Three essays on executive compensation

Other hypotheses are developed based on early exercise, vest date exercise, dividend related exercise, executive departure, the proportion o f vested options exercised, and the proportio

byBRANDON NEIL CLINE

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted Also, if unauthorized copyright material had to be removed, a note will indicate the deletion

®

UMI

UMI Microform 3201243 Copyright 2006 by ProQuest Information and Learning Company All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, Ml 48106-1346

degree o f Doctor o f Philosophy specializing in Finance.

Accepted on behalf o f the Faculty o f the Graduate School by the dissertationcommittee:

Robert E Brooks, Ph.D

Chairman

B iljyP Helms, Ph.D Department Chairman

/ ( ^ 0 /V )(

Ronald W R ogers/fh.D Dean o f the Graduate School

I am indebted to many people without whom this dissertation would be impossible These people are dear to me and the valuable encouragement they have provided have been a pillar o f support I am deeply grateful to each.

First, I would like to thank Professor Robert Brooks, m y advisor, who has been a tremendous help and encouragement throughout the dissertation process as well as the entire doctoral program It was his inspiration and ability to capture my interest as an undergraduate that motivated me to pursue these interests that made this endeavor possible I would like to also thank other members o f my dissertation committee: Professors Don Chance, Bruce Barrett, Harris Schlesinger, and Douglas Cook I am especially grateful to Professor Don Chance for his suggestions for the first essay o f this dissertation and the general guidance he has provided

I would like to acknowledge Professor Billy Helms, m y department head, and Dean Barry Mason for the departmental support that has been provided I am grateful to both o f you for providing me this opportunity Professor Helms has been ever gracious in providing me with funding for data and travel Dean Barry Mason has been the source o f great encouragement From my freshman year through my graduate years he always took the time to chat with a kid from a small town with little direction

I am grateful to all my friends in the doctoral program at the University o f Alabama Specifically, I would like to thank Melissa, Justin, Sahiba, and Jonathan for their friendship W ithout Melissa, enduring those first two years o f the program would

friendly competition To Jonathan I am extremely grateful for the patience exhibited getting me started with SAS Sahiba has provided more direction and insight than she could know Finally, I am glad that it was Chen whose cubical was behind me in Bidgood 339 Without her to help conjecture ideas and assist with seemingly mundane computer tasks I am quite confident I w ould still be on Essay One I have also been blessed with many true friends outside my department I offer a special thanks to Josh, Tyler, Clint, Gary, and Mac I also want to express my appreciation to Mr and Mrs Mann for their encouragement I cannot begin to express all that each o f you have done.

Above all, I want to thank the Lord for His blessings and love He has blessed me with two wonderful parents Randy and Elaine and a very special sister Dena I am confident that it is the work ethic instilled by my father and the dedication exhibited by

my mother that made this possible I dedicate this dissertation to you and to my wife Shanna- the most wonderful, loving and patient woman in the world

ACKNOWLEDGMENTS .iii

LIST OF TABLES viii

LIST OF FIGURES x

ABSTRACT xii

1 INTRODUCTION 1

2 EXECUTIVE STOCK OPTION EXERCISE, INSIDER TRADING, AND ABNORMAL STOCK RETURNS: EXECUTIVE CHARACTERISTICS 7

2.1 Introduction 7

2.2 Previous r esearch 13

2.2.1 Nontransferability 16

2.2.2 Portfolio rebalancing 17

2.2.3 Taxes 18

2.2.4 D ividends 19

2.2.5 Liquidity 19

2.3 Data and hypothesis development 19

2.4 M ethodology 25

2.5 Results 26

2.5.1 Results for the full sample 27

2.5.2 Results according to early exercise 28

2.5.3 Results according to vest date exercise 30

2.5.5 Results according to executive departure 33

2.5.6 Results according to proportion o f vested stock exercised 35

2.5.7 Results according to proportion o f stock sold 37

2.6 Conclusion 39

3 EXECUTIVE STOCK OPTIONS EXERCISE, INSIDER TRADING, AND ABNORMAL STOCK RETURNS: FIRM CHARACTERISTICS 71

3.1 Introduction 71

3.2 Previous Research 74

3.3 Data and hypothesis development 79

3.3.1 D ata 79

3.3.2 The hypotheses 80

3.4 M ethodology 83

3.5 R esults 84

3.5.1 Full sample 85

3.5.2 Sector analysis 86

3.5.3 Size 88

3.5.4 Growth characteristics 89

3.6 Conclusion 92

4 MANAGERIAL DELTA: OPTIONAL SENSITIVITY OF FIRM ’S MANAGEMENT TO ITS OWN STOCK 115

4.1 Introduction 116

4.2 Management incentives and option moneyness 118

4.2.1 Out-of-money options 120

4.3 Data and hypothesis development 121

4.3.1 D ata 121

4.3.2 Hypotheses 123

4.4 Results 124

4.5 Conclusion 126

REFEREN CES 134

M onthly FF-m om entum -adjusted returns based on the proportion o f vested options exercised 67

M onthly FF -m om entum -adjusted returns for m erged sam ple proportion o f acquired stock sold 69Monthly FF-momentum-adjusted returns for full sample 106

M onthly FF-m om entum -adjusted cum ulative abnorm al returns by com pany sector 108

M o n t h l y F F - m o m e n t u m - a d j u s t e d m e a n a b n o r m a l r e t u r n s b y f i r m characteristic 110

M onthly FF-m om entum -adjusted mean abnormal returns by company sector and firm characteristics 112Summary statistics company sector and firm characteristics 114Option grant size summary statistics by y e a r 128

4.3 Option grant size summary statistics by moneyness 130

4.4 Moneyness summary statistics by rank 131

4.5 Abnormal returns by moneyness rank 132

4.6 Abnormal returns by moneyness rank and grant s iz e 133

2.1 Cumulative abnormal returns for full sample o f all reported trades 41

2.2 Cumulative abnormal returns for full merged sam ple 42

2.3 Full sample early and maturity exercise cumulative abnormal returns 43

2.4 Full sample vest date and not vest date cumulative abnormal returns 44

2.5 Number o f exercises prior to ex-dividend date 45

2.6 Not dividend and dividend cumulative abnormal returns 46

2.7 Not depart and depart cumulative abnormal returns 47

2.8 Resigned and retired cumulative abnormal returns 48

2.9 Proportion o f vested options exercised cumulative abnormal return 49

2.10 Proportion o f stock sold cumulative abnormal returns 50

3.1 Cumulative abnormal returns for full sample o f all reported trades 94

3.2 Cumulative abnormal returns for the financial sector 95

3.3 Cumulative abnormal returns for the healthcare sector 96

3.4 Cumulative abnormal returns for the consumer non-durable sector 97

3.5 Cumulative abnormal returns for the consumer service sector 98

3.6 Cumulative abnormal returns for the consumer durables sector 99

3.7 Cumulative abnormal returns for the energy sector 100

3.8 Cumulative abnormal returns for the transportation sector 101

3.9 Cumulative abnormal returns for the technology sector 102

3.10 Cumulative abnormal returns for the basic industries sector 103

3.12 Cumulative abnormal returns for the public utilities sector 1054.1 Not depart and depart cumulative abnormal returns 127

This dissertation contains three essays in the broad area o f executive compensation In the first two essays we investigate whether corporate insiders use private information in their decision to exercise executive stock options.

The first essay examines executive stock option exercises to investigate corporate insider's use o f private information in their decision to exercise stock options In the year following exercise we document negative cumulative abnormal return of approximately 11% We find that the most informed insiders are those that exercise early, exercise after the vesting date, exercise a large proportion o f vested options, or sell a large proportion

of the stock acquired from exercise It is also shown for dividend paying firms that executives often exercise early just prior to dividend payments; these exercises are non- informative

In the second essay, three major hypotheses are developed The primary hypothesis states that insider trading differs according to company Second, we hypothesize that post-exercise abnormal performance differs according to company size Finally, we hypothesize that post-exercise negative abnormal returns are larger and more significant for growth firms than for value firms

The results o f this study support our primary hypothesis We document that the largest and most significant post-exercise negative abnormal returns occur in the technology, healthcare, and consumer service sectors while no significant negative

industries sectors Support is also found for our third hypothesis Our results indicate that growth firms experience more significant negative abnormal performance following stock option exercise than value firms.

Essay three investigates the optimal structure o f executive stock option policies

We argue that if managers are optimally motivated when options are struck at-the-money then it must be true that as options move away-from-the-money, and the convex payoff structure that once properly aligned the interest o f shareholders and managers becomes linear, the compensation package o f the company lacks proper motivation We hypothesize that this lack o f proper motivation s hould be evident in lower stock price performance for firms with executives’ holding deep in-the-money and deep out-of-the- money stock option portfolios

This dissertation contains three distinct essays in the broad area o f executive compensation.

The first essay is on insider trading and executive stock options (ESOs) In this essay, using a dataset that includes all exercises by corporate insiders reported to the SEC from 1996 to 2003, we investigate empirically whether corporate insiders use private information in their decision to exercise ESOs

Executive stock options (ESOs) represent a large portion o f many firms’ total compensation The dramatic increase in incentive based pay over the past two decades has spawned much debate over the valuation o f such compensation plans Often this debate has revolved around the valuation o f ESOs and why models used effectively to value publicly traded stock options (TSOs) are not useful in valuing ESOs Valuation of these two types o f options differs in that ESOs are subject to certain trading frictions.1 While most o f the literature has focused on the implications regarding vesting and transferability restrictions, others have addressed such factors as liquidity, performance based implications, the inability o f executives to properly diversify, and differences related to volatility However, there is another difference related to valuation that has received little attention: information

In this first essay we test the hypothesis that insiders use private information intheir decision to exercise ESOs This hypothesis is tested using a unique dataset which

1 These differences, among others, have been recognized by Noreen and W olfson (1981) and Foster,

consistent o f virtually all insider option exercises reported to the SEC between 1996 and

2003 We analyze 262,931 option exercises by 51,388 corporate executives from 7,475 firms and find evidence suggesting that insiders do effectively time the exercise o f their stock options

The primary hypothesis o f this study is that executives time ESO exercises prior

to periods o f declining stock The corresponding null hypothesis is that stock prices exhibit no significant movement after option exercises are made Other hypotheses are developed based on early exercise, vest date exercise, dividend related exercise, executive departure, the proportion o f vested options exercised, and the proportion o f stock sold at exercise

We find strong evidence supporting our private information hypothesis Specifically, for the full sample we document that over the 12 months following option exercise, firms experience significant negative cumulative abnormal returns (CARs) of approximately 11% To provide additional insight into ESOs and private information associated with their exercise insider exercises are disaggregated according to six factors

to distinguish which cohorts o f these insiders are most informed Specifically, the dataset

is partitioned according to 1) whether the executive exercised their stock option prior to maturity, 2) whether the executive exercised their stock option on the vesting date, 3) whether the exercise was dividend related, 4) whether the exercise was induced by executive departure from the firm, 5) the proportion o f options exercised to those vested that were not exercised, and 6) the proportion o f stock acquired from exercise that was sold This analysis provides further insight into valuation and how firms should approach

the hedging o f ESOs by considering which executives are most effective in timing the exercise o f their incentive based options Evidence presented here suggests that firms with insiders who exercise early experience large significant negative abnormal returns following option exercise, while firms with executives that exercise at maturity experience no significant abnormal returns Furthermore, firms with insiders who exercise early but after the vesting date experience larger negative abnormal returns than those exercised around the vesting date It is also shown for dividend paying, executives often exercise early just prior to dividend payments to capture dividends These dividend motivated exercises are non-informative.

In second essay we investigate the information contained in the decision of corporate insiders to exercise their executive stock options based on the company sector and other firm characteristics o f the executives’ initiating the transactions Analyzing all exercises reported to the SEC by insiders from 1996 through 2004 three major hypotheses are developed The primary hypothesis states that insider trading differs according to company sector and consequently some sectors experience larger and more significant negative abnormal stock performance following option exercise than do other sectors The corresponding null hypothesis is that post-exercise stock performance does not materially differ according to company sector The results o f this study support our primary hypothesis We document that the largest and most significant post-exercise negative abnormal returns occur in the technology, healthcare, and consumer service sectors while no significant negative abnormal returns occur in the financial, consumer non-durable, transportation, or basic industries sectors

In the second stage o f our analysis we develop hypotheses to gain insight to the sector results We hypothesize first that post-exercise abnormal performance differs according t o e ompany si ze T his h ypothesis i s p redicated o n r esults f rom t he S eyhun (1986), Seyhun (1998), and Lakonishok and Lee (2001) who find that insider trading in equity markets is more informative at smaller firm levels To test this hypothesis exercises are distinguished based on whether the transaction was initiated by insiders from small, medium, or large firms, where firm size is based on market capitalization Contrary to the results o f insider equity literature, we find that larger firms experience the most significant negative abnormal performance following option exercise.

Second, w e hypothesize that post-exercise negative abnorm al returns are larger and more significant for growth firms than for value firms This hypothesis is primarily predicated on the result o f Skinner and Sloan (2002) that growth stocks exhibit a more negative response to negative earnings surprises than do value stocks To test this hypothesis we partition our data according to characteristics that traditionally distinguish firms as either growth or value firms (Lakonishok, Shleifer, and Vishny, 1994) Specifically, we classify firms based on whether the exercise was initiated by an executive from firms that pay no dividend, pay a low dividend yield, or pay a high dividend yield; firms with low, medium, and high price-to-eamings ratios, and firms with low medium and high market-to-book ratios The findings o f this essay are consistent with the hypothesis that growth firms experience more significant negative abnormal performance following stock option exercise than value firms In summary, our results suggest that firms that experience the largest stock decline following an option exercise

by an insider are large firms with high price-to-eamings and high market-to-book ratios that pay no dividend.

Essay three investigates the optimal structure o f executive stock option policies Management compensation committees face many alternatives when deciding on compensation packages to provide its managers Types o f compensation vary from salary, cash bonuses, and restrictive shares o f stock to warrants and stock options

One i mportant consideration w hen d etermining c ompensation p ackages i s w hat properly motivates managers Traditional agency theory literature such as Jensen and Meckling (1976), Holmstrom (1979), Grossman and Hart (1983), and Holmstrom and Milgrom (1987) suggests that managers should be given incentives to maximize shareholder wealth (stock prices) when effort is non-observable or monitoring costs are high Most studies suggest that since managers are risk averse and shareholder are risk neutral, options are the best form o f incentives (Inderst and Muller, 2004)

The purpose o f this essay is to contribute to this ongoing debate regarding the optimal stock option expose a firm should subject to its manager We argue that if managers are optimally motivated when options are struck at-the-money then it must be true that as options move away-from-the money, and the convex payoff structure that once properly aligned the interest o f shareholders and managers becomes linear, the compensation package o f the company lacks proper motivation We hypothesize that this lack o f proper motivation should be evident in lower stock price performance for firms with executives’ holding deep in-the-money and deep out-of-the-money stock option portfolios

Contrary to our hypothesis we find that firms with outstanding executive stock options that are deep in-the-money along with firms with outstanding executive stock options that are deep out-of-the-money do not experience abnormal stock price performance that is significantly different from firms with outstanding options at or near- the-money.

2.1 Introduction

Executive stock options (ESOs) represent a large portion o f many firms’ total compensation These option plans typically award executives the right to purchase shares

o f the company stock at a specified price usually equal to the market value o f the stock

on t he d ay t he o ption i s g ranted, w ith m aturities r anging from f ive t o t en years T he dramatic increase in incentive based pay over the past two decades has spawned much debate over the valuation o f such compensation plans Often this debate has revolved around the valuation o f ESOs, and why models used effectively to value publicly traded stock options (TSOs) are not useful in valuing ESOs Valuation o f these two types o f

options differs in that ESOs are subject to certain trading frictions While most o f the literature has focused on the implications regarding vesting and transferability restrictions, others have addressed such factors as liquidity, performance based implications, the inability o f executives to properly diversify, and differences related to volatility However, there is another difference related to valuation that has received little attention: information Only corporate insiders hold ESOs thus it is reasonable to suggest that there might exist some information content in their exercise If this is true, then private information is yet another factor distinguishing ESOs from TSOs

2 These differences, among others, have been recognized by Noreen and W olfson (1981) and Foster,

practice Data vendors, such as CDA/Investnet, use insider trading data to predict returns, and money managers use insider activity to assist in asset allocation decisions

The demand for insider trading information is summarized in an article in Individual Investor (Feb 1998, p 54):

“Company executives and directors know their business more intimately than any

Wall Street analyst ever would They know when a new product is flying out the

door, when inventories are piling up, whether profit margins are expanding or

whether production costs are rising You always hear about the smart money

Generally, that is the smart money ”

With the large amount o f attention given to insider trading, and since ESOs are held by corporate insiders, there is reason to believe that information is yet another distinguishing factor that should be considered in the valuation o f ESOs In this article

we test the hypothesis that insiders use private information in their decision to exercise ESOs We test this hypothesis using a unique dataset which consists o f virtually all insider option exercises reported to the SEC between 1996 and 2003 We analyze 262,931 option exercises by 51,388 corporate executives from 7,475 firms and find evidence suggesting that insiders do effectively time the exercise o f their stock options

Executives o ften e xercise o ptions i n w ays t hat d o n ot c onform t o c onventional option pricing theory In particular, a large proportion o f ESOs are exercised early By exercising early, the holder loses the time value o f the option, thus giving up any potential appreciation in the option payoff Despite this fact, executives still exercise

early, commonly sacrificing approximately half o f the Black-Scholes value (Huddart and Lang, 1996) One possible explanation for this phenomenon is that a large proportion of early ESO exercises indicate that executives are timing the market, perhaps based on inside information To test this hypothesis we collect all reported ESO transactions in Thomson Financial Insider Filing Database (TFI) This dataset purports to include all exercises by corporate insiders reported to the SEC between 1996 and 2003.

The primary hypothesis o f this study therefore is that executives time ESO exercises prior to periods o f declining stock prices and that this at least partially explains the early exercise phenomenon The corresponding null hypothesis is that stock prices exhibit no significant movement after option exercises are made Other hypotheses are developed based on early exercise, vest date exercise, dividend related exercise, executive departure, the proportion o f vested options exercised, and the proportion of stock sold at exercise

This research differs from previous research in several ways First, it is the first paper to conduct analysis on insider stock option exercises based on the sector to which the executives’ firm belongs Second, this is the first research analyzing the exercise o f executive stock options to document larger and more significant negative abnormal returns for large firms Finally, to our knowledge, no other research investigates information content based on whether the executives firm is characterized as a growth or value firm

We find strong evidence supporting our private information hypothesis Specifically, for the full sample we document that over the 12 months following option exercise, firms experience significant negative cumulative abnormal returns (CARs) o f

approximately 11% To provide additional insight into ESOs and private information associated with their exercise we disaggregate insiders according to six factors to distinguish which cohorts o f these insiders are most informed Specifically, we partition our dataset according to 1) whether the executive exercised their stock option prior to maturity, 2) whether the executive exercised their stock option on the vesting date, 3) whether the exercise was motivates by the capture o f a dividend, 4) whether the exercise was induced by executive departure from the firm, 5) the proportion o f options exercised

to those vested that were not exercised, and 6) the proportion o f stock acquired from exercise that was sold This analysis provides further insight into valuation and how firms should approach the hedging o f ESOs by considering which executives are most effective in timing the exercise o f their incentive based options Evidence presented here suggests that firms with insiders who exercise early experience large significant negative abnormal returns following option exercise, while firms with executives that exercise at maturity e xperience n o significant abnormal returns Furtherm ore, firms w ith i nsiders who exercise early but after the vesting date experience larger negative abnormal returns than those exercised around the vesting date It is also shown for dividend paying firms that executives often exercise early to capture dividends; these exercises are non- informative

To conduct analysis on executive departure, proportion o f vested options exercised, and proportion o f acquired stock sold we construct a unique dataset comprised

o f all insider trades reported in the Table Two File o f the TFI where the insider appears

on the Compustat Executive Compensation (ExecuComp) database and the insider’s firm has return data on the Center for Research for Security Prices (CRSP) database for the

period 1996 through 2003 Using departure dates o f executives provided by ExecuComp,

we are able t o c ontrol fo r executives 1 eaving th e firm E xecuComp also provides th e value o f unexercised, exercisable options in the executive’s option portfolio This variable represents options that are vested and are at the executive’s disposal for possible exercise but remained unexercised at fiscal year end Using unexercised, exercisable options allows for the creation o f a proxy to control for exercises related to liquidity needs by calculating the percentage o f exercised options out o f those available for exercise One particularly interesting result is the ability to assess how the magnitude of mean abnormal returns changes as executives exercise a greater portion of the options at their disposal fo r exercise F inally, ExecuComp also allow s u s to follow a procedure similar to that found in Ofek and Yermack (2000) and determine the proportion o f stock acquired through option exercise that was sold

The purpose o f this article therefore is to investigate empirically, whether corporate insiders use private information in their decision to exercise ESOs We define private information as the ability to effectively time option exercise and do not allege that

it necessarily involves illegal activity on the part o f the executive We investigate first the information revealed in the exercise pattern o f insiders as a whole by examining all insider exercises reported to the SEC from 1996 to 2002 Evidence suggests that insiders

do effectively time their exercise Second, we disaggregate these insiders according to various measures to distinguish which cohort o f these insiders are most informed We find large negative abnormal post-exercise returns for options exercised early, supporting the hypothesis that insiders that exercise a large potion o f options early are more informed Furthermore, we find smaller and less significant abnormal returns for insiders

that exercise on the vesting date when compared to insiders that exercise after the vesting date b ut p rior t o m aturity, s upporting o ur h ypothesis t hat i nsiders t hat e xercise o n t he vesting date are less informed that those exercising after the vesting date We find no support for the hypothesis that exercises induced by executive departure are non- informative, instead results suggest both exercises forced by departure and those not related to departure are informative Dividing departure exercises according to whether the insider resigned or retired reveals that virtually all negative abnormal returns in departure exercises are explained by executive resignation.

Stock option grants are intended to align the long-term interests o f employees and managers with those o f shareholders; namely, to motivate employees to increase stock price.3 However, if managers are able to effectively time the exercise o f these options using private information and can thus extract firm wealth, then the role played by executive compensation is much more complex Because option recipients can benefit from the remarkable good timing o f their exercise o f options, their compensation may have less in common with effort, managerial skill and performance and have more to do with effective market timing A proper understanding o f the information contained in the exercise patterns o f executives is essential to adequately understand and to value incentive based options

The results o f this study have significant implications for valuation o f ESOs and

on hedging procedures o f firms that issue ESOs It is important to accurately assess the cost o f the options to shareholders If executives are effectively timing the market and expropriating shareholder wealth, then these ESOs are more expensive to the firm than initially expected If they are not effectively timing the market, incentive based

3 See Jensen and M eckling (1976), Haugen and Senbet (1981), Beck and Zorn (1982), and Murphy (1985).

compensation may be cheaper than initially expected and compensation committees should take this non-informed exercise pattern into consideration Specifically, the firm can liquidate the hedge position, pay the executive the value o f the exercised options, and still have some o f the value o f the hedge portfolio left over These findings are also important to the Financial Accounting Standards Board (FASB) as it develops accounting standards for incentive based pay.

The remainder o f the paper is organized as follows Section 2.2 discusses relevant literature In Section 2.3, we discuss our data and hypothesis development In this section we posit eight hypotheses Section 2.4 addresses methodology Section 2.5 summarized the empirical findings of our research, and Section 2.6 concludes

2.2 Previous research

There exists a vast amount o f research devoted to insider trading in equity markets It is generally agreed insider stock transactions contain information (see, e.g., Jaffe, 1974; Finnerty, 1974; Seyhun, 1986; Seyhun, 1988; R ozeff and Zamen, 1988; Lin and Howe, 1990) Bettis, Vickrey, and Vickrey (1997) find that by analyzing large insider transactions by top executives, investors can earn abnormal returns net of transaction costs The most conclusive evidence suggests that insider purchases contain the most information Jeng, Metrick, and Zeckhauser (2000) and Lakonishok and Lee (2001) find that insider purchases predict abnormal positive returns, but insider sales have no predictive power Lakonishok and Lee (2001) examine insiders’ trading activity for all companies traded on the NYSE, AMEX, and Nasdaq from 1975 to 1995 and determine that insider activity predicts stock returns for individual firms, with most o f the abnormal returns associated with smaller companies Controlling for size and book-to-

market, they find firms with extensive insider purchases outperform companies with extensive insider sales by 4.8%.

Although there is a plethora o f literature devoted to insider trading very little is devoted to testing for information contained in ESOs; furthermore, in studies that do investigate executives use o f private information in their decision to exercise their ESOs, the evidence is somewhat mixed We provide results suggesting significant information does exist in the exercise o f ESO by examining all reported exercises by corporate insiders from 1996 to 2003

Carpenter and Remmers (2001) test whether insiders use private information to time the exercise o f their ESOs by examining all reported exercises from 1984 to 1995 They study the period 1984 to 1991 and find that exercises preceded significantly positive abnormal returns, thus suggesting the use o f private information.4 For the period

1992 to 1995 they hypothesize that this informed trading should be revealed by negative abnormal returns following the exercise o f ESOs, as the current regime allows insiders to sell shares immediately upon exercise Their hypothesis was predicated on Ofek and Yermack (2000) who show managers typically sell all shares acquired through option exercise In contrast to their a priori hypotheses, for the period 1992 to 1995, they find little evidence o f such use o f information with the exception o f top managers at small firms However, due to data constraints Carpenter and Remmers (2001) were unable to eliminate exercises triggered by option expiration, they did not control executives leaving the company, nor did they control for executives exercising for liquidity purposes Our study differs from Carpenter and Remmers (2001) primarily because our data allows us

4 Prior to May 1991 insiders were required by law to hold the stock acquired through option exercise for at least six months In M ay 1991, the SEC changed the starting date o f Section 16b’s six m onth holding period to the grant date o f the option, thus allowing insiders to sell their acquired shares immediately.

to identify all exercises that are forced by maturity along with those that are forced by the departure o f the executive from the firm Furthermore, we control for exercises occurring

on the vesting date and for exercises induced by liquidity needs by distinguishing executives according to the proportion o f exercised option to the vested options available for exercise

In contrast to the findings o f Carpenter and Remmers (2001), Kyriacou and Mase (2003) find ‘that there is informational content in executive stock option exercises, contrary to the conclusions o f previous research ’ They analyze all executive transactions

in the United Kingdom from July 17, 1995 to July 3, 1998 and find that exercises overall

do not yield significant abnormal returns; however, when their sample is partitioned by the amount o f the acquired stock sold upon exercise, they find significant abnormal negative returns for those that sold a ‘high’ proportion o f the stock acquired Like Carpenter and Remmers (2001), Kyriacou and Mase (2003) are unable control for exercises forced by expiration or exercises forced by executive departure

Another study o f this nature is Huddart and Lang (2003) which considers option exercises by employees at seven firms and finds that when option exercises are high, stock returns are 10% lower than when option exercises are low They also show that exercise decisions o f junior employees contain just as much information as senior executives This finding is important since most studies o f ESO exercise only consider high ranking executives

Bartov and Mohanram (2004) study top-level executive exercises using ExecuComp data and document that firms with large option exercises experience negative abnormal returns Safdar (2004) examines returns at the quarterly level for

exercises by top executives at firms with high abnormal accruals and finds significant but small negative abnormal returns for only two quarters following option exercises.

As first shown by Merton (1973), a traded call option on a non-dividend paying stock should never be exercised before maturity A great portion o f ESOs are exercised early Are e xecutives i rrational? D o t hey i mplement s ub-optimal e xercise s trategies? Probably not since the market for ESOs is far from complete Exercise policies o f executives are dictated by publicly available information, trading restrictions, as well as executive-specific information including risk aversion, assets held in the executive’s personal portfolio, the executives’ demand for liquidity, and their expectations about the future movement in the stock

Previous literature suggests many rational explanations other than private information for the early exercise o f these options; thus, no conclusive evidence is given suggesting e xecutives u se p rivate i nformation i n t heir d ecision t o e xercise E SOs W e provide evidence that there is significant information content in early exercises Existing research offers the following alternative explanations as possible motivations for rational early exercise o f ESOs

2.2.1 Nontransferability

Cuny and Jorion (1995) note that executive departure typically forces early exercise of ESOs Executives holding ESOs cannot sell or transfer their options This is problematic when an executive departs from the company Thus, one possible explanation for the early exercise phenomena is that the executive may have left the company or is planning to leave the company before the option matures If the executive leaves the company, whether by choice or by force, she typically has three to six months

to exercise her option portfolios or else forfeit its entire value This being the case, options that are in-the-money will be exercised early, sacrificing the remaining time value and providing only the intrinsic value o f the option; options out-of-the-money will

be completely lost as they have no intrinsic value

It might also be true that the executive has inside information pertaining specifically to the executive Perhaps the executive knows that she is ill and that there exists a high probability o f death This should likewise be viewed as the executive leaving the company and would have the same implications as a forced or voluntary departure from the firm; namely, the executive would exercise all options that are in-the- money early, as it would otherwise be lost In some cases it may even be possible for the executive to pass her ESOs to her heirs In this event early exercise unrelated to information still may occur for the purpose o f building up needed cash reserves in anticipation o f health care costs

2.2.2 Portfolio rebalancing

Lambert, Larcker, and Verrecchia (1991) and Huddart (1994) focus on the inability o f executives to either sell the option or the underlying security as an explanation for the existence o f this early exercise phenomenon Typically, executives hold investment portfolios that are sub-optimally diversified, with the executive’s company stock constituting too large a portion o f the executive’s overall wealth Early exercise is beneficial to the executive o f the company since it eliminates the need to hedge in order to be adequately diversified Thus, a risk-averse executive may rationally choose to exercise early and sell the underlying stock if the diversification benefit from exercising is greater than the time value sacrificed from the early exercise Hemmerm,

Matsunaga and Shelvin (1996) examine 110 options exercised by executives in 1990 and find a positive relationship exists between option exercise and the risk inherent in holding

an un-hedged option position Huddart (1994) finds that risk-neutral executives exercise only at expiration, while risk-averse executives may find it optimal to exercise prior to option maturity Ofek and Yermack (200) find that managers typically sell all shares o f stock acquired through the exercise o f ESOs immediately

2.2.3 Taxes

Early exercise o f ESOs may also be attractive to executives because o f tax benefits Stock option awards provide a tax advantage to executives by supplying the executive with the option o f choosing when to exercise the stock option and incur the tax expense While the importance o f the difference in tax considerations for stock option awards and cash payments had influential effects in the 1950s (Holland and Lewellen, 1962), this difference has diminished in recent decades, (Yermack, 1995) If an executive believes the stock price will appreciate through the year, he may want to exercise the option and hold the stock, since income generated through option payoff is taxed a t a h igher r ate t han t he c apital gains received from s tock sa les Although t his argument offers intuitive appeal, Carpenter and Reemers (2001) show that this tax-based argument for exercise prior to positive stock performance is not merited Specifically, “if the executive expects the stock price to rise sufficiently, he is better o ff holding the option and buying additional stock with the money that he would otherwise have to pay

to exercise the option, namely, the strike price and the tax on the existing option profit.”

2.2.4 Dividends

Executives do not receive dividends on ESOs held in their portfolio However, the dividend can be captured by exercising and holding the acquired stock through the dividend date Thus, another possible reason for early exercise is to capture dividend payment by exercising the option and holding the underlying stock on the dividend date While intuitively appealing and theoretically sound, Carpenter and Reemers (2001) find that controlling for exercises that fall between a dividend announcement date and an ex- dividend date that their results were unchanged

2.2.5 Liquidity

Early exercise increases liquid wealth Because insiders holding ESOs cannot sell

or transfer their options, they may exercise options early for liquidity purposes For example, the executive may want to purchase a beach house, an expensive automobile, or

he may have a child going to college One would expect that if this were the case only a small portion o f the options at the executive’s disposal for exercise would actually be exercised

2.3 Data and hypothesis development

The primary dataset used in this study consists o f all option exercises by corporate insiders that were reported to the SEC between 1996 and 2003 The data was obtained from the Table Two File o f the Thomson Financial Insider Filing Data (TFI)

TFI defines corporate insiders as those that have “access to non-public, material, insider information” who are also required to file SEC forms 3, 4, or 5 when trading in their company stock as required by Section 16(a) o f the Securities and Exchange Act o f

1934 According to Section 16(a), insiders are required to report transactions by the tenth day o f the calendar m onth following the month in which the trade occurred We obtain all insider derivative transactions from TFI which begins in 1996 and contains all Table Two transactions and holdings information reported on SEC forms 3, 4, or 5.5 The information reported consists o f derivative transactions such as options, warrants, and convertible securities The data fields include open market derivative transactions as well

as information on the award, such as the type o f option received, number o f shares involved, strike price, date the options vest, date the options expire, exercise, and the expiration o f the stock options After the removal o f amended transactions, information

on 293,640 option exercises into common stock is obtained We restrict this sample to firms that have returns available on CRSP which further reduces our sample to 262,931 option exercises by 51,388 corporate insiders from 7,475 firms This data set is referred

to as the full sample and is used to test three hypotheses As noted previously, certain hypotheses require additional information For this purpose, we construct a subset o f the full sample that contains all insider trades in the original data set that can be matched with insiders’ compensation data reported in ExecuComp As we will see later, these restrictions reduce the total number o f transactions from 262,931 to 56,402 and include 7,662 executives from 1,907 firms This data set is referred to as the merged sample and

is used to test five hypotheses In all we examine eight hypotheses Formally, the first hypothesis is as follows:

5Form 3 - Initial statement o f beneficial ownership for all officers Form 4 - Contains information regarding changes in an insiders’ ownership position including purchases, sales, option grants, option exercises, gifts, or any other transaction that changes the insiders’ ownership position Form 5 - Annual statement o f change in beneficial ownership This form contains information regarding activity for exempt transactions not required in Form 4 Exempt transactions may include small transactions or small transfers within the company.

H I: Executives exercise options based on private information.

A straightforward test o f whether executives effectively time option exercise comes from measuring stock returns following ESO exercise We begin by defining an event as a month in which an insider o f a firm chooses to exercise a stock option For each stock option exercise, we use monthly stock return data from the CRSP database to estimate abnormal stock price performance from 12 months prior to 12 months after the event date Because managers typically sell all shares acquired through option exercise (Ofek and Yermack, 2000) we expect information to present itself in the form o f negative abnormal returns in the post-exercise period.6 Therefore, to test the hypothesis that executives exercise options based on private information we test for the presence of negative abnormal stock returns after the event dates

In the second stage o f our analysis we attempt to differentiate between exercises that c ontain i nformation a nd t hose t hat a re 1 ess informative S pecifically, w e d evelop seven hypotheses to distinguish option exercises that are likely to be associated with private information Hypotheses two and three state:

H 2: Options exercised early are based on private information and therefore exhibit

significant negative abnormal returns

H 3: Options exercised at maturity are not based on private information and therefore

exhibit no significant abnormal returns

To test hypotheses two and three we partition our sample according to whether the transaction date reported by the insider occurred prior to the maturity o f the option, where an exercise at maturity is defined as a reported exercise occurring within 30 days

6 This result is verified for our merged sample Executives in the sample sell 96% o f the stock acquired

o f the expiration date Since options not exercised upon maturity are lost, it is natural to assume that these transactions should not be based on private information Insiders that hold private negative information, positive information, or no information at all would rationally choose to exercise at this time.

Executives typically hold investment portfolios that are sub-optimally diversified, with the executive's company constituting too large a portion o f the executive's overall wealth This being the case, risk-averse executives may rationally choose to exercise ESOs to rebalance their portfolio Since ESOs cannot be exercised prior to vesting, one would expect a large amount o f portfolio rebalancing activity to occur on the vesting date Insiders that hold private negative information would still choose to exercise at this time and therefore some negative abnormal returns should be present; however, the private information effect should be distorted by portfolio rebalancing transactions which contain no private information To test this hypothesis we define an exercise on the vest date as one occurring zero to 30 days post-vest date, and partition our sample according

to whether the transaction date reported by the insider fell within this range Hypothesis four states:

H 4: Early exercises on the vesting date contain less private information and therefore

exhibit smaller negative abnormal returns than those exercised after the vesting date

Another rational reason for early exercise involves exercising to capture the underlying stock’s dividend payment Exercises that are dividend related should not contain private information To test this hypothesis we define a dividend related exercise

as one occurring within 15 trading days o f the ex-dividend date reported in CRSP and test

for the presence o f negative abnormal returns around the exercise dates in the full sample Hypothesis five therefore states:

H 5: Early exercises occurring just prior to the ex-dividend date contain less private

information and therefore exhibit smaller negative abnormal returns than those exercised after the ex-dividend date

For our three remaining hypotheses we construct a unique dataset This dataset is comprised o f all insider trades reported in the Table Two File o f TFI for the period 1996 through 2003 where the transactions can be matched with the insiders’ compensation data reported in ExecuComp We restrict this sample further to include only firms where CRSP returns are available Due to the smaller number o f firms in CRSP and fewer executives reported in ExecuComp these restrictions reduce the total number o f transactions from 262,931 to 56,402 which comprise o f 7,662 executives from 1,907 firms As noted previously, this sample is identified as the merged sample

Including the ExecuComp dataset is important for three reasons First, executives that left the company during the sample period are identified Using the departure date variable in ExecuComp options can be identified that were exercised early simply because the executive left the company and would otherwise have to forfeit them It is unlikely that early exercises forced by executive departure are exercised based on negative private information, we thus hypothesize that these trades are less informative than early exercises where the executive remains with the firm Hypothesis six therefore states:

H 6: Option exercises forced by executive departure are not based on private

information and therefore exhibit no significant abnormal returns

A second reason for including the ExecuComp d ataset is that it allows f or the calculation o f the proportion o f options exercised to those vested that were not exercised, thus providing a proxy for determining exercises that are in response to liquidity needs Using this proxy we can determine how the information content o f trades change as executives exercise more o f the options at their disposal If an insider exercises 100% o f the options available, one would expect this trade to likely contain more information than

a trade from an insider executing say only 10% o f the available options he could potentially exercise In other words, insiders that hold stronger private information are going to exercise more o f the options at there disposal This stronger private information should be revealed in higher negative abnormal post-exercise stock returns We therefore hypothesize that this proportion is positively related the amount o f private information contained in the exercise, as a larger proportion should indicate that execution was not liquidity related Hypothesis seven states:

H 7: Option exercises that are large relative to the amount o f vested options left

unexercised contain more private informative than options exercises that are small relative to the amount o f vested options left unexercised

Finally, the inclusion o f the ExecuComp database allows for the approximation of the amount o f stock acquired upon option exercise that was sold Exercises in which the stock acquired was retained in the executive’s portfolio should not have resulted from the executive holding negative information Hypothesis eight states:

H 8: Early exercises in which the executive retains a large proportion o f the stockacquired upon option exercise contain less private information and therefore

exhibit smaller negative abnormal returns than those in which the executive sold a large proportion o f stock acquired.

As noted in previous literature (see, e.g Carpenter and Remmers, 2001, and Kyriacou and Mase, 2003), it is inappropriate to use the estimation o f a standard market model to measure abnormal returns since events, which take place when options are in the money, are likely to occur after periods strong performance Therefore, we compute abnormal returns using the Fama-French (1993) three-factor model augmented by a fourth momentum factor as the benchmark (hereafter called FFM) The use o f this multifactor model is also important since we don’t consider CAPM to fully capture expected return Fama and French (1993) provide a time-series model describing excess returns as a function o f excess market returns, a small-minus-big market capitalization factor, and a high-minus-low market-to-book ratio factor Since exercises are preceded

by positive stock returns, we include momentum as a fourth factor as suggested by Carhart (1997) The four factor model is:

Rjt = cc + (3xRmt + /32SMBt + P 2HMLt + f3JJMDt + s jt (1)

where Rjt is the rate o f return o f the j th firm on month t ; Rmt is the rate o f return on the

market index on month t\ SMBt is the average return on small-minus-big market capitalization portfolios; HMLt is the average return on the two high book-to-market

portfolios minus the average return on the two low book-to-market portfolios, and

UMD, is the average return on two high prior return portfolios minus the average return

on two low prior return portfolios Each (5 parameter measures the sensitivity o f Rjt to