Audit tenure and audit quality the U.S. financial sector : a financial crisis analysis

Audit tenure and audit quality

the U.S. financial sector

A financial crisis analysis

Name: Sandy Franklin

Supervisor: Mr. Georgakopoulos

Second Examiner: Sanjay Bissessur

Date: June 23 2014

Abstract

This study analyzed the relationship between audit tenure and audit quality for U.S. financial institutions using auditor’s propensity to issue a going concern report and the degree that audit clients meet or beat earnings benchmark as audit quality proxies. Additionally this paper also analyzes the change in this relationship for the period before the financial crisis and the period after the financial crisis. These companies were more strongly affected by the crisis so an eventual change is more clearly observable. Previous research has found that independent auditors are more likely to issue going concern reports. Previous studies also found that auditors of high quality are better able to constrain earnings management. The result of this study show that for the period before the financial crisis long audit tenure did not affect auditors propensity to issue a going concern opinion. However for the period during the financial crisis results show that auditors were more likely to issue a going concern opinion. Furthermore the earnings management proxy show that for the period before the financial crisis audit tenure did not affect company’s ability to manage earnings. But again for the period during the financial crisis auditor were better able to constrain earnings management. Thus this study concludes that audit tenure does not negatively affect audit quality. This study assumes that reputation and litigation incentives were strong during the financial crisis.

Table of content

1. Introduction. 1.

2. Literature review. 5.

2.1 Determinants of audit quality. 5.

2.1.1 Auditors Independence. 5.

2.1.2 Audit Knowledge and expertise. 6.

2.1.3 Audit firm. 8.

2.2 Audit rotation 10.

2.3 Audit tenure 12.

3. Audit quality proxies. 13.

3.1 Propensity to issue going concern report. 13.

3.2 Degree of meeting or beating targets. 14.

4. Hypothesis Development 15.

5. Methodology 17.

5.1 Sample 17.

5.2 Research Design 18.

5.2.1. Propensity to issue going concern model 18.

5.2.2 Just meeting or beating earning benchmark models 20.

6. Results 22.

6.1 Results for propensity to issue going concern proxy 22.

6.2 Results of earning management proxy 23.

6.3 Summary Results 29.

7. Conclusion 30.

8. Bibliography 31.

9. Appendix 38.

1. Introduction

The role of the external auditor is to act as a third party to verify accuracy of financial statements. This is particularly important for public companies where there’s separation of ownership and

management. Managers are the ones running the business while the actual owners only get informed about the business through the financial statements. To protect the investors of public companies governments and regulatory authorities require that these financial statements be audited by an independent third party. According to Arens et al (2012) external auditor’s role is to provide reasonable assurance that information on the financial statements is adequate and reliable. Arens et al (2012) defines audit as “the accumulation and evaluation of evidence about information

to determine and report on the degree of correspondents between the information and established criteria “(Arens, 2012, p.24).

In the beginning of last decade a series of accounting scandals of several well respected public companies resulted in many new regulations (SOX) being passed (PCAOB Investors Advisory Group,2011). The purpose of these new rules and regulations was to the enhance auditors independence and restore the public perception regarding the integrity of companies financial statements (PCAOB, 2011). However according to the PCAOB Investors Advisory Group (2011) the recent financial crisis was a test for both the new legislation and the auditors who have the responsibility to practice these new rules. However both the auditors and the new rules were not able to pass the test (PCAOB Investors Advisory Group, 2011). Tens of well-known financial

institutions went bankrupt or had to be saved by the government. Those who invested in the stock market suffered enormous losses and many people lost their home and jobs (PCAOB Investors Advisory Group, 2011). As a consequence many questioned why auditors who are the public’s watchdog and who’s role is to verify the accuracy and reliability of financial statements did not see this coming. The question here is why the auditors didn’t alert society about the practice of the banks and other financial institutions that eventually led to the financial crisis. The PCAOB Investors

Advisory Group (2011) questions whether auditor did not have the ability to detect these practices or whether the auditors consciously help these institutions hide their risk or even help them design transactions to hide their practices. Many others had similar questions. For example in his 2009 paper Sikka (2009) questions what role did auditors have in the financial crisis and how was audit quality affected by this crisis. He discussed several examples of banks and other financial institutions including the Lehman Brothers that receives unqualified opinion shortly before they declared

bankruptcy or requested financial aid to the government. Thus distressed companies who should had received going concern opinions did not. This raises the questions of auditor’s independence and audit quality during the financial crisis.

However audit quality and auditors independence have always been a much discussed topic for audit professional, business manages, shareholders, governments and regulatory authorities. Deangelo (1981) defines audit quality as “the market-assessed joint probability that a given auditor will both

discover a breach in the financial statements and report the breach” (Deangelo, 1981, p.186).

According to Ewelt-Knauer et al (2012) many countries like Australia, Brazil, India, Italy, Holland, USA and many others continue to stipulate numerous regulations to improve auditor independence and audit quality. Without a doubt the most discussed regulations are mandatory audit rotation and the SEC’s prohibition of non-audit services. In U.S. SOX 203 requires audit partners to rotate every 5 years and have 5 year cooling off period after that (Arens, 2012). In The Netherlands a new law in Article number 42 of the Directive requires auditors to rotate every eight years with a cooling off period of two years (The directive 2006/43/ec of the European parliament and of the council).The purpose of mandatory audit rotation is to prevent long audit tenure between auditor and client that may increase familiarity, create dependence and thereby decrease audit quality. However there are some studies that do not support this belief. Some studies argue that audit rotation may actually decrease audit quality. For example Johnson et al (2002) concluded that for short term and medium length relationships there is evidence of reduction in audit quality and for long term relationships between auditor and client there is no evidence of audit quality reduction. Also Chen et al (2008) did not find any evidence that audit tenure reduces earnings quality and Carcello and Nagy (2004) did not find any evidence that long audit tenure results in higher probability of fraud.

The financial crisis was a tumultuous time for the whole economy especially for organizations in the financial sector. According to Gardo and Martin (2010) the financial crisis had a large effect on the European financial market segment. Stock markets booked large losses, exchange rates were

strongly affected, and banks had both a large decline in credit and deposit growth and big increase in non-profitable loans (Gardo and Martin, 2010). Aite group states that the U.S. financial sector suffered a similar faith. They experienced an increased charge-off of loans, credits were largely frozen and almost half on the financial institutions experienced a decrease in their activities (Aite Group, 2009). This resulted in the financial sector being exposed to even more rules and regulations, more supervision from government and financial authorities and even more scrutiny from the financial press. In this study I will make an empirical analysis to assess the effect of audit tenure on audit quality for exclusively companies in the U.S. financial sector and assess whether this

relationship changed during the financial crisis. To investigate this change data from the period before the financial crisis and during the financial crisis is collected. After a separate analysis of the results, a comparison is done. Thus the research question of this paper is:

To what extent does audit tenure affect audit quality during the financial crisis in the financial sector?

This paper offer significant contribution to prior research. Many papers like Carcello and Nagy (2004) and Carey and Simnet (2006) have focused on mandatory rotation, long audit tenure and audit quality. However the financial crisis has affected the business world and therefore also the

accounting and auditing world tremendously. Little research has focused on how the financial crisis did affect auditors independence and audit quality, thus there are calls for further research on this topic. Therefore analyzing the effects of audit tenure and the necessity of audit rotation during the financial crisis which holds the purpose to maintain independence of auditors will most certainly contribute to the accounting literature. The results of my study may help regulators get an insight on how audit tenure affects audit quality and whether this relationship is affected by the financial crisis. The results may also help financial authorities to assess whether mandatory firm or partner rotation needs to be maintained, improved or abolished. Thus they can create new and better regulations that are more fitted for the current situation. Also if it is proven that the positive relationship between audit tenure and audit quality is stronger in the period during the financial crisis, auditors can look within themselves and asses how this improved audit quality can be maintained in the future.

The remainder of the paper contains the following. Section 2 discusses a literature review, which discusses the determinants of audit quality, audit rotation and audit tenure. Section 3 elaborates on the audit quality proxies used in this paper followed by section 4 that present the hypothesis

development. Section 5 discusses the methodology used in this paper including sample and research design. In section 6 the results of the analysis are shown and discussed followed by a summary of the results and a conclusion in section 7.

2.

Literature review

2.1 Determinants of audit quality

As stated before Deangelo (1981) defines audit quality as” the market-assed probability that a breach

in the financial statements are both discovered and reported” (Deangelo, 1981, p186). This means

that audits are of high quality if auditors are able to identify accounting mistakes in the financial statement and willing to demand re-adjustments. There are many factors that could influence audit quality. Audit quality might be influenced by auditor’s independence and by their expert knowledge. However auditor’s litigation and Reputation risk might also be of influence on the quality of audit auditors provide. In this section several important determinants of audit quality will be discussed.

2.1.1 Auditor independence

Section 1 of Rule 101 of the American Institute of CPA’s states (AICPA ) that all member in the public sector should be independent (AICPA,1988). Section 2 of Rule 101 also discusses circumstances where the independence of the auditors is believed to be impaired. According to the AICPA auditors who have an material interest in their client, have joint closely held investments or have the

authority to make investment decision for the client are not considered to be independent

(AICPA,2014). Further if the audit partner’s or professional employee’s immediate family owns more than 5 percent of the clients equity, the partners is not considered independent (AICPA 2014). Finally if the audit partner during the period of the service was in any way employed by the client then this partner is considered not independent (AICPA 2014). To ensure that these rules are enforced the Independence Standards Board Standard No.1 requires auditor to disclose to the client’s audit committee all relationships the audit professional has with the company that might impair

independence (SEC 2014). To further ensure auditors independence the SEC prohibits the auditor to perform certain non-audit services to their clients and affiliates. These are (1) bookkeeping, (2)

design and implementation of clients information system, (3)management or human resource services, (4)legal and expert services that are not connected to the audit task,(5) internal audit services,(6) investment banking services and (7)valuation services (SEC 2014). The purpose of these prohibitions is to avoid situations where the auditor is auditing his own work or serve as managers or advocates for the client while they are performing audit services (SEC 2014).

Watts and Zimmerman (1983) define auditor’s independence as “the probability that the auditor will

report a discovered breach in the financial report” (1983, p. 615). However when addressing the

independence issue it is important to make a distinction between independent in appearance and independence in fact. According to Dopuch et al (2003) independence in fact is when auditors have an independent mind set when performing the audit task and therefore making the audit report unbiased. However this independent mind set is not easily observable and therefore difficult to regulate (Dopuch et al, 2003). An auditor with an material interest in an client that allows this client to manage earning upwards to increase firm value and therefore increase the value of his own interest is an example of an auditor that is not independent in fact (Almar and Olazabel,2001). On the other hand independence in appearance means that the auditor appears to be independent.

Meaning that the public perceives this auditor to be independent. According to Dopuch et al (2003) and (Almar and Olazabel, 2001) independence in appearance is the most important of the two. Because if an auditor is independent in fact but the public does not perceive the auditor to be independent then the auditor may encounter similar consequences as being independent. Some of these consequences are loss of reputation, litigation or loss of clients.

2.1.2 Audit knowledge and expertise

Auditor’s knowledge and expertise is also a great determinant of audit quality. According to Solomon et al (1999) auditors who are industry experts make more reliable audit judgements. This is because they have more experience in that particular sector compared to their non-experts competitors (Solomon, 1999). Auditors who have a larger market share in a particular industry have more opportunities to gain additional knowledge and become industry experts (Francis, 2004). Research have found that industry experts in the U.S. have over 50% of total audit fees of a particular industry while the number one non-expert has only 22% of total fees (Francis,2004). According to Francis (2004) research have found that industry experts are able to charge higher fees quality meaning that they have higher audit and earnings quality. Research has even found that in countries like Hong Kong and Australia industry experts earn a 10-30% premium above Big 4 firms (Francis, 2004). Finally research has also shown that companies who are audited by industry experts have smaller abnormal

accruals and experience less under-pricing during their IPO’s. Thus industry experts are better able to contain earnings management and have higher earnings quality.

In their 2004 paper Carcello and Nagy analysed the relationship between industry expert auditors and financial reporting fraud. They argue that previous research found that independent audit committees and board of directors are more likely to choose an auditor that is the industry specialist (Carcello and Nagy, 2004)). The idea behind this is that industry experts provide higher quality audits and are therefore better able to detect financial misstatements caused by error (unintentional) or by fraud (intentional) (Carcello and Nagy,2004). Furthermore according to them previous research has also found a positive relationship between industry experts and earnings response coefficient (ERC). They also found a negative relationship between industry specialist and discretionary accruals (Carcello and Nagy). In their own study Carcello and Nagy (2004) found a negative relationship between industry experts and financial reporting fraud. However their study also found that this relationship is weaker for larger clients. Meaning that because larger companies (clients) usually have more complex operations and are active in several industries makes it difficult for auditors to have this all around expert knowledge (Carcello and Nagy,2004).

According to Boner and Lewis (1990) there are 4 determinants of auditor expertise. The first is general domain knowledge which is general accounting and auditing information. This includes knowledge of general accepted accounting principles, general accepted auditing standards and knowledge of basic accounting transactions. According to Boner and Lewis (1990) audit professional gains this knowledge by formal training and work experience. The second determinant of audit expertise is subspecialty knowledge. Auditors gain this knowledge thru working experience in a certain industry and/or thru firm training that specialize in these areas or industries (Boner and Lewis, 1990). According to Boner and Lewis (1990) auditors are not able to gain subspecialty

knowledge through general working experience or training. Therefore this knowledge is not acquired by all auditors (Boner and Lewis, 1990). The third determinant of audit expertise is general business knowledge. For example understanding how managers react to certain compensation schemes compared to others (Boner and lewis, 1990). According to them this knowledge differs for auditors with both similar and different experience levels. Mostly because this knowledge is acquired thru personal experiences like client mix or thru personal interest like frequency of reading business articles (Boner and Lewis, 1990). The fourth and final determinant of auditor expertise is auditor’s problem solving abilities. This includes auditor’s analytical skills, ability to understand information and ability to recognise relationships (Boner and Lewis, 1990). Boner and Lewis (1990) illustrates the importance of these determinants by explaining that auditors who have the required general

accounting and auditing knowledge without the problem-solving skills will not be able to be experts in a particular task. The same goes the other way around.

Reichelt and Wang (2010) go even further in the expert knowledge and audit quality discussion by examining whether audit quality is higher for experts at firm-level or office level. According to them firm-level or national expert knowledge increase audit quality due to positive network synergies (Reichelt and Wang, 2010). These synergies are accomplished by sharing knowledge across offices, collective use of standardized audit systems customized for certain industry and extending the client mix by allowing audit professionals to travel to other national offices (Reichelt and Wang,2010). Experts knowledge at office level is develop thru close experience with the client and gaining client specific knowledge that cannot be transferred across offices. Reichelt and Wang (2010) measure audit by analysing auditor’s propensity to issue going concern opinion, degree of meeting or beating analyst profit forecast and magnitude of abnormal accruals. They concluded that client whose auditors are both city (office level) and national (firm-level) experts have smaller abnormal accruals (Reichelt and Wang, 2010). They also found the clients of auditors who are national and city expert are less likely to beet analyst forecast. Meaning that audit firm who are both office level experts and firm level experts are better able to constrain earnings management (Reichelt and Wang, 2010). The result for the propensity to issue going concern proxy is similar to the earnings management proxies. Auditors who are both city industry experts and national industry experts are more likely to issue a going concern report for distressed companies (Reichelt and Wang, 2010). So this evidence suggests that industry experts ability to provide higher quality auditors comes from the fact that they are both national experts (who can share information across offices) and because they are city experts (who share information within the local office) (Reichelt and Wang, 2010).

2.1.3 Audit firm size

It has long been argued that big 4 auditors have higher quality service and it is clearly observed that the companies in general favour big 4 auditors. According to Lennox (2003) there are several reasons for this. First, the market reacts more positively when companies switch to a big 4 auditor compared to a smaller auditor. Also there is a perception that larger audit firm are better able to identify financial trouble. Furthermore companies with high agency cost generally prefer big 4 auditors and companies who have big 4 auditors suffer less under-pricing during their IPO’s (Lennox, 2003). According to Deangelo (1981) larger audit firm are able to perform higher quality audits due to altered incentives. As mentioned before Deangelo (1981) defines audit quality as the probability that auditors will discover a breach in the financial statements and report this. However according to

Deangelo (1981) auditors sometimes have incentives to initially offer a certain high level of audit quality and later opportunistically reduce this level. According to her auditors are able to do this because it is very difficult and costly for audit clients to evaluate actual levels of audit quality and the probability of being caught is less than one. On the other hand auditors also have disincentives to cheat and these disincentives grow with the size of the audit firm (Deangelo 1981). The basis for this disincentive to cheat is that incumbent auditors earn quasi-rents that are client specific. However auditors who are caught cheating may suffer several negative consequences (Deangelo, 1981). Clients could reduce their audit fees or fire them. In other words, being caught will result in loss of client specific quasi rents (Deangelo 1981). Because larger audit firms have bigger amount of clients, they have bigger disincentives to not cheat. For this reason Deangelo (1981) believes that larger audit firm provide higher quality audits.

There are several empirical analysis that supports Deangelo’s theoretical analysis. For example Becker et al (1998) analysed and concluded that the discretionary accrual of big 6 client are smaller than discretionary accrual of non-big 6 clients. Thus big 6 auditors are better able to constrain earnings management then non big 4. Francis and Yu (2009) also support the argument that bigger firms offer higher quality audits. According to them experience and expertise is an important driver of audit quality and larger audit firms with larger local offices have more engagement hours and therefore more audit experience. That is why these bigger audit firms are better able to detect material misstatements in the financial reports of their clients (Francis and Yu, 2009). They also argue that larger audit firm have more peers to consult with, have better networking opportunities and are therefore able to perform higher quality audits then their smaller competitors. To test this

hypothesis Francis and Yu (2009) analysed whether big 4 accounting firms are more like to issue going concern reports, if these going concern reports are accurate and whether their clients engage in earning management. Their results show that big 4 auditors are more likely to issue going concern report and clients of big 4 auditors engage in less aggressive earnings management (Francis and Yu, 2009).

Two other well discussed explanation of higher audit quality by bigger firms is the Reputation and Deep pocket hypothesis (Lennox, 2003). The Deep pocket hypothesis argues that larger audit firms have more wealth then smaller firms. Therefore if they are sued by a party they have more to lose (deeper pockets).Thus they have bigger incentives to perform high quality audits to reduce litigation risk (Lennox, 2003). The reputation hypothesis is related to the Deangelo (1981) theory. According to this hypothesis audit firms have client specific quasi rents. If they perform low audits they are at risk of losing their reputation and therefore their quasi-rents (Lennox, 2003). Because large audit firm have more clients they have bigger incentives to perform high quality audits (Lennox, 2003). These audit firms want to avoid criticism that may damage their reputation (Lennox, 2003).

2.2 Audit rotation

The Sarbanes Oxley Act requires lead auditors of audit teams to rotate every 5 years. The act also requires them to wait 5 years before being able to audit the same client again. This rule was

introduces to improve and maintain auditors independence after major accounting scandals around that time (Arens et al, 2012). However mandatory firm rotation have also been proposed, and in some countries this has been implemented. According to Ewelt-Knauer et al (2012) countries like Brazil, India and Italy have introduced mandatory audit firm rotation and Countries like Spain and Canada did the same but abolished the rule after a few years. Ewelt-Knauer et al (2012) also state that countries like U.K and U.S. have discussed the implementation of this rule but concluded that there are more disadvantages then there are advantages. In their report Ewelt-Knauer et al (2012) discussed both the advantages and Disadvantages of Mandatory firm rotation. Their first argument in favour of firm rotation is the increased independence of auditors. They argue that it is believed that long audit-client relationships result in too much familiarity and auditors being less sharp and relying too much on last year’s audit results (Ewelt-Knauer et al 2012). The second argument for mandatory firm rotation Ewelt-Knauer et al (2012) mentions is the increased independence in appearance. According to them if audit firm rotate every few years this will increase financial statement users confidence in the audit work and therefore create positive market reactions. The third pro argument this report mentions is the increased market competition amongst audit firm. Now smaller firm are also able to participate in the proposals. If competition increases, auditors will have to differentiate themselves in terms of service to win the new clients which in turn will improve audit quality (Jackson et al, 2008).

However according to Gallagher and Richson (PWC, 2012) the costs of mandatory audit rotation are much larger than the benefits. They believe that the loss of knowledge of companies operations, risks, procedures and people, outweigh the benefits of rotation (Gallagher and Richson, PWC, 2012). According to them benefit of having a “fresh look “ every few years does not justify the loss of information. They also mention that the concern previous studies have found that fraudulent behaviour is more common in the early years of audit-client relationships (Gallagher and Richson, PWC, 2012). Furthermore they state that there is no consistent evidence in the accounting literature

that audit firm rotation does improve audit quality while the cost of mandatory firm rotation is very clear (Gallagher and Richson, PWC, 2012). They conclude by mentioning a few important

disadvantages of mandatory audit rotation. According to them mandatory audit rotation prevents the audit committee, who themselves are perfectly capable of choosing the right auditor, to exercise their best judgement and therefore reduce audit committee effectiveness. They also believe that mandatory firm rotation will distract managers and audit committee from their main task during the proposals (Gallagher and Richson, PWC, 2012).

Ewelt-Knauer et al (2012) also mentions disadvantages of mandatory firm rotation. The first argument against firm rotation is that short audit-client relations results in auditors being more tolerant of management and less critical of their practices. This is because auditors are more concerned about recouping the start-up costs. Ewelt-Knauer et al (2012) also mentioned that short tenure increases the risk of audit failure because auditors do not have the enough time to gather knowledge about clients operations, industry and risks to properly perform audits. Ewelt-Knauer et al (2012) argue that due to lack of expertise most audit failures happen in the early years of audit-client relationships. Myers et al (2003) support this argument. According to them research have found that a higher percentage of audit failure happens during the first few years of audit tenure and that auditors litigation risk is the highest during early years of client-auditor relationship. Ewelt-Knauer et al (2012) consistent with Gallagher and Richson (PWC, 2012) also mention the increased cost for both auditors and clients associated with audit rotation. According to them auditors incurred cost while trying to get to know and understand their new clients. While clients incur additional cost when supporting their new clients in learning their business. Finally Ewelt-Knauer et al (2012) argue that mandatory firm rotation may instead of increase competition among audit firm may reduce this competition. This is mainly because audit committees have the perception that smaller accounting firms do not have the resources to handle continuous short term changes and they do not have the capacity or expertise to quickly understand the complexities of these large organisations (Ewelt-Knauer et al 2012). They conclude by stating that while audit rotation does increase auditor’s independence in fact and appearance, regulators should recognise that audit rotation is very expensive and they should determine what the long term objectives are of such a costly rule. Especially because of the limited evidence in the literature that audit rotation does improve audit quality (Ewelt-Knauer et al 2012).

2.3 Audit tenure

Many research papers have argued that long audit tenure does not affect audit quality but instead helps auditors to perform high quality audits. For example Ewelt-Knauer et al (2012) argue that long audit tenure which creates client familiarity results in higher earnings quality. Myers et al (2003) argue that audit tenure is beneficial for the client because over time auditors gain client specific knowledge to help them better understand the client business and their complexities (Myers et al, 2003). Therefore the auditors can rely less on management estimations which should increase audit and earnings quality. According to Myers et al (2003) many managers are against short audit tenure because continuously changing auditor is costly and they do not believe new auditors are willing to put in the extra effort that is necessary to get to know the new client and their business. According to Jackson et al (2008) because of mandatory audit rotation the market also misses out on some

valuable market signal. For example if auditor and client were having some conflict over accounting treatments auditors resignation would send a signal to the market. Because rotation is mandatory the market cannot make a distinction between mandatory or voluntary rotation due to audit-client disagreements (Jackson et al, 2008).

Myers et al (2003) test results are inconsistent with the arguments that long audit tenure reduces earnings quality. Their results show that long audit tenure results in audit client having less discretionary accruals and income increasing accruals. Thus according to their results longer audit tenure in some cases reduces earnings management. Jackson et al (2008) also studies the effects of long audit tenure on audit quality. In their study they used propensity to issue going concern report and level of discretionary accrual as audit proxies. They concluded that audit tenure increases audit quality when this is measures as propensity to issue going concern opinion. On the other hand their study did not find any evidence that long audit tenure reduces audit quality when this measured as level of discretionary accruals (Jackson et al, 2008).

According to the auditor expertise hypothesis audit tenure actually increases audit quality because longer audit-client relationships reduces information inequality between the two parties. Meaning that with time auditors gain more client specific knowledge which results in higher quality audits (Monroe and Hossain, 2013).

3. Audit quality proxies

To analyse the relationship between audit quality and audit tenure, it is necessary to define and discuss the audit quality proxies used in this analysis. These are propensity to issue going concern opinion and degree of just missing or beating earnings benchmark. In the next section these two proxies will be discussed

3.1 propensity to issue going concern reports

Public Accounting Oversight Board (PCAOB) is a private non-profit organisation that oversees the audits of public companies in the U.S. The PCAOB’s goal is to protect investors and the public interest by making sure that information in the audited financial statements are accurate reliable and

informative (PCAOB, 2014). Therefore AU sec. 341 requires auditors to assess whether there is reasonable doubt that a company will not be able to continue as a going concern for in future. AU section 341 requires auditors to assess client’s ability to continue as a going concern in a step by step matter. The auditors should first evaluate if during the planning and performing of the audit, some events or conditions were identified that could affect the entities ability to perform in the future (AU sec. 341). According to Arens et al (2012) some of these event or conditions might be continuous operating losses, working capital problems, lost major client or not paying obligations on time. If that’s the case auditors should review managers plans to mitigate the consequences of these event and conditions and evaluate whether these plans can be effective (AU sec. 341). If the auditors still doubts an entities ability to continue as a going concern after reviewing manager’s plan, they should include an explanatory paragraph in the audit report to communicate this doubt (AU sec. 341). Because the role of auditors is continuously changing, auditors nowadays offer more than just audit services. Many audit firms provide a large range of other non-audit services like accounting and bookkeeping, tax, management consulting and forensic investigation services. In some cases the incomes from these non-assurance services might be a big part of the total fees the accounting firm receives from this client. If that’s the case auditors could be less willing to issue an unfavourable report such as going concern opinion when auditing client’s financial reports. It is in their better interest not to cause disagreements with these clients that bring large revenue to their firm.

Therefore it can be concluded that auditors who despite having these fees concern, are still willing to issue a going concern opinion when this needed are independent and perform high quality audits. DeFond et al (2002) also supports this taught. According to them many of those who believe that non-audit and audit fees could affects auditors independence do not consider the cost of auditor dependent behaviour (Defond et al 2002).They believe auditors are not willing to risk loss of reputation and litigations risk just to maintain a client.

Geiger and Rama (2006) also used propensity to issue going concern as an audit quality proxy. Geiger and Rama (2006) explain that healthy organisations that receive a going concern opinion (type 1 error) are very likely to leave the audit firm. On the other hand auditor who fail to issue a going concern report for a distresses company (type2 error) are risking both litigation and reputation consequences. Therefore auditor performing high quality audits will make sure that going concern error rates are low. In that sense willingness to issue a going concern and low going concern error is an indication of high audit quality

3.2 Degree of missing or beating targets

As stated before there is a separation of ownership and management in most large corporations. Thus the financial statements issued by managers are the primary source of information

shareholders have. However in most cases these financial statements are the basis for managerial bonus. Because managers are the ones making these financial statements they are able to

manipulate these numbers to satisfy their own interest without considering the interest of

shareholders and other stakeholders. Scott (2012) define earnings management as” the choice by a

manager of accounting policies or real actions, affecting earnings so as to achieve some specific reported earnings objective”. Scott (2012) identifies 4 patterns of earnings management. The first is

taking a bath were managers intentionally book a large cost in current year and thereby insuring a profit the next year. Managers may also manage earnings to lower the net income to reduce political attention when firms are highly profitable (Scott, 2012). In some cases managers engage in earnings management to smooth income and to receive constant compensation. However more often managers engage in earnings management to shift earnings upward and receive higher bonuses (Scott, 2012).

Auditors should be capable of detecting these maneuvers and prevent managers from managing earnings. Becker et al (1998) state that evidence from prior studies show that high quality auditors are better able to constrain earnings management. According to them audit reduces information

asymmetry between owners and managers by verifying the accuracy of financial numbers. They further state that the effectiveness and ability of audits to reduce earnings management depends on the quality of the auditor (Becker et al 1998). They argue that high quality auditors are better able to detect opportunistic earnings management, require adjustment or issue a qualified opinion.

According to them high quality audits reduces earnings management because managers know that detection of earnings management results in damaged reputation and reduced firm value (Becker et al 1998). Therefore Becker et al (1998) predicts those firms that have low quality auditors are better able to manage earnings. Furthermore research studies have concluded that auditors who specialize in certain sectors perform higher quality audits. Krishnan (2007) contributes to the high quality audit – low earnings management argument by stating that auditors who specialize in a certain industry are better able to constraint earnings management than non-specialist auditors. Thus if we assume companies degree of missing or beating targets as the degree of earnings management, then rate of missing and beating targets can also be used as an audit quality proxy.

Accounting rules and guidelines allow for some managerial discretion mainly in the recognition of accruals (Dechow 1994). According to Dechow (1994) managers can use this discretion to signal private information to the outside world because managers have superior information about a company’s operations. However this discretion can also be used opportunistically by managers to manage earnings (Dechow 1994). As sad before the ability of auditors to constrain opportunistic earnings management of managers depends on the quality of the external auditors. In their study Carey and Simnet (2006) measured the quality of financial statement by measuring to what extent auditors allow their clients to manage earnings. Thus auditors who allow more managerial discretion enable earnings management and therefore results in lower audit quality.

4. Hypothesis development

The SEC and other regulatory authorities around the globe require mandatory auditor rotation to assure auditors independence and improve audit quality. However some believe that longer audit tenure which results in increased client familiarity actually improves audit quality. Auditors who have long standing relationships with their clients have more knowledge of a client’s business model, operations, and industry. Therefore they are able to better understand a client’s risks and better

assess whether these clients can continue as a going concern during periods of financial distress. On the other hand as stated before in some cases auditors may be reluctant to issue a going concern report for distressed companies because they do not want to cause disagreements with managers. This is because these managers may be large clients who provide a large part of the firm’s total income. It is also believed that this familiarity with audit client may impair auditor’s independence. Advocators of audit rotation claim that long audit tenure reduces audit quality because auditors are less sharp, rely too much on past years result and are more easily manipulated by managers. Therefore if audit quality is measured as auditor’s propensity to issue going concern opinion I hypothesize that:

H1a: there is a negative relationship between audit tenure and propensity to issue going concern report when companies are distressed in the period between 2004-2007.

H1b: there is a negative relationship between audit tenure and propensity to issue going concern report when companies are distressed in the period between 2008-2011.

In many listed companies in the U.S. and around the world there is a separation of ownership and management. Unfortunately there is also a conflict of interest between owners and managers. Because managers are running the day to day business and have more knowledge of the operations then the actual owner, managers are able to manipulate the earnings to satisfy their own interest. According to Becker (1998) auditors who perform high quality audits are better able to constrain and detect earnings management. As stated before long audited tenure enables auditors to better understand their clients business, industry and risks. Meaning that they are better able to detect opportunistic earnings management of managers. Auditors that have long standing relationships with clients are better able to distinguish manager’s discretion to signal private information to the outside world and discretion that is used to opportunistically manage earning upwards to satisfy managers own interest. However if auditors rotate every few years there is a continuous “fresh look” from other auditors who can detect errors and misstatement that the previous auditor missed. Furthermore long audit tenure can result in client and fee dependency with the consequence of auditors overlooking certain things just to keep their clients. Meaning that they might allow too much managerial discretion without requiring readjustments when this is necessary. If audit quality is measured as audit degree audit clients just beat earning benchmark and degree audit clients just miss earning targets I hypnotize that:

H2a: There is a positive correlation between audit tenure and just beating earnings target for the period between 2004-2007.

H2b: There is a negative correlation between audit tenure and missing earnings targets for the period between 2004-2007.

H2c: There is a positive correlation between audit tenure and just beating earnings targets for the period during the financial crisis 2008-2011

H2d: There is a positive correlation between audit tenure and missing earnings targets for the period between 2008-2011

5. Methodology

5.1 Sample

To analyze the effect of audit tenure and audit rotation on audit quality, data of U.S. enlisted companies in the financial sector were collected. This study contributes to existing literature by comparing the effect of audit tenure on audit quality before the financial crisis and during the

financial crisis. To separate this research from what already has been done only the financial sector is analyzed. The comparison between the period before the financial crisis and during the financial

Crisis is done by collection data of the period between 2004-2007 and data from 2008-2011. The timespan 2004-2007 is chosen because this is the period right before the financial crisis and 2008-2011 is chosen because it is right in the middle of the financial crisis. Data about audit firms was collected from Audit Analitics. In this data base information was collected regarding audit fees, non-audit fees, non-audit opinions, current non-auditor, previous non-auditor last event date and operating cash flow. Compustat provided information for the rest of the variables for both models

To only collect data of companies in the financial sector I used NAISC codes between 52 and 525990 which were manually inserted. Using the ticker code of the companies compustat provided, data from Audit Analytics was collected. After merging information of both databases most companies without a complete set of data were dropped. Because this study only analyses the financial sector the sample size is relatively small. Because both Audit Analytics and Compustat in some cases had limited data of important variables information about these variables was collected manually for a

few companies. This was mainly for information about the IPO date, current auditor, total assets, total liabilities and market value. This was done primarily to prevent further reduction of the sample size. At the end there were a total of 2469 observations.

5.2 Research design

To analyses the effect of tenure on audit quality quantitative research will be done. In this study propensity to issue going concern and degree of meeting or beating targets will be used as proxies for audit quality. Once the statistical results are obtained, a comparison will be made during the two time periods analyzed to determine if long audit tenure has a different influence on audit quality during the financial crisis. In this paper I followed Carey and Simnet (2006) model who themselves followed Defond (2002) model to measure auditors propensity to issue going concern report .To measure degree of just meeting or beating profit benchmark I also use Carey and Simnet (2006) who themselves also followed Menon and Williams(2004.). Long audit tenure is measured by identifying auditor-clients relationship longer than 7 years and short audit tenure are relationships shorter than 2 years. Contrary to the Carey and Simnet (2006) model I removed the market adjusted return (RETURN) control variables because of the difficulty to reliable calculate this variable. I also removed the mining control variable they used to control for the large amount of mining companies in Australia. Because I’m only analyzing financial service companies in the U.S. this control variable will not have that much meaning.

5.2.1 Propensity to issue going concern report model

The following logistic regression model will be used to statistically analyze auditor’s propensity to issue going concern reports :

“OPINION=b0 + b1PBANK + b2SIZE + b3AGE + b4LEV + b5CLEV + b6RETURN + b7LLOSS +

b8INVESTMENTS + b9AUDFIRM + b10FEERATIO + b11CFFO + b12 TENURE≤2 + b13TENURE>7 + ᶓ Experimental Variable:

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 SIZE = natural logarithm of total assets of the company at financial year-end AGE = natural logarithm of number of years since listing on the U.S.

Stock Exchange

LEV = total liabilities divided by total assets CLEV = change in LEV during the year

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

INVESTMENT= short- and long-term Investment securities (measured s 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

TENURE<2 = 1 if audit partner is engagement partner on a client company for two Years or less, and 0 otherwise.”

Carey and Simnet (2006.p660):

The dependent variable in this model is OPINION. 1 is for if company did receive a going concern opinion and 0 otherwise. This variable was identified using Audit Analytics “going concern” variable. For this dummy variable the database showed 1 if that particular company in that particular year received a going concern opinion and 0 otherwise. So it was simple to identify which companies in which year received a going concern report. This model also has a range of control variables. The first control variable is PBank which is the probability of bankruptcy which was measured using the adjusted Zmijewsku score: -4.803 -3.6(net income/total assets) + 5.4(total debt/total assets) -

0.1(current assets/current liabilities).As stated before according to AU sector 341 auditor when doing their annual audit must assess the company’s ability to continue as a going concern and report on this. This is done by analyzing events, conditions and managerial plans that could lead to auditors to believe there is substantial doubt about a company’s continuance (AU sector 341). Therefore it is expected that companies with high probability of bankruptcy are more likely to receive a going concern opinion. The Second variable is SIZE which is measured as the natural logarithm of total assets. According to Carey and Simnet (2006) this control variable is included because larger

companies are less likely to become financially distressed and have better abilities to negotiate with auditors. Jackson et al (2008) also believe that firms who are larger in size have more opportunities to recover from difficult times. Therefore larger companies have a lower probability of receiving a going concern opinion. However as stated before Francis and Reynolds (2001) did not find any evidence that big five auditors tend to give bigger clients more favorable reports. Another control variable is AGE. This variable is measured as the natural logarithm of the total years companies are listed. According to Carey and Simnet (2006) this variable is included because young companies are more likely to file for bankruptcy and therefore have a higher chance of receiving a going concern opinion. According to Knechel and Vanstrealen (2007) larger companies have already shown their ability to operate and are therefore less likely to receive a going concern due to auditors doubt of the company’s ability to continue as a going concern. For these reasons a negative coefficient is expected for variable AGE. The company’s total liabilities in relation to their total assets may also be an

important factor affecting an organization’s ability to continue as a going concern. This variable is measured as total liabilities divided by total assets. This range varies between 0 to 1 and a lower LEV

is usually preferable. A higher range would mean that a company is highly leveraged and therefore would have a larger credit risk. This would result in company’s having more difficulty obtaining proper finance and therefore make their operations less stable. As a result a positive coefficient is expected. For similar reasons a negative coefficient is also expected for variable CLEV. Carey and Simnet (2006) define CLEV as the change in LEV. This variable was measured by calculating the difference between current and previous year LEV. According to Carey and Simnet (2006) this variable measures how the liabilities to assets ratio are changing which could be an indication that the company is obtaining unhealthy levels of debt. Whether the company is profitable or not could also affect auditor’s likelihood of issuing a going concern opinion. Therefore it is not surprising that a positive coefficient is expected for the variable LLOS. Carey and Simnet (2006) use the variable INVESTMENT as a measure of liquidity. Bryan et al (2002) predicted that companies with high solvency risk and high liquidity risk are less likely to emerge from bankruptcy. We therefore also expect a negative coefficient for this variable. If companies are not able to raise sufficient cash when needed they may have a higher risk of failure. Jackson et al (2008) also believe that companies with large investments are less likely to receive a going concern report. Furthermore there is considerable evidence in the audit literature concluding that audit quality of big4 or big6 firms is higher than those of non-big 6 audit firms. As stated before this is mainly because big6 audit firms have larger

reputation risk and litigation risks( deep pocket theory).Therefore we would expect auditor from big 6 firm to be more likely to issue a going concern opinion when this is required. Thus the control variable AUDFIRM is expected to have a positive coefficient. However auditors also perform non audit services. Sometimes the fees for these audit services are a large portion of the total fees collected by auditors. Evidence in the literature has suggested that non audit services may impair auditor’s independence (Carey and Simnet, 2006). However Svanstroom (2008) concluded that the provision of non-audit services does affect auditor’s independence. In this study however I expect the coefficient of the FEERATIO variable to be negative. Control variable CFFO which is measured by dividing total cash flow from operations total assets is also expected to have a negative coefficient. Companies with negative operating cash flow are more likely to fail and therefore more likely to receive a going concern opinion. Finally this model also controls for short audit tenure because it is believed that audit quality is lower during the first few years of audit tenure (Carey and Simnet, 2006). Myer (2003) concluded that restatements are more likely to occur during the early years of audit tenure. This coefficient is also expected to be negative.

5.2.2 Just meeting and beating earning benchmark model

The following logistic regression model was used to calculate degree of just beating or missing earnings benchmark:

“Y=b0 + b1AUDFIRM + B2PRANK + B3SIZE + b4AGE + b5MKTVAL + b6MINING + b7TENURE≤2 + b8TENURE≥7+ᶓ”

where Y is defined in four different models as Pr(BEATS_BE = 1), Pr(MISSES_BE = 1),Pr(BEATSLYR = 1), and Pr(MISSES_LYR = 1).

Control Variable:

MKTVAL = the natural logarithm of the market value of equity Carey and Simnet (2006, p.669)

For this analysis there are four Dependent variables and therefore resulting in four logistic regression models. According to Carey and Simnet (2006) logistic regression model the first dependent variable BEATS_BE= 1 when net income is lower than two percent of total assets and 0 otherwise. The second dependent variable MISSES_BE=1 when loss is less than two percent of total assets (Carey and Simnet 2006). Dependent variable number three BEATS_LYR= 1 if increase in profit or decrease in losses compared to last year’s profit or loss is lower than two percent of total assets. Finally the fourth dependent variable MISSES_LYR= 1 if decrease in profit or increase in loss compared to last year’s profit or loss is less than two percent of total asset (Carey and Simnet 2006).

I assume that clients of low quality auditor are able to manage earnings upward to either beat an earnings benchmark or avoid missing an earnings benchmark. If audit tenure reduces audit quality we would expect the coefficient of variable TENURE>7 to be positive for both Y=Pr(BEATS_BE) and Y= Pr(BEATS_LYR) because in that case auditor would allow clients more discretion as length of the relationship increases. Meaning that audit clients that have long audit tenure are allowed to manage earnings upwards to beat earnings benchmark. On the other hand we would expect a negative coefficient for the variable TENURE>7 for both Y=Pr(MISSES_BE) and Y= Pr(MISSES_LYR) if audit tenure reduces audit quality. In that case auditor would allow their long standing clients to manage earning and therefore result in a lower frequency of missed benchmarks.

For this model similar to Carey and Simnet I will use some of the control variables that were used for the OPINION model. Control variable AUDFIRM is included in the model because it is believed that high quality auditors are better able to constrain earnings management then lower quality auditors. Becker et al (1998) conclude that high quality auditors are better able to detect managed earnings and prevent their clients to report these numbers. For this reason a negative relationship is expected

between AUDFIRM and Y for the just beats analysis and a positive relationship for the "misses benchmark" analysis. Control variable SIZE is also included because according to Nelson et al (2002) the probability of auditors not requiring an earnings adjustment increases with size of the company. This is because as the company size increases so do the audit fees( Nelson et al 2002). Nelson et al (2002) ads that larger clients usually have more resource to both structure transactions and defend aggressive positions. Similar to the SIZE variable the AGE control variable is also included because older companies are usually larger in size, offer higher audit fees and auditors could therefore be more inclined to allow too much managerial discretion to these clients. TENURE<2 is also included due to believes that audit quality is lower during the first few years of the auditor-client relationship (Carey and Simnet 2006). PBANK is included as control variable because companies with high probability of bankruptcy are more willing to manipulate earnings to decreases loss or increase earnings. Therefore a positive coefficient is expected for the just beat benchmark analysis and a negative coefficient for the missed analysis. Finally Carey and Simnet(2006) also included the market value variable in their model. Similar to their study this variable will also be included in the model for this study. This model is included because companies that have high market value have larger incentives to keep beating earnings benchmark and avoid missing earning benchmark to therefore maintain or increase their market value. We therefore expect the coefficient for this variable to be positive for the just beat analysis and negative for the misses benchmark analysis.

6. Results

6.1 Propensity to issue going concern reports

The descriptive statistics of the sample of period 2004-2007 for the going concern proxy are reported in table 1. Out of the sample of 602 financial institutions analyzed less than 1 percent had an auditor-client relationship shorter than 3 year. Almost 1 percent of the audit opinions given in this sample included a going concern paragraph and 84 percent of the auditors have been with their current clients for longer than 7 years. Furthermore 7,5 percent of the companies in the sample reported a loss between 2004-2007 and only 39 percent of the companies were audited by big 6 audit firms. Finally the average total assets of these institutions were $6309 million and the companies have been listed 15 years on average.

Table 2 presents the results of the logistic regression of the period 2004-2007using Carey and Simnet (2006) model to measure audit quality by determining auditors propensity to issue going concern opinion. The model has an adjusted Rsquare of 0.26287 meaning that 26 percent of going concern opinions issued is explained by this model. Consistent with both Defond (2002) and Carey and Simnet (2006) the coefficients LLOS and INVEST is significant and in the predicted direction at p <0.05. This means that reporting losses and liquidity issues significantly affect probability of receiving a going concern opinion. Similarly with the Defond 2002 study the coefficient of PBANK is statistically significant in the predicted direction. This means that for the sample used in this study the adjusted Zmijewsku probability to go bankrupt significantly affect the likelihood of receiving a going concern opinion . However the CLEV and LEV variables also significantly affect the variation of the dependent variable. These variables are in the predicted direction. Thus unhealthy levels of debt does influence auditors believes about organization’s ability to continue as a going concern and therefore also the likelihood of receiving going concern opinion from the auditors. However consistent with Defond (2002) and Carey and Simnet(2006) the coefficient levels of variables SIZE and AGE are not significant at P<0.05 or at p<0.1. This means that consistent with other studies organizational size (log of total assets) and age (log of total years listed) are not strong indicators of probability auditors will issue a going concern opinion. Consistent Carey and Simnet (2006) the coefficients of AUDFIRM,FEERATIO and CFFO are not statistically significant at P<0.05 even though they were all in the predicted direction. Thus consistent with prior literature on auditor’s independence, non-audit service and non-audit fees which concluded that non-audit fees do not influence auditors likelihood of issuing a going concern opinion. Furthermore even though negative operating cash flow is a strong indicator

of a company being financially distressed neither this study nor Defond (2002) nor Carey and Simnet (2006) found this coefficient to be statistically significant. The Results in table 1 show that the coefficient of the AUDFIRM variable is not significant at p<0.05. However it is in the predicted direction. Finally the TENURE <2 variable is negative and both insignificant. Most importantly the coefficient of the variable TENURE>7 is negative but insignificant. Meaning that the results of this study does not provide sufficient evidence to conclude that long audit tenure results in lower

propensity of auditors to issue going concern report during the period right before the financial crisis (2004-2007). No evidence that long tenure results in lower audit quality. Therefore hypothesis 1a is rejected

The descriptive statistics of the sample of period 2008-2011 for the going concern proxy are reported in table 3. Out of the sample of 693 financial institutions analyzed 1,8 percent had an auditor-client relationship shorter than 3 year while 67 percent of the auditors have been with their current clients for longer than 7 years. In addition 1,2 percent of the audit opinions given in this sample included a going concern paragraph and Furthermore 25 percent of the companies in the sample reported a loss between 2008-2011 and only 45 percent of the companies were audited by big 6 audit firms.

Finally the average total assets of these institutions were $7253 million and the average age of these companies is 13 years.

The result of the logistic regression for period 2008-2011 which is the period in the middle of the financial crisis shown in table 4 are not completely consistent with the results of the previous regression. The results of this regression show that again consistent with Defond (2002) and Carey and Simnet (2006) the coefficient of variable LLOS is significant in the predicted direction at p<0.05. Also consistent with Carey and Simnet (2006) is the coefficient of variable LEV, which measures a company’s debt position, is significant at p<0.05. Similar to the regression for the period before the financial crisis the coefficient for LLOS is significant and positive as predicted at p<0.05. Meaning that both in the period before and in the financial crisis companies that book losses are more likely to receive a going concern opinion. However the coefficient of variables SIZE,AGE,CLEV,INVEST,

AUDFIRM,FEERATIO,CFFO are not significant. Meaning that for the period in the financial crisis being audited by a big6 or non-big6, having high non audit fees and being large in size did not affect the probability of auditors issuing a going concern report. However a remarkable difference in the regression models is the as Carey and Simnet (2006) called it the experimental variable TENURE >7. In the regression analysis for period 2008-2011 the variable TENURE >7 is both significant and positive. This contrary to Carey and Simnet(2006) who had a negative coefficient for this variable, means that auditors are more likely to issue a going concern report for clients with longer TENURE in that period. Meaning that according to this model audit quality increases when clients and auditors

have longer relationships. This result is consistent with those who believe that longer familiarity with the clients results in higher audit quality. This result also suggests that audit quality measured as propensity to issue going concern increases in the period during the financial crisis for those auditors with long audit tenure. This might be because auditors and clients were getting more attention from the financial press and more closely supervised by financial authorities. Auditors may have had larger reputation and litigation incentives to perform higher quality audits at that time.

Therefore this this analysis does not provide evidence to conclude that audit tenure reduces audit quality. Hypothesis 1b is rejected.

6.2 Degree of meeting and missing benchmark

Table 5 shows the descriptive statistics of the sample of period 2004-2007 for the earnings management proxy. The table shows that 70 percent of the time, the companies observed in this sample did beat their current year earnings benchmark while 4 percent missed their current year earnings benchmark. The table also shows that during the period between 2004-2007 59 percent of the observed sample did beat their breakeven (last year’s profit) while 36 percent missed it.

Furthermore 1,1 percent of the financial institutions had an auditor-client relationship shorter than 3 year and 85 percent of the auditors have been with their current client for longer than 7 years. The

market average value of the observed firm is $1743 million and only 38 percent of the companies were audited by big 6 audit firms. Finally the average total assets of these institutions were $6240 million and the companies have been listed 16 years on average.

Table 6 present the results of the logistic regression of Carey and Simnet (2006) model to measure the extent financial service companies meet or misses earning benchmark for the period before the financial crisis (2004-2007). This model shows that consistent with Carey and Simnet (2006) the coefficients for control variables PBANK, SIZE, MARKET VALUE and AUDFIRM are significant at p<0.05 and AGE is significant at p<0.1 for the “just beats analysis”. The coefficients of variables AGE and MARKETVALUE are both negative even though a positive coefficient was expected. This means that both firm’s that been listed on the stock exchange for a long time and firm with high market value do not tend to just meet earnings benchmark more often. In our study for the 2004-2007 “just beats

benchmark” analysis control variables PBANK and AUDFIRM are also significant in the predicted

direction at p <0.05. This means that as expected companies with higher probability of bankruptcy are more likely to manage earnings. This also means that consistent with a large scale of accounting literature big6 audit firms are better able to constrain earnings management. Finally the coefficient of variable TENURE>7 is negative, in the predicted direction, but insignificant.

For the” just misses benchmark” analysis of 2004-2007 none of the control variable are significant. Variable TENURE>7 is in the predicted direction but also insignificant. Therefore the logistic

regression of just beats and just misses model for 2004-2007 does not provide enough evidence to conclude that long audit tenure reduces audit quality.

The results for just “Beats breakeven last years” and just Misses breakeven last year for the period before the financial crisis (2004-2007) are similar and shown in table 7. None of the control variables are significant for either of the models. However even though the coefficient of variable TENURE>7 is not significant and not approaching significance, it is not in the predicted direction for the just beats

last year’s profit analysis. However the coefficient of TENURE > 7 is in the predicted direction but not

significant for the just misses last year profit analysis. Therefore the results of logistic regression that measures just meet or misses last year profit by more than two percent of total assets also concludes that there is not enough evidence to support the argument that longer audit tenure results in lower audit quality. Therefore hypothesis H2a and H2b is rejected.

Table 8 shows the descriptive statistics of the sample of period 2008-2011 for the degree of meeting or beating earnings benchmark proxy. The table shows that 55 percent of the time, the companies observed in this sample did beat their current year earnings benchmark while 14 percent missed their current year earnings benchmark. The table also shows that during the period between 2008-2011 54 percent of the observed sample did beat their breakeven (last year’s profit) while 43 percent missed it. Furthermore 1,8 percent of the financial institutions had an auditor-client relationship shorter than 3 year and 69 percent of the auditors have been with their current client for longer than 7 years. The average market value of the observed firm is $1383 million and only 38 percent of the companies were audited by big 6 audit firms. Finally the average total assets of these institutions were $10018 million and the companies have been listed 14 years on average.

The results of the logistic regression analysis for the models “just beats benchmark” and “just misses

benchmark” for the period during the financial crisis (2008-2011) are shown in table 9. The

coefficients of SIZE, MARKET VALUE and AUDFIRM are significant for the just beats model in the predicted direction at p<0.05 . Only the coefficient of MARKET VALUE is not in the predicted

direction. The coefficient of TENURE>7 for the” just beats benchmark” is positive at p <0.1 and in the predicted direction. For the” just misses benchmark” analysis only the coefficient of SIZE and

AUDFIRM are significant at p<0.1. The coefficient of TENURE>7 is not significant but in the predicted direction. Therefore the logistic regression for the model “ just beats benchmark “ providence evidence to conclude that audit tenure results in companies being less likely to beat earnings benchmark. These results provide evidence that auditor with long audit-client relationships are better able to constrain earnings management and therefore increase audit quality. The model “just

misses benchmark” has an insignificant coefficient for TENURE >7 and therefore does not provide

evidence that long audit tenure results in reduction of audit quality.