The effects of voluntary auditor rotation on audit quality : is there support for mandatory audit rotation proposed by the European Commission?

(1)

The effects of voluntary auditor rotation on audit quality.

Is there support for mandatory audit rotation proposed by the European

Commission?

Katarzyna Ożugowska Master Thesis in

10604669 Accountancy and Control

Supervisor: Alexandros Sikalidis Track: Accountancy 14th August 2014

(2)

2

Abstract

After the financial crisis, the world started to think of regulations which would improve the quality of audit in order to strengthen the independence and performance of the audit profession which would eventually aim to avoid further crises in the future. The European Commission presented proposals which should solve the issues. The project involved the introduction of mandatory audit firm rotation. This measure is highly controversial and there is no academic and political unanimity as to whether introduction of this measure would bring favourable effects or not. The hypotheses presented in this paper argue that the quality of audit will decrease after the change of audit firm. To illustrate that audit quality is represented by the level of real earnings management. As there are two main legislative environments in Europe – common-law (represented by the United Kingdom) and code-law (represented by Germany) settings, this paper investigates whether legal environment has significant effect on the use of real earnings management. Empirical research does not give significant results which would confirm the decrease of earnings quality after the change of auditor. Even though the differences between the countries were observed, the regression analysis showed that they are not significant.

(3)

3

1. Introduction

1.1. Background

The auditor’s opinion is an important source of information regarding the content of financial statements prepared by companies. The auditor is responsible for ensuring all stakeholders and shareholders that financial statements prepared by the company provide true and fair view of its financial position. In order to be fully informative, the auditor’s report has to be of high quality. Audit is believed to be of high quality if there is “a joint probability that a given auditor will both discover a breach in a client’s accounting system, and report the breach. Hence, this chance to detect and report the material misstatement depends on two things – auditor’s ability to detect the misstatement and his independence”. (DeAngelo, 1981, p. 186). It is also argued that bigger companies provide higher quality audit as no single customer is too important to lose or because they have more reputation to lose in case of a failure (DeAngelo, 1981).

Ideally the companies should be willing to improve the quality of financial statements as much as possible. In such a case, the switch of the auditor should be caused only by the desire to provide more credible information confirmed by higher quality audit. (Knechel et. al, 2007). However apart from above mentioned reasons researchers also identified that companies may have incentive to switch auditor if they want to hide their poor performance and obtain an unqualified opinion (Chow and Rice, 1982).

The auditor change is an important decision. Even if the company wants to improve the information content of its financial statements it may not so happen after auditor switch. Due to complex transactions and industry specific legal and procedural setting, it may take some time for the auditor to properly understand processes happening in the company. The AICPA (1992) found in their study that the risk of audit failures is increased because auditors initially do not have sufficient knowledge of the client’s business. Moreover, as mentioned above bigger audit companies offer higher quality audits (DeAngelo, 1981). However, there is a threat that even if newly appointed audit company is bigger (e.g. one of the Big 4), audit quality may not necessarily increase. This may be a result of for instance simultaneous provision of non-audit services, which is believed to cause a decline in auditor independence. The change of the auditor (voluntary as well as mandatory) can have positive effects too. A new auditor may be free from bias, have a fresh look and find deficiencies that were previously overlooked. Furthermore, auditor change can help to prevent potentially harmful

(5)

5

too close contacts between auditor and client’s managers which can pose a threat to audit quality. In order not to lose customer, the auditor may be more lenient towards recognized weaknesses of financial statement.

Since the outburst of accounting scandals such as WorldCom, Enron and Arthur Andersen at the beginning of XXI century there has been a debate around the world about what should be done to prevent such events in the future, how to re-build lost trust in audit companies and finally how to ensure high audit quality. One of the proposed solutions introduces concept of mandatory auditor rotation which defines maximum number of years certain auditor/audit company can examine financial statements of one particular customer. Proponents claim that mandatory audit firm rotation will lead to auditors being more independent and thorough. This concept has already been introduced in a few countries, including Italy, Brazil, South Korea, Singapore or Malaysia (Lim and Tan, 2010). The heated debate upon the adoption of this concept is still on, both in the US and the EU. In order to preserve auditor independence and increase investors’ utility of financial reports the 8th EU Company Law directive requires rotation of audit partners every seven years. Apart from that the European Commission is about to introduce mandatory audit firm rotation every 10 years. The project has been accepted in December 2013, now it has to be accepted by national governments and voted over in the European Parliament (EC 2013).

The remainder structure of the paper is as follows. Chapter 1 will be completed with the description of research question followed by motivation for this study. Chapter 2 will focus on overview of prior literature on audit quality and the development of mandatory audit firm rotation in the European Union together with arguments supporting and opposing the idea. Chapter 3 will discuss details of research design, specifying model and the sample used. Chapter 4 will present and discuss empirical results. Finally chapter 5 will give conclusions drawn from the whole research.

1.2. Research question

Based on the theory presented in prior literature, this study examines the influence of voluntary auditor rotation on the audit quality. It aims to discover whether in general companies switch to higher quality auditor in order to investigate whether proposal to introduce mandatory auditor rotation issued by the European Commission can be justified. As mandatory audit rotation is a controversial proposal in many EU countries, it is worth

(6)

6

examining if now, when companies change auditor voluntary, they do it to improve audit quality or to mitigate their deficiencies and still obtain unqualified opinion.

Mandatory audit rotation is perceived to be different from voluntary rotation (Ewelt-Knauer et. al 2013), however voluntary audit rotation and its impact upon audit quality can serve as a good outline as to what kind of effects of introducing mandatory audit rotation can be expected. If in general, the companies change auditor to the one of higher quality, then, if forced, they will be looking for an auditor of at least similar quality. Therefore the research question is as following:

“The effects of voluntary auditor rotation on audit quality.

Is there support for mandatory audit rotation proposed by the European Commission?”

1.3. Motivation of this study

A lot is written about the impact of auditor change on investors’ and thus stock market reactions on the information about auditor change. Also, there is significant amount of papers analyzing the effect of auditor tenure on audit quality. However, I haven’t found much regarding the impact of auditor switch during first years after the change and empirical investigation of reasons for the change. I would like to see if companies really aim at using services of higher quality audit firm. Therefore I would like to examine the influence of auditor change in first 3 years after the change compared with the quality of audit before the change.

Most of prior studies used the accrual-based earnings management as a proxy for audit quality (Myers et al. 2003, Kothari 2005, Lim and Tan 2010), however this type of earnings management can be limited not only by high quality audit but also by enhanced regulation. Over the last years there has also been developed a model using real earnings management which can shed new light on the auditing literature. As it is not widely used model yet, I would like to implement it. The way of managing the earnings presented in the model is harder to detect and is believed to be a substitute of accrual-based earnings management in case when manager’s ability to use the latter way of earnings management is limited. Previous research also shows that real activity management can be more harmful for the firm’s long-term welfare. High quality auditor should be able to limit not only accrual-based earnings management but also real earnings management. I believe this model is suitable to

(7)

7

my research as it would challenge prior results regarding audit quality and therefore would create value added to the existing literature.

Furthermore, the new rules aimed at strengthening auditor’s independence including mandatory audit rotation have been recently approved by the Permanent Representatives Committee. New regulations include a maximum tenure of audit company to examine financial statements of a customer to be 10 years (EC 2013). However, even though the project has been approved, it still needs to be voted over in the European Parliament and by national governments.

Since there are two opposing points of view on the impact of auditor change upon audit quality (quality is improved or worsened), I would like to examine which one holds true for German and the UK setting. I want to see if companies from both countries switch audit firms to improve the quality of performed audits or to worse audit firms to obtain unqualified opinion even though their accounts show deficiencies. The results would show if there is a need and if the proposed idea can really improve audit quality, thus if the mandatory audit rotation should be implemented in the whole EU.

I also chose the two countries because of societal reasons. They are two big economies having big influence on the decisions made by the EU. Moreover, Germany is currently economically most powerful country in the EU. Therefore its lobbying power is strong. On the other hand, British capital, London is the biggest and most important financial centre in whole Europe, and second in the world, recently being overtaken by the New York Stock Exchange. Moreover, especially in recent years the UK has become more and more anti-EU which resulted in plans to conduct nation-wide referendum in which the citizens will be asked if the country should stay in, or perhaps leave the European Union. Thus, the choice of the countries is not accidental. They belong to contrasting institutional and economical settings. Germany follows the so-called continental approach with code-law regulations regarding accountancy, whereas the UK is a representative of the Anglo-Saxon approach with common law usage. This division is still visible even though both countries adopted a lot of common European legislation. Therefore this study will also focus on comparison of how the audit firm rotation is perceived in both legislation systems based on the biggest European representatives perceptions.

(8)

8

2. Literature review and hypotheses

This chapter will present relevant theories concerning audit quality and mandatory audit firm rotation with regard to previous literature. This will be followed by the introduction of regulatory regimes existing in both examined countries. Last section of the chapter will give thorough development and state hypotheses tested with the empirical model in further sections.

2.1. Audit quality

2.1.1. Definition of audit quality

The aim of the auditor’s report is to express the opinion on the reliability and accuracy of the content of financial statements provided by the company (ISA 200). Wallace (1985) presents in his book two approaches to audit demand – exogenous and endogenous. Exogenous demand comes from the outside of the economy in form of outside intervention on the economy as for instance government regulations or specific standards on auditing. On the contrary, endogenous demand comes from within the economy. Wallace (1985) based the endogenous demand for (high quality) audits on three theories – the agency theory, the information theory, and the insurance theory. The agency theory assumes that one or more principals (shareholders) engage another person as their agent (managers, executives) to execute some tasks/service (managing the company) on their behalf. Such a relation requires delegation of decision rights to the agent. In the ideal world the interests of principal and agent are the same, and thus agent acts in the best interest of their principal. Therefore, the principals do not bear any cost of monitoring. However in reality the cost of monitoring exists as according to the agency theory agents try to maximize their own wealth first. (Shapiro 2005) Hence, the demand for assurance that the agent acts in the best interest of shareholders comes from the market system - audit serves as an important tool to align interests (Wallace 1985). The information theory assumes that the audited information is regarded as more reliable as it has been verified by an external auditor. Therefore it works as a mean of reducing risks of investments or improving both, external and internal, decision-making process. Wallace (1985) also recognized trading profits. Audited information can cause either immediate reaction which leaves no possibility to make abnormal profits based on publicly available information, or no reaction in case when audit just confirms expectations regarding the reporting company. Finally, insurance theory relates to management’s litigation exposure. Auditors are jointly responsible for the fraud with

(9)

9

auditees. Therefore, managers choose to audit their statements to divide the risk of disclosure of frauds and deficiencies of audited and published financial statement. In case of disclosure of a breach in accounting system once audit report is published the managers profit from auditor’s “deep pockets” – due to inefficiency in their work auditors have to cover part of the costs of discovery of financial statements’ deficiencies.

As mentioned before audit quality is mostly defined with reference to DeAngelo’s (1981) definition, i.e. audit is believed to be of high quality if there is a joint probability that a given auditor will both discover a breach in a client’s accounting system, and report the breach. This implies high independence of auditor from client. Deis and Giroux (1992) also supplement this statement by saying that auditor should be resistant to customer’s pressure put on him to issue unqualified opinion.

The definition of audit quality is rather vague and hard to unanimously present with regard to quantitative models. Assurance of audit quality can be performed by the assessment of auditor independence which is regarded as key factor in provision of high quality audit services. The .01 Rule 102 – Integrity and Objectivity embedded in the Code of professional Conduct and Bylaws prepared by AICPA (p. 2909) states that: “In the performance of any professional service, a member shall maintain integrity, shall be free of conflicts of interest and shall not knowingly misrepresent facts or subordinate his or her judgments to others”. It is important that auditor even though hired by the customer provides services not biased by company’s preferences or wishes (for instance to obtain unqualified opinion) but issues opinion that reflects the true view of company’s financial statements. Regulatory frameworks define this kind of independence as independence in fact (European Commission 2002). However as Pott et al. (2009) mention, it is also important to secure independence in appearance. International Federation of Accountants (2004) defined it as avoidance of significant facts and/or circumstances shedding negative light on informed third party’s perception of a firm’s or auditor’s objectivity, professional scepticism or integrity. Suspicion that firm’s or auditor’s independence is impaired can undermine reliability of auditing and financial reporting is undermined and leads to destabilisation of financial markets. This happens due to the fact that independence in fact is hard to observe and evaluate (Pott et al. 2009). This statement is confirmed by Dart (2011) who stated that auditing profession is strongly dependent on the credibility of auditor’s independence. If there is no trust in auditor’s independence their work is useless to all shareholders and interested stakeholders.

(10)

10

Therefore audit quality with regard to auditor independence has been measured in several papers. One of the measures used was the willingness to issue qualified opinions. Knechel and Vanstraelen (2007) reported that less independent auditor has less incentives and is less willing to issue qualified opinions even though there are prerequisites to do so. However, they noted that such a situation happens in the absence of strict regulatory requirement to issue going concern opinion in predefined circumstances.

2.1.2. Audit quality and earnings management

One of the most widely used proxies for audit quality is measurement of the level of earnings management which occurs when: “managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting practices” (Healy and Wahlen 1999, p. 368). Most of the studies measure earnings management with regard to discretionary accruals following Jones model (Jones 1991), or its modified versions developed later on (Dechow et al. 1995, Kothari 2005, Dechow and Dichev 2002). All of those studies show that high level of accruals which cannot be explained by the normal operating activities (Scott 2011), i.e. discretionary accruals, implies lower level of audit quality as it is an indicator of earnings management (Becker et al., 1998).

Roychowdhury (2006) and other authors following him (Kim at al. 2012, Chi, Lisic, Pevzner 2011) claim that earnings can be also managed through real activities management. Therefore, Roychowdhury (2006) defines real activities management as actions that deviate from usual business activities which are aimed to meet or beat certain earnings levels. Actions that can be classified as real earnings management include temporary price discounts, more lenient credit conditions, smaller profit margins, reductions in investments. Such actions can have negative influence on firm value, especially future value. Real earnings management can decrease firm value as decisions made in current period aiming at increasing current earnings may result negatively on the future cash flow. As Roychowdhury (2006) pointed out, clients who are granted price discounts may get used to them and expect further reductions in the future, threatening to choose contractor if the firm tries to go back to normal prices. As a result lower margins will exist also in the future. Furthermore, reductions in investments restrict firm’s growth and development.

(11)

11

Roychowdhury (2006) claims that real earnings management and accrual-base earnings management are substitutes. Basing on this assumption Zang (2007) points out that real earnings management is pursued when the regulatory regime is strict and litigation risk is high. This is the result of real earnings management being hard to detect as well as of the fact that when properly disclosed, it does not involve any violation of law. As proved by Ewert and Wagenhofer (2005), when accounting standards are stricter, which is shown by lower flexibility of accounting regulations, firms tend to engage more in real earnings management. As per Roychowdhury (2006), even if real earnings management is more costly in the long term and decreases firm’s development, managers do not want to rely only on accrual-based earnings management. They believe that they will bear higher personal and short-term costs when engaged in accrual-based earnings management on too big scale. Graham et al. (2005) in their study confirmed that executives show higher willingness to employ real earnings management rather than accrual-based earnings management. As accrual-based earnings management is put under scrutiny by regulatory bodies, it is more likely that the auditor will detect attempts to manage earnings this way (Roychowdhury 2006). Therefore real earnings management enables manipulation with earnings with high chances of not being discovered and still achieving short-term goals like for instance meeting or beating analyst forecasts or obtaining annual bonus (Graham et al. 2005).

2.2. Mandatory Audit Firm Rotation

This section focuses on the concept of mandatory audit firm rotation. The description of the evolvement if the mandatory audit firm idea in the legislation of the European Union will be followed by an outline of arguments supporting implementation of this concept which will precede presentation of arguments opposing enforcement of mandatory audit firm rotation in the EU.

2.2.1. Mandatory Audit Rotation in the EU

After accounting scandals such as Enron and Arthur Andersen which emerged at the beginning of the 21st century and recent financial crisis the European Union started to work on new regulations that would both strengthen the role of auditor and improve audit quality. Even though it is believed that auditing sector had “good crisis”, as the blame was put more on banks, rating agencies or financial regulations, there have appeared doubts of how could it happen that big institutions like Lehman Brothers bank in the USA which obtained unqualified audit opinions went bankrupt soon after receiving such an opinion (Humphrey et

(12)

12

al. 2011). According to the European Commission auditing is an essential element to establish trust and market confidence (Quick 2012). Therefore, the Commission decided to prepare new regulations which would strengthen already existing laws regarding auditing, auditor’s role as well as improve audit quality (Quick 2012).

At the moment, there are three directives regulating the audit market:

1.) The Fourth EU Directive from 1978 (78/660/EEC) implementing the requirement that companies have to have their financial statements audited by at least one person who is accredited by national law to perform audit,

2.) The Seventh EU Directive from 1983 (83/349/EEC) extending the requirement to consolidated statements, and

3.) The Statutory Audit Directive from 2006 (2006/43/EC) replacing former Eighth EU Directive aiming at improvement and harmonisation of audit quality by for instance implementing International Standards of Accounting as EU audit standards, or measures for independence and competence of auditors. It is worth noting that the Eighth EU Directive from 1984 (84/253/EEC) already introduced the concept of mandatory audit partner rotation – setting maximum period as 7 years of service. This rule has been sustained in the Statutory Audit Directive.

Green Paper: “Audit Policy: Lessons from the crisis” (EC 2010) prepared by Michel Barnier, the Commissioner for the Internal Market and Services, and issued on 13th October 2010 was an EU answer to growing doubts regarding audit quality. Furthermore, this was the first EU proposal including the implementation of mandatory audit rotation. The proposal, issued on 30th November 2011 (EC 2011), addressed the threat of familiarity as the one which can heavily impair audit quality. In order to minimise this risk the European Commission intended to introduce mandatory audit firm rotation after a period of 6 years of service which could be extended to 8 years in pre-specified rare conditions. In case of performing joint audits the duration of the engagements could be extended to 9 years, whereas under exceptional conditions the engagement could be extended up to 12 years. Furthermore, the European Commission wanted to introduce a so-called “cooling off” period which can be defined as a period before the audit firm can re-engage with the same customer.

The Green Paper (EC 2010) was accepted by the European Commission which started further development of the project resulting in approval of the preliminary agreement on the 17th December 2013 (EC 2013) published in a statement issued by Commissioner Barnier. Newly

(13)

13

accepted measures were not as ambitious and strict as proposed in the Green paper (EC 2010), yet the concept of mandatory audit firm rotation has been sustained. With regard to audit quality and mandatory audit rotation it was accepted that:

1.) A maximum period of 10 years during which a member state may allow an audit firm to continue auditing the same public-interest entity

2.) If the engagement is put out for public bid, the engagement can last for a maximum of 20 years

3.) In cases of joint audit, member states can extend the maximum tenure up to 24 years. In order to be effective, new rules still have to be voted over by the European parliament and national governments.

2.2.2. Arguments supporting mandatory audit firm rotation

Wallace (1985) recognised three sources of endogenous demands for audits 1) agency theory, 2) information theory and 3) insurance theory. These theories explain not only the demand for audit itself but also the demand for high quality audit. Low quality audit is not a solution to issues outlined in the three theories, it may even contribute to intensification of such problems. Thus, ideally firms switch audit firm to ascertain higher quality of financial statements from which they derive information whether managers did their best to represent shareholders’ interests therefore reducing agency costs or ensuring that information provided reflects company’s true and fair view. However, still some companies are switching auditor to the one of lower quality so that their actions mitigating deficiencies are not discovered. Lim and Tan (2010) claim that the auditor tenure is related to greater acquired expertise i.e. better understanding of customer’s business processes and risks. Yet, longer tenure tends to be also associated with reduced ability to spot deficiencies as auditors are getting too familiar with the client. Fear of too big familiarity causing the decrease of auditor’s professional scepticism and in turn lower quality of audit report were one of the main reasons why European Commission proposed to introduce mandatory audit firm rotation (EC 2010). Furthermore, Lu and Sivaramakrishnan (2009) claim that newly appointed auditor can bring fresh look and perspective to the audit approach. Deep analysis of business processes may bring their attention to issues that were omitted by the previous auditor but are important in appropriate risk assessment and evaluation of company’s statements. Ewelt-Knauer et al. (2013) recognize another threat connected with longer auditor tenure, namely routine. Excessive reliance on previous years’ working papers and templates diminishes auditors’

(14)

14

attention to details and may be a cause of suspicious transaction being overseen and thus not examined. As a result a potential fraud may not be discovered nor reported, which decreases the quality and usefulness of the auditor’s report. Another issue concerning longer audit firm tenure mentioned by Lim and Tan (2010) is the auditor becoming less independent because of the pressure to retain customer. According to Ewelt-Knauer et al. (2013) this may result in auditors conforming to customer’s demands in order to please them and thus secure revenues from future audits. Mandatory audit rotation therefore is one of the perceived remedies to the problem. It decreases the threat of routine and lower auditor independence due to their familiarity with the company. Moreover the reliance on long-term customer is lower, thus the auditor does not feel the pressure of losing the client in case of not conforming to their requests to mitigate the fraudulent actions.

Moreover, mandatory audit rotation is perceived to positively influence market concentration and market participation of smaller audit companies. Ewelt-Kanuer et al. (2013) mention that there will be possibilities for smaller audit firms to grow and benefit from forced changes of auditor. Green Paper (EC 2010) assumes that the development of the competition between audit firms will positively influence audit market. The European Commission desires to limit the Big 4 concentration as at the moment high concentration on Big 4 services constitutes a so-called “Big 4 bias” (Humphrey et al. 2011). Humphrey et al. (2011) define Big 4 bias as client’s limited choice of auditors because of the doubts that the capacities and capabilities of Non-Big 4 audit firms are underestimated or not recognised. European Commission in its Green Paper (2010) also draws attention to the fact that lower market concentration will decrease the risk of market failure in case of any market shocks following the possible future collapse of one of the Big 4 firms.

2.2.3. Arguments opposing mandatory audit firm rotation

Basing on previous sub-chapter it may seem that mandatory auditor change has many positive effects and should be regarded as positive phenomenon/regulation, however prior literature recognizes at least as many drawbacks of this conceptions as advantages.

To start with, the papers of Myers et al. (2003), Lim and Tan (2010) show that the longer auditor tenure, the higher audit quality due to greater acquired expertise, i.e. familiarity with business processes and risks encountered. Newly appointed auditor needs some time to acquire company/industry specific knowledge, and understand organizational structure as well as occurring processes which heightens the risk of audit failure. The results of Carcello

(15)

15

and Nagy (2004), Daniels and Brooker (2011) or Johnson et al. (2002) studies confirm this statement. They concluded that most fraudulent reports are prepared during first three years of cooperation between auditors and their new clients.

Ewelt-Knauer et al. (2013) suggest that greater acquired expertise allows auditors to be more independent as they already possess sufficient knowledge about company’s processes and do not need to rely on management that much. Moreover, audit quality can be assured with the use of means other than mandatory audit rotation. Deis and Giroux (1992) claim that audit quality can be improved by providing third party reviews of audit results, opinions and working papers. According to their study auditors have more incentives to withstand managers pressure when they know that their work will be reviewed by third party and they will suffer from losing their reputation and in turn customers in case of discovery of their poor quality work. Thus, instead of introducing mandatory audit rotation, peer reviews may serve as a solution to the risk of auditor being too familiar and too dependent on their customer. Markelevich and Rosner (2013) examine in their research four theories regarding audit quality. Their theory number 4 – audit risk theory sustains the reputation theory findings. According to audit risk theory presented by Markelevich and Rosner (2013) auditors charge higher audit fees if they perceive the risk of engagement to be higher. High engagement risk implies high audit failure and litigation risks. Therefore, the risk of losing auditor’s reputation is also higher and auditors increase their fees by risk premium. Furthermore, higher risk causes auditors to put more effort in the engagement, i.e. perform higher quality audit.

Market forces constitute another factor which fosters increase, instead of a decrease in audit quality. Deis and Giroux (1992) indicated that competing on auditor’s market makes auditors more aware that it is easier for companies to replace auditor than for auditor to replace customers, especially if their reputation is unfavourable.

Furthermore, Ewelt-Kauer et al. (2013) point out that mandatory audit rotation may discourage companies to specialize as after only few years the knowledge they acquired may not be useful any more. Having less specific knowledge they sometimes may be not able to detect frauds or potentially fraudulent actions. Moreover, having no incentives to specialize, audit companies will most probably also invest less in their employees, causing further deterioration of audit quality. Ewelt-Knauer et al. (2013) claim that worse perspectives of

(16)

16

self-development will attract less highly motivated and talented employees as they will not perceive their future perspective as auditors broad enough.

Humphrey et al. (2011) refer also to the European Commission argument about market concentration and Big 4 bias. They point out that the assumption of enhanced market structure will lead to higher quality of audit is unproven. Moreover they also notice that EC’s belief in the power of market is somewhat contradictory to their attempts to regulate, correct or enhance them by introducing mandatory audit firm rotation.

2.3. Regulatory environment

Regulatory environment of audit and financial services in Europe is still divided between the so-called common- and code-law systems. Even though the European Union has done much to unify and standarise those laws by implementing directives and laws which have to be abided by all of the Member States, the differences still exist on national levels. The majority of the countries of continental Europe follow the example of Germany and France, basing their regulations on code law, while the UK is a representative of and Anglo-Saxon approach, called common law regulations (Blij et al. 1998).

Audit profession in Germany is regulated by legally established body whose objective is to represent the profession, ensure high quality of profession, teach and conduct official exam for all candidates that fulfill the predetermined conditions. This body is called Wirtschaftsprüferkammer and it acts with regard to Gesetz über eine Berufsordnung der Wirtschaftsprüfer (the Act on the Profession of Auditors). According to §43 of this Act, auditor’s professional duty is to maintain independency, confidentiality and responsibility. WPK is also supervised by the Ministry of Economics of German government (Baker et al. 2001). The UK audit profession is regulated not by a quasi-governmental body as in Germany but by the recognized professional body. The UK auditors exercise their professional competence with regard to the Code of Ethics included in the Members’ Handbook issued by the Institute of Chartered Accountants in England and Wales which regulates audit profession in England (Chung et al. 2010)

The difference between continental and Anglo-Saxon economic approach is most evident in case of financing of business enterprises. Banks in Germany often have substantial part of ownership in firms and therefore it is common that their representatives are members of both bodies of the two-tier board. (Weber at al. 2008). The ownership in Germany is regarded to be concentrated and as shareholders share close ties with the company, they possess much of

(17)

17

the insider knowledge, therefore auditing is more necessary to assess how corporate government mechanisms works in contrast to the assessment of reliability of the content of financial statements as it happens in the UK. The difference originates from the fact that in the UK the ownership is dispersed and capital is provided mostly not by banks as in Germany but by stock markets. Therefore, to keep smaller investors well informed, auditing ensures that financial reporting presents true and fair view of the firm (Evans and Nobes 1998). Both countries have also different regulations with regard to auditor’s liability. In Germany it is very difficult to sue auditor for damages acquired due to misstatements (Weber et al. 2008). Clients have to prove that auditor made a mistake intentionally and the whole audit was conducted in a reckless way. Investors may also try to bring action under civil investigation, however in that case German Handelsgesetzbuch (Commercial Code) specifies a cap of auditor’s liability which is set at the level of €1,000,000 per audit and €4,000,000 in case of listed companies (Chung et al. 2010). Chung et al. (2010) present that in the UK auditor’s liability is specified by the Company’s Act. The regulations specified in the Companies Act 2006 address auditor’s liability to the client itself, not to any other third parties. As the Act also says that auditor can limit his/her liability by contract approved by shareholders, the liability with regard to third parties comes into effect only indirectly when shareholder’s agree on liability cap (Chung et al. 2010).

Regulatory environment differences can also affect the level of earnings management. Leuz et al. (2003) found out that in Anglo-Saxon model investors are more protected as it is easier to hold auditor responsible in case when financial reporting does not give the true and fair view of firm’s performance. Leuz et al. (2003) showed also that earnings management is lower in common- as opposed to code-law countries. This is caused by the ownership concentration, size of stock market and also by investor protection. The results of their study show that weak legal protection and high level of insider knowledge (as in case of Germany) results in higher level of accrual-based earnings management. One of the reasons why earnings are managed is to improve the perception of company’s performance with shareholder’s permission. Leuz et al. (2003) claims that another reason of higher earnings management in code law countries is the fact that insiders benefit from earnings management and try to hide this fact because if uncovered, the outsiders would take legal and disciplinary actions against them.

(18)

18 2.4. Hypotheses development

Central issue of this research focuses on detecting the relationship between the change of audit firm and associated with that switch a change in overall audit quality proxied by the level of real earnings management. Basing on prior literature it is hard to unanimously state whether audit firm change results in an increase or in a decrease of audit quality. As presented in previous section there are undeniable anticipated positive results of auditor switch, however more papers suggest that audit quality is significantly lower in the first years of engagement, and increases with the increase of audit firm tenure. It seems that acquired business specific knowledge outweighs possibly negative effects of too high familiarity and dependence between the auditor and auditee. The latter two issues are associated with lower auditor independence which is believed to be a key to high quality audits. Although previous papers recognised industry specialists having moderating effect on the audit quality even in case of audit firm switch, lack of company specific knowledge and high dependence on information retrieved from management still speaks in favour of a decrease on audit quality just after the change of audit firm (Ewelt-Knauer 2013). Therefore the first hypothesis is aimed to check whether this study can confirm that audit firm change has an influence on audit quality:

H1: Audit quality will be lower in first year after the change of auditor that in the year before the change.

Following expected increase in acquired knowledge of company’s processes, the quality of audit is expected to continue growing in the subsequent years. Numerous papers (Carcello and Nagy 2004, Daniels and Brooker 2011, Johnson et al. 2002, or Myers et al. 2003) show that the biggest number of audit failures occur in the first (three) years after the change of auditor. As this study concentrates also on the European Commission’s proposal of implementing mandatory audit firm rotation in the EU, it is essential to check whether the new regime would be beneficial for quality in general. If the audit quality in third year still remains to be lower than in the year preceding the change of auditor, then the assumed support for mandatory audit firm rotation is either weak or non-existing. The reason for that is such a result would mean that new auditor needs at least three years to acquire essential knowledge which has already been in the possession of previous auditor. Furthermore three out of ten years of a maximum allowed tenure constitute a significant part of the total length

(19)

19

of engagement. Therefore, referring audit quality in third year after the change to the quality in a year before the change, the second hypothesis is constructed as follows:

H2: The quality of audit in third year after the change of auditor is lower than in year preceding auditor change.

As Europe’s legislation is not totally unified and even though many laws implemented by the EU are common in all Member States, there is still much space for individual legislation. The differences are visible especially between common law and code law countries, represented respectively by the UK and Germany. Both of the countries have big influence upon European legislation and their regulatory setting affects their stances. Furthermore prior research has shown that level of earnings management and thus also audit quality depends also on the economic regulatory environment (Leuz, 2003). Therefore in order to check if there still is a difference between the two settings and how it may affect implementation of the lack of thereof of the EC’s proposal regarding mandatory audit firm rotation, the third hypothesis is formulated in the following way:

H3: Audit quality is higher in common-law regulatory environment when compared to code-law regulatory environment.

3. Research methodology

This chapter outlines the selection of sample relevant to this study and then it describes method used to test hypotheses formulated in previous section.

3.1. Sample selection

The study concentrates on public companies listed in Germany and the UK. A record of 485 firms has been collected from the Composite-DAX (CDAX) index. This is a composite index of all stocks traded on the Frankfurt Stock Exchange that are listed in the General Standard or Prime Standard market segments. Furthermore a the UK part of sample comprises of firms included in the FTSE 350 index This index incorporates 354 largest companies having primary listing on the London Stock Exchange, ranked by capitalisation. This index contains the FTSE 100 index of 100 largest companies by capitalisation and FTSE 250 index of subsequent 250 companies.

The data was collected for the 9-year period, i.e. years 2004-2012 to get enough firm-year observations. All companies for which the data was incomplete were excluded from the

(20)

20

sample. The same happened to firms belonging to the financial services, banks or insurance industries since their financial reporting significantly differs from this of non-financial industry based firms.

Audit-related data (excluding the name of audit firm) was retrieved from Datastream, whereas financial data was collected from both, Datastream and Compustat, depending on which database contained more accurate and complete data. Data regarding the name of audit firm was collected manually from companies’ annual reports.

3.2. Methodology

As mentioned before real earnings management can be defined as actions taken by managers which deviate from normal business practices undertaken to meet firm’s business objectives and expectations. High quality auditors should prevent aggressive earnings management aimed at increasing current period net income. Following Roychowdhury (2006), Chi et al (2011) and Kim et al. (2012) real earnings management can be indicated by abnormal level of cash flow from operations (Abn_CFO), abnormal level of costs of production (Abn_Prod), as well as abnormal level of discretionary expenses (Abn_DiscExp).

In this study, as presented by Kim et al. (2012) real earnings management will be proxied by the fourth, combined measure of real activities manipulation which will be obtained by aggregating above mentioned three individual real earnings management measures. Higher level of REM_Index indicate higher level of this type of earnings management and thus lower audit quality. Considering the direction and impact of each of the three measures, the combined REM_Index will be presented as following:

𝑅𝑅𝑅𝑅𝑅𝑅_𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 = 𝐴𝐴𝐴𝐴𝐼𝐼_𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐴𝐴𝐴𝐴𝐼𝐼_𝑃𝑃𝑃𝑃𝑜𝑜𝐼𝐼 + 𝐴𝐴𝐴𝐴𝐼𝐼_𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑅𝑅𝐼𝐼𝐷𝐷

(A) Following Roychowdhury (2006) current period cash flow from operating activities can be affected by the use of manipulation of sales. Abnormal sales manipulation can be presented as the residual of normal cash flow from operations, therefore the calculation of Abn_CFO will involve following equation:

𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1 = 𝑎𝑎1𝑖𝑖�1 𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷⁄ 𝑖𝑖,𝑖𝑖−1� + 𝑎𝑎2𝑖𝑖�𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷𝑖𝑖,𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1� +𝑎𝑎3𝑖𝑖�∆𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷𝑖𝑖,𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1� + 𝜀𝜀𝑖𝑖𝑖𝑖

(21)

21

Abnormal level of costs of production (Abn_Prod) are the second measure of real activities management and will derived similar as Abn_CFO, as a residual of following model:

𝑃𝑃𝑃𝑃𝑜𝑜𝐼𝐼𝑖𝑖𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1 = 𝐴𝐴1𝑖𝑖�1 𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷⁄ 𝑖𝑖,𝑖𝑖−1� + 𝐴𝐴2𝑖𝑖�𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷𝑖𝑖,𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1�

+𝐴𝐴_3𝑖𝑖(∆𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷_{𝑖𝑖,𝑖𝑖}⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷_{𝑖𝑖,𝑖𝑖−1})+ 𝐴𝐴_4𝑖𝑖�∆𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷_{𝑖𝑖,𝑖𝑖−1}⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷_{𝑖𝑖,𝑖𝑖−1}� + 𝜀𝜀_{𝑖𝑖𝑖𝑖} (C) Finally, manipulation of discretionary expenses, i.e. expenses which are not essential for the core operations of company are the last indicator of real activities manipulation used in this model. As in previous equations, the abnormal level of discretionary expenses (Abn_DiscExp) will be the residual of a following model:

𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑅𝑅𝐼𝐼𝐷𝐷𝑖𝑖𝑖𝑖⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1 = 𝐷𝐷1,𝑖𝑖�1 𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷⁄ 𝑖𝑖,𝑖𝑖−1� + 𝐷𝐷2,𝑖𝑖�𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷𝑖𝑖,𝑖𝑖−1⁄𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷𝑖𝑖,𝑖𝑖−1� + 𝑣𝑣𝑖𝑖𝑖𝑖 (D) Where:

CFO – cash flow from operations

Assetst – the total assets at the end of period t Salest – sales during period t

ΔSales = Salest – Salest-1

Prodt – sum of costs of goods sold and change in inventory in year t

DiscExpt – sum of R&D, Advertising and SG&A expenses

To test hypotheses, I employ model used by Kim et al. (2012), however I adjusted control variables with regard to prior literature so that they now include those which are most relevant to assess how audit firm rotation affects audit quality proxied by real earnings management.

𝑅𝑅𝑅𝑅𝑅𝑅𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 = 𝑎𝑎0 + 𝑎𝑎1𝐼𝐼𝐼𝐼𝐼𝐼𝑅𝑅𝐼𝐼𝐷𝐷𝑖𝑖+ 𝑎𝑎2𝐵𝐵𝐷𝐷𝐵𝐵𝐵𝐵𝑖𝑖+ 𝑎𝑎3𝐿𝐿𝐼𝐼𝑣𝑣𝑖𝑖−1+ 𝑎𝑎4𝑆𝑆𝐼𝐼𝑆𝑆𝑅𝑅𝑖𝑖−1+ 𝑎𝑎5𝐷𝐷𝐴𝐴𝑖𝑖+ 𝑎𝑎6𝐵𝐵𝐴𝐴𝐶𝐶𝑖𝑖 + 𝑎𝑎7𝐺𝐺𝑅𝑅𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖+ 𝑎𝑎8𝑅𝑅𝐶𝐶𝐴𝐴𝑖𝑖−1+ 𝑎𝑎9𝐿𝐿𝐼𝐼𝐵𝐵𝑅𝑅𝐼𝐼𝑣𝑣𝑖𝑖+ 𝑎𝑎10𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖+ 𝑎𝑎11𝐿𝐿𝐶𝐶𝑆𝑆𝑆𝑆𝑖𝑖+ 𝑎𝑎12𝑌𝑌0 + 𝑎𝑎13𝑌𝑌1 + 𝑎𝑎14𝑌𝑌3 + 𝜀𝜀𝑖𝑖

(22)

22 Where:

IndExp – indicator if an audit company is an industry specialist. Measure based on the market

share of an audit company in a given industry and year1. Following Fernando et al. (2010) a dummy equal to 1 if specialisation > 20%, 0 otherwise will be introduced;

BigN – dummy variable equalling 1 when the auditor is one of the Big4 firms (Ernst&Young,

KPMG, Deloitte, Pricewaterhouse&Coopers), 0 otherwise;

Lev – long-term debt scaled by total assets;

SIZE – natural logarithm of market value of equity for a firm;

DA – absolute value of discretionary accruals, where discretionary accruals are computed

using modified Jones model (1991)2;

NAF – sum of non-audit fees paid by client to audit firm in year t; GRWTH – growth in revenues scaled by total assets;

ROA – net income scaled by prior period assets;

LegEnv – Legal environment in which the company operates – dummy=1 if Germany (code

law), 0 if the UK (common law);

CFOt – cash flow from operations scaled by total assets in year t;

LOSSt – dummy variable equal 1 if the company suffered a loss in a given year, 0 otherwise;

Y0,Y1,Y3 – dummy variables =1 if it the year in which the audit was performed is a year

before the change of audit company (Y0), first (Y1) or third year (Y3) after the change.

Earlier studies show that auditor’s industry specialisation ensures extensive working knowledge of the client’s and sector’s business environment, its unique characteristics and process, as well as industry accounting practices, and potentially risky accounting actions.

1_{Market share will be calculated based on model presented by Jaggi (2012).}

𝑅𝑅𝑆𝑆𝑖𝑖𝑖𝑖=_{∑ ∑ 𝑆𝑆𝑆𝑆𝑆𝑆𝐼𝐼𝑆𝑆}∑ 𝑆𝑆𝑆𝑆𝑆𝑆𝐼𝐼𝑆𝑆𝑖𝑖𝑖𝑖𝑖𝑖

𝑖𝑖𝑖𝑖𝑖𝑖, where

MS – market share of an audit company

∑ 𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷 - sum of sales of all clients of an auditor in industry

∑ ∑ 𝑆𝑆𝑎𝑎𝑆𝑆𝐼𝐼𝐷𝐷 - sum of sales of all clients in industry summed over all auditor firms

2_{DA=TA-NDA, where:}

DA – value discretionary accruals, TA – value of total accruals,

NDA – value of non-discretionary accruals

Non-discretionary accruals are presented as:

𝐵𝐵𝐷𝐷𝐴𝐴𝑖𝑖= 𝑎𝑎1(1 𝐴𝐴𝐷𝐷𝐷𝐷𝐼𝐼𝐴𝐴𝐷𝐷⁄ 𝑖𝑖−1) + 𝑎𝑎2∆𝑅𝑅𝐼𝐼𝑣𝑣𝑖𝑖+ 𝑎𝑎3𝑃𝑃𝑃𝑃𝑅𝑅𝑖𝑖+ 𝜀𝜀𝑖𝑖, where:

Assetst-1 – total assets at the beginning of year t

ΔREVt – change in revenues (REVt-REVt-1) scaled by total assetst-1

PPEt – property plant and equipment (gross) scaled by total assetst-1

(23)

23

(Fernando et al. 2010). Furthermore, Gul et al. (2009) prove that the association between shorter auditor tenure and lower audit quality are weaker when industry specialist auditor is employed. They claim that industry-specific knowledge helps auditor to detect irregularities even though the auditor is not familiar with client-specific environment. Therefore, I expect negative relationship between IndExp and REM_Index.

DeAngelo (1981) claims that larger audit companies, especially BigN companies are regarded to provide higher quality audit. The same conclusions were drawn by Becker et al. (1998). Francis and Yu (2009) found that larger offices have benefit from bigger experience as they employ many auditors and perform many audits in various industries. Moreover, Choi et al. (2010) showed that large offices ten to rely less on a single client and are therefore more resistant to pressure put on auditor by client’s management. Choi et al. (2010) adds also that larger offices are able to charge higher fees and therefore provide higher quality audit as the difference in quality is priced in the premium. Thus I expect negative relationship between

BigN and REM_Index.

Chi et al. (2011) claim that real earnings management and accrual-based earnings management are substitutes. Zang (2007) shows that if the regulatory environment is strict and therefore the possibilities to use accrual-based earnings management are restricted, managers resort to real earnings management. It helps to manipulate earnings and is perceived to be safer as it does not violate any regulations if properly disclosed. In such a case where real earnings management and accrual-based management are each other’s substitutes, I expect higher level of accrual-based earnings management to be negatively correlated with REM_Index.

Markelevich and Rosner (2013) find in their study that higher non-audit fees lead to economic bonding, impairing audit quality. Therefore they reject knowledge spillover theory that says that if audit company performs services other than audit to one client, the increase in available information would spill over to auditor and hence increase audit quality. Also Frankel et al. (2002) point out that provision of non-audit services increases earnings management. Therefore NAF should be positively correlated to REM_Index.

Various control variables which may influence financial reporting and thus audit quality are included to minimise the risk of correlated omitted variables. Roychowdhury (2006) points that firm-specific growth measures and size of the firm may pose an explanation possible significant variation in earnings management. Bigger firms usually apply more internal

(24)

24

controls, they are also believed to posses better quality corporate governance, therefore both the incentive and chances of earnings management are lower. Therefore I expect the relation between the size of the company and REM_Index to be negative. Myers et al. (2003) showed that higher firm growth rate implies lower earnings management. Therefore I expect positive relationship between REM_Index and GRWTH. LEV variable is included in the model as previous literature shows that if firm has a relative high amount of debt, it may want to manage its earnings downwards to get more favourable contract renegotiations, especially if it is in financial distress (Becker et al. 1998). I also control for LOSS as firms suffering loss may have higher incentives to improve their financial position, and thus resort to higher level of earnings management.

Previous literature proved that regulatory environment affects the level of earnings management. It is believed that earnings management in the common-law regulatory setting is lower than in code-law regulatory setting (Leuz et al. 2003).

Following Roychowdhury (2006), I employ cash flow from operations scaled by total current year assets as one of the variables. According to this author firms which use real earnings manipulation have a lower cash flow from operations scaled by total assets. Therefore I expect negative relation between this variable and real earnings management.

ROA represents profitability of the firm, and is included in this research as one of the control variables because it is believed to be affected by real earnings management activities at the end year as specified by Zang (2012). Zang’s (2012) empirical research shows that ROA is positively associated with real earnings management, therefore I also predict that the higher ROA, the higher level of real activities manipulation.

This study involves running regressions which examine the level of audit quality in a year before and first and third year after auditor change to investigate whether the switch improved or decreased audit quality. Furthermore, separate regression will be run to examine differences between two legal environments, i.e. code law environment represented by German firms and common law environment represented by firms originating from the UK. According to previous literature audit quality is in general lower in the first (three) years after the change (Carcello and Nagy 2004, Daniels and Brooker 2011, Johnson et al. 2002, or Myers et al. 2003). According to the European Commission’s project the maximum audit firm tenure is set to be ten years which indicates that the quality of audit provided after the change will be lower for at least three out of ten years (30%) compared to the quality

(25)

25

delivered before the change. As 30% of total tenure constitutes its significant part I want to examine the empirical relation between the quality of audit provided in year before the change and in third year after the change. If it is significantly lower I would have to conclude that there is no support for mandatory audit rotation in the European Union.

4. Results

This section presents results obtained from empirical model specified in previous section. I do not present here the data regarding the estimation of discretionary accruals as they are not the core of my research. Nevertheless the adjusted R2 of this estimation model equaled to 27.7% which means that the model explains 27.7% of variation of estimated variables. In the following subsection, data regarding descriptive statistics and results of the estimation model presented in the previous section will be presented and analysed.

4.1. Descriptive statistics

This section gives most important information regarding the composition and characteristics of the sample and variables.

Final sample consisted of 170 companies, 102 from the UK, 68 from Germany, which resulted in 953 firm-year observations. Among those companies I identified 44 of them which decided to change audit firm during the examined period of 2004-2012. There were 20 firms emerging from the UK which have changed auditor, and for which I was able to gather all necessary data to conduct my research. In case of Germany, there were 24 of such firms. I observed 16 cases when the company changed its auditor from audit firm being described as non-industry specialist to industry specialist (9 cases in Germany, 7 in the UK). The opposite change was recorded by 13 sample firms (8 from Germany, 5 from the UK). Firms were less willing to switch audit companies from BIGN to non-BIGN and reversely. There were only 7 cases when company decided to appoint one of the BIGN firms after they used non-BIGN firm’s services (4 cases in Germany, 3 in the UK), whereas only one firm (from Germany) decided to switch from BIGN to non-BIGN audit firm.

As it can be seen from the table 1, the real earnings management is positive for majority of the sample. Positive mean value means that real activities management was used to increase current year income. Though dummy variables introduced only for the purpose of this section show that only 29.7% of all observations included upwards earnings management. This means that even though in majority the earnings were managed downwards, the companies

(26)

26

which decided to implement income increasing actions did that on a bigger scale and imposed outweighing effect. Contrasting this results with the accrual based earnings management, it turns out that in 46.2% of observations the earnings were managed upwards with the help of accruals, while 53.8% were managed downwards. Mean value is negative which this time is consistent with the motion of the majority.

Vast majority of companies used services of auditors working for one of the Big4 companies (96%), however only 53.8% used services of audit firms classified as industry specialists. 44.6% of observations belonged to firms originating from the UK, whereas German entities constituted 55.4% of the sample. Only 4.5% of firm-year observations indicated the change of auditor. The percentage of observations for third year is even lower (3.5%) as some firms decided to switch audit companies close to the end of examined period, therefore in those cases the third year after the change was not included in the sample. Furthermore, only in 8.4% of all firm-year observations there was a loss observed. On average, the leverage reached level of 21%. The sample consists of firm-year observations which represent fairly well prospering entities as the average ROA equaled 8.3%. The sample consists also of growing companies whose average growth rate reached 7%.

(27)

27

Table 2 shows the comparison of two groups contained in the whole sample, namely firm-year observations which from the firm-year before the change of audit firm, one and three firm-years after the switch. Second group encompasses all other firm-year observations which are not linked to the change of audit firm. Table 2 presents the results from two-tailed tests for real earnings management. The difference between change and no change group is significant at p=0.01. It is worth noting that among the group where the audit firm has changed, on average the earnings were measured upwards, whereas in the no change group, earnings were measured slightly downwards. The share of industry specialists as well as BIG4 auditors is significantly lower in case of change groups (43.7% of change companies were audited by industry specialist and 89.9% by one of the BIG4 firms compared to 56.6% among no change firms audited by industry specialist and 96.9% by one of the BIG4). Among significant differences between those two groups one can also find legal environment – 53.8% of change

Variable Obs Mean Std. Dev. Min Max

REM_Index 953 0.141965 6.354959 -7.76648 100.214 REM_up 953 0.296957 0.457157 0 1 REM_down 953 0.703043 0.457157 0 1 IndExp 953 0.549843 0.497771 0 1 BIG4 953 0.960126 0.195766 0 1 LegEnv 953 0.44596 0.497332 0 1 LOSS 953 0.083945 0.277452 0 1 y0 953 0.045121 0.207678 0 1 y1 953 0.045121 0.207678 0 1 y3 953 0.034628 0.18293 0 1 cfo_assets 953 0.6281695 6.255167 -5.73895 97.73829 ROA_t1 953 0.082564 0.09829 -0.6956 0.7944 Lev_t1 953 0.210975 0.198493 0 1.909 SIZE_t1 953 7.767085 1.514795 0 11.5104 GRWTH 953 0.070042 0.175803 -0.95397 1.443878 NAF 953 1704626 10300000 0 301000000 DA 953 -0.005421 0.075412 -0.42533 0.628837 DA_up 953 0.4617 0.498793 0 1 DA_down 953 0.5383 0.498793 0 1 Table 1 Summary of variables

(28)

28

companies originate from Germany, while in the no change group German firms constitute only 43.4%. Compared to no change firms, change companies suffered loss less frequently (6.7% against 8.6%), had lower leverage (20.4% against 21.2%), they also achieved lower return on assets (6.2% against 7.1%). Similarly as in case of real activities management on average change firms managed their earnings upwards with the use of accrual earnings management while no change firms used this form of earnings management to decrease their current year earnings. Nevertheless, the differences in LOSS, Lev_t1, ROA_t1 and DA are not significant neither at 1% nor at 5%, nor at 10% of significance level.

Table 3 also presents the results of the two-tailed t-test, however this time the differences between legal environments in which examined companies operate were compared. On average German companies recorded negative real earnings management results which means that they used income-decreasing real activities. On the contrary, among firms originating from the UK observed REM_Index shows negative mean value which indicates that activities employed by those firms aimed to increase current period earnings. The difference between the countries in that case is significant at 5% level of significance. German companies were more likely to choose industry experts as their auditor (64.9% against 47% of the UK firms). Although they were audited less frequently by one of the BIG4 firms, vast majority of German firms still chose one of the biggest 4 audit companies (94.3% against 97.3% in the UK). German firms suffered a loss significantly more often (11.1% against 6.3% in the UK). Furthermore UK companies on average report significantly higher return on assets

Change No change

Variable Mean Mean Difference t stat

REM_Index 1.599603 -0.066019 -1.665623 -2.6834*** IndExp 0.4369748 0.565947 0.1289725 2.6525*** BIG4 0.8991597 0.968825 0.0696653 3.655*** LOSS 0.0672269 0.086331 0.019104 0.7025 DA 0.0016372 -0.006428 -0.008065 -1.0915 Lev_t1 0.2042647 0.211933 0.0076679 0.394 ROA_t1 0.0620525 0.071182 0.0091297 0.00913 LegEnv 0.5378151 0.432854 -0.104961 -2.1579** *** - correlation significant at p=0.01 ** - correlation significant at p=0.05 * - correlation significant at p=0.1 Table 2

(29)

29

(10.3% against 5.7% in Germany), and their growth rate is significantly higher (7.9% against 5.9% in Germany). Table 3 also records insignificant differences observed between the two countries. Discretionary accruals are used in both countries to decrease current period earnings, and are slightly higher in case of German firms. Moreover firms originating from the UK are a bit more leveraged (21.12% against 21.06% in Germany). As mentioned before, there are more firm-year observations with regard to audit firm change in Germany. Such a change was recorded in 5.4% of all German observations, whereas in the UK observations including audit firm change constituted only 3.8%. This difference is yet insignificant. Similarly observations concerning third year after the switch of audit firm constitute 4.2% of all firm-year observations for Germany, and 2.8% of observations for the UK.

Table 4 shows the Pearson and Spearman correlation coefficients for all variables. It represents the strength of the relationship between variables. The majority of variables are significantly correlated to variable representing real earnings management (REM_Index). LOSS, SIZE_t1 and NAF are thus the exceptions.

Germany UK

Variable Mean Mean Difference t stat

REM_Index -0.4319121 0.603893 -1.035805 -2.508** IndExp 0.6494118 0.469697 0.1797148 5.6287*** BIG4 0.9435294 0.973485 -0.029955 -2.3536** LOSS 0.1105882 0.0625 0.0480882 2.6681*** y0 0.0541176 0.037879 0.0162389 1.2001 y1 0.0541176 0.037879 0.0162389 1.2001 y3 0.0423529 0.028409 0.0139439 1.1699 DA -0.0061465 -0.004837 -0.00131 -0.2664 Lev_t1 0.2106494 0.211237 -0.000588 -0.0454 ROA_t1 0.0572188 0.102965 -0.045746 -7.3374*** GRWTH 0.058546 0.079296 -0.02075 -1.8133* *** - correlation significant at p=0.01 ** - correlation significant at p=0.05 * - correlation significant at p=0.1 Table 3

(30)

REM_Index cfoassets IndExp BIG4 LegEnv LOSS y0 y1 y3 ROA_t1 Lev_t1 SIZE_t1 GRWTH NAF DA REM_Index 1.0000 0.3226*** -0.1807*** -0.0924*** -0.1620*** -0.0114 -0.0356 -0.0400 -0.0391 0.1122*** -0.0900*** 0.0386 -0.1849*** -0.0065 -0.0655** cfoassets 0.9863*** 1.0000 -0.0369 -0.0296 -0.2987*** -0.1535*** -0.0055 0.0105 -0.0370 0.4069*** -0.0358 -0.1118*** 0.1231*** -0.1429*** -0.1849*** IndExp -0.0534* -0.0302 1.0000 0.2252*** 0.1796*** 0.0303 -0.0777** -0.0370 -0.0247 -0.0304 -0.0361 0.1976*** 0.0288 0.0291 0.0670** BIG4 0.0055 0.0164 0.2252*** 1.0000 -0.0761** -0.0157 -0.1624*** -0.0074 -0.0201 0.0177 0.0103 0.0390 -0.0383 0.0853*** 0.0535* LegEnv -0.0811** -0.0777** 0.1796*** -0.0761** 1.0000 0.0862*** 0.0389 0.0389 0.0379 -0.2845*** 0.0471 0.3828*** -0.0704** 0.0555* -0.0261 LOSS -0.0086 -0.0021 -0.0303 -0.0157 0.0862*** 1.0000 -0.0111 0.0071 -0.0366 -0.1978*** -0.0680 0.0104 -0.0358 -0.0258 -0.1296*** y0 0.0423 0.0551* -0.0777** -.01624*** 0.0389 -0.0111 1.0000 -0.0473 -0.0412 0.0088 -0.0154 -0.0618* 0.0721** -0.0144 -0.0099 y1 0.0335 0.0307 -0.0370 -0.0074 0.0389 -0.0071 -0.0473 1.0000 -0.0412 -0.0149 0.0179 -0.0417 -0.0297 -0.0559* -0.0275 y3 0.0706** 0.0734** -0.0247 -0.0201 0.0379 -0.0366 -0.0412 -0.0412 1.0000 -0.0147 0.0060 -0.0348 -0.0613* -0.0117 0.0261 ROA_t1 0.0106 0.0028 -0.0385 0.0283 -0.2315*** -0.2244*** -0.0161 -0.0153 -0.0180 1.0000 -0.1064*** -0.0843*** 0.1279*** -0.1818*** 0.1440*** Lev_t1 -0.0706** -0.0715** -0.0624* 0.0273 -0.0015 -0.0441 -0.0297 0.0165 -0.0081 -0.0379 1.0000 0.0709** -0.0315 0.1107*** 0.0265 SIZE_t1 -0.0050 -0.0201 0.1951*** 0.0376 0.3597*** 0.0083 -0.0828* -0.0502 -0.0328 -0.0274 -0.0094 1.0000 -0.1526*** 0.3720*** -0.0087 GRWTH -0.0151 0.0098 -0.0035 -0.0485 -0.0587* -0.0251 0.0545* -0.0391 -0.0486 0.0848*** -0.0315 -0.1396*** 1.0000 -0.0588* 0.0251 NAF -0.0032 -0.0038 -0.0190 0.0037 0.0422 -0.0121 -0.0089 -0.0185 -0.0124 -0.0271 0.0037 0.0683** -0.0252 1.0000 -0.0352 DA 0.0095 0.0221 0.0221 0.0462 -0.0086 -0.1276*** 0.0266 0.0155 0.0161 0.1239*** 0.0426 -0.0074 0.0260 0.0080 1.0000 *** - correlation significant at p=0.01 ** - correlation significant at p=0.05 * - correlation significant at p=0.1 Table 4

(31)

4.2 Multivariate regressions results

This section presents the results observed from the analysis of multivariate regression. First two hypotheses are examined with the use of the same regression – dummy variables representing years allow to examine the comparison of level of real earnings management reported in years under scrutiny. Afterwards results regarding the effect of legal environment in which firms operate on real activities manipulation will be reported.

Table 5 reports empirical results of the regression analysis of equation [E] presented in section 3.2 The model depicts REM_Index – variable representing the level of real earnings management as a proxy for earnings quality well. The reported adjusted R2 equals to 0.9743 which means that 97.43% of variance in REM_Index is explained with the use of this model. Main variables of interest in table 5 are dummy variables representing the year. Only year 0 dummy is significant (at 5% level of significance). As specified in previous sections this variable takes value of 1 when the firm-year observation is made in the last year before the

Source SS df MS Number of obs 953

Model 37473.1925 14 2676.657 F( 14, 938) 2578.24

Residual 973.805897 938 1.0382 Prob > F 0

Total 38446.9984 952 40.3855 R-squared 0.9747

Adj R-squared 0.9743

Root MSE 1.0189

REM_Index Coef. Std. Err. t

cfoassets 1.002506 0.0053452 187.55 0.000 *** 0.9920162 1.012996 IndExp -0.3173446 0.0706501 -4.49 0.000 *** -0.4559951 -0.1786941 BIGN -0.2937966 0.1769275 -1.66 0.097 * -0.6410161 0.0534229 LegEnv -0.0700359 0.0751294 -0.93 0.351 -0.2174771 0.0774054 LOSS -0.1801153 0.1233648 -1.46 0.145 -0.4222183 0.0619877 y0 -0.3737514 0.1632234 -2.29 0.022 ** -0.6940767 -0.053426 y1 0.0645942 0.1604745 0.4 0.687 -0.2503363 0.3795247 y3 -0.1164347 0.1825527 -0.64 0.524 -0.4746937 0.2418243 ROA_t1 0.4987278 0.3576489 1.39 0.164 -0.2031568 1.200612 Lev_t1 -0.0598236 0.1680408 -0.36 0.722 -0.389603 0.2699558 SIZE_t1 0.0754982 0.0241141 3.13 0.002 *** 0.0281744 0.122822 GRWTH -0.835413 0.1914022 -4.36 0.000 *** -1.211039 -0.4597868

NAF -9.21E-10 3.22E-09 -0.29 0.775 -7.25E-09 5.40E-09

DA -1.026461 0.4453158 -2.31 0.021 ** -1.900391 -0.1525302 _cons -0.5273096 0.249795 -2.11 0.035 ** -1.017531 -0.037088 *** - correlation significant at p=0.01 ** - correlation significant at p=0.05 * - correlation significant at p=0.1 Table 5

Robust regression results: Equation [E]

The effects of voluntary auditor rotation on audit quality : is there support for mandatory audit rotation proposed by the European Commission?