Accrual based earnings management during financial crisis: a comparison of financial and non-financial industries

1 / 32

Master Thesis Accounting

Accrual based earnings management during financial crisis: a

comparison of financial and non-financial industries

Ying Li

18-06-2014

2 / 32 Abstract

Although earnings management has been widely investigated in the recent years, most of the prior research only focuses on industrial firms. Banks and other financial institutions are often excluded from relevant research due to their special characteristics, but banking industry is essential for the health of a country’s economy, especially during particular situation such as financial crisis. Consequently, this master thesis will study empirically the earnings management in both financial and non-financial industries from North American during the financial crisis period. While the loan loss provision is always seen as the main tool for banks to make earnings management in relevant studies, the earnings management in non-financial industries seems to be more complex. By investigating the influence of financial crisis on earnings management, some different reactions are found between financial industry and non-financial companies to financial crisis happened around 2007 in North American. In the end, social contributions are proposed as well by the message released from this thesis.

Name: Ying Li

Student number: 10042296

Date: 22-06-2014

Faculty: Economics and Business

Title: Accrual based earnings management during financial crisis: a comparison

of financial and non-financial industries

3 / 32

1. Introduction

Earnings management is one of the hottest topics in the field of accounting research. Walker

(2013) defines the earnings management as “the use of managerial discretion over within

generally accepted accounting principles (GAAP) accounting choices, earnings reporting choices, and real economic decisions to influence how underlying economic events are reflected in one or more measures of earnings.” There are two categories of earnings management, one is real earnings management and the other is accrual-based earnings management. Real earnings management is using real economic decisions to change earnings. For instance, offering discounts to customers can easily boost the amount of sales, therefore, firm’s earning gets increased. Whilst accrual based earnings management focuses on using discretionary accruals to manage earnings. According to Dechow and Skinner (2000), accrual based earnings management occurs when managers intend to hide or mask their real economic position by dealing with different accounting choices within the bound of GAAP.

Empirical researches of earning management were mostly conducted separately either in banking industry or other non-financial industry due to their different characteristics and different ways to manipulate earnings. Basically, there are two streams of research to examine the earnings management behavior of the banking industry. The first stream of research tries to find out whether banks indeed manage their earnings or not and through which way do they manage their earnings. For example, Greenawald and Joseph (1988) test the income smoothing hypothesis based on the data from 106 banks. The finding shows that banks smoothing earnings through loan loss provisions, therefore, it further confirms the income smoothing hypothesis in banking industry. Similarly, Collins et al. (1995) shows the significant positive relationship between earnings and loan loss provisions. Almost two thirds of the sample banks reduced loan loss provisions in the years with low non-discretionary earnings. The result is in consistent with the view that banks engage in earnings management and the loan loss provisions is served as a tool for earnings management.

Another stream of study tries to explore the influence of different economic situations on earnings management. Contradictory results are obtained from prior studies. Agarwal et al. (2007)

4 / 32

examines the behavior of earning management in Japanese banks during the period from 1985 to 1999. The research indicates that the behavior is only prevalent during high economic growth period and stagnant growth period. Surprisingly, it seems that during recession period, managers are less motivated in earnings management. However, some other literatures find the contradictory evidence that when firms face financial distress they are more likely to manage earnings. For example, Huizinga and Laeven (2009) perform a research to find the relationship between accounting discretions in banks and the financial crisis. Their study concludes that “banks use accounting discretions to overstate the value of distressed assets” (page 5). The book value of real estate-related assets is higher than the market value of assets during the US financial crisis. Authors also mention in their paper that financial reports do not represent the real health condition of banks during crisis period.

Nonetheless, no matter what type of research stream does the literature belongs to, the special accrual-namely the loan loss provision is always seen as the main tool for banks to make earnings management in relevant studies. Unlike the banking industry, the earnings management in non-financial industries seems to be more complex, since besides the accrual-based earnings management, they also engage in real-based earnings management. Although the non-financial industries use different ways to manipulate their earnings, the division of research streams is still valid.

One typical stream of research discusses how non-financial industries engage in earnings management. Fang et al. (2008) conclude that the accrual-based earnings management for non-financial industries arises when products are sold firstly on credit but payments are collected later. The managers’ right to discretionarily increase or decrease estimates of bad debt reserves and warranty costs provide the opportunity of earnings management. In contrast to accrual-based earnings, Graham et al. (2005) mention that managers are willing to make decisions that do not maximize firm’s performance and use actual business activities to smooth or boost earnings. For instance, managers may delay expenses for research and development, advertising, or reject project with a positive net present value. Some other reports also find that accrual-based earnings management and real-based earnings management are not conducted independently, and instead, there is a trade-off relationship between them (Zang 2012).

5 / 32

The economic influence on the earnings management of non-financial industries is also examined by some prior researches. Chia et al. (2007) examine the level of accrual-based earnings management during the Asian financial crisis in 1998, which confirms that companies engage more in earnings management during financial crisis period. Moreover, Habib et al. (2013) also show that financial distress has a large impact on accrual-based earnings management. Managers engage more in income-decreasing earnings management for firms in the situation of financial distress. Besides the evidence based on accrual-based earnings management in non-financial industries, Beaver (1966) argues that firms that face financial distress usually have less cash flow than those firm in good financial condition. Therefore, it is more likely for those financially distressed firms to manipulate their earnings through real activities such as reducing R&D expenditures. The research concludes that poor financial condition is positively related to real-based earnings management.

Prior literature has shown that the accrual-based earnings management is engaged in both financial and non-financial industries. However, few investigations have been conducted before to examine how different types of industry behave differently engaging in accrual-based earnings management. Moreover, the discrepant results concluded from prior research with regard to how economic situation influence the behavior accrual-based earnings management also inspire my motivation to do further research on this topic. What I would like to explore exactly in the thesis is the comparison of influences of financial crisis in 2007 in United Stated on accrual-based earnings management in financial and non-financial industries. It is a good opportunity to fill the research gap and to contribute some new insight into the prior literature.

1.2 Research question

The research question motivating this thesis is: What is the influence of financial crisis on accrual-based earnings management behavior of North American financial industry and non-financial industries. Research on earnings management has shown that managers have incentives to engage in accrual based earnings management in both financial and non-financial industries (e.g. Habib et al., 2013; Huizinga & Laeven, 2009; Chia et al., 2007). However, those prior studies either focus on financial industries or non-financial industries, and there is no such study examining both types of industries and trying to compare the difference. Moreover, prior

6 / 32

literature mostly focus on the influence of financial crisis on earnings management in Asia pacific area. This master thesis will contribute to literature by deeply examining the relationship between financial crisis occurred in United States in 2007 and accrual-based earnings management, and compare the different behaviors of financial and non-financial industries.

1.3 Motivation of this study

This thesis can contribute to the prior literature in the following ways. Firstly, this study focuses on the empirical study in both financial and non-financial industries. Although earnings management is a quite hot topic in recent years, most of the prior research only focuses on industrial firms. Banks and other financial institutions are often excluded from relevant research due to their special characteristics. However, banking industry is essential for the health of a country’s economy, especially during the particular situation such as financial crisis. Moreover, earnings management is crucially important for banks since it could influence the transparency of the firm’s financial position and may lead to the misinterpretation of financial information for investors. This study would provide a deeper understanding of the earnings management in banking industry, which is quite contributive. Moreover, this thesis will compare the accrual-based earnings management in financial and non-financial industries, which is also quite innovative.

Secondly, this study will examine the influence of financial crisis on earnings management. There are some other studies trying to examine the change of earnings management under special circumstances. Nonetheless, accounting regulations (Basel Accord, fair value regulation) and financial distress are always what prior literature looks at. Little attention has been paid to the association between recent financial crisis and earnings management. Therefore, this proposal could fill this gap and make certain contributions to prior literature.

Finally, this study has some social contributions as well. The research is relevant for investors, financial analysts. Investors and financial analysts have to be aware of the risks in earnings management especially in particular circumstances such as financial crisis, during which it is more risky to make investment. Earnings management may affect the transparency of a firm’s true position. Misinterpreting financial information due to the earnings management may lead to a wrong decision of investment. The auditors also have to be aware of the effect of

7 / 32

the earnings management tactics displayed by their clients to avoid the risk of litigation (Heninger, 2001).

2. Literature review and hypotheses

This part contains four main sections. In the first section, I am going to describe theories underpin this thesis and give a short introduction of earnings management. In the second section, I am going to introduce the models for measuring earnings management. The third section is about study of literature that relevant to our research topic. In the last section, I am going to develop hypotheses based on the prior literature.

2.1 Theory and concepts 2.1.1 Introduction of theories (1)Agency theory

Agency theory explains the relationship exists between owners of the firm (principle) and the management team (agent). Within this relationship, problems can arise because they have conflicting goals and both of them have incentives to maximize their own interests (Jensen and Meckling, 1976). One important assumption will be made: problems occurring between principle and agent are due to the information asymmetry. An agent always has access to the information superior to a principle, because management team has to deal with the business on a day-to-day basis, while most owners only receive information from the firm periodically. In this case, the agent can fool the principle by pursuing his own interest at the cost of the principle.

The principle agency problem can be mitigated by providing right incentives to align the interests of the principle and the agent. According to Shapiro (2005), compensation schemes like bonuses are often served as a tool to align the interest of agents and owners. When incentive contracts are provided, agents want to show their good performance in order to receive high bonuses and maximize their own utility. Earnings management can be involved in such process, and it also helps agents reach the predetermined goals which ensure the receiving of certain level bonuses.

8 / 32

The positive accounting theory (PAT) tries to explain and predict how managers choose accounting practices in different situations. Watts and Zimmerman (1986) provide the formal definition of PAT: “PAT is concerned with explaining accounting practice. It is designed to explain and predict which firms will and which firms will not use a particular method, but is says nothing as to which method a firm should use.” Similar to the agency theory, the positive accounting theory is also based on the assumption that individuals always maximize their own utilities and act in an opportunistic way to increase their own wealth at the cost of others. It explains that the reason for a manager to choose a particular accounting method is to raise their own income. The bonus plan hypothesis, the political cost hypothesis and the debt/equity hypothesis are the three hypotheses associated with the positive accounting theory. All those three hypothesizes explain the incentives for conducting earnings management.

The bonus plan hypothesis states that in the firms with bonus plans, managers will be more likely to select the accounting practice which gives the highest reported income (Watts and Zimmerman, 1990). The intention of selecting accounting practice to increase the bonuses that managers can receive can be viewed as earnings management.

The debt/equity hypothesis illustrates that firm with high debt/ equity ratio is more likely to select accounting practice to boost earnings. Watts and Zimmerman (1990) mentions that firms with high debt/ equity ratio will face tight debt covenant constraint. The tight debt covenant constraint will increase the probability of the covenant violation and may further lead the technique default to the firm. Consequently, selecting accounting practice to relax debt covenant constraint and reduce the possibility of technique default is also classified into the class of earnings management (Watts and Zimmerman, 1990).

The political cost hypothesis focuses on the size of the firm. It predicts that large firms tend to select accounting practices to reduce their reported profits. The reason to show their profit lower is to distract the attention from politicians (Watts and Zimmerman, 1990). Generally speaking, government will pay more attention to those firms with high earnings and place more regulations and laws on them to monitor their business. Therefore, political cost hypothesis implicating accounting practices can be manipulated in a way to reduce the political cost, which can also be classified into earnings management.

9 / 32 2.1.2 What is earnings management?

In this paragraph, the definition of earnings management will be provided.

Healy and Wahlen (1999, page 368) defines the earnings management as the followings: “Earnings management occurs when managers use judgment in financial reporting and in

structuring transactions to modify financial reports, in order to either mislead the stakeholders about the underlying economic health of the company or to influence contractual outcomes that depend on the reported accounting numbers.”

According to Schipper (1989), earning management is an intervention in the external financial reporting process, with the purpose of obtaining some private gain (as opposite to, for example, merely facilitating the neutral operation of process).

It is clear that managers try to affect reported earnings to influence stakeholders about the economic performance of the firm. These influences or earnings management stay within the boundaries of accounting standard. Accounting standards often leave room for flexible explanations/operations on how to follow the standards. However, in some case the measures taken by managers can be fraudulent, while others are not. In general, earnings management can be seen as a measurement of earnings quality. Higher quality earnings provide better information about the firm’s financial performance. Nevertheless, a few points need to be mentioned basing on those definitions. Firstly, Healy and Wahlen (1999) emphasizes that the main purpose of earnings management is to mislead stakeholders or to influence the contractual outcomes.

Moreover, managers can manipulate the quality of financial reports through different approaches, mostly via varying accounting methods.

2.1.3 Incentives for earnings management

In this subsection, the motivation to engage in earnings management will be presented. At first, I am going to list some main incentives for earnings management in companies from prior researches. Afterwards, description on incentives from bank industry is also going to be

presented.

The main incentive is consisted of three aspects as presented in the following. First, companies always have the consiscenious to manage earnings in order to avoid loss during the financial distress period (Burgstahler and Dichev, 1997). Secondly, the payment is linked to

10 / 32

financial reporting (Kedia, 2003; Deegan and Unerman, 2006). Third, debt covenants provide incentives of earnings management (Watts and Zimmerman, 1978).

Moreover, there are also other studies focusing on the incentives for earnings management in bank industry. Firstly, in order to achieve the regulatory capital requirements (Beatty, 1995). The capital incentive arises because regulators (government) monitor banks using accounting capital measures. Since banks are subject to capital requirement and earnings smoothing will happen as a result. Moreover, earnings smoothing through loan loss provisions become less costly after Basel I (Ahmed et al., 1999; Anandarajan et al., 2007). Second, disseminating enhanced earnings information to stakeholders can be considered as another motivation. From Healy and Wahlen’s (1999) study, banks will always take actions to prevent negative effects on the capital market, for example, the stock price decline due to negative financial reports. Earning management is used by bank managers to meet investors’ expectations (Scott, 2009). Last but not least, compensation contract incentive is also important in Bank industry, since the manager’s pay is partly based on the earnings (Healy, 1985; Gibbons, 1998).

2.3 Study of prior relevant literature

2.3.1 Influence of financial crisis on earnings management

In this part, the prior study of the impact of financial crisis on earnings management will be discussed. Surprisingly, limited prior research is available for the earnings management in banking industries, and studies about earnings management in distressed firms are considered as well.

Chia et al. (2007) analyze more than 300 firm-observations in their study from service-oriented listed companies in Singapore, where the market is highly regulated and during a period of financial crisis. The earnings management studies are extended to a more diverse and different institution settings so as to increase the generalizability of the results, as suggested by Kasanen et al. (1996). From the background study of prior literature together with that of Jones (1991), they document the direction of earnings management, insofar as it affects reported earnings, is dependent upon the incentives available to the managers, while the incentive of managers to manage earnings downward in order to benefit from the subsequent introduction of related governmental policies. With the awareness of the fact that Asian financial crisis will result in uncertainty and limit the earnings performance of companies, they propose two hypotheses for

11 / 32

testing: (1) Service-oriented companies engage in reduced earnings management during the Asian financial crisis, and (2) Service-oriented companies audited by Big-6 auditing firms engage in a greater level of reduced earnings management when compared with companies audited by non-Big-6 auditing firms. By adapting the iterative-seemingly-unrelated-regression methodology, their results conclude that service-oriented companies do engage in income decreasing earnings management during the Asian crisis.

Agarwel et al. (2007) investigates the behavior of earnings management in Japanese banks during three different economic periods: from high growth with bubble economy (1985-1990), to stagnant growth with financial distress economy (1991-1996), and severe recession with credit crunch economy (1997-1999). Motivated by the fact that since the middle 1980s Japanese banks experienced upturns and downturns in the economy as well as some structural changes in the function of its financial intermediaries, they investigated the earnings management behaviors from earlier period of the late 1980s to a later period of the late 1990s. As reported by Shrieves and Dahl (2003) on the discretionary accounting practices of Japanese banks during 1989-1996 period, Japanese banks used realized securities gains and loan loss provisions to smooth income, and capital-constrained banks, used earnings management to replenish regulatory during the financial duress period. Based on the methodology from Shrieves and Dahl (2003), they have estimated a simultaneous equation model to empirically assess the role of loan loss provisions and realized gains from securities portfolios on the earnings management practices of 78 banks from 1985-1999 in Japan. As mentioned above, the sample data was divided into three sub-periods, and the model is consisted of four decision variables: lending, securities gains, loan loss provision, and dividends. Their findings reveal that Japanese banks on average realized gains in order to offset the negative impact of loan loss provisions and consequently engaged in income smoothing throughout the whole period (1985-1999). However, they found that Japanese banks used loan loss provisions as a tool of managing earnings only during the high-growth and stagnant periods, and did not use loan loss provisions to smooth income in the face of severe economic recession.

Lin and Shih (2002) extended the investigation from discretionary accruals to mean discretionary accrual to determine whether certain firms engaged in earnings management. They specifically examine whether mean discretionary accruals taken by companies during a recession period are different from those in other periods within the period 1989-1993, including the three quarters of the 1990-1991

12 / 32

recession. They select 513 firms in compustat industrial files which meet their sample selection criteria, and by using a variation of the Jones model (1991), they measure the discretionary accruals for each firm. Empirical results find that firms took pervasive earnings management in the downward direction during the 1990-1991 recession. Mean discretionary accruals were proved to be negative and significant in the 4th quarter of 1990, when the recession was at its most severe stage, or in

quarters of very strong GDP growth; and mostly positive in other quarters. By regression fitting a quadratic function to the relationship of mean discretionary accruals with real GDP growth confirms an inverted U shape of the relationship. Besides, they found that the general direction of earnings management seemed to be highly predictable. Their result is consistent with the hypothesis that managers defer income in periods of very weak earnings (no chance of receiving bonuses which is linked to earnings) and very strong earnings (the bonus cap has been reached) to future periods and manipulate earnings upwards in other periods, under the consideration that the real corporate earnings (before smoothing) and real economic growth are likely to be highly correlated. Moreover, they also proved that the directions and extents of earnings management in individual firms were highly related to their performance during the 1990-1991 recession. Firms with significant revenues declines or increase during the distress period had the largest negative discretionary accruals, which agrees with the idea that earnings management is linked to how bonus plans are administered.

Chen et al. (2010) uses discretionary accruals as a proxy variable to study the earnings

management behaviour of financially distressed listed firms in china from 2002 to 2006, with the type of ultimate ownership and industry to which the company belongs functioning as

independent variables. Their empirical results show that government regulation in industry is likely to affect the earnings management of financially distressed companies, and that the desire to avoid continued special treatment (ST) status and the risk of being de-listed motivate firms to adopt different earnings management behavior.

New Zealand recently experienced a great number of finance company collapses which contributes indirectly to financial distress. In order to examine empirically the managerial earnings management practices of financially distressed firms, and to determine whether these practices changed during the global financial crisis (GFC), Habib et al. (2013) investigate the impact of distress on earnings management with Dechow et al.’s (1995) discretionary accruals (DA) model, under the hypothesis that: no association between financial distress and earnings

13 / 32

management proxied by discretionary accruals, no incremental association between firm distress and earnings management during the crisis period, and the market pricing of discretionary accruals for distressed firms during the GFC are not different from that of their non-distressed counterparts. They find that managers in distressed firms engage more in income-decreasing earnings management practices compared to their healthy firm counterparts. This paper also finds evidence of the global financial crisis effect on the association between financial distress and earnings management. In the end, fingerprint of positive market pricing of discretionary accruals in the non-crisis period, but a substantial reduction in pricing coefficients during the global financial crisis period was also found by the authors.

2.3.2 Relationship between accrual-based and real-based earnings management

In this section, the relationship between accrual-based earnings management and real-based

earnings management is going to be discussed base on prior literature. As mentioned above, non –financial industries engage both in accrual-based earnings management and real-based earnings management and financial industries only use accrual-based earnings management. Therefore, finding the relationship between those two earnings management strategies becomes particularly important to help us to compare accrual-based earnings management in different industries.

Ewert and Wagenhofer (2005) theoretically examine the relationship between earnings management and the tightness of accounting standard. The earnings management is divided into real-based earnings management and accrual-based earnings management in this research. In the paper they argue the accounting standards have large impact on accrual-based earnings

management however real-based earnings management are not influenced that much by accounting rules. Ewert and Wagenhofer (2005) conclude that there is a substitution effect between real-based earnings management and accrual-based earnings management especially when then accounting flexibility is limited.

Based on the work of Ewert and Wagenhofer (2005), Cohen et al. (2008) empirically investigate the influence of SOX on earnings management. Accrual-based and real-based earnings management are studied in U.S. listed firms from 1987 to 2005. According to this research, during pre-SOX period, accrual-based earnings management was mainly used to manipulate earnings, in contrast, real-based earnings management was only used at a very low level. However, after the implementation of SOX, accrual-based earnings management decreased

14 / 32

dramatically, whilst real-based earnings management increased significantly. The empirical results obtained by Cohen et al. (2008) consist with the conclusion made by Ewert and

Wagenhofer (2005) that substitution effect exists between real-based and accrual-based earnings management.

Zang (2012) further examines the substitution effect between accrual-based and real-based earnings management. The study confirms that there is a trade-off relationship between two earnings management. Managers take the cost of earnings management into consideration, if one kind of earnings management is relatively costly, they will switch to another kind of earnings management. Moreover, Zang (2012) suggests that managers will adjust the accrual-based earnings management based on the outcome of the real earnings management. In the paper, the author also mentions that while accrual-based earnings management is well known to be constrained by accounting standards, real-based earnings management is also restrained by the industrial competition, financial health of the company.

2.4 Hypothesis development

Based on the study and discussion from prior literature, I will develop the hypothesis that will be tested during this study.

As seen, the prior literature shows that different economic events have considerable influences on earnings management. For example, Chia et al (2007) concluded that companies engage more in earning management during financial distress, and Agarwel et al. (2007) shows that Japanese banks used realized securities gains and loan loss provisions to smooth income, and capital-constrained banks, used earnings management to replenish regulatory during the financial duress period. Lin and Shih (2002) also points out that firms took pervasive earnings management in the downward direction during the 1990-1991 recession, for example, managers defer income in periods of very weak earnings to future during financial crisis period. Habib et al. (2013) find that managers in distressed firms engage more in income-decreasing earnings management practices compared to their healthy firm counterparts. Therefore, I come up with the following hypothesis:

H1: The accrual-based earnings management of financial and non-financial industries is

15 / 32

As discussed in previous sub-section, Ewert and Wagenhofer (2005) and Zang (2012) both conclude that there is a trade-off relationship between accrual-based earnings management and real-based earnings management. Moreover, those prior studies also confirm that accounting standards and other codes has created a trend for non-financial industries to use more real-based earnings management instead of accrual-based earnings management. Therefore, the accrual based earnings management for those non-financial industries may react differently to financial crisis from financial industry since non-financial industries also uses real-based earnings management. Moreover, some of financial crisis indicators are more sensitive to financial industry than non-financial industries. For instance, Vaz et al. (2008) find that interest rate have large impact on the bank stock resturns. Therefore, the second hypothesis is:

H2: the accrual-based earnings management of financial and non-financial industries will react differently to financial crisis indicators: the lower the stock price of a firm, the more earnings management from the firm

3. Research methodology

The data for this study is going to be collected from Compustat North America database. The Compustat North America database contains large amount of financial information for stock listed North American companies and banks. Since this research is based on the information from US financial and non-financial industries, the largest 30 listed banks (available from Compustat) are selected as a sample to measure the accrual-based earnings management in US financial industry, while the largest 50 industrial firms (available from Compustat) are selected to measure the accrual-based earnings management of non-financial industry. The time period for this research ranges from 2004-2010 (meanwhile, data from 2003 is also included, as

explained in Section 4), which covers both the pre-financial crisis and post financial crisis period. The Bankscope database may also be used as a supplement for collecting bank-related

information if necessary.

3.2 Research design

Research methodology of this paper is quantitative. Main issues are: how to measure the earnings management not only in bank industry, but also in non-financial firms; and how to

16 / 32

evaluate the impact of financial crisis. The following sub-sections are going to address those issues.

3.2.1 Measure earnings management from US banks

McNichols (2000) discusses three approaches to measure accrual-based earnings

management, which are: (1) single accounting item approach based on specific accruals; (2) total accruals model; (3) those based on the distribution of earnings after management. In this study, the first approach of single accounting item is going to be used. Specifically, I am going to use discretionary loan loss provision as the estimator of earnings management. Nichols et al. (2009) demonstrates that discretionary loan loss provision is a good proxy to measure the earnings management in banking industry since loan loss provisions have a major discretionary impact on bank’s earnings.

3.2.2 Model to measure earnings management in banking industry

The multivariate model from Ahmed et al. (1999) is going to be applied here. This model has been successfully and widely used in prior literature (Betty et al., 2002; Anandrajan et al., 2007; Cohen et al., 2011). The basic principle is to regress:

LLP= β₀+ β₁∆NPL + β₂CAPR + β₃LNASSET + β₄LLA + ε (1) Where:

LLP: Loan loss provision as a fraction of average gross loans;

ΔNPL: Change in non-performing loans as a fraction of average gross loans; CAPR: Ratio of actual regulatory capital (Tier I Capital), divided by 100. LNASSET: Natural log of total asset

LLR: loan loss allowance as a fraction of the gross loans at beginning of each year. ε: residual from the regression.

The model is consisted of two parts. The first part estimates the non-discretionary part of loan loss provision. The second part is the residual of this regression model which measures the discretionary part of loan loss provision. In the next paragraph I am going to explain the reason for choosing those dependent variables to estimate the non-discretionary part of loan loss provision.

The ΔNPL are those loans which have the high possibility to a default, and it implies the

17 / 32

non-discretionary loan loss provision. The variable CAPB indicates the capital management incentive of using the loan loss provision. Banks with low capital will be more likely to use the loan loss provision to boost their capital. As discussed in Anandrajan et al. (2007), LNASSET is used as a measure of bank’s size. Banks with larger size involve higher level of business and face more risks. Thus, higher level of loan loss provision is expected for larger banks (Anandrajan et al. 2007). Beatty et al. (2002) state that companies with high level of loan loss reserve are expected to use more loan loss provision. Therefore, LLR is also included as an independent variable. The non-discretionary part of the loan loss provision (NDLLP) can be estimated with the help from independent variables in this regression model (Ahmed et al., 1999):

NDLLP = β₀+ β₁∆NPL + β₂CAPR + β₃LNASSET + β₄LLA (2)

Using the regression results (coefficients), discretionary accruals (DLLP) will be predicted based on the following equation (Dechow et al., 1995):

DLLP = LLP – NDLLP=LLP - ( β₀+ β₁∆NPL + β₂CAPR + β₃LNASSET + β₄LLA) (3) Where:

DLLP: Discretionary part of the loan loss provision. LLP: Loan loss provision as a percentage of gross loans. NDLLP: Non-discretionary part of the loan loss provision.

3.2.3 Measuring earnings management in US non-financial firms

Total accrual approach is always used to measure earnings management in non-financial firms, which focuses on multiple accounting items. There are many studies discussing about different models to detect the use of earnings management, and according to the prior research, the modified Jones model from Dechow et al. (1995) has been proved to be the most powerful model to measure the use of earnings management in non-financial firms. Total accruals are examined and categorized into discretionary (for example, sales and advertising expenses) and nondiscretionary accruals (e.g. salary expenses). Discretionary accruals are often used as a proxy for accruals earnings management since these accruals are directly influenced by managements’ actions. However, discretionary accruals are difficult to estimate, which leads to the developed Jones model (Dechow et al., 1995) and other models which use nondiscretionary accruals to estimate discretionary accruals.

18 / 32

Since this thesis investigates the use of earnings management over a period, a time series approach and a cross sectional method will be introduced in the modified Jones model. The time series approach uses the estimate of the specific firm parameters for every firm in the sample. To measure the use of earnings management in current period this approach uses data from a prior period. While the cross sectional approach is more year-specific and industry-related. A

disadvantage of using the time series is that a long time series is needed when estimating the parameters at the first step. Since it is assumed that these non-financial companies should provide financial information in an accurate way for the users of financial statements, it is not expected to encounter the issue of insufficient time series data in this thesis. In addition, when the use of earnings management is expected via bad debt accounts or revenue, the time series approach of the modified Jones Model are expected to be more proper to detect the use of earnings management.

The modified Jones Model (Dechow et al., 1995) to be used can be expressed in the following equation:

TAi,t = α1 (1/ Ai,t-1) + α2 (ΔREVi,t /Ai,t- 1) + α3 (PPEi,t / Ai,t-1 ) + εi,t (4)

Where:

TAi,t : total accruals of firm i at year t scaled by lagged total assets, calculated as the difference between

income before extraordinary items and operating cash flow in year t, as a percentage of the total assets in year t-1.

Ai,t-1: total assets of firm i in year t-1.

ΔREVi,t : change in revenues of firm i from year t-1 to year t.

PPEi,t : gross property, plant and equipment of firm i in year t.

α1, α2, α3: firm specific parameters.

t: a year subscription indicating a year in the estimation period.

To control the portion of the total accruals related to the non-discretionary depreciation expenses, gross property plant and equipment is included

After estimating the total accruals (TAi,t) and the firm-specific parameters (α1, α2 and α3) by

19 / 32

changes in revenue, receivables data of the period, the non-discretionary accruals are calculated by using the following formula :

NDAi,t = α1 (1/ Ai,t-1) + α2 ( ΔREVi,t - Δ RECi,t ) + α3*PPEi,t / Ai,t-1+ εi,t (5)

Where:

NDAi,t : non-discretionary accruals of firm i in year t scaled by lagged total assets

ΔREVi,t : change in revenues of firm i from year t-1 to year t

ΔRECi,t : change in net receivables of firm i when moving from year t-1 to year t-1

PPEi,t : gross property, plant and equipment of firm i in year t

Ai,t-1 : total assets of firm i in year t-1

α1, α2, α3: firm specific parameters

t: a year subscription indicating a year in the event period

ΔREV and ΔREC are included since it is based on the assumption that all changes in the credit sales in the event period are the results from the use of earnings management.

Once the non-discretionary accruals are calculated, the discretionary accruals can be obtained by:

DAi,t = TAi,t – NDAi,t = TAi,t – ((α1 (1/ Ai,t-1) + α2 ( ΔREVi,t - Δ RECi,t ) + α3*PPEi,t / Ai,t-1+ εi,t)) (6)

Where:

DAi,t: discretionary accruals of firm i in year t

TAi,t : total accruals of firm i at year t scaled by lagged total assets

NDAi,t : non-discretionary accruals of firm i in year t scaled by lagged total assets

3.2.4 The influence of financial crisis on the use of earnings management

As stated in the beginning, the last part of my thesis will be the test of how the financial Crisis affects the level of earnings management, and furthermore, this study will examine whether there is a difference, with respect to earnings management for both financial industry and non-financial industries.

20 / 32

Again, the discretionary accruals estimated above will be used. Therefore, the following model is developed based on control variables size, growth and leverage.

𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = 𝛽𝛽0 + 𝛽𝛽1( 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐼𝐼𝑟𝑟𝐼𝐼𝐼𝐼) + 𝛽𝛽2(𝐺𝐺𝐷𝐷𝐷𝐷) + 𝛽𝛽3 (𝑆𝑆𝐼𝐼𝑆𝑆𝑆𝑆𝑆𝑆 𝐷𝐷𝐼𝐼𝑃𝑃𝑆𝑆𝐼𝐼) + 𝛽𝛽4( 𝐼𝐼𝑃𝑃𝑠𝑠𝐼𝐼) + 𝛽𝛽5(𝐺𝐺𝐼𝐼𝑆𝑆𝐺𝐺𝐼𝐼ℎ) + 𝛽𝛽6(𝐷𝐷𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑟𝑟𝐿𝐿𝐼𝐼) + 𝜀𝜀 (7)

𝐷𝐷𝐷𝐷 = 𝛽𝛽0 + 𝛽𝛽1( 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐼𝐼𝑟𝑟𝐼𝐼𝐼𝐼) + 𝛽𝛽2(𝐺𝐺𝐷𝐷𝐷𝐷) + 𝛽𝛽3 (𝑆𝑆𝐼𝐼𝑆𝑆𝑆𝑆𝑆𝑆 𝐷𝐷𝐼𝐼𝑃𝑃𝑆𝑆𝐼𝐼) + 𝛽𝛽4( 𝐼𝐼𝑃𝑃𝑠𝑠𝐼𝐼) + 𝛽𝛽5(𝐺𝐺𝐼𝐼𝑆𝑆𝐺𝐺𝐼𝐼ℎ) + 𝛽𝛽6(𝐷𝐷𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑟𝑟𝐿𝐿𝐼𝐼) + 𝜀𝜀 (8)

Where:

DLLP: Discretionary loan loss provision for the bank.

DA: Discretionary accruals of the firm, obtained from the modified Jones model. Interest rate: level of nominal interest rate

GDP: real per capital GDP. Stock Price: stock market prices.

Size: the natural logarithm of total assets of the firm.

Growth: the growth of a company, measured by the ratio of market value of equity over book values of assets.

Leverage: ratio of long-term debt to total assets (normalized)

Variables of interest rate, GDP and Stock price are included because they are generally considered as indicators of financial crisis, as well known from literature. Since the use of earnings management is related to the growth of a company, the book-to-market ratio in the regression is included. The book–to-market ratio will be measured by the book–to-market capitalization divided by the shareholders equity value. Moreover, high amount of debt creates high cost of capital which may encourage companies to use earnings management to increase the reported income. Companies with high level of leverage might reach the point that they cannot pay of their debt. For the possible effects of leverage on the use of earnings management a measure of leverage is added to the regression. The leverage will be measured by dividing end of year long term debt by end of year total assets.

21 / 32

4. Empirical Results

This chapter focus on the discussion of the raw data obtained from COMPUSTAT and the regressed results after analysing, and finally, with the information from discussion I will test whether those two hypothesises mentioned before are true or not. Paragraph 4.1 shows the descriptive statistics of all variables and the change of stock price from North American firms during financial crisis. Paragraph 4.2 first lists the results from regression and estimation, and then discusses the reasonability of these

regression and estimated coefficients. In the end, a comparison of the reactions from financial and non-financial industries will be made.

4.1 Descriptive statistics of independent variables

Table 1 reports descriptive statistics for all variables for financial industries in USA. One remarkable point from Table 1 is that 1.1% of all loans has been used for the loan loss provision in the period of financial crisis, which is much higher compared to reports from literature. The 0.38% change in non-performing loans/assets indicates a small increase in non-non-performing loans. The Loan loss reserve is on average 1.86% of total loans, which is also in line with the loan loss provisions. The average ratio of actual regulatory capital (Tier I) is 10.5% during the whole financial crisis, which is higher than the minim requirement 4% for U.S. banks( which are required to maintain a minimum tier 1 capital ratio of 4% of risk weighted assets. This ratio must exceed 8% to be considered well capitalized).

Table 1: Descriptive Statistics (Financial industries-bank)

Variables Mean Std Min Max Observations

LLP .0111091 .0144316 -.004788 .0677482 210 DLLP -2.25e-07 .0077237 -.0409568 .0363812 210 NDLLP .0111094 .0121909 -.0061737 .0553292 210 ∆NPL . 0038819 .0115848 -.0429927 .053048 210 CAPR1 .1054319 .0264845 .068 .241 210 LNASSET (Size) 4.787113 .7482317 2.698402 6.486509 210 LLR .0186713 .010522 .0011684 .0604298 210 Growth .8827103 .577947 .2111123 3.055061 210 Leverage .0918497 .0578417 .0048249 .4140838 210 Stock Price 42.24871 33.69995 2.05 202.64 210 Notes:

LLP = Loan Loss Provisions as a fraction of average gross loans (average of gross loans at beginning and at end of each fiscal year).

22 / 32 DLLP = Discretionary part of the loan loss provision. NDLLP = Non-discretionary part the loan loss provision.

∆NPL = the change of non-performing loans/assets in one fiscal year. CAPR1= ratio of actual regulatory capital (Tier I), divided by 100. LNASSET = Natural logarithm of total assets.

LLR = Loan loss reserve as a percentage of gross loans at beginning of the year.

Growth = the growth of a company, measured by the ratio of market value of equity over book values of assets, book value is calculates as the difference between total assets and total liabilities.

Leverage = ratio of long-term debt to total assets.

Table 2 summarizes the descriptive results from non-financial industry (companies). As seen in the table, the mean value of total accruals in non-financial companies is negative during the sampled period, while the discretionary accruals is around 20% of total accruals. The change of REC and REV is positive in general, while the size of non-financial companies is in the same order as that of banks. The rest of the table includes growth, leverage, which are both positive and similar to case of banks. The stock price, which is considered as an important indicator of financial crisis, is higher than that of financial industry.

Finally, the changings of stock price for both banks and non-financial companies are as a function of fiscal year from 2004-2009, are plotted as well. Similar trends are found from the

Table 2: Descriptive Statistics (Non-financial industry)

Variables Mean Std Min Max Observations

Total accruals -.0520737 .0984159 -.6631702 1.341648 350 DA -.010351 .0010492 -.0161619 -.0006034 350 NDA -.0417227 .0983244 -.6470083 1.351628 350 ∆REC .0136714 .0446965 -.401596 .2612346 350 ∆REV .1521867 .2823025 -1.48261 1.977478 350 LNASSET (Size) 4.682979 .4288495 3.520267 5.901877 350 PPE .5769739 3760566 .0589621 1.801442 350 Growth .6149518 7.354949 -43.68467 122.9512 350 Leverage .1581547 .1167998 0 .5072194 350 Stock Price 56.51784 72.57489 .471 691.48 350

23 / 32

graph, while the stock price kept going up until 2008 when the financial crisis happened, a sharp drop was found and then the price recovered gradually.

4.2 Regression analysis

The study continues with the analysis of the regression from equation (1), (4), (7) and (8) using linear regression method. Table 3 shows the regression results for bank industry, where 71.36% of the variations in the loan loss provision (LLP) can be explained by these independent variables. This indicates that the regression model (1) well explains the variability in the loan loss provision. The variables ∆NPL, CAPRI, LLR are significant at 1%, while LAASSET (size) is significant at 10%.

24 / 32

Table 3 Estimating the non-discretionary portion of the loan loss provision F (4, 205) = 127.68, Prob > F = 0.0000, R-squared = 0.7136, Adj R-squared = 0.7080 Root MSE = .0078

LLP Coef. Std. Err. t P>|t| [95% Conf. Interval] ∆NPL CAPRI LAASSET LLR .3590215 .0292359 7.56 0.000 .2654251 .4526178 .1069305 .021392 5 0.000 .064754 .1491069 .0012174 .0007786 1.56 0.119 -.0003177 .0027524 .9366469 .00546168 16.82 0.000 .8268799 1.046414

Note: LLP= β₀+ β₁∆NPL + β₂CAPR + β₃LNASSET + β₄LLA + ε

As mentioned in previous studies (Ahmed et al., 1999; Beatty et al., 2002), the nonperforming loans (NPL) are loans that are in or close to a default, and the change of nonperforming loans (∆NPL) is a determinant of the non-discretionary component of the loan loss provision (Ahmed et al., 1999). As known, the non-discretionary component of the loan loss provision cannot be changed by a manager; ∆NPL is then also an important indicator of the loan loss provision. An increase in nonperforming loans should lead to an increase as well in loan loss provision and decrease in earnings. Therefore, positive coefficient is expected between LLP and ∆NPL. Similar expectation is considered for between LLP and LLR, since an increase of reserve of loan loss will normally result in the increase of loan loss provision, and the contribution from LLR in the regression model of LLP should be quite significant due to the similar level of LLP and LLR. Banks with greater credit risk in the loan portfolio always maintain higher

capitalization levels, in order to adsorb potential loan losses (Liu and Ryan 2006). Consequently,

a positive relationship between CAPRI and LLP is expected, as also found in the regression. LAASSET is supposed to be positively related to LLP as an increase in the natural log of total assets implies an increase in default risk (Beatty et al., 2002). In all, the regression model well describes the relationship between the dependent variable LLP and other independent variables as expected.

With the estimated coefficients for each variable from the regression, the non-discretionary part of the loan loss provision (NDLLP) can be predicated. The discretionary part of the loan loss provision (DLLP) is estimated by subtracting the non-discretionary part from the total loan loss

25 / 32

provision, and consequently, the DLLP is just estimation and couldn’t be calculated precisely. When the discretionary part of the loan loss provision has been estimated, the regression model (7) to test hypothesis can be estimated.

The regressed values for non-industry companies are also listed Table 4. While 14% percent of the variation in the total accruals can be explained by these 3 independent variables, the rest should be related to other factors which are not included in the model but indeed affect the total accruals. The variables Total assets and PPET is significant at 1%, while ∆Rev is not significant at all. The variable PPET is negatively associated with total accrual as expected, since the PPE controls for the non-discretionary part of total accruals and an increase of PPE means changes in operating activity and an increase of deprecation level (means less earnings).

Table 4 Estimating the non-discretionary accrual of the total accruals F (3, 311) = 18.22, Prob > F = 0.0000, R-squared = 0.1495, Adj R-squared = 0.1413 Root MSE = .0912

Total accruals Coef. Std. Err. t P>|t| [95% Conf. Interval] 1/Ai,t-1

Revi,t / Ai,t-1

PPETi,t / Ai,t-1

-431.2432 81.34288 -5.30 0.000 -591.2951 -271.1912 -.0234731 .021464 -1.09 0.275 -.065706 .0187599 -.0373723 .01371 -2.73 0.007 -.0643485 -.0103962

TAi,t = α1 (1/ Ai,t-1) + α2 (ΔREVi,t /Ai,t- 1) + α3 (PPEi,t / Ai,t-1 ) + εi,t

Next step is to use these industry parameters estimated previously in equation (4) to divide the total accruals into a discretionary part (DA) and a non-discretionary part (NDA), as described in equation (5) and (6).

4. 3Test hypothesis

When the discretionary part of loan loss provisions for banks and the discretionary accruals for non-financial companies are estimated with equation (2), (3) and (5), (6) respectively, the regression in model (7) and (8) can be performed and the estimated results are listed in Table 5 and table 6 respectively.

26 / 32

Table 5 Earnings management on discretionary loan loss provisions F (6, 203) = 8.55, Prob > F = 0.0000, R-squared = 0.2018, Root MSE = .007

DLLP Coef. Std. Err. t P>|t| [95% Conf. Interval] Interest rate GDP Price Size Growth Leverage .042399 .0292359 1.45 0.149 -.015246 .1000439 -1.93e-07 2.50e-07 -0.77 0.442 -6.86e-07 3.01e-07 -.0000374 .0000162 -2.30 0.022 -.0000693 -5.39e-06 -.0004375 .0006727 -0.65 0.516 -.0017639 .0008888 .0055564 .0010819 5.14 0.000 .0034233 .0076896 .00127 .0088462 0.14 0.886 -.0161723 .0187123 Notes:

Interest rate: level of nominal interest rate, the sum of real interest rate and the inflation rate. GDP: real GDP per capital

Price: return stock price

Size: the natural logarithm of total assets of the firm.

Table 6 Earnings management on discretionary accruals

F (6, 308) = 4.34, Prob > F = 0.0003, R-squared = 0.0779, Root MSE = .00102

DA Coef. Std. Err. t P>|t| [95% Conf. Interval] Interest rate GDP Price Size Growth Leverage -.0019725 .0031567 -0.62 0.533 -.0081839 .0042389 4.88e-08 2.59e-08 1.88 0.061 -2.22e-09 9.99e-08 -2.55e-06 8.28e-07 -3.08 0.002 -4.18e-06 -9.19e-07

.0003076 .000139 2.21 0.028 .000034 .0005811 -6.89e-07 7.88e-06 -0.09 0.930 -.0000162 .0000148 .0004359 .0005265 0.83 0.408 -.0006001 .0014719

Variables of interest rate, GDP and stock price are generally considered as indicators of financial crisis, as widely reported from literature. As seen from these regressed results, these three variables do correlate with discretionary accruals not only in financial industry, but also in non-financial companies. As interest rate and GDP are same for all the banks and companies in North America, not surprisingly, they don’t possess significant relationship with the variation of

27 / 32

discretionary accruals. In financial industry, the variables interest rate and GDP are not

significant, while in the case of non-financial companies, the variable GDP is significant at the level of 10%. Besides these ignorable correlations, the variable stock price does have a

significant coefficient at 5% and 1% for banks and companies, respectively. In general, a company’s stock is measured by the present value of its future earnings, investors and analysts look to earnings to determine the attractiveness of a particular stock. Companies with poor earnings prospects will typically have lower share prices than those with good prospects

(Rahman et al., 2013); consequently, managers from lower-stock-priced companies feel more

incentive to manage earnings in order to attract more investors and investments, especially during the financial crisis period and negative coefficient of the stock price on discretionary component of accruals is expected, which can also be regarded as an indication for earnings management. Under the analysis of regressed results from interest rate, GDP and stock price (the general indicator of financial crisis) the hypothesis 1 has been tested and it is proven to be correct that the accrual-based earnings management of financial and non-financial industries is associated with financial crisis.

From these summarized results in table 5 and table 6, it is visible that there are difference in coefficient for unique variable between financial industry (bank) and non-financial companies. While the variable Size doesn’t show significant relationship with the discretionary accruals for banks, it has a significant coefficient at the level of 5% for non-financial industry. The negative sign indicates that the size (or total assets) of a company is inversely proportional to the earnings management (the discretionary part of accruals), the bigger the total assets, the less earning management made by managers. Actually, this observation makes a lot of sense during the period financial crisis: the smaller the company, the more pressure or more motivation to smooth/manage earnings in order to obtain more bonuses for the shareholders and manager themselves. The variable Growth is positively related to the dependent variable DLLP for banks and is significant at 1%, but it doesn’t show significant relationship with DA for non-financial companies. The positive relationship can be explained by the factor that the fast growing

financial industry needs more earnings management during the financial crisis period, in order to meet the requirement from fast growing mode; for example, more money/cost to keep the

growing trend during financial crisis while earnings management seems to be the most efficient and direct approach. In the end, the variable leverage seems to be not import for discretional

28 / 32

accruals from the statistic point of view, for both banks and non-financial industry. This is a sign that the long-term debt during financial crisis doesn’t affect the discretionary accruals not only in banks, but also in companies.

In one word, the accrual-based earnings management of financial and non-financial

industries both reacted to financial crisis indicators, but the reaction was different. At this stage, the hypothesis 2 has also been tested and found to be true.

5. Conclusions

Although earnings management has been widely investigated in recent years, most of the prior research only focuses on industrial firms. Banks and other financial institutions are often excluded from relevant research due to their special characteristics. By concentrating on the empirical study of both financial and non-financial industries in North American during financial crisis period, this thesis provides a deeper understanding of the earnings management in both banking industry and non-financial companies, which is quite contributive to investors and analysts. Moreover, the influence of financial crisis on earnings management in financial and non-financial industry was investigated in detail via the study of the largest 30 banks and the biggest 50 companies in North American during the 2007 financial crisis period. From the empirical linear regression between discretionary accruals and several independent variables, including the interest rate, GDP and Stock market price and so on, solid association between financial crisis and accrual-based earnings management is found but different reactions to financial crisis are discovered between financial and non-financial industry.

6. References

Agarwal Sumit, et al., (2007). "Earnings management behaviors under different economic environments: Evidence from Japanese banks." International Review of Economics & Finance 16(3): 429-443.

Ahmed Anwer S., Carolyn Takeda, and Shawn Thomas. (1999). "Bank loan loss provisions: a reexamination of capital management, earnings management and signaling effects." Journal of Accounting and Economics 28(1): 1-25.

29 / 32

Anandarajan Asokan, Iftekhar Hasan, and Cornelia McCarthy.,(2007). "Use of loan loss

provisions for capital, earnings management and signalling by Australian banks." Accounting & Finance 47(3): 357-379.

Beatty, Anne, Sandra, L. Chamberlain, and Joseph Magliolo, (1995) "Managing financial reports of commercial banks: The influence of taxes, regulatory capital, and earnings." Journal of

Accounting Research 33(2): 231-261.

Beatty, Anne L., Bin Ke, and Kathy R. Petroni. (2002) "Earnings management to avoid earnings declines across publicly and privately held banks." The Accounting Review 77(3): 547-570. Burgstahler, David, and Ilia Dichev. (1997). "Earnings management to avoid earnings decreases and losses." Journal of accounting and economics 24(1): 99-126.

Chen, Y., C. Chen and S. Huang (2010) "An appraisal of financially distressed companies' earnings management: Evidence from listed companies in China", Pacific Accounting Review, 22.1 (2010): 22-41.

Chia, Yew Ming, Irvine Lapsley, and Hing-Wah Lee., (2007). "Choice of auditors and earnings management during the Asian financial crisis." Managerial Auditing Journal 22(2): 177-196. Cohen, Lee, et al., (2011). "Bank Earnings Management and Tail Risk during the Financial Crisis". Working paper, Boston College.

Collins, J., Shackelford,D., Wahlen, J., (1995). "Bank differences in the coordination of regulatory capital, earnings and taxes". Journal of Accounting Research 33(2), 263-292. Dechow, Patricia M., Richard G. Sloan, and Amy P. Sweeney., (1995). "Detecting earnings management." Accounting Review: 193-225.

Dechow, Patricia M., Skinner, Douglas J, (2000). "Earnings Management: Reconciling the Views of Accounting Academics, Practitioners, and Regulators", Available at SSRN: http://ssrn.com/abstract=218959.

Deegan, Craig, and Jeffrey Unerman, (2006). "Financial accounting theory: European edition". Maidenhead: McGraw-Hill.

30 / 32

Gibbons, Robert.,(1998). "Incentives in organizations". No. w6695. National Bureau of Economic Research.

Greenwald, Bruce and Stiglitz, Joseph E, (1988). "Pareto Inefficiency of Market Economies: Search and Efficiency Wage Models," American Economic Review, American Economic Association, vol. 78(2), pages 351-355.

Habib, Ahsan, Md Borhan Uddin Bhuiyan, and Ainul Islam, (2013). "Financial distress, earnings management and market pricing of accruals during the global financial crisis", Managerial Finance 39(2): 155-180.

Harry Huizinga and Luc Laeven. (2009) "Accounting discretion of banks during a financial crisis", 2009 International Monetary Fund Woring Paper, 09/207.

Healy, Paul M., (1985). "The effect of bonus schemes on accounting decisions." Journal of accounting and economics 7(1): 85-107.

Healy, Paul M., and James M. Wahlen., (1999). "A review of the earnings management literature and its implications for standard setting." Accounting horizons 13(4): 365-383.

Heninger, William G., (2001). "The association between auditor litigation and abnormal accruals." The Accounting Review 76(1): 111-126.

Jensen, Michael. C and Meckling, William. H, (1976). "Theory of the firm: managerial behavior, agency costs and ownership structure". Journal of Financial Economics 3: 305-360.

Jones, J. (1991), “Earnings management during import relief investigations”, Journal of Accounting Research, 29(2): 193-228.

Kasanen, E., Kinnunen, J. and Niskanen, K. (1996), “Dividend-based earnings

management: evidence from Finland”, Journal of Accounting and Economics, 22(2): 283-312. Kedia, Simi., (2003). "Do executive stock options generate incentives for earnings management? Evidence from accounting restatements." NBER working paper.

31 / 32

Kothari, S. P., Andrew J. Leone, and Charles E. Wasley. (2005). "Performance matched discretionary accrual measures." Journal of accounting and economics 39(1): 163-197.

Liu, C., and S. Ryan. (2006). "Income Smoothing over the Business Cycle: Changes in Banks’ Coordinated Management of Provisions for Loan Losses and Loan Charge-Offs form the Pre-1990 Bust to the 1990s Boom. " The Accounting Review 81: 421-441.

McNichols, M. F., (2000). "Research design issues in earnings management studies". Journal of Accounting and Public Policy, 19(313): 345.

Nichols, D. Craig, James M. Wahlen, and Matthew M. Wieland. (2009). "Publicly traded versus privately held: implications for conditional conservatism in bank accounting." Review of

accounting studies 14(1): 88-122.

Md. Musfiqur Rahman, Mohammad Moniruzzaman, Md. Jamil Sharif. (2013). "Techniques,

Motives and Controls of Earnings Management". International Journal of Information Technology and Business Management 11(1): 22-34.

Shih, Michael, and Zhi-Xing Lin. (2002). "Earnings Management in Economic Downturns and Adjacent Periods: Evidence from the 1990-1991 Recession." Available at SSRN 331400. Schipper, K., (1989). "Earnings Managemen", Accounting Horizons 3(4): 91.

Shrieves, R. E., and Dahl, D. (2003). "Staying afloat in Japan: Discretionary accounting and the behavior of banks under financial duress". Journal of Banking and Finance, 27: 1219−1243. Scott, William R., (2009). "Financial Accounting Theory", Pearson Prentice Hall.

Shapiro, S., (2005). "Agency Theory", Annual Review of Sociology, 31(1): 263-284. Walker, M. (2013). "How far can we trust earnings numbers? What research tells us about earnings management", Accounting and Business Research, 43(4): 445-481.

Watts, R.L., Zimmerman, J.L., (1978). "Towards a Positive Theory of the Determination of Accounting Standards", The Accounting Review 53: 112-134.

32 / 32

Watts, R. and Zimmerman, J., (1986). "Positive Accounting Theory", Edgewood Cliffs, NJ: Prentice Hall.

Watts, R.L., Zimmerman, J.L., (1990). "Positive Theory: A ten year perspective", The