Effects of Audit Fees on Audit Quality
Disclaimer: This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
In this part of dissertation, there is a grief introduction about this dissertation, which includes the following context: the background and overview of related researches in this paper; the motivation for doing this study and a introduction of the structure of this paper.
Firth (1997) notices that the responsibility of auditor is reporting comparative correct financial information to shareholders in an independent position. However, what makes regulators concern more about is the client-auditor relationship, especially the economic relation, which formed in the long-term cooperation may have impact on auditors' independent position. For a long time, because of the dependence of auditors' fees on client firms, therefore, the controversy about whether this economic dependence might impair auditors' independence position and audit quality make many empirical researches in the area. The impairment of audit independence caused by the economic relationship will result in the audit opinion shopping.
Previous studies consider about the affects of audit fees on audit quality in two ways: high audit fees paid to auditors may be the reflection of the complexity of auditing process and also increase auditors' effort. However, on the other hand, large audit fees paid to auditors might easy to build the economic bonding between client and auditors, thus, auditors are easy to compromise to audit independence, as the fear of losing highly profitable fees. Though a lot studies have researched the relationship between audit fees and audit quality, most of the studies just have blue results for their studies (Hoitash, Markelevich and Barragato, 2007).
Basing on the theory that examining the fees paid by client firms will better analyze the relation between audit independence and audit quality, this paper uses the methodology consistent with the way used in previous studies, i.e. Kinney and Libby (2002), Choi, Kim and Zang (2006), Hoitash et al (2007), in which develop a methodology that is depended on the notion that audit fees related to expected fees have effect on audit independence and quality. Therefore, being same as previous studies here will use to audit fees model to measure the expected fees in order to compare with the actual fees paid to auditor. Using the methodology in Choi, Kim and Zang's (2006) study, the differences between actual fees and expected fees are divided into two parts: the positive one and negative one, which helps us to compare whether there is any different reflection when auditors are paid less or excess their expectation. As for the proxy for examining audit quality, the discretionary accruals model which drawn from previous literature are selected. As suggested by Choi, Kim and Zang (2006), the modified Jones' model for discretionary accruals is used in this paper. The variables in models are selected according to previous studies which consider the influence on audit fees from perspectives of client firms' sizes, auditing risk, complexity and the size of audit firms (Chan, Ezzamel and Gwilliam, 1993). Beside dividing abnormal fees into positive subsample and negative subsample, the combined sample is cut into other two subsample which are based to the sizes of audit firms, which aims to examine the different behaviors to audit fees between BIG4 auditors and NON-BIG4 auditors. Thus, the objective in this paper is to prove whether abnormal fees might result in lower audit quality and whether there is different behavior to abnormal fees between BIG4 and NON-BIG4 auditors.
Continuing the previous study which is finished by Chan, Ezzamel and Gwilliam (1993), this study uses the data collected from UK market in order to find out related evidence from this market and 787 observations cover the period from 2006-2008 are collected from UK quoted public companies will help to examine audit fees paid to auditors. In addition to examine the relation between abnormal audit fees and audit quality, this study also follows the investigation which is finished by Ashbaugh, LaFond and Mayhew (2003) which focuses the study on whether the payment of non-audit fees has influence on audit independence. To extent this study, the paper here will test the ratio of non-audit fees to total fees paid to auditors which aims to find whether there is a positive relation between increased ratio of non-audit fees and decreased audit quality. Being different with Ashbaugh et al's (2003) research which analyzes non-audit fees in full sample, this non-audit fees here will be analyzed in subsample with positive abnormal fees and subsample of negative fees respectively, thus, whether non-audit fees have different influences on audit quality when auditors are paid higher or lower fees than their expectation should be examined.
The motivation for doing this study is based on the aim that finding out whether the economic bonding between auditors and client firms is existed really, as this relationship is one of the most controversial topic in public and academic area, especially after the Enron even and the failure of Anderson, this topic motivates more and more interesting in this area. However, according to the results of previous studies, many studies failed to prove the existent of this relationship between auditors and client firms, and the results in other relative studies are ambiguous. Moreover, because of active economic environment and accounting events, most studies paid attention on American market, as for other markets, seldom of relative investigations can be found. The aim in this study is try to find out relative evidence from UK market.
The structure of this paper follows: the next part is a literature review in related area which introduced both early and latest studies; and then is the development of hypotheses the resource of relative theory and the contribution of hypotheses are included in this part. The design of methodology is presented in chapter 4 which has a description of important models in detailed. Furthermore, the chapter after it is the description of sample and the result of test, which is also the center of this study. The limitation and conclusion will be displayed in the last chapter which gives an overview and comment of this study.
There is a literature review of relative studies in this part. The literature review covers the studies in relative area which includes the result of studies and the development of methodology related to the investigation. The primary previous studies are displayed and concluded in this chapter.
1. The importance of auditor independence
Since the collapse of Enron in US and the scandal of Andersen, the world's capital market confidence was reduced (Beattie and Fearnley, 2002). A large of the attention is paid on accounting and auditing practices, especially on the auditor independence. As some scholars (Beattie and Fearnley, 2002) point about that the independence of auditor is the base of the public confidence to audit process and the guarantee of the quality of financial information. In addition to this, rather than a benefit to investors, auditing also reduces the cost of information exchange for two sides (Douch, 1980 and Simunic, 1982). From these words, we can easily know that a high quality of auditing information might ensure the reliability of market information, therefore, many experts in this area focus their empirical researched on factors for the reduction of auditor independence in order to find out the reasons for impairing auditor independence. The economic bonding between client and auditor account for an important position in this kind of research.
2. Review on the payment of high audit fees on opinion shopping
DeAngelo (1981) shows the evidence through investigating the "low balling" phenomenon in audit market that when audit firms have a significant economic benefit on client firms, audit quality and independence are easy to be reduced. The same as DeAngelo's (1981) research, Magee and Tseng (1990) also have similar conclusion. Authors extend the previous research (DeAngelo, 1981) to find out in which condition the economic bonding may lead to the impairment of audit independence. The study provides a result that when auditors' compensation is tied to the decision of audit opinion and then, their independence is more likely to compromise to economic bonding. Frankel, Johnson and Nelson (2002) take the research in this area to provide empirical evidence for the relation between audit fees and earnings management. Their study is based on data selected from 3074 proxy statement listed in SEC in the period from Feb 5, 2001 to June 15, 2001. The evidence of this research confirms the assumption that there is an association between audit fees and the possibility of the compromise of audit independence. Kinney and Libby (2002) continue the research from Frankel, Johnson and Nelson (2002). Through analyzing the empirical research which has been done in the previous study, authors draw up a conclusion that auditors are probable to loss their independence in reducing their willingness to resist with client-induced biased in reporting audit opinion, when there is a strong economic benefit between auditors and clients.
However, some empirical tests in the same area conclude the opposite results. Craswell, Stokes and Laughton (2002) do the research to indentify whether fee dependence have impairment on audit independence. They do the research from both aspects of national market level and local market level, but they find evidence neither from national market level and local market level can demonstrate that fee dependence has negative impact on audit quality. In other words, the economic bonding is not existed in the auditor-client relationship.
In the following paragraphs, literature review about the impact of abnormal audit fees and non-audit fees on audit opinion shopping are shown.
2 .1 The review of abnormal audit fees on opinion shopping
From the literature review above, the evidence for the opinion shopping through the behavior of paying high non-audit fees is still absent. And some auditors (Kanodia and Mukferji,1994) point out that switching incumbent auditors cause clients loss initial engagement and negotiate fees, beside this, switching auditors is less likely for client firms to achieve clean opinion (Chow and Rice, 1982) and event more conservative audit opinion (Krishnan, 1994). Therefore, comparing with paying high non-audit services fees and switching auditors, paying auditors with higher audit fees is a more efficient and less risk approach in opinion shopping (Fang and Hong, 2004).
However, although there are some proofs can provide the assumption that higher audit fees, on some extent, can exert impact on audit quality, to analyze the relation between abnormal audit fees and audit quality, there are many factors should be concerned about, as these factors (no just economic bonding) also result in high fees. Some practitioners extend their research in this area by analyzing the association abnormal between audit fees and the changes in audit opinions.
Hoitash, Markelevich and Barragato (2007) consider the high audit fees paid to auditors on two aspects: one is the positive side which suggests paying high fees can increase auditors' effort in their jobs, thus, the quality also be improved; the other is the negative side which suggests that high audit fees paid to auditors may enhance the economic dependence on clients, thus, auditors' independence is easy to compromise to economic benefit and audit quality also reduced. In the research, they examine the fees paid to auditors between the period over 2000-2003, using two metrics to assess audit quality: the accruals quality measure and the absolute value of performance-adjusted discretionary accruals. They predict abnormal audit fees by using pricing model from previous studied Simunic (1980), and consider about the factors may have impact on fess pricing, such as risk, complexity, and company size. The outcome of this investigation shows the conclusion that abnormal audit fees might result in economic bonding and thus impair audit quality.
Similarly, Choi, Kim, and Zang (2006) have the same category of abnormal audit fees: the positive abnormal fees and the negative abnormal fees, the sample in this research constituted by the data collected from 9820 listed large companies' audit fees observations during 2000-2003 period. Study also choose model from Simunic (1980) to estimate the audit fees. The result of the regression model shows that evidence that the possibility of auditors' independence compromise to economic bonding is depending on whether clients may pay higher audit fees than the normal fees. In another word, audit quality will be impaired by abnormal fees.
Fang and Hong (2004) also issue the opinion about the abnormal fees that high fees paid to auditors may be caused by the real improvement in companies; therefore, the positive increase in audit opinion may also cause the increase in audit fees. Authors examine the relation between abnormal audit fees and audit quality by comparing the audit opinion in present year with the prior year, and find out the association between the abnormal fees and the changes of opinions. The database in this research is comprised by Chinese listed companies during the period from 2000-2002. Through the testing in regression model, authors find that the result is consistent with the positive relation between abnormal audit fees and improvement in audit opinion, which provides evidence that client companies succeed in opinion shopping through overpaying to auditors.
2.2 The relation between the size of audit firms and audit quality
But one point worth researcher to notice is the higher payment of audit fees does not mean the tendency of opinion shopping necessarily. Some evidences from other area can proof that the purpose of opinion shopping is not the only reason of high audit fees.
Simunic (1980) issues his opinion in the research of audit pricing that the competition in audit market is function of fees pricing. The same as this theory, the research on the low balling behavior (DeAngelo, 1981; Kanodia and Mukferji, 1994) confirm the opinion that competitive market will produce influence on audit fees. In addition to this, the audit firm sizes, the quality of auditors also have effects on audit price. Francis (1984) analyzes the effect from audit firm size on audit fees. Author researches for the evidences by comparing the differences of audit fees of Big-8 (Big-4 now) firms and of non-Big-8 firms over the period from 1974-1978 in Australian market. The research provides the evidence that there is larger size audit firms will result in higher audit fees, at the same time, study demonstrates that higher audit fee is consistent with higher audit quality. The similar research is done recently by Choi, Kim, Liu and Simunic (2008) through using a large sample from 15 countries and a cross-country regression. Beside this, study links the research to audit characteristic and legal environment. After analyzing the difference between the fees in Big-4 firms and non-Big-4 firms, authors achieve the conclusion that the relation between firm size and the level of audit fees is existed. According to the interview finished among auditors (Chan, Ezzamel and Gwilliam, 1993), the BIG 6 and NON-BIG6 (BIG4 now) auditors confirm that the BIG6 premium might exist if comparing with very small audit firms, but no medium size firms. The interview here reflects a phenomenon that the difference in audit fees between BIG4 and NON-BIG4 audit firms are not so obvious.
2.3 The payment of non-audit services fees on opinion shopping
The provision of non-audit services by audit firms to client firms is another controversial topic in opinion shopping. Policy makers argue that the provision of audit and non-audit services to same clients is possible to reduce the level of auditors' independence (Houghton and Ikin, 2001). At the same time, auditors protect themselves by arguing that supplying non-audit services does not impair their independence, because these works are often be done by different partners and staff. For a long time, researchers have never stopped investigating in the area.
3. The influence of non-audit fees on audit independence
As to the audit independence, there are many definitions for it. DeAngelo (1981a, p.186) defines audit independence as 'the conditional probability of reporting a discovered breach'; Knapp (1985) provides the definition as 'the ability to resist client pressure'; the definition from AICPA (1992) is 'an attitude/state of mind'; according to Magill and Previts' (1991) definition, audit independence is 'a function of character, with the attributes of integrity and trustworthiness being key' and ISB (2000) defines audit independence as 'the freedom from those pressure and other factors that compromise, or can reasonably be expected to compromise, an auditor's ability to make unbiased decisions'. Although there are somewhat differences between those definitions of audit independence, they have a common point that is the importance of objectivity and integrity (Beattie and Fearnley, 2002).
Beside provide the audit services to clients, audit firms also provide other services to clients, these services are called non-audit services, such as management advisory and consulting, but the compliance related services, such as taxation and accounting advice, are also included in them. However, as many experts (Beattie, Brandit and Fearnley, 1996) point out that this kind of non-audit services related closely to the annual reporting round. Therefore, the use of 'consultancy' for non-audit services is somewhat wrong.
What is discussing most in academy about the provision of non-audit services is the potential conflict of interest faced by audit firms who receive large non-audit fees from their audit clients. For example, after the Enron case, it was disclosed that Andersen received $25m in audit services fees and $27m for non-audit fees (Beattie and Fearnley, 2002). So, such high non-audit services fees paid to audit make public and scholars to suspect that the provision of non-audit services increases the economic bonding between auditors and clients. Many investigations are based on the assumption that the provision of non-audit fees reduces auditors' independence, as they fear for losing high profit engagement with clients in the future; therefore, they are probably to give up independence for high profit.
3.1 The review of the researches in association between non-audit fees and audit quality
In the early 1980s, researchers had found that the percentage the revenues from providing other services in audit firms had increased lot (Barkess and Simnett, 1994). The early empirical study is done by Simunic (1984). Author establishes his study on the assumption that those client companies who purchase non-audit services have a higher audit fees paid to auditors than audit fees from those companies who do not purchase non-audit services, and both of them hire the same incumbent auditors. In this investigation, author just focus the test in Big-8 (Big-4 now) firms, which avoids the differences on audit quality, and selects a database compromise of 397 US listed companies. From the research, he finds out than there is a positive relation between non-audit services and audit fees.
Simon (1985) continues previous research (Simunic, 1984) in this area by using more recent data in the period from 1978-1983. His research depends on the voluntary disclosure from proxy statements (the non-audit services fee is not disclosed that time). His research also produces the result that client firms who purchase non-audit services have higher audit fees than those companies without the engagement of non-audit services with audit firms.
The evidence from UK market (Ezzamel, Gwilliam and Holland, 1996) constitutes the research on the data from 314 UK listed companies. From the study, authors found out similar result consistent with previous researches (Simunic, 1984; Simon, 1985). To be different from above studies, this study joints non-audit services with other factors together to investigate the effects on audit pricing. Many previous researches have confirmed the phenomenon that the positive relationship between the purchase of non-audit services and higher audit fees paid by clients exists. But the existence of this relationship causes regulators, practitioners and public's attention about whether this economic bonding between client firms and auditors might impair audit independence.
Barkess and Simnett (1994) concern about independence on two aspects, one perspective is to examine whether those clients purchase other services from auditors are less likely to receive qualified opinion; the other one is the determination of the relationship between the provision of other services and audit service by comparing those companies who provide non-auditor services but do not change auditors and those companies change auditors. In the study, the sample is compromised by the Top 500 listed companies in Australia for each of the years from 1986 to 1990. From the study, they conclude the result that 85%of the companies in the research purchased non-audit services from the incumbent auditors and the increase in the percentage in stable in the period. In addition to this, their result supports the point that there is a positive relationship between audit fees and the provision of non-audit services. As far as the audit independence concerned, their research show that there is not identified relationship between the supply of non-audit services and the type of audit report, as they found that there were 308 qualified opinions in the total sample of 2094 audit reports. Therefore, they concluded that there was not enough evidence can demonstrate the hypotheses that auditors were less likely to issue the qualified opinion when the level of non-audit services to clients was higher. In other words, the provision of other services does not impair audit independence.
However, Wines (1994) concluded the opposite result by testing 100 public companies on the Australian Stock Exchange at 30th June 1980, in the sample, 24 of these companies were failed to do the test, therefore, and author examined the 76 left for over the period of ten years. From the test in these 76 listed companies, author drawn up the findings that 7 of 76 companies paid a higher level of remuneration for non-audit services to auditor than the provision of audit services and those companies with non-qualified opinion had higher payment of non-auditor services fees than those companies with qualified opinion (28 companies. Hence, Wines gives the summary that the evidences from the research in 76 listed companies associate with the assumption that the provision of non-audit services has impairment on auditors' independence. Wines (1994) also points out the limitation in his research is difficulty in assessing audit quality by considering the frequency with which auditing firms issue qualified opinions. However, one problem in Barkess and Simnett (1994) and Wine (1994) studies is that their sample is not large enough in test.
Craswell (1999) makes advantage in his study by using a larger data sample and makes the evidence related to auditors' actual decisions. The fiscal-year data chose by author to examine are obtained from 'who audit Australia', the sample is consisted by 885 public listed companies in 1984, 1477 in 1987 and 1079 in 1994, the results show that in each of the year (1984, 1987, 1994) the result does not associate with the assumption that non-audit services have negative effect on auditor independence, which is consistent with Barkess and Simnett (1994). But one limitation in his research is that this study just compares the companies with qualified opinions with those companies with unqualified opinions, while a better test should compare the companies with qualified opinions with such companies with clean opinions but experienced problems and likely to raise qualifications.
Although most of the investigation have done by researchers show that the provision of non-audit services does not impair audit independence, regulators still hold the assumption that auditors will prefer to give up their independence in order to obtain more non-audit services fees from clients (DeFond, Raghunandan and Subramanyam, 2002), especially after the Enron accounting scandal. In 2002, the Sarbanes-Oxley Act (the 'Act') imposed the prohibition on the provision of non-auditor services, which based on the direction of enhancing auditor independence, reducing 'conflict of interest' and the concern that 'all non-audit services were not created equal' (American Institute of CPAs, 2002). At the same time, the US Securities and Exchange Commission (SEC, 2002) made the revision the Commission's regulations related to the non-audit services, which are consistent with the content in Sarbanes-Oxley Act.
After the prohibition on non-auditor services added in Sarbanes-Oxley Act and SEC was carried out, many scholars continue the prior researches in this area. Frankel, Johnoson and Nelson (2002) used a sample constituted of 3074 proxy statements, used two indicators (discretionary accruals and the likelihood of firms meeting earning benchmark), to test whether audit independence would be reduced when the non-audit services grew. The consequence of the research indicates the evidence that there is an association for the assumption that auditors likely tend to sacrifice their independence when the non-audit services fee is high. Ashbaugh, LaFond and Mayhew (2003) continued this research, but the conclusion challenges the results made by Frankel, Johnoson and Nelson (2002). In the test, they used the same indicators as prior one. In the test, they find that there is no relation between positive discretionary accruals and auditor fee metrics; furthermore, their test proof that the relation between fee ratio and the likelihood that firms beat analysts' forecasts is not existed, in other words, auditors' independence will not compromise to clients' high non-audit services fees.
4. Review of the development of models
In empirical studies, the relative models will help to connect independent variables together in order to analyze the relation between each variable. In the study of the relation between audit fees and audit quality, the most important models are the model for measuring audit fees and audit quality. In this part, a review of the development and modification of two models is displayed.
4.1 Review of the development of audit fees model
Since Simunic (1984) develops the audit fees model in order to predict the expected audit fees, the model has been developed lot in these years. In the beginning, Simunic (1984) provides the theory that the audit fees level will be affected by the several factors, such as the client firms' sizes, the complexity of auditing process, audit firms' sizes and audit risk. The following scholars provide the related variables for model which the used to decide expected audit fee.
Chan, Ezzamel and Gwilliam (1993) point out in their study audit size is an explanatory variable which has important influence on the determinant audit fees. In the study, authors suggest to use the measurement of turnover to control audit size, which is also the measurement of client firms' sizes. However, the use of turnover as determining is not unproblematic because the definitions of turnover are varying widely between companies and industries. Therefore, to measure audit size and client firms sizes, many researchers choose the total assets as variable. To consistent with previous studies, Ashbaugh, LaFond and Mayhew (2003); Choi, Kim and Zang (2006); and Hoitash, Markelevich and Barragato (2007) use the nature log of total assets to control client firms' sizes and audit size. In addition to using total assets and turnover as proxies to audit size, on most investigations, likes Ashbaugh et al (2003) choose the number of employees to quantize audit size. Both of these scholars use number of business segments and geographic areas in measuring audit sizes.
As for the measurement of complexity of audit process which is another reason for increasing audit fees, Chan et al (1993) suggest to use the number of subsidiaries to measure it. According to previous studies (Ashbaugh et al ,2003; Hoitash et al ,2007), the proportion of foreign subsidiaries will have influence on the increase of audit fees, therefore, the ratio between foreign subsidiaries and total number of subsidiaries is selected to measure this effect on audit fees. Furthermore, Choi, Kim, Liu and Simunic (2008) use the ratio of the sum of inventories and receivables to total assets to present complexity. Lastly, the dummy of gain or loss before extraordinary items is a popular variable in latest studies, such as the study in Hoitash et al (2007).
The level of risk in audit processing is another factor causes increase in audit fees. Turley and Cooper (1991) provide the hypothesis that there is a positive relation between audit risk and audit fees. Chan et al (1993) predict higher risk makes consequence in higher audit fees is because auditors want to take the excess fees than normal level as an 'insurance' premium and this hypothesis is supported by interview findings. Most of previous studies focus on using such variables as liquidity ratio (the ratio between current liabilities and current assets), and gearing ratio (the leverage), to test the level of audit quality. However, as previous scholars point out that the audit risk which is the reflection of the nature of the business of the enterprise and the control of enterprise is difficult to measure. Therefore, the subjective judgments in measuring audit risk are hard to avoid.
The level of client firms' performances is will also has impact on audit fees. According to the interviews between audit partners (Chan, Ezzamel and Gwilliam, 1993), it is confirmed that there is a link between the level of client firms' profitability and the level of audit fees, and the association between them is negative. Furthermore, it is commonly agreed that when a client is facing with financial pressure is more likely to ask for controlling overhead costs wich might be result in higher audit fees. To measure the level of this variable, Chan et al (2003) use the return on equity to measure it, while other researchers like Hoitash et al (2007) and Chan, Kim, Liu and Simunic (2008) use the return on assets to measure this.
Beside variables above, there are other potential reasons which might also cause the changes in audit fees. However, seldom of previous studies pay attention on these potential variables (Chan, Ezammel and Gwilliam, 1993) according to the study finished by Chan et al (1993), the control of ownership is also one of the variables in audit fees model. The hypothesis in their paper makes the assumption that the extension of audit services will be a factor in the ownership control as companies with a diverse ownership structure are required a higher quality audit. Therefore, the audit fees are increased. However, it is difficult to measure the extension of ownership control directly. The timing variable is another reason in the fluctuation of audit fees (Chan et al, 1993). According to the audit season in UK market, the accounting year between 1 December and 31 March is the busy season, others are non-busy, and the former season will increases audit fees as auditors have a comparatively shorter deadline in it. Moreover, though test, authors find that the legal liability also have impact on the level of audit fees, as the evidences demonstrate that legal liability is a fee-increase factor. Lastly, other variables, such as the location of auditors, are seldom be used as variables in investigations.
4.2 Review of the development of discretionary accruals model
The evaluation of audit fees is a difficult topic in this area. Craswell,Stokes and Laughton (2002) model the expected audit opinions to compare with the actual opinions received by client firms in order to investigate whether audit fees have impairment on opinions. Fang and Hong (2004) the changes in audit opinions and use regression model to analyze the correlation between opinion changes and other independent variables. Most of other investigations tend to use accruals to measure earning management. Using discretionary accruals as a proxy to audit quality is based on the theory that examining managers' use of discretionary accruals to change income is fiscal year in order to evaluate whether there is any misstatement existed (Bartov, Gul and Tsui, 2001). In the history of the development of the discretionary accruals model, there are six popular models, and these model will be displayed in following words. The early model is developed by Healy (1985), in the model, the mean of total accruals (TAt) which are divided by total assets (At-1) in last year from the estimated period are use to measure nondiscretionary accruals. NDAt=1/nSTAt/At-1 The model provided by DeAngelo (1986) uses the last year's total accrual (TAt-1) and scaled by lagged total assets (At-2) to calculate the nondiscretionary accruals. And the model is looked as: NDAt=TAt-1/At-2 The discretionary accruals is the difference between total accruals and nondiscretionary accruals. The Jones model (1991) is the most popular and closed model used in investigations in these years. In his study, he separate discretionary accruals from nondiscretionary accruals. Being different from two previous models which just calculate nondiscretionary accruals by using total accruals divided by lagged total assets, in his model, he adds other independent variables, such as PPE (property, plant and equipment) and the changes in receivables to measure nondiscretionary accruals, the model is looked as: NDAt= ß+ß1(1/At-1)+ß2( RCEt/At-1)+ß3(PPEt/At-1) The discretionary accruals is the difference between total accruals and non discretionary accruals. To separate discretionary accruals from nondiscretionary accruals, Jones (1991) adds the residual in model, which represents the discretionary proportion of total accruals: DAt= ß+ß1(1/At-1)+ß2( RCEt/At-1)+ß3(PPEt/At-1)+e The modified Jones model is provided after it. What being different with Jones model is the addition of changed revenue to exercise discretion. And the modified model is looked as: DAt= ß+ß1(1/At-1)+ß2[( REVt- RCEt)/At-1]+ß3(PPEt/At-1)+e The most difference between former two models and later models is that Jones model and modified Jones model add coefficients which will help to investigate the correlation between independent variables and dependent variables. Dechow and Dichev (2002) suggest using firms' performances to measure earning management. The model is their study focuses on working capital accruals as cash flow realizations related to it happen in one year, which makes the theory and the empirics more tractable. Therefore, instead of using the changes in accounting receivables, changes in revenues and the total of property, plant and equipment to measure to expected accruals, scholars here use the total of cash flow in constant years in measurement. However, the problem in their study is that the model is fail to measure current accruals because of the long lags between noncurrent accruals and ultimate cash flow realizations. McNichols (2002) extent the D&D model by combing the fundamental variables included in the modified Jones model (i.e. PPE and the changed accounting receivables) which helps creates a better specific model and a better set of residuals. And the modified Jones model is used by Hoitash et al (2007) in study to investigate the relation between audit fees and audit quality. The more recently study done by Francis, LaFond, Olsson and Schipper (2005) provide a critical review of modified Jones model, in their paper, they find that there is an association between poorer accruals quality and larger costs of debt and equity; therefore, they suggest using the return on assets in order to control firms' performances. Kothari, Leone and Wasley (2005) focus their study on examining performance-matched discretionary accruals. Comparing with traditional modified Jones discretionary accruals model which uses the PPE (property, plant and equipment) to control discretionary accruals, Kothari et al (2005) replace it by using return on assets, which is consistent with the approach in Francis 's et al (2005) study.
5. Conclusion for this chapter
The sentences above describe the related studies in examining the relation between audit fees and audit independence (audit quality). From the literature about this area, we can realize that the relation between auditors and client firms is still a controversial topic in either public or academic area because of the potential economic bonding between them. Some previous researchers, such as Hoitansh, Marakelevich and Barragato (2007) and Fang and Hong (2004) find evidence to prove the impairment of abnormal audit fees to audit independence in American and Chinese market respectively; Choi, Kim and Zang (2006) detail the research by dividing the combined sample into two subsamples: the subsample with positive abnormal audit fees and subsample with negative abnormal audit fees, and prove the assumption that positive abnormal audit fees have impairment on audit independence while this possibility in negative abnormal fees is less. As for the association between non-audit fees and audit quality, the answers are unclear because most of investigations fail to prove the existence of positive relation between non-audit fees and decrease of audit quality. As for the calculation of expected audit fees and quantification of audit quality, majority of studies use the audit fees model and modified Jones model to calculate expected audit fees and discretionary accruals (audit quality) respectively.
The development of hypotheses
In this chapter, the hypotheses in this paper are displayed. Besides just describing hypotheses, the development of hypotheses and the origination of theory are also displayed in following paragraphs.
1. The difference between normal audit fees and abnormal audit fees
Actually, the total audit fees is constituted by two parts, one is the normal audit fee which is the reflection of audit effort the risk; the other one is the abnormal audit fee which might reflects the special relationship between auditor and client, and it is also the point in studies related to this area. How to price audit service, Simunic(1980) includes several factors which affect the normal audit service price. According to his theory, such factors as client size, client complexity, and client-specific risk should be included in analyzing the normal audit fee. Furthermore, Simunic (1984) develops the formula which models the normal audit fee expected by auditors. Therefore, many previous researches rely on examining the difference between actual audit fees and the fees modeled by formula. These studies are based on the opinion that the audit fees in excess of normal fees are possibly to create economic bonding between auditors and client (Choi, Kim and Zang, 2006).
Most prior researches just use the actual audit fees paid to auditors. However, Simunic (1980) has pointed out that the fee paid to auditors is the reflection of auditor effort and litigation risk. Therefore, the observation of differences in the level of actual fees between client and auditor is more likely to reflect the levels of audit effort and risk rather than the economic bonding. To overcome this limitation, Choi, Kim and Zang (2006) develop a new approach that the study should be established on the aspects of audit fees. One is the positive audit fee which is more than expected fee; the other one is the negative audit fee which is less than expected audit fee. Be different from previous studies, authors divide sample into two subsamples, thus, the cancellation between the two subsamples could be avoided effectively if the significant fee-quality relation just exists in one of the subsamples.
2. The relation between audit fee and audit quality
From the previous empirical studies we can easily know that most of those studies are built on the assumption that abnormal high audit fee might impair audit quality. For example, if auditor receives audit fee more than the expected fee and thus the economic bonding is easy to be created between auditor and client. That is because auditor may try to retain client from allowing substandard financial reports (Choi, Kim and Zang, 2006).
As far as concern about the abnormal audit fee, the provision of nonaudit fee should be included in test. Simunic (1984) considers the provision of nonaudit services as the reason for abnormal fee, therefore, for the purpose of keeping high profitability from clients auditors might choose to compromise to independence. The later empirical studies extent the researches in this area to analyze the association between audit fee and audit quality. Frankel, Johnson and Nelson (2002) test the relation between the provision of nonaudit services and earning management; Choi, Kim and Zang (2006) divide the total sample into two subsamples: the positive and negative abnormal fee respectively and test the significance between positive (negative) audit fee and discretionary accruals; Fang and Hong (2004) test the relation between abnormal audit fee and audit quality by making the assumption that clients with abnormal audit fees are more likely to receive better audit opinions than previous years.
To extent the previous prediction of the relation between abnormal audit fee and audit quality, the total sample in this research will be divided into two subsamples as the approach is used in Choi, Kim and Zang's study (2006). In this research, the following hypothesis will be tested:
H1: the clients with positive abnormal audit fees are more likely to receive better audit opinions than they should be, other things being equal.
On the other hand, if the clients provide auditors with less audit fees than auditors' expectation, then, auditors are less likely to compromise to independence. To test this prediction, the second hypothesis is made:
H2: the clients with negative abnormal audit fees are less likely to have better audit opinions than they should be, other things being equal.
Although dividing the total sample into two subsamples can avoid the cancellation between two subsamples, there is one possibility exist, that is the auditor who bear negative abnormal fee now in expectation of higher payment in the future might compromise to independence, consequently, the audit quality is also reduced (Choi, Kim and Zang,2006).
3. The association between audit firms' sizes and audit quality
As far as the association between audit firms' sizes and audit quality is concerned about, the differences between Big-4 audit firms and other non-Big 4 audit firms are worth to be discussed. First of all, according to Simunic's (1980) model for pricing audit fees, the employment of Big-4 audit firms is one factors for the increase in audit fees, therefore, the Big-4 audit firms' auditors are paid higher than other firms' auditors. In addition to this, Sori and Karbhari (2006) find the evidence that Big-4 auditors are better able to resist management pressure in competitive market because of the better technologies, resources they held. Li, Song and Wong (2005) examine the relationship between audit firms' sizes and audit quality in Chinese market and find that the audit firms with larger size are associated with higher quality. Furthermore, comparing the reputation loss and profitability loss, auditors are preferred to keep their reputation than sacrificing it to create higher profit (Defond, Raghunanadan and Subramanyam, 2002). Therefore, from the previous researches above, there is a speculation could be made that Big-4 audit firms' auditors are less likely to loss independence when they are paid abnormal audit fees. To demonstrate this prediction, the following hypothesis will be examined in this research:
H3: auditors from Big-4 audit firms are less likely to compromise to clients with positive abnormal fees.
On the other words, there is another prediction could be produced than because of the lack of competiveness and lower level of payment than Big-4 auditors, those auditors from non-Big-4 audit firms are possible to loss independence to abnormal fees. To find out the evidence to this prediction, the other hypothesis is made below:
H4: auditors from non-Big-4 audit firms are more likely to compromise to clients with positive abnormal fees.
The hypotheses here tend to analyze whether there is any difference of reflection between BIG4 and NON-BIG4 auditors. Being different from previous studies which put both sizes of audit firms together, the test in this paper focus on examining two groups of auditors separately, and thus, the effect between two can be avoided effectively.
The design of research
In this part of dissertation, there is an introduction of the design of methodology, which includes the measurement of abnormal audit fees and the assessment of audit quality. To follow the design of Choi, Kim and Zang's (2006) design of research, the data will be departed into two parts: the normal (expected) audit fees and abnormal (unexpected) audit fees.
1. The development of the measurement model of audit fees
Since Simunic (1980) began the investigation in the pricing audit fees, it is widely considered that the increase in audit fees is linked with several factors, such as the audit firm size, client firm assets, client complexity and client risk. Therefore, it is predicted that those client firms with larger sizes, more complexity and more risk have to pay higher audit fees. For connecting these factors together, the model is drawn from previous researches (Hoitash, Markelevich and Barragato, 2007; Choi, Kim and Zang, 2002). To consistent with previous researches (Hoitash, Markelevich and Barragato, 2007), the total assets of client firms are used to measure client firms' sizes; the number of subsidiaries, the number of business segments and the foreign operations are employed to assess the complexity of client firms; as for the risk of client firms, the liquidity, the return on assets ratios and company net loss are proxies for it.
To calculate the expected audit fees, the model is selected from previous studies (Hoitash, Markelevich and Barragato, 2007; Choi, Kim and Zang, 2002) and combine some variables from Chan, Ezzamel and Gwilliam's (1993) research:
LTFEE=a+ß1LNTA+ß2INVREC+ß3EMPLOY+ß4SUB+ß5EXORD+ß6LOSS +ß7LEVE+ß8ROA+ß9LIQUID+ß10ROE+ß11BIG4+e (1) In which, LTFEE=the nature log of audit fees paid to auditors
LNTA= the nature log of total assets
INVREC= the inventory and receivables divided by total assets;
EMPLOY= the square root of the number of employees;
SUB= the square root of subsidiaries;
EXORD= the dummy which is 1 if the firm reports any extraordinary
gains or losses, 0 otherwise;
LOSS= the dummy which is 1 if the firm reported a loss during the year, 0 otherwise;
LEVE= the leverage (the total liabilities divided by total assets);
ROA= the return on assets (income before extraordinary items divided by average total assets);
LIQUID= the current assets divided by current liabilities;
ROE=the return on equity;
BIG4= the dummy which is 1 if the auditors is one of Big-4, 0
According to the variables above, the total assets (LNTA) and number of employees (EMPLOY) are used to evaluate the size of client firms; the inventories (INVERC), subsidiaries (SUB) and extraordinary items are proxies for audit complexity; the losses (LOSS) in fiscal year, leverage (LEVE), the ratio of current assets to current liabilities (LIQUD), return on equity (ROE) and returns on assets (ROA) are employed to measure audit profitability.
2. The assessment of audit quality
One problem in testing the relation between audit fees and audit independence is the measurement of audit quality. Generally, there are two ways could be found from previous studies in assessing audit quality. One can be found in Fang and Hong's (2004) study which compares the changes of audit opinions in three constant years. This approach relies on the opinion that those companies pay higher audit fees are more likely to receive better audit opinions than the last year. However, this kind of research design has endogenous problem, as the higher payment of audit fees is related to the real increase in client firm's quality. To solve this problem, many scholars (Dechow and Dichev, 2002; McNichols, 2002; Hoitash, Markelevich and Barragato, 2007; Choi, Kim and Zang, 2006) choose the discretionary of accruals as the proxy for the assessment of audit quality. This approach is based on the assumption that discretionary accruals can reflect the biasness included in financial statements by management and allowed by auditing. Therefore, it is expected that if there is a positive relation between abnormal audit fees and discretionary accruals then the audit independence is impaired by abnormal audit fees, in other words, audit independence compromises to economic benefits.
As for the model of discretionary accruals, Jones' model (1991) separates discretionary accruals from nondiscretionary accruals. Although this model is popular in many previous studies, the considerable imprecision included in this model in assessing discretionary accruals brings criticism to it (Hoitash, Markelevich and Barragato, 2007). Especially, the Jones model is based on the assumption that accruals just sensitive to the current changes in sales, as to the lagged and future changes, it is not relevant to them. Ball and Shivakumar (2006) modify Jones model for controlling the asymmetric timeliness of realizing economic gain and loss. Be different from Jones (1991) model, Dechow and Dichev (2002) assess the accruals as a whole. However, Dechow and Dichev's (2002) model has limitation in calculating current accruals, as there are long lags between noncurrent accruals and ultimate cash flow (Hoitash, Markelevich and Barragato, 2007). Frankel, Johnson and Nelson (2002), Ashbaugh, LaFond and Mayhew (2003) adjust Jones model for firm performance. For controlling firm performance, Choi, Kim and Zang (2006) use PPE to measure it, as Hoitash, Markelevich and Barragato (2007) choose ROA in assessing firm performance. In this dissertation, the model is similar to the one in Choi, Kim and Zang (2006):
ACCR/TAt-1=ß1 (1/TAt-1) +ß2 [( REV- REC)/TAt-1] +ß3 (PPE/TAt-1) + ß4 (CFO/TAt-1) + ß5 DCFOt + ß6 [(CFO/TAt-1)*DCFOt] +et (2)
Where, the ACCR represents the total accruals; the TAt-1, REV and PPE denote the total assets in t-1 year, changes in revenues and gross property, plant and equipment respectively; CFO is the cash flow from operations; DCFO is dummy which equals 1 if CFO is negative and equals 0 otherwise; e is error and the REC is the change in net receivables. To consistent with Choi, Kim and Zang's (2006) research, equation (2) is used to obtain discretionary accruals. As for the estimation of total accruals, most of previous researches (Bartov, Gul and Tsui, 2001; Hoitash, Markelevich and Barragato, 2007) use the balance intcluded in Jones' (1991) study. The whole balance is displayed as follow:
TAt= CAt+ CASHt+ CLt+ DCLt+DEPt (3)
Where, the TAt is the total accruals in t year; CAt represents the change of current assets, which is the difference between the current assets in t year and it in t-1 year; CASHt is the change of cash from t-1 year to t year; CLt is the change of current liability; DCLt is the change of debt included in current liability and DEPt is the sum of deprication and amortization expenses
As to the measure of discretionary accruals, these values should be measured in two parts: the DA1 (discretionary accruals 1). The DA is computed in following ways: according to the recent research from Choi, Kim and Zang (2006), they point out that accounting accruals are a piecewise linear function of current period cash follows from operations, therefore, to control thie asymmetry between economic gain and loss into accrual model, they add three additional variables, which are CFO/TAt-1, DCFOt and CFO/TAt-1)*DCFOt, instead of using 1/TAt-1 ( REV- REC)/TAt-1, PPE/TAt-1 in traditional Jones; model (1991). Then, the total accrual used in equation (2) is measured by equation (3). The discretionary accrual is the difference between the total accruals which is measured by equation (3) and the value fitted in equation (2).
3. The model to test the association between abnormal audit fees and audit quality
To test the hypotheses in this article, the regression model to estimate the relation between abnormal fees and audit quality (discretionary accruals). The regression model is selected from the model used in Choi, Kim and Zang'(2006) study. In this article, it is added with the ratio of non-audit fee to total fees paid to auditors to test the influence of non-audit fees on audit quality (namely NONAUDITFEE). The whole model is seen as following:
DA=a+ß1ABFEE+ß2NONAUDITFEE+ß3LNTA+ß4BIG4+ß5LOSS+ß6LEVE +ß7CFO +e (4)
Where, the ABFEE represents the abnormal audit fees (the positive audit fees or negative audit fees) which is used to test the relation between abnormal audit fees and discretionary accruals (namely DA in the model). The previous researches related to this area show that there is a positive association between a stable operation and firm scale (e.g., Dechow and Dichev, 2002), which means that the larger firm is less likely to be reported a lower level of discretionary accrual; therefore, here use the LNTA (the nature log of total to test the scale of firms). Beside this, prior studies predict that Big-4 auditors are not so easy to compromise to abnormal fees; therefore, the dummy of BIG4 is chosen to test whether there is any difference between Big-4 auditors and NON-Big-4 auditors when they face with abnormal audit fees. In addition to this, the ratio of non-audit fees to total fees paid to auditors is meaningful in testing whether the non-audit fee has influence in audit opinions shopping, many previous studies (e.g., Ashbaugh, LaFond and Mayhew, 2003) deliver the point that the higher ratio of non-audit fee to total fees paid to auditors may result in lower audit quality, so, here use the NONAUDITFEE to test the effect of non-audit fees on audit quality. Furthermore, the loss dummy (LOSS) is included to test for the potential differences in earning management behavious between loss and profit firms. The firms with high leverage may more possible to boost reported earnings as they consider about the debt or private lending agreement violations (Becker et al, 1998), so, the LEVE is selected to control for the effect of this leverage. Last but not least, some previous researches provide evidence that discretionary accruals are positively associated with firm performance which means firm with low (high) cash flow is probable to have negative ( positive) discretionary accruals (Choi, Kim and Zang, 2006). To test this potential association between firm performance and audit quality, the CFO in equation (4) is used.
To test equation (4) in full sample, the data from 143 selected firms are used. Next, to make evidence for Hypothesis 1 and 2, the full sample will be divided into two subsamples: the sample with positive abnormal fees and the sample with negative fees, and then with sample will be tested in the regression model equation (4) to analyze the correlation between abnormal audit fees and audit quality. The prediction for the hypothesis 1 is that if there is a positive relation between positive relation between positive abnormal audit fees and discretionary accruals, then, the result is consistent with H1; if there is a negative relation between negative audit fees and discretionary accruals, then, the result is consistent with H2.
To test the assumption in Hypothesis 3 and 4, the full sample is separated by the kinds of audit firms: the Big-4 audit firms and NON-Big-4 audit firms, and the processing in testing is same as it is in the first testing (the test for full sample, Hypothesis 1 and 2), but the variable BIG-4 which is used in equation (1) and equation (4) is deleted in this step as for these two subsamples this variable could be seen as a constant, so it is meaningless in this test. The predictions for H3 and H4 is that if there is a negative relation between abnormal audit fees and discretionary accruals in the sample of Big-4 auditors, the assumption in H3 could be demonstrated; the positive relation between abnormal audit fees and discretionary accruals in the sample of NON-Big-4 auditors in consistent with H4.
4. The data collection and sample description
All the financial data and audit details about companies are collected from database: FAME and Orisis. At the first, 516 companies over the period from 2006-2008 (except for banks and insurance companies) are selected to construct the full sample, and then the firms with unavailable data are cancelled from the full sample in each year. At last, the companies with available data in each year are 267, 271 and 256, and all the companies are public quoted companies listed in London Stock Exchange Market. To convenient with the expression of data, all the data are expressed with GBP in million.
In this chapter, the results of the test and the evidence of hypothesis are displayed, in which include the prediction of results, description of data analysis and the evidence for hypotheses.
1. The data description
The Table 1 describes the descriptive statistics for the full sample in this study. What are displayed in table 1 are the non-deflated audit fees, non-audit fees and total assets over the period from 2006-2008. From the table, it is reflected that the averaged of total assets increased for around 42.9% in the full sample over the period; the audit fees (i.e. the fees for auditing financial statements) increases from the average of 967 in 2006 to the average of 1363.59 in 2008, which increased for around 41% in these three years; as for the averaged non-audit fees, it also increased gradually in these three years, which increased from 863.49 in 2006 to 889.68 in 2008. According to these significant increase happened in the fees paid to auditors and total assets, the increase in non-audit fees and audit fees mean the potential increased economic bonding from auditors to clients, as many empirical studies base the researched on the assumption that the higher level of non-audit fees (i.e. Barkess and Simnett, 1994; Ashbaugh, LaFond and Mayhew, 2003) and audit fees (i.e. Choi, Kim and Zang', 2006; Hoitash, Markelevich and Barragato, 2007 ) paid to auditors might emphasis the possibility of the loss of auditors' independence. At the same time, the increase in averaged total assets in these three years makes another possibility that those increased scale client firms might less likely to commit audit opinions shopping, as Dechow and Dichev ( 2002) point out that there is a negative association between client firms' scale ( which could be presented by the total assets) and the violation of audit quality, which gives the idea that the larger scale of client firms are less likely to accompanied with lower audit quality. However, to investigate the relationship between audit fees and audit quality and other relative variables, the data analysis is necessary. The detailed data analysis will be displayed in following paragraphs.
2. The result of estimated audit fees model
The table 2 expresses the result of the estimation of the audit fees model in the full sample. As what are displayed in the table, the R Square showed in the table represents that the model can explain about 77% of the determinants. Furthermore, the coefficients of audit fees model showed in the table 2 will be compared with the predictions before tests.
2.1 The predictions before test
Before the test of the relation between the variables and the fees paid to auditors in the audit fees model, it is predicted that these variables, like total assets (LNTA), the number of employees (EMPLOY), and the number of subsidiaries (SUB) which are used to control the size of client firms have positive relation with audit fees, because the larger the client firms' sizes are associated with higher audit fees (Choi, Kim and Zang, 2006). Next, the ration of the sum of inventories and receivables to total assets (INVREC), the gain or loss before extraordinary items (EXORD) are used to control the complexity of auditing and it is expected that those variables increase audit fees, as the higher these variables the more complexity in auditing process. Moreover, the gain or loss after taxation (LOSS), the ratio between liability and total assets (LEVE), the return on assets (ROA) and the ratio between current liability and current assets (LIQUID) are used to test the risk in auditing process, and with the prediction that there is a positive association between the former two variables (LEVE and LOSS) and audit fees, as the firms with higher potential risk are more likely to result in higher audit fees (Chan, Ezzamel and Gwilliam, 1993); as for the later two variables, it is expected that they are negatively associated with audit fees. In addition to these, the kind of audit firms (BIG4 or NON-BIG4) are included to capture the audit quality, and it is generally accepted that the BIG4 audit firms mean the higher audit quality and therefore result in the premiums of fees paid to auditors (Francis, 1984).
2.2 The compareson between prediction and the result of test
The Table 2 shows the test result of audit fees model (equation 1). Comparing with the prediction, the following result could be drawn up. The regression results which are shown in Table 2(a) reflect the coefficients of the determinants in audit fees model (equation 1). Following the unstandardized coefficients (the following results are all explained by unstandardized coefficients), they obviously show that there is a significant positive relation between total assets and audit fees (ß=0.511, t=24.004); the number of employees and audit fees (ß=0.003, t=7.038) and the number of subsidiaries and audit fees (ß=0.008, t=1.006), therefore, the positive association between audit fees and client firms' sizes could be demonstrated. Furthermore, the evidence for positive relation between audit fees and the auditing complexity is also shown in table which expresses that there is a positive relation between LNVREC (the ratio of the sum of inventories and receivables to total assets and audit fees (ß=0.362, t=2.117); the gain or loss (EXORD) between audit fees (ß=0.137, t=1.536), and all of these results consistent with the prediction. Being same as the prediction for this test, the relation between auditing risk and audit fees, a positive relation can be proved between the gain and loss after taxation (LOSS) and audit fees, the LEVE (the ratio between total liabilities and total assets) and audit fees, as for the relation between the return on assets (ROA) and audit fees, and the relation between LIQUID(the ratio between current liability and current assets) and audit fees, the relation is negative for each which is also consistent with the prediction. Last, the test also gives the evidence that the larger size of audit firms (the BIG4 auditors) will increase audit fees. All of the relations between each determinant in model (equation 1) and audit fees are same as previous studies (i.e. Hoitash, Markelevich and Barragato, 2007; Chan, Ezzamel and Gwilliam, 1993).
3. The correl
Cite This Dissertation
To export a reference to this article please select a referencing stye below: