Incentive pay and its influence on earnings management in American nonprofits

Incentive pay and its influence on earnings management in

American Nonprofits

Name: Floor de Ruiter Student number: 11121661 Thesis supervisor: V. O’connell Date: 24 June 2017

Word count (12.000 minimum): 12622, 0

MSc Accountancy & Control, specialization Control

Statement of Originality

This document is written by student Floor de Ruiter 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.

Abstract

This study focuses on the use of incentive pay in nonprofit organizations (NPOs) and its effect on earnings management and firm performance. IRS 990 forms from the United States tax authority were used to test if the use of variable compensation in annual bonus contracts had an effect on discretionary accruals and earnings. This research is the first to provide insights into the use of bonuses based on earnings, equity or revenue within the nonprofit sector. A significant relationship was found between the use of incentive pay and earnings, which imply that the use of incentives within NPOs can be beneficial. However, in contrast to for-profit organizations, the use of incentives does not seem to have an increasing effect on earnings management.

Table of Contents ! 1! Introduction ... 1! 1.1! Background ... 1! 1.2! Research Question ... 2! 1.3! Results... 2! 1.4! Contribution ... 3! 1.5! Structure ... 4!

2! Literature review and hypothesis development ... 5!

2.1! Earnings Management ... 5!

2.1.1! Accrual-based Earnings Management ... 5!

2.1.2! Incentives for Earnings Management ... 6!

2.1.3! Reducing Earnings Management ... 6!

2.2! Earnings Management in nonprofit organizations ... 8!

2.3! Incentive contracts ... 10!

2.4! Incentive contracts in nonprofit organizations ... 12!

2.5! Incentive contracts and firm performance ... 14!

2.6! Hypothesis development ... 15!

3! Research Methodology ... 17!

3.1! Data & Sample Description ... 17!

3.2! Method ... 19!

3.2.1! Accrual-based Earnings Management in the profit sector ... 19!

3.2.2! Accrual-based Earnings Management in the nonprofit sector ... 20!

3.2.3! Jones model ... 20!

3.2.4! Regression Models ... 21!

4! Results ... 26!

4.1.1! Discretionary Accruals ... 26!

4.1.2! Incentive compensation ... 26!

4.2! Regression results ... 28!

5! Summary and Conclusion ... 31!

5.1! Summary ... 31!

5.2! Conclusion ... 32!

5.3! Limitations ... 33!

References ... 34!

1! Introduction

Recently, nonprofit organizations (NPOs) are under increased attention of researchers. In this growing sector, accountability becomes more and more important. These days, NPOs receive an increasing amount of state funding, while total funds raised is still growing too (Jobome, 2006). NPOs have to deal with a lot of stakeholders, like the government, donors, public and clients, and thus increased pressure exists to show their impact on social problems (Boateng et. al, 2016). Moreover, NPOs need to show their social impact, and how this impact is achieved. This is why more researchers claim that NPOs should act more like profit organizations (Jobome, 2006). But whereas there is a lot known about profit organizations, literature about NPOs in certain areas is scarce. For example, there is extensive literature available about how managers of profit organizations try to influence their bonus pay by managing earnings (Guidry et. al, 1999). A commonly used proxy for earnings management are discretionary accruals. Discretionary accruals are part of total accruals and can be easily manipulated by managers, since they don’t represent real obligations or income. Managers observe earnings before discretionary accruals and the assumption made in literature (Healy, 1985) is that managers observe earnings before discretionary accruals and then use these accruals to create income increasing (decreasing) decisions if the earnings are below (above) the lower (upper) bound for earning a bonus (Guidry et. al, 1999).

However, although NPOs should act more like profit organizations, they differ from for-profit organizations in many ways. The main goal for-for-profit organizations is to make a for-profit and distribute these profits to shareholders or re-invest in the business. NPOs exists because of a philanthropic mission (Verbruggen and Christiaens, 2012) and distribute programs and services. This difference affects the behavior of NPOs in several ways: where for-profit organizations manage earnings mainly upwards, NPOs tend to manage their earnings towards zero (Leone and van Horn, 2005; Verbruggen and Christiaens, 2012). On the one hand, NPOs need to show that they are able to efficiently run the organization, which means that they tend to avoid showing losses. On the other hand, a high profit can be undesirable since it can be harder to attract funds from the government or donors.

Jobome (2006) claims that NPOs have come under increased attention and should act more like for-profit organizations, more skilled workers are needed. To be able to attract these skilled workers, salaries should be in line with the profit sector and thus incentive pay could be

desirable. It will be interesting to analyze how the use of incentives can affect the ‘drive towards zero’ behavior of NPOs, since traditional theory predicts that managers act in line with incentives, which usually will mean that they manage earnings upwards (Healy, 1985).

1.2! Research Question

The objective of this research is to learn more about the use of incentives within NPOs and how these incentives affects the existence of earnings management.

This leads to the following research question:

To what extent does the use of variable components in executive compensation contracts influence the existence of AEM within non- profit organizations?

The research question is schematically shown in figure 1.

Figure 1: Research Question

Additionally, I am interested if the use of incentives enhances the performance of NPOs. In recent years, several researchers concluded that the use of solely financial performance measures within for-profit organizations doesn’t enhances performance, since these metrics focus too much on short term gain. However, the use of incentives in NPO’s isn’t that developed yet, so here other mechanisms could exist.

I predict that NPOs tends to manage earnings towards zero, but when incentives are in place, managers will react in line with these incentives. Thus, the different types of incentives should give managers an incentive to manage earnings upwards, which results in earnings management. Furthermore, according to traditional agency theory, incentives are set to make sure the agent acts in the best interest of the firm. So, the use of incentives in NPOs should enhance the

Variable compensation for Directors in NPOs

Accrual-based Earnings Management

discretionary accruals as part of the total accruals. Within the profit research, discretionary accruals are seen as a proxy for accrual- based earnings management: this type of accruals is the easiest to manipulate for managers. However, in the recent years, this model is also increasingly used within the NPO research stream (Leone and van Horn, 2005; Verbruggen and Christiaens, 2012), and has been shown to be a reliable indicator of earnings management. After that, different types of tests were performed to see if different types of incentives (based on equity, revenue or earnings) influences the amount of discretionary accruals and earnings.

First, I provided more evidence for the theory that NPOs manage earnings towards zero, whereas for-profit firms manage earnings upwards. Results are similar to previous research (Leone and van Horn, 2005; Verbruggen and Christiaens, 2012) and shows that unmanaged earnings are significant negatively correlated to discretionary accruals.

After that, three incentive variables, equity-based, revenue-based and earnings-based incentives were regressed with discretionary accruals to test if the magnitude of discretionary accruals differ for NPOs using one of these incentives. No significant relationships were found between the use of these incentives and the magnitude of discretionary accruals within NPOs. In the same sense, the absolute amount of bonuses was regressed with discretionary accruals, to see how this influences the existence of discretionary accruals. Again, no significant relationship was found between the amount of bonuses paid and discretionary accruals.

Finally, I find that the absolute amount of bonuses is significantly related to earnings. This is line with my third hypothesis, which means that NPOs with incentives in place are performing better. However, overall results are mixed, and clearly more research is needed.

1.4! Contribution

First of of all, little is known about earnings management in nonprofit organizations (Verbruggen and Christianes, 2012). While widespread literature about EM in the profit sector is available, the literature about EM in the nonprofit sector is scarce. However, this type of organizations has become more and more important in our economies (Chenhall, 2003). Until now, several researchers showed that nonprofit organizations manage earnings towards zero, but this research focused mainly on Belgian nonprofit organizations or specific sectors within the nonprofit industry (Jegens, 2013; Verbruggen & Christianes, 2012; Leone and Van Horn, 2005). In this research, a large dataset of NPOs in the United States was used and provided more evidence in this field.

Second, even less is known about the use of variable components in executive compensation contracts in NPOs and its effect on discretionary accruals or earnings. Krishnan et al. (2006) was able to find a positive association between incentives and expense misreporting in American NPOs. However, they mainly focused on program spending ratios. Hallock (2002) also finds a positive relationship between program spending ratios and higher pay levels. To the best of my knowledge, my study is the first that looks at NPOs in all types of sectors, and the direct link between different types of incentives and its effect on discretionary accruals. Brickley and van Horn (2002) are, until now, to my knowledge, the only one that focused on incentive setting and its effect on performance of NPOs. However, their research mainly focused on large nonprofit hospitals whereas I provide research evidence for NPOs in all types of sectors. Until now, the use of incentives within the NPO sector is scarce. However, results imply that the use of incentives can benefit NPOs and can thus be desirable.

1.5! Structure

The structure of this paper is as follows. In the next chapter relevant literature will be discussed which results in the development of the hypotheses. I explained in chapter three how data was gathered and which regression models were used to test the hypothesis. In chapter four, the results of all analyses are discussed. Finally, a summary and conclusion are presented in chapter five.

2! Literature review and hypothesis development

As EM is a broad topic, and a lot of research has been done in this field, a lot of definitions of EM exist. According to Schipper (1989, p.92), EM is a ‘purposeful intervention in the external financial reporting process with the intent to obtaining some private gain.’ However, the most common and often used definition comes from Healy and Wahlen (1999, p.368), and is the following one:

Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to 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.

Executives can manipulate earnings in two ways: by manipulating accruals or by actually changing the business practices, with the objective to meet certain earnings criteria (Roychowdhury, 2006). Manipulating accruals, which is, Accrual-based EM (AEM), does not have any effect on the organizational cash flows. A way to apply AEM is to change the valuation method of assets in such a way that the depreciation costs will be transferred to a later period. Real earnings management (REM) occurs when executives make decisions that affect business practices, with the goal of report a certain amount of earnings. For example, Dechow and Sloan (1991) report that CEOs that are in their final year, are more likely to cut R&D expenditures. 2.1.1! Accrual-based Earnings Management

In summary, earnings are the sum of cash in- and outflows, but then matched to the appropriate time frame. If not, cash-flows are not an appropriate measure of firm performance, since some incoming cash flows can be the result of activities of last year, or other years than the reporting year, and give therefore distorted information about the profitability of a company in a given year. Thus, companies use accruals to assign in- and outgoing cash flows to the right time frame. However, executives can abuse these accruals to smooth earnings, since management has some discretion in assigning accruals (Dechow, 1994): this is called Accrual-based Earnings Management. The research of Graham et al. (2005) indicates that CFOs prefer a smooth earnings path above a constant cash flow. Motivation for this preference is that smooth earnings are more important for investors than constant cash flows in assessing the risk associated with a

firm. Another explanation is that stable earnings makes it easier for investors to predict future earnings, which will lead to higher stock prices.

2.1.2! Incentives for Earnings Management

Healy and Wahlen (1999) outline three reasons why managers engage in EM: (1) capital market expectations and valuation; (2) contracts written in terms of accounting numbers; and (3) antitrust or other government regulation.

First, investors and financial analysts are using accounting information to value stocks. When a manager can report higher earnings, the stock price performance will be influenced in a positive way: this will result in an increased value of the company. Second, variable compensation contracts can create an incentive for managers to engage in EM. To align managers’ incentives with that of the firms’ owners, accounting data are used to make sure the managers behave in the organization best interest. This creates an incentive for managers to influence these accounting data in a positive way. Third, different types of regulation can be a reason to engage in EM. For example, banking regulations require that banks own a minimal amount of assets to cover their loans. This minimum is written in accounting numbers and thus creates incentives to manage the income statement and balance sheet numbers in favor to regulators (Healy and Wahlen, 1999). Graham et al. (2005) surveyed and interviewed 400 executives, and found out that CFOs believes that earnings, are the most important key factor considered by outsiders: cash flows are considered less important. Executives often need to make a choice between delivering (target) earnings and long-term value. Most executives think they make the right choice when scarifying long-term value for short-term gains, since not delivering targets can be costly for the firm. This results in some form of EM.

2.1.3! Reducing Earnings Management

Although the fraud of Enron and Ahold occurred a while ago, the topic still remains relevant or other companies, like Mitra1 _{or Pon}2_{. EM is seen as something unethical and thus, recent} researchers try to find factors that can mitigate EM within organizations.

association between board independence, audit committee independence and abnormal accruals, which is a form of EM. Park and Shin (2002) look at Canadian firms, and find that directors who are officers of financial intermediaries, unlike ordinary outside directors, are helping the board to reduce EM. The reason for this is that they have more financial experience than an ordinary outside director. Supporting this, Xie et al. (2003) shows that firms that have board and audit committee members which have corporate and financial experience are less inclined to engage in EM. Also, board and audit committee meeting frequency seems to influence the magnitude of abnormal accruals within for-profit firms.

Another stream of research, looks at factors that influence social norms to behave ethically. Previous research showed that social norms are able to influence human behavior. For example, McGuire et al. (2012) provides evidence that firms that are headquartered in areas with strong religious social norms, experience less financial reporting irregularities. They conclude that religious social norms can act as an alternative monitoring mechanisms over corporate financial reporting, especially when external monitoring is low. Similarly, Du et al. (2015), shows, using a sample of 11,357 firm-year observations from the Chinese stock market, a negative association between religion and earnings management. Besides religion, Corporate Social Responsibility (CSR) is a factor that can mitigate Earnings management as well. Measuring CSR on seven different key areas (Community, Corporate Governance, Diversity, Employee Relations, Human Rights Environment and Product), Hong and Andersen (2011) shows that companies that score high on these seven key areas are less inclined to engage in EM. In the same spirit, Shafer (2015) conducted a survey among professional accountants, and findings suggest that ethical climate can influence accounting professionals to rationalize earnings management decisions by adjusting their attitude toward the importance of corporate ethics and social responsibility. In this sense, a strong ethical climate could work as a constraint for EM.

2.2! Earnings Management in nonprofit organizations

It can be argued that NPOs are less inclined to engage in EM compared to profit- organizations. Whereas for profit- organizations making profits is the main goal, since these profits can be distributed to the shareholders, for NPOs earnings are distributed to programs and services (Verbruggen and Christiaens, 2012). The results of these programs and services, which are in line with the organization’s philanthropic mission, are more important than making a profit.

However, NPOs may have different reasons to engage in EM, namely improvement in their efficiency ratios (Krishnan et al., 2006), avoiding taxes (Omer and Yetman, 2003), obtaining funding (Verbruggen and Christianes, 2012) and avoiding small losses (Leone and van Horn, 2005).

Baber et al. (2001) investigated program ratios, which is the part of the total expenses of a NPO that is committed to the charitable objective, divided by total expenses. The amount that is distributed to the program is seen as the amount of resources contributed to the charitable objective, less fund-raising expenditure. Program ratios are used by stakeholders to evaluate the organization: NPOs with a higher program ratio are evaluated more positively than organizations with a lower ratio: NPOs with a higher ratio distribute more expenses directly to its charitable objective. As a result, nonprofit managers have an incentive to manipulate efficiency ratios. In this spirit, Krishan et al. (2006) investigated NPOs in the United States and examined whether managers of NPOs misclassify expenses to boost their program ratios. In the United States, most of the NPOs are obligated to file an IRS 990 form, which is a tax form. In this form, total expenses are dividend in programs (the charitable objective), fundraising, and administrative categories. They examined a sample of NPOs which reported zero fundraising expenses and found, that in half of the these NPOs, fundraising activities took place that should be reported on the IRS 990 form. A second test indicated that in 38% of the observations, NPOs reporting lower fundraising expenses on their IRS 990 form than in their audited financial statements. They also find evidence that this misreporting is driven by managerial incentives, and thus some form of EM exists within NPOs.

2003). Omer and Yetman (2003) reports a ‘unusually large number of nonprofits that report taxable income profitability in the range of [-0.01, 0.1]’, which can be interpreted as intentional tax avoidance.

Another reason for NPOs to manipulate earnings is to avoid small losses. According to Leone and van Horn (2005), NPOs don’t have an incentive to avoid reporting earnings decreases as long as current period earnings are above zero for several reasons. First, reporting a small profit means that the organization can continue over time (going concern assumption). Second, a lot of NPOs raise capital by issuing debt. The cost of debt will be lower if the variance of earnings is low, since this means the risk for debt providers will be lower. This creates an incentive to minimize the earnings variance to a point just towards zero. Also, reporting an excessive profit will suggest that the organization is unable or unwilling, to distribute all its earnings to charitable programs. Investigating US hospitals, evidence was found that discretionary accruals are positive (when pre-managed earnings are negative (positive), which indicates that US hospitals are managing earnings towards zero. Verbruggen and Christiaens (2012) supports this ‘drive toward zero hypothesis’. They find that Belgian NPOs manage accruals to be able to report around zero profit. Leone and van Horn (2005) also found that the distribution of earning surrounding zero was non- normal just below zero, indicating that NPOs are avoiding small losses.

Verbruggen and Christiaens (2012) states that in order to increase donations, NPOs will manage earnings downwards to show they need funding: when they report high profits, subsidizing organizations may not feel the need to provide additional funding. But on the contrary, NPOs need to show positive results in order to be trustworthy. Otherwise, it may be socially and economically undesirable for organizations to subsidize an organization that reports losses. This is in line with Boateng et al. (2016), who also recognizes this pressure: ‘Charities have become increasingly important, as government and donors, clients and public have increased pressure on charities to demonstrate their impact on complex social problems.’ Thus, NPOs that receives a significant amount of governmental funding have an even greater incentive to manage earnings toward zero. Using data of Belgian NPOs, they provide evidence for this theory.

2.3! Incentive contracts

The well-studied principal-agent framework is based on the relationship between the principal (firm-owner) and the agent (executive). The agent receives he authority delegated from the principal to make decisions on behalf of this principal. If the actions of the agent cannot be observed by the principal directly, the agent may take actions that are not in the best interests of the principal. The principal can discourage this behavior by creating incentives for the agent to make sure he will not engage in actions that could harm the principal (Jensen and Meckling, 1976). Examples of these types of incentives are bonuses based on earnings or stock options that are tied to the stock price. In this way, the agent can maximize their wealth if the wealth of the principal increases too and some type of risk-sharing exists. However, important here is that the performance measures included in the variable contract give information about the manager’s actions (Bushman et al., 1996). Over the last decade, researchers tried to find, based on this agency and contingency theory, which incentives can be used best in different settings and what the behavioral consequences of different types of incentives are.

Traditionally, financial performance measures are most commonly and widely used within organizations (Ittner et al., 1997). However, in the recent years, using solely financial performance measures and not both financial and nonfinancial performance measures have been criticized for focusing too much on short- term gains. For example, Bergstresser and Philippon (2006) find that firms with higher levels of stock-based incentives are exposed to a higher level of AEM. Also, Burns and Kedia (2006) reports a higher propensity to misreport when the CEOs option portfolio is related to stock price. Guidry et al. (1997) investigated the use of discretionary accruals on business-unit level based on the ‘bonus-maximization hypothesis’, which states that managers manipulate discretionary accruals to maximize their own compensation (Healy, 1985). Using 179 business-unit years, they find evidence for this theory. Holthausen et al. (1995) provides evidence that executives manipulate earnings downwards if their compensation is at their maximum. However, Graham et al. (2005) points out that managers are interested in managing earnings, especially to influence stock prices and their own welfare via career concerns and external reputation. Incentives related to debt, credit ratings,

more weight on nonfinancial performance measures in their executive compensation contracts. Using solely financial performance measures within these type firms creates to much noise, and thus performance can’t be measured in a reliable way (Ittner, 1997). Product quality for example, can be an important for the long-term value for a firm, but is not easily captured in a financial performance measure such as earnings. Said et al. (2003) investigated if firms that use both financial and nonfinancial performance measures have significantly better performance than firms that use solely financial performance measures and provided evidence that firms that use both types of measures have a significant higher return on assets (ROA) and market return. When non-financial performance measures are tied to the variable compensation contract of executives, managers will be less motivated to focus on earnings or stock-price. Accordingly, Ibrahim and Lloyd (2011) provides evidence that firms that use both financial performance measures and non-financial performance measures in executive compensation contracts face a lower level of AEM than firms that use solely financial performance measures in their executive compensation contracts.

2.4! Incentive contracts in nonprofit organizations

While the literature about the use of financial and nonfinancial performance measures in incentive contracts within for-profit organizations is widespread, the literature about the use of incentive contracts in NPOs is scarce. For for-profit organizations, performance is captured in solely financial metrics or both financial and nonfinancial metrics. Until now, no clear metrics seems to exist to capture the performance of NPOs. However, there seems to be a consensus that a single measure isn’t able to capture the complexity of NPOs performance (Hallock, 2002) and thus multiple measures should be used (Boeateng et al., 2016).

Boeateng et al. (2016) identified through qualitative and quantitative research the most commonly used performance measures within British NPOs, which are programme spending to total income, quality of product/service and client satisfaction. The use of these measures is understandable since the core of NPOs is delivering services to clients, and these clients are thus the most important or often the only reason why the NPO exists. Five broad sets of factors seem to measure the performance of charities, which are: (1) financial measures, (2) client/customer satisfaction, (3) management effectiveness, (4) stakeholder involvement and (5) benchmarking. Nevertheless, profitability may be an important metric for for-profit organizations, but profit is not necessarily associated with performance in NPOs.

The absence of intensive focus on financial measures results in minimal use of incentive contracting in NPOs. In the UK, CEOs of charities receive on average about 6% of their base salary as bonuses, compared to 100% or higher for other major UK Companies (Jobome, 2006). However, there is still no agreement if incentive contracting is desirable in the nonprofit sector. In recent years, the nonprofit sector is growing (Chenhall, 2003), and this leads to a higher expectation of professional service delivery: NPOs have to act more like profit companies. To live up to this expectation, more skilled workers are needed, and to attract them, more competitive salaries are inevitable (Jobome, 2006). In the same spirit, Hallock (2002) investigated the compensation of top managers of nonprofits in the United States. He argues that it makes sense to study compensation for top managers for profit in the same way as for nonprofit firms

within NPOs. Accordingly, Baber et al. (2001) shows that managers of NPOs are compensated for performance with respect to increased program ratios.

However, Hallock (2002) also finds that pay level for nonprofit managers is significant lower than for managers working in for-profit organizations. He theorizes several reasons for this difference in salary. First of all, workers may be willing to receive a lower pay in exchange for social benefits. Second, heads of nonprofits may accept lower wages to signal that donations are used more effectively. Third, lower wages may be accepted due to other advantages, such as flexible hours or a slower pace or work. Fourth, NPOs may attract managers with different characteristics compared to profit managers. Jobome (2006) argues that nonprofit managers tend to have a high level of altruism, and paying salaries compatible with the profit sector may attract the wrong type of manager. He states that intrinsic motivated individuals need less supervision than extrinsic motivated individuals, since these individuals are motivated by a need for meaning, esteem and actualization, and are honest, show initiative, take responsibility and are self-directed in the service of objectives they are committed to. On the contrary, extrinsic motivated individuals need incentive pay as a control mechanism. Results of his research suggest that nonprofit managers are not motivated by pay and thus are intrinsic motivated managers. Similarly, Chen et al. (2014) shows that NPOs are using less performance-based incentives compared to for-profit organizations, but that employees are provided with better health benefits and insurance. In contrast, Brickley and van Horn (2002) examined the incentives of chief executive officers (CEOs) in large nonprofit hospitals. They find that incentives are tied to financial performance and thus the board of hospitals care about profit and not about altruistic activities. Research shows that some issues can arise when nonprofits use solely implicit incentives (Speckbacher, 2011: Ben-Ner et. al, 2011). NPOs that hire intrinsically motivated employees, may hire employees that have intrinsic motivation for some aspects of their tasks, yet ignore others. With the absence of any outcome contractibility, focusing the employee on all aspects can be problematic.

2.5! Incentive contracts and firm performance

Variable incentive contracts and the effect on firm performance have been widely investigated by researchers for decades now and has given mixed results. As discussed before, incentive contracts are designed to align the interest of the executives with that of the firm owners (Jensen and Meckling, 1976). The use of these incentives protect the interests of the shareholders and thus should result in better firm performance.

Murphy (1985) was one of the first that found a statistically significant positive relationship between executive compensation and corporate performance, which was measured by shareholder return and growth in firm sales. Evidence suggest that executives are rewarded (or punished) for taking actions that benefit (or harm) shareholders (Murphy, 1985). Mehran (1995) investigated 153 firms and results showed a positive relationship between the percentage of equity held by managers, percentage of their compensation that is equity based and return on assets (ROA).

On the contrary, Core et. al (1999) investigated board determinants of CEO compensation and its effect on future firm operating and stock market performance. They find a negative correlation between CEO compensation and firm performance in terms of return on assets (ROA) and stock returns. In the same spirit, Mishra et. al (2000) looked at the effectiveness of pay-for-performance. Results suggest that future performance will become more negative within higher risk firms. Executives are risk-averse, and by introducing incentive pay package, executives will try to improve firm performance and maximize their pay. However, when executives are exposed to too much risk, managers can become overly risk averse, and this in turn, will have a negative impact on firm performance (Mishra et. al, 2000).

2.6! Hypothesis development

Previous research provided consistent evidence that for-profit organizations have incentives to manage earnings upwards (Healy and Wahlen, 1999), whereas discussed earlier, NPOs have incentives to manage earnings towards zero. Literature provides several reasons for NPOs to engage in earnings management, namely: improve their efficiency ratios (Krishnan et al., 2006), avoiding taxes (Omer and Yetman, 2003), obtain funding (Verbruggen & Christiaens, 2012) and avoiding small losses (Leone and van Horn, 2005). Managing earnings upwards can have negative consequences for NPOs, such as paying too much taxes and funding withdrawal.

H1: Nonprofit organizations manage their earnings towards zero.

Until now, no consensus seems to exist about the desirable salary and incentive pay for employees within NPOs. However, it can be argued that managers of NPOs will react to explicit incentives. It is hardly possible that one employee will be intrinsic motivated for all the activities that employee performs in his/her job. Therefore, it can be desirable to use explicit incentives to align the employees’ behavior with the desired behavior for the activities that the employee isn’t intrinsic motivated for. According to the agency theory, organizations choose optimal compensation contracts (Chen et al, 2004), and if managers of NPOs receive bonuses based on net earnings or revenues, these metrics should be considered important for the NPOs. This is also in line with the results of the research of Brickley and van Horn (2002): when financial performance indicators are set, hospitals are acting like a profit organization. Another possibility is that the NPO wants the manager to pay attention to these metrics since the employee may not be intrinsic motivated for this part of his/her job. In this case, the ‘bonus-maximization hypothesis’ would also apply for managers of NPOs.

H2: Nonprofit organizations with executives with variable compensation contracts are more inclined to manage earnings upwards, compared to nonprofit organizations that don’t have executives with a variable compensation contract.

In line with the traditional agency theory (Jensen and Meckling, 1976), incentives are used to make sure the actions of the agent enhances firm performance and firms choose optimal incentive contracts. When NPOs are using incentives contracts based on for example, earnings, these earnings are considered important by NPOs. So, the NPO acts like a profit organization. this should enhance performance measured in ROA. This leads to the following hypothesis:

3! Research Methodology

To obtain (financial) data about NPOs I used publicly available data from the Internal Revenue Service (IRS), which is the United States government agency which is responsible for tax collection. In the United States, tax-exempt organizations which have a gross receipts equal or bigger than $200.000, or total assets that equal or are bigger than $500.0003 _{are required to file an} IRS 990 form. An IRS form is a return of organization exempt from income tax4_{. Every year, the IRS} creates a sample of these IRS 990 forms, ranging from 1 percent for small-asset classes to 100 percent for large-asset classes to create the Statistics of Income (SOI)5_{. This data, on individual} organization-level data, which contains hundreds of variables, is publicly available. The IRS 990 form, contains several schedules6_{, one of them is Schedule J, which contains} additional compensation information of the organization’s current and former officers, directors, trustees, key employees and highest compensated employees. Organizations which states “yes” to one of the following questions are obligated to complete Schedule J:

•! Did the organization list any former officer, director or trustee, key employee, or highest employee on line 1a (contains key information about compensation)?

•! For any individual listed on line 1a, is the sum of reportable compensation and other compensation and related organizations greater than $150.000?

•! Did any person listed on 1a receive or accrue compensation from any unrelated organization or individual for services rendered to the organization?

In Schedule J, under part I, organizations have to state if they pay or accrue any compensation contingent on the revenues or net earnings. Organizations are also required to state if their employees participated in or received payment from an equity-based compensation arrangement (See figure 1). I used the most recent data available, which were datasets from the years 2012-2013. Only organizations with a large-asset class are included in the sample, since I need multiple year data from the same organizations.

3 https://www.irs.gov/charities-non-profits/form-990-series-which-forms-do-exempt-organizations-file-filing-phase-in 4 _{https://www.irs.gov/charities-non-profits/exempt-organizations-forms-and-instructions} 5 _{https://www.irs.gov/uac/soi-tax-stats-charities-and-other-tax-exempt-organizations-statistics} 6 _{https://www.irs.gov/uac/about-form-990-schedules}

3.2! Method

3.2.1! Accrual-based Earnings Management in the profit sector

Healy (1985) was the first to identify EM in organizations through accruals. Accruals were defined here as the difference between reported earnings and cash flows from operations. The predicted sign of accruals given the managers’ bonus incentives were compared with the actual sign of accruals. For example, a manager has an incentive to manage earnings upwards if the net earnings are below the earnings target and hence accruals can be used to manage earnings upwards. A distinction is made between nondiscretionary accruals and discretionary accruals. Non-discretionary accruals are obligatory expenses that aren’t realized yet, but are already recorded as expenses in the general ledger. Non-discretionary accruals are harder to manipulate by management, since these expenses are obligatory expenses, such as salary costs or upcoming invoices. Discretionary accruals, on the other hand, are adjustments to cash flows selected by the manager and create opportunities for the manager to manipulate earnings. For example, a manager can choose different methods of depreciation to influence accruals to maximize his or her bonus compensation. In the same spirit, DeAngelo (1986) investigated conflicts in management “buy-out” transactions, which happens when a public corporation purchases all common stock held by outsiders. Here, an opportunity for inside-managers exist, to pay the outside stockholders less than fair value for their shares. Total accruals from a prior period were used to determine the “normal” total accrual. Then, the “abnormal” accrual was defined as the difference between the current total accruals and normal total accruals.

In 1991, Jones (1991), introduced a, now widely recognized, model to measure AEM. This study focused on discretionary accruals as the source of earnings management. In other words, the discretionary part of total accruals is used to detect EM. The total accruals are equal to the change in noncash working capital before income taxes payable, less depreciation expenses. The change in noncash working capital before taxes is calculated as the change in current assets less current liabilities. Cash, short-term investment, current maturities, long-term liabilities and income tax payable are not incorporated in the calculation. Jones (1991) recognized the effect of economic circumstances of the firm on accruals. So, firms facing declining revenues, can be expected to have a smaller amount of nondiscretionary accruals. So, she used the model of DeAngelo (1986), but took into account the change in economic circumstances of the firms.

Dechow (1994), introduced the modified Jones model. This modified model implicitly assumes that all changes in credit sales in the event period resulted from EM. The reasoning behind this is

that it is easier for management to manage earnings by manipulating the recognition of revenue of credit sales than it is to manipulating the recognition of revenue of cash sales. This resulted in adjustment for the change in receivables in the event period. Thus, Dechow (1994) assumed, in addition to Jones (1991), that discretion is exercised over revenues in order to manipulate earnings.

In the recent years, researchers tried to find other, or more refined models to estimate discretionary accruals. These adjusted Jones models, such as the performance matched Jones model (Kothari et. al, 2005), may increase reliability under certain circumstances. For example, the performance matched Jones model performances better in situations where the control firms are expected to not engage in earnings management (Kothari et. al, 2005). However, these circumstances are not relevant for my research, since I expect NPOs to engage in EM, and hence the original Jones (1991) model will be used.

3.2.2! Accrual-based Earnings Management in the nonprofit sector

Leone and van Horn (2005) examined the use of two specific accruals, namely third party allowances and doubtful accounts to investigate the existence of EM in non-profit hospitals. They also used the Jones model (1991) to estimate the discretionary accruals. To investigate if managers of NPO’s are avoiding to report small losses, they plotted histograms of performance and evaluate if the distributions around zero are normal. Ballentine et. al (2007) used a similar method. They tested of the distribution of reported income are discontinuous around zero for English hospitals. Verbruggen and Christiaens (2012) used the Jones model (1991) per sector to provide evidence that EM exists in Belgian NPOs. They used unmanaged earnings (earnings before discretionary accruals), reported earnings and the distribution characteristics of these variables.

3.2.3! Jones model

The Jones model (1991) uses discretionary accruals as proxy for accrual-based earnings management. As discussed earlier, the model estimates the nondiscretionary and discretionary accruals in year i using the following regression (1). The abnormal or discretionary accruals (DA)

TAit/Ait-1= αi[1/Ait-1] + β1i[∆REVit/Ait-1] + β2i[PPEit/Ait-1] + εit (1) where:

TAit = total accruals in year t for firm I

∆REVit = revenues in year t less revenues in year t – 1 for firm i PPEit = gross property, plant, and equipment in year t for firm i Ait – 1 = total assets in year t – 1 for firm i

εit = error term in year t for firm i

As mentioned earlier, Total accruals can be calculated as the change in noncash working capital before income taxes payable less total depreciation expense. This noncash working capital before taxes equals the change in current assets other than cash and short-term investments less current liabilities that are not current maturities or income taxes payable (Jones, 1991). Thus, TA are calculated using the following equation (2):

TAit= (∆CAit- ∆CASHit) – (∆CLit- ∆DLit) – DEAMit (2) where:

∆CAit = current assets in year t less current assets in year t-1 for firm i

∆CASHit = cash and short-term investment in year t less cash and short-term investments

in year t-1 for firm i

∆CLit = current liabilities in year t less current liabilities in year t-1 for firm i ∆DLit =debt included in current liabilities in year t less debt included in current liabilities in year t-1 for firm i

DEAMit = depreciation and amortization expenses in year t for firm i

3.2.4! Regression Models

I will use several regression models to test my hypotheses 1, 2 and 3. To test hypothesis 1, I need to establish whether NPOs manage earnings upwards (downwards) when unmanaged results are negative (positive). I will use a regression analysis (3) to test hypothesis 1:

DAit= β0 + β1EBDAit-1 + β2SIZEit+ β3LEVit+ β4LAGEARit-1 + β5LAGDAit-1 (3) where:

DAit = discretionary accruals in year t for firm i

EBDAit-1 = Earningst + interest expensest / Ait-1 – discretionary accruals in year t for firm i SIZEit = natural logarithm of the total assets at the end of year t for firm i

LEVit = total liabilitiest/ total assetst for firm i

LAGEARit-1 = lagged earningst-1 + interest expensest-1/ total assetst-2 for firm i LAGDAit-1 = lagged discretionary accruals in year t-1 for firm i

I predict a negative relationship between unmanaged earnings (EBDA) and DA: when the unmanaged earnings are positive, negative discretionary accruals will be used to manage earnings towards zero. On the other hand, to correct unmanaged losses, less positive discretionary

accruals need to be used.

With respect to hypothesis 2, I plan on using a different equation (4). Distinction is made between three types of incentives: equity, revenue or earnings based incentives. The same control variables will be used.

DAit = β0 + β1EBDAit-1 + β2EQUITYINCit + β3REVINCit + β4EARINCit + β5SIZEit +

β6LEVit+ β7LAGEARit-1 + β8LAGDAit-1 (4)

where:

DAit = discretionary accruals in year t for firm i

EBDAit = earningst+ interest expensest/ Ait-1 – discretionary accruals in year t for firm i EQUITYINCit = dummy: if equity based incentive is present: 1; if not: 0

REVINCit = dummy: if revenue based incentive is present: 1; if not: 0 EARINCit = dummy: if net earnings based incentive is present: 1; if not: 0 SIZE = natural logarithm of the total assets at the end of year t for firm i

these types of incentives are present, executives are inclined to manage earnings upwards, which results in a higher amount of discretionary accruals.

To provide more evidence for hypothesis 2, I plan on using a second regression model (5) with absolute amounts of incentive pay as variable, instead of the dummy variables. The absolute amount will be gathered from the filed schedule J, where NPOs have to state the amount of bonus or incentive compensation paid in year i. The downside of this regression is that a distinction isn’t made between the three types of incentive pay.

DAit = β0 + β1EBDAit-1 + β2 TOTALINC + β3SIZEit+ β4LEVit+ β5LAGEARit-1 +

β6LAGDAit-1 (5)

where:

DAit = discretionary accruals in year t for firm i

EBDAit = earningst + interest expensest / Ait-1– discretionary accruals in year t for firm i TOTALINCit = amount of incentive or bonuses paid in year t for firm i to member with highest compensation

SIZEit = natural logarithm of the total assets at the end of year t for firm i LEVit = total liabilitiest / total assetst for firm i

LAGEARit-1 = lagged earningst-1 + interest expensest-1/ total assetst-2 for firm i LAGDAit-1 = lagged discretionary accruals in year t-1 for firm i

To test my hypothesis 3, I use regression model (6). Similar as in regression (5), the absolute amount of incentive payment will be collected from schedule J. Based on the theory, I predict a positive correlation between the total amount of incentive payment and earnings.

EARNINGSit/Ait= β0 + β1 TOTALINC + β2SIZEit+ β3LEVit+ β4LAGEARit-1 +

β5LAGDAit-1 (6)

where:

EARNINGSit = net earnings + interest expenses in year t for firm i Ait = total assets in year t for firm i

TOTALINCit = amount of incentive or bonuses paid in year t for firm i to member with highest compensation

LEVit = total liabilitiest / total assetst for firm i

LAGEARit-1 = lagged earningst-1 + interest expensest-1 / total assetst-2 for firm i LAGDAit-1 = lagged discretionary accruals in year t-1 for firm i

The following items of the IRS 990 form are used for the variables in the equations or regression models (an example of an IRS form can be found in the appendix):

∆REVit = CORE Part I Item 12 (current year - prior year) PPEit = Schedule R Part VI Item 1A + 1B + 1C + 1D Ait – 1 = CORE Part I Item 20 (current year-prior year) ∆CAit = CORE Part X Items 1 to 9 (current year - prior year) ∆CASHit = CORE Part X Items 1 + 2 (current year – prior year) ∆CLit = CORE Part X Items 17 to 22 (current year – prior year) ∆DLit = CORE Part X Items 20 + 22 (current year – prior year)

DEAMit = CORE Part IX Item 22

EBDAit-1 = CORE Part I (Item 19 + interest expenses) / Item 20 -

discretionary expenses

Interest

expenses = CORE Part IX Item 20

EQUITYINCit = Schedule J Part I Item 4c REVINCit = Schedule J Part I Item 5a EARINCit = Schedule J Part I Item 6a TOTALINCit = Schedule J Part II Item b(ii)

In their review paper, Dechow et. al (2010) discuss different categories of variables that determine earnings quality, namely (1) firm characteristics, (2) financial reporting practices, (3) governance and control, (4) auditors, (5) equity market incentives, and (6) external factors. Due to the different nature of NPOs compared to for-profit firms, not all variables are relevant for my regression model. More reasons for not including variables are evidence weakness, prior

firm size is negatively associated with discretionary accruals (Watts and Zimmerman, 1978). Reason for this, is that bigger firms are more likely to be subjected to governmental interference and therefore face a higher degree of governmental regulation. However, other studies predict a positive relationship between firm size and discretionary accruals (Ball and Foster, 1982). They argue that the relative fixed costs associated with government regulation decreases rather than increases with firm size. Since evidence from previous research is mixed, I don’t predict the direction of the relationship between SIZE and DA.

LEV, or leverage, is another variable that influences the magnitude of discretionary accruals. With the view that higher leverage is indicative of a firm that is closer to a debt covenant restriction and therefore have an incentive to manage earnings upwards, a positive association between leverage and discretion accruals is expected (Watts and Zimmerman, 1991). DeFond and Jiambalvo (1994) provided evidence for this theory. They find abnormal working capital accruals in the year of debt covenant violation. Summarizing, I predict a positive association between LEV and DA.

LAGEAR, or lagged earnings, is often used as a control variable within the NPO research stream. Previous research showed a positive relationship between lagged earnings and discretionary accruals (Kothari et al., 2005; Leone and van Horn, 2005; Verbruggen and Christiaens, 2012).

LAGDA, or lagged discretionary accruals, (Leone and van Horn, 2005; Verbruggen and Christiaens, 2012) are incorporated in the regression model to control for autocorrelation.

4! Results

4.1! Descriptive statistics

As discussed before, observations were only included in the sample if the organizations were required to file a schedule J and if the total assets were equal to or higher than $50.000.000 at the end of the years 2012 and 2013. With respect to the compensation files, the extra files that are available when organizations have to file a schedule J, organization can have more than one highest compensated employee. For my sample, I selected the highest compensated employee per organization. Before the regression analyses, continuous variables have been winsorized to minimize the effect of outliers.

In the appendix (table 1), the spearman correlation coefficients between all variables used in the regression analyses are presented. Discretionary Accruals (DA) are significantly correlated to EBDA, REVINC, EARINC, TOTALINC, SIZE, LEV, LAGEAR and LAGDA. In line with predictions, EBDA is negatively correlated to Discretionary Accruals (DA). Results didn’t show high correlations between two independent variables, so no multicollinearity problems should arise.

4.1.1! Discretionary Accruals

The descriptive statistics of discretionary accruals (DA) are summarized in table 2. In the majority of the observations (66,42%), accruals were used to manage earnings upwards, which resulted in a slightly positive mean.

Obs. Mean Std. Dev % upwards % downwards

DA 6,560 .0014592 .1035544 66,42% 33,58 %

Table 2: Descriptive Statistics on Discretionary Accruals Discretionary Accruals (DA) are the residuals of equation (1).

4.1.2! Incentive compensation

DA Obs. Mean Std. Err. Std. Dev. P-value EQUITYINC 0 6,535 .0014909 .0012817 .1036124 1 25 -.0068091 .0176985 .0884926 combined 6,560 .0014592 .0012785 .1035544 diff .0083 .0207518 .6892 REVINC 0 6,360 .0018904 .0013081 .1043194 1 200 -.0122525 .0052446 .0741705 combined 6,560 .0014592 .0012785 .1035544 diff .0141429 .0054053 .0095* EARINC 0 6,238 .0023864 .0013066 .1031956 1 322 -.0165028 .0060684 .1088942 combined 6,560 .0014592 .0012785 .1035544 diff .0188892 .0059138 .0014**

Table 3: T-test results

Significance (two tailed) at 0.01 level (***), 0.05 level (**) and 0.10 level (*).

Discretionary accruals (DA) are the residuals of equation (1). EQUITYINC is a dummy variable that takes the value of one when incentive pay based on equity is present. REVINC is a dummy variable that takes the value of one when incentive pay based on revenue is present. EARINC is a dummy variable that takes the value of one when incentive pay based on earnings is present.

The mean of discretionary accruals for NPOs that don’t have incentive pay based on equity is .0014909 compared to -.0068091 for NPOs that do pay incentives based on equity. Although the mean is negative for NPOs without equity- based incentive pay, discretionary accruals does not differ significantly from NPOs that pay incentives based on equity (p = .6892). However, contrary to predictions, other results indicate that NPOs that pay incentives based on revenue

and earnings have significant lower (p = .0095 and .0014) discretionary accruals compared to NPOs that does not pay incentives based on revenue and earnings. The results suggest that executives that receive bonuses based on revenues and earnings are inclined to manage, contrary to predictions, earnings downwards. However, conclusions can’t be drawn from these results yet, since the tested variable is not incorporated in a regression model.

4.2! Regression results

The results of the regression analyses (3), (4) and (5) are shown in table 4. A Variance Inflation Factor Analysis (VIF) has been performed to check for any multicollinearity between independent variables (table 5). In line with the spearman correlation coefficients, no high multicollinearity between the independent variables seems to exist.

DA (H1) DA (H2) DA(H2)

Variables Exp. Sign

Coefficient P-value Coefficient P-value Coefficient P-value

EBDA - -.2532123 .000*** -.2523248 .000** -.2526913 .000*** EQUITYINC + .0032934 .867 REVINC + -.0055412 .453 EARINC + -.006916 .241 TOTALINC + -.000000 .280 SIZE +/- -.0038512 .001*** -.0037201 .000*** -.0035795 .000*** LEV + -.0550202 .000*** -.0547704 .000*** -.054692 .000*** LAGEAR + .0652406 .000*** .0651115 .000*** .0658203 .000*** LAGDA - .0713094 .000*** .0711366 .000*** .0711317 .000*** Observations 6,560 6,560 6,560

on revenue is present. EARINC is a dummy variable that takes the value of one when incentive pay based on earnings is present. TOTALINC is the absolute amount of incentive/bonus paid. SIZE is the natural logarithm of the total assets. LEV is total liabilities divided by total assets. LAGEAR is earnings in year t-1 divided by total assets in year t-2. LAGDA is DA for year t-1.

With respect to regression analysis 3, results show, as predicted, a negative significant (P = .000) relationship between EBDA and Discretionary Accruals (DA). This result provides more evidence for the theory that nonprofit organizations manage earnings downwards (upwards) when earnings are positive (negative). Results are similar to that of Verbruggen and Christiaens (2012), who also showed a negative relationship between EBDA and Discretionary Accruals (DA).

Furthermore, all control variables, SIZE, LEV, LAGEAR and LAGDA, have a significant relationship with Discretionary Accruals at the 0.001 level. However, LEV seems to have a negative relationship with Discretionary Accruals, while a positive relationship has been predicted. In line with previous research, LAGEAR is positively correlated to Discretionary Accruals (Kothari et al., 2005; Leone and van Horn, 2005; Verbruggen and Christiaens, 2012). LAGDA and Discretionary Accruals are negatively correlated which was not predicted. Results of regression analyses (4) and (5) don’t show a significant relationship between the different types of incentive pay, EQUITYINC, REVINC, EARINC, TOTALINC, and Discretionary Accruals (DA). Evidence suggest that incentive pay within nonprofits doesn’t influence the magnitude of Discretionary Accruals (DA). Hence, hypothesis 2 is rejected. This indicates that the used agency theory does not apply for NPOs. An explanation for the results may be that an employee is so strongly intrinsic motivated and therefore, behavioral change can’t be achieved by the use of explicit incentives (Jobome, 2006). Other means are necessary to drive change for these employees. Another possibility is that target setting within NPOs differ from for- profit organizations. For example, where the target for earnings in a profit setting is set as high as possible, for NPOs the target may be set around zero.

The explanatory power of all three regression models are similar, 10,39% and seems to be a lot lower than the comparable models of Verbruggen and Christiaens (2012) and Leone and van Horn (2005), with 57,2% and 33%, respectively. Reason for this lower explanatory rate may be the data difference, since the research of Verbruggen and Christiaens (2012) and Leone and van Horn (2005) is focused on Belgian NPO’s.

The results of regression analyses (6) are showed in table 6. All variables are significant related to EARNINGS. A significant positive relationship (p = .006) is found between the total amount of incentive payment and earnings. Organizations with executives which receives higher bonus

payments seem to have better firm performance, which is consistent with predictions and thus, hypothesis 3 is supported. This indicates that NPOs are acting more and more like profit organizations and earnings are thus considered important by NPOs. Exorbitant high earnings may not be desirable, but having an increased performance compared to others could have several advantages, for example in collecting funds from donors. SIZE and LAGEAR have a positive relationship with EARNINGS. The total variance of the data explained by the model is 19,19% which is higher than the previous models.

EARNINGS (H3)

Exp. Sign Coefficient P-Value

TOTALINC + .000000 .006*** SIZE +/- .0048696 .000*** LEV + -.0213641 .000*** LAGEAR + .3137757 .000*** LAGDA - -.0052822 .006*** Observations 6,560 Adjusted R2 19,19%

Table 6: Results of regression analysis (6).

Significance (two tailed) at 0.01 level (***), 0.05 level (**) and 0.10 level (*).

EARNINGS are the net earnings plus interest costs divided by total assets. EBDA is reported earnings (divided by lagged total assets) minus discretionary accruals. TOTALINC is the absolute amount of incentive/bonus paid. SIZE is the natural logarithm of the total assets. LEV is total liabilities divided by total assets. LAGEAR is earnings in year t-1 divided by total assets in year t-2. LAGDA is DA for year t-1.

5! Summary and Conclusion

The aim of this study was to investigate the effect of the use of incentives in compensation contracts on earnings management and enterprise performance in NPOs. Recently, NPOs are growing and becoming increasingly important in our economies (Chenhall, 2013). Thus, some researchers claim that NPOs should behave the same as for-profit organizations in some situations (Jobome, 2006).

Prior research shows that NPOs tend to manage earnings towards zero (Leone and van Horn, 2005; Verbruggen and Christiaens, 2012), since a big negative or high positive income is not desirable. Motivations for this ‘drive toward zero’ are increased opportunity to obtain funds from the government and donors and avoiding redundant tax payment (Leone and van Horn, 2005; Omer and Yetman, 2013; Verbruggen and Christiaens, 2012).

Research about the use of incentive contract in NPOs is scarce, but shows that incentive

contracting in the NPO sector is still rare (Jobome, 2006). However, according to the traditional agency theory, organizations choose optimal incentive contracts. So, if a certain incentive contract is chosen, for example, an incentive based on earnings, earnings should be considered to be important by the NPO. Then, a manager will always try to maximize his or her bonus, so will act in line with the incentive and thus will manage earnings upwards. In the same sense, an organization will only choose a particular contract if it enhances the firm performance.

IRS 990 forms from the United States tax authority where used to gather the data needed. 6,560 NPOs over the years 2011-2013 were analyzed. The existence of earnings management was measured through discretionary accruals. Discretionary accruals were estimated using the Jones (1991) model. Descriptive statistics shows that in the majority of the observations (66,42%), accruals were used to manage earnings upwards. Additionally, in line with Boateng et. al (2016), the use of incentive contracting still seems to be scarce within the nonprofit sector. Incentives based on equity are only present in 0,003% of the sample. Incentives based on revenue and earnings are increasingly used, with 0,03% and 0,05% respectively.

After that, T-test were performed to test if the mean of discretionary accruals differs if one type of incentive is in place; this seems the case for incentives based on revenue and earnings. However, contrary to predictions, results indicate that discretionary accruals seems to be lower for NPOs that use incentives based on revenue and earnings.

Regression analyses were performed to control for other factors as well. In the first regression discretionary accruals were regressed with the variable EBDA, and control variables SIZE, LEV, LAGEAR and LAGDEA to see whether NPOs manage earnings towards zero. A negative significant relationship was found between EBDA and discretionary accruals, which indicates that NPOs manage earnings towards zero. Improving efficiency ratio’s (Krishnan et al., 2006), avoiding taxes (Omer and Yetman, 2003), obtaining funding (Verbruggen and Christianes, 2012) and avoiding small losses (Leone and van Horn, 2005) could be reasons for this behavior. With respect to hypothesis two, two regression analyses were performed to see whether incentive payments had an influence on discretionary accruals. In one regression model, distinction was made between the three types of incentive payments (incentives based on equity, revenue or earnings) and in the other model the total amount of incentive payment was incorporated and regressed. No significant relationship was found between the incentive variables and discretionary accruals. The traditional agency theory does not seem to apply for NPOs (where the agent always will try to maximize his/her bonus) or target setting with respect to incentives within NPOs differs from for-profit organization.

Finally, earnings were regressed with the absolute amount of bonuses paid, TOTALINC, and control variables SIZE, LEV, LAGEAR and LAGDEA to test hypotheses 3. Results indicate a positive association between TOTALINC and earnings: thus, nonprofits with incentive payments seem to have better firm performance.

Overall, this research contributes to an emerging stream of research. With this thesis, more evidence is provided for the ‘drive towards zero’ hypothesis. Also, more insight is given in the way incentives are used within NPOs. However, results are mixed and more research is needed in this field.

5.2! Conclusion

In line with previous research (Leone and van Horn, 2005; Omer and Yetman, 2003; Verbruggen and Christiaens, 2012), results show that NPOs tend to manage earnings towards zero, instead of upwards. NPOs seems to manage earnings towards zero to avoid paying taxes, obtain funds

management. A possible explanation for this outcome may be found in the content of the incentive contracts. It may be possible that target earnings are not set as high as possible (which is usual for for-profit organizations), but around zero. This will logically influence the behavior of the manager: discretionary accruals are then not used to manage earnings upwards, but to manage earnings towards zero. Again, it would be interesting to know why certain incentives are set and how these incentives are set. After all, it seems a little bit strange for a NPO to use incentives based on earnings when the main goal of the NPO is to distribute money to programs and services.

Finally, a significant positive relationship was found between incentive payment and earnings, which indicates that the use of incentive contracting enhances the performance of NPOs. A side note here is that we don’t know yet if these higher earnings are desirable by the NPOs, since previous research shows that NPOs have a preference for earnings around zero. Again, more research is needed to indicate what the content of the contracts encompasses and when NPOs are considered better performer than other NPOs.

5.3! Limitations

One limitation of this research is that only large nonprofit organizations are included in the sample, which was necessary to conduct the research. Additionally, the results can’t be applied to specific sectors in which NPOs operates. Furthermore, the Jones (1991) model was used as a proxy for earnings management. Other models may result in different outcomes and the model can contain measurement errors. Furthermore, other factors than incorporated in my regression models have an effect on the variances in discretionary accruals and earnings. So, including other control variables may affect the outcomes too.

References

Baber, W.R., Roberts, A.A. and Visvanathan, G. (2001) Charitable organizations’ strategies and program-spending ratios. Accounting Horizons, 15(4), pp. 329-343.

Ball, R. and Foster, G. (1982) Corporate financial reporting: a methodological review of empirical research. Journal of Accounting Research, 20(supplement), pp. 161-234.

Ballantine, J., Forker, J. and Greenwood, M. (2007) Earnings Management in English NHS hospital trusts. Financial Accountability & Management, 23(4), pp. 421-440.

Ben-Ner, A., Ren, T. and Paulson, D.F. (2011) A sectoral comparison of wage levels and wage inequality in human service industries. Nonprofit and Voluntary Sector Quarterly, 40(4), pp. 608-633.

Bergstresser, D. and Philippon, T. (2006) CEO incentives and earnings management. Journal of

Financial Economics, 80(3), pp. 511-529.

Boateng, A., Akamavi, R.K. and Ndoro, G. (2016) Measuring performance of non-profit

organizations: evidence from large charities. Business Ethics: A European Review, 25(1), pp. 59-74.

Brickley, J.A. and van Horn, R.L. (2002) Managerial incentives in nonprofit organizations: evidence from hospitals. Journal of Law and Economics, 45(1), pp. 227-249.

Burns, N. and Kedia, S. (2006) The impact of performance-based compensation on misreporting. Journal of Financial Economics, 79(1), pp. 35-67.

Bushman, R.B., Indjejikian, R.J. and Smith, A. (1996) CEO compensation: the role of individual performance evaluation. Journal of Accounting and Economics, 21(2), pp. 161-193.

Chen, X., Ren, T. and Knoke, D. (2014) Do nonprofits treat their employees differently?

nonprofit management & leadership, 24(3), pp. 285-306.