Payout policy, agency conflicts, and corporate governance*

Kose John†

New York University

Anzhela Knyazeva

New York University

This version: March 2006

Abstract We examine the effect of corporate governance on the use of dividends and repurchases, composition of payout, and inclusion of a pre-commitment feature in the payout policy. We find that dividend payout is decreasing in governance quality. Corporate governance has a mixed effect on repurchases: good external governance prompts managers to repurchase more and to use repurchases as the only form of payout, but good internal governance lowers the need for repurchases. High agency costs due to weak corporate governance increase the likelihood of having cash distributions and decrease the reliance on discretionary payouts. The use of dividend pre-commitment in the payout policy serves as a substitute for strong corporate governance. JEL classification: G30, G34, G35 Keywords: Corporate Governance, Agency Conflicts, Payout Policy, Dividends, Repurchases, Pre-commitment.

* We are grateful to William Greene, Eli Ofek, David Yermack, and Bernard Yeung for helpful comments and suggestions. All errors and omissions are ours. Anzhela Knyazeva acknowledges the financial support of Stern School of Business, New York University. † Kose John, Stern School of Business, New York University, 44 W. 4th St., New York, NY 10012; kjohn@stern.nyu.edu. Anzhela Knyazeva, Stern School of Business, New York University, 44 W. 4th St., New York, NY 10012; akniazev@stern.nyu.edu.

Payout policy, agency conflicts, and corporate governance

This version: February 2006

Abstract

We examine the effect of corporate governance on the use of dividends and repurchases,

composition of payout, and inclusion of a pre-commitment feature in the payout policy. We

find that dividend payout is decreasing in governance quality. Corporate governance has a

mixed effect on repurchases: good external governance prompts managers to repurchase

more and to use repurchases as the only form of payout, but good internal governance lowers

the need for repurchases. High agency costs due to bad corporate governance increase the

likelihood of having cash distributions and decrease the reliance on discretionary payouts.

The use of dividend pre-commitment in the payout policy serves as a substitute for strong

corporate governance.

JEL classification: G30, G34, G35 Keywords: Corporate Governance, Agency Conflicts, Payout Policy, Dividends, Repurchases, Pre-commitment.

1

1. Introduction

Extensive empirical evidence shows that many firms pay dividends despite their cost, but a

growing fraction of firms is observed substituting stock repurchases for dividends. This paper

examines the effect of corporate governance on the use of dividends and repurchases,

composition of payout, and inclusion of a pre-commitment feature in the payout policy. We

propose an agency-driven explanation for the choice between dividends and repurchases and

provide empirical support for the hypothesized relation between corporate governance quality

and corporate payout.

We argue that dividend policy is a pre-commitment device that can be adopted to reduce

agency costs and raise firm value. A promise to distribute cash on a regular basis serves to

mitigate the manager-shareholder agency conflict. While dividends can increase firm value by

bonding the manager, they are associated with tax costs. Dividend pre-commitment also causes

the manager to bypass some positive-NPV projects or obtain costly external financing in the

periods of low cash flow. Repurchases allow ex post adjustments to the payout policy, thus

enabling the firm to take advantage of most of its investment opportunities. In exchange for

greater flexibility, repurchases do not represent an ex ante pre-commitment and are much less

effective in resolving the agency conflict. Poor corporate governance exacerbates the agency

costs, shifting the trade-off in favor of higher dividends and greater reliance on dividend pre-

commitment. Extending the dividends-driven analysis of corporate payout, we formulate and test

complementary hypotheses for share repurchases and total payout.

The free cash flow theory (Jensen, 1986) does not distinguish between the use of

different forms of payout to mitigate agency costs. By proposing a governance interpretation of

2

the design of payout policy, choice of dividends and repurchases, and the dividends-repurchases

tradeoff, we contribute to the existing agency and dividends literatures.

Our paper relates to the existing work on corporate payout. Easterbrook (1984) proposes

that dividends reduce the agency costs of free cash flow and minimize suboptimal managerial

behavior. The free cash flow theory can be extended to form predictions about dividends and

governance. Since good governance limits the potential for suboptimal managerial behavior, the

agency costs and the cash distributions required to offset them are lower. Optimal payout policy

design aimed at maximizing firm value would therefore indicate a negative relation between

governance quality and the level of payout. More generally, our study is related to Harford,

Mansi, and Maxwell (2004) and Dittmar and Mahrt-Smith (2005) that find a positive relation

between governance and the level of cash holdings driven by quicker dissipation of excess cash

reserves in the presence of entrenchment. Although dividends are shown to respond to low

investment opportunities, ownership structure, and CEO compensation, there is a lack of

empirical evidence on the role of governance in determination of dividends.

The counter-argument would suggest that ex post deviations from the optimal payout

policy can occur due to costly firing of CEOs and that dividends are decreasing in the degree of

managerial entrenchment. Examining differences in shareholder rights protection around the

world, LaPorta et al. (2000) find that dividends are lower in countries with weaker investor

protection. Consistent with it, cash holdings of firms are decreasing in the level of investor

protection (seem e.g., Dittmar, Mahrt-Smith, and Servaes, 2003). Our results are based on the US

sample and show the negative effect of governance on the level of dividends. Besides dividends,

we examine repurchases, total payout, the choice between dividends and repurchases and the

composition of payout.

3

Our work differs from the existing literature (for a survey of work on payout policy see,

e.g. Allen and Michaely, 2002; Baker, Powell, and Veit, 2002) in addressing several unanswered

questions about payout policy from the perspective of agency theory and governance quality.

The agency view does not explain the choice of repurchases over dividends. From the free cash

flow perspective, any form of distribution of excess cash to shareholders addresses the agency

issue. We propose and test an agency explanation for the dividends-repurchases tradeoff that

relies on the use of pre-commitment to dividends and conduct a systematic investigation of the

relation between corporate governance, different forms, incidence, levels, and composition of

firm payout. We arrive at the following empirical results.

First, we study the effect of good corporate governance on the incidence and level of

dividend payout. We adopt the indicator of external corporate governance that proxies for the

susceptibility of the manager to the market of corporate control (external governance). In

addition to the G index from Gompers, Ishii, and Metrick (2003), we employ an exogenous

measure of takeover vulnerability that aggregates several provisions from state antitakeover

laws. Controlling for investment opportunities, financial flexibility, firm size, and information

effects, greater susceptibility to takeovers - both at the firm and at the state level - lowers the

propensity to pay dividends and the likelihood of being a dividend payer. Good internal

corporate governance is also associated with lower dividends and lower incidence of dividend

payout.

Second, we analyze the impact of good governance on share repurchases. We contend

that stock repurchases, being a more flexible instrument of cash distributions, are largely a

discretionary tool of corporate payout rather than a form of pre-commitment. Existing empirical

evidence suggests that repurchases are more irregular than dividends and primarily serve to

4

distribute large temporary windfalls of cash. We find that managers in firms with strong external

governance are more likely to repurchase and on average will repurchase more. Absence of

antitakeover protections encourages managers to dispense cash to shareholders in a more

effective manner. Controlling for other factors, managers are more likely to make discretionary

payouts if they are vulnerable to the corporate control market. On the other hand, we find that, as

in the case of dividends, good internal governance reduces the incidence of repurchases. The

results on internal governance for the level of repurchases are weaker, but we find that the

inclusion of other proxies for monitoring does not affect the positive sign on the external

governance index.

We rely on this evidence to motivate our third set of results, on the structure and type of

corporate payout. A combination of good external and internal governance mechanisms lowers

the potential and incentives for suboptimal ex post managerial behavior. As a result, ex ante

design of payout policy can place a greater weight on less costly discretionary type of payout and

exclude pre-commitment to dividends. Firms with stronger external and internal governance are

more likely to rely solely on repurchases in their payout. We find strong empirical support for

the prediction that better governed firms gear their payout towards repurchases.

The remainder of the paper is organized as follows. Section II summarizes our

hypotheses. Section III describes the sample, variables, and methodology. Section IV presents

the empirical results and robustness checks. Section V concludes and discusses our results.

5

II. Corporate payout and governance: testable hypotheses

Governance and dividends

Below we formulate empirical hypotheses that relate governance quality to firm dividend

and repurchase behavior and payout policy design. Agency theory provides a general framework

for the role of dividends as a method of mitigating agency costs (Easterbrook, 1984). Dividends

reduce the amount of suboptimal investment, impose additional monitoring by forcing the

manager to address the external financing market, and increase managerial risk-taking (by

replacing leverage, dividends lowers the expected loss of human capital due to bankruptcy). The

free cash flow theory has found empirical support (see, e.g. Lang and Litzenberger, 1989; Smith

and Watts, 1992; Gaver and Gaver, 1993).

Our study of governance and payout is related to prior work that examined the effects of

ownership structure on payout policy. The existing empirical evidence is mixed. Rozeff (1982)

and Hu and Kumar (2004) find that firms with low insider ownership pay higher dividends. On

the other hand, Fenn and Liang (2001) find that payout is increasing in managerial ownership for

firms with high agency costs and Nam, Wang, and Zhang (2004) find a positive association

between post-tax cut dividend increases and managerial ownership. Amihud and Li (2002) find a

negative relation between institutional ownership and dividends focusing on the informational

aspects of dividends. Zeckhauser and Pound (1990) argue that institutional investors substitute

for the monitoring and signaling role of dividends. In contrast, Short, Zhang, and Keasey (2002)

find a strong positive relation. Controlling for the endogeneity of institutional ownership,

Grinstein and Michaely (2001) find no significant effect on cash distributions; however,

institutional investors prefer firms that pay lower dividends (but avoid non-payers) and

repurchase shares regularly. Gugler (2003) finds that dividends are higher for state-owned firms

6

than for family-owned firms using an Austrian sample. Hu and Kumar (2004) find that longer

CEO tenure, smaller ownership, larger cash compensation, and fewer stock options as well as

board independence and the absence of a large shareholder are associated with higher dividends,

but the results for total payout are less significant.

We use the free cash flow argument as the starting point and specifically focus on the role

of governance in determination of dividends controlling for other firm characteristics. Better

corporate governance induces more optimal use of firm resources by managers and lowers the

benefit of dividend use. On the other hand, dividends impose tax costs on the payer firm. Regular

dividend payments also incur the cost of forgone positive-NPV projects or the added cost of

raising external financing to fund them when internally generated cash is insufficient. Since

dividends are costly and better quality of corporate governance lowers the agency costs of free

cash flow, we expect the following to hold:

D1. Firms with better governance pay lower dividends.

Differently from the existing work, we focus on the effect of corporate governance on

payout policy of US firms and distinguish between the effects of external and internal corporate

governance mechanisms after controlling for miscellaneous dividend determinants, including

ownership structure and compensation.

The hypothesis above hinges on the assumption that payout policy maximizes firm value.

The manager either adheres to the ex ante set optimal policy or chooses the level of payout that

mitigates agency costs. However, in the presence of weak governance, firing is more costly, so

entrenched managers can make financial decisions in their own self-interest (e.g. Zwiebel, 1996).

In the case of leverage, entrenched managers take on less debt (Berger, Ofek, and Yermack,

1997). Brown and Caylor (2004) study governance and six performance measures, including

7

dividend yield, and find a positive association between an aggregate governance index and

dividend payout, but their analysis uses few controls related to dividends and employs OLS

framework. In LaPorta et al. (2000), country-level shareholder rights protection index is

associated with lower dividends and lower sensitivity of dividends to investment opportunities,

supporting the opposite hypothesis:

D2. Good governance increases dividends.

In an environment with strong legal protection, however, we expect that the possibility

for the entrenched manager to avoid dividends will be limited and the relation between dividends

and firm-level governance will not be positive. Since performance-based firing is relatively more

common and the stock market reacts strongly to dividend cuts and omissions, regular dividend

payouts can serve as a condition of staying with the firm with a weak governance structure. The

hypothesized negative relation between governance and dividends remains our main prediction.

Governance, repurchases, and total payout

We also examine the effect of governance on repurchases and total payout. On the one

hand, the free cash flow theory does not distinguish between dividends and repurchases, so the

substitution argument underlying our main dividends hypothesis can be extended to repurchases

and total payout. Although the propensity to pay dividends has decreased (Fama and French,

2001), with stock repurchases often replacing dividends (Grullon and Michaely, 2002),

repurchase behavior is consistent with the free cash flow explanation (see, e.g., Bagwell and

Shoven, 1989; Dittmar, 1999; Heron and Lie, 2000; Grullon and Michaely, 2004). Applying this

to our case, entrenched managers can be distributing more cash overall, either through

repurchases or dividends:

R1. Firms with better governance repurchase less.

8

T1. Firms with better governance distribute less cash to their shareholders.

On the other hand, compared to dividends, repurchases have greater flexibility and can be

viewed as largely discretionary managerial payouts. Ex post deviations from the previously

announced repurchase policy are more common and less consequential for the stock price than

dividend cuts, so entrenchment can be associated with a suboptimal repurchase policy.

Managerial entrenchment is more likely to affect ex post behavior with respect to irregular cash

distributions (repurchases) because the need to make them is less verifiable and no perceived

pre-commitment is made. Hence, when investment opportunities within the work and an excess

of cash is present, an entrenched manager is less likely to use the discretion to repurchase in the

shareholders’ interest. Instead, excess cash can be retained in the firm or channeled into

inefficient investment projects (Harford, Mansi, and Maxwell, 2004; Dittmar and Mahrt-Smith,

2005).

In the agency context, corporate payout serves to avoid wasting excess cash on negative-

NPV projects. Large cash excesses due to accumulation of free cash flow or large cash windfalls

can be undesirable even if investment opportunities are expected to improve in the future

because diminishing returns limit the portion of excess cash that can be optimally invested.

Shareholders will prefer a well governed manager to distribute cash if they can realize a higher

return on investment outside the firm. In the absence of a pre-commitment, better governed

managers can be expected to distribute cash more willingly than entrenched managers,

controlling for other firm characteristics.

The threat of dismissal can cause managers to make strategic share purchases to reduce

the monitoring pressure. Fluck (1999) shows theoretically that, without being afforded the

9

protection of entrenchment, managers will voluntarily limit private benefits of control when they

cannot effectively counter disciplinary actions of shareholders.

The presence of good governance lowers the manager’s potential gain from avoiding a

discretionary payout when firm investment opportunities are poor. Effective monitoring

mechanisms decrease the potential for deriving private benefits of free cash flow. Further, if the

manager who is not entrenched avoids the needed payout but the firm realizes a low return,

market value will fall over time, which increases the threat of firing through regular CEO

turnover or a takeover. Also, if the manager expects firm investment opportunities to improve,

she will use available excess cash to repurchase firm shares but will not sell any of her own

shares so that, when profits increase next period, the manager will hold a higher relative stake in

the firm (Ofer and Thakor, 1988). For an entrenched manager, the ensuing loss of private

benefits of free cash flow and increase in risk will reduce such gains from a firm repurchase

program. In other words, undervalued firm shares can be the best way to invest excess cash when

present investment opportunities are poor. We have discussed the motivation for our alternative

prediction concerning repurchases:

R2. Repurchases are increasing in the quality of governance

If the R2 hypothesis is supported, the sign of the effect of governance on total payout will

depend on the relative role of dividends versus repurchases. If dividends have a high share and

the D1 hypothesis holds, the relation between total payout and governance will be dominated by

the negative effect of governance on dividends. Otherwise, the sign may be reversed. We discuss

our hypotheses about payout composition later in this section.

External versus internal governance and payout

10

The discussion of managerial entrenchment and payout should take into account the

properties of the monitoring mechanisms in place. To our knowledge, existing work has not

distinguished between external and internal governance in the analysis of payout policy. Internal

governance can provide more accurate information about managerial behavior, firm investment

opportunities, and future firm prospects. It is therefore possible to tie the threat of dismissal to

instances of inefficient investment with a smaller margin of error and managers face a lesser

need to make strategic share purchases to counter the disciplinary pressure. Undertaking efficient

investment contributes to greater job security in the absence of entrenchment, reducing the need

for repurchases (hypothesis R1).

External monitoring by the corporate control market, on the other hand, is based on the

observed firm performance and provides a very limited look into the optimality of managerial

decisions. Firm performance can be affected by factors unrelated to suboptimal managerial

behavior, but a drop in market value makes the firm more susceptible to a takeover and increases

the risk of firing for the manager. Managers that are not afforded the protection from the

corporate control market will use discretionary payouts more frequently to sustain firm value

during periods of low investment opportunities. Payout can be used directly to thwart a pending

takeover threat that would have otherwise been mitigated by the firm’s antitakeover provisions.

Defensive payouts encourage existing shareholders to reject the raider’s offer and can discourage

the raider from proceeding by raising the current market value of the firm. Although dividends

lower the risk of a takeover (Dickerson, Gibson, and Tsakalatos, 1998), repurchases deter

takeovers more effectively by targeting shareholders with the lowest valuations and shifting up

the distribution of firm values among not tendering shareholders (Bagwell, 1991). Strong

external corporate governance is more likely to drive managers to repurchase (hypothesis R2).

11

At the same time, both good internal and external governance mechanisms lower the

optimal level of dividends (hypothesis D1) since they diminish the potential for suboptimal

managerial investment and the agency costs of free cash flow.

Governance, dividend pre-commitment, and payout policy design

We have separately considered dividends and repurchases as they relate to managerial

incentives arising from governance. Adding to previous work, we also examine the effect of

governance on the structure and design of payout policy and form hypotheses about payout

composition and policy type. One of the key payout policy decisions is whether to pre-commit to

dividends to address the agency conflict. The disadvantage of dividends stems from its tax status

and the lack of flexibility compared to repurchases. Dividend cuts are costly and rare, so if

investment opportunities improve in the future, they are more likely to be ignored or financed

externally rather than through downward adjustments to dividend policy. Discretionary payouts

through stock buybacks allow the manager to use more accurate information that arrives every

period in making payout decisions. However, regular dividend payments is a more effective

solution to severe agency manager-shareholder conflicts since entrenched managers will apply

excess cash to inefficient projects in the absence of pre-commitment.

Managers in firms with high agency costs will pre-commit to regular payments to raise

firm value and continue payouts to shareholders to avoid disciplinary action. Managers in firms

with better governance will avoid pre-commitment because the expected loss of firm value due

to the agency costs of free cash flow. When a firm with good governance faces cash excesses,

the manager will either store them in them for future investment or distribute them to

shareholders through discretionary payouts. We hypothesize that pre-commitment to dividends

will substitute for strong (external or internal) governance, yielding the following predictions:

12

P1. Firms with bad governance are more likely to pre-commit to dividends and use

dividends as the only form of payout; firms with good governance will be more likely to use

repurchases as the only form of payout.

P2. The share of dividends in total firm payout is increasing in governance quality.

The above hypotheses also imply that total payout will tend to be negatively associated

with governance even if repurchases are positively associated with governance because

entrenched managers tend to rely on dividends more and pay higher dividends.

Contributing to the existing literature, our analysis disaggregates corporate payout into

dividends and repurchases because of their different role in mitigating the agency costs of free

cash flow. We build separate agency explanations of managerial incentives to distribute cash

through these forms of payout and identify a potential difference in the effects of external and

internal governance. We also account for the pre-commitment feature of dividends and address

the issue of the effect of governance on the design of payout policy.

Governance and payout when agency costs are high

The last part of our analysis examines the effect of governance on payout in subsamples

of firms with low and high agency costs of free cash flow. We expect that the following

prediction holds:

A1. The effect of governance on payout is stronger in the presence of high free cash flow.

III. Data

Sample selection

Data on firm dividend policies and performance are obtained from Compustat Industrial

Annual for 1992-2003. Data for year 1991 is used when lags are needed to construct the

variables. The sample includes all (active and inactive) Compustat firms except those in the

13

regulated utilities and financial sectors (SIC codes 4000-4999 and 6000-6999). We further

eliminate observations with book value of total assets less than $20 mln., firms incorporated

outside the United States, and LBOs (Compustat stock code 4). We require that the data on

payout, firm-level control variables, and corporate governance be available. Our main sample

has 9,464 observations (6,974 obs. with available data on past and present share of repurchases

in total payout). The variables are explained below.

Variables

Our dependent variables include payout ratio, composition of payout, and payout policy

type dummies constructed on firm-year basis. Payout ratio is defined as the percentage of cash

dividends, share repurchases, or total payout in the book value of total assets of the firm. We use

the ratio based on the market value of the firm as a robustness check. The payout ratio variables

are censored at 0 from below. We compute the share of repurchases in total payout to measure

the composition of payout. The variable is censored at 0 from below and at 1 from above and is

defined only for payer firms. The presence of dividends, repurchases, or total payout is captured

by dummy variables equal to 1 if a positive cash distribution to shareholders is made and 0

otherwise. We also construct dummy variables for payout policy type: dividends as the only

form of payout, a combination of dividends and repurchases, and repurchases as the only form of

payout.

We use a set of explanatory variables to control for other determinants of corporate

payout that can be correlated with good governance. The lag of dividends is expected to enter

with a positive coefficient (Lintner, 1956). We include the lag of the dependent variable in all

payout analyses, but we expect to see the highest coefficient in the case of dividends. We include

the percentage of EBITDA in total assets to account for the positive relation between cash flow

14

and payout. Investment and growth opportunities are expected to enter with a negative sign since

growing firms with more positive NPV projects face less severe potential agency problems and a

higher opportunity cost of payout. Investment opportunities are measured by the growth of sales,

market-to-book ratio. As a robustness check, we use investment to sales ratio. The ratio of R&D

to sales captures investment opportunities as well as the information asymmetry about the firm.

We use share of intangible assets and the logarithm of the number of one-year-ahead analyst

earnings forecasts (obtained from I/B/E/S database) for robustness. Size is positively correlated

with dividends (Fama and French, 2001), so we control for the log of the book value of total

assets. Firm risk also increases the cost of sustaining dividends, so we expect a negative sign on

the annual standard deviation of excess return. We expect firms faced with a higher share of

income taxes in EBIT to have lower dividends. In robustness checks, we control for measures of

stock performance (average excess returns) and liquidity (bid-ask spreads) based on CRSP data.

Leverage (the ratio of the book value of debt to the book value of total assets) is expected to be

correlated with lower dividends and is used as a robustness check. A high level of debt

substitutes for the pre-commitment and takeover deterrence role of dividends, increases the cost

of sustaining dividends, and may involve covenants that restrict dividend payments.

The share of repurchases in total payout and the likelihood of adopting a repurchases-

only payout policy will also be influenced by various firm characteristics, including taxes,

financial flexibility demands (Jagannathan, Stephens, and Weisbach, 2000), information

asymmetry, option-based executive compensation (Fenn and Liang, 2001; Hu and Kumar, 2004),

high leverage, and the quality of investment opportunities. For instance, firms with better

investment and growth opportunities, R&D, high risk, tax burden, and leverage face a higher cost

of dividends. Larger firms can experience more severe agency problems or have a lower cost of

15

sustaining dividends. Managers awarded option grants are not compensated for the loss of value

of their portfolio due to dividends, so they prefer repurchases (Fenn and Liang, 2001).

Our governance variables are described below. Following Ishii, Gompers, and Metrick

(2003), we use the Investor Responsibility Research Center (IRRC) Governance database to

construct our measure of external governance. We rescale the variable by subtracting the G index

from 24 (the maximum possible value) so that higher values indicate better governance (as in

Cremers and Nair, 2005). The IRRC database covers the years 1990, 1993, 1995, 1998, 2000,

and 2002 of our sample period. We fill in the values for the gap years using the adjacent value or

the average of the two adjacent values. For robustness, we use the sample that contains only the

original IRRC years. A possible reservation regarding the use of the G index is the concern about

endogeneity because firm antitakeover provisions could be influenced by the payout policy.

Pugh and Jahera (1997) present univariate evidence on dividend increases following the adoption

of state antitakeover laws, so we construct the second external governance index using only

provisions of state antitakeover laws and use it as a robustness measure.

We construct several proxies for internal corporate governance. Yermack (1996)

identifies smaller boards as a factor in higher firm valuations. We obtain board size from the

IRRC Directors database. The data for early years (1992-1995) is reset to the 1996 level,

following the assumption that governance mechanisms are relatively stable over time. The

largest institutional blockholding variable is based on quarterly institutional ownership data from

Thomson Financial 13f filings. Quarterly data are averaged for each firm-year. Institutional

holdings data is reset to 0 if missing (as in Cremers and Nair, 2005). We construct an aggregate

index of internal governance by sorting the firms into quartiles by the largest institutional

16

blockholding, board size, and the proportion of independent directors on board (IRRC Directors),

described in Appendix A.

The stake of the CEO in the firm and the ratio of the Black-Scholes value of new stock

option grants to the firm market value are obtained from Execucomp, starting from 1992. We

expect a negative effect of stock options on dividends and a positive effect on repurchases

regressions. The sign of ownership effect is ambiguous.

Following Frank and Goyal (2003), we replace missing values for certain financial items

with zeros. All financial firm characteristics are winsorized at 0.5% of the sample distribution to

address the potential problem with extreme observations. We also winsorize internal governance

and analyst forecast variables. The external governance index is a categorical variable and is not

winsorized. Appendix A describes the construction of the variables. Descriptive statistics and

univariate correlations of the main variables are presented in Table 1, Panel A.

[insert Table 1]

Methodology

We begin with two-sided two-sample t-tests to assess the differences in the payout policy

of firms with governance quality above and below the sample median. Multivariate analysis

controls for additional firm characteristics. We employ Tobit regressions for payout ratio and

payout composition to avoid biases from using OLS estimation with censored variables. The

likelihood of being a payer or having a particular type of payout is predicted with the Logit

model. All regressions are performed in the panel data format and include year dummies2.

2 Inclusion of Fama-French industry dummies instead of firm effects does not affect the principal results; to save the space, this robustness check is not reported (but it is available from the authors upon request).

17

IV. Results

The correlations in Table 1, Panel B show that dividends, repurchases, and total payout

are positively associated with cash flow and negatively associated with growth opportunities and

firm risk. Dividends and total payout are also negatively associated with R&D, consistent with

the free cash flow intuition. Dividends are strongly negatively associated with both external and

internal governance measures. Repurchases are positively associated with external corporate

governance. Total payout is strongly negatively correlated with internal corporate governance.

The effect of stock options is negative for dividends, as could be expected.

Below we provide the results of t-tests of means. We divide the observations into

subsamples based on the quality of governance.

[insert TABLE 2]

Firms with better internal and external governance have lower dividends. Approximately

72% of firms with bad governance pay dividends, with 25% fewer observations in the good

external governance subsample and 15% fewer observations in the good internal governance

subsample recording positive dividends. The incidence of any form of payout is lower among

firms with good governance (10%-13%). These firms are also more likely to use repurchases as

the only form of payout. Firms with good governance have a higher share of repurchases in total

payout. Univariate evidence suggests that overall payout is decreasing in governance quality,

with some exceptions in the case of repurchases. The same relation preserves for the subsample

of firms with positive total payout. However, we have not controlled for other firm

characteristics that can be correlated with corporate governance as well as payout. To address the

issue, we use Tobit regressions using a set of firm-level controls described in the previous

section.

18

[insert TABLE 3]

We perform regressions separately for dividends and repurchases.

In dividends regressions, both external corporate governance index and internal corporate

governance index have strongly significant negative coefficients. Firms with poor governance

mechanisms are not able to curtail suboptimal managerial investment through corporate control

market or active internal monitoring, yielding high agency costs of free cash flow. The optimal

level of dividend payments is higher than it would be for a firm with good corporate governance.

In the presence of sophisticated shareholder rights laws, disclosure standards, and developed

financial markets applicable to the US sample, entrenched managers comply with dividend pre-

commitment and sustain overall higher dividends. Our results show a negative relation between

dividend payout and governance (consistent with hypothesis D1), regardless of monitoring type.

The results are different in the case of repurchases. External corporate governance is

positively associated with repurchases (hypothesis R2). Managers susceptible to the corporate

control market undertake higher discretionary payouts, controlling for other firm characteristics.

The internal corporate governance index is insignificant, with the substitution and entrenchment

effects possibly offsetting each other (hypotheses R1 and R2). We conclude that the type of

governance mechanism is important for determining the level of discretionary cash distributions

undertaken by the manager. A lack of takeover defenses increases the risk of firing when firm

value is low, so the manager is prompted to undertake discretionary cash distributions to respond

to the threat of dismissal.

Coefficients on the lag of payout ratio enter with positive and strongly significant signs,

with dividends showing a higher degree of persistence than repurchases. Consistent with the free

cash flow theory predictions, ROA enters with a positive sign and growth opportunities enter

with a negative sign, both for dividends and repurchases. The market-to-book ratio is

19

insignificant in most specifications, likely because our proxies for growth and investment

opportunities are correlated3. R&D has a negative effect on dividends and a positive effect on

repurchases (the latter could be due to informational considerations). Riskier firms have lower

dividends and repurchases due to a higher demand for flexibility and the resulting higher cost of

payout. Firm size also has the predicted positive sign, both for dividends and repurchases. Taxes

have an insignificant effect on dividends (the effect is marginally significant and positive for

repurchases). Incentive alignment mechanisms significantly affect payout policy. Managers

receiving more stock option grants pay lower dividends and undertake a higher level of

repurchases, consistent with prior work. CEO ownership is positively associated with dividend

payout, i.e. CEOs with a higher stake in the firm have a stronger financial interest in paying

dividends to shareholders. The effect of ownership on repurchases is insignificant.

The next table analyzes the level and composition of total payout in the Tobit framework

using the same set of controls.

[insert TABLE 4]

Opposite signs on external corporate governance and stock option grants variables in

dividends and repurchases regressions offset each other, resulting in a lack of significance of

these variables in the total payout case. However, internal corporate governance is negatively

correlated with total payout. We conclude that internal monitoring lowers the level of cash the

manager has to distribute to shareholders to address the free cash flow problem (hypothesis T1).

Total payout is less persistent than dividend payout or repurchases. The effect of growth

opportunities is negative. Firms with higher cash flow on average distribute more cash. The

positive effect of firm size and negative effect of risk preserve in the total payout case.

3 Market-to-book ratio is also potentially endogenous to payout because presence of dividends and repurchases affects firm market value.

20

Columns IV-VI of Table 4 use the share of repurchases in total payout as the target

variable. External governance has a positive effect on the share of repurchases (hypothesis P1).

Managers in companies with fewer antitakeover provisions distribute more cash through stock

repurchases. Internal corporate governance is insignificant, consistent with the negative sign

obtained both in dividends and repurchases regressions. Managers less entrenched by

antitakeover provisions are able to avoid the high cost of dividend pre-commitment and conduct

most cash distributions through repurchases.

One potential limitation of using the share of repurchases to characterize payout policy is

the potential for overstatement of the role of repurchases in the presence of mixed policies due to

the fact that repurchases tend to be larger than regular dividends. Below we use Logit analysis to

examine payout policy design and the presence of different payout components and their

combinations.

[insert TABLE 5]

Firms with good internal governance are less likely to distribute cash through any means,

which confirms the conjecture that active monitoring of the manager substitutes for payout in all

forms (hypotheses D1, R1, T1). Good external corporate governance is associated with lower

incidence of dividends (hypothesis D1), lower incidence of cash distributions in general

(hypothesis T1), and higher incidence of share repurchases (hypothesis R2). All of the effects are

statistically significant. We conclude that the likelihood of positive payout is higher in firms with

weak governance, but susceptibility to the corporate control market encourages managers to

repurchase more.

Consistent with expected signs on control variables, the incidence of dividends,

repurchases, or total payout is increasing in cash flow and firm size and decreasing in risk and

21

growth and investment opportunities. Stock option grants are associated with a lower probability

of having dividends, but CEO ownership increases the likelihood of dividends.

Columns IV-VI of Table 5 focus on the probability of specific types of payout policy.

The manager can use either dividends or repurchases as the only form of payout, adopt a mixed

policy with some combination of the two, or choose not to distribute cash at all. The probability

of adopting a mixed policy is decreasing in the quality of corporate governance. Repurchases are

more likely to be used as the only form of payout in the presence of good governance

(hypothesis P1). Firms with good external governance are less likely to use dividends as the only

form of payout. We conclude that firms with better monitoring tend to avoid pre-commitment to

dividends, consistent with our prediction.

The Logit results presented so far are based on the full sample of firms. Below we restrict

the sample to payer firms only (i.e. cases when total payout is positive). We consider pairs of

payout policy types to determine the effect of governance on the choice between mixed and

dividends-only policy, mixed and repurchases-only policy, and dividends-only and repurchases-

only policy.

[insert TABLE 6]

The use of dividends by payer firms is significantly negatively related to external

governance. The opposite holds true for repurchases: payer firms are more likely to use

repurchases if they have good external governance. Internal governance is less significant in the

case of payer firms.

Payer firms with better governance (internal and external) are more likely to use

repurchases as the only form of payout than to use dividends only, consistent with the pre-

commitment prediction (Column III). Firms are more likely to adopt a mixed policy rather than

22

use repurchases only when their governance (internal and external) is poor (Column IV).

Governance does not affect the choice between dividends as the only form of payout and a mixed

policy (Column V) for payer firms. These results lend further support to our pre-commitment

conjecture (hypothesis P2). Badly governed firms are more likely to pre-commit to dividends and

to avoid the use of repurchases as the only form of payout. However, the decision to use

repurchases in addition to dividend pre-commitment is influenced by other factors.

We conclude our main analyses with the consideration of subsamples by the severity of

agency costs. The high agency cost subsample includes firms with growth opportunities below or

at the sample median and cash flow above the sample median. Tobit regressions are repeated for

dividends and repurchases in each subsample. The variables used to define the subsamples are

omitted to avoid biased estimates.

[insert TABLE 7]

For the subsample with low free cash flow, internal governance is not significantly

related to dividend payout. Neither governance mechanism has a significant effect on

repurchases. The relation between governance and payout is significant in the high free cash

flow subsample: negative for dividends in the case of both internal and external governance,

positive for repurchases in the case of external governance, and negative for repurchases in the

case of internal governance. Governance is more important for the determination of payout when

the free cash flow problem is more severe (hypothesis A1).

Robustness checks

The results on payout and governance continue to hold after a number of robustness

checks.

23

One potential concern with the interpretation of external governance coefficients is the

endogeneity of governance quality due to inclusion of firm charter provisions. The agency

conflict can be mitigated simultaneously through an improvement in governance and pre-

commitment to dividends. Payout policy can influence governance quality. For instance, regular

dividends can weaken shareholder response if the manager attempts to increase the degree of

entrenchment by influencing the governance mechanisms in the firm. Managers can agree to a

reduction in the number of antitakeover provisions while continuing to repurchase actively as a

form of takeover deterrence.

To address this potential issue, we use the exogenous external governance measure based

only on state antitakeover law provisions (see, e.g., Garvey and Hanka, 1999; Pugh and Jahera,

1997). The correlation between the original external governance measure and the state law

measure is 0.30 and strongly statistically significant. As before, the state antitakeover laws index

yields strongly significant results for dividends (negative), repurchases (positive), and the share

of repurchases (positive).

[insert TABLE 8]

Columns I-IV of Table 9 also introduce alternative internal governance proxies that

control for the stake of the largest institutional blockholder and board size instead of the internal

governance index used previously. Institutional holdings and small board size are negatively

related to dividends, consistent with the previous results. Small boards are also associated with

lower overall payout and lower repurchases, but a higher share of repurchases. The signs and

significance are consistent with evidence presented in the main regressions.

In other robustness checks, we include additional controls for firm investment, intangible

assets, stock performance, bid-ask spread, analyst following, and leverage. The variables are

24

intended to account for informational characteristics of the firm and capital structure. The main

results do not change: governance quality is negatively related to dividends and overall payout,

but external governance determines the preference towards repurchases and a higher level of

repurchases.

[insert TABLE 9]

We further check our results for robustness using an alternative sample selection criterion

and a different definition of the payout ratio. Columns I-IV of Table 10 reproduce the results for

the sample of the original IRRC Governance years (without filling in the values for the gap

years). Coefficients on governance measures follow a similar pattern.

[insert TABLE 10]

The payout ratios used in the previous analysis were defined with respect to the book

value of total assets. Columns V-VII of Table 10 use the ratio of payout to the market value of

the firm to account for the potentially significant discrepancy between the book value and the

market value. For instance, repurchases can be tailored to the market value of the firm, so better

governed firms that realize market value gains will have overstated payout ratios according to the

book value measure that can be falsely attributed to the effect of good governance. The book

value and the market value measures have correlations of 0.79 for dividends, 0.81 for

repurchases, and 0.75 for total payout. The pattern of the signs and significance preserves.

VI. Conclusion This paper examines the effect of corporate governance on the use of dividends and

repurchases, composition of payout, and inclusion of a pre-commitment feature in the payout

policy. We find that the incidence of payout is significantly lower among firms with better

25

governance because these firms face lower agency costs of free cash flow. In the presence of

high agency costs, governance plays a greater role in determining corporate payout.

Dividends are more effective than repurchases at mitigating the agency costs of free cash

flow due to their pre-commitment nature and higher cost of deviations from the dividend policy.

Contrary to LaPorta et al. (2000), we find that firm dividend policies are a substitute for weak

corporate governance (at the internal and external level). In the US, where the legal environment

supports high overall standards of investor protection, entrenched managers follow through with

dividend pre-commitment. Firms with better governance tend to avoid the costs associated with

dividends in an attempt to achieve a more efficient investment policy.

The type of governance mechanism and its ability to detect suboptimal managerial

behavior is also important in understanding managerial incentives to make payouts, especially in

the case of discretionary cash distributions. The evidence on firm repurchase policy and

governance reflects a combination of the substitution and entrenchment predictions, with

significance attributed to the type of monitoring mechanism in place. The incidence and the level

of repurchases are higher for firms with better external governance. Managers susceptible to the

corporate control market repurchase shares to maintain firm valuation and address the higher

threat of dismissal. However, good internal corporate governance (board, institutional

blockholders) mitigates the asymmetry of information about managerial actions and reduces the

need for discretionary payouts. Managers in firms with strong internal monitoring are able to

sustain high market valuations by conducting efficient investment without resorting to costly

payout.

Finally, accounting for the pre-commitment feature of dividends helps us to better

understand the structure of payout and payout policy design. Entrenched managers are more

26

likely to include pre-commitment to dividends in their payout policy whereas well governed

managers are more likely to use repurchases as the only form of payout. We find that the use of

dividend pre-commitment mitigates the manager – shareholder agency conflict and substitutes

for good governance.

Our work has implications for future research. At the theoretical level, both governance

quality and payout policy can minimize agency costs. The tradeoff between costs of setting up

governance mechanisms and maintaining payout can depend on firm risk. Riskier firms bear

higher costs of pre-commitment through dividend policy or leverage and could prefer good

governance instead. At the same time, the level of firm risk could be endogenously determined,

with better governed managers both choosing higher risk and avoiding pre-commitment made

more costly by unpredictable cash flows. Understanding the causality in the relation between

good governance and the choice of payout, leverage, and firm risk is an important issue for

future research.

27

References

Allen, F., Michaely, R., 2003. Payout policy. In: Constantinides, G., Harris, M., and Stulz, R.

(Eds.), Handbook of Economics and Finance. Elsevier - North-Holland, Amsterdam, 337-

429.

Amihud, Y., Li, K., 2002. The declining information content of dividend announcements and the

effect of institutional holdings. Unpublished working paper. New York University.

Bagwell, L.-S., 1991. Share repurchase and takeover deterrence. The RAND Journal of

Economics 22, 72-88.

Bagwell, L., Shoven, J., 1989. Cash distributions to shareholders. Journal of Economic

Perspectives 3, 129-140.

Baker, H., Powell, G., Veit, T., 2002. Revisiting the dividend puzzle: do all of the pieces now

fit? Review of Financial Studies 11, 241-261.

Berger, P., Ofek, E., Yermack, D., 1997. Managerial entrenchment and capital structure

decisions. Journal of Finance 52, 1411-1438.

Brown, L.D., Caylor, M.L., 2004. Corporate governance and firm performance. Unpublished

working paper. Georgia State University.

Cremers, M., Nair, V., 2005. Governance mechanisms and equity prices. Journal of Finance 60,

forthcoming.

Dickerson, A., Gibson, H., Tsakalotos, E., 1998. Takeover risk and dividend strategy: a study of

UK firms. The Journal of Industrial Economics 46, 281-300.

Dittmar, A., 1999. Why do firms repurchase stock? Journal of Business 73, 331-355.

Dittmar, A., Mahrt-Smith, J., 2005. Corporate governance and the value of cash holdings.

Unpublished working paper. University of Michigan.

Dittmar, A., Mahrt-Smith, J., and Servaes, H., 2003. International corporate governance and

corporate cash holdings. Journal of Financial and Quantitative Analysis 38, 111-133.

Easterbrook, F., 1984. Two agency-cost explanations of dividends. American Economic Review

74, 650-659.

Fama, E., French, K., 2001. Disappearing dividends: changing firm characteristics or lower

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

28

Fenn, G.W., Liang, N., 2001. Corporate payout policy and managerial stock incentives. Journal

of Financial Economics 60, 45-72.

Fluck, Z., 1999. The dynamics of the management-shareholder conflict. The Review of Financial

Studies 12, 379-404.

Frank, M., Goyal, V., 2003. Testing the pecking order theory of capital structure. Journal of

Financial Economics 67, 217-248.

Garvey, G.T., Hanka, G., 1999. Capital structure and corporate control: The effect of

antitakeover statutes on firm leverage. The Journal of Finance 54, 519-546.

Gaver, J., Gaver, K., 1993. Additional evidence on the association between the investment

opportunity set and corporate financing, dividend, and compensation policies. Journal of

Accounting and Economics 16, 125-160.

Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly

Journal of Economics 118(1), 107-155.

Grinstein, Y., Michaely, R, 2001. Institutional holdings and payout policy. Unpublished working

paper. Cornell University.

Grullon, G., Michaely, R., 2002. Dividends, share repurchases, and the substitution hypothesis.

The Journal of Finance 57, 1649-1684.

Grullon, G., Michaely, R., 2004. The information content of share repurchase programs. The

Journal of Finance 59, 651-680.

Gugler, K., 2003. Corporate governance, dividend payout policy, and the interrelation between

dividends, R&D, and capital investment. Journal of Banking and Finance 27, 1297-1321.

Harford, J., Mansi, S., Maxwell, W., 2004. Corporate governance and firm cash holdings.

Journal of Finance (forthcoming).

Heron, R., Lie, E., 2000. On the use of poison pills and defensive payouts by targets of hostile

takeovers. Unpublished working paper. Indiana University.

Hu, A., Kumar, P., 2004. Managerial entrenchment and payout policy. Journal of Financial and

Quantitative Analysis 39, 759-790.

Jagannathan, M., Stephens, C.P., Weisbach, M.S., 2000. Financial flexibility and the choice

between dividends and stock repurchases. Journal of Financial Economics 57, 355-384.

Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American

Economic Review 76, 323-329.

29

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 2000. Agency problems and

dividend policies around the world. The Journal of Finance 55, 1-33.

Lang, L., Litzenberger, R., 1989. Dividend announcements: Cash flow signaling vs. free cash

flow hypothesis. Journal of Financial Economics 24, 181-191.

Lintner, J., 1956. Distribution of incomes of corporations among dividends, retained earnings

and taxes. American Economic Review 61, 97–113.

Nam, J., Wang, J., Zhang, G., 2004. The impact of the dividend tax cut and managerial stock

holdings on corporate dividend policy. Unpublished working paper. Pace University.

Ofer, A. R., Thakor, A.V., 1987. A theory of stock price responses to alternative corporate cash

disbursement methods: stock repurchases and dividends. The Journal of Finance 42, 365-

394.

Pugh, W., Jahera, J., 1997. State antitakeover legislation and firm financial policy. Managerial

and Decision Economics 18, 681-692.

Rozeff, M. S., 1982. Growth, beta and agency costs as determinants of dividend payout ratios.

Journal of Financial Research 3, 249-259.

Short, H., Zhang, H., and Keasey, K., 2002. The link between dividend policy and institutional

ownership. Journal of Corporate Finance 8, 105-122.

Smith, C., Watts, R., 1992. The investment opportunity set and corporate financing, dividend,

and compensation policies. Journal of Financial Economics 32, 263-292.

Yermack, D., 1996. High market valuation of companies with a small board of directors. Journal

of Financial Economics 40, 185-211.

Zeckhauser, R., Pound, J., 1990. Are large shareholders effective monitors? An investigation on

share ownership and corporate Performance. In: Hubbard, G. (Ed.) Asymmetric

Information, Corporate Finance, and Investment, University of Chicago Press, Chicago,

149–180.

Zwiebel, J., 1996. Dynamic capital structure under managerial entrenchment. The American

Economic Review 86, 1197-1215.

30

Appendix A. Variable definitions

Target variables

DIV_PAYOUT Ratio of cash dividends (#127a) to total assets (#6), multiplied by 100; censored from below at 0. Source: Compustat.

REP_PAYOUT Ratio of purchase of common and preferred stock (#115a) to total assets (#6), multiplied by 100; censored from below at 0. Source: Compustat.

TOT_PAYOUT

= DIV_PAYOUT + REP_PAYOUT; censored from below at 0. Source: Compustat.

REP_TOT Ratio of REP_PAYOUT to TOT_PAYOUT, multiplied by 100; defined only for firms with positive TOT_PAYOUT; censored from below at 0 and from above at 100. Source: Compustat.

DIV

Dummy variable equal to 1 if #127 is positive; equal to 0 otherwisea. Source: Compustat.

REP

Dummy variable equal to 1 if #115 is positive; equal to 0 otherwisea. Source: Compustat.

TOT

Dummy variable equal to 1 if #127 or #115 is positive; equal to 0 otherwise a. Source: Compustat.

DIV_REP

Dummy variable equal to 1 if DIV=1 and REP=1; equal to 0 otherwise. Source: Compustat.

REP_ONLY

Dummy variable equal to 1 if REP=1 and DIV=0; equal to 0 otherwise. Source: Compustat.

DIV_ONLY Dummy variable equal to 1 if DIV=1 and REP=0; equal to 0 otherwise. Source: Compustat.

Main explanatory variables

LAG_Y

Value of the target variable in the previous period. Source: Compustat.

CASH_FLOW Ratio of EBITDA (#13) to total assets (#6). Source: Compustat.

GROWTH

Relative change in net sales (#12) from the previous period. Source: Compustat.

MKTBOOK

Ratio of market value (#24*#25+#9a+#34) to total assets (#6). Source: Compustat.

SIZE

Natural log of book value of total assets (#6). Source: Compustat.

RISK Annual standard deviation of excess returns (holding period return net of value-weighted market return). Source: CRSP monthly stocks.

CEO_OWN Ratio of the aggregate number of shares held in the firm, excluding stock options (shrown, divided by 1000 to form mln.) to the number of common shares outstanding (shrsout, mln.), multiplied by 100. Source: Execucomp.

CEO_OPTIONS Ratio of the value of stock option grants to the CEO valued using S&P’s modified Black-Scholes methodology (blk_valu, divided by 1000 to form mln.) to the market value of the firm (#24*#25 + #9 + #34, mln.; #9 set to 0 if missinga), multiplied by 100. Source: Execucomp, Compustat.

TAXES

Ratio of total income taxes (#16) to EBIT (#178). Source: Compustat.

R&D

Ratio of research and development expenditure (#46a) to net sales (#12). Source: Compustat.

EXT_GOV (24 – G), where G is the Gompers-Ishii-Metrick (2003) index based on the sum of 24 antitakeover provisions in firm charter and state laws. Higher values indicate better external

31

governance. Source: IRRC Governanceb.

INT_GOV (12-I)/12, where I is the index of internal governance constructed as follows. For each year, firms are sorted into quartiles by INST (largest institutional holding; source: Thomson Financial 13f filings), BOARD (maximum number of directors on board in the sample minus actual number of directors on board divided by the maximum number of directors on board in the sample), and INDEP (the proportion of independent directors on board. Each observation is assigned a value N=1,2,3,4 for the Nth quartile by INST; by BOARD; by INDEP. The I index equals the sum of assigned values for three characteristics. Higher values indicate better internal governance (e.g., see Brown and Caylor, 2004; Yermack, 1996).

Robustness variables

DIV_PAYOUT(II) Ratio of cash dividends (#127a) to market value (#24*#25 +#9a +#34), multiplied by 100; censored from below at 0. Source: Compustat.

REP_PAYOUT(II) Ratio of purchase of common and preferred stock (#115a) to market value (#24*#25+#9a+#34), multiplied by 100; censored from below at 0. Source: Compustat.

TOT_PAYOUT(II) = DIV_PAYOUT(II) + REP_PAYOUT(II); censored from below at 0. Source: Compustat.

INVESTMENT

Ratio of capital expenditure (#128a) to total assets (#6). Source: Compustat.

INTANGIBILITY

Ratio of intangible assets (#33a) to total assets (#6). Source: Compustat.

BID-ASK Annual average of the absolute value of the difference between bid and ask price. Source: CRSP monthly

ANALYSTS Natural log of the number of one-year-ahead analyst forecasts of EPS for the current fiscal year. Source: I/B/E/S.

RETURN Annual average of excess return (holding period return net of value-weighted market return). Source: CRSP monthly.

LEVERAGE Ratio of book value of current and long-term debt (#9+#34) to total assets (#6). Source: Compustat.

EXT_GOV(II) (6 – S)/6, where S is the state laws index constructed by adding 1 for the presence of each of the following state laws: recapture of profits – antigreenmail law; business combination law; cash-out law; director’s duties law; fair price law; control share acquisition law. Source: IRRC Governanceb. Higher values indicate better governance.

INST The percentage held by the largest institutional holder, replaced with 0 if missing. Source: Thomson Financial 13f filings.

BOARD (20, maximum number of directors on board in the sample) - actual number of directors on board) / 20. Source: IRRC Directorsc.

a Compustat items #33, #115, #127, #128 set to 0 if missing, following Frank and Goyal (2003). Item #46 (R&D) set to 0 if missing. b Investor Responsibility Research Center (IRRC) Governance data set. Gap years (1992, 1994, 1996, 1997, 1999, 2001, 2003) were filled in using the closest value or the average of the two closest values for IRRC years (1993, 1995, 1998, 2000, 2002). c Investor Responsibility Research Center (IRRC) Directors data set. Early years (1992-1995) filled in using the values for 1996, the earliest available year.

32

Table 1. Descriptive statistics The sample includes all Compustat Industrial Annual firms excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The sample period is 1992-2003. The variables are defined in Appendix A.

Panel A. Summary statistics

Obs. Mean Median 25th percentile

75th percentile

Std. dev.

Target variables

DIV_PAYOUT 9,464 1.305 0.676 0.000 1.949 1.831 REP_PAYOUT 9,464 2.409 0.128 0.000 2.666 4.996 TOT_PAYOUT 9,464 3.848 1.799 0.229 4.706 6.014 REP_TOT 6,974 0.432 0.394 0.000 0.836 0.399 DIV 9,464 0.617 1.000 0.000 1.000 0.486 REP 9,464 0.561 1.000 0.000 1.000 0.496 TOT 9,464 0.792 1.000 1.000 1.000 0.406 DIV_REP 9,464 0.386 0.000 0.000 1.000 0.487 DIV_ONLY 9,464 0.231 0.000 0.000 0.000 0.422 REP_ONLY 9,464 0.176 0.000 0.000 0.000 0.381 Explanatory variables CASH_FLOW 9,464 14.749 14.814 10.058 19.882 10.257 GROWTH 9,464 0.106 0.074 -0.004 0.176 0.262 MKTBOOK 9,464 1.838 1.330 0.936 2.100 1.641 SIZE 9,464 7.229 7.085 6.207 8.109 1.413 RISK 9,464 11.302 9.728 6.920 13.835 6.534 CEO_OWN 9,464 0.027 0.003 0.001 0.016 0.061 CEO_OPTIONS 9,464 0.127 0.039 0.003 0.123 0.284 TAXES 9,464 0.225 0.298 0.191 0.363 0.652 R&D 9,464 0.055 0.006 0.000 0.046 0.212 EXT_GOV 9,464 14.718 15.000 13.000 17.000 2.702 INT_GOV 9,464 0.619 0.583 0.500 0.750 0.159 Robustness variables DIV_PAYOUT(II) 9,464 0.898 0.522 0.000 1.512 1.104 REP_PAYOUT(II) 9,464 1.420 0.098 0.000 1.742 2.792 TOT_PAYOUT(II) 9,464 2.397 1.464 0.157 3.166 3.357 INVESTMENT 9,464 0.077 0.045 0.027 0.076 0.129 INTANGIBILITY 9,464 0.110 0.052 0.000 0.173 0.142 BID-ASK 9,464 4.672 3.625 2.369 5.582 4.169 ANALYSTS 9,464 2.142 2.242 1.641 2.763 0.829 RETURN 9,464 0.407 0.302 -1.649 2.279 3.742 LEVERAGE 9,464 0.228 0.221 0.086 0.337 0.173 EXT_GOV(II) 9,464 0.715 0.833 0.667 0.833 0.209 INST 9,464 0.089 0.084 0.060 0.111 0.044 BOARD 9,464 0.531 0.550 0.450 0.650 0.130

33

Panel B. Correlations among variables

The sample includes all Compustat Industrial Annual firms excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The sample period is 1992-2003. The variables are defined in Appendix A. Pairwise correlations that are significant at 5% are in bold face. The correlations are based on 9,464 obs. (on 6,974 for REP_TOT).

DIV_PAYOUT

DIV

_PA

YO

UT

RE

P_PA

YO

UT

TO

T_P

AY

OU

T

RE

P_T

OT

CA

SH_F

LO

W

GR

OW

TH

MK

TB

OO

K

SIZ

E

RIS

K

CE

O_O

WN

CE

O_O

PTIO

NS

TA

XE

S

R&

D

EX

T_G

OV

REP_PAYOUT 0.14 TOT_PAYOUT 0.45 0.89 REP_TOT -0.29 0.54 0.37 CASH_FLOW 0.32 0.31 0.38 0.15 GROWTH -0.12 -0.07 -0.10 -0.03 0.16 MKTBOOK 0.12 0.22 0.24 0.17 0.32 0.25 SIZE 0.18 -0.02 0.02 -0.12 0.09 0.00 -0.06 RISK -0.36 -0.04 -0.15 0.22 -0.35 0.03 0.09 -0.28 CEO_OWN -0.03 -0.01 -0.01 0.04 0.05 0.04 0.06 -0.20 0.04 CEO_OPTIONS -0.20 0.00 -0.06 0.18 -0.24 -0.06 -0.06 -0.22 0.32 -0.02 TAXES 0.06 0.06 0.07 0.01 0.13 0.04 0.06 0.03 -0.08 0.02 -0.07 R&D -0.10 -0.02 -0.05 0.22 -0.43 0.02 0.17 -0.14 0.26 -0.04 0.18 -0.04 EXT_GOV -0.15 0.04 -0.01 0.19 -0.01 0.09 0.16 -0.20 0.20 0.19 0.09 -0.01 0.10 INT_GOV -0.22 -0.01 -0.07 0.10 -0.12 0.01 -0.02 -0.28 0.19 -0.07 0.13 -0.05 0.08 0.04

34 34

Table 2. Payout policy and governance: univariate analysis. Two-sample t-tests performed on payout policy characteristics based on the quality of governance. The firm belongs in the ‘Good’ governance subsample if its governance index value exceeds the sample median; ‘Bad’ otherwise. The measures EXT_GOV and INT_GOV are used separately. The null hypothesis is that the difference of the means is zero. The alternative hypothesis is that the difference of means is not zero (two-sided). The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

EXTERNAL GOVERNANCE

INTERNAL GOVERNANCE

GOOD BAD DIFF.

GOOD BAD DIFF.

ALL FIRMS

3,860 obs. 5,604 obs. 6,414 obs. 3,050 obs.

DIV_PAYOUT

0.948 1.552 -0.604** 1.064 1.812 -0.748**

REP_PAYOUT

2.558 2.307 0.251** 2.399 2.430 -0.030

TOT_PAYOUT

3.670 3.972 -0.302** 3.613 4.343 -0.730**

REP_TOT

0.519 0.384 0.135** 0.454 0.391 0.064**

DIV

0.466 0.721 -0.255** 0.567 0.721 -0.154**

REP

0.542 0.575 -0.033** 0.537 0.613 -0.077**

TOT

0.714 0.847 -0.133** 0.762 0.856 -0.094**

DIV_REP

0.294 0.449 -0.154** 0.342 0.478 -0.136**

REP_ONLY

0.248 0.126 0.122** 0.195 0.135 0.059**

DIV_ONLY

0.172 0.272 -0.100** 0.225 0.243 -0.018†

PAYERS ONLY (TOT=1)

2,453 obs. 4,521 obs. 4,501 obs. 2,473 obs.

DIV_PAYOUT

1.479 1.913 -0.433** 1.508 2.219 -0.710**

REP_PAYOUT

3.553 2.710 0.843** 3.113 2.814 0.299*

REP_TOT

0.519 0.384 0.135** 0.454 0.391 0.064**

DIV

0.718 0.882 -0.164** 0.794 0.880 -0.086**

REP

0.741 0.670 0.071** 0.689 0.706 -0.017

DIV_REP

0.459 0.552 -0.093** 0.483 0.586 -0.103**

REP_ONLY

0.282 0.118 0.164** 0.206 0.120 0.086**

DIV_ONLY

0.259 0.330 -0.071** 0.311 0.294 0.017

** significant at 1%; * significant at 5%; † significant at 10%

35 35

Table 3. Payout policy and governance: dividends and repurchases. Tobit analysis. Panel data Tobit regressions of payout ratio for dividends and repurchases (censoring at 0) on firm characteristics and governance. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. Standard errors are in the parentheses. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A. Payout through dividends Payout through repurchases

I

II

III

IV

V

VI

LAG_Y 0.946 ** 0.945 ** 0.942 ** 0.515 ** 0.514 ** 0.515 ** (.006) (.006) (.006) (.015) (.015) (.015) CASH_FLOW 0.016 ** 0.013 ** 0.013 ** 0.218 ** 0.228 ** 0.228 ** (.002) (.002) (.002) (.010) (.010) (.010) GROWTH -0.586 ** -0.604 ** -0.601 ** -5.068 ** -5.060 ** -5.050 ** (.051) (.051) (.051) (.328) (.327) (.327) MKTBOOK -0.043 ** -0.014 -0.016 0.094 † 0.036 0.035 (.009) (.010) (.010) (.052) (.055) (.055) SIZE 0.072 ** 0.073 ** 0.062 ** 0.161 ** 0.168 ** 0.147 * (.008) (.008) (.008) (.057) (.057) (.059) RISK -0.039 ** -0.036 ** -0.035 ** -0.062 ** -0.066 ** -0.064 ** (.003) (.003) (.003) (.015) (.015) (.015) CEO_OWN 0.684 ** 0.574 ** 0.465 ** -1.263 -1.034 -1.248 (.175) (.176) (.177) (1.223) (1.221) (1.231) CEO_OPTIONS -0.282 ** -0.245 ** -0.237 ** 0.917 ** 0.886 ** 0.897 ** (.055) (.055) (.055) (.272) (.272) (.272) TAXES 0.021 0.018 0.216 † 0.212 † (.017) (.017) (.111) (.111) R&D -1.283 ** -1.216 ** 1.957 ** 1.966 ** (.197) (.196) (.529) (.525) EXT_GOV -0.030 ** -0.029 ** -0.030 ** 0.079 ** 0.075 ** 0.074 ** (.004) (.004) (.004) (.029) (.029) (.029) INT_GOV -0.379 ** -0.682 (.070) (.481)

Number of obs. 9,464 9,464 9,464 9,464 9,464 9,464 AIC 18,301 18,254 18,226 38,612 38,606 38,606 BIC 18,458 18,425 18,405 38,769 38,778 38,785 Log-likelihood -9,128 -9,103 -9,088 -19,284 -19,279 -19,278

** significant at 1%; * significant at 5%; † significant at 10%

36 36

Table 4. Payout policy and governance: overall level and composition of payout. Tobit analysis.

Panel data Tobit regressions of total payout ratio (censoring at 0) and share of repurchases in total payout (censoring at 0 and 1) on firm characteristics and governance, with censoring at 0. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. Standard errors are in the parentheses. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

TOT_PAYOUT REP_TOT

I

II

III

IV

V

VI

LAG_Y 0.451 ** 0.451 ** 0.451 ** 1.157 ** 1.126 ** 1.126 ** (.012) (.012) (.012) (.018) (.018) (.018) CASH_FLOW 0.198 ** 0.200 ** 0.199 ** 0.006 ** 0.009 ** 0.009 ** (.009) (.009) (.009) (.001) (.001) (.001) GROWTH -4.380 ** -4.384 ** -4.361 ** -0.121 ** -0.119 ** -0.119 ** (.260) (.259) (.260) (.030) (.030) (.030) MKTBOOK 0.044 0.033 0.032 0.007 -0.018 ** -0.018 ** (.048) (.050) (.050) (.005) (.006) (.006) SIZE 0.201 ** 0.202 ** 0.165 ** -0.001 -0.003 -0.002 (.061) (.061) (.062) (.004) (.004) (.004) RISK -0.072 ** -0.073 ** -0.071 ** 0.004 ** 0.002 0.002 (.013) (.013) (.013) (.002) (.002) (.002) CEO_OWN -1.071 -1.032 -1.323 -0.083 0.007 0.011 (1.191) (1.191) (1.196) (.099) (.099) (.100) CEO_OPTIONS 0.270 0.270 0.284 0.252 ** 0.245 ** 0.244 ** (.232) (.232) (.232) (.034) (.034) (.034) TAXES 0.142 0.137 0.002 0.002 (.087) (.087) (.010) (.010) R&D 0.428 0.461 1.225 ** 1.221 (.595) (.592) (.131) (.132) EXT_GOV -0.038 -0.039 -0.040 0.015 ** 0.014 ** 0.014 ** (.030) (.030) (.030) (.002) (.002) (.002) INT_GOV -1.195 ** 0.015 (.447) (.039)

Number of obs. 9,464 9,464 9,464 6,974 6,974 6,974 AIC 48,018 48,019 48,014 8,560 8,467 8,469 BIC 48,183 48,198 48,200 8,711 8,632 8,640 Log-likelihood -23,986 -23,985 -23,981 -4,258 -4,210 -4,210

** significant at 1%; * significant at 5%; † significant at 10%

37 37

Table 5. Choice of payout policy and governance. Logit analysis (all firms) Panel data Logit regressions of the probability of having positive dividends, repurchases, payout, both dividends and repurchases, repurchases but not dividends, and dividends but not repurchases. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. Marginal effects are reported. Standard errors are in the parentheses. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

DIV REP TOT DIV_REP REP_ONLY DIV_ONLY

I

II

III

IV

V

VI

LAG_Y 0.914 ** 0.500 ** 0.488 ** 0.622 ** 0.350 ** 0.443 ** (.005) (.013) (.025) (.016) (.036) (.023) CASH_FLOW 0.005 ** 0.015 ** 0.004 ** 0.013 ** 0.002 ** -0.004 ** (.001) (.001) (.000) (.001) (.000) (.001) GROWTH -0.036 -0.274 ** -0.075 ** -0.265 ** -0.012 0.072 ** (.047) (.032) (.013) (.035) (.010) (.016) MKTBOOK -0.016 + -0.025 ** -0.010 ** -0.038 ** -0.004 + 0.002 (.009) (.006) (.003) (.007) (.002) (.004) SIZE 0.040 ** 0.029 ** 0.022 ** 0.043 ** -0.013 ** 4.8E-04 (.010) (.007) (.003) (.007) (.003) (.004) RISK -0.020 ** -0.008 ** -0.005 ** -0.018 ** 0.002 ** -0.003 ** (.002) (.002) (.001) (.002) (.000) (.001) CEO_OWN 0.533 ** 0.003 0.021 0.226 † -0.075 0.014 (.183) (.133) (.060) (.126) (.048) (.072) CEO_OPTIONS -0.122 ** 0.030 -0.022 † -0.064 † 0.015 * -0.084 ** (.046) (.027) (.011) (.035) (.007) (.022) TAXES 0.020 0.007 -0.002 0.017 -0.004 -0.010 † (.014) (.010) (.005) (.012) (.003) (.005) R&D -0.052 0.128 * -0.009 -0.143 4.8E-05 -0.619 ** (.154) (.051) (.027) (.153) (.018) (.086) EXT_GOV -0.016 ** 0.007 * -0.005 ** -0.010 ** 0.011 ** -0.008 ** (.005) (.003) (.002) (.003) (.001) (.002) INT_GOV -0.171 * -0.114 * -0.090 ** -0.180 ** 0.039 * -0.004 (.077) (.052) (.025) (.050) (.019) (.028)

Number of obs. 9,464 9,464 9,464 6,974 6,974 6,974 AIC 48,018 48,019 48,014 6,563 6,508 6,510 BIC 48,183 48,198 48,200 6,713 6,673 6,682 Log-likelihood -23,986 -23,985 -23,981 -3,259 -3,230 -3,230 ** significant at 1%; * significant at 5%; † significant at 10%

38 38

Table 6. Design of payout policy and governance. Logit analysis (payer firms) Panel data Logit regressions of the probability of a certain payout policy type among payers (TOT=1). Column I (Pr(DIV=1)): a policy including dividends vs. repurchases only. Column II (Pr(REP=1)): a policy including repurchases vs. dividends only. Column III (Pr(DIV=1), given that DIVREP=0): dividends only vs. repurchases only. Column IV (Pr(DIV=1) if REP=1): a mix of dividends and repurchases vs. repurchases only. Column V (Pr(REP=1), given that DIV=1): a mix of dividends and repurchases vs. dividends only. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. Marginal effects and standard errors are reported. The sample includes all Compustat Industrial Annual firms for 1992-2003 excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

Include DIV vs.

Only REP Include REP vs.

Only DIV Only DIV vs.

Only REP DIV & REP vs.

Only REP DIV & REP vs.

Only DIV I II III IV V LAG_Y 0.923 ** 0.525 ** 0.978 ** 0.933 ** 0.504 ** (.010) (.019) (.008) (.009) (.018) CASH_FLOW -1.1E-04 0.012 ** -0.010 ** 0.001 0.017 ** (.4E-03) (.001) (.004) (.001) (.002) GROWTH -0.012 -0.188 ** -0.039 -0.044 -0.285 ** (.011) (.031) (.103) (.030) (.043) MKTBOOK 4.6E-04 -0.023 ** 0.006 -0.001 -0.041 ** (.002) (.007) (.025) (.006) (.010) SIZE 0.002 0.013 * -0.026 0.006 0.031 ** (.002) (.006) (.023) (.005) (.008) RISK -0.002 ** -0.003 * -0.017 ** -0.006 ** -0.013 ** (.001) (.002) (.005) (.001) (.002) CEO_OWN 0.089 * 0.119 0.489 0.242 * 0.230 (.040) (.121) (.516) (.106) (.167) CEO_OPTIONS -0.029 ** 0.171 ** -0.425 ** -0.063 * 0.142 * (.010) (.040) (.106) (.026) (.059) TAXES 0.008 ** 0.010 0.068 ** 0.020 ** 0.021 (.002) (.010) (.024) (.006) (.014) R&D -0.035 1.207 ** -0.474 -0.062 1.085 ** (.041) (.170) (.495) (.108) (.275) EXT_GOV -0.002 * 0.010 ** -0.034 ** -0.007 ** 0.005 (.001) (.003) (.012) (.003) (.004) INT_GOV -0.033 † -0.030 -0.479 * -0.091 * -0.085 (.017) (.047) (.192) (.046) (.063) Number of obs. 6,974 6,974 3,350 4,848 5,750 AIC 1,126 5,635 637 992 5,211 BIC 1,298 5,806 790 1,147 5,377 Log-likelihood -538 -2,792 -294 -472 -2,581

** significant at 1%; * significant at 5%; † significant at 10%

39 39

Table 7. Payout policy and corporate governance: subsample analysis. Panel data Tobit regressions of payout ratio for dividends and repurchases (censoring at 0) on firm characteristics and governance. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. The criteria for the high free cash flow (FCF=1) subsample: GROWTH below sample median and CASH_FLOW above sample median; otherwise FCF=0. The variables used to define subsamples are excluded from the respective regressions. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

DIV_

PAYOUT REP_

PAYOUT Low FCF High FCF Low FCF High FCF LAG_Y 0.969 ** 0.926 ** 0.501 ** 0.634 ** (.011) (.008) (.028) (.018) CASH_FLOW GROWTH MKTBOOK 0.032 -0.006 1.126 ** 0.294 ** (.021) (.008) (.145) (.047) SIZE 0.057 ** 0.050 ** -0.146 0.071 (.018) (.009) (.136) (.061) RISK -0.053 ** -0.035 ** -0.045 -0.120 ** (.007) (.003) (.049) (.015) CEO_OWN -1.278 * -1.340 ** 6.002 -2.642 ** (.632) (.189) (4.279) (.613) CEO_OPTIONS 0.527 0.422 * -3.445 -0.165 (.385) (.197) (2.826) (1.297) TAXES -0.176 -0.269 ** 1.717 * 0.599 * (.147) (.058) (.840) (.275) R&D 0.325 ** -0.015 1.114 * 0.239 * (.058) (.018) (.450) (.108) EXT_GOV -0.031 ** -0.031 ** -0.021 0.094 ** (.009) (.004) (.066) (.030) INT_GOV -0.162 -0.450 ** 0.641 -1.042 * (.155) (.078) (1.136) (.509)

Number of obs. 1,936 7,528 1,936 7,528 AIC 4,637 13,616 10,751 28,104 BIC 4,770 13,775 10,879 28,264 Log-likelihood -2,294 -6,785 -5,353 -14,029

** significant at 1%; * significant at 5%; † significant at 10%

40 40

Table 8. Robustness checks: alternative governance measures Panel data Tobit regressions on firm characteristics and alternative measures of governance. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

DIV REP TOT REP_TP I II III IV LAG_Y 0.937 ** 0.512 ** 0.450 ** 1.130 ** (.006) (.015) (.012) (.018) CASH_FLOW 0.013 ** 0.228 ** 0.199 ** 0.009 ** (.002) (.010) (.009) (.001) GROWTH -0.593 ** -4.996 ** -4.327 ** -0.121 ** (.051) (.327) (.259) (.030) MKTBOOK -0.017 † 0.041 0.037 -0.017 ** (.010) (.055) (.050) (.006) SIZE 0.039 ** 0.077 0.050 0.003 (.009) (.065) (.067) (.005) RISK -0.035 ** -0.060 ** -0.067 ** 0.002 (.003) (.015) (.013) (.002) CEO_OWN 0.465 ** -0.443 -0.798 0.075 (.174) (1.213) (1.181) (.098) CEO_OPTIONS -0.227 ** 0.906 ** 0.319 0.237 ** (.054) (.272) (.232) (.034) TAXES 0.020 0.216 † 0.140 0.002 (.017) (.110) (.087) (.010) R&D -1.208 ** 1.985 ** 0.458 1.205 ** (.195) (.525) (.588) (.131) EXT_GOV(II) -0.172 ** 1.259 ** 0.261 0.095 ** (.048) (.360) (.387) (.027) INST -0.888 ** -1.528 -1.266 -0.041 (.238) (1.673) (1.523) (.133) BOARD -0.811 ** -1.450 * -3.531 ** 0.186 ** (.101) (.701) (.680) (.056) Number of Obs. 9,464 9,464 9,464 6,974 AIC 18,212 38,601 47,997 8,485 BIC 18,398 38,787 48,190 8,663 Log-Likelihood -9,080 -19,274 -23,971 -4,216

** significant at 1%; * significant at 5%; † significant at 10%

41 41

Table 9. Robustness checks: additional controls Panel data Tobit regressions on additional firm characteristics and governance. Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

DIV REP TOT REP_TP I II III IV LAG_Y 0.933 ** 0.509 ** 0.450 ** 1.115 ** (.006) (.015) (.012) (.018) CASH_FLOW 0.014 ** 0.226 ** 0.198 ** 0.008 ** (.002) (.010) (.009) (.001) GROWTH -0.557 ** -4.889 ** -4.190 ** -0.116 ** (.052) (.335) (.265) (.031) MKTBOOK 0.008 0.050 0.151 * -0.025 ** (.011) (.066) (.059) (.006) SIZE 0.140 ** 0.319 ** 0.385 ** -0.009 (.013) (.087) (.085) (.007) RISK -0.032 ** -0.051 ** -0.055 ** 0.002 (.003) (.016) (.013) (.002) CEO_OWN 0.260 -1.899 -1.858 -0.002 (.179) (1.235) (1.184) (.101) CEO_OPTIONS -0.221 ** 0.799 ** 0.145 0.250 ** (.055) (.274) (.233) (.035) TAXES 0.017 0.176 0.142 -0.002 (.017) (.111) (.087) (.010) R&D -1.251 ** 2.025 ** 0.565 1.042 ** (.202) (.583) (.625) (.132) INVESTMENT 0.006 -2.215 ** -2.452 ** -0.061 (.095) (.640) (.593) (.056) INTANGIBILITY -0.225 ** 0.511 -0.573 0.043 (.079) (.543) (.520) (.044) BID-ASK -0.008 * 0.006 -0.010 0.001 (.004) (.024) (.021) (.002) ANALYSTS -0.142 ** -0.169 -0.356 ** 0.032 ** (.020) (.134) (.124) (.011) RETURN -0.008 * -0.095 ** 0.110 ** -0.001 (.004) (.022) (.018) (.002) LEVERAGE -0.314 ** -3.391 ** -1.686 ** -0.236 ** (.074) (.492) (.459) (.043) EXT_GOV -0.032 ** 0.062 * -0.044 0.013 ** (.004) (.029) (.030) (.002) INT_GOV -0.367 ** -0.463 -0.980 * 0.027 (.070) (.483) (.444) (.040) Number of Obs. 9,464 9,464 9,464 6,974 AIC 18,161 38,532 47,950 8,434 BIC 18,383 38,754 48,179 8,647 Log-Likelihood -9,049 -19,235 -23,943 -4,186

** significant at 1%; * significant at 5%; † significant at 10%

42 42

Table 10. Robustness checks: alternative sample criteria and payout ratio measures Panel data Tobit regressions using the sample without fill-in for EXT_GOV (columns I-IV) and using the original sample with alternative target variables (columns V-VII). Firm effects are used. Year dummies and intercept are included in the regressions but omitted from the table. The sample includes all Compustat Industrial Annual firms for 1992-2003, excluding: firms incorporated outside the US; firms in financial and regulated utilities industries (SIC codes 6000-6999 and 4000-4999); firms with total assets (#6) less than 20 mln.; LBOs (stock code 4); firms with missing financial, compensation, or governance data. The variables are defined in Appendix A.

NO FILL-IN FOR EXT_GOV PAYOUT/MARKET VALUE

DIV_

PAYOUT REP_

PAYOUT TOT_

PAYOUT REP_TOT DIV_

PAYOUT(II) REP_

PAYOUT(II) TOT_

PAYOUT(II) I II III IV V VI VII LAG_Y 0.952 ** 0.575 ** 0.548 ** 1.087 ** 0.941 ** 0.404 ** 0.324 ** (.009) (.024) (.016) (.026) (.008) (.016) (.011) GROWTH -0.446 ** -4.590 ** -4.089 ** -0.119 ** -0.352 ** -2.988 ** -2.500 ** (.073) (.499) (.387) (.044) (.040) (.203) (.165) CASH_FLOW 0.011 ** 0.217 ** 0.179 ** 0.010 ** 0.011 ** 0.123 ** 0.095 ** (.002) (.017) (.014) (.001) (.001) (.006) (.006) MKTBOOK -0.022 -0.192 * -0.122 -0.043 ** -0.078 ** -0.362 ** -0.503 ** (.015) (.097) (.081) (.009) (.008) (.037) (.036) SIZE 0.069 ** 0.109 0.112 -0.001 0.040 ** 0.059 0.074 + (.013) (.089) (.079) (.007) (.006) (.036) (.042) RISK -0.027 ** -0.083 ** -0.090 ** 0.001 -0.029 ** -0.062 ** -0.070 ** (.004) (.023) (.019) (.002) (.002) (.009) (.008) R&D -0.662 ** 2.010 * 0.571 1.450 ** -0.716 ** 1.267 ** 0.157 (.244) (.829) (.763) (.197) (.152) (.304) (.380) CEO_OWN 0.592 * -1.095 -1.136 0.032 0.449 ** -0.367 0.231 (.267) (1.870) (1.591) (.147) (.139) (.761) (.796) CEO_OPTIONS -0.160 * 0.543 0.048 0.164 ** -0.172 ** 0.670 ** 0.264 † (.071) (.389) (.320) (.046) (.042) (.167) (.148) TAXES -0.020 0.126 -0.004 0.004 0.020 0.173 * 0.093 † (.025) (.175) (.132) (.015) (.013) (.068) (.055) EXT_GOV -0.035 ** 0.109 * -0.001 0.014 ** -0.021 ** 0.062 ** -0.030 (.006) (.043) (.038) (.003) (.003) (.018) (.020) INT_GOV -0.298 ** -0.449 -0.941 0.029 -0.343 ** -0.332 -0.836 ** (.106) (.742) (.621) (.059) (.055) (.297) (.293) Number of Obs. 4,113 4,113 4,113 3,061 9,447 9,447 9,447 AIC 7,889 17,385 21,389 3,706 15,397 33,647 41,293 BIC 8,003 17,499 21,509 3,815 15,576 33,826 41,479 Log-Likelihood -3,927 -8,675 -10,675 -1,835 -7,674 -16,799 -20,620

** significant at 1%; * significant at 5%; † significant at 10%