1 Amsterdam Business School
MSc Thesis
How does the adoption of IFRS affect the quality of earnings for
private companies in Europe?
Name: Gino Balke
Student number: 10423672
Thesis supervisor: Alexandros Sikalidis Date: 18 – 6 – 2016
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Gino Balke who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
Globalization is leading to ever increasing business worldwide. A part of this increased international business relates to investors seeking opportunities to invest their capital internationally. In order to make a well-founded investment decision however, they need good quality financial statements which are understandable to them. As there are a lot of differences in worldwide financial reporting, the IFRS was founded in 2005 to become the single set of internationally applied high-quality financial reporting standards. In order to be accepted as such, IFRS had to improve comparability as well as financial reporting quality. This research was performed to see whether IFRS indeed improved the reporting quality for private companies specifically by looking at reported earnings, a key metric used by investors to determine the value a company. The focus is on private companies as there is a gap in the literature on earnings quality for these companies, while the same research for public companies is widely performed. These results, however, are not readily transferable to private companies and this research therefore tries to answer the research question: How does the
adoption of IFRS affect the quality of earnings of private companies in Europe? This question
will be answered in an empirical archival research using cross-company panel data. The used dataset is split in a main sample of UK private companies and a sensitivity analysis on German private company data. The measures used to determine earnings quality are: ‘earnings persistence’, ‘timely loss recognition’, ‘abnormal accruals’ and ‘real earnings management’. The results indicate that overall, IFRS adopters show higher quality earnings than local GAAP users which indicates that IFRS may indeed facilitate international investors in their investment decisions as both comparability and reporting quality increase.
4 Table of contents
1 Introduction ... 5
2 Theoretical Background and measures used ... 9
2.1 Earnings quality ... 9
2.2 Reporting in Europe ... 11
2.2.1 The implementation of IFRS ... 11
2.2.2 Reporting by public companies ... 12
2.2.3 Reporting by private companies ... 14
2.2.4 The implementation of IFRS ... 15
3 Hypothesis Development ... 16
4 Dataset and Methodogy ... 17
5 Measures used ... 18
5.1 Earnings persistence ... 19
5.2 Timely loss recognition ... 19
5.3 Abnormal accruals ... 20
5.4 Real activities manipulation ... 22
6 Results ... 24
6.1 Descriptive statistics ... 24
6.2 Main results ... 28
6.2.1 Earnings persistence ... 28
6.2.2 Timely loss recognition ... 28
6.2.3 Abnormal accruals ... 30
6.2.4 Real earnings management ... 30
6.3 Sensitivity analysis ... 31
6.3.1 Earnings persistence ... 31
6.3.2 Timely loss recognition ... 32
6.3.3 Abnormal accruals ... 33
6.3.4 Real earnings management ... 34
6.4 Additional test ... 35
7 Discussion and conclusion ... 38
8 Limitations and further research ... 42
References ... 44
5 1. Introduction
Company earnings are a key metric for investors to base their investment decisions on. Investors therefore prefer reported earnings to be of high quality which can be defined as “Sustainable, Repeatable, Recurring, Consistent, Reflects long-term trend, Reliable and that has the highest chance of being repeated in future periods” (Dichev et al., 2013, p. 12; IASplus, 2016). Several tools can be used to improve the quality of earnings of which one of the most important is accounting standards. Starting 2005, International Financial Reporting Standards (IFRS) is mandatory for EU publicly noted companies and allowed for private companies in the preparation of their financial statements (Chan et al., 2015). The purpose of IFRS is to be expanded and applied internationally in order to increase comparability and quality of the financial statements of companies worldwide. The scientific interest on the effects of IFRS adoption on the quality of earnings management is ever increasing but is mostly focused on public companies. This article will therefore look at the effect of IFRS adoption on non-public (or private) corporations reporting quality, or more specifically, the quality of reported earnings. The problem that private company reporting faces is that the adoption of a current high-quality set of financial reporting standards (IFRS/US-GAAP) leads to the mandatory accumulation of a lot of information which is irrelevant to their investors (Cheney, 2012). Public companies have a broad base of investors who desire a lot of information to be reported as they cannot accumulate it themselves. Private companies, however, often accumulate capital from banks or other major investors who have access to private information (Burgstahler et al., 2006; Pope & McLeay, 2011). As reporting incentives differ, reporting quality following the adoption of IFRS cannot be assumed to be the same for public and private companies (Ball & Shivakumar, 2005). This research will therefore look at the effect that IFRS adoption has on the quality of reported earnings of private firms.
As argued by Pope & Mcleay (2011), the objectives of the promotion of IFRS by the IASB has two goals. The first is to increase comparability of financial reporting between countries; the second is to increase the quality of financial reporting trough a set of high quality financial reporting standards. As technology is expanding, the world is becoming ‘smaller’. Globalization is leading to ever increasing internationally oriented business and communication. Along with the increase in international business is the desire, as well as the ability, to globally acquire capital from investors and investing institutions. However, as reporting standards differ between countries, global capital allocation is hindered as financial information is not comparable and thus not informative to investors. The IASB tries to solve this problem by implementing (and stimulating the use of) IFRS which aims to be the globally
6
applied set of accounting principles. By using the same accounting standards in the preparation of the financial statements everywhere, regulators and standard setters hope that the comparability of the financial statements will be increased. IFRS thus facilitates international business and capital allocation by increasing international comparability of information. In order to be globally applied however, IFRS also has to fulfil the quality expectations. The quality requirements are set out in the fundamental and enhancing qualitative characteristics of useful financial information (IASplus, 2016). The first fundamental characteristic of relevance is expected to be fulfilled as earnings quality information should always be relevant to decisions based on earnings quality. The second fundamental characteristic of faithful representation means that information should be complete, neutral and free from material error. It relates to the information that is in the financial statements and therefore determines earnings quality. It is this fundamental characteristic of financial information that this research focusses on. The enhanced qualitative characteristics are out of the scope of this research (IASplus, 2016).
This article will perform research to find how IFRS fulfills its fundamental qualitative characteristic of faithful representation by looking at the influence that IFRS adoption has on the quality of reported earnings of non-public European companies. As argued by the SEC in a report on IFRS adoption: “a single set of high-quality financial reporting standards will help meet its goals of (1) improving financial reporting quality in the U.S. and (2) reducing differences in financial reporting across countries” (FASB, 2016; Sun et al. 2011). This article specifically looks at the first goal that IFRS is argued to fulfill (in a different region) by looking at the quality of reported earnings of European private corporations that report under IFRS, and compares these to other European private countries that still use local GAAP. The article focuses on nonpublic corporations as the literature on IFRS adoption for public corporations is already addressed in other research while non-public corporations have received less scientific attention. Private companies did not receive a lot of attention in previous literature while being highly important as it accounts for 75% of the European GDP and makes up 99% of the businesses in the EU (Cameran et al., 2014; Kaya & Koch, 2015). Recent literature argued that financial reporting quality cannot be assumed homogeneous between public and private companies because they have different motivations for reporting (Ball & Shivakumar, 2005; Cameran et al., 2014; Pope & McLeay, 2011). So results of earnings quality research on the adoption if IFRS for public corporations cannot be extended to answer the same question of what effect IFRS adoption would have on earnings quality of
7
private corporations, confirming the need for research as is performed in this article (Cameran et al. 2014).
The aim of this research is relevant and interesting as another paper written by Cameran et al. (2014) already looked into the effects of IFRS adoption on earnings quality for a sample of private Italian companies for the years 2005 until 2008. Their conclusion was that adoption of IFRS did not show higher earnings quality with local GAAP adopters. In fact, they found that companies which adopted IFRS had higher abnormal accruals and lower timely loss recognition which would indicate lower earnings quality. In 2009, the IASB issued IFRS for SMEs which are specific reporting rules for private companies. This version of IFRS is a simplified version of the complete IFRS specifically created for SMEs to reflect their needs and cost-benefit considerations (IFRS, 2015). The EU, however, rejects the option of reporting under IFRS for SMEs for private companies but leaves it to the member states to decide whether or not they use the full standards for private companies (Kaya & Koch, 2015). It is therefore likely that the EU sees the full version of IFRS to be of higher quality than IFRS for SME, possibly because IFRS has been improved over the years after adoption in 2005 as is argued in different other articles (Callao & Jarne, 2010; Barth et al., 2008). To find what the influence of IFRS adoption by private companies on earnings quality is we partially follow the approach of Cameran et al. (2014). However, there are differences in our approach. This research focusses on the UK instead of Italy and adds a small sample of German private IFRS adopters for a sensitivity analysis. Also, a different method to calculate abnormal accruals is used and earnings persistence and real earnings management are added to assess the quality of earnings.
To provide insight in the effect that IFRS adoption has on private the earnings quality of European private companies, this article looks at earnings quality through four different measures. The first measure used is earnings persistence, which states that more persistent earnings are more useful to investors to base their decision on because it allows them to better determine the value of a company (Dechow, 2010). The second measure used is timely loss recognition. Timely loss recognition requires the reporter to report losses when these losses are likely in order to increase earnings quality. This improves the quality of earnings to the investor as this provides more relevant information to the investment decisions of the investors. The third measure used is abnormal accruals. Accrual accounting relates to the ability of management to either recognize earnings sooner, or to postpone recognition until the next period (Dechow & Dichev, 2002, 2002). Accruals may be used to increase the quality of earnings. The abnormal part, however, decreases earnings quality. The fourth and last
8
measure used is real earnings management. Instead of using accruals to manage accounting earnings, real earnings management means that managers actually influence earnings by changing the cash flows (Kim et al., 2012). These measures will be combined to form an answer to the research question: How does the adoption of IFRS affect the quality of earnings
of private companies in Europe?
This article will provide empirical research to find an answer to the research question. To measure earnings quality, the aforementioned measures which are found in prior earnings quality research will be used to see what effect IFRS adoption has on earnings quality. The data used consists of private company data that is publicly available on Bureau van Dijk and is split up in two samples. The main sample for this research consists of UK private companies. The second sample is a smaller sample of German sample used in a sensitivity analysis. The results of the four different measures indicate that overall, earnings quality is higher for firms using IFRS. Earnings persistence is lower and real earnings management increased for the UK sample of private firms that adopt IFRS, which shows a negative effect. The use of abnormal accruals decreased and timely loss recognition increased for this same sample showing a positive effect. However, the results in the sensitivity analysis on German private firms using IFRS show a positive result on all four measures. Combining these results led to the conclusion that, overall, the adoption of IFRS has a positive effect on the quality of earnings.
This article proceeds as follows: the next section will give a theoretical background on earnings quality and the adoption of IFRS in Europe. After this theory is provided, the third section will state the hypothesis tested in this article. Section 4 contains information about the data and methodology. In section 5, the different measures used in this research are mentioned and explained. Next, section 6 will show the results of descriptive statistics and the regressions. Finally, section 7 contains a discussion on the results and a conclusion on this research after which section 8 discusses the limitations and areas of interest for future research.
9 1. Theoretical Background and Measures used
The theoretical background will be discussed in two parts. The first part gives a description of earnings quality and uses the IFRS conceptual framework to give a theoretical base. The second part discusses financial reporting by both private and public companies in Europe.
2.1 Earnings quality
There is a large amount of research on earnings quality already and, as said in the introduction, it is expanding vastly. Some of the reasons behind this increase are earnings management, the development of high quality standards and new models that are used for the determination of earnings quality in new ways (Defond, 2010). The concept of the quality of earnings is fundamental in accounting and financial economics because total earnings (or net income) are the primary product of financial reporting as a measure of performance of a corporation (Dechow et al., 2010). Investors use earnings in their valuation of a firm by discounting the future expected cash flows. Investors make their investment decisions based on this valuation. As company earnings are at the base of the valuation, a good quality of earnings is highly important to investors and other interested parties (Dichev et al., 2013; Dechow et al., 2010).
To conceptualize earnings quality, it is useful to look at the definition of ‘quality’ first. Quality in its most basic form, as defined by Hernon et al. (2015), can be seen as fulfilling the expectations of the users of the information. When quality is related to earnings, earnings quality can then be explained as: “higher quality earnings provide more information about the features of a firm’s financial performance that is relevant to a specific decision made by a specific decision-maker.” (Dechow et al., 2010, p. 344). Dichev et al. (2013) add three important characteristics to this explanation. They argue that high quality earnings should reflect: consistent reporting choices over time, long-term estimates should be avoided as much as possible, and high-quality earnings should be sustainable. According to the IFRS conceptual framework, there are two fundamental characteristics of information. The first is relevance, which argues that information is relevant when it is capable of having influence on the decision made by capital providers (IASplus, 2016). Relevance thus looks at whether the information is useful to the investment decision by capital providers for example. The criterion of relevance is assumed to be fulfilled as earnings quality information is relevant to the investment decision by definition. The second fundamental qualitative characteristic is faithful representation. Earnings information is faithfully represented when it is a depiction of
10
an economic phenomenon that is complete, neutral and free from material error (IASplus, 2016).
This research specifically looks at the quality of earnings. It does so by assessing whether earnings are faithfully represented and if earnings are sustainable and repeatable. These characteristics are important as earnings that are ‘complete, neutral and free from material error’ as well as ‘repeatable and sustainable’ provide a solid base for investors to value a firm on. Earnings quality is related to the financial performance of the company, and quality is perceived as being ‘good’ when it satisfies the requirements of the users of the information. However, earnings quality may be lower than desired by the users for several reasons. Economy-wide influences may lead to lower persistence of earnings and errors can be made in the preparation of financial information. This research, however, is more interested in intentional acts of earnings quality distortion. Intentionally distorting earnings quality has two requirements. The first is a motivation to misreport. A way of explaining this is through the agency problem. The agency problem occurs when there are both a principle and an agent who have diverging interests under an information asymmetry. In this case, the agent is the company’s management and the principle is the investor (Fama, 1980; Fama & Jensen, 1983). Managers can exploit the information asymmetry by taking actions that are in their own best interest but that do not benefit the investor while preparing the financial statements. Examples of this may be managing losses into profits to attract or retain capital, or reporting managing earnings to gain a bonus. The management of earnings negatively influences the financial statements. This distorts the quality of this information and thus lowers the value of the information to investors. The second requirement for diverging earnings quality is that, besides the motivation to misreport, management also requires an opportunity. The opportunity to misreport is provided by exploiting ‘flaws’ in financial reporting standards. The IASB argues that the IFRS is a set of high quality financial reporting standards which should essentially have less ‘flaws’. Also, more disclosures are required which lower the information asymmetry. However, the opposition argues that increased discretion can also be used to report to managements’ own benefits which could result in management reporting inflated earnings without real economic substance for their own benefits.
11
2.2 Reporting in Europe
2.2.1 The implementation of IFRS
Responding to an explicitly acknowledged desire for harmonization in accounting standards, the IASC was formed in 1973 by collaborating national standard setters (Pope & McLeay, 2011). The formation, however, was not an initiative triggered on the supply side of financial reporting regulation. The formation of this committee was in response to a call for international harmonization of accounting standards as well as from countries that did not have the capacity to develop high quality standards for themselves. There was an explicit need for new reporting standards to increase both quality and comparability. According to Pope & McLeay (2011), the European Union was perhaps the most important international organization that wanted convergence of accounting standards (IASplus, 2016). Initially, the European Union had the desire to create their own financial reporting standards through the Fourth and Seventh directives. They wanted to create a comprehensive European accounting system based on the French ‘plan comptable’ but the differences within the European accounting standards were too diverse and consensus was impossible. This led to different countries retaining their own national generally accepted accounting principles (local GAAP) (Haller, 2002). Eventually in 2002, the European Commission decided to adopt the International Accounting Standards (IAS) as issued by the IASC up until 2001 and the subsequent IFRSs as issued by the IASB that replaced the IASC in 2001 (Pope & McLeay, 2011; Armstrong et al., 2010). This meant that all publicly listed companies in the EU had to mandatorily report their financial statements under IFRS on or after the fiscal year January the first2005. The goal of the EU was to achieve reporting harmonization in the EU and with the rest of the world to improve capital market integration. The EU did, however, adopted an endorsement approach which allowed the EC to (partially) reject the adoption of any standard leaving them with significant power.
The adoption of IFRS provides more benefits than just harmonization. The new standards may fill in the flaws of the earlier local GAAPs by addressing issues that were not earlier addressed or by providing clearer standards to issues that are too complex for local standard setters to solve (Callao & Jarne, 2010). Also, IFRS requires more disclosure than other standards which improves reporting by solving the information asymmetry. There have also been negative aspects on the adoption of IFRS. The standards offer greater flexibility to managers which may lead to greater discretion and opportunistic behavior. Also, the use of professional judgement may lead to judgemental disagreements or errors (Ormrod and Taylor, 2004). These are viable critics but as argued by multiple articles, the switch from local GAAP
12
to IFRS is new to both standard setters as well as adopters. Significant (change in) expertise is required by both parties which may lead to an initial increase in discretionary accounting but which will eventually improve as the use of judgement and new standards is becoming more familiar (Callao & Jarne, 2010; Barth et al., 2008). Also, the initial increase in discretionary accounting may reflect a gradual shift into the use of IFRS by initially changing policies based on national standards. This argument may be supported by the fact that even though US GAAP was initially seen as being of higher quality because it involved less earnings management, the SEC now recognizes IFRS as equivalent to US GAAP. The SEC accepted the use of IFRS in foreign company reporting on US stock exchanges without reconciliation since 15 November 2007. This essentially means that the SEC acknowledges IFRS to be equivalently valuable as US GAAP which is the only other set of financial reporting standards recognized to be of high quality (Kaya & Pillhofer, 2013).
Ultimately, it will come down to a cost-benefit approach. While both exist, the need to converge to a more internationally oriented set of accounting standards was (and is) inevitable for local GAAP users as the business and investment climate is becoming more and more internationally oriented. While the cost of switching to IFRS may have been high, the cost of switching to US GAAP (which was the only viable alternative) would have been even higher because of the voluminous rules based standards aimed at US business instead of the international business environment (Pope & McLeay, 2011). Also, the rules-based US GAAP standards may not lead to optimal information accumulation in different regimes and cultures around the world. Accounting discretion within IFRS may be used to report the information in ways that is more in line with the regional differences while retaining international comparability.
2.2.2 Reporting by public companies
Public companies are companies in which ownership and control is separated. The equity of such a corporation is sold through an initial public offering (IPO) and those shares can subsequently be freely traded on a stock exchange (Bahadir et al., 2015).
As argued before, the literature on earnings quality for public companies that adopt IFRS is widespread. The results found on the subject are mixed but the general expectation, as argued by Ball & Shivakumar (2003), is that adopting high quality standards is a necessary condition for high quality information. Some literature confirms this as they show that the adoption of IFRS following local GAAP leads to improved quality of earnings (Zeghal et al., 2012; Barth et al. 2008). In these cases, IFRS adoption led to less earnings management, more
13
value relevance and more timely loss recognition. Other studies document a negative effect on earnings quality (Callao & Jarne, 2010; Cameran et al., 2014); or no effect at all (Van Tendeloo & Vanstraelen, 2005). Results showing a negative or no relation are often found in literature using relatively older data. Other reasons for the mixed results argued by Barth et al (2008), may be that firms already start to transition towards IFRS under their local GAAP. Another reason may be that developing economies lack the mechanisms to enforce the application of IFRS or that the incentives of firms or their economic environment may negatively influence the effect of IFRS adoption. Zeghal et al. (2008) argue that the positive results on the quality of earnings increase following the use of IFRS when more recent data is used. This may indicate that the negative effects of the adoption were due to its early stage, and that the positive effects of IFRS are increasing as IFRS, and the application of IFRS, is improving over time.
While there is a large body of research on the quality of reporting under IFRS for public companies, these results cannot be assumed to be the same for private companies (Ball & Shivakumar, 2005; Cameran et al., 2014). Fundamental differences between public and private companies may lead to different effects of IFRS adoption. The most important difference is the way in which companies acquire capital. Because public companies acquire capital by selling shares on public stock markets, the demand for high quality (public) financial information is high. The financial statements are heavily relied on by external investors as at arm’s length investors do not have access to private information (Burgstahler et al., 2006). One of the main indicators of company performance that these investors use is reported earnings. If the quality of this information source is assessed as low, investors will be reluctant to provide capital to corporations. Looking at the issue through agency theory, managers’ incentives are double sided. On the one hand, they want to provide high quality information to acquire capital from shareholders. On the other hand, they may try to mislead the investor providing inflated earnings numbers because of self-interested motives which is made possible by the information asymmetry. Stronger financial reporting standards may prevent managers from reporting misleading numbers by improved standards and improved disclosures.
14 2.2.3 Reporting by private companies
As argued before, the results of research on public company earnings quality after IFRS adoption may not be the same for private companies. This can be explained by the different motivations that private and public companies have for reporting. In public companies, ownership is separated from control while private companies often have relatively concentrated ownership structures with inside owners or outsiders that can acquire inside information. These concentrated ownership structures thus allow private communication to inform the shareholders through contracting and other private channels (Burgstahler et al., 2006; Pope & McLeay, 2011). Since there are other motivations behind corporate reporting, as well as other information channels, private company corporate reporting may lead to (and require) less informative earnings. Information dispersion through an insider model was also mentioned by Ball & Shivakumar (2005). They argue that public statements are not necessary in contracting with lenders, institutional investors, managers and other parties. Also, when ownership and control are not separated, there is no agency problem like in public corporations which leads to problems with earnings quality. Private companies accumulate capital through other channels like institutional investors instead of dispersed shareholders or banks. Institutional investors and banks often have access to private information and more information processing capabilities and information then small investors leading to less problems with the quality of reporting.
The question that rises here, however, is why do private companies then adopt IFRS which is expected to lead to higher quality financial reporting at higher cost? It seems that they want to increase the informativeness of their financial statements and this can be motivated by a changing culture with changing standards which both require and lead to more information in financial statements (Callao & Jarne, 2010). As argued, banks and institutional investors have more information procession capabilities and companies may adopt IFRS which is argued to be of higher quality to give a positive signal to their investors. Also, globalization is increasing the opportunities for capital accumulation worldwide. Reporting under internationally accepted and adopted reporting standards may increase comparability and understandability for a wider range of investors, thereby increasing the attractiveness for international investors. However, more informative and higher quality financial statements as well as harmonization in financial reporting are required for this to succeed. Another benefit of IFRS adoption is that it leads to more disclosures in the financial statements. While simple dispersed investors may only look at the numbers, banks and large institutional investors can better interpret these disclosures and determine the (financial) health of a company. Private
15
companies thus do have an incentive to report high quality earnings and therefore may be tempted to adopt IFRS. This research will therefore assess if the adoption of IFRS does indeed increase the quality of earnings (Ball & Shivakumar, 2005; Cameran et al., 2014).
2.2.4 Private reporting in the UK
The motivation to write this article was to find if reported earnings quality of private companies in Europe improved through the adoption of IFRS. As the data for UK private companies is the most complete and suitable for earnings management calculations, the regressions used focus on UK private company data. The results of the determined earnings quality in the UK are expected to be generalizable for several reasons.
First the European Union has a goal of achieving accounting harmonization in the EU and therefore adopted IFRS (IAS) (Pope & McLeay, 2011). The adoption of a single set of accounting standards increases comparability of reporting which also includes reported earnings quality. As earnings are determined in a largely similar way for every company, the effect of IFRS adoption on the reported earnings is likely to be similar as well. Secondly, different private companies throughout the EU are likely to report for the same motivations which were discussed in section 2.2.3. Reported earnings quality depends, amongst other things, on financial reporting standards quality as well as the motivation for reporting (Kaya & Pillhofer, 2013). Also, it is argued by Callao & Jarne (2010) that there is a significant shift in the philosophy of accounting in Europe from rules-based to principles-based accounting. It becomes apparent that changes in accounting (standards) happen for Europe as a whole instead of individual countries separately which makes the chance of (large) differences between these countries unlikely.
One difference between European countries worth both mentioning and looking into is the fact that there are differences between regulations. Two different regulatory clusters can be found, common-law and code-law. Common-law countries, like the UK, have a high degree of (outsider) investor protection (Leuz et al., 2003). Leuz et al. (2003) argue that earnings management, leading to low earnings quality, in these countries is lowest. Code-law countries, like Germany, have a lower degree of outsider protection and focus on insider protection. This allows insiders of a company, either managers or controlling investors, to manage reported earnings to their interest. However, private companies acquire most of their capital from non-controlling outsider investors like banks and institutional investors who have a higher ability to process information as well as the capability to acquire inside information (Burgstahler et al., 2006; Pope & McLeay, 2011). As common-law countries already protect
16
the outside investors, signaling that these private companies have high quality financial statements may be of less importance than in code-law countries as common-law countries already offer investor protection through regulation. The adoption of IFRS in code-law countries may therefore very well be a more credible signal that the company has an outsider focus as it offers more disclosures than the previously used GAAP in an attempt to show to the outside (international) investors that the reporting company is a credible investment.
3. Hypothesis Development
This section will provide the hypothesis used in this research. One main hypothesis is used to look at the overall effect of IFRS adoption on earnings quality. This hypothesis is then separated in sub-hypotheses for each variable that is used in this research. The sub-hypotheses shortly discuss the used measures and a more comprehensive explanation will be given in section 5.
The first hypothesis is the main hypothesis used for this research. It looks at the effect that (voluntary) adoption of IFRS has on earnings quality in private European companies. The expectation is that IFRS adopters display higher earnings quality as IFRS is generally seen as being of higher quality than local GAAP. This leads to the following main hypothesis.
H1: Private European companies that have adopted IFRS (overall) show higher earnings quality than private European companies that use local GAAP.
Four sub-hypotheses are used to measure earnings quality and for each we use one sub hypothesis to look at the effect of IFRS adoption through that specific measure. The first sub-hypothesis used is earnings persistence. It is argued that investors can better assess the future performance of a company when earnings are persistent. Earnings persistence has a positive effect on earnings quality so it is expected to increase with the use of IFRS.
H1a: Private European companies using IFRS show a higher degree of earnings persistence than private companies reporting under local GAAP.
The second sub-hypothesis relates to timely loss recognition. Investors prefer losses to be recognized when they are likely to happen; while managers want to show positive results and thus postpone the recognition of losses as long as possible (Ball & Shivakumar, 2005). Early recognition of losses provides timely information to investors to base their investment decisions on and thus improves earnings quality. It is therefore expected that private companies using IFRS show a higher degree of timely loss recognition than those reporting under local GAAP.
17
H1b: Private European companies using IFRS have more timely loss recognition than private companies reporting under local GAAP.
The next sub-hypothesis looks at accrual quality. Accruals can be of high quality when it maps earnings more closely with cash flows. However, accruals can also be used by management for opportunistic behavior. These accruals are called abnormal accruals. In this case, accruals are costly to the organization and negative to investors. We thus expect the use of abnormal accruals to be lower for private European companies that report under IFRS compared to local GAAP users.
H1c: Private European companies reporting under IFRS use less abnormal accruals than private companies reporting under local GAAP.
The last sub-hypothesis relates to real earnings management. Real earnings management is used when management manipulates earnings by manipulating the real cash flows of the organization instead of using accruals. Real earnings management has a negative effect on both the organization and the quality of earnings to investors. It is therefore expected that the use of IFRS leads to lower use of real earnings management.
H1d: Private European companies reporting under IFRS use less real earnings management than private companies reporting under local GAAP.
4. Dataset and methodology
This study will be performed as an empirical archival research using panel data in cross-company regressions for the years 2008 - 2014. As this research requires data for private European companies, the Bureau van Dijk database from Amadeus is used. Bureau van Dijk gathers both financial and non-financial data on private European companies and is therefore well suited for this research. Stata is used to perform regressions on the data collected. Companies that have adopted IFRS are compared to non-adopters to see the effects of private company IFRS adoption on earnings quality.
The initial sample collected contained data for the years 2006-2014. Two extra years were required to calculate lagged variables and yearly changes within variables while keeping data for the years 2008-2014. Also, the initial sample consisted of four European countries (Spain, France, Germany and the UK). After the data manipulations, however, only UK and German private company data remained. The final sample was biased with only relatively little German data (76 firms and 234 observations) and a relatively large UK dataset (2142 firms and 7769 observations). For this reason the data is split up. The main sample consists of
18
UK private companies only. The German data sample is used in a separate sensitivity analysis.
The data collection and manipulation process was as follows. UK (German) private company data was collected which provided an initial sample of 13871 (5417) firms with 83651 (36906) observations. After deleting financial firms and manipulating the data 3204 (140) firms and 12705 (489) observations were left. Then lagged variables and differences between yearly data were calculated leaving a final data sample of 2142 (76) firms with 7769 (234) observations. As the other two countries left no usable data, they are removed from the analysis.
Table 1: Data collection and manipulation.
UK Germany
Initial number of observations (firms) 83651 (13871) 36906 (5417)
Observations (firms) after deleting 12705 (3204) 489 (140) financial firms and blank data cells
Observations (firms) after preparing data for
regressions (final sample) 7769 (2142) 234 (76)
5. Measures used
To find whether the choice between IFRS and local GAAP has implications for earnings quality, this article measures earnings quality through four different methods. Three of these are found in an article on earnings quality by Dechow et al. (2010). These measures were chosen because both the article and the different measures used within the article were often used and cited in the earnings quality literature. The three variables used are earnings persistence, accrual quality, and timely loss recognition. The fourth measure on earnings quality used is real earnings management. This measure is also used often throughout the literature and it is interesting to compare its results to discretionary accruals to see whether managers only change the numbers or are prepared to change the actual cash flows of the organizations when (and if) they perform earnings management. As a lot of earnings management research focuses on public companies, the measures used often include data that is unavailable to private companies. Keeping this in mind, the measures used in this research were also selected for its relevance to private companies.
19
5.1 Earnings persistence
The first measure used is earnings persistence. According to Burnett et al. (2015) earnings persistence means that current earnings can be used as an indicator to predict future earnings. When a company is able to draw conclusions from current earnings to predict earnings in the future, the corporation has a good quality of earnings (Dichev et al, 2013; Lin et al., 2012). Dechow et al. (2010) add that earnings persistence is mainly used by investors to calculate the value of a company. Sustainable earnings are a good input for the valuation of a company. The downside, however, is that earnings persistence can be manipulated on the short term. Often, a simplified model of earnings persistence is used like in Dechow et al. (2010) who regress earnings on the previous years’ earnings. The model used is found in a study by Sloan (1996).
Earnings persistence:
Earnings𝑡+1= α + β1 Earnings𝑡+ 𝜀𝑡 (1)
Earnings𝑡+1= Net income for the next period scaled by assets Earnings𝑡 = Net income for the current period scaled by assets
𝜀𝑡 = Error term
As explained, this model calculates whether the earnings of a certain year are predictable based on the previous years’ earnings. The predictability of earnings is shown in β1. The regression is ran for the sample of IFRS and GAAP users separately to see how the choice of reporting regime influences earnings persistence. The influence is shown by the difference between β1 under both regimes.
5.2 Timely loss recognition
Timely loss recognition is a crucial attribute for earnings quality because it improves the quality of information for the user. Timely loss recognition is even more essential for private companies as it is important in loan agreements as argued by Ball & Shivakumar (2005). As a proxy for earnings quality, the authors argue that timely loss recognition has such a high quality as all balance sheet items (and all proxies calculated from them) flow through the income statements. When losses are recognized in the income statement on a timely base through conservatism, the user of the financial information is better informed for loan-pricing and debt covenant related repricing or breaches (Sun et al., 2011). The accrual-cash flow
20
model from Ball and Shivakumar (2005) is used which relates accruals to cash flows. This model is used as it was specifically implemented for a sample of private firms.
Timely loss recognition:
𝐴𝐶𝐶𝑡 = 𝛽0+ 𝛽1 𝐷𝐶𝐹𝑂𝑡+ 𝛽2 𝐶𝐹𝑂𝑡+ 𝛽3 𝐷𝐶𝐹𝑂𝑡𝑥 𝐶𝐹𝑂𝑡+ 𝜀𝑡 (2)
𝐴𝐶𝐶𝑡 = (assumed) linear relation between cash flows and accruals
= 𝐴𝐶𝐶𝑡 = ∆𝐼𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 + ∆𝐷𝑒𝑏𝑡𝑜𝑟𝑠 (+ ∆𝑂𝑡ℎ𝑒𝑟 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑎𝑠𝑠𝑒𝑡𝑠 − ∆𝐶𝑟𝑒𝑑𝑖𝑡𝑜𝑟𝑠 − ∆𝑂𝑡ℎ𝑒𝑟 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 − 𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛)
𝐷𝐶𝐹𝑂𝑡 = dummy variable which is 1 if 𝐶𝐹𝑂𝑡 is negative
𝐶𝐹𝑂𝑡 = Earnings before exceptional and extra-ordinary items less accruals
As argued by Ball & Shivakumar (2005), the timely loss recognition is shown in 𝛽3. This coefficient shows whether (and how much) companies actually recognize negative earnings when DCFO (earnings less accruals) is negative. In other words, 𝛽3 shows the amount of
negative income that is recognized on a timely base.
5.3 Abnormal accruals
The third measure used relating to earnings quality is abnormal accruals. Accrual accounting relates to the ability of management to either recognize earnings sooner, or to stall recognition until the next period (Dechow & Dichev, 2002). In other words, it is to managements discretion when they want to recognize earnings. According the Zeghal et al. (2012) accruals can be used to make earnings linked more closely to cash flows of the organization. When this is done, these are ‘normal’ accruals which improve the quality of earnings. Accrual accounting is valuable when it improves the quality of information transferred in the financial statements (Penman & Zhang, 2002) Earnings quality thus improves with (high quality) accruals that map earnings better into cash flows. However, accrual accounting can also be used by managers to fictionally improve their performance to receive higher performance related compensation. When this is the case, accrual accounting leads to the manipulation of earnings without any implication for real cash flows. When this is done, accrual accounting is costly for the organization as it decreases investor trust and increases management payments for performances that they didn’t deliver (Dechow et al., 2002). The measure used to find accrual quality is found in the article by Kim et al (2012). They use a slightly modified version of the Jones (1991) discretionary accrual model as a proxy for the regression model.
21
Income before extraordinary items are excluded from the regression as Bureau van Dijk did not provide enough information to calculate this.
Proxy:
𝑇𝐴 𝐴⁄ 𝑖𝑡−1 = 𝛼0+ 𝛼1(1 𝐴⁄ 𝑖𝑡−01 ) + 𝛼2 𝐼𝐹𝑅𝑆 + 𝛼3(∆𝑅𝐸𝑉𝑖𝑡− ∆𝑅𝐸𝐶𝑖𝑡) 𝐴⁄ 𝑖𝑡−1+
𝛼4𝑃𝑃𝐸𝑖𝑡⁄𝐴𝑖𝑡−1+ 𝜀𝑡 (3)
𝑇𝐴𝑖,𝑡 = Total accruals = (∆Current Assets - ∆Cash) – (∆Current Liabilities) - Depreciation and Amortization Expense
∆𝑅𝐸𝑉𝑖𝑡 = Change in net revenue from year t-1 to year t
𝑃𝑃𝐸𝑖𝑡 = Change in plant, property and equipment from year t-1 to year t
IFRS = Dummy variable, equals 1 if the company uses IFRS as reporting standards
𝜀𝑡 = Abnormal accruals (DA)
Residuals are used to determine the absolute value of abnormal accruals. A larger standard deviation of the residuals means that the (abnormal) accruals are used which lower earnings quality. The value of abnormal accruals (ABS_DA) is then used in the following regression to determine whether IFRS has an influence on the use of discretionary accruals:
Regression model:
𝐴𝐵𝑆_𝐷𝐴𝑡 = ∝0+∝1 𝐼𝐹𝑅𝑆𝑡+ ∝2 𝐶𝑂𝑀𝐵𝐼𝑁𝐸𝐷𝑅𝐴𝑀𝑡+∝3 𝑆𝐼𝑍𝐸𝑡+ ∝4 𝑅𝑂𝐴𝑡+ ∝5 𝐿𝐸𝑉𝑡+ ∝6 𝐹𝐼𝑅𝑀 𝐴𝐺𝐸𝑡+ 𝜀𝑡 (4)
ABS_DA = absolute value of 𝜀𝑡 from equation (3)
𝐼𝐹𝑅𝑆𝑡 = proxy for accounting practice, equals 1 if the firm uses IFRS
𝑆𝐼𝑍𝐸𝑡 = natural logarithm of total assets.
𝑅𝑂𝐴𝑡 = income scaled by lagged total assets
𝐿𝐸𝑉𝑡 = long-term debt scaled by total assets
𝐹𝐼𝑅𝑀 𝐴𝐺𝐸𝑡 = natural logarithm of (1+ number of years since the firm first appeared in BvD)
In this regression model ∝1 shows the influence that IFRS has on the use of abnormal
22
5.4 Real activities manipulation
Next to accrual accounting, which only alters the timing of earnings while not changing the cash flow of the organizations, organizations may also use real activities manipulation (RAM). Real activities manipulations are changes made by management in order to meet or beat earning thresholds (Kim et al., 2012). However, different from accrual accounting, real activities manipulation involves altering the actual cash flows of the organization instead of only the numbers. The empirical model used by Kim et al. (2012) is used to measure real activities manipulation. They use three measures and combine these three in a fourth combined measure of real activities manipulation. Their measures are: “(1) abnormal levels of operating cash flows (AB_CFO), (2) abnormal production costs (AB_PROD), (3) abnormal discretionary expenses (AB_EXP), and (4) a combined measure of real activities manipulation.” (Kim et al., 2012, p. 768). The combined RAM is calculated as follows: AB_CFO – AB_ PROD + AB_EXP. Some adjustments had to be made due to the unavailability of data requirements. As the measure from the study by Kim et al (2010) focus on public companies (with public data), some data was not available for our regressions and are excluded. Also, data for the last part of combined_RAM calculation (discretionary expenses) is unavailable for private companies, AB_EXP is therefore also excluded from the calculation. Proxies: 𝐶𝐹𝑂𝑡 𝐴𝑡−1 = ∝0+ ∝1 ( 1 𝐴𝑡−1) + 𝛽1 ( 𝑆𝑡 𝐴𝑡−1) + 𝛽2 ( ∆𝑆𝑡 𝐴𝑡−1) + 𝜀𝑡 (5)
𝐶𝐹𝑂𝑡 = Cash flow from operations in the year t
𝑆𝑡 = Sales in the year t
∆𝑆𝑡 = Change in sales in year t from year t-1
𝜀𝑡 = abnormal cash flow from operations (AB_CFO)
𝑃𝑅𝑂𝐷𝑡 𝐴𝑡−1 = ∝0+ ∝1 ( 1 𝐴𝑡−1) + 𝛽1 ( 𝑆𝑡 𝐴𝑡−1) + 𝛽2 ( ∆𝑆𝑡 𝐴𝑡−1) + 𝛽2 ( ∆𝑆𝑡−1 𝐴𝑡−1) + 𝜀𝑡 (6) 𝑃𝑅𝑂𝐷𝑡 = 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡𝑠 (= 𝐶𝑂𝐺𝑆𝑡+ ∆𝐼𝑁𝑉𝑡 ) 𝐶𝑂𝐺𝑆𝑡 = 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑔𝑜𝑜𝑑𝑠 𝑠𝑜𝑙𝑑 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡
∆𝐼𝑁𝑉𝑡 = change in inventory in year t from year t-1
23
∆𝑆𝑡 = Change in sales for the year t from t-1
∆𝑆𝑡−1 = Change in sales for the year t-1 from t-2
𝜀𝑡= abnormal production costs (AB_PROD)
The proxies described here are used to capture the abnormal part of Cash flow from operations and Production costs which are the error terms 𝜀𝑡 from equation (5) and (6). The abnormal values are used for the real earnings management regression model (7).
Regression model:
𝐶𝑂𝑀𝐵𝐼𝑁𝐸𝐷_𝑅𝐴𝑀𝑡 = ∝0+∝1 𝐼𝐹𝑅𝑆𝑡+ ∝2 𝐴𝐵𝑆_𝐷𝐴𝑡+∝3 𝑆𝐼𝑍𝐸𝑡+ ∝4 𝑅𝑂𝐴𝑡+ ∝5 𝐿𝐸𝑉𝑡+ ∝6 𝐴𝐺𝐸𝑡+ 𝜀𝑡 (7)
𝐶𝑂𝑀𝐵𝐼𝑁𝐸𝐷_𝑅𝐴𝑀𝑡 = AB_CFO – AB_ PROD + AB_EXP AB_CFO = 𝜀𝑡 from the CFO regression (5) AB_PROD = 𝜀𝑡 from the PROD regression (6)
𝐼𝐹𝑅𝑆𝑡 = proxy for accounting practice, equals 1 if the firm uses IFRS
ABS_DA = absolute value of 𝜀𝑡 from the discretionary accruals regression (4)
𝑆𝐼𝑍𝐸𝑡 = natural logarithm of total assets.
𝑅𝑂𝐴𝑡 = income scaled by lagged total assets
𝐿𝐸𝑉𝑡 = long-term debt scaled by total assets
𝐴𝐺𝐸𝑡 = natural logarithm of (1+ number of years since the firm first appeared in BvD)
By combining the proxies in the combined RAM, the regression of real earnings management can be executed. IFRS (∝1) shows whether the use of IFRS as accounting practice has an influence on real earnings management. A negative coefficient is desired as this shows that the use of IFRS lowers real earnings management.
24 6. Results
This section will show the results of this study. It will begin by shortly discussing descriptive statistics and correlations. Following this, the results of the regressions will be shown and interpreted. As explained, the two datasets were split up. The main sample of UK private firms will be discussed first and the German sample of firms will be discussed separately in a sensitivity analysis. Finally, an additional test is performed for earnings persistence.
6.1 Descriptive statistics
Table 2 shows the descriptive statistics on the variables used in this research. All variables were scaled by lagged total assets unless specified otherwise in the used literature. After scaling for firm size effects, the variables were winsorized on the 1% level to exclude significant outliers from affecting the regressions. Winsorization was used instead of dropping the outliers as this would remove too much of the data.
The mean for the IFRS dummy is 0.0919 which means that only 9.19% of the sample used has IFRS as accounting practice. While rather low, it does still provide significant results.
For earnings persistence, it can be observed that the mean (0.301) is a little bit to the right which indicates that firms more often have positive earnings than not. The negative mean (-0.0424) for ACCt from the timely loss recognition total accrual measure may indeed show that firms use negative accruals for timely loss recognition as suggested by Ball & Shivakumar (2005). We cannot really compare the mean to that in their research as they report a mean of 0.00 (-3.8) for private (public) companies but the number found is close to theirs.
The descriptive statistics show ABS_DA of 0.0877182. This number is comparable to the mean of ABS_DA reported in a study by Kim et al. (2012) (0.200). The mean indicates that abnormal accruals are positive on average, so that abnormal accruals are used to manage earnings upwards. AB_CFO and AB_PROD also have a different mean from the ones in the paper of Kim et al. (2012). AB_CFO is 0.129 in the paper while -0.0002 in this sample, resPROD is -0.096 in the article while 0.001 in this sample. This may show that the proxies for real earnings management both play a smaller and a different role in private companies compared to that of public companies. The descriptive statistics on COM_RAM show a mean of -0.0011 while that is 0.238 in the paper. Again, the number is smaller and opposite of the number found in Kim et al. (2012). This may be caused due to the lack of data for the
25
regression or simply because real earnings management plays a different or less important role in private companies compared to public companies.
Table 2: Summarized variables.
Table 1 Panel A, unmatched, FEA & NON
(1) (2) (3) (5) (6)
VARIABLES N Mean SD Min Max
IFRS 7,769 0.0919 0.289 0 1 NIt+1 7,769 0.0301 0.0786 -0.316 0.280 NIt 7,769 0.0295 0.0939 -1.453 1.453 ACCt 7,769 -0.019 0.1192 -0.3719 0.4501 CFO 7,769 0.0546 0.124 -0.704 0.656 DCFO 7,769 0.243 0.429 0 1 CFOxDCFO 7,769 -0.0227 0.0659 -0.704 0 T_ACC 7,769 -0.0424 0.136 -0.866 1.008 ∆REV 7,769 0.142 0.442 -5.754 5.049 ∆REC 7,769 0.000539 0.0682 -0.539 0.539 PPE 7,769 0.312 0.253 0.00145 1.083 COM_RAM 7,769 -0.0011 0.378 -2.932 4.276 AB_DA 7,769 0.0873 0.1019 0.0001 0.9710 AB_CFO 7,769 -0.002 0.0842 -0.0401 0.4479 AB_PROD 7,769 0.001 0.350 -4.345 2.635 SIZE 7,769 18.71 1.412 14.41 25.58 ROA 7,769 0.0360 0.0809 -0.254 0.307 LEV 7,769 0.206 0.247 0.000716 1.685 AGE 7,769 2.300 0.211 0.693 2.398
Variables are defined in appendix A.
The correlations are done for each regression separately. For earnings persistence, Earnings (t+1) is highly and significantly correlated with earnings (t) (0.5172). This is not strange, however, as the goal of the model is to predict future earnings by past earnings so correlation is desired in this model. Also found in this model are the variables related to the additional test to calculate earnings persistence. Sloan (1996), argues that earnings based on cash flows are more persistent than those based on accruals. The correlation tables show that earnings have a high correlation with cash flows (0.8583). This number, also, is higher than the correlation of earnings with accruals (0.1524). This means that a larger part of earnings moves with cash flows, which are persistent, compared to accruals which are less persistent.
26 Table 3 Earnings persistence correlations
NIt+1 NIt T_ACC CFO NIt+1 1,00 NIt 0.5173*** 1,00 0.0000 T_ACC 0.1524*** 0.0890*** 1,00 0.0000 0.0000 CFO 0.8583*** 0.4703*** 0.0370*** 1,00 0.0000 0.0000 0.0009 *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
The correlation matrix for timely loss recognition has several highly correlated variables. These correlations can be explained by the fact that the cash flows are predicted by earnings minus accruals. Accruals, thus, is part of the calculation of the cash flows and hence the correlations.
Table 4: Timely loss recognition correlations
ACCt DCFO CFO CFOxDCFO
ACCt 1,00 DCFO 0.5103*** 1,00 0.0000 CFO -0.7794*** -0.6741*** 1,00 0.0000 0.0000 CFOxDCFO -0.6355*** -0.6068*** 0.7428*** 1,00 0.0000 0.0000 0.0000 *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
27
The last correlation table combines the abnormal accruals and timely loss recognition correlations. COMBINED RAM is significantly reasonably highly correlated with ROA. This may indicate that more profitable firms are more likely to engage in real earnings management.
Table 5: ABS_DA and COMBINED RAM correlations
COM_RAM ABS_DA IFRS SIZE ROA LEV AGE
COM_RAM 1,00 ABS_DA -0.0245** 1,00 0.0287 IFRS 0.0047 -0.0432*** 1,00 0.6811 0.0001 SIZE -0.1786*** -0.0292*** 0.2222*** 1,00 0.0000 0.0090 0.0000 ROA 0.4401*** 0.0044 -0.0504*** -0.1212*** 1,00 0.0000 0.6921 0.0000 0.0000 LEV -0.0906* -0.0806*** 0.0706*** 0.2328*** -0.3536*** 1,00 0.0000 0.0000 0.0000 0.0000 0.0000 AGE -0.0214* 0.0128 .-0.1384*** -0.0030 0.1031*** -0.1795*** 1,00 0.0557 0.2526 0.0000 0.7906 0.0000 0.0000 *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
28
6.2 Main results
This section will discuss the results of the regressions on the different calculations for earnings quality. For each regression, a table with results and an interpretation of these results in relation to the hypothesis is given.
6.2.1 Earnings persistence
The first result discussed is that of earnings persistence. Earnings persistence increases information quality for decision making to investors; we would therefore expect earnings persistence to be higher when firms use IFRS as accounting practice. This, however, is not the case looking at the regression results of earnings persistence. For firms using IFRS, the coefficient is 0.265 while this is 0.453 for GAAP users indicating that GAAP users’ earnings in the next year are more predictable based on the current years’ earnings. From the regression, it becomes clear that earnings persistence is not better for firms using IFRS than for firms using GAAP. H1a is therefore rejected for the UK sample.
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
6.2.2 Timely loss recognition
Next is timely loss recognition. As discussed before, timely recognized losses increase information quality to investors. Increased timeliness of recognition of losses improves the decision making capability of investors as they can now base their decision on information that is more relevant and material to that decision. Looking at the regression results, dCFO looks at when earnings are negative. CFOxDCFO then measures how much losses a company actually recognizes when earnings are negative. Ball & Shivakumar (2005) argue that the
(1) (2)
VARIABLES NIt+1 IFRS NIt+1 GAAP
NIt 0.265*** 0.453*** (0.0263) (0.00837) Constant 0.0165*** 0.0171*** (0.00313) (0.000805) Observations 714 7,055 R-squared 0.125 0.293
29
coefficient should be positive as this shows that companies recognize negative earnings sooner. Both coefficients in this regression are negative indicating that the firms in this sample postpone the recognition of losses. The coefficient of IFRS, however, is less negative than that of GAAP users indicating that timely loss recognition does improve when firms adopt IFRS. As H1b predicted that timely loss recognition improves when firms adopt IFRS, this hypothesis can be confirmed.
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
(1) (2)
VARIABLES TLR IFRS TLR GAAP
DCFO -0.0212** -0.0138*** (0.00967) (0.00281) CFO -0.640*** -0.687*** (0.0367) (0.0116) CFOxDCFO -0.141* -0.257*** (0.0735) (0.0201) Constant 0.00476 0.0177*** (0.00525) (0.00155) Observations 714 7,055 R-squared 0.515 0.627
30 6.2.3 Abnormal accruals
The next model looks at the effects of IFRS adoption on the use of discretionary accruals. As can be seen from the model, IFRS is both significant and negative (-0.012). This indicates that ABS_DA are used less in firms that adopt IFRS as accounting practice which is as is predicted in H1c. As discretionary accruals are the part of accruals used by management that have a negative influence on earnings quality, using less discretionary accruals improves quality of reported earnings. The R-squared of the model is very low but the results significantly indicate that IFRS adoption reduces discretionary accruals. This means that earnings quality is improved and H1c is therefore confirmed.
(1) VARIABLES ABS_DA IFRS -0.0101** (0.00411) COM_RAM -0.0130*** (0.00413) SIZE -0.00167** (0.000848) ROA 0.0875*** (0.0177) LEV -0.0396*** (0.00426) AGE 0.00770 (0.00559) Constant 0.110*** (0.0200) Observations 7,769 R-squared 0.013
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
6.2.4 Real earnings management
The next and last regression done is that of real earnings management. As real earnings management is a way of management to influence earnings by altering real cash flows at their own discretion, it negatively influences earnings quality. It is therefore expected that the use of IFRS would decrease real earnings management which would be indicated by a negative coefficient for IFRS in this regression. The IFRS coefficient found in the regression, however,
31
is significant and positive (0.0459). This result indicates that H1d should be rejected and thus that the use of IFRS increases the use of real earnings management and thereby reduces earnings quality. (1) VARIABLES COM_RAM ABS_DA -0.0579* (0.0310) IFRS 0.0474*** (0.0112) SIZE -0.0357*** (0.00235) ROA 1.802*** (0.0417) LEV 0.129*** (0.0142) AGE -0.0694*** (0.0153) Constant 0.736*** (0.0543) Observations 7,769 R-squared 0.225
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
6.3 Sensitivity analysis
In this last section of the results the same regressions performed on the UK sample are performed on the sample of German private firms as a sensitivity. These regressions are separated from the main regressions as the sample of German private firms was too small to include in the main regressions. The UK is a common-law country while Germany is a code-law country. This sensitivity analysis is done to see whether the difference in regulatory regime affects the influence of IFRS adoption on earnings quality.
6.3.1 Earnings persistence
The first regression for the German sample is earnings persistence. The results show that earnings persistence is relatively similar for the GAAP adopters in the UK sample. The amount of earnings persistence for the IFRS adopters in the German sample, however, is a lot
32
higher than that of the UK sample. Regarding the hypothesis, earnings persistence is higher for IFRS adopters indicating that H1a should be confirmed for the German sample.
(1) (2)
VARIABLES NIt+1 IFRS NIt+1 GAAP
NIt 0.472*** 0.461*** (0.0931) (0.0627) Constant -0.00664 0.0212*** (0.00671) (0.00570) Observations 88 146 R-squared 0.230 0.273
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
6.3.2 Timely loss recognition
The second regression looks at timely loss recognition. Both coefficients for CFOxDCFO are negative like in the UK sample indicating that private firms in both countries postpone loss recognition. What also becomes clear is that IFRS adopters have a less negative coefficient meaning that they postpone loss recognition less. In other words, timely loss recognition is better for IFRS adopters which means that hypothesis H1b can be confirmed.
(1) (2)
VARIABLES TLR IFRS TLR GAAP
DCFO -0.0575*** 0.00313 (0.0207) (0.0229) CFO -0.986*** -0.657*** (0.122) (0.0710) CFOxDCFO -0.115 -0.191 (0.270) (0.157) Constant 0.0165 0.0107 (0.0137) (0.0106) Observations 88 146 R-squared 0.594 0.615
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
33 6.3.3 Abnormal accruals
The next regression for the German sample of private firms is that of abnormal accruals. Looking at the coefficient of IFRS, it can be seen that the use of IFRS is negatively related to ABS_DA meaning that firms that have adopted IFRS use less abnormal accruals. As abnormal accruals reduce the quality of earnings, the adoption of IFRS has a positive influence on the use of abnormal accruals. This was also predicted by H1c and this sub-hypothesis is therefore confirmed for the German sample.
(1) VARIABLES ABS_DA IFRS -0.0251** (0.0118) COM_RAM -0.0551** (0.0265) SIZE -0.00681* (0.00409) ROA 0.0885 (0.0785) LEV -0.00998 (0.0265) AGE 0.0304 (0.0247) Constant 0.153 (0.0934) Observations 234 R-squared 0.066
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Variables are defined in appendix A.
