Are Dividends Disappearing Dividend Concentration and the Consolidation of Earnings

ARTICLE IN PRESS

Journal of Financial Economics 72 (2004) 425–456

Are dividends disappearing? Dividend
concentration and the consolidation of earnings$
Harry DeAngeloa,*, Linda DeAngeloa, Douglas J. Skinnerb
a

Marshall School of Business, University of Southern California, Los Angeles, CA 90089, USA
b
University of Michigan Business School, Ann Arbor, MI 48109, USA
Received 4 September 2002; accepted 12 March 2003

Abstract
Aggregate real dividends paid by industrial firms increased over the past two decades even
though, as Fama and French (J. Financial Econ. 60, 3) (2001a) document, the number of
dividend payers decreased by over 50%. The reason is that (i) the reduction in payers occurs
almost entirely among firms that paid very small dividends, and (ii) increased real dividends
from the top payers swamp the modest dividend reduction from the loss of many small payers.
These trends reflect high and increasing concentration in the supply of dividends which, in
turn, reflects high and increasing earnings concentration. For example, the 25 firms that paid
the largest dividends in 2000 account for a majority of the aggregate dividends and earnings of
industrial firms. Industrial firms exhibit a two-tier structure in which a small number of firms
with very high earnings collectively generates the majority of earnings and dominates the
dividend supply, while the vast majority of firms has at best a modest collective impact on
aggregate earnings and dividends.
r 2003 Elsevier B.V. All rights reserved.
JEL Classification: G35; G32; M41
Keywords: Dividends; Payout policy; Earnings concentration; Losses

$
We would like to acknowledge the helpful comments of Jim Brickley, Ken French, Tom Gilligan, John
Matsusaka, Dave Mayers, Kevin J. Murphy, Micah Officer, Oguz Ozbas, Ed Rice, Jay Ritter, Ren!e Stulz,
Randy Westerfield, Fred Weston, and especially the referee, Eugene Fama. We are grateful for the
research assistance of Darcy Baker, Ido Dotan, Atul Gupta, Terry Lichvar, April Xu, and especially Mei
Feng and Aaron Lerner, and for the research support of the University of Southern California (Charles E.
Cook/Community Bank and Kenneth King Stonier Chairs) and the University of Michigan Business
School (KPMG Professorship).
*Corresponding author. Tel.: +1-213-740-6541; fax: +1-213-740-6650.
E-mail address: [email protected] (H. DeAngelo).

0304-405X/$ - see front matter r 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0304-405X(03)00186-7

ARTICLE IN PRESS
426

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

1. Introduction
In their intriguing study, ‘‘Disappearing Dividends: Changing Firm Characteristics
or Lower Propensity to Pay?,’’ Fama and French (2001a) document a large decline
over 1978–1998 in the number and percent of nonfinancial and nonutility firms
(hereafter, industrials) that pay dividends. Their analysis indicates that this dramatic
change in dividend practices is due both to changes in the population of firms that
are now publicly held (with many more public firms now exhibiting the
characteristics of firms that historically have not paid dividends), and to a reduced
propensity to pay dividends by firms whose characteristics historically would have
led them to distribute cash to stockholders. Although Fama and French carefully
state that their findings show a reduction in the number and percent of dividendpaying firms, their evidence is commonly interpreted as indicating that dividends
themselves are disappearing.1 The latter view seems more than plausible, given the
striking fact that the number of dividend-paying industrials has declined by more
than 1,000 firms (over 50%) over the last 20–25 years.
Although our evidence confirms a radical transformation in corporate dividend
practices over the last two decades, it does not indicate that dividends are
disappearing. Rather, dividends paid by industrial firms actually increased over
1978–2000, both in nominal and in real terms (by 224.6% and 22.7% respectively for
our sample). Why did aggregate real dividends increase despite a 50%-plus decline in
the number of payers? The answer is twofold: (i) the large reduction in payers
occurred almost entirely among firms that paid very small dividends, with the loss of
these firms’ dividends having at best a minor impact on the aggregate supply, and (ii)
dividends simultaneously increased substantially among the largest payers, reflecting
a marked increase in their real earnings. In short, the increase in real dividends paid
by firms at the top of the dividend distribution swamps the dividend reduction
associated with the loss of many small payers at the bottom.
These secular changes reflect high and increasing dividend concentration. For
example, the 25 largest dividend payers, all of which are ‘‘old line’’ established firms,
collectively supplied over half (54.9%) of aggregate industrial dividends in 2000.
Moreover, the earnings that underlie these high dividend payments are themselves
highly concentrated. The total earnings of the 25 top dividend payers constitute
51.4% of aggregate industrial earnings in 2000 and, in real terms, are more than
double their 1978 level. The dividends paid by these 25 firms in 2000 exceed their
1978 level by $9.2 billion in real terms ($24.3 billion in year 2000 dollars), an increase
1
For example, The New York Times and The Economist report that dividends have become less
relevant and perhaps irrelevant, citing the findings of Fama and French (2001a) as well as low dividend
yields and the popularity of stock repurchases (‘‘Dividends Are Fading as Market Signals, Too,’’ The New
York Times, November 7, 1999, ‘‘Shares Without the Other Bit: In Corporate America, Paying Dividends
Has Gone Out of Fashion,’’ The Economist, November 20, 1999, and ‘‘Economics Focus: Dividends
End,’’ The Economist, January 20, 2002). Time cites low dividend yields and an increased incidence of
dividend omissions by healthy electric utility firms as evidence that ‘‘dividends have become only slightly
more relevant than the gushing palaver in an annual report’’ (‘‘Disappearing Dividends? Ending Payouts
May Be a Good Thing for Investors,’’ Time, February 2, 1998).

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

427

that is greater than the $7.1 billion aggregate real increase for all industrial firms.
(Here and throughout the paper, we compare dividends and earnings for 1978 and
2000 by converting 2000 dollars into 1978 dollars using the consumer price index.)
This evidence shows that a relative handful of firms now both dominates the supply
of dividends and generates the preponderance of earnings, and that both dividend
and earnings concentration have increased substantially from the (already high) level
of two decades ago.
Changes in the cross-sectional distribution of earnings—especially among firms at
the top end of the distribution—are the fundamental reason why real dollar
dividends paid by industrial firms have increased even though, as Fama and French
(2001a) conclude, industrial firms now exhibit a reduced propensity to pay dividends
(i.e., they are now more likely to pay zero dividends, controlling for earnings and
growth opportunities). We find that 100% of the firms with at least $1 billion in real
earnings paid dividends in 1978, whereas 85.7% paid dividends in 2000, consistent
with Fama and French’s reduced propensity to pay. However, although a smaller
proportion of firms with high real earnings now pays dividends, top earners continue
to exhibit a very strong tendency to do so. And since top-end firms now produce so
much more in real earnings, on net this group shows a large increase in real dividends
even though a few very large earners, primarily technology firms, have been slow to
initiate dividends. The end result is that aggregate dividends paid by industrial firms
increased over 1978–2000 despite the reduced propensity to pay dividends.
Overall, the supply of dividends by industrial firms exhibits a two-tier structure in
which a small number of firms with very high earnings collectively generates the
majority of earnings and dominates the dividend supply, while the vast majority of
firms has at best a modest collective impact on aggregate earnings and dividends. We
discuss the implications of the two-tier structure for the dividend clientele and
signaling hypotheses and for the evolution of corporate payout practices in Section
7. Section 2 begins the paper by describing our sampling procedure, and it also
details the aggregate dividend increase from 1978 to 2000 for industrial firms. Section
3 documents the concentration of dividends and the consolidation therein that has
occurred over the last two decades, while Section 4 does the same for earnings.
Section 5 identifies the top payers, the top nonpayers, and the firms with the highest
earnings in 2000. Section 6 documents how many of the 1978 dividend payers
continued to pay dividends in 2000, the strong influence of these continuing payers
on the 2000 dividend supply, and what happened to the remaining firms (primarily,
they were acquired).

2. Sampling procedure and aggregate dividends, 1978–2000
Shoven (1986, Table 2) reports that US government data show a near doubling of
nominal dividends for the corporate sector as a whole over 1978–1985. And this
trend continues, with aggregate nominal dividends increasing 647.2% over 1978–
2000, nearly twice the contemporaneous 330.0% increase in GDP, and with
aggregate real dividends increasing 182.9% (all data from Economic Report of the

ARTICLE IN PRESS
428

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

President, 2001). These data paint a very different picture from the impression one
forms from Fama and French’s (2001a) finding that the number of dividend payers
has fallen by more than 50% since 1978 (see footnote 1 above). Of course, the
government population is not confined to publicly traded industrials, the sample that
Fama and French study, since it also includes private firms and publicly held
financials and utilities. However, as we show below, aggregate dividends also
increased for publicly traded industrials, thereby posing the conundrum that we
study in this paper: why have aggregate dividends increased in the face of a radical
decline in the number of payers?
Following Fama and French (2001a), we sample CRSP industrial firms with SIC
codes outside the ranges 4900–4949 and 6000–6999 (financials and utilities). We call
nonfinancial and nonutility firms ‘‘industrials,’’ while recognizing that this group
also includes service firms, conglomerates, and perhaps other companies not
conventionally labeled industrials. Like Fama and French (2001a), we restrict
attention to NYSE, AMEX, and NASDAQ firms with CRSP share codes 10 or 11
for at least one month of each year in question, and with nonmissing December share
price and quantity data. We consider only CRSP firms with dividends and earnings
on Compustat (the CRSP/Compustat sample). Our sample sizes differ slightly from
those of Fama and French because we place different requirements on the
availability of specific data items, and probably also because of differences in how
we implement sampling criteria regarding the monthly observation of CRSP share
and/or SIC codes. Fama and French examine trends over 1978–1998, while we
employ data that became available after publication of their study and therefore
examine trends over 1978–2000.
Table 1 shows that the large decline in the number of dividend payers from 1978 to
2000 is confined to industrial firms. While the number of dividend-paying industrials
fell 58.8%, from 2,250 in 1978 to 926 in 2000, the number of payers among financial/
utility firms increased by 9.5%, from 852 to 933. Although not shown in the table,
the proportion of dividend-paying financials/utilities on CRSP fell from 79.9% to
71.6% over 1978–2000, a decline that occurred because the increase in listed firms
proportionately exceeds the increase in these payers. Because the precipitous decline
in the number of dividend-paying industrials is not matched by a similar decline for
financials/utilities, it cannot simply reflect a general increase in managers’ reluctance
to pay dividends, but must instead relate to some underlying fundamental change(s)
largely confined to industrial firms. For example, income tax law changes that had
similar effects on nonindustrial and industrial firms’ incentives to pay dividends
cannot explain the secular trends in Table 1.
Fig. 1 tracks aggregate dividends, earnings, and losses for industrial firms on
CRSP/Compustat over 1950–2000, as well as total earnings for the subset of
dividend-paying industrial firms. The CRSP/Compustat population expands in 1962
(when CRSP added AMEX firms to its coverage of NYSE firms) and again in 1972
(when NASDAQ firms were added), but remains unchanged post-1978, when the
number of payers fell by more than 50%. The figure shows that (i) aggregate
dividends increased steadily over the full period 1950–2000, including the 1978–2000
subperiod, (ii) the long-term growth in aggregate dividends reflects the underlying

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

429

Table 1
Number of dividend-paying firms over 1978–2000: CRSP sample partitioned by industrial versus financial
and utility firms
For each year, the table includes NYSE, NASDAQ, and AMEX firms on CRSP that have CRSP share
codes 10 or 11 and that meet the other sampling criteria described in the paper. The sample of financial
and utility firms includes those with SIC codes in the ranges 4900–4949 or 6000–6999, while the sample of
industrial firms includes those with SIC codes outside these ranges.
Year

CRSP industrial
firms

CRSP financial
and utility firms

1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000

2,250
2,160
2,050
1,936
1,820
1,712
1,672
1,562
1,434
1,363
1,306
1,271
1,234
1,177
1,219
1,218
1,245
1,265
1,214
1,170
1,111
1,038
926

852
841
835
815
780
784
794
817
833
979
1,020
1,015
941
863
865
959
1,042
1,103
1,136
1,110
1,072
1,022
933

Absolute change
over 1978–2000
Percent change
over 1978–2000

1,324
58.8%

CRSP
total

NYSE industrials

NASDAQ
and AMEX
industrials

3,102
3,001
2,885
2,751
2,600
2,496
2,466
2,379
2,267
2,342
2,326
2,286
2,175
2,040
2,084
2,177
2,287
2,368
2,350
2,280
2,183
2,060
1,859

1,015
1,004
982
951
911
870
856
815
761
709
684
663
651
642
679
691
717
745
750
744
727
699
626

1,235
1,156
1,068
985
909
842
816
747
673
654
622
608
583
535
540
527
528
520
464
426
384
339
300

+81

1,243

389

935

+9.5%

40.1%

38.3%

75.7%

growth in earnings, although as expected, dividends grew more smoothly than
earnings, (iii) dividend payers account for the vast bulk of industrial earnings in all
years [as in Fama and French (2001a, p.18)], and (iv) aggregate losses increased
markedly from 1978 to the early 1990s, reaching massive levels in the late 1990s. In
the remainder of the paper, we compare aggregate industrial dividends for 1978 and
2000, since the former year marks the beginning of the long-term decline in the
number of payers and the latter year has the latest available data. As Fig. 1 shows,
there is nothing unusual about aggregate dividends in these two years. Rather, the
increase from 1978 to 2000 is simply part of a steady long-run uptrend in aggregate
dividends paid by industrial firms.

430

350

250

Total earnings of dividend payers
Aggregate dividends

150
100
50
0
1950
-50

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

-100

Aggregate losses
-150
-200

Fig. 1. Aggregate dividends, aggregate earnings, aggregate losses, and total earnings for dividend payers: Industrial firms on CRSP/Compustat, 1950–2000.
The sample includes NYSE, NASDAQ, and AMEX firms on CRSP that have share codes 10 or 11. Industrial firms are defined as those with SIC codes outside
the ranges 4900–4949 and 6000–6999 (financial and utility companies). The sample is restricted to firms for which Compustat has nonmissing dividends and
earnings before extraordinary items (Compustat items 21 and 18) for the year in question. For each year, the data in this figure are based on Compustat’s year
assignment, whereas the data in all other tables are generated under the Fama and French (2001a, p. 40) convention of assigning financial data to years based
on the calendar year in which the fiscal year-end falls. The alternative year-end conventions have little effect on the aggregate dollar magnitudes of the variables
that we analyze. Total earnings for dividend payers are the sum of earnings for all firms that pay a positive dividend during the year in question.

ARTICLE IN PRESS

Billions of dollars

200

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Aggregate earnings

300

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

431

Table 2
Aggregate dividends in 1978 and 2000 and related descriptive statistics: Industrial firms on CRSP/
Compustat
The sample includes NYSE, NASDAQ, and AMEX firms on CRSP that have share codes 10 or 11, and
SIC codes outside the ranges 4900–4949 and 6000–6999. The sample is restricted to firms for which
Compustat reports dividends and earnings before extraordinary items (Compustat items 21 and 18) for
each year in question. Real dividends in 2000 are nominal dividends converted to 1978 dollars using the
consumer price index. The CRSP/Compustat sample assigns each firm to 1978 (or 2000) based on the date
of its fiscal year-end, while the CRSP sample assigns a December 31 year-end for all firms. Since most firms
have a December 31 year-end, the number of dividend-paying firms on CRSP/Compustat can exceed the
number on CRSP (which is the case here for 2000 as a comparison of Tables 1 and 2 indicates).

1. Aggregate nominal dividends
($ billions)
2. Aggregate real dividends
($ billions, 1978 base)
3. Mean real dividend
($ millions, per dividend-paying firm)
4. Median real dividend
($ millions, per dividend-paying firm)
5. Number of dividend-paying
industrials on CRSP/Compustat
6. Percent of all CRSP/Compustat
industrials that paid dividends
7. Percent of dividend-paying
industrials that are NYSE-listed
8. Percent of total dollar dividends
accounted for by NYSE-listed firms

1978

2000

Absolute (%)
change

$31.3 billion

$101.6 billion

$31.3 billion

$38.4 billion

$14.4 million

$41.4 million

$1.4 million

$3.6 million

2,176

930

65.1%

19.4%

$70.3 billion
(+224.6%)
$7.1 billion
(+22.7%)
$27.0 million
(+187.5%)
$2.2 million
(+157.1%)
1,246
( 57.3%)
45.7%

45.0%

66.0%

+21.0%

94.7%

97.4%

+2.7%

Table 2 documents that aggregate nominal dividends increased by 224.6% for
CRSP/Compustat industrial firms, from $31.3 billion in 1978 to $101.6 billion in
2000, while aggregate real dividends (here and throughout denominated in 1978
dollars) increased by 22.7%, to $38.4 billion.2 The mean real dividend paid (per
dividend-paying firm) increased from $14.4 million in 1978 to $41.4 million in 2000,
while the median increased from $1.4 million to $3.6 million. The difference in mean
and median—and the large expansion in that difference over 1978–2000—reflects
substantial dividend concentration in 1978 and a large increase therein over 1978–
2000, both of which we document in the next section. NYSE firms account for 66.0%
2
Fama and French’s (2001a) approach of screening for firms with CRSP share codes 10 or 11 and
matching with Compustat data offers protection against the possibility that changes in Compustat’s
coverage over time drive the results. For example, since Compustat has recently added many foreign firms
with ADRs that pay large dividends, the aggregate dividend increase for the full Compustat population is
misleadingly large. The aggregate dividend comparisons in Table 2 do not suffer from this problem, as
they were generated using Fama and French’s sampling approach. Nor are they substantively affected by
the loss of CRSP firms with data unavailable on Compustat. Using CRSP dividend data for the latter
firms, we find that the $7.1 billion real dividend increase from 1978 to 2000 (reported in Table 2) narrows
by only $129 million, so that real dividends increased for the full set of CRSP firms.

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

432

of payers and 97.4% of aggregate dividends in 2000, up respectively from 45.0% and
94.7% in 1978. The fact that NYSE firms pay the overwhelming majority of
industrial dividends likely reflects the tendencies for older and more stable (thus
typically dividend-paying) firms to list their shares on the NYSE, and for young and
growing (thus typically not dividend-paying) firms to trade on NASDAQ.

3. Dividend concentration and the increase therein over the last two decades
Table 3 ranks dividend-paying industrial firms by cash dividends paid in 1978 and
2000, in groups of 100 firms. For each ranked group in 1978 and 2000, the first two
Table 3
Concentration of total dollar dividends paid by industrial firms in 1978 and in 2000
Firms are ranked from the largest to smallest total dollar dividends paid in each year (per Compustat). The
sample includes NYSE, NASDAQ, and AMEX firms on CRSP that have share codes 10 and 11 and SIC
codes outside the ranges 4900–4949 and 6000–6999. A firm is included only if Compustat reports dividends
and earnings before extraordinary items for the year in question (Compustat items 21 and 18). A firm’s
dividends are the amounts reported for the fiscal years ending in 1978 or 2000. For 2000, the row
corresponding to firms ranked from 901 to 1000 has 30 firms because there are 930 dividend payers that
meet our sampling criteria in 2000. Each cell amount is rounded to the nearest significant digit, which
explains a few minor discrepancies across row or column total figures.
Dividend
ranking

Top 100
101–200
201–300
301–400
401–500
501–600
601–700
701–800
801–900
901–1000
1001–1100
1101–1200
1201–1300
1301–1400
1401–1500
1501–1600
1601–1700
1701–1800
1801–1900
1901–2000
2001–2100
2100–2176
Total for all firms
Number of firms

Percent of
total dividends (%)

Cumulative % of
total dividends (%)

Real dividends
($ millions, 1978 base)

1978

2000

1978

2000

1978

2000

67.3
11.8
6.3
4.0
2.8
1.9
1.4
1.0
0.8
0.6
0.5
0.4
0.3
0.2
0.2
0.2
0.1
0.1
0.1
0.1
o0.1
o0.1
100.0

81.8
10.1
3.9
1.9
1.0
0.6
0.3
0.2
0.1
o0.1

67.3
79.1
85.4
89.4
92.2
94.0
95.4
96.4
97.2
97.8
98.3
98.7
99.0
99.3
99.5
99.6
99.7
99.8
99.9
99.9
99.9
100.0
100.0

81.8
92.0
95.8
97.7
98.8
99.4
99.7
99.9
99.9
100.0

$21,111
3,691
1,970
1,247
865
585
431
325
249
192
153
120
96
76
61
49
40
31
23
16
10
2
$31,343
2,176

$31,477
3,889
1,492
748
401
223
132
70
29
1

100.0

100.0

$38,461
930

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

433

columns report the percent of dividends paid, the middle two columns report the
cumulative percent, and the last two columns report total real dividends. Overall, a
relatively small number of firms pays the overwhelming majority of aggregate
industrial dividends, and this concentration has increased substantially over the last
two decades. For example, in 2000, the top 100 dividend payers distributed 81.8% of
dividends, up from 67.3% in 1978. The $31.5 billion in real dividends paid by the top
100 dividend payers in 2000 ($83.2 billion in year 2000 dollars) exceeds the $31.3
billion paid in 1978 by all 2,176 dividend payers. For the top 100, dividends
increased by $10.4 billion (49.1%), or more than triple the $3.3 billion dividend
decline for all ranks below the top 100. Since virtually all of the increased dividend
concentration is driven by the top 100 or 200 payers, this concentration increase is
obviously not an artifact of the reduction in the number of payers.
Table 4 summarizes the cross-sectional distributions of dividends in 1978 and in
2000, with dividend-paying firms categorized by real dollar dividends paid, ranging
from $500 million-plus per year to less than $1 million per year. The table documents
that the number of firms paying dividends of $100 million-plus increased by 34 over
1978–2000, for a dividend increase of $12.6 billion. Over the same period, the number
of firms paying less than $100 million decreased by 1,280 and this decline manifests
primarily in the two smallest dividend classes, with the 1,069 net reduction in firms
paying less than $5 million per year accounting for 85.8% of the overall net decline of
1,246 firms. And while the 1,069-firm reduction in small payers is large in number, the
dollar magnitude of the $1.1 billion in dividends lost is dwarfed by the $12.6 billion in
dividends gained from the 34-firm increase for the $100 million-plus category.
In sum, the top categories of dividend payers now contain more firms and these
firms now pay substantially higher total dividends, while the bottom categories now
contain many fewer firms and these firms now pay modestly lower total dividends.
The net result is a large decrease in the number of payers accompanied by an increase
in aggregate dividends, which reflects the fact that the top-end firms dominate, while
the bottom-end firms have little impact on the aggregate dividend supply. Black and
Scholes (1974) and Miller (1977) argue that what matters to investors is the
aggregate supply of securities with particular characteristics (e.g., dividends, taxable
interest returns, etc.), and not the number of firms delivering that supply or the
quantity delivered by any one firm. In their view, the decline in the number of payers
is of little consequence so long as sufficient dividends are supplied to meet the
demand for dollars delivered currently in that form. Since aggregate real dividends
increased over 1978–2000, the decrease in the number of payers was not caused by a
reduction in aggregate demand, but instead must reflect changes in the factor(s) that
determine firms’ dividend supply decisions.

4. Dividend and earnings concentration and payout ratios
Lintner’s (1956) finding that firms’ dividend supply decisions primarily depend on
earnings suggests that the high/increasing dividend concentration we observe may be
the result of high/increasing earnings concentration. The evidence reported in this

$500 million or more
$400–$499.9 million
$300–$399.9 million
$200–$299.9 million
$100–$199.9 million
$80–$99.9 million
$60–$79.9 million
$40–$59.9 million
$20–$39.9 million
$10–$19.9 million
$5–9.9 million
$1–$4.9 million
Less than $1 million

Number of
firms 2000

Change from
1978 to 2000

Percent change
from 1978 to
2000 (%)

Real
dividends
1978

Real
dividends
2000

Change from
1978 to 2000

Percent change
from 1978 to
2000 (%)

6
4
4
9
19
18
24
55
108
161
187
633
948

16
4
6
13
37
10
21
36
66
94
115
276
236

10
0
2
4
18
8
3
19
42
67
72
357
712

166.7
0.0
50.0
44.4
94.7
44.4
12.5
34.5
38.9
41.6
38.5
56.4
75.1

$9,095
1,717
1,441
2,099
2,630
1,591
1,649
2,642
3,003
2,233
1,344
1,523
375

$17,591
1,829
2,005
3,094
5,062
879
1,462
1,744
1,883
1,337
812
655
106

$8,496
112
564
995
2,432
712
187
898
1,120
896
532
868
269

93.4
6.5
39.1
47.4
92.5
44.8
11.3
34.0
37.3
40.1
39.6
57.0
71.7

Total

2,176

930

1,246

57.3

$31,342

$38,461

$7,118

22.7

$100 million and above
Less than $100 million

42
2,134

76
854

34
1,280

81.0
60.0

$16,982
14,360

$29,582
8,878

$12,600
5,482

74.2
38.2

Less than $5 million

1,581

512

1,069

67.6

$1,898

$761

$1,137

60.0

ARTICLE IN PRESS

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.

Number of
firms 1978

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Real dividend payment
(1978 dollars)

434

Table 4
Number of firms and real dividend payments in 1978 and in 2000 for samples of industrial firms that paid given amounts of real dividends ($ millions, 1978
dollars)
The sample is comprised of NYSE, NASDAQ, and AMEX firms on CRSP with share codes 10 or 11 and SIC codes outside the intervals 4900–4949 and 6000–
6999. A firm is included only if Compustat reports dividends and earnings before extraordinary items for the year in question (Compustat items 21 and 18). A
firm’s dividends are the amounts reported for the fiscal years ending in 1978 or 2000. Real dividends are given in millions of dollars, and are nominal payments
converted to 1978 dollars using the consumer price index. Each cell amount is rounded to the nearest significant digit, which explains a few minor discrepancies
across category totals.

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

435

and the next section strongly supports this hypothesis, indicating that (i) 1978
earnings are highly concentrated and 2000 earnings are considerably more so, (ii) the
cross-sectional distribution of 2000 earnings is dominated by firms at the extremes,
(iii) 28 firms with $1 billion-plus in real earnings generated the majority of 2000
aggregate earnings, (iv) these 28 top earners paid 50.1% of aggregate dividends in
2000, although 14.3% are nonpayers (up from 0.0% nonpayers among the top
earners in 1978), (v) 2000 aggregate earnings exceed those of 1978, with most of the
increase coming from firms at the top end of the earnings distribution, and (vi) for
dividend-paying firms, the typical payout ratio is little changed over the past two
decades.
Table 5 ranks dividend-paying firms by their earnings in 1978 and 2000, in a format
similar to that of Table 3. Like dividends, earnings were highly concentrated in 1978,
and substantial additional earnings concentration has occurred among dividend
payers over the last two decades. In 1978, the top 100 dividend payers generated
57.5% of the earnings of all payers, while cumulatively the top 200 payers generated
71.0%. The corresponding figures for 2000 are 74.0% and 86.0%. From 1978 to 2000,
the total real earnings of the 100 largest payers increased markedly, from $47.5 billion
to $80.2 billion. The next two groups also show real increases, albeit of considerably
more modest magnitudes, while all remaining groups show decreases. The net result is
a 31.0% increase in the real earnings of dividend payers, from $82.7 billion in 1978 to
$108.3 billion in 2000 (to $286.0 billion in year 2000 dollars).
4.1. The pooled earnings distribution of dividend payers and nonpayers
As background for our analysis of the earnings of dividend payers and nonpayers,
Table 6 summarizes the cross-sectional earnings distribution of all firms (payers and
nonpayers) combined. Panel A contains the pooled earnings distributions for 1978
and 2000. Since Lintner’s (1956) analysis indicates that dividends tend to be set in
response to long-run earnings rather than to a single (possibly aberrant) earnings
realization, panel B reports the pooled distribution of average five-year earnings
ending in 1978 and in 2000. The five-year averages are especially useful here because
in 2000 industrial firms reported remarkably large losses (see Fig. 1), which may be a
transitory phenomenon. If so, five-year average earnings better measure firms’ longrun ability to pay dividends than do one-year earnings realizations.
Table 6 indicates that earnings in both 1978 and 2000 are concentrated among a
relatively few firms at the top end of the distribution, and that such concentration is
notably greater in 2000 than it was in 1978. The aggregate real earnings of payers
and nonpayers combined increased from $85.0 billion in 1978 to $97.8 billion in 2000
($258.3 billion in year 2000 dollars), while aggregate five-year average real earnings
increased from $75.1 billion to $91.6 billion. These aggregate increases are driven by
firms at the top end of the earnings distribution, especially in the $500 million-plus
earnings categories, which show a total increase over 1978–2000 of $55.1 billion
(182.8%) in one-year earnings, and $30.2 billion (117.2%) in five-year average
earnings. These top-end increases far exceed the net aggregate increase in real
earnings for all industrials as a group, and they comprise more than three-quarters of

ARTICLE IN PRESS
436

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 5
Concentration of earnings of industrial firms that paid dividends in 1978 and in 2000
Firms are ranked from the largest to smallest total dollar dividends paid in each year (per Compustat). The
table reports the percent of total earnings generated by dividend-paying firms that are accounted for by the
top 100 dividend payers, the next largest 100 payers, and so on. The sample includes NYSE, NASDAQ,
and AMEX firms on CRSP that have share codes 10 and 11 and SIC codes outside the ranges 4900–4949
and 6000–6999. A firm is included only if Compustat reports dividends and earnings before extraordinary
items for the year in question (Compustat items 21 and 18). A firm’s dividends and earnings are the
amounts reported for the fiscal years ending in 1978 or 2000. For 2000, the row corresponding to firms
ranked from 901 to 1000 has 30 firms because there are 930 dividend payers that meet our sampling criteria
in 2000.
Dividend ranking

Top 100
101–200
201–300
301–400
401–500
501–600
601–700
701–800
801–900
901–1000
1001–1100
1101–1200
1201–1300
1301–1400
1401–1500
1501–1600
1601–1700
1701–1800
1801–1900
1901–2000
2001–2100
2101–2176
Total for all firms
Number of firms

Percent of total earnings
of dividend-paying
industrial firms (%)

Cumulative % of total
earnings of dividend-paying
industrial firms (%)

Real earnings
($ millions, 1978 base)

1978

2000

1978

2000

1978

2000

57.5
13.5
7.2
5.1
3.8
2.5
2.0
1.5
1.4
1.0
0.8
0.7
0.6
0.5
0.4
0.3
0.3
0.2
0.2
0.2
0.1
0.1
100.0

74.0
12.0
6.2
3.0
2.2
1.4
0.5
0.4
0.2
o0.1

57.5
71.0
78.2
83.3
87.1
89.6
91.6
93.1
94.5
95.5
96.3
97.1
97.7
98.2
98.6
98.9
99.2
99.4
99.6
99.8
99.9
100.0
100.0

74.0
86.0
92.2
95.1
97.3
98.8
99.3
99.7
99.9
100.0

$47,543
11,170
5,929
4,242
3,134
2,045
1,694
1,274
1,134
819
694
617
506
440
293
269
248
170
173
138
69
100
$82,701
2,176

$80,158
12,912
6,733
3,197
2,376
1,540
545
482
263
45

100.0

100.0

$108,251
930

the large total earnings increase among firms with positive earnings. A similar
picture emerges for firms with positive one-year earnings, a group that generated
$86.1 billion in 1978 and $152.8 billion in 2000 real earnings, for an increase of $66.7
billion, or 77.5%, while those firms with positive five-year earnings generated an
increase of $38.5 billion, or 50.9%.
The 2000 earnings distribution is concentrated at both the top and the bottom,
with a full 44.7% of firms reporting losses. Strikingly, the dollar masses at the
extremes are quite large and are roughly the same order of magnitude, although they
are generated by a radically different number of firms. The 28 firms at the top earned

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

437

Table 6
Cross-sectional distributions of firms’ real earnings (1978 dollars) in 1978 and in 2000:
Industrial firms on CRSP/Compustat
Panel A reports the distributions of real earnings in 1978 and in 2000. Panel B reports the distributions of
five-year average real earnings ending in 1978 and in 2000. For example, the panel B figure for a given firm
in 2000 equals the average of real earnings over the five years from 1996 to 2000 (or as many of those years
that Compustat reports earnings data on that firm). The sample is comprised of NYSE, NASDAQ, and
AMEX firms on CRSP with share codes 10 or 11 and SIC codes outside the intervals 4900–4949 and 6000–
6999. A firm is included only if Compustat reports dividends and earnings before extraordinary items for
the year in question (Compustat items 21 and 18). A firm’s dividends and earnings are those amounts
reported for the fiscal years ending in 1978 or 2000. Real earnings are nominal earnings before
extraordinary items converted to 1978 dollars using the consumer price index.
Panel A. Cross-sectional distributions of real earnings in 1978 or in 2000
Number of firms
Real earnings
(1978 dollars)

Real earnings
($ millions)

Real earnings as a
% of total (%)

1978

2000

1978

2000

1978

2000

$1 billion or greater
$500 million - $1 billion
$250–500 million
$100–250 million
$50–100 million
$25–50 million
$10–25 million
$0–10 million
Negative earnings

9
13
28
121
136
193
388
2,146
306

28
30
50
120
176
238
427
1,582
2,144

$20,781
9,377
9,716
18,235
9,327
6,814
6,261
5,601
1,148

$64,229
21,069
17,094
18,555
12,120
8,325
6,880
4,567
54,991

24.4
11.0
11.4
21.4
11.0
8.0
7.4
6.6
1.4

65.6
21.5
17.5
19.0
12.4
8.5
7.0
4.7
56.2

Total
Total positive earnings only

3,340
3,034

4,795
2,651

$84,964
$86,112

$97,848
$152,839

100.0
—

100.0
—

1.
2.
3.
4.
5.
6.
7.
8.
9.

Panel B. Cross-sectional distributions of five-year average real earnings ending in 1978 or in 2000
Five year average
real earnings
(1978 dollars)

Number of firms

Real earnings
($ millions)

Real earnings as a
% of total (%)

1978

2000

1978

2000

1978

2000

$1 billion or greater
$500 million - $1 billion
$250–500 million
$100–250 million
$50–100 million
$25–50 million
$10–25 million
$0–10 million
Negative earnings

9
10
22
106
128
179
348
2,147
391

21
21
45
92
149
213
367
1,742
2,145

$18,541
7,261
8,115
15,734
8,916
6,425
5,480
5,260
627

$42,709
13,328
15,230
14,663
10,243
7,424
5,798
4,862
22,685

24.7
9.7
10.8
20.9
11.9
8.6
7.3
7.0
0.8

46.6
14.6
16.6
16.0
11.2
8.1
6.3
5.3
24.8

Total
Total positive earnings only

3,340
2,949

4,795
2,650

$75,105
$75,732

$91,572
$114,256

100.0
—

100.0
—

1.
2.
3.
4.
5.
6.
7.
8.
9.

ARTICLE IN PRESS
438

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

$64.2 billion (65.6% of aggregate earnings), while the 2,144 firms at the bottom lost
$55.0 billion ( 56.2%). This $55.0 billion in losses far exceeds the $1.1 billion total
lost by 306 firms in 1978. This large upsurge in losses is consistent with the findings
of Hayn (1995), Burgstahler and Dichev (1997), Fama and French (2001a,b), and
Ritter and Welch (2002) who document a substantially increased loss incidence in
recent years. While in 2000 most firms (1,554, not reported in Table 6) lost less than
$10 million, 94 firms lost at least $100 million. Technology firms are prominent
among the latter group, which includes Amazon, Web MD, Webvan, Priceline,
Covad, Akamai, Ariba, JDS Uniphase, Earthlink, Broadcom, PSINet, MP3.Com,
and CMGI. DeAngelo and DeAngelo (1990) and DeAngelo, DeAngelo, and Skinner
(1992) document that losses play a key role in dividend cuts and omissions. These
findings, combined with recent years’ large increase in the incidence of losses, help
explain why so many fewer industrial firms now pay dividends.
4.2. The separate earnings of dividend payers and nonpayers
The strong link between losses and the failure to pay dividends is evident from
Table 7, which partitions the pooled distributions of real earnings from Table 6 into
separate distributions for dividend payers and for nonpayers. Virtually all (2,056 of
2,144, or 95.9%) firms that reported losses failed to pay a dividend in 2000 (row 9 of
panel A). The view that losses have driven many more firms to forego dividends is
further supported by the fact that in 2000 nonpayers as a group lost $10.4 billion,
while dividend payers as a group earned $108.3 billion.
Both panels of Table 7 document a strong positive relation between the level of
earnings and the proportion of firms paying dividends, and they also show that the
relation in 2000 is weaker than it was in 1978. For example, only 2.3% of the firms
with earnings of $100 million-plus failed to pay dividends in 1978, compared to
28.1% in 2000. The fact that a smaller proportion of firms with a given level of real
earnings paid dividends in 2000 than did so in 1978 is consistent with Fama and
French’s (2001a) conclusion that industrial firms now exhibit a lower propensity to
pay dividends (although, unlike the estimates of Fama and French, our numbers do
not control for growth opportunities). In 2000, a number of firms with large positive
earnings failed to pay dividends (see rows 1–3 in both panels of Table 7), whereas no
firms with comparably large earnings failed to do so in 1978.
Despite this reduced propensity to pay, aggregate real dividends increased by $7.1
billion from 1978 to 2000 (per Table 2). Several factors are jointly responsible. Most
fundamentally, aggregate real earnings increased from 1978 to 2000 and, while
earnings are concentrated in 1978, they are more so in 2000. Second, although the
percentage of firms with earnings of $1 billion-plus that pays dividends has fallen
from 100.0% in 1978 to 85.7% in 2000, the percentage remains high, and these firms’
earnings have increased substantially, from $20.8 billion in 1978 to $55.7 billion in
2000. Importantly, this group’s dividends increased by $8.9 billion (not reported in
the table), despite its reduced fraction of payers, and this amount exceeds the entire
$7.1 billion increase in aggregate dividends from 1978 to 2000. In sum, the
substantial increase in real earnings at the top end of the distribution, coupled with

1978 Number of firms
Payers

Non Percentage Payers Non Percentage
payers
from
payers
from
payers (%)
payers (%)

Panel A. Real earnings distribution for payers
1. $1 billion or greater
9
0
2. $500 million - $1 billion
13
0
3. $250–500 million
28
0
4. $100–250 million
117
4
5. $50–100 million
130
6
6. $25–50 million
185
8
7. $10–25 million
358
30
8. $0–10 million
1,284
862
9. Negative earnings
52
254
Total

2,176

1,164

65.1

930

3,865

19.4

2000 Real earnings ($ millions)

Payers

Non
payers

Percentage
from
payers (%)

Payers

$20,781
9,377
9,716
17,619
8,961
6,519
5,802
4,404
480

$0
0
0
616
365
295
459
1,197
668

100.0
100.0
100.0
96.6
96.1
95.7
92.7
78.6
41.8

$55,687
16,207
13,091
12,242
6,543
3,933
2,531
1,178
3,160

$8,542
4,862
4,003
6,313
5,577
4,392
4,350
3,389
51,831

$82,701

$2,263

97.3

$108,251

$10,403

100.0
100.0
100.0
99.0
96.1
95.4

$40,563
11,933
11,579
12,218
6,730
4,046

$2,145
1,395
3,651
2,445
3,513
3,378

distribution for payers and nonpayers ending in 1978 and in 2000
0
100.0
20
1
95.2
$18,541
$0
0
100.0
19
2
90.5
7,261
0
0
100.0
34
11
75.6
8,115
0
1
99.1
75
17
81.5
15,579
155
5
96.1
96
53
64.4
8,569
347
8
95.5
114
99
53.5
6,127
298

Non
payers

Percentage
from
payers (%)
86.7
76.9
76.6
66.0
54.0
47.2
36.8
25.8
5.7
—

95.0
89.5
76.0
83.3
65.7
54.5

439

Panel B. Five-year average real earnings
1. $1 billion or greater
9
2. $500 million - $1 billion
10
3. $250–500 million
22
4. $100–250 million
105
5. $50–100 million
123
6. $25–50 million
171

and nonpayers in 1978 and in 2000
100.0
24
4
85.7
100.0
23
7
76.7
100.0
38
12
76.0
96.7
79
41
65.8
95.6
95
81
54.0
95.9
113
125
47.5
92.3
154
273
36.1
59.8
316
1,266
20.0
17.0
88
2,056
4.1

1978 Earnings ($ millions)

ARTICLE IN PRESS

Real
earnings
(1978 dollars)

2000 Number of firms

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 7
Real earnings (1978 dollars) for industrial firms in 1978 and in 2000: Sample partitioned into dividend payers and nonpayers
Panel A reports the distribution of real earnings for payers and nonpayers in 1978 and in 2000. Panel B reports the distribution of average real earnings over
the five-year period ending with 1978 (or 2000), or over as many of those years as Compustat provides earnings data. The sample consists of NYSE,
NASDAQ, and AMEX firms on CRSP with share codes 10 or 11 and SIC codes outside the intervals 4900–4949 and 6000–6999. A firm is included in a given
year only if Compustat reports data on dividends and earnings (Compustat items 21 and 18). A firm’s dividends and earnings are the amounts reported for
the fiscal years ending in 1978 or 2000. Real earnings are nominal earnings before extraordinary items converted to 1978 dollars using the consumer price
index. The ‘‘percentage from payers’’ columns report the percent of total earnings that comes from dividend-paying firms. Amounts have been rounded to the
nearest significant digit, which accounts for a few minor discrepancies across category totals.

440

Table 12. (Continued)
1978 Number of firms
Payers

Non Percentage Payers Non Percentage
payers
from
payers
from
payers (%)
payers (%)

1978 Earnings ($ millions)

2000 Real earnings ($ millions)

Payers

Non
payers

Payers

Percentage
from
payers (%)

Non
payers

Percentage
from
payers (%)

7. $10–25 million
8. $0–10 million
9. Negative earnings

330
1,387
19

18
760
372

94.8
64.6
4.9

168
361
43

199
1,381
2,102

45.8
20.7
2.0

5,200
4,376
52

280
884
575

94.9
83.2
8.3

2,770
1,444
628

3,028
3,418
22,057

47.8
29.7
2.8

Total

2,176

1,164

65.1

930

3,865

19.4

$73,716

$1,389

98.2

$90,656

$916

99.0

ARTICLE IN PRESS

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Real
earnings
(1978 dollars)

2000 Number of firms

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

441

the continued strong tendency of top earners to pay dividends, is sufficient to
generate an aggregate dividend increase despite both the large reduction in the
number of small dividend payers and the modest increase in the number of firms with
very large earnings that fail to pay dividends.
4.3. Payout ratios and the propensity to pay dividends
The term ‘‘reduced propensity to pay’’ seems to imply that dividend-paying firms
now distribute a lower proportion of their earnings than they used to.3 That there
has not been a reduced propensity to pay in this sense is clear from the essentially flat
payout ratio data in Table 8. Row 1 of the table reports that the ratio of aggregate
dividends to aggregate earnings of payers and nonpayers combined increased
slightly, both when current year earnings are used in the denominator (from 36.9%
to 39.3%), and when five-year average earnings are used (from 41.7% to 42.0%).4 A
shortcoming of the row 1 payout ratios is that they combine the earnings of payers
and nonpayers and thus provide no information about the time series behavior of the
payout ratios of dividend-paying firms.
Table 8’s remaining payout ratios remedy this deficiency by incorporating only the
earnings of dividend-paying firms. Row 2 reports the ratio of aggregate dividends to
the total earnings of dividend payers, while row 3 presents the median of individual
firms’ payout ratios. Rows 4 and 5 report the same statistics for the ‘‘constant
composition sample’’ of 474 firms that paid dividends in both 1978 and 2000.
Although the latter sample suffers from a survivorship bias, it also likely captures
genuine changes in payout policy as opposed to differences over time in the
population of dividend-paying firms. The payout ratios in rows 3 and 5 give equal
weight to all observations, while those in rows 2 and 4 give more weight to firms with
large dividends and earnings.
Overall, payout ratios exhibit little change over the last two decades, with some of
the ratios in Table 8 exhibiting slight increases and others slight decreases. Row 2
shows that the ratio of aggregate dividends to total earnings of payers declined 2.4%,
from 37.9% in 1978 to 35.5% in 2000, based on a single year’s earnings, while the
ratio based on five-year average earnings decreased 0.1%, from 42.5% to 42.4%.
Row 3 indicates that the median payout ratio increased 2.2%, from 26.2% to 28.4%,
based on the one-year earnings measure, and decreased 0.7%, from 31.1% to 30.4%,
based on five-year average earnings. For the constant composition sample in rows 4
3
Fama and French (2001a) use ‘‘reduced propensity to pay’’ in a different sense, to represent the idea
that there are now more firms that pay zero dividends than would have done so historically based on their
economic fundamentals. The payout ratio evidence discussed below is consistent with that of Fama and
French (2001a, p. 38) who find no indication of a decline over 1978–1998 in the ratio of aggregate
dividends to the total earnings of payers.
4
When we calculate an aggregate dividend payout ratio using the summed earnings of all payers and
nonpayers with positive earnings, we find a decline from 36.4% in 1978 to 25.2% in 2000 (from 41.4% to
33.7% using five-year average earnings). But these declines reflect the fact that the earnings divisor of this
payout ratio includes the positive earnings of firms that paid no dividends in 2000, including a small
number of very large earners such as Microsoft, which we discuss in Section 5. Hence, these declines are
not evidence that firms that do pay dividends now have lower payout ratios.

ARTICLE IN PRESS
442

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 8
Aggregate and median dividend payout ratios for industrial firms on CRSP/Compustat, 1978 and 2000
The payout ratios in rows 1 and 2 are based on aggregate dividends paid by industrial firms in 1978 or in
2000. Row 1 takes the denominator to be the sum of earnings for all industrials (payers and nonpayers),
while row 2 takes the denominator to be the sum of earnings for payers only. Row 3 reports the median
firm’s payout ratio within the set of firms that paid dividends. The payout ratios in row 4 and 5 are based
on dividends and earnings for the constant composition sample of 474 firms that paid dividends in both
1978 and in 2000. Row 4 defines the payout ratio in a given year as (1) total dividends paid by firms in the
constant composition sample divided by (2) total earnings of all firms in that sample. Row 5 reports the
median firm’s payout ratio within the constant composition sample. The columns marked ‘‘One-year
earnings’’ report payout ratios based on earnings in the year in question. The columns marked ‘‘Five-year
average real earnings’’ report payout ratios based on a firm’s average real earnings over the five years
ending with the year in question (or as many of those years that Compustat reports earnings data for the
firm). For example, for a given firm in 2000, the earnings variable is the average of earnings over the five
years 1996–2000, with each year’s earnings converted to 2000 dollars using the consumer price index. The
full sample of industrial firms consists of NYSE, NASDAQ, and AMEX firms on CRSP with share codes
10 or 11 and SIC codes outside the intervals 4900–4949 and 6000–6999. A firm’s dividends and earnings
are the amounts reported for the fiscal years ending in 1978 or 2000. A firm is included in a given year only
if Compustat has data on dividends and earnings (Compustat items 21 and 18). Earnings are before
extraordinary items.
One-year earnings (%)

Five-year average
real earnings (%)

Payout ratio measure

1978

2000

1978

2000

1. Aggregate dividends/aggregate earnings
(payers and nonpayers pooled)
2. Aggregate dividends/total earnings
of dividend payers
3. Median firm’s payout ratio
(dividend payers)
4. Constant composition sample of firms that
paid dividends in both 1978 and 2000:
total dividends/total earnings of these dividend payers
5. Constant composition sample of firms that
paid dividends in both 1978 and 2000:
median firm’s payout ratio

36.9

39.3

41.7

42.0

37.9

35.5

42.5

42.4

26.2

28.4

31.1

30.4

41.3

37.7

47.0

43.6

27.5

33.0

33.7

31.9

and 5, three of four payout ratios decreased (by 3.6%, 3.4%, and 1.8%) while the
other increased (by 5.5%). For this last sample, the median change in the payout ratio
(as opposed to the change in the median, which is reported in Table 8) is 5.2% from
1978 to 2000 based on one-year earnings and 2.3% based on five-year average
earnings (details not in table). In sum, we find little indication of a substantial change
(up or down) over the last two decades in the payout ratios of dividend-paying firms.5
5
Fama and French (2001a, p. 35) study the recent upsurge in stock repurchases and find that repurchase
activity is dominated by dividend-paying firms, so that repurchases increase the ‘‘already high’’ cash
payouts of payers. Thus, although repurchases have increased in recent years (both absolutely and relative
to dividends), dividend payers have apparently used repurchases to increase the overall fraction of
earnings distributed to stockholders.

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

443

5. The identity of the top payers, nonpayers, and earners in 2000
Table 9 identifies the 25 industrials that paid the largest dividends in 2000, ranked
in descending order of dividends paid. These primary dividend suppliers are wellestablished firms such as Exxon Mobil and General Electric, with 14 of the 25
included in the Dow Jones Industrial Average (DJIA). The top 25 payers distributed
54.9% of aggregate industrial dividends in 2000, and their $9.2 billion real dividend
increase over 1978–2000 markedly exceeds the $7.1 billion aggregate increase for
industrials as a whole (per Table 2). Their dominance of the aggregate dividend
supply reflects these 25 firms’ $50.3 billion in real earnings ($132.8 billion in year
2000 dollars), which is 51.4% of aggregate industrial earnings, and represents a $26.2
billion real earnings increase from 1978 to 2000. Nineteen of the top 25 payers had $1
billion-plus in real earnings for 2000, and the real earnings increases of the top three
alone totaled $9.7 billion over 1978–2000. In short, a handful of ‘‘old line’’ firms now
both generates over half of industrial earnings and dominates the associated supply
of dividends.
While earnings concentration is striking among dividend payers, it is also very
high among nonpayers. Table 10 ranks the 25 nonpayers with the highest 2000
earnings, and reports their cumulative share of the earnings of all nonpayers with
positive income. Four nonpayers—Microsoft, the only nonpayer in the DJIA,
Oracle, WorldCom, and Cisco—reported real earnings of more than $1 billion in
2000, and these firms alone account for 20.6% of all positive earnings of nonpayers.
The top 10 and top 25 nonpayers respectively generated 31.0% and 43.2% of such
earnings, and just 39 nonpayers accounted for a majority (50.1%, not reported in
Table 10).6 That a handful of technology firms dominates the earnings of nonpayers
is clear from scanning the identities of the 25 firms in Table 10, and any dampening
of the aggregate dividend supply in recent years is due in no small part to their failure
to initiate dividends.
These firms’ decisions to forgo dividends more likely reflect their continued high
growth prospects than a reduced propensity to pay dividends and, as their growth
prospects decline, they will likely come under pressure to pay dividends. Microsoft,
widely viewed as a bellwether technology company, has recently announced plans to
institute a regular dividend, and Qualcomm and Fedex have also initiated dividends,
so that three of the top 25 nonpayers in Table 10 would now be reclassified as payers.
Additionally, the numerous corporate scandals uncovered in the last year or so may
prompt growth firms to initiate dividends at an earlier stage, in order to increase
investor confidence in their reported earnings. Moreover, President Bush’s January
2003 proposal to eliminate the dividend tax penalty would, if enacted, surely put
more pressure on firms to pay dividends. All these recent developments suggest that
a future reversal of the decline in the number of dividend-paying industrials is a
reasonable possibility.

6

The earnings of the top 25 nonpayers represent 11.7% of the earnings of all firms (payers and
nonpayers combined) with positive income, and 18.3% of aggregate industrial earnings.

444

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.

Exxon Mobil
General Electric
Philip Morris
Verizon
SBC
Merck
Ford
Pfizer
AT&T
Bristol Myers Squibb

Real earnings
($ millions, 1978 base)

Nominal dividends
and earnings
($ millions)

1978

2000

Change

1978

2000

Change

2000

2000

$1,472
570
125
—
—
132
417
82
3,038
77

$2,318
2,138
1,722
1,672
1,304
1,100
1,036
973
941
731

$846
1,568
1,597
1,672
1,304
968
619
890
2,097
654

$2,763
1,230
409
—
—
308
1,589
206
5,273
203

$6,054
4,822
3,222
4,093
3,017
2,583
2,048
1,408
1,768
1,551

$3,291
3,592
2,814
4,093
3,017
2,275
460
1,202
3,505
1,348

$6,122
5,647
4,548
4,416
3,444
2,905
2,736
2,570
2,485
1,931

$15,989
12,735
8,510
10,810
7,968
6,822
5,409
3,719
4,669
4,096

ARTICLE IN PRESS

Real dividends
($ millions, 1978 base)

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 9
Dividends and earnings in 1978 and in 2000 for the 25 industrial firms that paid the largest dividends in 2000
The table lists the 25 industrial firms on CRSP/Compustat that paid the largest total dividends in 2000, with firms ranked in descending order of dividends paid.
All but three of these firms are in our 1978 sample and also paid dividends in that year. [Verizon (Bell Atlantic), SBC, and BellSouth were subsidiaries of AT&T
in 1978, and are ‘‘baby bells’’ that were spun off in 1984. AT&T’s real dividends and earnings in 2000 would be well above their 1978 levels if we added back the
dividends and earnings of the spun off ‘‘baby bells.’’] Real dividends and earnings in 2000 are nominal values converted to 1978 dollars using the consumer
price index. A firm’s dividends and earnings are the amounts reported for the fiscal years ending in 1978 or 2000. The next-to-last and last columns respectively
give year 2000 dividends and earnings in nominal terms (i.e., denominated in year 2000 dollars).

101
435
215
223
348
—
1,713
207
47
116
543
234
1,763
2
65

653
639
638
636
551
539
490
455
446
439
370
348
344
337
304

552
205
423
413
203
539
1,223
248
399
322
173
114
1,419
335
239

299
1,106
375
512
787
—
3,508
348
149
277
852
563
3,111
22
194

1,817
1,963
824
1,341
876
1,598
1,686
341
1,055
1,158
962
703
3,064
2,111
917

1,518
857
450
829
89
1,598
1,822
690
906
880
110
140
46
2,089
724

1,725
1,688
1,685
1,680
1,455
1,424
1,294
1,202
1,178
1,159
977
919
909
890
803

4,799
5,184
2,176
3,542
2,314
4,220
4,453
901
2,786
3,058
2,541
1,857
8,092
5,575
2,422

Total for 25 firms
Total as a % of aggregate
for all industrials

$11,925
38.0%

$21,124
54.9%

$9,198

$24,084
28.3%

$50,300
51.4%

$26,219

$55,792
54.9%

$132,845
51.4%

ARTICLE IN PRESS

Johnson & Johnson
Chevron
Coca-Cola
Procter & Gamble
Du Pont
BellSouth
General Motors
American Home Products
Abbott Labs
Eli Lilly
Texaco
3M
IBM
Wal-Mart
Schering-Plough

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.

445

ARTICLE IN PRESS
446

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 10
2000 earnings for the 25 nondividend-paying industrial firms with the highest reported earnings
The table lists the 25 nondividend-paying industrial firms on CRSP/Compustat with the highest earnings
in 2000, ranked in descending order of earnings in that year. WorldCom is classified by Compustat as not
paying dividends, even though the tracking stock for its MCI unit did pay dividends in 2000. Our empirical
work employs Compustat dividend amounts for all sample firms. The recent accounting scandal at
WorldCom implies that its year 2000 Compustat earnings are overstated. Real earnings are nominal
earnings in 2000 converted to 1978 dollars using the consumer price index. A firm’s earnings for 2000 is the
amount reported for the fiscal year ending in that year. The last column gives the cumulative total earnings
as a fraction of the total earnings in 2000 of all nondividend payers with positive earnings in that year. All
earnings figures are before extraordinary items.

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.

Microsoft
Oracle
WorldCom
Cisco Systems
Applied Materials
Comcast
Cox Communications
Sun Microsystems
EMC
Dell Computer
Micron Technology
AOL Time Warner
Amgen
Safeway
Advanced Micro Devices
ADC Telecommunications
Federated Department Stores
Apple Computer
AMR
Tellabs
Agilent Technologies
Fedex
Qualcomm
Xilinx
Alliance Semiconductor
Total for 25 firms

Real earnings
in 2000
($ millions,
1978 base)

Nominal earnings
in 2000
($ millions)

Cumulative earnings as a
percent of total earnings of
all nonpayers with positive
earnings in 2000 (%)

$3,567
2,384
1,580
1,010
781
774
729
702
675
631
570
466
431
413
381
329
301
298
295
288
287
261
254
247
245

$9,421
6,297
4,174
2,668
2,064
2,045
1,925
1,854
1,782
1,666
1,504
1,232
1,139
1,092
1,006
868
795
786
779
760
757
688
670
652
648

8.6
14.4
18.2
20.6
22.5
24.4
26.1
27.8
29.5
31.0
32.4
33.5
34.5
35.5
36.4
37.2
38.0
38.7
39.4
40.1
40.8
41.4
42.0
42.6
43.2

$17,899

$47,273

43.2

Table 11 identifies the 28 firms with $1 billion-plus in real earnings in 2000, which
is just over $2.5 billion in year 2000 dollars. Together, these 28 firms account for
almost two-thirds (65.6%) of 2000 aggregate industrial earnings. In 2000, 24 of these
firms paid dividends, and in total their dividends represent a majority (50.1%) of the
industrial dividend supply. The 28 largest earners are primarily ‘‘old line’’ firms that

Exxon Mobil
General Electric
Verizon
Intel
Microsoft
Philip Morris
IBM
SBC
Merck
Oracle
Wal-Mart
Ford
Chevron
Johnson & Johnson
AT&T
Tyco
General Motors
BellSouth
WorldCom

Nominal dividends
and earnings
($ millions)

1978

2000

Change

1978

2000

Change

2000

2000

$1,472
570
—
0
—
125
1,763
—
132
—
2
417
435
101
3,038
1
1,713
—
—

$2,318
2,138
1,672
178
0
1,722
344
1,304
1,100
0
337
1,036
639
653
941
32
490
539
0

$846
1,568
1,672
178
0
1,597
1,419
1,304
968
0
335
619
205
552
2,097
31
1,223
539
0

$2,763
1,230
—
44
—
409
3,111
—
308
—
22
1,589
1,106
299
5,273
12
3,508
—
—

$6,054
4,822
4,093
3,989
3,567
3,222
3,064
3,017
2,583
2,384
2,111
2,048
1,963
1,817
1,768
1,711
1,686
1,598
1,580

$3,291
3,592
4,093
3,945
3,567
2,814
46
3,017
2,275
2,384
2,089
460
857
1,518
3,505
1,699
1,822
1,598
1,580

$6,122
5,647
4,416
470
0
4,548
909
3,444
2,905
0
890
2,736
1,688
1,725
2,485
85
1,294
1,424
0

15,989
12,735
10,810
10,535
9,421
8,510
8,092
7,968
6,822
6,296
5,575
5,409
5,184
4,799
4,669
4,519
4,453
4,220
4,173

447

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.

Real earnings
($ millions, 1978 base)

ARTICLE IN PRESS

Real dividends
($ millions, 1978 base)

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 11
Dividends and earnings in 1978 and in 2000 for the 28 industrial firms with at least $1 billion in real earnings in 2000
The table lists the 28 industrial firms on CRSP/Compustat that report at least $1 billion in real earnings in 2000, with firms ranked in descending order of 2000
earnings. Real dividends and earnings in 2000 are nominal values converted to 1978 dollars using the consumer price index. A firm’s dividends and earnings are
the amounts reported for the fiscal years ending in 1978 or 2000. The next-to-last and last columns respectively give year 2000 dividends and earnings in nominal
terms (i.e., denominated in year 2000 dollars). WorldCom is classified by Compustat as not paying dividends in 2000, even though the tracking stock for its
MCI unit did pay dividends in that year, and we follow Compustat’s classification. Verizon (Bell Atlantic), Microsoft, SBC, Oracle, BellSouth, WorldCom,
UPS, and Cisco are not in our 1978 sample. However, Verizon, SBC, and BellSouth are ‘‘baby bells’’ that were spun off from AT&T in 1984, and thus in 1978
were subsidiaries of AT&T (which is in the 1978 sample). The numbers are rounded to the nearest significant digit, which accounts for a few discrepancies across
category totals.

448

Table 12. (Continued)
Real dividends
($ millions, 1978 base)

Total for 28 firms
Total as a % of aggregate
for all industrials

1978

2000

Change

1978

2000

Change

2000

2000

77
82
14
223
40
116
—
47
—

731
973
242
636
53
439
298
446
0

654
890
228
413
13
322
298
399
0

203
206
153
512
140
277
—
149
—

1,551
1,408
1,348
1,341
1,169
1,158
1,111
1,055
1,010

1,348
1,202
1,195
829
1,029
880
1,111
906
1,010

1,931
2,570
639
1,680
140
1,159
787
1,178
0

4,096
3,719
3,560
3,542
3,087
3,058
2,934
2,786
2,668

$10,368
33.1%

$19,261
50.1%

$8,892

$21,314
24.8%

$64,229
65.6%

$42,916

$50,872
50.1%

$169,629
65.6%

ARTICLE IN PRESS

Bristol Myers Squibb
Pfizer
Hewlett-Packard
Procter & Gamble
Texas Instruments
Eli Lilly
UPS
Abbott Labs
Cisco

Nominal dividends
and earnings
($ millions)
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

20.
21.
22.
23.
24.
25.
26.
27.
28.

Real earnings
($ millions, 1978 base)

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

449

pay very large dividends, but they also include some profitable younger technology
firms that pay either modest or no dividends. Of the four large earners that failed
to pay 2000 dividends, Microsoft has since initiated a dividend, Oracle and Cisco
have said they would consider doing so, and WorldCom’s accounting fraud implies
that its #19 ranking in Table 11 substantially overstates its true earnings position in
2000.
Overall, industrial firms are characterized by a two-tier structure based on
earnings, with 25 or so firms (about one-half of 1% of all industrials) accounting for
most earnings and dividends in 2000, and with the vast majority of firms collectively
delivering small earnings and dividends. Tables 9–11 show that the top handful of
earners generated the bulk of corporate earnings and, despite some large earners like
Microsoft that did not pay dividends, also supplied the majority of industrial
dividends in 2000. We also know from Table 6 that 3,726 firms or 77.7% of all
industrial firms in 2000 individually earned $10 million or less, and these firms as a
group lost about $50 billion. These 3,726 firms paid total dividends of $2.1 billion, or
5.5% of the aggregate industrial supply. More than half of this $2.1 billion total was
paid by seven large firms with current losses or very low earnings, and more than
three-quarters was paid by 30 firms in a similar situation (including American Home
Products, Xerox, Lockheed, USX, Dow Jones, and Pennzoil). Therefore, in 2000,
some 3,700 firms (over three-quarters by number) collectively paid dividends of
about $0.5 billion (1.3% of the aggregate), whereas the top 25 payers (0.5% by
number) paid roughly 40 times that amount.

6. What happened to the dividend payers of 1978?
Table 12 classifies the 2,176 firms that paid dividends in 1978 (column (1))
according to whether they also paid dividends in 2000 (column (2)), remained listed
in 2000 but did not pay dividends in that year (column (3)), and the primary reasons
for delisting, either financial distress (column (4)) or acquisition (column (5)).
Financially distressed delists include all firms with CRSP delist codes 500–599
(delisted or stopped trading) or 400–499 delist codes (liquidations) for which The
Wall Street Journal Index (WSJI) provides no clear indication that the firm was
acquired. Acquisition delists include all firms with delist codes 200–299 (mergers), or
300–499 (security exchanges and liquidations) for which the WSJI indicates that the
company was acquired.7
Although much reduced in number, the set of 474 firms that paid dividends in
both 1978 and 2000 accounts for a large majority of industrial dividends and
earnings in 2000 (79.5% and 83.1% respectively)—another manifestation of the fact
7

This classification scheme is similar to that of Fama and French (2001b, Table 4), who treat codes 200–
399 as delisted due to merger and 400–599 as delisted for ‘‘cause’’. The 1,647 firm total of columns (3)–(5)
in our Table 12 falls 55 firms short of the 1,702 firm difference between columns (1) and (2). The reason is
that we omit 40 delists with codes 300–399 that the WSJI did not show as acquired and 15 firms without
dividends and earnings on Compustat for 2000. The former firms had paid $136 million in 1978 dividends,
while the latter had paid $254 million.

ARTICLE IN PRESS
450

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Table 12
Listing and dividend status in 2000 for the 2,176 industrial firms that paid dividends in 1978: Sample
partitioned by size of 1978 dividend payment
The sample includes all NYSE, NASDAQ, and AMEX firms on CRSP that have share codes 10 and 11
and SIC codes outside the ranges 4900–4949 and 6000–6999, and that have nonmissing values on
Compustat of dividends and earnings before extraordinary items for 1978 (Compustat items 21 and 18).
Column (2) contains firms that were in our sample in 1978 and in 2000, and that paid dividends in both
years. Column (3) contains firms that paid dividends in 1978 and that remained publicly traded in 2000,
but no longer paid dividends. Columns (4) and (5) contain dividend-paying firms that are in our 1978
sample, but that were delisted post-1978 due either to acquisition or financial distress, and thus are not in
the 2000 sample. The financially troubled delistings in column (4) include (i) all cases with CRSP delist
codes in the range 500–599 and (ii) those cases with delist codes in the range 400–499 for which we found
no evidence in the Wall Street Journal Index (WSJI) that the firm was acquired. The acquisition delistings
in column (5) include (i) all cases with CRSP delist codes in the range 200–299 and (ii) those cases with
CRSP delist codes in the range 300–499 for which we found evidence in the WSJI that the company was
acquired. The subsamples in columns (2) through (5) are mutually exclusive, but not exhaustive. They
exclude 40 firms with delist codes in the range 300–399 for which we found no evidence in the WSJI that
the firm was acquired. They also exclude 15 firms for which Compustat does not report dividends and
earnings data for 2000. A firm’s dividends and earnings are the amounts reported for the fiscal years
ending in 1978 or 2000. Real dividends and earnings in 2000 are nominal values converted to 1978 dollars
using the consumer price index.

Dividend payment
in 1978
1. $500 million or greater
2. $400–499.9 million
3. $300–399.9 million
4. $200–299.9 million
5. $100–199.9 million
6. $80–99.9 million
7. $60–79.9 million
8. $40–59.9 million
9. $20–39.9 million
10. $10–19.9 million
11. $5–9.9 million
12. $1–4.9 million
13. Less than $1 million
Total number of firms
(% of 1978 industrial total)
Total 1978 dividends
(% of 1978 industrial total)
Total 2000 real dividends
(% of 2000 industrial total)
Total 1978 earnings
(% of 1978 industrial total)
Total 2000 real earnings
(% of 2000 industrial total)

All
dividend
payers in 1978
(1)

Paid
dividends
in 2000
(2)

Listed, but
not dividend
payer
in 2000
(3)

Delisted
due to
financial
distress
(4)

Delisted
due to
acquisition
(5)

6
4
4
9
19
18
24
55
108
161
187
633
948
2,176 firms
(100.0%)
$31.3 billion
(100.0%)
—

6
2
3
6
14
12
9
19
37
58
51
143
114
474 firms
(21.8%)
$19.5 billion
(62.3%)
$30.6 billion
(79.5%)
$47.3 billion
(55.7%)
$81.3 billion
(83.1%)

0
0
0
0
0
0
5
4
7
5
9
43
86
159 firms
(7.3%)
$1.0 billion
(3.2%)
$0.0 billion
(0.0%)
$3.9 billion
(4.6%)
$2.0 billion
(2.0%)

0
0
0
0
0
0
0
1
1
9
7
46
175
239 firms
(11.0%)
$0.4 billion
(1.3%)
—

0
2
1
3
4
6
10
29
61
84
118
389
542
1,249 firms
(57.4%)
$10.0 billion
(31.9%)
—

$1.3 billion
(1.5%)
—

$28.9 billion
(34.0%)
—

$82.7 billion
(97.3%)
—

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

451

that dividends and earnings are highly concentrated among ‘‘old line’’ firms. The 159
firms that stopped paying dividends between 1978 and 2000 had paid only $1.0
billion (3.2% of the aggregate) in 1978. These firms’ 2000 nonpayments reflect
financial distress for at least some companies, since 2000 total earnings were nearly
50% below 1978 earnings for this group. Most of the 239 financially distressed delists
were small payers, with 221 firms (92.5%) having paid $5 million or less, and these
firms’ 1978 dividends totaled $0.4 billion, or 1.3% of the aggregate. So, while
financial distress and earnings difficulties together reduced the number of dividend
payers by 398 firms between 1978 and 2000, the associated dividend loss is just $1.4
billion. The $10.0 billion (31.9%) dividend loss from acquisition delists is markedly
larger, as is the loss in the number of payers due to acquisitions (1,249 firms, or
57.4% of the 1978 payers).
While acquisitions are thus the primary, and financial distress the secondary,
reason why so many 1978 payers no longer disburse dividends, these two factors
affect aggregate dividends in different ways. The dividends of financially distressed
firms are lost, whereas in many cases an acquired firm’s dividends are not lost, but
are simply relabeled. For example, the issuance of acquirer shares to target
stockholders increases the acquirer’s total dividend, and thereby continues the
target’s dividend at least in part.8 Exxon and Mobil respectively paid $4.0 billion and
$1.8 billion in 1998 dividends, while after their November 1999 merger, the new
Exxon Mobil paid $6.1 billion in 2000 dividends. In the aggregate, such acquisitions
do not eliminate dividends, but simply channel them to investors through a smaller
number of firms. Thus, acquisitions reduce the number of dividend-paying firms with
no necessary reduction in aggregate dividends, a point that carries special
significance here because the post-1978 decline in the number of payers occurred
during a massive merger wave.
In fact, the merger wave of the 1980s and 1990s explains a substantial portion
of the decline in the number of dividend payers over 1978–2000. The abnormal
number of delists attributable to the merger wave equals the actual merger delists
(1,249 firms, per Table 12) minus the number of mergers that would have occurred,
absent the wave. Fama and French (2001a, Table 2) report that dividend payers were
acquired at average annual rates of 3.9% over 1978–1999, 2.7% over 1963–1977,
and 0.6% over 1927–1962. The normal merger rate probably lies somewhere
between 0.6% and 2.7% since 1927–1962 is a wave-free period, while 1963–1977
includes the conglomerate wave. If 0.6% is the relevant benchmark, the abnormal
delists attributable to the recent merger wave is 979 firms, and it is 265 firms if 2.7%
is the appropriate benchmark.9 In either case, a reasonable portion of the actual
8

Acquisitions for stock were especially prevalent in the 1990s. Andrade et al. (2001, pp. 104–106) report
that 57.8% of acquisitions over 1990–1998 were for all stock, while 70.9% involved at least some stock.
They also report that, during the 1980s, 32.9% of acquisitions were for all stock, and 45.6% involved some
stock.
9
We calculate the expected attrition (compounded at either 0.6% or 2.7%) over 22 years starting from a
base of 2,176 firms, the number of 1978 payers. The abnormal number of acquisitions equals 1,249 minus
the expected attrition. The expected attrition at Fama and French’s 3.9% merger rate over 1978–1999 is
1,269 firms, which is virtually identical to the 1,249-firm decline in Table 12.

ARTICLE IN PRESS
452

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

merger delists, hence of the decline in the number of dividend payers, is due to
the recent merger wave. However, as Fama and French (2001a, p.11) point out,
since dividend payers and nonpayers were acquired at about the same rates during
the last two decades, acquisitions cannot explain the decline in the proportion of
payers.

7. Summary and implications
This paper reports evidence that industrial firms’ dividends are highly
concentrated, and that dividend concentration has increased over the past two
decades. We document that, while many fewer firms paid dividends in 2000 than did
so in 1978, aggregate real dividends increased over that period. The combination of a
decreased number of payers and increased aggregate dividends reflects high and
increased earnings concentration. In 2000, most firms with very high earnings paid
dividends, and the increased real earnings of the largest dividend payers is
responsible for the aggregate increase in dividends and the concomitant increase in
dividend concentration over 1978–2000. In 2000, nearly half of industrial firms
reported losses and, as one would expect, few of these firms paid dividends. The
decline over 1978–2000 in the number of dividend payers occurred predominantly
among firms that previously paid very small real dividends, and is due primarily to
acquisitions and secondarily to financial distress. For example, we find that 57.4% of
the firms that paid dividends in 1978 were subsequently delisted because they were
acquired, and that the abnormal level of acquisitions during the recent merger
wave accounts for much, but not all, of the decline in the number of payers. We also
find that the payout ratios of firms that pay dividends exhibit little change over
1978–2000.
These findings collectively suggest that the decline in dividend payers over 1978–
2000 is not attributable to factors that put across-the-board downward pressure on
dividends or on payout ratios. For example, income tax law changes cannot
plausibly underlie the decline in payers, since any increase in the dividend tax penalty
that led many more firms to pay zero dividends also should have led dividend payers
to markedly reduce their payout ratios, and we observe no such reduction. The same
logic implies that the sharp reduction in the number of payers was not caused by
demand- and/or supply related factors that generated a cross-sectionally pervasive
reduction in the marginal incentive to pay dividends. This inference is reinforced by
the fact that the earnings of nonpayers, like those of payers, are highly concentrated,
with 25 firms accounting for almost half the total earnings of all nonpayers with
positive earnings in 2000. Over one-third of these total earnings come from just 13
firms with a strong technology bent: Microsoft, Oracle, WorldCom, Cisco, Applied
Materials, Comcast, Cox Communications, Sun, EMC, Dell, Micron Technology,
AOL Time Warner, and Amgen. The decision to forgo dividends by high growth
technology firms with substantial earnings more likely reflects industry-specific
investment opportunities rather than economy-wide factors that reduced all firms’
incentives to pay dividends.

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

453

Our findings on dividend concentration cast doubt on the empirical importance of
the dividend clientele and signaling hypotheses (Allen and Michaely, 1995, survey
the extensive literatures on these two hypotheses). Clientele theories attribute
heterogeneity in dividend policies to the demands of different investors who, for tax
or behavioral reasons, prefer either to hold or to avoid dividend-paying stocks. Black
and Scholes (1974) discuss the difficulties of forming well-diversified portfolios of
stocks that pay either high or low dividends, and note the dearth of nondividend
payers to service the demands of the many investors who prefer capital gains. While
on the surface the large increase since the late 1970s in the number of nondividendpaying firms might appear to rectify this shortcoming, the attributes of the current
population of nonpayers suggest that well-diversified portfolios of their stocks are
not easily constructed. Among nonpayers in 2000, the majority of firms has negative
earnings averaged over 1996–2000, many are newly listed growth firms, and many
are from the technology sector. Even if some investors could construct welldiversified but dividend-free portfolios from this population, it is questionable
whether the aggregate demands of all clienteles seeking to invest in such portfolios
could be met, given the substantial dividend concentration that characterizes today’s
stock market.
If the demand to satisfy heterogeneous clienteles were truly a first-order
determinant of dividend policies, we would expect to observe substantial dividend
heterogeneity among the prominent firms whose securities are important components of well-diversified portfolios. As long as high tax bracket investors desire to
invest substantial amounts of wealth, we should observe a comparably large set of
prominent firms that do not pay dividends, and these nonpayers should be spread
across a broad spectrum of industries. And, within any given industry, we should see
a mix of large dividend-paying and nonpaying firms. What we observe, instead, is (i)
few firms with very high earnings fail to pay dividends, (ii) these firms are mainly
bunched in a narrow industry sector (technology), and (iii) very large firms in other
industries all tend to pay dividends. The fact that the market does not supply a broad
spectrum of dividend heterogeneity either across or within industries suggests that
pressure to satisfy heterogeneous clienteles is at best a second-order determinant of
dividend decisions. It would seem to follow that clientele pressure can have a major
impact only in unusual circumstances, for example, when a controlling stockholder’s
preferences shape a given firm’s dividend policy as, e.g., in DeAngelo and DeAngelo
(2000).
Our finding that dividends are highly concentrated among a small number of firms
with substantial earnings also raises doubts that signaling is a first-order determinant
of corporate dividend policy. If managers use dividends to communicate with
stockholders, dividend signaling should occur primarily in small, relatively unknown
firms with limited access to the financial press, Wall Street analysts, and other
conventional information dissemination outlets. But the vast majority of dividends
are paid by prominent corporations like Exxon Mobil and General Electric that
enjoy major coverage by analysts and journalists—exactly the firms whose managers
should have little need to use financial decisions to communicate with investors. How
much of aggregate dividends can be motivated by signaling when 92.0% of industrial

ARTICLE IN PRESS
454

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

dividends are paid by the top 200 dividend payers? While it is possible that signaling
motives may be important on the margin for some prominent dividend-paying firms,
it is hard to envision plausible scenarios in which a material portion of aggregate
dividends reflects signaling motives.
Our evidence on the high and increasing level of dividend concentration adds to a
growing body of empirical research that documents major changes in corporate
payout practices over the last 25–50 years. Prior studies have identified a number of
other important trends, including (i) a marked reduction in the 1970s in the
sensitivity of dividends to earnings, as evidenced by a decline in the typical Lintner
(1956) speed-of-adjustment coefficient coupled with target payout ratio stability
from the late 1940s through at least the mid-1980s (Choe, 1990, 1991), (ii) the virtual
disappearance of special dividends in recent years, despite their prominence in the
1950s and earlier (DeAngelo et al., 2000), (iii) a reduction in firms’ propensity to pay
dividends over the last two decades of the 20th century (Fama and French, 2001a),
(iv) the emergence of stock repurchase as a popular payout technique in the 1960s
and early 1970s (Dann, 1981; Masulis, 1980; Vermaelen, 1981), and (v) the massive
increase in repurchase activity in the mid-1980s (Bagwell and Shoven, 1989; Allen
and Michaely, 1995). The interplay of these trends and their possible relation to the
high and increasing concentration of earnings documented here are issues that merit
future investigation.
Finally, our evidence reveals that publicly traded industrial firms exhibit a two-tier
structure based on dollar earnings. The first tier contains a few very high earners,
most of which pay dividends, and these firms’ dividends collectively dominate the
aggregate supply. The second tier contains many firms which, individually and
jointly, have modest earnings and which collectively contribute little to the aggregate
dividend supply. In essence, the differing behavior of first- and second-tier firms
explains why aggregate dividends increased as the number of payers declined over
the past two decades. The two-tier structure also raises a possible inconsistency
between the findings of Lintner (1956) and Fama and Babiak (1968) and dividend
theories in which supply adjustments play a central role. For example, the Black and
Scholes’ (1974) equilibrium would seem to periodically require material dividend
changes by first-tier firms, since second-tier companies’ collective dividends are
small. But Lintner and Fama and Babiak show that such well-established firms tend
to adjust dividends only marginally and then primarily upward in response to their
own earnings increases. The interesting empirical question is whether the apparently
‘‘sticky’’ dividend practices of first-tier firms exhibit sufficient variation (in both
directions) to provide the supply adjustments critical to the Black and Scholes
dividend theory.
The two-tier structure is perhaps the signature characteristic that determines the
dividend supply of industrial firms, and it has been so for at least two decades (and
probably longer). Although our evidence is limited to dividends, we conjecture that
the small set of top-tier firms is also responsible for the majority of cash payouts via
stock repurchase. Three observations are consistent with this conjecture. First, Fama
and French (2001a) report that stock repurchases are primarily the province of
dividend-paying firms. Second, earnings are highly concentrated, with a handful of

ARTICLE IN PRESS
H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

455

firms generating the bulk of industrial earnings and dividends, and these firms may
also use their earnings to support repurchases. Finally, for S&P 500 firms in the late
1990s, Liang and Sharpe (1999, fig. 1) document substantial gross (as well as net of
stock option exercise) dollar repurchase volume, with gross repurchases the same
order of magnitude as dividends—facts consistent with our conjecture that top-tier
firms dominate aggregate repurchase activity. Whatever the ultimate verdict on
the concentration of repurchase volume, future analyses of payout policy
should recognize that a few large earners dominate the dividend supply, while the
vast majority of firms collectively contributes little to aggregate earnings and
dividends.

References
Allen, F., Michaely, R., 1995. Dividend policy. In: Jarrow, R.A., Maksimovic, V., Ziemba, W.T. (Eds).
Operations Research and Management Science, Elsevier Science, Amsterdam.
Andrade, G., Mitchell, M., Stafford, E., 2001. New evidence and perspective on mergers. Journal of
Economic Perspectives 15, 103–120.
Bagwell, L., Shoven, J., 1989. Cash distributions to shareholders. Journal of Economic Perspectives 3,
129–140.
Black, F., Scholes, M., 1974. The effect of dividend yield and dividend policy on common stock prices and
returns. Journal of Financial Economics 1, 1–22.
Burgstahler, D., Dichev, I., 1997. Earnings management to avoid earnings decreases and losses. Journal of
Accounting and Economics 24, 99–126.
Choe, H., 1990. Intertemporal and cross-sectional variation of corporate dividend policy. University of
Chicago, Ph.D. Dissertation, Chicago, IL.
Choe, H., 1991. Intertemporal variation of corporate dividend policy. Unpublished working paper, The
Pennsylvania State University.
Dann, L., 1981. Common stock repurchases: an analysis of returns to bondholders and stockholders.
Journal of Financial Economics 9, 101–138.
DeAngelo, H., DeAngelo, L., 1990. Dividend policy and financial distress: an empirical examination of
troubled NYSE firms. Journal of Finance 45, 1415–1431.
DeAngelo, H., DeAngelo, L., 2000. Controlling stockholders and the disciplinary role of payout policy: a
study of the Times Mirror Company. Journal of Financial Economics 56, 153–207.
DeAngelo, H., DeAngelo, L., Skinner, D., 1992. Dividends and losses. Journal of Finance 47, 1837–1863.
DeAngelo, H., DeAngelo, L., Skinner, D., 2000. Special dividends and the evolution of dividend signaling.
Journal of Financial Economics 57, 309–354.
Fama, E., Babiak, H., 1968. Dividend policy: an empirical analysis. Journal of the American Statistical
Association 63, 1132–1161.
Fama, E., French, K., 2001a. Disappearing dividends: changing firm characteristics or lower propensity to
pay? Journal of Financial Economics 60, 3–43.
Fama, E., French, K., 2001b. Newly listed firms: fundamentals, survival rates, and returns. Center for
Research in Security Prices working paper No. 530, University of Chicago, Graduate School of
Business.
Hayn, C., 1995. The information content of losses. Journal of Accounting and Economics 20, 125–153.
Liang, J., Sharpe, S., 1999. Share repurchases and employee stock options and their implications for S&P
500 share retirements and expected returns. Federal Reserve Board working paper, Washington, DC.
Lintner, J., 1956. Distribution of incomes of corporations among dividends, retained earnings, and taxes.
American Economic Review 46, 97–113.
Masulis, R., 1980. Stock repurchase by tender offer: an analysis of the causes of common stock price
changes. Journal of Finance 35, 305–319.

ARTICLE IN PRESS
456

H. DeAngelo et al. / Journal of Financial Economics 72 (2004) 425–456

Miller, M., 1977. Debt and taxes. Journal of Finance 32, 261–275.
Ritter, J., Welch, I., 2002. A review of IPO activity, pricing, and allocations. Journal of Finance 57,
1795–1828.
Shoven, J., 1986. The tax consequences of share repurchases and other non-dividend cash payments to
equity owners. In: Summers, L. (Ed.), Tax Policy and the Economy, Vol. I. NBER and MIT Press,
Boston, pp. 29–54.
Vermaelen, T., 1981. Common stock repurchases and market signaling. Journal of Financial Economics 9,
139–183.