Dividends on Unearned Shares and Corporate Payout Policy:

An Analysis of Dividend Equivalent Rights

Zi (Tingting) Jia

Don M. Chance

Draft: February 17, 2016

We investigate a little-known executive compensation device called dividend equivalent rights (DERs). DERs entitle an executive to receive dividends on unearned performance-based shares and options. Policies permitting these payments, known as dividend equivalents (DEs), are found in about one-fifth of S&P 500 firms and about 10% actually make such payments. While investors react negatively to announcements of DER policies or DE payments, DERs can, however, benefit shareholders by inducing a company to disgorge unproductive cash. If a firm allows DE payment, it is four times more likely to be a dividend payer and for firms already paying dividends, dividend payments are larger if DE payments are allowed. In the years prior to adoption of DER policies and initiation of DE payments, DER-policy adopters and DE payers accumulate an increasing amount of excess cash, a pattern that tends to cease following DER-policy adoption or payment of DEs. Thus, DERs can have the beneficial effect of encouraging a company to stop accumulating unproductive cash. JEL Classifications: G35, G34 Key words: dividend equivalent, dividend equivalent right, executive compensation, corporate governance, dividend policy. *The first author is Assistant Professor of Finance at the University of Arkansas at Little Rock. The second author is James C. Flores Endowed Chair for MBA Studies, Department of Finance, BEC 2900, Louisiana State University, Baton Rouge, LA 70803; zxjia@ualr.edu; dchance@lsu.edu. They would like to thank the faculty and students of the finance department research seminars at LSU, York, and Kent State and participants at the 2015 Vietnam International Conference in Finance, the 2015 World Finance Conference (Buenos Aires), the FMA and FMA-Europe (Maastricht) conferences. The authors would especially like to thank Wei-ling Song, Clifford Stephens, Brian Marx, Joseph Mason, Soku Byoun, Rehman Mian, Joe Legoria, Dan Bradley, Dmitry Shapiro, Jean McGuire, Tony Via, Steve Dennis, Kelly Pace, Melanie Cao, Mark Kamstra, Yisong Tian, Ji-Chai Lin, Shrikant Jategaonkar, and Tianxiang Xu for valuable comments and insights.

1

Dividends on Unearned Shares and Corporate Payout Policy:

An Analysis of Dividend Equivalent Rights

1. Introduction

In May of 2006, The Wall Street Journal reported that GE CEO Jeffrey Immelt had received more

than $1 million in dividends on shares of stock that he did not own (Thurm 2006). These payments, known

as dividend equivalents, were based on restricted stock and performance shares on which he had a claim,

but which were, nonetheless, subject to the achievement of future goals. Several months later, possibly in

response to the article, GE banned these types of payments. Nonetheless, knowledge of the payments

galvanized shareholders into proposing a ban, which was included in GE’s proxy statement of 2006 (GE

2006) even though the recommendation in the proposal was already in effect.

This paper examines this little-known form of executive compensation called the dividend

equivalent right (DER). DERs entitle executives to receive dividends on shares covered by their

performance-based equity awards, which are shares they may not own and, in fact, may never own. The

DER is the policy permitting such payments. The dividends themselves are referred to as dividend

equivalents (DEs) or dividend equivalent payments. We refer to the underlying shares that are not yet

owned by the executives as unearned shares. Because dividends, by definition, are payments to the owners

of a corporation, paying dividends on shares not owned seems counterintuitive. If the executive does not

own the shares, one could then reasonably question how the executive would be entitled to receive them

and how the firm could justify making such payments. Serious questions obviously arise as to whether these

payments are opaque forms of compensation.

Not only does the executive receive dividends on shares he does not own, but he is also likely to

have the ability to influence the per share dividend declaration. Obviously this arrangement poses a conflict

of interest, though one no greater than that posed if the executive actually does own the shares outright.

The unique aspect of dividend equivalent rights, however, is that the executive does not actually own the

shares. He may own them at some time in the future if performance conditions are met, but clearly there is

a possibility that he may not ever own the shares. Dividends on shares owned by executives are relatively

transparent. DE payments are often poorly disclosed and corporate policies regarding DERs in most cases

are not transparent. This instrument is unquestionably not widely known. There has been essentially no

academic research on DERs, and they are not commonly mentioned in trade publications.

These issues seem to suggest that DERs could be means of helping executives extract rents from

shareholders without much scrutiny. Yet, the potential agency problem created by DERs is only the

negative side of the story. The positive side is that DERs have the potential to reduce an agency problem.

2

For example, executive stock options are widely used as a form of compensation and incentives but are

subject to the criticism that they potentially discourage firms that have insufficient investment opportunities

for the cash they hold from paying dividends or at least paying dividends of the optimal size. Dividends

are well-known to reduce the value of a call option, and empirical evidence demonstrates that executives

holding stock options tend to keep dividends at what may be sub-optimal levels (Fenn and Liang 2001;

Lambert, Lanen and Larcker 1989; Verizon 2003). When dividends are restrained, executives may

sometimes hoard cash or engage in empire-building in the form of acquiring assets by investing in inferior

or, at best, value-neutral projects. Because dividend equivalents can be paid to executives only when

common dividends are paid to shareholders, DERs can encourage firms with excessive amounts of cash to

disgorge some of that cash by initiating dividends or by paying higher dividends to shareholders.

DERs are granted on the underlying shares of the incentive equity awards granted to executives.

These shares are either time-vested, performance-vested, or a combination of both. DERs granted on time-

vested shares such as time-vested restricted stock or units should not necessarily be viewed in the same

manner as those on unearned shares. Time-vested restricted stock or units are simply shares, or phantom

shares, that will eventually be owned if the holder stays employed by the firm. Thus, while there is the

potential that the executive will leave the firm and forfeit the shares or be fired, the condition that must be

met is not terribly onerous. So the shares are for all intents and purposes owned by the executive or will

almost surely become owned by the executive someday. Thus, they are more or less simply illiquid shares.

DERs granted on performance–based shares, however, are a different story altogether. Shares that

are subject to performance hurdles are earned by an executive only when specified results have been

achieved. Before these goals have been met, the shares are unearned shares. One form of performance-

based shares mainly includes unvested performance-vested restricted stock or units and unearned

performance shares. In addition, shares underlying stock options can be covered by dividend rights. Clearly

these shares may never be earned.

One may suspect that DEs paid on unearned shares are rare cases featuring little-known firms

paying their CEOs under the radar screen. In fact, granting DERs on CEOs’ unearned shares is not rare at

all, even for large and well-known U.S. firms. This study finds that about 10 percent of S&P500 firms

make such DE payments each year and more than 20 percent of S&P500 firms allow for this practice.

The primary objectives of this study are to identify the factors associated with the adoption of DER

policies and the payment of DEs, to determine whether shareholders approve or disapprove of DERs, to

examine whether the quality of corporate governance is related to the existence of DERs, to identify if there

is a relationship between stock returns and DE payments, to determine if DERs have the potential to induce

non-dividend paying firms to pay dividends and dividend-paying firms to pay higher dividends, and to

address whether DERs can induce firms to disgorge unproductive cash.

3

To our knowledge, this is the first empirical study on the dividend equivalent income that

executives earn from their unearned equity awards. The approach we take, however, is also somewhat

different from traditional studies. We dispose of the commonly used categorization of the equity awards

of executives and instead classify shares as either earned or unearned. This classification is based primarily

on the risk associated with the awards. Second, we contribute to the literature on corporate disclosures by

examining the question of how investors feel about this form of compensation. Third, this paper contributes

to the discussion of corporate dividend policy. We show that in addition to the commonly acknowledged

factors that are associated with the payment of dividends, dividend equivalents can influence the choice of

dividend payout levels and potentially make dividend payouts rise to more appropriate levels, given the

cash holdings and investment opportunities.

The remainder of the paper proceeds as follows. Section 2 reviews the literature, Section 3

describes the hypotheses and the sample, Section 4 discusses the primary results, including endogeneity

issues and robustness tests. Section 5 examines the question of whether dividend equivalents result in the

disgorgement of unproductive cash. Section 6 presents our summary and conclusions.

2. Related Literature

There has been little research on DERs or DE payments, but of course, considerable research on

executive compensation and incentives. The origin of that research is undoubtedly the classic Jensen and

Meckling (1976) model. The objective of an optimal compensation contract is to reduce agency costs by

aligning shareholder interests with executive interests (Murphy 1999). Equity-based compensation is found

to be a natural solution to reducing agency costs. For example, Jensen and Murphy (1990) find that stock

ownership is an effective means of providing incentives to CEOs, and Matsunaga and Park (2001) show

that performance bonuses encourage mangers to meet analyst earnings forecasts.

Dividend equivalents could be used in optimal contracts if they can provide incentives, of which

increasing dividend payouts may be the most important one. Empirical evidence has shown that the absence

of dividend protection on executive stock options is associated with a decrease in corporate dividends

(Arnold and Gillenkirch 2005; Lambert, Lanen and Larcker 1989). Interestingly, shareholders may

recognize this problem. For example, shareholders of Verizon expressed their concerns that the absence of

dividend protection on options “may discourage executives from increasing dividends” (Verizon 2003). As

noted above, DERs could help address this problem.

In contrast to the optimal contracting theory, the managerial power theory questions the assumption

that the board of directors acts in the best interests of shareholders. In fact, equity-based compensation

may well encourage managerial rent seeking rather than reduce it, examples of which are found in Yermack

(1997) (grant timing), Aboody and Kasznik (2000) (news releases), Callaghan, Saly and Subramaniam

4

(2004) and Chance, Kumar and Todd (2000) (option re-pricing), and Brooks, Chance, and Cline (2012)

(option exercise timing).

Although many public firms pay dividends and their CEOs typically hold large amounts of equity,

studies on the dividend income of executives are limited and the results are mixed. Zhang (2012) finds that

dividend payments on CEOs’ unvested restricted stocks can provide dividend incentives, and the market

reacts positively toward such news. In contrast, Minnick and Rosenthal (2010) find that firms that pay

dividends on CEOs’ restricted stock tend to have poor corporate governance and show inferior operating

performance. This evidence supports the managerial power theory that managers are able to influence their

own compensation packages and extract rents (Bebchuk and Fried 2004). Elsila (2013) documents a

substitution effect between CEOs’ cash compensation and dividend income in that the absence of dividend

protection on stock options tends to be made up by a premium in cash compensation. Kahle (2002) finds

that the possession of stock options motivates executives to favor share repurchase over dividends. Thus,

there is strong evidence that corporate dividend policy and managerial incentives are connected.

Although this paper also examines many aspects of CEO dividend income, it has two major

characteristics that differentiate it from other related studies. For one, it focuses on dividend equivalent

rights, whereas other studies examine dividend income in general. The second is that we do not classify

the underlying shares in the traditional way, such as restricted shares, performance shares, or options.

Instead, we categorize the shares covered by equity awards by whether CEOs need to achieve performance

targets to earn them. We distinguish unearned shares from earned shares, the latter of which are subject

only to time-based restrictions.1 This dichotomy is important in that the distinction is largely one of the

substantially different risks.

The choice of classifying equity awards in this paper also has a significant impact on the sample

structure and the interpretation of the empirical results. This categorization is inspired by the observation

that both firms and shareholders are concerned about CEOs’ rights associated with their unearned shares

For example, we find that when explaining shareholders’ rights on equity awards, firms classify shares as

either earned or unearned shares (Alcoa 2009; Mead WestVaco Coporation 2011; Fedex 2010). As noted

above, an excellent example of shareholders’ concerns over CEOs’ unearned shares is Shareowner Proposal

1Technically all performance based restrictions include time-based restrictions. Performance goals are set for a period of time. Thus, an executive might need to achieve a 10% increase in sales over the following fiscal year. Except at the last instant before that fiscal year ends, the performance goal has a time restriction. What we refer to as pure time restrictions, however, are merely related to the passage of time and not to any performance objective. Glagau (2013) notes that a 2010 survey of the incentive awards of almost 600 companies shows that about 82% paid dividends on restricted stock and 64% paid dividends on restricted stock units. Because our study is confined to dividends paid on performance shares and units, these percentages are much higher than what we found. That study includes shares and units that are earned merely by virtue of staying employed. More details about performance based stock are contained in Appendix A.

5

No.5 in GE’s 2005 proxy statement, in which the GE’s shareholders urged the firm to eliminate dividend

equivalents on CEOs’ unearned shares (GE 2006).

The second factor distinguishing this study from others is that the classification method used in this

paper allows us to include more types of executive compensation tools in the sample, especially those that

are widely used in firms but somewhat less discussed in the literature, such as performance stock or units,

phantom stock or units, and restricted stock units.

This strong bond of this classification of earned and unearned shares to performance provides the

convenience of relating the results in this paper to studies on the effectiveness of performance-based

compensation. Performance-based pay is distinct from its close relative, equity-based pay. Both of these

forms of incentives can tie a manager’s performance with his/her compensation. But they have a major

difference: Equity-based pay refers to compensation arrangements in which payment is made in equity-

based instruments such as shares of stock or stock options. It connects the manager’s compensation with

his/her performance through the stock price. Performance-based pay refers to compensation arrangements

that involve specified performance goals over a performance period that follows the award grant date. The

award is granted with a specified amount. The earned amount of the award is determined at the end of the

performance measurement period. The manager’s realized compensation is contingent on the achievement

of the specified performance goals and thus may be different from the granted amount. Performance-based

awards have been embraced by shareholders and policymakers alike, and they provide a useful distinction

for classification purposes. Performance-based equity awards and compensation awards with performance-

based vesting provision account for an increasing portion of executive compensation (Bettis, Bizjak, Coles

and Kalpathy 2010; Deloitte 2005; Kanter and Frederic W. Cook & Co. 2005). In the early 2000s a growing

number of firms adopted performance-based vesting provisions when granting stock option awards. In

addition, starting around 2000, firms began to reduce their use of conventional options, which was likely

due to the expensing requirements. Grants of restricted stock or units were becoming the first choice of

equity-based compensation among the U.S. public firms. Meanwhile, equity awards with performance-

based vesting provisions such as performance-vested stock options have also become popular (Deloitte

2005). Efforts have been exerted by policy makers and institutions to promote performance-based

compensation (IRS 1986). The ISS 2007 US Proxy Voting Guidelines Summary recommends that

shareholders vote for proposals advocating the use of equity awards that are contingent on the achievement

of performance goals (ISS 2007).

3. Hypothesis Development and Data

In this section, we specify a series of testable hypotheses and describe how the sample is obtained.

In addition, we provide basic descriptive statistics.

6

3.1. Hypotheses

Dividend equivalents are consistent with the managerial power theory in that they lack

transparency, and they are generally not performance-based.2 When shareholders do find out about them,

some have taken action, as in the aforementioned GE shareholder proposal. Yet, if DERs are beneficial to

shareholders, such actions are harmful to shareholders.

Thus, the existence of DERs raises a number of key questions. First, we propose to examine the

factors that may be associated with DER policies and DE payments. We will incorporate factors associated

with dividend payments, the existence of dividend payments themselves, governance factors, and fixed /

random effects as necessary to control for firm characteristics. We will also examine the market reaction

to initial disclosures of DER policies. After accounting for the conventional factors that correlate with

stock returns, we will attempt to determine if DE payments are related to monthly stock returns. We will

then examine whether DER policies explain whether companies pay dividends above and beyond other

well-known factors that explain whether companies pay dividends. For companies that pay dividends, we

will examine whether DE payments are associated with the level of dividends. Since share repurchases are

a form of dividend substitute, we will look at share whether DE payments are associated with share

repurchases. Finally, we will examine whether DER policies and payments can reduce the accumulation

of excess cash.

3.2 Data

We start with the set of all S&P 500 firms from 1993 to 2012 that meet six basic criteria: (1) an

S&P 500 constituent for at least three fiscal years, (2) included in both the ExecuComp database and the

Compustat data base, (3) at least five consecutive years of proxy statements available in the EDGAR (SEC)

database, (4) a non-financial (SIC code 6000-6999), non-utility and non-communication and transportation

(SIC code 4900-4949) company, (5) the firm-year observation must not have a stock split or special

dividend events, and (6) its policy regarding DER grants and DE payments must be identifiable.

The SEC began publishing proxy statements available beginning in 1994, which are generally filed

between January and March. For these firms, the executive compensation details provided in a proxy

statement in a given year are for the previous year. Therefore our sample period starts one year prior, 1993.

Information on the existence of DERs is manually collected from the publicly available proxy statements.

The sample selection process is described in Appendix B-1. The final sample has 5,376 firm-year

observations, of which 3,759 firm-year observations include the payment of common dividends. Some of

the tests require certain variables and others require different variables. In all cases, we use every available

2It is important to understand that the shares themselves are performance-based, but the dividend equivalents on those shares are usually not performance-based. Any that are performance-based are not included in our sample.

7

observation. For subsequent analysis, we will require data on a number of financial variables that capture

dividend payments, financial information and governance characteristics, and CEO characteristics and

compensation. Definitions and abbreviations of these variables are in Appendix B-2.

Annual financial data are obtained from the COMPUSTAT database. The industry definitions are

from Professor Kenneth French’s web site.3 We divide all firms into ten industries according to their four-

digit SIC Codes. If there are missing values in the firms’ accounting variables in COMPUSTAT, we then

manually collect the data from the firm’s regulatory financial statements (10-K, 10-Q, and DEF 14A).

Executive compensation data are from the ExecuComp dataset.

As noted, dividend equivalents data items are hand-collected from the companies’ proxy statements

that are available on the SEC web site.4 We collect four pieces of information for each firm-year

observation: the type of equity awards granted (restricted stock or units, performance stock or units, stock

options, etc.), the vesting condition of the equity awards (performance-based, time-based, etc.), the

performance period (if available)5, and the method of DE payment (cash, additional shares, etc.). The

majority of the observations in the sample are DERs granted with performance stock or units and

performance-vested restricted stock or units. The typical performance period is three to five years. DE

payments in the form of cash and additional shares can be made when the dividends on the firm’s common

shares are paid out or can be accrued in a separate account under the equity award holders’ name. When a

firm grants performance-based awards, it may specify a minimum, target, and maximum number of shares

to be earned.6 The number of shares that are eventually earned by the CEO can be anywhere between a

minimum and a maximum. Cash DEs are typically paid on the target number of shares. In cases where

DEs are accrued on such awards, firms may allow DEs to be accrued on the target or the earned number of

shares7. We do not differentiate among the various situations in which dividend equivalents are paid. As

long as a CEO is entitled to receive dividend equivalents immediately or on a deferred basis on shares that

3http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_48_ind_port.html 4http://www.sec.gov/edgar.shtml 5See Appendix A for more information about the performance period. 6For example, Marathon Oil Corporation (Marathon 2003) granted two performance share awards to the CEO in 2003. The performance periods were three and two years. The firm disclosed that “Vesting of these performance shares is based entirely on the achievement of pre-established performance measures related to corporate performance with payouts varying from 0% to 200% of target based on actual performance.” Dividends were paid on the target number of performance shares. See http://www.sec.gov/Archives/edgar/data/101778/000119312504036497/ddef14a.htm 7An obvious question about DEs accrued on performance-based awards is whether they are captured if the underlying awards ultimately are not earned by the executives. The disclosures about how firms treat DEs accrued on forfeited performance-based awards are very limited, especially before the late 2000s. The limited information suggests different situations: if the DEs (in the form of cash or shares) are delivered to the executives before the underlying awards are forfeited, firms do not recoup those DEs. If the DEs are accrued and kept in a separate account, they may or may not be forfeited if the underlying shares are never earned. The disclosures suggest that executives tended to forfeit DEs accumulated but not paid in the 1990s, while they ultimately received DEs accumulated but not paid in more recent years.

8

he or she has not yet earned, we count those dividend equivalents as dividend equivalents on the CEO’s

unearned shares. To simplify the terminology, the phrase “DER policy” in this paper denotes only the

existence of a firm’s policy that allows for DE payments on CEOs’ unearned shares. Having such a policy

does not mean that DEs are actually paid, and we do distinguish between firms that have DER policies and

the subset of those firms that actually make DE payments.

In their proxy statements firms disclose executive and board compensation for the three most recent

fiscal years. The SEC does not require firms to disclose the value of DEs paid, but they do require that

firms disclose whether DEs are paid on shares covered by long-term incentive awards in the footnotes of

the Summary Compensation Table. We collect information from the proxy statements and identify firms’

DER policies and DE payments in a retrospective manner.

For example, suppose a firm disclosed in its 1997 proxy statement that its long-term incentive plan

that came into effect in 1990 allows DE payments on executives’ unearned shares. This disclosure was the

earliest acknowledgement of the firms’ DER policy. Since 1993 is our first year, we then classify the firm

as a DER-policy firm from 1993 to 1997. After 1997 the firm would still be a DER-policy firm if no

changes are made to its DER policy. A firm is classified as a DER-payment firm in a particular fiscal year

if any of its proxy statements indicate that a DE payment on the CEO’s unearned shares was made in that

year. Detailed executive compensation information such as the usage of DERs is usually not in the firm’s

annual report or quarterly report. The proxy statements are the main, if not only, source of information

about DERs. Therefore, we assume that the market learns about the existence or usage of DERs from the

proxy statements. Thus, the day on which the proxy statement is filed with the SEC is the event day of the

disclosure.

The quality of disclosure in the proxy statements is less than ideal in some cases. For example, a

firm could have been paying DERs prior to 1994, when the proxy statements first became available on

EDGAR. A firm could have simply failed to disclose ongoing DE payments, since at one time it was not

required. A firm could have had a DER policy but paid no DERs since it paid no dividends over some

period. Finally, details could have been omitted. If we identify these observations as first-time disclosures,

and the information is already known, it will be far more difficult to draw inferences in the event studies

we will do. Thus, inclusion of these observations imparts a conservative bias, but only to the event studies.

In the most critical analyses we undertake, we need only know if the firm-year observation is associated

with a DER policy or DE payments, not whether it is the observation is the first year of a DER policy or

DE payment.

Table 1 summarizes the distribution of sample firms by time and industry. Panels A and B present

the distribution of sample firms over the period from 1993 to 2012, which we refer to as the Full Sample.

Panel A summarizes the sample firms by year. There are an average of 272 distinct firms in the sample in

9

a given year, 22% of which have policies that allow for the payment of dividend equivalents on CEOs’

unearned equity awards. Ten percent of the sample firms make actual dividend equivalent payments on at

least one type of the CEOs’ unearned equity awards. Both the number of DER-policy firms and the

percentage of DER-policy firms monotonically increase from 1993 to 2007 and then start to drop slightly

afterwards. DER-payment firms show similar patterns8. Panel B of Table 1 presents the distribution of

sample firms across industries. Using Fama and French’s 10-industry classification, the energy and high-

tech industries have fewer DER-policy firms and DER-payment firms. Panels C and D present two

subsamples, the Average Year Sample and the First Year Sample. The two sub-samples are discussed in

Section 4.5 in more details. In short, the Average Year Sample selects one year that most closely represents

the firm’s average year. The First Year Sample is the set of observations of the first year in which a given

firm appears in the overall sample. There is no obvious pattern in the number of DER-policy firms and the

percentage of DER-policy firms in the Average Year Sample. The average percentage of DER-policy firm

in the Average Year Sample is 29%, slightly higher than that of the full sample, while the percentage of

DER-payment firms is much higher than that of the full sample. The First Year Sample shows a similar

pattern with the Average Year Sample.

Table 2 provides a summary of basic descriptive statistics of the variables in this study.

Descriptions of the variables are in Appendix B-2. Although final inferences should not be made based on

univariate statistics, it is of preliminary interest to note that non-DER policy firms pay out an average of

1.77% of their assets in dividends, while DER-policy firms pay 2.34%. Non DER-payment firms pay out

an average of 1.80% of their assets, and DER-payment firms pay 2.84%. Similar observations are made

using other measures of dividends. Neither set of firms is distinct with respect to repurchases, which of

course are a substitute for dividends. DER-policy and DER-payment firms appear to be much larger with

respect to book value of assets (A) but not much different with respect to market cap (MEDecile). Further

and more rigorous analyses of the differences in these samples will be provided in the tables that follow.

4. Results

This section presents the statistical results from our analysis. The two principal issues we consider

are the relationship between DERs and corporate dividend policy and how shareholders react to

announcements related to DERs.

8Obviously the set of DER-policy firms is a larger set than the set of DER-payment firms. If a firm’s proxy statements indicate that it allows DE payments on unearned shares in year t or DE payments were made on CEO’s unearned shares in year t, then it is classified as a DER-policy firm starting at year t. The firm will still be a DER-policy firm in the following years unless it announces a suspension of the policy. A firm is classified as a DER-payment firm in year t if it discloses that DE-payments on CEO’s unearned shares were made in year t. Thus, the sample of DER-policy firms is one set, while the sample of DER-payment firms is a subset of the set of DER-policy firms.

10

4.1. Determinants of DER policy and DE payments

First, we take a look at the characteristics of companies that have DER policies and make DE

payments. We can use a logistic regression with DER policy or DE payment as the indicator dependent

variable, but we have to recognize that policies regarding dividend equivalents and dividend payouts are

naturally bonded. While DER policies can exist in companies that do not pay dividends, they would seem

far more likely to exist in dividend-paying companies. Of course, DE payments can exist only in firms that

pay dividends. So the most fundamental characteristic of dividend policy – whether a firm pays dividends

at all – is a necessity for the payment of dividend equivalents, though not an absolute requirement for firms

with DER policies. Nonetheless there is endogeneity in both cases, so we restrict the analysis to a dividend-

payer sample. By comparing DER-policy firms and non-DER policy firms within the dividend-payer

sample, we determine the characteristics that are related to DER policies, conditioned on the fact that all

firms in the sample are dividend payers.9

The logistic regression is specified as follows:

𝐷𝐸𝑅𝑑𝑢𝑚𝑚𝑦 = 𝛼 + ∑ 𝛽𝑖𝑋𝑖

𝐼

𝑖=1

+ ∑ 𝛾𝑖𝑌𝑖

𝐽

𝑖=1

+ ∑ 𝜃𝑖

𝐾

𝑖=1

𝑍𝑖 + 𝛿(𝐷𝐼𝑉/𝐴𝑆𝑆𝐸𝑇𝑆) + 𝑒

The DERdummy variable is an indicator dependent variable set at 1 if the firm has a DER policy and 0

otherwise. In a second set of regressions, the dependent variable is set at 1 if the firm makes DE payments

in that year and 0 otherwise. The X’s are control variables. Larger firms tend to have more complex

compensation tools such as performance-based equity awards and dividend equivalents so we include a

firm size measure, MEDecile as an explanatory variable. Dividend equivalent rights are discussed in firms’

long term incentive plans and can be related to long term performance goals. Therefore, we include

variables commonly used in studies of firm performance, such as return on assets, leverage, and

market/book. Elsila (2013) suggests that CEO dividend income from unvested restricted stocks can

substitute for cash compensation. Thus, we also include measures of cash compensation and total

compensation to control for possible substitution effects between DE payments and cash compensation.

We also use CEO tenure, which is the length of time that the person has been CEO, and a new CEO indicator

to identify CEOs with less than 18 months experience with the firm. CEO age was originally used to

describe the CEO but was dropped because it co-varies strongly with the time fixed effect.

The literature documents a crucial role of board structure in corporate governance. To encourage

more independent director involvement in the process of establishing executive compensation packages,

9A related but interesting question is whether the potential for making DE payments could induce a firm to be a dividend payer when it otherwise would not. Later we will address that question and account for a similar endogeneity.

11

the SEC increased disclosure requirements as of 1992 when insiders serve on compensation committees.

Research findings show little evidence that CEOs receive premium pay when insiders are members of

compensation committees (Anderson and Bizjak 2003), but the sensitivity of CEO compensation to firm

performance favors CEOs when insiders are on the compensation committee. We use the variable Interlock

to indicate whether a firm CEO is also on the compensation committee. Firms’ leadership structures as a

dimension of corporate governance are also related to this issue. Goyal and Park (2002) find that when the

CEO and chairman duties are vested in the same individual it is much more difficult to dismiss an ineffective

CEO. The variable Dualrole is used to capture this characteristic. Studies show that small boards are more

effective in monitoring managers and can result in better corporate governance (Yermack 1996), so we also

include board size (variable Boardsize) as a corporate governance measure. Thus, we have three governance

variables, but we use only one per regression. The Y’s in the equation above are year and industry fixed

effect variables, and the Z’s are firm random effect variables. We assume that firm random effects

have variance component (VC) variance-covariance structure.10 Finally, we incorporate the key variable of

interest, a measure of the magnitude of dividend payments, which is the ratio of dividends to total assets,

identified as DIV/ASSETS. This variable will indicate whether the percentage of the firm’s assets paid out

in dividends is associated with DER policy or DE payments.

The results are shown in Table 3.11 DER-policy firms are examined in models in Panel A while

DER-payment firms are examined in Panel B. We find that the dividend variable is significantly positively

related to the existence of a DER policy at the 5% level and significantly related to the DE-payment variable

at the 1% level. There is some modest evidence that governance variables are associated with the DER

policies and DE payments. The MEDecile variable is significant, meaning that larger firms are more

associated with DER policies and DE payments. There is some indication that leverage is inversely related

to DER policies but not DE payments. In addition, the higher the ratio of market to book, the less likely a

firm will have a DER policy or make DE payments. Thus, value firms, rather than growth firms, are more

likely to have DER policies and make DE payments.

10We assume firm random effects Z follow a normal distribution with a mean of zero. And the variance-covariance matrix of Z has a structure as follows:

[

𝜎𝐴2 0 0

0 𝜎𝐵2 0

0 0 𝜎𝐶2

]

where 𝜎2’s on the diagonal are variance terms. This structure assumes the heterogeneous variance and the covariance between two different firms is zero. 11Starting with Table 4 and in all remaining tables presenting statistical tests, *, **, and *** indicate that the results are significant at the 10%, 5%, and 1% levels, respectively.

12

4.2. Market reaction towards news of DER policy and DE payments

In this section we examine how the market reacts to two types of first time disclosures about

dividend equivalents. In the first type of disclosure, a company announces that it has been paying, is paying,

or is planning to pay DEs on CEO’s unearned shares. We refer to this type of disclosure as a positive

disclosure in that the board affirms the existence of past, current, or future DE payments. This type of

disclosure may or may not be followed by disclosures of actual DE payouts. In the second type of

disclosure, a company announces that it has never paid, no longer pays, or will not pay DEs on CEO’s

unearned shares, whether the DERs are associated with either existing awards or newly issued awards. We

refer to this type of disclosure as a negative disclosure. Negative disclosures can be preceded by actual DE

payments. In fact, negative disclosures can even be followed by DE payments. For instance, assume a

company paid DEs in the years prior to 2006 and the company’s fiscal year end is December. In the

company’s 2006 proxy statement, the company announced that it would not pay DEs on any unearned

performance-based stock awards that were granted in or after 2006. Nonetheless, holders of awards that

were granted prior to 2006 are still entitled to DERs. Given that the most commonly seen performance

period length is three years, it is possible that the firm still makes positive disclosures in year 2007 and

2008 for the stock awards with DERs granted in 2005. The sample has 93 positive disclosures and 179

negative disclosures.

In addition we estimate CARs around a first-time announcement of DE payments, of which there

are 49 such events. We estimate abnormal returns using the Fama-French three-factor model. The event

dates are the filing dates of proxy statements or the dates of first-time DE payments, and the event period

starts five days prior to the event date and lasts until one day after the event day.

The results are illustrated in Figure 1 and summarized in Table 4. In the figure we see that the two

groups exhibit similar behavior prior to the disclosure but markedly different behavior afterwards. Negative

disclosures result in improved risk-adjusted returns while positive disclosures result in weakening risk-

adjusted returns. Table 4 shows that on average a first-time positive disclosure generates a significant

negative abnormal market reaction, while a first-time negative disclosure has a positive abnormal market

reaction that is significant at the 5% level. The cumulative average return from day -5 to day +1 is -0.70%

for positive disclosures and 0.77% for negative disclosures. First-time payment announcements lead to a

negative CAR of -1.15%. Thus, these results indicate that investors do not favor positive disclosures, they

approve of negative disclosures, and they do not approve of DE payments. Thus, it appears that

shareholders do not view dividend equivalents in a favorable manner.

13

4.3. Monthly stock performance

In this section we examine the effect of DERs on monthly stock returns. We use the Fama-French-

Carhart four-factor model ((Fama and French 1993), (Carhart 1997) to assess monthly stock performance.

The sample we used in the OLS regressions is different from the event-study sample. During the sample

period from 1993 to 2012, there are 298 disclosures from 35 firms about actual DE payments on unearned

shares. We run separate regressions with dummy variables to indicate if the firm had a DER policy or made

DE payments. We incorporate the Fama-French factors, including momentum, and the Entrenchment index

(Bebchuk, Cohen and Ferrell 2009), which is the number of key anti-takeover provisions that firms adopt

and purports to measures the quality of a firm’s corporate governance. We also incorporate an indicator for

whether the firm is a dividend payer. Because the sample firms consist of only S&P 500 firms, there is a

potential survivor bias, so we use an indicator for whether the firm is in the S&P 500 in 1993. The

regression equation is as below:

𝑀𝑜𝑛𝑡ℎ𝑙𝑦 𝑟𝑒𝑡𝑢𝑟𝑛 = 𝛼 + 𝛿𝐷𝐸𝑅_𝑃𝑂𝐿𝐼𝐶𝑌 + ∑ 𝛽𝑖

4

𝑖=1

𝑋𝑖 + 𝛾𝑌 + 𝑒

where DER_POLICY is an indicator variable equal to 1 if the firm-observation has a DER policy and zero

otherwise. The X’s are the Fama-French-Carhart factors. Y is the control variable, which is either the

Entrenchment index, DIVDUM, or an indicator for being an S&P500 firm in 1993. Each regression includes

only one control variable. The standard errors are robust to heteroskedasticity.

The results are summarized in Table 5. The results show that firms that have DER policies

underperform the market by about 0.24% each month, which is significant at better than 0.05. The result

is relatively robust after controlling for variables that may be related to the firm’s DER usage. In separate

regressions, the E-index is significant, as is the dividend dummy, and the survivor bias variable. The

significance level of the DER policy variable does fall to the 10% level when the dividend dummy is

entered. Thus result is not surprising given the interaction between these two effects. The results in Panel

B regarding DE payments are not quite as strong. The DE payment variable is significant in all cases,

though only at the 10% level. These results suggest that the existence of a DER policy and possibly DE

payments has a negative impact on stock returns, even after controlling for other factors. 12

12We also tried to include all three control variables in one regression. DER_POLICY became insignificant with p-value of 0.15 in the regression. We decide it is more appropriate to run regressions for each control variable instead of use all the three due to two reasons: Running one regression for each control variable allows us to use all the available observations, while using all control variables in one regression limit the sample to the observations that have no missing values for any of the control variables. Also, the dividend dummy and SP500_93 have a correlation coefficient of 0.41, which could introduce multi-collinearity problem to the model.

14

4.4. DER policy and the propensity to pay

We now wish to disentangle the association between the DER policy and firms’ dividend payout

policies. A DER policy can affect its dividend payout in several ways. Consistent with the managerial

power theory (Bebchuk and Fried 2003), managers can influence the board to adopt a DER policy that can

benefit management by increasing managers’ dividend income on shares they may or may not own. The

board can also adopt a DER policy to compensate managers for losses in the value of their stock options or

other equity award holdings when dividends are paid. Thus, if a dividend policy insulates managerial

compensation from losses, managers will be more prone to encourage the firm to pay dividends.

Yet, what might appear to be a negative side effect from dividend equivalents can actually be a

positive. Consistent with optimal contracting theory (Murphy 1999), dividend equivalents can motivate

CEOs to increase dividends instead of investing in inferior projects, engaging in wasteful mergers, or

hoarding cash. Each competing theory suggests that ceteris paribus, DER-policy firms and DE-payment

firms could be associated with higher dividend payouts. We construct statistical tests to address this

question in two ways. First, we test whether DER-policy firms and DE-payment firms are more likely to

be dividend payers after accounting for factors that explain the payment and level of dividends in general.

Second, given that DER-policy and DE-payment firms are paying dividends, we examine whether they are

more likely to pay higher dividends.

We use logistic regression to examine how the existence of a DER policy is related to the likelihood

that a firm pays dividends. There is, however, an obvious endogeneity. Other than the possibility of a firm

having a DER policy, paying dividends, and then eliminating their dividends without eliminating its DER

policy, the existence of dividends is highly likely for firms having a DER policy. We are interested in a

subtle question: could DER policies induce firms to pay dividends when they otherwise would not? To

answer that question we need to first control for factors that explain why a firm would be a dividend payer.

Previous research shows that dividend payers are typically large and profitable firms that lack

sufficient investment opportunities and have large amounts of free cash flow (Fama and French 2001).

Thus, we incorporate measures of size, profitability, investment opportunities, and free cash flow. We use

the indicator DER_POLICY to measure whether a firm has a DER policy. Because a pooled logistic

regression has higher power, we use it instead of year-by-year regressions. The regression equation is as

follows:

𝐷𝐼𝑉𝐷𝑈𝑀 = 𝛼 + 𝛽0𝐷𝐸𝑅𝑃𝑂𝐿𝐼𝐶𝑌 + ∑ 𝛽𝑖𝑋𝑖

𝐼

𝑖=1

+ ∑ 𝛾𝑖𝑌𝑖

𝐽

𝑖=1

+ 𝑒

where DIVDUM is a categorical variable with 1 indicating that the firm paid a dividend in the firm-year

observation and 0 that it did not. DER_POLICY is the primary variable of interest and is measured with an

15

indicator variable that takes a value of 1 if the firm allows DE payments and 0 otherwise. The 𝑋s are the

control variables and include the following measures. MEDecile is the NYSE firm-size decile, that is, the

largest decile of NYSE firms that have the same or smaller market capitalization. This measure is not

affected by overall firm size growth over time and is thus more robust in capturing cross-sectional firm size

variation. Investment opportunities is measured by the market-to-book ratio, V/A, and the growth rate of

assets, dA/A. FCF/A measures free cash flow and is defined as operating income before taxes minus interest

expense, scaled by total assets. LEV is the firm’s financial leverage, ROA is the firm’s return on assets, and

the Y’s are fixed effect dummies that control for fiscal year fixed effects, industry fixed effects and firm

fixed effects. All observations are firm-year observations, and the variables are defined in Appendix B-2

with details.

Regressions of dividend policies, however, pose a particular problem. It is widely known that

because companies are reluctant to actively modify their dividend policies and amounts, dividend payments

tend to be smooth. Thus, the time variation in a firm’s dividend payouts is limited. This problem is more

pronounced in the regression described here in which the response variable to capture dividend policy is an

indicator variable. The value of the response variable will not change as long as a firm remains a dividend

payer or remains a non-dividend payer, nor will it reflect the changes in the amount of dividends paid. The

key explanatory variable DER_POLICY suffers from a similar problem. When a firm discloses that it

allows DEs to be paid on unearned shares, the value of the variable will not change unless the firm makes

a change to its DER policy. This lack of variation in variables can cause the statistical results to be

unreliable and exaggerated due to many similar data entries. Thus, we must address this problem with a

more conservative approach. One way to do so is, as noted in the previous paragraph, to use firm fixed-

effect models. In addition, we also use two subsamples based on an average year for a firm and on the first

year of the firm in the data set. This procedure is described further down.

The results are shown in Table 6. The first column contains the fitted coefficients using the full

sample of all available firm-year observations. The indicator DER_POLICY is positively associated with

dividend payout probability. The effect is positive significant at the 5% level, implying a strong association

between a firm’s DER policy and the likelihood of paying a dividend over and above the effect of other

factors that would ordinarily be associated with paying dividends. The odds ratio is 3.834, meaning that

all other things equal, a DER-policy firm is almost four times more likely to be a dividend payer than a

non-DER policy firm. The model has a concordant percentage of 99.3%, suggesting that it fits the data

quite well.13 Of course, as noted it is necessary to be a dividend payer to make DE payments, but we control

13We use SAS software to perform the logistic regression. SAS reports the “percentage concordant” in the output by default. SAS pairs up all observations with different response variables. In this case, it will be all possible combinations of a DER-policy firm and a non-DER policy firm. A pair is concordant if the observation with larger predicted value also has the larger actual value.

16

for this endogeneity by incorporating the factors known to explain why a firm would be a dividend payer.

Consistent with theory and previous studies, larger firms are more likely to be dividend payers while firms

with more investment opportunities are less likely to be dividend payers (Fama and French 2001). As

expected, the fitted coefficient of firm size (MEDecile) is positive and significant, while the fitted

coefficients of investment opportunities (V/A and dA/A) are negative and significant.

In addition, we construct an Average Year subsample, which contains 420 firm-year observations.

Each firm has at most two observations included in this sample. The basic idea is to select one observation

to represent a firm’s “average year,” that is, the year in which the firm’ size and investment opportunities

are the closest to its typical year. For each firm, we first separate observations into two groups, one for the

years during which its DER_POLICY takes a value of one and one for the years during which its

DER_POLICY takes a value of zero.14 We then calculate the average values of the variables MEDecile and

V/A for both groups.15 From each group, the observation in the year when the values of MEDecile and V/A

are the closest to the mean values is chosen for the Average Year sample. Because there are at most only

two observations per firm, it is not necessary to use firm fixed effects. We also construct an additional

subsample that is called the First Year subsample, and contains 420 firm-year observations. For each firm,

we choose the observation(s) from the first year that the variable DER_POLICY takes a value of one and

the first year in which the variable takes a value of zero. Thus, at most there are two observations per firm

and firm fixed effects are also not needed.

The results using the two subsamples in the second and third columns of Table 6 show that the

fitted coefficients of DER_POLICY remain positive and significant at the 5% level, confirming that there

is positive association between the policy allowing for DE payments on unearned shares and the dividend

payout probability, holding other factors constant. Thus, these subsample results agree with the results

from the overall sample regression.

4.5. Dividend Payments, Share Repurchases, and DER Policies

In the previous section, we examined whether the existence of a DER policy makes a firm more

likely to be a dividend payer, holding constant the other factors that lead firms to pay dividends. We find

that indeed it does. In this section we take a look at whether DER-policy firms that already pay dividends

tend to pay higher dividends.

14Some firms have never made any changes to their DE policies over the entire sample period. For these firms, only one observation will be chosen to be included in the Average Year sample. For those that do change their policies, two observations are chosen, with one from the years when the firms allow for DE payments and one from the years when the firms do not allow for DE payments. 15Although a firm’s size changes over time, its NYSE size decile is unlikely to change. This is true for the majority of the sample firms. Thus at least two variables are needed in finding an “average year.” We also examined MEDecile and dA/A to construct the sample. The results are still supported.

17

We use four measures of the size of dividend payments: (1) common dividends divided by total

assets, (2) common dividends divided by operating income, (3) common dividends divided by earnings

before interests and taxes, and (4) common dividends divided by net income. Measures (2), (3), and (4) are

sensitive to negative values in the denominator. When a negative value occurs, we drop the observation.

The regression equation is as follows:

𝑆𝑖𝑧𝑒 𝑜𝑓 𝑑𝑖𝑣𝑖𝑑𝑒𝑛𝑑 𝑝𝑎𝑦𝑚𝑒𝑛𝑡𝑠 = 𝛼 + 𝛽𝐷𝐸_𝑃𝐴𝑌𝑀𝐸𝑁𝑇 + ∑ 𝛾𝑖𝑋𝑖

𝐼

𝑖=1

+ ∑ 𝜃𝑖

𝐽

𝑖=1

𝑌𝑖 + ∑ 𝜌𝑖

𝐾

𝑖=1

𝑍𝑖 + 𝑒

where DE_PAYMENT is the key variable, X’s are the control variables, Y’s are the fixed effects dummies

and Z’s are the random effect dummies.

The results are reported in Table 7. Panel A summarizes the results of the OLS regression using

all available observations for a given regression, all of which are observations of dividend-paying firms.

Panel B contains the results with observations from the financial crisis of 2007 and 2008 excluded. Panel

C provides the results using the Average Year subsample, and Panel D reports the results using the First

Year subsample.

As shown in Panel A, the highest R-square, almost 49%, is obtained using the ratio of dividends-

to-assets. The other three measures give R-squares less than 10%. The estimated coefficient of the indicator

DE_PAYMENT is positive and significant at the 5% level when DIV/ASSET and DIV/EBIT are used to

measure the firm’s dividend payments. When DIV/OPERATINGINCOME is used, the variable is

significant at the 10% level, and when DIV/NI is used, it is not significant. The sizes of the coefficients can

be interpreted as the percentage changes in the dependent variable associated with the existence of DE

payments.

These results are somewhat supported when 2007 and 2008 observations are dropped as shown in

Panel B. Again, the R-square is highest when DIV/ASSETS is used (26%). The effect of DE_PAYMENT

is significant when DIV/ASSETS (10% level), DIV/OPERATINGINCOME (5%), and DIV/EBIT (5%) are

used. The regressions using the Average Year sub-sample show that the variable DE_PAYMENT has a

positive and significant effect on the dividend payout ratios when DIV/OPERATING INCOME and

DIV/EBIT (both at 5% level) are used as proxies. The effect is not significant for the measures

DIV/ASSETS and DIV/NI. The regressions using the First Year sub-sample show that when

DIV/OPERATINGINCOME and DIV/EBIT are used the effect of DE_PAYMENT is significant at 5% and

10% level, respectively. The First Year sub-sample suffers from a selection bias in that the majority of the

sample is from the year 1993. Thus the results may not be as reliable as the ones derived from the other

three samples. Overall the results are generally consistent with the argument that firms that pay dividend

equivalents on unearned shares tend to pay higher dividends. The results are not strongly significant at a

high level, but they are suggestive that DE payments seem to be associated with higher dividends.

18

Of course, dividends are only one means by which firms return cash to shareholders. The other is

share repurchases. Indeed, many firms use repurchases as a substitute for dividends. If the existence of a

DER policy has a positive effect on the payment of dividends, it could have a negative effect on share

repurchases. We examine this possibility using a model specified in Dittmar (2000). We measure common

share repurchases as Purchase of Common and Preferred Stock reported by Compustat less the decrease in

preferred stock, scaled by prior year end market value of equity. The determinants of stock repurchases

include a firm’s cash flow, cash reserve, investment opportunities, as measured by market–to-book,

dividend payout, information asymmetry, stock return performance, leverage, as well as whether the firm

is an object of a takeover attempt, and outstanding stock options. The takeover data are obtained from the

SDC database. We estimate the regression equation below to test the effect of DERs on common stock

repurchases:

𝑅𝐸𝑃𝑈𝑅 = 𝛼 + 𝛽𝐷𝐸𝑅𝑑𝑢𝑚𝑚𝑦 + ∑ 𝛾𝑖𝑋𝑖

𝐼

𝑖=1

+ ∑ 𝜃𝑖𝑌𝑖

𝐼

𝑖=1

+ 𝑒

where the 𝑋𝑖’s are the control variables in the Dittmar model, and the 𝑌𝑖’s represent year fixed effects.

We use both pooled regressions and fixed effects models.16 As shown in Table 8, the variables

DER_POLICY and DE_PAYMENT have significant negative effects on stock repurchases. So, while

dividend equivalents are positively related to dividends, they are negatively related to share repurchases, a

conclusion consistent with the substitution of dividends and share repurchases.

4.6. Endogeneity issues between dividend policies and DER policies

To this point the results indicate that a policy to pay dividend equivalents on CEOs’ unearned shares

is positively associated with dividends payouts. The causal relationship, however, is somewhat complex,

however, because both dividend policies and the policies regarding dividend equivalent rights are

endogenous and related choices. We have controlled for this factor, but we now take further measures to

ensure that endogeneity does not invalidate our findings. To disentangle this problem we use three

approaches: a lagged variable approach, a matched-sample approach and an exogenous shock approach. In

the interest of space, we provided abbreviated summaries of the tests with no tables and figures, but these

supporting items are available on request.

4.6.1. Lagged variable approach

If two events are strongly associated, the event that occurs at an earlier time is more likely to be the

cause and the one that occurs later is more likely to be the result. In this spirit, we use lagged values of the

16In the fixed effects model we include the year dummies. We also tried models with firm fixed effects or firm random effects. The firm fixed effects models show that DERs are not significant. The firm random effect models show that DER_POLICY is significant at 10% level but DE_PAYMENT is insignificant.

19

variable DER_POLICY and rerun the tests described in Section 4.4. The effect of a DER policy that was

established in earlier years on the current year dividend policy can provide support to the argument that a

DER policy can lead to changes in a firms’ dividend policy. The results show that a DER policy can affect

the likelihood of paying dividends up to three years. The one- and two-year DER_POLICY variables have

positive and significant effects on the probability of a firm being a dividend payer. The effects are

significant at 5% level. The three-year lagged DER_POLICY variable also has a modestly positive effect

with a significance level of 10%.

4.6.2. Matched-sample approach

Suppose we have two firms that are similar in the dimensions commonly associated with dividend

policy. One of the firms then initiates a DER policy and the other does not. The two firms thus start to

diverge on only one dimension, DER policy. If at a later date, their dividend policies also become different

from each other, this change is likely due to the only difference between them, their DER policies. In this

spirit we form a matched sample as described below.

We first construct a sample of treatment firms, which are firms that do not pay dividends in the first

year but initiate a DER policy in one of the remaining years and meet several other specifications necessary

for the tests. We then obtain a sample of control firms, which do not pay dividends in the first year and do

not initiate a DER policy during the remaining sample period. We measure the propensity to pay dividends

of each firm in both samples by the fitted dividend probabilities from the logistic regression in Section 4.4.

We require that the control firms have a dividend propensity greater than that of the treatment firms but

less than 0.05. Thus, the likelihood of the control firms to pay dividends is close to but slightly greater than

that of the treatment firms. Our final sample consists of firm-year observations for 24 treatment firms with

a corresponding sample that averages 20 control firms for each treatment firm.

The behavior of both samples over the entire sample period shows that they have a similar

likelihood of being a dividend payer after the DER policy initiation year of the treatment firms. We define

firms with propensity scores greater than the sample average (0.7) to be expected dividend payers, and

firms with propensity scores below average to be unlikely dividend payers. We assign a score of + 1 to

each expected dividend payer and -1 to each unlikely dividend payer. We then track the difference in the

scores of the treatment and control groups over time. When matched up, the average deviation values for

both groups are below zero. We observe, however, that the average deviation of the treatment sample

increases over time, but the mean deviation for the control sample is fairly stable at around -0.5. Seven

years after matching up, the gap between the two groups is as wide as 0.3. At that time, most of the

treatment firms have initiated dividends, but their control firms remain non-payers.

20

This evidence supports the notion that DER policy does seem to provide an incentive for expected

dividend payers to become actual dividend payers. This result does not, however, rule out the possibility

that the CEO anticipates the future dividend payout and influences the board to make self-serving DER

policies in the future. It is likely that an insider like the CEO who holds sufficient proprietary information

has a very good expectation of any upcoming dividend initiation.

4.6.3. The 2003 dividend tax reform as an exogenous shock

On May 23th, 2003, the United States enacted the Jobs and Growth Tax Relief Reconciliation Act

of 2003, which reduced the maximum tax rate on dividends from 38.6% to 15%. This large dividend tax

rate reduction greatly influenced corporate payout policies among U.S. firms. Companies whose executives

had high share ownership increased dividends, signifying the relationship between executive holdings and

payout policy (Brown, Liang and Weisbenner 2003). In addition, more firms initiated regular dividend

payouts, dividend paying firms increased their regular dividend payouts, and special dividends also

increased (Chetty and Saez 2005). These findings also show that firms’ responses to the tax cut was

strongest in firms in which executives had large share holdings and low option holdings, and in firms whose

shareholders had a strong preferences for dividends.

The 2003 tax cut provides a unique laboratory to disentangle the endogeneity between the dividend

policy and policies regarding DERs. The 2003 dividend tax cut as an external shock is unlikely to be

anticipated by firms.17 It can, however, lead to initiatives to change dividend policies. Other corporate

polices such as those regarding DERs, however, are not likely to be affected, at least not in the short run.

We use two-way tables to examine how a firm’s response toward the 2003 dividend tax cut is influenced

by its practices regarding DERs. A firm’s response is measured by any changes it might make in its

dividend policies around the time of the tax cut, such as an increase in dividends or initiation of regular

dividend payouts. The effect of DER practices can be tested by comparing the responses of DER-policy

firms and non-DER policy firms, and between DER-payment firms and non-DE payment firms.

Specifically, if a firm does not have a DER policy in 2002 but does in 2004, it is classified as a policy

initiator. If it does not have a DER policy in either 2002 or 2004, it is a non-policy holder. If it has a DER

policy in both 2002 and 2004, it is classified as a policy holder. If has a DER policy in 2002 but not in

2004, it is considered a policy terminator. We make the same classification decision for the firm based on

its payment of regular dividends. We then do the same classifications based on its DE payments. We

determine the totals in all categories for all groups and conduct a Fisher Exact Test, which reveals that how

a firm changes its dividend policy during the tax reform period is influenced by its DE payments. The two

17The key requirement is that the event is not anticipated well in advance, such as a year or so, and that the law would not be retroactive. In other words, we would not expect firms to alter their dividend policies in 2002 in anticipation of a lower and retroactive tax change in 2003.

21

sets of results using the DER policy measure and the actual payment measure are consistent with the

hypothesis that allowing for DE payments or making DE payments on unearned shares encourages firms

to pay dividends.

5. Are DERs Associated with the Accumulation of Excess Cash?

Even though DERs might encourage the payment of higher dividends, it is not necessarily true that

higher dividends are beneficial for shareholders. The question of the relevance of dividends is certainly

one of the classic issues in finance. Nonetheless, one thing is clear. It is an incontrovertible fact that

companies that hold more cash than they need to cover their costs, take advantage of investment

opportunities, and provide a reasonable liquidity reserve are not acting in the best interests of their

shareholders. If DERs encourage the return of unproductive cash to investors, they have a subtle benefit.

Thus, we wish to determine how DERs are related to the holding of unproductive cash. We approach this

question by examining whether the adoption of DER policies and the institution of DE payments follow

periods in which companies accumulate cash and whether the unproductive cash declines.

To investigate this problem, we first define and estimate excess cash. We start by specifying that

for a given company there is a benchmark amount of cash relative to total assets that can be defined as the

normal cash holding for firms with similar characteristics. We start with a sample that contains all firm-

year observations of all non-financial COMPUSTAT firms during the sample period. We estimate the

normal cash holding level for all non-DER policy firms using two methods, the first from DeAngelo,

DeAngelo, and Stulz (2010) and the second from Opler, Pinkowitz, Stulz, and Williamson (1999).

In the DeAngelo, DeAngelo, and Stulz (DDS) measure, each year the available firms are first sorted

into tertiles based on book value of assets. Within each tertile, firms are then sorted into tertiles based on

market-to-book. Thus, each observation is classified into one of nine groups based on size and market-to-

book. Each year within each of the nine groups, the industry median cash level is calculated using the

Fama-French 10 industry definitions. When calculating the industry median, we require at least three firms;

otherwise, we drop the observation. Each of these nine categories contains observations from the same

fiscal year, same industry, same size rank and same market-to-book rank. For each DER policy observation

we find the difference between the cash-to-asset ratio for that observation and the median cash-to-asset

ratio for that industry in the appropriate size and book-to-market category.

The second measure of normal cash is taken from Opler, Pinkowitz, Stulz, and Williamson (1999)

and is based on a regression of the log of the cash-to-net assets ratio on market-to-book, size, cash flow, net

working capital, R&D expenses, industry median cash flow volatility for last 10 years, leverage, capital

expenditures, a dummy for whether the firm is in a regulated industry and a dummy variable for whether

22

the firm pays dividends. The predicted value of that regression becomes the normal cash estimate for this

company.

We first conduct a visual analysis that will give a qualitative sense of the pattern of excess cash for

DER adopters. We classify every firm-year observation into one of three categories: cash rich, cash middle,

or cash poor based on the excess cash for that firm-year. We then determine the percentage of the sample

of DER policy and DE payment firms classified as cash rich for each year relative to the adoption or

payment year. These results are presented in Figures 2 and 3. Panel A of each figure is based on the DDS

measure with Panel B based on the OPSW measure. In Figure 2, Panel A, which reflects DER policy using

the DDS measure, we observe a steadily increasing portion of the sample of DER firms classified as cash

rich as the policy adoption date approaches. Following that date, the proportion begins to level off. A

similar result is shown in Panel B, using the OPSW measure. In Panels A and B of Figure 3, which are

based on DE payments, the pattern is slightly less clear but nonetheless somewhat similar. This visual

evidence is consistent with the notion that adoption of a DER policy or payment of DEs may encourage

firms to disgorge unproductive cash.

Classifying firms by whether they are cash rich, cash median, or cash poor is a relative measure of

excess cash. Next we conduct a more formal analysis to determine an absolute measure of excess cash and

whether DER policy or DE payments have any effect on the holding of excess cash. We regress the excess

cash-to-assets ratio using both the DDS and OPSW measures on market-to-book, size, cash flow, net

working capital, R&D, industry cash flow volatility, leverage, capital expenditures, a regulation factor, and

DIV/ASSETS as control variables. The formal regression equation is as follows

𝐸𝑥𝑐𝑒𝑠𝑠 𝑐𝑎𝑠ℎ = 𝛼 + 𝛽𝐷𝐸𝑅𝑑𝑢𝑚𝑚𝑦 + ∑ 𝛾𝑖𝑋𝑖

𝐼

𝑖=1

+ ∑ 𝜃𝑖𝑌𝑖

𝐼

𝑖=1

+ 𝑒

where the X’s are the control variables described above and the Y’s are fixed effect dummies.

Most of these variables were already employed in estimating normal cash, but for that purpose we

used an out-of-sample approach, that is, we omitted DER and DE firm-year observations when estimating

the two normal cash models. Those control variables should be significant, as these factors would help

explain dividend payments for DER companies as well as non-DER companies. The results are in Table

9, and indeed the control variables are almost all significant. We also find that cash flow and the

DIV/ASSETS variables are positive and statistically significant. Thus, not surprisingly, cash flow is

positively associated with excess cash, and firms that pay out higher dividends have more excess cash.

Clearly the payment of dividends does not, in and of itself, significantly reduce excess cash. The DER

policy dummy and the DE payment dummy, however, are negative and statistically significant. Thus, DER

policy and DE payments are negatively associated with excess cash, even after controlling for the payment

of dividends and other factors that can affect dividends.

23

These results are supportive of the notion that firms that adopt or pay DEs seem to experience a

strong buildup of excess cash in years prior to adoption or payment. The adoption of a policy or payment

of DEs does appear to reduce the amount of excess cash that companies hold. Thus, DEs have a positive

benefit in helping to reduce the hoarding of unproductive cash.

6. The Magnitude of DE Payments

An obvious question is how large are DE payments? Are they significant enough to motivate

executives? It would be interesting if data were available on the magnitude of all DE payments. Disclosure

requirements on DE payments have been variable. At one time, firms were required only to disclose the

dollar amounts if they were considered abnormal, such as paying an executive a higher dividend than the

shareholders, though it is unclear if any such inflated payments have ever been paid. Current regulations

require disclosure of the amounts paid if they have not been incorporated into the grant date fair value of

the award. In other words, they must be disclosed if they have not been already expensed. Whether they

have already been expensed with the grant award depends on the company’s estimate of the expected

portion of the DEs that will be ultimately be earned. Thus, the precise information needed is not available,

but we can make a rough estimate. Obviously the total dollar amount of dividends paid is a known figure,

but we need to know the number of unearned shares covered by the dividend equivalents. ExecuComp

reports the number of unearned shares (the variable “eip_unearn_num”) of the top five executives including

CEOs. The values of this item are reported by some firms in the Summary Compensation Table. This data

item is available for observations after fiscal year 2006.

In the definition we use in this study, unearned shares are those that are granted but have not been

earned by CEOs. They are subject to achievement of performance goals. We use the proxy statements to

collect the number of unearned shares. We compare the unearned shares reported by ExecuComp and our

sample and find a strong correlation with over 90% of observations identical. The observations that are

inconsistent are mainly due to the fact that a company fails to include performance-based restricted stock.

In this case ExecuComp’s measure would underestimate the unearned shares. Another source of bias is

that it is possible that some unearned shares are not specifically covered by a particular DE payment. Thus,

there is a small degree of error in this estimate, as it is with nearly all estimates.

For each DE payment firm-year observation, we estimate the DE payment to the CEO as the

product of the number of unearned shares and the common dividend per share. Of course, keep in mind

that the DE payment may be underestimated due to the first source of bias and overestimated due to the

second. It is difficult to determine which case dominates. Our overall estimate of the average DE payment

is found to be $308,542. To put this number in perspective, the average salary of a CEO from ExecuComp

24

in 2006-2012 for all DER firm-year observations is $1,263,430. Thus, this rough estimate is about a 25%

increment to a CEO’s salary.

7. Summary and Conclusions

Incentive-based compensation is widely accepted as a means of aligning shareholder and manager

interests. Yet when performance-based compensation is disconnected from performance, the effect is

offset. In this paper, we study dividend equivalent rights, which are a device in which compensation and

performance are largely disconnected. DERs, which allow executives to receive dividends on shares they

may never own, are permitted by about 20% of S&P 500 firms and payments occur in 10% of S&P 500

firms.

Our findings show that dividend equivalent rights, although viewed unfavorably by investors, can

provide valuable incentives. Even after accounting for other factors that make firms likely to be dividend

payers, firms that have policies allowing DE payments on CEOs’ unearned shares are more likely to be

dividend payers. Firms that make actual DE payments tend to pay higher dividends. In addition, we find

that firms that adopt DER policies and firms that initiate DE payments show a pattern of accumulating large

amounts of excess cash for as much as 10 years prior to adoption of a DER policy or initiation of DE

payments. After adoption of a policy or payment of a DE, the evidence suggests that DER policies and DE

payments reduce excess cash.

Thus, with DERs associated with the payment of dividends and the payment of higher dividends,

DERs can have a beneficial effect on shareholder wealth, even if the evidence to this point is that

shareholders do not necessarily believe it.

25

References

Aboody, David, and Ron Kasznik, 2000. CEO stock option awards and the timing of corporate voluntary

disclosures, Journal of Accounting and Economics 29, 73-100.

Alcoa, 2009. Alcoa Inc 2009 Proxy Statement.

Anderson, Ronald C., and John M. Bizjak, 2003. An empirical examination of the role of the CEO and the

compensation committee in structuring executive pay, Journal of Banking & Finance 27, 1323-

1348.

Arnold, Markus C., and Robert M. Gillenkirch, 2005. Stock options and dividend protection, Journal of

Institutional and Theoretical Economics 161, 453-472.

Bebchuk, Lucian Arye, and Jesse M. Fried, 2003. Executive compensation as an agency problem(National

Bureau of Economic Research).

Bebchuk, Lucian, Alma Cohen, and Allen Ferrell, 2009. What matters in corporate governance, Review of

Financial Studies 22, 783-827.

Bebchuk, Lucian, and Jesse Fried, 2004. Pay without performance(Harvard University Press).

Bettis, C., J. Bizjak, J. Coles, and S. Kalpathy, 2010. Stock and Option Grants with Performance-based

Vesting Provisions, Review of Financial Studies 23, 3849-3888.

Brooks, Robert, Don M. Chance, and Brandon Cline, 2012. Private information and the exercise of

executive stock options, Financial Management 41, 733-764.

Brown, Jeffrey R., Nellie Liang, and Scott Weisbenner, 2003. Executive financial incentive and payout

policy-firm responses to the 2003 dividend tax cut, The Journal of Finance 62, 1935-1965.

Callaghan, Sandra Renfro, P. Jane Saly, and Chandra Subramaniam, 2004. The timing of options repricing,

Journal of Finance LIX, 1651-1674.

Carhart, Mark M, 1997. On persistence in mutual fund performance, The Journal of Finance 52, 57-82.

Chance, Don M., Raman Kumar, and Rebecca B. Todd, 2000. The 'repricing' of executive stock options,

Journal of Financial Economics 57, 129-154.

Chetty, Raj, and Emmanuel Saez, 2005. Dividend taxes and corporate behavior-evidence from the 2003

dividend tax cut.pdf, Quarterly Journal of Economics 120, 791-833.

DeAngelo, Harry, Linda DeAngelo, and Rene M. Sulz, 2010. Seasoned equity offerings, market timing,

and the corporate lifecycle, Journal of Financial Economics 95, 275-295.

Deloitte, 2005. Options take a hit, but what will take their place- StockCompSurvey.

Dittmar, 2000. Why do firms repurchase stock? The Journal of Business 73, 331-355

Elsila, Anna, 2013. CEO's dividend income and its role in compensation decisions. Working paper,

University of Oulu.

26

Fama, Eugene F., and Kenneth R. French, 1993. Common risk factors in the returns on stocks and bonds,

Journal of Financial Economics 33, 3-56.

Fama, Eugene F., and Kenneth R. French, 2001. Disappearing dividends: changing firm characteristics or

lower propensity to pay?, Journal of Financial Economics 60, 3-43.

FedEx, 2010. FedEx 2010 Proxy Statement.

Fenn, George W., and Nellie Liang, 2001. Corporate payout policy and managerial stock incentives,

Journal of Financial Economics 60, 45-72.

GE, 2006. General Electric 2006 Proxy Statement.

Glagau, Denise, 2013. Taxation of Dividends and Dividend Equivalents on Stock Awards by U.S.

Corporations. http://f.datasrvr.com/fr1/813/16489/Glagau_Article.pdf

Goyal, Vidhan K., and Chul W. Park, 2002. Board leadership structure and CEO turnover, Journal of

Corporate Finance 8, 49-66.

IRS, 1986. Section 162(m) Internal Code of 1986.

ISS, 2007 Proxy Voting Guidelines Summary.

Jensen, Michael C., and William H. Meckling, 1976. Theory of the Firm: Managerial Behavior, Agency

Costs and Ownership Structure, Journal of Financial Economics 3, 305-360.

Jensen, Michael C., and Kevin J. Murphy, 1990. Performance pay and top-management incentives, Journal

of Political Economy 98, 225-264.

Kahle, Kathleen M., 2002. When a buyback isn't a buyback: open market repurchases and employee

options, Journal of Financial Economics 63, 235-261.

Kanter, Jeffrey M., and Frederic W. Cook & Co., Inc., 2005. Development and trends in compensating

officers and directors over the last year.

Lambert, Richard A., William N. Lanen, and David F. Larcker, 1989. Executive stock option plans and

corporate dividend policy, Journal of financial and Quantitative Analysis 24, 409-425.

Marathon Oil, 2003. Marathon Oil Corporation 2003 Proxy Statement.

Matsunaga, Steven R., and Chul W. Park, 2001. The effect of missing a quarterly earnings benchmark on

the CEO's annual bonus, The Accounting Review 76, 313-332.

Mead Westvaco Coporation, 2011. Mead Westvaco Corporation 2011 Proxy Statement.

Minnick, Kristina, and Leonard Rosenthal, 2010. Stealth Compensation : Do CEOs/directors increase their

pay by influencing dividend policy?

Murphy, Kevin J., 1999. Chapter 38 Executive compensation(Elsevier).

Opler, Tim, Lee Pinkowitz, Rene Stulz, and Rohan Williamson, 1999. The determinants and implications

of corporate cash holdings, Journal of Financial Economics 52, 3-46.

Thurm, Scott, 2006. Extra Pay: Many CEOs Receive Dividends on 'Phantom' Stock.

27

Verizon, 2003. Verizon 2003 Proxy Statement.

XcelEnergy, 2005. XcelEnergy 2005 Proxy Statement.

Yermack, David, 1996. Higher market valuation of companies with a small board of directors, Journal of

Financial Economics 40, 185-211.

Yermack, David, 1997. Good Timing: CEO Stock Option Awards and Company News Announcements,

Journal of Finance 52, 449-476.

Zhang, Dan, 2012. Managerial Dividend-paying incentives.

28

Appendix A. Classification and Conceptual Context

The most commonly seen underlying shares of DERs are stock units, which are the elements of a

grant valued in terms of company stock but in which shares are not issued at the time of the grant. Dividends

paid on executives’ incentive stock awards are referred to as dividend equivalents (DEs), and the right to

receive dividend equivalents is referred to as dividend equivalent rights (DERs).

Instruments that can be granted with DERs include executive stock options, restricted stock or

units, performance stocks or units, and other equity-based stock awards. Payments can be made in the form

of cash, common shares, or incentive shares of equivalent value. The provisions regarding DERs contain

details such as the amount of DEs paid, how and when the payments of DEs will be made, and whether the

DE payments are subject to certain time-based or performance-based conditions or both.

Our sample of unearned shares includes executives’ unvested performance-based restricted stock

or units, unearned performance stock or units, shares covered by unexercised executive stock options, and

shares covered by stock appreciation rights. Unvested shares include unvested time-based restricted stock

or units.

A typical performance stock award has four key elements: the award size, the performance goals,

the performance period and the vesting schedule. The award size is the number of shares that can be earned

by an executive if certain conditions are met. A small number of firms allow executives to earn more shares

than the award size based on exceptional performance. Performance goals are the criteria to which the

performance of an executive is compared. Performance benchmarks typically include stock market

measures as well as dividend yield and accounting variables such as total shareholder return, net income,

or sales growth. The performance period is the time interval over which an executive’s performance is

measured. In most cases, the performance period is three years. During the performance period the

underlying shares of the performance stock award are not owned by executives. As such the award is said

to be fully risky. At the end of the performance period, the number of shares earned by the executive is

determined, issued and credited to the executive as a shareholder. A vesting period is not a necessity. Some

stock awards have one while others do not. Most firms require that stock awards earned by executives vest

over a certain period of time. Earned performance stocks are similar to restricted stocks. Once the shares

are earned, the holder enjoys full rights as a shareholder. Dividend equivalent payments on performance

shares during the performance period are of interest in this study. Dividend equivalents on unearned

performance shares can be paid in cash or in additional shares.

Executive stock options are contracts that give the executives the right but not the obligation to buy

shares at an agreed-upon price. When the stock options are exercised, the executive receives the shares and

all rights as shareholders thereafter. Executive stock options sometimes have vesting periods during which

certain restrictions may apply. Stock options expire if the executives fail to exercise the options, which

29

will occur if the stock price fails to achieve the exercise price. The shares covered by executive stock

options are not owned by the executives before being exercised. Stock options as a means of compensating

executives are less common today than in the early 2000s. Many online resources still define DERs as the

dividend rights on the shares covered by executive stock options, showing that the combination of DERs

and executive stock options was once common. Motivated to some extent by rules forcing the expensing of

stock options that came into effect in 2004, restricted stock units and performance stock have greatly

reduced the use of stock options. Our sample collected from S&P500 firms shows that around two percent

of firms disclose that they have paid dividends on shares covered by executive stock options. Dividend

equivalents on options can be paid in cash, share, or additional stock options.

Restricted stock is an important alternative to stock options. The most common type of restricted

stock is time-vested restricted stock. The vesting condition of these restricted stocks is that the recipient

stays employed by the firm for specified period of time. Restricted stock is not transferable prior to being

vested. Time-vested restricted stock is owned by the recipients. A less common type of award is

performance-vested restricted stock. It is similar to unearned performance stocks in that the vesting

conditions include performance goals. For instance, Xcel Energy Inc. requires that the restricted stock units

granted after 2004 vest upon satisfaction of criteria including achievement of total shareholder return of 27

percent (XcelEnergy, 2005). Shares are subject to forfeiture if the pre-set performance goals are not met.

Therefore we consider unvested performance-vested restricted stock as unearned stock. If a firm pays

dividends on a CEO’s unvested performance-vested restricted stocks, we include it is a DER firm and

include the observation in the DER sample.

Restricted stock units are frequently referred to as “a sibling of restricted stock”. They share some

similarities in vesting conditions and vesting periods. The process of granting, vesting, and delivering

restricted stock units may or may not actually involve actual shares and may or may not be settled in cash.

The actual shares vested and delivered to the recipients may depend on how the performance goals are

reached. Restricted stock units have become popular in the last 10 years. Restricted stock units can be

time-vested or performance-vested. If a firm pays dividends on unvested performance-vested restricted

stock units, we include it is a DER firm and incorporate it in the DER sample.

Accounting standards prescribe that if the grant with which the DERs are associated is expensed at

the grant date, there is no further charge taken when the dividends are paid. The logic behind this rule is

that the present value of the grants is theoretically a component of the charge taken at the grant date.

Accounting rules, however, permit a certain allowance for expected awards that will never vest. Thus,

some dividend equivalents would not be expensed at the grant date. If the actual dividends subsequently

paid on the shares in question differ from the DEs expensed, an adjustment is made at a later date.

30

Appendix B. Sample Selection Process and Descriptions of Variables

B.1. Sample Selection Process

Criteria Firm-year

Observations

ExecuComp S&P 500 firm-year observations with CEOANN=’CEO’ and CUSIP identifiers

8,196

Less: Financial (SIC code 6000-6999), utility (SIC code 4000-4949), and observations either DER policies cannot be identified or no proxy statement is available

2,134

Firm-year observations included in the sample 6,062 Firm-year observations in which common dividends are paid 4,289 Firm-year observations in which common dividends are not paid 1,773

B.2. Variable Definitions

Variable Definition

DER_POLICY Categorical variable. =1 if the firm allows DE payments on CEOs’ unearned shares in year t; =0 otherwise

DE_PAYMENT Categorical variable. =1 if the firm makes DE payments on CEOs’ unearned shares in year t; =0 otherwise

Payout measures: DIV/ASSETS Total annual common dividends over total book value of assets DIV/EBIT DIV/EBIT where EBIT is earnings before interests and taxes DIV/NI DIV/NI where NI is net income DIV/OPERATING INCOME DIV/OIBDP where OIBDP is operating income before depreciation

DIVDUM Categorical variable = 1 if annual common dividend > 0; = 0 otherwise DIVYLD DIV/stock price at fiscal year end REPUR The repurchase payout ratio equal to annual expenditure on the purchase of

common and preferred stocks / End of prior year market value of equity Accounting variables: A Total book value of assets CAPEX Capital Expenditure CF Cash flow = (Operating Income Before Depreciation – Interests – taxes –

common dividends) / net assets Net assets = total assets

CF_VOL Industry cash flow volatility CF2 Cash flow as defined in Dittmar (2000)

(Net income before taxes + Depreciation + Changes in deferred taxes + other deferred charges) / Total Asset

CR Cash Ratio Cash and Short-Term Investments / Total Asset

31

dA/A Growth rate of firm assets in year t. dA in year t = Total assets in year t – Total Asset in year t-1; dA/A=( Total assets in year t – Total Asset in year t-1)/Total assets in year t

E/A Earnings Before Interest/Total Asset. Earnings Before Interest = Earnings Before Extraordinary Items + Interest Expense + Deferred Income Taxes (income)

FCF/A Free Cash Flow/Total Asset Free Cash Flow = Operating Income Before Depreciation – Capital Expenditures

LEV Leverage (Long Term Debt – Cash Holding)/Total Asset

LEVER A firm's leverage ratio in excess of its target leverage Leverage = LEV Target leverage ratio is the median leverage ratio of all firms with the same two=digit SIC code

ME Market value of equity, which equals stock price times common shares outstanding

MEDecile NYSE decile of market value of equity NWC Net Working Capital = Current Assets – Current Liabilities OPTIONS The percentage of shares outstanding held in reserve to cover executive stock

options R&D Research and Development Expenses REGULATION = 1 if a firm is in the regulated industry; = 0 otherwise RET Annual return - market return ROA Return on Assets. =Net income / Total Asset SIZE Firm size.

Ln (Total Assets) V/A Market value of the firm / Total Assets

Market value of the firm = Total Assets – Book Equity + ME; Book Equity = Stockholder’s Equity – Preferred Stock + Balance Sheet Deferred Taxes and Investment Tax Credit – Post Retirement Assets

Corporate governance measures: Boardsize Number of directors on board Dualrole = 1 if CEO is also the board chair; = 0 otherwise Interlock = 1 if CEO is also on the compensation committee; = 0 otherwise CEO characteristics and compensation: Cashcomp Cash compensation. = salary + bonus (in thousands) CEO age Age of the CEO New CEO = 1 if it is within the person’s first 18 months of being CEO; = 0 otherwise Tenure Number of years the person serves as CEO Totcomp Total compensation. = salary + bonus + restricted stock awards (in

thousands) Risk factors: HML High-minus-low. From Fama&French three-factor model MKT Market excess return. Market portfolio return – risk free rate

32

MOM Momentum Factor SMB Small-minus-big. From Fama&French three-factor model Others: SP500_93 Dummy variable. =1 if the firm is a SP500 firm in 1993; =0 otherwise TAKEOVER = 1 if a firm is the object of a actual takeover attempt or a rumored takeover

attempt; = 0 other wise

33

Table 1. Sample Characteristics

Panel A. Full Sample, firm-year observations by time

This panel summarizes the sample firms over the sample period from 1993 to 2012. The columns contain the fiscal years, the number of firms that are included in the sample each year, the number of firms that have a policy allowing for dividend equivalent payments on unearned shares (DER-policy firms), the percentage of DER-policy firms each year, the number of firms that make actual dividend equivalent payments each year (DE-payment firms), and the percentage of DE-payment firms each year, respectively.

Fiscal

Year

# of Sample

Firms

# of DE-

policy firms

% of DE-

policy firms

# of DE-

payment firms

% of DE-

payment firms

1993 223 22 10% 14 6% 1994 244 29 12% 17 7% 1995 252 33 13% 20 8% 1996 272 40 15% 20 7% 1997 284 43 15% 20 7% 1998 285 47 16% 21 7% 1999 295 49 17% 21 7% 2000 299 54 18% 25 8% 2001 301 56 19% 28 9% 2002 304 59 19% 24 8% 2003 308 67 22% 28 9% 2004 312 77 25% 32 10% 2005 322 86 27% 38 12% 2006 327 95 29% 44 13% 2007 338 102 30% 45 13% 2008 336 99 29% 44 13% 2009 341 97 28% 46 13% 2010 337 99 29% 47 14% 2011 339 99 29% 49 14% 2012 307 79 26% 35 11% Tot. 6026 1332 22% 547 9%

Avg. 301.3 67 21% 31 10% (Table 1, Panel B follows)

34

(Table 1 continued)

Panel B. Full Sample, firm-year observation by industry

This panel summarizes the sample firms across industries. We use the Fama-French 10-industry classifications and divide the sample into ten industries based on SIC codes (http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_10_ind_port.html). We exclude financial (SIC code 6000-6999), utility, communication and transportation (SIC Code 4900-4949) firms in the sample selection process. According to the Fama and French 10-industry classification, those SIC codes correspond to three categories: Telephone and Television Transmission, Utilities and Other (including Finance, Business Services, Mines, Construction, etc.). The Telephone and Television Transmission and Utilities categories are dropped from the sample. The category “Other” is retained because firms in this category except for the Finance firms are included in the sample. Thus eight industry categories are left in the sample. Industry # of Sample

Firms

# of DER-

Policy Firms

% of DER-

Policy Firms

# of DE-

Payment Firms

% of DE-

Payment Firms

Non-Durables 51 19 37% 6 12% Durables 8 2 25% 3 38% Manufacture 96 35 36% 18 19% Energy 43 11 26% 4 9% High-Tech 92 21 23% 7 8% Shops 57 19 33% 4 7% Health 43 12 28% 5 12% Other 36 12 33% 6 17% Total 426 131 31% 53 12% Average 53.3 16.4 30% 6.6 15%

(Table 1, Panel C follows)

35

(Table 1 continued)

Panel C. Average Year Sample, firm-year observations by time

This panel summarizes the Average Year Sample and the First Year Sample. The Average Year Sample contains data for a year chosen as each firm’s “average year.” For example, if we find that a firm remains a DER-policy firm for more than two consecutive years, we then average the firm’s MEDecile (their decile of market value of equity) and V/A (market/book) over that period and choose the firm-year observation that has the closest values of these measures to the average values. This observation becomes the firm’s average-year observation in the DER-policy Average Year sample. Similarly, if the firm stays in the non-DER policy firm sample for a few years, we choose its average-year observation in the Non-DER policy average year sample in a similar way. The sample is summarized in the same manner with Panel A.

Fiscal

Year

# of Sample

Firms

# of DER-

Policy Firms

% of DER-

Policy Firms

# of DE-

Payment Firms

% of DE-

Payment Firms

1993 13 3 23% 5 38% 1994 22 3 14% 1 5% 1995 25 2 8% 2 8% 1996 16 4 25% 3 19% 1997 20 5 25% 1 5% 1998 12 1 8% 3 25% 1999 12 2 17% 0 0% 2000 18 5 28% 1 6% 2001 24 5 21% 2 8% 2002 17 2 12% 3 18% 2003 21 6 29% 3 14% 2004 35 10 29% 3 9% 2005 26 12 46% 8 31% 2006 28 9 32% 7 25% 2007 27 9 33% 6 22% 2008 10 4 40% 4 40% 2009 25 11 44% 6 24% 2010 23 15 65% 10 43% 2011 24 12 50% 9 38% 2012 22 9 41% 8 36%

Average 21 6.45 29% 4.25 21% (Table 1, Panel C follows)

36

(Table 1 continued)

Panel D. First Year Sample, firm-year observations by time

This panel summarizes the First-Year Sample, which is the set of observations in which a given firm becomes a DER-policy firm or a non-DER policy firm for the first time. The sample is summarized in the same manner as with Panel A.

Fiscal

Year

# of Sample

Firms

# of DER-

Policy Firms

% of DER-

Policy Firms

# of DE-

Payment Firms

% of DE-

Payment Firms

1993 204 20 10% 20 10% 1994 23 10 43% 8 35% 1995 10 3 30% 0 0% 1996 31 11 35% 7 23% 1997 11 4 36% 2 18% 1998 5 4 80% 4 80% 1999 8 2 25% 2 25% 2000 11 3 27% 1 9% 2001 8 5 63% 3 38% 2002 14 7 50% 5 36% 2003 11 8 73% 3 27% 2004 10 8 80% 4 40% 2005 14 9 64% 8 57% 2006 15 12 80% 8 53% 2007 13 9 69% 4 31% 2008 10 4 40% 5 50% 2009 4 2 50% 4 100% 2010 4 3 75% 2 50% 2011 7 4 57% 6 86% 2012 7 3 43% 2 29%

Average 21 6.55 52% 4.9 40%

37

Table 2. Univariate Statistics

Descriptions of the variables are in Appendix B.

Non-DER Policy DER Policy Non-DE Payment DE Payment

Mean Std Mean Std Mean Std Mean Std

Payout Measures: DIV/ASSETS 1.77 3.17 2.34 2.88 1.80 3.13 2.84 2.85 DIV/OPERATING INCOME 9.55 17.94 15.50 47.14 10.24 28.34 16.61 16.74 DIV/EBIT 11.33 116.00 20.21 32.44 11.91 108.28 25.72 68.39 DIV/NI 20.32 237.91 32.44 115.30 21.55 222.24 36.17 158.14 DIVYLD 1.29 1.97 2.07 3.68 1.35 2.03 2.54 4.80 REPUR 4.04 8.72 4.45 7.08 4.19 8.64 3.70 5.52 DIVDUM 0.68 0.47 0.84 0.37 0.69 0.46 0.97 0.17 Accounting Variables: A 14,681 34,152 23,892 66,538 14,537 41,276 25,670 58,725 MEDecile 7.39 1.44 7.71 1.31 7.41 1.45 7.93 1.04 E/A 0.17 0.08 0.18 0.09 0.17 0.09 0.17 0.07 V/A 2.50 2.44 2.36 1.69 2.53 2.39 1.96 0.91 FCF/A 0.11 0.09 0.12 0.09 0.11 0.10 0.12 0.08 dA/A 0.10 0.20 0.07 0.16 0.09 0.20 0.06 0.14 ROA 0.07 0.13 0.08 0.08 0.07 0.13 0.07 0.06 LEV 0.06 0.25 0.08 0.21 0.06 0.25 0.11 0.19 Corporate Governance Measures: Interlock 0.03 0.19 0.04 0.18 0.03 0.18 0.04 0.21 Dualrole 0.36 0.48 0.49 0.50 0.38 0.49 0.52 0.50 Boardsize 10.19 2.47 10.93 2.18 10.26 2.44 11.70 2.06 CEO Characteristics and Compensation: New CEO 0.11 0.32 0.12 0.32 0.12 0.32 0.10 0.30 Tenure 7.31 6.68 6.68 5.61 7.20 6.57 6.92 5.37 Cashcomp 1,757 1,500 1,838 1,410 1,749 1,479 1,999 1,465 Totcomp 2,243 2,672 2,356 2,636 2,243 2,677 2,493 2,544

38

Table 3. Logistic Regression Analysis of the Characteristics of DER-policy firms and DER-

payment firms.

The response variables of the logistic regressions shown below are DER dummies. The response variable for the regressions in Panel A is DER_POLICY. The response variable for the regressions in Panel B is DE_PAYMENT. Variable definitions are in Appendix B-2. Panel A. Determinants of DER policies

(1) (2) (3) Intercept -5.4163 *** -3.1976 ** -3.6422 ** MEDecile 0.2565 ** 0.252 ** 0.2636 ** ROA 1.7769 -0.6377 0.0439 LEV -1.2173 ** -0.9975 -1.0366 * V/A -0.2431 *** -0.0835 -0.2157 ** Cashcomp 0.0001 0 0.0001 Totcomp -0.0001 0 -0.0001 * Tenure -0.0123 -0.0097 -0.0151 New CEO -0.1406 -0.0338 -0.1359 Interlock -1.2453 ** Dualrole 1.3362 *** Boardsize -0.0286 Year Fixed Effect YES YES YES Industry Fixed Effect YES YES YES Firm Random Effect YES YES YES

DIV/ASSETS 0.1205 ** 0.1215 ** 0.1351 **

# of observations 3,571 2,672 3,008

39

Panel B. Determinants of DE payments

(1) (2) (3) Intercept -5.473 *** -4.4176 ** 3.6928 ** MEDecile 0.2302 ** 0.3158 ** 0.2266 * ROA 2.0531 0.5515 1.1645 LEV -0.9292 -0.5893 -0.255 V/A -0.6807 *** -0.6961 *** -0.7532 *** Cashcomp 0 0 0 Totcomp 0 0 0 Tenure 0.0154 0.019 0.0111 New CEO -0.2709 -0.1983 -0.1642 Interlock -0.9054 * Dualrole 0.6159 ** Boardsize -0.045 Year Fixed Effect YES YES YES Industry Fixed Effect YES YES YES Firm Random Effect YES YES YES

DIV/ASSETS 0.3023 *** 0.3178 *** 0.3335 ***

# of observations 3,571 2,672 3,008

40

Table 4. Event Study Results The table below presents the statistics of abnormal returns for the DER announcements. The events being examined are three types of initial disclosures: announcing that a firm allows CEO’s to receive dividend equivalents on their unearned shares, or announcing that a firm has never paid, no longer pays, or will not pay dividend equivalents on CEO’s unearned shares, or announcing that a firm has paid dividend equivalents on CEO’s unearned shares during the year covered by the proxy statement. The event period is from day -5 to day +1. We use the Fama-French-Carter four-factor model.

# of observations Mean CAR (day -5 to day

+1)

Initial disclosure of DER policy 93 -0.70% ** Initial disclosure of no DER policy 179 0.77% ** Initial disclosure of DE payment 49 -1.15% **

41

Table 5. Regression Estimates of the Relationship between DER Policies and DE Payments to Stock Performance

Ordinary least squares regression of sample companies’ monthly stock returns. The sample includes S&P 500 firms between 1993 and 2012. The dependent variable is the raw monthly stock return minus the risk-free rate. The principal explanatory variable is an indicator for whether the company pays dividends on executives’ unearned shares (mainly unearned performance shares, unearned performance share units and shares covered by unexercised stock options). Other explanatory variables include the Fama and French (1993) factors: market excess return, HML, SMB, and the Carhart Momentum Factor, MOM. Explanatory variables also include the Bebchuck Entrenchment index, a dummy variable for firms that pay dividends, and a dummy variable for firms that were in the S&P500 in 1993. Standard errors are robust to heteroskedasticity. Panel A summarizes results for the DER policy, while Panel B contains results for DE payment.

Panel A. Results for DER Policy

Estimate Estimate Estimate Estimate

Intercept 0.4974 *** 0.6982 *** 1.2479 *** 1.0452 *** MKT 1.0204 *** 0.9805 *** 1.0217 *** 1.0218 *** HML 0.2005 *** 0.2131 *** 0.1991 *** 0.2004 *** SMB 0.1398 *** 0.0814 *** 0.1374 *** 0.1382 *** MOM -0.1199 *** -0.1181 *** -0.1197 *** -0.1187 *** DER_POLICY -0.2359 ** -0.2744 ** -0.0672 * -0.1732 ** E-index -0.0764 ** DIVDUM -0.8654 *** SP500_93 -0.7146 *** Adjusted R-squared 0.2089 0.2005 0.2100 0.2098 Number of Obs. 61,250 40,048 61,278 61,250

(Table 5, Panel B follows)

42

(Table 5 continued)

Panel B. Results for DE Payment

Estimate Estimate Estimate Estimate

Intercept 0.4848 *** 0.7832 *** 1.3463 *** 1.0603 *** MKT 1.0207 *** 0.9811 *** 1.0218 *** 1.0220 *** HML 0.2007 *** 0.2133 *** 0.1993 *** 0.2006 *** SMB 0.1396 *** 0.0812 *** 0.1374 *** 0.1381 *** MOM -0.1196 *** -0.1177 *** -0.1196 *** -0.1185 *** DE_PAYMENT -0.2170 * -0.1484 * -0.0392 * -0.1366 * E-index -0.0799 ** DIVDUM -0.8824 ** SP500_93 -0.7204 *** Adjusted R-squared 0.2089 0.2004 0.2100 0.2098 Number of Obs. 61,250 40,048 61,278 51,250

43

Table 6. Logistic Regressions to Determine the Effect of DER Policy on Whether a Firm Pays Dividends

The dependent variable is DIVDUM, which indicates whether a firm pays dividend in a year. Time, Industry, and firm’s fixed effects are dummies. Construction of the Average Year Sample is described in detail in the text, but it generally reflects the firm observations from the most typical years. The First Year Sample is constructed by choosing the first year observation when the firm begins to pay DEs, or when the firm stops paying DEs. To be included in the full sample, the firm-year observations must have positive net income and positive earnings before interests and taxes. The variables are described in Appendix B-2.

Firm Fixed

Effects

Average Year

Sample

First Year

Sample

Intercept -4.4004 *** -6.1110 *** 7.5713 DER_POLICY 1.3439 ** 0.7380 ** 1.3473 ** MEDecile 0.7281 *** 0.8411 *** 0.6351 *** V/A -0.4312 *** -0.6756 *** -0.5139 ** dA/A -1.6938 ** -2.7643 ** -4.7067 *** FCF/A 3.5768 ** 5.0199 * 1.7575 LEV -3.7848 *** 2.0293 ** 0.4592 ROA -2.0718 7.0171 * 8.8155 ** Odds Ratio of DER_POLICY 3.834 2.092 3.847 Time Fixed Effects YES YES YES Industry Fixed Effects YES YES YES Firm Fixed Effects YES # of Observations 5,442 420 420

% Concordant 99.3 90.0 94.4

44

Table 7. Regressions of Dividend Payout Measures on DE_PAYMENT

The regressions in Panel A use the full sample, the regressions in Panel B use the sample with observations from the year 2007 and 2008 excluded, the regressions in Panel C use the Average Year subsample, and the regressions in Panel D use the First Year subsample. Construction of the Average Year Sample is described in the text, but it generally reflects the firm observations from the most typical years. The First Year Sample is constructed by choosing the first year observation when the firm begins to pay DEs, or when the firm stops paying DEs. To be included in the full sample, the firm-year observations must have positive net income and positive earnings before interests and taxes. The response variables are described in Appendix B-2.

Panel A. Full Sample

(1)

DIV/ASSETS

(2)

DIV/OPERATING

INCOME

(3)

DIV/EBIT

(4)

DIV/NI

Intercept 3.2447 *** 27.8239 *** 60.0023 *** 70.0333 *** DE_PAYMENT 0.2031 ** 2.1756 * 7.0553 ** 7.9804 MEDecile -0.0796 * 0.0994 -2.2394 ** 0.8196 V/A 0.1196 ** 1.4359 ** 3.5313 *** 7.3810 *** dA/A -2.5907 *** -26.2548 *** -38.6729 *** -36.3266 ** FCF/A 4.2553 *** -57.0835 *** -60.6349 ** 162.0600 *** LEV 1.0905 *** -5.5670 -0.0116 -4.8574 ROA 6.1498 *** 11.0734 -75.2718 ** -718.9400 *** Time Fixed Effects YES YES YES YES Industry Fixed Effects YES YES YES YES Firm Random Effects YES YES YES YES R-squared 0.3693 0.0430 0.0153 0.0652 # of Obs. 3759 3759 3759 3759

(Table 7, Panel B follows)

45

(Table 7 continued)

Panel B. 2007-2008 excluded

(1)

DIV/ASSETS

(2)

DIV/OPERATING

INCOME

(3)

DIV/EBIT

(4)

DIV/NI

Intercept 3.3526 *** 32.8452 *** 59.5681 *** 70.9152 ** DE_PAYMENT 0.2129 * 2.6537 ** 8.0534 ** 2.2115 MEDecile -0.0877 ** -0.7360 * -2.1658 ** 1.1421 V/A 0.1438 *** 0.5675 3.6129 *** 8.5136 *** dA/A -2.1216 *** -13.9171 *** -39.0620 *** -31.1439 ** FCF/A 3.4940 *** -58.2044 *** -55.4954 ** 170.5500 *** LEV 0.8936 ** -0.0066 2.2484 -5.3974 ROA 6.5117 *** 34.8692 *** -79.8995 ** -788.9100 *** Time Fixed Effects YES YES YES YES Industry Fixed Effects YES YES YES YES Firm Random Effects YES YES YES YES R-squared 0.3988 0.0904 0.0180 0.0664 # of Obs. 3277 3263 3219 3173

(Table 7, Panel C follows)

46

(Table 7 continued)

Panel C. Average Year Sample

(1)

DIV/ASSETS

(2)

DIV/OPERATING

INCOME

(3)

DIV/EBIT

(4)

DIV/NI

Intercept -1.2954 10.5106 31.5555 93.5063 ** DE_PAYMENT 0.3409 6.2922 ** 19.6652 ** 0.3562 MEDecile 0.2415 * 0.7549 -0.8731 -1.1278 V/A 0.7331 *** 5.0991 ** 14.4915 ** 14.3036 ** dA/A -3.6240 ** -19.8472 * -56.3792 -88.2995 ** FCF/A 4.1732 -135.2700 *** -526.0100 *** 122.9000 * LEV 1.7615 ** 12.1109 * 56.3674 ** 38.3443 * ROA 19.8180 *** 159.4200 *** 581.0600 *** -469.4700 *** Time Fixed Effects YES YES YES YES Industry Fixed Effects YES YES YES YES Firm Random Effects NO NO NO NO R-squared 0.5335 0.2090 0.2297 0.1091 # of Obs. 272 271 267 266

(Table 7, Panel D follows)

47

(Table 7 continued)

Panel D. First Year Sample

(1)

DIV/ASSETS

(2)

DIV/OPERATING

INCOME

(3)

DIV/EBIT

(4)

DIV/NI

Intercept 2.6138 29.0086 32.9157 247.8200 DE_PAYMENT 0.3625 6.4771 * 4.6786 110.9000 * MEDecile 0.1669 0.5170 1.1381 13.3850 V/A -0.0327 -3.7165 -1.5960 30.1608 dA/A -13.9647 *** -73.6213 *** -114.1600 ** -21.7981 FCF/A 0.6559 -94.0133 ** -235.7300 ** 430.7600 LEV 0.9277 7.7511 33.8441 -67.8039 ROA 41.0061 *** 314.0100 *** 430.9200 *** -1833.3400 ** Time Fixed Effects YES YES YES YES Industry Fixed Effects YES YES YES YES Firm Random Effects NO NO NO NO R-squared 0.3796 0.1840 0.1137 0.0337 # of Obs. 272 272 270 267

48

Table 8. Regressions to Determine the Impact of DE Payments on the Amount of Stock

Repurchases The regressions in this table test the effect of DERs on repurchases of common shares using the full sample. The response variable is the dollar volume of repurchases scaled by prior year market value of equity. We use the pooled regression as well as the year-fixed effect regression models. Variable descriptions are in Appendix B-2.

Pooled Regression Fixed Effects Model Intercept -0.3545 *** -0.3577 *** -0.3758 *** -0.3734 *** DER_POLICY -0.0178 *** -0.0205 *** DER_PAYMENT -0.0206 *** -0.0227 *** CF2 -0.0067 * -0.0063 * -0.0083 ** -0.0079 ** CR 0.0899 *** 0.0911 *** 0.0907 *** 0.0940 *** V/A 0.0135 *** 0.0132 *** 0.0132 *** 0.0129 *** DIV/ASSETS 0.2216 ** 0.2264 ** 0.2309 ** 0.2338 ** SIZE 0.0528 *** 0.0527 *** 0.0520 *** 0.0520 *** RET 0.0028 * 0.0026 * 0.0035 ** 0.0033 ** LEVER -0.0527 *** -0.0522 *** -0.0439 *** -0.0431 *** OPTIONS 0.1452 ** 0.1527 ** 0.1962 ** 0.1918 ** TAKEOVER 0.1086 *** 0.1086 *** 0.0867 *** 0.0873 ***

49

Table 9. Effect of DERs on Excess Cash Holding

We test the effect of DERs on firms’ excess cash holding. We use an out of sample approach to estimate the excess cash, use it as the dependent variable, and then regress it on determinants of firms’ cash holding and DER dummies. The dependent variable excess cash is defined in two ways. The DeAngelo, DeAngelo and Stulz (DDS) excess cash measure defines the normal level as the industry median after controlling for firm size and book-to-market. The Opler, Pinkowitz, Stulz, and Williamson (OPSW) measure is the error term of a regression of the log of cash-to-net asset ratio on various explanatory factors. Variable definitions are in Appendix B-2.

DeAngelo, DeAngelo and Stulz Measure OPSW Measure (1) (2) (3) (4) Intercept -0.0483 *** -0.0539 *** -0.3092 ** -0.3399 ** DER_policy -0.0030 ** -0.4271 *** DER_payment -0.0088 ** -0.4331 *** V/A -0.0703 *** -0.0894 *** 0.0083 ** 0.0083 ** ME -0.0006 * -0.0006 * -0.0074 -0.0105 CF 0.2995 *** 0.2986 *** 2.6271 *** 2.5723 *** NWC -0.0927 *** -0.0928 *** -0.4626 *** -0.4588 *** R&D 0.3700 *** 0.3689 *** 4.1920 *** 4.1907 *** CF_VOL 0.0845 *** 0.0846 *** -0.6192 *** -0.6205 *** LEV 0.1653 *** 0.1647 *** -0.5475 *** -0.5328 *** CAPEX 0.0252 0.0246 -1.6231 *** -1.5464 *** REGULATION -0.0113 -0.0104 -0.1704 -0.1743 DIV/ASSETS 0.2260 *** 0.2342 *** 5.2112 *** 5.0860 *** R-Squared 0.3198 0.3010 0.1869 0.1789 # of Obs. 5,195 5,195 5,277 5,277

50

Figure 1. Cumulative Abnormal Returns First-Time Disclosure of DER Policy or DE

Payment

A. DER Policy Announcement

B. First-time DE Payments

-3.00%

-2.00%

-1.00%

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%

-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

CA

R

Event Day

Negative Disclosures

Positive Disclosures

-3.00%

-2.50%

-2.00%

-1.50%

-1.00%

-0.50%

0.00%

0.50%

-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

CA

R

Event Day

51

Figure 2. Portion of Cash Rich, Cash Median, and Cash Poor Firms vs. DER Policy

The results in this figure are based on using either the DeAngelo, DeAngelo, and Stulz (DDS) (Panel A), or Opler, Pinkowitz, Stulz, Williams (OPWS) (Panel B) methodologies for identifying normal cash. Each year, the top one-third of all eligible firms in the Compustat data base are classified as cash-rich, the middle third as cash-median, and the bottom third as cash-poor. We then compute the ratio for each DER policy adopter for each year starting with 10 years prior to adoption to 10 years afterwards. The percentages in the graphs are the percentages of DER policy adopters classified as cash-rich, cash-median, and cash-poor.

A. DDS measure of normal cash

B. OPSW measure of normal cash

0%10%20%30%40%50%60%70%80%90%

100%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Cash rich firms Cash median firms Cash poor firms

0%10%20%30%40%50%60%70%80%90%

100%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Cash rich firms Cash median firms Cash poor firms

52

Figure 3. Portion of Cash Rich, Cash Median, and Cash Poor Firms vs. DE Payment

The graphs in Figure 3 are constructed in the same manner as those in Figure 2, except that the event date is the date of initiation of DE payments.

A. DDS measure of normal cash

B. OPSW measure of normal cash

0%10%20%30%40%50%60%70%80%90%

100%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Cash rich firms Cash median firms Cash poor firms

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%100%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Cash rich firms Cash median firms Cash poor firms