Journal of accounting, auditing finance tập 27, số 03, 2012 7

In other words, if there are no benefits to the supply of nonaudit services, but onlypotential costs in terms of lower audit quality due to loss of audit independence, then limit-ing the

Auditing & Finance 27(3) 299–324

Ó The Author(s) 2012 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0148558X11409154

http://jaaf.sagepub.com

Reexamining the Relationship

Between Audit and Nonaudit

Fees: Dealing With Weak

Instruments in Two-Stage

Least Squares Estimation

Abstract

The authors introduce some new econometric tests and techniques for identifyingand overcoming the problem of weak instruments in the context of joint provision ofaudit and nonaudit fees The authors use this context because identifying appropriateinstruments is difficult due to the lack of theoretical guidance as well as due to the diffi-culty in intuitively identifying instruments that satisfy the econometric requirements Theauthors introduce a battery of empirical tests based on recent developments in econo-metrics to test for the appropriateness of the instruments The authors then illustratetwo approaches of using instruments from existing data: the size industry average portfo-lio approach and the synthetic-instrument approach Although the approach usingsynthetic instruments sidesteps issues of identifying proxies with desirable properties, itrequires some stringent assumptions that cannot be directly tested However, as amethodological alternative, this approach can be used for robustness tests The authorsfind that when the instruments are not weak, audit and nonaudit fees are positivelyassociated This relationship holds for audit and tax-related nonaudit fees as well Overall,the evidence suggests the existence of economies of scope benefits from the jointsupply of audit and nonaudit services Methodologically, the authors illustrate the impor-tance of testing the appropriateness of the instruments utilized when accounting forendogeneity

Keywords

audit fees, nonaudit fees, instruments, two-stage least squares estimation

A number of studies show a positive association between audit and nonaudit fees, ing that knowledge spillovers occur from one service to the other (see, for example, Abdel-Khalik, 1990; Bell, Landsman, & Shackelford, 2001; Davis, Ricchiute, & Trompeter, 1993;DeBerg, Kaplan, & Pany, 1991; Hay, Knechel, & Wongm, 2006; Palmrose, 1986; Simunic,1984) These studies use the ordinary least squares (OLS) procedure to estimate the associ-ation between audit and nonaudit fees However, Whisenant, Sankaraguruswamy, andRaghunandan (2003) consider the joint determination of audit and nonaudit fees using thetwo-stage least squares (2SLS) procedure and show no association between audit and non-audit fees In this study, we reexamine the association of audit and nonaudit fees and recon-cile the ‘‘seemingly’’ conflicting results.

suggest-Reconciling the conflicting results on the existence of knowledge spillover effectsbetween audit and nonaudit services is important both for gaining theoretical insights aswell as guiding policy on the provision of nonaudit services On the theoretical front,Simunic (1980) considers an analytical model of joint provision of audit and nonaudit ser-vices and shows that audit and nonaudit fees should be positively associated when there areeconomies of scope, that is, knowledge spillovers.1The intuition is that with economies ofscope providing audit and nonaudit services, only some of the savings of auditor’s costswill be passed onto clients This in turn will enhance the profitability of both segments ofthe auditor’s business.2 The empirical audit- and nonaudit-fee models relate the auditors’supply of effort to the fees Based on the knowledge spillover hypothesis, in the audit-feemodels, a positive association between audit and nonaudit fees would imply the existence

of a spillover effect, and no association would imply that the audit and nonaudit efforts areindependent and as such imply no knowledge spillover effects Accordingly, Whisenant

et al (2003) finding of no association between audit and nonaudit fees lead them to clude that,

con-The findings are not consistent with the existence of economies of scope for the jointperformance of audit and non-audit services Given the ongoing debate over thelevel of allowed non-audit services by auditors, the argument for joint provision ofaudit and non-audit services is less justified than if the joint-supply benefits had beendocumented (p 722)

In other words, if there are no benefits to the supply of nonaudit services, but onlypotential costs in terms of lower audit quality due to loss of audit independence, then limit-ing the nonaudit services would be a reasonable policy.3 Thus, reconciling the seeminglyconflicting findings of the association between audit and nonaudit services is important.Reconciling the conflicting results is also important in light of the recent developmentswith respect to nonaudit services.4Although nonaudit fees from management advisory ser-vices have been curtailed since 2002, recently the Public Company Accounting OversightBoard (PCAOB) has initiated efforts to curtail auditors of public companies from supplyingcertain types of nonaudit tax services Reacting to the extensive media coverage and con-gressional actions pertaining to tax shelters promoted by auditors for their public companyclients, the PCAOB conducted a roundtable in July 2004 and stated that ‘‘the Board hasdetermined that it is appropriate to consider the impact of tax services on auditor indepen-dence’’ (see PCAOB, 2004a) In December 2004, the PCAOB (2004b) proposed rules tocurb certain types of nonaudit tax services for their public company clients.5To summarize,although knowledge spillover is the benefit, loss of auditor independence is the cost of

providing nonaudit services It is therefore important to document whether there is a edge spillover benefit at all.

knowl-Audit and nonaudit services/fees are jointly determined because factors such as agencycost, complexity, size, risk, performance, and auditor characteristics influence the demandfor both (see Whisenant et al., 2003) It is well known that when audit and nonaudit feesare jointly determined, using the OLS procedure leads to inconsistent/biased coefficientestimates (Wooldridge, 2002) Accordingly, 2SLS procedure should be used Recent devel-opments in econometrics show that using weak instruments in 2SLS can lead to inappropri-ate conclusions (see Ebbes, Wedel, & Bockenholt, 2009, for an excellent summary) Assuch, in the context of joint provision of audit and nonaudit fees, we reconcile the mixedfindings by assessing the appropriateness of instruments, that is, whether the instrumentsare weak

We estimate the audit- and nonaudit-fee model using 2SLS with the instruments used inprior research, that is, reporting lag and new financing activity for audit and nonaudit fees,respectively We do this for three separate periods: 2000, 2001, and 2002 to 2006 We con-sider year 2000 separately so as to provide a benchmark of our sample with that of priorresearch that also considers the same year We consider 2002 onward separately because ofthe auditor reforms that went into effect from 2002 We find that audit and nonaudit feesare not associated with each other, that is, the findings in prior research that use 2SLS arerobust for each period However, the tests on the strength of the two instruments reveal thatthe instruments are ‘‘weak.’’

We thus consider additional instruments to examine the joint determination of audit andnonaudit fees with 2SLS We use the industry average audit and nonaudit fees as additionalinstruments for audit and nonaudit fees, respectively There is a rich tradition of usingindustry averages as instruments in the financial economics literature: for instance, Lev andSougiannis (1996), Bertrand and Mullainathan (2001), Murphy (2000), Hanlon, Rajagopal,and Shevlin (2003) In particular, we use the mean/median of the 10 closest firms in terms

of total assets operating in the same Fama–French industry group as the instrument Weshow that these instruments are not weak, and with these instruments, audit and nonauditfees are positively associated with each other This suggests that knowledge spillovereffects between audit and nonaudit services exist, that is, there are benefits to the joint pro-vision of audit and nonaudit services Although our research design does not consider thecost of such provision of joint services in terms of conflicts of interests and poor account-ing quality, our results suggest that the benefits should be compared with the costs in arriv-ing at policy decisions In other words, it is not the case that there are no benefits and onlycosts of allowing such joint provision of audit and audit-related services

We also consider synthetic instruments as additional instruments for audit and nonauditfees—synthetic instruments are instruments derived from the transformation of endogenousand exogenous variables, that is, the instruments are synthetically derived (see Lewbell,1997) We do this because even though there is a rich tradition of using industry averages

as instruments in the financial economics literature, recently Larcker and Rusticus (2010)argue that such methods may not be appropriate.6 It is widely recognized in the econo-metrics literature that finding suitable instruments may be difficult if not impossible; thus,creating such synthetic instruments are important for obtaining consistent and efficient esti-mates Ebbes et al (2009) and Lewbell (1997) show that candidates for synthetic instru-ments are (a) the product of the endogenous and exogenous variables and (b) the secondmoment of the endogenous variables We consider additional synthetic instruments to showthat if instruments are weak then we again obtain the result of no association between audit

and nonaudit fee We find that for the calendar year when the synthetic instruments based

on the product of the exogenous and endogenous variables are weak, the positive tion between the audit and nonaudit fee is also not statistically significant When we con-sider the second moment of the endogenous variables as additional instruments, theinstruments are not weak and the audit and nonaudit fees are positively associated Thisprovides indirect evidence on the importance of considering instruments that are not weak

associa-in assessassocia-ing the relationship between audit and nonaudit services

Finally, we examine the relationship between audit fees and tax-related nonaudit fees byconsidering the industry average–based instruments for these endogenous variables for theperiod of 2002 to 2006 We find that audit and tax-related nonaudit fees are positively asso-ciated This shows the existence of knowledge spillover effects across audit and tax ser-vices As discussed earlier, the PCAOB is considering the relationship between thenonaudit, tax service, and audit independence Kinney, Palmrose, and Scholz (2004) useaccounting restatements as a proxy for audit quality and show that the provision of nonau-dit services is not positively related to accounting restatements Although our researchdesign/test does not provide evidence on the relationship between the joint provision ofaudit and tax-related services, and auditor independence/quality, our evidence providesinsights into whether there are economies of scope in the provision of such services.Overall, our objective is to introduce some new econometric techniques using the con-text of joint provision of audit and nonaudit fees We use the ‘‘joint provision of audit andnonaudit’’ service setting because identifying appropriate instruments is difficult due to thelack of theoretical guidance as well as intuitively identifying instruments that satisfy theeconometric requirements We use a battery of empirical tests, developed recently in theeconometric literature, to test for the appropriateness of the instruments We then illustratetwo approaches of using instruments from data that are already available: the industry aver-age approach, which has been used in previous research, and the synthetic instrumentapproach Although the approach using synthetic instruments sidesteps issues of identifyingproxies with desirable properties, it requires some stringent assumptions that cannot bedirectly tested.7 However, as a methodological alternative, this approach can be used forrobustness tests More importantly, we use this approach to demonstrate the importance ofidentifying weak instruments

The rest of the article is organized as follows: Section titled ‘‘Summary of EconometricIssues with Instruments’’ provides a summary of the econometric issues with weak instru-ments, section titled ‘‘Empirical Analysis’’ presents the empirical analysis, and sectiontitled ‘‘Concluding Remarks’’ contains some concluding remarks

Summary of Econometric Issues With Instruments

In this section, we summarize the issues with instrumental variable (IV) methods drawing

on literature from statistics and econometrics To highlight the issues with IV methods,consider the following equation

where y is the dependant variable, x is the explanatory variable, and v is the structural errorterm If y is the audit fee and x is the nonaudit fee, and audit and nonaudit fees are jointlydetermined, then the nonaudit fee and the structural error are potentially correlated: theexplanatory variable x is not independent of the structural error term v It is well known

that estimating Equation 1 using OLS leads to an inconsistent estimate of a1 One approach

to mitigate this problem is to choose an IV, z, that is not correlated with the structural errorterm, v, and is correlated with the explanatory variable, x The 2SLS uses the IV approach.8

In the first step, the endogenous variable x is regressed on an instrument z as follows:

In the second stage, Equation 1 is estimated with x being replaced by the predictedvalue of the explanatory variable from Equation 2, ^x The probability limit (plim) of theestimator is given by (see Wooldridge, 2002) plimð^a1Þ5a11½covðz; vÞ=covðx; zÞ wherecov(a, b) is the covariance of a and b It follows that if the instrument z is correlated with xbut not with v then the second term converges to zero for sufficiently large samples Inother words, we can obtain consistent estimates of a1, if cov(z, v) is close to zero andcov(x, z) is nonzero and sufficiently large Correspondingly, an inappropriate instrumentwould have large cov(z, v) and/or sufficiently small cov(x, z), and such weak instrumentswill not provide consistent estimates The probability limit of the coefficient estimate pro-vides the basis for establishing the appropriateness of the instruments For this purpose,recently, various tests have been developed in the econometrics literature to validate theappropriateness of the instruments We provide a brief discussion of these tests

Relevance of Instruments

Partial R2 The bias introduced in the estimate using the 2SLS procedure depends on thecovariance of the IV and the endogenous variable (see Equation 1) Specifically, Larckerand Rusticus (2007) show that the IV method is valid only if the following inequality issatisfied:R2

suffi-Partial F test Estimates based on the 2SLS procedure are likely to have larger standarderrors than estimates obtained from the OLS procedure (Bartels, 1991) The asymptoticstandard error of the estimates obtained from the 2SLS procedure is greater than thatobtained from the OLS procedure: the difference in the asymptotic standard error increaseswith decreases in the correlation between the endogenous variable x and the instrument z

In other words, as the correlation between x and z approaches zero, the standard error usingthe 2SLS procedure approaches infinity.10 To see this, consider the following equation inplace of Equation 1:y5ao1a1x11a2x21 1akxk1u, where xkis the endogenous vari-able Let ^a to be the vector of 2SLS estimators using instruments Z The asymptotic var-iance (Avar) of the 2SLS estimator is Avarð^aÞ’ s2=S ^ k, where S ^ k is the sum ofresiduals of the first-stage regression, that is, the regression of the endogenous variable onall exogenous variables (Wooldridge, 2002) The denominatorS ^ k5S ^STkð1  ^R2

kÞ where

S ^STk is the total sum of squares of ^xk and ^R2

k is the R2of the regression of the predictedvalue of the endogenous variable ^xk on the exogenous variables in the second stage It

follows that if (1 ^R2

k) is small, Avar can be very high (Wooldridge, 2002) Simply put, ifthe endogenous variable is only weakly associated with the instruments over and above theother exogenous variables, then S ^STk is small leading to higher standard errors in thesecond-stage regression Weak instruments lead to small (1 ^R2k) and, most importantly, anonzero correlation between the endogenous variable, and the instruments alone is not suffi-cient to ensure that (1 ^R2

k) is not small The partial correlation in the first-stage regressionbetween the endogenous variable and the IV, that is, the correlation between the endogenousvariable and the instruments, after accounting for the correlation between the endogenousvariable and all other exogenous variables in the model needs to be sufficiently large Thus,good instruments need to have a sufficiently large partial F statistic in the first-stage regres-sion: the partial F statistic measures the strength of the correlation between the endogenousvariable and the instruments over and above that of all other exogenous variables

The partial R2and partial F tests do not provide an indication of how much correlation

is good enough Although Staiger and Stock (1997) provide F values that are considereddesirable for different sizes of finite samples, recently there have been various test statisticsdeveloped for validating the appropriateness of the instruments These tests are describedbriefly below

Underidentification Test

The underidentification test provides a canonical correlation version of the partial R2and Fstatistic tests That is, the endogenous variable x must be correlated with the IV z, whenthere are more than one endogenous and IVs The canonical correlations represent the cor-relations between the endogenous variable x and linear combinations of z The squaredcanonical correlations with multiple endogenous variables and instruments are calculated asthe eigenvalues, and the underidentification test requires that all canonical correlations aredifferent from zero Conversely, if one or more of the canonical correlations is zero, thenthe model is underidentified For the case where the errors are independent and identicallydistributed (i.i.d.), the Anderson’s (1951) canonical correlation LM statistic is computed.For the case where the errors are non-i.i.d., Kleibergen and Paap (2006) propose a robusterror correction for the Anderson’s statistic A failure to reject the null of zero correlationindicates that the model is underidentified

Weak Identification Tests

When the correlation between the endogenous variable and the IV is nonzero but small, theweak instrument problem arises (see discussion above) Staiger and Stock (1997) show thatfor large samples, the weak instrument problem can arise even when the correlationbetween x and z is significant at conventional levels of 5% and 1% Stock and Yogo (2005)develop an F statistic based on the Cragg and Donald (1993) F statistic for i.i.d errors, andKleibergen and Paap (2006) extend this, if the i.i.d assumption is violated The test statistic

is based on the rejection rate of various percentages, if the true rejection rate is 5% Stock–Yogo’s critical values are available for a range of the number of endogenous variables andinstruments If the computed F statistic is below the critical value, then the instruments areweak In our reported results, we use the rejection rate of 20%

Weak Identification Robust Inference Tests

The Anderson and Rubin (1949) test is used by substituting the relationship between theendogenous variable and the IV from Equation 2 in Equation 1 Thus, Equation 1 is esti-mated in its reduced form, and if the test fails to reject that the coefficient estimate on the

IV is zero, then the instrument is likely to be weak The idea with this test is that as theinstruments become weak, the relationship between the endogenous variable and the IVmust be weak, and thus, the relationship between the IV and dependant variable must also

be weak The beauty of this test is that it is robust to implicit weakness of the instruments.The Anderson and Rubin F statistic and the Stock and Wright (2000) S statistic providetests for the weak identification tests, when errors are non-i.i.d

Overidentifying Restrictions Test

In choosing instruments, we need to be reasonably assured that the covariance of the IV zwith the structural error term v in Equation 1 is zero In general, demonstrating this is diffi-cult because the structural error term is unobservable When the Generalized Method ofMoments (GMM) estimation is used, the Hansen’s J test can be used The null hypothesis

is that the instruments are not correlated with the structural error term in the second-stageregression That is, rejection of the null hypothesis provides evidence that the instrumentsare correlated with the error term In general, for large samples this test would lead torejection of the null hypothesis (see Larcker & Rusticus, 2007)

The Hausman Test

Hausman (1978) provides a formal test to assess whether the OLS estimates are cantly different from the 2SLS estimates: the Hausman test The Hausman test is a test forthe difference in OLS and 2SLS coefficients under the maintained assumption that theinstruments are not weak From the discussion of the partial R2 and partial F statisticabove, it follows that the Hausman test could reject the null or no joint determination ofthe endogenous variables if the instruments are weak Thus, the Hausman test should beconducted only if the instruments are not weak (also see Larcker & Rusticus, 2007)

signifi-In summary, we compute the following test statistics when we use the 2SLS estimation:(a) the partial R2, (b) the partial F statistic, (c) Anderson’s canonical correlation LM statis-tic for underidentification, (d) Cragg and Donald’s (1993) Wald F statistic for weak identi-fication, (e) Anderson–Rubin F statistic and Stock and Wright’s S statistic for weakinstrument robust inference, and (f) Hansen’s J statistic for the overidentification We usethe GMM estimation when we consider additional instruments: We substitute theKleibergen and Paap’s (2006) rk LM statistic and F statistic for Anderson’s canonical cor-relation LM statistic and Cragg and Donald F statistic, respectively (see Stock & Wright,2000; Stock, Wright, & Yogo, 2002)

Empirical Analysis

Following Whisenant et al (2003), we consider estimate Equations 3 and 4: These modelsdraw on a rich literature on the determinants of audit fees.11

Empirical Model

LogðNonaudit feeÞ5b1b1LogðAudit feeÞ1bD9Determinants1e; ð4Þwhere Log(Audit fee) is the natural logarithm of fees paid to the auditor for audit and audit-related services, and Log(Nonaudit fee) is the natural logarithm of fees paid to the auditorfor nonaudit services (fee data are obtained from Audit Analytics databases) Lag is thenumber of days between current fiscal year-end and earnings announcement date(Compustat report date of quarterly earnings) New Finance is an indicator variable thatequals one if the firm issues equity (Compustat Item 108 US$10 million) or long-termdebt (Compustat Item 111 US$1 million) in either the current or subsequent fiscal yearand is zero otherwise Assets is the natural logarithm of total assets (Compustat Item 6).Segments is the square root of the number of segments (Compustat Segment database).Employees is the square root of the number of employees (Compustat Item 29) DA ratio istotal debt (Compustat Item 181) divided by total assets (Compustat Item 6) Liquidity is theratio of current assets (Compustat Item 4) divided by current liabilities (Compustat Item 5).Inventory and AR turnover is inventory (Compustat Item 3) plus accounts receivable(Compustat Item 2), divided by total assets (Compustat Item 6) ROA is return on assetsdefined as operating income (Compustat Item 178) divided by total assets (Compustat Item6) Institutional ownership is the percentage of institutional holdings at the beginning of theyear Initial is an indicator variable that equals one if the auditor has been the firm’s auditorfor less than 2 years (Compustat Item 149) and is zero otherwise BIG5 is an indicator vari-able that equals one when the auditor is a member of the Big 5, and zero otherwise.Foreign Operations is an indicator variable that equals one if the firm has foreign opera-tions as indicated by foreign currency adjustments to income (Compustat Item 150) and iszero otherwise Loss is an indicator variable that equals one if the firm reported negativenet income (Compustat Item 172) in either of the two previous years and is zero otherwise.Sales growth is the growth rate in sales (Compustat Item 12) compared with the previousyear Volatility is variance of the residual from the market model over the current fiscalyear Opinion is an indicator variable that equals one if the firm received a modified going-concern audit opinion in either the current or previous fiscal year (Audit Analytics) andequal to zero otherwise Employee plans is an indicator variable that equals one if thefirm’s pension or postretirement plan assets or cost is greater than US$1 million and is zerootherwise Book to market is the book-to-market ratio at the beginning of the year.Discontinued operations is an indicator variable that equals one if the firm reports extraor-dinary items or discontinued operations (Compustat Item 48) and is zero otherwise.Distress probability is the 1-year change in Zimjewski’s (1984) probability of bankruptcyscore Stock return is the raw return for the fiscal year Restate is an indicator variable that

is one if net income or assets were restated and zero otherwise (data obtained fromGovernment Accountability Office [GAO] list of restatements’’)

Based on the knowledge spillover, we expect that a1 and b1in Equations 3 and 4 will

be positive, and based on Whisenant et al.’s (2003) results of no knowledge spilloveracross audit and nonaudit fees, we expect a1and b1 to be statistically no different fromzero (the null hypothesis) Note that if only a1(b1) is positive then theoretically speaking,

if we have the ‘‘correct and complete’’ model, the knowledge spillover is only from dit to audit (audit to nonaudit) This of course is a conclusion of tall order to draw fromour empirical analysis.

nonau-The instruments are Lag for audit fee and New Finance for nonaudit fee As discussed

in the previous section, the instruments Lag and New Finance should be correlated withaudit and nonaudit fees, respectively, and should not be correlated with the structural errorterm in the audit and nonaudit fee models, respectively.12

Sample and Initial Results

The sample contains all firms not in the financial services industry with data available inCompustat and Center for Research in Security Prices for fiscal years 2000 to 2006 Westop at 2006 because the instrument New Finance requires 1-year-ahead data The auditand nonaudit fees data are from Audit Analytics We consider three subsamples: year 2000,year 2001, and years 2002 to 2006 In particular, we consider year 2000 separately to pro-vide a benchmark for comparing our results with those of prior research We consider 2002

to 2006 together because there were regulatory changes involving nonaudit fees in 2002(see Sarbanes-Oxley Act [SOX] of 2002 and Securities and Exchange Commission Release

No 33-8183).13 As such, year 2001 is also considered separately There are 2,768; 3,812;and 20,173 observations for the subsamples 2000, 2001, and 2002 to 2006, respectively.Table 1, Panel A provides the descriptive statistics for the three subsamples Comparedwith the prior research that is typically based on hand collected data, the companies in oursample pay lower audit and nonaudit fees in 2000 This could be due to Audit Analyticscoverage of many smaller firms compared with the sample of earlier studies The meanaudit fee increases from about US$0.5 million in 2001 to about US$1.2 million in 2005;the median shows a similar substantial increase This suggests that the SOX requirementshave substantially increased audit fees in the later years of our sample Correspondingly,the nonaudit fees have substantially decreased The audit fees have increased over time,whereas the nonaudit fees have decreased reflecting changes in the regulatory environment

It is also interesting to note that even though the sample of firms covered in AuditAnalytics has increased over time, the fundamental firm characteristics are roughly stable,that is, the mean and median total assets, lag, segments, employees, debt to asset ratio,liquidity, turnover ratios, and institutional ownership are similar over the years, whereasprofitability, growth, and financial distress measures exhibit macroeconomy-wide trends.Table 2 contains the estimation of Equations 3 and 4 Panel A reports the results of theOLS estimation: similar to prior research, we find that the coefficients of Log(Audit fee)and Log(Nonaudit fee) are positive and significant, suggesting the presence of knowledgespillover effects Although statistically speaking the results are qualitatively similar tothose of prior research, compared with Whisenant et al (2003), in our 2000 sample, forEquation 3 the coefficient estimate is a little more than twice as large; these differences areclearly attributable to differences in the sample

Table 2, Panel B reports the results of using 2SLS estimation for Equations 3 and 4 Forthe 2002-to-2006 sample, we use the robust standard errors to compute the t statistic, withFama–French 48 industry groups as the cluster.14 Here again, as with the discussion ofTable 2, Panel A above, the results for year 2000 are qualitatively similar to those of priorresearch Most importantly, the OLS results and the 2SLS results together are consistentwith the conclusion that audit and nonaudit fees are not associated when we consider theirjoint determination This conclusion extends to the later years: 2001 and 2002 to 2006

Thus, the results of prior research are robust and carryover to different samples and ent periods.

differ-We test the validity/appropriateness of the instruments The F statistic for New Finance

as the instrument of Log(Nonaudit fee) are 0.57, 0.10, and 2.79 for the 2000, 2001, and

2002 to 2006 subsamples, respectively; the partial R2s are 02%, 00%, and 01% for thethree subsamples This suggests that the instrument New Finance is not appropriate forLog(Nonaudit fee) The F statistic for Lag as the instrument of Log(Audit fee) are 7.35,0.05, and 12.85, and the partial R2are 27%, 00%, and 06% for the 2000, 2001, and 2002

to 2006 subsamples, respectively Although the partial F statistic is statistically significantfor Lag as the instrument for Log(Audit fee), the partial R2indicates a close to zero correla-tion between the instrument and the endogenous variable This suggests that the instru-ments are not appropriate (see Larcker & Rusticus, 2007, 2010).15

We provide tests based on recent advances in econometrics for the appropriateness ofthe instruments (see discussion in section titled ‘‘Summary of Econometric Issues WithInstruments’’) The Anderson canonical correlation test for underidentification shows thatNew Finance is not appropriate for Log(Nonaudit fee), whereas Lag is reasonable forLog(Audit fee) for 2000 and 2002 to 2006 The Cragg-Donald F statistic for the weak iden-tification test for the instrument New Finance are considerably below the critical cutoffvalues for a rejection rate of 20%, whereas Lag is above the cutoff critical value for 2000and 2002 to 2006 subsamples The Weak Instrument Robust inference tests statistics showthat New Finance is marginally significant for 2002 to 2006, and Lag is strongly significant

in 2001 and 2002 to 2006 In summary, the tests of appropriateness of Lag as an instrumentfor Log(Audit fee) for 2000 and 2001 provide mixed support and show reasonable supportfor being an appropriate instrument for 2002 to 2006 For the instrument New Finance, thetests of appropriateness suggest that the instrument is weak.16 Because these instrumentsare weak, we next look for additional instruments that can alleviate the problem of weakinstruments

Additional Instruments

Although the existing audit literature supports the notion that audit and nonaudit fees arejointly determined, there is no theoretical model to guide our choice of instruments Forinstance, in the economics literature when the price and quantity of a commodity are con-sidered to be jointly determined, there is rich theory of demand and supply for the com-modity that guides the structural equation that needs to be estimated, which can guide thechoice of instruments Even with a rich theory, it is sometimes difficult to identify appro-priate instruments

We consider additional instruments based on two approaches used in the financial nomics literature First, we consider the industry average as the instrument (see Bertrand &Mullainathan, 2001; Hanlon et al., 2003; Lev & Sougiannis, 1996; Murphy, 2000) Second,

eco-we consider a method developed for mitigating the difficulty of identifying appropriateinstruments by creating synthetic instruments with higher moments of the combination ofendogenous and exogenous variables developed by Lewbell (1997) This approach hasbeen used in financial economics papers (Nagar & Rajan, 2005)

Industry average–based instruments In addition to Lag, we consider the industry average

of log of audit fee as an additional instrument for Log(Audit fee), and correspondingly, inaddition to New Finance, we consider the industry average of log of nonaudit fee as anadditional instrument for Log(Nonaudit fee) The instruments are developed using the

companion portfolio approach Industry groups are based on the Fama–French 48 industrygroups.17 For each firm-year observation, we identify the 10 closest firms in the sameindustry group in terms of total assets/size: thus the companion portfolio of firms belong tothe same industry group and are of similar size We choose size as the matching criteriabecause research in audit fees show that most of the variation in audit fees are explained

by the size variable The mean and median Log(Audit fee) and Log(Nonaudit fee) of thecompanion portfolio firms are the instruments for Log(Audit fee) and Log(Nonaudit fee).Larcker and Rusticus (2010) discuss two drawbacks of using industry averages: first isthe implicit assumption that the endogenous portion of the audit- and nonaudit fees onlyvary within the size industry, and second is the preclusion of the use of industry-fixedeffects, which are likely to be important in the audit- and nonaudit-fee context We try toovercome the first concern, by using the size-matched industry average Given that firmsize explains more than half the variation in audit fees, it appears reasonable that the endo-genous portion of the audit fees is contained in the size industry–matched portfolio.However, this may not be true for the nonaudit fees For the second concern, because weuse the size industry–matched portfolio approach, we can use the industry-fixed effects.Table 3, Panel A provides the results using Industry Mean as the additional instrument.For these estimations, we use the GMM procedure, with industry cluster robust estimator

of the variance and covariance matrix for computing the z statistics of the coefficient mates The GMM procedure is appropriate for our purpose because in the IV estimation,there is an implicit restriction that the errors are expected to be zero if and only if the coef-ficient estimate is the ‘‘true’’ estimate (see Amemiya, 1974) The Kleinbergen-Paap F sta-tistic of the instruments for Log(Nonaudit fee), that is, New Finance and Industry meanLog(Nonaudit fee) for year 2000 is 8.72, which is slightly below the critical value of 8.75.Thus, we cannot reject the hypothesis that the instruments are not weak Other than this theunderidentification, weak identification, and weak instrument robust inference tests allshow that the instruments for Log(Audit fee) and Log(Nonaudit fee) are not weak For theoveridentification test, the Hansen’s J statistic shows that the instruments for Log(Nonauditfee) are orthogonal to the structural error term, but the instruments for the Log(Audit fee)are not so As discussed before, for large samples, rejection of the null hypothesis is morelikely with the overidentification tests (see Larcker & Rusticus, 2007) In balance, the testsindicate that the industry mean–based additional instruments along with Lag and NewFinance are not weak In unreported analysis, the Hausman test indicates that Log(Auditfees) and Log(Nonaudit fees) are jointly determined

esti-Given that the set of instruments are not weak, the top rows of Table 3, Panel A showsthe coefficient estimates of the endogenous variables and the associated z statistic com-puted, correcting the standard error using industry groups as the cluster In Equation 3 withLog(Audit fee) as the dependant variable, the coefficient estimates of Log(Nonaudit fee) are0.6313, 0.6329, and 0.5020, for the subsamples 2000, 2001, and 2002 to 2006, respectively,and all these estimates are statistically significant In Equation 4 with Log(Nonaudit fee) asthe dependant variable, the coefficient estimates of Log(Audit fee) are 0.4458, 0.8593, and0.9175, for the subsamples 2000, 2001, and 2002 to 2006, respectively, and all these esti-mates are statistically significant This suggests that audit and nonaudit fees are positivelyassociated, indicating the presence of knowledge spillovers More interestingly, in Equation

3, the estimated coefficient of Log(Nonaudit fee) for 2002 to 2006 is lower than that in

2000 and 2001, which may be due to SOX provisions relating to nonprovision of nonauditservices

Table 3, Panel B provides the results of estimating Equations 3 and 4 with additionalinstruments based on industry medians The results are qualitatively similar to that dis-cussed in Table 3, Panel A Overall, the results are consistent with the existence of knowl-edge spillovers between audit and nonaudit services.

Synthetic instruments: Higher moment (HM) estimators Erickson and Whited (2002),Lewbell (1997), and Dagenais and Dagenais (1997) show that for models with measure-ment error, appropriate instruments can be constructed from available data by using higherorder moments Following Lewbell, we call the HM estimator as synthetic instrument toemphasize that the observations for the instruments are derived from transformations ofendogenous and exogenous variables, that is, the instruments are synthetically derived.Ebbes et al (2009) shows that such higher order moment methods are applicable for IVmethods as well In particular, Ebbes et al and Lewbell show that candidates for appropri-ate instruments are (a) the product of the endogenous and exogenous variables and (b) thesecond moment of the endogenous variable Lewbell shows that the HM estimators are sen-sitive to outliers As such, Ebbes et al suggest using robust estimators and GMM Nagarand Rajan (2005) use Lewbell’s HM estimator as the instrument The performance of themethod depends on the skewness of the endogenous variable, and the method can yieldweak instruments.18

Table 4, Panel A provides the results of estimating Equations 3 and 4 when the tional instruments are based on the second moment of the endogenous variables: the instru-ment for Log(Audit fee) is the second moment of the demeaned Log(Audit fee) and Lag,and the instrument for Log(Nonaudit fee) is the second moment of the demeanedLog(Nonaudit fee) and New Finance19 The weak instrument tests shows that the secondmoment is a valid instrument: this is because the Log(Audit fee) and Log(Nonaudit fee) areskewed Lewbell (1997) shows that the HM method is likely to be appropriate if the endo-genous variable is skewed However, the overidentification test largely fails to reject thenull hypothesis of no correlation between the instrument and the structural error term Asdiscussed earlier, the overidentification test, especially in large samples, is likely to rejectthe null hypothesis Overall, the synthetic instrument is appropriate In this case, both theaudit and nonaudit fees are positively correlated for each of the subsample periods

Table 4, Panel B provides the results of estimating Equations 3 and 4 when the tional instruments are based on the product of the endogenous variables and an exogenousvariable and the second moment of two exogenous variables In particular, the instrumentsfor audit fee are Lag, product of the demeaned log of total asset times demeaned Log(Auditfee), square of demeaned log of total assets, and square of demeaned square root of seg-ments, and the instruments for nonaudit fee are New Finance, product of the demeaned log

addi-of total asset times demeaned Log(Nonaudit fee), square addi-of demeaned log addi-of total assets,and square of demeaned square root of segments The instruments for nonaudit fee arebelow the critical value of 10.27 for the weak identification test for all subsample periods,whereas the instrument for audit fee is below the critical value for the subsample period

2002 to 2006 Here, we observe the relationship between the endogenous variables and theappropriateness of the instruments As suggested by Staiger and Stock (1997), in large sam-ples even if the F statistic and the partial R2s are reasonably high, the instrument may still

be weak The weak identification test demonstrates this point For nonaudit fee, the thetic instruments are weak for 2001: in this subsample, we find no association betweennonaudit and audit fee For the 2002 to 2006 period, the Kleibergen and Paap (2006) testshows that the instruments are weak, but the Anderson and Rubin (1949) and Stock andWright (2000) tests show that they are not weak Here the z statistic of the coefficient

estimate on nonaudit fee is 2.18, which is considerably lower than that when the instrument

is not weak A similar pattern can be observed in Equation 4: When the instrument foraudit fee is not weak, that is, for 2000 and 2001, the endogenous variables are positivelyassociated, but when the instruments are weak, that is, for 2002 to 2006, the z statistic indi-cates an association that is not significantly different from zero

Audit Fees and Tax-Related Nonaudit Fees

Although nonaudit fees from management advisory services have been curtailed since

2002, recently the PCAOB has initiated efforts to curtail auditors of public companies fromsupplying certain types of nonaudit tax services Reacting to the extensive media coverageand congressional actions pertaining to tax shelters promoted by auditors for their publiccompany clients, the PCAOB conducted a roundtable in July 2004 and stated that ‘‘theBoard has determined that it is appropriate to consider the impact of tax services on auditorindependence’’ (see PCAOB, 2004a, p 3) In December 2004, the PCAOB (2004b) pro-posed rules to curb certain types of nonaudit tax services for their public company clients

We thus examine the relationship between audit fee and tax-related nonaudit fee for theperiod 2002 to 2006 For this purpose, we use the instruments based on the companionportfolio approach, that is, industry mean and industry median as in Table 3 We find thatthe instruments are not weak, but as with large samples, the overidentification test rejectsthe null hypothesis of no association between the instrument and the structural error term.The results of this analysis are reported in Table 5 The association between the audit feeand tax-related nonaudit fee is positive and statistically significant, which shows the pres-ence of knowledge spillover effects between tax-related services and audit services

Concluding Remarks

We examined the joint determination of audit and nonaudit fees using the 2SLS estimationmethod We utilized tests to validate the appropriateness of the instruments We find thatwhen the instruments are not appropriate, that is, the instruments are weak, there is no rela-tionship between audit and nonaudit fees However, when we use instruments that are notweak, then audit and nonaudit fees are positively associated This shows the presence ofknowledge spillover effects across audit and nonaudit services We also show the existence

of such knowledge spillover effects across audit and tax services We thus document thebenefits arising from economies of scope, which in turn needs to be balanced with the cost

of loss of auditor’s independence (if any) for policy decisions on regulating nonauditservices

Acknowledgments

The authors gratefully acknowledge comments and suggestions from an anonymous reviewer, AshiqAli, Ferdinand Gul, Clive Lennox, Bin Srinidhi and Scott Whisenant

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the research, authorship, and/

or publication of this article

as Waste Management’s US$3.54 billion write-down of 1992 to 1997 profits and MicroStrategy’s US$55.8 million earnings restatement Economic bonding leading to loss of auditorindependence is not the focus of the research design used in this article Our objective is toexamine the presence of economies of scope benefits.

2 Other explanations posited for the positive association between audit and nonaudit fees are (a)nonaudit services could lead to extensive changes in organization structure that require additionalaudit effort, (b) clients that get nonaudit services are ‘‘problematic,’’ and (c) the noncompetitiveaudit market enables auditors to charge premiums (see Hay, Knechel, & Wongm, 2006)

3 A rich stream of literature examines the effect of nonaudit services on various proxies of auditquality and provides insights into the impact of such services on auditor independence DeFond,Raghunandan, and Subramanyam (2002) use the auditor’s going-concern qualification as a proxyfor audit quality and show that the provision of nonaudit services does not adversely impact theauditor’s propensity to issue going-concern opinion Kinney, Palmrose, and Scholz (2004) showthat audit quality is not impaired but improved by the provision of tax-related services: they userestatements as the proxy for audit quality Kinney et al.’s finding indicates the presence ofknowledge spillover Other studies such as Ashbaugh, LaFond, and Mayhew (2003) documentsimilar findings when they measure audit quality using abnormal accruals

4 Concurrent research by Francis and Lennox (2008) shows that results obtained from selection models of audit fees are not robust

self-5 Although this issue has been placed in the back burner, recent conflict of interest concernsregarding the Lehman collapse has reignited such discussions In addition, accounting firms aredeveloping their corporate governance and International Financial Reporting Standards–relatedconsulting practices in recent years; of course, their target market is not their current auditclients

6 Industry averages contain both the endogenous and the exogenous portions As such, Larckerand Rusticus (2010) state, ‘‘This approach will only work when the exogenous part of the origi-nal variable varies across industry, whereas the endogenous part varies only within industry.’’

7 Larcker and Rusticus (2010) summarize the challenge when they state, ‘‘There is no fool-proofway of dealing with the problem of endogeneity.’’ The synthetic-instrument approach for all its

‘‘heroic’’ assumptions is an additional method that we bring to the forefront

8 Even though the discussion is couched in terms of the two-staged least squares (2SLS) dure, the econometric problems apply to all methods that use the instrument (IV) approach Wethus refer the 2SLS and IV methods interchangeably In Equation 1, x is referred to as an endo-genous variable

proce-9 A nice interpretation of the partial R2 is provided by Shea (1997) and Godfrey (1999) Theyshow that the partial R2 is the ratio of the asymptotic variance of the coefficient estimateobtained from ordinary least squares (OLS) and IV, times the ratio of one minus the residual

sum of squares over the total sum of squares obtained from the second stage of IV and OLS.This demonstrates that the downside of using the IV method compared with the OLS method isthe increased standard error of the estimated coefficient In other words, improving the consis-tency of the estimate comes at a cost of increased standard errors.

10 Kennedy (1992, p 137) states, ‘‘The major drawback to instruments (IV) estimation is that thevariance-covariance matrix of the IV estimator is larger than that of the OLS estimator, by anamount that is inversely related to the correlation between the instrument and the regressor This

is the price paid for avoiding the asymptotic bias of OLS; the OLS estimator could be preferred

on the MSE criterion.’’

11 For example, see Ezzamel, Gwilliam, and Holland (1996), Felix, Gramling, and Maletta (2005),Ferguson, Francis, and Stokes (2003), Francis (1984), and Francis and Stokes (1986)

12 See Whisenant, Sankaraguruswamy, and Raghunandan (2003) for arguments on why Lag wouldnot be correlated with nonaudit fee and New Finance would not be correlated with the audit-feemodel However, as in any context without a solid theoretical underpinning (see Larcker &Rusticus, 2007, 2010), it could be argued that higher Lag would lead to more nonaudit opportu-nities so as to improve the accounting information system Similarly, it could be argued that NewFinance would be associated with higher audit fees because the auditors would be ‘‘more care-ful’’ due to legal liability concerns

13 The release is available at http://www.sec.gov/rules/final/33-8183.htm

14 All the results are qualitatively similar when we use years as the cluster

15 Theoretically speaking, the negative coefficients are difficult to interpret, and likely are tive of problems with the instruments

indica-16 The overidentification test is not applicable because Equations 3 and 4 are exactly identified

17 See http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data-library.html

18 The consistency of the higher moment approach requires that the exogenous portion of the auditand nonaudit fees, that is, the endogenous variables have a skewed distribution, and the endogen-ous portion has a symmetric distribution and is uncorrelated to the structural error term Theseassumptions cannot be directly verified and is one major limitation of this approach

19 The distribution of all the synthetic instruments is highly skewed (the mean of these variablesexceed the third quartile of their respective distributions The pairwise Spearman correlationcoefficients for all the IVs are highly significant (at less than 1% significance level) Additionaltables giving the distribution of the synthetic instruments and the correlation between the syn-thetic instruments are available from the authors

Bertrand, M., & Mullainathan, S (2001) Are CEOs rewarded for luck? The ones without principalsare Quarterly Journal of Economics, 116, 901-932.

Cragg, J G., & Donald, S G (1993) Testing identifiability and specification in instrumental variablemodels Econometric Theory, 9, 222-240

Dagenais, M G., & Dagenais, D L (1997) Higher moment estimators for linear regression modelswith errors in variables Journal of Econometrics, 76, 193-221

Davis, L R., Ricchiute, D N., & Trompeter, G (1993) Audit effort, audit fees, and the provision ofnonaudit services to audit clients Accounting Review, 68, 135-150

DeBerg, C., Kaplan, L., & Pany, K (1991) An examination of some relationships between non-auditservices and auditor change Accounting Horizons, 5, 17-28

DeFond, M L., Raghunandan, K., & Subramanyam, K R (2002) Do non-audit service fees impairauditor independence? Evidence from going concern audit opinions Journal of AccountingResearch, 40, 1247-1274

Ebbes, P., Wedel, M., & Bockenholt, U (2009) Frugal IV alternatives to identify the parameter for

an endogenous regressor Journal of Applied Econometrics, 24, 446-468

Erickson, T., & Whited, T M (2002) Two-step GMM estimation of the errors-in-variables modelusing higher-order moments Economic Theory, 18, 776-799

Ezzamel, D., Gwilliam, R., & Holland, K (1996) Some empirical evidence from publicly quotedU.K companies on the relationship between the pricing of audit and non-audit services.Accounting and Business Research, 27, 3-16

Felix, W., Jr., Gramling, A A., & Maletta, M J (2005) The influence of nonaudit service revenuesand client pressure on external auditors’ decisions to rely on internal audit ContemporaryAccounting Research, 22, 31-53

Ferguson, A., Francis, J R., & Stokes, D (2003) The effects of firm-wide and office-level industryexpertise on audit pricing Accounting Review, 78, 429-448

Francis, J R (1984) The effect of firm size on audit prices: A study of the Australian market.Journal of Accounting and Economics, 6, 33-151

Francis, J R., & Lennox, C S (2008) Selection models in accounting research (Working Paper).Hong Kong: Hong Kong University of Science and Technology

Francis, J R., & Stokes, D (1986) Audit prices, product differentiation, and scaled economies:Further evidence from the Australian market Journal of Accounting Research, 24, 383-393.Godfrey, L G (1999) Instrument relevance in multivariate linear models Review of Economics andStatistics, 81, 550-552

Hanlon, M., Rajagopal, S., & Shevlin, T (2003) Are executive stock options associated with futureearnings? Journal of Accounting and Economics, 36, 3-43

Hausman, J A (1978) Specification tests in econometrics Econometrica, 46, 1251-1271

Hay, D C., Knechel, W R., & Wongm, N (2006) Audit fees: A meta-analysis of the effect ofsupply and demand attributes Contemporary Accounting Research, 23, 141-191

Kennedy, P (1992) A guide to econometrics Cambridge: MIT Press

Kinney, W R., Palmrose, Z., & Scholz, S (2004) Auditor independence, non-audit services, andrestatements: Was the U.S Government right? Journal of Accounting Research, 42, 561-588.Kleibergen, F., & Paap, R (2006) Generalized reduced rank tests using the singular value decompo-sition Journal of Econometrics, 127, 97-126

Larcker, D F., & Rusticus, T O (2007) Endogeneity and empirical accounting research EuropeanAccounting Review, 16, 207-215

Larcker, D F., & Rusticus, T O (2010) On the use of instrumental variables in accounting research.Journal of Accounting and Economics, 49, 186-205

Lev, B., & Sougiannis, T (1996) The capitalization, amortization, and value-relevance of R&D.Journal of Accounting and Economics, 21, 107-137

Lewbell, A (1997) Constructing instruments for regressions with measurement error when no tional data are available, with application to patents and R&D Econometrica, 65, 1201-1213

Murphy, K J (2000) Performance standards in incentive contracts Journal of Accounting andEconomics, 30, 245-278.

Nagar, V., & Rajan, M V (2005) Measuring customer relationships: The case of the retail bankingindustry Management Science, 51, 904-919

Palmrose, Z V (1986) The effect of non-audit services on the pricing of audit services: Further dence Journal of Accounting Research, 24, 405-411

evi-Public Company Accounting Oversight Board (2004a, July 14) Briefing Paper: Auditor dence and tax services roundtable Washington DC: Author

Public Company Accounting Oversight Board (2004b, December 14) Proposed ethics and dence rules concerning independence, tax services, and contingent fees (PCAOB Release No.2004-015) Washington, DC: Author

indepen-Shea, J (1997) Instrument relevance in multivariate linear models: A simple measure Review ofEconomics and Statistics, 79, 348-352

Simunic, D (1980) The pricing of audit services: Theory and evidence Journal of AccountingResearch, 22, 169-190

Simunic, D (1984) Auditing, consulting and auditor independence Journal of Accounting Research,

Wooldridge, J M (2002) Econometric analysis of cross section and panel data Cambridge: MITPress

Zmijewski, M (1984) Methodological issues related to the estimation of financial distress predictionmodels Journal of Accounting Research, 22, 59-82

Auditing & Finance 27(3) 325–358

Ó The Author(s) 2011 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0148558X11409155

http://jaaf.sagepub.com

Do Takeover Defenses Impair

Equity Investors’ Perception of

‘‘Higher Quality’’ Earnings?

Abstract

Prior studies show that managerial entrenchment deteriorates the credibility of earnings,hence reducing the value relevance of earnings However, prior literature documents thatthe likelihood of earnings management is lower in firms with more antitakeover provisionsbecause entrenched managers pursue a ‘‘quiet life’’ instead of striving to maximize wealth

of shareholders Despite ‘‘higher quality’’ earnings of such firms, the authors find that over protection impairs the perception of equity investors on earnings quality The authorsattribute this contradictory result to the failure of management to take risky but value-enhancing projects owing to pursuits of a quiet life The authors also expect and find thatinvestments of more defensive firms are valued at a discount, suggesting that equity inves-tors expect such firms to take less advantage of their growth potentials The authors corro-borate this result by showing lower variability in firm value of more defensive firms

take-Keywords

takeover defenses, earnings informativeness, earnings quality, market perception

This study examines the effect of takeover defenses on earnings informativeness, which isestimated by the slope coefficient relating stock returns to earnings obtained from regressions

of annual returns on annual earnings The entrenchment view suggests that takeover defensesprotect managers from the potential threat of the market to corporate control, thereby aggra-vating agency problems (Ashbaugh-Skaife, Collins, & LaFond, 2006; Dittmar & Mahrt-Smith, 2007; Gompers, Ishii, & Metrick, 2003) By focusing on agency problems associatedwith takeover defense mechanisms, one may predict that the financial reporting quality willdecrease with the degree of takeover defenses given that agency problems associated withsuch mechanisms cause the credibility of earnings to deteriorate (Holthausen & Verrecchia,1988) Furthermore, consistent with previous studies linking the credibility of earnings toearnings informativeness, the deteriorating financial reporting quality of more defensive firms

2

The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong

Corresponding Author:

Woo-Jong Lee, M747, Li Ka Shing Tower, School of Accounting and Finance, The Hong Kong Polytechnic

University, Hung Hom, Kowloon, Hong Kong

Email: afwjlee@inet.polyu.edu.hk

may lead to the reduced value relevance of earnings (e.g., Fan & Wong, 2002; Francis,Schipper, & Vincent, 2005; Teoh & Wong, 1993; Warfield, Wild, & Wild, 1995).1

Although the entrenchment view of takeover defenses has been pervasive in literature,Zhao and Chen’s (2008a, 2008b, 2009) findings of a positive association between takeoverdefenses and earnings quality (i.e., lower magnitudes of absolute abnormal working capitalaccruals for more defensive firms) are likewise particularly intriguing Specifically, theyshow that takeover protection is associated with lower levels of absolute abnormal accruals(Zhao & Chen, 2008a, 2009) and a lower likelihood of accounting fraud (Zhao & Chen,2008b) Their findings suggest that takeover protection allows entrenched managers to enjoy

a ‘‘quiet life’’ rather than striving to maximize shareholder’s wealth, thus mitigating theirincentives to manage earnings despite their ability to do so At a glance, their findings differfrom the entrenchment view of takeover defenses, such that takeover protection is likely tomotivate incumbent managers to manage earnings to disguise their expropriation activities.With ‘‘higher quality’’ earnings in more defensive firms than in less defensive firms, highervalue relevance of earnings may be expected in the former because earnings quality is animportant determinant of value relevance of earnings (Holthausen & Verrecchia, 1988; Teoh

& Wong, 1993) Taken as a whole, our current knowledge of the association between agerial entrenchment and value relevance of earnings does not provide a clear answer Thisarticle thus aims to answer the question of how managerial entrenchment deteriorates thevalue relevance of earnings, even when the earnings are considered of higher quality.Specifically, the study analyzes the slope coefficient (i.e., earnings response coefficient[ERC]) relating annual returns to annual earnings levels and earnings changes as a proxyfor earnings informativeness By allowing ERC to vary across takeover defense mechan-isms, we examine differences in ERCs between less defensive firms (i.e., firms with fewerantitakeover provisions) and more defensive firms (i.e., firms with more antitakeover provi-sions) Using a sample of U.S firms covered by the Investor Responsibility ResearchCenter (IRRC) publications from 1990 to 2006, we find that more defensive firms exhibitsignificantly lower ERC compared with less defensive firms These results are robust tocontrol for other known determinants of ERC and other corporate governance mechanisms(e.g., insider ownership, board size, and the proportion of independent directors) The find-ings suggest that investors perceive takeover defenses as having unfavorable implications

man-on earnings quality, despite their tendency to lower the likelihood of earnings management

as reported by Zhao and Chen (2008a, 2008b, 2009)

Our finding raises the question whether more defensive firms do report higher qualityearnings and less defensive firms report ‘‘lower quality’’ earnings Under the traditionalentrenchment view of takeover defenses, entrenchment increases the likelihood of manage-rial expropriation of shareholder wealth, with the entrenched managers more likely tomanage reported earnings to minimize shareholder detection of the expropriation and toavoid subsequent legal penalties as well as other undesirable consequences Therefore, thisview leads us to cast doubt on Zhao and Chen’s (2008a, 2008b, 2009) proposition thatmore defensive firms are less likely to engage in opportunistic earnings management andconsequently report higher quality earnings than less defensive firms.2

Rather, it is likely that the lower absolute accruals for more defensive firms and thehigher absolute accruals for less defensive firms are related to the differential exploitation

of growth opportunities (Ball & Shivakumar, 2008; Fairfield, Whisenant, & Yohn, 2003).Existing literature maintains that the positive correlation between earnings growth andaccruals is stronger when accruals capture the fundamental investment information of afirm (Zhang, 2007) To the extent that accruals in less defensive firms capture the effect of

investment on future earnings, the earnings in such firms should not have been considered

of lower quality despite them containing higher level of accruals In fact, the threat of tile takeovers disciplines managers and motivates them to pursue positive net present value(NPV) projects; hence, less defensive firms are likely to undertake riskier but value-enhancing investments (Bertrand & Mullainathan, 2003; John, Litov, & Yeung, 2008; Low,2009) Accordingly, compared with more defensive firms, less defensive firms are likely toexhibit higher growth in production and sales, causing high growth in optimal working cap-ital levels, which naturally produces positive accruals and a high level of absolute accruals(Fairfield et al., 2003; Zhang, 2007) Therefore, Zhao and Chen’s (2008a, 2008b, 2009)finding of a high level of absolute abnormal working capital accruals in less defensivefirms can be attributed to their high growth potential and significant investments.According to the quiet life hypothesis, takeover protection provides managers with incen-tives to avoid making difficult investment decisions and the costly efforts required topursue value-generating projects Accordingly, more defensive firms invest in projects withless expected variance in future payoffs, which generate less need for accruals and thus alower level of absolute accruals.3

hos-Such risk avoidance of entrenched managers would raise concern among stock investorsregarding management’s incentives to pursue growth and profitability, which in turn haveimplications on the value relevance of earnings Low (2009) reports that the Delawareregime shift, which gives Delaware firms staggered boards for potent takeover defense,reduces shareholder wealth by discouraging managerial risk taking If shareholders expectthat the increased protection from takeovers allows managers to turn down risk-increasing,but positive, NPV projects, then the value relevance of earnings is likely to decline as cur-rent earnings are less likely to persist in the future The real option theory suggests that thevaluation model embeds the firm’s value-creating capital investment decisions within theset of available opportunities, as characterized by options to grow and to abandon (Chen &Zhang, 2007) Owing to low expected variance in future payoffs, investments of moredefensive firms would harm the option value of investors The impaired growth optionvalue can reconcile low value relevance of earnings in more defensive firms with the quietlife theory and with Zhao and Chen’s (2008a, 2008b, 2009) findings of the low level ofabsolute working capital accruals in such firms

To corroborate the growth- and investment-based explanation, we provide additionalevidence from four different tests First, we divide our sample into high- versus low-growth-opportunity groups to examine whether the negative association between takeoverdefenses and the ERC varies with growth opportunity The second test applies Chen andZhang’s (2007) value implication of accounting variables to stock returns They havedeveloped a model relating returns to accounting variables that measure a firm’s underlyingoperations Similarly, they have identified four cash flow–related factors and a discountfactor to explain returns, namely, earnings yield, capital investments, changes in profitabil-ity, changes in growth opportunities, and changes in the discount rate Analyses based ontheir model reveal that investors place less weight on capital investments when valuingmore defensive firms, which is consistent with the impaired growth option value of suchfirms Third, we present evidence on the value discount associated with investments usingstock price as a denominator of investment ratios (i.e., capital investments divided by thestock price) If investors perceive that takeover protections entrench managers to the detri-ment of outside shareholders due to inefficient investments, then more defensive firmsshould have higher investment-to-price ratios that represent the increased value discountassociated with investments Using several of the investment-to-price ratios suggested in

prior literature, we find that investments (e.g., capital expenditures) are valued at a discount

in defensive firms relative to less defensive firms Fourth, we expect that more defensivefirms are likely to reduce variability in firm value as a result of rejecting risky but value-increasing investment projects We find that more defensive firms do indeed exhibit alower degree of such variability (e.g., the SD of Tobin’s Q and stock return volatility),which suggests a deteriorated growth option value for shareholders Taken as a whole,these findings provide a better understanding of the reasons why equity investors valueearnings (and investments) of defensive firms at a discount

This article makes several contributions to the literature First, to our knowledge, thisarticle is the first to examine the impact of takeover defenses on earnings informativenessusing firm-level data.4 Our results are consistent with the entrenchment view as well aswith the quiet life view of takeover defenses One implication of the results is that equityinvestors view takeover defenses as having an unfavorable impact on earnings quality,even though earnings manipulation is less likely More importantly, this article examinesthe actual causes of reduced value relevance of earnings in more defensive firms Theresults imply that lower expected variance in future payoffs and lower levels of absoluteaccruals, which seem to yield higher financial reporting quality at a glance, do not necessa-rily lead to higher value relevance of earnings, given that they also reflect decreased man-agerial incentives to exploit growth potentials Beyond the mixed empirical evidence on therelationship between takeover defenses and financial reporting quality, this article extendsthe discussion to the investors’ perception of earnings quality

The rest of the article is organized as follows: The section titled ‘‘HypothesisDevelopment’’ discusses the economic links between corporate governance and the valuerelevance of earnings and develops the research hypothesis Section titled ‘‘VariableMeasurement and Research Design’’ describes the research design, variable definitions, andestimation models Section titled ‘‘Sample Description’’ introduces the sample selection pro-cedure and presents the summary of statistics Section titled ‘‘Empirical Results’’ providesthe empirical results Section titled ‘‘Summary and Conclusions’’ concludes the article

Hypothesis Development

Corporate Governance and Earnings Informativeness

Teoh and Wong (1993) have simplified the theoretical model of Holthausen and Verrecchia(1988) to show that more precise earnings signals lead to more stock price responses Teohand Wong argue that accounting earnings are more credible when earnings signals are pre-cise They attribute greater earnings informativeness of firms audited by large auditors togreater credibility of earnings reports Recent research extends these earlier studies byemphasizing the governance mechanisms of firms as an important determinant of precision

in earnings signals and the value relevance of earnings information (e.g., Fan & Wong,2002; Francis et al., 2005; Warfield et al., 1995)

Consistent with the precision or noise argument, numerous articles highlight the negativeassociation between corporate governance and earnings management (e.g., Dechow, Sloan,

& Sweeney, 1996; Xie, Davidson, & DaDalt, 2003) Such studies frequently use abnormalaccruals to proxy for earnings quality and demonstrate that sound corporate governancereduces the level of abnormal accruals and the occurrence of earnings management Thisviewpoint does not expect earnings signals of poorly governed firms to be a precise proxyfor underlying firm performance due to an opportunistic management of earnings

How then do investors form their perceptions of earnings quality? Durnev and Kim(2005) argue that firms with abundant investment opportunities and a greater need forexternal financing have stronger incentives to structure better governance mechanisms.Moreover, better investor protection inhibits managers from taking private benefits, whichdrives corporations to undertake optimal or value-enhancing investments (John et al.,2008) As a result, these features result in strong operating performance in well-governedfirms, whereas poorly governed firms will experience a significant degree of underperfor-mance in the future (Core, Guay, & Rusticus, 2006; Gompers et al., 2003) It is then likelythat reported earnings of more defensive firms are less persistent over time Given thatstock prices are mainly affected by permanent components of earnings, we expect lowervalue relevance of earnings in more defensive firms (Easton, Shroff, & Taylor, 2000).

In addition, firms with high growth potential and active investments likely experiencelow expected risk premium and hence, low discount rates (Fama & French, 1993) To theextent that accruals capture investment activities (e.g., Fairfield et al., 2003; Zhang, 2007),higher absolute accruals for less defensive firms are likely related to a lower discount rate(Hirshleifer, Hou, & Teoh, 2009; Kothari, Lewellen, & Warner, 2006).5 If less defensivefirms are able to take fuller advantage of profitable investment opportunities, their cashflow beta will be lower, which also brings about lower risk premiums (Campbell &Vuolteenaho, 2004) Furthermore, Lambert, Leuz, and Verrecchia (2007) argue that goodquality accounting reduces the cost of equity by (a) increasing the precision of earningsinformation and (b) directing corporate investments toward more productive uses Thus,because less defensive firms produce more persistent earnings due to exploitation ofgrowth opportunities, these firms are likely to exhibit lower cost of equity capital.6Takentogether, the earnings of less defensive firms are valued at a premium, leading to thehigher value relevance of their earnings

Takeover Defenses and Earnings Informativeness

The current literature examines the relationship between takeover defenses and earningsquality in several ways Zhao and Chen (2008a) use three measures: (a) the absolute value

of abnormal working capital accruals, (b) the absolute value of discretionary accruals, and(c) the timelier recognition of losses In another study, Zhao and Chen (2009) apply theintroduction of state antitakeover laws during the mid- to late 1980s as a natural experiment

to test the relationship between antitakeover protection and earnings quality They reportthat firms incorporating in states where the laws are passed tend to exhibit lower levels ofabsolute abnormal accruals and higher earnings informativeness Nevertheless, not one ofthe prior studies investigates the relationship between firm-level takeover defense measuresand earnings informativeness

Value relevance of earnings, or earnings informativeness, becomes a powerful measureespecially in investigating the information role of financial reporting (Holthausen & Watts,2001) Notably, the value relevance of earnings is a joint product of the supply (e.g., noise

in earnings signal) of and demand (e.g., perception of market participants) for earningsquality Our premise is that the lower value relevance of earnings in more defensive firmscan be attributed to stock investors’ impaired perception of earnings that are considered to

be of high quality in Zhao and Chen (2008a, 2008b, 2009) The firm-level analyses aredesirable to the extent that controlling for various firm characteristics is essential.7Amongothers, a proper control for growth opportunity is critical to mitigate potential measurementerrors in estimating abnormal levels of accruals

Given that Zhao and Chen’s (2008a, 2008b, 2009) findings are based on the absolutevalues of abnormal accruals, their results may have been driven by the presence of firmswith a high exposure to actual takeover threats (i.e., less defensive firms) rather than bymanagers’ pursuit of a quiet life in more defensive firms Less defensive firms are lesslikely to boost earnings because firms with high earnings are likely to be regarded as morepalatable targets by acquirers Rather, these firms may have an incentive to decrease earn-ings to thwart hostile takeovers (Watts & Zimmerman, 1986), which also leads to a greaterabsolute value of abnormal accruals However, this may not have resulted from a greaterrisk of subsequent detection, as income management by less defensive firms can attractenhanced scrutiny from various market monitors, including potential acquirers and regula-tors Relative to more defensive firms, less defensive firms therefore face higher expectedcosts from manipulating earnings.

However, as suggested by Zhao and Chen (2008a, 2008b, 2009) in their study, firmvalue of more defensive firms is lower than that of less defensive firms The quiet life viewsuggests that entrenched managers prefer to avoid costly efforts to maximize shareholderwealth, which results in low firm value Taken as a whole, we encounter a rare opportunity

to see the process by which managerial entrenchment affects the value relevance of ings, which arguably appears to be of high quality The value relevance of earnings incor-porates not only financial reporting quality itself but also managerial entrenchment thataffects the shares of firm value by an investor If the financial reporting quality dominatesthe managerial entrenchment in forming investor perception, we expect greater ERC inmore defensive firms On the contrary, if the managerial entrenchment dominates, weexpect a smaller ERC in more defensive firms We propose the following research hypoth-esis in the nondirectional form, which examines the association between takeover defensesand the value relevance of earnings

earn-Hypothesis 1: Ceteris paribus, takeover defenses are associated with the value vance of earnings

rele-Variable Measurement and Research Design

Variable Definitions

To mitigate concerns over the lack of consensus on an empirical measure that directly tures the agency conflicts between shareholders and management, antitakeover measureshave been investigated more thoroughly in recent years Gompers et al (2003), for exam-ple, have developed a unique measure of corporate governance (G-index) in which, forevery firm, one point is added for every antitakeover amendment that restricts shareholderrights This index serves as a proxy for the balance of power between shareholders andmanagers A higher G-index score indicates more provisions against shareholder rights;thus, a firm with a high score is considered to be poorly governed and vice versa As moreantitakeover provisions are likely to isolate managers from the market for corporate control,this lack of market monitoring can result in entrenched management, and hence severeagency problems Given the comprehensiveness and representativeness of this index, manyrecent articles have considered it to be a suitable proxy for a firm’s corporate governancequality (e.g., Ashbaugh-Skaife et al., 2006; Dittmar & Mahrt-Smith, 2007).8

cap-This article takes all the firms included in IRRC publications as its sample The IRRCcovers more than 1,800 large firms in six publication years: 1990, 1993, 1995, 1998, 2002,

and 2006 Consistent with prior research, we assume that the G-index changes little fromone publication year to the next We also obtain financial data and stock returns fromCompustat and the Center for Research in Security Prices (CRSP), respectively The datasource for institutional holdings is the CDA Spectrum, which collects information from theU.S Securities and Exchange Commission 13F filings.

Earnings Informativeness

We first estimate the traditional ERC model, which assumes that both the earnings leveland changes in earnings help to explain changes in stock prices Rather than including theearnings level and earnings changes, we make use of current and lagged earnings, a modifi-cation that allows for easier interpretation of the results because the coefficient of currentearnings captures both the permanent and transitory portions of earnings informativeness.9This modification still supports Easton and Harris’s (1991) argument that the earnings leveland the change in earnings complement, rather than substitute for, each other (Biddle,Seow, & Siegel, 1995) It is also consistent with the notion that multiple proxies for unex-pected earnings are likely to reduce ERC measurement error

The next step is to assess the earnings informativeness that is conditional on takeoverdefenses Previous literature uses ERC as an important measure of such informativeness(e.g., Collins & Kothari, 1989) The primary proposition of the present research is that cor-porate governance affects ERC through its impact on the perceived credibility of account-ing earnings Thus, the following model is estimated to examine the association betweenearnings informativeness and corporate governance

Rit5a01a1Eit1a2Eit11a3Git1a4Eit3Git1a5Eit13Git1Sb1*Controls

where

Previous literature reports that firm-specific factors affect ERC (e.g., Collins & Kothari,1989) We consider six determinants of earnings informativeness: firm size, leverage, lossincidence, growth, systematic risk, and earnings persistence The inclusion of Size is

Rt= 12-month buy-and-holding raw returns ending 3 months after the fiscal

year-end at t

Et= income before extraordinary items for year t, scaled by the market value

of equity at the end of t 2 1

Gt= G-index from Gompers et al (2003); this variable is standardized to fall

between 0 and 1Control variables

Sizet= natural log of total assets at t

Leveraget= firm leverage at t, measured by total liabilities divided by total assetsLosst= one if Et\ 0 and zero otherwise

motivated by the political cost theory Leverage is included because highly leveraged firmsare more likely to allow latitude in accounting to avoid debt-covenant violations (DeFond

& Jiamvalbo, 1994) The inclusion of Loss, M/B, Beta, and Pers is primarily motivated byvaluation considerations (Warfield et al., 1995) The inclusion of such factors as controlvariables serves to reduce the bias in the cross-sectional coefficients caused by correlatedomitted variables These factors are described in literature as the determinants of ERC(e.g., Cho & Jung, 1991) The coefficient of interest is a4 If a4 is greater (smaller) thanzero, it means firms with more takeover defenses exhibit higher (lower) value relevance ofannual earnings after controlling for typical determinants of ERC

To establish a connection between the G-index and ERC, the endogenous aspect of thedecision that firms make when choosing the extent of their takeover defenses is taken intoaccount To address the endogeneity issue, we first estimate a determinant model in whichthe dependent variable is G On the right-hand side of the regression model are the variablesthat are likely to influence firms’ decisions to include antitakeover provisions.10 A linearregression model does not fit well in terms of the raw value of the G-index because it is notcontinuous but rather discrete In this case, no assumption of the normality of the distur-bances can be made, and the prediction formulae that are deduced from the linear modelwould give impossible values (Gourieroux, Monfort, & Trognon, 1984) We thereforereplace the raw value of the G-index with the scaled decile rank and deflate it to ensure that

it falls between 0 and 1, which is expected to render interpretation much easier As a ness check, we also estimate a Poisson regression model to resolve these concerns about thelinear regression model Further to our discussion in the previous section, we hypothesizethat firms with fewer growth opportunities, less investment, more reliance on the debtmarket, and a higher degree of industry concentration are likely to be entrenched These pre-dictions are based on the notion that entrenched firms do not need to attract shareholdersand that the entrenchment decision is optimal for them For example, firms with moregrowth opportunities, and thus, a greater need for financing equity capital or less reliance onthe debt market, tend to prefer a shareholder-friendly governance structure.11 Throughoutthe article, our demonstration is based on the fitted rather than the raw value of G.Nevertheless, the use of the raw value of G produces qualitatively similar results.12

robust-In addition, we also take two important equity valuation issues into account First, weaddress the nonlinear association between stock returns and accounting earnings Althoughthe foregoing analysis is based on traditional linear returns-earnings regressions, prior stud-ies show that the stock return is a nonlinear function of earnings.13Hayn (1995), for exam-ple, demonstrates that a loss acts as a trigger for the liquidation option and that investorsrespond differently to losses and profits in pricing earnings Even though the profit versusloss binary classification is insufficient to reflect the convex relationships between returnsand earnings documented in prior studies, it represents a significant, simple, and powerfulearnings heuristic (Pinnuck & Lillis, 2007) To incorporate this value difference betweenprofits and losses, we replace the traditional ERC model with a piecewise linear regressionmodel by allowing the ERCs to vary across loss- and profit-reporting firms

Second, the negative association between takeover defenses and the value relevance ofearnings may result from different speeds of information processing across takeover defenseactivities Even though the ERC specification assumes market efficiency, the violation ofsuch efficiency with regard to takeover defense activities is common in literature For exam-ple, Gompers et al (2003) report that abnormal returns are greater for firms with strongshareholder rights than for those with weak shareholder rights Their findings are striking inthat, although strong shareholder rights may lead to good operating performance, no

relationship between shareholder rights and stock returns under the assumption of an cient market should exist (Core et al., 2006) With respect to this ‘‘governance-based anom-aly,’’ Gompers et al argue that weak shareholder rights result in agency costs and thatinvestors fail to fully anticipate these costs when they form their expectations of future cashflows.

effi-Thus, takeover defenses probably induce market mispricing of accounting earnings Thedelayed response to earnings of more defensive firms, in particular, may weaken the valuerelevance of earnings in such firms, thus inducing the downward biases in their ERC Torule out such potential mispricing, the Mishkin’s (1983) test is used, which aims to directlyexamine whether market participants fail to recognize the differential persistence of earn-ings across takeover defense activities The results of this test are discussed in the sectiontitled ‘‘Empirical Results.’’

Sample Description

The initial sample includes all firms in the Compustat and CRSP databases from 1990 to

2006 The missing observations are excluded as well as outliers for certain variables Theseoutliers are removed by excluding the extreme observations in the 1% left and right tails ofthe distribution for annual returns, earnings, and earnings components To estimate themarket beta, the firm-year observations that do not have at least 36 past monthly returnsare deleted The other variables are also winsorized at the top and bottom 1% The intersec-tion of these databases and the additional data requirements yields a sample that contains16,806 firm-year observations for the period 1990 to 2006

Table 1 presents the descriptive statistics of the data used in this article.14The summarystatistics reported in Panel A indicate that, on average, the sample firms have approxi-mately nine antitakeover provisions, consistent with Gompers et al (2003) The mean(median) annual return is 12.8% (9.5%) and that of earnings is 3.1% (5.2%) of the marketvalue in the previous year Consistent with previous studies, annual returns are rightskewed, whereas earnings are negatively skewed because of conservatism

The mean sales growth rate is 7.7% and that of the market-to-book ratio is 2.976, whichindicates the presence of substantial growth opportunities Capital investment and R&Dexpenditure have means of 6.5% and 3.1% of total assets, respectively Firm age rangeswidely from 3 to 19 years.15Firm size, a natural logarithm of market capitalization, rangesfrom 3.943 to 11.283, suggesting a substantial spread in firm size Loss incidence has amean of 16.6%, which indicates a small fraction of firms incurring a loss.16More than half

of the firms pay cash dividends, given that the median value of Div is greater than zero.Panel B of Table 1 reports the average values of the variables across different takeoverdefense activities We split the sample firm-year observations into quintiles (G1 to G5) ofthe G-index, which reveals a number of interesting patterns All of the variables, except for

R and HHI, show significant differences between G1 and G5 In terms of earnings quality,although discretionary accruals show no difference between the two extremes, earnings per-sistence clearly decreases with the governance quintiles The lower earnings persistence ofmore defensive firms affirms the growth- and investment-based explanation for their lowervalue relevance of earnings

The levels of operating cash flow, leverage, firm age, and cash dividends increase with

G, whereas sales growth, market-to-book ratio, R&D expenditure, and the market betadecrease with it The proportion of firms audited by big auditors is higher in the entrenchedgroups, which is consistent with the finding of Fan and Wong (2005) that firms exposed to

Table 1 Descriptive Statistics

Panel A Summary Statistics of the Variables (N = 16,806)

Panel C Correlation Matrix Between G-Index and Firm Characteristics

vari-*, *vari-*, and *** are significant at the 10%, 5%, and 1% levels, respectively.

greater agency risk tend to hire large auditors to mitigate concerns over agency problems.Firms with a high G-index score also tend to be at a mature stage in the corporate lifecycle; they report a lower market-to-book ratio, less capital investment, and less R&Dexpenditure These firms tend to be larger and concentrate more on debt financing thanequity financing Furthermore, they are likely to be less risky in terms of total risk Finally,these systematic differences are related to insiders’ choice of corporate governance devices.Panel C of Table 1 depicts the Pearson and Spearman correlation matrix of the variableswith the G-index and concurs with the implications of Panels A and B.

Overall, the results presented in Table 1 confirm that governance structure reflects ous firm characteristics Firms with fewer antitakeover provisions are generally youngerand smaller than those with more antitakeover provisions These results also suggest thatmore takeover defense activities are associated with fewer growth opportunities, less invest-ment, higher leverage, a higher earned-to-equity ratio, a greater propensity to pay cash divi-dends, and less systematic risk

vari-Empirical Results

Impact of Takeover Defenses On Earnings Informativeness

Table 2 presents the results of the main regression, which corroborate the fact that the stockmarket reacts more to the earnings of less defensive firms.17 Even after controlling for firmsize (Size), leverage (Leverage), loss incidence (Loss), market-to-book ratio (M/B), marketbeta (Beta), and earnings persistence (Pers), the results remain robust In the second column

of Table 2, the coefficient of corporate governance (G) and earnings (E) is 20.331 (t =22.010) To gain an understanding of the economic importance of the results, we use the esti-mated regression of Equation 2, set the level of earnings at the first quartile (0.023), and thenestimate how much the stock return would change if G increases from 0 to 1 We find thatwhen earnings increase from the first quartile by the interquartile value (0.052), the level ofstock returns increases by 30% in less defensive firms and by 12% in more defensive firms

To reflect the nonlinear association between R and E, we replace the traditional ERCmodel with a piecewise linear regression model by allowing the ERCs to vary across theloss- and profit-reporting firms in column 3 The results show that the impact of G-index

on ERC is not symmetric between profits and losses We find a lower value relevance ofprofits and a higher value relevance of losses in more defensive firms compared with theirless defensive counterparts This evidence suggests that investors value profits at a discountand penalize losses more in such firms The higher response coefficients of losses in moredefensive firms indicate that investors expect these losses to continue in the future (Basu,1997; Hayn, 1995)

Column 3 supports the nonlinear value implication of earnings It indicates that there aresignificant differences between pricing profits and losses and that the coefficient of theinteraction of G and E is significantly negative Meanwhile, the ERC of the positive earn-ings in the best-governed firms is 3.832 (t = 15.990), and the positive earnings of worst-governed firms are discounted by 21.353 (t = 23.670) The ERC of the negative earnings

in the less defensive firms drops to 0.539 (3.832 – 3.293) Interestingly, the three-way action variable (G 3 Loss 3 E) exhibits positive coefficients (1.304, t = 3.070), whichindicates that the losses reported by more defensive firms are penalized more severely.Previous studies report that the smaller slope ERCs in the lower profitability rangesresult from (a) closeness to the liquidation option of shareholders (e.g., Burgstahler &