This paper examines if non-audit service proviso impairs hearer independency, and if the magnitude of hearer independency has changed after the 2004 Procomp dirt in Taiwan. Sing the measurement mistakes of discretional accumulations, we propose an alternate independency proxy- sum audit accommodations defined as the difference between audited net incomes and un-audited net incomes scaled by entire assets. After commanding for the effects of fiscal purchase, runing and market public presentation, industry, company size and audit house size, arrested development analysis indicates that the coefficient for non-audit fees ratio is negative and important in 2003 and 2004.

But, the absolute value of the coefficient is significantly smaller for 2004 than for 2003. Using non-audit fees to replace non-audit fees ratio to carry on arrested development analysis outputs similar consequences. This determination is consistent with the impression that hearers make a tradeoff between deriving service fees and avoiding judicial proceeding and repute loss. Policy deductions are besides offered.Keywords: Non-audit service, Auditor Independence, Procomp Effect, Aggregate Audit Adjustments

Introduction

This paper examines if non-audit service proviso impairs hearer independency, and if the magnitude of hearer independency has changed after the Procomp dirt.

Low balling exists in audit markets all over the world,1 which has led CPA houses to render more non-audit services for lasting in the industry. Advocates of non-audit service proviso assert that synergisms of cognition spillover and audit efficiency arise from jointly supplying audit and no-audit service ( Simunic1984 ; Palmrose1986 ) . However, critics contend that rendering non-audit service additions hearers ‘ fiscal trust on the client. As a consequence, it may project a menace to auditor independency ( Wines 1994 ; Beeler and Hunton 2001 ; Frankel, Johnson and Nelson 2002 ; Firth 2002 ) . Prior empirical research has contributed small consensus on this issue. While some grounds suggests negative associations between non-auditor services and auditor independency ( Wines 1994 ; Frankel, Johnson and Nelson 2002 ; Firth 2002 ) , others province that the menace of cases and repute loss provides inducements for hearer independency and happen no via media of hearer independency ( Crasewell, Stokes and Laughton 2002 ; Defond, Raghunnandan and Subramanyam 2002 ; Ashbaugh, LaFond and Mayhew 2003 ; Chung and Kallapur 2003 ; Elder, Zhou and Chen 2003 ) .We believe that the inconsistent determination may be attributed to the usage of discretional accumulations as a placeholder of hearer independency.

Most anterior empirical research relies on modified Jones theoretical account to gauge discretional accumulations. DeFond et Al. ( 2002 ) argue that discretional accumulations is likely to be an indirect placeholder, and there are empirical jobs in mensurating discretional accumulations. Healy ( 1996 ) points out that the bing theoretical account can non adequately integrate the consequence of alterations in concern basicss.

Discretionary accumulations besides confuse audit quality with net incomes quality because the audited fiscal study is the joint merchandise of direction and the hearer. To prosecute the hearer independency issue, in this paper, we propose an alternate placeholder that is the difference between audited net incomes and un-audited net incomes scaled by entire assets.2 This figure ( hereafter sum audit judgements ) can more straight mensurate the audit accommodation and hence gaining control hearer independency.Despite the assorted findings, the Sarbanes-Oxley Act in the U.S. lists eight types of services that are “ improper ” if provided to a publically held company by its hearer after the Enron prostration. Three old ages after the Enron dirt, the Procomp dirt emerges in Taiwan and provokes public attending to auditor independency in supplying audit service ( see Appendix A for a description of the Procomp dirt ) . Although the Procomp dirt is non associated with non-audit service, the fast and unprecedented countenances by Financial Supervisory Commission ( FSC ) and judicial proceedings against hearers by the Securities Investor and Futures Trader Protection Center in Taiwan may take hearers to re-examine the balance between two types of inducements: service fees as opposed to judicial proceeding hazard and repute loss.

3 Whether hearers will compromise independency when supplying both audit and non-audit services to their client, and whether hearer independency will be enhanced after the Procomp dirt are of import but unsettled issues.This paper aims at analyzing whether supplying non-audit service to scrutinize clients will impair auditor independency, and whether hearer independency is enhanced after the Procomp dirt. In so making, our paper contributes to the literature by utilizing an alternate but more direct placeholder for hearer independency. In add-on, to our cognition, our paper is the first to utilize field-archive informations to look into the consequence of the Procomp dirt on hearer independency in the context of non-audit service proviso.

Finally, our survey will hold deductions for the amendment of CPA Law that is under manner in Taiwan. Our consequences indicate that non-audit fees ratio is significantly and negatively associated with audit judgement in both 2003 ( anterior to the Procomp event ) and 2004 ( after the event ) . But, as expected, the coefficient for 2003 is significantly smaller than that for 2004. Using non-audit fees ( instead than non-audit fees ratio ) as an independent variable suggests similar consequences.The balance of the paper is arranged as follows. The following subdivision reviews the relevant literature and develops our research hypotheses. Section III describes our sample and research design. Section IV presents the empirical consequences.

The last subdivision summarizes our findings.

Anterior Literature and Hypotheses

Recent concerns about hearer independency have focused on the proviso of non-audit services to scrutinize clients. Scholars concern that benefits either from cost nest eggs or from fees gross additions can beef up the economic bond between hearers and their clients, which can farther endanger hearer independency ( e.g. , Beck, Frecka and Solomon 1988 ; Beeler and Hunton 2001 ) . However, Arrunada ( 1999 ) argues that proviso of non-audit services increases the audit house ‘s investing in repute capital, which the hearer is non likely to endanger to fulfill the demand of any one client.

Following DeAngelo ‘s ( 1981 ) “ quasi rent ” statement, sing future quasi rent watercourse will be forfeited, the hearer will non be tempted to compromise their independency. In add-on, Dopuch, King and Schwartz ( 2004 ) province that proviso of non-audit service additions reputation capital since the chance of misstatement hazard is lowered.Empirical trials on the association between non-audit fees and independency mostly follow the research attack to net incomes direction via Jones theoretical account. If non-audit fees, but non the sum fees, have significantly positive association with the accounting accumulations, it appears that additions in non-audit fees influence hearer independency. Frankel et Al.

( 2002 ) foremost proxy hearer independency by discretional accumulations estimated with a cross-sectional modified Jones ( 1991 ) theoretical account and usage sample informations from proxy statements filed with the SEC between February 2001 and June 2001 to analyze this issue. They find a important and positive relationship between non-audit services and magnitude of net incomes direction and conclude that fiscal adhering from non-audit services impairs auditor independency. Studies subsequent to Frankel et Al. ( 2002 ) find contrast consequences because of different research designs for fees and accumulations measurings. For case, Chung and Kallapur ( 2003 ) argue that non-audit fees ratio can non reflect the grade of economic dependance. They suggest utilizing the ratio of client fees ( or non-audit fees ) to the audit house ‘s entire grosss as a step of client importance. They find no relationship between accumulations and proviso of non-audit services. Ashbaugh et Al.

( 2003 ) employ performance-adjusted steps of discretional accumulations and group them into income-increasing and income-decreasing accumulations. They find no association between the non-audit fees ratio and income-increasing discretional accumulations. But, income-decreasing discretional accumulations have important relationship with non-audit fees ratio. Ashbaugh et Al. ( 2003 ) conclude that hearers do non compromise their independency when clients pay high non-audit fees. Elder et Al.

( 2003 ) look into the relation between hearer size, non-audit services, and loan loss proviso of commercial Bankss audited by Large 5 CPA houses. They find a positive relation between non-audit services and loan loss commissariats. As a consequence, no grounds supports a relation between non-audit services and decreased hearer independency.

Other research workers use types of hearer ‘s sentiment as a alternate for hearer independency and non-audit and audit fees as independent variables to analyze their associations. Using audit making as a placeholder for hearer independency, Wines ( 1994 ) finds that hearers of companies non having an audit making of any type derive a significantly higher proportion of their wage from non-audit service fees than the hearers of companies having at least one audit making for a sample of publically listed companies in Australia. Firth ( 2002 ) , utilizing informations from U.

K. , finds the same consequences. Two possible grounds exist for the ascertained relationship: a deficiency of hearer independency, and/or the consultancy services uncluttering up uncertainnesss and dissensions prior to the audit.

DeFond et Al. ( 2002 ) effort to supply extra grounds by look intoing another index of hearer independence- the hearer ‘s willingness to publish a traveling concern audit sentiment. Using hard-pressed houses in U.S.

, they find no association between traveling concern sentiments and either entire fees or audit fees. The consequences are consistent with market-based inducements, such as loss of repute and judicial proceeding costs, ruling the expected benefits from compromising hearer independency. Craswell et Al. ( 2002 ) use a qualified audit sentiment as an index of the exercising of hearer independency and step fees dependance at both the national audit house degree every bit good as the local office degree. They find that the degree of hearer fees dependance does non impact hearer leaning to measure up their audit sentiments.In add-on, Kinney, Palmrose and Scholz ( 2004 ) assume that restatements of antecedently issued fiscal statement reflect low-level hearer independency and make non happen a statistically important and positive association between non-audit fees and restatements.Using publically disclosed informations on audit and non-audit fees, research workers in Taiwan besides investigate the association between non-audit service proviso and hearer independency. Lee, Xu and Chen ( 2003 ) find that hearers supplying both services will let higher magnitude of income-increasing accumulations, where they find no important relationship between income-decreasing accumulations and proviso of non-audit services.

Lu ( 2003 ) investigates whether the association between non-audit fees and auditor independency is moderated by the degree of engagement hazards. She finds no relationship between non-audit fees ratio and the absolute value of unnatural accumulations, but a important synergistic consequence of battle hazard and non-audit fees, proposing that high battle hazards leads hearer to stamp down clients ‘ net incomes direction. Using types of hearer studies as a dependant variable, Hu ( 2001 ) shows that proviso of non-audit service is related to the type of audit study and the alteration of hearers. The survey further indicates that the likeliness of hearer alteration becomes lower as the CPA house renders both services, connoting that hearer independency may be affected.In drumhead, anterior research outputs assorted findings on the contention that non-audit services impair auditor independency.

Methodological factors may account for the incompatibility. DeFond et Al. ( 2002 ) argue that the usage of net incomes direction as estimated by modified Jones theoretical account for alternates of hearer independency is likely to be an indirect placeholder, and there are empirical jobs in mensurating discretional accumulations. In add-on, Healy ( 1996 ) points out that the obvious lack of the bing theoretical account lies in its inability to adequately integrate the consequence of alterations in concern basicss. For illustration, a house ‘s phase in its life rhythm and the types of concern alterations are likely to be falsely classified as discretional accumulations by the current theoretical account. Discretionary accumulations besides confuse audit quality with net incomes quality because the audited fiscal study is the joint merchandise of direction and the hearer.Using audit sentiment as a placeholder for hearer independency, Joe ( 2003 ) examines if hearers are more likely to publish going-concern modified sentiments when the client has the topic of negative imperativeness coverage prior to the day of the month of the audit sentiment.

The consequences show that negative imperativeness coverage leads the hearer to modify the audit sentiment. Therefore, utilizing sentiment type as a placeholder besides suffers from measurement mistake. To extenuate the measuring job, we propose an alternate step as a placeholder for hearer independency. This step exploits the alone scene in Taiwan, which requires listed companies to unwrap audited and un-audited net incomes as good under certain fortunes. The difference between these two net incomes scaled by entire assets may more straight gaining control audit accommodation and hence auditor independency. In add-on, sing the judicial proceeding environment in Taiwan, hearer liability is far lower than that in the U.S.

while the corporate frauds occur. The countenances made by the regulative bureau besides tend to be inefficient due to the drawn-out processs. Therefore, we develop the undermentioned hypothesis:H1: The proviso of non-audit services will act upon hearer independency.Since the Enron prostration in 2001, there have been surveies analyzing the market reactions to Arthur Anderson clients and other Big 4 clients. Dunne et Al.

( 2005 ) find that their stock monetary values travel down and the market suspects the quality of information provided by hearers, bespeaking that the event impairs CPAs repute. In add-on, Lai ( 2003 ) finds increasing figure of traveling concern sentiments issued by hearers. Krishnani??2004i?‰supports the position that heightening net incomes conservativism becomes a common scheme to reconstruct repute for hearers. Hoitash, Markelevich and Barragato ( 2005 ) place a important positive association between non-audit fees and discretional accumulations in old ages 2000 and 2001, but no such association in station Enron old ages, reasoning that hearer independency is improved. The above findings suggest that the behaviour of stakeholders, including hearers, has been changed. In Taiwan, FSC instantly sanctions hearers involved in the Procomp dirt. The Securities Investor and Futures Trader Protection Center besides files a category action against the hearers and CPA houses.

The instant countenances and legal actions addition hearers ‘ judicial proceeding costs and repute losingss. Refering the higher cost of judicial proceeding and repute loss, hearers will re-examine the balance between the inducement of retaining clients and the inducement of take downing judicial proceeding hazard and repute loss, and therefore go more independent. Therefore, we propose the undermentioned hypothesis:H2: In the station Procomp period, the degree of hearer independency will be enhanced when jointly supplying audit and non-audit services to scrutinize clients.Analyzing H2 can supply penetrations into how the Procomp dirt has affected hearers in Taiwan to re-examine the balance between quasi rents and judicial proceeding loss. It will besides hold policy deductions about mechanisms for heightening hearer independency, aside from excluding non-audit services to scrutinize clients.

Research Method

Research DesignUsing discretional accumulations estimated from modified Jones theoretical account to surrogate for hearer independency has been a popular attack in the extant literature. However, it has serious measuring jobs as indicated by Healy ( 1996 ) and DeFond et Al.

( 2002 ) . As a consequence, we use an alternate step by ciphering the absolute value of differences between un-audited and audited net incomes scaled by entire assets. SFC in Taiwan requires that listed companies that have announced their net incomes forecast in a twelvemonth unwrap their un-audited net incomes by the terminal of January the following twelvemonth. If the un-audited net incomes pervert from net incomes forecast by a certain per centum, companies should register an account to the SFC and may be capable to SFC ‘s examination. Similarly, if audited net incomes pervert from un-audited net incomes by a certain per centum, the companies and hearers should register an account to SFC. The intent of this demand is to forestall companies from working net incomes prognosis to pull strings stock monetary values. However, this judicial admission could promote listed companies to pull off net incomes to run into the prognosis.

It could besides promote hearers to fix audit accommodations that result in a little divergence from un-audited net incomes. Given this inducement, differences between audited and un-audited net incomes will more straight capture hearer independency, with larger audit accommodations stand foring less via media of independency. Hsieh and Tsai ( 2005 ) suggest that the aggregative audit accommodation could be a placeholder of audit quality. We use absolute value of the difference scaled by entire assets at the terminal of the old twelvemonth as a placeholder of hearer independency. If the relationship between fees and audit accommodations is negative, it implies a via media of independency.For proving our hypotheses, following Frankel et Al. ( 2002 ) , Chung and Kallapur ( 2003 ) , Ashbaugh et Al. ( 2003 ) , we modify anterior theoretical constructions and organize our independency proving theoretical account:Diff = degree Fahrenheit ( fees, other control variables ) ( 1 )Where, Diff represents hearer independency measured by the absolute value of the difference between audited and un-audited net incomes scaled by entire assets at the terminal of the old twelvemonth.

Most anterior research uses the ratio of non-audit fees to entire fees as an explanatory variable. We farther disaggregate entire fees into constituents. It permits us to prove the separate inducement effects of audit and non-audit fees and extenuate correlated omitted variables bias as audit and non-audit fees are positively correlated ( Frankel et al. 2002 ) . In theoretical account 2, fees, hence, indicates the alternate specifications of fees measurings.We besides infuse audit house size as an extra placeholder of audit quality. Prior research suggests that Big5 hearers are less likely to let net incomes direction than non-Big5 hearers ( Frankel et al. 2002 ; Defond et al.

2002 ; Craswell et al.2002 ; Ashbaugh et Al. 2003 ) . Therefore, our survey includes a Large 4 ( labeled as Big4 ) index variables. In add-on, Reynolds and Francis ( 2001 ) suggest that companies holding fiscal hurt tend to pull strings their net incomes and therefore defy the accommodation proposed by hearers to diminish the chance of debt breach or their capital costs. We use the debt ratio ( Leverage ) and an index variable ( Loss, assigned one if the house reported a loss ) to mensurate fiscal hazard. Frankel et Al. ( 2002 ) , Ashbaugh et Al.

( 2003 ) and Hoitash et Al. ( 2005 ) papers that institutional investors can get information in low cost and have specific capablenesss for monitoring. As ownership held by institutional investors additions, they will non digest net incomes direction behaviour. Therefore, we control the per centum of portions held by institutional investors ( Ins % ) in the theoretical account. Dechow et Al.

( 1995 ) indicate that house public presentation related to net incomes quality. Francis and Ke ( 2002 ) , Frankel et Al. ( 2002 ) , Chung and Kallapur ( 2003 ) , Ashbaugh et Al. ( 2003 ) besides include steadfast public presentation in their survey.

We, thereby, utilize unnatural returns ( RET ) defined as stock returns minus market returns to command for public presentation effects. In add-on, because companies in the electronics industry late have more fiscal dirts and acquire legal countenances, we believe that they will hold high judicial proceeding hazard. Francis et Al. ( 1994 ) suggest that companies with high judicial proceeding hazard are more likely to pull off net incomes. We use an index variable ( ID ) to command for judicial proceeding hazard effects, where ID equals one if the company operates in the electronics industry as coded 13 by Taiwan Stock Exchange ( TSE ) and GreTai Securities Trading Center ( GreTai ) . Last, we control for the consequence of house size ( Size ) since size will be representative of any omitted variable ( Becker et al.

1998 ; Lee et Al. 2003 ) . Size is measured by the natural logarithm of gross revenues. Consequently, our complete theoretical account is as follows:( 2 )For proving H1, we analyze the above theoretical account 2 and gauge the arrested development coefficients for informations in 2003 and 2004, severally. We expect that the coefficients of I?1 will be significantly different from nothing. In order to prove H2, the Procomp effects, we use the F trial to analyze the coefficient difference between two old ages ( 2003I?1= 2004 I?1 ) .

Sample and Data SourceThis survey employs audit fees data publically disclosed by companies listed in TSE and GreTai for 2003 and 2004. Get downing in 2002, listed companies are required to unwrap non-audit fees information provided that these companies pay non-audit fees and audit fees to a same audit house with the ratio more than 25 % or the non-audit fees being more than NTD 500,000 ( about USD 15,295 ) .4 In add-on, companies run intoing one of the undermentioned two state of affairss are required to unwrap audit fees information: ( 1 ) The company changes its audit house and audit fees is less than that of the old twelvemonth. ( 2 ) Audited account fees are less than at least 15 % of the old twelvemonth ‘s sum. Companies run intoing one of the above state of affairss constitute a sample of 213 and 164 for 2003 and 2004, severally. However, companies in the latter two state of affairss do non needfully unwrap non-audit fees information ; we therefore exclude 25 and 38 observations severally.

Similarly, some companies unwrap merely non-audit fees without audit fees or have uncomplete revelation on other fiscal informations. These companies are farther excluded. In add-on, informations provided by fiscal service companies are excluded, because including the fiscal service industry in the analysis will writhe empirical consequences. Finally, we have 56 and 38 observations for each twelvemonth.Fees informations are extracted from footers of the one-year fiscal studies of each company. The one-year fiscal studies are posted on Market Observation Post System web site.

Other information on company features, fiscal information, and ownership construction for each company is obtained from the Finance File and Directorship File of Taiwan Economic Journal Data Base ( hereafter TEJ ) .Table 1 explains the informations choice method. Table 2 shows the distribution of sample houses across the industries as classified by TSE and GreTai. It shows that 70 % of our sample is from the electronics industry. Though we exclude fiscal service companies from analysis, the figure of companies in this sector is half of that in the electronics industry, bespeaking that both industries demand more non-audit services.Table 1. Sample choice20032004entireInitial Sample With Fees revelations213164377Missing Audited account Fees Data( 0 )( 1 )1Missing Non-audit Fees Datas( 25 )( 38 )63Firms in Financial Service Industries( 21 )( 22 )43Missing Aggregate Audit Adjustments Data( 95 )( 50 )145Other Information Incomplete( 16 )( 15 )31Entire563894Table 2.

Sample distribution by industriesIndustrycodification20032004entireCement1101Food2123Plastic3112Fabric4101Electric machinery5112Appliance overseas telegram6101Paper8101Steel Fe10123Rubber11213Car12235Electronicss13382563Construction14112Department shop18101Other20426Entire563894

Empirical Consequences

Sample FeaturesThe descriptive statistics for our variables are in Table 3 and Table 4. The entire fees are, on norm, about NTD 5,100 1000s ( about USD 156,011 ) for 2003 and 2004. The mean of audit fees is equal to NTD 3,408 1000s ( about USD 104,252 ) and NTD 3,276 1000s ( about USD 100,614 ) severally. That is 12 % higher than that of Chang and Tsao ( 2005 ) , which use informations for 2002. The ratio of non audit fees to entire fees represents about 32 % for both old ages. The non-audit fees are about to NTD 1,665 1000s ( about USD 50,933 ) and NTD 1,889 1000s ( about USD 57,785 ) severally. Average audit fees decrease by 4 % from 2003 to 2004, whereas non-audit fees represent a 14 % addition.

The fees tendency shows that audit houses become more dependent on non-audit services grosss.The norm of aggregative audit accommodations in 2004 is 0.016, which is twice higher than that of 2003 ( 0.007 ) . The higher sum of aggregative audit accommodations indicates that hearer independency is improved during the period of 2004. Therefore, the preliminary analysis provides grounds consistent with hypothesis 2 about the Procomp consequence.

As a proof, we besides find that the per centum of houses having qualified and modified unqualified sentiments is 54 % and 56 % for 2003 and 2004 severally, deducing that hearers lightly tend to go conservative in 2004.Sing the control variables, the mean steadfast size for 2003 and 2004 is severally 0.059 and 0.073 with important difference.

Liability to plus ratio is 49 % for both old ages. About 13 % of the companies incur loss in 2003, which is about four times the rate in 2004 ( 3 % ) . About 85-90 % of companies in the sample employ Big4 audit houses. Average ownership by institutional investors is 38 % for 2003 and 2004. Abnormal stock returns differ greatly between these two old ages.

On norm, companies earn a return that is 7 % higher than the market index in 2003, whereas companies earn a return less than market index by 11 % in 2004.Table 5 presents the correlativities of variables for pooling informations. The aggregative audit accommodation has negative correlativities with Big4 and unnatural return, but do non happen important correlativities with others. The correlativities among variables are by and large little in absolute value with the largest correlativity equal to 0.

25. We besides use two statistical methods to measure multicollinearity. They are tolerance statistics and the discrepancy rising prices factor ( VIF ) . The tolerance value for each independent variable in our theoretical account exceeds 0.1. Valuess of discrepancy rising prices factors for our forecasters are all less than 10. We therefore conclude that multicollinearity is non a distinguishable job in our survey.

Table 3. Descriptive statistics of 20032003 YearNitrogenMeanMedianVenereal diseaseSoapMinuteNon-audit Fees Ratio560.3080.2980.0160.6220.

065Entire Fees564984.6839043214.60135871532Audited account Fees563408.2525252487.537127041080Non-audit Fees561665.60511751416.3717400260Diff560.0070.

0010.0010.0760CPA Opinion560.54010.25310Size560.0590.0760.

1420.441-0.470Leverage560.4940.5030.1440.7730.

086Loss560.13000.11110Big4560.89410.3110INS %560.3780.3370.2230.

9530.038RET560.066-0.0830.4642.

046-0.451Industry560.77010.18110Table 4.

Descriptive statistics of 20042004 YearNitrogenMeanMedianVenereal diseaseSoapMinuteNon-audit Fees Ratio380.2880.2820.

0310.6830.065Entire Fees384655.85040182705.480124141337Audited account Fees383276.

0922689.502031.04912099300Non-audit Fees381888.6581197.5002035.

5431200350Diff380.0160.0030.0040.3730CPA Opinion380.56010.

25210Size380.0730.0860.1170.253-0.323Leverage380.

4870.4880.1410.7600.227Loss380.02600.02610Big4380.85010.

13410INS %380.3780.3260.2160.8020.

011RET38-0.117-0.0900.3150.

461-0.718Industry380.74010.18110Diff = the absolute value of difference between un-audited and audited net incomes scaled by assets.CPA Opinion = 1 if qualified or modified unqualified hearer studies, and 0 otherwise.

No-audit Fees Ratio = non-audit fees divided by entire fees.Entire Fees = the natural log of the entire fees.Audited account Fees = the natural log of the audit fees.Non-audit Fees = the natural log of the non-audit fees.Size = the natural log of net gross revenues.Leverage = debt over by assets.

Loss = 1if the house reports loss in the ascertained twelvemonth, and 0 otherwise.Big4 = 1 if the house is audited by Big 4 audit houses, and 0 otherwise.Ins % = ownership by institutional investors.RET = stock return subtraction market return.ID = 1 if industry codification is equal to 13, and 0 otherwise.Table 5. Pearson correlativityDiff = the absolute value of difference between un-audited and audited net incomes scaled by assets.

Non-audit Fees Ratio =non-audit fees divided by entire fees.Entire Fees = the natural log of the entire fees.Audited account Fees = the natural log of the audit fees.Non-audit Fees = the natural log of the non-audit fees.aˆ * , ** , *** are important degrees at 0.1, 0.05, and 0.

01respectively.Size = the natural log of net gross revenues.Leverage = debt divided by assets.

Loss = 1if the house reports loss in observe twelvemonth, and 0 otherwise.Big4 = 1 if the house is audited by Big 4 audit houses, and 0 otherwise.Ins % = ownership by institutional investors.RET = stock return subtraction market return.ID = 1 if industry codification is equal to 13, and 0 otherwise.Arrested development ConsequencesTable 6 demonstrates that the ratio of non-audit fees to entire fees ( i.

e. , non-audit fees ratio ) is negatively related to the aggregative audit accommodations for both old ages ( p=0.0298 ; 0.0518 ) with the magnitude of coefficients being greater in 2004 ( -0.098 ) than in 2003 ( -0.200 ) . Using non-audit fees instead than non-audit fees ratio as an explanatory variable indicates that the coefficient of non-audit fees is negative ( non-significant ) in 2003 while positive and important in 2004 ( p=0.

0164 ) . These consequences suggest that more non-audit fees lead to less aggregative audit accommodations, connoting hearers ‘ via media of independency. But, the inclination diminishes and hearers even become more independent despite the addition in non-audit fees in 2004.This determination is consistent with the impression that when hearers weigh more on non-audit service fees than on the judicial proceeding hazard and repute loss, they tend to compromise their independency in executing fiscal statement confidence.

This inclination occurs because judicial proceeding hazard and repute loss are low. However, when judicial proceeding hazard and repute loss are more terrible, hearers will re-consider the tradeoff between these types of inducements ( Arrunada1999 ; Palmrose 1999 ; Dopuch et al.2001 ) . As a consequence, hearers will go more independent by inquiring more conservative accounting interventions.

Therefore, we observe that hearers ‘ accommodation is negatively related to non-audit fees ratio ( and non-audit fees ) in 2003 in which the legal environment can non efficiently countenance hearers ‘ mis-behavior. However, the legal environment becomes more rigorous as a response to the Procomp dirt in 2004. We therefore observe that the coefficient of non-audit fees ratio becomes less negative in 2004. The coefficient of non-audit fees even becomes positive. F trials indicate that the differences in the coefficients for non-audit fees ratio and for non-audit fees between 2003 and 2004 are statistically important ( F values are 6.

09 and 4.24 severally ) . The consequences are consist with Krishnani??2004i?‰and Hoitash et Al. ( 2005 ) statements that Enron event leads hearers to go conservative. Overall, our consequences back up H1 and H2.With regard to other explanatory and control variables, we find that merely a few variables are important.

Of peculiar involvement is Big4 which has a negative relationship with aggregative audit accommodation. It might bespeak that hearers of Large 4 require less adjustment than those of non-Big 4. This account can non account for the magnitude of coefficients in both old ages. Although the coefficients for Big4 in both old ages are negative, the absolute value is larger for 2004 than for 2003 ( P values are & lt ; 0.03 ) . An alternate account is that companies self-select hearers and hence Large 4 hearers require less grade of accommodation for the clients that select Large 4 as hearers. In add-on, Big 4 hearers are more careful in choosing and retaining clients after the Procomp dirt and hence their clients are less aggressive in fiscal coverage. As a consequence, Big 4 hearers require even less accommodation for their clients in 2004.

RET has a positive but undistinguished association with aggregative audit accommodation in 2003, but the coefficient becomes negative and important in 2004. This might back up the old account that hearers are more conservative in choosing and retaining audit clients in 2004. The more selective the client showing, the lupus erythematosus is the needed audit accommodation.To prove the hardiness of our consequences, we replicate our analysis utilizing different samples: ( 1 ) a sample that excludes upward audit accommodation ; or ( 2 ) a sample of houses runing in the electronics industry. We retain merely the sample with downward audit accommodations because income-increasing net incomes direction is more concerned than income-decreasing net incomes direction by the populace and the regulative bureau ( Ashbaugh et Al. 2003 ; Lee et Al. 2005 ) .

In add-on, downward audit accommodation corresponds more closely to auditor independency. We include merely electronics companies as an alternate sample because they may be capable to higher judicial proceeding hazard. The consequences from all of the arrested developments ( presented in Panel A and Panel B of Table 7 ) study small difference from those in Table 6. Both tabular arraies systematically show that the coefficient for non-audit fees ratio is negative and important in 2003, but positive in 2004. The coefficient for non-audit fees is negative in 2003 but positive and important in 2004.

Therefore, our findings are non sensitive to our sample-selection standards.Table 6. Arrested developments of aggregative audit accommodations and audit andnon-audit feesModel:2003200420032004Non-audit Fees Ratio-0.200**-0.098**Audited account fees0.007-0.

032Non-audit Fees-0.0030.014**Size0.021-0.127-0.017-0.109Leverage-0.

0190.046-0.0230.037Loss0.009-0.0010.0070.006Big4-0.

015**-0.072***-0.018**-0.093***Ins %-0.005-0.

005-0.0070.004RET6-E4-0.054*7-E4-0.067**Industry0.0060.0240.

0060.023Adj-R Square0.1040.2770.

1110.324Model F1.812.821.783.03N56385638Hypotheses/F TestH0: I?2003= I?2004H0: I?2003= I?20046.

09 ***5.60** ( audit fees )4.24** ( non-audit fees )Diff = the absolute value of difference between un-audited and audited net incomes scaled by assets.Fees is defined as follows:Non-audit Fees Ratio = non-audit fees divided by entire fees.Audited account Fees = the natural log of the audit fees.Non-audit Fees = the natural log of the non-audit fees.Size = the natural log of net gross revenues.

Leverage = debt divided by assets.Loss = 1 if the house reports loss in the ascertained twelvemonth, and 0 otherwise.Big4 = 1 if the house is audited by Big 4 audit houses, and 0 otherwise.Ins % = ownership by institutional investors.RET = stock return subtraction market return.ID = 1 if industry codification is equal to 13, and 0 otherwise.aˆ * , ** , *** are important degrees at 0.

1, 0.05, and 0.01, severally.aˆ aˆ Hypotheses/ F trial is for proving the difference of coefficients for fees variables between 2003 and 2004.Table 7.

Arrested development analysis utilizing different samplesModel:2003200420032004Panel A. Sample of downward audit accommodationsNon-audit Fees Ratio-0.028**0.134Audited account Fees0.009**-0.

355Non-audit Fees-0.0060.686**Adj-R Square0.1200.3580.4670.469Model F1.94***2.

68***9.38***3.36**Hypotheses/F trial3.73**6.13*** ( Audit Fees )4.58** ( Non-audit Fees )Panel B. Sample of the electronics industryNon-audit Fees Ratio-0.

052**0.061Audited account Fees0.012**-0.036Non-audit Fees-0.

0100.007**Adj-R Square0.1530.

3580.4690.469Model F2.08***2.68***9.38***3.36**Hypotheses/F trial2.

87*6.83*** ( Audit Fees )2.79** ( Non-audit Fees )Diff = the absolute value of difference between un-audited and audited net incomes scaled by assets.Fees is defined as follows:Non-audit Fees Ratio = non-audit fees divided by entire fees.

Audited account Fees = the natural log of the audit fees.Non-audit Fees = the natural log of the non-audit fees.Size = the natural log of net gross revenues.Leverage = debt divided by assets.Loss = 1 if the house reports loss in the ascertained twelvemonth, and 0 otherwise.Big4 = 1 if the house is audited by Big 4 audit houses, and 0 otherwise.Ins % = ownership by institutional investors.

RET = stock return subtraction market return.ID = 1 if industry codification is equal to 13, and 0 otherwise.aˆ * , ** , *** are important degrees at 0.1, 0.05, and 0.

01, severally.aˆ aˆ Hypotheses/F trial is for proving the difference in coefficients for fees variables between 2003 and 2004.aˆ aˆ aˆ This tabular array merely reports consequences of fees variables in the theoretical account.

Decision

This paper aims at analyzing whether proviso of non-audit service to scrutinize clients impairs hearer independency, and whether the 2004 Procomp dirt has an consequence on heightening hearer independency. Using publically disclosed fees informations for financial old ages of 2003 and 2004, we through empirical observation test the associations between non-audit service proviso and hearer independency for both old ages.

We use the ratio of non-audit fees to scrutinize fees ( i.e. , non-audit fees ratio ) and non-audit fees as alternate alternates for the proviso of non-audit service. To extenuate the measuring job with the usage of modified Jones theoretical account to gauge discretional accumulations as a alternate for hearer independency, we adopt an alternate alternate. This step is calculated by taking the absolute value of the difference between audited net incomes and un-audited net incomes scaled by entire assets. Listed companies in Taiwan are required to denote their un-audited net incomes by the terminal of January if they have announced net incomes prognosis in the old twelvemonth. This demand provides an alone chance to acquire entree to the official un-audited net incomes informations.After commanding for the effects of fiscal purchase, runing and market public presentation, industry, company size and audit house size, arrested development analysis indicates that the coefficient for non-audit fees ratio is negative and important in both old ages.

But, the absolute value of the coefficient is significantly smaller for 2004 than for 2003. This determination supports the claim that proviso of non-audit service additions hearers ‘ fiscal trust on their audit clients and hence impairs their independency by necessitating less audit accommodation. Further, it indicates that hearer independency is enhanced after the Procomp dirt. Using non-audit fees to replace non-audit fees ratio to carry on arrested development analysis outputs similar consequences. But, the Procomp consequence is even stronger. While the coefficient for 2003 is negative, it becomes positive and important in 2004. Sensitivity analysis utilizing merely the companies whose net incomes are adjusted downward by hearers as the sample suggests qualitatively the same consequences. Another sensitiveness analysis with lone companies in the electronics industry as the sample besides indicates similar consequences.

This determination is consistent with the impression that hearers are faced with two types of inducements: deriving service fees and avoiding judicial proceeding and repute loss. When the judicial proceeding and repute loss is non terrible, they will act as if they were weighing more on service fees than on judicial proceeding and repute loss. It is precisely the legal environment in Taiwan prior to the Procomp dirt. Prior to this event, regulative countenances against hearers are inefficient due to the drawn-out processs and no civil case against hearers is successful. The investor protection is non effectual either. This legal environment provides hearers with small, if any, inducement to execute independently.

However, instantly after the event broke out, SFC expeditiously sanctions the involved hearers by suspending them from pattern for two old ages. The Securities Investor and Futures Trader Protection Center besides files case against hearers and audit houses. These unprecedented actions against hearers may hold made hearers re-examine the balance between service fees and judicial proceeding hazard when executing audit and non-audit services to the clients. In fact, there is some grounds that Large 4 houses are more careful and conservative in choosing audit clients as indicated by the per centum of Big 4 clients diminishing from 89.4 % in 2003 to 85 % in 2004. Because of the more selective pattern, the coefficient for Big4 becomes even more negative in 2004.

The differences in coefficients between 2003 and 2004 are important, irrespective of whether the independent variable is non-audit fees ratio or non-audit fees ( p values are & lt ; 0.01 ) .Our survey contributes to the literature by suggesting an alternate alternate for hearer independency. It captures the alone revelation ordinance in Taiwan that requires companies to officially and publically unwrap informations on un-audited net incomes, which is frequently hard to obtain in other states without such a demand. Furthermore, this step straight observes the magnitude of aggregative audit accommodations made by hearers. In add-on, to our cognition, this survey is the first to analyze the impact of the Procomp dirt on hearer independency.

Such an enquiry non merely provides empirical grounds on how hearers balance different types of inducements, but it besides has of import deductions for policy devising. After the Enron dirt, hearers in the U.S. are barred from supplying eight types of non-audit services to their audit clients. The Certified Public Accountants Law in Taiwan is presently under amendment. The bill of exchange proposes that hearers are proscribed from executing non-audit services to their audit clients whenever these non-audit services impair independency. Although the bill of exchange does non name specific non-audit services, as the U.S. does, the kernel of the statute law is to forbid non-audit service. Based on our findings, we believe that beef uping regulative bureau efficiency and jurisprudence enforcement will be a better option. In fact, old research indicates that jointly supplying audit and non-audit services to the clients can increase audit efficiency via cognition spillover ( Simunic1984 ; Palmrose1986 ) . As an economic agent, hearers will do a balance between service fees and judicial proceeding hazard and loss.The survey has the undermentioned restrictions. First, the sample in our survey consists of companies paying non-audit fees to their hearers at least NTD 500,000 ( about USD 15,295 ) , or at least 25 % as a per centum of non-audit fees to entire fees. Whether the findings apply to other companies requires farther survey. Finally, our determination is based on the current institutional background ; farther research may analyze hearers ‘ behaviour after the transition of the amended CPA Law.

Notes

Harmonizing to China Times ( 2004/07/19 ) , audit fees in Taiwan are lower than fees in the other Asiatic states. It reports that audit fees in Taiwan are merely one 3rd every bit much as fess in Hong Kong.It is a alone institutional scene in Taiwan that listed companies are required to denote their un-audited one-year net incomes by the terminal of the following twelvemonth ‘s January if companies have of all time announced net incomes prognosis in a twelvemonth. Listed companies are required to denote their audited one-year net incomes by the terminal of April regardless of whether or non they have announced net incomes prognosis.Since the Procomp dirt took topographic point in June, 2004, FSC instantly suspended hearers from pattern for two old ages and Securities and Futures Investors Protection Center filed a category civil action against hearers. Procomp dirt is non associated with proviso of non-audit services to the audit client, but these terrible and fast countenances and legal actions against hearers were ne’er occurred in Taiwan before.The NT dollar to US dollar transition rate is 32.69:1. In the undermentioned, we use this transition rate to show US dollars.