DOES THE ROTATION OF HEAD OF AUDIT OFFICE ACROSS OFFICES IMPACT AUDIT QUALITY? STUDY CASE OF THE AUDIT BOARD OF REPUBLIC OF INDONESIA

ABSTRAK


APAKAH ROTASI PIMPINAN ANTAR KANTOR AUDIT MEMENGARUHI KUALITAS AUDIT? STUDI KASUS DI BADAN PEMERIKSA KEUANGAN REPUBLIK INDONESIA
This study examines the effect of head of audit office rotation on auditing quality.This study is motivated by two propositions in private sector auditing.First, that audit partners from larger audit offices are argued to be more experienced compared with audit partners from smaller offices.Second, the audit partners from larger audit offices also argued to be more capable in handling the rotation challenges.This study aimed to examine these propositions in public sector setting by using the Supreme Audit Board of the Republic of Indonesia (BPK) as a case study, where the head of representative office as the audit partner equivalent.This study tested these propositions by examining audit quality across BPK's representative audit office sizes and the effect of rotation and types of rotation of the head of the representative audit office on audit quality.This study used the number of audit findings in the financial audit reports as the proxy of audit quality.This study found that audit quality varies across office sizes in the BPK, where the larger representative offices tend to have a better audit quality.This study also found that the rotation of the head of the audit office is negatively associated with audit quality.Lastly, this study found that the negative effect of rotation is caused by the negative effect of changes in the office sizes, which outweigh the positive effect of rotation.

INTRODUCTION
This study examines the effect of head of audit office rotation on auditing quality.This study is motivated by two existing auditing topics: audit office size effects and mandatory audit partner rotation.First, in the private sector, an audit partner from a larger audit office size is argued to be more experienced compared with an audit partner from a smaller office (Francis, Michas, & Yu 2013).Therefore, larger audit offices have a higher audit quality than their smaller counterparts (e.g., Choi et al. 2010;Francis & Yu 2009;Francis, Michas, & Yu 2013;Sundgren & Svanstrom 2013).Second, auditor rotation is argued to be a useful way to improve auditor independence, which is expected to improve audit quality (Bamber & Bamber 2009;Doti 2011).
The existing auditor rotation literature shows mixed results.Several prior studies found that mandatory audit partner rotation impaired audit quality in the initial years of rotation because of a deficiency in client knowledge (e.g., Cameran, Francis, & Marra 2014;Litt et al. 2014;).The negative effect of auditor rotation has also been found within public sector auditing (e.g., Cagle & Pridgen 2015;Schelker 2008).Conversely, other studies suggest that audit partner rotation improves audit quality through a positive peer review and a 'fresh eyes' perspective on the audit (e.g., Hamilton et al. 2005 In each office there will be one auditor who is assigned as the head of the representative audit office (HRO).The HRO's responsibilities are similar to those of signing or reviewing partners in private sector auditing.In the several audit offices HROs responsibilities are not only as the reviewing partner but also the signing partner.In the HRO rotation context, Francis, Michas and Yu's (2013) finding implies that HROs from smaller office sizes have less experience compared with HROs from larger office sizes, and not only have less audit experience but also an inferior understanding and knowledge of the new client in the larger office.Therefore, HRO rotation from smaller to larger office sizes might result in a lower audit quality in the new office.From the opposite view, HRO rotation from smaller to larger office sizes could indicate promotion, with the opposite rotation pattern implying demotion.In the private sector, these methods are commonly used by the companies to place their employees into the most suitable job based on their past performance (Campbell, 2008).This implies that the promoted HRO enjoyed exceptional performance in the past and is expected to perform in a similar manner in the larger and more complex office.In contrast, the demoted HRO may indicate deficient performance in the previous office, and thus may be expected to perform better in the smaller and less complex office.
Although the promoted HROs may demonstrate outstanding past performance, they have less audit experience due to fewer audit assignments in the previous office.This study argues that a greater number of clients and the presence of more complex problems in the larger office will have a negative effect on the promoted HRO's audit performance.Conversely, the demoted HRO may demonstrate an inferior past performance, but they possess more audit experience, moving from a larger office.This study contends that moving to fewer clients with less complex problems will have a positive effect on the demoted HRO's audit performance.This lead to the second and third hypotheses in this study, which are: (2) HRO rotation is negatively associated with audit quality; (3a) HRO rotation from a larger to smaller office is positively associated with audit quality; and (3b) HRO rotation from a smaller to larger office is negatively associated with audit quality.
This study followed Lennox, Wu and Zhang's (2014) approach in measuring audit quality, and used the changes in the number of audit findings in audit reports as the proxy of audit quality.Audit findings occur when the auditor detects and reports incompatibility between accounting regulations and the client's accounting practices.The auditor's ability to detect and disclose such discrepancies in audit reporting reflects DeAngelo's (1981) concept of audit quality: the joint probability of detecting and disclosing a client's deviations from accounting systems.The number of audit findings in this study consists of the number of audit findings in the audit report of internal control system and the audit report of compliance to standards.
This paper is organised as follows.The next section addresses the research methods, the third section presents the empirical results and discussion.Last section discusses the research conclusion and implications for policy-makers.

Research Samples
The sample in this study comprises data on HRO rotation and local government financial audit reports.The local government audit reports consist of the audit reports for province, district and municipality.These data are drawn from the BPK RI database for the period 2011−2014 11 .This study used one full sample and is summarised in Table 1.
Data in this study were manually identified and hand-collected from each local government audit report and HRO rotation data.This study also compared each audit report date with HRO data rotation to differentiate audit reports produced by previous and new HROs.Further explanation of each variable that collected from each financial audit report will be presented in section 3.

Research Design
This study used three linear models to test the hypotheses.Table 2 presents the summary of variables definition and expected sign.

The office size and rotation effect on audit quality
The purpose of Model 1 is to test the first and second hypotheses.This model simultaneously 1 The audit report for a given fiscal year is published in the following year.
examines the effect of office size and rotation.This model does not distinguish between different types of rotation (i.e., promotion, demotion or position).This model only classifies the sample into rotation and nonrotation.This regression model combines and modifies the models in Francis, Michas and Yu (2013) and Lennox, Wu and Zhang (2014).
DeltaAF, the dependent variable, is the measurement of audit quality.In this study audit quality is measured by changes in audit findings for audit reports i in year t.OfficeSize is the size of the BPK RI representative office that conducts the financial audit for audit report i at year t.Rotation is an indicator to distinguish the audit report i in year t, produced by the new HRO.Gen, YoS and ClientSize are control variables.Each variable in this model is explained as follows.
This study follows Cagle and Pridgen (2015) in using the changes in the number of audit findings as a proxy for audit quality.The changes in the number of audit finding, hereafter DeltaAF, is measured as the differences in the number of audit findings at year t and the number of audit findings at year t−1.Audit findings are the number of findings that is reported by the auditors in their audit reports, which are report of internal control system and compliance to standards.These findings occur when the auditor detects incompatibilities between accounting regulations and the client's accounting practice.Where: Pro is coded 1 if the audit report i in year t is produced by the promoted HRO, or an auditor who experiences vertical promotion to become an HRO.Pos is coded 1 if the audit report i in year t is produced by an HRO who is rotated to the same size office.Dem is coded 1 if the audit report i in year t is produced by a demoted HRO.This study expects the Pro sign to be negative and Dem to be positive.Non-rotation samples are captured by the constant.In this model, OfficeSize1, OfficeSize2, Gen, YoS and ClientSize are the same variables as in Model 1.
Model 2a uses OfficeSize1 and interaction between OfficeSize1 with Pro.Model 2b uses OfficeSize2 and interaction between OfficeSize2 with Pro.This model includes the interactions between promotion and demotion with office size.The Pro represents the HRO who is rotated to the larger office.Pro captures two effects.First, Pro indicates that rotation occurred in that particular year.Second, it indicates that rotation is to a larger office.Hence, there are two effects in Pro: a rotation effect and a larger office size effect.Therefore, the interaction between promotion and office size controls the effect of office size on promotion.Conversely, the Dem captures the effect of rotation and smaller office size.The effect of office size and the effect of rotation types may be contradictory.The effect of office size reduces the effect of rotation, or vice versa.There is also the possibility that the effect of office size and the effect of rotation are in the same direction, with the effect of office size compounding the effect of rotation.Therefore, this study does not state specific expectations of sign for these two interactions.

RESEARCH RESULT
Table 4 presents the regression results of Model 1.The regression results show that Office-Size1 is positive and significant at ρ<0.05, and OfficeSize2 is positive and significant at ρ<0.1.Rotation is negative and strongly significant at ρ<0.01.Gen is negative and significant at ρ<0.1.YoS and ClientSize are insignificant.The R-squares for both models are similar.
OfficeSize1 and OfficeSize2 are the test variables for the first hypothesis.The positive sign of OfficeSize1 and OfficeSize2 suggest that the larger the audit office size, the greater the number of audit findings.Since the proxy of audit quality in this study is the number of audit findings, this result suggests that larger offices will have a higher audit quality compared with smaller offices.This finding indicates that audit quality varies across office sizes, with a tendency for larger offices to have a higher quality.This result supports the first hypothesis, and is consistent with prior study results that found that larger offices had a better audit quality compared with smaller offices (e.g., Choi et  Dem*OfficeSize Interaction between demotion and office sizes ?
Table 3.The Summary of Variables Definition and Sign Expectation caused by the categorisation of office sizes.Therefore, the discrepancy of audit quality between office size categories is wider.
Rotation is negative and significant at ρ<0.01.This suggests that the number of audit findings will decrease in the year of rotation.This result corroborates prior studies that found that auditor rotation impaired audit quality due to the auditor's inferior knowledge of new clients (e.g., Cameran, Francis, & Marra 2014;Litt et al. 2014).The negative coefficient suggests that the new rotated HROs are more sceptical and need time to learn and become familiar with the new client's characteristics and risks before they can perform high quality audits.
Among control variables, the Gen variable shows consistently negative sign under the two regressions.However, the sign of Gen is not as (1) The dependent variable used in all models is DeltaAF, which is the change in audit findings.OfficeSize1 is the size type category of the BPK RI representative offices based on Decree 45/2011.OfficeSize2 is the actual number of audit clients in each audit office.Rotation is an indicator variable that is coded 1 if the audit report is produced by the new HRO.Gen is the HRO's gender.YoS reflects the HRO's experience.ClientSize is measured by the logarithm natural of client total assets.2014).Intuitively, local governments with larger government expenditures have more diverse and complex problems.Therefore, the HROs in Type A offices are more likely to encounter more difficult audit problems than HROs in Types B and C offices.These audit problems shape and enhance the HRO's audit performance, which results in a more experienced HRO.The coefficient of OfficeSize1 is greater than OfficeSize2.This indicates that the difference of audit quality across offices is more pronounced in OfficeSize1.This may be variables, Gen is significant and negative at ρ<0.1 in Model 2b.Dem, Pos, YoS and ClientSize are insignificant.Regression under decree benchmark has a larger R-square than regression under actual number of client benchmark.
The Model 2a results show that Pro is positive and significant at ρ<0.01.Since DeltaAF is the audit quality proxy in this study, this association suggests that promotion improves the quality of audit.This result is consistent with the proponents of auditor rotation (e.g., Hamilton et al. 2005;Lennox, Wu, & Zhang 2014).In contrast, the interaction of Pro*OfficeSize1 is negative and significant at ρ<0.01.In Model 2 there are two interactions of Pro*OfficeSize1.This is because OfficeSize1 is a categorical variable.In the results, Type C offices (small offices) are used as a base because the effect of office size is likely to be stronger in the medium (Type B) and large (Type A) offices.The Pro*OfficeSize1TypeB coefficient is −6.373, significant at ρ<0.01.The Pro*OfficeSize1TypeA coefficient is −8.184, significant at ρ<0.01.The interaction coefficient of Pro*OfficeSizeTypeA is larger than Pro*OfficeSizeTypeB.This result suggests that, when an HRO is promoted to a larger office, the larger the new office the greater the incremental negative effect.Hence, the larger the office size, the more effort required from an HRO to catch up his or her discrepancy of audit experience and inferior understanding of the new clients.This result is consistent with Bills, Swanquist and Whited (2016), who suggest that an increase in the number of audit assignments impairs audit quality.
The net effect of Pro is negative.This is because the coefficients of the interaction between Pro*OfficeSize1TypeA and Pro*OfficeSize1TypeB are larger than the positive coefficient of Pro.The net effect for promotion to a Type B office is −0.918 33 , and the net effect for promotion to a Type A office
predicted.The negative sign suggests that the female HRO produces fewer audit findings than male HROs.Although contrary to predictions, a plausible explanation for this negative relationship is the risk-averse characteristic of female HROs.The female HROs may produce fewer audit findings because they focus on the substance of the finding.Their findings may be more significant in terms of materiality; that is, their findings may be more substantial in content than in the overall number of findings.Therefore, this result should be interpreted cautiously.
Table 4 presents the regression results of Model 2. This model examines Hypotheses 3 a and 3 b .Model 2 classifies Rotation into three different indicator variables: promotion (Pro), demotion (Dem) and position (Pos).In the regression, this study also interacts with these rotation types with OfficeSize1 and OfficeSize2.The purpose of this interaction is to examine the extent of the office size effects.The new office size may have negative or positive effects on the new HRO's audit performance.
Model 2a uses OfficeSize1.Model 2b uses OfficeSize2.The dependent variable used in all models is DeltaAF.DeltaAF is the change in audit findings.OfficeSize1 is the size type category of the IAO representative offices based on decree number 24 year 2011 (BPK RI, 2011).Type C office is the base level in this regression.OfficeSize2 is the actual number of audit clients in each audit office.Rotation is an indicator variable that is coded 1 if the audit report is produced by the new HRO.Pro is promoted HROs.Pos is HROs rotated to the same size office.Dem is demoted HROs.Gen is the HRO's gender.YoS reflects the HRO's experience.ClientSize is measured by the logarithm natural of client total assets.The results show that Pro is positive and significant at ρ<0.01 in Model 2a and significant at ρ<0.05 in Model 2b.Pro*OfficeSize1 and Pro*OfficeSize2 interactions are negative and significant at ρ<0.01.Among control diminished by the size of the new office.The larger the new office is, the more severe its detrimental effect.This result gives two insights that relevant to private sector auditing.First, lower audit quality in the initial year may be caused by new clients that are larger and more complex than the audit partner's previous clients, similar to the promoted HRO in this study.Experience with past clients is not taken into account by prior studies when examining the effect of auditor rotation.Second, the audit partners' new clients may be in a different industry to their previous clients, similar to a vertical promoted HRO in this study.The audit partners' preferences to learn a new industry instead of rotating to other offices have been documented by Daugherty et al. (2012).Therefore, these audit partners also experience the negative effect of auditor rotation.

CONCLUSION
This study addresses the following research questions: (1) Does public sector auditing experience vary audit quality across office sizes, as in private sector auditing?; (2) What is the effect of rotation and type of rotations (promotion and demotion) on audit quality?
This study investigated these issues in the context of the BPK RI.First, this study tested whether the larger representative offices in is −2.279 4 .The net effect shows that promotion to a Type A office has a larger detrimental effect on audit quality than promotion to a Type B office.The Pro net effect supports H 3b .
The Model 2b result shows that Pro is positive and significant at ρ<0.05.This result is similar to Pro in Model 2a.The interaction of Pro*OfficeSize2 is negative and significant at ρ<0.01.In contrast to Model 2a, the negative coefficient of Pro*OfficeSize2 is smaller than the positive coefficient of Pro.This result in the positive net effect of Pro for 1.700. 5This suggests that the positive effect of rotation is larger than negative effect of office sizes.This result does not support H 2b .
Although the final effect of Pro is different under the two models, the interaction shows that the new office size does affect the HRO's audit performance.This study argues that the difference in the final effect for Pro is caused by the noisy measurement of OfficeSize1.Therefore, Pro*OfficeSize2 is m 4 ore reflective of the rea 5 l condition.The interactions between promotion and office size results support Bills, Swanquist and Whited's (2016) proposition that increasing workload has detrimental effects on audit quality.In this study, the increase in workload is through an increase in the number of audit clients, which is experienced by the promoted HRO.
These interactions result may also provide an explanation for the mixed results of the effect of auditor rotation on audit quality in the private sector.Prior studies on audit partner rotation have focused on the rotation effect on the firm's financial reporting quality, and did not take into consideration the past experience of the audit partners.In contrast, this study considered the past experience of the HRO and took into account the next audit engagement (i.e., the new office).As mentioned above, promotion has a positive effect, but this is

5
(1.840 + (−0.140)) = 1.700.across offices size cannot be circumvented; however, the detrimental effects of such rotation can be minimised.This study shows that the larger the new office size is, the more severe the effect will be.The BPK RI can apply the rotation across office sizes in consideration of the difference in the size between the previous and new office.The smaller the difference, the less negative the effect is.In this way, the positive effect of rotation will outweigh the negative effect of office size.This will eventually give a greater benefit for the BPK RI itself.
Despite this study's findings, this study acknowledges its research limitations.This study used changes in the number of audit findings as a proxy for audit quality.This study acknowledges that this proxy is noisy.It is possible that an auditor found fewer audit findings because of the client's strong accounting practices, instead of the auditor's inferior competency.It is also possible that the number of findings may decrease, but the significance and materiality of findings may increase.In this scenario, the audit report will have fewer audit findings.This study interpreted this scenario as a lower audit quality, while it is not.Therefore, future studies may use a better audit quality proxy to avoid this problem.The future study may use changes in the number of significant findings instead of changes in the number of audit findings.
Furthermore, the results of this study suggest future research opportunities.Future research can examine other audit quality proxies that are less noisy.The research findings of the present study can also be tested in the private sector by examining audit partner rotation, taking into account the audit partner's previous and new clients.

Acknowledgement
I am very grateful for the guidance and valuable advices from my supervisors Dr Greg Shailer the BPK RI have a higher audit quality than the smaller representative offices.Second, this study examined the effect of rotation on audit quality.Third, this study classified the rotation variable into three different indicator variables (promotion, demotion and position) and examined the effect of these variables on audit quality.
Using changes in the number of audit findings as the proxy for audit quality, this study found that audit quality varies across office sizes in the BPK RI.The larger audit offices have relatively higher audit quality in comparison with smaller offices.This study also found that the HRO rotation had a negative and significant association with audit quality.Further, this study found mixed results regarding the effect of promotion on audit quality.This mixed result is caused by the measurement of the office size.Measuring the size of audit offices based on the BPK RI decree, this study found that promotion has a positive and significant effect on audit quality; however, the negative effect of office size exceeds the positive effect of promotion.Therefore, the net effect of the promotion on audit quality is negative and significant.In contrast, measuring office size based on its actual number of clients showed that the office size effect is smaller than the positive effect of promotion.Therefore, based on this measurement, promotion has a positive and significant effect on audit quality.
Despite the result of the final effect, this study found that there is a trade-off effect between the effect of rotation and office size.This result suggests that the size of the office does matter when the rotation is across different office sizes.This result can be extended as an additional explanation regarding audit partner rotation research in the private sector.The mixed results of audit partner rotation research in the private sector might be influenced by the size of previous and new clients.

Table 1 .
Francis, Michas, & Yu 2013;Gul, Wu, & Yang 2013)se more audit findings, than their male counterparts.Consistent with prior studies, this study expects that female HROs are positively related with DeltaAF.Gen is coded 1 if the audit reports i in year t is produced by the HRO who is female, and 0 otherwise.This study follows prior studies in using experience and ClientSize as control variables (e.g.,Francis, Michas, & Yu 2013;Gul, Wu, & Yang 2013).This study uses year of experience or YoS as a proxy of the HRO experience.It is expected that the more experienced HRO will produce more audit findings.Therefore, YoS is expected to have a positive sign.ClientSize is measured by the logarithm natural of client total assets.The purpose of this variable is to control the possibility of larger clients experiencing more accounting problems that will lead to more audit findings.In this model, this study also controls for the fixed effect of entity type (provincial, city or district).This study will use the same control variables and fixed effect indicator for the other two models.This study defines promotion, demotion and position based on the BPK RI decree number 13 year 2014 about job level and remuneration(BPK RI, 2014).The promoted HROs are those who experience an increase in the job level (i.e., promoted to a higher job level).The demoted HROs are those who experience a decrease in the job level.The positioned HROs are those who are rotated to a different office but remain at the same job level.The promotion, demotion and position determined using these methods are denoted as Pro, Dem and Pos, respectively.
2This study excluded one audit office for 4 year observations due to an anomaly.This audit office only had one client, but is categorized as a Type A office because of the size of its client.Consistent with H 1 , OfficeSize1 and OfficeSize2 are expected to have a positive sign.The positive sign implies that the HRO from larger representative offices have a superior ability to find incompatibilities between accounting regulations and a client's accounting practices compared with smaller representative offices.This study follows prior studies (e.g.., Cameran, Francis, & Marra 2014; Litt et al. 2014) in using an indicator variable to distinguish the rotation effect on audit quality.Rotation, in this study, is an indicator variable that is coded 1 if the audit report is produced by the new HRO, and 0 otherwise.The new HRO needs time to adjust and become familiar with clients in the new office.Therefore, it is expected that the new HRO will produce fewer audit findings than the existing HRO.Consistent with H 2 , this study expects that Rotation has a negative sign.3b .In this model, Rotation from Model 1 is classified into three indicator variables: promotion, demotion and position.

Table 4 .
The Office Size and Rotation Effect on Audit Quality Table 5 provide mixed results in terms of support for the third hypothesis.The results under the two models do not support H 3a .In contrast, this study finds mixed results regarding hypothesis H 3b .Pro represents the rotation from smaller to larger offices, and under both regressions Pro is positive and significant.The net effect of Pro under Model 2a is negative, which supports hypothesis H 3b .Conversely, the net effect of Pro under Model 2b is positive, which does not support hypothesis H 3b .