The impact of audit quality on cash incentive compensation and cost of capital

The impact of audit quality on cash incentive compensation and cost of capital

I use the agency theory framework to model the effect of audit quality on cipal agent dynamics The model shows that as audit quality increases, the weightplaced on accounting earnings in determining managers compensation increases Ialso show that the returns expected by the investors decline with increasing au-dit quality Hence the cost of capital to the firm decreases with increasing auditquality I find empirical evidence to support the two propositions The empiricalresults confirm that the weight placed on accounting earnings in determining theincentive pay of both CEOs and non-CEO managers increase with audit quality.The evidence also confirms that cost of capital of firms declines as audit qualityincreases although this effect is only observed in small firms I further test theimpact of audit tenure and auditor opinion on cost of capital and find that cost

prin-of capital is inversely proportional to auditor tenure, and that cost prin-of capital forthe client will increase if the auditor issues any opinion other than an unqualifiedopinion

UMI Number: 3295518

3295518 2008

Copyright 2007 by Fernando, Guy D.

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346 All rights reserved.

by ProQuest Information and Learning Company

Copyright 2007 Guy D Fernando

All rights reserved

1.1 Introduction 1

1.2 Literature Review 5

1.2.1 Compensation 5

1.2.2 Cost of Capital 10

1.2.3 Audit Quality 12

1.2.4 Other audit and auditor attributes that impact Cost of Capital 16 1.2.5 Summary 20

2 Analytical model - Effect of Audit quality on Incentive Compen-sation and Cost of Capital 21 2.1 Model 21

2.2 Results 22

3 Empirical Study 1 - Effect of Audit Quality on Incentive Com-pensation 27 3.1 Hypothesis development 27

3.2 Data and Research Design 28

3.2.1 Data 28

3.2.2 Research Design 28

3.3 Results and Analysis 34

3.3.1 Descriptive Statistics 34

3.3.2 Control Variables 36

3.3.3 Specialist auditors 37

3.3.4 BigX auditors 39

3.3.5 Analysis with both Specialist and BigX variables 40

3.3.6 Audit quality and non-CEO Executive compensation 41

3.3.7 Sensitivity Analysis 42

3.3.8 Fixed Effects Regression 43

3.4 Alternative Explanations 44

3.4.1 Governance 44

3.4.2 Earnings Quality 46

3.5 Summary and Discussion 48

4 Empirical study 2 - Effect of Audit Quality on Cost of Capital 49 4.1 Hypothesis development 49

4.2 Data and Research Design 50

4.2.1 Large Firms and Small Firms 51

4.2.2 Specialization 51

4.2.3 Cost of Capital - re 52

4.2.4 Regression Models 53

4.3 Empirical Results 55

4.4 Robustness checks 61

4.5 Summary and Discussion 64

8 Appendix B - Computation of the Economic Impact of Auditor

9 Appendix C - Data Definitions for tables 1 - 11 87

10 Appendix D - Data Definitions for tables 12 - 17 88

List of Appendices

Appendix A - Detailed model derivation 78Appendix B - Economic Impact of Auditor Specialisation on

Compensation 83Appendix C - Data Definitions for tables 1 - 11 87Appendix D - Data Definitions for tables 12 - 17 88

List of Tables

Table 1-Descriptive Statistics 89

Table 2 - Correlation Coefficients 90

Table 3 - Regression of CEO compensation against Control variables Full sample of clients 91

Table 4 - Regression of CEO compensation against audit quality BigX clients only 92

Table 5 - Regression of CEO compensation against audit quality Full sample of clients 93

Table 6 - Regression of CEO compensation against audit quality Full sample of clients 94

Table 7 - Regression of non-CEO Executive compensation against audit quality BigX clients only 95

Table 8 - Regression of Change in CEO compensation against audit quality BigX clients only 96

Table 9 - Regression of CEO compensation against audit quality Fixed Effects model BigX clients only 97

Table 10 - Regression of Change in CEO compensation against audit quality Fixed Effects model BigX clients only 98

Table 11 - Regression of CEO compensation against audit quality after controlling for Governance and Earnings Quality BigX clients only 99

Table 12 - Breakdown of the dataset 100

Table 13 - Descriptive statistics 101

Table 14 - Correlation Coefficients 103

Table 15 - Cost of Capital regressed against BigX Full sample 104

Table 16 - Cost of Capital regressed against SPX BigX clients only 105

Table 17 - Cost of Capital regressed against both BigX and SPX 106

I would like to express my deepest gratitude to my advisor and mentor, ProfessorAlex Thevaranjan His guidance and support contributed immeasurably to mycareer at Syracuse University My thanks also go out to the rest of my committee,Professors Bill Brown, Hsihui Chang, Randy Elder and David Richardson

I acknowledge the funding I received from Syracuse University over the lastfour years, which enabled me to continue with my studies without having to worryabout finances

I would also like to express my appreciation to the rest of the faculty in mydepartment, both past and present; notably Professors Mo Onsi, Badr Ismail,David Harris and Anver Ahmed

My time as a PhD student at Syracuse University was made a lot easier by myfellow PhD students My thanks go out to all of them, especially Scott Duellmanand Ahmed Abdel-Meguid

I gratefully acknowledge my wife Jasmine who stood by me during these fouryears, and our daughter Jayathri who appeared during the middle of this periodand turned my life upside down Thank you Amma (mother) Thaththa (father)and Nangi (sister) for your support and encouragement throughout my life Specialthanks also go out to my mother-in-law who came over to help us during the mosthectic phase of my PhD

Finally I thank all of you who had faith in me I will do my best to fulfill yourexpectations

of a firm can be mitigated Accordingly, in this paper I investigate the effect thataudit quality would have on managerial incentives and cost of capital of the firm.Motivating an agent to take actions in the best interest of the principal is

an important area of research in several academic disciplines including finance,accounting, economics, management etc It is widely accepted that performancelinked compensation is the most effective way of obtaining such congruence, whenthe managerial effort is unobservable by the principal (Jensen and Meckling 1976;Murphy 1999) Accounting measures of performance, most notably earnings arewidely used in compensation contracts to link pay to performance (Lambert 1993).But accounting earnings can be easily manipulated by the management (De-chow et al 1995) In fact Healy (1985) finds that managers engage in earningsmanagement specifically to impact their incentive payments However, if there is

an external auditor, who can assure the quality of earnings, the principal will havemore faith in earnings and place greater weightage in that measure in designingthe managers’ compensation contract Prior research shows that specialist audi-tors perform a higher quality audit1 than non specialist auditors (Krishnan 2003;

1 DeAngelo (1981) defines audit quality as ”‘the ability to detect and the willingness to port breaches in the clients accounting system.” Hence, audit quality becomes a function of the technical expertise of the auditor, which will enable him to detect ’breaches’ and the auditors political power vis-a-vis the managers of the firm that will enable him to report the ’breaches’

re-to the shareholders.

Balsam et al 2003) It also shows that higher quality auditors ensure a lowerlevel of earnings management resulting in higher quality earnings (Becker et al.1998) According to agency theory the improvement in earnings quality makes theearnings measure more attractive for contracting purposes Hence, I argue thatspecialist auditors will cause the principal to place a greater weight on earnings indetermining the managers’ compensation compared to non-specialist auditors.

I also investigate the effect of the two audit quality attributes, size and industryspecialization, on the cost of capital of client firms Specifically, I examine whichroles of auditing affect the cost of capital of the client firm The four largest audi-tors dominate today’s auditing market In such a setting, auditing firms will rely

on quality differentiating strategies that enable them to expand and/or maintaintheir client base One of these strategies is industry specialization Thus my studygoes beyond the traditional BigX -non-BigX size dichotomy quality differentiation,

to the more timely issue of quality differentials among the BigX group.2

The auditor provides ”reasonable assurance” that the financial statements arefree from ”material misstatements” Auditing dilutes the adverse effects of theseparation of ownership and control (Jensen and Meckling 1976) Thus auditing

is a means of reducing information risk for users of financial statements Thisrisk reduction should be matched by a reduction in the cost of capital for thefirm (Lambert et al 2006) The demand for auditing in capital markets could beanalyzed from three different perspectives (i.e auditing roles); a monitoring role,

2 BigX firms refer to the large public accounting firms that perform most of the audits for publicly traded firms The BigX were called the Big8 audit firms in the 1970s and the 1980s The Big8 consisted of Arthur Andersen, Arthur Young, Coopers & Lybrand, Ernst & Whinney, Deloitte, Haskins & Sells, Peat Marwick International, Price Waterhouse & Touche Ross In

1989, Ernst and Whinney merged with Arthur Young to become Ernst and Young and Deloitte, Haskins & Sells merged with Touche Ross to become Deloitte & Touche From 1989, the BigX were referred to as the Big6 Price Waterhouse merged with Coopers & Lybrand in 1998 to become PricewaterhouseCoopers and the Big6 became the Big5 Arthur Andersen collapsed after the firm’s indictment for obstruction of justice involving Enron in 2002, and the Big5 became the Big4 Peat Marwick International became KPMG Hence the Big4 now consists of PricewaterhouseCoopers, KPMG, Ernst & Young and Deloitte & Touche.

an information role and an insurance role (Wallace 1980) How the auditor fulfillsthese roles determines the level of audit quality.

DeAngelo (1981) posits that the BigX auditors provide better audit qualitythan non-BigX firms Audit quality is defined as the joint probability that theauditor will (1) detect ”breaches” in the client’s accounting system, and (2) willreport them It is widely accepted in the literature that Big X auditors provide,

or are perceived to provide, higher levels of audit quality (e.g., Teoh and Wong

1993, Becker et al 1998, Francis et al 1999) After a series of mergers amongthe Big X and the recent collapse of Andersen, the Big 4 audit the majority ofpublicly traded companies in the US Thus an interesting and relevant question to

be addressed is whether there are quality differentials among the BigX in terms ofthe effectiveness of performing the three roles discussed above

Khurana and Raman (2004) (hereafter KR) show that clients of BigX auditorshave significantly lower cost of capital compared to clients of non-BigX auditors

in the US but not in other Anglo-American countries They argue that the ception of BigX performing higher quality audits than non-BigX is a function ofthe litigation environment Their conclusion is that the threat of litigation is astronger driver than reputation behind perceived audit quality, proxied by cost ofcapital This conclusion implies that the investing public primarily perceives auditquality in terms of the BigX auditor’s ”deep pockets” Thus their paper primarilystudies the firm size audit quality attribute from an insurance role perspective.This role is magnified in a more litigious environment like the US, but may be ofless importance to investors in other environments (i.e UK, Australia or Canada).The main purpose of the multi-country analysis used by KR was to examinehow litigation environments moderate the effect of auditor size on perceived auditquality KR analyze audit quality from a litigation exposure versus reputationperspective I extend their work and attempt to disentangle insurance from non-

per-insurance roles of auditing I argue that the monitoring and information roles ofauditing are related to the technical aspects of auditing These technical aspectscorrespond to both the auditor’s competence and independence as described byDeAngelo (1981) On the other hand the insurance role is related to the compen-satory aspects of the audit Thus the latter role is merely a function of the level

of financial endowments of the auditor regardless of its technical qualities

I further analyze the type of firms that may benefit from having a higher qualityaudit The effect of firm size in the available information environment of the firm

is well documented (e.g., Atiase 1985; Bamber 1987; Llorente et al 2002) Thesepapers show that larger firms have a better information environment compared

to smaller firms Hence, the incremental value of audit quality will be greater forsmall clients than for large clients I analyze the KR results to examine whetherclient size plays a role in the effect of audit quality on cost of capital

By holding the litigation environment constant, any differences in the level ofperceived audit quality (i.e cost of capital) is likely to be a function of the mon-itoring and information roles rather than the insurance role Therefore I select

an audit quality attribute that only relates to the first two non-insurance roles,namely industry specialization Industry specialization refers to industry-specificknowledge and expertise an auditor gains from extensive auditing in any one indus-try Empirical evidence shows that industry specialization is an important auditquality attribute that varies among Big X auditors (Balsam et al 2003; Krishnan2003; Francis et al 2005)

I find that small firms in the sample drive the results shown in KR Lower cost

of capital associated with clients of BigX auditors is limited only to small clients.Furthermore, the results confirm that the ex − ante cost of capital is lower forsmall clients who engage specialist BigX auditors compared to clients who engagenon specialist BigX auditors

I further examine the impact of different attributes of the auditor and the audit

on the cost of capital of the client firm I find that the cost of capital is inverselyproportional to the length of the auditor tenure I also find that the client firmscost of capital increases if the auditor issues any opinion other than an unqualifiedopinion Again these results are driven by the smaller firms

This study contributes to the literature in the following ways First, I show thatthe weight placed on accounting earnings in determining executive cash compen-sation increases with audit quality Next, I show that audit quality lowers the cost

of capital for small clients Finally, I examine the effect of industry specialization

as an audit quality attribute in two new settings, namely executive compensationand the cost of capital

The paper is structured as follows The remainder of Chapter 1 will containthe literature survey Chapter 2 describes the analytical model Chapter 3 consists

of the empirical analysis of compensation and audit quality, including the data,empirical results, sensitivity analysis and a discussion of the results Chapter 4contains the empirical analysis of cost of capital and audit quality Chapter 5discusses the conclusion

3 Murphy (1999) says that ”CEO pay research has grown even faster than CEO paychecks”.

4 A search on Lexis-Nexis for CEO pay for May 2006 shows 22 hits.

academics, practitioners, regulators and the public I attempt to provide additionalinsight to this evolving field of executive compensation.

In the modern business environment, the functions of ownership and ment are separate In such circumstances, classic agency theory states that theowner (principal) can get the manager (agent) to maximize the principal’s utilitythrough monitoring and incentive alignment (Jensen and Meckling 1976) I inves-tigate the effect of audit quality on the monitoring aspect and incentive alignmentaspect of managerial performance

manage-Theoretical research into agency theory shows that to motivate risk aversemanagers into exerting effort so as to maximize the utility to the principal, themanagers’ incentives should be aligned to the performance of the firm Holm-strom (1979) shows that when the ultimate outcome alone is observable (and notmanagers effort), the optimal contract will be the ’second best contract’, meaningthe contract will be based on the observable outcome Banker and Datar (1989)show that different signals of firm performance can be aggregated into an optimalmeasure of performance on which to base the compensation contract Jensen andMurphy (1999: p242) summarize the essence of agency theory as ”‘Agency theorypredicts that an optimal contract will tie the manager’s expected utility to prin-cipal’s wealth; therefore agency theory predicts that CEO compensation policieswill depend on changes in shareholder wealth”’

There is a long stream of literature that empirically analyzes the sensitivity ofCEO pay to firm performance Smith and Watts (1982) state that incentive plansexplicitly tie CEO’s performance to changes in firm value They also state thatcash incentives are tied 0ex − ante0 to some measure of firm profit Lambert andLarcker (1987) analyze a sample of firms from the F orbes annual compensationsurvey for the period 1970-1984 and show that profits are strongly related to cashcompensation Jensen and Murphy (1990) show that CEO compensation is directly

proportional to increases in shareholder wealth, and that the pay increases by

$3.50 for every $1000 increase in shareholder wealth Sloan (1993) investigates therelative weights placed on earnings and stock returns in CEO cash compensation,and show that the relative weights depend on the noise of the two signals Morerecently, Core et al (2003) document the positive relationship between earningsand CEO pay

CEO compensation broadly consists of three attributes (Murphy 1999; Conyon2006) One is the base salary, which is one of two cash based components The basesalary is generally a fixed amount negotiated by the CEO beforehand The othercash component is the bonus, which is generally based on the ex-post performance.Finally there is a component of equity and long term stock awards In addition tothe stock options awarded annually, the CEO will also benefit from the increase invalue of his existing equity and options of the firm

According to Lambert and Larcker (1987) the cash component comprised around80%-90% of the total CEO compensation in the mid 1980s However, with the ex-plosive growth in CEO options, this proportion has been declining Conyon (2006)shows that the cash component (base salary + bonus) of CEO compensation was52% in 2003 Due to this decline, there is a tendency to ignore the importance

of the cash component in compensation However, it is important to note that asignificant proportion of compensation is in the form of cash, and there are stillsome parts of the puzzle that are not totally understood

The incentives for cash compensation and equity compensation are different.Cash compensation is generally linked to past performance Bonuses are mostlylinked to the prior period’s accounting numbers (Murphy 1999) Hence cash com-pensation shows a strong positive relationship to accounting numbers (Lambertand Larcker 1987; Sloan 1993) This indicates that the cash component of com-pensation is a reward for past performance

On the other hand, much theoretical work hypothesizes that equity sation is intended to induce risk-averse managers to take risks on behalf of share-holders (Jensen and Meckling 1976; Copeland et al 2005 pg 487; Murphy 1999).Equity compensation will provide incentives for the managers to maximize share-holder wealth by tying the managers’ compensation directly to increases in thestock price (Core et al 2003; Hanlon et al 2003) Hence, it is apparent that eq-uity compensation is a method to induce risk taking by the managers Smith andWatts (1992) show that firms in industries with greater investment opportunitieswill be more likely to use stock options in compensating their CEOs Rajgopaland Shevlin (2002) analyze the risk taking behavior of a sample of managers fromthe oil and gas industry and show that equity compensation is related to risk tak-ing If so, it would mean that equity compensation will not be related to pastaccounting performance Instead, it will be related to future performance of thefirm Hanlon et al (2003) show that equity compensation is related to futureearnings, while Baber et al (1996) and Core et al (1999) find that there is no re-lationship between equity compensation and past accounting performance Hence,prior research illustrates two distinct and separate motives for cash and equitycompensation.

compen-Lewellen et al (1987) explicitly states this dichotomy in the composition ofCEO compensation

”Consider first, the role of immediate forms of compensation - salary and currentbonus5- in the pay package By whatever formula these payments are established, thepayoff to the executive/recipient necessarily reflects only the firms revealed performance

up to the payment date; the amounts involved are fixed when awarded and as received areindependent of the firm’s subsequent performance Moreover bonus awards are typicallybased on short-term performance measures such as current year profits or return onequity The time horizon relevant to shareholders however is in principle unlimited sinceall future residual cashflows the firm is expected to generate should be impounded inshare prices Managers therefore may need to be given an explicit claim to those futurecash flows in order to encourage proper attention to decisions that will favorably affectthem

5 Salary and bonus is the cash component of compensation

This can be accomplished either by conveying to managers an equity interest in thefirm through some type of restricted stock compensation or by deferring a portion ofcash compensation.” - Lewellen et al (1987) pg 289.

This leads me to conclude that since past earnings will be important only in termining cash compensation, earnings quality will be important to cash compen-sation, but not to equity compensation Audit quality will impact past accountingreturns And since I am investigating the effect of audit quality on compensation,

de-I posit that audit quality will have greater impact on the cash component of CEOcompensation than on the equity component

Non-CEO Executives

Agency theory posits that the effect predicted for managers would apply tonon-CEO executives as well However, the non-CEO managers cannot directlycontrol the accounting profits of a firm, since each may be responsible only for asingle functional or geographical area of performance On the other hand, whenall non-CEO executives are considered in the aggregate, the common feature oftheir performance will be the impact on firm profitability Therefore, the questionarises whether it is fair to let the overall firm performance (calculated in terms ofROA or ROE) impact the compensation of non-CEO executives?

There has been relatively little research on non-CEO executive compensation,with most of it appearing in management literature When it comes to deter-mining the compensation of non-CEO executives, Balkin and Gomez-Mejia (1990)find that in a study of compensation of business units within firms, corporateprofitability is linked to the pay levels Fisher and Govindarajan (1992) analyzethe compensation of profit center managers and find that Return on Equity of thefirm is significantly and positively related to the compensation level of profit centermanagers

Lambert et al (1993) conduct a study on the compensation of 4 distinctorganizational levels (ranging from plant manager to CEO) on a sample of 303

publicly traded firms during the years 1982 - 1984 They use the overall firm ROA

as a control variable in the analysis and find that ROA is positive and significant.More recently, Carpenter and Wade (2002) find that prior ROA is also significantlyrelated to the compensation level of non-CEO executives

Several studies derive their hypothesis for executive compensation using agencytheory, and assume that firm profitability is a component of executive compensa-tion But the studies do not limit the executive to CEOs Antle and Smith (1986)use the 3 most highly paid executives of the firms in their sample Lewellen andHuntsman (1970) derive their hypothesis that the executives’ pay is related in alinear fashion to both profits and sales of the firm without making any distinctionbetween CEOs and non-CEO executives Although they only use the single highestpaid executive’s pay for each firm, they explain this by stating

”‘ As a matter of both convenience and efficiency, the compensation of the singlehighest paid executive in each firm in each year is taken into account here as the depen-dent variable measuring the size of the firm’s managerial pay package While it may seemmore appropriate that the remuneration of all the senior policy-making individuals in acorporation be tested for a relationship to company performance, it happens that thepay of a firm’s top man is a suitable surrogate for the pay of his closest subordinates interms of their relative standing vis-a-vis corresponding officials in other firms - Lewellenand Huntsman (1970), pg 714”’

The above review shows that even non-CEO executive compensation is affected

by the overall firm performance In this study, I attempt to answer the question;

if the quality of the performance measure improves (as will accounting measures

in the face of better quality audit), will there be a greater weight placed on thatperformance measure in determining the compensation of non-CEO executives aswell?

1.2.2 Cost of Capital

Cost of capital to the firm is the return expected by the investors when they investmoney in the firm (Lambert et al 2006) The rational investor will expect a return

in accordance with the CAPM model, that will cover the risk free return he couldobtain by investing in risk free securities and a risk premium for investing in therisky assets of a firm A firm’s risk is a function of external forces (GDP growth,inflation, political situation etc) as well as internal forces (managerial performance,agency problems etc).

The separation of ownership and control of the firm leads to information metries between the owners and manager of the firm (Jensen and Meckling 1976)

asym-In addition rational expectation theory and agency theory suggest that the cipals (investors) and agents (managers) would have divergent interests resulting

prin-in a moral hazard problem Wallace (1980) argues that prin-investors price-protecttheir investments resulting in a reduced stock price, which implies a higher cost ofcapital

As information asymmetries are reduced via more transparent and reliable formation, the risk assumed by investors should decline This would ultimately lead

in-to less price protection Barry and Brown (1985) suggest that better informationcan reduce the rate of return demanded by the investors by reducing estimationrisk Merton (1987) suggests that better information reduces cost of capital byimproving risk sharing Lambert et al (2006) state that better information willlead to lower cost of capital due to better alignment between the firm’s investmentopportunities and its investment choices Easely and O’Hara (2004) show in theiranalytical paper that investors will demand higher returns (and hence increase thecost of capital) from firms which have lower public information compared to theprivate information available

Several empirical studies have reinforced the analytical work mentioned above.Botosan (1997) show that the level of disclosure provided in the annual reports of

a sample of 122 manufacturing firms in 1990 is negatively related to cost of capital.This result is confirmed in Botosan and Plumlee (2002) Francis et al (2004) find

that firms with better quality earnings enjoy a lower cost of capital.

Standard setters have stressed the connection between improved informationquality and reduced cost of capital Arthur Levitt (1998), the former chair ofthe Securities and Exchange Commission stated that ”high quality accountingstandards [ ] reduce capital costs.” More recently, John M Foster (2003), a formermember of the FASB mentions in the FASB report that

” Numerous academic studies have concluded that more information in the ketplace lowers the cost of capital Upon reflection, although it is nice to have empiricalsupport, academic studies are not really necessary to reach this conclusion; it is intu-itive More information always equates to less uncertainty, and it is clear that people paymore for certainty Less uncertainty results in less risk and a consequent lower premiumbeing demanded In the context of financial information, the end result is that betterdisclosure results in a lower cost of capital.”

mar-Hence, since higher information levels lead to lower cost of capital, I investigatethe impact of audit quality in improving the information environment, and thuslowering the cost of capital for firms

1.2.3 Audit Quality

The auditor provides ”reasonable assurance” that the financial statements arefree from ”material misstatements” Auditing, to a certain extent, mitigates theagency problems arising out of the separation of ownership and control (Jensen andMeckling 1976) Wallace (1980) describes three main roles of auditing, namely themonitoring role, the information role and the insurance role

The Monitoring Role of Auditing and Cost of Capital

The monitoring role for auditing implies that auditing will ensure better use

of resources entrusted to the agent by the principal Jensen and Meckling (1976)state that one component of the agency cost is the cost of monitoring the man-agers Chow (1982), in a study of firms publicly traded in 19266, found that the

6 The use of a sample from 1926 controls for the mandatory engagement of an auditor that was later required by the 1933 and 1934 Securities acts.

probability of engaging an auditor is increasing in the degree of conflict amongthe stakeholders of the company In addition, Francis and Wilson (1988) showthat the demand for quality differentiated audits, proxied by auditing firm size ispositively related to the company’s agency costs Thus I posit that the monitoringrole of the audit is effective in mitigating the agency problem arising out of theseparation of ownership and control of the firm The effectiveness of the auditor

as monitor on behalf of the owners will be reflected in the risk premium demanded

of the firm by the owners, and will lead to a lower cost of capital I also posit thatthis monitoring effect will manifest itself in the CEO compensation awarded bythe firm

The Information Role of Auditing and Cost of Capital

As previously mentioned, one of the adverse effects of the separation of ship and control is information asymmetries between managers and investors Asinformation asymmetries are reduced via more transparent and reliable informa-tion, the risk assumed by investors should decline This would ultimately lead toless price protection Barry and Brown (1985) suggest that better information canreduce the rate of return demanded by the investors as it reduces estimation risk.Merton (1987) finds that better information reduces cost of capital by improvingrisk sharing Lambert et al (2006) state that better information will lead to lowercost of capital due to better alignment between the firm’s investment opportuni-ties and its investment choices Principal agent theory states that if investors hadperfect information about the managers and the states of the world, an enforceablecontract can be implemented However, in environments of imperfect information,the level of information will be tied to incentives based on observable performancemeasures The better the quality of the performance measure, the greater theweight placed on the performance measure

owner-The Insurance Role of Auditing and Cost of Capital

The insurance role postulates that risk to the investors will be reduced becausethe investors will have another source of compensation, namely the auditor, in theevent of failure of the firm Menon and Williams (1994) argue that this insurancefactor is built into the share price of the client firm They examined the impact ofthe bankruptcy of Laventhol & Horwath (L&H) on the stock prices of its clients.They document a significant negative effect on the stock prices and attribute this

to the deterioration of the insurance capacity of L&H Within the context of cost

of capital, KR document a significant negative relation between engaging a Big4auditor and the cost of capital of the client This negative relation was only found

in the US but not in other Anglo-American countries They attribute this finding

to the differences in litigation exposure between the US and the other countries.While this role does not have much bearing on managers’ compensation, it has

a significant bearing on the cost of capital of a firm The first two roles of the auditindicate that higher audit quality will lead to lower cost of capital for the clientfirm (technical aspect of auditing) The third role of the audit indicates that themore financial resources the auditor has, the lower the cost of capital of the clientfirm (compensatory aspect of auditing) Based on the above discussion, I arguethat the monitoring and information roles of auditing are ex ante roles Theseroles dilute the negative effects of agency problems on the pricing of stocks Onthe other hand, the insurance role of auditing is an ex − post role, which increases

in importance after some form of failure in highly litigious environments

Investors consider the deep pockets of auditors as safety nets in case of ring losses on their investments The number of audit failures and the associatedlawsuits are small relative to the number of audits of publicly traded companies.Therefore the insurance role of auditing, although important, should not over-shadow the importance of the monitoring and information roles of auditing thatdirectly relate to it technical aspects Thus I examine how the three roles of au-

incur-diting suggested by Wallace (1980) are captured by two main attributes of auditquality, audit firm size and industry specialization.

Measurement of audit quality

DeAngelo (1981) argues that BigX auditors provide better quality audits thannon-BigX auditors, which is supported by extensive subsequent empirical research.Teoh and Wong (1993) find higher earnings response coefficients for clients audited

by BigX firms compared to those audited by Non-BigX firms Becker et al (1998)and Francis et al (1999) demonstrate that BigX auditors are better at constrainingearnings management in their clients compared to non-BigX auditors by showingthat the latter have higher levels of discretionary accruals Elder et al (2004)show the same results in the context of commercial banks

In addition to the auditor size (or BigX non-BigX dichotomy) proxy for auditquality, another proxy that has been extensively studied in recent times is theauditor specialization Casterella et al (2004) describe auditor industry special-ization as ”A differentiation strategy whose purpose is to provide auditors with

a sustainable competitive advantage over nonspecialists” Krishnan (2003) andBalsam et al (2003) find that there is less earnings management in clients of spe-cialist BigX auditors compared to non-specialist BigX auditors by analyzing thediscretionary accruals of client firms Dunn and Mayhew (2004) find that clients

of specialist BigX auditors have significantly better AIMR (Association for vestment Management and Research) rankings than clients of non-specialist BigXauditors, signifying that the former have better quality financial reports

In-There is also empirical evidence from the governmental sector that supports theargument that industry specialization is an important audit quality attribute Deisand Giroux (1992) document a negative relationship between auditor specializationand quality control review outcomes In a similar study, O’Keefe et al (1994) find anegative relation between auditor specialization and Generally Accepted Auditing

Standards (GAAS) violations.

1.2.4 Other audit and auditor attributes that impact Cost of CapitalAuditor Opinion

Statement on Auditing Standards (SAS) No 59, ”‘The Auditor’s Consideration

of an Entity’s Ability to Continue as a Going Concern”’, requires auditors toevaluate whether substantial doubt exists about an audit client’s ability to continue

as a going concern It stresses that this information is an essential signal for users

of financial statements Prior research has indicated that the issuance of a goingconcern opinion by the auditor is likely to be preceded by increasing probability

of bankruptcy (McKeown et al 1991) Jones (1996) compares 68 audit reportsthat disclose going concern uncertainties during the period 1979 to 1988 with 86similar financially distressed firms that receive unqualified opinions during thesame period He finds that firms that receive a going concern opinion had negativeabnormal returns whereas the financially distressed firms that did not receive agoing concern report had positive abnormal returns over a 5 day window aroundthe release of the going concern opinion Carlson et al (1998) confirm these results.Furthermore, Firth (1980) document that a going concern opinion would impair thecredit rating of client firms The existing research indicates that a going concernopinion increases the risk to the investors in the client firm, which should result

in increased cost of capital Geiger and Raghunadhan (2001) explicitly speculatethat going concern opinion might result in increased cost of capital to the client

On the other hand, Bryan et al (2005) find that amongst firms that havefiled for bankruptcy, there is more probability of them emerging from bankruptcy

if they are issued a going concern opinion than if they are not The authorsspeculate that this is due to the signalling effect of the going concern opinion, thatwill bring the problems in the firm out into the open, and make the stakeholders

take action to resolve those problems Hence, it could be argued that bankruptfirms that receive a going concern opinion is actually a less risky proposition thanone that has not Therefore, it may not be possible to predict the direction of therelationship between going concern opinion and cost of capital.

I test this assumption empirically I find that any opinion other than an qualified opinion results in increased cost of capital for the client

un-Auditor Tenure

There has been a perception amongst the regulatory authorities that tors, over time, will develop stronger relationships with the clients, resulting in adeterioration of the audit quality This has led to the imposition of mandatoryauditor retirement in some countries (See Geiger and Raghunandan (2002) for adiscussion on this issue) However, academic research into this area finds contraryresults Research has found that there are more audit failures in the early years ofthe auditor-client relationship (Geiger and Raghunandan 2002) and shorter audittenure is associated with lower earnings quality (Johnson et al 2002; Myers et al.2003) Investors, too apparently acknowledge this fact and reward long auditor-client relationships with lower cost of debt (Mansi et al 2004) and better earningsresponse coefficients (Ghosh and Moon 2005)

audi-If longer audit tenure results in a better quality audit, and if the financialmarkets realise this, then the markets should reward the client firms with a lowercost of capital I find that as the auditor’s tenure increases, there is a significantdecline in the cost of capital for the client firm

Effect of audit quality on compensation

Agency theory states that performance based incentives can reduce the agencyproblem However accounting based performance measures, which are extensivelyused in compensation based contracts can be easily manipulated by the managers

In fact, research shows that managers manipulate earnings for the specific purpose

of influencing their bonus payments Healy (1985) in a seminal study finds thatmanagers will manage earnings around bonus caps Hence, if the earnings go abovethe cap, the managers will manage earnings downwards7 These results have beenvalidated by subsequent research such as Gaver et al (1995), Holthausen et al.(1995) and Guidry et al (1999).

In the context of the principal agent model, the effect of information asymmetrybetween the principal and the agent could be reduced through the use of highquality auditors The main effect this would have will be to place a greater weight

on the incentives based on observable performance measures Better monitoringeffect could have two opposite effects on the level of agent compensation Onthe one hand, better monitoring could reduce rent extraction by the agent, thusreducing the compensation level On the other hand, better monitoring mightassure the principal that unseen rent extraction (i.e: use of corporate jets) has beeneliminated or minimized and will enable higher levels of managers compensation.Furthermore, improved audit quality will lead to more good effort expended by themanagers, leading to increased desired outcome for the firm If the firm decided toshare some of it with the managers, the compensation level of the managers willincrease Hence, it is difficult to make an a priori prediction

Effect of audit quality on cost of capital

The three roles put forward by Wallace (1980) combined with literature lishing the connection between information quality and risk indicate that investorswill value a higher quality audit over a lower quality audit Balvers et al (1988)and Beatty (1989) document a negative relation between underpricing in IPOsand engaging a BigX auditor Pitman and Fortin (2004) show that clients of BigXauditors have a lower cost of interest immediately after an IPO than clients of

estab-7 The managers are putting the excess earnings away in the ’cookie jar’ to be used in the lean years Healy interprets this result as ’strong association between accruals (Healy’s measure of earnings management) and managers’ income-reporting incentives under their bonus plan’ (p 106).

non-BigX auditors BigX auditors have the financial resources to better fulfilltheir monitoring, information and insurance roles than non BigX auditors In thecontext of this study this better fulfillment of these roles is expected to be reflected

in a reduced cost of capital

Given the above contentions, cost of capital of a client audited by a BigX tor could be lower due to the monitoring role, the information role, the insurancerole or a combination of all three roles Disentangling the effects of such roles is

audi-a difficult taudi-ask, which requires very unique circumstaudi-ances For exaudi-ample, Menonand Williams (1994) were able to isolate the insurance role of auditing within thecontext of an auditor’s bankruptcy Another approach would be to identify anaudit quality attribute that is unlikely to be related to a specific auditing role Iargue that industry specialization is not related to the insurance capacity of theauditor On the other hand, specialization would be related to the monitoring andinformative abilities of the auditor

Small firm effect on cost of capital

Small firms have poorer information environments compared to large firms ase 1985; Bamber 1987; Llorente et al 2002) Larger firms have higher analystfollowing (Christensen et al 2004; O’Brien and Bhushan 1990) and higher per-centages of institutional ownership (O’Brien and Bhushan 1990) Furthermore,there is more media attention showered on larger firms Hence, the marginal ef-fects of higher audit quality will be greater for small firms Casterella et al (2004)show that specialist auditors can charge a fee-premium to small clients, while largeclients enjoy a fee discount I posit that small clients pay this premium due to thelower cost of capital that will result from higher audit quality

(Ati-1.2.5 Summary

To summarize this section, CEO compensation consists of a cash component and

an equity component The cash component, although declining in importance, isstill significant and accounts for close to a third of the total CEO compensation

I highlight that the separation of ownership and control in a modern firm createsthe agency problem Then I discuss auditing as a means to mitigate this problem,and finally I analyze the proxies available for audit quality

The three roles of auditing specified by Wallace (1980) would all have a role inreducing the cost of capital of firms KR highlights the role of insurance provided

by the large auditors in reducing the risk premium and hence cost of capital toclient firms However, I argue that information and monitoring roles of the auditare as important I identify a proxy for audit quality that will be independent ofauditor size, namely auditor specialization, and posit that specialist auditors willresult in a lowering of cost of capital to the client firms

In the next section, I derive an analytical model that enables me to developthe hypothesis for empirical tests

2 Analytical model - Effect of Audit quality on

Incentive Compensation and Cost of Capital

I use the LEN framework to study the effect of audit quality on compensation.LEN stands for Linear compensation contract to the agent, Exponential utility forboth the agent and the principal, and Normally distributed outcomes

I model the scenario where an entrepreneur is looking to attract investors toinvest in his firm The entrepreneur announces the salary he will pay himself.This salary will be based on accounting returns The investors will assume thatthe entrepreneur will try to maximize his utility The return to the investor will

be the outcome to the firm minus the compensation that the entrepreneur payshimself Therefore, if the investor is to invest in the firm, the expected utility

of this residual return from the firm to the investor has to be greater than theinvestor’s reservation utility Both the investors and the entrepreneur are assumed

after the contracting period as:

where S is the (fixed) salary, and By1 is the (variable) bonus

The entrepreneur has a cost of effort, C(a, b) which is convex As the managersexpend more effort, the cost of effort increases in a convex fashion The cost ap-plies to both the bad effort and good effort However, I posit that audit qualitywill increase the cost of bad effort Given that auditors seek to detect earningsmanipulation, the difficulty or cost of achieving a given level of earnings manip-ulation will naturally increase with audit quality I define audit quality as q and

model the said effect as follows.

The investor invests A amount of capital in the firm A is a fixed quantity Theresidual left to the investors, or investors’ outcome (IO) will be,

Traditional principal agent theory posits that the managers will receive their vation wage, and the principal selects B so as to maximize his returns However,the ’management power’ view of compensation literature (Bebchuk and Fried 2004)mentions the close relationship between the managers and their compensation com-mittees to posit that firms maximize the managers’ utility rather than the investors’utility I model the situation where the entrepreneur (who is also the manager) ismaximizing his utility subject to meeting the reservation utility of the investors

the entrepreneur IR refers to the individual’s rationality constraints and IC refers

to the individual’s compatibility constraint of the entrepreneur The investors’ IRconstraint implies that the investors will not invest in the firm unless the utility ofreturns to the investors is greater than the utility of returns from an alternativeinvestment The entrepreneur’s IC constraint implies that the entrepreneur willdecide on the amount of good effort a and bad effort b so as to maximize his utility.Solving the ICentrepreneur gives the following expressions for a and b

The productive (good) effort expended by the entrepreneur increases with theincentive payments The unproductive (bad) effort also increases with incentivecompensation, but decreases with audit quality The intuition for this is as follows.Both the good effort and the bad effort equally contribute to y1 Hence, whenincentives are based on y1, both efforts are likely to increase

Given that I assume that capital markets are perfectly competitive, in librium, the investors will be able to obtain more than their reservation utilities.Equating their utilities to their reservation utilities and solving the IRInvestor con-straints results in

As audit quality (q) increases, the denominator of B decreases, and hence B creases Since B is the weight placed on the observable performance measure, thisgives us our first proposition.

The model also provides the following corollaries

Corollary 1

As audit quality improves, the good effort expended by the managers increases

As q increases, so does B From (7) as B increases, a increases

Corollary 2

As audit quality increases, the bad effort expended by the principal will decline

As q increases, the primary effect on b is an increase in the cost of earningsmanipulation, resulting in a decline in the bad effort There will be a secondaryeffect caused by the increase in incentives (B), which will cause b to increase andnegate the primary effect somewhat, but it will not be sufficient to dominate theprimary effect Hence, as q increases, b will decrease

Corollary 3

As audit quality improves, the firm’s desired outcome also increases

Since firm’s desired outcome is only a function of the managers’ good effort, asthe good effort increases with audit quality, so does the desired outcome

Proposition 2

As audit quality increases, expected return to the investor will decline.Hence, cost of capital to the firm will decline

The intuition is as follows From equation (9) the expected returns to theinvestor is in the form of a fixed component plus a risk premium, similar to theCAPM model10 As audit quality increases, the entrepreneur pays himself more

of his compensation in the form of incentives (i.e Since B increases, the variablecomponent of compensation increases) In other words, as q increases, the firm ispassing more of the risk to the managers Hence, as audit quality increases, therisk premium of the investors declines (since the entrepreneur is bearing a largerportion of it) As risk premium declines, overall expected returns decline, and cost

of capital to the firm also decline

10 The Capital Assets Pricing Model shows cost of capital as the risk free rate plus a risk premium (Copeland et al (2005), pg 152)

3 Empirical Study 1 - Effect of Audit Quality

on Incentive Compensation

In this section, I test the predictions on compensation derived from the sitions in the analytical model as well as those derived from the literature survey

From the literature survey, I conclude that the predictions of the model will apply

to payment made on observable performance The literature identifies this payment

as the cash component of CEO performance Proposition 1 of Chapter 2 statesthat as audit quality increases, the weight placed on the observable performancemeasure in determining the compensation of managers will increase The literaturesurvey identified that specialist BigX auditors have a higher audit quality compared

to non-specialist BigX auditors I identify the manager as CEO of the firm andstate my first hypothesis as follows

H1: There will be more weight placed on an internal performance measure indetermining CEO cash compensation, if the auditor is a specialist auditor

The literature survey on non-CEO executive compensation shows that firmprofitability is a factor in non-CEO executive compensation as well Hence, Iidentify the managers from proposition 1 as non-CEO executives and state mysecond hypothesis

H2: There will be more weight placed on an internal performance measure indetermining non-CEO executive cash compensation, if the auditor is a specialistauditor

3.2 Data and Research Design

3.2.1 Data

I use the Execucomp database to obtain the CEO compensation data, and theCompustat database to obtain the data for control variables Execucomp data isavailable from 1992 I extract the data using the variable ’PCEO’ to obtain com-pensation data relevant to CEOs The Execucomp database contains informationfor 19,373 firm years of data for CEOs Once this data set is merged with theCompustat Database and firm years with missing data attributes are deleted, Iend up with a total of 12,449 firm years, representing 2,085 distinct firms for theperiod 1992-2004

Of the 12,449 firm years of data, 12,136 firm years are for BigX clients, whileonly 313 firm years, representing 69 distinct firms, are for non-BigX clients Thismeans that out of the total sample, only 2.51% of the data is for non-BigX clients.Out of the total of 11,142 firm years of BigX clients, 6,091 firm years are for clients

of non-specialist auditors, and 6,045 are for clients of specialist auditors

I also obtain compensation data for non-CEO executives from the ExecuCompdatabase I obtain a total of 68,158 firm-executive-years of data for non-CEOexecutives, for whom data is also available on Compustat These are all firms thatare audited by BigX auditors In that sample, I have 2,148 distinct firms and19,666 non-CEO executives

3.2.2 Research Design

Firm performance

In accordance with previous studies (Lambert and Larcker 1987; Murphy 1999;Core et al 2003) I use return on assets (ROA) as the measure of accounting per-formance Most studies also use stock returns as an indicator of firm performance

(Core et al 1999; Core et al 2003) Hence I use stock returns (RET) as anotherindicator of firm performance, and as such, a determinant of CEO compensation.However, I do not expect audit quality to impact RET, since the market shouldhave already priced in the impact of audit quality into the stock price Therefore

in determining CEO compensation, I do not expect the weight placed on stockreturns to change with audit quality

ClientSales - denotes Client sales revenue

i - denotes audit firm

j - denotes client firms

k - denotes industry category

Jik - denotes number of clients of the ith auditor in the kth industry

Ik - denotes number of firms in the kth industry

I use a specialization cutoff level of 20% in my analysis.11 This means thatevery auditor who has more than 20% market share in a specific industry in agiven year (in terms of client sales revenue) is classified as an industry specialistfor that particular industry, in that particular year Hence, I define a dummy

11 Craswell et al (1995) use a 10% cutoff and Krishnan (2003) uses a cut-off 15% for sation More recently, Casterella (2004) use a 20% cutoff I too, use a 20% cutoff since this is a more appropriate number due to consolidations in the audit industry