two essays in corporate finance

This is in contrast to other performance-based compensation, e.g., restricted stock, long-term incentive plans LTIP and equity holdings that have a symmetric payoff with respect to the s

TWO ESSAYS IN CORPORATE FINANCE

The Ohio State University

2003

Business Administration

Copyright by Natasha A Burns 2003

induced by options Finally, we examine the market reaction to the announcement of a restatement It is more negative for those restating firms in which the exercise of options was greater during the misreported period, providing support for the idea that options provide a

‘camouflage’ for insider trading

The second essay, presented in Chapter 3, examines the role of cross-listed stock in foreign acquisitions of U.S firms Recent research states that there are important disclosure and legal implications for foreign firms that cross-list on a U.S exchange By cross-listing, a foreign firm reduces its cost of an acquisition made with equity by enhancing the rights of its minority investors and by decreasing barriers to ownership of its shares by U.S investors

Cross-listed firms using equity to finance an acquisition pay, on average, 10% less than cross-listed firms paying with cash, as measured by the target’s premium around the

non-announcement of the acquisition Despite this benefit, cross-listed firms use equity less often than U.S firms because of factors affecting foreign firms but not U.S firms, e.g., home bias and investor protection Using a sample of forty-four cross-listed bidders acquiring U.S targets with equity, the essay shows that cross-listed firms from countries with poorer investor protection use equity less often than those from countries with greater investor protection Moreover, they pay a higher premium when using equity Thus, even after bonding to the U.S legal regimes via its cross-listing, investors are still wary of the firm’s home country legal protections We test Hansen’s (1987) hypothesis that equity financed acquisitions enable the bidder to share the risk with the target that the bidder overpaid We find that acquirers are more like ly to use equity in acquisitions involving targets with greater growth opportunities that are more difficult to value, supporting Hansen’s hypothesis Finally, we find that while non-cross-listed firms use favorable exchange rate movements to bid more aggressively, cross-listed firms do not

ACKNOWLEDGMENTS

I would like to thank my advisor René Stulz and my committee members Jean Helwege and Andrew Karolyi for their guidance and insightful feedback that helped make this dissertation possible

I also thank Craig Doidge, Jim Hsieh, Jan Jindra, Anil Makhija, Christof Stahel, Siew

Hong Teoh, Karen Wruck and seminar participants at Auburn University, The Ohio State

University and The University of Georgia for helpful comments and suggestions

Finally, I am grateful to W.C Benton, Steve Buser and Jean Helwege for their words of encouragement and advice; Melanie Roy for help with data; to my family for their support; and to

my precious daughter, Shefali, for keeping the joy in my life during this time

VITA

August 24, 1971 … Born – Columbus, Ohio

1993 Bachelor of Science in Computer Information Science,

Ohio State University, United States

1996 Master of Business Administration, Michigan State University,

United States

FIELDS OF STUDY Major Field: Business Administration

Concentration: Finance

TABLE OF CONTENTS

ABSTRACT ii

ACKNOWLEDGMENTS iv

VITA v

LIST OF TABLES ix

LIST OF FIGURES xi

CHAPTER 1 1

CHAPTER 2 3

2.1 Introduction .3

2.2 A primer on restatements 6

2.3 Literature review and development of hypotheses 8

2.3.1 The effect of option compensation .8

2.3.2 Convexity .12

2.4 Data and Methodology 15

2.4.1 Introduction of sample .15

2.4.2 Summary Statistics .16

2.4.3 Characteristics of restating firms 17

2.4.4 Discretionary accruals 19

2.5 Measures of CEO compensation .21

2.5.1 Option sensitivity .21

2.5.2 Sensitivity of restricted stock and equity holdings 24

2.5.3 Long term incentive plans .24

2.6 The effect of compensation .24

2.6.1 Convexity .30

2.6.2 Abnormal compensation .31

2.7 Announcement reaction .32

2.8 Conclusion 34

CHAPTER 3 37

3.1 Introduction .37

3.2 The Determinants of financing .40

3.2.1 Growth opportunities and the limit of debt financing .40

3.2.2 Risk sharing and signaling 41

3.2.3 Benefits of cross-listing .41

3.2.4 The role of exchange rates 45

3.3 Data and Results .46

3.3.1 Variables .49

3.3.2 U.S vs cross-listed acquirers 52

3.3.3 Cross-listed acquirers 53

3.3.4 Premiums 56

3.4 Conclusion 61

CHAPTER 4 63

LIST OF REFERENCES 65

APPENDIX A: Tables 71

APPENDIX B: Figures 101

APPENDIX C: Estimation of discretionary accruals 104

APPENDIX D: Estimation of equity incentive deviation from optimal equity incentives 105 APPENDIX E: Probit used in Heckman estimation 108

LIST OF TABLES

Table 1: Restatements for the period 1995 - 2002 71

Table 2: Measurement of variables 74

Table 3: Characteristics of restating firms 75

Table 4: Time series of discretionary accruals around the first year of alleged manipulation .76

Table 5: Firm-year characteristics .77

Table 6: Logit regressions of determinants of restated year 78

Table 7: The effect of Gamma .82

Table 8: Deviations from optimal compensation 84

Table 9: Announcement reactions .85

Table 10: Distribution of sample by year and country 87

Table 11 Description of financing .90

Table 12: Probit regressions of factors affecting method of payment for cross-listed and U.S bidders of U.S firms 91

Table 13: Probit regressions on the use of equity in acquisitions by cross-listed firms 92

Table 14: Cumulative abnormal returns 95

Table 15: Cross-sectional weighted least-square regression results of premiums to shareholders of US targets of cross-listed and non-cross listed foreign acquirers 96

Table 16: Cross-sectional weighted least-square regression of premiums to target shareholders of U.S and cross-listed acquirers 97

Table 17: U.S Target Premiums of cross-listed acquirers 98

Table 18 Determinants of equity incentives 107 Table 19: Probit regression used in Heckman estimation 108

LIST OF FIGURES

Figure 1 Timeline for a restating firm that announced in January 2000 that it is restating

financial statements for fiscal years 1998 and 1999 101 Figure 2 Median Discretionary Accruals (DCA) of restating and non-restating firms in the three years around the initial year of GAAP violation 102

Figure 3 Cummulative Abnormal Returns in the 120 days around the announcement of the

restatement 103

CHAPTER 1

INTRODUCTION

Recently, there has been an increase in the number of restatements issued by larger firms Previous literature has focused on market-based incentives affecting a manager’s choice to use aggressive accounting practices that might result in restatements However, it does not seem likely that a manager would make such a choice absent personal incentives because managers face personal risks for making these choices In the first essay, presented in Chapter 2, we examine the CEO’s personal incentives to make such choices

We examine the role played by the different components of the CEO’s compensation package, with a focus on options We focus on options because the increase in restatements has coincided with the increase in the use of options as a tool to align the interest of managers with shareholders More importantly, options have an asymmetric payoff with respect to stock price that mitigates the downside risk of a stock price movement This is in contrast to other

performance-based compensation, e.g., restricted stock, long-term incentive plans (LTIP) and equity holdings that have a symmetric payoff with respect to the stock price

The results show that the CEO’s incentives from options are positively associated with the likelihood of misreporting Despite the fact that equity, restricted stock and long-term

incentive plans expose managers to the downside risk of a stock price movement, these

components of the CEO’s compensation package are not significantly associated with

misreporting Results are also consistent with the notion that options provide ‘camouflage’ in that the market reaction to the announcement of the restatements is significantly and negatively associated with the exercise activity of the CEO in the misreported period

Chapter 3 presents an essay on the factors affecting the method of payment in a foreign acquisition In addition to the factors that U.S acquirers must take into account in an acquisition

of a U.S target, foreign firms must also take into account their home-country legal environment, disclosure standards and the home bias effect When a firm cross-lists, it effectively “bonds” itself to the U.S legal regime and disclosure standards Because U.S investors are accustomed to greater legal and disclosure standards than offered by most countries, a U.S investor may not want to accept the shares of a cross-listed firm if cross-listing does not fully “bond” the firm to U.S legal standards

Results indicated that after cross-listing, home country legal standards, but not disclosure standards, still affect the likelihood of the foreign firm using equity Foreign firms from countries with poorer legal standards use equity less often They are also charged a higher premium by U.S target shareholders for their shares We examine other factors that determine the use of equity and affect the premiums in foreign acquisitions U.S firms We find that non-cross-listed foreign firms bid more aggressively when the dollar has depreciated relative to the home country currency, in contrast to cross-listed firms that do not Another important finding is that the growth

opportunities of the target affects the choice of equity for cross-listed firms, in that they are more likely to use equity when the target has greater growth opportunities This supports Hansen’s (1987) risk-sharing hypothesis that equity may be used in an acquisition to cause target

shareholders to share in any post-acquisition revaluation effects

manipulate their accounting numbers. 1 Back in 1997, when large option grants were beginning

to become more prevalent, L Dennis Kozlowski, former CEO of Tyco (currently under

investigation for defrauding shareholders), characterized options as a “free ride… a way to earn megabucks in a bull market with a hot company.”2 Is the earnings management to which Levitt is speaking motivated by firms trying to look like a “hot company” in order to earn those

“megabucks” while in a bull market? Is the recent increase in the number of restatements

coinciding with the increased use of large option grants evidence of what these men said? In this paper, we examine restatements that are due to ‘purposeful’ accounting choices in order to understand whether and how management’s incentives, through their compensation contracts, affects the likelihood of engaging in unusual accounting practices that result in restatement

1

Levitt (1998)

2

Lublin, J., Executive Pay (A Special Report)- View From the Top: A CEO discusses his unusual pay package

with a shareholder activist The Wall Street Journal, April 10, 1997

If aggressive accounting affects stock prices, then executives, whose wealth is tied to equity valuation, will be motivated to maximize their wealth through aggressive accounting.3

Bebchuk, Fried and Walker (2002) argue that options enable managers to extract rents, encouraging focus on short-term price at the expense of long-term fundamental value by allowing easy exit strategies Holding option compensation causes executive wealth to be a convex

function of the stock price Because of this convexity, executive’s exposure to the stock price from options changes with firm value Thus, increasing the stock price benefits management through their option holdings, but the magnitude of their loss is limited in the event of a decline in the stock price Management is rewarded in the good times, but not hurt as much in the bad times This asymmetric feature of options may cause focus on the short-term in periods in which it is more profitable to do so If shareholders have a short-term focus, then they, too, may benefit from encouraging management to focus on the short-term, especially in periods of speculation (Bolton, Scheinkman, and Xiong (2003))

We compare restating firms and non-restating firms in the S&P 1500, over the period from 1994-2002 Our contribution is in examining the personal incentives that CEOs have in relation to their compensation to make aggressive accounting choices In particular, we examine equity-linked compensation: options, stock holdings, restricted stock and long term incentive plans, as these are more likely to cause managers to focus on stock price We find that the

sensitivity of compensation from options, as measured by the change in value of the option for a percentage change in stock price (delta), is greater in restated firm-years than non-restated firm-years This provides evidence consistent with the hypothesis that incentives from option

compensation encourages misreporting that results in restatement We also examine the effect of the convexity of the CEO’s option portfolio, by examining gamma (the second derivative of the

3

See Sloan (1996); also Teoh, Welch and Wong (1998) for information on how accruals affect price

option with respect to stock price) because the delta of the option does not directly take into account the convexity of an option Misreporting may be more likely when there is a larger change in the value of options for a change in the stock price, as captured by gamma We find that misreporting is more likely when the convexity is greater

In contrast to options, equity and restricted stock performance have a symmetric relation

to the stock price; thus, the downside risk of using aggressive accounting is greater for these relative to options However, aggressive accounting may still be beneficial to equity holdings if managers can sell before a decline in the stock price Restricted stock and long-term incentive plans (LTIPs) may also reduce incentives to use aggressive accounting because they make

executive compensation a function of longer-term firm value We did not find evidence to support the hypotheses that LTIPs and restricted stock reduce the propensity to misreport This may be due to the smaller portion of compensation that long-term incentive plans and restricted stock have relative to other compensation We find that equity holdings do not encourage misreporting

to the extent that options do

We examine whether restating firms have ‘abnormal’ equity incentives We estimate

‘normal’ or optimal incentives using Core and Guay’s (1999) model of the CEO’s optimal equity incentives, a mode l that estimates equity incentives from theoretically and empirically motivated factors We estimate abnormal option and stock incentives as the deviation of the firm equity incentives from optimal incentives We find significantly abnormal incentives from option

compensation in restated years

We test the implication from the Bolton, et al (2003), that managers are more likely to use aggressive accounting in periods in which it is most profitable to do so We find weak

evidence consistent with this, finding that restated years were more likely in 1998 and 1999,

periods with inflated market valuations However, an aggregate market valuation measure does not explain restatements

Finally, we examine whether the market reacts differently to firms based on the CEO’s exercise of options during the time of alleged manipulation If options provide the camouflage suggested by the rent extraction view, then the market will under-react to the exercise of options

in the periods which had to be restated We find the announcement reaction is more negative for firms with greater exercise activity in those periods, providing evidence that options provide the camouflage suggested by proponents of the rent extraction view

We also examine how restating firms differ from non-restating firms with regards to characteristics identified in other studies We find evidence that restating firms have higher leverage and greater market valuations than non-restating firms in the S&P 1500 We find that restating firms use discretionary accruals more aggressively than non-restating firms in the first year that had to be restated, confirming similar results in Richardson, Tuna and Wu (2002) However, we do not find that their use of accruals is significantly greater in the years leading to the restatement

The organization of the paper is as follows: Section 2.2 provides a primer on

restatements Section 2.3 reviews literature on compensation, and develops hypotheses Section 2.4 introduces and reports characteristics of our sample; Section 2.5 describes the measurement of key variables Section 2.6 presents analysis of factors that may enhance or mitigate incentives to engage in aggressive accounting that may lead to restatements Section 2.7 presents analysis of the market reaction Section 2.8 concludes

2.2 A primer on restatements

A restatement is a firm’s admission to the fact that its accounting statements are not in conformance with Generally Accepted Accounting Practices (GAAP) They are generally

associated with a strong market reaction at the time of announcement, although speculation that a firm might have to restate is often in the news before the announcement of the restatement (Palmrose, Richardson and Scholz (2001)) While restatements can be filed due to a change in accounting practices, merger and acquisition, stock split, or one-time error, we exclude these from analysis, because they are not the result of purposeful accounting practices outside of GAAP and are generally not associated with significant market reactions (Palmrose, et al., (2001)) When a firm restates, it may restate one or more financial statements previously filed with the SEC These may be previously filed 10-Qs or 10-Ks

A restatement involves the misreporting of an item (or number of items) on a financial statement An example of misreporting involving revenue recognition is as follows:

Xerox improperly recognized $1.5 billion in pretax earnings over

the 1997 to 2000 fiscal years, by recording revenue from lease

contracts immediately, as opposed to over the life of the contracts 4

The use of aggressive accounting can be used, as in the Xerox example, to increase revenue in a particular period by recording revenue too soon However, in times when earnings are good, management may also choose to set aside reserves- that is, recognize revenue in later periods rather than the current period This may be accomplished by taking larger than needed one-time charges. 5

One can see the similarities between the above example and the use of accruals in which revenues might be shifted forward (reversed) and expenses delayed (forward).6 As a

spokeswoman for Xerox, Christa Carone, said regarding the revenue recognition, “It’s just going from one place (in Xerox’s books) to another”.7 However, the use of accruals does not lead to

restatements as they occur within the confines of GAAP The previous examples violate GAAP and therefore, when identified, result in a restatement Thus, the study of restating firms is useful

in evaluating whether mode of compensation gives incentives to managers to make aggressive accounting choices because it is less questionable that these are aggressive choices

Implicit in the arguments that aggressive accounting can be used to affect stock price is the assumption that manipulative accounting fools the market The large impact of a restatement

on firm value is an indication of the fact that the market was not able to sort out the aggressive accounting Moreover, it is hard to argue that the market appropriately adjusted Enron, Qwest and Worldcom’s stock prices for their undisclosed debt, swap-like sales8 and other forms of

misreporting Given that the market uses accounting information to infer firm valuation and given the link between a manager’s compensation and stock price performance, it is reasonable to investigate the relation between compensation and accounting choices

2.3 Literature review and development of hypotheses

2.3.1 The effect of option compensation

Adoption of performance-sensitive compensation plans and the adoption of stock option plans are often greeted positively by the market (Morgan and Poulsen (2001)) This is attributed

to the idea that it is necessary to tie management’s wealth to shareholder wealth (Jensen and Meckling (1976)) in order to reduce agency conflicts The dependence of a manager’s wealth on firm performance causes uncertainty in management’s compensation As a result, risk-averse management may forgo risky projects Smith and Stulz (1985)) assert that stock options may be used to mitigate the effects of risk-aversion by making a manager’s compensation a convex

8

A swap sale occurs when two firms in similar industries swap the use of a revenue generating asset For instance, in the telecom industry, firms might swap the use of phone network connections This allows each firm

to record revenue See Berman, D., J Angwin and C Cummins, As the Bubble Neared its end, Bogus Swaps

Padded the books The Wall Street Journal, Dec 23, 2002

function of firm value Thus, traditionally, options are viewed as an influential means to align manager’s interest with shareholders by inducing management to take on more positive net present value risky projects

The assumption that options are used solely to align the interests of management with

shareholders has come under scrutiny with the increase in the number of restatements by large firms Bebchuk, Fried and Walker (2001), proponents of the rent extraction view, argue that options enable management to extract rents in the form of excessive compensation Especially important to the ability of management to extract rents is their ability to make it less identifiable that rent extraction is taking place

A camouflaging feature of options, introduced by Bergstresser and Philippon (2002), is that they enable managers who are exercising stock options in order to take advantage of

information asymmetry to pool with those exercising stock options for liquidity reasons.9

Bebchuk and Bar-Gill (2003) argue that the ability of management to take advantage of

information asymmetry depends on the amount of shares that managers may sell relative to the number of shares that management would be able to sell for liquidity purposes (or diversification purposes) Since managers tend to sell the equity acquired through stock options (Ofek and Yermack (2000)) as they vest, then managers who are manipulating can pool themselves with managers who are sellin g options for liquidity reasons If the market is unable to undo the aggressive accounting during the periods in which managers are selling, then managers benefit from selling at a price that is higher than it would be otherwise

It is important to remember that a manager’s utility from holding options increases with the stock price even if they are not planning to immediately exercise (Core and Guay (1999)) The potential for options to increase wealth comes from their sensitivity to the underlying stock

9

Consistent with this hypothesis, they find that managers exercise more options in periods with greater

discretionary accruals

price Gao and Shrieves (2002) find that earnings management intensity is significantly and positively associated with the sensitivity of the CEO’s option portfolio to stock price While, they attribute this to the manager’s attempt to exploit the non-linearity in the payoff of option

compensation, they do not test for this empirically

H1: Aggressive accounting associated with restatements is more likely in firms in which executives have more wealth affected by options

If managers who take advantage of information asymmetry are pooled with those

exercising for liquidity reasons, then the market will not react significantly to the exercise of options in the manipulated period Thus, the announcement reaction to the restatement will be related to the aggregate amount of exercise activity of executives during the period of alleged manipulation That is, the market will not have incorporated into the stock price, at the time of

manipulation, the exercise of options that occurred in the misreported years This argument rests

on the assumption that other information is released during the misreported year that is consistent with the financial statements that are misreported at fiscal year end.10 If the aggressive accounting continues to work after the release of the misreported financial information, then we expect the exercise activity to continue until the announcement of the need for restatement, so we also look

at the subsequent year

H2: The announcement reaction to a restatement is related to the exercise activity of executives in the period for which misreporting occurred

10

For instance, quarterly financial numbers should approximately add up to annual numbers

Bolton, Scheinkman and Xiong (2002) model a world in which rational current

shareholders want management to maximize the short-term stock price, even at the expense of long-term fundamental value In this model, there exists a speculative market in which a subset of investors is overly optimistic This creates an option-like feature for incumbent shareholders- one

in which they can sell to the new “gullible” shareholders.11 As a result, in a speculative market, incumbent shareholders compensate managers in such a way as to increase their focus on

maximizing stock prices in the short-term, even at the expense of long-term value Bolton et al (2003), argue that shareholders do this by granting stock options that are easier to sell, allowing executives to sell shares to overoptimistic investors within the laws of insider trading In contrast

to the rent extraction view, managers are not exploiting information asymmetry with rational shareholders for their own benefit, but are induced by their compensation structure and the speculative components of the stock market In this way, executive’s interests are aligned with a subset of investors- those with a short-term horizon who wish to focus a manager’s attention to a particular time

The Bebchuk and Bar-Gill (2003) model also has implications for the timing aspects of misreporting Their model implies that the relative weight the market puts on future estimated growth versus current cash flows enables firms to have more opportunity to benefit from

misreporting This may be more likely to occur in periods of speculation, or excessive optimism

by the market They also conclude that misreporting is more likely to occur when executives have more shares that they can sell in a particular period This is likely to coincide with the number of exercisable options, as these can be sold under the guise of liquidity purposes, as described earlier

11

More specifically, firm value is the sum of its fundamental value plus a project that may be overvalued in a speculative market Rational investors price the probability of short -term profits from selling to irrational investors - a speculative component of the firm’s stock price Managers are encouraged to exploit the irrational investors

H3: Aggressive accounting is more likely to occur during speculative periods, or in periods with higher market valuations

H4: Aggressive accounting is more likely when managers have a greater number of

Moreover, the asymmetric payoff structure of options mitigates a negative wealth effect

in the event of a stock price decline If managers engage in aggressive accounting, and that aggressive accounting has the effect of causing the stock price to increase, then both shareholders and managers will benefit If the aggressive accounting stops working, management will be less affected by the decline in share price than if their exposure had been through stock holdings.13Therefore, options partly insulate managers from the negative effects of their discretion.14 We

14

The Sarbanes -Oxley Act of 2002 attempts to address these concerns, forcing managers to pay back incentive based compensation earned during the 12 month period following the filing of a financial statement that had to

be restated (Section 304)

expect that managers will be more likely to misreport when incentives are more sensitive to stock price, as measured by gamma (the second derivative of the option with respect to stock price).15,16

H5: Aggressive accounting is more likely to occur when a small change in price causes a larger change in incentives or when convexity is greater

In contrast to options, we expect that performance-sensitive compensation with a

symmetric payoff structure will reduce the incentive to use aggressive accounting With a

symmetric payoff structure, a CEO may be less likely to engage in aggressive accounting,

because the downside risk on CEO compensation is greater Both the executive’s stock holdings and restricted stock have a linear payoff with respect to the stock price Because of its linear relationship with the stock and its restrictions imposed by vesting requirements, restricted stock may intensify executives aversion to risk taking, even more so than stock (Bryan, Hwang and Lillien (2000))

Both restricted stock and equity may also increase the manager’s time horizon relative to the short-term focus encouraged by options Restricted stock increases the manager’s time

horizon due to vesting restrictions, while equity increases the time horizon because it does not offer the ease of selling that options do (Bebchuk et al., (2002), Bergstresser and Philippon (2002)) Shleifer and Vishny (2002) describe the importance of a manage r’s time horizon,

suggesting that managers might want to “get out” of their ownership positions due to closeness to retirement, ownership of stock options or desire to cash-out while market valuations are high

15

Jensen (2001) finds evidence that non-linear payoff structures provide incentive to manipulate accounting profits When far from a performance threshold, incentives to reach that target are diminished; when close to it or above the target, incentives are greater If manipulation of accounting statements affect prices, then in the context

of equity compensation, non-linearities will induce similar behavior

16

Previous studies measure convexity as the derivative of the option value with respect to volatility (Vega) However, Vega measures incentives to increase volatility We focus on incentives to increase price

Thus, the effect of a manager’s equity ownership is mitigated by his personal time horizon in which case short-term focus will not be significantly reduced To the extent that an executive can sell his equity holdings prior to an adverse change in stock price, there is still a benefit to equity from using aggressive accounting However, the benefit of aggressive accounting will not be as valuable to a holder of equity as to a holder of options because there is greater exposure to the downside risk from the equity holdings

H6: The exposure of executive wealth to restricted stock and long-term incentive plans will be negatively related to the propensity to use aggressive accounting relative to options

H7: Equity holdings do not encourage misreporting to the extent that options do

There is mixed evidence that performance based compensation has given management

incentive to take action to influence its compensation (based on the firm’s stock price) that is unrelated to managers impact on firm value In regards to firms that are subject to AAERs (Accounting and Auditing Enforcement Releases),17 Dechow, Sloan and Sweeney (1996) find that insider trading does not occur more often than in control firms prior to the infraction

However, Beneish (1999) finds that management is more likely to exercise their stock

appreciation rights18 and be net sellers of their holdings in periods for which earnings are

overstated Each of Beneish’s (1999) and Dechow, et al.’s, (1996) samples focus on periods prior

to 1993,19 periods in which the use of options were not as prevalent We may find that using data after 1993 clarifies the role of options with respect to restatements

In contemporaneous work, Kedia (2003) finds that the top five executives exercise significantly more options in the two through five years prior to the announcement of a

restatement, consistent with the idea that they are taking advantage of information asymmetry

2.4 Data and Methodology

2.4.1 Introduction of sample

We start with firms in the ExecuComp database ExecuComp covers the S&P small cap (600), mid cap (400) and large cap indices (500) We use announcements of restatements, for full

or partial years, identified by the General Accounting Office over the period from 1997 to June of

2002 We supplement this dataset with a search of Lexis-Nexis20 for other S&P 1500 firms that restated in the period from 1995 to 1997 and through the end of 2002 We also collect data on the fiscal years (or partial years) for which the firm had to restate The total time period covered for announcements of restatements by S&P 1500 firms is 1995 to 2002.21 This dataset does not contain restatements for benign reasons, for instance, due to accounting changes, mergers and acquisitions or one-time errors, but for purposeful accounting choices To the extent that our dataset does contain restatements caused by error rather than management’s discretion, then we are less likely to find support for the hypotheses This yields a sample of 215 restating firms in the analyzed period from 1994 through 2002 Due to data availability the number of firms is reduced in some analysis, thus we report the number of firms used

19

Beneish (1999) covers the period from 1987 to 1993 Dechow, Sloan and Sweeney (1996) covers the period from 1982 to 1992 Moreover, laws have changed in 1994 allowing executives to immediately sell shares acquired through the exercise of options Previously, they were required to hold the stock from the exercise of options for six months following exercise See Carpenter and Remmers (2000) for more details

Conditioning on the availability of data for firms in ExecuComp forces us to examine firms that are larger and probably more mature than the average restating firm is However, these restatements are more likely to concern policy makers and investors Moreover, recent studies indicate that the size of the average restating firm has increased over time (Richardson, Tuna and

restatements, the CEO’s sensitivity to firm performance has increased over time (and options are especially responsible for this increase in sensitivity).23 It is likely that these increases have a relation

2.4.2 Summary Statistics

Table 1 presents summary statistics for our sample Panel A reports the distribution of the announcements of the 215 restatements from 1995 through 2002 There is an increase in the number of restatements made by firms in the S&P 1500 over time, consistent with the general increase in the number of restatements reported by Richardson (2002), et al.24 Panel B reports the distribution of the fiscal years for which the alleged manipulation occurred (the fiscal year that had to be restated).The number of restated years is greater than the number of announcements

Richardson et.al (2002) use a sample of 225 earnings restatements from 1971 to 2000 They do not condition

on firms being available in ExecuComp, thus they have a larger sample

because some firms announce that they are restating multiple years The number of restated years

is increasing over time, with a concentration in the fiscal years of 1998 through 2000

Figure 1 of Appendix B presents a timeline for a restating firm that announced in

January 2000 that it is restating financial statements for fiscal years 1998 and 1999

For our sample, the mean (median) amount of time lapsed between the manipulation and the announcement of the manipulation is 1.67 (1.25) years, as reported in Panel C, with a range of 8 years at the first quartile, to 2.5 years at the 3rd quartile

The industry distribution of the restating firms is shown in panel D Restatements are concentrated in computer equipment, electrical and measuring instruments industries as well as services The industries in which restatements are more prevalent are growth industries that seemed to be affected by investor optimism regarding technology Financial firms are not

included (SIC 60-69) due to data that is inappropriate for calculating leverage and book; variables that we use as controls in later reported regressions

market-to-2.4.3 Characteristics of restating firms

Previous literature concentrates on market based pressures as causes for executive’s management of financial statements resulting in restatement For instance, Dechow, Sloan and Sweeney (1996) find that significantly more firms engaging in fraudulent accounting are in violation of debt covenants However, Beneish (1999), by contrast, finds that earnings

overstatements do not seem to be motivated to meet debt covenants or to reduce the cost of capital Richardson, Tuna and Wu (2002) find restating firms to be growth firms, and that

restating firms are more likely to have accessed external capital during their restated years

We examine proxies of firm growth opportunities: Earnings to price and market-to-book

We also examine factors that have been associated with motivations for aggressive accounting

related to financing Restating firms might be manipulating accounting numbers in order to access external capital markets at a lower cost, thus, we examine whether the firms have accessed equity

or debt markets to a greater extent than non-restating firms Specifically, we examine restating firm’s equity finance raised and long term debt issued If restating firms use aggressive

accounting to reduce their cost of capital in the short-term, then we expect more issuing activity relative to non-restating firms

We also examine leverage and size Restatements may be the result of an attempt to avoid debt covenant violations We follow Dechow et al (1996) by examining leverage as a proxy for closeness to debt covenant violations, or financial distress Size can proxy for information

asymmetry, with large firms having lower information asymmetry Large firms may be less likely

to use risky accounting that results in restatements as they are under more scrutiny and thus more likely to be caught At the same time, the increased analyst following experienced by larger firms may lead to more aggressive accounting (Richardson, et al (2002)) We examine the dispersion

of analyst one-year forecasts in the year prior to the first reporting violation Greater dispersion may indicate that analysts recognize a problem with the firm Finally, we examine Earnings per Share as well as Fully Diluted EPS

We are interested in the incentives in place during the years restated as opposed to at the time of announcement Therefore, we measure the firm’s characteristics relative to the violation period as in Table 2

Table 3 presents univariate characteristics of restating firms compared to non-restating firms Data for restating firms are as initially reported for the fiscal year; that is, the revised accounting numbers are not used Panel A presents characteristics of restating firms relative to the first year of aggressive accounting In panel B of Table 3, we compare restating firms to the characteristics of the median non-restating firms Industry-year adjusted values for restating firms

are calculated by deducting the median non-restating firm’s characteristic in the corresponding industry-year of the restating firm Non-restating firms are firms that have not restated in the 1995-2002 period that we analyze Industry is defined by the two-digit SIC code Mean and median values of firm characteristics are reported

The average restating firm is large, with a market capitalization equaling $10 billion (we use the first restated year if there is more than one year restated) However, the size distribution is skewed, with the median market value of $1.5 billion Overall, the univariate results in Panel A and B indicate that the restating firms in our sample are larger than those of Richardson, Tuna and

Wu (2002) This likely reflects the fact that we are conditioning on firms in ExecuComp (S&P 1500).25 Restating firms also have greater market valuations relative to non-restating firms, if measured by market-to-book However, earnings-price ratios are not significantly different from non-restating firms We find that restating firms are more likely to issue debt or equity in the manipulated year and have greater leverage.26 The dispersion of analyst one-year forecast is greater for restating firms than non-restating in the year prior to the violation period, indicating that some analysts may be foreseeing a problem We further explore these results in the

regressions described in Section 6

likely to have high accruals in the year for which they restate This is not surprising as firms that issue equity sometimes use accruals in the years they access the market (Teoh, Welch and Wong (1998)) Beneish (1999) suggests that firms which violate GAAP do so because they have run out

of alternative ways to represent their firm’s performance in a positive way, implying that firms use accruals in the years prior to the restated years We expect that restating firms use aggressive accounting within U.S GAAP, in the form of discretionary accruals, in the year of alleged

manipulation However, restatements are not necessarily the result of accruals, so there may be no relationship

We examine the use of discretionary accruals for restating firms and non-restating firms around the initial year for which financial statements were restated Appendix C describes the measurement of discretionary accruals We compare restating firms to portfolios of non-restating firms that are also in ExecuComp to facilitate comparison of compensation characteristics Non-restating firms are from event-year zero For the firm described in Figure 1, we select non-

restating firms in 1998

Table 4, panel A reports the time series of discretionary accruals surrounding the first year of alleged manipulation Results indicate that restating firms, on average, have significantly more positive discretionary accruals in the first misreported year The difference in median discretionary accruals is not significant In year two, the difference is negative, perhaps indicating

a reversion in accruals or the effect of a restatement on the current year’s financial statement The figure shows the increasing use of discretionary accruals prior to the first misreported year, similar to the results reported by Dechow, Sloan, and Sweeney (1996) While discretionary accruals are increasing in the years prior to restatement, they are not significantly different from benchmark portfolios except in the initial year and second year relative to the initia l year of

current accruals may be managed upwards (in which case earnings are greater than cash flow) by reporting income early or delaying expenses (amounts of revenue and expenses may also be manipulated)

restatement This provides some evidence consistent with the notion that executives in restating firms are using aggressive accounting within GAAP

2.5 Measures of CEO compensation

In order to analyze the CEO’s personal incentives to use accounting that violates U.S GAAP, we need to take into account his portfolio of equity incentives We separate the

components of equity incentives to enable us to discern the effect of each component

2.5.1 Option sensitivity

In order to test the hypothesis that CEOs with greater fractions of their compensation linked to options are more likely to engage in actions that result in restatements, we need a measure of the sensitivity of the CEO’s option compensation to the stock price We use Core and Guay’s (2001) method to measure the sensitivity of the CEO’s option portfolio to stock price, the delta of the stock options Core and Guay (2001) report that this accounts for approximately 99%

of the variation in option portfolio sensitivities;29 the method is commonly used in studies of compensation It measures the change in value of the option for a percentage (one percent) change in stock price The measure uses proxy statement data contained in ExecuComp on unvested and vested previously granted options as well as newly granted options It is measured

as the Black-Scholes delta multiplied by price divided by 100 Multiplying by price/100 changes the units of delta from a dollar change in option value per dollar change in price, to a dollar change in option value per percentage change in price The measure is calculated as follows:

Sensitivity of option to stock price =

* 100

price price

value Option price

price

value

Where Z=[ln(S/X)+T(r-d+σ2

/2)]?σT(1/2) Φ??(z) is the cumulative normal distribution, S is the price

of the stock; X the exercise price of the option??; σ the volatility of the option, r the risk free rate (from T-bills/bonds) corresponding to the maturity of the option; T the time to maturity; d the natural log of the dividend yield These variables are obtained from ExecuComp, with the risk free rate from the Federal Reserve at St Louis We measure the price at fiscal year end; volatility

is taken from ExecuComp.30 Because ExecuComp does not offer details on previously granted options, assumptions need to be made regarding the time to maturity and exercise prices Time to maturity for unvested options is calculated as the time to maturity of the recent grant of options minus one year The time to maturity of the vested options is estimated as the time to maturity of unvested options minus three years Exercise prices are calculated as the price at fiscal year-end minus the profit per option.31

The change in value for a percentage change in stock price is estimated for each option and summed up across all options to account for the fact that executive’s incentives are magnified for greater numbers of options

We measure the sensitivity of the CEOs portfolio due to options in a second way Core and Guay (1999) show that by multiplying the previous measure by 100 divided by market value (in thousands) will convert it to the Jensen and Murphy (1990) measure of a dollar change in CEO wealth per a $1000 change in firm value

30

Annualized volatility is measured over 60 months prior to the fiscal year It is the value used to calculate the

Black Scholes value of the option as reported by ExecuComp

31

Profit per option is calculated as the realizable value of options divided by the number of options at fiscal year end

The Jensen and Murphy (1990) measure is motivated by Jensen and Meckling’s (1976) model of agency costs whereby an agent that owns less than 100% of the firm has incentive to consume perks as he benefits from these but does not bear the full cost of the perks The greater his ownership the greater the costs he bears; thus agency costs are negatively related to the percentage of ownership However, this does not take into account the CEO’s wealth or the size

of the firm A small amount of ownership of a large firm is a large amount of CEO wealth Thus, large dollar holdings of equity have powerful incentive effects despite that fractional holding is small (Hall and Liebman (1998)) The Core and Guay (1999) measure approximates for these equity incentives by computing the dollar change in wealth for a percentage change in firm value These measures take into account the value of vested and unvested options The value of vested options may seem most relevant as only these can be exercised to realize value However,

unvested options are also a source of wealth, increasing manager’s utility, and sometimes offering collateral for debt.32

Baker and Hall (2001) argue that the first measure (change in wealth for a percentage change in price) is more useful when examining incentives to change entire firm value, for instance, a change in the accounting system They suggest that percentage ownership (the second measure) is more important in explaining incentives to expropriate through perquisite

consumption, e.g., purchase of a corporate jet for personal reasons While we use both measures, the Core and Guay (2001) measure is most relevant for our purposes, as we are assessing the effect of an aggressive accounting choice on the manager’s wealth Therefore, we focus on this measure in our analysis

32

See Lublin, J and J Sandber, Deadbeat CEOs Plague Firms as Economy and Markets Roll., The Wall Street Journal, August 1, 2002

2.5.2 Sensitivity of restricted stock and equity holdings

We also measure the CEO’s sensitivity to stock holdings in two ways Consistent with the Core and Guay (2001) measure of option incentives, we measure the stock incentives as the number of shares held multiplied by the price divided by 100, setting the derivative of equity with respect to price to one The same measure is used for restricted stock That is:

stock of y

∂

∂

=Our second measure is consistent with the Jensen and Murphy (1990) measure of

incentives caused by equity It is the market value of shares held divided by the market value of the firm The difference in the two measures is as in the previous discussion The incentives related to restricted stock are measured in the same way

2.5.3 Long term incentive plans

The effect of long term incentive plans (LTIP) is measured as payout from long-term incentive plans divided by total compensation including option grants LTIPs are based on a three

to five year moving average of firm performance ExecuComp does not provide information on what the payout consists of, only the value

2.6 The effect of compensation

We examine the impact of compensation on the propensity to restate As described in section 2, restating firms may restate more than one year (or partial year) Table 1 shows that for the sample of 215 restating firms, we have a total of 266 restated years Following Richardson, et

al (2002), we compare restatement firm-years to non-restatement firm-years; that is, we treat each year as an observation First, we present summary statistics on compensation in restating firm-years and non-restating firm-years Next, we present results of a logistic regression with

restated year as the dependent variable to test for those factors associated with the propensity to engage in aggressive accounting that might result in a restatement For the logistic regressions, using firm-years has an advantage over using a control firm approach because using a control firm approach overstates the likelihood of a restatement

Table 5 reports summary statistics of characteristics for restated and non-restated years We focus on the Core and Guay (1999) measure of option sensitivity, as it is most relevant for our purpose (described earlier) For restating firm-years, the median (mean) sensitivity of CEO’s wealth to a one percent change in stock price is quite high at $132,000 ($561,000) This is significantly different from that of non-restating firms, with a median (mean) of $80,000

firm-($264,000) The fact that restating firms are larger than non-restating firms may partially explain this large difference in the magnitude of compensation Recall that the incentives from stock holdings are measured as the number of shares held multiplied by 1% of the price The incentive from stock compensation is $69,000 at the median ($759,000 on average) for restating firm-years, and is significantly skewed Thus, in regression results we also winsorize the data to check the results Stock incentives for restating firm-years are significantly different from that of non-restating firm-years at the median, only For options to play a more significant role in a manager’s decision then equity, it must be the case that characteristics of options affect manager’s decision making more so than equity These features are the convexity of options and the ease of selling through vesting schedules that they offer relative to equity, as discussed in section 2 Not

surprisingly, long-term incentive plans and restricted stock are not a significant portion of

compensation

Table 5 also reports the age of the CEO Age is available for 143 of the firm-years, so we also report the number of years credited toward retirement because this variable is available for

more firms The average and median age of CEOs in misreported years is less than that of

non-restating firms We will later use these variables as proxies for the manager’s time horizon

Table 6, panel A, presents results of logistic regressions with restated-year as the

dependent variable (=1) and characteristics of compensation as independent variables The first logistic regression presents results including the compensation variables and control variables previously described: market value, market-to-book and leverage.33 Conditioning on the

availability of lagged control variables reduces the number of observations to 242 restated-years and 8017 non-restated years We use the log of the option sensitivity as in Core and Guay (1999) because this measure increases at a decreasing rate with firm size (Baker and Hall (1998))

Results indicate that the CEOs sensitivity of wealth from options is significantly and positively related to the propensity to use aggressive accounting, providing results consistent with hypothesis, H1 Stock options encourage aggressive accounting by tying management’s wealth to stock price while limiting downside risk associated with the accounting no longer working We also examine other components of compensation Recall that H6 argues restricted stock

ownership and long-term incentive plans, with their linear payoffs and function of long-term firm value, should mitigate the effect of stock options because management will be exposed to the downside from using aggressive accounting Thus, we expect a negative relationship The sensitivity of compensation from restricted stock and payout from long-term incentive plans are each insignificant Thus, we do not find support for H6 This may be due, in part, to the fact that restricted stock and long-term incentive plans are a small part of compensation The coefficient

on the sensitivity of equity also indicates that equity holdings do not significantly affect the propensity to misreport

33

We also tried total assets, market-to-book value of equity and leverage; results do not change substantially

We also include controls in the logistic regressions Market value is positively significant, reflecting the fact that our sample of restating firms tends to be large Results also indicate that leverage is positively associated with the propensity to use aggressive accounting, lending

support for the argument that CEOs are trying to avoid debt covenant restrictions.34 It is

surprising that growth opportunities, proxied by the market-to-book ratio, is not significantly related to the propensity to restate It is also not significant when we do not include size and leverage (in unreported results).35 Industry and year dummies are included in regressions Firms

in the services, electrical and computer equipment industries are more likely to engage in

aggressive accounting, as we observed in the univariate results, perhaps due to the fact that it is harder to verify that a service or other intangible asset has been provided than a tangible asset To test this more directly, we us a measure of intangible assets: (R&D expense + advertising

expense)/assets; however, the coefficient is insignificant

Providing some support for Bolton et al.’s (2002) implication that CEOs are more likely

to use aggressive accounting in years in which it is more likely to be profitable to do so- that is, years with inflated market valuations (H3)- the coefficients on fiscal years 1998 and 1999 years are significant and positive There is greater use of aggressive accounting associated with

restatements in these years, confirming the univariate results As the market was declining in

2000, we would expect that firms manipulate less in those years, yet results indicate the opposite effect Because Bolton et al.’s (2002) and Bebchuk and Bar-Gill’s (2003) also imply that

restating is associated with the firm’s environment, we include the median market-to-book ratio

of S&P 1500 firms in the restating firm’s industry-year to proxy for the its environment

(regression 4) Specifically, we measure the median market-to-book ratio for all firms in

ExecuComp (the S&P 1500) by year and industry This variable is insignificant; thus, we do not find support that market valuation is a driving factor for restatements

We treated each restated firm-year as an observation in the previous regressions;

however, some firms restated multiple years so it may not be accurate to treat each year

independently Therefore, in regression (5), we include a dummy variable equal to one if the firm restated the previous year Results indicate that firms are more likely to misreport in a subsequent year if they misreported in the previous The significance of option incentives is reduced to the ten percent level In regression (6), we proxy for the manager’s time horizon by the number of years the CEO has credited towards retirement If a manager is closer to retirement, then he may

be expecting to sell his shares, and we should find a positive relation between the propensity to restate and closeness to retirement However, results indicate that the number of years credited towards retirement is negatively related to the propensity to restate Perhaps younger managers are more likely to be aggressive.36 If younger managers do not have the experience of more seasoned managers, then they are probably in their positions, in part, because they are more aggressive and take more risks

In regression (7), we examine whether the aggregate value of options exercised during the fiscal year, as measured by the value of options exercisedt/market valuet-1, is associated with the restated year While the coefficient is positive, it is not significant.37 Regression (8) reports a similar specification, but using the contemporaneous value of options exercised plus the lead value of options exercised divided by the market value at t-1 The intuition being that to the extent that the misreported information continues to work prior to being discovered (at