrically large negative price response to negative earnings surprises.

ability to detect such a relation will be greatest close to thresholdpoints Consequently, we condition our empirical tests on theproximity of pre-managed earnings to zero earnings and zerochange in earnings We predict that board monitoring willconstrain income-increasing earnings management when pre-managed earnings undershoot these thresholds.

The preceding discussion focuses on the link between boardmonitoring and the incidence of income-increasing earningsmanagement Earnings management, however, is not restrictedsolely to income-increasing behaviour Prior research indicatesthat, under certain conditions, managers engage in income-decreasing earnings management For example, Degeorge et al.(1999) find that managers appear systematically to manipulatereported earnings downwards when pre-managed earningsexceed threshold earnings by a substantial amount.5 Similarly,Healy (1985), Gaver et al (1995) and Holthausen et al (1995)report evidence of income-decreasing accounting choices by USfirms when managers’ accounting-based bonuses are at theirmaximum There are several plausible explanations for theseincome-decreasing accounting choices One possibility is thatmanagers prefer to shift abnormal positive earnings forward

in time in order to make the thresholds easier to cross andtargets easier to meet in the future Another possibility is thatmanagers are reluctant to report large gains because of fearsthat it will result in increased earnings-based performance tar-gets for them in the future If downward manipulation imposessignificant costs on external parties, boards should be as con-cerned with income-decreasing manipulations as they are withincome-increasing ones Therefore, we also test whether boardmonitoring is negatively associated with income-decreasingearnings management when pre-managed earnings exceedthese thresholds by a large margin

5 The big bath hypothesis predicts income-decreasing earnings management when managed earnings fall well below the threshold We have examined the issue of big bath accrual choices for our sample Untabulated results provide no indication that our below-target firms were engaged in big bath accounting One reason might be that the incidence of big baths is dependent on other factors, such as whether or not there has been a change in management This is an issue that falls outside the scope of our study and is therefore an aspect of earnings management that we do not investigate any further in this paper.

pre-Our empirical tests explicitly examine both upward anddownward manipulations However, we anticipate that boardswill have asymmetric loss functions with respect to earningsmanagement because the penalties associated with overstatingearnings (e.g., loss of reputation) are likely to exceed the costs

of understating earnings This leads us to conjecture that theincentives for boards to monitor income-increasing earningsmanagement are greater than those to monitor income-decreasing earnings management Whether this is the caseremains an empirical issue on which we aim to shed light inthis paper

(iii) The Intervening Effect of Managerial Share Ownership

The need for board monitoring depends on the extent to whichmanagers’ interests diverge from those of shareholders and otherinvestors We expect share ownership by managers to lead to acloser alignment of interests US research has found that manage-rial ownership is associated with lower levels of earnings manage-ment (Dhaliwal et al., 1982; and Warfield et al., 1995) Also, there isevidence in both the US and the UK of a negative associationbetween managerial share ownership and both the proportion ofoutside board members (Weisbach, 1988; and Peasnell et al., 2003)and the presence of an audit committee (Pincus et al., 1989; andCollier, 1993) This prior research suggests that the demand forboard monitoring declines with managerial ownership and thatmanagerial ownership interacts with board monitoring of earningsmanagement More precisely, we hypothesize that the constrainingassociation between earnings management and (i) the proportion

of outside directors and (ii) the existence of an audit committee will

be more pronounced when the level of managerial share ship is low

owner-3 METHODOLOGY(i) Measuring Earnings Management

Manipulation of operating accruals is likely to be a favoredinstrument for opportunistic earnings management because

they generally have no direct cash flow consequences6 and arerelatively difficult to detect A more costly method to manipulateearnings is by changing the way the firm does business Thefirm could, for example, boost reported profit by cutting back

on advertising and research and development (Bushee, 1998).There are many more possibilities: selling assets it would other-wise keep (Bartov, 1993; and Poitras et al., 2002); cutting back

on staff development and essential equipment maintenance;channel stuffing; the list is almost endless All such actions arecostly, in the sense that they have negative effects on the firm’sfuture cash flows Such manipulations reduce the value of thefirm and as such are more costly than mere accounting manip-ulations We therefore expect that the manipulation of accrualswill be the instrument chosen first, before management resorts

to more costly ones involving real changes in investment andoperating activities We focus only on the accountingmanipulations

We use abnormal accruals as our proxy for earnings ment.7 Several methods have been proposed in the literaturefor separating operating accruals into abnormal (or dis-cretionary) and normal (or non-discretionary) components.(See Dechow et al., 1995, for a review of these models.) Themost frequently used methods are the Jones (1991) model andthe modified-Jones model (Dechow et al 1995) Both methodsinvolve estimating parameters for normal accrual activity byregressing a measure of accounting accruals on proxies fornormal business activity These estimated normal accrual para-meters are then combined with event-period data to generateestimated unexpected accrual activity

manage-We estimate abnormal accruals using the modified-Jones J) model Dechow et al (1995) present evidence that the m-Jmodel is more powerful at detecting sales-based manipulations

(m-6 Accruals can have tax effects, but these are likely to be of second-order importance in the UK, given that financial reporting and tax accounting are governed by different measurement rules.

7 Accruals-based measures are theoretically appealing because they aggregate into a single measure the net effect of numerous recognition and measurement decisions, thereby capturing the portfolio nature of income determination (Watts and Zimmerman, 1990) Moreover, to the extent that boards treat earnings from operations and cash from operations as key indicators, a simple comparison of the two numbers will provide them with direct insight into the aggregate effect of operating accruals.

than the original Jones (1991) model.8 We estimate the m-Jmodel on a cross-sectional basis in order to maximize our sam-ple size and to avoid the survivorship bias problem inherent inthe use of a firm-specific time-series approach (Becker et al.,1998; and DeFond and Subramanyam, 1998) In contrast toprior studies that model total operating accruals, we focus spe-cifically on the working capital component Prior studies havegenerally defined total operating accruals as working capitalaccruals plus a key long-term accrual, depreciation We ignorethe long-term component of total accruals, for several reasons.

As Beneish (1998) points out, depreciation offers limited tial as a tool for systematic earnings management since changes

poten-in depreciation policy cannot be made very frequently withoutattracting adverse attention from the auditor or investors.However, this is not the case with some other long-termaccruals, such as defined benefit pension obligations and certainenvironmental liabilities where the amounts are very sensitive tokey assumptions Nevertheless, we follow prior practice andexclude these non-depreciation long-term accruals from ouranalysis because of their very complexity No prior study hasincluded long-term accruals other than depreciation Lacking amodel of what drives those other long-term accruals, it is diffi-cult to distinguish between normal and abnormal long-termaccruals The power of our tests will be adversely affected bythis omission

The m-J model parameters are estimated with the followingcross-sectional OLS regression:

where WCit is working capital accruals for firm i in year t,defined as the change in non-cash current assets minus thechange in current liabilities,
REVit is the change in revenuefor firm i in year t, !0and !1are regression coefficients, and
iisthe (assumed i.i.d.) regression residual Model estimates areobtained for each industry and year combination For estima-tion purposes, we scale the variables by beginning-of-periodtotal assets, TAi,t
1, and suppress the intercept.9 Industry-yearportfolios with less than ten observations are excluded from theanalysis.

Following Dechow et al (1995), abnormal accruals (AA) isdefined as follows:

REVit by
RECit is designed to help capture credit salesmanipulations by treating ^!1
RECit as a component of abnor-mal accruals (Dechow et al., 1995)

Our empirical tests seek to isolate the incentives for earningsmanagement by comparing pre-managed earnings (PMEt) withbenchmark or ‘target’ earnings levels As explained in sub-sec-tion 2(ii) above, we employ two benchmarks for current periodearnings: (i) zero and (ii) earnings reported in the previous year(EARNt
1) We predict that managers will seek to manipulateearnings upwards when either PMEt <0 or PMEt <EARNt
1

We also predict that managers will seek to manipulate earningsdownwards when PMEtexceeds the zero and EARNt
1 thresh-olds by a large margin, so as to increase the likelihood of meet-ing the respective threshold in future periods We then examinewhether the level of board monitoring influences the amount ofupward (downward) earnings management when pre-managedearnings undershoot (significantly exceed) target earnings

9 We have also replicated all the tests reported in the paper using a version of (1) that replaces the first term on the right-hand side with a constant The results are substan- tially the same.

We use cash flow from operations (CFOt) as the instrument forPMEt As we explain in the Appendix, the problem with using the

‘obvious’ measure of PME, namely EARNt
AAt, is that it issubject to a backing-out problem, where error in estimating AAwill lead automatically to equal error in the estimation of PME.This in turn could possibly induce spurious correlation between

AA and PME We have repeated all analyses using EARNt
AAtwith substantially similar – indeed somewhat stronger – results tothose reported below However, because of the possibility of spur-ious correlation, we err on the side of caution and report only theresults that use CFO as the proxy for PME

(ii) The Model

We test the impact of board monitoring on the extent of ings management by estimating the following pooled OLSregression separately for each earnings threshold (zero andEARNt
1, respectively):

earn-AAit¼ 0þ 1BELOWitþ2HIGHitþ 3OUTit

þ 1OUTit BELOWitþ 2OUTit HIGHit

in a large one In other words, the underlying model is in deflatedform BELOW is an indicator variable taking the value of one

if our proxy for pre-managed earnings is below thresholdearnings – PMEt< 0 and PMTt <PMTt
1 respectively – andzero otherwise Holding board monitoring effects constant, wepredict that 1will be positive due to income-increasing earningsmanagement behaviour

HIGH is an indicator variable taking the value of one ifour proxy for pre-managed earnings (CFO) exceeds zero orEARNt
1, as the case may be, by a specified margin and zero

otherwise For the regressions where PME is benchmarkedagainst zero, we define HIGH as one if PME for firm i in period

t exceeds the 75th percentile of the distribution of positive managed earnings and zero otherwise For the regressionswhere PMEt is benchmarked against last year’s reported earn-ings, HIGH is set equal to one if PMEtminus EARNt
1for firm i

pre-in period t exceeds the 75thpercentile of that variable’s tion of positive pre-managed earnings changes, and zero other-wise In the absence of board monitoring, we are expectingfirms to engage in income-decreasing earnings management;

distribu-we therefore predict that 2will be negative

OUT is the proportion of outside board members and AC is

an indicator variable taking the value of one if the firm has anaudit committee and zero otherwise If outside directors andaudit committees fulfill a monitoring role and seek to reduceearnings management, we predict that 1 and 3 will be nega-tive and that 2and 4will be positive

Equation (3) also includes proxies for other factors that mightaffect the level of earnings management These include otherelements of the governance system such as board size, the pre-sence of a dual Chairman and CEO, ownership structure andauditor type, other earnings management stimuli such as lever-age and financial performance CFO (scaled by total assets) isincluded as a control for errors in the measurement of abnor-mal accruals We predict CFO will be negatively associated with

AA As we explain in the Appendix, we also use CFO to proxyfor PME; hence any hypothesis about the sign of the CFOcoefficient is therefore a joint one involving both governanceand measurement error factors Our monitoring theory pre-dicts a zero association between AA and PME when the latter

is outside the critical zones identified by the BELOW and HIGHindicator variables Measurement error considerations and ourmonitoring theory therefore both predict that the coefficient onCFO will be non-positive

4 DATA AND SAMPLEOur empirical tests are conducted using data for UK listed firmswith fiscal year ends between June 30, 1993 and May 31, 1996

The m-J model is estimated for each industry (Datastreamlevel-6) and year combination, using all firms on the DatastreamActive and Research files with available accruals data.10The num-ber of firms used to estimate the models in 1993, 1994 and 1995were 620, 651 and 657, respectively The maximum number ofobservations for any given industry-year combination is 56(general engineering, 1995) while the mean (median) estimationportfolio size is 21 (18) observations.

Board data were hand-collected using the following samplingprocedure For each year we selected the largest 1000 listedfirms based on market capitalization at December 31, asreported in the London Share Price Database We then excludedall financial firms (SIC codes 60–69) because they have funda-mentally different accruals processes that are not captured bythe m-J model We also excluded all regulated utilities (SICcodes 40–44, 46, 48–49) because of differences in their incen-tives and opportunities to manage earnings The PriceWaterhouse Corporate Register is used as the source of boardcomposition data, and firms’ annual reports are used to identifythe presence or absence of an audit committee.11The CorporateRegister is also the main source of managerial share ownershipdata, supplemented where necessary by annual reports TheCorporate Register reports the number of shares held by eachexecutive board member, classified into beneficial and non-ben-eficial categories, together with the total number of the firm’sshares outstanding at the fiscal year-end Since it is often impos-sible to determine the voting and control rights of non-benefi-cial shareholdings, we restrict our definition of ownership tobeneficial shareholdings only For the same reason, we alsoexclude all family and family-related holdings, together withshares held in employee pension and share option plans Data

on external share ownership were collected from the StockExchange Official Yearbook Data on auditor identity were

10 The sample of firm-years used to test the association between earnings management and board effectiveness is a subset of the firm-year observations used to estimate the m-J model.

11 The Corporate Register is published quarterly by Hemmington Scott Ltd and includes data for all firms listed on the London Stock Exchange Hemmington Scott constructs their board composition database using information from the London Stock Exchange and Reuters.

collected from the Corporate Register All remaining data, ing that used to calculate abnormal accruals, were obtainedfrom Datastream.

includ-The final sample consists of 1,271 firm-years with the plete data needed to estimate equation (3) The sample com-prises 559 firms drawn from 34 Datastream level-6 industrialgroups.12 The distribution of firms across sample years is asfollows: 125 firms are included in only one year, 156 firms areincluded in two years, and 278 firms have data for each of thethree sample years Annual sample sizes are 406, 453 and 412for 1993, 1994 and 1995, respectively When pre-managedearnings are benchmarked against zero, 134 firm-year observa-tions (10.7%) are classified as below target, while 260 (20.5%)are classified as well-above target Alternatively, when pre-man-aged earnings are benchmarked against EARNt
1, 269 firm-year observations (21.6%) are classified as below target, while

com-177 (13.9%) are classified as well-above target The difference inthe below-target sample sizes reflects the fact that fewer firmsreport losses than earnings declines

Descriptive statistics for the explanatory variables arereported in Table 1 The average board in our sample containseight directors, of which approximately 43% are outsiders.Audit committees are present in 85% of firm-year observations.The fact that so many firms in our sample have audit commit-tees suggests that the inclusion of an indicator variable in ourregressions may not be sufficient to capture the effectiveness ofthis particular governance mechanism

Regarding the additional control variables, the roles of man and chief executive are combined in 24% of cases Thetypical sample firm has median managerial (institutional) own-ership of 2% (19%) These levels are comparable to thosereported in previous studies of the UK (e.g., Short andKeasey, 1999) Over half of the sample (53%) has at least oneexternal blockholder whose stake exceeds 10%

chair-12 Firms are distributed fairly evenly across the industrial groups The largest industry

is Engineering with 115 firm-year observations, representing 9% of the final sample No other industrial group contains more than 6% of the sample.

5 RESULTS(i) A Signalling Explanation

As explained in sub-section 2(ii), before considering whetherboard monitoring influences the extent of earnings manage-ment, we must first establish that such earnings management

is not in shareholders’ interests If managers are using theiraccounting discretion to signal private information about future

Table 1 Descriptive Statistics for Explanatory Variablesa

LEV is total debt over total assets.

CFO is operating cash flow over beginning-of-year total assets.

AC is an indicator variable taking the value of one if the firm has an audit committee and zero otherwise.

BLOCK is an indicator variable taking the value of one if the firm has an external stockholder owning 10% of the outstanding shares and zero otherwise.

DUAL is an indicator variable taking the value of one if the roles of Chairman and CEO are combined and zero otherwise.

AUD is an indicator variable taking the value of one if the firm has a Big 5 auditor and zero otherwise.

REL is an indicator variable taking the value of one if earnings before abnormal accruals, scaled by beginning-of-period total assets, are less than industry median scaled reported earnings and zero otherwise.

All values are measured at fiscal year-end, unless otherwise indicated.

earnings performance, then it is unclear why boards shouldseek to prevent such behaviour We therefore present evidence

on the extent to which abnormal accrual activity is attributable

to signalling If a signalling rationale explains abnormal accrualactivity, then we would expect to observe a positive associationbetween current period abnormal accruals and future earningschanges

We test this prediction using a logistic regression, where thedependent variable takes the value of one if our proxy for one-period-ahead pre-managed earnings is higher than reportedearnings in the current period and zero otherwise:

log probðPMEtþ1 >EARNtÞprobðPMEtþ1  EARNtÞ

¼ 0þ 1CFOtþ 2NAtþ 3AAt: ð4Þ

In addition to abnormal accruals, the set of explanatoryvariables also includes measures of current period operatingcash flow and normal accruals Separate regressions are esti-mated for the below- and above-threshold sub-samples Theresults for equation (4) appear in Table 2 Regardless ofwhether pre-managed earnings are benchmarked against zero(Panel A) or EARNt
1 (Panel B), the estimated coefficients

on AA are significantly negative when PME is above the hold and indistinguishable from zero when PME is below thethreshold These findings do not support a signalling explan-ation We therefore conclude that it is appropriate to proceed asthough abnormal accruals are potentially costly to marketparticipants

thres-(ii) Board Monitoring

Equation (3) provides our basic model for capturing the impact

of board monitoring on abnormal accrual activity We considertwo versions of the model, M1 and M2 M1 includes only ourtest variables OUT and AC, together with the indicator variablesBELOW and HIGH and their interactions with the test vari-ables M2 is the full model, inclusive of the vector of controlvariables

The control variables are as follows DUAL is an indicator ofCEO power, set equal to one if the roles of chairman and chiefexecutive are combined, and zero otherwise BRDSIZE is thenumber of directors and serves as a measure of board effective-ness BRDOWN is the proportion of shares owned by insidedirectors and proxies for the congruence of interests of share-holders and management We use two indicators of the mon-itoring activities of important shareholders: INSTOWN is the

Table 2 Logistic Regression Models Relating Future Pre-managed Earnings Changes to Measures of Current Period Accruals and Cash Flowsa

log probðPMEtþ1>EARNtÞ

probðPME tþ1  EARN t Þ

¼  0 þ  1 CFO t þ  2 NA t þ  3 AA t

Panel A: Earnings Threshold ‡ 0

PME below threshold

Panel B: Earnings Threshold ‡ EARNt{1

PME below threshold

b All probability values are for two-tailed tests.

AA is current period abnormal accruals scaled by beginning-of-period total assets, estimated using the m-J model.

NA is current period normal accruals scaled by beginning-of-period total assets, ured as the difference between total working capital accruals and AA.

meas-CFO is current period operating cash flow scaled by beginning-of-period total assets.