The impact of the financial crisis in 2008-2009 on accrual accounting earnings quality in U.S. profit and non-profit organizations

(1)

The impact of the financial crisis in 2008-2009 on accrual

accounting earnings quality in U.S. profit and non-profit

organizations

Name: Yahia Akhribech Student number: 11095822

Thesis supervisor: Alexandros Sikalidis Date: 26 June 2017

Word count: 14.708

MSc Accountancy & Control, specialization [Accountancy] Faculty of Economics and Business, University of Amsterdam

(2)

2 Statement of Originality

This document is written by student Yahia Akhribech who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

This study examines the relationship between earnings management and the financial crisis of 2008 in profit and non-profit organizations. It aims to know if the level of earnings management decreases during crisis periods compared to pre-crisis periods in both profit and non-profit organizations. Prior research has shown that in time of crisis the use of earnings management can vary in non-profit organizations than in pre-crisis times, which is the same as in the profit sector. Prior literature shows that during crisis year’s organizations are a subject to increased monitoring from different stakeholders, such as auditors and creditors. This could result in managers having less incentives to manage their earnings. Prior literature also indicated that due to the increased uncertainty about future outcomes, it will motivate different market forces to demand more conservative earnings in crisis periods. The findings show that the level of earnings management through accruals is lower in crisis periods compared to pre-crisis periods in non-profit organizations. In addition, the findings also shows that the decrease in the use of earnings management in profit organizations during the crisis periods compared to pre-crisis periods was not significant.

(4)

4

5 5 Empirical results ... 25 5.1 Hypothesis 1 ... 25 5.1.1 Descriptive statistics ... 25 5.1.2 Regression ... 26 5.1.3 Robustness testing ... 30 5.2 Hypothesis 2 ... 32 5.2.1 Descriptive statistics ... 32 5.2.2 Regression ... 33 5.2.3 Robustness testing ... 37 6 Conclusion ... 39

6.1 Research question and Summary ... 39

6.2 Limitations ... 40

6.3 Further research ... 40

(6)

6

1 Introduction

This chapter is an introduction to the subject of this research. Hereby I describe the background of the important elements of the research. The emphasis is placed on earnings management, the financial crisis and the profit and non-profit sector. Furthermore, I will explain the development of my research question. Lastly, I provide the motivation of this research. In short I want to look what impact the recent financial crisis had on the level of earnings management in profit and non-profit organizations.

1.1 Background

My research is built on three key fundamentals, the recent financial crisis, earnings management and the non-profit sector. All these elements have been examined in various ways in other studies. Some of these elements have been deeply examined and thus provide a broad literature, such as earnings management. To gain better insight into my research, I will provide the background information of these elements.

Over the past years, the financial crisis has attracted the attention of different parties, both academics and practitioners. The financial crisis had major effects on the level of economic activities. From the ‘70s until 2008, the world has experienced an uninterrupted period of economic growth with continuous positive GDP growth 3. But unfortunately in 2009 for the first time in 40 years the GDP decreased. Especially in Europe the crisis was particularly serious because the GDP growth decreased to -4,30 % versus -2,05 % for the world at large. Typically, the period before the financial crisis is the global strong economic growth the world has known. This was mainly due to the strong economic growth, low inflation and an increase in international trade, in particular the financial flows (Obstfeld and Rogoff, 2009). In this regard, Obstfeld and Rogoff (2009) distinguished between three clear trends that undermine the seemingly favorable balance of the period before the crisis. First, they mention the rising values of real estate in many countries at a high rate. Secondly, they mention the existence of some simultaneous high and rising deflections on the current account in some countries. Finally, the relationship between the equity and the debt accumulated in many sectors around the world to exceptionally high levels. On the basis of this distinction they indicate that these trends have been important factors for the emergence of the credit crisis.

According to Verbruggen and Christiaens (2012) earnings management is a vital area of accounting research. In this regard Healy and Wahlen (1999) gave a definition that sets the tone for different papers on earnings management:

(7)

7 “Earnings management occurs when managers use judgment in financial reporting and in structuring transactions alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers” (Healy and Wahlen, p. 365).

There exist a constant stream of literature with regard to the earnings management in different companies. There are also different motives for earnings management that have been described over the past years, such as manipulation of stock markets, decline in the levels of taxation, avoiding political cost and retention of CEO reputation (Verbruggen and Christiaens, 2012). Recently the extent of the earnings management research has expended to a level to also include the non-profit and public sector. Healy and Wahlen (2009) used in their definition as I mentioned above only “companies”, but there are no motives or techniques that suggest that earnings management is only restricted to profit organizations. Verbruggen and Christiaens (2012) on the contrary, indicate that since the economic performance is being controlled by the society that clearly demands accountability, it implies that earnings management is quite possible important in the non-profit sector. Furthermore, the growing importance of the non-profit sector implies that an assessment of the quality of financial reporting to many donors, government agencies, tax authorities, staff and volunteers, and also accounting standard setters is relevant.

The dependence of the stability of non-profit organizations is related to their ability to acquire and maintain resources. Non-profit organizations have some main sources of income, such as the sale of goods and services, private contributions and government subsidies. As it is known that governments often call for cuts in difficult times, which leads to the cut of general budgets by the government. This is due to the declines in the fiscal revenue and individual donors are also reducing their donations to the non-profit organizations. This implies that susceptibility is playing a major role in non-profit organizations during the financial crisis. Since non-profit organizations are susceptible to effects of a financial crisis and the associated economic depression (Verbruggen and Christiaens, 2012).

1.2 Research question

There are several studies focusing on the different core elements that I have described. To my knowledge there is not really done any research on the effects of the financial crisis on earnings management in the non-profit sector. To contribute to the prior literature I want to prove statistically if the level of earnings management in profit and non-profit organizations decreases during the crisis periods compared to the pre-crisis periods. To gain better insight into the research, I have shown the research question schematically in figure 1 which consists of the three core elements. On this basis, I have formulated the following research question that will be answered in

(8)

8 this study: To what extent does the recent financial crisis has an impact on the earnings management level in profit and non-profit organizations?

Figure 1: research question schematically

1.3 Motivation

My research can be viewed from different angles, such as the scientific point of view and the societal point of view. One of the important objectives that I want to achieve with this research is to contribute to the existing literature in different ways. To begin with the scientific point of view, it will contribute through research that focuses specifically on the level of earnings management. Although earnings management is a broad topic in science I still think it can make a small contribution. I’m going to do it by measuring the level of earnings management that emerges from the impact of a particular event (financial crisis). To make it more specific I will focus on two sectors, which are the profit and non-profit sector. This will thus (possibly) entail new findings, which will show that the use of earnings management in profit and non-profit organizations will decrease in crisis periods compared to pre-crisis periods. Therefore it can be used in various ways, such as for future research, because there is an existing gap in the literature with regard to the research about the decrease of earnings management during crisis periods compared to pre-crisis periods in non-profit organizations.

From the societal point of view, the research can certainly contribute. This is possible through two core elements I have described previously, which are the recent financial crisis and the non-profit sector. These two elements are closely related to the society because both elements also have an effect on society. The financial crisis had an impact on almost all aspects of the society. Including this element in my research will thus be able to contribute. The non-profit sector will also contribute. This is mainly due to the short distance between the non-profit sector and the society. An important source of income for non-profit organizations is for example from donors. Important questions that will be asked are, what happens if the donations fall?

Recent financial crisis Level of earnings management Non-profit organizations Profit organizations

(9)

9

2 Literature review

In this section I will describe the literature relevant to the core elements. This chapter of the thesis is also classified on the basis of the core elements of the research. Its purpose is to collect the key literature elements. The literature includes a description of the recent financial crisis, earnings management, non-profit organizations, the positive accounting theory, the positive accounting motives and some earnings management strategies. My research will focus on profit and non-profit organizations from the United States of America. Because of this I will describe the literature in the context of this country.

2.1 Financial crisis

It is known that the recent recession actually began in 2007, is the longest and probably the deepest economic recession since the Great Depression that people have known in the 1930s. The Committee of the National Bureau of Economic Research has defined a recession as “a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in production, employment, real income, and other indicators” (Determination of the December 2007 Peak in Economic Activity, 2008, p. 1). Isidore (2008) indicated that the National Bureau of Economic Research believes that domestic production and employment are the leading conceptual measures of economic activity. An indication to know when the recession begins is based on when the economy reaches a peak of activity and ends when the again at the moment when the economy reached its lowest point. This is based on the payroll employment measure, real GDP declines, industrial production and wholesale and retail trade measure. In this regard December 2007 was identified as the peak month, because the subsequent decline in economic activity was big enough to categorized it as a recession (Determination of the December 2007 Peak in Economic Activity, 2008).

In general most people will agree that the recession was caused by the outbreak of the financial crisis. This was largely being driven by the housing downturn which started in 2006. This was also the result of the mortgage crisis that actually started in the early 2000s. In this period the housing and the credit bubbles started to build. Ackerman (2008) indicated that there an extraordinary amount of subprime mortgages were provided at the beginning of the new millennium. These mortgages were provided to individuals who actually were not creditworthy. Additionally Mizen (2008) added that these loans had a low interest rate at the beginning but after two years the interest rates increased. This leaved many people unable to meet their mortgage obligations. Thereafter the mortgages were bundled with mortgage-backed securities and resold in a complex set in the financial market. The advantages of this construction was that a mortgage was changed from a credit rating C to triple A. The triple A rating only occurred for government bonds,

(10)

10 but with a higher interest rate (Mizen, 2008). At the moment of the increase in interest rates many people could no longer meet their repayment obligations and were forces to sell their home. This had major implications for the housing market in the United States of America. Ackerman (2008) also indicated that due to the higher supply than the demand, housing prices went down sharply and the housing market came into a downward spiral. As a result, banks made heavy losses on their mortgages. This led to mutual distrust between banks and escalated in the financial crisis of 2008. Ackerman (2008) emphasized that the financial crisis led to substantial write-downs on financial assets. Banks provided no loans and therefore companies could no longer make investments which led to a recession. At the end of 2008, one of the largest Wall Street financial services provider Lehman Brothers was hit by a liquidity crisis. The company was not able to borrow the money that is needed to meet daily requirements. After Lehman Brothers collapsed, the financial market on which banks and corporates trusted to finance daily operations froze. Investors tried to sell what they could to raise money, as a result the value of investments of financial institutions feel even more. The breakdown of the financial system threatened not only the most robust institutions, but also had a negative impact on the broader economy (Taylor, 2009).

In response to this disaster, the government (USA) began in taking some urgent measures keeping credit flows in the economy. The Congress established the Troubled Asset Relief Program (TARP). This program allowed the US government to by the assets and equities from financial institutions. With the aim to bring the markets back from the collapse by injection of capital. They also established the stimulus program to increase public spending, that they would add to the expected payrolls and enhancing economic output (Paletta and Enrich, 2009). Yellen (2009) indicated that there were also other measures, such as the cut of the federal to zero (this is what banks are charging each other for overnight loans) by The Federal Open Market Committee, they also purchased large quantities of treasury debt (long-term) and the Federal Reserve also bought major parts of securities that were issued by the giant mortgage companies. The steps I have mentioned before demonstrated assistance to the housing sector to help push conforming mortgage interest rates to near-record low levels (Paletta and Enrich, 2009). Many people believes that these measures have helped to reach a stabilization in the financial market.

2.2 Earnings management

In the literature there are different concepts and names for earnings management. Mulford and Comiskey (2002) called it creative accounting and Schipper (1989) called it disclosure management. The difference between earnings management and manipulation is often made, because earnings management is within the limits of the law and earnings manipulation exceeds that limit. As I

(11)

11 mentioned previously in the introduction the definition of earnings management according to Healy and Wahlen (1999), is earnings management also defined in other ways, such as the definition of Mulford and Comiskey (2002):

“The active manipulation of earnings towards a predetermined target. That target may be set by management, an analyst’s forecast, or an amount that is consistent with a smoother and more sustainable earnings stream” (Mulford and Comiskey, 2002, pp. 15-16).

When analyzing the definitions it is clearly that the definition of Healy and Wahlen can be translated freely as misleading of financial reports. On the other hand, the definition of Mulford and Comiskey can be summarized as the purposeful manipulation of figures in order to achieve set targets. In general, it concerns the adjustment of financial information, which can be achieved by adjusting accruals. Accruals are items by which expenditure and revenue that belongs to an accounting period will be assigned to other periods. For example, products that are produced but not purchased yet or already paid are booked in the current year instead of the period of a later year. Over the past years the acceptable definitions appear not enough anymore. Graham, Harvey and Rajgopal (2005) indicate that more companies in addition to accrual-based earnings management also choose for real-based earnings management. The real-based earnings management means that there will not be accruals to adjust to make the firm perform better, but instead long-term investments will be postponed in order to achieve a goal in an accounting year or period. A definition which is more in line with the forms of earnings management I have described above would be the definition of Schipper (1989):

“A purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain” (Schipper, p. 92).

This study will focus on the accrual-based earnings management in which I will follow Bouwens et al. (2004), Leone and Van Horn (2005), Ballantine et al. (2007) and Verbruggen and Christiaens (2012). There are several reasons that accruals are allowed in financial statements, although sometimes abuse is made of. Dechow (1994) indicate that the result of organizations need to be correctly reflected. It is possible to use multiple indications or standards, but they are not all equally well. The cash flow is a suitable example, because the generated cash flow will often provide a distorted view of noise in it. This has to do with the moment of realization, which is not always gradual. Accruals take the noise away, therefore the results of organizations per year are less different, or so it does not differ in any event by incorporation of actual cash flow. Dechow’s (1994) investigation shows that at short intervals, accruals provide a better prediction of earnings per share than actual cash flow. As the interval which was measured widened the lead took off,

(12)

12 what was expected: at a greater interval with more cash flows, individual cash flows have less impact on the result. The second examined point was the comparison between the earnings on the basis of accruals and the cash flow realized in the case that a major fluctuation was in need for working capital, investments and financing policies. In this case the realized cash flow had big problems with the timing, where the accruals provided a good reflection of the organization. Dechow’s (1994) research has demonstrated that certain accruals played no role in the process. The long-term operational accruals would in the short-term even deteriorate the reflection of firms.

Healy and Wahlen (1999) indicates that it is difficult to prove earnings management. Firstly, the financial statement should be prepared, which is a difficult task because there are many ambiguities in the preparation. Thereafter at the founded differences with the reported financial statements motives have to been found, to decide that one can speak of a conscious act, or that errors were made during the preparation. Often a research is done on a conscious motive to trace the excessive outcomes for that motive. In many of the studies on earnings management, it uses the same models to prove the earnings management in individual situations. The most commonly used models for the literature that I have used in this research are the Jones model (1991) and the Modified Jones model (Dechow et al., 1995).

In contrast to what was customary before the Jones model, this model uses the proportion of discretionary accruals relative to the total accruals. Before the Jones model the subjective portion of one accrual were determined, or according to Jones that they wrongly used the total accruals, where the objective accruals are included. In this case changes in the economic situation would be part of the model, which is not desirable. In 1995 Dechow, Sloan and Sweeny created a modified Jones model and examined what model would be better. The motives examined in the literature are almost examined through accruals.

Verbruggen and Christiaens (2012) emphasize that although the research on earnings management in the non-profit sector is relatively scare in comparison to that concerning the profit sector, some authors have well documented its existence. Non-profit organizations are notified to adjust accounting numbers for different reasons, such as to improve the efficiency ratio, to avoid taxes and to avoid small losses.

2.3 Earnings management techniques

It is very clear that earnings management is nowadays popular within the accounting world. As I described in the previous paragraph that this way is used by managers to manage their earnings. There are a lot of researches on this subject, but it is still complicated. Managers reach their earnings by making certain accounting decisions or by operating decisions. This has to do with the flexibility

(13)

13 that managers have by making decisions. Managers make certain decisions by using specific earnings management techniques. There are different kind of earnings management techniques, but in this research I will only mention four known earnings management techniques. In his book Scott (2009) describes and explains the four known techniques regarding earnings management, which are: big bath accounting, income minimization, income maximization and income smoothing.

2.3.1 Big bath accounting

Organizations that have a year with bad results are using this kind of earnings management techniques, by trying to manipulate the profit and loss statement in a certain way. Managers are doing this by making the poor financial result in the current year more badly, so in the next year the results and the bonuses are increased and secured. This technique is mostly used in situation when there is a change in the management staff. There are some studies that found that this technique was used in a period in which there was a change in the management. Leone and Van Horn (2005) found the same in their study, were managers in non-profit hospitals used this technique. The reason for using this technique in a period of management change is that losses can be referred to the previous management, while future profits can be attributed to the performance of the new management. This is the idea behind taking a bath, which means beginning with a clean slate.

2.3.2 Income minimization

In contrast to other techniques, this earnings management technique is used to manage the earnings of an organization downwards. This technique is suitable for use in the non-profit sector, because non-profit organizations can use income minimization, for example to continue receiving grants from the government. In addition, there is also another regulatory motive, which is to keep the tax exempt status. This technique can be performed in different ways, for example by increasing the costs, such as costs for research and development or by changing certain accounting policies. While a known way to change some accounting policies is through the lifetime of assets by increasing the depreciation costs. In this way the organization can influence earnings by minimize it.

2.3.3 Income maximization

While managers who perform income minimization seek to manage their earnings downwards, managers who perform income maximization strive to manage their earnings upwards. Managers are doing this to increase the organizations profit. This earnings management technique is often

(14)

14 used in the for-profit sector, because in this sector most organizations are using a bonus system to reward their managers. In most cases this bonus is based on a fixed percentage of the profit, this is why managers try to manipulate the earnings by using this technique. This means that the higher the profits the higher the bonus for the managers. In addition, up warding the earnings means better performance, which leads to a good reputation for the organizations. In the Positive Accounting Theory the regulatory motivations try to avoid income maximization in non-profit organizations, because it is not the achievement of these organizations to have high profits.

2.3.4 Income smoothing

Managers can choose for income smoothing by skimming off the profits to make the losses less severe over a period of time. Managers can assume that if they can increase the profits through accruals at the current period, they can decrease the profits with the same amount in a subsequent period. They can also create accruals in years with a good result, this will lead to a decrease of profits in that year. When there are years with low results managers can decide to release these created accruals to show a better result in that year. In fact these organizations and managers are a kind of cookie jar reserve. Using unrealistic assumptions about assets and liability accruals is the achievement of the creation and releasing of these accruals. This earnings management technique direct to a stabilization in the future flow of income. This will lead to a stable impression of the organization towards her stakeholders. The stability is also a positive element for the reputation of the managers. This technique can also be used for regulatory motivations as I have mentioned before.

2.4 Non-profit organizations

The United States counted more than 1.9 million non-profit organizations in 2005. This number covers both the organizations registered with the IRS (about 1.4 million) and an estimation of the number of religious congregations that are not registered with the IRS. The number of non-profit organizations would increase further if small organizations (annual revenues less than $ 5,000) were to be counted, but these organizations are not required to register with the IRS (Blackwood, Wing and Pollak, 2008). From the period 1995 to 2005 revenues and assets for reporting non-profit organizations grew at least with 54%. While the GDP in the same period increased with about 35% after the correction for inflation. The non-profit sector in 2006 accounted for a number of 5.0% of the GDP. The non-profit sector employed about 12.9 million individuals, which is about 9.7% of the U.S. economy. This was greater than the number of people employed by the financial sector (Blackwood, Wing and Pollak, 2008). The independent sector is the collectively name that is given

(15)

15 to a major part of the non-profit organizations. This emphasizes the unique role that these organizations are playing in the society, distinct from government and business. The total number of independent sector non-profit organizations has been about doubled in the last 15 years (Hodgkinson, 1992).

Human service organizations, health care, education and most arts are forms that refers to public charities. From the total number of non-profit organizations that are registered with the IRS almost 63% were labeled as public charities. People mostly refer to public charities when the use the term “non-profit” (Blackwood, Wing and Pollak, 2008). The total revenue of the public charities that were registered in 2005 came almost to a number of $1.14 trillion. This was a little bit more than the total expense in the same year, which was $1.05 trillion. The estimation of the total assets was at $1.98 trillion. From the contributors, health and education are the dominant contributors. The total revenue, expenses and assets of these contributors is accounting for more than half of all public charities. 32.2% of all the public charities in 2005 is granted by the human service organizations, which is also the greatest number of reporting public charities. This number is followed by the education organizations with a percentage of 18.7 in the same year (Blackwood, Wing and Pollak, 2008). The sources of revenue for reporting public charities is also playing a major role, because it is important to know where the money is coming from. In 2005 about 50% of the revenue came from fees for the sale of goods and services from private sources. Approximately 30% of the total revenue for reporting public charities came from government sources, which means that the government is still playing an important role in public charity sector. The remaining sources came from private contributions, investment income and other income (Blackwood, Wing and Pollak, 2008).

2.5 Profit organizations

According to Verbruggen and Christiaens (2012) there are no motives or techniques that suggest that earnings management is limited to for-profit organizations. If we look at the literature we can clearly see that literature with regard to earning management is focused on earnings management in the profit sector. For example, Van Herck (2003) expects in his research less incentives for the use of earnings management in non-profit hospitals in comparison with the for-profit sector. Mostly this is needed to attract financings. It is also clear that in most cases for-profit organizations typically have different characteristics and operate in very different industries than non-profit organizations. There is also a difference between non-profit and for-profit organizations, because non-profits are supposed to spend all their available resources on current operations. There are different managerial incentives that can lead to earnings management in for-profit organizations,

(16)

16 which has to do with the self-interest of the managers, such as bonus compensation. In non-profit organizations there are other motives and incentives for using earnings management. I will describe this topic in more detail in the next paragraph.

2.6 The Positive Accounting Theory

Watts and Zimmerman (1986) created the Positive Accounting Theory. During the years this theory was used in many studies by many researchers, such as Scott (2009). This theory tries to explain and predict real world events and translate them to accounting transactions. This means that the Positive Accounting Theory investigates the interest and behavior of managers. Watts and Zimmerman (1986) created three important parts of this theory, which are translated into three hypotheses: bonus plan hypothesis, debt-equity hypothesis and political cost hypothesis. I will explain each of these hypotheses below.

2.6.1 Bonus plan hypothesis

Firms that have a bonus system which is used for the remuneration of managers will lead to manager’s incentive for accounting. In this case managers will tend to manipulate the accounting policies, methods and figures to show the accounting performance better than it should be. They do this to benefit themselves through the bonus system. For example, managers try to use different depreciation methods which shows lower profits at the beginning and higher profits at the end. This way will create a certain image in the financial statements of these firms and the same way the managers can benefit from the bonuses. This theory is only relevant with regard to the profit organizations in this research.

2.6.2 Debt-equity hypothesis

In this part of the theory managers are trying just like the bonus plan hypothesis to show a better performance by improving the liquidity position of the organization. They do this to pay the interest and principal of the debt they have built in the business. The intention of the managers is to show better profits in certain periods. In general the higher the debt-equity ratio the more managers will tend to make use of some accounting methods, policies and figures to increase the profit in their advantage. In this way managers exercising discretion by choosing accounting method I have mentioned before. This theory is only relevant with regard to the profit organizations in this research.

(17)

17 2.6.3 Political cost hypothesis

This part of the positive accounting theory is the opposite of the two parts, because in this part there is an assumption that organizations tend to show lower profits by using certain accounting methods, policies and figures to not attract the attention of the politics. The reason of this situation is that politicians will have an eye on high profit sectors. In this way to keep the profits lower it will also keep any attention from the public and the government, because in most cases they will implement stringer regulation on high earnings organizations. This part of the theory is mostly relevant for this research, because political costs are applicable. Non-profit organizations aim not to report maximum profits, but they want also not attract any focus from political institutions.

2.7 Motivations for the Positive Accounting Theory

Healy and Wahlen (1999) describe in their study three types of motivations for the use of earnings management under the Positive Accounting Theory. In their study they make a distinction between contracting motivations, capital market motivations and regulatory motivations. I will describe each of these motivations to create a clear picture about the motivations and therefore to make a link to this research.

2.7.1 Contracting motivations

Healy and Wahlen (1999) indicate that accounting data is playing an important role with regard to the contracts, because the accounting data helps to regulate and monitor the contracts between the organization and their stakeholders. The alignment of the incentives of management and external stakeholders are used for the management compensation contracts. Furthermore, lending contracts are created to limit actions of the managers. These actions were intended for actions that benefit shareholders of the organization at the expense of their creditors. As I mentioned in the paragraph about the Positive Accounting Theory Watts and Zimmerman (1978) made a suggestion that incentives for earnings management were created by the contracts I mentioned before. The cost to undo earnings management by compensation committees and creditors are high. With regard to the non-profit sector, managers of non-profit organizations according to Leone and Van Horn (1999) can reach a higher pay when replacing from a small non-profit hospital to a large non-profit hospital in their study. They found that bonuses are less applicable in the non-profit sector as I mentioned before.

(18)

18 2.7.2 Capital market motivations

In most cases it can clearly be seen that motives for putting earnings management into practice can mostly be found in a capital market. In this case, the base carrying is that shareholders, financial analysts and potential investors are using financial information from the financial statements to determine the present value of future cash flows and try to predict the future value of the organization. It is important that an organization is sufficient covered by analysts to benefit from more investment activity. That is why managers can play a huge role in this case, because they can influence the preparation of the financial statements. In this way the managers can give another (misleading) view of the financial situation or value of the organization, which is mostly higher than the situation or value in reality. That is why Healy and Wahlen (1999) find that higher values of the organization and a positive coverage by analysts will lead to more investment activities by new investors. This theory is only relevant with regard to the profit organizations in this research.

2.7.3 Regulatory motivations

The literature with regard to earnings management has examined the effects of two important forms of regulation, which are industry-specific regulation and anti-trust regulation. During the years accounting standard setters have shown that there is an interest in earnings management to bypass certain industry regulation (Healy and Wahlen, 1999). Watts and Zimmerman (1978) claimed that managers of organizations vulnerable to anti-trust research or any other adverse political consequences have certain incentives. These incentives lead to managing their earnings to be less profitable. It is most likely that managers in non-profit organizations have the same kind of incentives. Leone and Van Horn (2005) emphasized that accounting policies are demanded by the regulator for a certain type of industry and the organizations need also to comply with the regulation. In contrast, most studies on earnings management are giving a strong suggestion that regulatory considerations induce organizations to manage their earnings. Healy and Walhen (1999) could only find little evidence on whether this behavior is widespread or rare.

(19)

19

3 Hypotheses development

On the basis of the literature that I have mentioned before, I want to develop hypotheses to test. Most literature or theories provide a kind of prediction in which direction you have to develop the hypotheses. In time of crisis the use of earnings management can vary in non-profit organizations than in pre-crisis periods, which is the same as in the profit sector. The difference is in the direction of the level of earnings management. It is therefore important to test whether the level of earnings management will increase or decrease compared to the period before the time of crisis. Assuming that in time of crisis non-profit organizations are struggling, then there is no need to adjust their figures aiming to give a better image of the organization. It is just in favor of non-profit organizations to show poor results in difficult times. Jegers (2012) indicated that the fact of the manipulation of earnings assumes that certain earnings levels are better than other levels. In addition, there are some costs that were perceived by managers of non-profit organizations when disclosing profits or losses (Leone and Van Horn, 2015). Costs that are a result of reporting losses leads to reputation losses in the non-profit sector. In contrast, reporting profits can lead to a change in the tax exempt status of the non-profit organization as I mentioned in the literature review. Reporting profits can also affect the crow out donations and induce increased pressure by third-party payers. There are a lot of studies that suggest that periods of economic downturn are related to a higher level of earnings management. In contrast, there are several reasons to believe that periods of crisis are less advantageous to use earnings management than in expansion periods. Chia et al. (2007) indicated that during crisis year’s organizations are a subject to increased monitoring from different stakeholders, such as auditors and creditors. This could result in managers having less incentives to manage their earnings. Jenkins et al. (2009) gave another reason with regard to the behavior around earnings management. They suggest that the litigation risk is probably higher during period of crisis. Therefore, managers should respond in a natural way to this risk increase by limiting earnings management. In this manner crisis periods should be associated with less earnings management and thus leading to more conservative earnings, which means more timely. There is a study about the influence of litigation risk on conservatism by Huijgen and Lubberink (2005) in which they show how organizations report more conservative earnings in high legal liability regimes. Jegers (2012) and Krishnan, Yetman and Yetman (2006) did a study in the same direction. That is why it is important to know the status of non-profit organizations with regard to earnings management before and during crisis years. Based on this I want to test the following hypothesis:

Hypothesis 1: The level of accrual-accounting earnings management in U.S. non-profit organizations will decrease in crisis years relative to the years before the crisis.

(20)

20 As I mentioned before in time of crisis the use of earnings management can vary in profit organizations than in normal times, which is the same as in the non-profit sector. In order to get a broader picture of the impact of the financial crisis I want also to investigate what will happen with de level of earnings management in profit organizations. There are different studies that suggest that firms will use more earnings management during economic downturn, such as Ahmad-Zaluki et al. (2011). They indicated that the lower earnings should motivate managers to engage in income-increasing earnings management to compensate the decrease of the operational performance.

In contrast, there are other studies that believe that crises periods are less favorable to earnings management than normal periods. Ball and Shivakumar (2005) believed that lower levels of earnings management in recession periods can also be a consequent of an increased demand for conservative earnings. Crises have a transitory nature and that is why earnings reported in such periods are less persistent, which are less useful for predictions. In association with this assertion, different studies show that the value relevance of earnings varies across the business cycle (Brown et al., 2006). Jenkins et al. (2009) indicated that due to the increased uncertainty about future outcomes, it will motivate different market forces to demand more conservative earnings in crisis periods. Therefore it will discourage organizations to manipulate reported earnings. As I mentioned above Ahmad-Zaluki et al (2011) suggest another side of their study by indicating that under consideration in crisis periods, the market is more inclined to tolerate poor performance. Therefore, it leads to less incentives by firms to engage in earnings management activities.

In a the study of Bertomeu and Magee (2011) they created a model, in which they investigate the dynamics between accounting standards, the quality of financial reporting and the state of the economy. The results of their study show that financial reporting quality reaches its maximum when the economy is good, decreases as the economy conditions become less favorable, and increases again if the economy becomes recessionary. This means that the financial reporting quality is non-monotonic with the state of the economy. Even if their model addresses to regulator point of view, the model still assumes that the regulator passes the reporting quality level which is supported by a majority of the agents in the economy according to Wagenhofer (2011).

It is important to know the status of profit organizations with regard to earnings management before and during crisis years. Based on this I want to test the following hypothesis: Hypothesis 2: In profit organizations accrual-accounting earnings management decreases after the burst of the 2008 financial crisis relative to the pre-crisis years.

(21)

21

4 Research methodology

In this part of the paper I will describe the initial phase of the research methodology. I will focus on the various components of the research, such as the type of research, the different variables I will use for my research, which data (sample) I will use and over what period will be the research. The variables are focused on the measurement of the level of earnings management and the determination of the financial crisis. For this research I will collect quantitative data. The data that will be used is derived from United States databases.

4.1 Model to detect earnings management

First I have to measure the earnings management, which is the level of earnings management. As I mentioned before in this paper, in this research I will focus on the accrual-based earnings management. Furthermore, I will follow Bouwens et al. (2004), Leone and Van Horn (2005), Ballantine et al. (2007) and Verbruggen and Christiaens (2012) in this research. I will measure the level of earnings management on the basis of the modified Jones model (Dechow et al., 1995). This model is widely used in different studies. I think this model seems appropriate for this research. The Jones model (1991) had some weaknesses, because it did not take the account balance receivables under consideration. This was the reason for Dechow et al. (1995) to create a modified Jones model. In this modified model they solved this weaknesses of the normal Jones model (1991). The modified Jones model is more capable to detect earnings management in comparison with the normal Jones model. That is why I will focus on this model and explain the calculation of the total accruals and discretionary accruals.

 The first step is the calculation of the total accruals, which is as follows: 𝑇𝐴𝐶𝐶_𝑡 (Total accruals) = NI (net income) – CFO (operating cash flow)

 The second step is the estimation of the modified Jones model, which is as follow:

𝑇𝐴𝐶𝐶𝑡 𝐴_𝑡−1 = 𝛼1 1 𝐴_𝑡−1+ 𝛼2 (∆𝑅𝐸𝑉𝑡− ∆𝑅𝐸𝐶𝑡 ) 𝐴_𝑡−1 + 𝛼3 𝑃𝑃𝐸𝑡 𝐴_𝑡−1 + 𝜀𝑡

𝑇𝐴𝐶𝐶𝑡 = Total accruals in year 𝑡 divided by total assets in year 𝑡 − 1

∆𝑅𝐸𝑉_𝑡 = Revenues in year 𝑡 less revenues in year 𝑡 − 1

∆𝑅𝐸𝐶_𝑡 = Delta revenues in year 𝑡 less delta net receivables in year 𝑡 − 1

𝑃𝑃𝐸_𝑡 = Gross property plant and equipment in year 𝑡

𝐴𝑡−1 = Total assets in year 𝑡 − 1

𝛼1,𝛼2, and 𝛼3 = Parameters to be estimated, namely alphas

(22)

22

 The third step is to calculate the discretionary accruals, which is as follow:

𝐷𝐴𝐶𝐶_𝑡 = 𝑇𝐴𝐶𝐶_𝑡− 𝑁𝐷𝐴𝐶𝐶_𝑡

The non-discretionary accruals can be calculated with the following formula:

𝑁𝐷𝐴𝐶𝐶_𝑡 𝐴𝑡−1 = 𝛼̂1 1 𝐴𝑡−1+ 𝛼̂2 (∆𝑅𝐸𝑉_𝑡− ∆𝑅𝐸𝐶_𝑡) 𝐴𝑡−1 + 𝛼̂3 𝑃𝑃𝐸_𝑡 𝐴𝑡−1

𝑁𝐷𝐴𝐶𝐶𝑡 = Non-discretionary accruals divided by total assets in year 𝑡 − 1 The use of (∆𝑅𝐸𝑉𝑡− ∆𝑅𝐸𝐶𝑡 ) makes it possible to manage the non-discretionary part of the accruals. The changes in current assets and current liabilities give an explanation of the raise or reducing of the business activities which are associated and that leads to higher or lower earnings. This implies that more business activities guide to a higher earnings, which means that it will lead to higher non-discretionary accruals. This means that a positive coefficient for 𝛼2 (see second step of model) is expected. In addition, the amortizations costs are also playing an important role, because they are another part of non-discretionary accruals and can be managed through the use of 𝑃𝑃𝐸𝑡 . The amortizations costs will lead to a lower earnings, which will lead to a negative coefficient for 𝛼3 (see second step of model).

4.2 Regression model

When the discretionary accruals are determined for all organizations (profit and non-profit) in the sample, the relationship between the discretionary accruals and the crisis can be explained. In order to know what impact the financial crisis had on the earnings management in profit and non-profit organizations I have to compare between two periods. These two periods are distinguished in this research in pre-crisis period and during crisis period. The intention is to see if there are differences between the levels of earnings management in these periods. Since I will do this for two types of organizations I will be also able to compare between the coefficients of the regression for profit and non-profit organizations. In this manner I will be capable in giving a broader picture of the impact of the financial crisis of 2008-2009 on two different types of organizations.

To compare between the two periods I mentioned before I have to create dummies. One dummy will be for the crisis period and one dummy for the period during the crisis. The pre-crisis period is about the years 2005 to 2007 and the period during the pre-crisis is about the years 2008 and 2009. First I have to determine the earnings management level for both types of organizations as I mentioned before in paragraph 4.1. After that I have to examine the difference, which means that I have to perform a multivariate analysis. In this analysis I will regress the discretionary accruals per period based on a dummy variable, which suggest pre-crisis and during crisis and some control

(23)

23 variables for both regressions. These control variables are for the control of the differences between pre-crisis and during crisis periods. In order to run the regression for both types of organizations, which are based on the hypotheses of this research, I created a regression model for my research:

 Regression for non-profit organizations, which is as follow:

𝐷𝐴𝐶𝐶_𝑖𝑡= 𝛼0 + 𝛽1 Crisisit + 𝛽2 Sizeit + 𝛽3 Growthit + 𝛽4 Leverageit + 𝜀𝑡

 Regression for profit organizations, which is as follow:

𝐷𝐴𝐶𝐶𝑖𝑡= 𝛼0 + 𝛽1 Crisisit + 𝛽2 Sizeit + 𝛽3 Growthit + 𝛽4 Leverageit + 𝜀𝑡 𝐷𝐴𝐶𝐶_𝑖𝑡 =Discretionary Accruals for firm I and year t

Crisisit = Dummy variable which has a value of 1 if it meets the criteria of the period during the crisis and a value of 0 if it meets the criteria of the pre-crisis period Sizeit = Total assets at the end of year t

Growthit = Mutations by change in sales for organization I between t-1 and t

Leverageit = Leverage for organization I at the end of year t, calculated total debt in relation to total assets.

In addition to the independent dummy variable that shows whether the company in question is in a pre-crisis or crisis period, the regression models include three control variables to reduce the influence of results by other factors of discretionary accruals.

The size of the organization plays an important role in the earnings management study. Larger organizations prefer accounting choices that are based on income decrease, while smaller organizations don’t have a preference. This means that smaller organizations use both income increasing and income decreasing on earnings management practices (Sun and Rath, 2008). According to Myers et al. (2003), larger organizations have more opportunities to spread their profits than smaller organizations, which makes larger organizations have more stable accruals.

There are a lot of studies that found a relation between the indications of an organizations growth and the activities regarding earnings management. According to Sun and Rath (2008) organizations with a high growth are significantly related with a high level of discretionary accruals.

In organizations that uses debt to finance its operations are also associations between the level of debt and earnings management. Organizations that have financial problems are tend to use some accounting policies to increase income in order to remove the constraints that are affected by the high debt level (Dichev and Skinner, 2002). This is based on the debt-equity hypothesis

(24)

24 under the Positive Accounting Theory that I have mentioned in the literature review (Watts and Zimmerman, 1990).

4.3 Data and Sample Selection

During my research the database that I have used to collect data for the calculation of the level of earnings management for non-profit organizations is the National Center for Charitable Statistics (NCCS). I have used historical data that is already available on this database. I have used the financial statements of various non-profit organizations to collect data for the different variables. The data that I needed to determine the earnings management level for non-profit organizations was very difficult to collect, that was the reason to use hand collected data in this research. I made use of the 990 forms of the RSI (Internal revenue Service) from the United States. The National Center for Charitable Statistics (NCCS) is the national storage place of data on the non-profit sector in the United States. This database cooperate closely with the Internal Revenue Service (IRS) and other government agencies, private sector service organizations, and the scientific community. NCCS develop uniform standards for reporting on the activities of charitable organizations. The other part of the data I have used during this research is data of profit organizations. This data is collected from the Orbis database through WRDS (Wharton Research Data Services). The data for profit organizations is excluded from financial organizations.

I will follow Verbruggen and Christiaens (2012), by using data from organizations in the non-profit sector at large. This means that I will not only focus on earnings management within a specific kind of organizations. The sample consists of non-profit organizations from educational institutions, healthcare institutions, environment organizations, human services and other public charities. In addition, the data for profit organizations consists of both listed and unlisted organizations. The data is based on one country, which is the United States. The testing of hypotheses will be performed for the period 2005-2009 as a whole.

My research is based on the longitudinal model. By using this model the data include observations of several phenomena produced over several periods for the same organizations. The data set consisted of a total of 6.953 financial year observation, which is divided in 320 financial year observation for non-profit organizations and 6633 financial year observation for profit-organizations. To standardize this data I had to use winsorization and log transformation, because of some extreme values. This means that the original data was a little bit larger. The sample size will be 64 non-profit organizations and 1326 profit organizations

(25)

25

5 Empirical results

In this chapter of the paper I will explain the statistical test for earnings management. The descriptive statistics, the correlation matrix and the hypothesis will be described. I will provide the interpretation of the results about the regression with regard to earnings management, profit and non-profit organizations and crisis periods. The results will be divided over the two hypotheses of this research. All hypotheses are stated in the alternative form.

5.1 Hypothesis 1

H1: The level of accrual-accounting earnings management in U.S. non-profit organizations will decrease in crisis years relative to the years before the crisis.

5.1.1 Descriptive statistics

Descriptive statistics non-profit data Variable Pre-crisis

or during crisis

Mean Median Std. Deviation

Minimum Maximum Observation

DA_Win 0 0,1159 0,1908 0,29216 -1,49 0,48 160 1 -0,0383 0,1009 0,39700 -1,49 0,48 160 Log_Size 0 4,6880 4,6971 0,61315 2,83 5,95 160 1 4,7587 4,7224 0,56387 3,18 5,94 160 Growth_Win 0 5015,93 1452,00 16720,625 -27613 118893 160 1 -2519,86 501,50 28872,881 -130665 11892 160 Leverage_Win 0 0,30472 0,16979 0,445797 0,000 2,404 160 1 0,27527 0,19045 0,352359 0,000 2,404 160 Table 1: Descriptive statistics

Definition of variables:

DA_Win = Discretionary accruals (winsorized)

Log_Size = Size control variable (log transformation)

Growth_Win = Growth control variable (winsorized)

Leverage_Win = Leverage control variable (winsorized)

0 = pre-crisis period

1 = during crisis period

The important statistics about the non-profit data are described in table 1 above. The first statistic to deal with is the mean, because the mean is the most common measure of central tendency. The mean of the discretionary accruals (DA_Win) is as (0,1159) shown in the table for the pre-crisis

(26)

26 period. During the crisis period, the mean changed to a negative mean (-0,0383), which just means that the use of discretionary accruals by non-profit organizations decreased in the crisis period compared to the pre-crisis period in this dataset. Leverage control variable shows a mean of (0,30472) over 30% in the pre-crisis period. This implies that a few non-profit organizations have a high debt ratio. During de crisis period it is shown that the mean of the leverage control variable has decreased slightly. This can be explained by the strict lending conditions of the banks during the crisis periods. The standard deviation for the size control variable is higher than 50 percent (0,50), which means that there are more deviations from the average observations. It is shown in the table that the difference between the median and the mean are slightly greater in some variables. This has to do with the unusual values, because they affect the median less than they affect the mean. The growth control variable shows the mutations by change in sales for organizations, so that is why great differences are shown between the minimum and the maximum values of this variable. The table shows a minimum value of (-27613) -27,6 million and a maximum value of (118893) 118,8 million in the pre-crisis period. During the crisis period it is shown that both the minimum (-130665) value and the maximum (11892) value decrease due to the decline in sales during the crisis. The number of observations for the crisis period and during the crisis period is equal in this case (160). This is due to the winsorizing of unusual values on the side of the pre-crisis period, which therefore has led to a number of observations of 160 for the crisis and 160 during the crisis.

5.1.2 Regression

The first step before analyzing the regression results I will explain the relationship between the variables I have used in this regression for the non-profit organizations (In appendix 1 the histogram of the dependent variable DA_WIN is displayed). This relationship is showed as Pearson’s Correlation Coefficient. This is a measure of the strength of the association between the two variables. The correlations are showed in a table on the next page.

(27)

27

Correlations non-profit data

Variable DA_WIN Pre-crisis or during crisis dummy

Log_Size Growth_WIN Leverage_ WIN

Pearson Correlation DA_WIN 1,000 -0,217 0,405 -0,187 -0,268 Pre-crisis or during crisis dummy -0,217 1,000 0,060 -0,158 -0,037 Log_Size 0,405 0,060 1,000 0,002 -0,350 Growth_WIN -0,187 -0,158 0,002 1,000 -0,079 Leverage_WIN -0,268 -0,037 -0,350 -0,079 1,000 Table 2: Correlation matrix (Pearson’s R)

Based on the correlation matrix, the following correlations are shown. If the discretionary accruals (DA_WIN) changes with 1%, pre-crisis or during crisis variable changes with -0,217. In this manner the Pearson’s R is negative, which implies that as one variable increases in value, the second variable decreases in value. If the discretionary accruals (DA_WIN) changes with 1%, the size (Log_Size) changes with 0,405. In this manner the Pearson’s R is positive, which means that as one variable increases in value, the second variable also increases in value. This implies that when the level of discretionary accruals increases, the size control variable also increases. The conclusions is that the correlation is positive, which means that there is a positive relationship. Furthermore, the correlation matrix shows that when the discretionary accruals changes with 1%, the (Growth_WIN) growth control variable changes with -0,187 and the (Leverage_WIN) leverage control variable changes with -0,268. In this manner both Pearson’s R are negative, which means that as one variable increases in value, the second variable decreases in value. When discretionary accruals increases, the growth and leverage control variables decreases. This indicates that when non-profit organizations uses more discretionary accruals, at the same time they have lower mutations by change in sales and they have a lower debt ratio. There are also some correlation coefficients between the control variables. The table shows that when the size control variable changes with 1%, the growth control variable changes with 0,002 and the leverage control variable changes with -0,350. In this manner there is a positive relationship and a negative relationship between the variables. The correlation of the growth variable is not strong because it is very close to the 0. The leverage control variable shows a negative relationship, because when the size control variable increases, the leverage control variable decreases.

(28)

28

Correlations

Variable DA_WIN Pre-crisis or during crisis

dummy

Log_Size Growth_WIN Leverage_ WIN

Sig. (1-tailed) DA_WIN . 0,000 0,000 0,000 0,000 Pre-crisis or during crisis dummy 0,000 . 0,142 0,002 0,256 Log_Size 0,000 0,142 . 0,485 0,000 Growth_WIN 0,000 0,002 0,485 . 0,079 Leverage_WIN 0,000 0,256 0,000 0,079 . Table 3: Correlation matrix (Sig. value)

Based on the table above it will show if there is a statistically significant correlation between the variables. There are statistically significant correlations between discretionary accruals variable and size control variable (0,000), growth control variable (0,000) and leverage control variable (0,000), because all the Sig. values are less than 0,05.

The statistical outcome of the regression on the basis of the modified Jones model for non-profit organizations data is:

Model Summaryb

Model R R Square Adjusted R Square Std, Error of the estimate

1 0,548a _0,300 _0,291 _0,30005

a. Predictors: (Constant), Leverage_WIN, Pre-crisis or during crisis dummy, Growth_WIN, Log_Size b. Dependent Variable: DA_WIN

Table 4: Model summary

In the model summary it is shown that the R2 _{is 0,300, this means the proportion of variance in the}

dependent variable (discretionary accruals) which can be explained by the independent variables (pre-crisis or during crisis dummy, size control variable, growth control variable and leverage control variable). This indicates an overall measure of the strength of association and does not reflect the extent to which any particular independent variable is associated with the dependent variable. In this manner the R2 _{indicates a percentage of 30%, which is not a really strong}

relationship between the dependent and the independent variables. The adjusted R2 _{is shown as}

(29)

29 explained by only the independent variables that actually affect the dependent variable (discretionary accruals). In this manner the R2 _{and the adjusted R}2 _{are almost equal.}

ANOVAa

Model Sum of

squares

df Mean Square F Sig

1 Regression 12,175 4 3,044 33,808 0,000b

Residual 28,359 315 0,090 Total 40,534 319

a. Dependent Variable: DA_WIN

b. Predictors: Predictors: (Constant), Leverage_WIN, Pre-crisis or during crisis dummy, Growth_WIN, Log_Size

Table 5: ANOVA

The table above shows that the p-value (0,000) of the f-statistic (33,808) is extremely small, which means that it is smaller than 0,001. The F-statistic is only one measure of significance in an F-test. This is the reason why I will also use the p-value in this case, because even if the p-value is less than 0,05 I have to study the individual p-values to find out which of the individual variables are statistically significant. I will do this by using the next table of coefficients.

Coefficientsa Model Unstandardized Coefficients Standardized Coefficients _t _Sig. Collinearity Statistics

B Std. Error Beta Tolerance VIF 1 (Constant) -0,850 0,150 -5,656 0,000 Pre-crisis or during crisis dummy -0,202 0,034 -0,284 -5,928 0,000 0,970 1,030 Log_Size 0,220 0,030 0,363 7,209 0,000 0,875 1,143 Growth_WIN -3,675E-6 0,000 -0,246 -5,133 0,000 0,967 1,034 Leverage_WIN -0,151 0,045 -0,171 -3,376 0,001 0,870 1,149

a. Dependent Variable: DA_WIN Table 6: Parameters estimates (coefficients)

The table shows that the coefficient for pre-crisis or during crisis dummy is -0,202, which means that for every unit increase in the pre-crisis or during crisis dummy a -0,202 unit decrease in the

(30)

30 discretionary accruals is expected, holding all other variables constant. Because pre-crisis or during crisis dummy is coded 0/1 (0=pre-crisis, 1=during crisis), the interpretation is: for during crisis, the predicted discretionary accruals would be 2 points lower than for pre-crisis. The coefficient for size control variable is 0,220, which means that for every unit increase in the size control variable, 0,220 increase in the discretionary accruals is expected, holding all other variables constant. The coefficient for growth control variable is -3,675, which means that for every unit increase in growth control variable a -3,675 unit decrease in the discretionary accruals is expected, holding all other variables constant. The coefficient for leverage control variable is -0,151, which means that for every unit increase in leverage control variable a -0,151 unit decrease in the discretionary accruals is expected, holding all other variables constant.

To test if the given coefficients are significantly different from zero I will have to check the t-statistics and their p-values in the table above. The coefficient for pre-crisis or during crisis dummy (-0,202) is significantly different from the 0 because its p-value is 0,000, which is smaller than 0,05. The coefficient for size control variable (0,220) is significantly different from 0 because its p-value is 0,000, which is smaller than 0,05. The coefficient for growth control variable (-3,675) is significantly different from 0 because its p-value is 0,000, which is smaller than 0,05. The coefficient for leverage control variable (-0,151) is significantly different from 0 because its p-value is 0,001, which is smaller than 0,05. Based on the significance of the p-values it can be concluded that the level of earnings management in U.S. non-profit organizations decreases during crisis periods relative to pre-crisis periods. Therefore hypothesis 1 can be accepted.

5.1.3 Robustness testing

In order to do a robustness check I want to examine how certain core regression coefficient estimates behave when the regression specification is modified by adding or removing some ingredients. To find out if the estimate is different from the results of the plausible model. In order to do an appropriate robustness check I have chosen the total assets as variable, because the total assets variable is used in the regression model I have created to determine earnings management. This variable is used as a control variable which indicates the size of the organizations that are used in the sample.

(31)

31 Percentiles Percentiles 5 10 25 50 75 90 95 Weighted Average (Definition 1) Assets Total 3357,90 10494,80 24193,75 48827,00 113157,00 320147,10 439627,20

Tukey’s Hinges Assets Total

24268,50 48827,00 112951,00

Table 7: Robustness testing (Percentiles)

In order to determine if the coefficients change after the regression, I will use Tukey's Hinges by making use of the total assets variable. I will take only the lower hinge (25th_{percentile) till the upper}

hinge (75th_{percentile) in consideration, which indicates the values higher than (24268,50) 24,2}

million and lower than (112951,00) 113 million. In this case I only took into account the medium-sized organizations in my sample. After the regression the results remained the same, which means that the significance of the variables did not change.

The impact of the financial crisis in 2008-2009 on accrual accounting earnings quality in U.S. profit and non-profit organizations