1

. . . 5,! ,
m,»%«.bb€:»k
“flu” ». {W

a
x.
.V.

 

.. ,,
. 3.
. . 1:1». 53”....
V .J warm V
V .4333; V .. 7!;
. . V V V. V . .V {inerTu in? a... way?
552:“. V V . .
, I .
. V . ,
213...”?
2451;553:de . ,
.2 at}! it.

x,

,.:n: V V ‘ ,V
. .me 5.: Qvumfl
pump. 3%.. .9. 143$

J31

E: h: . Jul. WWW—“4;“.

 

r .

.. ..(<.C ..
u D . 3:2 L1.
3‘ x

v . , r!
v. list-é“. _ ., p
«.3 ....«§r,himmr
‘ d 0.”?
9.. .2 5w a
390! 2.: 19
v .-V Iii .
.x. . F
I:
5.2!. .7 5..
3?}.2
‘22:: .3.
i... 5th): .1.
55...; ..
32.1. t) u .f: ‘ V
13:...- OA .. .3 l
1!st} $1911 1. £1.
.. $.52.
1‘ a1" 2

 

. 7......) .1. a V
I . y .1 V
that-30}!
1...?
.I x. x : A.

if):

Inn-.113? 1
. .3! t. L
Hi .31. . 2.1%..

n .
I.| '0’:
3 1i1i3¥%«s .
.1...‘ .i . n .

. OI. . \iy.‘
1.....13 tuna... vi.
uxixi.311{§\t ,5}.
\EV nkii .. CZ...
5I’|I¢d: 3‘11: .- '- ‘9
.wsvufinunvds .4
~ quxvsc. £6...“ 9).
I . ,3 3
,2 1.39.3. .
A
l
... $54. ‘-
: i!
{5%
. . km"...
.3
ii:..;‘3{\
4) I. 1", .

a Ln". 5.
. ... lib»... 6.13....4? I
:5... . I
A: 0A. n Airfrlfixu
u... :9. :31...)
. 1!. ,4

,0 6
E.v.,1.tv:vr\I).!
.. x . 5.331.];
. :1 . .V ‘
1.31... 30.25
.. . r: .751! A

 

nufiulvrwvmou -
.1969. 2m... .. .J.V...o._..../..._.yfl..m2 V 1 EH“. its”). Em..." . . . V
«gray .. innerhmgr , . uwdfieflkmn .55.”... if. i, fine ..... we“; , .E

 

 

i 25 UBRARY
W Michigan State
University

 

 

This is to certify that the
dissertation entitled

The Credit Market Perception of Employee Stock Options

presented by
Yen-Jung Lee

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Accounting and Information
Systems

KW Pm'

Major Professor’s Signature
7 // l O 5

Date

 

MSU is an Affirmative Action/Equal Opportr‘mity Institution

_'

""‘ .4- - 'A

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

FEBfi “09'?

{fl

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DateDue.indd-p.1

THE CREDIT MARKET PERCEPTION OP EMPLOYEE STOCK OPTIONS
By

Yen-Jung Lee

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment Of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Accounting and Information Systems

2005

ABSTRACT

THE CREDIT MARKET PERCEPTION OF EMPLOYEE STOCK OPTIONS
By
Yen-Jung Lee

This study empirically examines whether the credit market perceives outstanding
employee stock options (ESOs) as a liability-like or equity-like Obligation when assessing
the credit risk of the issuing company. ESOs create a potential obligation for a company
to transfer its own shares at a price lower than the prevailing market price in the future.
Since this obligation is settled with equity but not the company’s assets or services, ESOs
do not meet the definition of liabilities under the FASB’s current conceptual framework.
Critics argue that the settlement of ESOs involves a sacrifice Of economic benefits and
prOpose that ESOs be accounted for as a liability. Using a sample 913 firm-years
covering 338 firms for the period 2001-2003 and a sample Of 309 new bonds issued by
195 firms from January 2001 to October 2004, I find that a higher value of outstanding
ESOs is associated with a higher cost of debt, as proxied by the firm’s credit rating and
yield spread. The positive relationship between outstanding ESOs and the cost Of debt is
robust to controls of information risk, operating risk, various corporate governance
mechanisms and known determinants of corporate credit risk. In addition, the adverse
impact of outstanding ESOs on credit quality is more pronounced for firms that maintain
a policy of expending cash to repurchase shares in response to ESO exercise. Overall,
these results are consistent with the notion that credit market participants perceive
outstanding ESOs as having a liability-like impact on the company’s credit risk and thus

demand a higher risk premium to compensate for the added risk associated with ESOs.

Dedicated to my loving and supportive family.

iii

ACKNOWLEDGEMENTS

I would like to express my deepest gratitude to the members of my dissertation
committee — Dr. Kathy Petroni (chair), Dr. K. Ramesh, Dr. Marilyn Johnson and Dr. Ana
Herrera — for their continuous support and guidance during the many stages of this
dissertation. I am especially indebted to Kathy, who has been the most generous and
caring mentor a Ph.D. student could wish for and aspire to one day emulate. She opened
my eyes and mind to the excitement and challenges of academic research. She taught me
how to think and write. Her incredible patience, understanding and encouragement kept
me on track and inspired me to explore, learn and grow.

I would also like to thank all other faculty members and supporting staff in the
Department of Accounting and Information Systems. I have had the great fortune to get
to know and interact with these intelligent and supportive people over the course of my
program. I am also indebted to my fellow doctoral students with whom I went through
the ups and downs of the past five years. They have never failed to share their enthusiasm
for and comments on my work.

Finally, I am especially thankful to my family — my parents, Mou-Hsiung Lee and
Shu—Yu Lin, and my sisters, Yin-Yao Lee and Chi-Yen Lee — for always being there for
me. Their unconditional love and support help keep my life in perspective and balance.
They have been a constant source of inspiration throughout the years and have always
believed that I could accomplish what I set my mind to. It is to them that I dedicate this

dissertation.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... vi
CHAPTER 1: INTRODUCTION ....................................................................................... 1

CHPATER 2: MEASUREMENT ISSUES AND DEBT VERSUS EQUITY

CLASSIFICATION OF EMPLOYEE STOCK OPTIONS ..................................... 8
2.1 MEASUREMENT DATE DEBATE AND MARK-TO-MARKET ACCOUNTING CONTROVERSY 8
2.2 LIABILITY VERSUS EQUITY CLASSIFICATION AND THE MEASUREMENT OF ESO-

RELATED EXPENSE ........................................................................................................... 9
CHAPTER 3: RELATED LITERATURE AND HYPOTHESIS DEVELOPMENT ...... 11
3.1 RELATED LITERATURE ............................................................................................. 11
3.2 HYPOTHESIS DEVELOPMENT ..................................................................................... 15
CHAPTER 4: RESEARCH DESIGN ............................................................................... 20
CHAPTER 5: SAMPLE SELECTION AND DESCRIPTIVE STATISTICS ................. 29
CHAPTER 6: EMPIRICAL RESUTIS ........................................................................... 36
6.1 CREDIT RATING ANALYSES ..................................................................................... 36
6.2 YIELD SPREAD ANALYSES ....................................................................................... 39
6.3 ADDITIONAL ANALYSES AND ROBUSTNESS CHECKS ................................................. 42 A
6.3.1 Change Specification ....................................................................................... 42
6.3.2 Executive Versus Non-Executive Stock Options .............. > .............................. 44
6.3.3 The Cost of Debt and Corporate Governance .................................................. 45
6.3.4 Grant Date Value and Post-Grant-Date Value Change .................................... 49
CHAPTER 7: CONCLUDING REMARKS .............................................. . ...................... Sl
APPENDD( A: INTUITION BEHIND THE FAIR VALUE OF A CALL OPTION ...... 55
APPENDD( B: CREDIT RATINGS’ NUMERICAL CONVERSIONS ......................... 56
APPENDIX C: VARIABLE DEFINITIONS ................... 57
APPENDIX D: TEST FOR ENDOGENEITY OF OPTIONS IN THE CREDIT
RATINGS MODEL (RIVERS-WONG 1988 PROCEDURE) ....................................... 60
REFERENCES 91

LIST of TABLES

TABLE D.l: OLS REGRESSIONS OF THE DETERMINANTS OF OPTIONS ........... 64
TABLE D.2: OLS REGRESSIONS OF THE DETERMINANTS OF OPTIONS ........... 65
TABLE E.l: OLS REGRESSIONS OF THE OPTIMAL OPTION GRANTS .............. 68
TABLE 1: SAMPLE SELECTION ................................................................................. .69
TABLE 2: DESCRIPTIVE STATISTICS ........................................................................ 72
TABLE 3: CORRELATION MATRIX ............................................................................ 77
TABLEA: ORDERED PROBIT REGRESSIONS OF THE EFFECT OF
OUTSTANDING EMPLOYEE STOCK OPTIONS ........................................................ 79
TABLE 5: OLS REGRESSIONS OF THE EFFECT OF OUTSTANDING EMPLOYEE
STOCK OPTIONS ON YIELD SPREADS ..................................................................... 82
TABLE 6: SPECIFICATION AND SENSITIVITY TESTS ........................................... 83

vi

CHAPTER 1: INTRODUCTION

This study empirically examines whether credit market participants perceive
outstanding employee stock options (ESOs) as a liability-like or equity-like claim when
assessing the credit risk of the issuing company. This investigation is motivated by the
long—standing controversy over whether employee stock option grants are a liability or
equity of the company.I This distinction is important because it not only determines the
timing and magnitude of the ESO expense (FASB 1990; AAA FASC 1993; AAA FASC
2004) but also provides a foundation for assessing the appropriateness of mark-to-market
accounting for outstanding ESOs. Despite the crucial implications it has for many ESO
measurement decisions, the liability versus equity classification of employee stock
options has received relatively little attention in the extant empirical literature.

ESOS are financial contracts that give employees the right to purchase the
company’s own shares at a pre-determined price and at a date of the employees’
choosing. While there is no doubt that ESOs create equity claims upon exercise, it is less
clear whether during the interim period before ESOs are exercised, the company’s
continuing obligations to deliver shares at employees’ request represent a liability or
equity of the company. Because ESOs will only be exercised when the prevailing market
price of the underlying stock exceeds the exercise price of the options, advocates of the

“options-as-liability” view argue that this price difference represents an outflow of

 

' For example, in paragraph 14 of the Financial Accounting Standards Board’s (FASB) 1990 Discussion
Memorandum entitled “Distinguishing between Liability and Equity Instruments and Accounting for
Instruments with Characteristics of Both ”, the FASB states that: “. . .the Board decided that it could not
resolve accounting for stock compensation without first considering the more fundamental issues in the
broader project on distinguishing between liabilities and equity. . .”. For other examples, see AAA FASC
(1993), Balsam (1994), Rubinstein (1995), Penman (2003), Kirschenheiter, Mathur and Thomas (2004),
AAA FASC (2004) and Penman and Ohlson (2005).

economic benefits and therefore should be recognized as a liability (AAA FASC 2004).
However, under the FASB’S current conceptual framework, liabilities are defined as
“probable fitture sacrifices of economic benefits arising from present obligations of a
particular entity to transferassets or provide services to other entities in the future as a
result of past transactions or events” (Statement of Financial Accounting Concepts
(SFAC) No. 6, paragraph 35, emphasis added). Since ESOS are settled in the company’s
own shares but not the assets or services of the company, they lack an essential
characteristic of a liability and can only be classified as equity under the current
dichotomous classification framework within the balance sheet.2

This dissertation adopts an information users’ perspective to shed light on the
liability versus equity controversy of outstanding ESOS. Specifically, I investigate which
of the two views, the “options-as-equity” or “options-aS-liability” view, is more
descriptive of credit market participants’ perception of outstanding ESOS when assessing
the credit risk of the issuing company. I hypothesize that outstanding ESOS are likely to
be viewed by the credit market as having characteristics of both liabilities and equity.
This is because on the one hand, ESOS are commitments to create future equity equal to
the amount of the option exercise price if ESOS are exercised. This aspect of ESO

contracts suggests that they could have an equity-like impact on the issuers’ credit quality.

 

2 SFAS 150, Accounting for Certain Financial Instruments with Characteristics of both Liabilities and
Equity, mandates that certain obligations that require the company to convey to the holder of the financial
instrument a fixed monetary amount of value known at inception to be accounted for as liabilities because
such obligations do not establish an ownership relationship. An example given in SFAS 150 is “a financial
instrument that requires settlement by issuance of $100,000 worth of equity shares establishes something
more akin to a debtor-creditor relationship than to an ownership relationship, because it requires that the
issuer convey a fixed amount of value to the holder that does not vary with the issuer’s equity shares”
(SFAS 150, paragraph B 13). This requirement conflicts with the definition of liabilities under SFAC No. 6.
To resolve this conflict, the FASB has agreed to amend SFAC No. 6. In phase two of its liabilities and
equity project, the FASB proposes to distinguish liabilities from equity using a settlement and an ownership
criterion. ESOS will still fall under the category of equity under the FASB’s proposal.

On the other hand, however, the settlement of ESO obligations typically involves a real
cash outflow in the form of share repurchases in order to make shares available to fulfill
ESO obligations. In addition, the additional shares issued upon ESO exercise trigger the
need for the company to manage outstanding shares in order to meet its leverage or
earnings per share (EPS) targets. Since the exercise of ESOS is associated with potential
future cash outflows, outstanding ESOS are likely to be viewed as a contingent liability in
substance if not in form.

The empirical tests of this study are motivated by a debt-equity continuum
framework used by credit rating agencies to assess a debt—equity hybrid security’s impact
on the issuers’ capital structure as well as creditworthiness. This framework predicts that
if creditors consider outstanding ESOS liability-like (equity-like) obligations of the
issuing company, one would expect an adverse (favorable) effect of outstanding ESOS on
the company’s perceived credit risk, as reflected in the required rate of return for the
company’s debt securities (i.e., the cost of debt). In other words, this framework suggests
that the essential characteristic that creditors use to distinguish liabilities from equity is
whether a financial contract results in a negative impact on the company’s debt service
ability.

In this study, two proxies are used to capture the company’s cost of debt. The first
proxy is the issuing company’s credit rating, which represents the credit rating agency’s
opinion about the issuer’s overall creditworthiness. Keenan et a1. (2000) and Czamitzki
and Kraft (2004) document that credit ratings are an important predictor of the bond
default probability. As a result, it is not surprising that credit ratings are the most

important determinant of the corporate bond pricing (Gabbi and Sironi 2002) and are

commonly used as an ex ante measure of the cost of debt. The credit ratings literature,
however, also documents that credit rating agencies do not appear to incorporate all
pricing-relevant information in their rating decisions. Several studies have shown that
public accounting information exhibits incremental explanatory power beyond credit
ratings in explaining bond yields (e.g., Ziebart and Reiter 1992; Khurana and Raman
2003; Czamitzki and Kraft 2004). As a result, it is likely that credit market investors hold
a different view on the nature of outstanding ESOS than credit rating agencies. To
investigate this possibility, I use the yield Spread as my second proxy for the firm’s cost
of debt. The yield spread is the difference in yield between the corporate bond and the
risk free Treasury bond of approximately the same maturity. This spread is the actual risk
premium demanded by the credit market investors to compensate for the corporate bond’s
default risk.3

My empirical results based on a sample of 913 firm-year observations
representing 290 S&P 500 Industrial firms and 48 non-S&P 500 firms that are potential
heavy option users for the years 2001-2003 and a sample of 309 new bonds issued by 195
firms over the period January 2001 to October 2004 are consistent with the “options-as-
liability” hypothesis. Specifically, I find that the credit ratings assigned by Standard and
Poor’s are negatively associated with the fair value of outstanding ESOS after controlling
for information risk, operating risk, firm performance and other known factors that affect
corporate credit ratings. This relationship holds in both the “levels” and “change”
specifications, is robust to controls of various corporate governance mechanisms, and is

more pronounced for firms that are more likely to buy back shares in response to ESO

 

3 Using credit-default swap and corporate bond data, Longstaff et al. (2004) find that the default risk
component accounts for the majority of the corporate-Treasury yield spreads across all credit ratings.

exercise. The results from yield spread tests corroborate those from credit rating analyses
and suggest that firms with higher values of outstanding ESOS incur higher costs of
financing in the form of a wider yield spread at new bond issuance. Further analyses
reveal that the adverse impact of ESOS on yield spreads is driven by a subsample of
bonds that have a maturity longer than outstanding ESOS’ expected life an_d are issued by
firms with a propensity to repurchase shares around ESO exercise. This last finding
suggests that creditors are concerned about ESOS’ liability-like impact only when the
cash outflow associated with the settlement of the ESO obligations is expected to take
place before their loans mature.

Partitioning the value of the outstanding ESOS into the component held by top-
five executives and that held by non-executive employees, I find that non-executive
ESOs continue to explain credit ratings while executive ESOS do not.4 This finding
suggests that the negative impact of outstanding ESOs on credit ratings is not driven by
the credit market’s potential aversion to large executive ESO programs. As further
evidence on the direction of causality, I document that the change in outstanding ESOs
helps to predict the one-year-ahead change in credit ratings while the change in credit
ratings does not predict the one-year-ahead change in outstanding ESOS. Finally, I
provide evidence that the change in fair value subsequent to the grant date explains credit
ratings and yield spreads above and beyond the grant date fair value of outstanding ESOs,
which lends support to the view that credit market participants find mark-to-market

information useful when making credit risk assessments.

 

‘ For the purpose of this dissertation, top-five executives are defined as the five highest-paid executives and
non-executive employees are those other than the five highest-paid executives.

This study contributes to the ESO literature by demonstrating that the fair value of
outstanding ESOs provides value relevant information to credit market participants. Prior
studies indicate that managers have incentives to manipulate Black-Scholes assumptions
in order to underreport the value of the pro forma ESO expense (Aboody et al. 2004b;
Bartov et al. 2004; Hodder et al. 2004; Johnston-Wilson 2004). The existing literature
also documents, however, that the disclosed ESO information is sufficiently reliable to be
reflected in the frrm’s stock price (Aboody 1996; Aboody et al. 2004a; Bell et al. 2002;
Li 2002). This study extends equity market research on the value relevance of ESOs to
the credit market setting and demonstrates that credit market participants factor the fair
value of outstanding ESOs into their credit risk and bond pricing decisions.

The finding that the credit market considers outStanding ESOS a liability-like
claim that affects the issuing firm’s creditworthiness implies that the credit market looks
beyond the corporate obligation’s contractual settlement method and looks into the cash
flow consequences of these transactions. This implication should be of interest to
standard setters since the FASB has been criticized for placing too much weight on the
contractual provisions rather than economic substance of accounting transactions (e.g.,
AAA FASC 2001 and 2004). Moreover, under the Statement of Financial Accounting
Standards No. 123R (SFAS 123R), Share—Based Payment, which took effect for most
public companies for the fiscal year beginning on or after June 15, 2005, companies are
not required to disclose the fair value of outstanding ESOS. To the extent that the credit
market participants regard ESOs as imposing an increased credit risk on the issuing
company and incorporate the fair value of ESOS in their credit risk assessments, SFAS

123R’s current disclosure requirements may not provide the most decision-useful

information to financial statement users. Although evidence presented in this study does
not provide and is not intended to provide a definitive answer to the “ESOs-as-liability-
or—equity” controversy, it nevertheless complements the conceptual approach undertaken
by the FASB in determining whether a financial instrument should be classified as a
liability or equity. In addition, evidence provided by this study also has implications for
how well the standard setters’ conceptual classification of financial instruments connects
to information users’ perspective on accounting information.

The rest of the dissertation proceeds as follows. Chapter 2 provides institutional
background regarding current ESO measurement debates and describes why the debt
versus equity classifiCation is crucial to these debates. Chapter 3 reviews the related '
literature and develops my main hypotheses. Chapter 4 presents the empirical research
design. Chapter 5 describes the sample selection and provides descriptive statistics.
Chapter 6 discusses the empirical findings and robustness tests. Chapter 7 concludes this

dissertation.

CHAPTER 2: MEASUREMENT ISSUES AND DEBT VERSUS EQUITY

CLASSIFICATION OF EMPLOYEE STOCK OPTIONS

2.1 Measurement Date Debate and Mark-to-Market Accounting Controversy

The Financial Accounting Standards Board (FASB) has recently revisited
accounting for ESOs as part of its effort toward international convergence.s On
December 16, 2004, the FASB issued SFAS 123R, which is a revision of SFAS 123,
Accounting for Stock-Based Compensation, and supersedes APB No. 25, Accounting for
Stock Issued to Employees. Under FASB’S SFAS 123R, the amount of ESO
compensation expense is determined at the grant date using an option pricing model with
model inputs that are appropriate for the conditions of the option grants (grant-date
measurement). The ESO compensation is recognized as an expense over the option’s
vesting period and is matched by a corresponding increase in stockholders’ equity. There
is almost no ex-post adjustment of expenses over time. The only exception is the
adjustment for differences in the expected and realized options forfeited during the
vesting period.

Critics are generally concerned about the emphasis that grant-date measurement
places on the reliability of managers’ grant-date estimates. Since estimation errors

introduced at the grant date under grant-date measurement are not updated as time goes

 

5 The existing accounting rules governing ESOs are established by the Accounting Principles Board
Opinion No. 25 (APB 25) and the Statement of Financial Accounting Standards No. 123 (SFAS 123).
Under APB 25, the accounting charge for ESOs equals the intrinsic value, defined as the difference
between the fair market value of the underlying stock and the exercise price of the stock option, at the date
when both the number of shares and the exercise price are determinable or “fixed”. As a result, no expense
is recognized for options granted with an exercise price equal to the grant-date market price (i.e., intrinsic
value equals zero). In 1995, the FASB issued SFAS 123, which recommends that options be expensed at
the “fair market value” but also allows firms to continue reporting under APB 25, provided that additional
information about the value of the options granted is disclosed in a footnote to the financial statements.

by, both unintentional and intentional bias will accumulate in retained earnings and never
reverse in a future period. In light of the concern that estimate manipulation will go
undetected due to a lack of ex-post adjustments, analysts and academic researchers (e. g.,
Balsam 1994; Rubinstein 1995; AAA FASC 2004, Kirschenheiter et al. 2004; Wei 2004;
Nellans 2004) have called for an exercise-date measurement approach. Under this
approach, options would still be expensed when granted but the value of the options
would be recorded as a liability. This ESO liability and therefore the ESO compensation
expense will be remeasured (i.e., marked to market) at each reporting date up until the
exercise or final settlement date. Advocates of exercise-date measurement accounting
argue that this measurement approach minimizes the information damage created by
inaccurate valuation during the life of the options.
2.2 Liability Versus Equity Classification and the Measurement of ESO-Related Expense
The liability-versus-equity classification of ESOS has implications for the
measurement date controversy as well as the determination of ESQ-related expenses
because under the FASB’s Concepts Statement No. 6, comprehensive income is defined
as the change in equity of a business enterprise during a period from transactions from
nonowner sources. If one regards ESOS as equity of the company, the ownership
relationship between employees and the company is established at the date of grant. Since
a company cannot profit from transacting with its owners, ESO expense should be
measured at the grant date and any value change subsequent to the grant date should not
result in a performance statement effect. If, on the other hand, one considers ESOs a
liability that obligates the company to transfer economic resources because of the current

compensation transaction, employees are not considered owners of the company until

options are exercised. Therefore, the final amount of ESQ-related expense should be
determined at the exercise date and all changes in the value of ESOS between the grant
date and exercise date should be marked to market and flow through the income
statement each period. Consequently, fundamental ESO measurement questions such as
whether ESO compensation should be measured at the grant date or exercise date and
whether outstanding ESOs should be marked to market to reflect post-grant-date price
fluctuations depend on whether outstanding ESOS are treated as a liability or equity of the

company.

10

CHAPTER 3: RELATED LITERATURE AND HYPOTHESIS DEVELOPMENT

3.] Related Literature

This study is closely related to a stream of research that examines the liability
classification of different corporate claims. Most of the liability-equivalence studies,
however, adopt a systematic risk analysis, which focuses on how a particular corporate
claim affects the equity market’s aSsessments of the firm’s systematic risk (i.e., equity
beta). The reason for this focus is that finance theory has established the link between
debt-like obligations and equity beta. Under the assumptions of perfect markets, riskless
debt and no taxes, several studies Show that debt-like obligations affect systematic equity
risk through financial leverage (Hamada 1972; Rubenstein 1973; Bowman 1979;
Mandelker and Rhee 1984). Using the systematic risk framework, Bowman (1980) and
Dhaliwal (1986) demonstrate that unrecorded capital lease commitments and unfunded
pension liabilities are debt equivalents by showing that these two off-balance-sheet items
explain systematic equity risk. In a similar vein, Kimmel and Warfield (1995) and Cheng
et al. (2003) report that redeemable preferred stock does not have a debt-like impact on
the firm’s systematic risk. Cheng et al. (2003) further demonstrate that non-redeemable
preferred stock and minority interests are viewed as debt-like and equity-like claims,
respectively. More recently, Cheng et al. (2003) and Linsmeier et al. (2004) employ an
equity valuation analysis that links security prices with various financing instruments
with attributes of both liabilities and equity to investigate the equity market’s perceptions
of the liability/equity classifications of these instruments. These studies conclude that

whether equity market investors view financing instruments as liabilities or equity

11

depends on the issuing firm’s characteristics such as size, profitability and insolvency
risk.

This study differs from the above-cited literature in that I take a credit market’s
perspective to examine the liability versus equity debate in the ESO setting. Adopting a
credit market as opposed to an equity market approach has the advantage of providing
insights into ESOS’ claim against the issuing company rather than their claim against pre—
existing common equity holders. As analytically demonstrated by Galai and Schneller
(1978), ESOS dilute common shareholders’ ownership interest by selling employees
stock for a price less than its fair market value, which results in a transfer of economic
value from current shareholders to potential future Shareholders. Therefore, ESOS
represent an economic claim or “liability” against pre-existing common equity holders.
Consistent with Galai and Schneller’s (1978) analysis, Aboody (1996) and Li (2002) find
that equity market investors place a negative valuation multiple on outstanding ESOS in
an empirical model that regresses the market value of equity on the book value of equity,
net income (or residual income) and outstanding ESOs.

While ESOS represent a future claim on common equity holders, it is less clear
whether creditors, another important group of financial information users identified by
the FASB, perceive outstanding ESOS as equity-like or a liability-like claim against the
company.6 This is because the wealth transfer between current and future shareholders
might not reduce total economic resources available to outside stakeholders at the firm
level. In fact, the existence of ESOs suggests that the firm will have at least as much

equity cushion and in some states will have more when ESOs are exercised.

 

6 Paragraph 34 of Concepts Statement No. l contends that equity investors and creditors are primary users
of financial statements.

12

In addition, understanding the credit market’s perception of outstanding ESOS is
important in itself because the credit market provides substantially more funds to
companies than does the equity market. In 2002, US. corporations raised more than $1.2
trillion in the bond market compared to $102 billion in the equity market. Even in days
when most equity securities were considered overvalued, debt issuance was still much
larger than equity issuance.7 Given the essential role that the corporate credit market
plays in the US. economy, regulators, standard setters as well as corporate managers may
want to know whether outstanding ESOs affect the credit market’s evaluation of the
company’s debt service ability.

This study also relates to the research on the value relevance of ESO disclosures.
This strand of research has shown that although not formally recognized in the financial
statements, the pro forma ESO expense disclosed in the SFAS 123 footnote is inversely
related to the stock price, suggesting that investors consider ESO compensation as a
corporate expense that is measured with sufficient reliability (Aboody et al. 2004a; Li
2002). In addition, Aboody (1996) and Li (2002) both find a negative association
between the stock price and the value of outstanding ESOs. Li (2002) theoretically
demonstrates that this negative relationship is a reflection of the fact that current option
holders have a claim on the company’s future earnings, which reduces the value of
current equity holders’ residual interest. This study extends this line of literature by
demonstrating that the credit market also incorporates the fair value of outstanding ESOS
in their credit risk assessments even though the fair value of outstanding ESOS is not

currently disclosed in footnotes. Hence, while prior literature suggests that the

 

7 In 2000, corporate debt issuance amounted to more than $1.2 trillion while equity issuance was less than
$200 billion. (Speech by SEC Staff: Remarks at the Bond Market Association, Corporate Credit Markets.
Available at http://www.sec.gov/news/speech/spch041003a|b.htm)

13

information contained in outstanding ESOS is relevant for equity valuation, the findings
in the current study indicate that this information is also relevant for default risk
assessments by creditors.

Research on the consequences of the balance sheet classification suggests that the
debt-equity distinction affects financial statement users’ as well as prepares’ behaviors.
For example, Hopkins (1996) experimentally demonstrates that the accounting
Classification of mandatorily redeemable preferred stock (MRPS) on the balance sheet
affects buy-side analysts’ stock valuation judgments. Specifically, he finds that analysts
presented a balance sheet with mandatorily redeemable preferred stock classified as a
liability (equity) made common stock price judgments as if the firm had issued debt
(equity). He also documents that analysts appear to incOrporate more attribute-type and
less category-type information in stock price judgments when MRPS are classified in a
mezzanine zone. Existing research also demonstrates that financial statement preparers
incur a significant cost in order to avoid the liability-classification on the balance sheet.
Engel et al. (1999) examine trust preferred stock, which is treated as preferred stock for
financial reporting purposes and as debt for tax purposes, and is thus a tax-equivalent
source of capital to regular debt. They find that firms are willing to pay between $10 and
$43 million to replace debt with trust preferred stock in order to reduce their reported
leverage ratio by approximately 12.8 percent, on average.

Finally, extant research also seeks to identify the accounting treatment that best
depicts the economic reality of employee stock options. Kirschenheiter et al. (2004) and
Ohlson and Penman (2005) argue that an accounting method that treats ESOS as a

liability at the grant date and continuously marks the options liability to market produces

14

accounting measures that are more reflective of the fundamental firm performance and
common equity value. Landsman et al. (2004) theoretically and empirically substantiate
that the aforementioned accounting treatment reflects ESOs’ dilution effect on existing
shareholders’ value more faithfully than the accounting treatments prescribed by the APB

25, SFAS 123 and SFAS 123R.

3.2 Hypothesis Development

To investigate whether credit market participants perceive outstanding ESOs as a
liability-like or equity-like claim, I rely on credit rating agencies’ debt-equity-continuum
framework to formulate empirical predictions. When assessing a debt-equity hybrid
security’s impact on the issuer’s capital structure, credit rating agencies (e.g., Standard &
Poor’s, Moody’s, Fitch and AM. Best) typically adopt a framework, which places the
financial contract on a debt-equity continuum based on the contract’s cash flow flexibility
relative to straight debt and common stock. Credit rating agencies argue that common
equity enhances the firm’s creditworthiness because it affords the issuer the highest
possible level of cash flow flexibility in terms of delaying periodic and principal
payments and providing a protective cushion in the event of bankruptcy (S&P Corporate
Ratings Criteria and Moody’s Fundamental Rating Methodologies). Consistent with the
rating agencies’ argument, Elliott et al. (2004) find that bondholders experience positive
abnormal returns on the announcements of seasoned equity offerings. The abnormal bond
returns are larger for bonds with longer maturities and poorer bond ratings, consistent
with the notion that bondholders benefit from the reduction in credit risk through an

increase in common equity. Therefore, if credit market participants regard a financial

lS

contract as equity of the issuing company, one would expect the contract to have a
favorable impact on the company’s credit profile. In contrast, from the creditors’
perspective, straight debt is the least preferred source of corporate financing because it
calls for fixed interest payments and requires repayment at maturity. The need to fund
these payments with cash flows creates a liquidity risk and thus has a negative impact on
the issuers’ creditworthiness. With straight debt and common stock as two extremes of
the debt-equity continuum, the location of a financial contract on the continuum
summarizes credit rating agencies’ Opinion about how debt-like (equity-like) a financial
contract is. The closer a financial contract gets to the equity (debt) end, the more (less)
“equity credit” it receives and the more (less) favorable impact it has on the company’s
credit quality. Given that ESOS may carry characteristics of both liabilities and equity,
the debt-equity continuum provides a useful framework to examine whether the debt-like
or equity-like attributes of ESOs dominate in credit market participants’ credit risk
decisions.

The issuance of ESOs can also be thought of as selling future equity to employees.
The issuing company collects the first installment of the transaction price (i.e., the
premium or fair value of the options) in the form of employee service. If employees
decide to pay the second installment of the transaction price in cash equal to the amount
of the option exercise price, new shares are issued. Even if employees decide not to pay
the second installment, the company still retains the benefit of the employee services
without having to expend cash.8 As a result, ESOs can be viewed as part of the new

equity that will potentially be created in the future, which suggests that outstanding ESOs

 

8 Alternatively, one can think of the issuance of ESOs as a means of financing compensation expenses by
selling promises to create equity in the future. Firms with outstanding ESOS should have more cash to
service debt than they would otherwise do if they had chosen pay employees in cash.

16

are likely to be perceived as equity—like contracts of the company. If credit market
participants consider outstanding ESOs equity-like in their credit risk assessments, the
fair value of ESOS at the grant date, which represents the premium of the options paid for
by employees in service or the amount of “equity” contributed by employees with service
in installments, should have a favorable effect on the company’s credit profile.

Unlike the case of common equity, however, the existence of outstanding ESOS
does not have an unequivocal credit-enhancing effect on the company’s financial position
because firms are obligated to deliver shares to employees upon the exercise of ESOs.
These obligations resemble the repayment requirement of the straight debt and the need
to fund ESO ‘maturity” imposes an increased credit risk on issuers. Companies typically
have two means of satisfying ESO exercise. They can either distribute pre-existing shares
acquired from the open market or issue new shares directly to employees. For’firms that
repurchase shares to redeem ESOS, the settlement of ESO obligations clearly involves a
cash outflow, which reduces net assets available for debt service and thereby directly
increases firms’ credit risk. For firms that choose to issue new Shares to employees, the
settlement of the ESO obligations could still have an indirect impact on their credit risk.
Graham and Harvey’s (2001) survey evidence suggests that firms typically have a target
capital structure and are concerned about the EPS dilution when issuing equity. The
issuance of common stock as a result of ESO exercise increases the number Of shares
outstanding and moves the capital structure and EPS number away from the firm’s
targets. Therefore, companies have the incentive to launch share repurchases to achieve
target capital structure or to offset the dilutive effect of ESO exercise. Consistent with

above survey findings, anecdotal (e. g., Standard and Poor’s 2005) as well as empirical

l7

evidence suggests that the use of ESOS is often accompanied by ongoing share
repurchase programs. Weisbenner (2000) provides evidence that firm’s outstanding ESOS
are a strong predictor of subsequent share repurchases. He also demonstrates that firms
engage in ongoing repurchases to undo much of the dilution to earnings per share (EPS)
resulting from past stock option grants. Kahle (2002) finds that firms are more likely to
announce a share repurchase program when total exercisable options are high and when
many options have recently been exercised. Bens et al. (2003) document that firms
increase repurchase intensity when ESOs’ impact on EPS is more dilutive. Moreover,
Bens et al. (2002) find that firms divert cash from R&D and capital expenditures to fund
Share repurchase programs in order to mitigate EPS dilution from large ESO exercise.
To the extent that companies repurchase shares in response to option exercise, the
settlement of ESOS involves future cash outflows equal to the excess of the cost of
reacquiring shares over the proceeds from option exercise. These net cash outflows can
be approximated by the intrinsic value of the ESOS at the exercise date.9 Since the
exercise-date intrinsic value depends on unknown future stock price, the fair value of an
stock option, which is computed as the present value of the probability-weighted
exercise-date intrinsic value, serves as a reasonable estimate of the potential cash
outflows associated with the settlement of the existing ESO obligations at any point in
time (see Appendix A for the intuition behind the fair value of a call option). Ideally,
creditors will treat outstanding ESOS as an economic liability only when the company
chooses to repurchase shares in response to option exercise. In reality, creditors do not

have the perfect foresight as to whether ESO-related repurchases will take place in the

 

9 An employee stock option is in fact a call option issued by a company that gives the holder the right but
not obligation to purchase the underlying stock. The intrinsic value of a call option is defined as the
maximum of zero and the difference between option exercise price and the share market value.

18

future. Given creditors’ tendency to focus more on the lower end of cash flows,10

outstanding ESOS are likely to be viewed as a contingent liability, which could trigger a
call on future cash flows and impair the company’s debt service ability.

As a result of the above discussion, my first hypothesis is as follows, stated in
alternative form:

H I : Ceteris paribus, the fair value of the outstanding employee stock options is

negatively (positively) associated with the issuing company’s perceived credit risk

if the credit market perceives outstanding employee stock options as equity-like

(liability-like) obligations of the issuing company.

If creditors are unsure about whether a firm will engage in ESQ-related
repurchases in the future, they are likely to price ESO obligations based on their belief
about the likelihood that future repurchases will occur. Companies that historically
repurchase shares around option exercise are more likely to continue the practice in the
future. As a result, ESOs issued by firms with a propensity to repurchase shares in
response to option exercise are more likely to be regarded as liability-like obligations by
credit market participants. This leads to my second hypothesis as follows, stated in
alternative form:

H2: Outstanding employee stock options have a more adverse impact on the

perceived credit risk when the issuing company maintains a policy of
repu rchasing shares in response to ES 0 exercise.

 

'0 Creditors care more about the frrm’s downside risk than upside potential because when the frrm’s cash
flows are sufficient to cover the loan, creditors only received promised interest and principal payments and
do not participant in further firm success. However, when the firm is unable to generate sufficient cash flow
to cover the loan, creditors lose the entire contracted amount (Watts 2003).

19

CHPATER 4: RESEARCH DESIGN

To test my first set of hypotheses on the credit rating sample, I estimate the
following ordered probit model to assess the impact of outstanding ESOs on credit
ratings. An ordered probit model is selected over an ordinary least square (OLS) model
because the former takes into account the discrete and ordinal nature in the dependent

variable. My credit rating model is constructed as follows:

RATINGI.’ t = alOPTIONSl.’ t 4- (12,4 C CQUALITYI.’ t + a3STDROAi’ t + a4STDCF0i, t

+0155 TDRETi, t + a6R0Ai, t + a7 CASHFLOWSl.’ t + aSLEVERAGEi, t + aQSIZEiJ

BETA. + a
t,t

+6! 11

10 CAP_INTENi,t+a

121NT_COVi,t +0:

13 t

32M + 8i,
where i, t are firm and year indexes, respectively. (Credit Rating Model)

RATING is the long-term issuer credit rating assigned by Standard & Poor’s six
months after the fiscal year end. I allow a six-month lag between the close of the fiscal
year and the production of credit ratings to ensure that rating agencies have sufficient
time to incorporate the stock option information in their credit rating decisions.ll
Following Ashbaugh et al. (2004), I recode the credit ratings data (Compustat Quarterly
data SPDRC‘) to remove unassigned codes and collapse the ratings into seven categories,
with a higher value of RATING indicating a better credit rating. Appendix B provides the
numerical transformation of credit ratings.

OPTIONS is the fair value of outstanding ESOs as of the end of the fiscal year
scaled by total assets. Following Core and Guay (2002) and Li (2002), I value a firrn’s

outstanding ESOS using the Black-Scholes option valuation model and assume that they

 

” My results are not sensitive to the use of a three-month lag instead of a six-month lag between the end of
the fiscal year and the production of the credit ratings in my empirical analyses.

20

were a single option grant (see Appendix C for more details about the measurement of
OPTIONS). If credit rating agencies view outstanding options as a liability-like claim, I
expect a, to be negative, which suggests that, other things being equal, a higher level of
outstanding ESOs is associated with a lower credit rating.

A growing body of literature documents that stock-based compensation,
particularly ESOs, creates perverse incentives for managers to misrepresent true financial
performance. For example, Bartov and Mohanram (2004) demonstrate that top executives
inflate reported earnings to boost Short-term stock price in order to increase the cash
payout of ESO exercise. Cheng and Warfield (2005) find that managers with more equity
incentives are more likely to smooth earnings to meet or just beat analysts' forecasts.
Erickson et al. (2004) document that firms granting more executive stock-based
compensation are more likely to commit accounting frauds. Efendi et al. (2004) report
that Chief Executive Officers’ (CEO) Stock Option compensation is a strong predictor of
accounting restatements. Collectively, this line of literature suggests that higher levels of
executive stock-based compensation are associated with poorer accounting reporting
quality. To the extent that executive option holdings are positively correlated with
outstanding ESOs, the positive correlation between outstanding ESOs and the cost of
debt could reflect the risk premium charged by the credit market for the higher
information risk associated with poorer financial reporting quality (i.e., Sengupta 1998;
Francis et al. 2005).

To control for this alternative explanation, I include in the credit rating model the
firm’s accrual quality, ACCQ UALTY, to proxy for the information risk derived from the

firm’s financial reporting quality. Francis et al. (2005) demonstrate that accrual quality is

21

priced by the debt and equity markets and therefore is a valid proxy for information risk
that cannot be diversified away. Following Francis et al. (2005), ACCQUALITY is
measured based on the standard deviation of the residuals from the cross-sectional

Dechow-Dichev (2002) regression and modified by McNichols (2002) as follows:

TCAi, t = CF 0i, t —l CF01; t CF01; t +1 AREVi’ t PPEI.’ t
Assets. 0 1 Assets. 2 Assets. 3 Assets. 4 Assets. 5 Assets.
l,l I,t I,t t,t l,l I,t
4-81., t
(1)
where TCAL, = firm i’s total current accruals in year t, defined as (ACA;,,-ACL,-,, -ACash.-,,
+ASTDEBT.-,,),
ACAL, = firm i’s change in current assets (Compustat #4) between year t-l and
year t,

ACLL, = firm i’s change in current liabilities (Compustat #5) between year t-l and
year t,

ACashl-J = firm i’s change in cash (Compustat #1) between year t-l and year t,

ASTDEBTL, = firm i’s change in debt in current liabilities (Compustat #34)
between year t-l and year t,

CF 0,; t = NIBEL, — TAM = firm i’s cash flows from operations in year t,

NIBEL, = firm i’s net income before extraordinary items (Compustat #18) in year
t.

TAM = firm i’s total accruals in year t, defined as (ACAiJ-ACLL, -ACash,-,t
+ASIDEBTiJ-DEPRLI), '

DEPRL, = firm i’s depreciation and amortization expense (Compustat #14)
between year t-l and year t,

Assets“ = firm j’s total assets at the end of year t,

ARE Vi, , = firm i’s change in revenues (Compustat #12) between year H and
year t,

PPEiJ =firm i’s gross value of property, plant and equipment (compustat #7).

22

Equation (1) is estimated cross-sectionally for each two-digit SIC industry group
with at least 15 firms for each year.‘2 Since a larger standard deviation indicates poorer
accrual quality, ACCQ UALITY is therefore measured as the negative of the standard
deviation of the resulting firrn- and year-specific residuals, calculated over year t-4
through t. Consistent with Francis et al.’s (2005) finding that better accrual quality is
associated with a higher credit rating, I expect the coefficient on ACCQUALITY to be
positive.13

My next set of explanatory variables controls for the operating risk associated
with ESO programs. Because the value of ESOS increases in stock price volatility,
several studies have hypothesized and found that the use of ESOS is associated with
greater management risk-taking behaviors. Raj gopal and Shevlin (2002), Li (2004) and
Coles et al. (2003) document that ESO incentives are positively correlated with the
possibility that firms choose riskier projects. Consistent with the notion that ESOs induce
risk-taking, the extant research also documents that CEOs’ ESO incentives are positively
correlated with contemporaneous return volatility (Guay 1999; Cohen et al. 2000), future
return volatility (Coles et al. 2003; Hanlon et al. 2004), future cash flow volatility
(Rajgopal and Shevlin 2002) and future earnings volatility (Hanlon et al. 2004). Since
higher operating risk may increase the probability that the firm’s cash flow realizations at

any given period will fall below the required debt service level (Minton and Schrand

 

'2 Firms that are excluded from my final sample because of insufficient observations in their two-digit SIC
industries to estimate the Dechow and Dichev (2002) regression do not appear to concentrate in any
particular industry. Industries that drop out of the Dechow and Dichev (2002) regression include mining,
construction, tobacco, retail lumber and person services.

'3 As a robustness check, I also include the option sensitivity measure described in Core and Guay (1999)
to proxy for managers’ earnings management incentives since Cheng and Warfield (2005) Show that this
measure is associated with managers’ tendency to smooth earning to meet or beat analysts’ earnings
forecast. Option sensitivity is defined as the expected change in top-five executives’ option wealth from a
one percent change in stock price. The coefficient on the option incentive measure is insignificant and none
of my results is affected by including this incentive measure in credit rating and yield spread models.

23

1999), creditors are unlikely to favor the increased operating risk associated with ESO
programs. Consistent with the above discussion, DeFusco et al. (1990) find that
announcements of new management ESO plans are associated with negative abnormal
bond returns. The authors also document an increase in stock return variance after the
approval of executive Option plans. Consequently, the positive correlation between
outstanding ESOs and the perceived credit risk could be a manifestation of creditors’
aversion to operating risk. To control for this possibility, I include three operating risk
proxies in the credit rating model: (i) the standard deviation of stock returns (S TDRET);
(ii) the standard deviation of income before extraordinary items deflated by total assets
(STDROA); and (iii) the standard deviation of operating cash flows deflated by total
assets (STDCFO), all calculated over the five-year-period between t-4 and t.14 The signs
for these operating risk proxies are expected to be negative.

One of the most commonly cited reasons to compensate employees with ESOS is
that stock options encourage employees to think like owners and thus mitigate agency
problems (Ittner et al. 2003). Consistent with this argument, Hanlon et al. (2003)
document that the use of executive stock options is associated with better future operating
performance. Therefore, I include in the credit rating model firm profitability (ROA) and

cash flows from operating activities ,(CASHFLOWS) to control for the realized

 

“ I also estimate credit rating and yield spread models including the measure of CEOs’ ESO risk-taking
incentives used in Guay (1999), Raj gopal and Shevlin (2002), Coles et al. (2003), and Hanlon et al. (2004).
Specifically, CEO’s ESO risk-taking incentives are calculated as the Black-Scholes partial derivative of
option value with respect to a one percent change in stock-retum volatility and then weighted by the
number of options held by the CEO. The inclusion of this variable does not change the direction and
significance of the coefficient on OPTIONS and the coefficient on E80 risk- taking incentives is
insignificant In the credit rating and yield spread models. In addition, including R&D expenditures as an
additional proxy for the operating risk (Shi 2003) does not alter my results.

24

motivational benefits of ESOS.” In addition, I also control for other firm characteristics
that have been identified by prior research as determinants of corporate credit ratings
(e.g., Kaplan and Urwitz 1979, Ziebart and Rieter 1992, Bhojraj and Sengupta 2003,
Ashbaugh et al. 2004). These firm characteristics include leverage ratio (LEVERAGE);
firm size (SIZE); systematic equity risk (BET A); capital intensity (CAPJNTEN); interest
coverage ratio (INT _COV) and book to market ratio (BZM). See Appendix C for the
measurement of these control variables. The expected signs for these firm characteristics
based on prior literature are presented in Table 4 through Table 6.

The yield spread model is similar to the credit rating model with two exceptions.
First, the yield spread model is estimated using the OLS regression because the
dependent variable, yield spread, is now a continuous variable. Second, aside from the
issuers’ characteristics identified in the credit rating model, I also controlled for four
issue-specific characteristics that have been demonstrated by prior studies to affect yield
spreads of new bond issues (Sengupta 1998; Khurana and Raman 2003). My yield spread

model is estimated as follows:

SPREADi t =a +alOPTIONSi t +6: ACCQUALITYl. t +0: STDROAi t+a SIDCFOi t

O 2 3 4
+aSSTDRET;.’ t + a6R0Ai, t + a7 CASHFLOWSM + a8LEVERAGEl.’ t + a9SIZEi, t
+a1 OBETAI', t + a1 lCAP _ INTENI.’ t + a1 2INT _ COVi, t + a1332M + a14MATYRSl.’ t
+011 SISSUESIZEI.’ t + a1 6CALLABLE1.’ t + a17SUBORDl.’ t + 8i, t

(Yield Spread Model)

 

'5 My empirical results are unaffected by the inclusion of Tobin’s Q and analysts’ long-term growth
forecasts as proxies of the expected incentive benefits of ESOS. In addition, my results are robust to the
inclusion of the pro forma ESO expense in the credit rating and yield spread models. This latter finding
ensures that outstanding ESOS do not simply serve as a proxy for the ESO expense in credit market
participants’ credit risk assessments.

25

SPREAD is the difference between the yield to maturity on newly issued
corporate bonds and the yield to maturity on US. Treasury bonds of approximately the
same maturity at the issue date.“5 This Spread captures the default risk premium charged
by bondholders and comprises the largest component of the firm’s cost of debt (Sengupta
1998). I examine new bond issues because many seasoned corporate bonds are not
actively traded. Therefore, the prices for these bonds are established based on the matrix
pricing methodology, which estimates the bond's value by comparing the bond to other
bonds with similar provisions that are traded at about the same time, and may not fully
reflect the credit market’s actual beliefs toward the issuers. In addition, since new bond
issues are typically preceded by intensive data acquisition and analyses by the credit
market, the use of new issues ensures that offering yields impound the most up-to-date
information so that the usual concern about bond price inefficiency (Katz 1974) is
mitigated. In order to allow the credit market investors enough time to process financial
reporting information, I match yield spread data of bonds issued in year t+1 with the
issuer’s year t financial and stock Options information.'7 I expect a; to be positive,
suggesting that higher levels of ESOS are associated with a wider yield spread and
therefore a higher cost of debt.

The four bond issue characteristics included in the yield spread model are: (1)
years to maturity (MA TYRS); (2) the log of the issue size in million dollars (ISSUESIZE);

(3) whether the issue is callable (CALLABLE); and (4) whether the bond is a subordinated

 

'6 Yield to maturity is the discount rate necessary to make the present value of the bond’s principal and
interest payments equal to the issue price. As a result, a higher yield to maturity translates into a lower issue
price and thus a higher cost of debt.

7 Since firms are not required to disclose detailed information about ESO plans such as the number and the
weighted average exercise price of outstanding ESOS in quarterly financial reports, I am unable to match
new bond data with quarterly ESO information.

26

debt (S UBORD). I expect a positive Sign for MA TYRS because the yield curve is
generally upward sloping during the 2001-2003 period.18 As argued by Khurana and
Raman (2003), the sign for ISSUESIZE is ambiguous because on the one hand, large
bond issues could lower the transaction costs due to the economies of scale in the
underwriting process. On the other hand, the large issue Size might result in the difficulty
in placing the bond issue and therefore increase the borrowing cost. Given these two
conflicting effects, I make no prediction on the signfor ISSUESIZE. The Sign for
CALLABLE is expected to be positive since the bondholders will demand a premium for
giving the company an option to redeem the debt prior to maturity. I also expect the Sign
for S UBORD to be positive because subordinated debts have substantially less protection
in the event of bankruptcy.

To test H2 that outstanding ESOS have a more adverse impact on the perceived
credit risk when the company maintains a policy of repurchasing shares in response to
option exercise, 1 augment the credit rating and yield spread models by introducing an
indicator variable, REPUR, and an interactive term, REPUR*OPTIONS. REPUR is set
equal to one if the firm has a tendency to repurchase Shares surrounding the exercise of
E808 and zero otherwise. Identifying companies with such a tendency, however, is no
easy task because companies can buy back shares for various reasons, from signaling
undervaluation to fending off unwanted takeovers, from distributing excess free cash
flows to maintaining target leverage ratios. Moreover, companies are under no obligation
to disclose their intentions behind repurchase programs, which makes it harder to

extrapolate the likelihood of future ESO-related cash outflows from historical repurchase

 

'8 Data source: US. Treasury. Available at: http://www.treasury.gov/ofiices/domestic-finame/debt-
management/interest-rate/yield-hist.htrnl.

27

behaviors. To better identify firms that engage in ESO-related repurchases, I estimate the
following Tobit model for each Tum using data from 1992 to 2003. The sample period

starts in 1992 because it is the first year for when ExecuComp data is available.

REP” = ,60 + ,BIEXERM + gm (2)

where REP is the dollar amount Of repurchases, computed as Compustat reported
purchase of common and preferred stock for year t (Compustat #115) redUced by the
decrease in preferred stock (Compustat #130) from year t-l to year t.'9 EXER is the total
value of ESOS exercised (ExecuComp SOPTEXER) in years t and t+l summing across
top five executives.20 The rationale behind this Tobit model is that if a firm repurchases
shares in response to current and upcoming ESO exercises, its contemporaneous and one-
year ahead ESO exercises should exhibit some predictive power about the dollar amount
it spends on common Share repurchases. Therefore, REPUR is coded One if [31 is
significantly positive at the 5% level and zero otherwise. I expect a negative (poSitive)
coefficient on REP UR *OPTIONS in the credit rating (yield spread) model because the
potential cash outflows as a result Of the settlement of ESO obligations are more likely to

occur when the firm has a propensity tO buy back shares around Option exercise.

 

'9 Stephens and Weisbach ( 1998) point out that Compustat #1 15 includes the purchases Of treasury stock,
the conversions of preferred stock into common stock, the conversions Of class A, B and special stock into
common stock, and the retirement and redemption of both common stock and preferred stock. Since this
dissertation is interested only in the dollar amount spent on repurchasing common stock, adjusting for the
decrease in preferred stock removes some noise contained in Compustat #1 15 (Dittrnar 2000).

2° Ideally, I would like to use the value of ESOs exercised by all employees as the predictor of the dollar
amount spent on repurchasing. This variable, however, requires costly manual data collection for an
extended period of time necessary to estimate firm-specific regressions. As a result, I use the value of ESOs
exercised by top five executives as a proxy for the value Of ESOs exercised at the firm-wide level.

28

CHAPTER 5: SAMPLE SELECTION AND DESCRIPT IVE STATISTICS

Panel A of Table 1 summarizes the sample selection process for my credit rating
sample. The initial sample begins with 497 S&P 500 firms from the ESO database
maintained by Jack Ciesielski of RC. Associates, Inc. and 139 non-S&P 500 firms that
are potential heavy Option users from the ExecuComp database. This sample is chosen to
ensure that sample firms have substantial outstanding ESOS, which reduces the
possibility that immaterial ESO Obligations Obscure creditors’ assessment Of the ESO’S
impact on firms’ creditworthiness.21 The sample period is limited to the three-year period
2001-2003 because detailed stock Option information has to be manually collected from
companies’ 10-K filings for non-S&P 500 firms. I also hand collect additional stock
Option information necessary for my empirical analyses that is not available in RC.
Associates’ ESO database for S&P 500 firms.

TO identify the non-S&P 500 heavy Option users, I divide the number of
individUal executive Option grants (ExecuComp data N UMSECUR) by the percentage of
total Option grants in a given year represented by the executive Option grants
(ExecuComp data PC TTOTOP) to Obtain the number of total Option grants for the year at
a firm-wide level. An option usage ratio is calculated by dividing the number of total
Option grants by outstanding common shares at the end Of the year and averaging over the
three-year period between 2001 and 2003.22 All non-S&P 500 firms are ranked within

their S&P index groups (i.e., S&P MidCap, S&P SmallCap and Non-S&P 1500 firms) by

 

1' To assess whether my credit rating results are driven by heavy option users, I re-estimate the credit rating
model using only S&P 500 companies. The tenor of my results remains qualitatively similar.

22 To be included in my sample, firms must have executive compensation data for year 2003 in
ExecuComp. ExecuComp is updated four times each year in January, April, July and October. The data
used in this study are extracted from ExecuComp after the July 2004 update.

29

the Option usage ratio. I retain firms that are ranked in the top 80 within their S&P index
groups and exclude from this sample fnms that do not have at least five years of data on
Compustat and ExecuComp because some of my variables are measured over the past
five years. These procedures reduce my non-S&P 500 sample to 139 firms.

I exclude firms in the utilities (SIC=4900-4999) and financial services
(SIC=6000—6999) sectors because of their unique industry and risk characteristics. My
credit rating analyses further require firms to have credit rating information on Compustat
and all other necessary data on Compustat, CRSP and ExecuComp for my credit rating
model. Finally, I eliminate firm-years that have voluntarily chosen to report ESOS on
financial statements following SFAS 123’s fair value method to ensure that the equity
labeling of ESOs on the balance sheet mandated by SFAS 123 does not confound
creditors’ perception of outstanding ESOS. 23 These procedures yield a final sample of
913 firm-years, representing 338 firms. S&P 500 firm-years constitute 86.4% Of the
credit rating sample, followed by S&P MidCap, Non-S&P, and S&P SmallCap firm-
years, which makes up 6.4%, 4.8% and 2.4% of the sample, respectively. There is some
industry concentration in my credit sample with durable manufacturers, computers and
retail industries making up over 50% of the sample. As discussed later in the dissertation,
the empirical results in this dissertation are robust to the inclusion Of industry controls.
See Panel B of Table l for the industry and S&P membership distributions for the credit

rating sample.

 

23 Recall that Hopkins (1996) shows that buy-side analysts price a firm’s common stock as if the firm had
issued a debt (equity) security when the security is classified as a liability (equity) on the balance sheet.
This finding suggests the need to exclude firm-years that voluntarily report under the SFAS 123’s fair value
method because information users’ responses to outstanding ESOS are likely to be influenced by the equity
classification on the balance sheet mandated by SFAS 123.

30

Panel C of Table 1 summarizes the sample selection procedures for my yield
spread sample. My initial sample starts with 1039 straight (nonconvertible), fixed
(nonfloating) rate bonds issued by US industrial companies from January 2001 to
October 2004. Data pertaining to new bond issues are drawn from the Securities Data
Company (SDC). If the company has more than one bond issue in a particular year, I
retain only the bond with the largest principal. After excluding bond issues without all
necessary data for the yield Spread model and those that voluntarily adopt the fair value '
method under SFAS 123, my final yield spread sample comprises 309 new bonds issued
by 195 firms, 78%, 14%, 3% and 5% of which are issued by S&P 500, S&P MidCap,
S&P SmallCap and Non S&P firms, respectively. The industry distribution in Panel D Of
Table 1 indicates that durable manufacturers and retail industries are two of the most
representative industries in the yield-spread sample.

For each of my sample firm-years, I obtain stock Option data from either R.G.
Associates’ ESO database or the firm’s SFAS 123 footnote from its 2000 to 2003 lO-K
filings. These data items include the number and weighted average exercise price for ’
outstanding stock options at the end of the year, the maximum contractual life of the
Options, the weighted average remaining contractual life of outstanding stock Options,
and inputs to the Option valuation model (i.e., expected price volatility, expected Option
life, risk-free interest rate and dividend yield).

Panel A of Table 2 reports descriptive statistics for my credit rating sample. As
with Ashbaugh et al. (2004), this study classifies the credit ratings into investment and
speculative (BB-I- or worse) grades to facilitate univariate analyses and allow more

intuitive interpretation Of the economic significance of my empirical results. The median

31

investment- (speculative) grade firm-year has a credit rating of about 5 (3),
corresponding to the S&P letter grade between A- and A+ (BB- and BB+). The fair value
of outstanding Stock Options as a percentage of total assets, OPTIONS, is significantly
higher for firm-years that receive a speculative-grade rating than those that receive an
investment-grade rating, with a mean (median) value of 9.0% (3.2%) and 4.2% (2.5%),
respectively. This difference is significant at the 1% level both in terms of mean and
median. Partitioning OPTIONS into the component held by top-five executives,
MGTESO, and that held by non-executive employees, NONMGTESO, I find that the
mean MGTESO and NONMGTESO are both higher for speculative—grade than for
investment-grade firm-years although the median NONMGTESO is not significantly
different across these two rating groups. Investment-grade firm—years have significantly
better mean and median accrual quality than speculative grade firm-years. The mean
(median) firm with a speculative grade has Significantly higher volatility in terms of its
accounting income, cash flows, and stock returns, which are consistent with the notion
that credit rating agencies dislike Operating uncertainties. Consistent with prior research,
firms receiving better credit ratings are those with lower financial leverage, higher
profitability, more abundant operating cash flows, lower systematic equity risk, higher
levels Of capital intensity, greater growth Opportunities, higher interest coverage ratios
and are larger in size. About 21.3% (15.7%) Of the investrnent- (speculative-) grade finn-
years are classified as having a tendency to repurchase shares in response to ESO
exercise (REP UR).

Panel B of Table 2 reports summary statistics for selected variables in my yield

spread sample. Similar to Panel A of Table 2, I Split the entire sample in half based on

32

yield spreads to facilitate univariate comparisons. Firm-years having above (below) the
sample median yield spread are labeled as high (low) spread firm-years in Panel B Of
Table 2. Turning first to the bond issue characteristics, the mean yield to maturity for the
high- (low-) spread bonds is 267.6 (89.9) basis points above the yield on US. Treasury
bonds Of similar maturity at the time of bond issuance, with a median of 232.0 (90.5)
basis points. As expected, low-spread bonds are substantially more likely to have an
investment grade rating than high-Spread bonds.24 The mean (median) time to maturity
for the high- and low-spread bonds is 8.5 (10) and 7.7 (7) years, respectively, with the
high spread bonds having a slightly longer maturity. High-Spread bonds are more likely
to be subordinated and callable bonds. The size Of the bond, however, does not differ
across high- and low-spread subsamples.

Inconsistent with the result from the credit rating sample, the high spread sample
has a smaller OPTIONS than the low-Spread sample and this difference is significant for
both the mean and the median. For the mean high- (low-) spread sample, the fair value Of
outstanding ESOS is about 3.1% (4.3%) of total assets, with a median of 1.9% (2.5%).
There is no significant difference in the value of executive ESOS scaled by total assets
across high- and low- spread firm-years. Consistent with Panel A of Table 2, low spread
firm-years have better accrual quality and lower operating risk. Univariate tests on other
control variables are generally consistent with prior studies. Specifically, the issuing
firms are generally profitable and have positive Operating cash flows. The low-spread
firm-years have lower LEVERAGE, higher ROA, higher CASHFLOWS, higher

INT _COV, and are larger in size. The percentage Of firm-years exhibiting a tendency to

 

7‘ Firms in the new bond sample enjoy higher issuer credit ratings than firms in the credit ratings sample.
On average, firms in the credit ratings (yield spread) sample has a credit rating of BBB+ (A-).

33

repurchase shares around ESO exercise does not differ across high- and low—spread firm
years.

Panel A Of Table 3 reports pairwise correlations among variables used in my
credit rating model. Pearson product-moment (Spearman rank-order) correlations are
presented in the upper right (lower left) triangle. OPTIONS is negatively correlated with
RATING in terms of the Pearson correlation at the 1% level but the Spearrnan correlation
between these two variables is insignificant. ACCQUALITY is positively correlated with
RATING, which is consistent with Francis et al.’s (2005) finding that higher accrual
quality is associated with a better credit rating. Consistent with Panel A Of Table 2 and
the prior literature, I find that ROA, CASHFLOWS, SIZE, CAP_INTEN, and INT _COV
are positively associated with credit ratings while STDRET, STDROA, STDCFO, BETA,
LEVERAGE and BZM are negatively correlated with credit ratings. OPTIONS is
positively correlated with Operating risk measures, consistent with the notion that stock
options give employees risk-taking incentives and/or firms that are more volatile in
nature use stock options to attract less risk-averse employees. The correlation between
LEVERAGE and OPTIONS is negative and Significant, which suggests that firms with
lower financial leverage tend to use more ESOS to compensate their employees.
Interestingly, OPTIONS is positively correlated with CASHFLOWS, which is inconsistent
with the conventional wisdom that cash-constrained firms are more likely tO use ESOs to
substitute for cash compensation.

The correlation matrix for the yield spread sample presented in Panel B Of Table 3
is generally consistent with the results from univariate tests in Panel B of Table 2.

Specifically, the Pearson and Spearman correlation coefficients between OPTIONS and

34

SPREAD are negative and significant at the 1% level, which is inconsistent with the
finding from the credit rating sample. Consistent with prior literature, SPREAD is
positively correlated with LE VERAGE, BETA, 32M, and all Operating risk proxies and
negatively correlated with ROA, CASHFLOWS, SIZE and INT_COV. All of these
correlations are significant at the conventional levels in terms of the Pearson and
Spearrnan correlation coefficients. Better accrual quality is associated with a lower yield
spread although this relationship is significant only in terms of the Spearrnan correlation

coefficient.

35

CHAPTER 6: EMPIRICAL RESUTLS

6.1 Credit Rating Analyses

In this section, I examine whether the fair value of outstanding ESOS is associated
with a higher or lower perceived credit risk after controlling for information risk,
Operating risk and firm characteristics that have been shown to affect rating agencies’
credit rating decisions. Before proceeding to the empirical analyses, I first test for
endogeneity of OPTIONS in the credit rating model because of the concern over the
direction of causality between RATING and OPTIONS. This study hypothesizes that a
higher level Of Outstanding ESOS leads to a poor credit rating. It is likely, however, that
the causality also runs in the Opposite direction. Because granting ESOs requires no cash
outlay, firms facing liquidity problems are likely to use Options in lieu of cash
compensation to conserve cash. Although this prediction has not been documented
consistently in the literature” and researchers have questioned the economic justification
and efficiency Of such a practice (Hall and Murphy 2003; Oyer and Schaefer 2005) 26,

one cannot rule out the possibility that firms with poor credit ratings use employees to

 

25 Yermack (1995) and Core and Guay (2001) document that firms award more ESOS when facing liquidity
constraints. However, Matsunaga (1995), Ittner et al. (2003) and Kroumova and Sesil (2003) do not find
such a relationship. In addition, Rebeiz’s (2003) survey evidence shows that a majority of the compensation
committee members he surveyed do not agree that ESOS are necessarily a cheap source of capital.
Moreover, contrary to the prediction that firms grant ESOS to conserve cash, Ittner et al. (2003)-find that
new economy firms with greater cash flows use employee stock Options more extensively.

2‘ While some people argue that compensating employees with ESOs helps firms conserve cash to finance
future growth, researchers note that this argument, while intuitively appealing, lacks a clear economic
justification (Hall and Murphy 2003; Oyer and Schaefer 2005). Hall and Murphy (2003) point out that
substituting options for cash compensation is essentially borrowing from employees the amount that
otherwise would have been paid in cash. The existing literature has demonstrated that options are worth
less to employees than their economic cost to the issuing company because employees are risk averse and
poorly diversified (Lambert et al. 1991; Meulbroek 2001; Hall and Murphy 2002). This begs the question
of whether risk-averse and undiversified employees are an efficient source of financing, compared to banks
and other private capital providers such as venture capitalists who are better at managing risk (Hall and
Murphy 2003; Oyer and Schaefer 2005).

36

help finance the firms because these firms face a higher cost of accessing the public debt
market. As discussed in greater detail in Appendix D, results from the Rivers-Vuong
(1988) test procedure provide little evidence that OPTIONS is endogenous in the credit
rating model. As a result, the empirical analyses in this subsection are based on regular
ordered probit models.27 I provide further evidence on the direction of causality in
Chapter 6.3.1.

Panel A of Table 4 presents the test results Of the first hypothesis — whether credit
market participants perceive ESOS as a liability-like or equity-like Obligations in their
credit risk assessments — for the credit rating sample. The third column in Panel A Of
Table 4 reports the analysis using all available firm-year observations. Since each of the
sample firms can appear in the regression model up to three times as long as it has no
missing data for all Of the sample years, the serial correlation in residual terms could
potentially Obscure statistical inferences. However, given that the state Of the art
econometric techniques are unable to estimate a fixed effect ordered probit model
(Wooldridge 2002, P. 484), the pooled cross-sectional results are supplemented with a
year-by-year analysis.

Overall, the credit rating model correctly classifies over 60% of observations into
the seven levels of credit ratings, suggesting that the credit rating model is reasonably

specified. COnsistent with the “options-as-liability” hypothesis, the coefficient on

 

27 Another reason this dissertation chooses to use the regular ordered probit (OLS) to estimate the credit
rating (yield spread) model is the difficulty in selecting appropriate instruments that are correlated with
OPTIONS but are exogenous in the credit rating (yield spread) model. There are reasons to believe that the
determinants of option granting behaviors proposed by prior research such as monitoring difficulty, growth
opportunities, cash flow availability and firm performance also affect credit risk assessments. Larcker and
Rusticus (2005) demonstrate that instrumental variable estimation with weak and/or endogenous
instruments invites more statistical inference problems than it solves. After evaluating the quality of
potential instruments for OPTIONS, I choose to retain the regular ordered probit and 01.8 regressions for
the credit rating and yield spread models, respectively. See Appendix D for more discussion.

37

OPTIONS is negative and significant at the 5% level or better in the pooled cross-
sectional as well as year-by-year models. The information risk proxy, ACCQUALITY,
does not appear to affect credit ratings in the multivariate framework despite its
significant pairwise correlation with RATING in Panel A Of Table 3. Two Of the three
operating risk proxies, STDROA and STDRET, are negative and significant, indicating
that Operating uncertainties have an unfavorable impact on the firm’s credit profile. The
coefficients on most of the control variables are significant and in the predicted direction
except for capital intensity (CAPJNT EN). Ashbaugh et al. (2004) find a positive
association between CAP JNT EN and credit ratings (RA TING). While'I also find a
positive and significant pairwise correlation between these two variables in Table 3, the
positive correlation turns negative after controlling for other variables in the credit rating
model.

Panel B Of Table 4 displays the results from the test of H2. The coefficient on
REPUR *OPTIONS is negative and significant as predicted across all regressions. This
finding is consistent with H2 that outstanding ESOs have a more negative effect on credit
ratings when firms have a proclivity toward buying back shares surrounding the exercises
Of ESOs. Since OPTIONS is interacted with REP UR in Panel B Of Table 4, a negative
coefficient on OPTIONS indicates that ESOs have an adverse impact on credit ratings
even for firms that do not exhibit a historical tendency to engage in ESQ-related
repurchases.28 All other variables behave similarly across Panel A and Panel B of Table 4.

Given that the marginal effects of independent variables in an ordered probit

model are not easy to interpret (Long 1997), I follow Ashbaugh et al. (2004) and partition

 

7’8 This finding is consistent with the notion that creditors believe that past repurchase behaviors do not
guarantee future ESO settlement methods and there is always the possibility that companies will incur real
cash outflows to redeem outstanding ESOs regardless of historical policies.

38

the seven levels of credit ratings into two classes - investment grade and speculative
grade - and assess the economic significance of the impact of OPTIONS on credit ratings
through an examination of the marginal effect from the probit model. The marginal effect
is measured as the change in the probability of receiving an investment-grade rating for a
change in the independent variable from the 25"I percentile to the 75th percentile of the
variable in the sample, holding all other independent variables at their mean values. Panel a
C of Table 4 details the results from the economic significance analysis. The dependent
variable in Panel C of Table 4 is INVGRADE, which takes on the value of one if credit
rating is BBB- or above, and zero otherwise. To facilitate the interpretation and
comparison Of the economic impact across independent variables, the marginal effect
analyses are estimated based on the baseline credit rating model without the REPUR
interaction term. The probit model results are Similar to those of the ordered probit model
in Panel A of Table 4. The marginal effect of OPTIONS is -0.036, indicating that when
the value of outstanding ESOs increases from the 25th percentile to the 75‘h percentile in
the sample, the probability of receiving an investment-grade rating decreases only by
3.6%. This result suggests that outstanding ESOs do not have an economically significant
influence on the probability that the firm receives an investment-grade rating, eSpecially
when compared to other firm characteristics such as return Volatility and firm size. A
change in return volatility (firm size) from the 25‘h percentile to the 75th percentile
decreases (increases) the probability of receiving an investment-grade rating by 21.8%
(20.7%). Interestingly, the marginal effect of the firm’s on-balance-sheet debt,
LEVERAGE, is -0.030, a magnitude similar to that of OPTIONS.

6.2 Yield Spread Analyses

39

Table 5 presents the results for the yield spread sample. Model 1 of Table 5
presents the test results of the first set of hypotheses. Consistent with the “Options-as-
liability” hypothesis, the coefficient on OPTIONS is significantly positive, indicating that
credit market investors charge a higher risk premium for firms with a higher level Of
ESOS. The coefficients on STDCFO and STDRET are significantly positive, which again
demonstrates that operating volatilities have unfavorable implications for the perceived
credit risk. Inconsistent with my predictions, the coefficients on maturity and
subordination Status of the bond are not significantly different zero.29 Note, however, that
only about 3% of the sample bonds are subordinated debt. Thus, the lack of variation in
the S UBORD variable might explain its insignificant coefficient. To determine whether
outstanding ESOS and balance sheet debt have comparable impacts on yield spreads, I
multiply the coefficients from Model (1) of Table 5 by the interquartile range of
OPTIONS (0.0360) and LEVERAGE (0.1620). The results Show that an increase from the
25m percentile to the 75‘h percentile in OPTIONS and LEVERAGE increase the yield
spread by about 12.8 and 14.1 basis points, respectively. This finding, in conjunction with
the results in the previous section, suggest that outstanding ESOs have a negative impact
on the firm’s perceived credit risk Similar to that of the recorded balance sheet debt. To
provide perspective on the economic significance of OPTIONS, I multiply the 12.8 basis
point difference by the median bond issue size, which is 300 million in my yield spread
sample. The yield spread difference translates into a higher financing cost of $384,000

($300 M*0.128%) for firms in the 75th percentile of OPTIONS compared to firms in the

 

29 Following prior literature, I include only the level of the bond’s maturity in the yield spread model.
However, because the shape of the yield curve is generally nonlinear, I also include the squared term of
maturity in the yield spread model as a robustness Check. Neither the level nor the squared term of maturity
is significant when I include both terms in the model and the coefficients on other variables are virtually the
same as those reported in Table 5.

4o

25th percentile of OPITONS. Given that the median total interest expense (interest
expense on newly issued bonds)30 is $172 ($18) million for my sample firm-years, the
economic magnitude of this financing cost difference is arguably modest.

Model (2) of Table 5 displays the test results of H2. Inconsistent with H2 that
OPTIONS has a more negative impact on the perceived credit risk for firms that
repurchase shares in response to the settlement of the ESO Obligations, the coefficient on
REPUR *OPTIONS is not significantly different from zero. Recall that while RATING is
credit rating agencies’ opinion about the company’s overall debt serviceiilitv. the bond
yield spread is the credit market’s assessment of the firrn’s capacity to meet a particular
bonjd obligation. Therefore, the impact Of outstanding ESOs on yield spread should be a
function of not only the likelihood of ESO-related repurchases but also the expected life
of ESOS in relation to the maturity of a specific bond. The existence of ESO Obligations
should not have a negative impact on bondholders’ chance of receiving guaranteed
principal and interest payments if the company does not repurchase shares in response to
ESO exercise g if the bond has been repaid before ESO obligations become due. Put
differently, outstanding ESOS would adversely affect the yield spread of a particular bond
only when the company engages in ESO-related repurchases and when the maturity of
the bond is longer than the expected life of outstanding ESOS. To test this conjecture, I
construct an indicator variable, LONG, which takes on the value of one if the maturity of
the bond is longer than the expected life of outstanding ESOs at issuance, and zero

Otherwise.“ Following Li (2002), I measure the expected life of outstanding ESOs as the

 

3° Interest expense on newly issued bonds is estimated as the amount of proceeds multiplied by the offer
yield to maturity.

' Since the weighted average maturity of a company’s total outstanding debt is not publicly available, I am
unable to introduce the LONG variable to my credit rating model.

41

expected life of the new ESO grants minus the age of outstanding ESOs, which is
calculated as the difference between the maximum contractual life and the remaining
contractual life of outstanding ESOs. Model (3) presents the results when the impact of
OPTIONS on the credit risk is conditioned on the bond’s maturity relative to the expected
life of ESOs. The coefficient on LONG*OPT IONS is not significant, indicating that
OPTIONS does not have an increased adverse impact on the cost of debt for bonds with a
maturity longer than ESOS’ expected life. However, as reported in Model (4) of Table 5,
when the impact of OPTIONS is conditioned on both the issuers’ ESQ-related repurchase
propensity and the bond’s maturity in relation to the expected life of outstanding ESOs,
the coefficient on REPUR*LONG*OPTIONS is significant and positive at the 10% level
while other interaction terms remain insignificant. This finding suggests that outstanding
ESOS impair the company’s debt service ability only when the issuing company is prone
to repurchasing shares around ESO exercise _a_nd_ when the cash outflows associated with
ESO exercise take place before bondholders received their guaranteed payments under
the bond contract.
6.3 Additional Analyses and Robustness Checks
6.3.] Change Specification

To further enhance the robustness of the credit rating results, I also estimate the
credit rating model using a “change” specification to examine whether a change in the
firm’s outstanding ESOs is positively associated with a change in the firm’s credit
ratings. This test controls for firm-specific factors that are constant over time and thus
mitigates the concern over the autocorrelation in residual terms. Since the “change”

specification requires all variables in the credit rating model to be first differenced, I lose

42

one year of observations for year 2001. The results of the “Change” analyses are
presented in Model (1) in Panel A of Table 6. The dependent variable, DOWNGRADE, is
an ordinal variable that measures the change in credit ratings from year H to year t.
DOWNGRADE is coded 1 (-1) when the firm’s credit rating is downgraded (upgraded)
and coded zero when the firm’s credit rating remains the same. Of the 566 firm-year
observations that have sufficient data for the “Change” model, 82 (45) firm-years receive
a downgraded (upgraded) rating and 439 firm-years experience no change in their credit
ratings. Consistent with the results in Table 4 and the “Options—as—liability” hypothesis,
the coefficient on the first difference of OPTIONS (DIFF_OPTIONS) is positive and
significant at the 1% level, suggesting that the tenor of my main results in Table 4 is
unaffected by the “change.” specification.32

To provide further evidence on the direction of causality, I also examine whether
the change in outstanding ESOS explains the one-year-ahead change in credit ratings. The
one-year-ahead “Change” model further limits my analysis to observations with necessary
differenced variables in year 2002. This is because I lose year 2001 data in the process Of
first differencing and the one-year—ahead change in credit ratings for year 2003 was not
available when empirical analyses were conducted. The results of the one-year—ahead-
change analysis are displayed in Model (2) in Panel A of Table 6. The coefficient on

DIFF_OPTIONS remains positive and significant at the 1% level.

 

32 In addition, I also perform several other sensitivity checks to ensure the robustness of my results. In my
primary analyses, I convert Standard & Poor’s credit ratings using Ashbaugh et al.’s (2004) coding scheme,
which collapses all letter grade ratings into seven categories. To examine whether my results are sensitive
to the rating conversion process, I recode the credit ratings following Klock et al. (2004) to allow credit
ratings to range from 1 (D ratings) to 22 (AAA ratings). See Appendix A for the conversion schedule. The
negative relationship between RATING and OPTIONS is robust to this alternative rating conversion. In
addition, to mitigate the concern that the results in this dissertation are driven by extreme outliers or
industry-specific risk factors, I winsorize all independent variables at the 1% and 99% levels and include
industry controls in the credit rating and yield spread models. The results are qualitatively similar to those
reported in Table 4 and Table 5.

43

In addition, I also examine whether the change in RATINGS predicts concurrent
and one-year ahead change in OPTIONS. For the purpose of this analysis, the level of
OPTIONS is modeled as a function of monitoring cost, cash flow availability and fum
performance as described in Appendix D. I take the fnst difference of all variables in the
OPTIONS model and include the change in credit ratings (DIFF_RA TING) as an
additional predictor in the “change” specification. Results from Panel B of Table 6 show
that the change in credit ratings does not predict either contemporaneous or one-year-
ahead change in OPTIONS. Taken together, findings from this subsection suggest that the
direction of causality appears to run from OPTIONS to RATINGS as hypothesized but not
the other way around.

6. 3.2 Executive Versus Non-Executive Stock Options

By tying executive remuneration to firm performance, ESO programs are
designed to align management’s interests with those of shareholders. However, if
executives’ personal wealth ties too closely to the firm’s stock price, managers might
have Opportunistic incentives to maintain or inflate the short-term stock price at the
expense of the company’s long-term economic success. If creditors, like many critics,
believe that the cost of the perverse incentive from executive ESO compensation
outweighs the incentive alignment benefit, they may be more wary of firms with large
executive ESO programs and price protect themselves against potential value-destroying
behaviors in these firms. To the extent that the value Of executive Option holdings is
positively correlated with the value of total outstanding ESOs, OPTIONS could simply be

a proxy for the size of the executive ESO program.

To investigate this possibility, I partition the value of outstanding ESOs into the
part that accrues to top-five executives, MGTESO, and the part that belongs to non-
executive employees, NONMGTESO. For the median firm-year in the credit rating (yield
spread) sample, NONMGTESO is about 4.7 (5.2) times larger than MGTESO, suggesting
that top-five executives hold about 17.5% (16.1%) of the value of outstanding ESOs. If
creditors are Opposed to the adverse incentive effect of executive stock option programs
rather than the liability-like impactof total outstanding ESOs, we should expect a
negative (positive) coefficient on MGTESO but not a significant negative (positive)
coefficient on NONMGTESO in the credit ratings (yield spread) model. Panel C of Table
6 reports the results from this partition. Interestingly, the coefficient on MGTESO is not
statistically different from zero in both the credit rating and yield spread regressions
while the coefficient on NONMGTESO continues to be significant in the predicted
direction. These results indicate that the negative relationship between OPTIONS and the
cost of debt is not driven by credit market participants’ potential aversion to large
executive ESO programs.

6.3.3 The Cost of Debt and Corporate Governance

Although the existing empirical research has not established the link between
governance structure and the Size of the firm-wide ESO program (Oyer and Schaefer
2005), the popular press and researchers have generally contended that the excessive use
of ESOs is a symptom of poor corporate governance. Jensen and Murphy (2004) argue
that corporate governance and compensation policies are highly inter-related. Inefficient
pay practices can generally be traced to weak corporate governance. Hall and Murphy

(2003) posit that firms face growing pressure to push stock options down throughout the

45

organization as grants for top executives increase. As a result, the size of the ESO
program could also be proxying for the strength of the firm’s governance structure. Since
prior studies have documented that the firm’s governance strength is inversely related to
its cost of debt (Ashbaugh et al. 2004; Bhojraj and Sengupta 2003; Anderson et al. 2004),
the positive relationship between OPTIONS and firms’ credit risk might be driven by the
link between governance strength and the cost of debt. To rule out this possibility, I re-
estimate both my credit rating and yield spread models including proxies for corporate
governance structure. Because a firm’s corporate governance process tends to be sticky
over adjacent years, a pooled cross-sectional analysis could inflate the significance of
governance variables. As a result, the credit rating'analyses in this subsection are based
on year-by-year regressions.33
My first set of governance proxies are the corporate governance mechanisms
identified by Ashbaugh et al. (2004) as being associated with the cost of debt. As with
Ashbaugh et al. (2004), I use the Board Analysts database as my primary source of
corporate governance variables. I use Board Analysts data from the 2003 proxy season,
which represents the governance Structure for fiscal 2002, and assume that the
governance variables for 2002 serve as reasonable proxies for the governance structure
for 2001 and 2003. Therefore, I use the same set of governance variables in all three year-

by-year regressions. As reported in Panel D of Table 6, the coefficient on OPTIONS

continues to be significant at the 1% level and none of the governance mechanisms

 

33 If one believes that governance structure is sticky over a short period of time, the “change” model
specification also helps to alleviate the concern that the adverse impact of OPTIONS on credit ratings is
driven solely by the omission of governance proxies in the model.

46

appears to be significant.34 Moreover, the p-values of the joint significance tests of
governance variables suggest that the potential multicollinearity does not explain the
insignificant coefficients on governance variables.

It is possible, however, that outstanding ESOS are proxying for other dimensions
of corporate governance structures that are priced by the credit market but not captured
by the governance mechanisms identified by Ashbaugh et al. (2004). Therefore, my
second proxy for governance structure is the Gov-Score from Brown and Caylor (2004).”
Brown and Caylor (2004) created Gov—Score based on 51 factors provided by
Institutional Shareholder Services as of February 1, 2003. These 51 factors can be
divided into eight governance categories including: audit, board of directors,
charter/bylaws, director education, executive and director compensation, ownership,
progressive practices, and state of incorporation. Brown and Caylor (2004) show that
firms with higher Gov-Score are more profitable, have higher market valuation, and pay
out more cash to their shareholders. In addition, Gov-Score is better linked to firm
performance than is Gompers et al.’s (2003) G-Index. Therefore, Gov-Score is potentially
a more comprehensive summary measure of a firm’s governance strength. The results of
this analysis are reported in Panel B of Table 6. The coefficient on Gov-Score is
insignificant in the credit rating model and significant in the opposite direction in the
yield spread model. Importantly, the inclUsion of this governance proxy does not affect

the direction and significance of OPTIONS.

 

3" Unlike Ashbaugh et al. (2004), I do not include in this analysis the percentage of independent directors
that hold seats on other boards and the percentage Of the directors that own stock in the company because
almost all of the directors in my sample firms hold some stock of the company and sit on at least one other
board.

3’ Brown and Caylor make GOV_SCORE publicly available on the following website:
http://robinsongsu.edu/accountancy/gov scorehtml.

47

My third proxy for corporate governance structure is motivated by Core et al.’s
(1999) finding that CEOs earn greater compensation when governance structure is
weaker. This finding suggests that the excessive ESO compensation of which the
cumulative effect is reflected in large outstanding ESOs could be a potential indication of
ineffective governance mechanisms. If the negative impact of OPTIONS on the perceived
credit risk simply communicates creditors’ concern over the lavish ESO payments
stemming from ineffective governance structure, OPTIONS should have a negative
impact on the cost of debt only when the company consistently compensates employees
with excessive ESO grants. To investigate this possibility, I first create a proxy for
excessive ESO grants. As described in Appendix E, I model the benchmark level of ESO
grants as a function of idiosyncratic firm risk, growth opportunities, the strength of
external monitoring, firm size, cash flow availability, and firm performance. An indicator
variable, EXCESS_ESO, is set equal to one when the five-year average residual from the
benchmark option grant model is positive, suggesting persistent ESO overpayrnents, and
zero otherwise. If the negative relationship between outstanding E803 and the perceived
credit quality is driven by excessive ESO grants, the coefficient on OPTIONS, which
represents ESOs’ impact on the credit risk for firms that are not classified as granting
excessive ESOS, should not be significantly different from zero while the coefficient on
the interaction term, OPTIONS *EXCESS_ESO, which reflects ESOs’ impact on credit
risk for companies that give away excess ESOS, should be significantly negative (positive)
in the credit rating (yield spread) model. Panel F of Table 6 reports this analysis.
OPTIONS remains significant in the expected direction while OPTIONS *EXCESS_ES O

is never significant across all regressions. Collectively, results from this subsection

48

suggest that the unfavorable impact of OPTIONS on the cost of debt is unlikely to be
driven by the association between the strength of governance structure and the cost of
debt.
6.3.4 Grant Date Fair Value and Post-Grant-Date Fair Value Change

The “options-as-liability” hypothesis argues that ESOs are liability-like
obligations whose fair value at any point in time provides useful information about the
present value of the amount of net cash outflOws likely to occur upon the settlement of
ESO obligations. To further support this argument, it is important to also demonstrate
that creditors find the updated information about the ongoing ESO'Obligations beyond the
grant date helpful in assessing credit risk. In the next analysis, I divide the fair value of
outstanding ESOs into the grant date fair value (GRANT _VALUE) and the post-grant-date
fair value change (VALUE_CHG). Outstanding ESOS at a specific point in time comprise
a portfolio of options granted in different periods with distinct exercise prices and
. differing terms and conditions. Since the detailed information about each individual
Option’s grant date, its exercise price and the grant date market value of the underlying
Stock is not available in the SFAS 123 footnote, I estimate the grant date fair value as if
outstanding ESOS were a single option grant with an exercise price equal to their
weighted average exercise price. Since ESOs were typically granted at the money to
avoid any accounting charges, the weighted average exercise price for outstanding ESOS
also serves as a good estimate Of the weighted average market price of the underlying
stock at the grant date. In addition, I assume that the Black-Scholes assumptions
disclosed in the SFAS 123 footnote for new ESO grants apply to all outstanding ESO

grants. As a result, the grant date fair value per option is estimated using the Black-

49

Scholes Option pricing model with the footnote disclosed Black-Scholes assumptions as
model inputs and assuming that outstanding ESOS were all granted at the money. The fair
value per option is then multiplied by the number of outstanding E803 and scaled by
total assets to obtain GRANT_VALUE. The value change subsequent to the grant date
(VALUE_CHG) is simply the difference between the estimated current fair value per
Option and grant date fair value per option, multipled by the number of outstanding E803
and scaled by total assets. Results from Panel G of Table 6 Show that the coefficients on
GRANT _VALUE and VALUE_CHG are both significant in expected directions, indicating
that the value change between the grant date and valuation date provides useful
information incremental to that contained in the grant date fair value in explaining credit
market participants’ credit risk decisions. In other words, this last finding implies that
creditors consider mark-tO-market information of outstanding ESOS decisionerelevant and
that they do not regard the value change’in ESOs as a simple wealth transfer among

shareholders.

50

CHAPTER 7: CONCLUDING REMARKS

This dissertation empirically examines whether credit market participants view
outstanding ESOs as a liability-like or equity-like obligation when assessing the issuing
company’s credit risk. I hypothesize that, from a creditors’ perspective, outstanding ESOS
could have both equity-like and liability—like impacts on the company’s credit quality. On
the one hand, ESOs are financial commitments to create equity infusion in the future and
the fair value of ESOs can be viewed as the first installment of the capital contributed by
employees with service. Companies retain the benefits of employee services irrespective
of whether the second installment of the capital contribution, namely the option exercise
price, is eventually collected. This aspect of ESOS suggests that they are equity-like
contracts that increase the equity base and enhance the issuer’s debt service ability. On
the other hand, the settlement of the ESO contracts could involve potential cash outflows
because companies typically launch Share repurchase programs in response to ESO
exercise. The need to fund the ESO “maturity” with cash flows or through refinancing
imposes a liquidity risk and makes ESOS an economic liability of the issuing company.

For a sample of 913 firm-year Observations from 338 firms over the years 2001-
2003, my investigation indicates that firms with higher value of outstanding ESOS
receive lower credit ratings assigned by Standard and Poor’s after controlling for
information risk, operating risk, firm performance and other firm Characteristics that are
known to predict corporate credit ratings. This relationship holds true in both the “levels”
and “change” model specifications and is robust to controls of various corporate

governance mechanisms. In addition, the change in outstanding ESOS is associated with

51

the one-year-ahead change in the firm’s credit ratings, which provides further evidence
that outstanding ESOs do affect credit rating agencies’ evaluation of the firm’s debt
service ability. Moreover, the unfavorable impact of outstanding ESOS on the firm’s
credit rating appears to be larger for firms that are more likely to engage in share
repurchases surrounding the exercise of ESOs.

To investigate whether credit market investors hold the same belief toward the
nature of ESO obligations as credit rating agencies, I perform yield spread tests on a
sample of 309 new bonds issued by 195 firms over the period January 2001 to October
2004. The results from yield spread tests are generally consistent with those from credit
rating tests. Specifically, firms with higher levels of outstanding ESOs incur higher costs
of financing in the form of wider yield spreads. Moreover, I find that OPTIONS has a
more adverse impact on yield spreads for bonds that have a maturity longer than the
expected life of outstanding ESOS gn_d are issued by firms with a propensity to repurchase
shares in response to ESO exercise. Taken together, the results from credit rating and
yield Spread tests are consistent with the notion that credit market participants perceive
outstanding ESOS as a liability-like claim, although the economic magnitude of ESOS’
impact on the cost of debt is arguably modest. Finally, I demonstrate that the change in
the fair value of outstanding ESOS subsequent to the date of grant provides useful
information beyond the grant date fair value in explaining the company’s perceived credit
risk. This latter finding identifies information useful for creditors to make credit risk
assessments and therefore provides insights into the kind of disclosure creditors may find

useful if the grant-date measurement and equity classification of ESOs are mandated.

52

This study contributes to the ESO literature by demonstrating that outstanding
ESOs provide value relevant information that affects the credit risk premium charged by
the credit market for the added default risk associated with ESOS. Given that my
empirical analyses focus on the perception of only one group of information users, two
caveats are in order. Examining creditors’ perception is not the optimal means for
assessing the appropriate balance sheet classification of outstanding ESOs. Nor does it
speak to whether standard setters should mandate the recognition of the fair value of
ESOS on the balance sheet. Nevertheless, this study sheds light on the liability versus
equity classification debate. The results from this study indicate that in assessing ESOS’
impact on the credit risk, information users appear to respond to the fair value
information that is typically provided by accountants for some liabilities but not equity.
However, under SFAS 123R, the fair value of ESO grants is computed just once at the
grant date. Firms are not required to disclose the fair value of outstanding ESOS, nor the
fair value changes in these stock options. In order to keep track of the ESO obligations,
credit market participants have to come up with their own estimates every period. As a
result, findings in this study, in conjunction with equity market evidence on the value
relevance of outstanding ESOs, suggest that SFAS 123R might not mandate sufficient
disclosure that provides the most decision-useful information to accounting information
users. One could argue that whether or where the fair value and value changes of ESOS
are disclosed is irrelevant because the credit (equity) market already incorporates the fair
value of ESOS in their credit risk (equity valuation) assessments. This argument,
however, fails to consider whether it is informationally efficient to have numerous

investors produce their own estimates (Dhaliwal 1986; Lipe 2001), which may or may

53

not be accurate. After all, managers are in the best position to generate the fair value of
ESOs, given their private access to the information necessary to compute this value.
Finally, this dissertation suggests several other interesting issues for future
research. Of particular interest is whether the recognition of ESOS as equity on the
balance sheet changes the credit market’s belief about ESOs’ impact on the issuer’s
creditworthiness. In addition, since my sample is concentrated in years when the stock
market experienced a downturn and when ESOs received more negative publicity, the
results in this study might not be generalized to other periods. Therefore, future research
might want to examine whether the credit market’s reaction to ESOS changes over time

as more firms start to adopt or discontinue ESO programs.

54

Appendix A: Intuition behind the Fair Value of a Call Option

The fair value of a call option can be thought of as the present value of the expected
payoff at the expiration date. At expiration, the payoff of a call option holder is the
difference between share market value and the option exercise price (i.e., the intrinsic
value) if the Option is in or at the money and zero if the call option is out of the money.
As a result, at expiration (time T), the value of a call Option is:

CT: max [0, ST - X] (Al)

Where C is the fair value of the call Option, S is the market value of the underlying stock
and X is the exercise price of the option.

Since the stock price at expiration date is unknown, the expected value of the option at
expiration can be written as follows:

EICTl = E{max [0. ST - X] l , (A2)

L] is the expectation operator.
At any time before expiration, the fair value of the call option is simply the present value .
of the expected intrinsic at expiration as follows:
C1: PV { E{max [0, ST - X]} /(1+y)t} (A3)

Where 7 is the rate at which the expected value is discounted to calculate a present value
and t is the difference between T and

To satisfy ESO exercise, the company typically engages in open market
repurchases to make shares available for E80 exercise. Although the actual repurchase
does not necessarily coincide with option exercise, the Stock price at exercise date serves
as a reasonable proxy for the cash that the company has to pay out to acquire shares.
Since an ESO will only be exercised when it is in the money, the settlement of the ESO
Obligation always results in a net cash outflow equal to the excess of the underlying stock
price over the option exercise price at the settlement date, which is what equation (A1)
measures. Hence, the fair value Of outstanding ESOs as calculated in equation (A3)
provides an estimate of the present value of the expected net cash outflows that will incur
upon the settlement of outstanding ESOS.

55

Appendix 8: Credit Ratings’ Numerical Conversions

 

 

S &P Compus tat Conversion Conversion
Letter Quarterly Data Number Number Grade
Ratings S P D R C (Ashbaugh et al. (Klock et al.
2004) 2004)

AAA 2 7 22 Investment
AA+ 4 6 21 Investment
AA 5 6 20 Investment
AA- 6 6 19 Investment
A+ 7 5 18 Investment
A 8 5 17 Investment
A- 9 5 16 Investment
BBB+ 10 4 15 Investment
BBB l 1 4 14 Investment
BBB- 12 4 13 Investment
33+ 13 3 12 Speculative
BB 14 3 1 1 Speculative
BB- 15 3 10 Speculative
8+ 16 2 9 Speculative
B 17 2 8 Speculative
B- 18 2 7 Speculative
CCC+ 19 1 6 Speculative
CCC 20 l 5 Speculative
CCC- 21 l 4 Speculative
CC 23 1 3 Speculative
C 24 1 2 Speculative
D 27 1 1 Speculative
Total

 

56

 

as» news? 05 E Be 20 a as: 33 3 38... was 05305 a Samaensaae E as... as 2:.

Mba<>lh2<-0

 

.0th0: 0:58 m20fim0 9 20st

0302202

 

88.0.3258 Le Ease 2: E 82% 8.. 8%805 9.0.03 aesdEEEa 232:5: +63%me Ea_

5.6 300500me .30» 05 00 0:0 05 La 20: 038090 05 L05 gouge 003.025 van 0030., .«o 0035:: 05 mo :33 05 9 .

E300 £023 6028005 020.3“ 3 20.. 63 36:53.5 .6 .5309: 05 863 «.2230 8 008850 E 0035028 020
A§ B @5800 h20E0 mo .8qu 05. .388 .88 3 00—08 83305 032.3. 9 $8.58 0.29.50 we 5qu 0E.

0&502

 

demos..— eogabEe .550: 03:2:an
05 E CZ .050» 5 mega 05 we 0.:— 008090 05 E L 55.0.9, 00qu 0—003 .355 0200qu 05 E e NEE» 0:02.50
3:58 690098 05 E u ”88 $0085 coho—EC 05 E .. ”gouge 05 we 00% 0E0._0x0 E N ”0:6 8:02? 8 00% 0.090. E m

Dvaoune ecu cvbtcinond+vucifimv==u_e 000:3»NEZ t-0 X - 032 m u 0

”30:8 0.0 E 2:508 cot02-00_o:0me_02m BREE—w .530 :03

Beam defiance?“ 3:800 mama: 08030—00 0029 05 55, 003—230 Emma 08 85 0020.; BE» 0050:. 30an 0350 05
36 8038820 380 3:0 0:0 050 5:9: 009.: mu 0:335:30 .002: .58 0.6% 05 can .003 0.00% 20.80» 38:
05 .3230 memeesaso 08 00% 0E000x0 093.6 002$?» “acme—0E0 05 mama: Efiw :ouao 0&5». a 0003 »05 n: an 28:90
052558 WEE e as? L .888 5 an 388 36 es. 28 9.36:8 Ramada see 9:880me seam 8:8
0.090. 0.65005 .«0 020.» E0.“ 05 89:50 2 9:00:me 3 e02. 085 08 8.0m» 3:00ng A3 .0250 A0 53833

5 080328 0.:— 080098 05 9 3:86 .3032: 8 08: wEEaEB a 55 0055000. D300; .w.D :0 20m» 05 3 3.5808
E 88 $0085 00¢-me 05 A0 .0858“ m2 m<mm 05 Set 005050 on =00 :02? mo 53 .00: 3300.280 9:52:00 05
.05 0.:— EBoabcoo 8:552: 05 50.53 005.02% 05 we 00550.8 E 00mm wfiucsszo we 0? 2E. .53 mum—0:50.80
.«o own 05 BEE 3:3» 0%— 30: we 03 00809.0 05 m0 A800 5 9.36:8 082560 E 0.0mm mam—0:838 we 0:— @800qu
A0 .0558 8 Lead 05 00.6 0050.008 E can N“ we .08 0.35.... 05 0.05: 3.2.5 0.0.6200 02a Amy—O 35:08 00 855mg
ESE 05 .«o 59500 3205.6 05 ma m2 mafia we 0052:» 05 9 330800 003850 E DEE—0Z0 deduce—.30
009.: 080530 9:30:00 05 £3 wee—a .5563.“ 0:022». .52 2580a 2 833 5:02 3 .0588 m0 .009: corms—9,
condo $3: 00.28-0605 05 :o 0093 022228 E 80» 30mm 05 we 0:0 08 .«o 8 cacao com 030.» Emu 0E. .30» 38c
05 00 0:0 :05 an wEeesano 8230 00 09:5: 05 3 002303 Sumo 00.— 02.3 :3 05 8 0059.80 E 88 mam—0:00.008
no 09:; En.— 05. .Gfi Lam—.9808 300.8 38 3 00—000. 000» .80.: 05 .«o 30 05 .8 ma 00mm wfieqsaso me 020., Emu 05.

@2050

 

35%

 

 

wade E020 NEE 05 E 03%: 5 E 205 .3 0:0 023mg 93 $58 5006 ME: 05
5 038.0 0: E 205 E 80s a 2 E .80» 89a 00.089520 E 958 £080 EEG 06 a 0:0 9 .260 00. 030%? .0520 .2

NQ§0>§0Q

 

.050 note 05 .0 3:32: Luz—Em .23 econ
3508p. 0 :0 Drama 8 2E» 05 05 econ 88088 003E »_30: 05 co brag: 9 2E» 05 50.503 00:05.06 05.

0322.3.

 

#88 .a Bananas. E 033.8 SEES 2: no Ban 025$ e. 8:028
E 020%. 6:0 30» 38¢ 05 00.00 .0585 Em 620.2,». 800 b.0230 “Susan—cow macs. £020 BEE 0300520 58.33

02:52

 

 

 

flat-33.3

 

 

82.2.8 222...; "0 5.2.8.2

57

 

.:0 0::: 30:00: 0:: 0: 00050: E 0:0 30:00:: 00m 0:: E8: 005050 E :00? 00:03:: :0 _::0E 0: 0:00 :0 :0:E:Z
00000000 02: 0: 08:00 EEEsm :00 £326: 0:: 0:0 :00 ::0:::0 0:: E «8:05:00 E00338 0006:000 0.me
:0 020> 0:: :0 E00 05"”:me .820 :0:E::E00 E 0000000 .2 2: 20:39:00: "mm—M 06:3 :0 +5550: +0: u Emmy:
.88 0: ~00: E8: 0:00 9:3 E:: :000 :0: 30:00:00 0808000 E 600:: E:.:. .0E3:0::0 0:00 0:0 :0>0: 00m 0:: :0

   

   

v.5: <2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0:: :0 0:0 0:: :0 $0000 0:0: »: 00:00: 0:00 ::0:» 0:0 0: E03033. 00mm 30:0:an :0 0::0> :0: 0:: E 0w:0:0 0::.

::00::_:w:m E 600E :38. mEBozo: 0:: E0:: 032:; ::00:0:00:_ 0:: :0 ::06:.::000 0:: :_ 0:0 00000 03000.» 5:00:05 :< 0:00:02
:833 ll
0:39:00: »:600 :0 020> :0::0E 0:: .3 00—00: 80:: :8000E00: »:600 :0 0:03 0:00: 0:: 00 005002: .000: :00—0:: 0: :_00m EN0
:2: 0::: :50: 8: :0 :23: n 80:30 :20 :52: sou: 80:8: 00 05:0: 088:: 002000 30:5:
.3: :50: E00: 0:800 00:0: : 00—00: 2.: :80: E00: ::0E 0: 0 0:0 ::0_ . 0: 08:0 2022Im<0
90% :0>0
000:: :0:_:0E 003903.03? 0:0 2:30: 060:: »=00 :0 0000: 600:: :0:_:0E 0:: 30:: 0006200 :2: »:_:_:0 00058:»: 0::. EM:
.820 :00: 080., 000 0::. :80: s :o 020.. 3:05 0:: :o 55:0 o: 2:. mm:
00:: : 0: :1: 0:: :0>0 00 0:0>0 0:0 80:00 000: n 0200: $090 :50: E00: 0006000 :00:0 0 E0:: 030.: :000 ”03:03.0
:6:
.30 0::. Eco: :38 :08: a 00020 000 :32: 050+ 00 :82. 500: 500 :08: 00555.:
.m::=0E 8 :00 0:: :0>0 00:00:20.: 8:30: _:::0E «5:0 :0 :000600 0:00:80 0::. $005.0
1 A .: g
1: :00» :0>0 00:00:28 0:800 00:0: 3 0200: 30m: :Sm0QE00: 0006000 m:00:0:0 :8: 0.50:: :000 :0 50250 0:00:00 0::. 0 :00 km.
.30 :53 Eco: :80 :08: a 00:30 0::: :38 :80: was:
»:0E0:00::x0 0:0:0: 0E00£ m0 005:00 E <0: 06:? .: :00» :w=0::: 0.: :00» :0>0 02030—00 <00 :0 :000300 0:00:80 0::. «0:95.
:0: .0580 :000:
00:86:: E00800 :20: < .: :00: .0025 I :00: :0>0 00:28:00 .680: 22.2202 3 8:09: 8 8:08:00:
@000 565-3280 0:: :0 0:000E000 650000-086 E0:: E00030: 0:0 :0 :00E>00 0:00:07. 0:: :0 0>00m0c 0::. 0:20:50000.
830:0.» .90....
030530.00 05 0.202% 083:0: 8:80:00 2:: :0 85:00:: E :: :00 80020.,
0:01:33:

 

 

.000<>I&2§0 E030 0: :00 :000:_0> 0:: :0 0:0 0:: :0 8030 _0:0: : 00—00:

0:0 00mm w::0:0:m:=0 :0 6:50: 05 »: 00:00.:E :05 E 00000 :0: 020.» :0: 0::. .»0:0E 0:: :0 02.5%
:0 0:03 00mm w::0:0:m::0 :0:: mEEamfi 0:0 33:: 600E m0 0:000anm0 00_o:0m-:_00_m 080.80 08500:
0:: 50: 600E E60: 50:0 00:0:0w0605 0:: 90:: 00:05:00 E 50:0 :0: 0:03 :0: 0:00 ::0:m 0::.

 

 

 

 

58

.:0..0>.: 0.05.. 0.0 0:00 05:39.0... 4000......»5 0: 08.0 000 .000 0:. 000000... 60:00 .00 0.03 :5: 0:. ..
20.. 000: 02... 2003 .0... 0.00.5.0: 05000 05.0. .0 00.0> .008... .00 0.: 5 :80: 8.0.00.5 0: ...3 5.50.5550 0..... ..0.0:3 0:00... 5 00.00.5500 0: :.3

05.83: 600.0500 m. 5.50 :00 0.5 .. $.50 50: 5.0.00.0. ..0 00.. 0. 530.5: 0:. A.30.... .0:. :5: 0 5 5.050.... :00 .0 0..... 0 u.

.8 0.23.900... < a

 

 

 

8.3550
0.00 :00 .08.. .05 0.00.0.9... me 03080.5 05.0.05: 5.52:... .0>...5.. m. m. 5:00:90. 0. :0:..08: .05... .00.» 5:00
...0...:0.5: 0... .00.. .0028. 09.53. 500.03. 0:. .. 0.5 .0 00.0> 0:. 5 8:0. :0.:3 0.:0...0> 5.00:5. 00 m. Owniwmmbxm

 

......:.0500I>0 \ 0080000003500 08.....0...” :
.0..0:03 5.30:0. 0:. 5 032.03. 20.30.. MKOUMUKVO 00.0... .0300 :00 039m .88 .. 0.03:0... .0
m0 80.50m .0:.0:0.0:m .05..0...m... .3 :0:.>0... 05.00. .m 5 :80: .33. 5.00... :00 c305 0: 5.00.0 m. mxobmfi >09

 

OwMIMMMUNM

szUWI>QU

 

 

.803 w GE 0.— EmOGdfla m 5 0

a

.0 00.0> 50.0.... 5.5.. .0 9:550: :: 5.08 00.0.. D0055 05.0 05.0: 0.50... .00 0. 050:0 0:. m. maz< 9.3550
0.00 :00 ovmmwz .. 0.5 0.00.00 mmoq .Amm.¢.0.00..E00U*§ 5.02.800. 3.000 .0 00:3 5.5:. 5.5.. .0 w........w0:

a: 5.000 8.... 5.000.000. 0.00.. b0...:500..0 05.0: 0.50... .00umaz 0.00% 0.0.0:... :000 0.5.3 0...... m. .000.

.0 w...>0.. 5 .05...:..00 00005 900:... m: :000....0E0n. :000 0.5.3 2.000.500-0005 :0.0E..8 0. 5.0850. 0:. 05:3
$050+0000582£+50£+maz.menus...

”00.0850. 5.30:0. 0:. .00.. .0028.

5.00... 0:. .0 0>..0w0.. 0:. m0 :0.....0: m. .meg 0.0.5.5000: .0505... .0 0.0000... 0:. 5.88 .0 .0 :w00::m< 5.30:0”.

 

.5 .00 E00 .0 5.0.0> mooN .... .4. ”00.009 0.00.. 0.0.... 0... .0 05... 5 00m. 030 .0... 05:55.02: .0 5:50.. 0.....

 

 

 

 

0.0.83: .05..0...m.: : :_0: 8.0:. .0 0 0.000. 0:. 0. 30.0030 .05..0...m...

OtkkaNWZN
Deaths:

 

 

00.35:.0 0.00
:00 00:0: 00000.05» .085. 0 m0: 0..... 0 .. 05 .0 00.0> 0:. 5 5.0.0. 0.:0..0> .050: 0 m. .0.00.:... 00...... 00000550

>030...

 

 

.5 w... 08 0:. .0... 000.0550 :000
.0. 5.0.... 0.5 :00 :.00: 0:. .0 ...0:0 0:. 8.0 m. own. 0:. .. .05.. 05 83000. E... 0 0.0:3 050m 0:50.50 0 m. 530.. OmU

 

 

.:.00: 0:. 5 05.00:: .00.... 0:... .0 0 0.50.0 0:. w. 0000:: 0:... :.00m

3:30.:
QZNIQQ

 

 

.mOmm. 0.9.0.800 .0 0... 50.00050 0.0.00.0. 0:. :00 0... 50.00.58 80.0.0.0...

0:. 503.0: 0055...: 0:. 00 5.0.5.00 w. :0.:3 £me w...:..0.m.00 .0 0?. 0:. 00...... 0.005 me— 30.. 0:. .0 0... 5.0095
0:. 00 00mm. 9:50.800 .0 0... 5.0095 0:. 0.0000... . doom. .: 5.30:0”. 00.3550 050 :00 600008. :5: 30.. .0 0.0:
0:. .0 00mm w...:..0.m.00 .0 0... 5.0095 0:. 00:. 550. m. :5: 0:. .0 3.5.0... 0:. .. 05 0. .0000 .00 0.:0..0> 5.00.:... ..<

 

 

00.3550 050 :00 .:0.0.::5:0m m. :5: 0:. .. 05 .0 00.0> 0:. 5 0.0.0. 0.:0..0> .0.00.:... ..<

 

 
 

00.0.5. .:00 20:3. 00.00.: 0 m0: .. 8.. ...0.0E .....0 :0:00
0: 5.0.00 .0: 5.0.5... :00 0 m0: .. Am. .5530... :00 0.. m0: .. 2. 5...:50 5.30:0. 0:. .0 b... 0.00... .. .. 0.:0:00-5..
00 :0.....0: m. :008. :5: < 3.3550 050 :00 .0.:0=00 m. 008. :5: 0:. .. . .0 00.0> 0:. 5 5.0.0. 0.:0...0> .0.00.:... ..<

 

 

 

58:80: 09m 0:. .00.. :0:.0.:0 m. 00.0 008. 0:.: 80:0: 5....E 0. 00.0 0000. 0:. .0 .05.... 0. 0:...

023
kambh

MQMESU

 

NNNWMDMM.

 

59

Appendix D: Test for Endogeneity of OPTIONS In the Credit Ratings Model
(Rivera-Vuong 1988 Procedure)

This appendix provides details about the Rivers-Vuong (1988) procedure that I
use to test for endogeneity of OPTIONS in the credit rating model. The implementation
of the Rivers and Vuong procedure requires me to first identify instrumental variables for
OPTIONS and regress OPTIONS on instrumental variables as well as all exogenous
variables from the structural equation in a first stage regression. The residual from the
first stage regression will then be included in the credit rating model as an additional
explanatory variable. A significant coefficient on the first stage residual suggests that the
OPTIONS variable is endogenous in the credit rating model and the usual ordered probit
standard errors are not strictly valid (W ooldridge 2002, p.474). In this case, a consistent
estimate of the outstanding ESOs’ coefficient and standard error can still be obtained by
running the ordered probit model with the first stage residual as an additional explanatory
variable in the credit rating model and bootstrapping the correct standard errors.

. Drawing upon recent research on the determinants of option grant behaviors (e. g.,
Core and Guay 2001, Ittner et al. 2003, Hanlon et al. 2003 and Kroumova and Sesil
2003), I first model the level of OPTIONS as a function of the firm’s monitoring cost,
cash flow availability and the firm’ s accounting as well as stock performance.

OPTIONS: f (MONITORING COST, CASH FLOWAVAILABILITY, FIRM

PERFORMANCE).

Given that these economic factors are also likely to play a role 1n credit rating
agencies’ credit risk decisions, I use the industry median of each economic factor detailed
below as the instruments for OPTIONS in the Rivers-Vuong procedure. The rationale for
the inclusion of each economic determinant and the empirical proxies for the economic
determinants are summarized as follows:

Monitoring cost. Corporate governance literature has long argued that compensation
systems are important control mechanisms that align the interests of the employees with
those of the owners (e.g., Eisenhardt 1989). In contexts when direct monitoring of
employee efforts is costly, firms are likely to use more equity incentives to mitigate the
agency problem. Core and Guay (1999) predict that shareholders face a higher
monitoring cost when there is a greater uncertainty in the firm’s operating environment
(Demsetz and Lehn 1985), when the firm has greater growth opportunities (Smith and
Watts 1992) and when the firm is larger and less centralized (Jensen and Meckling 1976).

Following Core and Guay (1999), I use the logarithm of idiosyncratic risk (RISK)
as a proxy for the uncertainty in the firm’s operating environment. The idiosyncratic risk
is measured as the standard deviation of the market model residuals calculated using the
firm’s stock return data for the preceding 36 months. I also include the standard deviation
of return on assets to control for the total risk of the firm (Core et al. 1999). Similar to
Core and Guay (1999), Hanlon et al. (2003) and Ittner et al. (2003), I use the following
variables to capture the firm’s growth opportunity: (1) R&D intensity (RDINT EN), which
is measured as R&D expenditures (Compustat #46) scaled by sales (Compustat #12).
Consistent with Hanlon et al. (2003), I predict that firms with greater R&D intensity will
use more stock options to attract highly qualified and less risk averse employees; (2)
Growth option per employee (GROW_EMPL), which is measured as the difference
between the market value of equity and the book value of equity divided by the number

60

of employees. This variable is used to capture the importance of human capital. The firm
is more likely to grant options throughout the organization when firms’ growth
opportunities depend largely on attracting and retaining high quality employees; and (3)
Book to market ratio (82M), which is calculated as the book value of equity (Compustat
#60) divided by the market value of equity (Compustat #25*Compustat #199). 32M is
included in the model to control for the growth opportunity at the firm-wide level.
Consistent with prior literature, I expect that firms with lower book to market ratio are
likely to have more growth potential not reflected in the book and thus are more likely to
compensate employees with stock options. Ittner et al. (2003) argue that shareholders
would use less equity-based compensation when there are alternative monitoring
mechanisms in place. Therefore, I include LE VERAGE, defined as the ratio of total debt
to total assets, to proxy for external monitoring by debtholders. Finally, I follow Hanlon
et al. (2003) and use the logarithm of sales (LOGSALES) to capture the monitoring
difficulty arising from firm size.

Cash flow availability. As discussed before, firms facing financial constraints are likely
to substitute equity-based compensation for cash pay to conserve cash. Therefore, I
include two variables to capture the firm’s cash flow availability. The first variable,
CONSTRAIN, is based on Core and Guay (1999) and measured as the three-year average
of cash flows from investing activities plus common and preferred dividends minus cash
flow from operations, all deflated by total assets. The second variable, CASH_EMPL,
measures the firm’s ability to compensate employees with cash. This measure is
calculated by first taking cash flow from operating activities (Compustat #308) minus
cash dividends (Compustat #127), capital expenditures (Compustat #128) and R&D
expenditures (Compustat #46). The resulting measure is then divided by the number of
employees to arrive at CASH_EMPL.

Firm Performance. While the above factors explain the size of the ESO programs, the
value of outstanding ESOs also depends on the fair value per option. The value of the
options increases when the firm performs well. As a result, I include two firm .
performance measures as the determinants of OPTIONS. Specifically, I include the firm’s
AVGROA, calculated over the three-year period between t-2 and t to capture the firm’s
accounting performance. I also include the three-year buy-and-hold stock returns,
RET3YR, calculated using CRSP monthly returns over the t-2 to t period, to measure the
firm’s stock performance.

My OPTIONS model is as follows:

OPTIONS. =5 +5 RISK. +5 RDINTEN. +5 GROW_EMPL. +5 STDROA.
z,t O l 1,: 2 t,t 3 l,t 4 1,t
+5 82M . +5 LOGSALES. +5 CASH __EMPL. +5 CONSTRAIN.
5 1,t 6 1,t 7 1,1 8 1,t

+59AVGROAI.’ t + 510RET3YRi, t + 51 lLEVERAGEi, t + 8i, 1

(OPTIONS Model)
To determine how well the factors identified in this appendix explain the variation
in OPTIONS for my credit rating sample, I report the results of the OPTIONS model in
Table D.l. Consistent with my predictions, firms with greater idiosyncratic risk, total
firm risk, higher levels of firm-wide as well as per-employee growth opportunities, and
better recent stock return performance have greater levels of OPTIONS. However, larger
firms and firms that are more cash-constrained do not appear to use more ESOs to

61

compensate employees. This latter finding is consistent with Ittner et al.’s (2003) result
that less cash constrained new economy firms make larger equity grants. In addition, the
coefficient on CASH_EMPL is of mixed signs, which is inconsistent with the
conventional wisdom that ESOs are used to substitute for cash compensation to preserve
scarce cash. The adjusted R-squared of OPTIONS model ranges from 38% to 61%,
suggesting that the economic factors identified in this appendix exhibit a reasonable
statistical power in explaining OPTIONS.

As pointed out by econometricians, identifying appropriate instruments that are
correlated with the endogenous variable but are exogenous in the structural equation is
the most difficult task when adopting the instrumental variable (IV) approach to applied
research. Larcker and Rusticus (2005) demonstrate that using instruments that are only
weakly correlated with the endogenous variable results in a considerable loss of statistical
power associated with IV estimation. In addition, when the instruments selected by the
researchers are not substantially ‘more exogenous’ then the endogenous variable, the IV
estimate is likely to have a larger asymptotic bias than the bias in the OLS estimate.

Although the second stage regression in the Rivers-Vuong procedure is an
ordered probit as opposed to an OLS regression, the importance of the IV quality carries
over. However, there are reasons to believe that creditors’ credit risk assessments are not
independent of the economic determinants of option grant behaviors discussed above. In
fact, the firm-specific economic determinants of OPTIONS fail the overidentifying
restrictions test when these variables are used as instruments for OPTIONS in the two-
stage least squares estimation for the yield-spread sample.37 As a result, I use the industry
medians of these economic determinants based on the two-digit SIC code as my
instrumental variables in the Rivers-Vuong procedure because industry medians are less
likely to determine firm-specific credit ratings and yield spreads but are still likely to
affect the firm’s ESO compensation policies.

Table D2 reports the results of the Rivers-Vuong first stage regression. The
partial R-squared of the instruments are rather modest, indicating that the selected
instruments are only weakly associated with the suspected endogenous variable,
OPTIONS. Based on the benchmarks developed by Stock et al. (2002), the small partial
F-statistics in Table D2 indicate the presence of a weak instrument problem.38 Therefore,
I repeat the Rivers-Vuong procedure replacing the industry medians with the firm-
specific OPTIONS determinants as the instrumental variables for OPTIONS. The results
are reported in the bottom section of Table D2. Compared to industry median
instruments, firm-specific instruments have much higher partial R-squared and partial F-
statistics, suggesting that they have a higher correlateion with OPTIONS. However, the
inferences drawn from the Rivers-Vuong procedures are not sensitive to the set of
instruments used in the first stage regression. Specifically, none of the first-stage

 

37 The fact that the determinants of OPTIONS are likely to be endogenous in the structural equation
suggests that these variables are potential correlated omitted variables in my credit rating and yield spread
models. As a sensitivity test, I re-estimate my credit rating and yield spread models including all of the
OPTIONS determinants identified in this appendix. The positive relationship between outstanding E805
and the cost of debt remains significant at the conventional levels even after controlling for these potential
correlated omitted variables.

38 Stock ct al. (2002) suggest that researchers suffer from the weak instrument problem if the partial F-
statistics from the first stage regression are less than 8.96, 11.59, 12.83, 15.09 and 20.88 when the number
of instruments is 1, 2, 3, 5, and 10, respectively.

62

residuals appears to be significant in the second stage ordered probit model. For example,
the p-value of the first stage residual is 0.99 (0.54) when industry medians (firm-specific
characteristics) are used as instruments and all available firm-year observations are
included in the Rivers-Vuong procedure. Taken together, the results from Rivers-Vuong
(1988) procedure fail to reject the null hypothesis that OPTIONS are exogenous in the
credit rating model, which suggests a regular ordered probit model is appropriate. In
addition, since industry-medians are more likely to be exogenous in the structural
equation but suffer from the weak instrument problem, and the firm-specific
characteristics are better correlated with OPTIONS but likely to have the endogeneity
problem, the instrumental variable estimation is unlikely to yield more reliable statistical
inferences than regular ordered probit and OLS estimation. As a result, the empirical
analyses for the rest of the dissertation are based on the ordered probit and OLS
regressions for the credit rating and yield spread models, respectively.

63

. Table D.1
OLS Regressions of the Determinants of OPTIONS (Dependent variable=OPTIONS)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Predicted OLS Re essions of the Determinants of OPTIONS
Sign All firm-years Year 2001 Year 2002 Year 2003
INTERCEPT ? 0.182*** 0.157* 0.0290.101*** 0262*"
(0.038) (0.093) (0.023) (0.062)
RISK + 0.038*** 0.055* 0.018***** 0.029*
(0.012) (0.031) (0.007) (0.016)
STDROA + 0.160** 0.207* 0.134*** 0.166
(0.067) (0.106) (0.038) (0. 120)
RDINTEN + -0.025 0029 0.061 *** 0.062***
(0.071) (0.093) (0.029) (0.080)
GROW_EMPL + 0.002" 0004*“ 0.002*** 0.001
(+100) (0.001) (0.001) (0.0006) (0.001)
BZM - -0.038*** -0.035* -0.017*** -0.07 1*"
(0.01 l )) (0.021) (0.006) (0.017)
LEVERAGE - -0.078*** -0.013 -0.023 -0.107***
(0.022) (0.038) (0.020) (0.028)
LOGSALES + -0.006** -0.003 -0.005*** -0.014***
(0.003) (0.005) (0.001) (0.003)
CONSTRAIN + 007 8* -0.041 * -0.046** -0.07 1
(0.043) (0.078) (0.023) (0.043)
CASH_EMPL - -0.029** -0.015 0.017" 0006
(+100) (0.043) (0.024) (0.007) (0.010)
AVGROA + 0.081 0.208* 0.080* 0.132
(0.061) (0.121) (0.063) (0.082)
RET 3 YR + 0.035*** 0.047" 0.013*** 0.008
(0.013) (0.020) (0.004) (0.005)
Sample Size 867 298 309 260
Adjusted R-StEared 0.4566 0.6056 0.4031 0.3758

 

*,**,*** significant at the 0.1, 0.05 and 0.01 levels (two-tailed).
Heteroskedasticity-robust standard errors are in the parentheses. For the pooled cross-sectional

regression, the standard errors are also adjusted for serial correlation per Rogers (1993).

RISK is the logarithm of idiosyncratic risk, measured as the standard deviation of the market model
residuals calculated using the firm’s stock return data for the preceding 36 months. STDROA is the

 

standard deviation of return on assets calculated over the three-year period between t-2 and L RDINTEN
is the R&D intensity, measured as R&D expenditures (Compustat #46) scaled by sales (Compustat #12).
GRO W_EMPL is growth option per employee, measured as the difference between the market value of

‘ equity and the book value of equity divided by the number of employees. BZM is the book to market
ratio, calculated as the book value of equity (Compustat #60) divided by the market value of equity
(Compustat #25 *Compustat #1 99). LE VERAGE is defined as the ratio of total debt (Compustat
#9+Compustat #34) divided by total assets. LOGSALES is the logarithm of sales. CONSTRAIN is
measured as the three-year average of cash flows from investing activities plus common and preferred
dividends minus cash flow from operations, all deflated by total assets. CASH_EMPL equals the cash
flow from operating activities (Compustat #308) minus cash dividends (Compustat #127), capital
expenditures (Compustat #128), and R&D expenditures (Compustat #46), all scaled by the number of
employees. AVGROA is the average ROA calculated over the three-year period between t-2 and t, where
ROA is income before extraordinary items divided by total assets. RET3 YR is the three-year buy and hold
returns calculated using CRSP monthly returns over the t-2 to t period.

Table D.2
OLS Regressions of the Determinants of OPTIONS (Dependent variable=OPTIONS)

 

OLS Re essions of the Determinants of OPTIONS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“names All firm-years Year 2001 Year 2002 Year 2003
Instruments:
INTERCEPT . .0024 -0.145 -0.016 -0092
(0.050) (0.118) (0.036) (0.075)
INDRISK 0.012 .033 -0005 -0.041*
(0.015) (0.034) (0.013) (0.023)
INDSTDROA -0.218 -O.178 0090 0.397
(0.151) (0.303) (0.140) (0.443)
INDRDINTEN 0.358** 0.474 0.201 * 0574*
(0.165) (0.387) (0.118) (0.344)
INDGROW_EMPL (+100) 0.004** 0.01 1 -0003 -0004
(0.002) (0.013) (0.005) (0.003)
11/032114 -0.045*** -0.046 0019 0.011
(0.016) (0.035) (0.016) (0.034)
INDLOGSALES -0.006 0.003 -0.006** 0012***
(0.004) (0.010) (0.003) (0.004)
INDCASILEMPL (+100) 0.020 0.008 0.030 0.015
(0.022) (0.129) (0.043) (0.060)
INDCONSTRAIN 0.014 0.0004 0.025 . 0.016
(0.073) (0.211) (0.061) (0.143)
mm VGROA -0010 -0.261 0.023 0.382
(0.114) (0.400) (0.082) (0.262)
INDRET3YR 0.032*** 0.031 0.014 0.016
(0.009) (0.037) (0.009) (0.011)
INDLEVERAGE 0.031** 0047* 0.012 0.023
(0.013) (0.027) (0.008) (0.280)
Control variables:
ACCQUALITY 0.005 0.091 0.0004 0.034**
(0.019) (0.127) (0.010) (0.016)
37191104 0.245*** 0.339*** 0.156*** 0.038
(0.051) (0.124) (0.030) (0.131)
srocro 01 18 -0249 -0.068 0.065
(0.145) (0.396) (0.067) (0.156)
STDRET 0.550*** 1.100 0.304*** 0.522***
(0.157) (0.536)** . (0.068) (0.177)
ROA 0.178*** 0.255*** 0.130*** 0.271***
(0.040) (0.084) (0.020) (0.077)
CASHFLOWS 0.189** 0.125 0.099** 0.1 11
(0.074) (0.135) (0.050) (0.103)
LEVERAGE 0.030 -0.069 0.009 .0051
(0.022) (0.054) (0.018) (0.034)
SIZE 0.003 0.002 0.002 0.0003
(0.003) (0.005) (0.002) (0.004)
BETA 0.013 ' -0001 0.002 0.023
(0.013) (0.029) (0.007) (0.015)
CAPJNTEN -0.018** -0.026* -0.006 0.002
(0.009) (0.015) (0.006) (0.015)
INT_COV (+100) 0.030 0.040 0.030** 0.005

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(0.021) (0.044) (0.0001) (0.012)
82M -0.034*** -0.08 l *** -0.012** -0.055***

(0.01 1) (0.030) (0.005) (0.016)
Industry median instruments
Sample Size 913 321 320 272
Adjusted R-Squared 0.4460 0.4523 0.6706 0.6558
Partial R-guared 0.0342 0.024 0.062 0.090 1
Partial F-statistic_(p-value) 2.86 (0.001) 0.66 (0.773) 1.78 (0.056) 2.23 (0.013)
Rivers-Vuong test p—value 0.99 0.54 0.29 0.83
Firm-specific instruments
Sample Size 865 297 308 260
Adjusted R-Squared 0.6285 0.8065 0.7326 0.7196
Partial R-squared 0.3383 0.6376 0.2241 0.2217
Partial F-statistic (p-value) 42.60(<0.001) 46.54(<0.001) 9.44(<0.001) 8.77(<0.001)
Rivers-Vuongtest p—value 0.54 0.61 0.12 0.28

 

*,**,*** significant at the 0.1, 0.05 and 0.01 levels (two-tailed).
Heteroskedasticity-robust standard errors are in the parentheses. For the pooled cross-sectional

regression, the standard errors are also adjusted for serial correlation per Rogers (1993).

IND# is the industry median of variable # calculated based on the two-digit SIC code, where # is one of
the variables in the OPTIONS model.

66

 

Appendix E: Estimation of Excessive ESO Grants

To identify firms that grant excessive ESOs requires a model of benchmark 580
grants. Using the economic determinants of option- grant behaviors identified 1n
Appendix C, I first estimate the benchmark ESO grant as follows. 39

ESOGRANTS“ = 50 + 51mm” -1 + 52RDINTENI.’ t _1 + 5BGROW _ EMPLI.’ t _1

+5 4STDROAi, t -l + 6532Mi, t _1 + 56LOGSALESiJ“l + 5.7 CASH _ EMPLI.’ t _1 +

58CONSTRAINi’t _1 + 59AVGROAI.’ t _1 + 510RET3YRi, t _1 + 51 ILEVERAGEI.’ t _

24 2003
+ Z ¢NINDUSTRYNJJ + Z aymm
N =1 1’: 1992
(ESOGRANT S Model)

where ESOGRANTS is the fair value of the firm-wide ESO grantsawardedin year
t and scaled by total assets. ESOGRANTS is measured as the Black-Scholes value of
individual executive option grants estimated by ExecuComp (ExecuComp N UMSECUR)
divided by the percentage of total option grants in a given year represented by the
executive option grants (ExecuComp PCTTOTOP). All economic determinants are
measured at the end of year H. In addition, I also include industry and year dummies in
the cross-sectional option grant model in order to control for any industry or time-specific
variation in ESOGRANTS not captured by the identified independent variables.

The residual from the benchmark model represents the ESO grants that are not
explained by the included economic factors and 1s therefore used as a proxy for excessive
ESO grants. 40To create the indicator variable of excessive ESO compensation,

EX CESS_ESO regression residuals are averaged over the five- -year period from year t-4
to year t in order to smooth out the temporary fluctuations in the hypothesized economic
determinants.

The ESOGRANTS model is estimated using all firm-years with necessary data in
ExecuComp. The final sample consists of 12,092 firm-year observations from 1992 to
2003. To mitigate the undue influence from extreme outliers, all variables are winsorized
at the 1% and 99% levels. Coefficients on industry and time dummies are suppressed for
expositional convenience. Results presented in Table E.1 indicate that coefficients on

 

39 While there are reasons to believe that monitoring cost and cash flow availability are associated with the
value of both outstanding ESOs and new ESO grants, it is not immediately clear how firm performance as
described in Appendix C would be associated with new ESO grants for a given year. I include firm
performance in my ESO grant model because agency theory predicts that compensation is an increasing
function of firm performance. Core and Guay (2001) argue that if options are used to substitute for cash
compensation, option grants are expected to be greater when contemporary and prior firm performance 1s
better. It 1s worth noting that although I use the same set of firm characteristics to predict the value of
outstanding ESOs and new ESO grants, the fact that the adjusted R2 from the OPTIONS model and
ESOGRANTS model are both greater than 0.4 suggests that this set of variables appears to exhibit
reasonable explanatory power in both models.

Admittedly, the residual term is only a noisy proxy for excessive ESO compensation due to the potential
model misspecification or measurement errors.

67

LOGSALES, CASH_EMPL, CONSTRAIN and AVGROA are inconsistent with my
predictions. I find that smaller fnms and firms that have more cash are more likely to use
ESOs to compensate employees. While these results are contrary to my predictions, they
are not inconsistent with the empirical findings in prior research (i.e., Hanlon et al. 2003
and Ittner et al. 2003, respectively). Recent accounting performance, however, does not
explain new ESO grants for a given year. The adjusted R2 is 52% for the overall
benchmark model and 45% for the model without industry and year dummies. Overall,
the economic factors identified in this appendix explains a fair amount of the variation in
ESO grants. ‘

Table E.l
OLS Regressions of the optimal option grants (Dependent variable=ESOGRANTS)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OLS Regressions of the Determinants
Variables Predicted Sign of ESOGRANTS
All firm-years
INTERCEPT ? 0.0460***
(0.0043)
RISK + 0.0113***
(0.0014)
STDROA + 0.0721***
. .. (0.0132)
RDINTEN + 0.0522***
(0.0071)
GROW_EMPL (+100) + 0.0014***
(0.0001)
BZM - -0.0077***
(0.0010)
LEVERAGE - -0.0267**?"
(0.0026)
LOGSALES + -0.0017***
(0.0003)
CASH_EMPL (+100) - (100324111111:
(0.0011)
CONSTRAIN + ~0.0087*
(0.0046)
AVGROA + -0.0159
(0.0118)
RET3YR + 0.0019***
(0.0004)
Sample Size 12092
Adjusted R-Squared overall 0.5160
Adjusted R-squared without industry and year 0 4530
dummies '

 

 

 

*,‘**,*** significant at the 0.1, 0.05 and 0.01 levels (two-tailed).

The standard errors in parentheses are adjusted for serial correlation and heteroskedasticity per Rogers
(1993). Coefficients on industry and time dummies are not shown. GROW_EMPL and CASH_EMPL are
divided by 100 to preserve significant digits of the coefficients. All variables are as defined in Table

DJ.

68

 

Table 1
Sample Selection

Panel A: Credit Rating Sample from 2001 to 2003
No. of Firm- No. of F inns“

 

 

 

 

 

 

 

Years
All S&P500 firms in R.G. Associates’ stock option database 1491 497
Less: firms not listed in Compustat (3) (1)
Add: non-S&P 500 firms that are potential heavy option users 417 139
Less: utilities and financial services firms (414) (138)
Initial credit ratings sample 1491 497
Less: firm-years without credit ratings on Compustat (411) (147)
Less: firm-years without necessary Compustat, CRSP and (116) (59)
ExecuComp data for the credit rating model
Less: firm-years voluntarily adopting the fair value method (51) (45)
under SFAS 123
Number of firm-years used in the credit ratings analyses 913 338
Panel B: Industry and S&P Index distribution for the credit rating sample
Industry Number of % S&P Index Number of %
, firm-years firm-years
(unique firms) .
Mining and construction 14 1.53 S&P 500 789 (290) 86.42 (85.80)
Food 49 5.37 S&P 58 (21) 6.35 (6.21)
MidCap
Textile, printing and 68 7.45 S&P Small 22 (10) 2.41 (2.96)
publishing Cap
Chemicals 52 5.70 Non_S&P 44 (17) 4.82 (5.03)
Pharmaceuticals 54 5.91
Extractive Industries 56 6.13
Durable manufacturers 233 25.52
Computers 119 13.03
Transportation 65 7.12
Retail 131 14.35
. Services 70 7.67
Others 2 0.22 .
Total 913 100 Total 913 (338) 100 (100)

 

 

’0 Note that the adjustments in the ‘number of firms’ column do not add up to the number offirms in the
final sample. This rs because when firms have missing data m one of the three sample years, they will show
up as adjustments. However, as long as firms have at least one firm-year remaining in the final sample, they
will not be excluded from the bottom-line number of firms 1n the final sample.

69

Table 1
Sample Selection

Panel C: Yield Spread (New Bond) Sample from January 1, 2001 to October 30, 2004
No. of Bond No. of Firmsin

 

 

 

 

 

 

Issues
Total number of new industrial bond issues from January 2001 . 1039 317
to October 2004
Less: bonds having a maturity date that is earlier than the issue (91) (62)
date on SDC
Less: firms having multiple bond issues during a given year (459) (0)
(only retain the largest issue) '
Less: firms not listed in Compustat _ (22) (18)
Less: firms not listed in ExecuComp (79) (59)
Less: firms having missing or negative credit spreads (5) (5)
Less: firms lacking necessary Compustat, CRSP and (63) (42)
ExecuComp data for the credit spreads model
Less: firms following Canadian GAAP (3) ( 1)
Less: firm-years voluntarily adopting the fair value method (8) (8)
under SFAS 123 -
Number of new bond issues used in the yield spread analyses 309 195
Panel D IndustrLand S&P Index distribution for the credit spread sannrle
Industry Number of % S&P Index Number of %
firm-years firm-years
(unique firms)

Mining and construction 2 0.65 S&P 500 242 (141) 78.32 (72.31)
Food 24 7.77 S&P 42 (35) 13.59 (17.95)

MidCap ,
Textile, printing and 31 10.03 S&P Small 10(8) 3.24 (4.10)
publishing Cap _
Chemicals 28 9.06 Non_S&P 15 (11) 4.85 (5.64)
Pharmaceuticals 16 5.18
Extractive Industries 28 9.06
Durable manufacturers 61 19.74
Computers 19 6.15
Transportation 32 10.36
Retail 44 14.24
Services 20 6.47
Others 4 1.29
Total 309 100 Total 309 (195) 100 (100)

 

The industry membership is assigned based on Lu’s (2003) classification: Agriculture (SIC= 0100-0999),
Mining and construction (SIC= 1000-1999 except 1300-1399), Food (SIC=2000-2111), Textile, printing
and publishing (SIC=2200-2799), Chemicals (SIC=2800-2824 and 2840-2899), Pharmaceuticals

 

" Note that the adjustments in the “number of firms’ column do not add up to the number of firms in the
final sample. This is because when firms issue multiple bonds but have missing data necessary for
estimating the yield spread model, they will show up as adjustments. But as long as they have at least one
bond issue remaining in the final sample, they will not be excluded from the bottom-line number of firms in
the final sample.

70

2840-2899), Pharmaceuticals (SIC=2830—2836), Extractive industries (SIC=2900-2999 and
1300-1399), Durable manufacturers (src=3000-3999 except 3570-3579 and 3670-3679),
Computers (SIC=7370-7379, 3570-379, and 3670-3679), Transportation (81c=4000-4899),
Utilities (49004999), Retail (SIC=5000-5999), Services (SIC=7000-8999, except 7370-7379)
and Other (SIC>=9000).

71

Table 2

Descriptive Statistics

Panel A: Descriptive statistics for all firm-year‘observations in the credit rating sample across credit rating

grade (2001-2003)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Speculative Grade Investment Grade
(n=223) (n=690)
Mean Two-sided p-value for
(Median) t-test
[Standard Deviation] (W ilcoxon Test)
RATING 2.6413 4.6319 <.0001
(3.0000) (5.0000) (<.0001)
[0.4900] [0.7273]
OPTIONS 0.0900 0.0423 <.0001
(0.0316) (0.0253) (0.0318)
[0.1758] [0.0530]
MGTESO 0.0196 0.0079 <.0001
(0.0066) (0.0040) (0.0002)
[0.0384] [0.01 15]
NONMGTESO ‘ 0.0703 0.0344 0.0003
(0.0239) (0.0190) 0.1097
[0.1449] [0.0461]
ACCQUALITY -0.1068 -0.0859 0.0953
(-0.0681) (-0.0441) (<.0001)
[0.1607] [0.1683]
STDROA 0.1080 0.0341 <.0001
(0.0566) (0.0249) (<.0001)
[0.2061] [0.0333]
STDCFO 0.0575 0.0400 <.0001
(0.0516) (0.0332) (<.0001)
[0.0338] [0.0286]
STDRET 0.2052 0.1 107 <.0001
(0.1922) (0.1047) (<.0001)
[0.0727] [0.0306]
LEVERAGE 0.3358 0.2674 <.0001
(0.3081) (0.2618) (<.0001)
[0.1894] [0.1335]
ROA -0.0612 0.0574 <.0001
(0.0010) (0.0559) (<.0001)
[0.3535] [0.0628]

 

72

 

Table 2

Descriptive Statistics

Panel A (Continued): Descriptive statistics for all firm-year observations in the credit rating sample across

credit rating (2001-2003)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Speculative Grade Investment Grade
(n=223) (11:690)
Mean Two-sided p-value for
(Median) t-test
[Standard Deviation] (W ilcoxon Test)

CASHFLOWS 0.0787 0.1235 <-0001
(0.0766) (0.1205) (<-0001)
[0.0590] [0.0543]

SIZE 7.6624 9.1259 <.0001
(7.7763) (8.9981) (<.0001)
[1.2319] [1.1698]

BETA 1.5518 0.8669 <-0001
(1.4603) (0.8654) (<-0001)
[0.7185] [0.4311]

CAP_INTEN 0.5129 0.6190 0.0002
(0.3969) (0.5423) (<.0001)
[0.3722] [0.3517]

m'r_cov 7.9131 19.6794 <.0001
(3.5447) (9.8132) (<.0001)
[24.9985] [34.0368]

BZM 0.4957 0.3906 0.012(1)
(0.4061) (0.3383) (000 >
[0.6149] [0.2769]

REPUR 0.1570 0.2130 0.0535
(0.0000) (0,0000) (0.0689)
[0.3646] [0.4098]

 

All variables are as defined in Appendix C.

73

 

Table 2

 

 

 

 

 

 

 

 

 

 

 

 

Descriptive Statistics
Panel B : Descriptive statistics for the yield spread sample across high and low spread firms (January 2001
October 2004)
Low Spread High Spread
(n=154) (n=155)
Mean Two-sided p—value for
. (Median) t-test
[Standard Deviation] (W ilcoxon Test)
SPREAD 89.8571 267.6323 <.0001
(90.5000) (232.0000) (<.0001)
[27.9926] [115.4479]
INVGRADE 0.9935 0.7677 <.0001
(1.0000) (1.0000) (<.0001)
[0.0806] [0.4236]
MATYRS 7.7273 8.4774 0.0343
(7.0000) (10.0000) (0.0357)
[3.2944] [2.8950]
ISSUESIZE 5.6704 5.6458 0.8122
(5.7664) (5.7038) (0.5741)
[0.9540] [0.8615]
CALLABLE 0.0260 0.1 161 0.0020
(0.0000) (0.0000) (0.0023)
[0.1596] [0.3214]
SUBORD 0.0000 0.0645 0.0014
0.0000 0.0000 (0.0015)
0.0000 0.2465
OPTIONS 0.0433 0.0308 0.0162
(0.0252) (0.0188) (0.0085)
[0.0518] [0.0374]
MGTESO 0.0057 0.0074 0.1265
(0.0032) (0.0032) (0.5411)
[0.0082] [0.01 10]
NONMGTESO 0.0376 0.0235 0.0017
(0.0193) (0.0141) (0,0008)
[0.0472] [0.0289]

 

 

 

 

 

74

 

Table 2

 

 

 

 

 

 

 

 

 

 

 

 

Descriptive Statistics
Panel B (Continued): Descriptive statistics for the yield spread sample across high and low spread firms
January 2001-October 2004)
Low Spread High Spread
(n=154) (n=155)
Mean Two-sided p-value for
(Median) t-test
[Standard Deviation] (W ilcoxon Test)

ACCQUALITY -0.0474 -0.0794 0.0244
(-0.0343) (-0.0418) (0.0453)
[0.0486] [0.1687]

STDROA 0.0243 0.0347 00013
(0.0175) (0.0246) (0.0019)
[0.0213] [0.0335]

STDCFO 0.03 19 0.0396 0.0055
(0.0268) (0.0317) (0.0005)
[0.0234] [0.0250]

STDRET 0.0978 0.1223 <.0001
(0.0940) (0.1440) (<.0001)
[0.0228] [0.0400]

LEVERAGE 0.3052 0.3450 0.0032
(0.3020) (0.3441) (0.0023)
[0.1175] [0.1176]

ROA 0.0749 0.0384 <.0001
(0.0660) (0.0384) (<.0001)
[0.0564] [0.0440]

CASHFLOWS 0,1235 0.0998 <.0001 .
(0.1270) (0.0949) (<.0001)
[0.0520] [0.0508]

SIZE 9.7516 8.3014 <.0001
(9.6377) (8.1951) (<.0001)
[1.3048] [1.1510]

BETA 0.6666 0.6673 0.9875
(0.6795) (0.5441) (0.5692)
[0.3878] [0.4554]

 

 

 

 

 

75

 

Table 2

 

 

 

 

 

 

 

Descriptive Statistics
Panel B (Continued): Descriptive statistics for the yield spread sample across high and low spread firms
(January 2001-October 2004)
Low Spread High Spread
(n=154) (n=155)
Mean Two-sided p-value for
(Median) t-test
[Standard Deviation] (W ilcoxon Test)

CAP_INTEN 0.6687 0.7432 0.0749
(0.5678) (0.7049) (0.1655)
[0.3315] [0.3981]

INT_COV 14.8556 6.9613 <.0001
(9.6715) (5.8824) (<.0001)
[15.0222] [5.3384]

82M 0.3283 0.5404 <.0001
(0.2723) (0.4739) (<.0001)
[02518] [0.3730]

REPUR 0.2662 0.2065 0.2175
(0.0000) (0.0000) (0.2181)
[0.4434] [0.4061] -

 

 

 

 

 

All variables are as defined in Appendix C.

76

 

 

 

 

 

 

 

 

 

88$ :88 88$ 88$ 68$ 88$ $8 88$ :68 no.8 8$ 88$ 68$ :8

«+8 8... c3 :18 3.8 Na... 88 and S8 88 25 S8 88
A808 828 68$ 89$ 88$ 88$ 88$ 88$ :38 88$ 5.8 898 88$ 88
28 88 28 and 8.8 :8 $8 28 :8 :8 88 :3 and E
6.8 8.8 88$ §8 68$ 58 88$ 88$ 89$ 88$ 888 88$ 88$ E
3.8 _ .8 28 88 8.8 8.8 as 88 28 28 2.8 $8 2.8 5.0
5.8 $8 88$ 88$ 88$ 88$ A88 88$ save 88$ 88$ a~8 88$ fine
85 8.8 38 28 88 9.8 :8 ”3 8.0 m3 :8 3.8 $8
88$ 88$ 88 69$ 88$ 88$ sore 88$ 89$ 88$ 88$ 88$ 888 use.
:8 ”we 58 :8 “no and 88 £8 £8 28 NS 2.8 Rd
8.8 8.8 8.8 sev- seve sex sex 8.8 see save 58 save save $8..
28 n3 :6 £8 :5 .ed 88 :8 85 88 8.8 NS :3 -55
88$ 88$ 88.8 88$ 89$ 89$ 88$ 68$ 89$ §$ 58 save 88$ <9.
:8 c; 8.8 38 and and 28 88 88 38 a... Se and
A88 “8.8 898 3.8 88$ 88$ 88.8 :88 28.8 5.8 88$ 88$ 88$ 855m:
88 $8 :8 88 88 38 28 38 8.8 88 a; «N8 88
898 5.8 sci 88$ §$ 88$ 88$ A28 898 save 688 88$ 88$ 55.5
8:. 88 88 $8 38 88 «+8 Ed 98 cod Z8 _ 3 $8
E8 888 A88 88$ 898 3.8 89$ 388 save 88$ save 898 88$ 285
88 8.8 88 88 :8 3.8 :8 3.8 “3 e2 38 2.8 $8
8N8 818 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ «.955
88 88 88 m2 :8 :8 $8 2.8 :3 2.8 88 26 :8
888 28.8 88$ 88$ :88 8.8 §8 A38 88$ 88$ 89$ 88$ 88$ Ease
S8 88 85 88 88 8.8 8.8 8.: S8 88 28 m8 :5 8<
88$ 88$ 82 88.8 8.8 888 G~8 8.2 898 888 A808 5.8 E8 mzeEo
:8 N3 :8 +3 8.8 :5 3.8 28 one 28 +8 88 3o
88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 88$ 0252
38 +3 88 8.8 cc... :8 “3 28 $8 38 28 :6 28

e8 rszz. mBQ: E150
2mm .9: .70 SEE me -=m<U <01 mega: Edam 0..—UPS <O¢Em -UU< mZOEO 02F:

82 8175 03:13» . 8 “680 05 ecu 5.58 done—goo n< 35$

5...: 58859

n 035.

 

77

.ggq fl 8.. 858.»... 6.98.... 8238 8.5.. u... 3 8.. 8.50808 note—oboe Anna—89$ .828.— .0 £825... 3 3:89.. 3 2.. 8.9.8.5 =<

 

 

 

 

  

 

 

 

89$ :...8 89$ 89$ 89$ 89$ :98 89v. 8...... 8...... .:.8 89$ 89$ 2a.

~98 2... .~... 898 898 898 8.8 89.. 8... ~..8 8... 8.8 3...
88$ 8...... 2...... 88$ 88$ 88$ 88$ 88$ 8.9... :8... 88$ 88$ 88$ boo
3.8 89.. 28 :.... :.... a... 898 .88 8... .98 8.8 S... 8.8 E.
898 898 89$ .98 88$ :98 898 .38 8...... :98 89$ .~...... 898 ma...
8 .8 . .8 -8 ~..8 89.. 1.... 8... 88 8... e . .8 .~... :8 8... .5
.....$ 898 89$ 8.8 89$ 88$ 898 89$ 88... 898 89$ 89$ 8 .8 :....
2... +98 38 88 8~8 38 :8 en... 8... 8... :8 ~98 8...
89$ 89$ 89$ .88 89$ 89v. 89$ 89$ 89$ :98 :...8 89$ 89$ 8....
$8 8.... :8 ~38 on... no... ..~8 898 -8 28 2... 8... ~98
89$ 89$ 89v. 89$ 89$ 89$ 89$ 89$ .98 898 898 89$ 89$ 38...
.98 8... 3.. 38 .8... .8... 2.8 :8 ~.... ~..8 88 8+... 3.8 .86
89$ 89$ 898 89$ 89$ 89$ 89$ 89$ .88 898 82.. 89$ 89$ <9.
.98 8... .98 .98 ~..8 2... n~8 :.8 8... .98 :.8 ~98 8.8
8+... 89$ 898 898 89$ 8...... 89$ 88... 8~8 898 89$ :98 89$ 838.
88 .48 ~..... ...8 .88 ~..8 8.8 8... 8.8 2.8 8.... :8 .~...
89$ 89$ .. 98 89$ 89$ 89$ 89$ .88 89v. 89$ 89$ 8...... 89$ 8.88
a... .~8 e98 . 9.... .88 898 .88 :... n... .8... -8 8.8 8....
898 898 89... 898 89v. .~...... 8...... .. 98 89$ 89$ 89$ :98 89$ 9.88
8... 8... ~..8 8.... 8.8 2... 8... .98 e~... 49.. .88 8.8 a...
8.8 898 .~...... :..8 898 :...... ....~.8 898 89$ 89$ 89$ .~...... 89$ 3.88
.88 8... .~.... 8... :8 8.... :.... 8.8 2... ~48 ~q8 8... n...
8.8 8.8 898 898 898 898 898 .~...... .38 8...... 898 .~...... :...8 r58...
8... 8.8 89.. ...8 .98 8... .98 n . .8 .9... :8 . .8 m . .8 8.8 .82
89$ 89$ 898 89$ 89$ 88$ 89$ 89$ 898 8~8 :98 8.8 89$ 82656
2.8 8.... :8 8.8 89.. :.... 8+... 88 :.8 8... 2... 88 m~8
89$ 89$ 88... 89$ 89$ 89$ 89v. 89$ 89$ 89$ 88... 8.8 89$ 99:8
~..... :18 8... :... e98 898 $8 -... ~98 n~... .~... 88 88

88 ”ME. 833.. Eng
28 be 3 (Ba gm -=m<o <9— mog 595 9.8.5 (CE .09. 9555 035:8

26. o... 3.. x38. 8.8.880 .m .68..
5.5.2 noun—2.80

n 051—.

78

Table 4
Ordered Probit Regressions of the Effect of Outstanding Employee
Stock Options (Dependent Variable=Rating)

Panel A: Ordered probit analyses of the credit ratianodel

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent Predicted Sign Estimated Coeficient
Variables All firm years Year 2001 Year 2002 Year 2003
OPTIONS ? 2727"" 2571"” 5309” 8.528"
(0.591) (0.750) (2.198) (1.421)
ACCQUALITY + 0.372 1.247“ 0.202 0.504
(0.250) (0.639) (0.346) (0.587)
TDROA - -3.489"”"' -3.248’ -3.291 -4.393"
(1.220) (1.873) (2.550) (2.348)
STDCFO - 0.155 -3.958 0.474 2.987
(1.600) (2.780) (2.994) (2.81 1)
STDRET - -16.275”" -l4.647‘” -16.151““ -l7.830‘””
(1.351) (2.367) (2.479) (2.548)
ROA + 2.612’” 3.017” 2.189‘ 3.423"
(0.755) (1.229 (1.293) (1.847)
CASHFLOWS + 6400*“ 5.552‘" 6908*" 6.102'"
(0.955) (1.592) (1.691) (1.975)
LE VERAGE - -l.970“" -2.586"'” -l.9l9"‘” -l.8l9"‘
(0.330) (0.562) (0.584) (0.622)
SIZE + 0.582‘” 0506*" 0678”" 0.571""M
(0.04 I) (0.067) (0.073) (0.081)
BETA - 0234" 0362“ -0.229 -0.067
(0.102) (0.159 (0.224) 40.217)
CAP_INTEN + 0423‘“ -0.433"'I -0.506“ -0.354
(0.122L (0.212) (0.206) (0.233)
INT_COV + 0.047 0.221 0.001 -0.071
(+ 100) (0.143) (0.248) (0.003) (0.233)
@ZM - -0.028 -0.548“ 0.154 -0.342
(0.127) (0.268) (0.172) (0.353)
Sample Size 913 321 320 272
Psuedo R- 0.4332 0.4330 0.4501 0.4440
Squared
Likelihoo ratio 1 166.85‘" “0.02"“ 425.24‘" 353.25‘“
chi-square
Percentage 63 .09 62 .93 63.13 63.97
correctly
[predicted

 

‘,","" significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses

All variables are as defined in Appendix C.

79

 

Table 4

Ordered Probit Regressions of the Effect of Outstanding Employee Stock Options

(Dependent Variable=Ratiug)

Panel B: Ordered probit analyses of the credit rating model conditioning on the issuing company’s

tendency to rmurchasc shares in response to ESO exercise.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent Variables Predicted Estimated Coefficient
Sign All firm-years Year 2001 Year 2002 Year 2003
OPHONS ? -2.077"* 2225*“ -3.109 -2.539"'
(0.595) . (0.730) (2.332) (1.469)
ACCQUALITY + 0.285 ‘ 1.043 0.091 0.494
(0.253) (0.657) (0.352) (0.490)
STDROA - -3.508”" —3.681 " -3.348 -4.336“
(1.232) (1.908) (2.572) (2.354)
SYDCFO - -0.024 -3.637 -0.364 2.777
(1.616) (2.800) (3.046) (2.841)
STDRET - -15.935“* -l4.160“" -15.627*” -17.687”"
@358) (2.391) (2.493) (2.558)
ROA + 2666"" 2.820“ ‘ 2.291“ 3.471“
(0.762) (1.232) Q .32) (1.855) .
CASHFLOWS + 6761‘“ 5992*” 7.486‘" 6128““
(0.962) 41.607) (1.714) (1.978)
LEVERAGE - 2194“" 2943“" 2042““ -1.983“"
(0.338) (0.579) (0.598) (0.637)
SIZE + 0584"" 0.499‘” 0.689‘“ 0.578‘"
(0.041) (0.067) (0.074) (0.082)
BETA - 0277““ 0413"“ -0.309 -0.083
(0.102) (0.158) (0.227) (0.218)
CAP_INTEN + 0443‘" 0464" 0522” -0.345
(0.122) (0.214) (0.208) (0.233)
JNT_COV (+100) + -0.006 0.157 0.007 0.091
, (0.001) (0.250) (0.308) (0.235)
32M - 0.002 0579*" 0.213 -0.300
(0.127) (0.270) (0.174) (0.356)
REPUR ? 0.669‘" 0793*" 0.624“ 0.694"
(0.146) (0.239) (0.252) (0.307)
REPUR ‘OPHONS - -23.454"‘" -20.057"“ -32.479”* 44.781"
. (4.888) (7.056) (10.869) (10.081)
Sample Size 913 321 320 272
Psuedo R-Square 0.4429 0.4455 0.4600 0.4521
Likelihoo ratio chi- 1193.01‘" 421.82"'“ 434.6?" 359.76‘“
square
Percent correctly 64.07 63.86 65.00 66.18
redicted

 

‘,”,"“ significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses

All variables are as defined in Appendix C.

80

 

Probit Regressions of the Effect of Outstanding Employee Stock Options on Credit

Panel C: Probit Wes of ESOs’ impact on credit ratings (Dependent Variable=INVGRADE)

Table 4

Ratings

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent Variables Predicted Sign Probit
Coefficient Marginal Effect
OPTIONS 3533‘" -0.0362
(1.372)
ACCQUALITY -0.301 -0.00394
(0.477)
STDROA -4.000* -0.02933
12042
STDCFO 4.082 0.028721
(2.866)
STDRET -19.102""' -0.21835
(2.295)
ROA 0.132 0.004797
(1.282)
CASHFLOWS 4.103” 0.057939
(1.812)
LEVERAGE -2.079""“' -0.02954
(0.570)
SIZE 0649"" 0.2071 19
(0.088)
BETA 0646*” -0.08809
(0.193) ,
CAP_INTEN -0.306 0.0320?
40.224)
INT_COV (+100) 0006‘ 0.014341
(0.004) .
32M 0.115 0.008112
(0.218)
Sample Size 913
Psuedo/Adj. R-Square 0.6229
Likelihoo ratio chi-gum 632.29‘"

 

*,",*** significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses.

INVGRADE takes on the value of one if credit rating is BBB- or above (i.e., investment grade), and

zero otherwise.

81

 

Table 5

Ordinary Least Squares Regressions of the Effect of Outstanding Employee Stock Options
on Yield Spreads (Dependent Variable=SPREAD)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Predicted Model (1) Model (2) Model (3) Model (4)
“dwm‘vamb'e‘ sign Coefficient Coefficient Coefficient Coefficient
INTERCEPT 7 290.722m 298.715m 308.284m 316.085m
(78.992) (78.219) (77.618) (76.506)
OPTIONS 7 355.847m 377.198m 106.421 223.054
(1 16.126) (143.129) (204.132; (179.854)
ACCQUALITY - 42.518" 48.765" 42.433" 47.33955
(29.546) (26.499) (24.500) (26.129)
STDROA + 228.054 272.289 212.301 264.500 *
(216.238) (222.553) (217.635) (223.439
STDCFO + 521.160‘ 557.0715 537.290- 569.355“
(305.565) (292.491) (304.972) (292.053)
STDRET + 585.242" 632.176‘" 583.969" 627.932
(246.378) (230.035) (246.138) (229.437)m
ROA - 460.116 493.445 464.959 496.841
(153.309) (158.791) (155.379) (160.342)
CASHFLOWS . 456.654" -360.313* -343.428‘ -349.684"'
(186.464) (186.580) (185.961) (186.562)
11517517405 + 87.200‘ 64.662 84.265“ 63.609
(48.840) (48.470) (48.428) (47.900)
SIZE - -34.796““ -35.358*** -34.849"‘" -35.250m
(5.187) (5.264) (5.168) (5.235)
BETA + -5.969 -905 -5210 -8.084
(15.623) (15.038) (15.873) (15.372)
CAP_HVTEN - 3.942 4.063 4.073 -0.620
418.284) (19.104) Q8278) (19.195)
INT_COV . 0.097 -0.l6l 0.070 -0.144
(0.388) (0.431) (0.399) (0.428)
32M + 48.1 13* 51 .234W 47.183‘ 50.544"
(25.681) (25.51 g (24.510) (24.351)
MATYRS + 4.544 4.106 4.029 -0.641
(1.751) @664) (1.904) (1.839)
ISSUESIZE 7 17.643m 16.461m l8.084"* l6.700*"
(5.409) (5.183) (5.566) (5.340)
CALLABLE + 87.4261m 87.825‘” 88.320m 88.435‘“
(30.881) (30.706) 431.035) (30.921)
SUBORD + 57.241 53.528 57.695 54.485
(37.879) (37.492) (38.150) (37.663
REPUR 7 41.882" 42.954"
(18.713) (18.900)
REPUR ‘OPTIONS 7 -53.362 -541 .860
(182.380) (329.309)
LONG 7 -26.318 -26.747
(21.717) (21.363
LONG*OPTIONS 7 277.808 165.256
(207.732) (198.324)
REPUR‘LONG‘OPTIONS + 512.841 *
(307.509)
N 309 309 309 309
Adjusted R-Squared 0.5608 0.5750 0.5598 0.5734

 

‘,“,”"’ significant at the 0.1, 0.05 and 0.01 levels (two-tailed)
The standard errors in parentheses are adjusted for serial correlation and heteroskedasticity per Rogers

(1993).

All variables are as defined in Appendix C.

82

 

Table 6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specification and Sensitivity Tests
Panel A: Ordered Lrobit analyses on the “change” giecification of credit ratinflnodel
Independent Variable Expected Siln Model (1) Model (2)
Contemporaneous One-year-ahead
DOWNGRADE DOWNGRADE
DIFF. 7 1.662m 6.105m
0””0’“ (0.607) (1.569)
DEF. - 0945* -0.426
ACCQWW (0.498) (0.563)
DIFF. + 0.455 -8.228”"
STDROA (1.463) (3.017)
DIFF. + -4.968 14.712m
3790’" (3.298) (5.403)
DIFF. + 21.679"" 42.269" ~
3773“” (4.435) (6.477)
DIFF. - 4.191‘1 0.137
“04 (0.710) (1.297)
DIFF. - 43.921m 40.177*
CWLOW (3.494) (5.302)
01”. + 1.013 -2.195
LEVERAGE (0.798) (1.334)
DIFF. - 0886““ 4.231m
3125 (0.178) (0.283)
01”. + 0.216 0.7435"
”m (0.170) (0.247)
DIFF. - 0.263 0.152
CAP-”V75” (0.702) (0.868)
DIFF. - .0003 0.006
”VT-C0” (0.003) (0.005)
DIFF__ + -0.274 0.71 1""
32” (0.219) (0.245)
Sample Size 566 286
Pseudo 0.1911 0.1902
R2

 

 

 

 

 

*,“,“"‘ significant at the 0.1, 0.05 and 0.01 levels (two—tailed)
Standard errors are in parentheses.
DOWNGRADE is an ordinal variable set equal to one if the firm’s credit rating is downgraded fiom year H
to t, zero if there is no change in the firm's credit rating, and minus one if there is an upgrade in the firm’s

1 credit ratings.

DIFF_X is the change in the variable X from year H to t, where X is one of the variables in the credit

rating model.

83

 

Panel B: Ordered probit analyses on the “change” glecification of OPHONS model

Table 6

Specification and Sensitivity Tests

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable Expected Sign Model (1) Model (2)
Contemporaneous One-year—ahead
DIFFOPHON DIFFOPHON
0012*" 0.012m
”RC5" - (0.004) (0.003)
DIFF. 9 .0010 -0.012
“Am/G (0.009) (0.013)
DIFF. + 0.045" .0029
”9”“ (0.018) (0.023
DIFF. + 0.147 0.3537"
57W“ (0.242) (0.131)
DI”. + 0150*" 0.033
ROME” (0.054) (0.044)
DIFF. + 0.002" 0.001"
GROW_EMPL“! 00) @901) 40-0006L
Dina - .0014 0.006
32” (0.014) (0.008)
mm; - -0.080 -0.045
LEVERAGE (0.099) (0.037)
DIFF. + .0019 .0019
LOGSALES (0.023) (0.016)
DIFF. - 0.027" -0.001
CASH—EMPHHW) (0.012) 40.004)
DIFF. + 0.031 0.063
CONSTRAIN (0.091) (0.069
DIFF_ + -0120 0.106
AVGROA (0.144) (0.121)
DIFF_ + 0.05011" 0015"
”573’” (0.016) (0.006)
Sample Size 541 236
Pseudo 0.4295 0.2226
R2

 

 

 

 

 

‘3‘)“ significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

DIFF_Y is the change in the variable Y fi'om year H to t, where Y is one of the variables in the OPHONS
model described in Appendix D.

The standard errors in parentheses are adjusted for serial correlation and beteroskedasticity per Rogers
(1993).

84

Panel C: Sensitivity analysis partitioning OPTIONS into the component accruing to top-five executives and

Table 6

Specification and Sensitivity Tests

that accruing to non-executive emplolees

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable RATING SPREAD
Expected Sign Coeflicient Expected Sign Coefficient
(Standard Error) (Standard
Error)
Mama ? 3.340 ? -39.204
(2.996) (557.008)
NONMGIESO ? 4427*" ? 429.13?“
(1.016) (138.068)
ACCQUALITY + 0.387 - 43.035“
(0.251) (30.034)
STDROA . 3353*" + 240.286
(1.236) (214.079)
SIDCFO - 0.151 + 507.976
_L (1.602) (308.186)
STDRET - -l6.266"”""' + 584.848‘“
(13$ (245.497)
ROA + 2417*“ - -l48.297
(0.761) (154.114)
CASHFLOW + 6756"" - -363.041"'
(0.972) (187.336)
LEVERAGE - 2011"" + 88200"
(0.331) (48.727)
5125 + 0.598m - -36.016*"
(0.042) (5.462)
BETA - 0224“ + -5.766
(0.102) (15.630)
CAP_INTEN + 0424*” - 3.637
(0122) (18.268)
INT_COV + 0.001 - 0.094
(0001) (0.379)
82M - 0.024 + 46794“
(0.127) (25.715)
Sample Size 913 309
Pseudo 0.4348 0.5857
R2

 

 

 

 

 

 

‘,“"","'" significant at the 0.1, 0.05 and 0.01 levels (two-tailed). The standard errors for the SPREAD
sample are adjusted for serial correlation and heteroskedasticity per Rogers (1993) For ease of exposition,
the intercept of the SPREAD model and the coefficients on bond issue characteristics (MA TYRS,
ISSUESIZE, CALLABLE and SUBORD) for the yield spread model are not shown.

85

Panel D: Sensitivity

analysis using Ashbaugh et al. ’8 (2004) corporate governance variables

Table 6

Specification and Sensitivity Tests

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent RATING SPREAD
Variable
Year 2001 Year 2002 Year 2003
Expecte Coefficient Coefficient Coefficient Expected Coefficient
(1 Sign (Standard (Standard (Standard Sign (Standard
Error) Errog Error) ErroQ
0P770N5 ? -2.589‘”"" -4.735"' -2.779“ ? 415.319“
(0.909) (2.430) (1.537) (137.269)
ACCQUALITY + 3.069” -0.053 0.484 - -61.766
(1.297) (0.407) (0.521) (45.616L
STDROA - -3.965“ -3.524 -3.313 + 59.623
(2.000) (2.728) (2.423) (266.000)
S7DCF0 - -2.604 1.817 2.188 + 595.016
‘ (3.044) (3.222) (3.546) (372.518.)
STDRET - -17.948“* -19.656*" -20.174"'" + 801.889"
(2.808) 42 .843) (3 .001) (267.201)
R04 + 3.019” 3.204“ 4.130“ - -187.465
(1.332) (1.418) (2.030) (158.329)
CASHFLOW + 6634"" 6703*" 6285““ - -500.091"
(1.813 (1.845) (2.091) (205.136)
LEVERAGE - -2.959"'" -2.083”" 1977“" «1» 73.953
(0.642) (0.642) (0.685) (56.590)
SIZE + 0464*" 0.675‘" 0.503‘" - 40.853‘“
(0.085) (0.090) (0.095) (5.955)
BETA - -0.256 0.071 0.075 + -17 .271
(0.173) (0.248) (0.244) (16.223)
CAP_HVTFN + 0716"" 0657"“ -0.466‘ - -4.927
(0.241) (0.225) (0.250) (19.293L
INT_COV + 0.001 0.000 -0.001 - 0.513
(0.003) 40.003) (0.002) (0.494)
32M - -0.647" 0.154 -0.279 + 44.412“
(0.318) (0.186) (0.373) (25.279
BDJND + 0.870 0.562 0.702 - 20.033
(0.557) (0.559) (0.599) (48.515)
POWER - 0.312‘" 0.103 0.057 + -6.983
(0.106) 40.105) (0.1 15) (7.681)
POLICY + -0.003 -0.084 0.000 - 15.344
(0.163) (0.159) 40.170) (15.387)
”V3770“ ? -0.325 -0.1 10 -0.779 ? -45.267
(0.612) (0.594) @672) (49.286)
[”5105um - -0.154 0.355 -0.441 + 0.973
(1 .091) (1.073) 4.188) (68.894)
BLOC/(HOLDERS ? 0.018 -0.042 -0.033 ? -0.991
(0.040) (0.040) (0.042) (3.7 13)
(UNDEX ? -0.013 0.019 0.043 ? -1.059
(0.034) (0.033) (0.036) (2.740)
ACQWNSP + -0.050 -0.202 0.01 l - -8.968
(0.132) (0.838) (0.1 14) (8.948)

 

 

 

 

 

 

 

 

 

86

 

 

 

Sample Size - 283 292 248 274

 

PseudozlAdj. 0.4601 0.4790 0.4606 0.5844
R

 

p-values of the 0.1103 0.7802 0.4887 0.7744
joint significance
test of
governance
variables

 

 

 

 

 

 

 

 

 

‘3’!“ significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses. The standard errors for the SPREAD sample are adjusted for serial
correlation and heteroskedasticity per Rogers (1993). For ease of exposition, the intercept of the
SPREAD model and the coefficients on bond issue characteristics (MA TYRS, ISSUESIZE, CALLABLE
and SUBORD) for the yield spread model are not shown. All variables are as defined in Appendix C.

87

Table 6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specification and Sensitivity Tests
Panel B: Sensitivity analysis using Brown and Caylor’s (2004) GOV_Score as an alternative corporate
g9_vernance prox
Dependent RATING SPREAD
Variable
Year 2001 Year 2002 Year 2003
Expected Coefficient Coefficient Coefficient Expected Coefficient
Sign (Standard (Standard (Standard Sign (Standard
Error) Error) Error) ' Error)
0P770NS ? 4.874“ 8145"" -5.789"'“ ? 413.246""
(0.916) @741) (1.837) (119.413)
ACCQUALITY + 0.132 0.166 0.448 - 494.529“
(1.163) (0.682) (0.709 (102.564)
STDROA - -5.398""" -6.232* -2.204 + 299.079
(2.525) (3.665) (3.425) , (262.823)
STDCFO - -4.397 -4.723 -2.207 + 699.298
(3.904) (4.298) (4.579) (486.936)
STDRET - -17.628"“ -20.493“"‘ -19.790“" 4» 724.064‘""‘
(3.014) (3.288) (3.375) (289.199
R04 + 0.953 3.799“ 2.048 - -225.052
(1.784) (2.247) (2.024) (191.898)
CASHFLOW + 7.474‘" 10.156*" 8609"" - -247.925
(2.248) (2.585) (2.588) (210.831)
LEVERAGE - 2567““ 4.710“ -1.871"'" + 81.039
(0.714) (0.681) @728) (54.646)
SIZE + 0438*” 0613"” 0560*" - -33.328"""
(0.086L 40.092) (0.100) (6.905)
BETA - 0567‘" -0.055 «0.181 + -6.169
(0.196) (0.297) (0.277) (17.407)
CAP_INTEN + -0.401 0722"" 0689“ - -4.585
(0.260) (0.264) (0.290) (20.760)
INT_COV + 0.004 0.003 —0.001 - —0.473
(0.005) (0.004) (0.003) (0.462)
32” - -0.819‘"" -0.054 -0.681 + 51.328
(0.328) (0.224) (0.448) (33.145)
GOV_SCORE - 0.024 0.022 0.007 - 3.216“
(0.021) @122) (0.024) (1.757)
Sample Size 228 226 192 218
1’88!ng 0.4555 0.4980 . 0.4575 0.5805
R

 

 

 

 

 

 

 

 

‘,“,"" significant at the 0.1, 0.05 and 0.01 levels (two-tailed)
Standard errors are in parentheses. The standard errors for the SPREAD sample are adjusted for serial
correlation and heteroskedasticity per Rogers (1993). For ease of exposition, the intercept of the SPREAD
model and the coefficients on bond issue characteristics (MA TYRS, ISSUESIZE, CALLABLE and
SUBORD) for the yield spread model are not shown. All variables are as defined in Appendix C.

88

 

Table 6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specification and Sensitivity Tests
Panel F: Sensitivity tests using excessive ESO compensation as an alternative corporate givernanee proxy
Dependent RATING SPREAD
Variable
Year 2001 Year 2002 Year 2003
Expected Coefficient Coefficient Coefficient Expected Coemcient
Sign (Standard (Standard (Standard Sign (Standard
Error) Error) Error) Error)
OPUONS ? -2.950"'“ -3.090 6.180“ ? 327.782’
(1.150) (3.019) (2.185) (168.055)
ACCQUALITY + 1.441” 0.653 0.510 - 63.844”
‘ (0.656) (0.473) (0.505) (30.039)
STDROA - -3.419" -2.622 -3.178 + 208.810
(1.916) (2.670) (2.488) (221.390)
STDCFO - -2.486 -0.306 2.325 + 508.914‘
(2.874) (3.134) (2.921) (305.283)
SIDRET - -15.415"“ -l8.676”'" ~19.469“"' + 554.610“I
(2.609) (2.690) (2.800) (240.756)
804 + 3.382‘” 3.093" 2.977 — -136.901
(1.25Q (1.394) (1.833) (154.102)
CASPU’LOW + 5.146‘“ 7.226‘“ 6922““ - 415.152"
(1.641) (1.745) (2.028) (195.887)
LEVERAGE - 2481"” -1 .416“ -l .472“ + 104.632“
‘ (0.592) (0.637) (0.672) (58.104)
SIZE + 0.534‘" 0690“" 0598"" - -34.448”"
(0.070) (0.077) (0.085) (5.078)
BETA - 0362” 0.069 -0.092 + -7.808
(0.163) (0.240) (0.228) (15.734)
CAP_INTEN + -0.492“ 0609"” -0.416‘ - 8.022
(0.218) (0.215) 40.237) (18.100)
INT.C0V + 0.003 0.056 0.089 - 0.161
(+100) (0.003) (0.315) (0.243) (0.395)
52” - -0.345 0.244 -0.265 + 55.075“
(0.292) (0.208) (0.371) (24.423)
“€593.30 - 0652‘“ -0.300 0526“ + -25.076"'
(0.235) 40.213) (0.236) (15.116)
EXCESS_EYO‘ + 8.148 -2.288 11.116 - 151.079
0pm” (6.718) (7.955) (7.293) (172.115)
Sample Size 299 299 261 309
Psetgdo 0.429 0.4613 0.4562 0.5614
R

 

 

 

*,","” significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses. The standard errors for the SPREAD sample are adjusted for serial
correlation and heteroskedasticity per Rogers (1993). For ease of exposition, the intercept of the SPREAD
model and the coefficients on bond issue characteristics (MA TYRS, ISSUESIZE, CALLABLE and
SUBORD) for the yield spread model are not shown. All variables are as defined in Appendix C.

89

Table 6

Specification and Sensitivity Tests

Panel G: Sensitivity tests partitioning OPTIONS into the grant date fair value and post grant date fair

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

value change
Dependent Variable RATING SPREAD
Expected Sign Coefficient Expected Sign Coefficient
(Standard Error) (Standard
Error)
GRANTJALUE ? 2534"" ? 764.321""
(0.718) (227.890)
VALUE_CHG 2 . 4,9347!" 7 231.686‘
(0.733) (127.506)
ACCQUALITY + 0.369 - -72.31 1"
(0.250) (30.08Q
STDROA - 3495"" + 215.033
(1.219) (218.61)
SIDCFO - 0.163 + 534.824‘
(1.600) (304.053)
STDRET - -l6.387”‘ + 440.262‘
(1.373) (248.447)
ROA 1 + 2653‘“ - -142.934
L (0760) (150.95)
CASHFLOW + 6294*" - 389.542"
((1.980) (192.367)
LEVERAGE - -1.959"""' + 89055"
(0.330) (53.691)
SIZE + 0.581‘“ - -37.669"'“
(0.041) (5.5693
BETA - 0246” + -6.793
(0.105) (15-192)
CAP_INTEN + 0417"" - 9.521
(0.123) (18.173)
M_COV(+100) + 0.045 - -2.846
(0.144) (40.071)
82M - 0028 + 52.473"
(0.127) Q5515)
Sample Size 913 309
Adj/Pgeudo 0.4333 0.5662
R

 

 

 

 

 

 

‘,”,‘" significant at the 0.1, 0.05 and 0.01 levels (two-tailed)

Standard errors are in parentheses. The standard errors for the SPREAD sample are adjusted for serial
correlation and heteroskedasticity per Rogers (1993). For ease of exposition, the intercept of the
SPREAD model and the coefficients on bond issue characteristics (MA TYRS, ISSUESIZE, CALLABLE
and SUBORD) for the yield spread model are not shown. All variables are as defined in Appendix C.

90

 

REFERENCES

Aboody, D. 1996. Market Valuation of Employee Stock Options. Journal of Accounting
and Economics 22: 357-391.

Aboody, D., M. Barth, and R. Kasznik. 2004a. SFAS No. 123 Stock-Based
Compensation Expense and Equity Market Values. The Accounting Review 79
(2): 251-275.

Aboody, D., M. Barth, and R. Kasznik. 2004b. Do Firms Understate Stock-Based
Compensation Expense Disclosed under SF AS 123? Working Paper, University
of California at Los Angles and Stanford University.

American Accounting Association (AAA) Financial Accounting Standards Committee ' '
(FASC). 1993. Response to the FASB Discussion Memorandum "Distinguishing
between Liability and Quality Instruments and Accounting for Instruments with
Characteristics of Both". Accounting Horizons 7 (3): 105-1 13.

American Accounting Association (AAA) Financial Accounting Standards Committee
(FASC). 2001. Evaluation of the FASB's Proposed Accounting for Financial
Instruments With Characteristics of Liabilities, Equity, or Both. Accounting
Horizons 15 (3): 387-400.

American Accounting Association (AAA) Financial Accounting Standards Committee
(FASC). 2004. Evaluation of the IASB's Proposed Accounting and Disclosure
Requirements for Share-Based Payment. Accounting Horizons 18 (1): 65-76.

Accounting Principles Board (APB). 1972. APB Opinion NO. 25 :Accounting for Stock
Issued to employees. Norwalk, CT: APB.

Anderson, R., S. Mansi, and D. Reeb. 2004. Board Characteristics, Accounting Report
Integrity, and the Cost of Debt. Working Paper.

Ashbaugh, 11., D. Collins, and R. LaFond. 2004. The Effects of Corporate Governance on
Firms' Credit Ratings. Working Paper, University Of Wisconsin-Madison.

Balsam, S. 1994. Extending the method of accounting for stock appreciation rights to
employee stock options. Accounting Horizons 8 (4): 52-60.

Bartov, E., P. Mohanram, and D. Nissim. 2004. Stock Option Expense, Forward-Looking
Information, and Implied Volatilities of Traded Options. Working Paper, New
York University and Columbia Business School.

Bartov, E. and P. Mohanram. 2004. Private Information, Earnings Manipulations, and
Executive Stock-Option Exercises. The Accounting Review 79 (4): 889-920.

91

Bell, T., W. Landsman, B. Miller, and S. Yeh. 2002. The Valuation Implications of
Employee Stock Option Accounting for Profitable Computer Software Firms. The
Accounting Review 77 (4): 971-996.

Bens, D., V. Nagar, and M. Wong. 2002. Real Investment Implications of Employee
Stock Option Exercises. Journal Of Accounting Research 40 (2): 359-393.

Bens, D., V. Nagar, D. Skinner, and M. Wong. 2003. Employee Stock Options, EPS
Dilution, and Stock Repurchases. Journal of Accounting and Economics 36: 51-
90.

Bhojraj, S. and P. Sengupta. 2003. Effect of Corporate Governance and Bond Ratings and
Yields: the Role of Institutional Investors and the Outside directors. The Journal
of Business 76: 455-475.

Black, F., and M. Scholes. 1973. The pricing of options and corporate liabilities. Journal
Of Political Economy 81 (3): 637-654.

Bowman, R. 1979. The Theoretical Relationship between Systematic Risk and Financial
(Accounting) Variables. Journal of Finance 34 (3): 617-630.

Bowman, R. 1980. The Debt Equivalence of Leases: An Empirical Investigation. The
Accounting Review 55 (2): 237-253.

Brown, L. and M. Caylor. 2004. Corporate Governance and Firm Performance. Working
Paper, Georgia State University.

Cheng, Q., P. Frischrnann, and T. Warfield. 2003. The Market Perception of Corporate
Claims. Research in Accounting Regulation 16: 3-28.

Cheng, Q. and T. Warfield. 2005. Equity Incentives and Earnings Management. The
Accounting Review 80 (2):441-476.

Cohen, R., B. Hall, and L. Viceira. 2000. Do Executive Stock Options Encourage Risk
Taking? Working Paper, Harvard University.

Coles, J ., N. Daniel, and L. Naveen. 2003. Executive Compensation and Managerial
Risk-Taking. Working Paper, Arizona State University and Georgia State
University.

Core, J. and W. Guay. 1999. The Use of Equity Grants to Manage Optimal Equity
Incentive Levels. Journal Of Accounting and Economics 28: 151-184.

Core, J. and W. Guay. 2001. Stock Option Plans for Non-Executive Employees. Journal
of Financial Economics 61: 253-287.

92

Core, J ., and W. Guay. 2002. Estimating the value of stock option portfolios and their
sensitivities to price and volatility. Journal of Accounting Research 40(3): 613-
630.

Core, 1., R. Holthausen and D. Larcker. 1999. Corporate Governance, Chief Executive

Officer Compensation, and Firm Performance. Journal of Financial Economics
51: 371-406.

Czamitzki, D. and K. Krafi. 2004. Are Credit Ratings Valuable Information. Working
Paper.

Dechow, P. and l. Dichev. 2002. The Quality of Accruals and Earnings: The Role of
Accrual Estimation Errors. The Accounting Review 77 (supplement): 35-59..

DeFusco, R., R. Johnson, and T. Zorn. 1990. The Effect of Executive Stock Option Plans
on Stockholders and Bondholders. Journal Of Finance 45 (2): 617-627.

Demsetz, H. and K. Lehn 1985. The Structure of Corporate Ownership: Causes and
Consequences. Journal of Political Economy: 1 155-1 177.

Dhaliwal, D. 1986. Measurement of Financial Leverage in the Presence of Unfunded
Pension Obligations. The Accounting Review 61 (4): 651-661.

Dittrnar, A. 2000. Why Do Firms Repurchase Stock? The Journal Of Business 73 (3):
331-355. . . .

Efendi, J., A. Srivastava, and E. Swanson. 2004. Why Do Corporate Managers Misstate
Financial Statements? The Role of Option Compensation, Corporate Governance,
and Other Factors. Working Paper, Texas A&M University.

Eisenhardt, KM. 1989. Agency Theory: An Assessment and Review. Academy Of
Management Review 14: 57-74.

Elliot, W., A. Prevost, and R. Rao. 2004. The Announcement Impact of Seasoned Equity
Offerings on Bondholder Wealth. Working Paper.

Engel, E., M. Erickson, and E. Maydew. 1999. Debt-equity hybrid securities. Journal of
Accounting Research 37 (Autumn): 249-274.

Erickson, M., M. Hanlon, and E. Maydew. 2004. Is There a Link between Executive
Compensation and Accounting Fraud? Working Paper.

Financial Accounting Standards Board. 1985. Statement of Financial Accounting
Concepts NO. 6: Elements of financial Statements. Norwalk, CT: FASB.

Financial Accounting Standards Board. 1990. Discussion Memorandum: Distinguishing
between Liability and Equity Instruments and Accounting for Instruments with
Characteristics ofBoth. Norwalk, CT: FASB.

93

Financial Accounting Standards Board. 1995 Statement of Financial Accounting
Standards NO. 123: Accounting for Stock-Based Compensation. Norwalk, CT:
FASB.

Financial Accounting Standards Board. 2004. Statement of Financial Accounting
Standards NO. 123R: Share-Based Payment. Norwalk, CT: F ASB.

Financial Accounting Standards Board. 2004. Statement Of Financial Accounting
Standards NO. I 5 0 Accounting for Certain Financial Instrument: with
Characteristics of Both Liabilities and Equity. Norwalk, CT. FASB.

Francis, J., R. LaFond, P. Olsson, and K. SChipper. 2005. The Market Pricing of Accruals
Quality. Journal ofAccounting and Economics 39 (2): 295- 327.

Gabbi, G. and A. Sironi. 2002. Which Factors Affect Corporate Bond Pricing? Empirical
Evidence From Eurobonds Primary Market Spreads. Working Paper.

Galai, D. and M. Schneller. 1978. Pricing of Warrants and the Value of the Firm. Journal
Of Finance 33(5): 1333-1342.

Gompers, P., J. Ishii, and A. Metrick. 2003. “Corporate Governance and Equity Prices”.
Quarterly Journal of Economics 1 18: 107-155.

Graham, J. and C. Harvey. 2001. The Theory and Practice of Corporate Finance:
Evidence from the Field. Journal Of Financial Economics 60 (2): 187-243.

Guay, W. 1999. The Sensitivity of CEO Wealth to Equity Risk: an Analysis of the
Magnitude and Determinants. Journal Of Financial Economics 53: 43-71.

Hall, B and K. Murphy. 2002. Stock Options for Undiversified Executives. Journal of
Accounting and Economics 33: 3-42.

Hall, B. and K. Murphy. 2003. The Trouble with Stock Options. Journal Of Economic
Perspectives 17 (3): 49-70.

Harnada, R. 1972. The Effect of the Firm's Capital Structure on the Systematic Risk of
Common Stocks. Journal Of Finance: 43 5-452.

Hanlon, M., S. Rajgopal, and T. Shevlin. 2003. Are Executive Stock Options Associated
with Future Earnings? Journal of Accounting and Economics 36: 3-43.

Hanlon, M., S. Rajgopal, and T. Shevlin. 2004. Large Sample Evidence on the Relation
between Stock Option Compensation and Risk Taking. Working Paper,
University of Washington.

Hodder, L., W. Mayew, M. McAnally, and C. Weaver. 2004. Employee Stock Option
Fair-Value Estimates: Do Managerial Discretion and Incentives Explain
Accuracy? Working Paper.

94

Hopkins, P. 1996. The effect of financial statement classification of hybrid financial
instruments on financial analysts’ stock price judgments. Journal of Accounting
Research 34 (Supplement): 33-50.

Ittner, C., R. Lambert, and D. Larcker. 2003. The Structure and Performance
Consequences of Equity Grants to Employees of New Economy F inns. Journal of
Accounting and Economics 34: 89-127.

Jensen, M. and W. Meckling. 1976. Theory of the Firm: Managerial Behavior, Agency
Costs and Ownership Structure. Journal of Financial Economics 3: 305-60.

Jensen, M and K. Murphy. 2004. Remuneration: Where We've Been, How We Got to
Here, What Are the Problems, and How to F ix Them. Working Paper, European
Corporate Governance Institute.

Johnston-Wilson, D. 2004. Managing Stock Option Expense: the Manipulation of
Option-Pricing Model Assumptions. Working Paper, Colorado State University.

Kahle, K. 2002. When a Buyback Isn't a Buyback: Open Market Repurchases and
Employee Options. Journal of Financial Economics 63 (2): 235-261.

Kaplan, R. and G. Urwitz. 1979. Statistical Models of Bond Ratings: a Methodological
Inquiry. Journal of Business 52 (2): 231-261.

Katz, S. 1974. The Price Adjustment Process of Bond to Rating Reclassifications: a Test
of Bond Market Efficiency. Journal of Finance 29 (2): 551-559.

Keenan, S., D. Hamilton and A. Berthault. 2000. Historical Default Rates of Corporate
Bond Issuers, 1920-1999. Moody’s Investors Service, Special Comment.
Available at: http://www.moodyskmv.com/research/whitepaper/52453.pdf.

Khurana, I. and K. Rarnan. 2003. Are Fundamentals Priced in the Bond Market?
Contemporary Accounting Research 20 (3): 465-494.

Kimmel, P. and T. Warfield. 1995. The Usefulness of Hybrid Security Classifications:
Evidence From Redeemable Preferred Stock. The Accounting Review 70 (1): 151-
167.

Kirschenheiter, M., R. Mathur, and J. Thomas. 2004. Accounting for Employee Stock
Options. Accounting Horizons 18 (2): 135-156.

Klock, M., S. Mansi, and W. Maxwell. 2004. Does Corporate Governance Matter to
Bondholders. Working Paper.

Kroumova, M. and J. Sesil. 2003. Intellectual Capital, Monitoring and Risk: What
Predicts the Adoption of Broad-Based Employee Stock Options? Working Paper.

95

Lambert, R., D. Larcker, and R. Verrecchia. 1991. Portfolio Considerations in Valuing
Executive Compensation. Journal Of Accounting Research 29 (1): 129-149.

Landsman, W, K. Peasnell, P. Pope, and S. Yeh. 2004. The Value Relevance of
Alternative Methods of Accounting for Employee Stock Options. Working Paper.

Larcker, D. and T. Rusticus. 2005. On the Use of Instrumental Variables in Accounting
Research. Working Paper, University of Pennsylvania.

Li, F. 2004. Executive equity holdings and merger and acquisition efficiency. Working
Paper, University of Michigan. '

Li, H. 2002. Employee Stock Options, Residual Income Valuation and Stock Price
Reaction to SF AS 123 Footnote Disclosures. Working Paper, the University of
Iowa.

Linsmeier, T, C. Shakespeare and T. Sougiannis. 2004. Liability/Equity classifications
and Shareholder Valuation. Working Paper, Michigan State University.

Lipe, R. 2001. Lease Accounting Research and the G4+1 Proposal. Accounting Horizons
15 (3): 299-310.

Long, J. 1997. Regression Models for Categorical and Limited Dependent Variables,
Sage Publications, Thousand Oaks.

Longstaff, F., S. Mithal, and E. Neis. 2004. Corporate Yield Spreads: Default Risk or
Liquidity? New Evidence From the Credit-Default Swap Market. Working Paper,
National Bureau of Economic Research.

Mandelker, G. and S. Rhee. 1984. The Impact of the Degrees of Operating and Financial
Leverage on Systematic Risk of Common Stocks. Journal of Financial and
Quantitative Analysis 19 (1): 45-57.

Matsunaga, S. 1995. The Effects of Financial Reporting Costs on the Use of Employee
Stock Options. The Accounting Review 70 (1): 1-26.

McNichols, M. 2002. Discussion of ‘The quality of accruals and earnings: the role of
accrual estimation errors.’ The Accounting Review 77 (Supplement): 61-69.

Meulbroek, L. 2001. The Efficiency of Equity-Linked compensation: Understanding the
full cost of Awarding Executive Stock Options. Financial Management 30: 5-30.

Merton, R., 1973. Theory of rational option pricing. Bell Journal of Economics and
Management Science 4: 141 -1 83.

Minton, B. and C. Schrand. 1999. The Impact of Cash Flow Volatility on Discretionary
Investment and the Costs of Debt and Equity Financing. Journal of Financial
Economics 54: 423-460.

96

Nellans, M. 2004. Accounting for the Real Cost of Stock Options. Morningstar.com.
2004/9/15.

Oyer, P. and S. Schaefer. 2005. Why Do Some Firms Give Stock Options to All
Employees? An Empirical Examination of Alternative Theory. Journal Of
Financial Economics 76 (1): 99-133.

Ohlson, J. and S. Penman. 2005. Debt vs. Equity: Accounting for Claims Contingent on
Firms’ Common Stock Performance with Particular Attention to Employee
Compensation Options. Columbia Business School, CEASA, White Paper No. 1.

Penman, S. 2003. The quality of financial statements: Perspectives from the recent stock
market. Accounting Horizons 17: 77-96.

Raj gopal, S. and T. Shevlin. 2002. Empirical Evidence on the Relation between Stock
Option Compensation and Risk Taking. Journal Of Accounting and Economics
33: 145-171.

Rebeiz, K. 2003. Employees' stock options fi'om a shareholder's perspective. Journal of
the Academy Of Business and Economics 3 (l).

Rivers, D., and Q. Vuong 1988. Limited Information Estimators and Exogeneity Tests for
Simultaneous Probit Models. Journal of Econometrics 39: 347-366.

Rogers, W. H. 1993. sg17: Regression standard errors in clustered samples. Stata
Technical Bulletin 13: 19-23.

Rubinstein, M. 1973. A Mean-Variance Synthesis of Corporate Finance Theory. Journal
OfFinance 28(1): 167-181.

Rubinstein, M. 1995. On the Accounting Valuation of Employee Stock Options. Journal
Of Derivatives Fall: 8-24.

Sengupta, P. 1998. Corporate Disclosure Quality and the Cost of Debt. The Accounting
Review 73 (4): 459-474.

Shi, C. 2003. On the Trade-Off between the Future Benefits and Riskiness of R&D: A
Bondholders’ Perspective. Journal of Accounting and Economics 35: 227-254.

Smith, C., and R. Watts. 1992. The Investment Opportunity Set and Corporate Financing,
Dividend, and Compensation Policies. Journal of Financial Economics 7: 117-61.

Standard and Poor’s. 2005. Press Release: 2004 Buybacks Soar to Record Levels, Says
S&P. Available at http://www2.standardandpoors.com/spf/pdf/index/O40805.pdf.

Stephens, C. and M. Weisbach. 1998. Actual Share Reacquisitions in Open-Market
Repurchase Programs. Journal of Finance 53: 313-341.

97

Stock, J., J. Wright and M. Yogo. 2002. A Survey of Weak Instruments and Weak
Identification in Generalized Method of Moments. Journal Of Business &
Economics Statistics 20: 518-529.

Watts, R. 2003. Conservatism in Accounting - Part I: Explanations and Implications.
Accounting Horizons 17 (3): 207-221.

Wei, L. 2004. Tracking the Numbers/Outside Audit: In Expensing Options, Devil Is in
the Details -- As FASB Writes Draft Rule Treating Them as Equities, Liabilities
Fans Press Case. The Wall Street Journal 2004/8/ 19.

Weisbenner, S. 2000. Corporate Share Repurchases in the 19908: What Role Do Stock
Options Play. Working Paper, Federal Reserve Board.

Wooldridge, J. 2002. Econometric Analysis OfCross Section and Panel Data. The MIT
Press, Cambridge Massachusetts.

Yermack, D. 1995. Do Corporations Award CEO Stock Options Effectively? Journal Of
Financial Economics 39: 237-269.

Ziebart, D. and S. Reiter. 1992. Bond Ratings, Bond Yields and Financial Information.
Contemporary Accounting Research 9: 252-282.

98

1'

 

11:11)(liggilejylujggfu