The Nature of Employee Stock Option Costs and Other Major Expenditures: An Empirical Examination of Their Link to Executive Compensation Thesis Proposal by Xu Yue Faculty of Business Administration National University of Singapore 1 CHAPTER 1 INTRODUCTION 1.1 Background of the Study Many generally accepted accounting principles (GAAPs) result in accounting treatments in conflict with the underlying economic reality. Employee stock options are one glaring example. Many investors and analysts including the famed Warren Buffet have pointed to the continuing treatment of the cost of employee stock option grants as a non-expense as one of the biggest failures of the accounting profession and regulators. As another example, many including Stern Steward of the EVA fame have argued that R&D expenditures should not be treated as an expense, since such expenditures provide future benefits and should be treated as an asset. The confusion created by the criticism against certain accounting principles and the reporting of income numbers based on different rules raises an interesting question: where do firms themselves stand on the issue? 1.2 Objective of the Study This study attempts to find out the TRUE position taken by the companies themselves on the issue: whether employee stock option and R&D expenditures should be accounted as the cost of the period they happen? I use chief executive officer (CEO) compensation as a tool and try to find out whether firms determine executive compensation based on income number before these controversial items or after them. I focus on the CEO compensation because prior research documents a strong empirical relation between top executive compensation and earnings (Lambert and Larcker, 1987, Defeo et al., 1989, and Sloan, 1993). CEO is responsible for the firm performance, including accounting earnings. To fairly reflect CEO’s effort on firm performance, most firms relate earnings with CEO compensations. However, the earnings used by firms for compensation may be different from the earnings shown in the income statement. For those items controllable by CEO, only when firms do not consider some items in income statement as a true cost of the same period, are their effect eliminated from establishing CEO compensation plan. In this sense, from the relationships between CEO compensation components and the controversial items, we can find out whether firms consider employee stock option and R&D as true cost or not, even though they are expenses against profits in the income statement complying with accounting principle. The negative relationship would tell us firms do think the item as true cost of the current year, vice versa. While agency theoretic models related total pay to performance, my study focuses on the components of total compensation. Compensation agreements with executives 2 may be structured and administered so that components of total pay, such as salary, bonus and stock-based awards, respond differently to observable performance measures, such as market and accounting returns. Thus, I estimate an empirical model that relates CEO compensation to firm performance through bonus, cash compensation as well as total pay. 1.3 Potential Contributions of the Study By finding out whether controversial items affect CEO compensation, this study can disclose the firms’ true standpoint about whether these items are cost or not. Their support or objection for the current accounting measurement would be clear. This may give accounting regulators and researchers some useful information about the firms’ attitude about financial disclosure. They could find whether information about those items has relevance for financial statement users and thereafter better improve accounting reporting principle. Another contribution of the study is to start a new way to do accounting research on firms’ position on some controversial items. It could be useful in promoting future research in this area. Finally, the study expends the research on executive compensation. Prior research usually examines the relationship between executive compensation and firm performance and their mutual effects. This study further discovers something new by using the same modeling. 1.4 Organization of the Thesis The remainder of the paper is organized as follows: Chapter Two gives an overview of the accounting treatments for the two special items and their corresponding controversy. A review of empirical research on the executive compensation factors’ effects are also presented in this part. Chapter Three identifies the development of model conducted in this study and several hypotheses are put forward. The empirical research methodology, data sources, and sample selection procedure are outlined in Chapter Four. Chapter Five will present the empirical findings and analyze research results. Finally, Chapter Six will sum up the main findings in this thesis, present the implications of the study, explain the weakness of the study, and suggest areas for future research. 3 CHAPTER 2 LITERATURE REVIEW 2.1 Overview This chapter starts with a review of the two special items: employee stock option and R&D expenditures. Their accounting treatments and the corresponding controversy are presented. This is followed by a review of empirical research on the factors affecting executive compensation. 2.2 Employee Stock Option 2.2.1 Incentive Compensation Plans A stock option is a right to buy a number of shares of stock at or after a given date in the future (the exercise date) at a price agreed upon at the time the option is granted (usually at the current market price or less). Employee stock options provide a non-cash substitute for part of the wage compensation the firm must provide to attract and retain employees. Especially, those new entrepreneurial firms may not be able to provide enough cash for compensation. Instead, they can offer stock options (Malkiel and Baumol, 2002). The major motivational benefit of stock option plans is that they direct managers’ energies toward the long-term, as well as the short-term, performance of the company. Given that agency problems1 exist, the principal’s concern is to devise an appropriate incentive and monitoring system that induces the agent to act accordingly to the principal’s desire (Starks, 1987). Stock compensation gives executive incentive to keep stock price high. Haugen and Senbet (1981) also showed that the use of executive stock options in executive compensation may emerge as solutions to the agency problems analyzed by Jensen and Meckling (1976). Stock option is also a way to motivate employees. Businesses that want to attract and retain top-notch talents look at options as a primary vehicle. By stock options, they can accomplish this goal with limited cash resources but significant stock appreciation potential --- including startups and turnaround situations (Plishner, 1993). In price-declining stock market, many stock options are for restricted stock (stock that vests over a period of time) as a tool to retain employees (Kroll, 2002). Option incentives have been widely used by new companies and innovative managements. 1 Jensen and Meckling (1976) defined an agency relationship as a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. The separation of ownership and control in corporations produced a condition where the interests of the owner and that of the ultimate manager might and often did, diverge (Berle and Means, 1932). As the managers in the pursuit of self-interest would not always act in the interests of the shareholders, agency costs arise. Another source of agency cost might come from the divergence between managers and shareholders in risk preference. Because of the decreasing utility for wealth and the large amount of agent capital that is dependent on the company, managers are assumed to be risk averse. On the other hand, the owners can easily diversify away risk and are risk neutral. 4 According to reports “The State of Employee Stock Options - 2002”, an electronic survey conducted jointly by WorldatWork and Sibson Consulting/The Segal Co, stock options are alive and well at least for now. Two major examples: “The goals of option programs are unchanged. Most of the nearly 300 companies responding to the survey report they use option programs to attract and retain talent, motivate performance, focus employee attention on organizational performance, and create a culture of ownership, just as their peers in 2000 did.” “The specific employee groups eligible for options, the percentage of the firm's employees eligible for options, and the value of options in the overall compensation mix changed relatively little in the past three years.” 2.2.2 Accounting Measures for Employee Stock Option Accounting for employee stock options remains a highly emotive issue and the self-interest based lobbying that was evident in the US in the early 1990s continues internationally today. Issued in October 1995 by Financial Accounting Standards Board, Statement of Financial Accounting Standards (SFAS) No. 123, "Accounting for Stock-Based Compensation" established a fair value based method of accounting for stock-based compensation plans and encourages entities to adopt that method of accounting for their employee stock compensation plans. This pronouncement also allows an entity to continue to measure compensation cost for those plans based on Accounting Principles Board Opinion No. 25 (APB 25) (issued in October 1972), "Accounting for Stock Issued to Employees" and disclose the pro forma net income and net income per share in footnotes as if the fair value method had been applied in measuring cost. The fair value of a stock option (or its equivalents) of a publicly-traded company shall be estimated using an option pricing model (e.g., Black-Scholes2), which includes expected volatility on prices of the underlying stocks. However, SFAS 123 does not provide specific guidance on the methodology for determining fair value for such an arrangement or the measurement date on which the fair value of the equity instrument is determined. Under APB 25, compensation cost is determined based on the market value of the stock and is recognized for awards of shares of common stock or stock options to directors, officers and employees of the Company and consolidated subsidiaries only 2 A model for pricing call options based on arbitrage arguments developed by Fischer Black and Myron Scholes in 1973. Black-Scholes Option Pricing Model is the basic instrument used for the determination of a stock option's estimated value at grant. It uses the stock price, the exercise price, the risk-free interest rate, the time to expiration, and the expected standard deviation of the stock return. Some of the assumptions of the model are: the price of the underlier is lognormally distributed with constant mean and volatility; there are no transaction costs or taxes; markets trade continuously, i.e. there are no sudden jumps in prices; the risk-free rate is constant and the same for all maturities. 5 if the quoted market price of the stock at the grant date (or other measurement date, if later) is greater than the amount the grantee must pay to acquire the stock. 2.2.3 Controversy on the Accounting Measurement Under current accounting rules, companies do not have to count the cost of stock options against profits as compensation costs. Instead, the costs of such option grants -- as calculated using standard pricing models -- must merely be cited in a footnote in companies' annual financial reports. The footnotes were the result of heated battles between options-inclining technology companies and accountants. In 1995, accounting regulators tried to force companies to expense employee stock options against profits. However, the complaint and objections from companies made the regulators retreat and only require companies to disclose the "pro forma" effect of options on earnings in a footnote. Some firms have elected to apply APB 25 and related interpretations in accounting for their employee stock options, only disclosing the pro forma net income in footnotes as if compensation cost had been determined based on SFAS 123. Very few companies will voluntarily adopt the fair-value method, as it always results in compensation expense. They believe that the models available to estimate the fair value of employee stock options do not provide a reliable single measure of the fair value of employee stock options. Moreover, such models required the input of highly subjective assumptions, which can materially affect the fair value estimates. Therefore, they measure compensation cost using the intrinsic value method3 and do not recognize compensation expense on the issuance of its stock options because the option terms are fixed and the exercise price equals the market price of the underlying stock on the grant date. Some support the SFAS 123: Because employee stock options have value for employees, they represent a cost to employers, and thus should be recognized as an expense in the financial statements. As famous Warren Buffet said, "If options aren't a form of compensation, what are they? If compensation isn't an expense, what is it? And, if expenses shouldn't go into the calculation of earnings, where in the world should they go?" Though Warren Buffett and other critics suggest that the income statement should reflect an "expense" to the firm measured by the cash-equivalent value of the options, Malkiel and Baumol (2002) point out two problems with this view. They said that firstly there is no way to measure that "cost" -- the value of the options at the time they are issued -- with reasonable precision. Even the Nobel Prize winning Black-Scholes formula does not provide reliable estimates for longer-term options, 3 The intrinsic-value method under APB 25: Compensation costs attributable to stock option and similar plans are recognized based on any difference between the quoted market price of the stock on the date of grant and the amount the employee is required to pay to acquire the stock. 6 such as those lasting six months to one year, and market prices often differ substantially from predicted values. The second problem with "expensing" stock options is that they can have both positive and negative effects on share prices. They tend to reduce earnings per share when measured on a "fully diluted basis." On he other hand, stock prices preponderantly benefit from the issue of employee options. Because the market believes that the positive benefits of the options make the “pie” grow faster than the dilutive effect that shrinks current shareholders' percentage piece of the “pie”. Quinn (2002) refutes Malkiel and Baumol’s points in arguing that stock option would be accounted for similarly to salary compensation. He objects unscrupulous use of options, earnings timing, and opaque accounting standards and argues that if options are good in the long run for corporations, they will still be good when accounted appropriately. The controversy that attends the share-based payment debate centers largely on when this expense should be recognized in the accounts, and the basis on which measurement should be made, that is the how. The problem is that there is no clear, objective method by which to calculate fair value. Typically firms use a (traded) options pricing model such as the Black-Scholes to estimate employee stock option expense, but employee stock options are quite different to traded options in that they cannot be transferred and they have characteristics that may well lead to early exercise. Currently, most stock options are treated as a gift from the gods; that is, as entailing no "cost". The employee is richer, but the company, without recording an expense, is none the poorer. The "costless" compensation has hardly discouraged boards from ladling out options (Lowenstein, 1995). Accounting analysts estimate that option grants issued by high-tech companies in 1997 would have slashed net income at those companies anywhere from 10% to 100%, if they were expensed. As MacDonald and McGough (1999) pointed out, if Wall Street started paying more attention to options' implicit cost, stock prices of some companies could come under pressure. It has been argued that a requirement to charge employee stock options to the profit and loss account could hit US company profits by approximately 9%, the impact on the high-technology alone being closer to 33% (Grey, Cotter and Barnes 2002). A new study by Bear Stearns & Co, similarly, indicates that option grants would completely wipe out corporate profits and operating income at some fast-growing high-technology companies, which tend to rely heavily on generous option grants to attract top talent. For example, counting the cost of stock-option grants would have erased fiscal 1997 operating income at companies such as Silicon Graphics and Digital Equipment. According to Grey, Cotter and Barnes (2002), the accounting treatment of employee stock option also has implications for wider corporate finance issues. "As companies 7 increasingly rely on stock options to compensate workers, not expensing them inflates corporate earnings and thus overstates companies' performance, which leads to higher stock prices, which leads to more valuable options," says Pat McConnell, senior managing director at Bear Stearns. Firms that grant employee stock options and thus avoid a charge against income, artificially boost earnings. Managers may favor share repurchases over conventional dividend payments as they buoy up share price with positive implications for the fair value of managers' stock options. Forcing firms to expense employee stock option costs could affect the roles of dividend, compensation and financing policies. If volatile share prices, short vesting periods and low revenues/large losses are typical of high-technology firms in general, it is not surprising that reporting firms are vigorously opposed to recognizing employee stock option expense. Even for firms without the twin characteristics of share price volatility and low revenues/incomes, expensing employee stock options is likely to impact on reported earnings substantially. Broadly speaking, the spirit of the responses to the FASB is that preparers are opposed to recognition of employee stock option expense, and users may favor it. Recognition of share-based payments is currently the subject of much debate. An important dimension to the discussion is whether the benefits to users of expensing employee stock options are greater than the possibility that firms would be constrained from using them to attract and retain employees (especially for higher risk and emerging firms) (Zeff, 1978). This is a key issue worth addressing. “Although its decision should rest --- and be seen to rest --- chiefly on accounting considerations, it must also study --- and be seen to study --- the possible adverse economic and social consequences of its proposed actions”. 2.3 R&D Expenditures 2.3.1 R&D and Its Accounting Measurement Research and development (R&D) is defined in the Federal Acquisition Regulation (FAR) in terms of four elements: 1. Basic research is directed toward increase of knowledge in science; it seeks a fuller knowledge or understanding of the subject under study, rather than any practical application. 2. Applied research attempts to exploit the potential of scientific discoveries or improvements in technology, materials, processes, methods, devices, or techniques, and to advance the state of the art. 3. Development is the systematic use of scientific and technical knowledge in the design, development, test, or evaluation of a potential new product or service or of an improvement in an existing product or service. 8 4. Systems and other concept formulation studies are analyses and study efforts either related to specific R&D efforts or directed toward identifying desirable new systems, equipments or components, or modifications and improvements to existing systems, equipments, or components. Professional accountants have tried to determine how to treat research and development (R&D) expenditures for many years. At first, the accepted practice was to defer R&D costs, but with changes in revenue laws in the 1950s, many companies began to write off R&D expenditures immediately. Based on the belief that a direct relationship between R&D costs and specific future revenue generally has not been demonstrated, the Financial Accounting Standards Board issued SFAS No. 2, "Accounting for Research and Development Costs" in 1974, specifies that all R&D charges should be written off immediately. This preserved the standard that most US companies use today. GAAP require that R&D costs be expensed in the period incurred. Some costs can be appropriately capitalized, carried as assets, and depreciated to research expense. They include research facilities, equipment, and machinery. 2.3.2 Controversy on the Accounting Measurement The controversy about the accounting measurement of R&D still exists. The point lies in whether R&D should be considered as a kind of investment or an expense. National Association of Insurance Commissioners (NACI) supports R&D costs to be expensed, “because the future benefit is rarely certain or measurable and the future period over which deferred costs might be allocated is usually arbitrary, deferral is not consistent with the conservatism concept included in the Statement of Concepts.” By contrast, many persons look R&D as one form of investment. Newman (1988) argued that at the macroeconomic level, the large amounts of money spent on R&D each year enhance profits for companies in later years; therefore, costs should somehow be matched against future revenues. Horwitz and Ronald Zhao (1997) examines whether the capital market ignores the mandated expensing of firms' R&D to make its own judgments about current and future cash flows. The study finds that cash flow statements reconstructed to assume different proportions of expensed and capitalized R&D can result in a better association of cash flow variables with security returns than that provided under the GAAP requirement of 100% expensing. That is, the market does regard a good amount of R&D outlays as investments, even though these outlays are actually classified as operations expenses under U.S. GAAP. Another question arisen now is the inconsistency on accounting for costs required completing a purchased R&D program after it is acquired. It makes little sense to capitalize and amortize the acquisition costs of R&D while expensing the subsequent 9 costs of completion. The lack of consistency is likely to cause confusion among investors. If companies that purchase R&D programs capitalize and amortize these projects while other companies immediately expense internal R&D activities, earnings for the two companies are not comparable. Thus, differential treatment of R&D based on internal programs versus external purchases potentially confuses interpretation of operating results. In a word, as Clem and Jeffrey (2001) indicated, the key point in the controversy is whether R&D meets the definition of an asset, and, if so, whether information about that asset has relevance for financial statement users. R&D is undertaken with the expectation that it will provide future benefit to the firm. While not every R&D project initiated generates benefits, the expectation is that on average some will, resulting in assets for the firm. Yet meeting the definition of an asset and demonstrated relevance are not adequate for recognition in the financial statements. To be recognized, an item must have a relevant attribute that is reliably measurable. Widely accepted valuation models for R&D have not been developed, and questions about valuation approaches are critical to current deliberation. 2.4 Executive Compensations 2.4.1 Components of Executive Compensations In the organization of a public company, control rights are vested with the directors who authorize the exercise of control rights by the chief executive officer (CEO) and other top executive officers. The board of directors develops a compensation policy that enables the firm to attract, retain and motivate the CEO and other top executives through negotiated compensation agreements. Thus, compensation can be used to motivate directors and other senior executives towards meeting a company's or its shareholders' goals. The chief executive officer’s compensation is usually discussed by the compensation committee of the board of directors. The percentage of incentive compensation applied to the CEO compensation is an important signal about how the board regards the CEO’s performance. Forbes magazine publishes annually a list of the compensation of hundreds of CEOs, compared with the profits for which they are presumably responsible. The general message is that compensation is indeed highly relative to performance. An executive’s total compensation package consists of three components: (1) salary, (2) benefits (principally pension and health benefits, but also perquisites of various types), and (3) incentive compensation. The three components are interdependent, but the third is specifically related to the management control function and approved by the board of directors according to the executive’s performance. 10 Incentive compensation plans can be divided into (1) short-term incentive plans, which are based on performance in the current year, and (2) long-term inventive plans, which relate compensation to the longer-term accomplishments. The representative of the former one is bonus, and the representative of the latter is stock option. Much empirical research investigating relations between executives’ compensation and their firms’ market and accounting performance has focused on the cash components (which is defined as base salary plus annual bonus) of total compensation because cash compensation represented a material portion of a CEO’s yearly compensation and dominated other forms of compensation during the periods studied (Murphy 1985, Lambert and Larcker 1987). In recent years, the stock-based share of total compensation has increased significantly and now makes up about one-third of executives’ total compensation. 2.4.2 Executive Compensations Determinants CEO compensation is related with his/her performance. How to measure CEO performance? How do these measurements determine the compensation? A variety of factors have been investigated to find their roles in determining executive compensations. Firm Performance Accounting researchers are interested in the information properties of market and accounting performance measures in the context of executive performance evaluation. Empirical research has investigated how the sensitivity and precision of the performance measures (Banker and Datar, 1989) influences the relative weights placed on market and accounting returns (Lambert and Larcker, 1987, Sloan, 1993), whether the relative influence of accounting returns has changed over time (Bushman, Engel, Milliron, and Smith 1998a), how the properties of reported earnings affect pay-performance sensitivities (Bushman, Engel, Milliron and Smith 1998b), and whether relative performance evaluation (Holmström, 1982) is used to remove noise in the performance measures (Antle and Smith, 1986; Gibbons and Murphy, 1990; Janakiraman, Lambert and Larcker, 1992). The evidence is that both security returns and accounting earnings are used to set compensation contacts. Both measures are informative of the agent’s unobservable actions in the agency context. Lambert and Larcker (1987) document a positive association between the cash compensation of CEOs and their firms' contemporaneous earnings performance. In his research, he uses accounting return on equity (ROE) and security market return (RET) as performance measures. Sloan (1993) provides evidence in support of the hypothesis that earnings-based incentives help shield executives from fluctuations in firm value caused by market-wide factors that are beyond their control. The paper demonstrates that earnings reflect firm-specific changes in value, but are less sensitive 11 to market-wide movements in equity values. The hypothesis is shown to explain cross-sectional variation in the use of earnings-based incentives. Researchers have found that accounting returns have significant incremental explanatory power to market returns in estimations that relate cash compensation to performance. Bushman et al. (1998a) found that, while the weight placed on earnings in determining cash compensation has not declined over time, the importance of earnings relative to other information captured by stock returns has declined over time. He also found that changes in pay-earnings sensitivities are positively associated with changes in price-earnings sensitivities (earnings response coefficients) and the noise in market returns and negatively associated with the noise in earnings (Bushman et al., 1998b). Most of these studies consider only cash compensations. Very few studies have examined relations between total compensation or non-cash components of compensation and firm performance measures. Murphy (1985) found significant positive relations between the change in total compensation and stock returns and between the change in cash compensation and stock returns, but he found no relationship between the change in stock-based compensation and stock returns. Baber et al. (1996) found significantly positive coefficients on market return in their total pay, cash (salary and bonus), and long-term pay (option, restricted stock and LTIP pay-outs) equation, but they found a significantly positive coefficient on accounting return in their cash equation only. Veliyath (1999) examines the linkage between two separate components of executive compensation (i.e. cash compensation and stock options) and market return performance, among a selected sample of US pharmaceutical company CEOs and COOs. The study finds that market returns were not instrumental influences on the levels of both compensation components. Cash Flow and Accruals As for accounting-based performance, the investigation of the degree to which components of earnings are informative about managerial actions, and therefore priced in the managerial labor market, parallels the inquiry into the role of cash flows and accruals in firm valuation. Clinch and Magliolo (1993) examine the possibility that components of earnings relate differently to the CEO’s performance and do not enter into the compensation function in the same way. Results indicate that income from discretionary transactions accompanied by cash flow effects is reflected in the CEO compensation function. Natarajan (1996) investigates the role of components of earnings in CEO compensation contracts. It argues that shareholders will use components of earnings as additional performance measures whenever the components provide information, over and above earnings, about managerial decisions. Results indicate that earnings and cash flow measures together have a better association with cash compensation 12 paid to CEOs of U.S. companies than aggregate earnings alone. The evidence also suggests that current accruals and cash flows from operations are aggregated for performance evaluation. The study by Gaver and Gaver (1998) also investigates the role of alternative earnings components in the CEO cash compensation function. It finds that cash compensation is significantly positively related to above the line earnings, as long as results are positive. Similarly, nonrecurring transactions that increase income flow through to compensation, while nonrecurring losses do not. In a word, compensation committees distinguish among the transactions comprising net income in determining executive pay. However, accounting numbers can be “managed” or “manipulated” via changing accounting methods and can be subject to falsification by management. Discretion reduces the effectiveness of accounting-related transactions as performance measures (Firth, Tam, and Tang, 1999). Size Murphy (1998) provides a review of the compensation literature. Many studies, including most of those cited above, report a strong link between firm size and managerial rewards (Roberts, 1956; Ciscel and Carroll, 1980; and Agarwal, 1981), and size (book value of total assets) is a major determinant of executive compensations (Chung and Pruitt, 1996 and Veliyath, 1999) Ryan and Wiggins (2001) find that the executive compensation has a positive relation with firm size. Bliss and Rosen (2001) provide several theoretical explanations for this positive correlation. Compensation can be used to motivate effort among lower-level managers who view the top job as spoils that go to the winner of an intra-firm tournament (Nalebuff and Stiglitz, 1983; Rosen, 1992). A bigger firm represents a larger tournament, and therefore demands a commensurate high prize. Also, managing a bigger firm might involve more skill and job complexity than managing a smaller firm. Compensation is thus used to solve the adverse selection problem in choosing a competent manager. These explanations for the correlation between compensation and size imply that better managers control bigger firms rather that making a firm larger should increase the compensation of an existing manager. CEO Age The age of executives, as a proxy variable for general training levels and/or experience, may affect their remuneration. However, the direction of this relationship cannot be determined with certainty. The relationships are also different among the components of compensations. 13 Bryan, Hwang, Lilien (2000) find an unexpected negative relationship between CEO age and stock option. One potential explanation is that older CEOs facing imminent retirement prefer to minimize the uncontrollable effects of the market-wide factors on their wealth. Another explanation is younger CEOs are frequently found in high-growth firms that tend to rely heavily on stock option awards. Under the premise that older and younger CEOs have the incentives to focus on short-term outcomes, Ryan and Wiggins (2001) find a concave relation between cash bonus and age, suggesting that firms pay the youngest and oldest managers less short-term bonus. They also find a negative linear relation between options and age, but a concave relation between restricted stock and age. On the surface, these two relations seem counterintuitive since firms appear to be reinforcing short-term tendencies of older CEOs in the case of options and both older and younger CEOs in the case of restricted stock. In contrast, Chung and Pruitt (1996) cannot find any effect of CEOs age on their compensation, which rejects the hypothesis that age, as a proxy variable for general training or experience, affects executive compensation. It seems that the positive relationship between age and income does not apply to executive labor market. Executive positions are characteristically different from other jobs and general experience may not increase a CEO’s total productivity. CEO Tenure CEO's tenure, defined as the number of years he has been CEO, is considered to be related with incentive compensations. However, prior researches have got different conclusions about the relationship. Consistent with previous research (Murphy, 1986; Barro and Barro, 1990), Clinch and Magliolo (1993) observe that for operating earnings and the discretionary cash-flow-based earnings component, the length of CEO tenure is negatively associated with the strength with which these items enter the compensation function. Chung and Pruitt (1996) find the number of years as CEO is significantly and positively related to executive stockholding. This is consistent with the view that the cumulative value of incentive compensation received in the form of stock options may be larger for executives with longer employment histories. Large tenure means more experience, which deserves greater levels of compensation. Ryan and Wiggins (2001) find CEO tenure is negatively related with stock options at the 10% significance level. This finding is inconsistent with an entrenchment argument, but consistent with the premise that CEOs with long tenures have accumulated more stock and therefore require less equity-based pay than newer CEOs. 14 Ownership Structure When CEOs hold a large fraction of their firm’s equity, their demand for further stock-based compensation is like to be reduced, since the interests of CEOs and the shareholders are already relatively aligned (Jensen and Meckling, 1976). Furthermore, CEOs are typically unable to diversify the risk associated with their wealth (Smith and Watts, 1992). The constraint affects their tolerance for additional risk, that is, they prefer cash compensation to stock option. Bryan, Hwang and Lilien (2000) find a significant negative relationship between CEO stockholding and stock option, and the negative association is somewhat stronger for mid-cap and small-cap firms. Ryan and Wiggins (2001) use the percentage of a firm's shares held by the chief executive to measure CEO stock ownership, and find a negative relation with all forms of incentive compensation. It is consistent with the notion that as CEOs own more stocks, their interests become more aligned with shareholders and there is less need for incentive compensation. Growth Many prior papers have documented that high-growth firms are more likely than low-growth firms to use stock-based compensation rather than salary and bonus compensation in the pay packages of their senior executives (Lewellen, Loderer, and Martin, 1987; Clinch, 1991; Smith and Watts, 1992; Gaver and Gaver, 1993). Growth firms are expected to have high investment opportunities. Gaver and Gaver (1995) use market-to-book assets, market-to-book equity, R&D expense to assets, and total return variance to measure the investment opportunity and find that firms with abundant investment opportunities pay higher levels of total compensation to their executives. Moreover, executives of growth firms receive a larger portion of their compensation from long-term incentive compensation (such as performance awards, restricted stock grants, and stock option grants), while those of non-growth firms receive a larger portion of their pay from fixed salary. They conclude that higher manager-shareholder information asymmetry in growth firms leads these firms to emphasize long-term incentive compensation in order to motivate managers to act in their shareholders' interests and reduce agency costs. It is also consistent with long-term incentive contracts motivating managers to take actions to seek out and exploit new investment opportunities. Industry Industry also has considerable effect on the structure of executive compensations. Smith and Watts (1992) examine industry-level data and find that high-growth firms are more likely than low-growth firms to use stock-based compensation plans. Executive compensation studies suggest that many of these same factors are also associated with the relative weight placed on accounting measures. Ely (1991) finds 15 that the choice between alternative accounting measures varies by industry, suggesting that these measures must be tailored to reflect industry-specific value drivers and competitive environments. 16 CHAPTER 3 MODEL DEVELOPMENT 3.1 Overview This chapter introduces the development of model conducted in this study. The variables measures and hypotheses development are also described in detail. 3.2 Model Development The confusion from the criticism against certain accounting principles is pervasive in the accounting research field. This raises a question: where do firms themselves stand on the issue? My way to test the firms’ position is to find whether firms determine executive compensation based on income number before these controversial items or after them. I focus on the CEO compensation because early research documents a strong empirical relation between top executive compensation and earnings (Lambert and Larcker, 1987, Defeo et al., 1989, and Sloan, 1993). CEO is responsible for the firm performance, including accounting earnings. To fairly reflect CEO’s effort on firm performance, most firms relate earnings with CEO compensations. However, components of earnings relate differently to the CEO’s performance (Clinch and Magliolo, 1993; Natarajan, 1996; and Gaver, 1998). The earnings used by firms for compensation may be different from the earnings shown in the income statement. One reason is firms may not consider some items in income statement as a true cost of the same period. As a result, these items are not taken into consideration when evaluating CEO performance, even though they are expenses against profit in the income statement complying with accounting principles. Another reason is these items are uncontrollable by CEO and thus not related with CEO performance. Option cost and R&D expenditures do not belong to this type of items, since CEO is involved in deciding employee stock option as well as the R&D project of the firm. In this sense, we can find out whether firms consider employee stock option and R&D as true cost or not through the relationships between CEO compensation and these controversial items. A negative relationship may tell us that firms do think the item as true cost of the current year. On the other hand, if significant negative association does not exist, one big possibility is that firms do not look the item as an expense. 3.3 Measures and Hypotheses Development 3.3.1 CEO compensations The agency theoretic models that motivate empirical research in executive 17 compensation derive total compensation as a function of available performance measures (Holmström 1982, Banker and Datar 1989). Total compensation paid to an executive includes cash payments such as salary (fixed cash payments) and bonus (variable cash payments), and the ex ante or grant date value of stock-based awards including stock options and restricted stock. James Cathro (1996), a principal with the international compensation consulting firm William M. Mercer, commented: “There seems to exist a public perception that option grants are made over and above what is already adequate compensation in the form of salary and bonus. This is rarely the case. Typically, today’s compensation committee will establish a mix of pay that includes the estimated value of the long-term incentives as part of a total compensation package.” When compensation is paid in two or more forms, their relative mix depends on factors that influence the marginal costs and benefits to the firm and the marginal utility to the employee (Woodbury, 1983). In prescribing an approach for compensation committees, William White (1992), National Director of Compensation Services for Ernst and Young, stated: “Determine the overall pay philosophy that underlies the company’s pay mix and levels, its pay-for-performance guidelines and the use of various vehicles, balancing executive safety and risk factors. …Once this assessment has been made, pay specific attention to the level of predictable pay (base salary) versus that of variable pay (annual and long-term incentives).” These observations suggest that firms determine the total compensation required to satisfy the executive’s reservation utility (based on alternative opportunities in the executive labor market) and the mix of compensation components that will provide appropriate incentives to motivate and retain the executive. Compensation agreements with executives may be structured and administered so that components of total pay, such as salary, bonus and stock-based awards, respond differently to observable performance measures, such as market and accounting returns. Thus, I estimate an empirical model that relates total compensation to firm performance through the cash bonus and stock-based shares, as well as total pay. 3.3.2 Independent Variables My independent variables determining executive compensation are divided into three parts. Executive compensation should be determined by the costs and benefits of the executive's actions during a period. I thus first measure the costs by employee stock option and R&D expenditure. I pick them up from profit in order to find their links with compensation. Corresponding to the cost, I also need to include benefits of the executive's actions during the same period. I measure the benefits by such accounting measures as cash flow (before R&D expenditure) and accruals, and such market measures as shareholder returns in one (shorter term) and three years (longer term). 18 My cost part of independent variables includes only employee stock option cost and R&D expenditure. There are other cost items such as goodwill amortization, restructuring charges and extraordinary item that may affect executive’s compensation in different ways. I drop them from my list of cost variables due to a desire to obtain a large sample because not many firms have those cost items at the same time. They are not the objects in this study and their influences on the CEO compensation have been incorporated in the benefits items. Other than the costs and benefits, compensation is supposed to be determined by the personal characteristics of the executive and firm characteristics. This forms the third part of independent variables. The personal characteristics that I include in my study are CEO age, CEO tenure and CEO ownership. The firm characteristics are size, growth opportunities and industry. Employee Stock Option The share-based payment debate centers largely on whether this is an expense and should be recognized in the accounts of the current period. If firms consider option as a true cost of the current year, they would reduce the option from the profit when determining CEO compensation. As a result, there should be a negative relationship between CEO compensation and the option cost. Otherwise, no significant relationship will be found. Since till now, most firms would not like to expense employee stock option (only two companies in the S&P 500 include employee stock option grants as an expense in their income statements), I make the following hypothesis: H1: There is no significant negative relationship between employee stock option and the bonus or the total compensations. R&D Expenditures The controversy about the accounting measurement of R&D lies in whether R&D should be considered as a kind of investment or an expense. If firms think the R&D expenditure is related with future benefit and should be looked as one kind of investment, they would like to capitalized it rather than expense it in the current year. Even though GAAP asks firms to expense the R&D, firms may not consider it as a true cost when evaluating CEO’s performance. Consistent with the objections to implemented GAAP, the hypothesis is: H2: Firms look R&D as investment and do not expense it when determining CEO compensations. Performance of the Firm 19 To reduce the biases for the relationships between CEO compensation and controversial items, I control for as many other factors that may also affect executive compensation as possible. The intention is to control the effects from other factors and capture the true effect of items examined. Both accounting-based measure and market-based measure have conceptual and methodological weaknesses as measures of performance. Accounting measures are subject to management manipulation and may not correlate significantly with firm value (Lubatkin & Shrieves, 1986). On the other hand, a company's stock market performance is also sensitive to numerous factors beyond the control of management, and so market-based measures may also be an inadequate indicator of CEO performance (Jensen & Murphy, 1990). To avoid the biases inherent in using either method alone, we used both accounting-based and market-based measures of performance in this study. Both measures were based on rates of return to facilitate comparisons to each other. Operating cash flow and accruals represent main part of accounting earnings. Although CEO compensation is positively related with earnings according to previous empirical analysis, components of earnings do not enter into the compensation function in the same way (Clinch and Magliolo, 1993; Natarajan, 1996; Gaver, 1998). Since accruals can be “managed” or “manipulated”, it may play a less important role to the CEO’s performance than cash flow. It leads to the following hypotheses: H3a: CEO compensation increases with cash flow and accruals. H3b: Cash flow has more weight than accruals in determining CEO compensation. To measure the market-based performance, I use one-year and three-year market return to shareholders. The return includes stock price appreciation and dividends. I expect higher market return is one indication of CEO good performance and therefore a higher compensation could be observed. H4: CEO compensation increases with market return. Growth Opportunity Previous research documents greater use of stock-based pay for firms with more extensive growth opportunities in their investment opportunity set (Smith and Watts, 1992; Gaver and Gaver, 1993). According to Gaver and Gaver (1995), the investment opportunity set is unobservable, and it is likely to be imperfectly measured by any individual proxy. Therefore, they use an ensemble of variables to measure the investment opportunity. I follow the approach and include the market to book ratio of equity and sales growth rate as proxies for growth opportunity in this study. Market to book ratio is the growth measures used most frequently by earlier researchers in 20 previous studies. It is the ratio of the market value of the firm (share closing price times outstanding shares) to the book value of its equity and relies on current stock price to assess the firm’s growth potential. Another growth measurement is firm’s growth rate in net sales (Bathala and Rao, 1995). I use 3-year least-squares4 growth rate of sales. According to prior research, I predict: H5: CEO compensation is positively related with firm growth opportunity. Other Factors Determining CEO Compensations Other exogenous variables are included in the compensation equations based on economic theory and prior empirical research. They include firm size (sales revenue in the fiscal year), CEO age and tenure, CEO ownership (the percentage of a firm's shares held by the chief executive), and industry effect (industry dummy variables are used to control for the possibility of spurious correlations among the variables operating through industry effects). 4 The least-squares trend is a commonly used growth indicator. It has the following advantages: firstly, it takes into account each of the observations under consideration, unlike geometric (or compound) growth rates, which only consider the first and last observations; secondly, it measures the stability of observed growth; finally, unlike the arithmetic average of annual growth rates, it takes into account the sequence of different growth rates over time. 21 CHAPTER 4 RESEARCH DESIGN AND DATA 4.1 Methodology 4.1.1 Basic Regression Many researches (Lambert and Larcker, 1987; Sloan, 1993 and Dechow et al., 1994) have documented a positive association between the cash compensation of CEOs and their firms’ contemporaneous earnings performance. Accordingly, to examine the relationships between CEO compensations and the controversial items, the following cross-sectional OLS regression is estimated for the CEO compensation of 1999 on the contemporaneous firm performance of 1999. CEOCOMP 99 0 1OPT99 2 RD99 3CF99 4 ACCRUAL99 5 MRKYR 1 6 MRKYR 3 7 MBV 8 SALESGROWT H 9 SIZE 10 AGE 11TENURE 12CEOSH 13 IND1 ... 20 IND8 <Equation 1> Since the effect of factors may be varied among the different components of CEO compensations, I examine the components of CEO compensations separately. The CEOCOMP99 has three conditions: (1) Cash compensation: Salary + Bonus; (2) Stock Option; and (3) Total Pay: Salary + Bonus + Stock Option + Other Annual Pay. The description of independent variables: OPT --- Option Cost. The fair market value of all awards of stock-based compensation at the grant date, computed as the difference of pro forma net income and reported net income, deflated by total assets5. RD --- Annual R&D expenditure to book value of total assets, deflated by total assets. CF --- Cash flow before R&D expenses, deflated by Total Assets. ACCRUAL --- Total accruals defined as income before extraordinary items minus operating cash flows, deflated by Total Assets. MRKYR1 --- 1-year market return to shareholders, used to measure market performance of the firm. MRKYR3 --- 3-year market return to shareholders, used to measure long term market 5 In my regression, the option cost, R&D expenditure, cash flow and accruals are deflated by total assets in order to enhance cross-sectional comparability and reduce heteroscetisticity. 22 performance of the firm. MBV --- Growth or investment opportunities, measured by market value of the firm to the book value of its total assets. SALESGROWTH --- 3-year least-squares growth rate of sales, used to measure firm’s growth opportunities. SIZE --- Firm size, measured by the natural logarithm6 of the sales of the fiscal year. AGE --- Age of CEO who is in office. TENURE --- Number of years since the time of CEO appointment. CEOSH --- CEO ownership, measured by the percentage of a firm's shares held by the chief executive. IND1~8 --- Industry dummy variables are used to control for the possibility of spurious correlations among the variables operating through industry effects. The number of dummy variables is decided by the number of industries my sample belongs to. According to four-digit SIC codes of 10 industry categories7, all sample firms come from 9 categories. 4.1.2 Sensitivity Tests 4.1.2.1 Time-series Relations Murphy (1985) observed that cross-sectional differences in characteristics of firms and their executives are likely to influence relations between compensation and performance. Failure to control for these factors would cause an omitted variable problem and lead to biased estimates in empirical models that relate compensation levels to performance. He controlled for this problem by using a first-difference specification of the compensation variable, reasoning that the lagged compensation values would be informative about the characteristics of the firm and the executive. In the first difference model, changes in compensation are related to surprises in performance. It suggests that the changes in compensation persist over time. While there may be a dynamic response of pay to surprises in performance, it does not imply a permanent change in pay except under the assumption that the manager’s true productivity evolves as a random walk. Anderson et al. (1999) included the lagged value of total pay as an independent variable in the pay-performance equation and 6 The objective of taking the natural logarithm of sales is to mitigate the influence of extreme observations. Four-digit SIC codes partition firms into 10 industries: 1. Agriculture, Forestry, and Fishing; 2. Mining; 3. Construction; 4. Manufacturing; 5. Transportation, Communications, and Utilities; 6. Wholesale Trade; 7. Retail Trade; 8. Financial, Insurance, and Real Estate; 9. Service; 10. Public Administration. 7 23 found that compensation responses to performance shocks are not persistent. Baber et al. (1999) found that the sensitivity of compensation to earnings levels is concave in persistence. These findings indicate that earnings changes are essential to properly specify relations between CEO compensation and earnings performance. To specify the relations between CEO compensation and controversial items as well as resolving the missing variable problem sited by Murphy (1985), I regress the change of compensation on the change of option cost, change of R&D, and change of performance for year 1999 and 1998. CEOCOMP 0 1OPT 2 RD 3 CF 4 ACCRUAL 5 MRKYR 1 6 SIZE <Equation 2> Similar to the first equation, I examine three conditions of CEOCOMP change: cash compensation, stock option and total pay. 4.1.2.1 One-year Lagged Relations One possibility is that the actual reward is based on the degree to which company financial goal and individual performance objective established early in the fiscal year have been achieved. Accordingly, the bonus and stock-based compensation are paid at the beginning of the next year for each eligible executive. It implies that earnings performance relative to bonus plan bounds can be inferred from ex post bonus payments in the following year (Holthausen, Lacker and Sloan, 1995). Therefore, I regress the current year’s bonus and stock option on the previous year’s performance. The model examined represents a one-period lagged relationship in which company performance in 1998 is the independent variable and the CEO compensation (bonus and stock option) from 1999 is the dependent variable. BONUS 99 OPTION 99 0 1OPT98 2 RD98 3CF98 4 ACCRUAL98 5 MRKYR 1 6 MRKYR 3 7 MBV 8 SALESGROWTH 9 SIZE 10 AGE 11TENURE 12CEOSH 13 IND1 ... 20 IND8 <Equation 3> 4.2 Sample Selection and Data Sources I use cross-section data for the study. In this session, the sample selection method and data are explained. I present the data sources and selection method in the first part. It is then followed by descriptive statistics on the executive compensation, controversial items, and firm and industry characteristics. 24 4.2.1 Data Sources I use US publicly held companies for my sample. CEO compensation data and CEO information are collected from the COMPUSTAT Executive Compensation files. Except option cost, all the company information and financial statement data are obtained from the COMPUSTAT annual industrial files with supplements from DATASTREAM for missing data. Option costs of the companies are obtained from SEC online EDGAR (Electronic Data Gathering, Analysis, and Retrieval system) archives. Some selection criteria are considered when collecting data: (1) The firm has both accounting data and CEO compensation data reported in COMPUSTAT files. (2) The firm has a December fiscal year-end each year. This criterion is used because Smith and Pourciau (1988) show that firms with December fiscal year-end and non-December fiscal year-end have significant differences in financial characteristics. (3) SEC online EDGAR archives contain the firm’s annual report (10-K form), from which employee stock option cost data are collected8. (4) The firm has both employee stock option cost and R&D expenditure different from zero for the sample year. (5) For sensitivity tests of Time-series Relations and One-year Lagged Relations, the firm must have the same CEO who was in office for both 1998 and 1999. Finally, 407 firms are included in the cross-sectional regression, and 301 firms are included in the sensitivity tests. 8 Most companies choose to apply APB 25 and disclose the pro forma net income in footnotes as if compensation cost had been determined based on SFAS 123. Since the option prices were greater than or equal to the market prices at the date of grant, most companies have not recognized the compensation cost for stock options. Thus, the difference between pro forma net income and net income as reported in income statement can be considered as compensation expense for stock option (after-tax effect) amortized in the current period. In this study, I use this difference from 10-K form to approximate the employee stock option cost of a company. 25