ofX$&
HD28
.M414
n
THE EFFECTIVENESS OF ACCOUIITIKG-BASED
DIVIDEND COVENANTS
Du
14
THE EFFECTIVENESS OF ACCOUNTING-BASED
DIVIDEND COVENANTS
Paul M. Healy
School of Management
Massachusetts Institute of Technology
Krishna G. Palepu
Graduate School of Business Administration
Harvard University
lAff
February
1989
Forthcoming: Journal of Accounting and Economics
[MAY
13
1991
i
The paper has benefited from the comments of Joy Bagley, Bob Holthausen, Bob Kaplan,
Katherine Schipper, Ross Watts, Karen Wruck, the conference participants at the University of
Rochester, and seminar participants at Duke University, the University of California at Los
Angeles, and the Wharton School. We would like to thank Jeffrey Liu and Ken Hao for their
research assistance and the Division of Research, Harvard Business School for funding the
research.
THE EFFECTIVENESS OF ACCOUNTING-BASED
DIVIDEND COVENANTS
This paper examines whether firms that are close
to violating their
lending covenants
make
income-increasing accounting decisions to avoid cutting dividends. While firms close to
adopt income-increasing pension accounting changes, these appear to
be motivated by cash management concerns. There is no evidence that these firms make other
accounting changes to circumvent the dividend restriction. However, there is a substantial
increase in the frequency of dividend cuts and omissions for these firms. The magnitude of the
dividend cuts is related to the tightness of the dividend constraint. Thus despite the accounting
their dividend restriction
flexibility
available to managers, accounting-based covenants appear to be effective
bondholders
to restrict firms' dividend policies.
means
for
DBNEV
M.
MAY
1
3 1991
1.
INTRODUCTION
Dividend restrictions, the most
reduce potential
conflicts of interest
common covenant
in
restrictions,
GAAP, managers can
and avoid dividend cuts
through accounting decisions.
response
an increase
to
in
(to
Since there
restricting
restrictions
considerable accounting
is
covenant
potentially circumvent dividend
the detriment of bondholders) by increasing earnings
This paper examines accounting and dividend policy changes
in
the tightness of the dividend covenant restriction to assess whether
accounting earnings-based dividend covenants are effective
themselves by
These
to
terms of accounting earnings, typically under
generally accepted accounting principles (GAAP).
available under
debt contracts, are designed
between bondholders and stockholders.
define funds available for dividend payments
discretion
in
means
managers from using dividend
for
policy
bondholders
to
transfer
to protect
wealth
to
stockholders.
The
covenant
changes
results,
based on a sample
restrictions
in
126 firms which were close
of
to violating their dividend
1981-1985, indicate that there are no significant
during the period
While the sample firms
accounting methods surrounding the near-violation.
firms'
adopt income-increasing pension accounting assumptions during the four years surrounding the
near-violation
of
the
dividend
constraint,
considerations, rather than the increase
the year of the near-violation there
and omissions,
reductions
for
is
changes
is
in
they
appear
to
be motivated by cash flow
tightness of the dividend constraint.
a substantial increase
persisting for several subsequent years.
in
earnings and operating cash flows.
A
The magnitude
small
number
restructure the debt that contains the binding dividend restriction.
accounting decisions play only a limited
accounting discretion available
to
role
in
sample
contrast,
in
the frequency of dividend cuts
related to the degree of tightness of the dividend constraint,
in
In
We
firms' overall
even
of
of the
dividend
after controlling
sample firms also
therefore conclude that
responses.
Despite the
managers, earnings-based dividend covenants appear
to
be
effective
means
The
bondholders
rest of the
paper
is
to constrain firms' dividend policies.
organized as follows. Section two describes the dividend covenant
Section three discusses the sample and the data.
restriction.
are presented
section
for
five.
The accounting response
results
section four, and dividend and debt restructuring responses are reported
in
The
final
in
section presents a discussion of the findings.
DIVIDEND COVENANT RESTRICTIONS
2.
Covenant
restrictions
in
debt contracts are designed
reduce the potential
to
conflict of
interest
between bondholders and stockholders (see Jensen and Meckling (1976), Myers
(1977),
Smith
and Warner (1979),
Kalay
(1982),
While
and Leftwich (1983)).
debt
contracts include a variety of accounting-based restrictions on the firm, dividend covenants are
the
most frequent.
Duke and Hunt (1989) examine
187 firms from Moody's
Industrial Manual,
covenant,
35% have
18% have
a stockholders' equity covenant.
and
a working capital covenant,
Since stockholders
(or
the debt covenants for a
55%
report that
28% have
of the
random sample
of
sample have a dividend
a debt-equity ratio covenant, and
managers appointed by them)
control
the
firm,
they have
incentives to pursue dividend policies which transfer wealth from bondholders to stockholders.
For example, shareholders have an incentive
thereby reducing the collateral available
is
particularly severe
when
the firm
of debt reflects the probability of
is
to
to deplete existing
bondholders'
in
performing poorly.
the event of default.
In
such opportunistic behavior.
reduce the resulting cost by precommitting
high frequency of dividend covenants
in
assets by paying out dividends,
This incentive
rational debt markets, the price
Managers and stockholders can
to limit the level of future dividend
payments.
The
debt contracts suggests that bondholders consider this
conflict to
be important, and believe
bondholders' interests without
As Kalay (1982)
this
that the firms'
managers
not cut dividends to protect
will
covenant.
reports, dividend covenants typically define an
available for dividends (inventory of payable funds or IPF) over the
definition of the inventory for period
t
(Kalay (1982), and Watts and
t
IPFt =
kl
issues
in
period
The formula
t,
Dx
for
is
Inc. in
t,
k
S,
-
x=0
is
ID
t
a constant (0<k<1),
in
Sx
is
is
:
the proceeds from stock
period x, and F
is
a fixed number.
depends on accumulated retained
issued since the date of the debt issue (t=0). 1
following description of the dividend restriction faced by Gulf
a 1966 public debt issue
(1986))
x=0
implies that the inventory of payable funds
new stock
Zimmerman
+ F
dividends and share repurchases
earnings and the amount of
The
period
X
The usual
of the debt.
t-1
t
ET +
x=0
where, E t are earnings
life
inventory of funds
illustrates the
& Western
Industries,
above formula:
(The) company may not pay cash dividends or acquire capital stock in excess of
consolidated net income after December 31, 1965; plus net proceeds from the
sale of capital stock after such date, up to an amount not exceeding the amount
capital shares; plus
expended after such date to purchase or redeem
$7,000,000.
(Moody's Industrial Manual, 1985,
Moody's also reported
that at July 31, 1984, Gulf
p.
436).
& Western had $329
million of unrestricted
retained earnings (or inventory of funds payable) available for dividends under the firm's most
restrictive
covenant.
In
1984, the company paid $65.5 million
in
cash dividends and used
requiring the
Dividend payments are also constrained indirectly through covenants
paper
focusses
only
This
worth.
maintenance of a minimum level of working capital or a minimum net
on direct dividend covenants.
1
$170.4
million
for
share repurchases.
Debt covenants
principles
typically
use earnings computed under generally accepted accounting
GAAP
(GAAP) (see Smith and Warner (1979) and Leftwich (1983)).
variety of accounting policies to record the
private lending agreements,
same economic
make adjustments
to
However, debt covenants generally do not constrain
GAAP
firms'
event.
to
Some
permits a
covenants, especially
in
reduce management's discretion.
choices of accounting policies within
GAAP.2
Watts and
Zimmerman
(1986) hypothesize that
when
a firm
is
close to exhausting
its
inventory of payable funds, managers have an incentive to exploit the accounting discretion
available under
GAAP
dividend payments.
and make income-increasing accounting decisions
A few
and Daley and Vigeland (1983) report
that
firms
Bowen, Noreen and Lacey
that
capitalize
research and development expenses have lower inventories of funds available
firms that
avoid reducing
studies test this hypothesis by examining the relation between
dividend covenant restrictions and firms' accounting choices.
(1981),
to
expense these items.
In
for
interest,
dividends than
contrast, Holthausen (1981) finds no relation
inventory of payable funds and depreciation accounting changes.
A number
and
between the
of other studies use
the debt/equity ratio as a proxy for tightness of covenant restrictions and conclude that the
probability of a firm selecting income-increasing accounting procedures
2
Leftwich (1983) argues that
accounting policies.
it
is
optimal to allow
managers some
is
positively related to
flexibility
in
the choice of
closeness
to
covenant constraints. 3
The above evidence suggests
be an
effective
to induce
means
managers
rather than to
debt,
covenants based on accounting earnings may not
of protecting bondholders' interests. 4
The
intent of dividend
change accounting
policies.
This paper provides descriptive evidence on the
to their
The
dividend constraint.
findings are useful
assessing whether accounting-based dividend covenants are effective means
3.
is
accounting policy changes, dividend reductions, and debt renegotiations
a sample of firms that are close
restrict
covenants
of firms that are close to their constraint to cut dividends or renegotiate the
relative frequencies of
for
that dividend
managers from using dividend
for
bondholders
in
to
policy to transfer wealth to stockholders.
SAMPLE SELECTION AND DATA
The sample comprises
dividend constraint during
(unrestricted
retained
reported
in firms'
base,
used
is
ratio of
in
firms that experience a sharp increase
the
earnings)
1981
period
under the most
financial statement footnotes,
the
sample
funds available
for
to
the tightness of their
The inventory
restrictive
of
in
payable funds
dividend covenant, which
is
and on the Standard and Poor's Compustat data
Tightness of the dividend constraint
selection.
dividends
1985.
in
a given year to dividends
in
is
the previous year.
available for dividends are defined as unrestricted retained earnings at year
and preferred cash dividends paid and stock repurchases during the
estimated by the
year.
Funds
end plus common
These funds represent
See, for example, Deakin (1979), Dhaliwal (1980), Bowen, Noreen and Lacey (1981),
Zmijewski and Hagerman (1981), Dhaliwal, Salamon and Smith (1982), Lilian and Pastena (1982), and
Daley and Vigeland (1983).
3
recent study of accounting changes by Lilien, Mellman, and Pastena (1988) concludes that
despite increased regulation of accounting choices by the FASB and SEC, managers are able to modify
4
A
reported income through accounting changes.
the
maximum
violating
dividends and stock repurchases that a firm can pay during the year without
new
dividend constraint or issuing
its
previous year's
common and
preferred cash dividends
years that a firm can maintain
future earnings or
common
defined to be close to
its
its
stock issues.
first
files.
years of unrestricted retained earnings available
For the
remaining firms
retained earnings for 1981 to 1985.
A
unrestricted retained earnings
below two
became
all
falls
in
one
firm
of the five
are identified from this process.
dividend covenant
is
for
we
dividends
included
is
in
in
one
was
in
0).
first
the
of near-violations for
for
cutoff of
tests, with the costs of
two years
is
Data
of
of years of
The sample thus
in
breaching
to
years
-1, 0, 1,
these firms by year
is
collection costs are significant
in
this
1980, but
thirty-three firms
for
is
its
and
which
relatively
somewhat
selected to tradeoff sample size, and therefore the
data collection.
number
sample comprises the 126 firms
Selecting firms with less than two years of funds available for dividends
A
the
number
comes close
Annual reports
reports are available 6
The frequency
if
One hundred and
which a firm
final
1979 from
in
not close to binding
The
5
is
1980 and discard firms with
of these five years.
subsequent years.
The year
in
we compute
the sample
2 are collected for the sample firms.
arbitrary.
of
monitor the number of years of unrestricted
defined as the event year (year
.
less than two, a firm
is
pay cash dividends
For these firms
firms for which the dividend constraint
close to binding
of years
identify all firms that
the 1986 Compustat Industrial and Research
comprises
number
the
dividend constraint. 5
To select the sample, we
less than two years.
is
(and repurchases) without additional
number
this
If
of this variable to the
ratio
and stock repurchases
of dividends
level
The
equity.
power
of the
study since data
are largely hand-collected through a detailed reading of companies' annual reports.
6
in these companies financial statements discuss the dividend and other
covenants facing the firm.
The most common restriction discussed is the dividend
Working capital
constraint, which defines unrestricted retained earnings available for dividends.
covenants, which restrict cash available for dividend payments, are also discussed but there is no
restrictive
The debt footnotes
20%
evenly distributed across the sample period:
in
1984 and
20%
Table
earnings available for dividends
medians. 7
In
30%
1981,
16%
1982,
in
in
1983,
14%
1985.
in
presents descriptive
1
in
years -5
in
years -5
this variable is
to
on the number of years of unrestricted retained
statistics
to
2 for the sample firms and their industry
defined as unrestricted retained earnings at year end
plus cash dividends
and stock repurchases, divided by cash dividends and stock repurchases
the prior year.
years
In
and 2 the number
1
of
years
for
defined as unrestricted retained
is
earnings at year end plus cash dividends and stock repurchases, divided by cash dividends and
stock repurchases for year -1.8
For years -5 to
the
-1,
sample firms have on average seven
to
nine
unrestricted retained earnings available for dividends, similar to their industries.
number
of years for the
firms' industries
do
sample
not experience a decline
two subsequent years, there
The sample
dividends.
indication for our
the footnote
firms falls sharply to 1.2
is
sample
0.
In
no further deterioration
in
sample
sudden increase
in
firms'
in
the annual reports,
we
verify the value
sample
year
In
the
funds available
for
in
0.
the tightness of their dividend
firms that they consider violation of these constraints to
information
The mean
contrast, the
funds available for dividends
in
firms thus experience a
year
in
years of
of
be imminent.
Using
unrestricted retained earnings
reported by Compustat.
7
Industry medians are calculated for
dividends and are
the
same
four-digit
all
Compustat
Industrial
constraint.
Defined
and Research firms
that
pay
Standard Industry Code as the sample firms.
this way, the variable measures the number
absent
dividends
a change in dividend policy subsequent
8
for
in
of
to
years of retained earnings available
the near-violation of the dividend
constraint, which
is
The sample
dramatic decline
bolster income.
in
unmatched by
their industry counterparts. 9
firms can take a
number
their dividend-paying
Alternatively, they
we read
the
management's
reports of each
sample
We
and
(iii)
responses
firms'
report, the financial statements,
firm
in
years
-1
to
-2;
2.
(ii)
and
to
to the dividend constraint,
and footnotes from the annual
From these sources we
changes
in
collect data on:
accounting procedures
dividend covenant waivers, debt retirements or restructurings
Compustat
also collect the following data from
following, the
dividend payments, and/or renegotiate the debt
To analyze
accounting procedures used during year
to 2;
anticipation of,
in
They can make accounting decisions
abilities.
may reduce
has the binding dividend constraint.
that
actions
for
years -5
to 2:
(i)
in
in
years
years
-1
(i)
-1
to 2.
earnings, operating
cash flows, non-working capital accruals (depreciation and deferred taxes), and working
capital accruals
assets;
(iv)
(ii)
(changes
common
in
receivables, payables, and inventory) as a percentages of sales
dividends per share;
(iii)
and
earnings and operating cash flows per share; and
beginning of year stock price.
To
1980
to
we
construct a population benchmark,
1986, and accounting policies
Techniques (hereafter ATT).
ATT
is
for
on accounting method changes from
collect data
1979
to
1983 from Accounting Trends and
published annually by the American Institute of Certified
Public Accountants from a survey of the financial reports of 600 major public corporations.
The
the
firms surveyed by
ATT
are typically larger than those
sample firms have annual revenues
We
of less than
in
$500
our sample.
million
Sixty-four percent of
compared
to
31%
for
ATT.
examine unrestricted retained earnings for the sample firms in years -5 to -2 to
determine whether the dividend constraint had been binding prior to the sample period. None of the
Two firms in year -5, one
firms had less than one year of funds available for dividends in these years.
firm in year -4, and no firms in year -3 and -2 had less than two years of funds available for dividends.
9
also
Thus, the dividend constraint was not binding
in
the few years prior to the test period.
9
12%
Further, only
ATT.
of our
sample have revenues greater than $2
Earlier studies find that there
a
summary
of these studies).
therefore biases our tests
accounting decisions relative
in
The
size difference
ATT
for
between our sample and the ATT population
make income-increasing
firms.
ACCOUNTING DECISIONS AND DIVIDEND COVENANT RESTRICTIONS
4.
In this
increase
section
we examine whether
degrees
of
freedom available
dividend constraint.
to
respond
methods
to the
in
Next,
we
estimates,
an
firms already selected
all
they
response
in
may have no
increased tightness of the constraint. Therefore,
we examine changes
also investigate accruals
in
further
in
to the
increased tightness of the
accounting procedures
years
in
-1
in
to 2.
accrual
these years.
Accounting Procedure Choices
In
table 2
we
tabulate the frequency of accounting procedure choices for the test firms
The accounting procedures examined
-2.
extent of LIFO
amortization
use,
investment tax credit method,
period for pension
prior
10
some
in
are: depreciation methods, inventory pricing,
period service
amortization
costs,
assumptions. 10 Where available, we also report the frequency
since
If
prior years,
to
decisions
since managers can also influence reported earnings through changes
Finally,
year
make accounting
analyze accounting procedures selected two years prior
first
4.1
firms
the tightness of their dividend covenant constraint.
in
available income-increasing accounting
we
36%
to
and size (see Watts and Zimmerman (1986)
favor of finding that the test firms
to the
compared
a negative relation between the likelihood that a firm
is
selects income-increasing accounting procedures,
for
billion,
period
and pension
of choices
for
intangibles,
rate
made by
the
of
ATT
return
firms.
The frequencies for depreciation, inventory, and investment tax credit do not add to 100%
more than one method. For example, some firms use both LIFO and FIFO, and are
firms use
therefore included
in
both categories.
1
A
two-tailed Binomial test
The
likely to
is
used
to
test results, reported in
compare the frequencies
Table
2, provide
for
these samples.
some evidence
use income-increasing accounting procedures than the
that the test firms are
ATT
firms
most firms use Straight-Line depreciation, where there are multiple methods
more
firms are
than the
firms
ATT
likely to
firms.
use the Units
of test firms
of
These
over 40 years.
earlier, the test firms are
use, the test
in
use Accelerated methods
less likely to
LIFO and Average Cost than the ATT
use the Flow-Through method
their intangibles
and
While
-2.
where multiple inventory accounting methods are used, the
Similarly,
have lower frequencies
of Production,
year
in
more
of Investment
results should
A
firms.
test
higher proportion
Tax Credit accounting, and amortize
be interpreted with caution.
on average smaller than the ATT
firms.
differences could be size-related rather than due to the differences
As noted
Therefore, the documented
in
nearness
to the dividend
constraint.
There appear
accounting methods
to
to
be several areas where the sample firms have opportunities
increase earnings.
Average Cost, and may be able
maximum
many
of the test firms
increase earnings by changing to FIFO.
to
firms could increase the period
For example,
used
to
Also, a
entail higher tax
likely to
change
use LIFO or
number
of test
amortize pension prior period service costs to the
Both inventory and
allowed, or increase the pension rate of return assumptions.
pension changes are
to
increase reported income.
However, the inventory changes may
payments, whereas the pension changes can reduce cash outlays by reducing
funding requirements.
In addition,
earnings through changes
in
some
of the
sample firms may be able
to
depreciation and goodwill amortization methods.
increase reported
1
1
4.2
Changes
To
test
in
Accounting Procedures
whether firms that are close
to
their dividend constraint are
change accounting procedures, we examine the frequency
firms during years -1 to 2.
more
to
likely
changes by the sample
of reported
Voluntary changes examined include changes
in
pension accounting
assumptions and/or cost methods, LIFO adoptions or extensions, other inventory changes,
depreciation
changes.
method changes, changes
We
in
depreciable
and Investment Tax Credit method
lives,
also examine mandatory accounting changes since firms have discretion over the
timing of their adoption.
The mandatory changes
(compensated absences), SFAS 52
(pension accounting).
relevant to the sample period include
To provide a population benchmark, the frequency
mandatory accounting changes during these years are also reported
tailed Binomial test is
SFAS 87 and 88
currency translation), and
(foreign
used
to
significantly from that for the
for the
assess whether the frequency of changes
ATT
SFAS 43
of voluntary
and
A
two-
ATT
firms. 11
for the test firms differs
population.
Positive accounting theory predicts that, given an increase
in
the tightness of the
dividend constraint and the potential cost that imposes on stockholders, managers of the test
firms are likely to switch accounting
methods
frequency of voluntary accounting changes
for the test firms
likely to differ
The
than
for the
ATT
firms.
in
The
to
avoid being forced to cut dividends.
years
-1
to 2 is therefore
expected
timing of mandatory accounting
to
The
be higher
changes
is
also
across these samples.
results of the analysis are reported
in
Table
3.
With the exception of changes
in
pension assumptions and/or cost methods, the frequency of voluntary accounting changes by the
" ATT
frequencies are computed for each sample firm
the corresponding calendar years.
Table 2 reports the means
in
years
-1
to
2 using survey data from
of these frequencies.
12
test
firms
is
very small, and not different from the
proportion of the sample and
However, the
2.
test
in
firms
change
frequencies
firms'
significantly larger than the
methods are reported
ATT
ATT
proportion of test firms and the
ATT
indicates that the test firms do not
3.
pension assumptions
The
results of adoptions of
There
is
differently
in
a
2.13
significant
each of years
-1
to
(37%) are both
1
mandated accounting
no significant difference between the
firms adopting these
behave
1
(33%), and year
year
in
frequencies.
Panel B of Table
their
ATT benchmark.
methods
from the
ATT
in
the sample years.
This
firms regarding the timing
of these adoptions.
While the above results indicate that the change
in
general not accompanied by changes
firms
making
significant
changes
in
their
in
in
closeness
by $383,000) and adopted FIFO
in
dividend constraint
is
accounting methods, there are individual cases of
accounting policies during the sample period.
example, Raymark Corporation adopted Straight-Line depreciation
income by $229,000), changed pension
to the
actuarial
assumptions
in
in
year
year
-1
For
(increasing net
(increasing net income
year 2 (increasing earnings by $5,951,000).
While
it
is
possible that the above accounting changes (perhaps combined with less visible accrual
policies)
were motivated by the increased
probability
of contract breach,
Raymark
is
an
Since a LIFO adoption or extension change typically decreases reported earnings, we would
expect the frequency of changes to be lower for the test firms than the ATT sample. We are unable to
reject the hypothesis that there is a difference in the frequency of these changes for the two samples.
12
13
Table 2 indicates that ATT firms are more likely to have chosen Accelerated depreciation
and the Deferral ITC method at the beginning of year -1. Th6y therefore have more opportunities than
We repeat the tests in
the test firms to switch to income-increasing alternatives in subsequent years.
switched to Straighthave
Table 3 after adjusting for these differences. Of the test firms that could
-1
the
ATT firms that could
In comparison, 17% of
Line depreciation, only 12% switch in years
to 2.
have switched did so in these years. For the ITC, 10% of the test firms and 22% of the ATT firms that
could have switched to the Flow Through method did so in years -1 to 2. There is no evidence that,
after
controlling
for
higher frequency of
accounting flexibility for the two groups, the test firms have a
income-increasing accounting changes than the ATT firms.
differences
in
13
case among the 126 sample
isolated
Further Evidence on Pension Accounting Changes
4.3
The evidence
constraint are
we
section
more
in
section 4.2
likely to
of different
income-increasing change
changes
0,
17%
assumptions.
Changes
in
in
is
number
1,
used
and 18%
in
to calculate the
year
2.
the
Table 4 provides descriptive
In
year
-1
statistics
The most frequent
9%
sample
of the
The proportion
pension expense.
making these changes ranges from 4
of the firms
economic significance
on reported earnings
statistics
for
The
62%
in
on the income
pension methods, and
earnings. 14
of
The second most frequent change
each
of the pension
firm that
of these firms report that the accounting
summary
A
In this
to
is
27%
is in
actuarial
10%
per year.
the amortization period for prior service costs are rare.
(28%, 60%, 50%, and
1.
Panel
the rate of return assumption.
in
The percentage
To assess
their effect
year
further.
ATT sample.
the
types of pension changes for the sample firms.
firms increase the rate of return
year
indicates that firms that are close to their dividend
change pension accounting methods than
investigate these
on the frequency
in
firms.
for
years
-1
effect of
to
2
changes, we
collect information
announced an accounting change.
change has no material
respectively).
pension changes
for
effect
A
on
large
on earnings
Panel B of Table 4 reports
all
firms that
change
their
those that disclose that the pension changes have a material effect on
effect of the
changes on
after-tax earnings are deflated by
dividends
in
year
-
This ratio measures the effect of the accounting change on the number of years of
unrestricted retained earnings available for dividends.
The income effect of pension changes is assumed to be zero for firms that report that the
immaterial. The average effect reported in Table 3 for all pension changing firms is therefore
14
effect
is
likely to
be understated.
4
1
For
all
pension accounting change firms the median earnings effect
dividends, and zero
occurs
year
in
0.
In
values suggest that
the remaining years the
effect for a
few
(920%).
1
on earnings
is
mean
mean ranges from
8%
-1
earnings effect
to
is
is
6%
of
34% and
18%. The small median
evidence that the accounting changes have a significant
The mean earnings
effect
effect for firms that disclose that the
83%
of dividends in
year
pension changes
There are
0.
five firms for
These are PSA
Inc.
Corp. (72%), Gates Learjet Corp. (75%), Portec Inc. (158%), and Vertipile
Inc.
which the earnings
GAF
is
largest
year
pension changes do not have a significant effect on
of these firms,
However, there
firms.
have a material
(55%),
most
for
dividend restrictions.
The
the three subsequent years.
in
in
effect is greater than
The mean
5
effect in the other years
The above pension
significant effect
in
year
of
-1
ranges from
results indicate that:
frequency of pension accounting changes
change has a
50%
the
(i)
years
-1
dividends.
12%
to
35%.
sample firms have a
to 2;
and
(ii)
for a
relatively high
few firms the pension
on the tightness of the dividend constraint.
This evidence
is
consistent with the hypothesis that sample firms' pension accounting changes are motivated by
dividend covenant considerations.
However, there
is
A
a competing explanation.
reduction
in
pension expense, from changing pension assumptions, not only increases reported earnings but
also decreases cash outflows.
the pension changes
may
If
the
sample
firms experience declines
also be motivated by cash
in
operating
management concerns,
cash flows,
rather than the
tightness of the dividend covenant per se. 16
15
and one
Of these
firm
firms,
maintained
its
two omitted
common
dividends
in
year
0,
two decreased
their dividends,
dividend.
Evidence presented later in the paper shows that the sample firms have lower operating
than
and 1.
cash flows
their industries in years
16
1
To examine
closeness
the relation
between the earnings
pension accounting changes and the
effect of
to the dividend constraint after controlling for
5
changes
operating cash flows,
in
we
estimate the following regression:
PEFF = B
where, PEFFj
t
the
is
RiACF
+
jt
change
in
jt
+
f3
2
DR
jt
+
C jt
(1)
net income per share (cash flow per share) from the pension
accounting changes as a percent of the stock price
change
in
beginning of the year;
at the
is
the
operating cash flows per share excluding the effect of the pension change also as a
percent of the stock price at the beginning of the year; DRjt
unrestricted retained earnings available for dividends; and
respectively. 17
will
ACFjt
If
j
and
is
t
the
number
of years of
represent firms and years
pension accounting changes are motivated by cash management concerns, Bi
be negative, whereas
if
they are motivated by dividend covenant considerations, B2 wi " De
negative.
Equation
(1) is
estimated for firms that
the year of the change.
cash flow variable (BO
less than the
5%
level.
The
is
results are presented
-0.025.
The
make
Although
this
a pension accounting change using data
in
Table
value
is
5.
The estimated
relatively small,
coefficient for the dividend restriction (B2)
is
in
coefficient for the
it
is
significant at
insignificant at the
17
Operating cash flows before the effect of pension changes are Net Income + Depreciation
Deferred Income Taxes - Changes in Inventories and Receivables + Changes in Payables - After-tax
Change in Pension Expense due to the Accounting Change.
1
10%
level. 18
management
We
expand
the
These
results
suggest that the pension accounting changes are motivated by cash
considerations, rather than the dividend constraint.
also estimate a
sample
using data for years
number
to include
-1
to 2.
all
one
and zero otherwise.
estimated coefficient
Third,
The
for the deflated
is
in
in
cash and marketable securities as
of years of funds available for dividends
cash flow (R^
is
is
The
negative and significant, and
insignificant.
firms
make pension accounting changes
tightness of the dividend constraint,
dividend constraint
4.4
pension-changing firms,
results are similar to those reported in Table 5.
'S
we
thus no evidence that the dividend constraint influences the pension changes.
Even though the sample
increase
level of
First,
reconstruct the dividend restriction variable as a
number
change
the dividend restriction coefficient (R2)
There
we
the
if
specifications of equation (1).
firms, rather than just the
Second, we include the
variable, taking the value
less than two,
of alternative
sample
an additional independent variable.
dummy
6
it
is
in
years surrounding the
not possible to attribute these to the
itself.
Accrual Management Decisions
Changing accounting methods
is
but
one means
the probability of violating a dividend constraint.
accelerate revenues or defer expenses.
of
managing reported earnings
Managers can also
to
reduce
select accruals that
Accruals are less visible than method changes and are
18
Specification tests are conducted to assess whether the residuals are homoskedastic (see
White (1980)), and normally distributed.
We cannot reject the hypotheses that the residuals are
homoskedastic and normally distributed at the 0.05 level. Belsley, Kuh and Welsch (1980) diagnostics
for the effect of extreme observations on the coefficients, and for multicollinearity indicate that the
reported estimates are not influenced by extreme observations, and that the independent variables are
not collinear.
1
therefore
difficulty
more
to
manage
earnings.
Of course,
low
their
visibility
and the
specifying an accruals expectations model also increases the difficulty for the
in
researcher
be used
likely to
7
earnings management. To test whether managers use accruals to reduce the
to detect
we examine
probability of breaching dividend constraints,
accruals, earnings
and cash flows
in
years surrounding the increased tightness of the dividend constraint.
Panel A of Table 6 reports median values of operating cash flows, non-working capital
accruals, working capital accruals,
years -5
to
2J9
and earnings as percentages
Panel B presents medians of industry-adjusted
difference between ratios for the
sample firms and
experience significant declines
in
years
and
-1
capital
0,
and outperform
accruals
Earnings, the
sum
fall
sharply,
earnings
in
their industry
ratios,
sample firms
computed as
medians. 20
The sample
and non-working
capital
accruals
In
firms
in
to
2 working
relatively
constant.
years
are
-1
and accruals, therefore experience even greater declines
Hence, there
is
in
the
operating cash flow ratios relative to their industries
their industries the following year.
of cash flows
these years than cash flows.
to increase
of sales for the
in
no prima facie evidence that managers use accruals
years surrounding the increased tightness of
their
dividend constraint. 21
19 Working capital accruals are Changes in Receivables and Inventories - Changes in Payables.
Non-working capital accruals are Depreciation + Deferred Income Taxes. The results presented in
Table 6 are for accruals deflated by sales. We replicate the analysis using book values of assets rather
than sales as a deflator, and the conclusions are unchanged.
20
Industry
pay dividends
21
is
no
median
ratios are
year
and are
prior to
0,
Compustat Industrial and Research firms
the same four-digit SIC industry as the sample firms.
computed
in
for all
that
These conclusions are based on a random walk expectation model for accruals. Since there
model of accruals, this model is ad hoc. The findings should therefore be
theoretical expectation
interpreted with caution.
1
bias.
possible that the findings
is
It
managers
If
flexibility,
they
policies.
In
covenant
will
may reduce
the extreme,
the
when
in
probability
year
sample
0.
If
will
this
of accounting
covenant
this
is
some
subsample are
not important. 22
against finding
To address
violation.
which do not
more than one year but
accounting
circumvent the dividend
to
be sampled and
flexibility will
have violated
firms that
flexibility
in
this issue,
we
their dividend
an increase
to
repeat the analysis
covenant
in
year
different
in
the
sections 4.1
is,
The
firms with
results for
sample, indicating that the sample selection bias
sample selection bias
also believe that our results are not driven by
because Holthausen (1981) uses a
in
(that
less than two years of funds available for dividends).
full
covenant
that year, their inclusion in the
accounting responses
violate the dividend
similar to those for the
We
little
flexibility
paper consists of firms with less than two years of unrestricted
results
to 4.4 using only firms
covenant violation by changing accounting
changes.
these firms have no accounting
bias the
probability of a
covenants and have accounting
the sample selection requires firms to violate the dividend
retained earnings and therefore contains
in
of
Only firms that have
show no unusual number
The sample
sections 4.1 to 4.4 are influenced by a sample selection
that exercised their accounting
be excluded.
will
in
anticipate problems with their dividend
companies
restriction,
they
Sample Selection on Results
Effect of
4.5
8
sample selection procedure and arrives
at the
same
results.
We
22
and
for firms that
differences
in
subsamples.
to 2,
any of years
significant
There are no
also repeat the analysis for firms that cut dividends per share
increase or maintain their dividend rate
accounting changes
in
years
-1
to 2,
and
in
in
these years.
in
the accounting choices
in
year -2
for the
two
1
Dividend and Debt Renegotiation Responses to Dividend Constraint Violation
5.
Dividend Change Responses
5.1
If
dividend covenants are effective
policies, firms
means
experiencing sudden increases
To
forced to cut dividends.
in
for
bondholders
to regulate firms dividend
the tightness of the constraint are likely to be
test this hypothesis
we examine
the frequencies with which the
sample firms increase, decrease, omit, and pay constant common dividends per share
4
to 2.
Firms that are classified
As a benchmark, we compute mean
each sample
Compustat
the
firm,
same
in
and Research
four-digit
SIC
years
-
the omit category each year they pay zero dividends.
For
industry frequencies for these categories as follows.
the frequency of dividend changes
Industrial
in
the increase, decrease, and constant dividend categories pay
in
Firms are included
non-zero dividends.
in
9
firms that
industry.
is
computed
in
years -4 to 2
for all
pay dividends (preferred and/or common) and are
Cross-sectional
means
of
these frequencies are then
computed.
Table 7 reports the frequency of changes
firms
and mean frequencies
prior to the
for their industries in
per share
years -4
common
to 2.
dividends
each
In
sample
for the
of the four years
increased tightness of the dividend constraint a majority of the sample firms as
well as their industries increase
of dividend
in
changes
this year, the
for the
common
dividends per share.
However,
the frequency of dividend decreases or omissions goes up from
dramatic: increases
fall
in
the frequency of dividend changes
from
61%
This pattern of dividend changes
least two further years.
The
to
year
the pattern
firms diverges markedly from their historical pattern.
sample
incidence of dividend increases for the sample firms
an industry-wide change
in
15%
in this
falls
57%
to
20%, and
56%. While there
is
also
year, the magnitude
is
less
to
49%, and decreases and omissions
for
from
In
rise
from
13%
to
23%.
both the sample firms and their industries persists for at
results therefore indicate that for
many
firms the
change
in
20
closeness
to the
dividend constraint
The decreases
sample
in
is
associated with a major change individend policy.
firms'
dividends
tightness of their dividend constraints as illustrated
Table
prior
6, the
years
in
sample firms also experience a decline
Fama and Babiak
research (see
Figure
in
2 coincide with the increased
1.
However, as documented
reported earnings
in
in
these years. Since
in
(1968)) indicates that current and lagged earnings
declines tend to be accompanied by dividend decreases,
firms' dividends
to
it
is
possible that the changes
in
sample
are due to their poor performance rather than increased tightness of the
dividend constraint.
To examine
the relation
changes,
after controlling for earnings
we
between dividend changes and dividend
restrictions
estimate the following regressions:
*
ADIVj, = B
+ RiAEARNjt + B 2 DRj t +
ADIV
= B
+
= B
+
Jt
BijAEARN jt + B 2 DR
jt
Hjt
(
2
)
+ Hjt
(
3
)
(
4
)
5
ADIV
Jt
I
B k AEARN j>t k+1 + B 2
.
DR
+
jt
ji jt
k=1
ADIVj
t
is
the
change
in
common
dividends per share
stock price at the beginning of the year;
AEARNj
t
in
is
year
the
t
for firms
change
percent of the stock price at the beginning of the year; and DRj t
unrestricted retained earnings available for dividends
the earnings coefficients
will
be positive.
If
policy, the dividend restriction coefficient will
The
in
year
t.
in
is
j
as a percent
of the
earnings per share as a
the
number
Prior research
of years of
suggests that
dividend constraints have an effect on dividend
be
positive.
coefficients in equations (2) to (4) are estimated using data for
all
sample
firms
in
21
years -4 to 2.
Equation
assumes
(2)
constant across firms, and that
only current earnings changes are important.
assumptions are relaxed separately
allowed to vary across firms
included
equation
in
positive relation
in
equations
equation
(3)
and
(3)
and
(4).
The earnings
results
and
are presented
significant at the
dividend changes, there
1%
in
Table
8.
24
In
level, confirming
equation
1%
is
level.
(2)
changes are
restriction coefficient is
Therefore, after controlling for the effect of earnings on
a positive relation between the
relaxes the assumption that the earnings coefficients are the
coefficient
is
number
of
years of unrestricted
0.017, significant at the
1%
same across
level.
An
the
Equation
sample
F-test
earnings coefficients differ across firms, inconsistent with the assumption
this
the earnings
prior findings that there is a
retained earnings available for dividends and the magnitude of the dividend change.
However, relaxing
These two
coefficients are
four years' lagged earnings
between earnings and dividend changes. The dividend
0.1, also significant at the
mean earnings
is
(4). 23
The regression
coefficient is 0.014
in
between earnings and dividend changes
that the relation
firms.
(3)
The
indicates that the
in
equation
assumption does not change the magnitude or the significance
(2).
of the
dividend constraint coefficient.
23
We
also estimate several other models.
First,
we
estimate models with lagged earnings
not present the results here for brevity since they
changes for one, two and three prior years but do
do not change the conclusions presented in the text.
intercept to vary across firms.
Second, we estimate equation (2) allowing the
An F-test indicates that intercept terms do not vary across firms.
24
Specification tests are conducted to assess whether the residuals are homoskedastic (see
We cannot reject the hypotheses that the residuals are
White (1980)), and normally distributed.
homoskedastic and normally distributed at the 0.05 level. Belsley, Kuh and Welsch (1980) diagnostics
for the effect of extreme observations on the coefficients, and for multicollinearity indicate that the
reported estimates are not influenced by extreme observations, and that the independent variables are
not collinear.
22
We
include the
number
also estimate a
change
in
variable, taking the value
less than two,
One
equations
and
it
Equation
the
securities as
number
funds available
of years of
The conclusion
for
dividends
is
that the dividend restriction variable
is
(4)
provides evidence contrary
level.
Further,
to this.
The estimated
to current earnings
However, the introduction
not alter the magnitude of the coefficient
1%
in
captures the effect of lagged earnings changes which are omitted
three prior years. 25
significant at the
and marketable
reconstruct the dividend restriction variable as a
changes are related not only
indicate that dividend
in
if
the level of cash
we
First,
the significant coefficient on the dividend restriction variable
for
(3) is that
from these models.
changes
one
specifications of equation (2).
unchanged.
is
explanation
(2)
we
Second,
and zero otherwise.
highly significant
and
operating cash flows,
additional independent variables.
dummy
of alternative
of the
on the dividend
coefficients
changes, but also
to
lagged earnings variables does
restriction variable,
which
is
still
since dividend changes are not related to earnings changes
four years prior to the current year,
it
unlikely that the dividend
is
restriction
variable
is
proxying for lagged earnings from earlier years.
The above
when
results indicate that a large
their dividend
covenants become more
attributable to earnings declines.
significant relation
dividend cuts.
25
have
However,
between the tightness
number
of
sample firms reduce
restrictive.
their
dividends
These dividend cuts are
in
part
after controlling for the earnings' effect, there is
of the dividend constraint
and the magnitude
Therefore, accounting-based dividend covenants appear to be effective
Equation
(4)
does not allow the earnings coefficients
sufficient time-series observations per firm.
to
vary across firms since
we do
a
of
in
not
23
restricting
firms'
dividend policies. 26
Debt Renegotiation Responses
5.2
In addition to
reducing dividends, a number of the sample firms restructure the debt that
contains the binding restriction.
the covenant, retiring the debt,
The debt
amending the lending agreement, and/or issuing new equity
increase unrestricted retained earnings.
9. 27
In
of the
years
sample
-1
in
and
The frequencies
the most frequent response
year-1
and 6.5%
in
year
9%
the firms retire the debt.
and 13%
One
of the
these responses are shown
to obtain
to
is
in
in
years
firms reports that
to
Table
a waiver of the covenant (3.3%
In
the two
or restructure the debt.
retire
sample firms
sample
of the
is
of
usually for preferred dividends.
0),
subsequent years, the most frequent response
restructurings occur for
responses include obtaining a waiver of
restructuring
it
and
1
2,
and
issued equity
5%
to
and
Debt
7%
of
increase the
funds available for dividend payments.
6.
Summary and Discussion
Dividend covenant restrictions
conflicts of interest
Some
debt contracts are designed to reduce potential
between bondholders and stockholders.
funds available for dividend payments
26
in
in
These
restrictions typically define
terms of accounting earnings defined under
conference participants argued
that
managers
of the
sample firms chose
GAAP.
to
cut
dividends rather than change accounting policies because dividend cuts are costless to the shareholders.
However, a number
of earlier studies
show
that dividend cuts are associated with significant wealth
losses for shareholders (see Dielman and Oppenheimer (1984), and Healy and Palepu (1988)). These
wealth losses occur for our sample also. Firms that omit dividends have a -7% two day abnormal stock
return at the omission announcement; dividend decreases (excluding omissions) are accompanied by a
-3% two day abnormal announcement return.
bondholders believe that managers are reluctant
Presumably dividend covenants exist because
reduce dividends given these wealth
to voluntarily
losses to shareholders.
27
Frequencies of liability restructurings are
Techniques and are therefore not shown in Table 9.
not
reported
by
Accounting
Trends and
24
Since there
is
considerable accounting discretion available under
potentially circumvent
dividend covenant restrictions, and avoid dividend cuts by increasing
This paper provides evidence on this hypothesis by
earnings through accounting decisions.
examining accounting and dividend policy changes
of violating the dividend
The
GAAP, managers can
covenant
in
response
to
an increase
in
the probability
restriction.
results indicate that there are
no significant changes
surrounding the increased tightness of the dividend constraint.
accounting methods
in firms'
While the sample firms adopt
income-increasing pension accounting assumptions during the four years surrounding the nearviolation
the
of
dividend constraint, they appear to be
considerations, rather than the increase
the year of the near-violation there
and omissions,
reductions
sample
is
is
tightness of the dividend constraint.
a substantial increase
persisting for several
related to the
in
subsequent years.
degree of tightness
management
motivated by cash
in
In
contrast,
in
the frequency of dividend cuts
The magnitude
of the dividend constraint.
A
of the dividend
small
number
firms also restructure the debt that contains the binding dividend restriction.
therefore conclude that accounting decisions play only a limited role
in
sample
of
We
firms' overall
responses.
The
high frequency of dividend covenants
in
debt contracts suggests that bondholders
consider this conflict to be important, and believe that the firms' managers
dividends to protect bondholders' interests without
this
questions about the effectiveness of these covenants given
managers.
effective
The
means
covenant.
Earlier
will
not cut
studies
raise
the accounting discretion available to
findings of this study suggest that earnings-based dividend covenants are an
for
bondholders
to constrain firms'
dividend policies.
25
A number
of
bond covenant constraints other than
accounting variables.
affirmative covenants.
These can be
the dividend restriction also use
two categories: negative covenants, and
classified into
Negative covenants, such as the dividend
restriction
examined
in this
paper, prevent managers from taking actions which transfer wealth from bondholders to
stockholders.
Affirmative covenants, such as working capital, interest coverage,
covenants, are designed
to
and net worth
increase the security of bondholders. Negative covenants violations
can be avoided by taking actions which are managers
control.
For example, violation of the
dividend covenant can be avoided by cutting dividends, even though
it
is
costly for shareholders.
In
contrast, avoiding violations of affirmative covenants usually requires the firm to improve
its
operating performance, which
may
therefore be
examined
in this
more
paper.
likely
This
is
to
is
not completely under managers' control.
trigger accounting
an interesting topic
These constraints
responses than the dividend covenants
for future research.
26
References
Belsley, D.A., E. Kuh,
Bowen,
R., E.
and R.E. Welsch, "Regression Diagnostics," (Wiley,
New
York, NY).
Noreen, and J. Lacey, "Determinants of the Corporate Decision to Capitalize
Journal of Accounting and Economics 3 (August 1981), pp. 151-179.
Interest,"
Daley, L.A., and R.L. Vigeland, "The Effects of Debt Covenants and Political Costs on the Choice of
Accounting Methods: The Case of Accounting
Economics 5 (December 1983),
pp.
for
R&D
Costs," Journal of Accounting
and
195-211.
Deakin, E.B., "An Analysis of Differences Between Non-Major Oil Firms Using Successful
Efforts and Full Cost Methods," Accounting Review 54 (October 1979), pp. 722-734.
Dhaliwal, D., "The Effect of the Firm's Capital Structure on the Choice of Accounting Methods,"
Accounting Review 55 (January 1980), pp. 78-84.
and E. Smith, "The Effect of Owner Versus Management Control on the
Choice of Accounting Methods," Journal of Accounting and Economics 4, (July 1982),
pp. 41-53.
Dhaliwal, D., G. Salamon,
Dielman
T.E., and H. R. Oppenheimer, "An Examination of Investor Behavior During Periods of
Large Dividend Changes," Journal of Financial and Quantitative Analysis, 1984, pp.1 97-
216.
Duke,
Fama,
C, and
"An Empirical Examination of Debt Covenant Restrictions and
Accounting-Related Debt Proxies," Journal of Accounting and Economics, (forthcoming).
J.
E. F.
H. G. Hunt
III,
and H. Babiak, "Dividend
Statistical
Policy:
An
Empirical Analysis,"
Journal of the American
Association 63, (1968), pp. 1132-1161.
Healy, P.M. and K.G. Palepu, "Earnings Information Conveyed by Dividend Initiations and
Omissions," Journal of Financial Economics, 21 (1988), pp.1 49-1 75.
Holthausen, R.W., "Evidence on the Effect of Bond Covenants and Management Compensation
Contracts on the Choice of Accounting Techniques: The case of the Depreciation Switch
Back," Journal of Accounting and Economics 3 (March 1981), pp. 73-1 09.
Jensen, M.C., and W.H. Meckling, "Theory of the Firm: Managerial Behavior, Agency Costs and
Ownership Structure," Journal of Financial Economics 3 (October 1976) pp. 305-360.
Kalay, A., "Stockholder-Bondholder Conflict and Dividend Constraints," Journal of Financial
Economics 10
(July
1982), pp. 211-233.
"Accounting Information in Private Markets: Evidence from Private Lending
Agreements," Accounting Review 58 (January 1983) pp. 23-42.
Leftwich,
R.,
27
M. Mellman, and V. Pastena, "Accounting Changes: Successful Versus Unsuccessful
Firms," Accounting Review 63, (October 1988), pp. 642-656.
Lilien, S.,
and V. Pastena, "Determinants of Intra-method Choice in the Oil and Gas Industry,"
Journal of Accounting and Economics 4 (December 1982), pp. 145-170.
Myers, S.C., "Determinants of Corporate Borrowing," Journal of Financial Economics 5
(November 1977) pp. 147-175.
Lilien, S.,
Smith, C.W., and J.B. Warner, "On Financial Contracting: An Analysis of
Journal of Financial Economics 7 (June 1979) pp. 117-161.
Bond Covenants,"
Watts, Ross and Jerold Zimmerman, "Positive Accounting Theory", Prentice Hall, Englewood
Cliffs, NJ, 1986.
White, Halbert, "A Heteroskedasticity-consistent covariance matrix estimator and a direct test
for heteroskedasticty, Econometrica 48, pp. 817-838.
3
8
2
3
Table 2
Frequency
of
accounting procedures used by firms two years prior
agreement dividend constraint 8
to near-violation of their lending
Accounting
Frequency
of accounting
procedures used by
methods'1
Acctg. Trends &
Techniques sample
Test sample
Depreciation Method
Straight-line
93.7%
93.6%
Accelerated method
18.
C
25.9
Units of production
15. 9«
9.1
0.0
0.4
Other
Inventory pricing
58. 7C
LIFO
FIFO
Average cost
Other
Use
66.3
62.9
57.9
21.4C
39.8
7.9
9.1
of LIFO
All
inventories
50%
more
or
of inventories
Less than 50% of inventories
Not determinable
16.
C
6.5
37.
C
51.1
24.4
21.6
23.2
19.0
94.4 C
7.9
0.0
88.8
10.2
40.5
37.4
2.7
Investment Tax Credit
Flow-through
Deferral
No reference to Investment Tax Credit
Amortization of Intangibles
40 years
10-30
years
1.1
6.3
Other
10.
C
5.2
Not reported or no intangibles
42.9 C
54.7
Amortization of Prior Period Service Cost
40 years
13.5
12.7
20.6
17.5
4.8
30.9
30-40 years
30 years
10-30 years
Other
No pension plan or
prior period service cost reported
Pension Rate of Return
6%
13.5
or less
6.1%
7.1%
8.1%
-
7%
8%
19.1
25.4
13.5
3.2
25.3
more
or
Other
Not reported or no pension plan
8
A
firm is
defined as close to violating
retained earnings.
The
lest
its
dividend constraint
sample comprises 126
If
firms that are
less than two years of dividends are available In unrestricted
newly classified as close to the constraint between 1981 and
1985.
b
The frequencies
one method.
c
for depreciation. Inventory,
and investment
Significantly different from the frequency for Accounting
Binomial
lest.
tax credit
do not add
to
100%. since some
Trends and Techniques sample
at the
5%
firms
use more than
level using
a two-tailed
Table 3
of accounting procedure changes by year for 126 firms
close to violating their lending agreement dividend constraint.
Accounting Trends and Techniques' frequencies in parentheses 3
Frequency
Accounting
procedure,
change
Year
-
relative to
near violation of dividend constraint
1
Panel A. Voluntary Changes:
Pension assumptions
and/or cost method
LIFO adopted or extended
Other inventory changes
Depreciation method
Depreciable lives
Investment credit
14
1
2
Table 4
Summary
statistics
126 firms close
on type and magnitude
to violating their lending
Year
-
1
of
pension accounting changes
relative to
near violation of dividend constraint
1
Panel A: Frequency of types of pension accounting changes
Changes
in
rate of return
Reduction
assumptions
for
agreement dividend constraint 8
2
Table 5
Tests of the relation between earnings effects of changes in pension accounting
and the closeness to dividend constraints after controlling for changes in
operating cash flows for firms that are close to violating their dividend constraint
a
and make pension accounting changes
policies,
PEFFj
t
=
f?o
BiACFj
+
t
+ G 2 DR jt +
cj
b
t
R2
t
•
A
firm is defined
as close
to violating its dividend constraint
available in unrestricted retained earnings.
accounting changes (140 firm-years
°
PEFF,
is
(-2.0)
(2.5)
statistic
the change
in
net
in
-0.030
-0.025
0.80
Coefficient
Obsevations are
if
0.05
(-0.4)
less than two years of dividends are
for firm-years
where there are
pension
total).
income per share from the pension accounting change as a percentage
beginning of year stock price; ACFjj
is
the change
in
operating cash flow per share for firm
the effect of the pension change as a percentage of the beginning of year stock price; DRj{
years of unrestricted retained earnings available
for
dividends for firm
j
in
year
t.
is
j
in
the
of the
year
t
before
number
of
c
jD
Table 8
Tests of the relation between changes in dividends, and the closeness to
dividend constraints after controlling for changes in earnings for a sample
of 126 firms that are close to violating their dividend constraint 3
Estimate
Coefficient
Equation 2:
ADIVjj - B
+
BiAEARNj
+
t
B2
-0.600
0.014
0.100
R2
0.13
B
B,
Equation 3: ADIV; t - Bq +
BiAEARNjt
t
Mean B^
0.017 c
Median
B2
0.012
0.090
R2
0.40
B,j
2 DR jt
+ B 2 DRjt *
-0.500
B
(3
statistic
c jt
-8.1
4.5
9.2
c
b
jt
-6.9
8.0
Table 9
Frequency
of liability
restructurings by year for 126
firms close to violating their lending
Year
-
Waiver
of
covenants
1
agreement dividend constraint 3
relative to
near violation of dividend constraint
1
2
Figure
1
Number
of years of unrestricted retained earnings available for dividends and
frequency of dividend cuts and omissions, and in years surrounding nearviolation of dividend covenant for 126 firms
Panel A: Number of years of unrestricted retained earnings available
3
-
Year
2
-
o
1
relative to near-violation of dividend
constraint
Panel B: Frequency of dividend cuts and omissions
60.00% j
50.00%
40.00% 30.00%
20.00%
-•
10.00% -
..«,..
3
-
Year
2
-
1
relative to near-violation of dividend
constraint
1
for dividends
Date Due
Lib-26-67
MIT
3
LIBRARIES DUPL
1060 0075A7T5
3