To what extent does corporate social
responsibility (CSR) affect earnings quality?
Author’s name and student number: Ka Yiu Lee (5941334)
Research field: Financial Accounting and Social Accounting
Supervisor UvA: Georgios Georgakopoulos
Supervisor KPMG: Han Poesiat (assistant manager)
Date:
Number of words:
12 August 2015 9761
1 Abstract
This thesis investigates whether the CSR-related features of 208 companies listed on the U.S. stock exchanges have a positive or negative effect on the quality of their publicly released financial information during the period 2006 to 2011. The independent CSR-variables ‘regulatory problems’, ‘climate change’, ‘pollution prevention’ and ‘clean energy’ are regressed on the dependent variables ‘change in receivables’ and 'change in return on assets’. These two dependent variables are two components of the F-score by Dechow et al. (2011). The F-score is a new measure of the likelihood of earnings management. This research uses a panel data model, specifically the pooled OLS model. Using this model, this paper found the following results: there is only one significant result for the regulatory problems on change in receivables. The positive sign of the estimated coefficient of regulatory problems on change in receivables indicated that firms that recently have paid fines or penalties for violations of air, water or other environmental regulations are more likely to engage in earnings management, and thus the earnings quality is lower. However, there are no significant results found for climate change, pollution prevention and clean energy on change in receivables and change in return on assets.
2 Contents
1. Introduction ... 3
2. Literature review and hypotheses ... 5
2.1 CSR ... 5
2.2 Earnings quality ... 6
2.3 Theories ... 8
2.4 Hypotheses on CSR and earnings quality ... 9
2.5 Summary ... 16
3. Research methodology ... 18
3.1 Data source and data selection ... 18
3.2 Measurement of variables ... 19
3.3 Descriptive statistics ... 21
3.4 Empirical Model ... 23
4. Empirical results and discussion ... 26
5. Conclusion ... 30
Appendix 1: STATA descriptive statistics ... 32
Appendix 2: STATA result F-test ... 35
Appendix 3: STATA result of the pooled OLS model ... 36
Appendix 4: STATA performed command code-list ... 37
3 1. Introduction
Currently, there is an upward trend of (voluntary) reporting corporate social responsibility (CSR). CSR reports incorporate non-financial information, e.g. the social and environmental aspects of a firm in addition to financial information reported in the financial statements. CSR reports are demanded by investors, regulators and other stakeholders as they seek transparency about a firms’ business activities. They rely on the information provided in the financial statements and CSR reports in order to assess the performance of a firm (Dechow et al., 2011).
Corporate financial reports are prepared using accrual accounting. Accrual accounting allows to record economic transactions on the basis of expected, not necessarily actual cash flows. Accrual accounting is demanded by investors as it provides more complete information on a firms’ performance on a periodic basis. However, accrual accounting is subjective and relies on a variety of assumptions. The task of making the appropriate estimates and assumptions is entrusted to the firm’s managers because they have inside knowledge of their firm’s business. However, managers have incentives to use their accounting discretion to distort accounting numbers for self-serving and reputation motives.
For instance, their compensation and job security rely on the reported profits, so they have incentive to manipulate the company’s earnings. They could have motivation for income smoothing. This means that companies try to provide stable earnings by overstating and understating earnings in bad and good years respectively (Palepu et al., 2007).
This paper investigates whether firms who report environmental strengths behave more social responsible on the use of earnings management and therefore have a higher
earnings quality than firms who report environmental concerns. Specifically, this paper
Statement of Originality
This document is written by student [Ka Yiu Lee] 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.
4
investigates what effect CSR has on earnings quality. Therefore, the main research question in this paper is:
“To what extent does Corporate Social Responsibility (CSR) affect earnings quality?”
When managers use CSR opportunistically, the relation between CSR and earnings management is positive associated, which will lower the earnings quality. Firms may adopt CSR to cover up the impact of some corporate misconducts (Hemingway & Maclagan, 2004, p.35).
Previous studies showed mixed results on the relation between CSR and earnings management. Trébucq & Russ (2005) found a positive relationship between CSR and earnings management, thus a lower earnings quality. However, Prior et al. (2008) found in their results that there is a negative association between CSR firms and earnings quality. It is unclear from prior literature what the relation between CSR reporting and earnings quality is.
This research aims to provide more explanatory power of the relationship between CSR and earnings quality. This is examined by using the components; change in receivables and change in return on assets of the F-score by Dechow et al. (2011) as a new proxy for earnings management. Many accounting researchers (e.g. Trébucq and Russ, 2005; Prior et al., 2008; Hong and Andersen, 2011) use measures of “discretionary accruals” as their proxy for earnings management. The F-score uses financial statement information beyond accruals i.e. performance variables like change in return on assets, for identifying earnings management. The F-score that measures the likelihood of earnings management offers researchers a complementary and supplementary measure to discretionary accruals for identifying low and high quality earnings firms (Dechow et al., 2007, p.77).
This thesis investigates whether the CSR-related features of 208 companies listed on the U.S. stock exchanges have a positive or negative effect on the quality of their publicly released financial information during the 2006–2011 period. For the analyses in this thesis, financial data are collected from the Compustat database and CSR related variables are obtained from the KLD database.
Independent CSR-variables such as; regulatory problem, climate change, pollution prevention and clean energy are regressed on the following dependent variables which are two
5
components of the F-score; change in receivables and change in return on assets. This research uses a panel data model, specifically the pooled OLS model.
The remainder of this thesis is organized as follows. Chapter two provides the literature review with more background information about the terms “CSR” and “Earnings quality” and discusses the hypotheses that are tested in this thesis. Chapter three elaborates the research methodology. In addition, it discusses the measurement of variables, provides the descriptive statistics and presents the empirical model. Chapter four presents and discusses the empirical results. Chapter five provides the conclusion and limitations of this research and provide some recommendations for future research.
2. Literature review and hypotheses
This section elaborates a description of the literature on which the research question is built. The first two paragraphs explain the concepts CSR and earnings quality. Then, the hypotheses on the relationship between CSR and earnings quality are discussed.
2.1 CSR
The concept of CSR is most of the time cited according to the definition of Gray et al. (1987): “...the process of communicating the social and environmental effects of organizations’
economic actions to particular interest groups within society and to society at large. As such it involves extending the accountability of organizations (particularly companies), beyond the traditional role of providing a financial account to the owners of capital, in particular, shareholders. Such an extension is predicated upon the assumption that companies do have wider responsibilities than simply to make money for their shareholders.” (p. 3.)
In this definition, Cecil (2008) elaborates that the part of ‘the process of communicating the social and environmental effects of organisations’ is provided by disclosures (p.2). Normally, companies choose to disclose social and environmental impacts of their economic actions in their annual reports. But more companies chose to disclose separate standalone reports for each social and environmental aspect in a CSR report (p.3).
6
The upward trend of (voluntary) reporting CSR can be explained by three underlying theories according to Dawkins and Ngunjiri (2008). The first one is to manage the perceptions of key stakeholders, as reporting CSR have a signalling function according to the signalling theory. The second one is to convey the organization’s values to the public according to the impression management theory. And the last one emphases that the organization’s activities are in line with social norms according to the legitimacy theory.
According to Adams (2004), two organizations have an huge influence on CSR reporting. These are the Global Reporting Initiative (GRI) and AccountAbility. Both organizations are non-profit organizations, international oriented and are interested by multi-stakeholders. They developed frameworks to help auditors in determining the credibility of CSR reports. The GRI developed the GRI framework and the AccountAbility developed the AA1000AS assurance framework.
2.2 Earnings quality
The concept of earnings quality have many definitions in prior literature (Teets, 2002; Hermanns, 2006). Teets (2002) stated that earnings quality is closely related to earnings management, but they are not the same concept. The author stated that earnings management is an aspect that have an influence on earnings quality (p.356). Earnings is used as a summary measure of the performance of a company (Dechow, 1994). Earnings quality is used by investors “as a conditioning variable to extract valuation-relevant information from earnings patterns” (Francis et al, 2003a, p.1). Probable reasons for that earnings quality have different definitions is that earnings quality is a vague concept according to Hermanns et al. (2006, p.4).
The following main definitions of earnings quality can be found in the literature: Richardson et al. (2001) stated that earnings quality is the degree to which earnings performance persists into the next period. Mikhail et al. (2003) regarded earnings quality as the degree to which a firm’s past earnings are associated with future cash flows. Another definition is used by Schipper and Vincent (2003), they define earnings quality how it corresponds and represents to the Hicksian income (p.98). Penman and Zhang (2002) defined earnings of good quality when earnings quality is a good indicator of future earnings. Ball and Shivakumar (2005) provided a more general view to earnings quality. They defined earnings
7
quality as usefulness to investors, creditors, managers and other parties. Dechow et al. (2010) defined earnings quality borrowed from the Statement of Financial Accounting Concepts (SFAC No.1) as “Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.” They also stated that the term ‘earnings quality’ alone is meaningless, and that earnings quality is defined only in the context of a specific decision model (p.1). To summarize, earnings quality is not clearly defined in prior literature.
Watts and Zimmerman (1978) defined earnings management as “managers exercising their discretion over the accounting numbers (p.118)”. Earnings management can be defined further in two types: real and accrual-based earnings management. Real earnings management refers to the influenced message conveyed to external users by modifying the actions that companies take, for instance decreasing discretionary costs like advertising costs to boost earnings. The aim of accrual-based earnings management refers to the adjustment of accruals by the firm without influencing real economic impacts, for example changing estimates of warranty estimates (Hong & Andersen, 2011, p.462).
Dechow et al. (2010) further defined earnings quality proxies into three types: the first one is properties of earnings, the second one is investor responsiveness and the third one is external indicators of earnings management. The first category of properties of earnings consists of accruals, earnings persistence and earnings smoothness. The second category consists of examination of the earnings response coefficient and the explanatory power of an earnings-return model. And the third category examines the earnings misstatements and internal control deficiencies reported under the Sarbanes Oxley Act (p.345).
It is important to research more about how earnings management behaviour affect different parties since lower earnings management can lead to more transparent and reliable information for investors, regulators, analysts and other stakeholders. Previous research that is based on discretionary accruals as proxy for earnings management has shown mixed results concerning earnings management in CSR firms. Also there is a limited research of how earnings management influenced CSR firms.
Another research method that is used recently is the new measure of Dechow et al. (2011). The new measure of the likelihood of earnings management that they developed is used as supplement or complement to discretionary accruals (p.77). As a result, their research introduces a new measure of the likelihood of earnings manipulation: the F-score.
8
This paper focusses on the two components of the F-score as proxy for earnings management. The F-score falls within the first type of properties of earnings. The F-score consists of three regression models. These variables contains the following earnings properties: accruals quality variables (change in receivables); performance variables (change in return on assets); non-financial variables (abnormal change in employees); off-balance sheet variables (existence of operating leases) and the market-related variables (market-adjusted stock returns). The three variables; non-financial variables, off-balance sheet variables and the market-related variables can be related to the degree of earnings persistence, hence earnings properties (Ge, 2007).
2.3 Theories
Theories can also explain the relation between CSR and earnings management. A few studies focused on the opportunistic use of CSR within an agency theoretical framework (Prior et al., 2008, pp. 161-162; Kim et al. 2012, p.763). The agency theory refers to the problems that are associated with the separation of ownership and control within a firm. According to Prior et al. (2007) “the existence of information asymmetry give the possibility of opportunistic action by agents (the managers) who may have different objectives than the principals (the owners), hence pursue self-serving goals (the agency problem) (p.161).” Earnings management is regarded as a type of agency costs, since costs are involved when managers prioritize their own interest by engaging in CSR practices above the interest of the principals (Jensen & Meckling, 1976, p.5). Earnings management exists when agents have information advantage above principals. This is applicable for the self-serving and reputation motives. When managers use CSR opportunistically, the relation between CSR and earnings management is positive associated, which will lower the earnings quality. This is because managers are more likely to mislead stakeholders regarding firm value and financial performance (Kim et al. 2012, p.766). Firms may adopt CSR to cover up the impact of some corporate misconducts (Hemingway & Maclagan, 2004, p.35).
9
Other theoretical frameworks predict a negative relation between CSR and earnings management. Kim et al. (2012) stated that “Ethical theories emphasize that firms must accept social responsibility as an ethical obligation and require CSR firms to give attention to the legitimate interest of all stakeholders in reference to some guiding moral principle.” Political theories suggest that firms must operate in such a way that it simultaneously improves the community. And integrative theories argues that firms have to integrate social demands, because success depends on society (p.765).
However, these three theories of CSR assumed that managers caries incentives to behave honest, trustworthy and ethical. Managers with incentives of these characteristics for CSR will be more likely to constrain earnings management and maintain transparency in financial reporting. This result in an expectation that CSR would be negatively (positively) associated with earnings management, thus earnings quality.
2.4 Hypotheses on CSR and earnings quality
The study of Trébucq and Russ (2005) examined the relationship between corporate social records and earnings management. This study was considered one of the first studies that examined this relationship. They examined the total strengths and concerns of CSR dimensions on accrual-based earnings management. They used U.S. data from the KLD database, and their sample consists of 587 U.S. firms. Both the strengths and concerns were negatively correlated with earnings management. However, these results were not significant. Also the research of Chih et al. (2008) examined the relation between CSR and earnings management. The research of Chih, Shen and Kang (2008) uses cross-country data to examine the association between CSR and earnings management. In the period 1993-2002, they investigated for 1653 corporations in 46 countries CSR-related features and financial information that are published publicly. The firms that the researchers selected come from the COMPUSTAT Global Vintage database. There are three types of earnings management within the database: earnings smoothing, earnings aggressiveness and earnings losses and decreases avoidance. In their methodology are some equations for each proxy used, then they calculated the results in multivariate regressions. Further, they found the relation that greater commitment to CSR leads to a mitigation of earnings smoothness. This will reduce the avoidance of earnings losses and decreases. However, earnings aggressiveness increases.
10
Overall, the results show that the relation between earnings management with CSR are unclear (pp.187-195).
Following the agency theory, some studies examined the relation between CSR and financial reporting behaviour which focusses on the opportunistic use of CSR. Prior et al. (2008) examined whether firms use CSR as a strategy to mask earnings management. They found a positive relation between earnings management and CSR for regulated firms. This can be explained by the finding that regulated firms have less discretion in accounting choices. But for the relation between CSR and unregulated firms were found insignificant results, which counted for more than 80 percent of the sample. So the results of their study are less powerful and still contains mixed results.
Kim et al. (2012) studied the ethical concerns for CSR that drift corporate financial reporting. They researched how motives for CSR in general influence financial reporting practise. Also they based on unregulated firms and discretionary CSR activities as charitable giving, environmental policies, and diversity hiring providing more general evidence on the relation between CSR and financial reporting practise (p.764). This will lead to the following hypothesis:
H0: There is no relationship between CSR and earnings quality.
Kim, Park and Wier (2012) examined three different proxies for earnings management on the relationship with CSR: (1) discretionary accruals, (2) real activities manipulation and (3) the incidence of Accounting and Auditing Enforcement Releases (AAERs). They examined the relationship with a sample of U.S. firms in the period between 1991 to 2009. CSR-related variables were collected from the KLD database. Furthermore, they included firms in the Domini 400 Social Index from the KLD database for CSR firms.
The researchers proxied CSR firms in CSR scores that contained five specific dimensions of KLD. They used SIC (Standard Industrial Classification) codes to exclude observations of financial institutions. In total, they had 18,160 firm-year observations.
The Kothari et al. (2005) model is useful for the discretional accrual proxy for determining the absolute value of total accruals scaled by total assets. In addition, real activities manipulation is derived from abnormal levels of four measures. These are production costs, discretionary expenses, operating cash flows and a combined measure of real activities manipulation.
11
The proxy earnings quality illustrates firms that are under SEC enforcement actions for a certain fiscal year. Multivariate regressions are useful for concluding on a specific relationship, especially regarding after including of specific control variables to reduce the problem of correlated omitted variables (pp.767-769). In their findings is stated that “social responsible firms were less likely (1) to manage earnings through discretionary accruals, (2) to manipulate real operating activities and (3) to be subject of SEC investigations, as evidenced by AAERs against top executives (p.761).” This means that there is a significant negative relation between CSR firms and earnings management, thus higher earnings quality.
Hong and Andersen (2011) also investigated the association between CSR and earnings management. They made use of the COMPUSTAT database to compute their sample of non-financial U.S. firms for the period 1995-2005. Then they combined data of the KLD database. There are no financial firms included, because their earnings are not comparable to non-financial firms. They observed 8078 firm year data and derived a CSR index by subtracting the sum of concerns from the sum of the strengths within seven specific areas using KLD data. The result can be interpreted as follows; if the CSR index is negative or equals to zero, then the firm is classified as less socially responsible. And if the CSR index is positive, then the firm is classified as being socially responsible.
The proxies used for accrual quality measured the error term of the total accrual equation of the modified Jones model by a multivariate regression. The outcomes of their research in CSR of the total accruals and discretionary accruals are measured. Hereby are the abnormal measures of three other variables used in multivariate regressions to proxy for real earnings management. The main difference in absolute value of accruals is only significant with the discretionary accruals as a proxy, and results in lower values of accruals for social responsible firms. They also found that the standard deviation is significantly lower for CSR firms, and will lead to higher accruals quality. Furthermore, the proxy used for determining real earnings management shows a positive association between the degree of social responsibility and accrual quality. Hence, according to Hong and Andersen (2011) more socially responsible firms have higher quality accruals and less earnings management-based activities (pp. 464-467). This shows a positive association between CSR and earnings management, thus higher earnings quality. This will lead to the following hypothesis:
12
H1: There is a positive relationship between CSR and earnings quality.
However Prior et al. (2008) stated in their results that there is a negative association between CSR firms and earnings quality, since CSR firms are more likely to manage their earnings. Their study investigated the relevance of distinguishing whether investments in CSR affect the firms’ earnings. CSR could be a part of managerial strategy to obtain support of stakeholders after having employed practices damaging to shareholders’ interests like earnings management. They used financial data from COMPUSTAT and data of 26 countries of the SiRi Pro database. They chose a sample of 593 industrial firms between 2002 and 2004. Again, their research is based on the proxy of discretionary accruals as measure for earnings management. Their main aim of their paper is to a warning signal that CSR can be linked to value-destroying practises like earnings management, and that these practices could lead to lower firm’s returns. Thus, managers who manipulate earnings can deal with stakeholder activism and vigilance by resorting CSR practices. Earnings management practices damage the collective interests of stakeholders (p.172).
Prior et al. (2008) uses the Kothari et al. (2005) model in order to extract the effect of performance on computing discretionary accruals. The model of Kothari et al. (2005) is different from the modified Jones model, by incorporating a non-deflated constant term and a term to control for the impact of firm performance on discretionary accruals; return on assets (p.161-162). To determine whether firms behave corporate social responsible, they derive a weighted sum of each score based on the ratings of SiRi-analysts and the corresponding weights. When the sum is larger than the mean, then the firm is considered to be a CSR firm. They used a multivariate test to test the connection between the two concepts. There is a positive relation between CSR and earnings management in the conclusion (pp.164 and 171-173), thus a lower earnings quality.
The study of Choi, Lee and Park (2013) discussed the explanations for the mixed results of prior research. They focused on highly concentrated ownership and the amount of long-term institutional investors. They investigated the ownership structure in Korean firms, these are multinational conglomerates and private firms. The data consists of 2042 firm year observations collected from the period between 2002 to 2008. They used the KEJI Index to proxy for CSR ratings and is matched to the Data Guide to provide financial information. Also, on this research is found that the earnings quality is measured by using the absolute
13
value of abnormal discretionary accruals derived from the modified Jones model. The Kothari et al. model was used for performing robustness tests. At last, they achieved results by executing a multivariate regression and a two-stage least squares method, as controlling test. To conclude, the results indicate a positive association between earnings management and CSR performance (pp.447-449). In other words, earnings quality is negatively associated with CSR in this case. This will lead to the following hypothesis:
(-)
H2: There is a negative relationship between CSR and earnings quality
To summarize, prior studies showed mixed results on the relation between CSR and earnings management. Different researches show inconsistent results, which makes it hard to find the relationship. A clear overview of this related paper prior research is given in the summary table here below:
Author(s) Object of study Sample Research Methodology Results related to earnings quality Trébucq & Russ (2005) CSR – Earnings Quality - 587 firms - U.S. public firms - 1991-2001 - KLD and COMPUSTAT database - Multivariate regressions - Proxy CSR: KLD data - Proxy earnings quality: discretionary accruals (modified Jones +/- CSR Earnings Quality
14 model) Prior, Surroca & Tribó (2008) CSR – earnings management - 593 industrial firms - 26 countries - 2002-2004 - SiRi and COMPUSTAT database - Multivariate regressions - Proxy CSR: SiRi data - Proxy earnings quality: discretionary accruals (Kothari et al. model) -
Chih, Shen & Kang (2008) CSR – earnings management - 1653 public firms - 46 countries - 1993-2002 - FSTE indexes and COMPUSTAT database - Multivariate regressions - Proxy CSR: FSTE All-World Developed Indexes and FSTE4Good Indexes - Proxies earnings quality: - Earnings smoothing; aggressiveness, losses and decreases avoidance +/- Hong & Andersen (2011) CSR – earnings management - 8078 firm years - U.S. public non-financial firms - Multivariate regressions - Proxy CSR: KLD data - Proxies earnings +
15 - 1991-2009 - KLD and COMPUSTAT database quality: total and discretionary accruals (Jones model); real-earnings management Kim, Park &
Wier (2012) CSR – earnings quality - 18,160 firm years - U.S. public non-financial firms - 1991-2009 - KLD and COMPUSTAT database - Multivariate regressions - Proxy CSR: using KLD data - Proxies earnings quality: discretionary accruals - (Kothari et al. model); real-earnings management; AAERs +
Choi, Lee & Park (2013) CSR – earnings quality - 2,042 firm years - Korean public and private - firms - 2002-2008 - KJI database and Data - Guide - Multivariate regressions - Proxy CSR:
using KJI Index - Proxies earnings quality: - discretionary accruals - (modified Jones model, - robustness: Kothari et al. model) -
16
Table 1: Overview literature review
2.5 Summary
Based on the mixed results in this part of the literature review, there are two major views about the relationship between CSR and earnings quality.
Prior literature showed competing views about the relationship between CSR and earnings quality. There are two different types of view based on several theories: the first one is a positive view, which states that there is a positive relationship between earnings quality to CSR. This implies that limiting earnings management will result in more transparency in financial reporting. This view is congruent with ethical, integrative and political theories.
The second one is a negative view, which states that there is a negative relationship between earnings quality to CSR. This implies that managers pursue their own interest rather than that of the firm. So decisions of managers could influence the reputation of the firm with the information asymmetry for the principals. This view is congruent with the agency theory. Or there is also an chance that there might be no relationship at all.
Summarized, the following hypotheses can be formulated: H0: Earnings quality is not affected by CSR.
H1: Earnings quality is positively affected by CSR. H2: Earnings quality is negatively affected by CSR.
These hypotheses will be investigated to answer the central question of this study:
“To what extent does corporate social responsibility (CSR) affect earnings quality?”
Therefore, this research investigates whether socially responsible firms behave in a more responsible way to deliver more transparent and reliable financial information by constraining earnings management or earning misstatements. In the previous section, previous research e.g. Trébucq et al. (2005), Prior et al. (2008) and Hemmingway (2004) showed mixed results. Therefore there is more evidence needed to determine the relationship between CSR and earnings quality. The two components of the F-score model introduced by Dechow et al.
17
(2011) would provide more explanatory power to this relationship. The Libby boxes in figure 1 show the relationship between the explanatory variable and the explained variable.
Explanatory variable Explained variable
Conceptual
Operational
Figure 1: Explanatory variable and explained variable on a conceptional operational level
Corporate social responsibility
Earnings quality
CSR-environmental strengths (pollution prevention and clean
energy) and environmental concerns (regulatory problems and climate
change
(using KLD-database)
Change in receivables and change in return on
assets of the F-scores model 1 Dechow et al.
(2011) (Using
18
3. Research methodology
This section provides a more specific description of the data source and data selection, the measurement variables to employ, the descriptive statistics and the empirical models that measure the main theoretical constructs.
3.1 Data source and data selection
This paper consists of 1248 firm-year observations for 208 firms in the time period 2006 to 2011. The time period 2006 to 2011 was chosen, as these years had the most available data for CSR related variables. The following steps were taken: first, this paper uses U.S. data to increase the power of the test. The list of U.S. listed firms was collected from the Orbis database. Four criteria were chosen: (1) All active and inactive companies (2) Location: world regions; United states of America (U.S.) (3) Stock data; public listed companies (4) Accounts type & availability; availability of accounts; years with available accounts. On the last criteria, Orbis stated that the years with available accounts were the years 2005 to 2014. This resulted in 1308 U.S. public listed companies. In addition, the US primary Standard Industrial Classification (SIC) codes were added in the list to filter out the financial firms and utilities. Eventually, the list was exported into an excel-format sheet.
The list of U.S. financial data of the firms listed on the U.S stock exchanges contained a SIC code for each firm. This SIC code was given by the U.S. government to identify the primary business. The financial firms with SIC codes between 6000 and 6999 were excluded from the sample, because they may carry cash to meet capital requirements rather than for the economic reasons studied here. Also utilities with SIC codes between 4900 and 4999 are excluded from the sample because their cash holdings can be subject to regulatory supervision in a number of states. (Bates, 2009, p.1990; Kim et al., 2012, p. 767).
After the export of the data list, the given International Securities Identification Number (ISIN) codes were converted to useful “the Committee on Uniform Securities Identification Procedures” (CUSIP) codes in order to create a list of U.S. firms, which the WRDS database uses for identifying and matching the companies in that database. Then data
19
was collected from the COMPUSTAT North America database and CSR data of the U.S. listed firms were extracted from The Kinder Lydenburg and Domini (KLD) database. The KLD database is used to extract CSR data because the database contains a lot of useful data of social responsibility and the database is available to researchers (Hillman and Keim, 2001, pp.130, 135).
The next step was to delete firm-year observations that were not available or not rated. Also firms with errors in the output were deleted, e.g. firms that firms that have a negative total assets while it cannot be negative. Those firms were also trimmed. Remarkably, some observations deviated from the other firm-year observations. Therefore, the outliers were winsorized within the 1% and 99% percentiles. Winsorizing is a technique to make the distribution not suspected to outliers. The observations at the end of the distribution were replaced by the highest value, this is the value where 99% observations are below this value. The same technique was used for the observations at the begin of the distribution and were replaced by the lowest value where 99% of the observations are above this value (Frank and Goyal, 2008). As a result, only firms that contain available data were taken in the sample size.
Each of the continuous variables is winsorized at 1% and 99% percentile to mitigate outliers Dechow (2011, p.57). This yielded a panel of 1248 firm-year observations for 208 firms in the time period 2006 to 2011. The cross-sectional dimension and the time-series dimension make the sample a panel data set. Cross-sectional data means having observations at only one in time. And time-series data means having observations over different points in time for one firm. Each firm for every year has observations for the period between 2006 to 2011.
3.2 Measurement of variables
The sample data set includes six variables of 208 firms measured at the COMPUSTAT database and the KLD database. The six variables consist of the two dependent variables: change in receivables (ch_rec) and change in return on assets (ch_roa), which are the two different components of the F_SCORE to proxy earnings management; and the four independent KLD variables, which are two KLD environmental strengths variables: Pollution Prevention (PP), Clean Energy (CE), and two KLD environmental concerns variables:
20
Regulatory Problems (RP), Climate Change (CC). The KLD environmental variables consist of dummy variables. These four variables are measured as follows by Chatterji et al (2009):
KLD environmental strengths
Pollution prevention: the company has notably strong pollution prevention programs including both emissions reductions and toxic-use reduction programs.
Clean energy (previously called Alternative fuels). The company has taken significant measures to reduce its impact on climate change and air pollution through use of renewable energy and clean fuels or through energy efficiency. The company has demonstrated a commitment to promoting climate-friendly policies and practices outside its own operations. KLD environmental concerns
Regulatory problems. The company has recently paid substantial fines or civil penalties for violations of air, water or other environmental regulations, or it has a pattern of regulatory controversies under the Clean Air Act, Clean Water Act, or other major environmental regulations.
Climate Change. The company derives substantial revenues from the sale of coal or oil and its derivative fuel products, or the company derives substantial revenues indirectly from the combustion of coal or oil and its derivative fuel products. Such companies include electric utilities, transportation companies with fleets of vehicles, auto and truck manufacturers, and other transportation equipment companies.
The two dependent variables are measured as follows: Change in receivables (ch_rec)
Dechow et al. (2011) stated that receivables are directly linked to revenue recognition and cost of goods sold. This will affect the gross profit which is a key performance metric. From the results of Dechow et al. (2011), the change in return on receivables have a significant positive relationship with earnings management. E.g. firms with more change on receivables in the balance sheet are more likely to engage in earnings management.
21
The change in receivable is measured by the percentage of accounts receivables (RECT) set off to the average total assets (AT), where average total assets is defined as the average amount of assets of the current year and the previous year.
Ch_rec Δ Accounts Receivable (RECT) ⁄ Average total assets (AT)
Change in return on assets (ch_roa)
According to Dechow et al. (2011), the change in return on assets is significant lower for misstating firms. Managers prefer to show positive growth in earnings. Therefore, according to Dechow et al. (2011), managers tend to show positive increases in earnings. The change in return on assets is computed as follows:
Ch_roa [Earningst (IB) ⁄ Average total assetst ] - [Earningst-1 (IB) ⁄ Average total assetst-1 (AT) ]
3.3 Descriptive statistics
The data can be summarized with the descriptive statistics, which are the mean, standard deviation, minimum, maximum and the number of observations. See table 2. The raw results from STATA are presented in appendix 1.
ch_rec ch_roa RP CC PP CE Mean - 0.0077 0.0001 0.0610 0.0490 0.0385 0.0658 SD 0.0326 0.0935 0.2394 0.2159 0.1925 0.2481 Min - 0.1495 - 0.3745 0 0 0 0 Max 0.0851 0.4289 1 1 1 1 Skewness -1.0577 0.3958 3.6687 4.1806 4.79567 3.5022 Kurtosis 8.020 11.0101 14.4597 18.4771 23.9984 13.2656 N 1246 1246 1246 1246 1246 1246
Table 2: Descriptive statistics
The distribution of the sample can be obtained more precisely by computing the skewness coëfficiënt and kurtosis coëfficiënt.
The skewness coëfficiënt indicates the level of assymetry of the distribution of the data. The coëfficiënt is zero when the distribution is perfectly symmetric. In other words, a
22
zero value means that both tails are equal where asymetries are even out. The coefficient is negatively when the distribution is negatively skewed. This means that the histogram contains more data on the right side of the distribution and has a longer tail on the left side. The skewness of the change in receivables is -1.057718, thus negative.
The coëfficiënt is positively when the distribution is positively skewed. This means that the histogram contains more data on the left side of the distribution and has a longer tail on the right side. The skewness of the change in return on assets, regulatory problems, climate change, pollution prevention and clean energy are positive. Pollution prevention has a distribution that is the most positively skewed.
The kurtosis coëfficiënt indicates whether the distribution is more peaked or more flat compared to a symmetric distribution. The kurtosis for all variables are greater than three, indicating that the distribution is more sharply peaked than a symmetric distribution. The distribution of pollution prevention is the most sharply peaked.
It is also examined whether the independent variables are correlated with two or more independent variables within the multiple regression model. There is the possibility that one or more independent variables affect the other independent variables in the regression model. This is called multicollinearity. To examine this problem, the Pearson correlation coefficient will be performed and generated using STATA. The Pearson correlation coëfficiënt examines the strength between two variables. This measure reflects the level of linearity between the two variables and is designated with the greek letter 'rho' (r). The coëfficiënt lies between -1 to +1. A r of +1 indicates a perfect linear relationship between the two variables. A negative coëfficiënt means that there is a negative correlation between the variables. And a positive coëfficiënt shows that there is a positive correlation between the variables. An correlation coëfficiënt that is zero, means that there is no correlation at all.
ch_rec ch_roa RP CC PP CE ch_rec 1.0000 ch_roa -0.1659 1.0000 RP 0.0625 -0.0109 1.0000 CC -0.0379 -0.0128 0.0820 1.0000 PP 0.0327 -0.0028 0.1407 0.0899 1.0000 CE 0.0113 0.0147 0.0811 0.0298 0.4010 1.0000
23
Table 3: Correlation matrix
In this research, the results of the Pearson correlation coëfficiënts are presented in table 3. In this table, the coefficients are less than 0.5. This implies that there is no multicollinearity between the variables.
3.4 Empirical Model
The firms in the Excel spreadsheet was ranked and given a number on it. This can be used in STATA by assigning the company numbers as an “id” variable in STATA. The second row was used to combining the associated years to the “id”. The associated years were named as the variable “t” in STATA. Then the variables ch_rec and ch_roa were assigned as the dependent variables ylist, and the environmental CSR related variables Regulatory Problem (RP), Climate Change (CC), Pollution Prevention (PP) and Clean Energy (CE) as the independent variables xlist.
The next step is to set data as panel data in STATA. The KLD independent variables are lagged one year to avoid reverse causality. Which means that the independent variable of the previous year explains the dependent variable of the current year. This has a consequence that the observations of one year are lost (Zou and Xiao, 2006).
There are three panel data models used in this research. These are the (1) pooled OLS model (2) the fixed effect model (3) random effect model. The models differ from each other in the sense that the pooled OLS model takes out the individual effects, whereas the fixed model and random effect model incorporate individual effects. In section four will this three models tested what model is appropriate for this research.
If individual effects are incorporated, it means that firms in model are different from each other in contrast to the pooled OLS model. The individual effect is not incorporated in the independent variables. The individual effect can be fixed or random. The μi captures the fixed or random effect.
In the pooled OLS model, the μi is zero. This means that each firm has the same intercept as the same slope for every independent variable and contains the same error term. The pooled OLS model has the following equation:
24
(1) yit= α + β1RPit-1+ β2CCit-1 + β3PPit-1+ β4CEit-1 + εit Where:
yit = one of the proxies of earnings management for firm i at time t α = the intercept
εit = the error term for firm i at time t.
In general, the estimated coefficient β of the independent variable illustrate the impact of the independent variable on the dependent variable. The larger the β, the larger the effect of the independent variable on the dependent variable. The sign indicates the direction of the effect of the independent variable on the dependent variable.
In the fixed effect model, every firm have a different intercept. The slopes for the independent variables are the same for each firm, and share the same error term. This model makes it possible to correlate the individual effects μi with the independent variables. The fixed effect model has the following equation:
(2) yit = (α + μi) + β1RPit-1 + β2CCit-1 + β3PPit-1 + β4CEit-1 + εit
In the random effect model each firm has a different error term. However, each firm shares
the same intercept and slopes for the independent variables. The individual effect is part of the error term. The error term consists of two parts. The traditional error term εit and the
individual specific error μi. This model makes it not possible to correlate the individual effects μi with the independent variables. The random effect model has the following equation:
(3) yit = α + β1RPit-1+ β2CCit-1 + β3PPit-1+ β4CEit-1 + (μi+ εit )
The F-test is used in order to examine whether fixed individual effects are present. If the null hypothesis is zero, then there are no individual effects. H0: μ1 = μ2 =…= μn-1 = 0. If the null hypothesis is not zero then there are fixed individual effects available in the model. In this case, the fixed effect model is more appropriate than the pooled OLS model.
To test whether the random individual effect model is more appropriate than the pooled OLS, the Breusch and Pagan’s Lagrange multiplier (LM) test will be used. If the null hypothesis of the model is zero then random individual effects are not present. If the null hypothesis is rejected, then random individual effects are present and will the random effect model be more appropriate than the pooled OLS model.
25
In the case that the F-test and Breuch Pagan’s LM test the null hypotheses are rejected, then the Hausman test will be used. The Hausman test finds out whether the fixed effect model or the random effect model is appropriate. When the individual effects are not correlated with the independent variables, then H0: E (μi | Xi) = 0, where μi is the individual specific effect. When the individual effects are correlated with the independent variables, then H1: E (μi | Xi) ≠ 0. The fixed model will be chosen when the null hypothesis is rejected. Otherwise, the random effect model will be chosen if the null hypothesis is accepted.
26 4. Empirical results and discussion
The F-test was conducted to examine whether fixed individual effects were present. The null hypothesis was accepted (for the change in receivables prob > F = 0.6501; for the change in return on assets prob > F = 0.9468), therefore there are no fixed individual effects available in the model during the sample period 2006 to 2011. In this case, the pooled OLS model is more appropriate than the fixed effect model. See table 4 and appendix 2 for the F-score:
ch_rec ch_roa RP 0.0085* (0.030) -0.0040 (0.720) CC -0.0069 (0.110) -0.0051 (0.678) PP 0.0050 (0.341) -0.0039 (0.799) CE -0.0006 (0.888) 0.0072 (0.538) Constant -0.0081 (0.000) 0.0002 (0.932) F-test F(207,1034) = 1.78 Prob > F = 0.6501 F(207,1034) = 0.57 Prob > F = 0.9468 R2 0.0065 0.0006 Adjusted R2 0.0033 -0.0027 N 1246 1246
Table 4: Results of estimating the Pooled OLS model. The p-values are presented in the parentheses. ***, **, * indicate statistical significance at the 0.01, 0.05 and 0.10 levels respectively.
So the pooled OLS model will be used. In this model, the R2 is 0.0065 for change in receivables and 0.0006 for change in return on assets. This means that the KLD independent variables explain 0.65% of the variation in the change in receivables and 0.06% of the variation in the change in return on assets. There is a difference between R2 and adjusted R2. This is that R2 takes as assumption that every independent variable in the model explains the variation in change in receivables and change in return on assets. The adjusted R2 illustrates the percentage of the variation that have a significant relationship with the dependent variable. In the adjusted R2, independent variables are eliminated that have not a significant relation
27
with the dependent variable. So R2 will be higher than the adjusted R2 in each model. In this pooled OLS model, the R2 is higher for the change in receivables than the R2 than the change in return on assets. In appendix 3 the raw results are presented.
Regulatory problems (RP)
The estimated coefficient of regulatory problems on change in receivables (ch_rec) in the pooled OLS model is 0.0084921. The p-value is 0.030 and is less than 0.05. This implies that the null hypothesis, H0 = β1 = 0, is rejected. H1 = β1 ≠ 0 applies. The presence of regulatory problems does have an effect on change in receivables. Regulatory problems is an environmental concern. The company could have recently paid substantial fines or civil penalties for violations of air, water or other environmental regulations, or it has a pattern of regulatory controversies under the Clean Air Act, Clean Water Act, or other major environmental regulations. The sign of the estimated coefficient of regulatory problems on the change in receivables is positive, which implies that there is a positive effect of the presence of regulatory problems on change in receivables. Regulatory problems is a proxy for earnings management. Dechow et al. (2011) found a positive and significant relationship between change in receivables and earnings management. The positive sign of the estimated coefficient of regulatory problems on change in receivables indicates that firms that recently have paid fines or penalties for violations of air, water or other environmental regulations are more likely to engage in earnings management, and thus the earnings quality is lower.
The estimated coefficient of regulatory problems on change in return on assets (ch_roa) in the pooled OLS model is –0.0040295. The p-value is 0.720 and is more than 0.05. This implies that the null hypothesis, H0= β1 = 0, is accepted and that H1= β1≠ 0 is rejected. The presence of regulatory problems does not have an effect on change in return on assets, and thus no effect on earnings quality.
Climate change (CC)
Climate change is an environmental concern. The company could derive substantial revenues from the sale of coal or oil and its derivative fuel products, or the company could derive substantial revenues indirectly from the combustion of coal or oil and its derivative fuel products. The estimated coefficient of climate change on change in receivables (ch_rec) in the pooled OLS model is -0.0068712. The p-value is 0.110 and is more than 0.05. This implies
28
that the null hypothesis, H0 = β1 = 0, is accepted and that H1 = β1 ≠ 0 is rejected. The presence of climate change as environmental concern does not have an effect on change in receivables, and thus no effect on earnings quality.
The estimated coefficient of climate change on change in return on assets (ch_roa) in the pooled OLS model is –0.0051328. The p-value is 0.678 and is more than 0.05. This implies that the null hypothesis, H0 = β1 = 0, is accepted and that H1 = β1 ≠ 0 is rejected. The presence of climate change as an environmental concern does not have an effect on change in return on assets, and thus no effect on earnings quality.
Pollution prevention (PP)
Pollution prevention is an environmental strength. The company could have notably strong pollution prevention programs including both emissions reductions and toxic-use reduction programs. The estimated coefficient of pollution prevention on change in receivables (ch_rec) in the pooled OLS model is 0.0050319. The p-value is 0.341 and is more than 0.05. This implies that the null hypothesis, H0 = β1 = 0, is accepted and that H1 = β1 ≠ 0 is rejected. The presence of pollution prevention as environmental strength does not have an effect on change in receivables, and thus no effect on earnings quality.
The estimated coefficient of climate change on change in return on assets (ch_roa) in the pooled OLS model is -0.0038746. The p-value is 0.799 and is more than 0.05. This implies that the null hypothesis, H0= β1 = 0, is accepted and that H1= β1≠ 0 is rejected. The presence of pollution prevention as an environmental strength does not have an effect on change in return on assets, and thus no effect on earnings quality.
Clean energy (CE)
Clean energy is an environmental strength. The company could have taken significant measures to reduce its impact on climate change and air pollution through use of renewable energy and clean fuels or through energy efficiency. The company could have demonstrated a commitment to promoting climate-friendly policies and practices outside its own operations. The estimated coefficient of clean energy on change in receivables (ch_rec) in the pooled OLS model is -0.005739. The p-value is 0.888 and is more than 0.05. This implies that the
29
null hypothesis, H0 = β1 = 0, is accepted and that H1 = β1 ≠ 0 is rejected. The presence of clean energy as environmental strength does not have an effect on change in receivables, and thus no effect on earnings quality.
The estimated coefficient of climate change on change in return on assets (ch_roa) in the pooled OLS model is 0.0072065. The p-value is 0.538 and is more than 0.05. This implies that the null hypothesis, H0 = β1 = 0, is accepted and that H1 = β1 ≠ 0 is rejected. The presence of clean energy as an environmental strength does not have an effect on change in return on assets, and thus no effect on earnings quality.
30 5. Conclusion
This thesis investigates whether the CSR-related features of 208 companies listed on the U.S. stock exchanges have a positive or negative effect on the quality of their publicly released financial information during the period 2006 to 2011. In other words, this paper examines whether firms with environmental strengths behave more social responsible on the use of earnings management and therefore have an higher earnings quality than firms with environmental concerns. Specifically, this paper investigates what affect CSR has on earnings quality. Therefore, the main research question in this paper is:
“To what extent does Corporate Social Responsibility (CSR) affect earnings quality?”
For the analyses in this thesis, financial data are collected from the Compustat database and CSR related variables are obtained from the KLD database.
The independent CSR-variables regulatory problem, climate change, pollution prevention and clean energy are regressed on the following dependent variables which are two components of the F-score; change in receivables and change in return on assets. This research uses a panel data model, specifically the pooled OLS model. Using this model, this paper found the following results: there is only one significant result for the regulatory problems on change in receivables. The positive sign of the estimated coefficient of regulatory problems on change in receivables indicated that firms that recently have paid fines or penalties for violations of air, water or other environmental regulations are more likely to engage in earnings management, and thus the earnings quality is lower. However, there are no significant results found for climate change, pollution prevention and clean energy on change in receivables and change in return on assets.
This research is based on the assumption of two significant components of Dechow et al. (2011) that is used for his F-score model. Dechow et al. (2011) stated that the two components, change in receivables and change in return on assets, are positively related to earnings management. This is considered as a limitation, because the two components are not tested whether they are positively and significantly related to earnings management. More measures of the F-score model could have been used to increase the explanatory power of the research. Most of the results were not significant, this
31
could be explained by the fact that more independent variables could have been added to the research model. At last, the list of U.S. listed publicly firms using the Orbis database is only available from 2005 to 2014. The years before 2005 cannot be used for research.
For future research, it is recommended that the F-score model of Dechow et al. (2011) must be tested and also to add more independent variables to the research model.
32 Appendix 1: STATA descriptive statistics
35 Appendix 2: STATA result F-test
STATA result F-test for change in receivables
36
Appendix 3: STATA result of the pooled OLS model
STATA result of the pooled OLS model for change in receivables
37 Appendix 4: STATA performed command code-list
38 References
Adams, C. A. (2004). The ethical, social and environmental reporting-performance portrayal gap. Accounting, Auditing and Accountability Journal, 17(5), 731- 757.
Ball, R., and L. Shivakumar,. (2005). “Earnings quality in UK private firms: comparative loss recognition timeliness.” Journal of accounting & Economics 39, pp. 83-128.
Bates, T. W., Kahle, K. M., & Stulz, R. M. (2009). Why do US firms hold so much more cash than they used to?. The Journal of Finance, 64(5), 1985-2021.
Cecil, L.(2008). Corporate social responsibility reporting in the United States. McNair
Scholars Research Journal, 1(1), 43-52.
Chatterji, A., Levine, D., & Toffel, M. (2009). How well do social ratings actually measure corporate social responsibility? Journal of Economics and
Management Strategy, 18 (1), 125-169.
Chih, H., Shen, C., & Kang, F. (2008). Corporate social responsibility, investor protection, and earnings management. Some international evidence.
Journal of Business Ethics, (79), 179-198.
Choi, B. B., Lee, D., & Park, Y. (2013). Corporate social responsibility, corporate governance and earnings quality. Evidence from Korea. Corporate
Governance, 21(5), 447-467.
Dawkins, C., & Ngunjiri, F.W. (2008). Corporate social responsibility reporting in South Africa: A descriptive and comparative analysis. Journal of Business
Communication, 45(3), 287-307.
Dechow, P.M., (1994).“Accounting earnings and cash flows as measures of firm performance: the role of accounting accruals”, Journal of Accounting and
Economics, No.18, pp.3-42.
Dechow, P.M., Ge, W., Larson, C.R., Sloan, R.G. (2011). Predicting material
accounting misstatements. Contemporary Accounting Research, 28(1), 17-82. Dechow, P.M., Ge, W., Schrand, C. (2010). Understanding earnings quality: A review
of the proxies, their determinants and their consequences. Journal of Accounting
and Economics, 50, 344-401.
Francis, J., LaFond, R., Olsson, P., & Schipper, K. (2003). Earnings quality and the pricing effects of earnings patterns. Available at SSRN 414142.
Frank, M.Z., Goyal, V.K., (2008). Trade-off and Pecking Order Theories of Debt. In Espen Eckbo (editor), The Handbook of Empirical Corporate Finance, Volume 2, Chapter
39
Ge, W. (2007). Off-balance-sheet activities, earnings persistence and stock prices:
Evidence from operating leases. Retrieved from http://ssrn.com/abstract=816044 or
http://dx.doi.org/10.2139/ssrn.816044
Gray, R. H., Owen, D.L., & Maunders, K. T. (1987). Corporate social reporting.
Accounting and accountability. Hemel Hempstead, UK: Prentice Hall.
Hemingway, C., & Maclagan, P. (2004). Managers’ personal values as drivers of corporate social responsibility. Journal of Business Ethics, 50 (1), 33-44.
Hermanns, S., (2006). “Financial Information and Earnings quality: a literature review.” Universitaires Notre-Dame de la paix; faculté des sciences économiques
Hillman, A.J., & Keim, G.D. (2001). Shareholder value, stakeholder management, and social issues. What's the bottom line? Strategic Management Journal,
22(2), 125-139.
Hong, Y., & Andersen, M.L. (2011). The relationship between corporate social responsibility and earnings management. An exploratory study. Journal of
Business Ethics, 104, 461-471.
Jensen, M., & Meckling, W. (1976). Theory of the firm. Managerial behavior, agency cost and capital structure. Journal of Financial Economics, 3 (4), 305-360. Kim, Y., Park, M.S., & Wier, B. (2012). Is earnings quality associated with corporate
social responsibility? The Accounting Review, 87(3), 761-796.
Kothari, S.P., Leone, A.J., & Wasley, C.E. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39, 163-97.
Mikhail, M.; Walther, B.; Willis, R.,.(2003) “Reactions to dividend changes conditional on earnings quality”, Journal of Accounting, Auditing and Finance, Vol.18, No.1, pp.121-151.
Palepu, K. G., & Healy, P. M. (2007). Business Analysis and Valuation: Ifrs Edition-Text
Only. Cengage Learning EMEA.
Penman, S., and X. Zhang, (2002). “Accounting conservatism, the quality of earnings and stock returns.” The Accounting Review 77 (2): 237-264.
Prior, D., J. Surroca, and J. Tribo. (2008). Are socially responsible managers really ethical? Exploring the relationship between earnings management and corporate social responsibility. Corporate Governance 16 (3): 160–177.
Richardson, S.; Sloan, R.; Soliman, M.; Tuna, I.,.(2001) “Information in accruals about the quality of earnings”, working paper university of Michigan business school, ann arbor, pp.52
Schipper, K. and L. Vincent, (2003). “Earnings quality.” Accounting Horizons, supplement
40
Teets, W.R. (2002). Quality of earnings. An introduction to the issues in accounting education, special issue. Issues in Accounting Education, 17 (4), 355-360. Trébucq, S., and R. Russ. (2005). The Case of Earnings and Stakeholder Management:
Towards an Integrated Theory of Managerial Behaviour. Working paper, Université
Montesquieu Bordeaux IV.
Watts, R.L., & Zimmerman, J.L. (1978), Towards a positive theory of the
determination of accounting standards. The Accounting Review, 53 (1), 112- 134.
Zou, H., Xiao, J.Z., (2006). The Financing Behaviour of Listed Chinese Firms. The British
