0
21
thof June, 2015
Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam
Supervisor: Dr. Reka Feleg
Msc Accountancy & Control, variant Accountancy
Earnings management in Europe: before and after
mandatory IFRS adoption
Mark van Tol
1
Statement of Originality
This document is written by Mark van Tol, 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
2
Abstract
In 2005 IFRS became mandatory for listed firms in Europe. I examine the effect on both accrual based and real earnings management practices after this framework became mandatory. My sample consists of firm observations from 1999 till 2011 and I used data from firms that were listed on the EUROSTOXX 600 index in the year 2001. For the analysis of the effect on both earnings management strategies I followed the work of Cohen et al. (2008). For the calculation of accrual based earnings management variables I used the modified Jones model (Dechow et al., 1995) and for the calculation of real earnings management I followed the methods first used by Roychowdhury (2006). The results that were obtained from this research point to the conclusion that the use of accrual based earnings management
increased after IFRS became mandatory, which is in support of my first hypotheses. This while the use of real earnings management practices shows a small decline after 2005, which in turn does not support my second hypotheses but the evidence here is not strong enough to fully reject this hypotheses. These results could indicate a substitution effect, which was earlier observed in other studies from Zang (2012) and Cohen et al. (2008).
3
Table of contents
1. Introduction 4
2. Literature review and hypotheses development 8
2.1 IFRS adoption 8
2.2 Impact of IFRS adoption on accrual based earnings management 11 2.3 Impact of IFRS adoption on real earnings management 15
3. Methodology and sample selection 17
3.1 Sample selection 17
3.2 Accrual based earnings management – research design 19 3.3 Real earnings management – research design 20
4. Results 24
4.1 Descriptive statistics 24
4.2 Time trend analysis 27
4.3 Substitution effect of earnings management 32
5. Conclusion 34
4 1. Introduction
In July 2002 the European Commission decided in Regulation nr. 1606/2002 that from the year 2005 all listed firms in the EU, whose securities are traded on regulated markets, are obligated to report their financial statements according to the rules and principles of International Financial Reporting Standards, also called IFRS (EC, 2002). This transition from local GAAP’s to a comprehensive set of rules posed a significant challenge for over 7000 firms. According to the regulators this adoption would increase the transparency of financial statements and It should be very beneficial for the users of those financial statements (CESR, 2003). The reason for this lies in the fact that IFRS applies fair value accounting. This means that firms have to value the amounts in their financial statements at their current fair value. This should protect investors to unseen decreases in the value of a firm. By looking at Leuz et al. (2003) I can conclude that an increase in this protection for investors should lead to enhanced quality of the financial reporting.
To assess whether the mandatory adoption of IFRS indeed leads to a better quality of reported earnings I investigate a widely used measure of reporting quality, namely the amount of earnings management. Earnings management is defined in the paper of Healy and Wahlen (1999). They define these practices as a process in which management uses their discretion in financial reporting and in structuring their transactions to alter financial reporting of the firm. This is done to mislead both
investors and stockholders of the firm about the underlying economic performance of the firm, or to influence some contractual arrangements that are based on certain accounting numbers (Healy and Wahlen, 1999).
From this I can conclude that there are two strategies to manage earnings. In the first case, firms alter the timing and structuring of real economic events to achieve a certain monetary goal. By doing this a firm can influence their own net income for a particular purpose, like meeting the expectations of investors. Research from
Roychowdhury (2006) and Cohen et al. (2008) shows that a lot of firms engage in this type of earnings management. This strategy of altering real economic events is called real earnings management. The second strategy that firms use to manage their earnings is called accrual based earnings management. In this case, managers use their judgment discretion in the estimated values within the financial statement
5 numbers to mislead the users of the financial statements about the economic
performance of the firm. Most of the times this is done with the intention of a temporary monetary benefit (Healy and Wahlen, 1999).
According to work from Barth et al. (2008) and Cohen et al. (2008) earnings management practices lead to a decrease in the quality of financial reporting. This is due to the fact that earnings management induces high costs of misallocation. When reporting does not give a fair view of underlying economic reality, earnings
management can lead to the case in which resources are allocated in a wrong way. Because of this the future value of a firm can suffer from earnings management, especially if management sacrifices future profits and future cash flows to increase the amount of income in the current reporting period (Gunny, 2005).
With the adoption of IFRS regulators try to enhance the comparability of
financial statements, improve corporate transparency and raise the quality of financial reporting (EC Regulation Nr. 1606/2002). This should lead to a better understanding of the financial statements by the users of those statements. Under the principles of IFRS it becomes less easy for firms to make opportunistic forecasts about values within their reporting because this set of rules puts a limit on managerial discretion which in theory should increase the quality of the reporting. So, when the standards under IFRS are compared to most of the local GAAP’s that were used before the adoption of IFRS it seems logical that this adoption leads to a decrease in the amount of earnings management within the financial statements (Doukakis, 2014). However, there is a lot of discussion about this. For example, work from Sunder (2009) points out that these new principles under IFRS lead to more volatility and subjectivity in reporting because fair values are mostly based on estimates made by managers. This could be seen as an incentive for managers to conduct in more earnings management.
Because of the controversy about the impact of mandatory IFRS adoption on earnings management practices this thesis investigates these effects on financial reporting. In this thesis I try to make a distinction between both accrual based
earnings management and real earnings management. According to Doukakis (2014) it is important to understand the different effects of mandatory IFRS adoption on both earnings management strategies. First, I investigate whether the mandatory adoption
6 of IFRS in Europe has led to a change in accrual based earnings management
practices. To assess this I use data from non-financial firms that are listed on the EUROSTOXX 600 - Index, which consists of the 600 biggest firms in Europe. I shall use data from three time periods. Namely, a period till the announcement in 2002, a period from the announcement till the mandatory adoption of IFRS in 2005 and the period after this adoption. I investigate whether there has been a change in accrual based earnings management practices by making use of the Jones model (Jones, 1991). This model looks at the amount of discretionary accruals to measure earnings management within financial reporting. The reason for this lies in the fact that
management has the opportunity to use their discretion in these accruals to manage their earnings with a certain purpose. This model is modified by work from Dechow et al. (1995), in which they assess different models for measuring accrual earnings management practices. This adjusted model is called the Modified Jones Model, and it is used in my thesis to predict the level of accrual based earnings management in both time periods. After that I investigate the change in real earnings management. To test this I use earlier studies to establish proxies for this type of earnings
management. Here I follow work from Roychowdhury (2006), Cohen et al. (2008) and Zang (2012). These papers used abnormal levels of cash flows from operations (CFO), discretionary expenses and the abnormal costs of production as proxies for the amount of real earnings management within financial reporting.
But these studies are focused on data from the US, where my study is focused on data from European firms. They all conclude that these activities, that I take as proxies in my research, are used to manage earnings. In work from Cohen et al. (2008), Zang (2012) and Cohen and Zarowin (2010) they also come to the
conclusion that both earnings management strategies are used as substitutes. I test for this possibility in my research to see whether this is also the case for Europe. Work from Doukakis (2014) shows no change in the use of both earnings
management strategies, but he uses a different method than I do. He compares levels of both earnings management strategies before and after IFRS adoption. He does this with the use of his own regression model. Instead of comparing both periods I perform a time trends analysis to investigate how the level of earnings management changes over the chosen time period.
7 The results that were obtained from my research show that from the year 2005, when IFRS became mandatory in Europe for listed firms, the use of earnings management practices are indeed influenced by it. It seems that IFRS adoption has increased the amount of accrual based earnings management, which is showed by an increase in discretionary accruals. The results for real earnings management
show a small decrease but this evidence is not very overwhelming, so it could also be that IFRS adoption has little influence on real earnings management practices.
This study contributes to prior literature by assessing the impact of mandatory IFRS adoption on both earnings management strategies. This is done using a longer time period then earlier papers do, which improves the quality of the results because it rules out short term effects on reporting quality. Also, there is much literature
written on the impact of IFRS on accrual based earnings management in Europe, but there is just few literature written about the impact on real earnings management. This distinction that I make in my thesis is important because it is widely known in literature that real earnings management is much more harmful for a firm than accrual based earnings management practices are (Cohen et al., 2008;
Rowchudhury, 2006). Also this is the first paper that uses the method first used by Roychudhury (2006) on firms from Europe to assess the effect of IFRS adoption. And I contribute to prior literature by investigating the possible trade-off between both earnings management strategies in an European context.
The remainder of my thesis is as follows. In the second section I give a review of the earlier literature written on this topic. In the third section I explain the methods that are used and the sample that is selected in my thesis. In section four the results of the tests are presented. In section five I state my conclusion about the research question that I try to answer. And the sixth section gives references that were used in this thesis.
8 2. Literature review and hypotheses development
In this chapter of my thesis the known literature on earnings management in Europe is investigated. First I look at some papers written about the adoption and impact of IFRS on reporting quality. Here I also look at whether earnings management
practices are used to avoid earnings decreases and losses, and whether these practices changed after the adoption of IFRS. After this I investigate papers concerning the effect of IFRS adoption on the amount of accrual based earnings management. The literature on this earnings management strategy is used to derive a hypotheses about the impact of mandatory IFRS adoption on accrual based
earnings management. After that I discuss some literature on the effect of IFRS adoption on real earnings management practices. But, due to the fact that there is just few literature written on this I also look at research from outside Europe. From this I derive a second hypotheses about the impact of IFRS adoption on real earnings management practices.
2.1 IFRS adoption
First it is a good thing to know more about these International Financial Reporting Standards, or so called IFRS. These standards are issued by the
International Accounting Standards Board (IASB) and they should become the global standard used in preparing the financial reporting of all public firms. With this
comprehensive set of accounting principles the IASB wants to issue standards that are of high quality, understandable and enforceable for people from all countries of the world. This should lead to highly transparent and comparable information to help users of financial statements in capital markets (Ball, 2006). So, from this I can conclude that the main goals of IFRS are increasing reporting quality and to bring convergence in financial reporting.
The first paper that is investigated in this literature review is that from Barth et al. (2008) in which they look at whether the adoption of IFRS leads to higher
accounting quality. Like Ball (2006), they also state that the goal of these standards was to develop a set of high quality financial reporting standards that should be acceptable in many different countries. To achieve this the IASB came up with
9 standards that are not based on rules, but on principles. This should better reflect the financial position and performance of a firm, while it should also increase the
comparability between firms in different countries. These standards should also limit the possibilities that managers of firms have to use their discretion for earnings management practices (Barth et al. 2008). So, from this viewpoint it seems that this comprehensive set of principles should lead to higher reporting quality and less earnings management within financial reporting because it puts a limit on the amount of discretion that managers have in accounting estimates.
This claim that a tighter accounting system leads to less earnings
management and more relevance within financial reporting is investigated by Ewert and Wagenhofer (2005). In their paper they use a rational-expectations model to assess the effect of more stringent accounting principles. They underline the statement that limiting the discretion that managers have in reporting leads to a decrease in the amount of earnings management. This should mean that the adoption of IFRS should increase the quality of reporting. But they also state that standard setters can only influence accrual based earnings management practices with their accounting principles. The consequence of this is that tighter principles can lead to an increase in the amount of real earnings management because firms use this as a substitute for accrual based earnings management (Ewert and Wagenhofer, 2005). So, if this is indeed the case the adoption of IFRS will not lead to a decrease in the use of earnings management practices.
Another paper that investigates earnings management practices in Europe is that from Goncharov and Zimmerman (2008). They look at three different accounting frameworks in Germany to see which of these leads to the highest amount of
earnings management. The three tested frameworks are US GAAP, German GAAP and IFRS. They conclude that the chosen accounting system indeed influences the amount of earnings management practices. Their results point out that earnings management is less under US GAAP, but under the other two frameworks the level of earnings management stays broadly the same (Goncharov and Zimmerman, 2008). So, for the purpose of this thesis these results should mean that the
mandatory adoption of IFRS in 2005 would not lead to a significant change in the use of earnings management practices.
10 The last paper that I investigate in this part of the literature review is the paper from Jeanjean and Stolowy (2008). In their research they concentrate on three
countries in Europe, namely the United Kingdom, France and Austria. They analysed the distribution of earnings from firms in those countries to investigate whether they used earnings management to avoid small earnings decreases and losses. They conducted their research in accordance with the methods used in the paper of Burgstahler and Dichev (1997). Following this method they look for irregularities in these earnings distributions. They find out that after IFRS was implemented in Europe the pervasiveness of earnings management stayed the same in both the United Kingdom as in Austria. They even found an increase in the use of earnings management for the firms in France (Jeanjean and Stolowy, 2008). For the purpose of my thesis it seems that these results slightly contribute to the results of the other papers that are investigated in this literature review, namely that the mandatory adoption of IFRS has not led to a decrease in earnings management practices.
From this part of the literature review I can conclude that this framework of IFRS is mandated to firms in Europe to bring greater convergence in reporting and to increase the quality and financial reporting. With tighter principles and less discretion for managers to make opportunistic estimates the amount of earnings management practices should decrease, with the intention to increase the relevance and reliability of the reported earnings. So, in theory the mandatory introduction of IFRS should lead to a decrease in the use of both accrual based earnings management and real earnings management practices within European firms. When management has less opportunity to engage in opportunistic estimates it would be more difficult to manage earnings towards a certain goal or benchmark. However, it is a fact that IFRS still contains a lot of principles that only give guidance to management in how they should make reporting estimates. So, there is still room for discretion in these estimates which gives opportunities for earnings management. By looking at
literature written on this topic it seems that this is indeed the case. The majority of the research written about this states that the mandatory adoption of IFRS has not
accomplished his goal regarding earnings management. I can conclude from this literature that the amount of earnings management has not decreased after the mandatory adoption of IFRS. In fact, the literature suggests that the adoption IFRS can even increase the use of earnings management practices in some countries if it
11 is compared to the domestic GAAP (Jeanjean and Stolowy, 2008; Goncharov and Zimmerman, 2008). Now I investigate papers written on the impact of IFRS adoption on solely accrual based earnings management practices to derive my first
hypotheses.
2.2 Impact of IFRS adoption on accrual based earnings management
The first literature on accrual based earnings management that I investigate is the paper of Callao and Jarne (2010). In their research they investigate earnings management practices in Europe by comparing discretionary accruals in the periods before and after the mandatory adoption of IFRS. Their results point out that the amount of accrual based earnings management has intensified after the change in regulation. They report for all countries that they have investigated a significant rise in the use of long-term discretionary accruals. The only exception that they noticed was in Italy, where the results were not significant. From this research they conclude that mandatory IFRS adoption has encouraged the use of accruals to manage earnings and to conduct in some opportunistic behaviour, which has of course negative consequences for the quality of financial reporting (Callao and Jarne, 2010). So, for my thesis these results indicate that the use of discretionary accruals to manage earnings have increased after the introduction of IFRS.
Another paper that investigates the impact of IFRS adoption on earnings management in Europe is that from Van Tendeloo and Vanstraelen (2005). They look at adoption of IFRS within German firms. In their research they compare firms that adopted IFRS with firms using German GAAP. They come to the conclusion that when both accounting frameworks are compared, IFRS does not impose a significant constraint on earnings management practices. On the contrary, their results point out that it even increases the magnitude of discretionary accruals. German firms that implemented IFRS engaged in more earnings smoothing than firms using German GAAP, although this effect is less significant for firms that use Big-4 auditors. However, if they include the use of hidden reserves, which are allowed under German GAAP, their results show that IFRS adopters do not present a significant increase in earnings management behaviour compared to firms using German GAAP (Van Tendeloo and Vanstraelen, 2005). These results are in accordance with those
12 from Goncharov and Zimmerman (2008). Also, their research points to the same conclusion as that made by Callao and Jarne (2010), namely that IFRS leads to an increase in the use of accrual based earnings management.
Other research that looks at accrual based earnings management in Germany is the paper from Lin and Paananen (2009). They investigate a possible change in the quality of financial reporting after the mandatory adoption. In their research they look at earnings smoothing, timely loss recognition and value relevance of earnings to come up with a conclusion about earnings quality. They find out that after the mandatory adoption of IFRS the level of earnings quality decreases. Their results point out that the amount of earnings management increases after the mandatory adoption (Lin and Paananen, 2009). This conclusion is in accordance with other papers that I investigated (Van Tendeloo and Vanstraelen, 2005; Goncharov and Zimmerman, 2008).
Christensen et al. (2008) also investigated IFRS adoption in Germany. They do this by analysing two dimensions of earnings quality, namely earnings
management and timely loss recognition. And they operationalize earnings management with the use of discretionary accruals. In accordance with prior literature they also find that for voluntary adopters the amount of earnings management decreases. But their results do not indicate such an significant decrease for firms that were obligated to implement IFRS after 2005. Also, their results suggest that the reason that IFRS cannot achieve its objective to decrease earnings management lies in the fact that mandating these standards, which are indeed of high quality, will only lead to an improvement in earnings quality if firms have incentives to do so (Christensen et al., 2008). For the purpose of my literature review, these results are in accordance with the results reported by Van Tendeloo and Vanstraelen (2005), Goncharov and Zimmerman (2008) and Lin and Paananen (2009).
Another important paper that investigates the impact of IFRS adoption in Europe is that from Capkun et al (2012). In their research they investigate this impact by looking at firms from 29 countries. They notice an increase in the use of earnings management after countries adopt IFRS, but only for firms which adopted IFRS on a mandatory basis after it was obligated by the European Commission in 2005. They
13 noticed that firms who voluntary adopted IFRS before it became mandatory in 2005 engaged in less earnings management after they adopted IFRS (Capkun et al., 2012). So, it seems that early voluntary adoption of IFRS does not have an
increasing effect on the use of accruals to conduct earnings management. Only the obligation to firms to implement this reporting framework has this negative effect of an increased use of accrual based earnings management.
The results of Capkun et al. (2012) are in accordance with work from Ahmed et al. (2013). They looked at earnings data obtained from a broad set of firms from 20 countries and they to investigate preliminary effects of mandatory IFRS adoption on accounting quality. They looked at three accounting quality metrics, namely income smoothing, accrual aggressiveness and timely loss recognition and they compared these data with data from a benchmark sample of firms that did not adopt IFRS. The results that they come up with are slightly the same as those from Capkun et al. (2012). Ahmed et al. (2013) also finds evidence of an increase in income smoothing for firms that adopted IFRS on a mandatory base. They notice a significant increase in the aggressive reporting of accruals used by the firms to manage their earnings.
So, most of the papers that looked at the impact of mandatory IFRS adoption on accrual based earnings management in European countries find significant evidence of an increase in the use of discretionary accruals to manage earnings when firms are obligated to adopt this reporting framework on a mandatory base (Callao and Jarne, 2010; Capkun et al. ,2012; Van Tendeloo and Vanstraelen, 2005; Christensen et al., 2008). But, there is also some literature that contradicts these results. In the paper of Chen et al. (2010) they investigate effects of mandatory IFRS adoption on earnings data from publicly listed firms from 15 countries in Europe. They use five accounting metrics as a proxy for earnings quality. And they find evidence that IFRS adoption lead to less managing of earnings toward a certain target. Also they notice a decrease in the magnitude of absolute discretionary accruals and they find that the quality of these accruals increased after IFRS was adopted by these firms. However, they also recognize that these firms engage in more income smoothing practices than before IFRS was obligated to them. As a reason for this, there results suggest that IFRS puts a limit on managers opportunistic discretion by reducing available accounting alternatives (Chen et al., 2010).
14 But, except for this paper of Chen et al. (2010) I could not find much support for a decrease in earnings management after IFRS adoption. So, from this
perspective it seems that the outcomes of the mandatory adoption of IFRS are not entirely in accordance with the goals that the IASB had. The reason for this lies probably in the fact that standards are not the most important determinant of financial reporting quality. The incentives that firms face are more important in this respect (Leuz et al., 2003). A goal of IFRS was to improve the comparability of financial reports provided by European firms. In this way regulators and analysts should be better able to assess the choices in reporting that firms make, which should lower the incentives for accrual based earnings management (Doukakis, 2014). However, due to different management incentives in reporting and the flexibility that is provided by the principle-based nature of IFRS, it seems to me that the goal of decreasing
accrual based earnings management practices is probably not accomplished with the mandatory introduction of IFRS.
So, It seems that prior literature written on this topic shows mostly the same results. Most of the papers that I investigated report evidence of an increase in accrual based earnings management practices in countries across Europe after the mandatory adoption in 2005. Examples of these papers are Callao and Jarne (2010), Capkun et al. (2012), Ahmed et al. (2013). Also, other papers that investigated this effect found the same results for Europe’s biggest economy, namely Germany (Christensen et al.,2008; Van Tendeloo and Vanstraelen, 2005; Lin and Paananen, 2009). They conclude that data from firms that are mandated to adopt IFRS show this increase in accrual based earnings management. I also found literature suggesting that this type of earnings management decreased in European countries after the adoption of IFRS became mandatory in 2005 (Chen et al., 2010). But this is a
relatively small amount of evidence. Besides this big amount of empirical evidence, I also base my first hypotheses on theory about different management incentives and the greater flexibility given by the principles-based nature of IFRS, when it is
compared with rules-based domestic standards which were used in some European countries before IFRS became mandatory. So, because of this I suppose that mandatory IFRS adoption does not have a decreasing effect on accrual based earnings management practices, which contradicts the goals that the IASB had with
15 the issuance of these standards. Therefore, I use the following hypotheses for the first part of my research:
H1: The mandatory adoption of IFRS has not significantly decreased the amount of accrual based earnings management practices over time in European firms.
2.3 Impact of IFRS adoption on real earnings management
While there is much research done on the impact of mandatory IFRS adoption on accrual based earnings management practices, this is not the case for real
earnings management. There is just few literature written on this second earnings management strategy. There are some papers that try to give an answer on whether or not it occurs and what the consequences are. But there is only one paper that investigates the impact of IFRS adoption on real earnings management in Europe, namely that of Doukakis (2014). From other papers I can conclude that this indeed happens within firms to achieve some monetary targets (Cohen et al., 2008;
Roychowdhury, 2006; Zang, 2012). They all recognize the existence and importance of real earnings management practices, but as I said they do not look at the impact of mandatory IFRS on it and there research is focused on the US market instead of the European market, which is the scope of this thesis.
The first who did research on this impact was Doukakis (2014). He used data on firm’s earnings from 22 countries in Europe that adopted IFRS in 2005. He conducted a differences-in-differences design to look for changes in earnings management practices in the periods before and after the mandatory adoption. These data were compared to a control sample of firms that voluntarily adopted the accounting framework to test for significant changes in both real and accrual based earnings management. His empirical findings suggest that the mandatory adoption of IFRS had no significant impact on both real and accrual based earnings
management practices (Doukakis, 2014). His findings on accrual based earnings management are inconsistent with the literature that I discussed in the first part of this literature review (Callao and Jarne., 2010; Capkun et al., 2012; Ahmed et al., 2013). Because they came to the conclusion that mandatory IFRS adoption has a significant increasing impact on accrual based earnings management.
16 Other literature on real earnings management suggests that both strategies are substitutes of each other. Like Zang (2012), which investigated the existence of this trade-off between real and accrual based earnings management. This trade-off is also recognized in work from Cohen et al. (2008) and Zarowin and Cohen (2010). And from this I can conclude that when one of the two increases the other should decrease. In my opinion this finding should mean that if accrual based earnings management has not decreased after the mandatory adoption of IFRS, the amount of real earnings management should also stayed broadly the same over time. Because otherwise the substitution effect that is recognized in these papers would not apply. But, as I stated earlier these studies are focused on the US market, so maybe this substitution effect does not hold on the European market. Off course, this is investigated in my research.
Another factor that has an influence on this type of earnings management is the greater comparability and transparency arising from IFRS. Jo and Kim (2007) state in their research that higher transparency should lead to more easy detection of real earnings management, which should lower the incentives for managers to
engage in these practices. But I believe that this increased transparency should be of little effect because it is too difficult for outside users of financial statements to assess the use of real earnings management practices. So, in accordance with the findings of Doukakis (2014) I do not believe that IFRS has an decreasing effect on real earnings management. So, for the second part of my research I use the following hypotheses:
H2: The mandatory adoption of IFRS has not significantly decreased the amount of real earnings management practices over time in European firms.
17 3. Methodology and sample selection
In this part of my thesis I further elaborate on how the two hypotheses that were developed in the last section are tested. First, I introduce the set of data that is used for these tests. After that I explain the methods used to measure the amounts of real and accrual based earnings management. And I introduce the model that is used to investigate whether these earnings management practices increased after IFRS became mandatory in Europe.
3.1 Sample selection
To conduct my research I obtained data from a sample of firms that are listed on the EUROSTOXX 600 index. The reason that I used these firms as a sample is because this index contains the 600 biggest European firms in terms of market value. I obtained these data using Compustat and Datastream. For my sample I select the firms that were listed on the EUROSTOXX 600 index in the year 2001 to look at how earnings management practices within these firms behave over the chosen time period. I choose the year 2001 because I want a set of firms from before 2002, when the adoption of IFRS was announced and these firms will provide enough data during the rest of my sample period. Therefore I believe this to be a good proxy for firms that are listed within Europe. But I do acknowledge that these firms may contain different earnings management incentives due to the fact that these incentives can differ across the various countries in Europe. However, because the potential influences of these different incentives lie outside the scope of this thesis I assume here that the different earnings management proxies that are used in my research capture all or at least most of the earnings management practices within these firms although the underlying incentives may differ.
In my research I use data from 1999 till 2011, and I divide this time span into three periods. Namely, from 1999 till the mandatory adoption of IFRS was announced in 2002. From 2002 till 2005, when this framework became mandatory in Europe. And the last period goes from 2005 till 2011. To rule out the effect of firms that
already adopted IFRS on a voluntary bases I exclude the data of these firms from the data set. Also, I exclude the results of financial institutions from my sample because
18 these firms have specific accounting requirements that are not applicable to the rest of the commercial firms within the sample. This is in accordance with research done by Cohen et al. (2008) and Burgstahler and Dichev (1997). Furthermore, I require each firm year observation to have enough data to measure the real and accrual based earnings management models that are used in my research. And with the intention to delete possible outliers that bias the outcomes of my research I drop the data in the first percentile and the data in the last percentile of my sample. After modifying my dataset I end up with a sample of in total 446 firms from 19 European countries, which leads to a set of 5,331 firm year observations. Table 1 presents the distribution of the firms for the different countries.
Table 1. Sample firms distribution
Sample countries Firms
United Kingdom 130 France 68 Germany 50 Sweden 31 Switzerland 30 The Netherlands 24 Spain 19 Italy, Finland 15 Denmark 14 Ireland 11 Belgium 10 Norway, Iceland 7 Luxemburg 5 Austria 4 Portugal 3 Greece 2 Czech republic 1 Total 446
19 3.2 Accrual based earnings management – research design
To measure the amount of accrual based earnings management within my sample I looked at work from Dechow et al. (1995). In their research they explain the various methods that can be used to determine the level of accrual based earnings management. Their results point to the fact that no method will give a complete view of the total amount of these earnings management practices. But in their opinion the Modified Jones Model gives the best reflection of this amount (Dechow et al., 1995). Therefore this model is used in my thesis to measure accrual based earnings
management in Europe. This is in accordance with the paper from Cohen et al. (2008), on which most of my research is based. In the Modified Jones Model accrual based earnings management is measured using discretionary accruals as a proxy. These discretionary accruals are calculated by subtracting non-discretionary accruals from total accruals. First the amount of total accruals is estimated with the use of the model. The modified Jones model is estimated for each two-digit SIC-year grouping as follows: 𝑇𝐴𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 ∆𝑅𝐸𝑉𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
3 𝑃𝑃𝐸𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ ɛ
𝑖𝑡(1)
Where in this equation TA represents the amount of Total Accruals, which are defined as: TA = EBXI – CFO, where EBXI is the earnings before extraordinary items and discontinued operations ( Compustat data item IBC) and CFO is the cash flows from operations, which is obtained from the statement of cash flows (annual
Compustat data item OANCF – annual Compustat data item XIDOC); Where Assets is the total amount of assets (annual Compustat data item AT); Where ∆𝑅𝐸𝑉 is the change in revenues from the preceding year ( Compustat data item SALE); and where PPE is the gross value of property, plant and equipment (annual Compustat data item PPEGT).
The estimates of the coefficients that are obtained from equation (1) are then used in the following regression to estimate the firm-specific Normal Accruals for the firms within the sample:
20
𝑁𝐴
𝑖𝑡= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 (∆𝑅𝐸𝑉𝑖𝑡−∆𝐴𝑅𝑖𝑡) 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
3 𝑃𝑃𝐸𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1(2)
Where in this equation NA represents the amount of Non-discretionary Accruals; where ∆𝐴𝑅 is the change in accounts receivable from the preceding year (annual Compustat data item RECT). While computing the amount of normal accruals I adjust the revenues of the firms within the sample for the change in their accounts
receivables. This is done to capture the possible discretion within revenues that arise from credit sales. This method is in accordance with the paper from Cohen et al. (2008). After both the amount of total accruals (TA) and normal accruals (NA) are estimated I can use these estimates to measure the amount of discretionary accruals (DA). These are measured by the difference between total accruals and normal accruals, defined as:
𝐷𝐴
𝑖𝑡= (
𝑇𝐴𝑖𝑡𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
) − 𝑁𝐴
𝑖𝑡.
The estimates that areobtained from this equation are used as a proxy for the amount of accrual based earnings management practices.
3.3 Real earnings management – research design
To test for the amount of real earnings management practices I again base my research on the paper of Cohen et al. (2008). In their research they rely on work from Roychowdhury (2006), an in accordance with that paper they measure real earnings management with the use of three proxies. Namely, the abnormal level of cash flows from operations (CFO), the discretionary expenses and the production costs. I follow this method by conducting three different regression models. First I investigate the abnormal level of cash flows from operations. By accelerating the timing of sales through increased price discounts or less strict credit terms a firm can temporarily increase their sales. This increase in sales will lead to higher earnings in that reporting period. To obtain an estimate of the abnormal cash flow from operations I first calculate the normal level of CFO’s. Therefore I use the following regression, which is obtained from Dechow et al. (1998) and also used by Roychowdhury (2006) and Cohen et al. (2008):
21 𝐶𝐹𝑂𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
3 ∆𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝜀
𝑖𝑡(3)
Where CFO is the cash flows from operations, which is obtained from the statement of cash flows (annual Compustat data item OANCF – annual Compustat data item XIDOC); Where Assets is the total amount of assets (annual Compustat data item AT); Where Sales is the amount of total sales; and where ∆𝑆𝑎𝑙𝑒𝑠 is the change in sales. With the use of the estimate of the coefficient obtained from equation (3) I calculate the level of abnormal cash flows from operations. This is done by
subtracting the estimate of normal cash flows from operations from the actual cash flows from operations of the firms.
After that, I perform a second regression to test for real earnings management by reporting lower cost of goods sold through an increase in the production. This is done by increasing the amount of products that are produced. In this way the fixed overhead costs can be divided over more products, which lowers the costs per each product. This leads to a decrease in the cost of goods sold and therefore it leads to an increase in the margin on each product. To obtain a measure of the amount of this type of real earnings management I calculate the normal amount of production costs for the firm. These total production costs are calculated by taking the sum of both the change in inventory and the cost of goods sold. For each of these two I use a
regression to measure their normal levels. For the normal cost of goods sold (COGS) I use the following regression:
𝐶𝑂𝐺𝑆𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠1,𝑡−1+ 𝑘
2 𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝜀
𝑖𝑡(4)
Next I measure the normal level of inventories. This level is measured with the use of the following regression:
∆𝐼𝑁𝑉𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 ∆𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
3 ∆𝑆𝑎𝑙𝑒𝑠𝑖𝑡−1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝜀
𝑖𝑡 (5)With the use of equations 4 and 5 I estimate the normal level of production costs as follows: 𝑃𝑟𝑜𝑑𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
3 ∆𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
4 ∆𝑆𝑎𝑙𝑒𝑠𝑖𝑡−1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝜀
𝑖𝑡 (6)22 In the equations that are presented here COGS are the cost of goods sold
(Compustat data item COGS); Where Prod are the production costs, defined as the sum of COGS (annual Compustat data item COGS) and the change in inventories (annual Compustat data item INVT).
The last regression that I perform is to measure the amount of real earnings management by decreasing discretionary expenses. These discretionary expenses are defined as the sum of advertising expenses, R&D expenses and SG&A
expenses. When these kind of expenses are decreased it will lead to an increase in the earnings of that reporting period. The normal level of these expenses are
measured with the use of the following regression:
𝐷𝑖𝑠𝑐𝐸𝑥𝑝𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1
= 𝑘
1𝑡 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝑘
2 𝑆𝑎𝑙𝑒𝑠𝑖,𝑡−1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑡−1+ 𝜀
𝑖𝑡 (7)Where DiscExp are the discretionary expenditures, which are defined as the sum of advertising expenses (annual Compustat data item XAD), R&D expenses (annual Compustat data item XRD) and SG&A (annual Compustat data item XSGA). With the use of the regressions that are presented here in this part of my thesis I calculate the levels of the proxies that I use to measure the amount of real earnings management. These proxies were the level of abnormal cash flow from operations, abnormal production costs and abnormal discretionary expenses. The levels of abnormal CFO (R_CFO), abnormal production costs (R_PROD) and abnormal discretionary
expenses (RDISX) are obtained by calculating the difference between the actual levels and the normal levels. Where these normal levels are obtained from the predicted estimates arising from equations (3), (6) and (7).
Following earlier research it seems likely that firms that engage in earnings management to increase their earnings would have one or all of these (Cohen et al., 2008): lower cash flow from operations compared to the normal level, lower
discretionary expenses compared to the normal level and/or higher production costs compared to the normal level. To measure the impact of this set of variables, I combine those in one single variable as a proxy for total real earnings management. This variable is called RM_PROXY, and it is defined as the sum of the three variables that were calculated, namely R_CFO, R_PROD, R_DISX.
23 The results from all these four proxies are presented, and I perform a time trends analysis to investigate how these four proxies behaved over the chosen time period. Therefore, I perform a regression on each of the different earnings
management variables on a time trend variable. This variable is called Time, and it is defined as the difference between the year of the observation and the first year in my sample, namely 1999. Also, I include two dummy variables in this regression. The first one is IFRS_ANN which is the period from the announcement in 2002 till IFRS became mandatory in 2005 and this dummy will be one in this period and zero in the other two periods. The second one is IFRS_MAN which is the period after the
mandatory adoption in 2005 and it will be one in the period after the adoption and zero in the other periods. This to see whether the amount of real and accrual based earnings management increased after the announcement in 2002 and after the issuance in 2005, when IFRS became mandatory in Europe. Also I perform a test on the correlation of the proxies to see whether firms use both earnings management practices as substitutes for each other, which is stated in earlier research (Zang, 2012; Zarowin and Cohen, 2012). The methodology that is used in my thesis is based on and in accordance with the research methods of Cohen et al. (2008).
24 4. Results
In this chapter of my thesis I present the results that I have obtained from my research. First I present and discuss the descriptive statistics in my sample that I used to calculate values for the key variables and proxy estimates. After that I shall analyse results and graphs on the time trend that these variables follow during the sample period. This to see whether the accrual-based and real earnings
management proxies have changed after the adoption of IFRS. The last part of this chapter will present results regarding the correlations of the different variables with each other. I analyse these results to obtain evidence about the possibility of a substitution effect.
4.1 Descriptive statistics
Statistics regarding the variables used to calculate earnings management proxies and these obtained values are presented in table 2 below:
Table 2
Descriptive statistics
25th Perc. Mean Median 75th Perc. Std. Dev. Total Assets 1658 108690 5341 19394 219604 Sales 1406 53474 4487 14449 96761 Sales growth - 0.01 0.12 0.05 0.15 0.17 Leverage 0.14 0.26 0.24 0.35 0.17 Total Accruals - 0.08 - 0.05 - 0.04 - 0.01 0.11 DA - 0.03 0.00 0.001 0.03 0.06 Positive_DA 0.00 0.02 0.001 0.03 0.03 Negative_DA - 0.03 - 0.02 0.00 0.00 0.04 ABS_DA 0.01 0.04 0.03 0.05 0.04 R_CFO - 0.04 0.00 - 0.01 0.03 0.07 R_PROD - 0.11 0.00 0.02 0.13 0.26 R_DISX - 0.09 0.00 - 0.02 0.06 0.14
25 RM_PROXY - 0.28 0.001 0.01 0.07 0.25
Table 2 continued Variable definitions:
Total assets = Annual Compustat data item AT;
Sales = Annual Compustat data item SALE;
Sales growth = The change in the level of SALE divided by the lagged AT;
Leverage = Total liabilities (Compustat data items DLTT + DLC) divided by AT;
Total accruals = Net operating cash flow (Compustat data item OANCF) adjusted for the extraordinary items and discontinued operations (Compustat data item XIDOC) minus income before extraordinary items
(Compustat data item IBC) divided by lagged AT;
DA = The total of discretionary accruals calculated using the modified Jones model;
Positive_DA = The positive value of discretionary accruals calculated with the use of the modified Jones model;
Negative_DA = The negative value of discretionary accruals calculated with the use of the modified Jones model;
ABS_DA = The absolute value of discretionary accruals calculated with the use of the modified Jones model;
R_CFO = The level of abnormal cash flows from operations obtained from the regression model;
R_PROD = The level of abnormal production costs obtained from the regression model and defined as the sum of cost of goods sold (Compustat data item COGS) and the change in inventories (Compustat data item INVT); R_DISX = The level of discretionary expenses obtained from the regression model
defined as the sum of the advertising expense (Compustat data item XAD) R&D expense (Compustat data item XRD) and SG&A expenses
(Compustat data item XSGA);
RM_PROXY = The proxy for total real earnings management, defined as the sum of the three earnings management proxies, R_CFO, R_PROD and R_DISX .
From these descriptive statistics I can conclude that the average level of
discretionary accruals lies at zero. Table 2 also shows that the means of both positive discretionary accruals and negative discretionary accruals are the same. Here the negative discretionary accruals are higher in the 25th percentile, while positive ones
have the same value in the 75th percentile. This shows that the earnings decreased
26 same amount as they decreased. A very crucial variable within my research
regarding the effect on accrual based earnings management is the absolute value of discretionary accruals. Table 2 shows positive values for this variable with a mean of 0.04 and a median of 0.03 and a variance of 0.04. From these values I can conclude that the amount of accrual based earnings management raised during the chosen sample period. This can be seen as some evidence for the support of my hypotheses that the introduction of IFRS has not decreased the amount of accrual based
earnings management practices. In fact, these results even show an increase in the use of this earnings management strategy.
My proxies for real earnings management are presented by the values of the abnormal cash flows, abnormal production costs and the abnormal discretionary expenses. Table 2 shows average values of zero for all of these three proxies. Here it seems clear that the average values for accrual based earnings management are greater than those values for real earnings management. This result is supported by work from Graham et al. (2005) and Cohen et al. (2008). The reason for these greater values lies in the fact that real earnings management is far more costly for firms than accrual based earnings management. Therefore firms are more likely to engage in this type of earnings management. This while the 25th percentile values
are all negative and the 75th percentile values are all positive. The variances for the
different proxies are not small but only the variance of the abnormal production costs is especially big compared to the other two. For my proxy of total real earnings management it seems that is slightly positive with only a negative value for the 25th
percentile, but the positive values are very small. But this could give some evidence that supports my second hypotheses, namely that IFRS introduction has not
27 4.2 Time trend analysis
Now I present the results of the time trends analysis of both accrual based earnings management variables and real earnings management proxies. As I stated in the methodology chapter of this theses I regress the different earnings
management variables and proxies with a variable for time, this to analyse the behaviour of these values over the chosen time period. This variable Time will be defined as the year of the observation within my sample minus the first year of my sample period, which is 1999. I also make use of two dummy variables which were presented earlier in this thesis. Namely as IFRS_ANN, defined as the period from the announcement date in 2002 till IFRS was obligated in 2005 and IFRS_MAN, which is defined as the period after 2005 when IFRS became mandatory.
Table 3
Earnings management variables: time-trend analysis
𝐷𝑒𝑝
𝑗𝑞= a + b x Time + c x IFRS_ANN + d x IFRS_MAN Dependent Variables a b c d Adj. R2 --- --- --- --- --- --- ABS_DA 0.051*** 0.002*** - 0.003** 0.005 0.014 Positive_DA 0.026*** 0.001** - 0.002 0.002** 0.043 Negative_DA - 0.021*** - 0.001** 0.003 - 0.001** 0.056 R_CFO 0.056*** - 0.001* - 0.002** 0.001 0.055 R_PROD 0.287*** - 0.014** 0.003 - 0.006*** 0.128 R_DISX 0.139*** - 0.004** - 0.018** - 0.018* 0.026 RM_PROXY 0.161*** - 0.013** - 0.003** 0.001 0.013*, **, *** means significant at 10 percent, 5 percent and 1 percent;
Variable definitions:
ABS_DA = Absolute value of discretionary accruals obtained by modified Jones model;
Positive_DA = Positive value of discretionary accruals obtained by modified Jones model;
Negative_DA = Negative value of discretionary accruals obtained by modified Jones model;
RM_PROXY = The sum of the three real earnings management proxies;
Time = The time variable, defined as the year of observation minus 1999 (first year);
IFRS_ANN = A dummy variable which has value of 1 in years 2002, 2003, 2004, 2005;
IFRS_MAN = A dummy variable which has value of 1 in years 2006 till 2011.
28 Now I analyse the results in Table 3 to see what this time trends analysis tells about my earnings management variables and proxies. First I look at the behaviour of the absolute values of discretionary accruals to give an opinion about whether or not accrual based earnings management practices have decreased over the chosen sample period. The estimated coefficient for the variable Time, which is shown in the first row under (b), provides evidence of a small but significant increase in the use of discretionary accruals during the sample period. The estimates of the two dummy variables gives a better view on the behaviour of the discretionary accruals. It seems that they decrease a little in the period of the announcement as can be seen under (c), but after that they seem to increase a lot as can be seen under (d), which leads to an overall increase in the amount of discretionary accruals. But here I have to
acknowledge the fact that the results for the second dummy are not significant at a desired significance level.
In the second and third row the values for the positive and negative discretionary accruals are presented. These results are almost symmetric, but it seems that in the period after the adoption the positive values were twice as much as the negative ones. This contributes to the fact that the absolute values increased in this period. Also, it seems that both the values obtained for positive and negative discretionary accruals are significant at a 5 percent level. So, for the purpose of this thesis it seems that these results point to the fact that especially after IFRS became mandatory the amount of discretionary accruals increased. For my research this means that the use of earnings management practices did not significantly decreased after the adoption. On the contrary, the use of this earnings management strategy seems to increase given these results. Which in turn supports my first hypotheses, namely IFRS adoption has not significantly decreased accrual based earnings management practices.
Next, I look at the estimates for the real earnings management proxies. By looking at the values under the Time variable it seems that there is a small decrease for the levels of R_CFO and R_DISX , while the level of R_PROD decreased more significantly. This leads to a decrease in the value of the combined proxy for total real earnings management RM_PROXY. The values for the two dummy variables, which represent the different time periods, show a decrease during the announcement period for all proxies, except R_PROD. But this value is not significant at the desired
29 significance level. Therefore, it seems that the use of these proxies has decreased during the period of the announcement till the actual adoption of IFRS. But the decrease in the Time variable seems to be caused by the reduction during the announcement period. The values for the mandatory adoption period under (d) show declines for both the R_PROD and for the R_DISX. But RM_PROXY shows no decline for this period. The value for R_CFO does not decline in this period but this result is not significant at a desired level. So, by looking at the trend for the values of real earnings management proxies during the sample period it seems that there is a small decrease in the use of this earnings management strategy but this is mainly during the announcement period. There is not much evidence to support my hypotheses after IFRS was adopted in 2005, and based on these results I also cannot fully reject the hypotheses in this respect.
Now I present some graphs for the values of my earnings management
metrics to give graphical illustrations of the results that are obtained through the time trend analysis. These graphs are displayed in the figures below. Figures 1,2 and 3 will show the results for the accrual based earnings management variables. In figure 1 the estimated means for absolute discretionary accruals are graphically presented by their fiscal year.
Figure 1: A plot of the mean values of absolute discretionary accruals per each fiscal year calculated
using the modified Jones model.
This graph provides evidence of the results that were obtained using the time trend analysis. The graphical illustration of the absolute discretionary accruals shows
0,03 0,032 0,034 0,036 0,038 0,04 0,042 0,044 0,046 0,048 0,05 199920002001200220032004200520062007200820092010
ABS_DA
ABS_DA30 that the use of accrual based earnings management practices increased from 1999 till 2001. After that year the graph shows a decreasing trend during the years till 2005. This supports the results that during the period from the announce till the actual adoption the amount of accrual based earnings management decreased. However, the time trend analysis shows an increase in the use of discretionary accruals after 2005, when IFRS was mandatory adopted in Europe. This is also illustrated by the graph because it shows a clear upward trend in the period after 2005. Therefore, this graph supports my conclusion that accrual based earnings management has not decreased significantly after IFRS became mandatory.
Figures 2 and 3 provide the graphical illustration of the mean values of both positive discretionary accruals (2) and negative discretionary accruals (3) for each different fiscal year. This to give a more clear view and opinion about their behaviour during the chosen sample period.
Figure 2: A plot of the mean values of positive discretionary accruals per each fiscal year calculated
using the modified Jones model
This graphical illustration of the positive discretionary accruals shows the same results as the time trend analysis. Namely, an increase over the chosen sample period. In the period of the announcement date till IFRS became mandatory the graph shows a decreasing trend which is supported by the results obtained from the time trend analysis. After this decrease, which occurs in the years 2002, 2003 and 2004, graph shows an upward trend again which is most severe in the years 2006
0,015 0,016 0,017 0,018 0,019 0,02 0,021 0,022 0,023 0,024 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
POS_DA
POS_DA31 and 2007. After that the graph flattens a bit, but overall the positive discretionary accruals increase.
Figure 3: A plot of the mean values of negative discretionary accruals per each fiscal year calculated
using the modified Jones model
This graph represents the use of negative discretionary accruals to manage earnings. It shows a small drop in the use of these accruals in the period of the announcement, which is followed by a small rise in the use during the first years of the mandatory adoption. This can be seen in the years 2005, 2006 and 2007. This supports the results obtained for the other two graphs and the results that were obtained through the time trend analysis. All pointing at the same conclusion, namely that the use of discretionary accruals to manage earnings certainly did not decreased after the implementation of IFRS. Which in turn supports the first hypotheses.
In figure 4 a plot is presented of the summarized proxy for total real earnings management to give a better view of the trend that this proxy follows during the chosen sample period. It shows graphical illustrations of the mean values that are obtained for the sum of abnormal cash flows from operations, abnormal production costs, discretionary expenses. Therefore the line represents the summarization proxy, RM_PROXY, of these real earnings management. This graph underlines the results that were obtained. There is a clear decrease in the use of real earnings management till 2004, 2005. After that the graph shows an increase, which is
-0,03 -0,025 -0,02 -0,015 -0,01 -0,005 0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
NEG_DA
NEG_DA32 according to the results obtained through the time trend analysis. Which also shows a very small increase in the years after IFRS became mandatory.
Figure 4: A plot of the mean values of the summarization proxy, RM_PROXY per each fiscal year
So, for the purpose of my thesis these results tell me that real earnings management did not decreased significantly after 2005, when IFRS became mandatory. Only during the announcement period there was a drop in the use of real earnings management but this mostly reversed in later years.
4.3 Substitution effect of earnings management
In the following part I present a correlation matrix consisting of the outcomes for my calculations of the correlations among the different earnings management metrics that are used within my thesis. Here, I calculate correlations among the four real earnings management proxies and for both the discretionary and absolute discretionary accruals. By looking at the results for my time trend analysis it seems that there arises some substitution effect. They show an increase in accrual earnings management accompanied with a decrease in the use of real earnings management, which could point towards the substitution effect that is also noticed in work from Cohen et al. (2008) an Zang (2012). The results for the correlations are presented in
-0,04 -0,03 -0,02 -0,01 0 0,01 0,02 0,03 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
RM_PROXY
RM_PROXY33 Table 4 below and based on these results I give my opinion on whether this
substitution effect indeed exists for my data.
Table 4
Earnings management variables: correlation
DA ABS_DA R_CFO R_PROD R_DISX RM_PROXY --- DA 1 ABS_DA - 0.06*** 1 R_CFO - 0.03** 0.25 1 R_PROD - 0.05*** - 0.06** - 0.29** 1 R_DISX - 0.05* - 0.07* 0.23 - 0.42* 1 RM_PROXY - 0.01*** - 0.03** 0.03** 0.37 0.18 1
*, **, *** means significant at 10 percent, 5 percent and 1 percent;
The results in Table 4 give a clear view on the occurrence of a substitution effect. The correlations of the discretionary accruals with the real earnings
management variables show a negative relationship, which is significant at the 1 percent level for abnormal production costs and my summarized real earnings management proxy. And significant at the 10 and 5 percent for discretionary expenses and abnormal cash flows of operations. The correlations among the
absolute value of discretionary accruals and the real earnings management variables show mostly the same results. These are also all negative, except for the abnormal cash flows of operations but this correlation is not significant. This indicates that there indeed exist a substitution effect because the results in Table 4 show that when accrual based earnings management increases, this has a decreasing effect on real earnings management. Thereby, supporting the conclusions drawn by Cohen et al. (2008), Zang (2012) and my own results regarding the existence of the substitution effect. The correlations among the three real earnings management variables are also mostly negative, which shows a shift between these strategies. But here I have to acknowledge that these results are not at the desired significance levels.
34 5. Conclusion
With this thesis I contribute to the literature on the effects of mandatory IFRS adoption. I analyse these effect by examining the changes in both accrual based earnings management and real earnings management around the years that IFRS became mandatory in 2005. To analyse the effects on accrual based earnings management I calculated values for absolute, positive and negative discretionary accruals which are commonly used in the literature when this type of earnings management is investigated. To quantify the effects of IFRS adoption on real earnings management I calculated the abnormal cash flows from operations, the abnormal production costs and the discretionary expenses. Also I combined these in one summarized real earnings management proxy. Here I followed research done to U.S. firms from Cohen et al. (2008), Roychowdhury (2006) and Gunny (2005). This also underlines my contribution, because these methods have not been used before to analyse the effects of mandatory IFRS adoption in Europe.
My results indicate an increase in the use of accrual based earnings management through an increase in discretionary accruals over the chosen time period. My analysis shows that these accruals decrease a bit in the time period from 2002, when IFRS adoption was announced, till 2005, when this framework became mandatory. But after that my results show a significant increase in their use which leads to an overall increase in accrual based earnings management practices. Therefore, I can conclude that my first hypotheses is supported by the results from my research, namely that mandatory IFRS adoption has not led to a significant decrease in the use of accrual based earnings management. On the contrary, it even increased over the chosen sample period.
The results obtained for the real earnings management show a slight decrease in the use of these variables which could lead to the conclusion that mandatory IFRS adoption has decreased the use of these practices. This would mean that firms try to substitute both earnings management strategies. Which is observed earlier in prior earnings management literature. This observation is also supported by the results that are obtained through the calculation of the correlations among the different earnings management variables. These results were shown in a correlation matrix and they also point to the fact that firms are likely to shift from one of these two
35 earnings management strategies to another. Thereby leading me to the conclusion that the substitution effect, which was earlier noticed in prior work regarding this subject, indeed exists between both earnings management practices. But, the results for the real earnings management variables are less significant and are possibly not strong enough to fully reject my second hypotheses upon. So, it is also likely that mandatory IFRS adoption has a relatively small effect on real earnings management practices.
A limitation of this research lies in the fact that there are other incentives that cause firms to manage their earnings. These incentives differ across the different countries in Europe that were used in my sample. In this thesis I suppose that the mandatory adoption of IFRS in 2005 and the announcement of this is 2002 were the main incentives for earnings management in Europe. And also the biggest reasons for possible changes in the amount of these practices.
For further research it would be a good idea to investigate these incentives for each specific European country. And then rule out the possible effects of these other incentives that differ for each country. This would give a better view on whether changes in earnings management practices in each separate country are fully caused by mandatory IFRS adoption or induced by other incentives to manage earnings.
