auditor switch decision of malaysian listed firms: tests ... papers/jssh vol. 8 (2) sep....

Pertanika J. Soc. Sci. & Hum. 8(2): 77 - 90 (2000) ISSN: 0128-7702© Universiti Putra Malaysia Press

Auditor Switch Decision of Malaysian Listed Firms: Tests ofDeterminants and Wealth Effect

HUSON JOHER*, M. ALI, SHAMSHER M., ANNUAR M.N. 8c M. ARIFFDepartment of Accounting and Finance, Faculty of Economics and Management,

Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia

Keywords: Audit switch, audit quality, revaluation effect

ABSTRAKKajian ini menilai rasional ekonomi untuk mengganti juruaudit oleh firma-firma tersenaraidengan meneliti kesan keputusan ini terhadap perubahan harga saham firma-firma tersebut (ataudikenali juga sebagai kesan harta). Keputusan pihak pengurusan firma untuk mengekal ataumenggantikan juruaudit ini melibatkan satu perubahan ke atas firma-firma audit yang berbezakualiti. Kualiti audit ditakrifkan dengan mengklasifikasikan firma-firma audit kepada firma-firmaTahap 1 (Big-5) dan firma-firma Tahap 2 (bukan-Big 5). Ciri yang membezakan antara duakumpulan produk audit dipercayai menjadi kredibiliti yang dibawa oleh setiap kumpulan dalamperjanjian audit. Faktor-faktor yang berhubung dengan pilihan firma audit dan perubahan untukciri-ciri firma berkaitan pilihan juruaudit disiasat menggunakan model regresi logistik. Hasilkajian menunjukkan bahawa penggantian juruaudit oleh firma-firma tersenarai sebahagiannyaditerangkan oleh perubahan dalam pengurusan dan pertumbuhan perolehan. Perubahan dalamciri-ciri firma seperti pertumbuhan aset, pembelian aset tetap kepada jumlah aset, kemampuanmempengaruhi orang lain dan perubahan dalam aktiviti kewangan menerangkan penggantianjuruaudit. Hal ini menunjukkan tiada bukti kesan harta signifikan daripada pengumumanpenggantian juruaudit.

ABSTRACTThis article examines the economic rationale for auditor change by Malaysian listed firms byexamining audit switch effect on share prices. The auditor change decision by management toretain or to change involves a switch across audit firms with different quality. Audit quality isdefined by classifying the audit firms into Tier 1 (Big-5) firms and Tier 2 (non-Big 5) firms. Thedistinguishing attribute between the two groups of audit products is believed to be the credibilitythat each group brings to the audit engagement. Factors associated with the choice of audit firmand changes for firm characteristics associated with auditor choice were investigated using thelogistic regression model. The findings show that the auditor switch of Malaysian listed firms ispartly explained by changes in management and turnover growth. Changes in firms' characteristicssuch as asset growth, purchase of fixed asset to total asset, leverage and changes in financingactivities explain auditor switches. There appears to be no evidence of significant wealth effectfrom auditor switch announcements.

INTRODUCTION of financial reporting and cost of monitoringAccounting literature on auditor change decision management activities is well documented in theand its implications on firm's value, credibility literature emanating from the developed

countries. Auditor switch decision involves

The first author expresses his gratitude to Universiti Pulra Malaysia for providing the financial grant and scholarly assistancethrough qualified supervisors to complete this major study as his Ph.D thesis. He wishes to record his appreciation to thereviewers of this paper for selecting this as the best paper for the award of the first prize at Malaysian Finance Associationmeeting in May, 2000. The authors jointly thank the anonymous reviewers for their decision to accept this paper in theproceedings.

Huson Joher, M. Ali, Shamsher M., Annuar M.N. &: M. Ariff

change of incumbent auditor resulting in thechoice of quality differentiated audit firms torealign the characteristics of the audit firm withthe growing needs of clients under changingcircumstances. Changes in management,perceived expertise of audit firms anddeterioration of financial health of clients havebeen found to be associated with auditor change/switch decisions. Changes in a firm's activitiesand perception of advances in audit technologyhave been shown to be associated with the choiceof quality differentiated audit firms.

Changes in management might result inreplacement of the incumbent auditor with aview to imbibe fresh ideas to enhance the firm'sexpansion policy under a changed management.Similarly, auditor replacement will be initiated ifthe existing audit firm lacks the expertise tokeep up with the firm's expansion policies andits changed internal control systems. Firmsexperiencing consistent deterioration inperformance may also decide to replace theincumbent audit firm with a more compliantauditor in an attempt to evade a qualified reportdetrimental to the value of the firm.

Change in firm's activities (expansion,contraction, financing, performance, etc.) andaudit technology creates demand for the choiceof quality differentiated audit firms. The rationalefor choosing a relatively higher quality auditfirm might be due to the growing needs of thefirm, to take advantage of the audit firm'sreputation. The choice of a lower quality auditfirm might be prompted by a sudden contractionof business activities, to gain an ability tonegotiate audit comments to reflectmanagement's view rather than an unsolicited"fair view" as well as a desire to lower costs ofengaging audit services.

Due to asymmetry of information betweenprincipals and management, management ofgrowing firms might redirect resources, aspecuniary and non-pecuniary benefits on thejob, at the expense of shareholders. Theshareholders have to incur costs to ensure thatmanagement's activities are consistent withshareholders' objectives. Management of highlylevered firms might be tempted to transfer wealthfrom their shareholders by engaging in riskyinvestments beyond that sanctioned byshareholders. Engaging relatively higher qualityaudit firms mitigates against these agency costs

elements of management but which areultimately borne by shareholders.

Revaluation effect of auditor switch has beenan issue of interest among investors and unlikecorporate dividend and earnings announcements,which reflect a real change in expected corporateperformance, auditor change announcementsconvey no direct apparent economic information.The economic effect from the latter event is thesignal associated with different investors'interpretation about the quality of audit servicesprovided by the auditor. Investors are observedto utilise the auditor change/switchannouncements to revise their expectation ofthe firm's expected future cash flows, and henceits share prices. A change to higher prestigeauditors might be perceived as an improvementin audit services and hence an expected positiverevaluation effect may result. Similarly, a changeto lower prestige audit firms might be perceivedas negative news. Evidence (Nichols and Smith1983; Eichensher et al 1989) suggests that largeraudit firms provide higher quality audit servicesby offering greater credibility to clients' financialstatements than the small audit firms.

Though there is substantial documentationon determinants and revaluation effect of auditorswitch announcements in developed markets,there is a hardly any documented evidence onsimilar issues in developing markets, likeMalaysia. This research examines thedeterminants and the revaluation effect ofauditor change announcements of firms listedon the KLSE. Section 2 presents literature onthe economic rational for auditor switch. Section3 provides discussion on methodology and datacollection. Section 3 is further divided into testmodel, abnormal returns measures and statisticaltests. Section 4 provides discussion on findingsfor simple parametric test, logistics regressionand event study methodology. The final sectionsummarises the findings of the paper.

LITERATURE REVIEW

The theory of the firm as amended to includeAgency Problem emphasises the importance ofmonitoring management activities. Jensen andMeckling (1976) suggest that auditing is onemonitoring device that can mitigate agency costs,implying a need for independent audit services.Based on Watts and Zemmerman's (1978) work,DeAngelo (1981a; 1981b) developed a demandand supply rationale for audit quality. Audit

78 PertanikaJ. Soc. Sci. 8c Hum. Vol. 8 No, 2 2000

Auditor Switch Decision of Malaysian Listed Firms

quality is defined as the probability that anauditor will both discover the breach of contract(material mis-statement) and subsequentlyactually report it. It is implied that auditorsspecialise in supplying various level of auditquality and audit firm size is an effective surrogatefor audit quality. Firms change their auditors toensure a desired quality of audit service.

An analogy from product differentiatedhypothesis is that firms use auditor choice as asignalling device to reveal their desirablecharacteristics. Investors incorporate the arrivalof new information (choice of quality auditor)and re-evaluate the firm's value. Investors arewilling to pay a relatively higher price for betterperforming firms. Holthausen and Verrecchia(1990) suggest that firms appear to signal theirex ante uncertainty by hiring a higher prestigeaudit firm to perform their audit. This signal iscredible to the market since the auditor'scompensation is higher exhibiting firm-specificreputation capital. Firms with unfavourableinformation would prefer a lower quality auditor.

The literature on auditor changedocumented in the developed markets offersseveral explanations for factors affecting bothswitching and its affect on share revaluation.Early work on these issues by Burton and Robert(1967) and Carpenter and Strawser (1971)provide evidence on the determinants of auditorswitch decisions. They documented a positiverelationship from changes in management,changes in new financing and switching auditor.

Qualified audit reports are important indetermining auditor switch. Managersstrategically use switch decisions to avoid anyunfavourable information release to investors(Chow and Rice 1982; Crawswell 1988; Dye 1991;Citron and Tafler 1992). However, the findingsof Gul et al (1991) and Takia et al (1993) didnot support this notion. Other factors includethe demand for additional audit service (Burtonand Robert 1967; Lurie 1977), firms* growth(Lingbeck and Rogow 1978), financial distress(Schwart and Menon 1985; Dhaliwal andSchwartzberg 1993), and the importance of auditfee to corporate management decision(Bedingfield and Loeb 1974; Ettredge andGreenburg 1990).

There is evidence of a significant relationshipbetween firm size, growth and choice of auditor(Healy and Lys 1986; Johnson and Lys 1986;

Simunic and Stein 1987). In general, firm sizeincreases contribute to agency costs since itcreates a vast opportunity for managers toconsume non-pecuniary benefits thus resultingin a demand for a quality audit firms (Tier 1)(Fama and Jensen (1983a; 1983b)). Alternatively,

Johnson and Lys (1986) argue that fixedinvestment in the auditor error detectiontechnology leads to specialisation in marketsegment and difference in technologies and costfunction across market segments are likely to bereflected by difference in audit firm's size(Francis and Wilson 1988). Palmrose (1984),Eichenseher and Shields (1986), Johnson andLys (1990) showed a positive association betweenleverage and choice of Tier 1; negative associationfor Tier 1 audit firms which underwent mergeractivities (Healy and Lys 1986). Healy and Lysalso assert that clients who issue new debtsecurities remain with Tier 1 audit firms to takeadvantage of its reputation and thereby lowerinvestors' information costs in assessing theinvestment quality. Francis and Wilson (1988)provide support for an hypothesised associationbetween agency costs and choice of brand nameafter controlling for growth and client size.

Evidence of market reaction on auditorswitch decision is inconclusive. Fried and Schiff(1981) examined the disclosure requirement bySEC and the degree of market reaction to suchdisclosures surrounding the auditor changes.The findings suggest a negative effect on average.The literature offers several explanations fornegative revisions in stock prices, which, amongothers include, substantial direct and indirectcost associated with auditor switch and investorperception of poor economic prospect of firm'soperating, financing and performance. Dupuchand Simunic (1982) suggest that firms switchingto higher prestige audit firms will yield a positiveresponse while switching to lower prestige auditfirms will have negative response from marketparticipants.

Smith and Nichole (1982) documented adispute over accounting and auditing principleswith auditors prior to the auditor switch andthose of client firms which did not disclose anydispute. A systematic price decline was reportedsurrounding the auditor switch for a client firmwhich reported a dispute with the auditor.

Johnson and Lys (1990) examined themarket reaction to voluntary auditor changes

PertanikaJ. Soc. Sci. & Hum. Vol. 8 No. 2 2000 79

Huson Joher, M. Ali, Shamsher M., Annuar M.N. & M. Ariff

and reported no price reaction. Davidson andGribbin (1995) documented a negative abnormalreturn to the announcement of auditor changeand postulated that it might be due to themarket's lack of confidence about the motivefor the change. John et al (1999) showed anegative market reaction to auditor resignationand suggested that auditor resignation fromoffice is likely to be a cost signal for audit firmsparticularly when a client firm is a listed company.

DATA AND METHODOLOGYOne hundred and thirty-five firms that switchedtheir auditors over the period 1986 to 1996 weresampled. The complete data set for all analysiswas available for 108 firms. The sample wasverified using annual reports and announcementdates for auditor changes were obtained fromthe minutes of the annual general meeting. Therevaluation effect of auditor switch was analysedusing stock prices and Composite Index valuesextracted from the daily diary of KLSE.

Following Zurada et al (1998), the logisticregression model is used to analyse the decisionto change, retain auditor or (switch to higheror lower prestige audit firms). This model avoidsnormality assumptions when the dependentvariable is dichotomous and produces highestclassification accuracy for the traditionaldichotomous response variables. The functionalform of a logistic cumulative density function:

P (Y= 1IX) = exp (XPk Xk)/[l+exp(5:pk Xk)](1)

The unknown parameters (a, p) areestimated using Maximum Likelihood Estimators(MLE) in contrast to ordinary regression modelswhich are estimated by the method of LeastSquares Estimators (OLS). Since the likelihoodequations for logit equations are non-linear inthe parameters to be estimated, algebraicsolutions are not obtainable and thereforeapproximation by standard iterative algorithmsis used.

Test ModelParametric Test

The parametric test of the differences in themean value of the characteristics of sampledfirms (firms changing their audit firms) and

control firms (client firms that did not changetheir audit firms) was conducted. Thecharacteristics are turnover, average asset,acquisition, of fixed asset return on asset, leverageand liquidity position of the firms. A similar testwas also conducted to examine the differenceamong client firms associated with qualitydifferentiated audit firms.

Auditor Change ModelThe stepwise logistic regression technique wasselected to ascertain the important determinantsof audit switch decision. The functional form ofthe regression equation is as follows:

(2)

Z= A with A = 1 or 0 indicating that a client firmdid (1) or did not switch auditors (0). X = thevariables identified for the model. These aremanagement change (MGTCH), averageacquisition of fixed asset to total asset (ACQUI),turnover growth (GROWTH) both prior andafter the auditor switch, liquidity (LIQ), firmsleverage (LEV), average returns on asset (AROA),average earnings per share (EPS), qualified auditreport both prior and after the auditor switch.

Change in management could serve asprincipal-agents contractual arrangement as newmanagement could demand for the replacementof an incumbent auditor with a new one withwhom it has favorable dealings in the past andwho will bring new ideas that is instrumental tothe firm's expansion policy. This is measured bytaking value of one if there is a change inmanagement or zero otherwise. Rapid growthcould be a measure of principal-agent contract.Clients who are constantly acquiring subsidiariesand expanding into new markets would demandnew auditors who are more effective indischarging auditing service. Rapid growth ismeasured by percentage changes in turnovergrowth three years prior and three years afterthe auditor switch. Auditor effectiveness ismeasured by the size of the audit firm, that iswhether the audit firm is a member of higherprestige auditor or otherwise prior to the auditorchange. This measured by taking the value ofone if pre-switch audit firm was a member ofhigher quality (Tier 1) audit firm or 0 forotherwise. Client firms whose reputation istarnished by its poor performance, corporatemanagement will try to change auditors to avoid

80 PertanikaJ. Soc. Sci. & Hum. Vol. 8 No. 2 2000


any unfavourable information disseminated tothe capital market. A qualified report, averagereturn on asset, average earnings per share andliquidity of the firms are used as proxy forclient's reputation. Qualified audit report is abinary variable which takes the value of 1 ifauditor issued qualified report one or two yearsprior to or after auditor switch or otherwise. Anoperational variable such as audit fees takes thevalue of 1 if there is a reduction in audit feesubsequent to auditor switch or otherwise.

Auditor Choice Model

The analysis of the firms' characteristics and thedirection of auditor changes (Tier 2 to Tier 1audit firms and vice versa) are done using logisticregression model. Previous studies (Johnson andLys 1990; Francis and Wilson 1988) used similarmodels to determine the characteristics of thefirms which are associated with direction of theauditor changes. The hypothesised relationshipmay be expressed as follows:

(3)

whereY = 1 indicating firms switching to higherprestige (Tier 1) audit firms and 0 indicatesfirms switching to less prestige (Tier 2) auditfirms.X. = predictor (independent) variables, and(Aj, Xr \v...,Xk): the coefficient of the predictorvariables.

Variable MeasurementsThe frequently used variables to proxy for thefirm's change in activities over time are assetgrowth, asset size, turnover growth, changes inacquisition, firm's leverage, changes in financing,changes in operating cash flow, and averagereturns on asset.

Expansion: Expansion entails increasing inscope, geographical dispersion and volume ofclient's activities. The corresponding increase inquantity and complexity of accountingtransactions results in economies for largerauditors, which provide high quality audit service(De Angelo 1981). The expansion or contractionis proxied by four operational variables namelyannual growth in total assets three years prior toand three years after the switch: it is indicated asGRTHB and GRTHA respectively. Changes inaverage acquisition of fixed to total asset is

abbreviated to CHACQ and annual growth ofsales prior to the switch is abbreviated toTURNGRTHB. Therefore, the larger the size ofthe client's growth, the greater the demand forthe services of larger audit firms.

Financing: The operational variable to proxyfinancing is estimated from newly issued debtand equity ratios measured as "Long term debt+ Equity)/Total Asset" abbreviated to CHFA.Firms that change to larger audit firms arepredicted to exhibit a higher level of post-auditchanges in financing compared to ones thatchange to smaller audit firms (Johnson and Lys1990). We expect a positive correlation betweena firm's financing activity and the choice ofhigher prestige audit firms.

Profitability: The profitability of the firm ismeasured by two operational variables: averagereturns on asset (AROA) and average cash flow(ACFL). If poor returns and cash flows areexhibited prior to the event, client firms arelikely to change to smaller audit firms. Therefore,the profitability prior to the auditor changeshould be positively correlated with auditor size.

Audit Risk: The audit risk relates to theprobability of an auditor issuing unqualifiedopinion on materiality of mis-stated financialstatements. It is difficult to measure audit riskobjectively and accurately. No single proxy foraudit risk is considered adequate. However, itappears to be related to client's business risk(Simunics and Steins 1987). The business risk isproxied by two operational variables namely,client firm's size (SIZE) measured by total assetsand leverage (LEVR) both prior to and afterauditor changes. An increase in client size entailsa wider geographical dispersion and scope;therefore clients need the services of largeraudit firms that have competitive advantage overthe smaller firms. Higher leverage client firmswould pose higher levels of financial risk,therefore, it is likely that firms with higher riskwill engage the services of larger audit firms thathave greater expertise to analyse the situationresulting in greater credibility to the reports.

Market ModelThe standard Market Model (Sharpe 1964) isused to estimate the expected returns andaverage excess returns. The model expressed asfollows:

Rt = a + Pj Rmt + u t (4)

PertanikaJ. Soc. Sci. & Hum. Vol. 8 No. 2 2000

Huson Joher, M. AH, Shamsher M., Annuar M.N. &: M. Ariff

where

x 100

c - c x 100

Rt : the rate of return of the ith stock on theperiod t

Pu : stock price i at period tPU1 : stock price i at period t-1D.t : Cash dividend paid to the shareholdersa : the constant average return while market

yields zero returnsP : beta estimateD.t : Residual or random noise term assumed

to have property of Uk~ (o\a2)Rmt : the rate of returns on the market portfolio

(Composite Index) for period t Ci andCt { are the values of Composite Index atperiod t and t-1.

To estimate the parameters of the marketmodel, 60 monthly observations from outsidethe analysis period (event window) are used toavoid any misestimates of the market returnaround the event dates. Market Modelparameters are adjusted for non-synchronoustrading problem caused by thin trading usingScholes and Williams (1977) two lag and twoleads model.

Abnormal Returns Measures

Abnormal returns or residual returns areprediction errors. The abnormal returns for agiven share price at any time period is thedifference between the actual returns and theexpected returns.

u t = R - (a + (3Rrn) (5)

The average excess returns are:

AR = 1/N 1 U

N : number of sample companies across thesub-sample

AR t: average abnormal returns for companiesat period t

If ARl > 0 and statistically significant, itindicates that the market on average reactspositively to the event and thus increases the

wealth of the shareholders. To observe thecumulative effect, cumulative abnormal returns(CARs) were calculated by summing up the ARt

over various time periods of interest:

+KCAR = S ARt

-K(6)

whereCAR K t = is the cumulative abnormal returns

for cut-off point over the windowperiod from K to L,

-K,...+K refer to event window surroundingauditor changes.

Statistical TestsIndividual Coefficient EstimatesTo measure the relationship between theexogenous variables, X, and dichotomousresponse variable, individual estimate is tested.Thus this test statistics is defined as

Where the Sk is the standard error of thecoefficient and Bk is the coefficient of theindividual variable in the model.

Goodness of Fit TestIn normal regression analysis, F statistics can beused to test the joint hypothesis that allcoefficients except intercept is zero. Acorresponding test in logistic regression thatserves the same purpose is based on LikelihoodRatio. The functional form of Likelihood Ratiois as follows:

where

In I <P) is the value of the likelihood^ functionfor full (unrestricted) model and l($*) is themaximum value of the likelihood function if allcoefficients except the intercept (restricted), arezero.

The method produces a statistics that followsapproximately a Chi-square distribution with k-1(k being the number of independent variables)degree of freedom if the joint null hypothesis istrue. If the alternative hypothesis were to be

82 Pertanika J. Soc. Sci. & Hum. Vol. 8 No. 2 2000


accepted, XLR becomes larger. If null hypothesisis to be accepted, XLR < %2

RESULTSDifferences in Characteristics of Switch and Non-switch Firms

Table 1 presents the test results on thecharacteristics of client firms that switched theirauditors and those of control firms that did notswitch their auditors over a period of five years(2 years proceeding and 2 years after the auditorswitch). These are based on mean differencesrespectively for (a) size, (b) turnover growth, (c)returns on assets, (d) leverage of the firms, (e)acquisition of fixed asset to total asset and finallythe liquidity position of the two groups. A simpleparametric test was used to observe thedifferences in the firm's characteristics associatedwith switch and non-switch groups. The resultssuggest that both switch and non-switch groupsare distinctly different from one another in anumber of dimensions. For instance, the turnovergrowth of firms that switched their auditors issignificantly larger than those that did not switchauditors over the same period. The mean valuesof the turnover growth over the 5-year (twoyears prior and two after the auditor) period forthe two groups were recorded as 130 percentand 70 percent, respectively. Meanwhile, theaverage return on assets (ROA) of the two groupsover the same period is 3.4 percent for firmsthat switched their auditors and 5.1 percent fornon-switch firms, though not statisticallysignificant (t-value = -1.4). The observeddifferences on average acquisition of fixed assetsto total assets registered a marginally higher ratefor firms that switched auditors, for example,the average acquisition of the two groups was 7.5percent and 6.1 percent respectively. Thedifferences on asset sizes, leverage and liquidity

of the two groups were small and not significantat the conventional level.

Determinants of Auditor SwitchTo provide an objective framework, the variablesfor the determinants of auditor switch werederived from agency theory and others in theaccounting literature. These are turnover growth(TGROWTHB) prior to auditor switch and after(TGROWTHA), average acquisition of fixedassets to total assets (ACQ), return on assets(ROA), average earning per share (EPS), changein audit fees (AUDF), management change(MGTCHG), audit report both prior (RPORTB)and after (PRORTB) the switch, firms leverage(Leverage), liquidity of the firms (LIQ) andaudit type (AUTYPE)

Table 2 presents the results of the logisticregression model explaining the determinantsof auditor switch firms. Initially 13 variableswere analysed using maximum likelihoodestimation procedure in stepwise logisticregression based on centred data. In initial step,stepwise regression identified GROWTHB,GROWTHA, MGTCHG AND ROA as significantvariables. However, in the final step, theprocedure selected only three variables(GROWTHB, GROWTHA, and MGTCHG)which met the 0.10 and 0.05 levels of significancefor inclusion in the final model. The chi-squarevalue for overall model was 25 with 3 degrees offreedom (significant at the .0001 level). Basedon the findings in Table 2, the joint nullhypothesis (that is, all the slope coefficients aresimultaneously zero) cannot be accepted. Theresults support the notion that auditor switchdecisions of listed firms in Malaysia are mainlydetermined by management change, andturnover growth both prior and after auditorchange. The coefficient of the explanatory

Simple parametric test

Characteristic

Size (RM)Sale growthROALeverageLiquidityAvAcq

TABLEfor mean difference

Mean Switch

617890 (000)1.30.034

0.43251.82.075

1between switch and

Mean Non-switch

558508 (000),70.051

0.42211.76.06

non-switch sample

t-value

0.281.736*-1.4510.309.259.78

* Marginally significant at 10 percent level



variables are consistent with theory and findingsas reported in Burton and Robert (1967),Linbeck and Rogow (1978) and Takiah et al(1993). Burton and Robert document asignificant association between change inmanagement and replacement of new auditor.Consistent with Takiah et al (1993) in theMalaysian context, this study could not establishany significant relationship between qualifiedopinion and subsequent auditor switch. It alsoconfirms the conclusion drawn by Takiah et al(1993) that having profit or losses over the yearsdoes not necessarily influence the switch ofauditor in Malaysia.

It must be noted that though qualified auditopinion was most strongly associated with auditorchange in the US (Chow and Rice 1982),Australia (Craswell 1988) and Hong Kong (Gulet al 1991), it is not a significant determinant ofauditor change in Malaysia, Similarly, the findingscould not establish any significant relationshipbetween audit fee and change in audit firm,inconsistent with documented findings

(Eichenseher and Shields 1983); Bedingfield andLoeb 1974).

Changes in Firm's Characteristics and Choice of Audit

Firms

Table 3 summarises the descriptive statistics forfirms that switch to Tier 1 audit firms and thosethat switch to Tier 2 audit firms. The results arefor mean differences of the following variables:turnover growth, asset size, growth of asset,leverage, returns on assets, financing activitiesand average acquisition to total assets. There aresome noticeable differences. The averageturnover growth of firms that switched to Tier 1auditor are comparatively higher than firms thatswitched to Tier 2 auditor recording 54 percentand 45 percent respectively, 2 years precedingthe auditor change.

Meanwhile, the average asset growth beforethe auditor change for firms that switched toTier 1 audit firms is higher than firms thatswitched to Tier 2 audit firms, recording at 50percent and 42 percent respectively. And the

TABLE 2Regression results on determinants of auditor switch

Vars p-value Model specification Percent

MGTCHGTGROWTHBTGROWTHA

.05**.07*

.012**

Ch-SquareClassification ratePrediction RateSwitch GroupNon-Switch Group

25.00**64.00

72.0051.43

Characteristics Mean Mean

(p=.000)

significant at 5 percent level. * significant at 10 percent level.

TABLE 3Test of differences between switch to tier 1 and switch to tier 2 firms

t-value

Turnover growth beforeSize before ('000)Size afterAsset growth beforeAsset growth afterLeverage beforeLeverage afterROA beforeFinancing beforeFinancing afterAcquisiticn before

.54556246241108536

.506

.597.4436.4436.041.452.421

.07938

.45144632316577

.418.58

.3496.37.05.489

.40081.0539

.272.52*2.35*.3117.0491.7**

1.6.455-.529.3931.565

* 5 percent significant level**10 percent significant level



size of the asset for client firms that switch toTier 1 are significantly larger than firms thatswitched to Tier 2 audit firms. The averageacquisition before the auditor switch is recordedat 7,9 percent for firms switching to Tier 1 auditfirms and 5.39 percent for firms that switched toTier 2 audit firms. Furthermore, firms thatswitched to Tier 1 auditor exhibited higherleverage than those that switched to Tier 2 auditfirms, significant at 10 percent over. The returnon assets for firms switching to Tier 2 audit firmsis higher, registering 5 percent over thoseswitching to Tier 1 audit firms recording 4.1percent, but not statistically significant at theconventional level. This finding suggests somesignificant differences in the characteristics offirms that switched to Tierl and Tier 2 auditorfirms respectively.

Table 4 summarises the results of changesin firm's characteristics and choice of auditorsusing logistic regression. Initially asset size, assetgrowth, turnover growth, return on assets, changein operating cash flow, leverage, change infinancing activities and changes in acquisitionwere included in the analysis.

The stepwise procedure retained 4 variables(LEVA, CHFA, GRTH, CHACQ which arestatistical signal) in the analysis. The resultsindicate that the choice of auditor exhibits asignificant positive association with changes infinancing activities, leverage after the auditorchanges, and growth in assets before the switch,while a significant negative association is reportedfor change in acquisition. Though asset size forclient firms that switched to Tier 1 significantlydiffers from client firms that switched to Tier 2audit firm, the regression analysis fails to exhibita significant association between asset size andaudit choice. It is only significant at 21 percentlevel. The coefficients of the variables are

consistent with theory except for turnover growth.The negative coefficient for change in averageacquisition demonstrates that firms that switchto Tier 1 auditor exhibit a higher level of averageacquisition to total asset during the pre auditorchange period compared to the post period,consistent with the summary findings in Table 2.

The significant positive coefficient forleverage after choice of auditor indicates thathigher leverage firms pose a higher level offinancial risk and increases in agency cost ofdebt. To allow for this possibility, client firmswould engage the services of high quality (Tier1) audit firms, who have greater expertise toanalyse the situation and give greater credibilityto the financial reporting than a small audit firmwould. Meanwhile, evidence of a positiverelationship between changes in financingactivities and choice of Tier 1 audit firms showedthat firms switching to Tier 1 audit firms exhibita higher level of post-auditor change financingto increase the marketability of new securities(both debt and equity). Furthermore, thedocumentation of positive relationship betweenfirms' asset growth and choice of auditor suggestthat rapid growth entails substantial increases intraction volume and accounting complexity, anddecentralisation of financial controlling systemthus requiring the services of larger audit firmspresumably having the expertise to providespecialised services. The large audit firms dohave a cost competitive advantage over smalleraudit firms.

The summary results show that the averageacquisition for firms that switch to Tier 1 auditorsare relatively higher than firms that switched toTier 2 auditors, although the average acquisitiontends to decline for former group in the post-switch period. Thus the joint-hypothesis (all theslope coefficients are simultaneously zero) can

TABLE 4Result of the logistic regression analysis

Variables p-value Model specification Percent

CHACQUICHFAGRTHBLEVATURNGBSIZEB

.008

.03

.024

.048

.288

.221

Chi-SquareClassification Rate:OverallSwitch to TierlSwitch to Tier2

17.4

81.295.844.4

1 significant at 5 percent level

Pertanika J. Soc. Sci. & Hum. Vol. 8 No. 2 2000


be rejected with a chi-square value of 17.46 witha 6 degree freedom (p=. 0069). The modelcorrectly classifies for 81.2 percent. Earlierstudies on auditor choice have documentedinconsistent results on the association betweenclients' characteristics and direction of auditorchange. The findings of this study are moreconsistent with the hypothesis that firms that areexpected to raise debt financing demand theservices of high quality auditors to monitormanagement activities that are detrimental tothe bondholders. The leverage was hypothesisedto be positively associated with the choice ofTier 1 by Palmrose (1984), Eichenseher andShield (1986). The findings of a positivecoefficient for the change in financing activitiesafter the auditor change indicates that firmswhich are expecting to issue securities in thenear future demand the services of Tier 1auditors to attest credibility to the financialreporting to market participants. This isconsistent with the findings of Carpenter andStrawser (1971). They asserted that firms maychange auditors especially from a Tier 2 to Tier1 auditor to increase the marketability of thenew securities (debt and equity issue). Consistentwith the study of Johnson and Lys (1990), thisstudy also documents asset growth before andafter auditor change, change in financingactivities and change in acquisition as the majordeterminants of choice of auditors. However,contrary to Johnson and Lys (1990), this studydocumented a negative association betweenchange in acquisition and choice of auditor.The finding of negative coefficient indicates thatpre-switch acquisition for clients firms that switchto Tier 1 audit firms is comparatively higherthan clients firms that change to Tier 2 auditfirms.

The Wealth Effect of Auditor Sioitch Decision

Table-5 summarises the average abnormal returns(ARs) and cumulative abnormal returns (CARs)around the announcement day over a window of81 days. Average daily excess returns andcumulative abnormal returns were examined forstatistical significance using standard testprocedure.1 Findings indicate that auditorchange on average are not associated with

significant price adjustments in Malaysia. Averageabnormal returns on the day of announcementitself and the 3-day (-1 to +1) excess returns are0.092 percent and 0.0461 percent respectively.These are not statistically significant. Thecumulative abnormal returns over the days (-60to -8) and (-8 to -1) are 0.018 and 0.62 percentrespectively. Post-announcement CAR over thedays (1 to 8) and (8 to 20) are 0.43 percent and0.25 percent respectively. However, none arestatistically significant at the conventional level.

The client firms that switched to higher(lower) quality audit firms experienced positiveexcess returns at day zero of 0.12 percent and0.69 percent respectively. The 3-day (-1 to +1)excess returns for firms that switched to lowerquality audit firms recorded -0.29 percent.However these are not significant at theconventional level. The pre-announcement CARfor client firms that switched to higher qualityaudit firms over the days (-8 to -1) recorded anet gain of 2.25 percent with a t-value of 1.84.However, the CAR at post announcement periodover the days (1 to 8) and (8 to 20) declined,recording cumulative abnormal returns ofpercent and -2.00 percent respectively. But noneare statistically significant.

Market on average reacted negatively toclient firms that switched to a lower qualityauditor. The CAR over the day (-60 to -8)recorded a cumulative 0.12 percent, which isnot statistically significant. However, pre-announcement CAR over the days (-8 to -1)recorded a net loss of 4.56 percent, which ismarginally significant at 10 percent level. CARin post-announcement period over the days (1to 7) and (8 to 20) recorded a net gain of 1.68percent and 0.13 percent respectively. However,these are not statistically significant at theconventional level.

The revaluation of auditor change typewithin classes is more ambiguous and there is noclear-cut direction of price changes. However,overall it appears to suggest a common stockprice decline surrounding the auditor change.The average abnormal returns on the day ofannouncement and three days (-1 through +1)excess returns for client firms that switched fromhigher prestige to higher prestige audit firms

1. t-AR = ARt-0/SE(AR), t-CAR=CARia/SE (CAR^), where SE(AR) = standard error of AR and SE(CAR)CAR and (K.L) = cut-off point from K to L during window period

standard error of



TABLE 5Market reaction to auditor switch announcements

Trading day

-10-9-8-7-6-5A-3-2-1012345678910CAR(-60 to -8)(-8 to -1)(1 to 7)(8 to 20)

Full sampleAR

0.00146-0.001470.00019-0.003920.002100.0054*-0.00073-0.000190.00390-0.000650.002280.00095-0.00042-0.00087-0.002130.003400.00503-0.00119-0.00040-0.003020.00038

0.000180.006180.00430.0025

Tier 2-Tier 1AR

-0.00296-0.000530.00311-0.002470.007810.00686-0.000450.01151-0.00093-0.004150.001160.000160.00022-0.00484-0.003260.004740.00514-0.00329-0.00611-0.00306-0.00347

.0595*

.0225*0.00002-0.0288

Teirl-Tier 2AR

-0.00294-0.00812-0.014**-0.01710-0.003650.01126-0.019**-0.024140.021**0.000050.00690-0.00431-0.006000.000780.00794-0.005090.022**0.00202-0.00072-0.005360.00575

0.00117-0.04568*

0.01680.00132

Tier 1-Tier 1AR

-0.00275-0.00126-0.00034-0.001820.00015-0.00173-0.003800.001860.00034-0.00291-0.001610.001110.001800.00133-0.005970.005480.00403-0.000010.00820*-0.005560.00371

-0.031-0.00980,00780.0168

Tier 2-Teir 2AR

0.00392-0.00344-0.013630.007130.01354-0.004360.02316-0.00629-0.004800.001320.00014-0.00295-0.016570.0128*-0.00999-0.021160.013350.00385-0.006210.002270.00479

-0.03580.00462-0.02062-0.04628

* significant at 10 percent level** significant at 5 percent level

recorded at -0.161 percent and -.34 percentrespectively. These are small and insignificant.The CAR over the days (-60 to -8) and (-8 to -1) recorded a loss of 3.1 percent and .98 percentrespectively, which are not statistically significant.The revaluation effect of auditor change fromTier 2 to Tier 2 reported a weak negative marketreaction. Though significant positive and negativeabnormal returns were reported, none of theday zero and three-day (-1 to+1) excess returnswere significant, recording at 0.014 percent and-0.034 percent respectively. The CAR over theday (-60 to -8) and (-8 to -1) registered acumulative return of -3.5 percent and 0.46percent respectively for the Tier 2 to Tier 2switch sample. CAR during the postannouncement recorded over the (1 to 8) and(8 to 20) were -2.06 percent and -2.5 percentrespectively. However these findings are notstatistically significant.

To substantiate existing literature, furtheranalysis was done to determine whether firmsbelonging to different levels of financialcondition, and switched audit firms resulted indifferent market reaction. The financially healthyfirm that switched audit firms resulted in apositive market reaction while financiallyunhealthy firms that switched audit firms resultedin a significant negative reaction surroundingthe auditor changes. For financially healthy firms,the ARs for the day of announcement and 3-day(-1 to +1) excess recorded at -0.5 percent and -0.03percent respectively. Pre-announcementCAR over the days (-60 to 8) and (-8 to -1)registered a net gain of 0.63 percent and 1.53percent respectively, but these are statisticallyinsignificant. Post-announcement CAR over thedays (1 to 8) and (8 to 20) were at 1.09 percentand -1.15 percent respectively. While CAR forfinancially unhealthy firms that switched auditors

PertanikaJ. Soc. Sci. & Hum. Vol. 8 No. 2 2000

Huson Joher, M. AH, Shamsher M., Annuar M.N. & M. Ariff

over the (-60 to -8) and (-8 to -1) recorded anet loss of 15,8 percent and 1.4 percentrespectively. These are not statistically significantat 10 percent level.

The revaluation effect of auditor change forclient firms that received a clean opinionreported a weak positive market reactionsurrounding the auditor change. The ARs forannouncement day and 3 days (-1 to +1) were at0.16 percent and .009 percent respectively. Thepre-switch CAR over the days (-60 to -8) and (-8 to 1) are recorded as net gains of 1.5 percentand 0.25 percent respectively. But none arestatistically significant. The post-switch CAR overthe interval (1 to 20) reported a net gain of 0.11percent. This is apparently consistent with Teoh's(1992) contention that firms will experience apositive reaction after a clean opinion thanqualified opinion, because high value retentionis more common after clean than qualifiedopinion. But none are statistically significant.

Judging from the market reaction to auditorchanges, there is weak evidence that the marketindeed perceives auditor change as a signal.Thus, auditor switch in this emerging capitalmarket conveys information value associated withauditor change, but due to unknown reasons,are not producing the significant effect normallyreported in some developed markets. Thedemonstration of weak positive market reactionreflects that an increase in firm value appears tooccur, and it is not a negative market reactiondocumented in earlier literature from thedeveloped markets. Observing significantcumulative abnormal returns for client firmsthat switch to Tier 1 audit firms prior to auditorchange reflects a confirmation of quality shiftalso observed in other markets.

CONCLUSIONThe issue of auditor has been of interest toacademics, researchers and industry experts dueto its strategic implication for firm value,credibility of financial reporting and monitoringcosts to curtail agency costs. Despite the concernsshown in developed economies, little attemptappears to have been made in Malaysia toexamine such an important issue in this fastgrowing economy. Thus, this paper is a modestfirst attempt that ascertains the determinants ofauditor switch decision and its effect on sharevaluation of firms listed on the Kuala Lumpur

Stock Exchange. Logistic regression and eventstudy methods were used to analyse the data.

In general, findings appear to suggest thatauditor switch in Malaysia is determined bychanges in management and higher turnovergrowth. Changes in firms* characteristics such asasset growth prior to auditor switch, changes inaverage acquisition of fixed assets to total assets,firm's leverage, and changes in financing activitieswere found to be significantly associated withchoice of quality differentiated audit firms.

Auditor change in general is not associatedwith any significant price adjustment coincidingwith the announcement of auditor switch, despitea positive trend in upvaluation of such firms.However, once portfolios were formed based onthe auditor change types, different resultsemerged. Firms that switched to higher qualityaudit firms experienced positive (though weak)market response while negative reaction isobserved for firms that switched to lower qualityaudit firms. The revaluation effect from shiftswithin classes exhibits weak negative abnormalreturns. An interesting difference in the findingsof this study and those of similar studies reportedin developed economies is that there is a weakpositive abnormal market reaction anomalous tothose reported in the developed economies.This could be due to the positive developmentat firm's level and significant upsurge in theMalaysian economy, which had registered averageGDP growth of 8-9 percent over the test period.Alternatively, there are some still unknownmissing variables confounding the results.

REFERENCESBEDING FIELD, J.P and S. E. LEOB. 1974. Auditor

change: an examination. Journal of AccountancyMarch: 66-69

BURTON, J. and W. ROBERT. 1967. A study ofauditor changes. Journal of Accountancy: 31-36

CARPENTER, C.G and R.H STRAWSER. 1971.Displacement of auditor when the clients gopublic. Journal of Accountancy June: 55-58

CHOW T. and S. RICE. 1982. Qualified audit opinionand auditor changes. The Accounting ReviewApril: 326-335.

CITRON, D. B and R.J . TAFFLER. 1992. The auditorreport under going concern uncertainty: anempirical analysis. Accounting and BusinessResearch Autumn: 337-351.



CRASWELL, A.T. 1988. The association betweenqualified audit opinion and auditor switches.The Accounting Review April: 326-335.

DAVIDSON, W.N. and D.C. GRIBBIN. 1995. The broadof director's and audit opinion's on the stockmarket reaction to announced auditor change:an agency cost perspective. Working Paper,Southern Illinois University.

DE ANGELO, L.E. 1981a. Auditor independence,low bailing, and disclosure regulation. Journalof Accounting and Economics 3:113-127.

DE ANGELO, L.E. 1981b. Auditor size and auditorquality. Journal of Accounting and EconomicsDecember: 183-199.

DHILAWL, D. S and J. SCHWARTZBERG. 1993. Ananalysis of the economic factors related toauditor-client disagreements preceding theauditor changes. Auditing: Journal of Practiceand Theory 12 (2): 22-47.

DYE, R.A. 1991. Informational motivated auditorreplacement. Working Paper, University ofWashington.

DUPUCH, N. and D. SIMUNIC. 1982. Competition inauditing: an assessment, pp:401-450. Symposiumin Auditing Research IV, University of Illinois.

EICHENSEHER, J.W, M. HAGIGI and D. SHIELD. 1989.Market reaction to auditor changes by OTCcompanies. Auditing: A Journal of Practice andTheory Fall: 29-40.

EICHENSEHER, J. and D. SHIELD. 1986. The correlatesof CPA-firms change for publicly heldcorporation. Auditing Journal of Practice andTheory Spring: 23-32.

FAMA, E. F. and M.C.JENSEN. 1983a. Separation ofownership and control. Journal of Law andEconomics 26:301-325.

FAMA, E. F. and M.C.JENSEN. 1983b. Agency problemsand residual claims. Journal of Law and EconomicsJune: 327-349.

ETTREDGE, M. and R. GREENBURG. 1990, Determinantof fee cutting on initial audit engagement.Journal of Accounting Research Spring: 198-210.

FRANCIS J.R. and E.R. WILSON. 1988. Auditorchange: a joint test of theories relating toagency cost and auditor differentiation.Accounting Review October: 663-683.

FRIED D. and A. SCHIFF. 1981. Audit switches andassociated market reaction. The AccountingReview April: 324-341.

GUL, F. A., D. LEE M. and M. LYNN. 1991. Auditqualification and switches: the Hong Kongevidence. Unpublished manuscript. TheChinese University of Hong Kong.

HEALY, P. and T.LYS. 1986. Auditor changesfollowing big eight takeover of non-big-eightaudit firms. Journal of Accounting and PublicPolicy Winter: 251-265.

HOLTHAUSEN, R.W. and R.E. VERRECCHIA. 1990. Theeffect of informedness and consensus on priceand volume behaviour. The Accounting Review:191-208

JENSEN, M. and MECKLING. 1976. Theory of thefirms: managerial behaviour, agency cost, andownership structure. Journal of FinancialEconomics October: 305-360.

JOHNSON, W,B. and T. Lvs. 1986. The market foraudit service: evidence from voluntary auditservice. Working Paper, NorthwesternUniversity.

JOHNSON, W.B. and T. LYS. 1990. Market for auditservice: evidence from voluntary auditorchanges. Journal of Accounting and EconomicsJanuary: 281-308

JOHN, D., H. DAVID and M. ANDREW. 1999. Marketreaction to auditor resignation. Journal ofAccounting and Business Research 29:95-108

LINGBECK, R. and R. Rocow. 1978. An analysis ofauditor change. Paper presented at theSoutheast Regional Meeting of the AmericanAccounting Association.

LURIE, A. G. 1977. Selecting your auditor. FinancialExecutive July: 50-58.

NICHOLS D.R. and D.B. SMITH. 1983. Auditorcredibility and auditor changes. Journal ofAccounting Research Autumn: 534-544

PALMROSE, Z. 1984. Demand for qualitydifferentiated audit service in an agency costsetting: an empirical investigation. Paperpresented at Sixth University Illinois AuditingResearch Symposium.

SHARPE W. F. 1964. Capital asset pricing: a theoryof market equilibrium under condition of risk.Journal of Finance 19: 425-442.

SCHOLES, M. and J. WILLIAMS. 1977. Estimating betafrom non-synchronous data. Journal of FinancialEconomics 5:309-328.



SCHWARTZ, K,B and K. MENON. 1985, Auditor switchby failing firms. The Accounting Review April:248-261.

SIMUNIC, D.A and M. STEIN. 1987. Productdifferentiation in auditing: a study of auditorchoice in the market of unseasoned new issue.Canadian Certified General AccountantsResearch Foundation, 1987.

SMITH, D. B. and D. R. NICHOL. 1982. A market testof investor reaction to clients auditorsdisagreement. Journal of Accounting andEconomics October: 109-120.

TAKIA M.I, M. SAVED and W.S. GAZALI. 1993. Incidenceof qualified opinions and the effect on auditorswitching: an empirical study in Malaysia. JurnalPengurusan December: 53-63.

TEOH, S. 1992. Auditor independence, dismissalthreats and the market reaction to auditorswitch. Journal of Accounting Research 30:1-31

WILLIAMS, D. D, 1988. The potential determinantsof auditor change. Journal of Business FinancecV Accounting XV: 243-261.

ZURADA, J. M., F. BENJAMIN, T.J. WARD and R.M.BAKER. 1998. Neural networks versus logitregression models for predicting financialdistress response variables. Journal of AppliedBusiness Research 15(l):21-30.

WATT, R.L. and ZIMMERMAN. 1978. Positiveaccounting theory: a ten years perspective. TheAccounting Review 65:131-156.

Pertanika J. Soc. Sci. & Hum. Vol. 8 No. 2 2000

auditor switch decision of malaysian listed firms: tests ... papers/jssh vol. 8 (2) sep....

Documents