carey and simnett - 2006 - audit partner tenure and audit quality in australia [mapr]

For long tenure observations we find a lower propensity to issue a going-concern opinion and some evidence of just beating missing earnings bench- marks, consistent with deterioration i

Audit Partner Tenure and Audit Quality

Author(s): Peter Carey and Roger Simnett

Source: The Accounting Review, Vol 81, No 3 (May, 2006), pp 653-676

Published by: American Accounting Association

Stable URL: http://www.jstor.org/stable/4093109

2006

pp 653-676

Peter Carey Monash University

Roger Simnett The University of New South Wales

ABSTRACT: Rotation of audit partners is one of the main policy initiatives that has been implemented in many jurisdictions around the world to deal with concerns about audit quality The basis of any requirement limiting the tenure of audit partners is that there is a reduction in audit quality associated with long periods of tenure Using data from Australia, where the audit partner can be identified and for a period where partner rotation was not mandatory, we examine the association between audit quality and long audit partner tenure The three measures of audit quality examined are the audi- tor's propensity to issue a going-concern audit opinion for distressed companies, the direction and amount of abnormal working capital accruals, and just beating (missing) earnings benchmarks For long tenure observations we find a lower propensity to issue

a going-concern opinion and some evidence of just beating (missing) earnings bench- marks, consistent with deterioration in audit quality associated with long audit partner tenure There is no evidence of an association of long audit tenure with abnormal working capital accruals

Keywords: audit partner tenure; audit quality; qualifications; earnings management

Data Availability: All data has been extracted from publicly available sources

I INTRODUCTION policy of mandatory rotation of audit partners for a particular audit client is sug-

gested as a means of improving audit quality and therefore increasing the quality

of general-purpose financial statements The benefits normally espoused are that the independence of the engagement partner is maintained or that a new perspective on the audit may result in the identification of issues that have been overlooked on previous audits These benefits therefore rely on the belief that extended audit partner tenure is associated

of Australia and New Zealand Conference, Boston Area Research Colloquium (BARC), University of Alberta,

Ana de Mates, Kim Huynh, Amna Khalifa, Bibi Moore, and Linda Pellegrino Financial assistance was provided

by a Monash University and University of New South Wales research grant

Editor's note: This paper was accepted by Terry Shevlin

Submitted January 2004 Accepted November 2005

653

654 Carey and Simnett

with decreases in audit quality This research study provides evidence on the extent to which long audit partner tenure is associated with decreases in audit quality

In countries (such as the United States) where a policy of partner rotation is already

in place, it is impossible to undertake an empirical evaluation of whether this policy is beneficial, as it is impossible to identify the consequences of not having (and therefore the benefits associated with implementing) this policy The basis for the policy, that extended audit partner tenure is associated with decreased audit quality, is best informed by under- taking the analysis in an environment and at a time where the policy is not mandatory and the audit partner can be identified Australia provides such an environment There has been

a statutory requirement for the auditor of listed Australian companies to sign the audit report in their own name as well as in the name of the audit firm since the 1970s (section

324 (10) of Australian Corporations Act) In Australia there was neither a legal nor pro- fessional requirement to rotate the audit partner until 2001.1

The basis for a policy of audit partner rotation is that there is a diminution in audit quality over the tenure of an audit partner Such a policy assumes that a sufficiently high level of audit quality is achieved in the years up to recommended rotation It is inferred that after a long period of tenure, continuing relationships may impact a partner's indepen- dence and time may erode a partner's capacity for critical appraisal If these potential factors result in a reduction in audit quality or in audit failure, then they will impose a cost on society and the profession.2

This study investigates if there is a diminution in audit quality associated with long partner tenure using three common measures of audit quality: the auditor's propensity to issue a going-concern audit opinion; an examination of the signed and absolute amount of abnormal working capital accruals; and an analysis of the extent to which key earnings targets are just beaten (missed)

We undertake a cross-sectional analysis where we trace the length of partner tenure for all listed companies in Australia for a specific year to identify any association with audit quality The year we choose is 1995, before a rotation policy became mandatory (2001) and before it was voluntarily adopted by any accounting firms (1997, refer to footnote 1) This time period allows for the identification of long tenure situations and therefore facil- itates an examination of the theoretical underpinnings of an audit partner rotation policy; that there is a diminution in audit quality associated with long audit partner tenure For long partner tenure observations we find that there is a lower propensity to issue

a going-concern opinion There is no evidence of an association of long partner tenure with either the signed or absolute amount of abnormal working capital accruals For the third audit-quality measure we find that there is some evidence of just beating (missing) earnings benchmarks for long partner tenure observations The findings with regard the auditor's propensity to issue a going-concern audit opinion and the extent to which key earnings targets are just beaten (missed) are consistent with a deterioration in audit quality associated with long partner tenure Sensitivity analyses demonstrate that identified deterioration in audit quality is specifically associated with non-Big 6 audit firms

credibility (Grant et al 1996)

The Accounting Review, May 2006

The paper is structured as follows Section II outlines recent key policy initiatives regarding audit partner rotation in some of the major jurisdictions around the world Section III outlines the literature review and development of theoretical arguments This is followed

by separate sections outlining the research methodology and the results for each of the three audit-quality measures Section X concludes and outlines the implications of the findings

II RECENT POLICY DEVELOPMENTS REGARDING AUDIT PARTNER

ROTATION IN MAJOR JURISDICTIONS There is considerable variation in international policy regarding periodic audit partner rotation The issue continues to attract policy attention in many countries The policy first emerged in the U.S in the 1970s, at which time the AICPA Practice Section mandated periodic audit partner rotation after seven years of tenure (this became a requirement for U.S Securities and Exchange Commission registered clients) Recent corporate events led

to a revision of this policy with the signing of the Sarbanes-Oxley Act of 2002 and there

is now a requirement that the lead audit partner and audit review partner be rotated every five years on public company engagements

In the United Kingdom, the Cadbury Committee (Committee on the Financial Aspects

of Corporate Governance 1992) argued in favor of partner rotation and regulations were subsequently adopted that followed the U.S approach of requiring rotation of the audit engagement partner after seven years for listed companies In January 2003, the Department

of Trade and Industry and the U.K Treasury issued a final report on review of audit and accountancy regulations The report recommended that the maximum period before rotation

be reduced from seven to five years for the lead audit partner The report also welcomed the recommendation from the EU that the partner rotation requirement should extend to other key audit partners (such as those involved in second partner review) It recommended

a maximum seven-year period before rotation of other key audit partners

More than 80 countries use the international auditing standards and Code of Ethics prepared by the International Federation of Accountants (IFAC) as the basis for their stan- dards In 2001 IFAC recognized in their Code of Ethics that prolonged use of the same lead engagement partner on an audit may create what they termed a "familiarity threat." The code proposed that the lead engagement partner be rotated after a pre-defined period, normally no more than seven years, and that the rotating partner not resume the lead engagement partner role until an additional two years have elapsed

In Australia, the earliest references to partner rotation arose in a report issued by the professional accounting bodies (Australian Society of CPAs and The Institute of Chartered Accountants in Australia 1994) On the basis of concerns about audit quality they advo- cated, among other reforms, the rotation of audit engagement partners every seven years This suggestion was not adopted by the profession In 1996 an Australian Commonwealth Government working party released a review of the requirements for the regulation of auditors, which recommended mandatory rotation of audit partners every seven years Again

this suggestion was not adopted, although the independence standard at the time, AUP32, was revised to require "consideration" of a rotation policy In 2001 the Australian profes- sion's independence standard was revised in line with changes to the IFAC Code of Ethics

It stated in paragraph 2.51 that the lead partner be rotated after a predefined period of no longer than seven years In 2004, legislation was introduced that requires the audit partner

of listed companies to be rotated after no more than five years There was no empirical evidence provided to support the benefits of either the introduction of a policy of partner

656 Carey and Simnett

rotation or a change in the conditions of this policy As outlined earlier, the basis of any policy of audit partner rotation is that there is a diminution in the quality of the audit associated with long audit partner tenure This study examines the existence of such an association

III LITERATURE REVIEW AND DEVELOPMENT OF

TESTABLE HYPOTHESIS While there are a number of recent studies on audit firm tenure (Ghosh and Moon 2005; Myers et al 2003; Davis et al 2003; U.S Government Accounting Office [GAO] 2003; Geiger and Raghunandan 2002; Johnson et al 2002), there is very little empirical examination of the relationship between audit partner tenure and audit quality The only published paper to date to examine this issue is Chi and Huang (2005), who, in an exam- ination of the Taiwanese audit market, find evidence consistent with lower earnings quality

in the early years of audit tenure (either firm or partner) and the later years of tenure (firm), where earnings quality is measured as the level of abnormal accruals However, their results are contingent on the method used to calculate the level of abnormal accruals

The literature on audit firm tenure suggests deterioration in audit quality in the early years of tenure due to a loss of client-specific knowledge and expertise DeAngelo (1981) identifies a "learning curve" that gives incumbent auditors a comparative quality advantage Continuity on an audit is said to reduce audit risk due to a familiarity with the client's system and an understanding of risks associated with the client's business/industry envi- ronment (Financial Reporting Commission [Ryan Commission Report] 1992; AICPA [Cohen Commission Report] 1978)

While there is some empirical evidence of a heightened risk of reduced audit quality during the first years of an audit firm's tenure (AICPA 1992; Johnson et al 2002; St Pierre and Anderson 1984) and higher audit quality with longer audit firm tenure (Myers et al 2003), the extent to which these relationships translate to audit partner tenure is not clear The risk of audit failure in the years immediately after partner rotation is likely to be less severe than the risk in the first years of an audit firm's tenure due to other quality controls that an audit firm can establish to mitigate risks associated with a new partner's lack of familiarity (e.g., second partner review and standard new engagement familiarization pro- cedures for incoming partner) and the maintenance of knowledge and expertise within the firm (e.g., continuity of field staff, the carrying forward of working papers, partner famil- iarity with existing audit methodology and client databases)

Association between Long Audit Partner Tenure and Audit Quality

The two (related) primary arguments supporting a negative association between long audit partner tenure and audit quality are (1) erosion of independence that may arise with the development of personal relationships between an auditor and their client, and (2) deterioration in the audit partner's capacity to effect critical appraisal (described in the IFAC Code of Ethics as a familiarity threat)

The development of personal relationships between client management and the audit partner may threaten the independence of the auditor Mautz and Sharaf (1961, 231) state that the auditor "must be aware of the various pressures, some obvious some subtle, which tend to influence [their] attitude and thereby erode slowly but surely [their] independence."

In most instances "the greatest threat to [their] independence is a slow, gradual, almost casual erosion of [their] honest disinterestedness" (Mautz and Sharaf 1961, 208) This threat

to independence does not only occur at the time of reporting, but also has the potential to

The Accounting Review, May 2006

impact judgments made during the audit (Dopuch et al 2003; Bazerman et al 1997) Thus,

it is alleged that extended personal relationships to the extent of developing bonds of loyalty

or emotive relationships will consciously or subconsciously impact the auditor's indepen- dence and objectivity Specific arguments as to the consequence of diminished independence with long partner tenure include the increased likelihood of the auditor yielding to the inevitable client pressure in an audit conflict situation and, at the extreme, the possibility that excessive familiarity might result in auditor/client collusion (McLaren 1958)

It is also argued that long audit partner tenure can lead to a familiarity threat that can result in a restriction on the creative audit-testing approaches that are often seen in the early years of an audit engagement (AICPA 1978; Hoyle 1978; McLaren 1958) The audit program may become stale and routine as the auditor begins to anticipate the condition of the client's systems and the presence of control procedures (Hoyle 1978) Shockley (1981, 789) asserts "complacency, lack of innovation, less rigorous audit procedures and a devel- oped confidence in the client may arise after a long association with the client." There is

a tendency to anticipate results instead of being alert to subtle and often surreptitious, although important, anomalies (Arrunada and Paz-Ares 1997)

There are also counter-arguments that suggest that long audit partner association with

a particular client can result in higher audit quality These arguments highlight the higher audit costs associated with early periods of auditor tenure and the increase in client and industry knowledge gained over repeated audits (Myers et al 2003)

The increased prevalence of mandatory partner rotation policies shows that the regu- lators believe that the potential costs associated with long periods of partner tenure outweigh the potential benefits The Cadbury Committee (1992) argues that partner rotation enhances the technical rigor of an audit by encouraging a "fresh viewpoint." The AICPA's early requirement to change the partner in charge of auditing a public company at least every seven years "was adopted for the specific purpose of periodically bringing a fresh per- spective to each audit" (AICPA 1992, 4) The IFAC Code of Ethics (2003, para 8.151) outlines that "using the same lead engagement partner on an audit over a prolonged period may create a familiarity threat." Partner rotation is therefore prescribed as a quality control procedure that overcomes the problems associated with long periods of partner tenure The prior discussion of a potential association between audit partner's tenure and audit quality suggests that in the early year(s) of tenure the partner is most likely to be indepen- dent and bring to the audit a fresh perspective While the incoming engagement partner is developing the necessary knowledge and expertise, there is a possible increased risk of failure to detect material misstatements The audit firm will attempt to reduce this risk through quality control processes designed to ensure the new engagement partner gains (or has access to) adequate knowledge and expertise After this initial period has passed and the partner has gained the necessary client or industry knowledge and expertise, there is a period where audit quality is expected to be at its maximum The relatively small number

of years of association is not expected to have eroded independence or the partner's capacity

to undertake critical appraisal The length of this period is unknown, but the recent devel- opments in policy initiatives outlined earlier in this study suggest that this period is com- monly viewed as either five or seven years After this period, the literature suggests a reduction in audit quality associated with long audit partner tenure through the development

of personal relationships and deteriorating capacity for critical appraisal (familiarity threat) This association is examined by the following hypothesis:

Hi: There is a negative association between audit quality and long audit partner tenure

658 Carey and Simnett

In the analyses undertaken in the paper, long audit partner tenure is defined as situations where the auditor has been a partner in charge for more than seven years (TENURE>7) This is consistent with the period of concern currently espoused by the IFAC Code of Ethics, and the period of partner tenure that was commonly used in 1995 in other jurisdic- tions as the basis for a partner rotation policy We also anticipate that there may be an initial period of familiarization for the audit partner Thus we include the first two years of partner tenure as a control variable (TENUREs2) This approach means that the basis of comparison for the later partner tenure is the period where audit quality is expected to be

at its maximum (for this paper, between years 3 to 7 of partner tenure) Other possible specifications of long partner tenure are considered as forms of sensitivity analyses where appropriate This includes the use of a continuous measure for partner tenure, TENURE>5, and to further accentuate the effect of long periods of tenure, tenure squared

IV METHODOLOGY Data Collection

The data collection for this study involves the review and analysis of published infor- mation for public companies listed on the Australian Stock Exchange (ASX) in 1995 We select 1995 as the year for analysis as this is before the introduction of firm, professional,

or legal requirements to limit audit partner tenure in Australia For all listed companies in

1995 we trace back the partner tenure, as well as collecting other relevant information for the analyses of the audit-quality measures The population of companies listed on the ASX

at 30 June 1995 is 1,152 (Craswell 1996) Of these, 40 foreign companies are removed (the audit report is signed in an overseas jurisdiction and does not necessarily identify the audit partner) and 66 are removed due to missing data with respect to incomplete financial information or an inability to trace the inception of audit partner tenure The 25 observations

in the banking and insurance industry are also excluded as their total asset base and financial structure is not comparable to those of the other companies This results in a final sample

of 1,021 Australian-domicile companies

Research Method

To identify whether there is a negative association between long audit partner tenure and audit quality, we use three common measures of audit quality.3 First, the auditor's propensity to issue a going-concern opinion for distressed companies; second, the amount

of abnormal working capital accruals; and third, the extent to which key earnings targets are just beaten (missed) These are all examined over short, medium, and long periods of audit partner tenure In accordance with the predicted directional effect, we utilize one- tailed testing in evaluating the impact of audit partner tenure The comparison of measures

of audit quality for the medium (when audit quality is expected to be at its maximum) with the long tenure periods allows us to identify any diminution in audit quality associated with long tenure The examination of multiple measures of audit quality allows us to build greater confidence in any observed relationship between audit quality and long partner tenure The research methods associated with each of these measures are discussed with their results

3 Other measures considered are the number of audit failures as identified by legal action undertaken against the auditor, or companies placed in bankruptcy soon after the issue of an unmodified audit report The very small number of legal actions and bankruptcies in Australia means that these proxies are not appropriate

The Accounting Review, May 2006

V DESCRIPTIVE RESULTS AND DISCUSSION Descriptive statistics are reported in Table 1 Of the 1,021 Australian-listed companies,

15 percent (n = 152) have greater than seven years of continuous partner tenure.4 Forty- four percent of companies (n = 447) have less than two years of continuous partner tenure The companies have been listed on average 12.81 years and have A$299 million of assets The Big 6 audit firms audit 64 percent of these clients and 7 percent of the audit opinions are modified relating to the going-concern issue An examination of the correlation matrix reveals that apart from the expected high correlations of PBANK with LEV (0.627) and CFFO (-0.532), there are no other correlations greater than 0.50

VI ANALYSIS 1: ISSUING GOING-CONCERN AUDIT OPINIONS

Method

The first measure of audit quality examined in this study is that of the auditor's pro- pensity to issue a going-concern modified audit opinion Although it is not easy to associate

TABLE 1 Descriptive Statistics

SIZE = total assets of the company at financial year-end;

LEV = total liabilities divided by total assets;

LLOSS = 1 if company reported a loss for the previous year, and 0 otherwise;

CFFO = cash flow from operations over total assets at the end of the fiscal year; and

MINING = 1 if mining industry, and 0 otherwise

660 Carey and Simnett

any particular type of audit opinion with a level of audit quality, there is a well-established body of literature that aids the prediction of going-concern modified opinions (see for example, DeFond et al 2002; Mutchler et al 1997; Hopwood et al 1989) DeFond et al (2002) argue that in issuing a going-concern opinion, the auditor must objectively evaluate company performance and withstand client pressure to issue a clean opinion This suggests

a correlation between the issuing of going-concern opinions and auditor independence Thus, if after controlling for other factors that influence the propensity to issue a going- concern modified audit opinion, the probability of issuing going-concern reports is inversely related to long audit partner tenure, this suggests audit quality is diminishing with long partner tenure

Consistent with DeFond et al (2002) and on the basis that going-concern problems are more salient among financially distressed companies, we restrict our analysis to situations where the company reports negative earnings and/or operating cash flows for the current fiscal year Also, consistent with the approach of DeFond et al (2002), the following logistic regression model estimates the auditor's probability of issuing a going-concern modified opinion to a financially distressed client:

OPINION = PO + PIPBANK + 32SIZE + P3AGE + 34LEV + P5CLEV

+ 36RETURN + P7LLOSS + 38INVESTMENTS + 39AUDFIRM + Bo1FEERATIO + PICFFO + 312MINING + P13TENURE<2

TENURE>7 = 1 if audit partner is engagement partner on a client company for greater

than seven years, and 0 otherwise;

Control Variables:

PBANK = probability of bankruptcy as measured by adjusted Zmijewski score;5 SIZE = natural logarithm of total assets of the company at financial year-end; AGE = natural logarithm of number of years since listing on the Australian

Stock Exchange;

LEV = total liabilities divided by total assets;

CLEV = change in LEV during the year;

RETURN = market-adjusted return over the fiscal year;

model calculated by Carcello et al (1995) who use the weightings b* = -4.803 -3.6(net income/total assets)

The Accounting Review, May 2006

LLOSS = 1 if client reported a loss for the previous year, and 0 otherwise; INVESTMENTS = short- and long-term investment securities (measured as current assets

less debtors and inventory) divided by total assets;

AUDFIRM = 1 if audited by Big 6 firm, and 0 otherwise;

FEERATIO = ratio of nonaudit fees to total fees paid to the incumbent auditor; CFFO = operating cash flow divided by total assets;

MINING = 1 if mining industry, and 0 otherwise; and

TENURE?2 = 1 if audit partner is engagement partner on a client company for two

years or less, and 0 otherwise

The choice of control variables in the model is informed by the literature that aids the prediction of a going-concern opinion (in particular, DeFond et al 2002) As outlined earlier,

TENURE_2 is used to control for instances of possible lower audit quality in the first two years of partner tenure due to a lack of client familiarity affecting auditor expertise The other variables are consistent with the model of DeFond et al (2002).6 PBANK is the probability of bankruptcy score based on Zmijewski (1984), with higher values indicating

a higher probability of bankruptcy SIZE is included because large companies have greater negotiating power and are less likely to end up in bankruptcy AGE captures the fact that younger companies are more likely to encounter financial distress LEV captures risk as- sociated with higher levels of debt, and CLEV captures movements in leverage that may move companies to unacceptable or unsustainable levels of debt RETURN is a market- based measure of risk and company performance LLOSS is included as companies with continued losses are more likely to fail INVESTMENTS is a liquidity measure that captures the ability to quickly raise cash, while AUDFIRM captures any differences in the propensity

of the big audit firms to issue going-concern modifications to the audit opinion It is sug- gested that FEERATIO is associated with reduced audit quality (Frankel et al 2002) and poor operating cash flows (CFFO) are commonly associated with a higher likelihood of financial distress (DeFond et al 2002) The variable MINING is included because of the large number of mining companies listed in Australia and their potentially different financial profiles (Butterworth and Houghton 1995)

Results

The restriction of our analysis to situations where the client reports negative earnings and/or operating cash flows in 1995 results in the identification of 559 companies The average partner tenure of these observations is 3.75 years (46 percent are in either years 1

or 2 of partner tenure and 11 percent are long tenure observations of greater than seven years) and we observe 66 (12 percent) going-concern modified audit opinions

Table 2 presents results from the logistic regression model for going-concern modi- fied audit opinions The model adequately distinguishes the going concern decision (x2

= 114.15, df = 14, p = 000), and an overall fit of 88.6 percent is achieved Consistent

with DeFond et al (2002), we find significance in the predicted direction at p < 05 for

the coefficients LLOSS and INVESTMENTS, while SIZE, CLEV, RETURN, CFFO, and FEERATIO are not found to be significant predictors However, in contrast to DeFond et

al (2002), we find significance in the predicted direction for LEV and opposite to the predicted direction for AGE We do not find significance in the predicted direction of the

risk and return volatility and an indicator variable for ability to raise future financing These variables are not

662 Carey and Simnett

TABLE 2 Propensity to Issue a Going-Concern Opinion for Financially Distressed Companies

(n = 559)

+ 37LLOSS + P3INVESTMENTS + 39AUDFIRM + 310FEERATIO

otherwise;

and 0 otherwise;

LEV = total liabilities divided by total assets;

CLEV = change in LEV during the year;

LLOSS = 1 if company reported a loss for the previous year, and 0 otherwise;

CFFO = cash flow from operations over total assets at the end of the fiscal year; and

MINING = 1 if mining industry, and 0 otherwise

coefficients for AUDFIRM and PBANK The estimated coefficient for TENURE<:2 is pos- itive but is also not significant (p = 072, one-tailed)

The negative estimated coefficient for TENURE>7 is significant (p = 042, one-tailed) and in the expected direction This result suggests that long-serving audit partners have a

lower propensity to issue a going-concern modified opinion This provides evidence that is

The Accounting Review, May 2006

consistent with Hi, that long audit partner tenure is negatively associated with audit quality

To determine the economic significance of this result, we change the value for TENURE> 7 from 0 to 1 while setting other variables in model (1) equal to their sample mean This results in a 5.30 percent decrease in the probability of receiving a going-concern modified opinion

Sensitivity Analyses

To control for the impact of a prior year going-concern opinion on an auditor's pro- pensity to issue a going-concern opinion in the current year, we first replicate the primary logistic regression analysis, substituting first-time going-concern opinions as the dependent variable, as in DeFond et al (2002) Of the 66 going-concern modifications there are 33 first-time going-concern opinions.' The model is a significant predictor of first-time going- concern modifications (x2 = 49.54, df = 14, p = 000) and results are broadly consistent with the analysis containing all going-concern modifications The TENURE> 7 variable has

a negative coefficient but its significance is now marginal (p = 075, one-tailed) Second,

we rerun the primary analysis including a dummy variable for whether the company re- ceives a prior-year going-concern modification Similar results to those of the analysis of the first-time going-concern opinions are obtained, with TENURE>7 having a negative coefficient and again being marginally significant (p = 098, one-tailed)

Identifying that not all loss making or cash flow negative firms will receive a going- concern modified opinion (only 66 of the 559 companies received going-concern modifi- cations), we extend this analysis to consider those companies most likely to receive a going- concern modified opinion We therefore examine the most severely distressed quartile (140 companies) of the financially stressed companies.8 The going-concern modified opinion is issued in 17 of 43 (39.5 percent) instances for short tenure, 27 of 80 (33.8 percent) instances for medium tenure, and 2 of 17 (11.8 percent) instances for long tenure (greater than seven years) Using a test of proportions, we identify that there is no significant difference between the proportions for short and medium tenure (z = 0.64, p > 05), while the difference between medium and long tenure is significant (z = 1.72, p = 043, one-tailed) This provides evidence of a lower propensity to modify the audit opinion for the most severely financially stressed companies associated with long audit partner tenure

To further investigate if the auditor's propensity to issue a modified audit report di- minishes with tenure, we substitute the two tenure variables (TENUREs2, TENURE>7) with a single linear TENURE variable, while controlling for instances of possible lower audit quality in the first two years of partner tenure by excluding the 258 TENURE?2

observations TENURE remains negative and significant (p = 012, one-tailed),9 indicative

of a lower propensity to issue a going-concern modification during long partner tenure

To explore the potential interaction between AUDFIRM and long partner tenure, we include an interaction term AUDFIRM*TENURE in model (1) TENURE is measured as a continuous variable'0 and observations where tenure is less than two years are excluded The inclusion of the interaction term results in TENURE remaining significant and negative

(3 = -.509, p = 005) while the interaction term is significant and positive (P = 544,

distressed companies, the coefficient remains negative and highly significant (p = 003, one-tailed)