THE CHANGING NATURE OF CORPORATE DISTRIBUTIONS

AND ITS IMPLICATIONS FOR INVESTORS

by Susana Yu*

Abstract
Because cash dividends no longer appear to be a complete measure of corporate distributions, tests of
dividend irrelevance based on cash dividends may not be well specified. We compare the performance
of long/short investment strategies based on portfolios of high, low-, and zero-payout firms and on
each of three measures of corporate distributions. The strategy with the highest risk-adjusted returns is
long in high-payout/short in low-payout stocks formed on the basis of the most comprehensive
measure of payout, including dividend yield, and net share repurchases. We show that corporate
profitability and investment opportunities are significantly associated with firms’ payout policies, and
this helps explain the relatively strong performance of the zero-payout stocks.
Keywords: dividends, distribution policy, share repurchase, share issuance, value investing strategy

1. Introduction across different industries. We supplement this

with a similar analysis across different firm sizes.
In academic research, there has been increasing Overall, we verify the increasing reliance of U.S.
interest in analyzing the implications of the shift firms on share repurchase (and share issuances) to
from cash dividends towards share repurchase. One distribute cash to shareholders since 1973. How-
of the most important issues is what the shift may ever, there are some exceptions, including firms in
mean for investors. According to Modigliani and the telecommunications and utilities sectors, and
Miller’s argument that dividend policy is irrelevant, the smallest firms.
the shift should not matter to them. However, We next evaluate the effectiveness of invest-
their irrelevance proposition holds only under very ment strategies utilizing three measures of share-
restrictive assumptions, including the absence of holder payout within different economic sectors
taxes and transactions costs, and perfect information. and size categories. These measures are (1) the
Because cash dividends no longer appear to be a traditional dividend yield; (2) a payout yield that
full measure of distributions, it is not clear whether adds repurchases to dividends; and (3) a still more
tests of dividend irrelevance versus dividend pref- comprehensive measure, called “net payout yield,”
erence based on cash dividends are well specified. which also takes cash flows into the firm from new
This consideration motivates our inquiry into the stock issues into account.
general issue of dividend irrelevance, and into We find that high payout portfolios significantly
several more specific questions of corporate prac- outperform low payout portfolios for all three mea-
tice, and effects of policy on firm valuation and sures. In particular, we show that an investment
stock performance. To test for this, we compare strategy that is long in high-payout stocks and
stock performance of high, low-, and zero-payout short in low-payout stocks results in significant
firms. risk-adjusted profits. The specific strategy with
We begin our empirical analysis with a compre- the highest risk-adjusted returns is the one based
hensive overview of changes in corporate distribu- on the most comprehensive measure of payout,
tion policies from 1973–2007. We further analyze which includes dividend yield, share repurchases,
these trends across ten major sectors in order to and share issuances. These results apply not only
determine whether the general trend from cash to the largest sample of firms on average, but
dividends to share repurchase applies uniformly that it also applies fairly consistently across the

* Associate Professor of Finance, Montclair State University

Vol. 56, No. 1 (Spring 2011) 89

different sectors. Overall, the evidence is inconsis- reaction tends to follow shares of firms that an-
tent with the dividend policy irrelevance theory nounce to start or increase (decrease) cash divi-
and supports the traditional view that shareholders dend or to buy back (stop buying back) their
have a preference for stocks that pay something shares.
out, whether in the form of dividends or share Wansley, Lane, and Sarkar (1989) suggest,
repurchases. according to their survey results, that management
By contrast, when we compare the high payout are drawn to use share repurchase to acquire
portfolios to a portfolio of stocks with zero pay- undervalued stock and to show investors their con-
outs, there is no difference in performance. To help fidence about their companies’ earnings outlooks.
explain this finding, we determine the extent Baker, Powell, and Veit (2003) reach a similar
to which corporate profitability and investment conclusion following their survey. Ikenberry,
opportunities drive corporate distribution policies Lakonishok, and Vermaelen (1995) document em-
across all sectors and sizes. We find that both fac- pirically that shares of repurchase announcing
tors are significantly associated with firms’ payout firms are usually undervalued at the time when
policies, and conclude that this helps explain the repurchases on their shares are announced and that
fact that the stock performance of firms with zero average abnormal four-year buy-and-hold return
payout policies perform as well on average as measured after the initial share repurchase an-
stocks with high-payout policies. nouncement is 12%.
The rest of the paper is organized as follows. Agency problems are a byproduct of the separa-
We review related literature and develop our test- tion of corporate ownership and management.
able hypotheses in Sections 2 and 3. We describe Shareholders are burdened with the costs of moni-
our data and sample in Section 4, and then docu- toring or limiting management investment activ-
ment trends in the use of repurchase and cash divi- ities. According to the agency cost hypothesis,
dends by industry sector and firm size. We test investors, anticipating management’s tendency of
whether a profitable trading strategy can be formed overinvestment, are willing to pay higher prices for
based on high and low payout firms in Section 5A. shares of companies with generous shareholder
We test for the relation between payout and firm distribution records, and announcements of divi-
profitability and growth opportunities in Section dend initiation or increases are associated with
5B. The final section concludes. positive abnormal returns.
In a survey by Baker, Gallagher, and Morgan
(1981), a majority of their respondents agree that
2. Review of Related Literature share repurchase is an effective way for corporation
to use idle cash, especially when the firm’s share
Three areas of existing financial literature help price is under pressure. Easterbrook (1984) argues
motivate this study: the information content of that as companies pay out cash dividend, they may
shareholder distribution policy and policy changes, need to raise new capital to fund new projects and,
the recent growth in the use of share repurchase as therefore, invite more scrutiny by potential investors
a cash payout vehicle, and the impact of firms’ to their investment decisions. Grullon and Michaely
characteristics on corporate distribution policy. (2004) suggest that the market reacts more favorably
Shareholder distribution policy and changes in to share repurchase announcements made by compa-
distribution policies have long been proven in- nies with declining investment opportunities than
formative and can impact a firm’s share price. similar announcements made by other firms. As
The signaling and agency cost hypotheses are profitable investment alternatives dwindle, an in-
two of the most important theories of corporate crease in cash payout to shareholders signals man-
distributions. agement’s commitment to reduce the agency costs
According to the signaling hypothesis, dividend of free cash flows.
policy and share repurchase programs are reflec- Although financial managers generally do not
tions of management expectations regarding future believe share repurchase can fully replace cash
earnings and cash flows of their firms and can dividends, the once dominant role in shareholder
cause investors to adjust their expectations. A di- distribution policy played by cash dividends is now
rect implication is that positive (negative) price shared with repurchase decisions. According to

90 THE AMERICAN ECONOMIST

Fama and French (2001), the proportion of firms Although cash dividends are no longer dominant
paying cash dividends falls from one third of all over share repurchase, these methods of distributing
firms in 1978 to 21% in 1999 in the U.S. In a cash to shareholders are not perfect substitutes. For
research survey and synthesis study, Weigand and one, there are some differences in the information
Baker (2009) conclude that improved economic conveyed by dividends and repurchases. According
conditions (share prices), and the flexibility of to Ikenberry and Vermaelen (1996), dividends signal
repurchases relative to dividends may have con- financial managers’ long-term optimism about the
tributed to this trend. Grullon and Michaely permanence and expected stability of earnings be-
(2002) study the rising trend of share repurchase cause they represent a firmer commitment to return
in the U.S. and find that not only has share repur- excess capital to shareholders. Julio and Ikenberry
chase become an important form of payout, it also (2004) find that repurchases convey more informa-
has directly competed for funds that could other- tion about short-term cash flows and allow manage-
wise have been paid out as cash dividends. ment to signal their belief regarding their
Noting the shift from dividend to repurchase in undervalued shares. As tax code, regulation, and
shareholder distribution policy, practitioners have economic conditions evolve, this declining trend for
gradually incorporated the share repurchase infor- cash dividends may be reversed. Julio and Ikenberry
mation in developing value investment strategies. (2004) observe a resurgence of dividend-paying
Originally dividend yield is one of several factors firms between 2002 and 2004.
used by value investors seeking undervalued Firm characteristics also affect firms’ distribu-
stocks because they expect an undervalued stock’s tion policies. According to Baker, Farrelly, and
price will eventually rise to its intrinsic value. Edelman (1985), managers of utility firms see
Keim (1986) endorses the U-shaped relation dividend policy differently from managers of
between dividend yields and stock returns, in manufacturing and wholesale/retail firms. They
which only stocks paying either zero dividends or suggest that regulation is the main cause behind
high dividend yields experience abnormally high this. Examining the long-run stock performance
returns. According to O’Shaughnessy (2005), port- following share repurchase announcements by
folios composed of high dividend-yield stocks out- firms in the 1980’s, Ikenberry et al. (1995) con-
perform portfolios of low dividend-yield stocks. clude that positive and significant abnormal returns
For instance, stocks in the highest dividend yield after share repurchase announcements occur to
decile generate an average annual return of ‘value’ stocks, but not to ‘glamour’ stocks, as the
15.98%, 2% higher than the 13.91% from stocks market is more inclined to believe that under-
in the lowest dividend yield decile in the 1951– valuation is the reason behind a share repurchase
2003 period.1 Acknowledging that there are times announcement made by a ‘value’ firm than a simi-
when financial managers favor one over another lar announcement made by a ‘non-value’ firm.
between cash dividend and share repurchase, According to Fama and French (2001), small firms
O’Shaughnessy (2005) combines the dividend face more serious issues of information-asymmetry
yield of the stock and the net share buyback and higher costs of issuing securities than larger
activity by the company to create an overall mea- firms, and are, therefore, less likely to pay
sure of shareholder yield. He has since developed dividends. They also note that dividends have
value strategies based on shareholder yields and become increasingly concentrated among large-
managed portfolios following the revised value capitalization companies.
strategies. Researchers also acknowledge indirectly that
In a more recent study, Boudoukh, Michaely, industry and other characteristics of firms affect
Richardson, and Roberts (2007, hence BMRR) sug- these firms’ distribution policies. By design,
gest that combining dividend yield, share repurchase O’Shaughnessy (2005), when testing the effective-
yield, and new issuances contains more information ness of value strategy based on dividend yield,
about the cross section of expected stock returns excludes utility stocks from his sample because of
than information containing only dividend yield. their exceptionally high dividend yields. Fama and
They show that regression of returns on either pay- French (2001) exclude utilities to avoid the criti-
out yield or net payout yield generates higher R2’s cism that their dividend decisions are a byproduct
than regression of returns on dividend yield. of regulation.

Vol. 56, No. 1 (Spring 2011) 91

 3. Development of Hypotheses H3Null: There is no association of corporate
profitability and investment opportu-
In view of this literature, we focus on three nities with a firm’s distribution policy.
related research issues. The first is to establish
empirically the extent of differences in distribution If we are able to reject this null hypothesis, we will
policies across firms in different industries and of be able to verify the significance of these factors in
different sizes. Next we consider whether there are firms’ distribution policies, and further, suggest an
differences in stock performance between portfo- explanation for the fact that the stocks of zero-
lios based on high, low, and zero payout patterns.2 payout firms perform as well as those of high-
The following hypotheses are based on the irrele- payout firms on average.
vance theorem of Miller and Mogidliani (1961).
Our first null hypothesis is: 4. Data and Preliminary Analysis

H1Null: The stock performance of firms is not We begin our empirical analysis with an over-
different whether they have low payout view of the major characteristics of the distribution
or high payout distribution policies. policies of the firms in the U.S. In particular, we
document the average percentage of firms with
If we reject this hypothesis, we are able to dividends, repurchases, or equity issuances in a
conclude that distribution policy is related to firm given fiscal year by sector and by firm size. To
performance. If we find that stocks with higher illustrate trends in distribution policies over time,
payouts have better performance, then this would we employ a time series regression model to illus-
suggest that a profitable investment strategy can be trate how U.S. firms have shifted their distribution
formed. Interestingly, this strategy underlies the policy from cash dividends to share repurchase and
“Dogs of the Dow” theory. According to it, an issuances over our sample period.
investor should buy the ten stocks in the Dow Data adopted in this study mainly come from
Jones Industrials index with the highest dividend the CRSP monthly return database and the merged
yield. The rationale is that among the large firms in CRSP/COMPUSTAT database with fiscal years
this index, the stocks with the highest yield are between 19733 and 2007. We include only com-
stocks whose prices are cyclically low. Our second mon stock, those issues with CRSP share codes of
null hypothesis is: 10 or 11. The two-digit Global Industry Classifica-
tion System (GICS) sector codes are from COM-
H2Null: The stock performance of firms is not PUSTAT.4 The GICS system has ten sectors:
different whether they have zero pay- Energy (ENRG), Materials (MTRL), Industrials
out or high payout distribution policies. (INDU), Consumer Discretionary (CD), Consumer
Staples (CS), Healthcare (HLTH), Financials (FIN),
If we find no difference in firm performance, Information Technology (TECH), Telecommunica-
then no profitable investment strategy is possible. tion Services (TELS), and Utilities (UTIL).
Instead, we need to broaden our view of firm’s We define three shareholder distribution yields
motivations for their particular policies. This leads according to the study of BMRR. They are:
to our next question, which concerns factors that 1. Dividend yield ¼ dividends / market capitali-
are associated with corporate distribution policies. zation
We hypothesize that firms that have higher distri- 2. Payout yield ¼ dividend yield þ repurchases /
butions are characterized by two factors: on aver- market capitalization
age they are more profitable and (or) they have 3. Net payout yield ¼ payout yield – equity issu-
fewer investment and growth opportunities. Fur- ances / market capitalization
ther, we allow for differences across industries,
not only because profitability and growth differ, Market capitalization at the end of each fiscal year is
but also because firms in different industries are a product of the number of common shares outstand-
subject to different degrees and types of regulation. ing (COMPUSTAT mnemonic CSHO) and the clos-
Our hypothesis in null form is: ing price (COMPUSTAT mnemonic PRCC_F).

92 THE AMERICAN ECONOMIST

Dividends include the total dollar amount of divi- market value of equity by sectors and size quintiles
dends declared (COMPUSTAT mnemonic DVC). are also summarized in this table.
Repurchases are defined as the total expenditures In Panel A, Table 1, average annual number of
on the purchase of common and preferred stock firms increases rapidly from 3,525 in the first sub-
(COMPUSTAT mnemonic PRSTKC) plus any re- period to 6,085 in the third sub-period before de-
duction in the value of the preferred stocks out- clining to 4,822 in the final sub-period, a reminder
standing (COMPUSTAT mnemonic PSTKRV). of the burst of the technology bubble. While the
Equity issuances are defined as the sale of common average annual number of firms is lower in the
and preferred stock (COMPUSTAT mnemonic 2000–2007 period, the average annual aggregate
SSTK) minus any increase in the value of the pre- market value of equity increases from $7,033
ferred stock outstanding (COMPUSTAT mnemonic billion in the third sub-period to $12,933 billion in
PSTKRV). the 2000–2007 period (Panel B, Table 1).
We use COMPUSTAT data to form measures Although INDU (17.3%), CD (20.0%), HLTH
of profitability and investment (or investment (9.9%), FIN (15.3%), and TECH (15.4%) have the
opportunities), as in Fama and French (2001). The five highest average percentages of firms for the
first profitability measure of a group of firms, entire period (1973–2007) in Panel A, the actual
Et /At, is defined as the ratio of aggregate earnings winners are HLTH, FIN, and TECH, as their per-
before interest5 to aggregate assets (COMPUSTAT centages at least double between the first and the
mnemonic AT). The second profitability measure, last sub-period. Moreover, their annual average
Yt/BEt, is defined as the aggregate earnings avail- percentages of market value of equity in Panel B
able for common stock6 over aggregate book also more than double between the first and the
equity.7 Asset growth rate is the first measure fourth sub-periods. For example, the average mar-
of investment opportunities of a group of firms, ket value of equity of the TECH sector increases
dAt/At, and is defined as the aggregate change in 18.0% in the 2000–2007 sub-period from 8.9% in
assets over aggregate assets. The second measure the 1973–1979 sub-period. At the same time, the
of investment opportunities, Vt /At, is the ratio of other seven sectors consistently lose share through
the aggregate market value8 to the aggregate book the four sub-periods in Panel A and Panel B.
value of assets. Panel C reveals that the top 20% of firms (largest
Following BBMR, we merge all fiscal year-end quintile) account for around 90.4% of aggregate
accounting variables in year t  1 from COMPU- market value of equity and the top 40% of firms
STAT with the CRSP monthly returns for July9 of account for at least 97.0% of the entire market value
year t to June of year t þ 1 to ensure that the of equity. A critical implication is that the small firm
accounting information is known prior to the stock effect may bias the empirical results from portfolios
returns. A buy-and-hold strategy is taken, so portfo- consisting of stocks with equal dollar investments.
lios are held for exactly one year until the following How U.S. firms shift their distribution policy
June, when portfolios are reconstructed on the basis from dividend to share repurchase and issuances is
of new accounting information. To ensure consisten- shown in Table 2. We employ a least squares regres-
cy across analyses, we also apply the same trims of sion model to estimate the annual (fiscal year: t)
extreme values as BBMR. We trim the upper 5% of decay (or change) in percentage (PCNTt) of firms
the dividend yield and payout yield distributions, with dividends, repurchases, or equity issuances:
and the upper and lower 2.5% of the net payout yield
distribution, to mitigate the effects of outliers. PCNTt ¼ at þ bt ðt  1973Þ þ et : ð1Þ
From 1973 to 2007, the structure and composi-
tion of the U.S. equity market has shifted, accord- Negative estimated b coefficients are expected
ing to Table 1, from industrials and manufacturing for the case of dividends, while positive estimated
to services and technology. The average annual b coefficients are expected for the case of repur-
number of firms and average annual aggregate chases or issuances, according to Fama and French
market value of equity are summarized for the (2001) and other prior researchers.10
entire period (1973–2007) and four sub-periods The consistent negative estimated slope terms in
(1973–1979, 1980–1989, 1990–1999, and 2000– Panel A and widespread positive estimated slope
2007). The distributions of the number and the terms in Panels C, D, and E support the conjecture

Vol. 56, No. 1 (Spring 2011) 93

 TABLE 1.
Characteristics of the Sample by Sector and Firm Size
Panel A: Average Annual Number of Firms and Percentage of Firms in Each Sector
Fiscal Years ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
1973–2007 4,904 5.3 6.3 17.3 20.0 5.6 9.9 15.3 15.4 1.3 3.7
1973–1979 3,525 5.6 9.7 22.1 24.6 8.9 4.8 9.6 8.8 0.8 5.2
1980–1989 4,754 6.7 6.9 19.2 21.0 5.6 7.7 13.4 14.2 1.0 4.3
1990–1999 6,085 4.3 5.2 15.2 19.0 4.6 11.9 18.1 17.1 1.6 3.0
2000–2007 4,822 4.4 4.0 13.1 15.9 4.0 14.7 19.3 20.5 1.7 2.4
Panel B: Average Annual Total Market Value of Equity (ME) ($billions) and
Percentage of ME in Each GICS Sector
Fiscal Years ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
1973–2007 5,896 10.1 6.3 12.5 13.9 9.8 9.9 13.6 12.3 5.8 6.0
1973–1979 774 16.0 9.7 13.2 14.6 9.2 6.9 8.6 8.9 5.7 7.2
1980–1989 1,915 12.2 7.4 14.6 14.8 8.8 7.8 10.7 9.7 6.1 7.9
1990–1999 7,033 6.6 5.2 11.5 13.8 11.4 11.2 15.0 12.8 7.2 5.4
2000–2007 13,933 6.8 3.2 10.4 12.2 9.5 13.5 19.7 18.0 3.6 3.3
Panel C: Average Annual Total ME ($billions) and Percentage of ME in Each Size Quintile
Fiscal Years ALL Largest Qntl 5 Qntl 4 Qntl 3 Qntl 2 Smallest Qntl 1
1973–2007 5,896 90.4 6.6 2.1 0.8 0.2
1973–1979 774 88.8 7.5 2.4 1.0 0.3
1980–1989 1,916 89.5 7.2 2.2 0.8 0.2
1990–1999 7,033 91.7 5.7 1.8 0.6 0.2
2000–2007 13,933 91.4 5.9 1.9 0.7 0.2
All firms covered in the COMPUSTAT Fundamental Annual Data (COMPUSTAT) with CRSP share code of 10 or
11 between fiscal years 1973 and 2007 are included. Each firm is assigned to one of ten sectors according to the
Global Industry Classification System (GICS). They are Energy (ENRG), Materials (MTRL), Industrials (INDU),
Consumer Discretionary (CD), Consumer Staples (CS), Healthcare (HLTH), Financials (FIN), Information Tech-
nology (TECH), Telecommunication Services (TELS), and Utilities (UTIL). Each firm is also assigned to a size
quintile based on its fiscal-year end market value of equity. The column for “ALL” firms contains the average
annual number of firms or average total market value of equity, while other columns contain the percentage of firms
assigned to each sector or size quintile.

that more firms adopt share repurchases and share tors or size quintiles. Judged by the estimated
issuances to supplement their dividend policy. slopes in Panel A, CD, HLTH, and TELS lost
Only a few non-positive estimated slopes occur dividend-paying firms at a pace faster than other
for share repurchases in TELS and UTIL (Panel sectors did between 1973 and 2007.
B), and the smallest size quintile (Quintile 1) (left Large firms are more likely to pay dividends,11
panel of Panel D), but no similar exception occurs buy back their shares of common stocks, and issue
for share issuances (Panel C and right panel of equity than small firms, according to Panel D. In
Panel D). An R2 of 0.10 or less appears only 7 out the entire period (1973–2007), an average 78.0%
of 43 cases in Table 2; they are FIN in Panel A, (16.5%) of firms in the largest (smallest) quintile
MRTL and UTIL in Panel B, UTIL in Panel C, and pay dividends, an average 49.8% (25.8%) of firms
size quintiles 1 (smallest) through 3 in the left- in the largest (smallest) quintile buy back shares
hand column of Panel D. (as seen in left side of Panel D), and an average
Firms in MTRL (59.9%), CS (58.9%), FIN 69.0% (44.4%) of firms in the largest (smallest)
(66.1%), UTIL (90.1%), and largest size quintile quintile issue equities (as seen in right side of
(Quintile 5) (78.0%) are more likely to pay divi- Panel D). Large firms clearly provide larger distri-
dends to their shareholders than firms in other sec- butions to their shareholders than small firms.

94 THE AMERICAN ECONOMIST

 TABLE 2.
Percentage of Firms with Dividends, Repurchases, and New Issuances by Sector and Firm Size
Panel A: Average Annual Percentage of Firms with Dividends by Sector
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average % 44.1 33.8 59.9 46.6 40.4 58.9 23.3 66.1 19.4 43.9 90.1
R2 0.66 0.44 0.50 0.74 0.65 0.77 0.77 0.00 0.73 0.77 0.77
Intercept 63.32 46.31 71.99 66.58 61.13 79.16 46.79 67.33 38.02 77.41 97.13
Slope 1.13 0.73 0.71 1.18 1.22 1.19 1.38 0.07 1.10 1.97 0.42
Panel B: Average Annual Percentage of Firms with Repurchases by Sector
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average % 34.9 34.3 39.9 37.9 38.1 41.8 28.8 24.7 30.9 40.7 56.8
R2 0.30 0.17 0.07 0.21 0.44 0.57 0.47 0.26 0.57 0.17 0.09
Intercept 30.38 30.09 36.67 33.43 29.22 33.38 23.76 13.63 22.51 45.68 62.12
Slope 0.27 0.25 0.19 0.26 0.52 0.50 0.30 0.65 0.49 0.29 0.31
Panel C: Average Annual Percentage of Firms with Issuances by Sector
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average % 60.8 61.2 59.3 63.1 61.2 58.7 77.6 26.6 77.3 73.0 74.4
R2 0.77 0.40 0.65 0.82 0.86 0.81 0.83 0.40 0.76 0.33 0.04
Intercept 41.23 45.37 43.81 43.36 37.02 39.97 59.43 3.97 58.15 63.98 71.18
Slope 1.15 0.93 0.91 1.16 1.42 1.10 1.07 1.33 1.13 0.53 0.19
Panel D: Average Annual Percentage of Firms with Repurchases/Issuances by Size Quintile
Percentage of Firms with Repurchases Percentage of Firms with Issuances
Largest Qntl Qntl Smallest Largest Qntl Smallest
Qntl 5 4 Qntl 3 2 Qntl 1 Qntl 5 4 Qntl 3 Qntl 2 Qntl 1
Average
% 49.8 37.3 32.7 28.9 25.8 69.0 67.9 64.8 58.0 44.4
R2 0.70 0.44 0.05 0.01 0.02 0.73 0.92 0.76 0.61 0.66
Intercept 36.07 30.12 30.89 27.96 26.85 54.11 45.97 45.17 38.08 22.77
Slope 0.81 0.42 0.11 0.06 0.06 0.88 1.29 1.16 1.17 1.27
All firms covered in the COMPUSTAT Fundamental Annual Data (COMPUSTAT) with CRSP share codes of 10 or
11 between fiscal years 1973 and 2007 are included. Dividends include the total dollar amount of dividends declared
(DVC). Repurchases are defined as the total expenditure on the purchase of common and preferred stock (PRSTKC)
plus any reduction in the value of the preferred stocks outstanding (PSTKRV) in COMPUSTAT. Equity issuances
are defined as the sale of common and preferred stock (SSTK) minus any increase in the value of the preferred stock
outstanding. A least squares regression model is adopted to estimate the annual (fiscal year: t) decay (or change) in
percentage (PCNTt) of firms with dividends, repurchases, or equity issuances.
PCNTt ¼ at þ bt ðt  1973Þ þ et

5. Empirical Results separate the firms that did not distribute any cash
to their shareholders in a fiscal year from the
A. Trading Strategies Based on ones that did distribute some cash, whether in
Payout Patterns the form of cash dividends or share repurchases.
We then sort these firms into deciles, for each
In this section we examine the performance size group or sector. Finally, we form three pay-
differentials between high, low, and zero payout out sub-groups: high payout (top 3 deciles), me-
portfolios. We first form the group of zero pay- dium payout (middle 4 deciles) and low payout
out firms in the entire universe, as well as in (bottom 3 deciles). All stocks in a portfolio re-
each size group or sector. This enables us to ceive an equal dollar amount of investment, and

Vol. 56, No. 1 (Spring 2011) 95

the portfolios are rebalanced annually in June, Dogs of Dow theory is to be supported, the high
following BBMR. payout portfolio must outperform both the low and
Monthly returns of a high payout portfolio are zero payout portfolios.
compared to those of the corresponding low payout First, test results based on the all common
portfolio to determine whether the theory the Dogs stocks in the COMPUSTAT/CRSP database
of the Dow works. A significant average monthly (Panel A) and a non-Financials/Utilities subset of
return difference supports the theory. A three- common stocks (Panel B) are presented in Table 3.
factor Fama-French regression model is applied to The results of the three-factor Fama-French regres-
test the risk-adjusted returns (portfolio return R1t, sion model generally confirm that high payout
or R2t versus the risk-free Rft) of each portfolio, portfolios outperform low payout portfolios with a
using the market excess return (RMt  Rft), the positive intercept which is significantly different
small minus big factor return (SMBt), and the high from zero, but do not outperform zero payout port-
minus low factor return (HMLt). folios consistently. The significant negative esti-
mated coefficients of both RMt  Rft and SMBt
R1t  Rft ¼ a þ b1 ½RMt  Rft  are as expected since high payout stocks are gener-
ð2Þ
þ b2 SMBt þ b3 HMLt þ et ally larger in market capitalization and have lower
betas than low or zero payout stocks. The signifi-
ðR1t  Rft Þ  ðR2t  Rft Þ ¼ ðR1t  R2t Þ cant positive estimated coefficients of HMLt in
Table 3 can be explained by why value investing
¼ a þ b1 ½RMt  Rft  þ b2 SMBt ð3Þ strategy tends to include high payout stocks (for
þ b3 HMLt þ et their high BV/MV ratios). As results in Panel B
(non-FIN/UTIL) closely match those in Panel A
We also evaluate the difference between returns of of Table 3, we continue to use all COMPUSTAT/
a high payout portfolio (dividend, payout, or net CRSP common stocks in the following empirical
payout) and those of a zero payout portfolios. If the analysis. Additionally, since high payout portfolios

TABLE 3.
Monthly Return Differentials among High, Low, and Zero Payout Portfolios
Panel A: Based on All COMPUSTAT/CRSP Common Stocks
Dividend Yield Payout Yield Net Payout Yield
High vs. Low High vs. Zero High vs. Low High vs. Zero High vs. Low High vs. Zero
Intercept 0.21% *** 0.18% 0.25% *** 0.18% 0.57% *** 0.02%
RMt  Rft 0.2472 *** 0.2843 *** 0.2117 *** 0.2184 *** 0.2568 *** 0.1141 ***
SMBt 0.1243 *** 0.7804 *** 0.1936 *** 0.6255 *** 0.5641 *** 0.4645 ***
HMLt 0.2718 *** 0.3640 *** 0.2736 *** 0.2423 *** 0.6038 *** 0.0328
Panel B: Based on All Non-FINANCIALS/UTILITIES COMPUSTAT/CRSP Common Stocks
Dividend Yield Payout Yield Net Payout Yield
High vs. Low High vs. Zero High vs. Low High vs. Zero High vs. Low High vs. Zero
Intercept 0.14% ** 0.08% 0.28% *** 0.22% * 0.58% *** 0.07%
RMt  Rft 0.1388 *** 0.1731 *** 0.1706 *** 0.1672 *** 0.2247 *** 0.0750 **
SMBt 0.0652 *** 0.6925 *** 0.1183 *** 0.5165 *** 0.4765 *** 0.4193 ***
HMLt 0.3373 *** 0.3953 *** 0.2772 *** 0.2168 *** 0.5736 *** 0.0368
High (low) payout portfolio consists of stocks in the top (bottom) 30% yield distribution, while zero yield portfolio
consists of stocks with no cash dividend, repurchase, or share issuance. Portfolios are rebalanced in June from 1974
and 2008 and held from July until the following June. An equal-dollar investment is made in each stock in each
portfolio. A three-factor Fama-French model is then adopted to test the statistical significance of the intercept term:
ðR1t  Rf t Þ  ðR2t  Rf t Þ ¼ ðR1t  R2t Þ ¼ a þ b1 ½RMt  Rf t  þ b2 SMBt þ b3 HMLt þ et

***(,**(,*)): Significant at 1(,5(,10))% level.

96 THE AMERICAN ECONOMIST

fail to outperform zero payout portfolios in general, share repurchases, and share issuances. The first
we analyze them separately.12 expectation is that more profitable firms will re-
Table 4 documents return differential between turn more cash to their shareholders than less
high and low payout portfolios formed within each profitable firms. The second expectation is that
of the ten GICS sectors. In Panel A, using dividend firms with fewer investment opportunities will re-
yield as the payout measure, the long high divi- turn more cash to their shareholders than firms
dend yield and short low dividend yield strategy with more investment opportunities. We employ a
only works for two sectors because the intercept least squares regression model to study the rela-
term is significantly positive for only two sectors tionship between payout (whether dividend yield,
of stocks: INDU (0.17%) and TECH (0.33%). payout yield, or net payout yield) and measures of
Contrary to other sectors, the TECH sector has a corporate profitability and investment opportu-
significant positive estimated SMBt coefficient of nities jointly. We estimate the slope coefficients
0.1344. in Equation (2) and test for averages (across 35
The long high payout and short low payout fiscal years: 1973 to 2007 where i is defined as
strategy works for more sectors when payout yield the particular firm in the sample) of the regression
(Panel B) and net payout yield (Panel C) are used intercepts and slopes for statistical significance:
to rank stocks in each sector. The strategy works
very well for ENRG, CD, HLTH, and TECH in Yieldi ¼ a þ b1  Profitabilityi
Panel B, but does not work for FIN, TELS, and
UTIL. The strategy based on net payout yield in þ b2  Investment Opportunitiesi þ ei
Panel C works well for all but the TELS. ð4Þ
Overall, it helps to broaden the definition of
cash distribution to shareholders in a value invest- The results presented in Table 6 (overall sample
ing strategy. Also, a value strategy based on cash and by sector) and Table 7 (by size quintile) are
distribution to shareholders seems to work better in based on the following independent variables: the
sectors that are not FIN, TELS, or UTIL, the three firm’s ratio of earnings to assets (Ei,t/Ai,t) and its
traditionally highly regulated sectors. market to book ratio (Vi,t/Ai,t).13 Both Tables 6 and
Similarly, net payout yield provides the more 7 show averages (across 35 fiscal years, 1973 to
support to the theory of the Dogs of the Dow than 2007) of the regression intercepts and slopes and
dividend yield and payout yield do in Table 5 t-statistics for the averages. The null hypothesis is
and the improvement of the intercept term is espe- that firms that are more profitable (larger Ei,t/Ai,t)
cially true for smaller size quintiles. While the or with less investment opportunities (smaller
intercept term of the largest size quintile raises to Vi,t/Ai,t) tend to have more generous distribution
0.27% in Panel C from 0.22% in Panel A, the policies.14
intercept term of the quintile 2 increases from In Panel A, Table 6, when dividend yield is the
0.19% in Panel A, to 0.47% in Panel B, and to dependent variable, all average estimated intercept
0.91% in Panel C. Even for the smallest quintile, terms (ALL plus 11 sectors), ranging from 0.65 in
the intercept term becomes positive in Panel B and TECH to 6.27 in UTIL), are significantly greater
Panel C from negative in Panel A. It suggests than zero at 1% level according to corresponding
that share repurchases and issuances by firms t-statistics. The average estimated b1 coefficients
are associated with the performance of their of corporate profitability are also significantly
shares, especially for firms with smaller market greater than zero at 1% or 5% level for 9 out of
capitalizations. 11 cases, except for FIN and UTIL, two highly
regulated sectors. The average estimated b2 coeffi-
cients of corporate investment opportunities are
B. Key Factors that Drive Payouts: also significantly negative at 1% or 5% level for
Profitability and Growth Opportunities 9 out of 11 cases, except for ENRG and MTRL,
which are significantly negative only at 10% level.
In this section, we test whether corporate prof- In all, corporate dividend policies are influenced
itability and investment opportunities are asso- by firms’ profitability and investment opportu-
ciated with corporate policies regarding dividend, nities and their regulatory environment.

Vol. 56, No. 1 (Spring 2011) 97

98
TABLE 4.
Monthly Return Differentials Between High and Low Payout Portfolios, by Sector
Panel A: Between High and Low Dividend Yield Portfolios
ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Intercept 0.22% 0.02% 0.17% * 0.10% 0.17% 0.12% 0.09% 0.33% * 0.21% 0.01%
RMt – Rft 0.1342 ** 0.0596 * 0.1529 *** 0.1340 *** 0.1198 *** 0.1213 *** 0.1729 *** 0.3191 *** 0.4312 *** 0.2017 ***
SMBt 0.0281 0.1050 ** 0.0809 *** 0.0836 ** 0.0560 0.1516 *** 0.0209 0.1344 ** 0.1233 0.1023 **
HMLt 0.3617 *** 0.1776 *** 0.2344 *** 0.1871 *** 0.1248 *** 0.2171 *** 0.1569 *** 0.5023 *** 0.2305 ** 0.2064 ***
Panel B: Between High and Low Payout Yield Portfolios
ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Intercept 0.40% ** 0.31% ** 0.23% *** 0.41% *** 0.24% * 0.39% ** 0.11% 0.34% ** 0.16% 0.07%
RMt – Rft 0.1581 *** 0.0509 0.1594 *** 0.1902 *** 0.1457 *** 0.1383 *** 0.0799 *** 0.2037 *** 0.4122 *** 0.0606 **
SMBt 0.1142 * 0.1645 *** 0.1028 *** 0.1455 *** 0.2361 *** 0.0701 0.0078 0.0306 0.2453 ** 0.1584 ***
HMLt 0.1852 *** 0.1334 *** 0.1807 *** 0.0746 ** 0.0447 0.0416 0.1795 *** 0.1228 ** 0.0466 0.3114 ***
Panel C: Between High and Low Net Payout Yield Portfolios
ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Intercept 0.47% ** 0.59% *** 0.66% *** 0.72% *** 0.80% *** 0.47% *** 0.32% *** 0.64% *** 0.56% 0.40% ***
RMt – Rft 0.1112 ** 0.0599 0.1868 *** 0.1841 *** 0.2189 *** 0.1317 *** 0.1273 *** 0.2802 *** 0.5439 *** 0.1485 ***
SMBt 0.2025 *** 0.3384 *** 0.3219 *** 0.3352 *** 0.4021 *** 0.6039 *** 0.1506 *** 0.3200 *** 0.6418 *** 0.2009 ***
HMLt 0.2214 *** 0.1992 *** 0.2671 *** 0.2063 *** 0.0928 * 0.3771 *** 0.2124 *** 0.3041 *** 0.3825 *** 0.0867 **
High (low) payout portfolio consists of stocks in the top (bottom) 30% yield distribution in each sector. Portfolios are rebalanced in June from 1974 and 2008
and held from July until next June. An equal-dollar investment is made in each stock in the portfolio. A three-factor Fama-French model is then adopted to test
the statistical significance of the intercept term:
ðR1t  Rf t Þ  ðR2t  Rf t Þ ¼ ðR1t  R2t Þ ¼ a þ b1 ½RMt  Rf t  þ b2 SMBt þ b3 HMLt þ et
The GICS sector structure consists of 10 sectors – Energy (ENRG), Materials (MTRL), Industrials (INDU), Consumer Discretionary (CD), Consumer Staples
(CS), Healthcare (HLTH), Financials (FIN), Information Technology (TECH), Telecommunication Services (TELS), and Utilities (UTIL).
***(,**(,*)): Significant at 1(,5(,10))% level.

THE AMERICAN ECONOMIST

 TABLE 5.
Monthly Return Differentials Between the High and Low Payout Portfolios, by Size Quintile
Panel A: Between High and Low Dividend Yield Portfolios
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Avg. Mth Return 0.30% 0.06% 0.10% 0.05% 0.07%
Intercept 0.22% ** 0.13% 0.16% 0.19% * 0.05%
RMt  Rft 0.2134 *** 0.2630 *** 0.1940 *** 0.1972 *** 0.1025 ***
SMBt 0.0528 * 0.1847 *** 0.1928 *** 0.1735 *** 0.1129 **
HMLt 0.4685 *** 0.2652 *** 0.2027 *** 0.0033 0.1235 **
Panel C: Between High and Low Payout Yield Portfolios
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Avg. Mth Return 0.36% 0.10% 0.15% 0.32% 0.12%
Intercept 0.22% ** 0.15% 0.22% ** 0.47% *** 0.27% *
RMt  Rft 0.2045 *** 0.2648 *** 0.2255 *** 0.2155 *** 0.2100 ***
SMBt 0.0437 0.1363 *** 0.1733 *** 0.2687 *** 0.1236 **
HMLt 0.6078 *** 0.2835 *** 0.2111 *** 0.0565 0.0496
Panel C: Between High and Low Net Payout Yield Portfolios
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Avg. Mth Return 0.46% 0.56% 0.61% 0.78% 0.32%
Intercept 0.27% ** 0.52% *** 0.63% *** 0.91% *** 0.46%
RMt  Rft 0.2082 *** 0.3564 *** 0.3170 *** 0.3061 *** 0.2917
SMBt 0.2803 *** 0.3950 *** 0.4870 *** 0.5636 *** 0.3136
HMLt 0.8957 *** 0.7679 *** 0.6367 *** 0.4237 *** 0.2150
High (low) yield portfolio consists of stocks in the top (bottom) 30% yield distribution in each size quintile.
Portfolios are rebalanced in June from 1974 and 2008 and held from June until next June. An equal-dollar
investment is made in each stock in the portfolio. A three-factor Fama-French model is then adopted to test the
statistical significance of the intercept term:
ðR1t  R2t Þ ¼ a þ b1 ½RMt  Rf t  þ b2 SMBt þ b3 HMLt þ et

***(,**(,*)): Significant at 1(,5(,10))% level.

When payout yield or dividend yield plus variable, the estimated intercept term is 1.38
share repurchase yield is used as the dependent (t-statistic ¼ 1.07) for TECH and positive but
variable, the corporate profitability becomes less not statistically significant for INDU, HLTH, and
influential according to Panel B, Table 6. The TELS, a direct result of their dependence on share
average estimated b1 coefficient is –1.06 (t-statis- issuances (see Panel C, Table 2). Among these
tic ¼ 0.29) in the ALL column and is signifi- sectors, average estimated coefficients of corporate
cantly positive at 1% or 5% level only in 6 profitability are also not significantly positive for
sectors: ENRG, MTRL, INDU, CD, CS, and INDU, TECH, and TELS, and average estimated
FIN. The average estimated b2 coefficients of coefficients of corporate investment opportunities
corporate investment opportunities are still signif- are not significantly negative for INDU and TELS.
icantly negative at 1% or 5% level for almost all All other columns receive significantly positive
columns but HLTH. Factors other than corporate average coefficients for corporate profitability fac-
profitability and investment opportunities become tor and significantly negative average coefficients
more influential in share repurchase decision than for corporate investment opportunities with only
in divided decision. one exception. The average estimated corporate
In Panel C, Table 6, when net payout yield (or profitability coefficient of CS is not significantly
payout yield plus share issuances) is the dependent different from zero.

Vol. 56, No. 1 (Spring 2011) 99

 TABLE 6.
The Relation among Distribution Yields to Shareholders, Corporate Profitability, and Investment
Opportunities, by Sector
Panel A: Dependent Variable is Dividend Yield
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average a 1.84 1.18 2.29 1.62 1.68 1.96 0.69 2.97 0.65 2.60 6.27
t_statistic 12.49 7.31 15.13 11.87 10.23 13.66 9.12 13.52 8.27 7.54 9.96
Average b1 1.61 4.47 3.42 3.80 2.96 4.71 1.49 0.78 0.83 4.37 0.07
t_statistic 10.55 2.08 4.96 5.08 6.67 4.80 5.14 0.69 5.00 2.57 0.01
Average b2 1.00 3.07 2.85 1.82 2.40 2.96 0.89 1.33 0.61 3.34 3.43
t_statistic 6.62 1.48 1.64 3.26 2.43 2.84 4.45 2.13 3.13 3.63 4.03
Panel B: Dependent Variable is Payout Yield
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average a 4.42 4.31 4.58 4.24 5.32 4.44 2.84 4.66 2.93 8.39 8.25
t_statistic 24.41 8.26 15.53 16.56 9.45 15.06 10.50 16.81 13.51 4.14 11.74
Average b1 1.06 7.13 6.13 6.16 4.69 3.79 0.60 3.62 8.10 15.59 5.50
t_statistic 0.29 2.33 5.36 4.73 4.15 2.99 0.58 2.47 0.79 0.84 0.69
Average b2 3.88 8.40 6.83 7.20 7.65 6.75 0.88 4.28 6.84 5.72 13.86
t_statistic 4.76 2.74 3.45 4.72 5.47 4.89 1.27 4.26 2.18 2.53 3.13
Panel C: Dependent Variable is Net Payout Yield
ALL ENRG MTRL INDU CD CS HLTH FIN TECH TELS UTIL
Average a 2.39 1.02 2.42 3.86 2.98 4.03 0.31 3.94 1.38 36.21 4.42
t_statistic 1.70 1.97 7.65 0.61 4.40 2.09 0.77 10.34 1.07 1.18 4.88
Average b1 9.99 18.53 16.95 7.35 16.61 8.52 11.65 9.76 22.73 17.21 25.51
t_statistic 1.93 4.64 9.44 0.61 8.57 0.32 7.28 3.47 1.22 0.79 1.85
Average b2 7.55 16.69 13.76 0.40 16.38 17.49 14.77 7.76 18.04 16.37 26.50
t_statistic 3.05 4.88 5.23 0.02 8.36 5.73 11.37 4.90 6.46 0.48 4.16
Corporate profitability, Ei,t/Ai,t, is defined as the ratio of a firm’s earnings before interest to its assets. Growth
opportunities, Vi,t/Ai,t, are defined as the ratio of the firm’s market value to the book value of its assets.
The following least squares regression model is adopted to explain the three distribution yields to shareholders:
dividend yield, payout yield and net payout yield.
Yieldi ¼ a þ b1  Ei;t =Ai;t þ b2  Vi;t =Ai;t þ ei
The table shows averages (across 35 fiscal years: 1973 to 2007) of the regression intercepts, slopes, and t-statistics
for the averages.
Note: Critical values of one-sided student t-test: 1.310 [90%], 1.697 [95%], and 2.457 [99%]; with 34 degrees of
freedom for errors.

In Panel A, Table 7, when dividend yield is the may be caused by management’s attempt to follow a
dependent variable, average estimated intercept sustainable dividend policy. In Panel B, the average
terms, ranging from 3.74 in the largest size quintile estimated coefficient for corporate profitability fac-
to 1.02 in the smallest size quintile, are significantly tor is not significantly positive for both the largest
greater than zero at 1% level according to size quintile and the smallest size quintile of firms
corresponding t-statistics for all five columns. and the average estimated coefficient for corporate
Regarding the corporate profitability factor, the aver- investment opportunities is not significantly nega-
age estimated coefficients are 4.04 (t-statistic ¼ tive to the smallest size quintile of firms. In Panel
3.29) for the 20% largest firms and 1.75 C, the average estimated intercept term and both
(t-statistic ¼ 1.90) for the next 20% largest firms, estimated slope coefficients are not significantly dif-
which negates the null hypothesis that more profit- ferent from zero for the smallest size quintile of
able firms should have higher dividend yields. This firms.

100 THE AMERICAN ECONOMIST

 TABLE 7.
The Relation among Distribution Yields to Shareholders, Corporate Profitability, and
Investment Opportunities, by Firm Size
Panel A: Dependent Variable is Dividend Yield
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Average a 3.74 2.77 1.88 1.26 1.02
t _statistic 10.13 8.72 9.58 10.10 12.40
Average b1 4.04 1.75 1.52 1.54 1.84
t _statistic 3.29 1.90 8.29 7.93 4.08
Average b2 4.41 3.04 2.36 0.64 0.96
t _statistic 8.08 8.15 6.87 4.49 2.62
Panel B: Dependent Variable is Payout Yield
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Average a 5.82 4.70 4.06 3.51 4.25
t _statistic 16.34 15.69 16.56 17.17 4.67
Average b1 0.09 3.24 2.63 2.13 10.81
t _statistic 0.06 1.38 2.96 3.68 0.65
Average b2 9.87 5.74 3.91 1.67 3.50
t _statistic 9.80 8.51 4.04 3.31 0.99
Panel C: Dependent Variable is Net Payout Yield
Largest Quintile 4 Quintile 3 Quintile 2 Smallest
Average a 4.62 3.02 1.72 0.91 6.88
t _statistic 11.96 7.57 4.69 2.25 0.89
Average b1 3.50 7.95 8.76 11.39 11.22
t _statistic 2.62 3.10 8.07 13.55 0.68
Average b2 15.52 15.49 14.44 11.99 3.02
t _statistic 11.91 17.42 11.31 10.16 0.43
Corporate profitability, Ei,t/Ai,t, is defined as the ratio of a firm’s earnings before interest to its assets. Growth
opportunities, Vi,t/Ai,t, are defined as the ratio of a firm’s market value to the book value of its assets.
The following least squares regression model is adopted to explain each of the three distribution yields to share-
holders: dividend yield, payout yield and net payout yield.
Yieldi ¼ a þ b1  Ei;t =Ai;t þ b2  Vi;t =Ai;t þ ei
The table shows averages (across 35 fiscal years: 1973 to 2007) of the regression intercepts, slopes, and t-statistics
for the averages.
Note: Critical values of one-sided student t-test: 1.310 [90%], 1.697 [95%], and 2.457 [99%]; with 34 degrees of
freedom for errors.

6. Conclusion chase, we perform these tests for each of three

measures of payout: cash dividends, a measure that
According to Modigliani and Miller’s theoreti- includes share repurchase, and a comprehensive
cal analysis, a firm’s dividend policy is irrele- measure that also includes share issuances.
vant when there are no taxes, transactions, and We first compare the stock performance of high-
full information. By contrast, shareholders are of- and low-payout firms. In particular, we test whether
ten believed to value cash dividends. Because cash an investment strategy that is long in high-payout
dividends no longer appear to be a full measure of stocks and short in low-payout stocks results in
distributions, it is not clear that tests of dividend significant risk-adjusted profits. We find that the
irrelevance versus dividend preference that are strategy with the highest (significant) risk-adjusted
based on cash dividends are well specified. To returns is the one based on the most comprehensive
account for the growing importance of share repur- measure of payout (which includes dividend yield,

Vol. 56, No. 1 (Spring 2011) 101

share repurchases, and share issuances). Hence, we 4. Standard & Poor’s and MSCI jointly launched
confirm the benefit of broadening the definition of the GICS structure in 1999. For firms without
cash distribution. In addition, this result is consistent GICS information in the COMPUSTAT
with the traditional view that shareholders have pos- annual database, a Standard Industrial Classi-
itive preferences for distributions. Moreover, it em- fication (SIC) 4-digit industry code is
pirically confirms the importance of including net converted into the 2-digit GICS sector code.
share repurchase as a means of distribution in tests 5. Earnings before Interest (EBI) ¼ Earnings be-
of distribution policies. fore Extraordinary Items (COMPUSTAT mne-
We next compare the stock performance of high- monic IB) þ Interest Expenses (COMPUSTAT
and zero-payout firms. We test a strategy that is mnemonic XINT) þ Deferred Taxes (COMPU-
long in high-payout stocks and short in zero-payout STAT mnemonic TXDI)
stocks. This strategy does not result in 6. Earnings Available for Common Stock (Yt) ¼
significant returns. It implies that there is no differ- Earnings Before Extraordinary Items – Pre-
ence in performance between these two groups of ferred Dividends (COMPUSTAT mnemonic
stocks. This test reflects the dividend-as-residual DVP) þ Income Statement Deferred Taxes.
policy, which models the corporate payout decision 7. Book Equity (BE) ¼ Stockholder’s Equity
as one that is made after the firm sets its preferences (COMPUSTAT mnemonic SEQ) – Preferred
for investment and financial structure. To help ex- Stocks (COMPUSTAT mnemonic PSTKL,
plain this finding, we establish the extent to which PSTKRV, or PSTK).
corporate profitability and investment. Our analysis 8. Market Value of Firm (Vt) ¼ Assets (At) -
shows that both factors are significantly associated Book Equity (BEt) þ Market Equity (MEt);
with firms’ payout policies. We conclude that this where MEt ¼ Stock Price (COMPUSTAT
helps explain the fact that the stock performance of mnemonic PRCC_F) * Shares Outstanding
firms with zero payout policies perform as well on (COMPUSTAT mnemonic CSHO).
average as stocks with high-payout policies. 9. The use of June as a cutoff period may not be
Overall, our results are consistent with the no- suitable for a very limited number of firms
tion that share repurchases or issuances by firms with fiscal years ending in May because,
with smaller market capitalization (maybe except according to COMPUSTAT, annual statistics
for the smallest ones) are more effective in influ- in fiscal year 2000 includes firms with fiscal
encing their share prices. We show that growth year-end month falls in June 2000 through
opportunities facing companies are a key factor in December 2000 and in January 2001 through
corporate distribution policies as companies with May 2001.
more investment opportunities are found to have 10. Although not reported, a p-value higher than
lower distributions than other companies across 0.100 for either the estimated intercept term or
almost all sectors and size quintiles. the slope term in Table 2 usually corresponds
to the model’s R2 of being 0.10 or less.
11. Average percentage of dividend-paying firms:
Notes size quintile 5 (78.0%), quintile 4 (55.2%),
quintile 3 (40.2%), quintile 2 (30.5%) and
1. The average annual return is 14.79% for all quintile 1 (16.5%).
stocks included in the study. 12. Comparative results regarding high payout
2. As indicated in Boudoukh, Michaely Richardson portfolios and zero payout portfolios are avail-
and Roberts (2007), a zero-payout portfolio able from the author upon request.
exhibits unique characteristics. Hence, we assign 13. Similar conclusions are reached when other
firms with zero yields to their own portfolio. three independent variable combinations (Yi,t/
Also, due to the size of our sample, we sort the BEi,t and Vi,t/Ai,t, Ei,t/Ai,t and dAi,t/Ai,t, Yi,t/BEi,t
firms into three portfolios, as opposed to decile and dAi,t/Ai,t) are used. These results are avail-
portfolios as in Boudoukh et al. (2007). able from the author upon request.
3. COMPUSTAT cash flow items are made 14. Critical values of one-sided student t-test: 1.310
available from 1971. There is a significant [90%], 1.697 [95%], and 2.457 [99%]; with 34
increase in in coverage in 1973. degrees of freedom for errors.

102 THE AMERICAN ECONOMIST

 References Ikenberry, David L., Josef Lakonishok, and
Theo Vermaelen, 1995, Market underreaction
Baker, H. Kent, Gail E. Farrelly, and Richard B. to open-market share repurchases, Journal of
Edelman, 1985, A survey of management views Financial Economics, 39(2–3), 181–208.
on dividend policy, Financial Management, Ikenberry, David L. and Theo Vermaelen, 1996,
14(3), 78–84. The option to repurchase stock, Financial Man-
Baker, H. Kent, Gary E. Powell, and E. Theodore agement, 25(4), 9–24.
Veit, 2003, Why companies use open-market Julio, Brandon and David L. Ikenberry, 2004,
repurchases: A managerial perspective, Quar- Reappearing dividends, Journal of Applied
terly Journal of Economics and Finance, 43(3), Corporate Finance, 16(4), 89–100.
483–504. Keim, Donald B., 1986, Dividend yields and the
Baker, H. Kent, Patricia L. Gallagher, and Karen January effect, Journal of Portfolio Manage-
E. Morgan, 1981, Management’s view of stock ment, 12(2), 54–60.
repurchase programs, Journal of Financial Lakonishok, Josef, Andrei Shleifer, and Robert W.
Research, 4(3), 233–247. Vishny, 1994, Contrarian investment, extrapo-
Boudoukh, Jacob, Roni Michaely, Matthew lation, and risk, Journal of Finance, 49(5),
Richardson, and Michael R. Roberts, 2007, On 1541–1578.
the importance of measuring payout yield: Miller, Merton H., and Franco Modigliani, 1961,
Implications for empirical asset pricing, Journal Dividend policy, growth, and the valuation of
of Finance, 62(2), 877–915. shares, Journal of Business, 34(4), 411–433.
Fama, Eugene F. and Kenneth R. French, O’Shaughnessy, James P., 2005, What works on
2001, Disappearing dividends: Changing Wall Street, New York, McGraw-Hill.
firm characteristics or lower propensity to Stewart, Samuel S. Jr., 1976, Should a corporation
pay? Journal of Financial Economics, 60(1), repurchase its own stock? Journal of Finance,
3–43. 31(3), 911–921.
Grullon, Gustavo and Roni Michaely, 2002, Wansley, James W., William R. Lane, Salil Sarkar,
Dividends, share repurchases, and the substitu- 1989, Managements’ view of share repurchase
tion hypothesis, Journal of Finance, 57(4), and tender offer premiums, Financial Manage-
1649–1684. ment, 18(3), 97–110.
Grullon, Gustavo and Roni Michaely, 2004, Weigand, Robert A. and H. Kent Baker, 2009,
The information content of share repurchase Changing perspectives on distribution policy,
programs, Journal of Finance, 59(2), 651–680. Managerial Finance, 35(6), 479–492.

Vol. 56, No. 1 (Spring 2011) 103