THE FIRM LIFE CYCLE HYPOTHESIS AND
CORPORATE PAYOUT RESPONSES TO THE 2003 TAX CUT
By
Lu Tan

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Economics - Doctor of Philosophy
2014

ABSTRACT
THE FIRM LIFE CYCLE HYPOTHESIS AND
CORPORATE PAYOUT RESPONSES TO THE 2003 TAX CUT
By
Lu Tan
This dissertation examines changes in corporate dividend policies in response to the taxregime change resulting from the Jobs and Growth Tax Relief Reconciliation Act of 2003
(JGTRRA). It is shown that firms at different stages of the life cycle have significantly different
responses in dividend policy to the change in dividend taxation, after controlling for profitability,
investment opportunities, free cash flow, firm size, leverage, liquidity, and the rate of economic
growth. Low- and moderate-payout firms responded to JGTRRA by increasing their payout
ratios, while the highest-payout firms reduced their payout ratios after 2003. A significant
portion of the increased propensity to pay dividends in the post-JGTRRA period can be
explained by the increase in the propensity of growing companies to pay dividends as they
mature. This result is consistent with the firm life-cycle hypothesis.

Copyright by
LU TAN
2014

To my parents and to all who have stayed close to me through good and bad times.

iv

ACKNOWLEDGEMENTS

I am grateful to my committee members, Professor Leslie Papke, Professor Stacy
Dickert-Conlin, and Professor Edmund Outslay, for their continuous support and
encouragement. Charles Ballard has been an inspiration for me. I have been extremely
lucky to have him as my mentor. He is everything a professor should be.
I would also like to express my deep gratitude to all the faculty of the Department
of Economics.

v

TABLE OF CONTENTS

LIST OF TABLES..............................................................................................................viii
LIST OF FIGURES.............................................................................................................. ix
CHAPTER ONE: INTRODUCTION ................................................................................... 1
CHAPTER TWO: LITERATURE REVIEW AND THE JOBS AND GROWTH TAX
RELIEF RECONCILIATION ACT OF 2003 ...................................................................... 5
2.1. Theories of Dividend Taxation .......................................................................... 5
2.2. Agency Models and Signaling Theory .............................................................. 6
2.3. The Firm Life-Cycle Hypothesis .................................................................... 10
2.4. The Jobs and Growth Tax Relief Reconciliation Act of 2003 ......................... 13
CHAPTER THREE: THEORETICAL PREDICTIONS .................................................... 15
CHAPTER FOUR: DATA AND VARIABLES ................................................................ 21
CHAPTER FIVE: EMPIRICAL EVIDENCE .................................................................... 27
5.1. The Time-Series Behavior of Corporate Distributions ................................... 27
5.2. Comparison Tests of Dividend Payout Ratios................................................. 29
5.3. Effects of JGTRRA ......................................................................................... 32
5.4. An Inverse-U-Shaped Relationship between Payout Ratios and Firm Age .... 34
5.5. Reactions of High Dividend Payers and Low Dividend Payers ..................... 36
5.6. Multinomial Logit Regressions ....................................................................... 38
5.6.1. Probabilities of Firm Exit and Dividend Policy by Firm Age.......... 40
5.6.2. The Effects of Within-Firm Control Variables ................................. 41
5.6.3. The Effects of Other Control Variables ............................................ 45
CHAPTER SIX: SENSITIVITY ANALYSIS AND ADDITIONAL TESTS ................... 47
CHAPTER SEVEN: CONCLUSION ................................................................................. 50
APPENDICES .................................................................................................................... 54
APPENDIX 1: Tables.............................................................................................. 55
APPENDIX 2: Figures ............................................................................................ 63
REFERENCES.................................................................................................................... 79

vi

LIST OF TABLES

Table 1. Variable Definitions .............................................................................................. 56
Table 2. Descriptive Statistics ............................................................................................. 57
Table 3. Matched-Pairs t-Tests on Dividend Payout Ratios ............................................... 58
Table 4. The Effects of JGTRRA and Firm Age on Dividend Payout Ratios .................... 59
Table 5. Multinomial Logit Regression .............................................................................. 60
Table 6. Marginal Effects of Multinomial Logit Regression .............................................. 61
Table 7. Sensitivity Test ..................................................................................................... 62

vii

LIST OF FIGURES

Figure 1(a). The Separating Equilibrium with Dividend Signaling .................................... 64
Figure 1(b). The Simplified Separating Equilibrium .......................................................... 65
Figure 1(a). The Dividend Expectation of Growing and Mature Firms ............................. 66
Figure 1(b). Impact of JGTRRA on Corporate Dividend Payout ....................................... 67
Figure 2(a). Number of Firms and Dividend Payers and Percentage of Firms Paying
Dividends, 1978-2010 ......................................................................................................... 68
Figure 2(b). Percentage of Firms Initiating Dividends, 1978-2010 .................................... 69
Figure 2(c). Median Firm Age, 1978-2010 ......................................................................... 70
Figure 2(d). Median Listing Age, 1978-2010 ..................................................................... 71
Figure 3. Count of Firms by Number of Years Paying Dividends...................................... 72
Figure 4(a). Average Dividend Payout Ratios, 1993-2010 ................................................. 73
Figure 4(b). Average Dividend Payout Ratios for Quintiles of Firms ................................ 74
Figure 4(c). Average Dividend Payout Ratios for Growing and Mature Firms.................. 75
Figure 4(d). Average Yield Ratios for Growing and Mature Firms ................................... 76
Figure 5. Predicted and Observed Probabilities of Firm Exit and Dividend Payout Policy
by Firm Age ........................................................................................................................ 77
Figure 6. Mean Reversion in Probabilities of Firm Dividend Payout Policy by Firm Age
............................................................................................................................................. 78

viii

CHAPTER ONE
INTRODUCTION
The effects of dividend taxation on corporate behavior have been debated by
scholars for decades. The debates do not appear to have achieved a consensus, due largely
to the lack of compelling tax variations (Auerbach (2010)). The most recent cut in
dividend and capital-gains tax rates provides an important opportunity to study the effect
of taxes on dividends and corporate financial policy, and to draw empirical lessons of
relevance for tax policy.
In 2003, the U.S. Congress reduced the tax rates on dividends and capital gains.
This law is known as The Jobs and Growth Tax Relief Reconciliation Act of 2003
(JGTRRA). The objective of this study is to provide insights into the determinants of
firms’ payout policies after the tax-policy change. Specifically, the first goal of the paper is
to quantify the impact of JGTRRA on the propensity of firms to pay any dividends, as well
as on dividend payout ratios. The second goal is to investigate the way in which the
responses of corporate payout policies to JGTRRA vary for firms at different stages of the
firm life cycle.
In the four decades before the passage of JGTRRA, there was a substantial decrease
in the percentage of firms that pay any dividends. From 1963 to 1967, 71.6 percent of
NYSE, AMEX, and NASDAQ non-financial non-utility firms paid dividends. This

1

proportion declined to 36.1 percent in 1983-87, and to only 20.8 percent in 1999
(DeAngelo, DeAngelo, and Skinner (2004)). Fama and French (2001) have referred to this
as the phenomenon of “disappearing dividends”. They attribute the reduction in the
proportion of firms that pay dividends to the surge in the number of newly listed small
firms that did not pay any dividends.
The trend of disappearing dividends reversed in the last decade. The proportion of
firms paying dividends increased to an average of 30.5 percent from 2006 to 2010,
although it remained far below the level of a few decades before. However, this occurred
at the same time as a sharp reduction in the total number of publicly traded companies.
The firms that exited the market had a disproportionate tendency to pay no dividends.
Over 2000 firms exited the market from 2000-Q2 and 2005-Q2, only 2 percent of which
were dividend payers. Thus even if the surviving companies did not increase their
distributions at all, the average payout ratio would still be higher. Moreover, as the
surviving companies grow older, their increasing financial stability may increase the
probability that they would make distributions. As a result, in analyzing this change in
dividend behavior, it is important to deal with issues of sample selection. Without taking
into consideration the reduction in the total number of companies and their growth
patterns, it is difficult to distinguish the net effect of JGTRRA from the selection of
companies and time-driven life-cycle effects.
To account for these econometric issues, I add the choice of exiting the market into
a model of corporate dividend policy, and estimate the relative probabilities among

2

alternatives with a multinomial logit model. Furthermore, I include an exogenous proxy
for firm life-cycle stage that is preferable to those used in previous research designs.
Specifically, I use the age of the firm to distinguish between young firms and mature firms,
to identify the differences in their responses to a change in dividend taxation. Thus the
estimated aggregate effect of a dividend tax change depends on the cross-sectional
distribution of the age of firms.
My main findings are as follows. First, the marginal effect of the tax-policy change
on the propensity to pay dividends is significantly different for firms at different stages of
the life cycle. In the pre-JGTRRA period, mature firms were more likely to pay dividends
than young firms. In the post-JGTRRA period, however, there was mean reversion in the
probability of paying dividends. Young firms became more likely to pay dividends in the
post-JGTRRA period than in the pre-JGTRRA period, while the probability of paying
dividends declined for mature firms. Second, mature firms did not increase the proportion
of their earnings paid out in dividends after JGTRRA. In fact, after controlling for payout
history and profitability, the average payout ratio decreased for firms over the age of 42.
Finally, the average payout ratio increased in the post-JGTRRA period. This is due to a
large increase in payout ratios for moderate-dividend-payout firms, as well as the initiation
of dividend payments by many firms that had never paid dividends before JGTRRA.
These findings of tax-policy effects are important, but it is also true that dividends exhibit a
great deal of stickiness. Many firms in the sample have not paid dividends during the
whole sample period of 1993-2010, and thus these firms were not responsive to any

3

changes in dividend tax policy.
Overall, the dividend payout ratio increased among young firms in the postJGTRRA period, while the propensity to pay dividends decreased among mature firms.
These patterns are consistent with the theoretical predictions developed in Chapter Three.
The remainder of this dissertation proceeds as follows. In Chapter Two, I review the
relevant literature and summarize the main components of the 2003 Tax Act. In Chapter
Three, I develop a theoretical prediction of the effect of JGTRRA. In Chapter Four, I
describe the data and the variables used in the analysis. I discuss the empirical results in
Chapter Five, and present additional sensitivity tests in Chapter Six. The dissertation
concludes with Chapter Seven.

4

CHAPTER TWO
LITERATURE REVIEW AND THE JOBS AND GROWTH TAX
RELIEF RECONCILIATION ACT OF 2003
A large body of theoretical work on dividend payments has been developed in the
fields of public finance and corporate finance. In this chapter I will summarize the main
theories of dividend taxation. These are agency models, signaling theory, and more
recently, the firm life-cycle hypothesis.

2.1. Theories of Dividend Taxation
Historically, dividend income has most often been taxed at a higher rate than
capital-gains income.1 This tax penalty on dividend income for individual taxpayers leads
to the question of why firms pay dividends. Economists have long been puzzled that firms
pay dividends instead of repurchasing shares, and that firms issue new equity
simultaneously with paying dividends. Several theories have been developed to explain
the “Dividend Puzzle”. Miller and Modigliani (1961) argue that a firm’s dividend payout
policy is irrelevant to its market valuation, but most of the empirical evidence suggests the
opposite. For example, Sialm (2009) provides direct evidence that taxes are capitalized
into asset prices. Dhaliwal, Krull, Li, and Moser (2005) show that the magnitude of the

1

The only exceptions are the short period of 1988-1990 following the Tax Reform Act of
1986, and the years after 2003.

5

dividend tax capitalization is a function of the marginal investor’s tax attributes.
Under the “traditional view” of dividend taxation, new shares are the marginal
source of finance. See, for example, Harberger (1962) and Feldstein (1970). Under the
traditional view, a reduction in the rate of dividend taxation would reduce the costs of
dividends, stimulate corporate investment, and raise dividend payments. Poterba and
Summers (1985) provide support for the traditional view, using data from the United
Kingdom.
The “new view” of dividend taxation proceeds from the assumption that retained
earnings are the marginal source of funds for investments. See Auerbach (1979), Bradford
(1981), and King (1977). According to the new view, the benefits of dividend tax deferral
effectively offset the future dividend tax liability attributable to the investment financed by
retained earnings. Therefore, dividend taxes do not increase the tax burden on investments
financed with retained earnings. See Auerbach (1979) and Zodrow (1991). Desai and
Goolsbee (2004) find a statistically insignificant relationship between investment and
dividend taxation, thus providing empirical evidence in support of the new view.

2.2. Agency Models and Signaling Theory
Lintner (1956) conducted a survey of managers, and found that they decided the
future level of dividends based on the current level of dividends. Managers tended not to
reduce dividends unless there were no other alternatives, and increased dividends only if
the new dividend levels were sustainable. This empirical finding is known as dividend

6

smoothing, which is, at least in part, related to agency costs and signaling effects (Brav, et
al. (2005)).
The “informational content of dividends” was first discussed by Miller and
Modigliani (1961), when they relaxed the assumption of perfect information in their taxirrelevance theory. Within the context of signaling theory, the payment of dividends may
provide information that is not revealed by financial statements, audit reports, or firm
announcements. Thus the payment of dividends can serve as a signal of the financial
health of the firm. See John and Williams (1985) and Bernheim and Wantz (1995). Within
the context of dividend signaling theory, the major effect of the dividend tax rate on the
market reaction to dividend distributions is to provide credible signals of profitability and
effective corporate governance for firms that pay dividends. If we assume that dividend
distribution involves some costs, and that these costs are higher for low-quality firms, then
in the perfect Bayesian equilibrium high-quality firms are willing to pay higher dividends
than low-quality firms. As a result, investors can reasonably regard dividend levels as a
signal of profitability and effective corporate governance. Since the costs of funds are
partly determined by the dividend taxes, it follows that, at a higher rate of dividend
taxation, a high-quality firm can pay a lower level of dividend to deter imitation by a lowquality firm.
In the context of agency models, one major explanation for why firms pay
dividends is the free-cash-flow hypothesis. See Jensen (1986) and Blanchard, Lopez-deSilanes, and Shleifer (1994). This hypothesis is based primarily on the existence of a

7

conflict of interest between managers and shareholders. For example, managers may be
tempted to invest free cash flows in pet projects with negative net present value. Thus
dividend payments can serve as a method to discipline managers by reducing free cash
flows.
Lang and Litzenberger (1989) find that the average announcement return of a large
dividend change is significantly higher for firms with low Tobin’s Q than for firms with
high Tobin’s Q. Empirically, a low Tobin’s Q is associated with a lower level of
investment opportunities. This implies that a firm with a low Tobin’s Q may suffer from
overinvestment. This is consistent with the free-cash-flow hypothesis, in that an increase
in dividends by a firm with low Tobin’s Q could be interpreted as a signal of management’s
intention to mitigate overinvestment problems, and could therefore receive a larger
reaction from the market.
After comparing the effects of dividend taxation in the new view, signaling, and
agency models, Gordon and Dietz (2008) conclude that the agency model is most likely to
fit the empirical evidence. Chetty and Saez (2010) develop a theoretical agency model,
and argue that cash-richness may be associated with a tendency to invest in pet projects.
Their model suggests that firms in which managers put more weight on profit
maximization (either by having a large proportion of equity incentives such as stock
options, or because they are being monitored carefully by institutional shareholders) are
more responsive in dividend payout following a dividend tax cut.
To reduce the agency problems in corporate governance, institutional investors,

8

such as public and private pension funds, mutual funds, are generally considered to be
more influential monitors of corporate governance. The involvement of large institutional
investors increased dramatically during the mid-1980s with the advent of public pension
fund activism (e.g., Gillan and Starks (2007)). Since the corporate scandals of 2001 and
2002, institutional investors have tended to become more active. It should also be noted
that the institutional investors are not affected by dividend taxes because their dividend
income is tax-exempt. Auerbach (1983) and Baker and Wurgler (2004) show in theoretical
models of the tax-clientele hypothesis that wealth-maximizing strategies are different for
firms with different shareholders, so it is important to know the stock ownership structure
of the firms. Auerbach predicts that clientele sorting occurs with respect to dividend
payout rates.
Scholz (1992) and Graham and Kumar (2006) show empirical evidence consistent
with the tax-clientele hypothesis. For example, Scholz (1992) uses the 1983 Survey of
Consumer Finances and finds that investors choose portfolios based on their tax rates.
Graham and Kumar (2006) find that retail investors prefer non-dividend-paying firms,
while institutional investors prefer dividend-paying firms. They also show that investors
who experienced an increase in their tax rates as a result of the 1993 tax law change
reduced high-dividend-yield stocks in their portfolios, and the difference in dividend
holdings between taxable and tax-deferred accounts is consistent with tax-motivated
preferences. Based on the tax-clientele theory, a firm whose shareholders are primarily
institutional investors is less likely to be impacted by a dividend tax cut than a firm owned

9

by individuals.

2.3. The Firm Life-Cycle Hypothesis
The theories discussed above do not explicitly refer to the firm life cycle.
However, each of these theories has implications for firms at different stages of the life
cycle. Therefore, each of these theories can be understood partly in terms of the life-cycle
hypothesis. In this section, we will move on to the studies that explicitly depend on the
firm life cycle.
Grullon, Michaely, and Swaminathan (2002) propose the “maturity hypothesis”,
which states that a firm would tend to increase dividends as it moves from a growth stage
to a mature stage. Companies with high growth opportunities have less free cash flow.
Therefore, a high dividend level is not needed to control free cash flow for a rapidly
growing young firm. On the other hand, mature firms may face a dearth of profitable
projects in which to invest. As a result, mature firms tend to have high free cash flows.
Thus mature firms tend to be able to make high dividend payments. Therefore, a firm’s
dividend policy appears to depend on its life-cycle stage. This is also known as the “lifecycle hypothesis” (DeAngelo, DeAngelo, and Stulz (2006)).
Sinn (1991) argues that it is optimal for the firm to retain its profits for an extended
period of internal growth (which necessarily follows the issue of new shares), and not to
pay dividends until the firm’s growth rate slows down. He suggests a life-cycle model in
which firms progress from the traditional view (which assumes that new equity issues are

10

the marginal source of funds for investments) to the new view (which assumes that
retained earnings are the marginal source of funds for investments). The effective cost of
capital declines through the firm’s life cycle as retained earnings accrue. Sinn concludes
that, in the presence of high dividend taxes, high dividend payments signal that firms are
mature and face a low cost of capital, because investments can be financed with dividend
reductions. On the other hand, low dividend taxes associated with low dividends signal a
lack of funds. Sinn argues that dividend tax distortions apply primarily to the immature
firms that retain most of their profits. Following Sinn’s arguments, the payout responses to
a reduction in dividend taxation should be greater among immature firms.
Korinek and Stiglitz (2009) incorporate different financing constraints at different
stages of the firm’s life cycle into Sinn’s model. They also assume an absence of agency
problems (i.e., they assume that retained earnings are allocated optimally by the manager).
Korinek and Stiglitz establish new results on the intertemporal tax-arbitrage opportunities
for mature firms. The key insight of Korinek and Stiglitz is that temporary dividend tax
changes allow mature firms to engage in intertemporal tax arbitrage, by shifting dividend
payments from high-tax periods to low-tax periods.
According to the free-cash-flow hypothesis, there are notable differences in the
patterns of resource allocation between firms at different stages of the life cycle.
DeAngelo, DeAngelo, and Stulz (2006) show that dividends are concentrated among the
largest and most profitable firms. This is consistent with the fact that free cash flows are
accumulated disproportionately by mature firms. Young firms are typically trying to gain

11

market share in their rapidly evolving industries. As a result, they are expected to invest
heavily in expenses such as research and development, advertising, and capital
expenditures. If they go public, they are likely to continue to need external financing to
sustain their high growth rates and large expenses. Since young firms are more likely to
need new capital, they do not benefit from high levels of dividends as much as mature
firms (Easterbrook (1984) and Jensen (1986)).
The empirical study of Ittner, Lambert and Larcker (2003) has also shown that, as a
result of cash constraints, young firms use a significantly higher proportion of equity
incentives. The agency model of Chetty and Saez (2010) suggests theoretically that firms
with a higher proportion of equity incentives are expected to be more responsive to a
dividend tax cut. Thus, this theoretical result of Chetty and Saez implies that young firms
will be more responsive to a dividend tax cut.
Thus the firm life-cycle hypothesis is consistent with a substantial variety of
theoretical literature. However, the effects of heterogeneity in firms’ life-cycle stages has
not been tested in the prior empirical literature related to the 2003 Tax Act. In this
dissertation, I show that when firm age is taken into account, dividend taxation has firstorder effects on firm-level dividend payout. These effects are different for firms at
different stages of the firm life cycle. When I include a quadratic in firm age, the estimated
coefficients are such that young firms increase dividends. However, for firms that are
sufficiently old, dividends decrease.
2.4. The Jobs and Growth Tax Relief Reconciliation Act of 2003

12

The effects of dividend taxation on corporate behavior have been debated by
scholars for decades. See the survey in Hanlon and Heitzman (2010). The debates do not
appear to have achieved a consensus, due largely to the lack of compelling tax variations
(Auerbach (2010)). The most recent cut in dividend and capital-gains tax rates provides an
important opportunity to study the effect of taxes on dividends and corporate financial
policy, and to draw empirical lessons of relevance for tax policy.
The passage of JGTRRA in 2003 equalized the tax rates for capital gains and
dividends for the first time since 1990. The new tax law decreased the top dividend tax
rate from 38.6 percent in 2002 to 15 percent, and from 10 percent to 5 percent for lowerincome individuals. The top capital-gains tax rate was reduced from 20 percent to 15
percent. For lower-income individuals, the capital-gains tax rate was reduced from 10
percent to 5 percent (and to zero percent in 2008).2
Several researchers have already studied the effects of JGTRRA. This research has
found that JGTRRA had a significant and positive impact on dividend payments by U.S.
firms. For example, see Julio and Ikenberrey (2004); Chetty and Saez (2005); Brown,
Liang, and Weisbenner (2007); and Blouin, Raedy, and Shackelford (2011). Yet the
precise explanation of the positive effect remains in dispute. Julio and Ikenberrey (2004)
use a constant-size sample of 1,000 firms, selected on the basis of market capitalization.
They find that dividend distributions began to increase starting in the late 1990s. Because
2

These provisions of JGTRRA were scheduled to expire after 2010. However, in
December 2010, the reduced tax rates were extended to 2012 by the Tax Relief,
Unemployment Insurance Reauthorization, and Job Creation Act of 2010.

13

the trend began prior to 2003, Julio and Ikenberrey conclude that the increase in dividend
distributions could not be attributed entirely to the 2003 tax cut. Chetty and Saez (2006)
also use a constant-size sample—including the top 3785 firms by market capitalization in
each quarter, up to the second quarter of 2005. However, Chetty and Saez find that the
trend of “disappearing dividends” stops precisely at the last quarter of 2002. Therefore,
they argue that the 2003 tax cut had a clear, strong causal effect. Chetty and Saez also
conclude that without a direct measure of the entry and exit effects, it is difficult to make
precise inferences about the effects of the 2003 tax cut on the level and concentration of
corporate distributions.
Brown, Liang, and Weisbenner (2007) test the effects of executive stock ownership
on firm payouts, and find that executives with higher stock ownership are more likely to
increase dividends after the 2003 tax cut, while no relation is found before the tax cut.
They also find no difference in the dividends paid on stocks held by tax-exempt institutions
before and after the tax cut. Blouin, Raedy, and Shackelford (2011) focus on the
managerial stock ownership, as well as individual stock ownership. They use simultaneous
equations to estimate both investor and managerial responses to the 2003 tax cut, and find
that firms adjust their payout distribution in a manner consistent with the altered tax
incentives for individual investors.

14

CHAPTER THREE
THEORETICAL PREDICTIONS
I will first develop a simple theory of separating equilibrium, to describe the payout
behavior of young and mature firms in the presence of dividend taxation. This model
generates testable hypotheses regarding the effect of JGTRRA on the dividend payout
behavior of young firms and mature firms. The theory is based on the separating
equilibrium in Bernheim and Wantz (1995). They show that the impact of dividend payout
on the market value of the firm increases monotonically with the marginal cost of
dividends, which is partly determined by the dividend tax rate, as shown in the market
valuation function below,
v ( y ) = [λτ + (1 − λ )(τ + ξ )1/ 2 ] y .

(1)

where v is the market valuation, y is the dividend, τ is the dividend tax rate, λ is a number
between 0 and 1, and ξ is a positive constant.
In the separating equilibrium, bad firms will pay no dividends, and investors will
correctly identify non-dividend-paying firms as bad firms. This is illustrated in Figure
1(a). Assume that dividends are costly and that a typical manager cares about his firm’s
actual value (which is negatively related to dividends), as well as its perceived value by
investors. The fundamental reason that dividends can serve as a signal here is that the
marginal cost of dividends is higher for bad firms than good firms. Under these

15

assumptions, we can depict managers’ indifference curves on the y − v surface. In a
separating equilibrium, bad firms will pay no dividends and be perceived with a market
value of v (which denotes the value of a bad firm paying no dividends under perfect
information), and good firms will pay a dividend of y and be perceived with a market
value of v (which denotes the value of a good firm that pays dividends of y under perfect
information). The choice of (0, v ) is the optimal choice for bad firms because in the
separating equilibrium, the indifference curve of bad firms will be at the highest possible
level among all other choices.
Since the choices of good firms and bad firms can be both included in the
indifference curve of a bad firm, only the indifference of a bad firm will be kept for the
further analysis (see Figure 1(b)).
In Figure 1(c), I use the signaling theory to set up the hypotheses for firms’ payout
responses after a dividend tax cut. The dividend payout and perceived firm valuation are
denoted by y and v respectively. The figure depicts the indifference curve of a bad firm at
the growing and mature stages. In the most efficient separating equilibrium of mature
firms, a good firm will pay the level of dividends at y , and investors will identify firms
paying y as good firms and value them at v . On the other hand, bad firms pay no
dividends, and they are valued at v by investors.
The cost of paying dividends (or the “bang for the buck” in Bernheim and Wantz
(1995)) is shown in the brackets of the market valuation equation shown above. As the

16

dividend tax increases, the cost of paying dividends increases, and good firms do not have
to pay as large dividends in order to deter imitation of bad firms. This is illustrated as the
flatter indifferent curves of the growing and mature firms, and higher payout to maintain
the same perceived firm values in a separating equilibrium.
While it appears that the signaling theory suggests that firms will increase payout
after the dividend tax reduction, there are a number of complications to this proposition.
For example, the signaling theory does not account for firm heterogeneity. Since firms
have different sources of funds and different compositions of the sources, their marginal
costs of funds may have different correlations with the dividend tax rate. Because of these
complications, most papers examining the effects of 2003 tax cut have been empirical.
The majority of the studies use indirect tests, which try to infer the characteristics of firms
making marginal dividend payments. In this paper, I develop the theoretical hypotheses
after taking into consideration the effect of firm heterogeneity on the cost of dividends.
According to the life-cycle hypothesis, dividends are more costly for growing firms, for a
variety of reasons. These include the higher opportunity cost of investment opportunities,3
financing costs,4 and dividend tax distortions. The separating equilibrium for growing
firms can be shown by the steeper indifference curve in Figure 1(c). The efficient

3

Grullon, Michaely, and Swaminathan (2002) argue that firms increase dividend payout as
they grow older because the marginal return of capital declines.
4

Opler, et al. (1999), and Bates, Kahle, and Stulz (2009) provide evidence that when
agency costs or asymmetric information make it more difficult to raise external capital, the
precautionary incentive to conserve cash is increased.

17

separating outcome for the good growing firm will be at the point of ( y ', v ' ).
As a result of the dividend tax cut, the marginal costs of paying dividends are
reduced, and each good firm is expected to increase its dividend payout. In Figure 1(a),
the decrease in the marginal costs of paying dividends is demonstrated by a reduction in
the slopes of the dotted indifference curves.
Moreover, in the life-cycle hypothesis, young firms have a higher proportion of
funds from new issues of equity than retained earnings, and the firms relying on the new
issues of equity are considered to bear heavier dividend taxes. The extrapolation of the
life-cycle hypothesis predicts that the payout responses to a reduction in dividend taxation
should be greater among growing firms. Under this hypothesis, growing firms face greater
dividend tax distortions than mature firms in the pre-JGTRRA period, and thus the payout
responses to a reduction in dividend taxation in the post-JGTRRA period should be greater
among growing firms. In other words, the payout responses to a reduction in dividend
taxation should be greater among growing firms, since the dividend tax distortion is
reduced by a larger magnitude for growing firms than for mature firms.
Figure 1(d) is derived from Figure 1(c). If we draw the indifference curve for firms
at each age and find the efficient separating dividend payout in Figure 1(c), we can connect
the points on the firm-age/dividend-payout plane and get the curve S. The signaling theory
predicts higher dividend payments after the dividend tax cut (as shown by the curve S’).
In addition, for mature firms that are sufficiently old, when the ability to generate
cash flows slows down and the agency costs are very high because of the separation of

18

ownership and control, the dividend payout might decline. In this case, the separating
equilibrium curve would rotate from S’ to S’’. Compared to mature firms, growing firms
are more responsive in dividend payout to a reduction in dividend taxation, since they face
greater reduction in dividend tax distortions than mature firms in the post-JGTRRA period.
The prediction to be tested would be an increase in dividend payout from growing firms,
and a decrease in dividend payout from mature firms.
In the following chapters, I will examine both the changes in the probability of
paying any dividends and the changes in dividend payout ratios.5 The results are
consistent with the predictions: In response to JGTRRA, growing firms increased both the
propensity to pay dividends and the average payout ratios. On the other hand, after
controlling for profitability and payout history, the 2003 tax cut is associated with a
decrease in the likelihood that mature firms pay any dividends, and the high-dividendpaying firms reduced average payout ratios in the post-JGTRRA period. Although I did
not have a direct measure of the tax characteristics of firms’ shareholders, it should be
noted that firms whose shareholders are primarily institutional investors are less likely to
be impacted by a dividend tax cut than firms owned by individuals. Thus with a large and
growing proportion of institutional investors, such as public and private pension funds, and
growing charitable foundations, and tax-deferred individual accounts such as IRAs, the
impact of the 2003 tax cut on dividend payout will be expected to have a smaller

5

The dividend payout ratio is defined as the percentage of a company’s earnings that is
paid in dividends. Details can be found in Chapter Four.

19

magnitude.

20

CHAPTER FOUR
DATA AND VARIABLES
The data for firm characteristics are taken from the CRSP/COMPUSTAT merged
database. This database merges historical price and distribution data from the Center for
Research in Security Prices (CRSP) and a variety of other company financial data from
Standard and Poor’s COMPUSTAT database. The sample includes all firms in fiscal year

t that have values for total assets of at least USD 0.5 million, total sales of at least USD 20
million, and book equity of at least USD 0.25 million. Utilities (SIC codes 4900-4949)
and financial firms (SIC codes 6000-6999) are excluded since they are subject to special
regulations.
Figure 2(a) shows the aggregate number of firms in the sample and the number of
firms paying dividends in each year from 1978 to 2010. The percentage of dividend
initiators (i.e., the firms that began paying dividends for the first time) gradually declined
during the 1980s and 1990s. Figure 2(b) shows that the percentage of dividend initiators
increased dramatically in 2003, consistent with the findings of Chetty and Saez (2006).
Moreover, over 2000 firms exited the market from 2000-Q2 and 2005-Q2. Only 2
percent of the exiting firms were dividend payers. Since most of the firms that exited the
market were young firms that had never paid dividends, excluding the option of exiting the
market could overestimate the effect of the tax-law change on firms’ propensity to pay
dividends.

21

Both dividend payout ratios and dividend yield ratios are used as measures of
corporate dividend behavior. The dividend payout ratio is calculated as cash dividends
(COMPUSTAT DATA 26) divided by earnings before interest and taxes. The dividend
yield ratio shows the amount of dividends that a company pays to its investors in
comparison to the market price of its stock, computed as the ratio of dividends per share to
the share price at the end of the year.
DeAngelo, DeAngelo, and Stulz (2006) proxy for firm maturity by using retained
earnings as a proportion of total equity. This measure is potentially endogenous. Instead, I
use the direct and exogenous measure of firm age, computed as the difference between
year t and the first year in which the firm appeared on CRSP/COMPUSTAT, plus one
( firm age = t − t0 + 1 ). Moreover, to control for the heterogeneity in the length of the firm
life cycle across industry, I use a dummy variable equal to one if the firm’s age is greater
than or equal to the median age of the firms in its 3-digit SIC industry.
The earliest year in which a firm appeared on CRSP/COMPUSTAT is 1962.6 Thus
the maximum age is 17 years at the end of 1978, and 49 years at the end of 2010. Figure
2(c) shows that the median firm age decreased in the 1990s during the Internet boom, and
increased rapidly after the climax of the Internet bubble in 2000.
On average, firms are older after JGTRRA than before JGTRRA. The average age
increased from 11.77 years in the pre-JGTRRA period (1993 – 2002) to 15.88 years in the

6

This measurement truncates at 1962. Fortunately, among the 909 firms, only 58 were
founded before 1962.

22

post-JGTRRA period (2003 – 2010). (See Table 2.) An alternative measure of firm age—
the listing age—is the number of years between the current year and the date of the firm’s
initial public offering, plus one. The median listing age is less than the median firm age,
but it follows a similar pattern of decreasing in the 1980s and1990s, and increasing in the
2000s.7 (See Figure 2(d).)
Following the previous literature, I control for firm characteristics by using
profitability (measured by the return on assets), a measure of the size of assets, Tobin’s Q
(measured by the ratio of market value to book value), a measure of leverage, cash
holdings (as a measure of free cash flow), and a measure of liquidity. I also control for an
important industry characteristic by using the median age of the firms in the industry,
where the industry is defined as the three-digit SIC. All control variables are lagged by
one year to reduce the possibility of endogeneity in estimation. The next few paragraphs
contain detailed descriptions of each of the explanatory variables. These definitions are
also summarized in Table 1. Descriptive statistics are summarized in Table 2.
Return on Assets: Return on Assets is a measure of firm profitability, computed as
the ratio of the firm’s operating income before depreciation (DATA13), divided by the
book value of assets (DATA6). The average firm profitability was slightly lower after
JGTRRA than before JGTRRA.
Tobin’s Q: Tobin’s Q (or the ratio of market value to book value) is included as a

7

The decrease in the median listing age in the 1980s and 1990s is due to the concentration
of new listings in the time periods of 1986-1987 and 1995-2000.

23

proxy for investment opportunities. Tobin’s Q is calculated as the market value of the
firm’s assets at the end of the year, divided by their book value (DATA6). The market
value of assets is calculated as the book value of assets (DATA6), minus the book value of
common equity (DATA60), plus the market value of common equity (DATA25 x
DATA199). In the post-JGTRRA period, the average level of Tobin’s Q was lower than in
the pre-JGTRRA period.
If the agency-costs/free-cash-flow hypothesis is correct, firms with greater
investment opportunities will be less likely to pay dividends, because firms with greater
investment opportunities will have fewer free cash flows.
Free Cash Flow: Dividend payments depend on the availability of a firm’s cash
flow. Following Lehn and Poulsen (1989), the firm’s free cash flow is calculated by
adding net income, depreciation, and deferred taxes,8 and then subtracting dividends paid9
and capital expenditures (computed as DATA13 - DATA15 - (DATA16 - DATA35) DATA19 - DATA21). This measure of free cash flow is then divided by the market value of
the firm’s assets.
The free cash flows of the median firm are similar in the pre- and post-JGTRRA
periods. However, the distribution of free cash flows is more positively skewed in the
post-JGTRRA period, which means that the average free cash flow rose after JGTRRA.
8

Depreciation and deferred taxes are deducted in calculating net income (for tax purposes),

although they do not directly affect cash flows.
9

Free cash flow is net of dividends because it is used to predict the dividends for the
following year.

24

Liquidity: Liquidity, like Free Cash Flow, is a measure of the corporate financial
resources that could potentially be the source of dividend payments. The difference is that
liquidity is a broader measure than Free Cash Flow. Liquidity is calculated as the firm’s
cash and short-term assets (DATA1), divided by the market value of assets. The average
level of liquidity was higher in the post-JGTRRA period than before JGTRRA. Although
including the correlated variables of investment opportunities, free cash flow, and liquidity
may reduce the statistical significance of the estimated coefficients, it will reduce the bias
problem caused by omitting potentially important predictor variables.
Leverage: Leverage is measured by the ratio of total liabilities (DATA181) to the
market value of assets (as defined above). Bagwell and Shoven (1989) predict that firms
with high indebtedness are less likely to make corporate distributions, all else equal. On
average, the level of leverage was lower in the post-JGTRRA period than before
JGTRRA.10
Industry age: Industry age is measured by the listing age of the median firm in the
3-digit SIC industry in year t. The average industry age increased from an average of 9.39
years before JGTRRA to an average of 13.60 years in the post-JGTRRA period.
Firm Size: In the agency model, larger firms have higher agency costs and lower
financing costs, and are thus more likely to pay dividends. Firm Size is calculated as the
natural logarithm of the ratio of the firm’s market capitalization (DATA25 x DATA199) to
10

JGTRRA increased the relative cost of debt to equity by reducing the tax rates on
dividends and capital gains. For a discussion of the relationship between the debt-equity
choice and tax costs, see Auerbach (1979).

25

the equal-weighted average market capitalization of all firms in year t. The average size of
firms increased after JGTRRA.
GDP Growth Rate: The final explanatory variable is the annual growth rate of real
GDP, from the U.S. Bureau of Economic Analysis. According to the agency models, a
firm may increase its dividend payments to control excessive spending, either because
there is an increase in cash flow (which is good news for the firm), or because of a
decrease in investment opportunities (which is bad news for the firm). The nature of the
information revealed about the firm may vary depending on conditions in the aggregate
economy.
The summary statistics shown in Table 2 indicate that in the post-JGTRRA period,
which experienced a lower rate of economic growth, the average firm had fewer
investment opportunities (lower average Tobin’s Q), accumulated more cash flows, raised
less debt (lower leverage), and had more liquidity.

26

CHAPTER FIVE
EMPIRICAL EVIDENCE
5.1. The Time-Series Behavior of Corporate Distributions
Two samples will be used in this chapter. The unrestricted sample includes all
firms that have values for total assets of at least USD 0.5 million, total sales of at least
USD 20 million, and book equity of at least USD 0.25 million, and excludes the Utilities
(SIC codes 4900-4949) and financial firms (SIC codes 6000-6999) that are subject to
special regulations. The summary statistics were described in Chapter Four. The restricted
sample consists of a balanced dataset that includes all the firms that maintained their
corporate identities for the entire 18-year period from 1993 to 2010, and for which we have
information on dividends and earnings before interest and taxes.
The restricted sample includes 909 firms, so that the sample has (909 firms) * (18
years) = 16,362 firm-year observations. Among these firms, 289 firms paid dividends
every year, while 389 firms did not pay dividends in any year. Figure 3 presents evidence
on the persistence of dividend payments, consistent with the literature. The majority of the
firms either keep paying dividends every year or never pay dividends in any year. Lintner
(1956) cites the managers’ belief that the market places a premium on dividend stability or
gradual growth. The existence of dividend stability is empirically supported by several
authors. For example, Fama and Babiak (1968) find that dividend payouts are unchanged
for 25% of their sample between 1947 and 1965, while 33% of Pettit’s (1972) sample of

27

634 firms had constant dividends for 18 quarters. More recently, Leary and Michaely
(2011) find that firms smoothed dividends more in the 1990s and in the 2000s than they
did in the 1960s and 1970s.
I test two kinds of corporate payout behaviors: (1) dividend-paying vs. nondividend-paying vs. exit, (2) the payout ratio (i.e., the ratio of dividends to net earnings).
The unrestricted/unbalanced sample will be used in the rest of Section 5.1 for descriptive
analysis and then used after Section 5.5 to test the first behavior. The restricted sample
will be used in Sections 5.2 to 5.5 to measure the second behavior. Since 389 firms never
paid dividends, the variation in payout ratios comes exclusively from the 289 dividendpaying firms and the 231 firms that paid dividends in some years, but not in others.11 The
descriptive details of payout ratios will be discussed in Section 5.2.
Figure 4(a) plots the time-series behavior of dividends per share, earnings per share,
and the payout ratios for the firms in the unrestricted sample. On average, dividends per
share decreased from $0.36 in 1993 to $0.24 in 2002. This trend reversed after 2002, and
the level of dividends per share has increased steadily since then. In 2010, the mean level
of dividends per share reached $0.44. 12 Earnings per share are more volatile than
dividends per share. Earnings per share dropped substantially in 2001 and 2002, and again

11

Among the 231 firms, only 6 firms initiated dividends in 2003 and made payments every
year afterwards.
12

These figures are nominal numbers without adjustment for inflation. Thus the decrease
in real dividends before 2002 is more pronounced than these figures would suggest, while
the increase in real dividends after 2002 is less pronounced.

28

in 2008 and 2009. Despite the fluctuations in corporate earnings, dividends per share
remained relatively stable. Thus dividend payout exhibits a substantial degree of
stickiness.
The plot of aggregate data in Figure 4(a) shows that the payout ratios remained
within the range between 0.078 and 0.108 until 2003. The firms in the sample increased
their payout after JGTRRA, on average. In the next section, a matched-pairs t-test will be
performed to test the statistical significance of this observation.

5.2. Comparison Tests of Dividend Payout Ratios
The results of tests for the differences between the dividend payout ratios before
and after 2003 are presented in Table 3. The pre-JGTRRA period includes 10 years from
1993 to 2002, while the post-JGTRRA period includes 8 years from 2003 to 2010. For
each firm, the pre- and post-JGTRRA dividend payout ratios are calculated as the average
of the annual payout ratios in the respective period. Consistent with the prediction that a
reduction in the relative tax costs of dividends reduces the aversion of average investors
toward dividends, 36 percent of the firms in the sample increased their dividend payout
ratios following JGTRRA. The average dividend payout ratio rose from 8.9 percent before
JGTRRA to 11.8 percent after JGTRRA. The one-tailed t-test shows that the differences
between the payout ratios pre- and post-JGTRRA are statistically significant from the null
hypothesis of zero at the five-percent level.
To look further into the dividend behaviors of different groups of firms, the

29

restricted sample is divided into five quintiles, based on their average payout ratios in the
pre-JGTRRA period. (Quintile 1 has the lowest pre-JGTRRA average payout ratio.) The
results reveal significant differences among the quintiles. Each of the first four quintiles
experienced an increase in its average payout ratio after JGTRRA (from 0, 0, 0.006, and
0.115 before JGTRRA to 0.038, 0.033,13 0.050, and 0.164 in the post-JGTRRA period). On
the other hand, the average payout ratio of the firms in the highest quintile declined from
0.324 to 0.307. The mean differences are statistically significant for all quintiles.
The time path for each quintile is shown in Figure 4(b). The average payout ratio
for quintile 5 declined after the peak in 2001 as the firms grew older, despite the short-term
rise in 2008 and 2009.14 Quintile 4 has moderate average dividend payout ratios before
JGTRRA, but shows the largest increase in average payout ratios after JGTRRA. (See
Table 3.) Quintiles 1, 2, and 3 follow a similar pattern: These quintiles consist primarily or
exclusively of non-dividend payers in the pre-JGTRRA period, while having positive
average payout ratios after JGTRRA. Specifically, 22, 20, and 30 percent of the firms
within Quintiles 1, 2, and 3 increased their average payout ratios. Some 78 percent of the
firms in the first quintile and 80 percent of the firms in the second quintile continued to pay
no dividends. In the two higher quintiles, 62 and 48 percent of the firms increased their

13

Since the first and second quintiles are composed of firms that did not pay dividends before
JGTRRA and the rank among non-dividend payers is random, any differences between these two
quintiles should be considered as random.
14

Dividend smoothers would be expected to have an increase in the payout ratio in those two years,
because of the economic recession. If profits are down, and if dividends stay the same, the payout
ratio will increase.

30

average payout ratios, respectively.
The fourth quintile has the largest number of firms with an increase in their
dividend payout ratios (62 percent), as well as the highest average increase in payout ratios
(an increase of 4.9 percentage points). In the third quintile, only 30 percent of the firms
increased their payout ratios, while 14 percent of the firms reduced their payout. The
average increase in payout ratios is 4.4 percentage points, which is nearly as large as the
average increase in the fourth quintile. Thus the third quintile is characterized by relatively
large increases in dividend payout ratios for a relatively small number of firms. The fifth
quintile has the highest aggregate dividend payout, both before and after JGTRRA.
Although 48 percent of the firms in the fifth quintile increased their dividend payout ratios
after JGTRRA, the overall average dividend payout ratio was lower, because 51 percent of
these high-dividend-payout firms reduced their dividend payout ratios during the period,
and the reductions by the 51 percent were slightly larger than the increases by the 48
percent.
In the discussion that follows, I distinguish between “mature” firms, which are
older than the median firm in their 3-digit industry, and “young” firms, which are younger
than the industry median. The low- and medium-dividend-payout quintiles (Quintiles 1, 2,
and 3) are mostly composed of young firms. (Young firms make up from 51 percent to 57
percent of the firms in these quintiles.) On the other hand, 73 percent of the firms in
Quintile 4 are mature, along with 79 percent of the firms in Quintile 5. The quintilespecific evidence is consistent with the prediction that mature firms are able and willing to

31

pay more dividends when facing high tax burdens (a top marginal tax rate of 38.6 percent
before JGTRRA) to signal maturity. The separate plots of dividend payout ratios for
young and mature firms in Figure 4(c) manifest a gap between the dividend payout ratios
for young firms and mature firms, although the gap narrowed in the post-JGTRRA period.
This reduction was the result of the increase in average dividend payout ratios of the lowand moderate-dividend-payout young firms, along with the reduction in average payout
ratios of the highest-dividend-payout mature firms.
A similar comparison is made on the differences of yield ratios between young and
mature firms (Figure 4(d)). Yield ratios are calculated as the ratio of dividends to share
prices. The average yield ratio of mature firms moved downward slightly in the preJGTRRA period, and increased after JGTRRA. Thus the graph indicates that mature firms
provide a higher rate of return in dividends, relative to investors’ costs, than young firms.
The spike of yield ratios in 2008, and the fact that in 2009 the ratios don’t fall back to their
pre-2008 levels, are a result of both dividend-smoothing and the plunge in stock prices
during the financial crisis.

5.3. Effects of JGTRRA
To test the significance of the effects of JGTRRA on dividend payout ratios, I
modify Lintner’s (1956) model and include a firm-specific error term to control for
unobserved firm-level heterogeneity.
Payoutit = β 0 + β1 Payouti ,t −1 + ( β 2 + δ Postt ) * ROAi ,t −1 + ε i + uit ,

32

(2)

where Payoutit and Payouti ,t −1 are the current and lagged payout ratios; Postt =1 if t >
2002, and zero otherwise, and ROA is return on assets. In equation (2), β1 measures the
stickiness of dividend payout ratios. It is expected to be positive because of the otherwise
adverse signaling effects of dividend cuts. The effects of JGTRRA on dividend payout
ratios are measured by ߜ. A positive (negative) estimate of ߜ would provide evidence that

payout ratios were raised (lowered) in the post-JGTRRA period. The results of ordinary
least squares regressions, controlling for firm fixed effects, are listed in column (1) of
Table 4. All standard errors have been corrected for firm clustering.
The coefficient of the lagged dividend payout ratio, β1 , is positive and statistically
significant at the one-percent level. This is consistent with the theoretical predictions, as
well as Lintner’s finding of persistence in dividend payments and the well-documented
observation of dividend smoothing. The coefficient of the interaction term between
profitability and the post-JGTRRA period, δ , is positive and statistically significant at the
five-percent level. This indicates that the average payout ratios increased substantially in
the post-JGTRRA period.
As expected, the relation between profitability ( ROAi ,t −1 ) and the payout ratio, β 2 ,
is significantly negative. This suggests that an increase in dividends is associated with a
declining return on assets. This is consistent with the stickiness of dividend payments
since, as profits drop, the payout ratios will increase if dividends payments remain
constant. This is also consistent with the prediction of the life-cycle hypothesis that a firm

33

would tend to increase dividends as it moves from a growth stage to a mature stage, as
mature firms face fewer investment opportunities and have more free cash flow. For
example, DeAngelo, DeAngelo, and Skinner (1994) find that more than two-thirds of the
145 NYSE firms in their sample responded to stalled earnings growth by increasing
dividends.
In addition, the sum of β 2 and δ is positive, which indicates that profitability has a
net positive effect on corporate dividend payout after JGTRRA. This result holds after
controlling for firm maturity, and also for regressions within all of the subgroups of firms
that are discussed in the following sections.
It should be noted that since the lagged dependent variable is used in the regression,
it may cause bias in the estimates as the error term is correlated with the lagged dependent
variable. This bias may be mitigated with the relatively large sample size used in the
regression. An alternative estimation strategy, involving differencing and instrumental
variables, would be a fruitful avenue for further research.

5.4. An Inverse-U-Shaped Relationship between Payout Ratios and Firm Age
Since firms tend to pay more dividends as they grow older15, the estimate of the
effect of JGTRRA on dividend payout ratios could be overstated. As shown in Figure 2(c),
the median firm age increased monotonically from 1996 to 2010. Thus the estimate of δ
15

A discussion of interpreting the results as an age effect or a vintage effect can be found in
Section 5.5.

34

(the parameter for the effect of JGTRRA) could also be capturing the effect of firm age on
dividend payout ratios. To control for the effect of firm age on dividend payout ratios, I
include a quadratic in the age of the firm, with firm age and its squared form
( AGE 2 / 100 ):
Payoutit = β 0 + β1 Payouti ,t −1 + ( β 2 + δ Postt ) * ROAi ,t −1 + β 3 AGEit + β 4 ( AGEit2 / 100) + ε i + uit

.

(3)

The results in column (2) of Table 4 are consistent with the life-cycle hypothesis
that older firms pay more dividends out of earnings. As predicted, the magnitude of the
coefficient δ is reduced after firm age is included in the model. Moreover, the result
shows an inverse-U-shaped relationship between dividend payout ratios and firm age.
Dividend payout ratios initially increase with firm age, at a rate of about 0.5 percent per
year. However, the negative coefficient on the age quadratic means that the positive
relation between firm age and dividend payout ratio is reversed as firms age beyond a
certain point. The estimates shown in column (2) of Table 4 indicate that firms older than
42 years16 will reduce their average payout ratios in the following years, holding
profitability constant. This “aging effect” could reflect the higher agency costs associated
with larger size of organization as firms grow older.17
16

The critical firm age at which dividend payout reaches its maximum is given by
AGE * = − β3 / (2γ /100) = −0.007888 / (2*0.00929 /100) = 42.45.

17

For example, DeAngelo, DeAngelo, and Skinner (1994) study a sample of firms that
reduce dividends. They find that within the three years after the dividend reduction, 87
percent of sample firms engage in contractual renegotiations—with lenders, unions,
government, and/or management.

35

In column (3) of Table 4, we see the results of a regression in which a dummy
variable for mature firms is added to equation (3). The dummy variable for mature firms is
equal to one when a firm is older than the median firm in its 3-digit industry. As predicted
by the life-cycle hypothesis, the coefficient of the dummy variable for mature firms is
positive. It is significant at the one-percent level. On average, the dividend payout ratio is
6.6 percentage points higher for mature firms (firms older than the industry median) than
for young firms (firms younger than the industry median).

5.5. Reactions of High Dividend Payers and Low Dividend Payers
Columns (4) to (8) of Table 4 show the results of quintile-specific regressions. No
significant change is found among firms in Quintile 4 in the post-JGTRRA period. The
firms in Quintile 5 have a decrease in the dividend payout ratio (significant at the tenpercent level). By contrast, the firms in Quintiles 1, 2, and 3 have significant increases in
payout ratios. The estimates of δ are positive and significant for Quintiles 1, 2, and 3, at
the one-percent level.
The similarity of the results for the lower three quintiles is unsurprising given their
similar composition of young and mature firms, as shown in Table 3. To gain further
insights into the difference in payout behavior between young firms and mature firms, I
divide the sample into two subgroups –young and mature firms –and apply model (2) to
the two groups of firms separately. The results are presented in columns (9) and (10) of

36

Table 4.
The estimates of δ are significantly positive (at the one-percent level) for young
firms (i.e., firms younger than the industry median). This finding is consistent with the
theoretical prediction that young firms will increase their dividend payout in response to
the 2003 Tax Act. However, the estimates of δ are insignificant for mature firms (i.e.,
firms older than the industry median). In other words, mature firms do not appear to have
responded significantly to JGTRRA. This can be interpreted in light of the dividend
smoothing theory that increased dividend levels would be difficult to maintain for high
dividend payers. Overall, the regression results are consistent with the matched-pairs ttests.
The results in columns (9) and (10) also help differentiate the age effect from the
alternative interpretation. One alternative interpretation of the age effect discussed in
Section 5.4 could be a vintage effect that firms born before a certain date pay more. If the
vintage effect were the driver of the increases in dividend payout, AGE would be
statistically significant only for young firms (but not for mature firms) because the young
firms have a higher correlation between age and vintage than the mature firms. In contrast,
columns (9) and (10) show that AGE is a statistically significant predictor of dividend
payout for mature firms, while insignificant for young firms. The results cast a great deal
of doubt on the vintage effect as an alternative interpretation for the increases in dividend
payout as firms grow older.

37

5.6. Multinomial Logit Regressions
According to Baker and Kennedy (2002), the 10-year survival rate of firms trading
on the New York Stock Exchange and the American Stock Exchange is only 61 percent. In
the previous literature (for example, Chetty and Saez (2006)), researchers have used
balanced samples to estimate the effects of tax-code changes on corporate dividend policy.
Until now, the analysis here has also used a balanced sample. However, the firms that
survive are more likely to be financially stable and mature firms, and the fact that fewer
public firms survive after 1997 (see Figure 2(a)) could lead to an overestimation of the
effects of policy changes made during the period, including JGTRRA. Chetty and Saez
also conclude that without a direct measure of the entry and exit effects, it is difficult to
make precise inferences about the effects of the 2003 tax cut on the level and concentration
of corporate distributions.
A firm’s dividend policy is not observable if the firm in question has exited the
market, either because of takeover, failure, or other reasons. If a firm that exited the
market had been able to stay in the market, and if the best dividend policy had been to not
pay dividends, then nonlinear least squares regressions would be inefficient, because the
sample variance is conditioning on the observability of the firm’s dividend policy. The
conditional variance could be heteroskedastic, and any model for the conditional variance
would be arbitrary (Wooldridge 2002). Moreover, a Tobit model would be inappropriate
because firms could choose to exit the market due to liquidation, merger, takeover,
privatization, etc., but not necessarily because of the potential that dividends could be

38

negative or zero. For example, if a firm exits because of a merger, it is entirely possible
that, without the merger, the firm still would have paid dividends.
One way to solve this estimation bias problem is to add the choice of firm exit into
the model of corporate dividend policy, and estimate the relative probabilities among
alternatives. In the following maximum likelihood regressions, three different choices can
be made by firms each year: exit, pay no dividends, or pay positive dividends. The
probability of exiting the market is normalized, so that comparisons can be made across the
choices. In year t , let j = 0 if the firm exits the market, j = 1 if the firm pays no
dividends, and j = 2 if the firm pays dividends. The probability of the firm exiting the
market is
2

P ( yi = 0 | x) = 1 / [1 + ∑ exp( xi' β h )] .

(4)

h =1

The probability of the firm being a non-dividend payer or dividend payer is
2

P ( yi = j | x) = exp( xi' β j ) / [1 + ∑ exp( xi' β h )], j=1, 2 .

(5)

h =1

A log-likelihood estimator is used to estimate the parameters. The log-likelihood
function to be maximized is
N

2

ln L = ∑∑ d ij ln P ( yi = j ) ,

(6)

i =1 h = 0

where d ij = 1 if alternative j is chosen by firm i, and 0 if not. For each i , only one of the
d ij ’s is equal to one.

39

The model implies that the log-odds ratios can be computed as

 P( yi = j ) 
'
'
ln 
 = xi ( β j − β k ) = xi β j if k = 0 .
 P( yi = k ) 

(7)

Therefore the estimated odds ratio, Pj / P0 , implies the relative probability of
alternative payout policies, relative to the choice of exiting the market. Odds ratios greater
(less) than one correspond to positive (negative) parameter estimates. All standard errors
have been adjusted for heterogeneity and firm clustering.

5.6.1. Probabilities of Firm Exit and Dividend Policy by Firm Age
To give an intuitive illustration of the different payout responses of young firms and
mature firms to the 2003 Tax Act, Figures 5(a) and 5(b) show the probabilities of a firm
exiting the market and paying dividends, before and after JGTRRA. The probabilities are
estimated by the linear prediction with parameters from logit regressions of probabilities
on firm age and its square. As the graphs show, as firms age, they are monotonically more
likely to pay dividends and less likely to be non-dividend payers. This result, combined
with the previous result that emerged from the earlier analysis of payout ratios, indicates
that as the surviving firms grow older, they are more likely to pay dividends and also pay
more dividends, until after a certain age where the trend between the dividend payout ratio
and firm age reverses.
The probability of firm exit increases in the early stages of the firm life cycle, and
declines in the later stages. The inverse-U-shaped curves in Figure 5(a) and 5(b) indicate

40

that younger firms have higher probabilities of exit, due to competition, financial distress,
lack of managerial experience, or other factors. This pattern is particularly pronounced in
high-tech industries. Older firms that have overcome these problems are less likely to exit.
In Figure 6(a), the probabilities of a firm paying dividends in the pre- and postJGTRRA periods are compared by firm age. Consistent with the theoretical predictions,
young firms are more likely to pay dividends after the 2003 Tax Act. On the other hand,
older firms become less likely to pay dividends than before. This mean reversion is
reflected as a flatter probability curve in the graph. Figure 6(b) shows a similar pattern in
comparing the probabilities of non-dividend-paying around the 2003 Tax Act.

5.6.2. The Effects of Within-Firm Control Variables
Tables 5 and 6 present the estimates of odds ratios and the marginal effects from
pooled multinomial logit regressions, with standard errors corrected for heterogeneity and
firm clustering. The results show that the coefficients for firm age are statistically
significant at the one-percent level among non-payers and payers in both periods. Before
JGTRRA, as the average firm with a 12-year history grows older, it is 0.74 percent more
likely to pay dividends in the next year, and 0.81 percent less likely to be a non-payer,
holding other characteristics constant. After JGTRRA, the average firm with a 12-year
history is 0.65 percent more like to pay dividends in the next year, and 0.73 percent less
likely to be a non-payer.
Comparing the control variables in the pre- and post-JGTRRA periods, the average

41

firm is older in age (12 to 15 years), less profitable, and larger in size of market
capitalization. It has fewer investment opportunities (Tobin’s Q declined from 2.34 to
2.07), more free cash flows, lower leverage, and higher liquidity. The average rate of
economic growth declined substantially, from 3.61 percent before JGTRRA to 1.63 percent
after JGTRRA.
In Tables 5, 6, and 7, I use standardized ROA (abbreviated in the tables as S ROA).
To get standardized ROA for a firm, I subtract the industry average ROA, and then divide
by the industry standard deviation. This controls for the substantial degree of industry
heterogeneity in ROA. As shown in Table 5, the coefficients for standardized ROA and
size of market capitalization are both positive and significant. Their marginal effects
(Table 6) indicate that firms with either higher profitability or a larger size of market
capitalization are more likely to pay dividends and less likely to exit the market or to pay
no dividends.
The marginal effect of profitability on the probability of paying dividends is higher
in the post-JGTRRA period, while the marginal effects on the probabilities of exiting the
market or not paying dividends are both reduced. The marginal effects of size of market
capitalization are slightly different. Although the probability of paying dividends is still
higher for larger firms, the marginal effect of larger size is slightly lower in the postJGTRRA period. These changes in marginal effects imply that smaller and more profitable
firms are the most responsive to changes in dividend policy, while larger and less
profitable firms are less likely to pay dividends after JGTRRA. This is consistent with the

42

mean reversion found in the comparison of younger and older firms.
The coefficient of Tobin’s Q is negative (positive) and significant in the case of
dividend payers (non-payers), which suggests that firms with fewer investment
opportunities are more likely to pay positive dividends. This is consistent with the
implication of the firm life-cycle hypothesis, that dividend increases are often associated
with a shrinking investment opportunity set. The average level of investment opportunities
decreased after JGTRRA. However, the estimated relationship between investment
opportunities and dividend payment still holds. The magnitudes of the marginal effects
turn out to be larger than before JGTRRA.
The coefficient of free cash flow is significant only in the case of non-payers in the
pre-JGTRRA period, and only in the case of dividend payers in the post-JGTRRA period.
The marginal effects imply that free cash flow is negatively related to the probability of
paying dividends. One interpretation of this result is that free cash flow is reduced by past
dividend payments, and is thus negatively related to current dividend payments. For the
same reason, liquidity—a broader measure of financial resources than free cash flow—is
also negatively related to the probability of paying dividends.
As shown in Tables 5 and 6, the significant negative coefficient of leverage in the
probabilities of both dividend policies indicates that firms with high indebtedness are more
likely to exit the market. The marginal effect of indebtedness on the probability of firm
exit is even larger in the post-JGTRRA period. Before JGTRRA, leverage is positively
correlated with the propensity to pay dividends. However, after JGTRRA, firms with high

43

indebtedness are less likely to make corporate distributions.18 This conclusion holds even
though firms have reduced the average level of leverage in the post-JGTRRA period.
The relation between the availability of financial resources and the propensity to
pay dividends is consistent with the firm life-cycle hypothesis. For young firms facing
abundant investment opportunities and a lack of available funds (e.g., limited sources of
external capital), the payout policy was simply cut to fit the available funds. It is also
consistent with the notion of free cash flow being a measure of the ability of a company to
generate internal growth, and young firms are expected to have higher internal growth than
mature firms, and make more corporate distributions in the future. The differences
between mature and young firms are less significant in the post-JGTRRA period. This is
consistent with the mean reversion between young firms and mature firms as the young
firms mature. As the marginal costs of equity decreased after JGTRRA, the difference in
their payout responses between young firms and mature firms has been effectively
narrowed.

5.6.3. The Effects of Other Control Variables
The measure of industry age is significant in most cases, but not in the regression
for the probability of not paying dividends in the post-JGTRRA period. Before JGTRRA,

18

This is consistent with the finding in Bagwell and Shoven (1989) that firms with high
indebtedness are less likely to make corporate distributions.

44

firms in older industries were more likely to pay dividends and firms in younger industries
were less likely to pay dividends. (See Table 6.) After JGTRRA, firms in older industries
are still significantly more likely to pay dividends (at the five-percent level). However, the
difference between firms in younger and older industries has narrowed as the average firms
in most industries have grown older. The marginal effect of an additional year of industry
age on the probability of paying dividends is reduced from 0.8 percent to 0.6 percent, and
its effect on the probability of not paying dividends is changed from -0.9 percent to -0.5
percent.
Firm maturity has a significant effect in most of the regressions. This result
indicates that in the pre-JGTRRA period, holding firm age constant, firms that are
relatively mature in their industry are more likely to pay dividends than firms that are
relatively young in their industry. Depending on the magnitude of the difference in
industry age, if we hold constant the age of the firm, a mature firm in a one-year younger
industry could be more likely to pay dividends than a young firm in a one-year older
industry. An average 12-year-old mature firm in a young industry is 1.0 percent more
likely to pay dividends than a young firm at the same age in an old industry.
However, as we have seen, the coefficients suggest that there is mean reversion
between the payout propensities of mature and young firms in the post-JGTRRA period.
For example, an average 15-year-old young firm in an old industry is 3.2 percent more
likely to pay dividends than a mature firm at the same age in a young industry, holding
other characteristics constant. This result is significant at the ten-percent level. Thus,

45

holding constant firm age and other firm characteristics, young firms in older industries
were more likely to pay dividends than mature firms in younger industries in the postJGTRRA period.
The rate of economic growth has a positive effect on the probability of paying
dividends, and it is significant at the one-percent level in all cases. Unsurprisingly, the
lower rate of economic growth in the post-JGTRRA period is associated with a lower
average level of investment opportunities (implied by Tobin’s Q). Furthermore, the
significant relationship between economic growth and the propensity to pay dividends
indicates that firms are more likely to pay dividends in a good economy, all else equal.
This is consistent with the prediction in Chapter Three that in a good economy, an increase
in dividend payments is more likely to be interpreted as an increase in cash flow (good
news) rather than a decrease in investment opportunities (bad news).
Combining these findings with the above analyses of firm characteristics, we can
summarize the characteristics that are most likely to be associated with paying dividends in
the post-JGTRRA period. Dividend payments are more likely for young firms in older
industries with a longer firm history and a longer industry history, with higher profitability,
lower investment opportunities, larger size of market capitalization, lower free cash flow
(or higher payments in history), and lower leverage.

46

CHAPTER SIX
SENSITIVITY ANALYSIS AND ADDITIONAL TESTS
If the assumption of irrelevance of independent alternatives (IIA) is violated,
multinomial logit regression is inappropriate. Therefore, the assumption of irrelevance of
independent alternatives is tested using Hausman-McFadden and Small-Hsiao tests. The
Hausman-McFadden test statistics are negative. As suggested by Long and Freese (2006),
a negative Hausman-McFadden statistic often provides evidence that the assumption of
independence of irrelevant alternatives has not been violated. The Small-Hsiao test
suggests that omitting the category of payers does not violate the IIA assumption, while
omitting the category of non-payers violates the IIA assumption. Wald tests for combining
alternatives are implemented, and the results reject the assumption that any pair of
alternatives can be combined. These results suggest that it is appropriate to use
multinomial logit techniques in this context.
I also looked into the importance of firm age in high-competition and lowcompetition industries, where firms have differing exit rates. Following Giroud and
Mueller (2010), the degree of competition in the firm’s industry (3-digit SIC) is measured
with a Herfindahl index, H E : H E = ∑ iN=1 si2 , where N is the number of firms in the same 3digit industry, the subscript i identifies the firm, and si is the firms’ market shares,
expressed in percentage points. A higher Herfindahl index indicates lower competitiveness

47

in the industry.
Table 7 replicates the analysis of Table 5 by comparing firms in high-competition
and low-competition industries (i.e., industries with a Herfindahl index below or above the
median for all industries). Fewer firm characteristics are significantly correlated with
dividend policy in the low-competition industries. One interpretation of this result is that,
for an average non-dividend payer in a low-competition industry, the choice between
exiting the market and staying in the market without paying any dividends is less sensitive
to the changes in observed firm characteristics.
In the post-JGTRRA period, the negative relationship between firm age and the
probability of not paying dividends in low-competition industries is reduced in magnitude.
The positive relationship between firm age and the probability of paying dividends in lowcompetition industries is also smaller in magnitude. This implies a mean-reversion pattern
in low-competition industries, as observed in Figure 6(a) and 6(b). Since low-competition
industries are likely to be older industries,19 this result is consistent with the finding in the
last section that young firms in older industries are more responsive to the tax-law change.
Firms in high-competition industries show a different pattern. Firm age has a
smaller positive effect on the probability of paying dividends for firms in high-competition
industries. Firm age also has a larger negative effect on the probability of not paying
dividends. This is probably because the probability of firm exit is larger in industries with

19

In high-competition industries, the average industry age is 9.7. In low-competition
industries, the average industry age is 12.7.

48

higher competition.
To test the sensitivity of the results with respect to the way in which firm age is
measured, multinomial logit regressions similar to the regressions in Table 5 are also
implemented using the listing age (the number of years since the IPO date). The marginal
effects, which are available on request, are qualitatively consistent with the effects reported
above.

49

CHAPTER SEVEN
CONCLUSION
In this study, I investigate whether corporate dividend payout policies responded to
Jobs and Growth Tax Relief Reconciliation Act of 2003 (JGTRRA), and how the responses
varied according to the life-cycle stage of the firm, as well as other firm and industry
characteristics. In particular, I tested (1) whether the decrease in dividend tax rates was
associated with changes in dividend payout, and (2) whether the corporate dividend policy
responses differed among firms of different ages and in different industries. Dividend
payout ratios and the propensity to pay dividends are tested separately.
The test of the payout responses to the dividend tax cut has a lot of complications.
I have included a number of variables that may be relevant to control for the payout
responses of firms, but there is still a great deal of variation not controlled for in the
models. For example, the dividend payout shows strong stickiness, especially for the
established mature firms, due to managers’ concerns regarding market reaction. Also, with
a growing proportion of tax-exempted institutional shareholders and tax-deferred
individual shareholders, firms become less sensitive to dividend taxes. Moreover, the
marginal cost of funds can vary a lot depending on the composition of the marginal sources
of funds. In a low-interest-rate environment and where leverage is high, the cost of funds
is less correlated with the dividend taxes. Another factor that would affect corporate
payout behavior is the expectation of future tax rates. If dividend tax rates are expected to

50

increase after a certain time period, firms may choose to accelerate the payout and move
payments from the high-tax period to the low-tax period. Without controlling for all the
firm attributes, it is difficult to have a measure of the direct impact of dividend tax policy
on corporate payout.
Nevertheless, the findings of this study are supportive of the predictions. First, as
predicted by signaling theory, a lower dividend tax rate is associated with higher dividend
payments. Also, the reduction in the dividend tax rate meant that surviving firms became
more likely to pay dividends. Since the overall survival rate declined after 2003, the
impact of the 2003 tax cut on the likelihood of dividends-paying would be overestimated
without any estimation of the exit of firms. To reduce the selection bias, I include firm exit
as an alternative choice for the firm and use the multinomial logit model to estimate the
likelihood of dividend paying.
These findings are consistent with the signaling theory. But the signaling theory
does not differentiate among firms in the cost of funds. I find the new view and traditional
view of dividend taxation theories are useful in understanding the variation across firms.
As firms have different sources of funds and different composition of the sources, their
marginal costs of funds may have different correlations with the dividend tax rate, and thus
may lead to different reactions to the 2003 tax cut. A firm with a higher proportion of
retained earnings is more likely to behave according to the prediction of the new view,
while a firm with a higher proportion of new issues of equity is more likely to behave
according to the prediction of the traditional view.

51

Among mature firms, whose marginal source of funds is more likely to be from
retained earnings than young firms, and for which the cost of funds is not as much affected
by the dividend tax, the propensity to pay dividends does not appear to be impacted by the
2003 tax cut. This is consistent with the new view of dividend taxation. In the data, firms
in the highest payout ratio group (Quintile 5) experienced a significant decline in their
payout ratios, while firms in Quintile 4 experienced positive but insignificant changes in
payout ratios. This phenomenon is consistent with the predictions of the firm life-cycle
hypothesis that firms tend to pay more dividends as they grow to a mature stage, but lower
dividends as the firms grow to a stage where profitability declines and/or they face more
agency problems. A separate test was performed for mature firms (i.e., firms that are
above the median firm age in their 3-digit industry). This test showed no significant
changes in payout ratios after 2003, but it did find a significant inverse-U-shaped relation
between firm age and payout ratios.
In contrast to the responses of mature firms, increases in dividend payout are
observed in young firms, whose marginal source of funds is more likely to be from new
issues of equity. The marginal source of funds of young firms is close to the assumption of
the traditional view of dividend taxation, and their payout response to the 2003 tax cut is
consistent with the traditional view. Moreover, as predicted by the firm life-cycle
hypothesis, young firms increase their dividend payout when they grow older, as well as in
response to reductions in earnings growth and in the level of investment opportunities. In
the post-JGTRRA period, profitability and investment opportunities were lower, and

52

average firms were older, and young firms made higher dividend payments. For example,
following JGTRRA, low- and moderate-payout firms (Quintiles 1, 2, 3) exhibited larger
increases in dividend payout ratios.
The findings show that firms responded to the realignment of dividend tax rates and
capital-gains tax rates by adjusting their corporate dividend policies. This conclusion is
consistent with the findings in previous papers, which also found that JGTRRA led to
increases in dividends. This relationship still holds after taking the choice of firm exit into
consideration. However, this paper differs from previous papers by differentiating the
responses of firms of different ages. The results show that the increase in aggregate
dividend payout is largely driven by the behavior of young and growing firms, rather than
by the tax-arbitrage behavior of older and mature firms.
Since the corporate payout increased with the reduction in the dividend tax rate, the
total dividend tax revenue would not have decreased as much as the reduction in the
dividend tax rate. However, the number of firms reduced rapidly after the 2003 tax cut,
and this also had a negative effect on the total dividend tax revenues. Therefore, the tax
policy implications of the 2003 tax cut cannot be traced out simply by looking at the
effects on total dividend tax revenues.

53

APPENDICES

54

APPENDIX 1

Tables

55

Table 1. Variable Definitions
Variables
Firm Age

Definitions
Firm age is computed as the difference between year t and the first year in
which the firm appeared on the CRSP/COMPUSTAT Merged Database, plus
one. An alternative measure, the listing age, is the number of years between
the current year and the date of the firm’s initial public offering, plus one.

ROA

Return on Assets (ROA) is computed as the ratio of the firm’s operating
income before depreciation (DATA13) dividend by the book value of assets
(DATA6).

Free Cash Flow

The measure of free cash flow is calculated as the firm’s free cash flow
weighted by its market value of assets. The amount of free cash flow is
calculated by adding net income with depreciation and deferred taxes and
then subtracting dividends paid and capital expenditures. Following Lehn and
Poulsen (1989), the firm’s amount of free cash flow is computed as DATA13
- DATA15 - (DATA16 - DATA35) - DATA19 - DATA21.

Tobin’s Q

Tobin’s Q (or market-to-book value) is calculated as the market value of the
firm’s assets divided by their book value (DATA6).

Market Value of
Assets

The market value of assets is calculated as book value of assets (DATA6)
minus book value of common equity (DATA60) plus market value of
common equity (DATA25xDATA199).

Size

Size is calculated as the natural logarithm of the firm’s market capitalization
(DATA25 x DATA199) to the equal-weighted average of all firms’ market
capitalization in year t.

Leverage

Leverage is the ratio of total liabilities (DATA181) to market value of assets
(as defined above).

Liquidity

Liquidity is calculated as the firm’s cash and short-term investment
(DATA1) divided by the market value of assets.

Industry Age

Industry age is the median listing age of all firms in the same 3-digit SIC
industry in year t.

Maturity

Binary variable equal to 1 if the firm age is greater than or equal to industry
age.

GDP Growth

The relative change in the U.S. gross domestic product. The data are from the
U.S. Bureau of Economic Analysis.

56

Table 2. Descriptive Statistics
Pre-JGTRRA

Firm Age
Return on Assets
Tobin’s Q
Free Cash Flow
Size
Leverage
Liquidity
Industry Age
Maturity
GDP Growth
No. of Observations

Mean

SD

11.77
0.05
2.34
0.91
-2.54
32.95
10.42
9.39
0.51
3.61
39,271

10.09
0.26
3.45
111.20
2.19
23.93
16.28
5.20
0.50
1.11

Overall

Post-JGTRRA

Median

95%

Mean

SD

Median

95%

Mean

SD

Median

8.07
0.11
1.50
0.08
-2.66
28.63
4.98
7.98
1.00
4.10

33.91
0.28
6.29
0.52
1.33
78.39
37.86
21.16
1.00
4.80

15.88
0.04
2.05
5.08
-2.38
31.67
13.85
13.60
0.52
1.75
27,518

11.81
0.27
1.94
726.34
2.18
22.35
18.82
6.12
0.50
1.99

12.31
0.10
1.53
0.08
-2.43
27.49
8.04
12.11
1.00
2.50

41.92
0.27
4.96
0.48
1.40
75.72
46.12
26.00
1.00
3.50

13.55
0.05
2.21
2.72
-2.47
32.39
11.91
11.22
0.52
2.80
66,789

11.06
0.27
2.89
485.36
2.19
23.27
17.51
5.99
0.50
1.81

10.08
0.10
1.51
0.08
-2.55
28.12
6.23
10.02
1.00
3.00

57

95%
37.13
0.28
5.69
0.51
1.37
77.48
41.50
23.54
1.00
4.80

Table 3. Matched-Pairs t-Tests on Dividend Payout Ratios of Restricted Sample
Differences

Means
No. of
Obs

PreJGTRRA

PostJGTRRA

%
Positive

Overall

16,362

0.089

0.118

36

Quintile
1
2
3
4
5

3,272
3,272
3,273
3,272
3,273

0.000
0.000
0.006
0.115
0.324

0.038
0.033
0.050
0.164
0.307

Life-Cycle Stage
Young
6,954
Mature
9,408

0.047
0.121

0.084
0.144

%
Negative

No. of
Mature
Firms

%
Mature
Firms

Diff>0

9,408

57

Diff>0
Diff>0
Diff>0
Diff>0
Diff<0

1,397
1,419
1,613
2,378
2,601

43
43
49
73
79

Mean
Difference

Std.
Dev.

t-statistic

H0

21

0.029

0.154

24.16***

22
20
30
62
48

0
0
14
38
51

0.038
0.033
0.043
0.048
-0.017

0.109
0.101
0.127
0.155
0.230

19.94***
18.69***
19.37***
17.71***
4.22***

29
42

12
37

0.037
0.023

0.137
0.165

22.75*** Diff>0
13.48*** Diff>0

Note: *** denotes significance at the 1% level.
** denotes significance at the 5% level.
* denotes significance at the 10% level.

58

Table 4. The Effects of JGTRRA and Firm Age on Dividend Payout Ratios of Restricted Sample
The table estimates fixed effects OLS regressions with standard errors corrected for heterogeneity and firm clustering. The dependent variable is
current dividend payout ratios Payoutt. t-statistics are in parentheses. The symbols ***, **, and * indicated statistical significance levels of 1%, 5%,
and 10%, respectively. The sample period is 1993-2010.

(1)
Payoutt-1

0.112

(2)
***

(4.78)

ROAt-1

-0.091
0.136

***

(4.70)
***

(4.07)

Post*
ROAt-1
Age

0.110

(3)

-0.041

(5.19)

0.061

***

***

Age /100

-0.005
(2.34)

Mature
Constant

(28.19)

R-squared
No. of Obs.

0.061

***

0.005

***

-0.005

0.117
15,406

***

Quintile 2

Quintile 3

Quintile 4

Quintile 5

(4)

(5)

(6)

(7)

(8)

0.115

0.084

0.018

0.049

0.132

(1.61)

(1.03)

(0.29)

(1.32)

(4.66)

(1.91)

-0.043

-0.107

-0.073

-0.069

-0.049

(1.64)

(2.43)

(0.97)

(0.34)

(3.38)

0.085

-0.345

(2.85)

(1.48)

(1.80)

0.002

0.005

0.018

-0.024

***

(2.35)
***

***

(4.76)
**

(2.33)
0.066
(13.24
)

0.095

**

(2.73)

(4.77)
2

-0.041
(2.09)

(2.73)
0.005

***

(4.70)

(2.09)
***

0.110

Young
Firms
(9)

Quintile 1

0.121

***

0.083

***

0.125

**

***

(4.63)

(2.94)

0.000

0.003

(0.33)

(2.66)

(1.17)

(1.61)

(1.97)

0.006

0.001

0.004

-0.002

-0.020

(1.46)

(0.52)

(0.95)

(0.57)

-0.025

***

***

*

0.084

0.074

Mature
Firms
(10)
*

0.119
(4.45)

***

-0.021
(0.47)

***

0.034

(4.01)

(0.63)

0.001

0.008

(0.77)

(3.49)

0.006

-0.009

(1.74)

(1.15)

(2.61)

-0.015

-0.079

0.023

*

*

***

0.017

-0.025

(1.38)

(2.28)

(1.82)

(2.50)

(1.62)

(0.23)

(0.52)

(2.14)

(0.64)

0.137
15,406

0.101
15,406

0.066
3,114

0.021
3,066

0.011
3,076

0.011
3,090

0.007
3,060

0.101
6,545

0.125
8,861

**

-0.020

*

-0.027

59

**

***

**

-0.023

***

***

Table 5. Multinomial Logit Regression
This table presents estimates of pooled multinomial logit regressions with the base outcome of firm exit
using the unrestricted sample. All standard errors have been adjusted for heterogeneity and firm clustering.
The symbols ***, **, and * indicated statistical significance levels of 1%, 5%, and 10%, respectively.
The sample period is 1993-2010.
Pre-JGTRRA
Odds
Coefficient
z-statistic
Ratios
Case 1: Non-Dividend Payer
Within Firm
Firm Age
-0.021
0.979
5.24 ***
S ROA
0.111
1.117
4.60 ***
1.94 *
Tobin’s Q
0.019
1.019
Free Cash Flow
-0.004
0.997
2.69 ***
Size
0.033
1.034
2.16 **
Leverage
-0.014
0.986 13.51 ***
Liquidity
-0.001
0.999
0.50

Coefficient

Post-JGTRRA
Odds
z-statistic
Ratios

-0.020
0.115
0.068
-0.004
0.058
-0.014
-0.004

0.980
1.122
1.070
0.996
1.060
0.986
0.996

4.88
4.39
2.94
1.34
3.43
10.44
3.81
1.50
0.51

***
***
***
***
***
***

Cross Firm
Industry Age
Maturity

-0.026
-0.485

0.974
0.616

4.38 ***
7.40 ***

0.010
0.039

1.010
1.039

Macro
GDP Growth
Constant

-0.151
4.306

0.859

6.79 ***
35.85 ***

-0.080
3.215

0.923

5.22 ***
23.12 ***

Case 2: Dividend Payer
Within Firm
Firm Age
S ROA
Tobin’s Q
Free Cash Flow
Size
Leverage
Liquidity

0.045
0.659
-0.370
-0.046
0.631
-0.010
-0.006

1.046
1.932
0.691
0.955
1.880
0.990
0.994

8.71
13.72
4.75
1.64
25.56
3.63
1.64

0.035
0.585
-0.278
-0.054
0.542
-0.013
-0.025

1.036
1.794
0.757
0.948
1.719
0.987
0.975

6.79
12.37
5.44
2.52
23.29
5.89
7.04

Cross Firm
Industry Age
Maturity

0.049
-0.358

1.051
0.699

6.73 ***
3.65 ***

0.050
-0.194

1.052
0.823

6.20 ***
1.81 *

Macro
GDP Growth
Constant

-0.105
3.238

0.900

4.50 ***
11.11 ***

-0.035
3.215

0.965

2.13 **
23.12 ***

Observations
Log Likelihood
Pseudo R-squared

39,271
-21,680.32
0.280

***
***
***
***
***

27,518
-17,280.19
0.223
60

***
***
***
**
***
***
***

Table 6. Marginal Effects of Multinomial Logit Regression
Panel A: Marginal Effects of Independent Variables on Firms’ Probability of Not Paying Dividends
Pre-JGTRRA

Post-JGTRRA
Marginal
Effects

Mean

-0.081
-0.051
0.148
0.544
-0.136
-0.027
0.008

-0.008
-0.052
0.042
0.004
-0.062
-0.001
0.001

15.288
-0.005
2.070
5.630
-2.450
31.444
13.944

11.509
0.976
1.996
765.18
2.168
22.50
19.26

-0.094
-0.053
0.096
0.709
-0.128
-0.018
0.046

-0.008
-0.054
0.048
0.006
-0.059
-0.001
0.002

5.144
0.500

-0.048

-0.009
-0.039

13.137
0.509

5.744
0.500

-0.027

-0.005
0.032

1.104

-0.014

-0.013

1.633

2.032

-0.020

-0.010

Mean

SD

Within Firm
Age
S ROA
Tobin's Q
Free Cash Flow
Size
Leverage
Liquidity

11.575
-0.009
2.339
0.914
-2.616
32.795
10.545

9.947
0.981
3.453
111.4
2.162
32.95
16.44

Cross Firm
Industry Age
Maturity (0->1)

9.364
0.505

Macro
GDP growth

3.608

-+SD/2

SD -+SD/2

Marginal
Effects

Panel B: Marginal Effects of Independent Variables on Firms’ Probability of Paying Dividends
Pre-JGTRRA

Post-JGTRRA
Marginal
Effects

Mean

SD

-+SD/2

Marginal
Effects

0.074
0.063
-0.156
-0.609
0.152
0.008
-0.010

0.007
0.064
-0.044
-0.005
0.069
0.000
-0.001

15.288
-0.005
2.070
5.630
-2.450
31.444
13.944

11.509
0.976
1.996
765.18
2.168
22.50
19.26

0.087
0.065
-0.095
-1.000
0.149
-0.002
-0.056

0.008
0.067
-0.047
-0.007
0.068
-0.000
-0.003

5.144
0.500

0.043

0.008
0.010

13.137
0.509

5.744
0.500

0.033

0.006
-0.032

1.104

0.004

0.004

1.633

2.032

0.011

0.005

Mean

SD

Within Firm
Age
S ROA
Tobin's Q
Free Cash Flow
Size
Leverage
Liquidity

11.575
-0.009
2.339
0.914
-2.616
32.795
10.545

9.947
0.981
3.453
111.4
2.162
32.95
16.44

Cross Firm
Industry Age
Maturity (0->1)

9.364
0.505

Macro
GDP growth

3.608

-+SD/2

Note: Following the multinomial logit regression results in Table 5, the marginal effect were calculated as
the change in probability associated with a marginal change in one explanatory variable at the mean
values of the other variables.

61

Table 7. Sensitivity Test 																																										
High-Competition vs. Low-Competition Industries
The table estimates pooled multinomial logit regressions with the base outcome of firm exit using the
unrestricted sample. An industry with a Herfindahl index lower than the median Herfindahl in any given
year is categorized as High-competition industry. All standard errors have been adjusted for heterogeneity
and firm clustering. The symbols ***, **, and * indicated statistical significance levels of 1%, 5%, and
10%, respectively. The sample period is 1993-2010.
Pre-JGTRRA
Low-Competition
Case 1: Non-Dividend Payer
Within Firm
Firm Age
-0.025
S ROA
0.110
Tobin’s Q
-0.000
Free Cash Flow
-0.006
Size
0.026
Leverage
-0.014
Liquidity
-0.002

***
***

***

Post-JGTRRA

High-Competition

-0.016
0.115
0.035
-0.003
0.040
-0.014
0.001

**
***
**
***
*
***

Low-Competition

-0.021 ***
0.119 ***
0.083 **
-0.007
0.054 **
-0.015 ***
-0.004
*

High-Competition

-0.021
0.120
0.073
-0.000
0.057
-0.013
-0.004

***
***
**
***
**
***
***

Cross Firm
Industry Age
Maturity

-0.029
-0.508

***
***

-0.013
-0.466 ***

Macro
GDP Growth
Constant

-0.137
4.374

***
***

-0.167 ***
4.135 ***

-0.084 ***
3.367 ***

-0.075 ***
2.980 ***

Case 2: Dividend Payer
Within Firm
Firm Age
S ROA
Tobin’s Q
Free Cash Flow
Size
Leverage
Liquidity

0.040
0.549
-0.430
-0.053
0.638
-0.015
-0.004

***
***
***

0.052
0.905
-0.293
-0.060
0.628
-0.005
-0.003

0.036
0.530
-0.242
0.032
0.564
-0.020
-0.014

0.032
0.708
-0.213
-0.062
0.522
-0.003
-0.027

Cross Firm
Industry Age
Maturity

0.027
-0.245

***
*

0.101 ***
-0.474 ***

0.031 ***
-0.052

0.048 ***
-0.335 **

0.105 ***
3.999 ***

-0.116 ***
2.136 ***

-0.049 **
3.445 ***

-0.035
2.201 ***

Macro
GDP Growth
Constant

***
***

62

***
***
**
***

0.002
0.066

0.025 **
0.025

***
***
***
***
***
***

***
***
***
**
***
***

APPENDIX 2

Figures

63

Figure 1(a). The Separating Equilibrium with Dividend Signaling

64

Figure 1(b). The Simplified Separating Equilibrium

65

Figure 1(c). The Dividend Expectation of Growing and Mature Firms

66

Figure 1(d). Impact of JGTRRA on Corporate Dividend Payout

67

Figure 2(a). Number of Firms (left scale); Number of Firms Paying Dividends
(left scale); Percentage of Firms Paying Dividends (right scale), 1978-2010
100
90

6,000

80
5,000

70
60

4,000

50
3,000

40

2,000

30
20

1,000
0
1978

10
0
1983

1988
1993
1998
2003
2008
Number of Firms
Number of Firms Paying Dividends
Percentage of Firms Paying Dividends

68

Percentage of Firms Paying Dividends

Number of Firms and Dividend Payers

7,000

Figure 2(b). Percentage of Firms Initiating Dividends, 1978-2010

4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1978

1983

1988

1993

69

1998

2003

2008

Figure 2(c). Median Firm Age, 1978-2010
16
14

Firm Age (in Years)

12
10
8
6
4
2
0
1978

1983

1988

1993

70

1998

2003

2008

Figure 2(d). Median Listing Age
16
14

Firm Age (in Years)

12
10
8
6
4
2
0
1978

1983

1988

1993

71

1998

2003

2008

300
200
100
0

Count of Firms

400

Figure 3. Count of Firms by Number of Years Paying Dividends,
1993-2010

0

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18

Number of Dividend-Paying Years during 1993-2010

72

Figure 4(a). Average Dividend Payout Ratios (left scale); Dividend Per Share
(right scale); Earnings Per Share (right scale) 1993-2010
3.5

0.16

3.0

0.14
2.5

Payout Ratios

0.12
0.10

2.0

0.08

1.5

0.06

1.0

0.04
0.5

0.02

0.00
0.0
1993 1995 1997 1999 2001 2003 2005 2007 2009
Payout Ratios
Div per share
Earnings per share

73

Dividends and Earnings Per Share

0.18

Figure 4(b). Average Dividend Payout Ratios for Quintiles of Firms,
1993-2010
0.45
0.40
0.35
Payout Ratios

0.30
0.25
0.20
0.15
0.10
0.05
0.00
1993

1995

1997

1999

2001

2003

Quintile 1

Quintile 2

Quintile 4

Quintile 5

74

2005

2007

2009

Quintile 3

Figure 4(c). Average Dividend Payout Ratios for Young and Mature Firms
0.20
0.18

Payout Ratios

0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
1993 1995 1997 1999 2001 2003 2005 2007 2009
Young Firms

Mature Firms

75

Figure 4(d). Average Yield Ratios for Young and Mature Firms
0.020
0.018
0.016

Yield Ratios

0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
1993 1995 1997 1999 2001 2003 2005 2007 2009
All Firms

Young Firms

76

Mature Firms

Figure 5. Predicted and Observed Probabilities of Firm Exit and Dividend
Payout Policy by Firm Age		
Figure 5(a) Probabilities of Firm Exit and Dividend Policy (Pre-JGTRRA)

Figure 5(b) Probabilities of Firm Exit and Dividend Policy (Post-JGTRRA)

77

Figure 6. Mean Reversion in Probabilities of Firm Dividend Payout Policy by
Firm Age	
Figure 6(a) Dividend Paying by Firm Age (Pre- and Post- JGTRRA)

Figure 6(b) Non-Dividend Paying by Firm Age (Pre- and Post-JGTRRA)

78

REFERENCES

79

REFERENCES

Auerbach, Alan, Michael Devereux, and Helen Simpson. 2010. “Taxing Corporate Income.” In
James Mirrlees, et al. (eds.) Dimensions of Tax Design: 837-93. Oxford, UK: Oxford University
Press.
Auerbach, Alan. 1979. “Wealth Maximization and the Cost of Capital.” Quarterly Journal of
Economics 93(3): 433-46.
Auerbach, Alan. 1983. “Shareholder Tax Rates and Firm Attributes.” Journal of Public
Economics 21(2): 107-27.
Bagwell, Laurie S. and John B. Shoven. 1989. “Cash Distributions to Shareholders.” Journal of
Economic Perspectives 3(3): 129-40.
Baker, George P. and Robert E. Kennedy. 2002. “Survivorship and the Economic Grim Reaper.”
Journal of Law, Economics, and Organization 18(2): 324-61.
Baker, Malcolm and Jeffrey Wurgler. 2004. “A Catering Theory of Dividends.” Journal of
Finance 59(3): 1125-65
Bates, Thomas W., Kathleen M. Kahle, and René M. Stulz. 2009. “Why Do U.S. Firms Hold So
Much More Cash Than They Used to?” Journal of Finance 64(5): 1985-2021.
Bernheim, B. Douglas and Adam Wantz. 1995. “A Tax-Based Test of the Dividend Signaling
Hypothesis.” American Economic Review 85(3): 532-51.
Blanchard, Oliver, Florencio Lopez-de-Silanes, and Andrei Shleifer. 1994. “What Do Firms Do
With Cash Windfalls?” Journal of Financial Economics 36(3): 337-360.
Blouin, Jennifer, Jana Raedy, and Douglas Shackelford. 2011. “Dividends, Share Repurchases,
and Tax Clienteles: Evidence from the 2003 Reductions in Shareholder Taxes.” The Accounting
Review 86(3): 887-914.
Bradford, David. 1981. “The Incidence and Allocation Effects of a Tax on Corporate
Distributions.” Journal of Public Economics 15(1): 1-22.
Brav, Alon, John Graham, Campbell Harvey, and Roni Michaely. 2005. “Payout Policy in the
21st Century.” Journal of Financial Economics 77(3): 483-527
Brown, Jeffrey R., Nellie Liang, and Scott Weisbenner. 2007. “Executive Financial Incentives
and Payout Policy: Firm Responses to the 2003 Dividend Tax Cut.” Journal of Finance 62(4):
1935-65.
80

Chetty, Raj and Emmanuel Saez. 2005. “Dividend Taxes and Corporate Behavior: Evidence
from the 2003 Dividend Tax Cut.” Quarterly Journal of Economics 120(3): 791-833.
Chetty, Raj and Emmanuel Saez. 2006. “The Effects of the 2003 Dividend Tax Cut on Corporate
Behavior: Interpreting the Evidence.” American Economic Review 96(2): 124-29.
Chetty, Raj and Emmanuel Saez. 2010. “Dividend and Corporate Taxation in an Agency Model
of the Firm”, American Economic Journal: Economic Policy 2(3), 1-31.
DeAngelo, Harry, Linda DeAngelo, and Douglas J. Skinner. 1994. “Accounting Choice in
Troubled Companies.” Journal of Accounting and Economics 17(1-2): 113-43.
DeAngelo, Harry, Linda DeAngelo, and Douglas J. Skinner. 2004. “Are Dividends Disappearing?
Dividend Concentration and the Consolidation of Earnings.” Journal of Financial Economics
72(3): 425-56.
DeAngelo, Harry, Linda DeAngelo, and René M. Stulz. 2006. “Dividend Policy and the
Earned/Contributed Capital Mix: A Test of the Life-Cycle Theory.” Journal of Financial
Economics 81(2): 62-82.
Desai, Mihir and Austan Goolsbee. 2004. “Investment, Overhang and Tax Policy.” Brookings
Papers on Economic Activity 2004 (2): 285-338.
Dhaliwal, Dan, Linda Krull, Oliver Zhen Li, and William Moser. 2005. “Dividend Taxes and
Implied Cost of Equity Capital.” Journal of Accounting Research 43(5): 675–708.
Easterbrook, Frank. 1984. “Two Agency-Cost Explanations of Dividends.” American Economic
Review 74(4): 650-59.
Fama, Eugene F. and Harvey Babiak. 1968. “Dividend Policy: An Empirical Analysis” Journal
of the American Statistical Association 63(324): 1132-61.
Fama, Eugene F. and Kenneth R. French. 2001. “Disappearing Dividends: Changing Firm
Characteristics or Lower Propensity To Pay?” Journal of Financial Economics 60(1): 3-43.
Feldstein, Martin. 1970. “Corporate Taxation and Dividend Behavior.” Review of Economic
Studies 37(1): 57-72.
Gillan, Stuart, and Laura Starks. 2007. “The Evolution of Shareholder Activism in the United
States.” Journal of Applied Corporate Finance 19(1), 55-73.
Giroud, Xavier and Holger M. Mueller. 2010. “Does Corporate Governance Matter in
Competitive Industries?” Journal of Financial Economics 95(3): 312-31.
Gordon, Roger and Martin Dietz. 2008. “Dividends and Taxes.” In Alan J. Auerbach and Daniel
N. Shaviro (eds.) Institutional Foundations of Public Finance: Economic and Legal Perspectives:
204–24. Cambridge, MA: Harvard University Press.

81

Gourio, Francois, and Jianjun Miao. 2011. “Transitional Dynamics of Dividend and Capital
Gains Tax Cuts.” Review of Economic Dynamics 14(2): 368-383.
Graham, John R. and Alok Kumar. 2006. “Do Dividend Clientele Exist? Evidence on Dividend
Preferences of Retail Investors.” The Journal of Finance 61(3): 1305-36.
Grullon, Gustavo, Roni Michaely, and Bhaskaran Swaminathan. 2002. “Are Dividend Changes a
Sign of Firm Maturity?” Journal of Business 75(3): 387-424.
Hanlon, Michelle and Shane Haitzman. 2010. “A Review of Tax Research.” Journal of
Accounting and Economics 50(2-3): 127-78.
Harberger, Arnold. 1962. “The Incidence of the Corporate Income Tax.” Journal of Political
Economy 70(3): 215-40.
Ittner, Christopher D., Richard A. Lambert and David F. Larcker. 2003. “The Structure and
Performance Consequences of Equity Grants to Employees of New Economy Firms.” Journal of
Accounting and Economics 34(1-3): 89-127.
Jensen, Michael. 1986. “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.”
American Economic Review 76(2): 323-29.
John, Kose and Joseph Williams. 1985. “Dividends, Dilution, and Taxes: A Signalling
Equilibrium.” Journal of Finance 40(4): 1053-70.
Julio, Brandon and David Ikenberry. 2004. “Reappearing Dividends.” Journal of Applied
Corporate Finance 16(4): 89-100.
King, Mervyn. 1977. Public Policy and the Corporation. London, UK: Chapman and Hall.
Korinek, Anton and Joseph E. Stiglitz. 2009. “Dividend Taxation and Intertemporal Tax
Arbitrage.” Journal of Public Economics 93(1-2): 142-59.
Lang, Larry H.P. and Robert H. Litzenberger. 1989. “Dividend Announcements: Cash Flow
Signalling vs. Free Cash Flow Hypothesis?” Journal of Financial Economics 24(1): 181-92.
Leary, Mark T. and Roni Michaely. 2011. “Determinants of Dividend Smoothing: Empirical
Evidence.” Review of Financial Studies 24(10): 3197-3249.
Lehn, Kenneth and Annette B. Poulsen. 1989. “Free Cash flow and Stockholder Gains in Going
Private Transactions.” Journal of Finance 44(3): 771-87.
Lintner, John. 1956. “Distribution of Incomes of Corporations among Dividends, Retained
Earnings and Taxes.” American Economic Review 46(2): 97-113.
Long, J. Scott and Jeremy Freese (2nd ed.). 2006. Regression Models for Categorical Dependent
Variables Using STATA. College Station, TX: Stata Press.

82

Miller, Merton H. and Franco Modigliani. 1961. “Dividend Policy, Growth, and the Valuation of
Shares.” The Journal of Business 34(4): 411-33.
Opler, Tim, Lee Pinkowitz, René Stulz, and Rohan Williamson. 1999. “The Determinants and
Implications of Corporate Cash Holdings.” Journal of Financial Economics 52(1): 3-46.
Pettit, R. Richardson 1972. “Dividend Announcements, Security Performance, and Capital
Market Efficiency.” Journal of Finance 27 (5): 993-1007.
Poterba, James M. and Lawrence H. Summers. 1985. “The Economic Effects of Dividend
Taxation.” In Edward Altman and Marti Subrahmanyam. (eds.) Recent Advances in Corporate
Finance: 227-284. Homewood, IL: Richard D. Irwin Publishers.
Robin, Ashok. 1991. “The Impact of the 1986 Tax Reform Act on Ex-Dividend Day Returns.”
Financial Management 20(1): 60-70.
Scholz, John. 1992. “A Direct Examination of the Dividend Clientele Hypothesis.” Journal of
Public Economics 49(3): 261-85
Sialm, Clemens. 2009. “Tax Changes and Asset Pricing.” The American Economic Review 99(4):
1356-83.
Sinn, Hans-Werner. 1991. “The Vanishing Harberger Triangle.” Journal of Public Economics
45(3): 271-300.
Wooldridge, Jeffrey M. 2002. Econometric Analysis of Cross Section and Panel Data.
Cambridge, MA: MIT Press.
Zodrow, George. 1991. “On the ‘Traditional’ and ‘New’ Views of Dividend Taxation.” National
Tax Journal 44(4): 497-509.

83