CEO compensation based on non-GAAP performance measures

(1)

1

CEO Compensation

Based on Non-GAAP

Performance Measures

Master Thesis

Name: Sam Dijkshoorn (10858490) Thesis Supervisor: mw. N. (Nan) Jiang MSc

Date: June 22, 2018 Word Count: 9577

MSc Accountancy & Control, Specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by student Sam Dijkshoorn 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)

Abstract

There is a lot of discussion going on about the use of non-GAAP performance measures in the compensation plans of CEO’s. Some argue that managers might manipulate non-GAAP earnings to report a higher non-GAAP earnings number and extract additional compensation from the firm. On the other hand, prior research has shown that non-GAAP earnings can give a better reflection of the real performance of a firm and that it’s a better indicator of future performance. This indicates that non-GAAP performance measures can be a useful and fairer measure to motivate and compensate managers. The purpose of this study is to investigate what the effect is of the use of non-GAAP based performance measures for CEO compensation on the level of total CEO compensation and future firm performance.

To answer this question, I hand collect data about whether U.S. based firm use non-GAAP performance measures in their CEO compensation plan and perform two tests. First, I test whether the use of non-GAAP performance

measures has a positive effect on the level of CEO compensation. Secondly, I test whether the predicted excess compensation, arising from the dummy variable non-GAAP, is associated with future stock returns and future return on assets.

I find that firms that do use non-GAAP performance measures in their CEO compensation plan, pay more compensation to their CEO. I do not find evidence that this higher total compensation is associated with better future firm performance.

The results indicate that managers misuse the freedom they have in determining the non-GAAP earnings number to extract additional payment from the firm. This insight can especially be useful for standard setters and compensation committees. Compensation committees must be cautious to use non-GAAP performance measures to motivate and compensate their CEO.

(4)

1 Introduction

In spite of the fact that the manager is supposed to make decisions that maximizes value for shareholders, it is in the manager’s best interest to maximize his own wealth. In other words, the interests of the manager and the shareholders are not aligned. This is called the agency problem. To solve this problem and better align management incentives with the interest of shareholders, the firm can establish a compensation plan with appropriate incentives for the agent. In recent years, it has become very common for companies to use non-GAAP performance measures to motivate and compensate managers. In a recent article (Accountant.nl, 2016), Hoogervorst states that it is alarming that many CEO compensations are based on non-GAAP earnings, because these earnings can be manipulated.

In accordance with Hoogervorst, Black et al. (2015) find that managers can use the discretion in defining non-GAAP earnings to mislead investors. Managers have a lot of freedom in deciding which components to exclude from non-GAAP earnings. Therefore, they might label more costs as nonrecurring cost, to report a higher non-GAAP earnings number and extract additional compensation from the firm. On the other hand, prior research has shown that non-GAAP earnings can give a better reflection of the real performance of a firm and that it’s a better indicator of future performance (Frankel, McVay & Soliman, 2011). This suggest that compensation based on non-GAAP performance measures can be a helpful and fairer method to motivate and compensate managers.

Because little is known about CEO compensation that is based on non-GAAP performance measures, it’s unclear whether it is a good way to better align the interest of the manager with the interest of the shareholders. The purpose of this study is to investigate what the effect is of the use of non-GAAP based performance measures for CEO compensation on the level of total CEO compensation and future firm performance. This study contributes to the literature about non-GAAP earnings, because it is the first study that directly examines the effect of using non-GAAP based performance measures in CEO compensation plans on the level of CEO compensation and future firm performance.

The research question answered in this thesis is: What is the effect of using non-GAAP based performance measures in CEO compensation plans on the level of CEO compensation and future firm performance? The sample used in this thesis

(6)

6

consists of 100 random U.S. firms over a time period of 8 years. I use a dummy variable to measure the use of non-GAAP performance measures in the compensation plan of the CEO. This data is hand collected. First, I test whether the use of non-GAAP performance measures has an effect on the level of CEO compensation. Because a higher level of CEO compensation can be explained by either rent extraction or efficient contracting, I will perform a second regression. In this regression I test whether the predicted excess compensation, arising from the dummy variable non-GAAP, is associated with future stock returns and future return on assets.

I find that the use of non-GAAP based performance measures in CEO compensation plans has a significant positive effect on CEO compensation. Firms that use non-GAAP performance measures to motivate and compensate managers, pay a higher level of total compensation to their CEO. I do not find evidence that this higher compensation is associated with better performance in future periods. These findings suggest that managers misuse the freedom they have in determining the non-GAAP earnings number to obtain a higher level of total compensation. The insight that managers might manipulate non-GAAP measures to obtain a higher level of total compensation is especially relevant for standard setters and compensation committees. Compensation committees must be cautious to use non-GAAP performance measures in their CEO compensation plan.

The remainder of this thesis is structured as follows. In section 2 I discuss prior literature about the agency problem, efficient contracting, rent extraction and non-GAAP measures. The hypothesis development is described in section 3. In section 4 I explain my research design in more detail. The results for the first test are presented in section 5. The research design of the second test and the results of this test are shown in section 6. Section 7 contains the discussion and conclusion.

(7)

7

2 Literature Review

In this section I review the existing literature. First the Agency Problem is explained. Compensation plans can limit the agency problem. There are two theories concerning compensation plans, the Efficient Contracting Theory and Rent Extraction Theory. These theories are explained in paragraph 2.2 and 2.3 respectively. The difference between GAAP and non-GAAP measures is explained in paragraph 2.4.

2.1 Agency Problem

According to the agency theory, an organization is a nexus of contracts. Because the ownership and control are often separated in firms, one important contract is the contract between the owners and the controlling chief executive officer (Jensen & Meckling, 1976). In the agency theory, the owners are seen as the principal and the chief executive officer is seen as the agent. The CEO is supposed to make decisions that will maximize value for shareholder, but it’s in the CEO’s best interest to maximize his own wealth. This is called the agency problem. Because of the separation in ownership and control, agency costs arise as a result of the divergence between the interest of the manager and the interest of the outside shareholders. In other words, the interests of the shareholders are not aligned with the interests of the manager. As a result, the manager may spend more money on luxurious offices or expensive business trips because the manager bears only a fraction of the costs from consuming these perks (Jensen & Meckling, 1976). Also, a manager may engage in empire building (Jensen, 1986). This means that the manager invests more resources than optimal in projects or firms, just to build a larger firm and obtain a higher status. This is not in the interest of the shareholders.

To limit these problems, the principal can reduce the misalignment of interest by monitoring the agent. The costs associated with monitoring are called monitoring costs. By incurring these monitoring costs, the shareholders can limit the consumption of perks by the manager (Jensen & Meckling, 1976). Another way to limit the misalignment of interest is by establishing a compensation plan with appropriate incentives for the agent. The costs associated with this compensation plan are called bonding costs. Although these measures will limit the agency problem, it’s impossible to perfectly align the interest of the shareholders with those

(8)

8

of the CEO. There will always be some divergence between the decisions of the CEO and those decisions that maximizes value for shareholders. This is called the residual loss. The agency costs are defined as the sum of the monitoring costs, the bonding costs and the residual loss (Jensen & Meckling, 1976). In this thesis, the focus will be on compensation plans. There are two views on executive compensation plans, the efficient contracting theory and the rent extraction theory. These theories are explained in the next two paragraphs.

2.2 Efficient Contracting Theory

As already mentioned, one solution to the agency problem is the alignment of interests through the structure of the CEO’s compensation plans (Fama & Jensen, 1983; Jensen & Meckling, 1976). Firms try to structure the chief executive officer’s compensation plan in a way that provides managers with efficient incentives to maximize value for shareholders (Bebchuk & Fried, 2003). This is called optimal contracting. To structure the compensation in the most optimal way, different elements are used. The total compensation of a CEO mainly consists of a base salary and a bonus that depends on performance targets. Prior research has shown that several factors can explain the level of CEO compensation. In this thesis the focus is on the three most important factors: firm performance, firm size and growth opportunities, and firm risk.

The first factor is firm performance. According to Sloan (1993), the compensation of the CEO is often linked to firm performance to align incentives of the CEO with the incentives of the shareholders. To motivate and compensate the CEO in the most optimal way, a combination of different performance measures and targets is used. Traditionally, firm performance have been measured by using financial performance metrics like net earnings or return on investments (Ittner, Larcker & Rajan, 1997). There are two types of financial measures, stock-price based measures and accounting based measures. Firms use both stock-price based incentives and accounting-earnings based incentives to motivate managers, because these measures do not always capture firm performance in the same way. For example, Berk and DeMarzo (2013) state that stock-price based measures are, in contrast to accounting-based measures, more forward looking and include investors’ expectations about the future. Prior research show that motivating and compensating managers with bonuses based on financial performance measures can

(9)

9

result in better firm performance. Firth et al. (2007) find a positive relation between CEO compensation and return on assets. In addition, Crumley (2008) find that stock return is significant positive related to CEO compensation. Also, non-financial measures, like product quality or customer satisfaction, are often used to evaluate and reward the performance of a manager (Ittner et al., 1997). According to Kaplan and Norton (1992), these measures are more forward looking than financial measures and therefore able to complement financial measures. In addition, Banker, Potter and Srinivasan (2000) analysed incentive compensation plan of hotel managers and found that customer satisfaction measures are significantly associated with future financial performance. In recent years, it has become very common for companies to also motivate and compensate managers based on non-GAAP performance measures, but there is no evidence that non-GAAP performance based compensation will result in better firm performance or not. In paragraph 2.4 I explain the difference between GAAP and non-GAAP earnings in more detail.

The second factor that can explain the level of CEO compensation is firm size and growth opportunities. Smith and Watts (1992) state that managers of larger firms deserve a higher compensation, because a manager’s decisions affect more resources. Therefore, managers of larger firms have a higher value added and thus deserve a higher compensation. Larger firms with more growth opportunities will demand for higher quality CEOs and these managers will ask a higher compensation.

Also, firm risk can be an important factor. Cyert, Kang and Kumar (2002) find that firms with more firm risk pay more compensation to their CEO. Also, Haggard and Haggard (2008) find that the level of CEO compensation is significant positively associated with firm risk. According to Murphy (2000), for firms with more firm risk, the outcomes are more uncertain. This results in a lower willingness of the CEO to invest in risky projects. To solve this problem, firms try to structure CEO compensation plans in a way that motivates risk-averse managers to bear higher risks and undertake value-increasing projects. These results show that the level of CEO compensation is partly explained by firm risk.

2.3 Rent Extraction Theory

Another view on executive compensation is the Rent Extraction Theory. Under this approach, executive compensation is not only seen as a mechanism to limit agency problems, but also as part of the agency problem itself (Bebchuk & Fried, 2003).

(10)

10

They argue that if the CEO has more power, he is able to negotiate a higher level of compensation. Several other studies have shown that managers can use their power to extract additional compensation from the firm. This results in compensation plans that are not structured in a way that maximizes value for shareholders.

For example, Core, Holthausen and Larcker (1999) find that CEO compensation is higher when the CEO is also the board chair, the board is larger or when there is a greater percentage of the board composed of outside directors. The CEO compensation is also higher if the outside directors are appointed by the CEO itself, grey, old or serve on more than three other boards. They concluded that if firms have weak corporate governance structures, managers are able to influence the board and extract higher compensation. This is called rent extraction. They also find that these companies have greater agency problems and perform worse in the future periods. Furthermore, Holderness and Sheehan (1988) states that managers that are owning more than half of the shares of a public company receive marginally higher compensation than other managers.

In addition, research by Bebchuk and Fried (2004) indicates that directors, who negotiate with the CEO about his compensation, do not always do this in the interest of the shareholders. Directors have incentives to go along with the demands of the manager because of social and psychological mechanisms like collegiality and friendship. Also, the CEO can influence the compensation of directors.

These studies indicate that managers can use their power to extract additional compensation from the firm.

2.4 GAAP and Non-GAAP Measures

As mentioned before, the most standard way to motivate and compensate CEO’s is by using financial performance metrics like net earnings or return on investments (Ittner, Larcker & Rajan, 1997). These financial performance metrics are determined in accordance with the General Accepted Accounting Principles, designed by the Financial Accounting Standards Board (FASB). These principles consist of uniform rules about how to record for various types of transaction in the financial statements. Also, these numbers are audited by an accountant and therefore reliable (Stolowy et al., 2013).

(11)

11

In recent years, it has become very common for companies to voluntarily disclose additional financial information that does not comply with the GAAP guidelines (Baik, Billing & Morton, 2008). These non-GAAP earnings are often made public through a press release (Frankel et al., 2011). Other names for non-GAAP earnings are pro forma earnings, street earnings or adjusted earnings (Young, 2014).

Research by Bradshaw and Sloan (2002) shows that investors, managers, analysts and the press are increasingly relying on non-GAAP earnings instead of GAAP earnings. Moreover, Graham et al. (2005) states that managers find non-GAAP earnings one of the most important measures of firm performance. In addition, Bradshaw and Sloan (2002) state that by determining the non-GAAP earnings, certain components are excluded. These exclusions are nonrecurring expenses or revenues. They argue that these nonrecurring items are unpredictable and therefore not relevant for the assessment of the performance of the company on the core activities. Also, managers argue that non-GAAP earnings better reflect the real performance of a firm and that it’s a better indicator of future performance by excluding these one-time nonrecurring items (Frankel, McVay & Soliman, 2011).

On the other hand, research by Black et al. (2015) shows that managers can use the discretion in defining non-GAAP earnings to mislead investors. Also, research by Bhattacharya et al. (2003) shows that sometimes costs, that are directly related to the core business activities of a firm, nevertheless are excluded from non-GAAP earnings. In addition, Bradshaw and Sloan (2002) state that managers can use non-GAAP earnings to show the results more positively than they actually are. They also find that managers increasingly identify costs as nonrecurring, resulting in increasing differences between GAAP and non-GAAP earnings. Research by Bhattacharya et al. (2004) shows that depreciation costs are the most common exclusions. Also, costs related to stock-based compensation, research and development and acquisitions are often not included in non-GAAP earnings. Furthermore, Fields, Rangan and Thiagarajan (1998) state that non-GAAP reporting is unreliable because non-GAAP reporting is, in contrast to GAAP earnings, not bound to consistent rules and is also not controlled by an independent accountant. Therefore, the management has a lot of freedom in deciding which components to exclude.

The discussed literature above indicates that there is a discussion going on whether non-GAAP measures are informative or not.

(12)

12

3 Hypotheses Development

In this section I explain the development of my hypotheses. The hypothesis for the first test is described in the first paragraph. In the second paragraph I explain my hypothesis for regression 2 and 3.

3.1 Non-GAAP Performance Measures and CEO Compensation

Although, the manager is supposed to make decisions that will maximize value for shareholder, it’s in the manager’s best interest to maximize his own wealth. This is called the agency problem. The interest of the shareholders is not aligned with the interest of the manager. The board of directors tries to limit the agency problem by structuring the CEO compensation plan in a way, so that the interest of the CEO is aligned with the interest of the shareholders.

Prior literature show that non-GAAP earnings can give a better reflection the real performance of a firm and that it’s a better indicator of future firm performance (Frankel, McVay & Soliman, 2011). On the other hand, managers have a lot of

freedom in deciding which components to exclude from non-GAAP earnings, because there are less strict rules and there is no audit on non-GAAP earnings (Thiagarajan, 1998). In addition, Bradshaw and Sloan (2002) find that managers increasingly identify costs as nonrecurring. Moreover, Bhattacharya et al. (2003) show that costs are sometimes excluded from non-GAAP earnings, even though these costs are directly related to the core business activities of a company. Also, Hoogervorst is worried about the fact that many CEO compensations are based on non-GAAP earnings, because these earnings can be manipulated (Accountant.nl, 2016).

Based on prior literature, I argue that because managers have a lot of freedom in deciding which components to exclude from non-GAAP earnings, they might label more costs as nonrecurring cost to report a higher non-GAAP earnings number and obtain a higher total compensation. This is in line with the rent extraction theory. It is also possible that managers who have non-GAAP performance measures in their CEO compensation plan are more motivated and therefore have better future performance. This indicates that non-GAAP performance-based compensation can be a useful and fairer measure to motivate and compensate managers. This is in line with the efficient contracting theory. Managers deserve a higher compensation because of better future performance. In both cases I expect the use of non-GAAP

(13)

13

performance measures to be positively associated with the level of CEO compensation. Therefore, I state my first hypothesis as follows:

H1: The use of non-GAAP performance measures has a positive effect on the level of CEO compensation

3.2 Predicted Excess Compensation and Future Firm Performance

In order to make a distinction between the two possible explanations presented in paragraph 2.1, I investigate the effect of the use of non-GAAP performance measures in CEO compensation plans on future performance.

If non-GAAP performance based compensation is a useful and fairer measure to motivate and compensate managers, the predicted excess compensation is expected to be positively associated with future performance. If managers manipulate non-GAAP based performance measures to obtain a higher level of compensation, the predicted excess compensation is not expected to be associated with future firm performance. Because it can be both ways, I formulate my hypotheses in null form:

H2: Predicted excess compensation is not related with future stock returns H3: Predicted excess compensation is not

related with future return on assets

(14)

14

4 Research Design

In this section I explain my research design. First, I discuss the methodological approach. Then I talk about the sample I use. In paragraph 2.3 and 2.4 I explain which dependent and independent variables I use. In the last paragraph I explain for which variables I proxy and why.

4.1 Methodological Approach

The research question of this thesis is: What is the effect of using non-GAAP based performance measures in CEO compensation plans on the level of CEO compensation and future firm performance? In other words, are non-GAAP performance measures in CEO compensation a good motivator for managers to act in the best interest of the company, or will managers misuse the freedom they have in determining the non-GAAP earnings to obtain a higher bonus.

To examine the effect of using non-GAAP performance measures in the compensation plan of the CEO on the compensation of the CEO I use an OLS regression model. This model is partly based on the model used by Core et al. (1999). This model is a good starting point for my model, because this model examines the effect of certain variables on CEO compensation and future performance. In image 1 I present the Libby box to test my first hypothesis.

Image 1, Libby Box Hypothesis 1

Compensation Based on

Non-GAAP Performance Measures CEO Compensation

Dummy Variable = 0 or 1

Total CEO Compensation and Non-equity Incentive Plan

Compensation

Independent Variable (X) Dependent Variable (Y)

Conceptual

(15)

15

The independent variable is a dummy variable which indicates whether a firm use non-GAAP performance measures in their CEO compensation plan or not. The dependent variable is CEO compensation measured by total CEO compensation. In addition, control variables are included for firm size and complexity, firm performance, firm risk, investment opportunities and debt. I do not include board and ownership structure variables, because this data is unavailable for most firms in my sample. This results in the following model:

𝑇𝑂𝑇𝐴𝐿 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁𝑡 = β0+ β1𝑁𝑂𝑁_𝐺𝐴𝐴𝑃𝑡+ β2𝑆𝐼𝑍𝐸𝑡+ β3𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁𝑡+

β₄𝑆𝑇𝐷𝐸𝑉 𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁 + β₅𝑅𝑂𝐴_𝑡+ β₆𝑆𝑇𝐷𝐸𝑉 𝑅𝑂𝐴_𝑡+ β₇𝐺𝑅𝑂𝑊𝑇𝐻 𝑂𝑃𝑃_𝑡+ β₈𝐷𝐸𝐵𝑇_𝑡

If non-GAAP performance measures in CEO compensation are a good motivator for managers to act in the best interest of the company, only those variables that justify a higher level of CEO compensation like size, complexity, growth opportunities, firm performance and firm risk should have explanatory power. The dummy variable NON-GAAP should not be significant. If a CEO misuse the freedom they have in determining the non-GAAP earnings to obtain a higher compensation, I expect the coefficient for the dummy variable Non-GAAP to be positively associated with total compensation. It’s also possible that the dummy variable will proxy for underlying economic determinants not captured by the control variables in my model. Therefore, I do a second regression between future firm performance and the predicted excess compensation arising from the dummy variable Non-GAAP. This regression is described in section 6.

4.2 Sample

I randomly selected 100 firms from the database used in the paper ‘Disentangling Managers’ and Analysts’ Non-GAAP Reporting’ by Bentley, Christensen and Gee (2017). I do not use the data from this database itself. I only use the link presented in this database to quickly access the proxy fillings. The database of Bentley, Christensen and Gee consists of all consolidated firms that are available in the CRSP, Compustat, and I/B/E/S universe with fiscal years ending in 2003 through 2012. They exclude real estate investment trusts, because it is customary for these companies to report a standardized, industry-specific funds from operations metric.

(16)

16

They also exclude firm quarters that report extraordinary items and discontinued operations and firm-quarters for which they could not identify an 8-K earnings announcement.

The sample used in this thesis consists of 100 random U.S. firms over a time period of 8 years for which DEF 14A proxy statements were available. I start my sample period in 2006, because since then a new standard is effective, forcing firms to provide more information about their CEO compensation structures. To be able to measure the firm performance for the next three years, I end my sample period in 2013. This result in 800 firm years. The descriptive statistics of all variables can be found in table 2.

4.3 Dependent Variable

For CEO compensation I use Total Compensation as reported in SEC because this number is available on Compustat for most firms. The Total Compensation consist of salary, bonus, stock awards, option awards, non-equity inventive plan compensation, change in pension value and nonqualified deferred compensation earnings and all other compensation. Because in this research I will only focus on the compensation of the CEO, I use the dummy variable Annual CEO Flag from Compustat to filter out directors that are not the CEO of a firm. I also deleted the firms for which there was no data available about the Total CEO compensation. This resulted in 530 firm years.

4.4 Independent Variable

In 2006 the SEC released a new standard for executive compensation and related person disclosure (U.S. Securities and Exchange Commission, 2006). This new standard was published to make proxy statements easier to understand and to provide investors with a clearer and more complete picture of the compensation earned by a company’s chief executive officer. This standard is effective since November 2006. In this standard a new Compensation Discussion and Analysis section is adopted. In this section the compensation structure is explained, including the financial objectives for the firm. Because there is no database available containing information about the financial objectives used in the compensation plans, I hand collect this data. For the hand collection I use the Compensation Discussion and Analysis section in the proxy statement to identify whether firms use non-GAAP performance measures in their CEO compensation plan. I also search on

(17)

17

the most used key words associated with non-GAAP earnings in the entire proxy statements. In table 1 an oversight of the most used key words associated with non-GAAP earnings is provided based on the paper by Isidro and Marques (2013).

Table 1

Keywords Associated With Non-GAAP Earnings

Frequence in %

Non-GAAP Earnings per Share (EPS) 16.26

Earnings before Interest and Tax (EBIT) 16.26

Non-GAAP Net Income 13.24

Earnings before Interest, Tax, Depreciation and Amortization (EBITDA)

12.47

Non-GAAP EBITDA 9.39

Free Cash Flow 9.04

Non-GAAP EBIT 8.97

Non-GAAP Income from Operations 6.80

Other Cash Measures 4.27

Non-GAAP Income from

Continuing Operations

1.68

Non-GAAP Earnings per Share from Continuing Operations 1.05

For the identification I followed the definition of non-GAAP measures used by the U.S. Securities and Exchange Commission (SEC). According to the SEC, A non-GAAP financial measure is a numerical measure of a registrant's historical or future financial performance, financial position or cash flows that: excludes amounts, or is subject to adjustments that have the effect of excluding amounts, that are included in the most directly comparable measure calculated and presented in accordance with GAAP in the statement of income, balance sheet or statement of cash of the issuer; or includes amounts, or is subject to adjustments that have the effect of including amounts, that are excluded from the most directly comparable measure so calculated and presented (U.S. Securities and Exchange Commission, 2002).

Earnings measures like Adjusted Earnings, EBIT, and EBITDA, Operating Income or earnings for which several costs have been excluded are labelled as non-GAAP measures. Also, ratios that were calculated using earnings numbers that are not in accordance with the General Accepted Accounting Principles, like adjusted earnings per share, I labelled as non-GAAP measures. Firms that use non-GAAP

(18)

18

measures in their CEO compensation plan are assigned a value equal to 1 for the dummy variable Non-GAAP. Firms that do not use non-GAAP measures in their CEO compensation plan are assigned a value equal to 0. As shown in table 2, in 60.8% of the firm years in my sample a non-GAAP performance measure is used.

The selection of my sample may influence the results slightly. Compared with the sample used in a recent study about gender-diverse compensation committees and CEO compensation by Bugeja, Matolcsy and Spiropoulos (2016), the average total assets in my sample is lower. The mean of the natural logarithm of total assets in my sample in 7.85 and the mean of the natural logarithm of total assets in the sample used by Bugeja et al. (2016) is 9.58. The mean of the total compensation in my sample, $5,594,000, is similar to the mean of the total compensation in the sample used by Bugeja et al. (2016), $5,248,000. Both the stock return and the standard deviation of stock return are lower in my sample. Also, the return on assets in my sample is slightly lower. The standard deviation of return on assets is in both samples low. The differences found can be explained by the fact that Bugeja et al. (2016) only focus on U.S.-listed firms and I also include smaller firms. In addition, the sample period in Bugeja et al. (2016) starts in 2002 and ends in 2009 and my sample period start in 2006 and ends in 2013. Although, there are some differences, I think my sample is random and I do not think the differences will influence my final conclusions. The differences are small. Also, the results in my first test are significant at the 1 percent levels and the results in my second and third test are clearly insignificant.

4.5 Control Variables

Prior research indicates that larger firms with more growth opportunities will demand for higher quality CEO’s (Smith & Watts, 1992). These managers will require a higher compensation. Therefore, I proxy for firm size and complexity with total assets. This data is available in the Compustat database. I use the natural logarithm of total assets to control for extreme values. This is the most common method used in prior literature to proxy for size. To control for growth opportunities, I use a firm’s year-end market-to-book ratio averaged over the previous five years (Core et al., 1999). The year-end market-to-book ratio is calculated by dividing the closing stock price at year end by the book value per share at year end. This ratio can become extremely large if the denominator is close to zero. To reduce the influence of the few

(19)

19

extreme observations, I winsorized this variable at the 1st_{and 99}th_{percentiles of their}

distribution. This data has been extracted from the Compustat database. The ratio is averaged over the previous five years to control for fluctuations.

Holmstrom (1979) states that when both stock returns and earnings are informative about the performance of a CEO, both measures should be used to determine the compensation of the CEO. Sloan (1993) finds a positive relation between stock returns and CEO compensation. Firm performance has a positive effect on Total CEO Compensation. Therefore, I will control for firm performance. To measure firm performance, I use a market-based measure and an accounting-based measure because these measures do not always capture firm performance in the same way (Fryxell & Barton, 1990). For example, stock returns are more forward looking and include investors’ expectations about the future (Berk & DeMarzo, 2013). The market-based measure I use is stock return and the accounting-based measure I use is return on assets. The stock returns are calculated by dividing closing stock price minus opening stock price plus annual dividends per share by the opening stock price. The closing stock prices and the dividends per share are available on Compustat. Because the data over dividends per share is on a monthly basis, I summed all monthly payments to calculate yearly dividends per share. Return on assets is calculated by dividing net income by total assets. The data about net income and total assets are available in the Compustat database. Both stock return and return on assets are ratios that can become extremely large if the denominator is close to zero. To reduce the influence of the few extreme observations, I winsorized these variables at the 1st_{and 99}th_{percentiles of their distribution.}

Cyert, Kang and Kumar (2002) find that CEO compensation is higher for firms with more firm risk. Therefore, I proxy for firm risk with the standard deviation of stock return and return on assets over the prior five years (Core et al., 1999). The calculation of stock return and return on assets is already explained. To calculate the standard deviation of the sample I use the ‘’n-1’’ method. The standard deviation is a measure of how widely values are dispersed from the average value. This is a common method to control for firm risk.

I also proxy for debt because this can be an important firm characteristic. The debt ratio is calculated by dividing long term debt by total assets. This is a common method to proxy for debt in prior literature. The debt ratio can become extremely small if the numerator is close to zero. To reduce the influence of the few extreme

(20)

20

observations, I winsorized this variable at the 1st_{and 99}th_{percentiles of their}

distribution. Long term debt and total asset numbers are available in the Compustat database.

Table 2

Descriptive Statistics for CEO Compensation its Economic Determinants and the Dummy Variable Non-GAAP

Mean Median St. Dev.

Total CEO Compensation $ 5,594.25 $ 3,791.99 $ 5,553.97

Non-GAAP 0.61 1 0.49

Firm Size 7.85 7.73 1.84

Stock Return 0.13 0.07 0.43

Return on Assets 0.05 0.05 0.08

St. Dev. Stock Return 0.59 0.37 1.68

St. Dev. Return on Assets 0.06 0.03 0.18

Growth Opportunity 2.59 2.24 1.96

Debt Ratio 0.15 0.10 0.18

Observations 503

Definitions of Variables:

Total CEO Compensation: total of salary, bonus, stock awards, option awards, non-equity inventive plan compensation, change in pension value and nonqualified deferred compensation earnings and all other compensation (x1000)

Non-GAAP: dummy variable for whether CEO compensation is based on non-GAAP performance measures

Firm Size: natural logarithm of total assets

Stock Return: calculated by dividing closing stock price minus opening stock price plus annual dividends per share by the opening stock price

Return on Assets: calculated by dividing net income by total assets

Standard Deviation Stock Return: standard deviation of stock return over the prior five years

Standard Deviation Return on Assets: standard deviation of return on assets over the prior five years Growth Opportunity: year-end market-to-book ratio averaged over the previous five years

(21)

21

5 Results

The relation between the level of compensation for the chief executive officer and the dummy variable for non-GAAP based CEO compensation, firm size, growth opportunity, firm performance, firm risk and debt, is examined using an ordinary least squares regression. Initially, the sample consisted of 800 firm years. Because data about the CEO’s compensation was not available for all firm years, several observations have been lost. Also, for some observations, certain data field were empty. After dropping observations for which compensation data was not available and observations with missing data field, the sample consists of 503 firm years. The variables for firm performance, growth opportunities and debt are winsorized at the 1st_{and 99}th_{percentiles of their distribution to control for extreme values. The results}

of the OLS regression are presented in table 3. To show that the results are not significantly affected by winsorization of certain variables, results without winsorization are shown in Appendix I.

The adjusted R-squared found for this model is 56,8%. This means that 56,8% of the variation in total compensation is explained by the model. The regression results show that total CEO compensation is significant positively related to firm size and growth opportunity. Bigger firms and firms with more investment opportunities pay more compensation to their CEO. This is consistent with the regression results of Core et al. (1999). According to Core et al. (1999), larger and more complex firms pay more CEO compensation reflecting their demand for a higher quality chief executive officer. Also, I find return on assets to be significant positively associated with CEO compensation. Although, the coefficient of stock return is positive, this result is not significant. This is in contrast with Core et al. (1999). They do not find a significant relationship between return on assets and CEO compensation and the coefficient for stock return is in their research significantly positive. Even though, these results are not consistent, we could say that firms that perform better, pay more CEO compensation. This is in line with my expectation based on the paper by Sloan (1993). The coefficients for the proxies for firm risk, the standard deviation of stock return and the standard deviation of return on assets, are negative but not significant. In contrast with Cyert, Kang and Kumar (2002) and Haggard and Haggard (2008), I find no evidence that more risky firms pay higher CEO

(22)

22

compensation to compensate managers for uncertainty. Also, I do not find a significant relationship between CEO compensation and the debt ratio.

The coefficient for the dummy variable Non-GAAP is positive and significant related to total CEO compensation. Firms that do use non-GAAP performance measures in their CEO compensation plan, pay more compensation to their CEO. On average, a firm that use non-GAAP performance measures to motivate and compensate the CEO pay $1,752,180 more compensation. This is a difference of 31% compared to the average total compensation of $5,594,000. Therefore, I argue that the results are also economically significant. Using non-GAAP based performance measures in CEO compensation plans has a positive effect on the level of total CEO compensation. There are two possible explanations for this result. The first explanation is that managers might misuse the freedom they have in determining the non-GAAP earnings number to obtain a higher compensation. The second explanation is that managers who have non-GAAP performance measures in their CEO compensation plan are more motivated and have better future performance. Prior research has shown that non-GAAP earnings can give a better reflection of the real performance of a company and is more forward looking than GAAP earnings. This indicates that non-GAAP performance-based compensation can be a useful and fairer measure to motivate and compensate managers. To make a distinction between the two possible explanations I do a second regression between the predicted excess compensation, arising from the dummy variable non-GAAP, and future performance. This regression is described in the following section.

(23)

23

Table 3

Regression of Total CEO Compensation on its Economic

Determinants and Whether CEO Compensation is Based on Non-GAAP Performance Measures Coefficient (t-statistic) Non-GAAP 1752.18*** (5.11) Firm Size 2161.38*** (23.88) Stock Return 11.40 (0.03) Return on Assets 10297.20*** (5.10)

Standard Deviation Stock Return -19.39

(-0.20)

Standard Deviation Return on Assets -126.70

(-0.13) Growth Opportunity 347.23*** (4.05) Debt Ratio 603.96 (0.61) Observations 503 Adjusted R-squared 0.568

*** Indicate statistical significance at the 1 percent levels

This table reports the results from the OLS regression based on the following model:

Stock Return: calculated by dividing closing stock price minus opening stock price plus annual

dividends per share by the opening stock price

Debt Ratio: calculated by dividing long term debt by total assets

𝑇𝑂𝑇𝐴𝐿 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁_𝑡 = β₀+ β₁𝑁𝑂𝑁_𝐺𝐴𝐴𝑃_𝑡+ β₂𝑆𝐼𝑍𝐸_𝑡+ β₃𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁_𝑡 + β₄𝑅𝑂𝐴 + β₅𝑆𝑇𝐷𝐸𝑉 𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁_𝑡 + β₆𝑆𝑇𝐷𝐸𝑉 𝑅𝑂𝐴_𝑡+ β₇𝐺𝑅𝑂𝑊𝑇𝐻 𝑂𝑃𝑃_𝑡+ β₈𝐷𝐸𝐵𝑇_𝑡

(24)

24

6 Excess CEO Compensation and Future Firm Performance

In this section I describe the second test I do in order to make a distinction between the two possible explanations for the results in paragraph 5. In the first paragraph I explain the methodological approach. In the following paragraph I present the results.

6.1 Methodological Approach

The goal of this paper is to analyse whether non-GAAP performance measures in CEO compensation are good motivators for managers to act in the best interest of the company, or that these non-GAAP performance measures can be misused by managers to extract additional payment from the firm. As shown in section 5, firms that do use non-GAAP performance measures in their CEO compensation plan, pay more compensation to their CEO. There are two possible explanations for this result. The first explanation is that managers might misuse the freedom they have in determining the non-GAAP earnings number to obtain a higher compensation. The second explanation is that managers who have non-GAAP performance measures in their CEO compensation plan are more motivated and have better future performance.

In order to make a distinction between the two possible explanations, I test the second and third hypothesis explained in section 3. I perform a second and third OLS regression between the predicted excess compensation arising from the dummy variable non-GAAP and future performance. In image 2 I present the Libby box to test these two hypotheses.

(25)

25

Image 2, Libby Box Hypothesis 2 and 3

I do a regression of predicted excess compensation on both future stock returns and future return on assets, because these measures do not always capture firm performance in the same way (Fryxell & Barton, 1990). I do not control for firm size, firm risk, growth opportunity and debt because the total compensation explained by these variables is already captured in the first test. In the second and third test I only focus on the expected part of total compensation that is explained by the dummy variable non-GAAP. The independent variable is predicted excess compensation. The predicted excess compensation is calculated using the estimated coefficient for the non-GAAP dummy variable reported in table 3. Because I analyse the effect of predicted excess compensation on scaled return measures, I also scale the predicted excess compensation by dividing it by total compensation. This results in a predicted excess compensation ratio for every firm year.

𝑃𝑅𝐸𝐷𝐼𝐶𝑇𝐸𝐷 𝐸𝑋𝐶𝐸𝑆𝑆 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁 𝑅𝐴𝑇𝐼𝑂 = β1𝑁𝑂𝑁_𝐺𝐴𝐴𝑃𝑡

𝑇𝑂𝑇𝐴𝐿 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁 The dependent variable in the first model is average stock return over the next three years after the compensation is awarded. This results in the following model:

𝐹𝑈𝑇𝑈𝑅𝐸 𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁_𝑡= β₀+ β₁𝑃𝑅𝐸𝐷𝐼𝐶𝑇𝐸𝐷 𝐸𝑋𝐶𝐸𝑆𝑆 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁_𝑡

Excess CEO Compensation Future Firm Performance

Predicted Excess Compensation

Average Stock Return & ROA over the Following 3 Years

Independent Variable (X) Dependent Variable (Y)

Conceptual

(26)

26

In the second model, I use the average return on assets over the next three years after the compensation is awarded as the dependent variable. This results in the following model:

𝐹𝑈𝑇𝑈𝑅𝐸 𝑅𝑂𝐴_𝑡 = β₀+ β₁𝑃𝑅𝐸𝐷𝐼𝐶𝑇𝐸𝐷 𝐸𝑋𝐶𝐸𝑆𝑆 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁_𝑡

If non-GAAP performance based compensation is a useful and fairer measure to motivate and compensate managers, I expect predicted excess compensation to be positively associated with future stock return and future return on assets. If managers misuse the freedom they have in determining the non-GAAP earnings number to extract additional payment from the firm, I expect to find no or even a negative relation between the predicted excess compensation and the variables for future firm performance.

6.2 Results

The relation between predicted excess compensation and future firm performance is examined using two ordinary least squares regressions. The dependent variables in these regressions are future stock return and future return on assets. In independent variable in both models is predicted excess compensation. The sample used in this regression consists of the same observations used in the regression described in chapter 5. For firms that do not use non-GAAP performance measures in their CEO compensation plan the predicted excess compensation ratio is equal to zero. Therefore, we drop observation which have a predicted excess compensation equal to zero. Also, for some observations certain data field about the future performance of a firm were empty. After dropping observations with a predicted excess return equal to zero and observations with missing data fields, the sample consists of 294 firm years. The results of the OLS regression are presented in table 4 and 5. To show that the results are not significantly affected by winsorization of certain variables, results without winsorization are shown in Appendix II.

The results in table 4 and 5 show that the predicted excess compensation is not significantly associated with future firm performance. In both models the coefficient for predicted excess compensation is negative, but not significant. The additional payment CEO’s receive when their compensation is (partly) based on non-GAAP performance measures, is not explained by a better performance in the three

(27)

27

years after the compensation is awarded. In other words, managers receive more compensation without performing better. This suggests that managers misuse the freedom they have in determining the non-GAAP earnings number to extract additional payment from the firm, in contrast to the alternative explanation that managers who have non-GAAP performance measures in their CEO compensation plan are more motivated and have better future performance.

Table 4

Regression of Future Stock Returns on Predicted Excess Compensation

Coefficient (t-statistic)

Predicted Excess Compensation -0.005

(-0.07)

Observations 294

Adjusted R-squared -0.003

* Indicate statistical significance at the 10 percent levels

Future Stock Return: average stock return over the next three years after the compensation is awarded

Predicted Excess Compensation: calculated by dividing the estimated coefficient for the non-GAAP dummy variable from model 1 by total compensation

(28)

28

Table 5

Regression of Future Return on Assets on Predicted Excess Compensation

(-1.09)

Observations 294

Adjusted R-squared 0.0006

Future ROA: average return on assets over the next three years after the compensation is awarded Predicted Excess Compensation: calculated by dividing the estimated coefficient for the non-GAAP dummy variable from model 1 by total compensation

(29)

29

7 Conclusion

The purpose of this study was to investigate the effect of the use of non-GAAP based performance measures for CEO compensation on the level of total CEO compensation and future firm performance. The research question of this paper was: What is the effect of using non-GAAP based performance measures in CEO compensation plans on the level of CEO compensation and future firm performance? I find the coefficient for the dummy variable Non-GAAP to be positively and significantly associated with total CEO compensation. Firms that do use non-GAAP performance measures in their CEO compensation plan, pay more compensation to their CEO. I find no evidence that the predicted excess compensation, explained by the dummy variable Non-GAAP, is associated with future firm performance. The additional payment CEO’s receive when their compensation is (partly) based on non-GAAP performance measures, is not explained by a better performance in the three years after the compensation is awarded.

These results suggest that managers misuse the freedom they have in determining the non-GAAP earnings number to extract additional payment from the firm, in contrast to the alternative explanation that managers who have non-GAAP performance measures in their CEO compensation plan are more motivated and have better future performance. This study contributes to the literature about non-GAAP earnings, because it is the first study that directly examines the effect of using non-GAAP based performance measures in CEO compensation plans on the level of CEO compensation and future firm performance. The insight that managers might manipulate non-GAAP measures to extract additional payment from the firm is especially relevant for standard setters and compensation committees. Compensation committees must be cautious to use non-GAAP performance measures in the CEO compensation plan.

A limitation of this study is the limited sample size. This is mainly due to a lack of time. Hand collecting the data about non-GAAP performance measures was very time consuming. Also, data about CEO compensation was not available for all firms. In addition, this study only focusses on US-based firms, because since 2006 it is mandatory for these firms to disclose information about their CEO compensation structure. Based on this research it is only possible to say something about US based firms. Another limitation of this study is that I do not control for board and

(30)

30

ownership structure variables, because this data was unavailable for most firms in my sample. These corporate governance factors might be an important factor with regard to my research question. It could be that firms with low quality board and ownership structures more often use non-GAAP performance measures in their CEO compensation plan. In this case, the manager might be able to force the compensation committee to use non-GAAP performance measures in his compensation plan, and manipulate the non-GAAP earnings to obtain a higher compensation.

For further research, it might be informative to include corporate governance factors, because these factors might influence the results. Also, it would be good to perform the same regressions using a bigger sample. In addition, it would be interesting to examine non-GAAP performance measures in other countries and see if the results found in this thesis also hold in other countries.

(31)

31

References

Accountant.nl. (1996, May 12). Hoogervorst laakt toenemend gebruik alternatieve winstbegrippen. Retrieved June 5, 2018, from

https://www.accountant.nl/nieuws/2016/5/hoogervorst-laakt-toenemend-gebruik-alternatieve-winstbegrippen/

Baik, B., Billings, B. K., & Morton, R. M. (2008). Reliability and transparency of non-GAAP disclosures by real estate investment trusts (REITs). The Accounting Review, 83(2), 271-301.

Banker, R. D., Potter, G., & Srinivasan, D. (2000). An empirical investigation of an incentive plan that includes nonfinancial performance measures. The accounting review, 75(1), 65-92.

Bebchuk, L., & Fried, J. (2004). Pay without performance (Vol. 29). Cambridge, MA: Harvard University Press.

Berk, J., & DeMarzo, P.(2013). Corporate Finance (Third Edition). London: Pearson

Bhattacharya, N., Black, E. L., Christensen, T. E., & Mergenthaler, R. D. (2004). Empirical evidence on recent trends in pro forma reporting. Accounting Horizons, 18(1), 27-43.

Black, D., Christensen, T., Ciesielski, J., & Whipple, B. (2017). Non-GAAP earnings:

A consistency and comparability crisis (Vol. 2759312). Working paper,

https://papers.ssrn. com/sol3/papers. cfm? abstract_id.

Bradshaw, M. T., & Sloan, R. G. (2002). GAAP versus the street: An empirical

assessment of two alternative definitions of earnings. Journal of Accounting Research, 40(1), 41-66.

Bugeja, M., Matolcsy, Z., & Spiropoulos, H. (2016). The association between gender-diverse compensation committees and CEO compensation. Journal of

(32)

32

Core, J. E., Holthausen, R. W., & Larcker, D. F. (1999). Corporate governance, chief executive officer compensation, and firm performance1. Journal of financial economics, 51(3), 371-406.

Crumley, C. R. (2008). A study of the relationship between firm performance and CEO compensation in the US commercial banking industry. Journal of Applied Management and Entrepreneurship, 13(2), 26.

Cyert, R. M., Kang, S. H., & Kumar, P. (2002). Corporate governance, takeovers, and top-management compensation: Theory and evidence. Management

Science, 48(4), 453-469.

Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. The journal of law and Economics, 26(2), 301-325.

Fields, T. D., Rangan, S., & Thiagarajan, S. R. (1998). An empirical evaluation of the usefulness of non-GAAP accounting measures in the real estate investment trust industry. Review of accounting Studies, 3(1), 103-130.

Firth, M., Fung, P. M., & Rui, O. M. (2007). How ownership and corporate

governance influence chief executive pay in China's listed firms. Journal of Business Research, 60(7), 776-785.

Frankel, R., McVay, S., & Soliman, M. (2011). Non-GAAP earnings and board independence. Review of Accounting Studies, 16(4), 719-744.

Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate financial reporting. Journal of accounting and economics, 40(1), 3-73.

Haggard, D. L., & Haggard, K. S. (2008). Executive compensation: does industry risk matter? International Journal of Organization Theory & Behavior, 11(4), 451-470.

(33)

33

Holderness, C. G., & Sheehan, D. P. (1988). The role of majority shareholders in publicly held corporations: An exploratory analysis. Journal of financial economics, 20, 317-346.

Isidro, H., & Marques, A. (2013). The effects of compensation and board quality on non-GAAP disclosures in Europe. The International Journal of Accounting,

48(3), 289-317.

Ittner, C. D., Larcker, D. F., & Rajan, M. V. (1997). The choice of performance measures in annual bonus contracts. Accounting Review, 231-255.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behaviour, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.

Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American economic review, 76(2), 323-329.

Kaplan, R. S., & Norton, D. P. (1992). In Search of Excellence: der Maßstab muß neu definiert werden. Harvard Manager, 14(4), 37-46.

Murphy, K. J. (2000). Performance standards in incentive contracts. Journal of

Accounting and Economics, 30(3), 245-278.

Sloan, R. G. (1993). Accounting earnings and top executive compensation. Journal of accounting and Economics, 16(1-3), 55-100.

Smith Jr, C. W., & Watts, R. L. (1992). The investment opportunity set and corporate financing, dividend, and compensation policies. Journal of financial

Economics, 32(3), 263-292.

Stolowy, H., Lebas, M. J., & Ding, Y. (2013). Financial Accounting and Reporting: A Global Perspective (4e ed.). Boston, Amerika: Cengage Learning Emea.

(34)

34

U.S. Securities and Exchange Commission. (2002). Final Rule: Conditions for Use of Non-GAAP Financial Measures. Retrieved from:

https://www.sec.gov/rules/final/33-8176.htm

U.S. Securities and Exchange Commission. (2006). Executive Compensation and Related Person Disclosure. Retrieved from:

https://www.sec.gov/rules/final/2006/33-8732a.pdf

Young, S. (2014). The drivers, consequences and policy implications of non-GAAP earnings reporting. Accounting and Business Research, 44(4), 444-465.

(35)

35

Appendix I

Table 3 (Without Winsorizing)

Regression of Total CEO Compensation on its Economic

Determinants and Whether CEO Compensation is Based on Non-GAAP Performance Measures Coefficient (t-statistic) Non-GAAP 1714.24*** (4.87) Firm Size 2123.95*** (22.93) Stock Return -167.13 (-0.52) Return on Assets 5910.22*** (3.94)

Standard Deviation Stock Return -32.06

(-0.32)

Standard Deviation Return on Assets 314.94

(0.32) Growth Opportunity 90.60** (2.02) Debt Ratio 322.51 (0.34) Observations 503 Adjusted R-squared 0.544

*** Indicate statistical significance at the 1 percent levels

Stock Return: calculated by dividing closing stock price minus opening stock price plus annual

dividends per share by the opening stock price

Debt Ratio: calculated by dividing long term debt by total assets

𝑇𝑂𝑇𝐴𝐿 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁_𝑡 = β₀+ β₁𝑁𝑂𝑁_𝐺𝐴𝐴𝑃_𝑡+ β₂𝑆𝐼𝑍𝐸_𝑡+ β₃𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁_𝑡 + β₄𝑅𝑂𝐴 + β5𝑆𝑇𝐷𝐸𝑉 𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁𝑡 + β6𝑆𝑇𝐷𝐸𝑉 𝑅𝑂𝐴𝑡+ β7𝐺𝑅𝑂𝑊𝑇𝐻 𝑂𝑃𝑃𝑡+ β8𝐷𝐸𝐵𝑇𝑡

(36)

36

Appendix II

Table 4 (Without Winsorization)

Regression of Future Stock Returns on Predicted Excess Compensation

(-0.07)

Observations 294

Adjusted R-squared -0.003

Future Stock Return: average stock return over the next three years after the compensation is awarded

Predicted Excess Compensation: calculated by dividing the estimated coefficient for the non-GAAP dummy variable from model 1 by total compensation

Table 5 (Without Winsorization)

Regression of Future Return on Assets on Predicted Excess Compensation

(-1.09)

Observations 294

Adjusted R-squared 0.0006

Future ROA: average return on assets over the next three years after the compensation is awarded Predicted Excess Compensation: calculated by dividing the estimated coefficient for the non-GAAP dummy variable from model 1 by total compensation

𝐹𝑈𝑇𝑈𝑅𝐸 𝑆𝑇𝑂𝐶𝐾 𝑅𝐸𝑇𝑈𝑅𝑁_𝑡 = β₀+ β₁𝑃𝑅𝐸𝐷𝐼𝐶𝑇𝐸𝐷 𝐸𝑋𝐶𝐸𝑆𝑆 𝐶𝑂𝑀𝑃𝐸𝑁𝑆𝐴𝑇𝐼𝑂𝑁_𝑡

CEO compensation based on non-GAAP performance measures