Peter Joos.

Costly dividend signaling : the case of lost firms with negative cash flows online

. (page 1 of 2)
Online LibraryPeter JoosCostly dividend signaling : the case of lost firms with negative cash flows → online text (page 1 of 2)
Font size
QR-code for this ebook


3 9080 02618 4603


' .1,

ma f"

)\r'u, ii/i

i';f f ^itih f ^ I


no MM^*^'^^ DEWEY

MIT Sloan School of Management

MIT Sloan Working Paper 4474-04
January 2004

Costly Dividend Signaling:
The Case of Loss Firms with Negative Cash Flows

Peter R. Joos and George A. Plesko

© 2004 by Peter R. Joos and George A. Plesko.

All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted

without explicit permission, provided that full credit including © notice is given to the source.

This paper also can be downloaded without charge from the

Social Science Research Network Electronic Paper Collection:


DEC 3 im


Costly dividend signaling:
The case of loss firms with negative cash flows

Peter Joos and George A. Plesko

First Draft: November 2003
This Draft: January 2004


We examine the dividend-signaling hypothesis in a sample of firms for which dividend
increases are particularly costly, namely loss firms with negative cash flows. When
compared to loss firms with positive cash flows, we find the predictive power of dividend
increases for future return on assets to be greater for loss firms with negative cash flows,
consistent with the predictive power of the dividend signal being stronger when its cost is
higher. Our results provide support for the dividend-signaling hypothesis and have
broader implications since loss firms comprise a large and increasing share of publicly-
traded firms.

Contact information:

Sloan School of Management

Massachusetts Institute of Technology

E52-325, 50 Memorial Drive

Cambridge, MA 02142-1347

Joos: 617-253-9459, [email protected] (corresponding author)

Plesko: 617-253-2668. g [email protected]

We thank Paul Asquith, Joe Weber, and Peter Wysocki for helpful comments.

Whether firms signal future prospects through dividend changes has been a source
of debate and research in the corporate finance literature since the early papers by Lintner
(1956) and Miller and Modigliani (1961). Despite considerable research, the debate over
the empirical validity of the dividend-signaling hypothesis remains alive in the literature.
Nissim and Ziv (2001) present evidence consistent with the dividend-signaling
hypothesis by showing dividend increases (but not decreases) relate to fiiture
profitability. However, two recent papers come to different conclusions. Grullon et al.
(2003) argue the results in Nissim and Ziv (2001) follow from a misspecification of the
earnings expectations model used to predict expected earnings. They find the evidence
supporting the dividend-signaling hypothesis disappears when the earnings expectations
model accounts for non-linear patterns in the behavior of earnings.' In a more general
review of dividend policy. Skinner (2003) concludes structural changes in dividend
policy and the nature of corporate earnings over time rule out signaling, at least in recent
decades. He finds dividends have become too smooth and earnings too volatile for
dividend changes to be an informative signal for future earnings changes.

Although not conclusive, this recent empirical evidence appears to be moving
towards rejecting the dividend-signaling hypothesis.^ In this paper, we contribute to the
debate with a different approach to test the dividend-signaling hypothesis. Instead of
examining dividend behavior for all firms in the market, we examine the dividend-
signaling hypothesis in a setting where use of dividends to signal is particularly costly to
the firm.

Spence (1973) argues the cost of sending an economic signal determines its
informativeness, therefore we test for dividend signaling in a sample of firms that

increase their dividend payment (i.e., cash outflow) while experiencing current losses
caused by negative cash flows. ^ We argue that, since investors can readily observe the
current loss and its components, management will need to send a strong and credible
signal to convince market participants that performance will improve. We assume that
increasing cash dividends at a time the firm has a negative cash flow constitutes a strong
and costly signal of future performance for two reasons. First, the increase in current
cash dividends will immediately affect the liquidity of the firm. Second, an increase in
the cash dividend implies a strong commitment to maintain the higher level of dividends
in the future, given previous studies document a reluctance of managers to cut dividends
(e.g., Lintner 1956 and Brav et al. 2003).

We test our hypothesis by comparing the predictive power of dividend increases
between loss firms with positive and negative cash flow components for future
performance. We collect a sample of loss observations from 1970-2001 and test whether
a dividend increase provides incremental information in predicting firms' return on assets
beyond that contained in current earnings and a number of control variables.

Our main results show that, condifioning on cash flows, the signaling power of
dividend increases for loss firms exists only for negative cash flow firms, consistent with
the hypothesis that the costlier the signal is the more information it contains. We verify
our main results in additional analyses focusing on subsamples of firms with multiple
losses for which increased dividend pajmients are increasingly costly, and on augmented
specifications of our basic model. Although some results are consistent with a loss being
sufficient for a dividend increase to improve forecasts of future returns irrespective of the
sign of the cash flow, all robustness analyses demonstrate the predictive power of

dividend increases is larger for loss firms with a negative cash flow than for loss firms
with a positive cash flow.

Our study extends the dividend signaling literature by identifying a particular
segment of firms for which we hypothesize the decision to increase dividends is
particularly costly. Consistent with Spence's criterion for informative signals, we find
dividends help to predict a firm's fiiture performance when the dividend signal is costly.
We also extend previous research on the relation between losses and dividends by
focusing on the particular quality of losses that renders the dividend signal costly and
credible, namely the cash flow component of the loss.

In the next section, we discuss related research and motivate our study. In section
II we provide descriptive statistics of the sample and present our empirical model.
Sections III and IV contain our main results and the results of robustness analyses. We
conclude in a final section.

I. Background and motivation

To test the dividend-signaling hypothesis, we evaluate the predictive power of an
increase in cash dividends for future firm performance in a sample of firms that report
current losses. We argue the cost of the dividend signal will determine its
informativeness and distinguish between losses with a negative versus a positive cash
flow component to capture the differential cost of the dividend increase across loss firms.

We are not the first to investigate the relafion between dividends and losses. Like
DeAngelo et al. (1992) [hereafter, DDS], we focus on loss firms to study the dividend-
signaling hypothesis, arguing that dividends will have information content when current

earnings are an unreliable indicator of future profitability, and that losses provide such a
special context. In a sample of 167 firms over the period 1980-1985, DDS show a loss is
a necessary, but not sufficient, condition for a firm to decrease dividends. They find
firms that decrease dividends experience more severe and more persistent losses than
firms that do not. Further, unusual income items (e.g., special items) are a larger factor in
the earnings of loss firms that do not reduce dividends than of the firms that do. Focusing
explicitly on dividend signaling, they find dividend decreases provide incremental
information to predict future earnings, although their forecasting power diminishes in the
presence of unusual income items.

Using a larger sample over a long time period. Skinner (2003) finds that when
firms paying large dividends experience a loss, the loss is more likely caused by special
items, and more likely to reverse than a loss reported by a firm that does not pay large
dividends. In related work, Joos and Plesko (2003) examine a large sample of loss
firms, and the timing of loss reversals. They show the losses of firms that continue to pay
dividends are more likely to reverse than those of non-dividend paying firms, and that
eliminating a dividend is associated with a decrease in the likelihood the loss will reverse
in the immediate future.^

We extend this line of research by comparing whether increases in cash dividends
by loss firms signal future performance better when negative cash flows rather than
negative accruals drive the loss. Whereas previous research focuses specifically on the
role of special items when studying the relation between dividends and losses, we
differentiate between losses with a positive and negative cash flow component to capture
the relative cost of the decision to increase dividend cash outflows. The evidence in Joos

and Plesko (2003) showing losses have become more persistent in recent years, often due
to persistent negative cash flows, emphasizes the potential cost of an increase in cash
dividends for loss firms.

The most recent papers on the dividend-signaling hypothesis find a reduced
signaling role for dividends over time, consistent with findings that firms have changed
their dividend-paying behavior (see Skinner 2003)7 In light of the evidence of the
reduced signaling role of dividends in a general cross-sectional time-series context, we
complement the literature by focusing on a narrower setting that provides a powerfiil test
of the dividend-signaling hypothesis.

II. Sample construction and descriptive statistics

Our sample consists of firm-year observations from Compustat's Industrial and
Research Annual Data Bases for the years 1971-2000. Consistent with Hayn (1995) we
define our earnings variable as income (loss) before extraordinary items and discontinued
operations or IB (annual Compustat data item #18). We define two main variables to
capture dividend payments by the firm. We define DIVCF as the total dollar amount of
cash dividends paid by the firm (annual Compustat data item #21).^ We focus on
(changes in) the total amount paid rather than dividends per share to be consistent with
our view that the total dividend cash outflow establishes the cost to a firm already faced
with both a loss and a negative cash flow in a particular year. Since it is possible that the
dividend cash outflow in a particular year increase without dividends per share being
affected if the firm's number of outstanding shares changes, we define a second variable
to measure dividends paid per share {DIVPS or annual Compustat data item #2 1 scaled by

data item #25).

Table I presents descriptive information for the sample. Panel A shows our initial
sample contains 206,420 firm-year observations: 146,394 profit observations and 60,026
loss observations (29.08% of the total). The panel flirther shows a relation between
dividend payments and loss occurrence: consistent with the evidence in Skinner (2003)
we find dividend-paying firms are less likely to incur losses than non-dividend-paying
firms. Focusing first on the dividend payments of our firm-year observations in the year
prior to the current observation, we observe 117,618 firm-year observations with no
dividends and 88,802 firm-year observations with cash dividends. Of the firms that pay
no dividends 44.43% incur a current loss; by contrast, of the dividend paying firms only
8.74% incur a current loss. The contrast between dividend-paying and non-dividend-
paying firms becomes sharper when we focus on the contemporaneous relation between
dividend payments and firm profitability: of the firm-year observations not paying a
dividend, 45.90% have a contemporaneous loss, compared to only 6.69% of firms
currently paying dividends.

Panel B provides a description of dividend changes occurring in our sample. In
the ftiU sample the majority of firms never change their dividend payments: measured as
total cash outflow (or per share) 56.49% (54.40%) of firm-year observations do not
change dividends in a given year, 33.00% (28.85%) increase dividends, and 10.51%
(16.75%) decrease dividends. The percentages change significantly when we partition
the sample between profit and loss firm-year observations: profitable firms increase
dividends payments far more often than loss firms. By contrast, the large majority of loss
firms do not change their dividend payments (86.98% or 86.04% depending on whether

we measure dividends as total cash outflow or per share). The high percentage reflects
the fact that loss firms are less likely to pay dividends, and that only a small fraction of
loss firms that pay dividends increase dividend payments (4.20% or 3.80% depending on
how we measure the dividend variable).

Panel C in Table I cross-tabulates our two measures of dividend changes. The
diagonal percentages in panel C show that in the vast majority of cases both proxies
reflect the same direction of dividend change. However, changes in the number of shares
outstanding, with or without a constant dividend per share, can lead to non-zero off-
diagonal percentages. For example, we observe that in 20.54%) of cases firm's total cash
dividend payments increase in a particular year, yet the dividend per share variable shows
a dividend decrease. Such a combination is the result of an increase in the number of
shares in the same year (e.g., as a result of equity issuances or stock option exercises).
The two variables therefore complement each other as proxies for dividend decisions by
management. While we focus primarily on the dollar value of dividend payments, since
it best captures the amount of cash the firm is using, we present results using both
variables to illustrate the signaling role of dividend increases.

In Table II, we present evidence for our main variables of interest. Since we
hypothesize that the sign of the cash flow component of negative earnings will determine
the relative cost of a dividend increase we present our descriptive statistics for a sample
of loss observations partitioned by the sign of the cash flow component of the losses. We
define the cash flow component of earnings (CFO) as cash flow from operations,
measured as net income (annual Compustat data item # 1 72) less accruals. We measure
accruals as {ACurrent Assets (data item #4) - ACash (data item #1) - ACurrent Liabilities

(data item #5) + ADebt in Current Liabilities (data item #34) + Depreciation and
Amortizations (data item #14).

Panel A of Table II shows the mean, standard deviation, and median for four
variables of interest. First, we define Size as market value of the firm (stock price times
the number of outstanding shares, or annual Compustat data item #199*data item #25).
Second, we define ROA as earnings before extra-ordinary items and discontinued
operations (or IB as defined before) scaled by lagged total assets {TA, annual Compustat
data item #6). Third, we define CFO as before. Finally, we include SPI, or special items
(annual Compustat data item #17) scaled by lagged assets {TA,.i), in the Panel since
previous research singles out SPI as the specific component of losses related to both the
quality of losses (i.e., degree of permanence) and the dividend-paying behavior of firms
(Skinner 2003).

Panel A reports significant differences between the means and medians of the two
subsamples (based on two-sided t-tests and two-sided Wilcoxon tests for the mean and
median) as a fiinction of the sign of their cash flow component. Generally speaking, loss
observations with a positive cash flow are larger and exhibit stronger profitability
(smaller losses) than loss observations with a negative cash flow component. Positive
cash flow loss firms on average also report less negative SPI, with both types of loss
firms having median SPI of zero though. All differences between means and medians are
statistically significant.

Panel B of Table II provides descriptive statistics on the incidence of dividend
increases in the sample of loss observations as a function of the sign of the cash flow
component of earnings. We define ADIVCFJUP {ADIVPSJUP) as an indicator variable

that takes the value of 1 if the change in dividend cash outflow (dividend payout per
share) is positive, and zero otherwise, with DIVCF and DIVPS as defined before. Panel
B shows a significantly smaller proportion of loss observations with a negative cash flow
component increases dividends, consistent with a dividend increase being costly:
regardless of the dividend variable used, the percentage of positive CFO loss firms
increasing their dividends is more than twice the percentage of negative CFO loss firms.

Table 11 provides evidence consistent with loss observations being different as a
fiinction of the sign of the cash flow component of the loss: a negative cash flow
component suggests a greater deterioration in profitability and a lower incidence of
dividend increases. In Table III, we formally test the relation between current
profitability, the sign of current cash flows, and (changes in) dividend payments in the
sample of loss observations. Specifically, we estimate a logistic regression to evaluate
the relation of profitability and its components to the decision to increase current
dividends. Focusing on our two dividend variables we estimate the following four

Prob (ADIVCFJJP) = og + a,ROA, + asCFONEG, + as ROA,*CFONEG, +

a^LSize, + Si (1)

Prob (ADIVPSJJP) = bg + biROA, + b2CFONEG, + bj ROA,*CFONEG, +

b4LSizet + 82 (2)

where ADIVCFUP, ADIVPSUP, ROA are as defined before; CFONEG is an indicator
variable equal to one if the firm reports a negative cash flow, and zero otherwise;
ROA*CFONEG is the interaction between ROA and CFONEG. Besides our main

variables of interest, ROA, CFONEG, and the interaction between both variables, we
include a control variable for the size of the firm in each specification since Hayn (1995)
and Joos and Plesko (2003) relate the size of the firm to the persistence of the loss and
therefore the potential cost of a dividend increase. Our size variable is LSize, the log of
the market value of the firm.

Table III reports the results of estimating models (1) and (2) using the method
detailed by Fama and Macbeth (1973). In both models the coefficient on ROA is positive
and highly significant, consistent with a relation between higher profitability and
dividend increases. However, the negative coefficient on CFONEG indicates that, on
average, loss firms with negative cash flows are less likely to increase their dividend.
Further, the negative and significant coefficient on ROA*CFONEG shows that the
relation between higher profitability and dividend increases in the full sample is smaller
for firms with a negative cash flow. The size control variable has a positive and
significant coefficient, suggesting that larger firms are more inclined to increase
dividends in the current loss year regardless of the sign of the cash flow.^

In sum, the analyses in Tables I through III suggest a positive relation between a
firm's profitability and its propensity to increase dividends. Focusing on loss
observations in particular, we find the presence of a negative cash flow component of the
loss reduces the probability of a dividend increase, consistent with negative cash flows
from operations increasing the cost of a dividend increase.


III. Do dividend increases forecast future profitability?

To examine whether costly dividend increases constitute strong signals of fiature
profitability we estimate an earnings forecasting model in our sample of loss
observations. Since we argue that increases in dividend outflows are more costly when
cash flows are negative, we predict the decision to increase dividends is a stronger
predictor of future profitability for negative cash flow loss firms than for positive cash
flow loss firms. We consider two forecast horizons, one and three years, and focus on
future accounting profitability by estimating the following parsimonious models: '"

AROA,+r = ao+ a, CFONEG, + aj ROA, + aj ADIVCF_UP,+

a4 ADIVCF_UP*CFONEG, +05 SPI, + a^ LSize, + 83 (3a)

AROA,^, = Po+ /^i CFONEG, + yff, ROA, + y9j ADIVPS_UP,+

P4 ADIVPS_UP*CFONEG, +^5 SPI, + Pe LSize, + £3 (3b)

We define future profitability as average ftiture ROA over the forecast horizon:
AROA,+^=(Z,+r ROA,+/t, where t=1 or 3) and estimate models (3a) and (3b). The first
specification (3a) focuses on increases in DIVCF, and the second (3b) on increases in
DIVPS. ROA^ SPI, CFONEG, LSize are defined as before. Our main variables of interest
in model (3) are dividend increases (ADIVCFUP or ADIVPSUP), and dividend
increases interacted with the negative cash flow indicator variable CFONEG. If our
prediction that the decision to increase dividends is a stronger predictor of future
profitability for negative cash flow loss firms than for positive cash flow loss firms, a^
and ^4 will both be positive and significant.

We include controls for current profitability (ROA), special items, and size. We
include special items {SPI) for the reason mentioned earlier, namely that previous


research relates SPI to both the quality of losses (i.e., degree of permanence) and the
dividend-paying behavior of finns (Skinner 2003). We include LSize to control for
potentially omitted variables such as risk or growth of the firm. We estimate both
specifications using the Fama-MacBeth methodology.

Table IV presents the results of the estimation of equations (3a)-(3b). All four
estimation results (columns (1) through (4)), focusing on different dividend measures and
forecast horizons, show the same result for the dividend increase variables: the
coefficients as and /5j on the dividend increase variables are not significant, indicating
that a dividend increase for loss firms with a positive cash flow component does not
signal future profitability controlling for other factors in the model. By contrast, the
coefficients a^ and y9^ on the dividend increase variable interacted with CFONEG are
positive and significant in all specifications. In untabulated analysis, we also find the
sum of Oi+a^ and ^3+^4 are positive and statistically significant in all specifications. The
evidence is consistent with dividend increases signaling future profitability, even after
controlling for other factors, when the cash flow component of losses is negative. This
finding supports the hypothesis that dividend increases constitute an informative signal
when the cost of the signal is relatively high.

Table IV also shows the coefficients on CFONEG {aj or ^/) are negative and
significant in all four specifications, consistent with losses with negative cash flows
signaling persistent profitabihty problems (see also Joos and Plesko 2003). By contrast,
the coefficients on ROA (a2 or ^2) are positive and highly significant in all specifications,
consistent with the previous findings on the serial correlation and mean reversion of ROA
(e.g., Sloan 1996). The coefficients on SPI (as or (is) are negative in all four


specifications, but the level of significance varies depending on the forecast horizon: the
coefficients are marginally significant in the one-year horizon models (columns (1) and
(3)), but highly significant over the three-year horizon (columns (2) and (4)), suggesting
special items affect firm profitability more over the longer horizon, and are less
informative over the shorter horizon. Finally, size predicts fiiture profitability only one
year ahead (columns (1) and (3)), but not three years ahead (columns (2) and (4)).

In summary, the results for both dividend variables and both forecast horizons are
consistent with a dividend increase providing information on the future performance of
loss firms only when current cash flows are negative. We interpret the results to indicate
the usefulness of a dividend increase to signal future firm performance is directly related
to the expected cost of the dividend increase.

IV. Robustness analyses

We carry out three (unreported) analyses to test the sensitivity and robustness of
our findings. In our first analysis, we focus on a subsample of firms with more than one


Online LibraryPeter JoosCostly dividend signaling : the case of lost firms with negative cash flows → online text (page 1 of 2)