A Comparison of Dividend Cash Flow and Earnings Approaches To Equity Valuation
A Comparison of Dividend Cash Flow and Earnings Approaches To Equity Valuation
A Comparison of Dividend Cash Flow and Earnings Approaches To Equity Valuation
Stephen H. Penman
Walter A. Haas School of Business
University of California, Berkeley
Berkeley, CA 94720
(510) 642-2588
and
Theodore Sougiannis
College of Commerce and Business Administration
University of Illinois at Urbana-Champaign
Champaign, IL 61820
(217) 244-0555
January, 1995
Revision: April, 1996
over various horizons, with and without terminal value calculations, are
compared with (ex ante) market prices to give an indication of the error
these errors show that accrual earnings techniques dominate free cash
the accounting that makes it less than ideal for finite horizon analysis
present value but little attention has been given to the specification
finite horizons. The problems this presents for going concerns are well
immediate future are often not related to value so the forecast period
the firm. However this substitution solves the practical problem only
In discounted cash flow (DCF) analysis the terminal value often has
future forward relative to cash flow analysis, but this claim has not
horizon analysis. What techniques work best for projections over one,
cash flows are "real" and the accounting introduces distortions, but is
forecasts are not observable for all payoffs, intrinsic values are
3
averaged in portfolios and portfolio values are then pooled over time to
on which the techniques are compared. Both mean errors and the
prices with which calculated intrinsic values are compared are efficient
payoffs.
methods with different rules for recognizing payoffs, and their relevant
value in DCF analysis. This brings the focus to the critical practical
analysis) are identified and the error metrics for the techniques are
calculated over departures from this ideal. Thus the aspect of the
perform particularly well over five to eight year horizons. These are
Section III outlines the research design and the data sources, and
Pt = ∑ ρ E( ~d )
-τ
t+τ (DDM) (1)
τ =1
t+τ, ρ is one plus the discount rate (equity cost of capital), indicated
for all τ. In this "clean surplus relation," Bt+ τ is the measured stock
of value ("book value") at t+τ, Xt+ τ is the measured flow of added value
recognized (in Preinreich (1938), Edwards and Bell (1961) and Peasnell
(1982), for example) that, solving for dt+ τ in the CSR equation and
that for the dividend discount formula. The expression over which the
technique and a (pro forma) accounting system (for equity valuation) are
Koller, and Murrin (1990), Hackel and Livnat (1992) and Cornell (1993)).
(CCE):
Ct+τ - I t+τ ≡ d t+τ - F t+τ , all τ , (CCE) (4)
equity contributions (as before). Let FAt denote the present value of
assets). Then, solving CCE for dt+ τ and substituting into (1),
Pt = ∑ ρ E(~C -
-τ
w t+τ
~I )
t+τ + FA t , (5)
τ =1
where Ct+ τ - It+ τ is called "free cash flow" and FAt is usually indicated
also be derived from the stocks and flows equation (CSR). Thus (5) is a
identifies Bt+ τ ≡ FAt+ τ and Xt+ τ ≡ Ct+ τ - It+ τ + it+ τ , all τ, where it+ τ is cash
and which is negative for net debt. Thus the clean surplus equation,
that tracks financial assets (or debt). Free cash flows are invested in
financial assets (or reduce debt) and dividends are paid out of
financial assets. This merely places the CCE flow equation on a stocks
~ ~ ~
∑ ρ E (C- I+ i ) - (ρ -1)~FA
.
-τ
PTt = FA t + t+τ t+τ -1 (6)
τ =1
*
Replacing it+ τ with i t+τ such that
∑ ρ E(i )= ∑ ρ ( ρ -1)E(~FA ),
~ -τ * -τ
t+τ t+τ -1 (7)
τ =1 τ =1
∑ ρ E(~C - ),
then
-τ ~I
PTt = FA t + w t+τ t+τ (DCFM) (8)
τ =1
The book value of equity is the value of the bonds and the technique for
with no financial assets or debt (an "all equity" firm, for example),
free cash flow, Ct+ τ - It+ τ ≡ dt+ τ , by CCE, and hence the target is the
same as in the dividend discount formula with the same problems induced
liabilities) which are accounted for as OAt+ τ = OAt+ τ -1 + It+ τ + oat+ τ where
are booked as part of operating assets rather than part of the value
from (3),
∑
τ =1
[
)]
(RIM) (9)
- (ρ -1)( ~
FA t+τ -1 + ~
OA t+τ -1
∑ ρ E[OI - (ρ -1)OA ],
~ -τ ~
PTt = FA t + OA t + w t+τ t+τ -1 (10)
τ =1
Equation (10) reflects that financing is at zero net present value and
therefore drops out. The target, operating income less a charge against
~
VTt ≡ ( ρ T -1 ) -1E ∑ X + ∑(ρ -1) ~
T- τ
t+τ d t+τ (CM) (11)
τ =1 τ =1
earnings within the firm for earnings from reinvesting the dividends
paid out and capitalizing the aggregated cum-dividend flow at the cost
value of the expected residual income in (3) rather than its present
X and B and the free cash flow and accrual accounting specifications are
special cases. Easton, Harris and Ohlson (1992) show that the
according to GAAP, are highly correlated with stock returns over five to
projected "to infinity," and this equals the valuation for the infinite-
what conditions.
∑ ρ E(d )+ ( ),
~
PTt = -τ
t+τ ρ -T E ~
Bt+T (12)
τ =1
that is, the present value of forecasted dividends to t+T plus the
present value of the expected t+T stock. As, for DCF analysis,
Bt+T ≡ FAt+T and for RIM, Bt+T ≡ FAt+T + OAt+T , the two valuations differ for
assets, and are the same only when operating assets are projected to be
∑ ρ E(d )+ ( )
~
Pt = -τ
t+τ ρ -T E ~
Pt+T (13)
τ =1
T
comparing (12) and (13)), and the error of P t is ρ -T E (~ Bt+T).
Pt+T - ~
13
Accordingly, the DCF analysis will yield the correct valuation only if
market value), and RIM will yield the correct valuation if expected t+T
horizon, and the error is given by the present value of the expected
change in premium (Ohlson (1995)). The zero error conditions for both T
P
T
and Vt have the feature that the accounting brings the future forward
t
sufficient for subsequent flows (and for expected price at t+T) and for V
T
aggregated (cum-dividend) flows to t+T are sufficient for projecting
t
be used for firms with continuing operations and Ou and Penman (1995)
+ρ (ρS -Ks) -1 E ∑ X~ t+T+ τ + ∑ ρS-τ -1 ~dt+T+τ - (Ks -1) B~t+T -E ~Bt+T
-T
τ =1 τ =1
14
restated as τ =1
S S (14a)
( )
S
τ =1 τ =1
capitalizes at the rate ρ-K1 where K1 is the one period growth rate.
DDM. As the last term in (14) gives the error, E(~ ~ ), then E(t+T )
Pt+T - Bt+T
in (13) is supplied:
15
~
( ) E ∑ X~ + ∑ (ρ )~
-1
+ ρ -T ρ S -Ks t+T+τ
S-τ
-1 d t+T+τ - (Ks -1) Bt+T
(15)
τ =1 τ =1
= PT*
t
for the DDM and it is this calculation that is the determining one.
becomes
Pt = ∑ ρ E(d
τ =1
~ -τ
t+τ ) + ρ [(ρ -1)
-T -1
E(X~ t+T+1 )] (15a)
as ∑
τ =1
( ) [ ]
(15b)
[ )],
T
∑
τ =1
( ) [ ]
(15c)
[(ρ -1) )]
T
= ∑ ρ E( ) +
τ =1
-τ ~
d t+τ ρ -T
w
-1
E( ~
OIt+T+1 )+( ~
FA t+T
and so for S > 1 and Ks > 1. Thus, given the premium (error) condition
under which (14) yields the price for the accrual accounting model, the
DCF valuation will also yield the same price for the same horizon (only)
Pt = FA t + ∑ ρ E (C-
τ =1
~ ~
wI) +
-τ
t+τ
ρ -Tw [( ρ -1) E(OI~ )]
w
-1
t+T+1 (15d)
this is not cash flow analysis at all, but rather accrual accounting,
and contrasts to the pure DCF analysis in (15b) which, stated in the
with the accommodation for S > 1 and Ks > 1. As Ct+T+1 - It+T+1 ≡ dt+T+1 -
accounting against the pure DCF analysis with the understanding that
RIM and CM over various horizons, with and without the terminal value
models up to various t+T+1 and these are then compared with actual
traded price at t.
from ex-post data. We assume that (a) average realizations are equal to
with the aim of averaging out unexpected realizations and any market
then under various circumstances where the accounting may affect the
horizon over which analysis is done. The analysis over all conditions
conditioning circumstances.
18
errors and the variation in errors are then calculated over all 18
years.5
The data used in this study are taken from the COMPUSTAT Annual
and Research files which cover NYSE, AMEX, and NASDAQ firms. The
termination. The files cover the period 1973 to 1992. Financial firms
(industry codes 6000-6499) are not included in the analysis. The number
of firms available for each year (with prices, dividends, and accounting
data for that year) range from 3544 in 1973 to 5642 in 1987, with an
average of 4192 per year. As there are no data after 1992, the number
ten-year horizons (T=10), there are 10 years (1973-82) and for T=1,
the risk free rate (the 3-year T-Bill rate p.a.) for the relevant year
plus an equity risk premium of 6% p.a. for all firms (approximately the
the beginning of the sample period); the cost of capital given by the
CAPM using the same risk free rate and risk premium with betas estimated
for each firm; and the cost of capital for the firm's industry based on
the Fama and French (1994) three factor (beta, size and book-to-price)
model.6 These all were updated each year. Finally, we used a 10% rate
for all firms in all years. We report results with CAPM estimates (and
the notation, ρ, will imply this) but little difference in results was
A. Unconditional Analysis
was 210, and the mean (over the 20 portfolios) of the (arithmetic) mean
portfolio per-share market prices (over years) was $14.29, with a range
estimated beta, 1.13 (with a range of 1.12 to 1.14). The mean ex ante
give, for each t+T, the cum-dividend earnings yield per dollar of price
increased more than at the calculated average ex ante rate of 12.8% per
misspecification of this rate but also reflects the bull market of the
sample period. In other words, the data period is not long enough to
The t+1 figures for dividends, free cash flows, and earnings
indicate that the average annual yield of these payoffs was less than
the 12.8% rate during the period, but each increased at the average over
cum-dividend prices. However, the increase was less than that of the
lead" payoffs. The yields of ex post dividends and free cash flows were
less than that of GAAP earnings. As free cash flows are returns to
payoff in the time t price (by which these realizations are initialized)
as
[ ]
Error T (•) = P pt - PTpt (• ) / P pt (17)
T
where P pt (⋅) is the portfolio intrinsic value at t calculated from
outliers. The results here and elsewhere are based on the latter
actual price at t. They are thus the market's forecasting errors, and
for any of these reasons and which one would expect to observe for a
23
inefficiencies (at the portfolio level) at t+T also and these are not
Rows two through five of the panel give valuation errors for the
dividend discount model (DDM), the discounted cash flow model (DCFM),
the residual income model using GAAP earnings and book values (RIM), and
with the target projected to the relevant t+T without a terminal value.
carrying values).8 Free cash flow is after income taxes so the tax
benefit of debt is included. Errors for the DCFM and RIM with terminal
values are given lower in the panel. These are calculated according to
(14a) with S = 1 and K1 , the annual "growth rate," set to 1.0 and 1.04
for the DCF model (for going concerns) and 1.00 and 1.02 for the RIM
∑ ρ E(d ) + ρ [(ρ - K ) ]
model calculated with a terminal value as
~-τ -T -1 ~
Pt = t+τ 1 E( d t+T+1) (DDMA) (18)
τ =1
The errors for the dividend discount models are large and positive
for short horizons but decline over t+T towards the benchmark errors as
calculation.10 The errors for the DCF calculation are also positive and
large over all horizons, indeed greater than 150% of actual price.
These errors reflect the missing accounting for operations. With the
terminal value calculations, the errors are still large for all t+T,
though declining with higher values of K1 . (When K1 was set to 1.06 the
mean error for t+8 was -0.076.) In contrast, the errors based on GAAP
accounting in RIM and CM are lower for all horizons and much closer to
portfolio errors from these means were also calculated and the rankings
over techniques were similar to that for means. In no case did earnings
B. Conditional Analysis
distinguishing firms from the market and we now examine how errors
differ over firms (for varying horizons) when the alternative techniques
are applied. The analysis proceeds as before except that firms are
applying the DDM over finite horizons. This difficulty is acute when a
request. Predictably, the DDM and DDMA valuations varied over payout
ex ante values. Errors for short horizons were typically large. Those
for the DCF techniques were also large for all horizons, though
and how competing techniques perform under the same conditions. The
in (3) (and of the finite horizon model in (14)) arises when the
accounting system accounts for Bt such that Pt = Bt (and the other terms
in (3) and (14) are zero). Here the horizon is T=0, all the future is
brought forward into the current book value, and current book value is
sufficient for all expected future payoffs (by applying the cost of
not satisfied, there is missing value in the current stock and one has
to project the future to discover this value, and thus T>0. The ratio
firms on this ratio for DCF accounting and GAAP accounting and examine
the implied horizons (to capture the missing value) over deviations from
the ideal.
portfolios formed from ranking firms on FAt /Pt . FAt is the DCF stock and
stock (PS). Only results for horizons t+1, t+5 and t+8 are reported;
valuation errors for six models are given as indicated. Results with
(Debtt + PSt )/Pt for each portfolio, the GAAP B/P ratio at t and free
e e *
cash flow to equity, FCF t/Pt where FCF t ≡ Ct - It + i t (with i
*
dividends), and the GAAP E/P ratio at t. These are ranking variables in
current price (the price model in the first panel) are negative and
portfolios with very high leverage firms performing better than average,
minus the value of debt plus preferred stock, the ranking ratio captures
the value of the omitted operating assets in the DCF stock. Over all
levels of this condition the DCFM errors are positive and large for all
errors. The payoff in free cash flow is too low to justify the price
case of the high debt firms. The "terminal value correction" with
K1 = 1.04 reduces these errors but they are still large and the
28
as these are pure equity firms where free cash flow equals dividends.
accounting, one has to forecast future free cash flows but the results
indicate that this calculation does not bring the future forward within
horizons less than nine years. GAAP book values include a measure of
are much closer to the benchmarks. They are in the order of the
benchmarks but still higher, indicating value payoffs are not entirely
(RIM(TV:1.0)) reduces the errors for the lower portfolio numbers, but
increases them for the higher ones (as explained with the next table).
The CM errors also are lower than DCFM but are typically higher in the
calculated for each portfolio (overtime) and these are also considerably
Table 3, firms are ranked on GAAP B/P (the GAAP stock to price) and this
between B/P and the price model errors describes the positive
29
in Fama and French (1992), among others. This could indicate superior
ex post performance for high B/P firms or higher risk, but also may
phenomena.12
deviation of B/P from unity in portfolio 13. However, those for high
B/P are close to their benchmark errors for t+5 and t+8. It is the low
B/P firms for which the errors are relatively high and, as the ex ante
error for RIM is given by E(t+T - t+T ), these are firms for which the B/P
supplies the missing value for the low B/P firms (and of course more so
with a growth rate), but its errors for high B/P are actually higher
than those for RIM. These are portfolios which on average had negative
indicate that the horizons for the firms are too short and that the
The errors for CM are also ordered on the benchmarks except they
are higher for both low and extremely high B/P firms. The error of this
B/P that are associated with the biggest changes in premiums (Ou and
30
Penman (1995)). The errors for DCFM without a terminal value are very
tables. It is clear from the DCFM (TV:1.04) results reported that DCF
analysis, even with a growth rate of .04 for the horizon correction,
positive correlation between (Debt + PS)/P and B/P, because the table
also indicates that FCF tends to be negative for low B/P firms.
price and book value to grow at the cost of capital and accordingly
residual income). Thus portfolio 13's RIM valuation errors, just like
those for the price model, represent systematic unexpected errors due to
errors due to unexpected value appreciation. The errors for t+5 and t+8
are higher than those for the price model and this is consistent with
are incorporated into price before being recognized in earnings and book
value. Errors for other portfolios reflect the phenomenon and thus
Rather than the current stock being sufficient for valuation, the
flows are projected by applying the cost of capital to the current flow.
ρ
Pt + d t = X, (19)
( ρ -1) t
that is, cum-dividend price is the capitalized current flow and the
this ideal all the future is pulled into the current flow calculation
maximizes the dispersion from the ideal (for Xt ≡ FCFt to equity and Xt =
The errors for DCFM (TV:1.04) are indeed relatively small for these
errors. They are particularly high for negative FCF firms where the
problem of using DCF analysis is acute. The errors for RIM, with and
32
without the terminal value, are much lower but, as with those for CM,
they are higher for portfolios where the reported B/P are low.
E/P closest to .113 and thus represent the ideal in (19). By the same
The CM errors increase from this benchmark as the spread from the ideal
However, they are higher for low E/P portfolios. This is so for the
RIM calculations with and without terminal values. The DCFM errors are
again large.
33
The difference between free cash flows and GAAP earnings is the
the subsequent cash (in flows) from the investment. This is apparent in
is large relative to cash flow from operations, the DCF valuation errors
the error in DCF valuation. However, accruals are by fiat and may
Table 6 ranks firms on GAAP E/P minus FCFe /P. This difference is
price. The greater the absolute difference between FCF and GAAP
produces the largest correction to DCF analysis when free cash flows are
extreme and when the difference between earnings and free cash flows is
finite horizon analysis. The results suggest that rather than adjusting
earnings forecasts to get back to cash flows, one is better served (for
results also indicate conditions where GAAP models do not perform well
B/P (Table 3) and low E/P (Table 5). These conditions are associated
with the central portfolios in Table 6 where again the RIM errors are
the highest. Given that these findings are not due to market
clean surplus accounting, then the zero error conditions for (3), (11)
and (14) are not satisfied in these circumstances for the horizons
investigated.
The results for B/P and E/P involve conditions where the
other conditions involve low book values and earnings to price, one
Median B/P ratios are less than unity in the sample, reinforcing this
impression. The low B/P are cases where the conservatism is likely to
price change (return) at the horizon, that is, the conservatism does not
premium condition for K1 = 1.0 will not be satisfied. This has a formal
Table 7 displays mean valuation errors for various joint values of B/P
and E/P. In each year firms are ranked on B/P and those with
.95 ≤ B/P < 1.05 assigned to portfolio 12. Then firms with B/P < .95
are ranked on E/P and assigned to portfolios 1-11 from this ranking, and
those with B/P ≥ 1.05 are also ranked on E/P and assigned to portfolios
capital. Thus this portfolio describes results for both a normal book
value and a (close to) normal P/E ratio in (19) and the errors for the
benchmark errors given the systematic ex post price errors. For low B/P
low B/P and low E/P. The GAAP models do not perform well in conditions
that are associated with conservatism in the accounting for book values
errors for low B/P and E/P portfolios are considerably lower than those
of RIM (TV:1.0) for the longer horizons. However, even with this
37
adjustment, the errors are higher than the benchmarks. If one considers
then longer forecasting horizons are required for these firms. In any
conditions.
V. CONCLUSION
GAAP accounting has the feature of bringing the future forward in time
the accounting for the payoff. This facilitates valuing firms from
technique will work well (or otherwise). The analysis is couched within
analysis, but they also indicate conditions (associated with high P/E
and high P/B firms) when the GAAP accounting is unsatisfactory. GAAP
might promote better techniques and better accounting. Also, the paper
business schools. The typical valuation book "backs out" accruals from
suspect. The results here modify that view. However, the paper shows
corrects the errors from forecasting free cash flow to the horizon to
get back the accruals. One questions the efficiency of going through
this exercise (of taking out accruals and then adding them back in) when
forecasting the accrual numbers produces the same result, and indeed
H-THS.17-13N
40
Appendix A
Dividends
(COMPUSTAT item 26) adjusted for stock splits and stock dividends over
and Smith (1993) and Shoven (1986) to discover them. We searched the
CRSP monthly returns file for information on shares outstanding and each
price at the end of the month preceding the decrease. This amount was
(6,117 for NYSE and AMEX firms and 1,542 for NASDAQ firms). This number
seems plausible for our sample period given that Comment and Jarrell
(1991) report 1,303 stock repurchases for the 4-year period 1985-1988.
41
Two free cash flow calculations were made. Results are similar
for the two calculations, but those reported are based on the second.
first appeared in 1971 when APB Opinion No. 19 mandated the preparation
code = 3). In 1987 the FASB issued Standard No. 95, "Statement of Cash
Flows" mandating the reporting of cash receipts, cash payments, and net
For firms with format code = 7.000 in a given year, cash from
under GAAP. We did not exclude them as, for the DCF calculation, we
or
2
I t+τ = Change in Property, Plant and Equipment-Total
(Net) (change in item 8)
+ Depreciation and Amortization (item 14)
+ Change in Investments and Advances-Other
(change in item 32)
+ Change in Intangibles (item 33)
- Capitalized Interest (item 147).
The change in working capital was modified for the change in Debt in
I.D, ∆OAt+ τ = It+ τ + oat+ τ and oat+ τ = OIt+ τ - Ct+ τ . Thus Ct+ τ - It+ τ = OIt+ τ -
operating activities. This is a problem only if these are not mean zero
(FAt ) are identified as (minus) the sum of debt and preferred stock at
of debt is the book value of long-term debt (COMPUSTAT item 9) plus the
These calculations of free cash flow are after tax but include tax
total dollar basis and the total dollar intrinsic price at t was placed
GAAP Accounting
(COMPUSTAT item 53), adjusted for stock splits and stock dividends over
time. Book value per share is influenced by share issues so its value
value per share at t plus accumulated earnings net of cash dividends per
adjusted for stock splits and stock dividends over time. That is, dt+ τ
does not include cash distributions from stock repurchases as these are
Appendix B
obtained from the CRSP monthly returns files and the COMPUSTAT files.
The CRSP files provide delisting codes indicating the reason for
1973-1992 (1,792 delistings for NYSE and AMEX firms and 1,563 for NASDAQ
firms). Out of this total, 1,851 delistings were due to mergers, 261
due to acquisitions, 124 due to liquidations and the remaining 1,119 due
detected 1,736 cash and 1,013 non-cash terminal distributions for our
sample (some firms had both cash and non-cash distributions). The mean
$20.62. A total of 402 NYSE and AMEX firms and 797 NASDAQ firms did not
price of the delisting firm in the CRSP files. The mean terminal price
the sum of dividends per share, cash distribution per share from stock
all adjusted for stock splits and stock dividends. Tracking subsequent
the last market value on CRSP and the last book value on COMPUSTAT. The
mean per share terminal gain for the 3,355 delistings in our sample was
FOOTNOTES
This contrasts with Kaplan and Ruback (1995) and Abarbanell and Bernard (1995) where prices
are compared to values calculated from forecasts of cash flows or earnings. That approach is
limited by the availability of dividend and cash flow forecasts and of earnings forecasts for
longer horizons. Further, it assumes that the analyst forecasts identified are unbiased. On
this point see Frankel and Lee (1995).
Rubinstein (1976) derives the model under no-arbitrage conditions. In that derivation the
discount for risk is in the numerator which is then discounted to present value at the risk-
free rate. The common textbook form is stated here as this is usually how the model is applied
in practice.
Firms with different fiscal year ends in the same calendar year were assigned to portfolios
together and portfolio prices were based on firm prices of fiscal year end.
The Fama and French estimates are based on risk premiums estimated on data after our
portfolio formation dates. For CAPM rates, we also used an 8% equity premium which was the
historical rate at the end of the sample period. (This of course produced lower errors as we
define them.)
Precisely, those used in Kaplan and Ruback (1995) with unlevered betas calculated from
estimated equity betas and debt and preferred stock betas assumed to be 0.25. The income tax
rate was set at the prevailing top federal corporate rate in the relevant year plus 4% for state
and other taxes.
Financial assets are not netted out as one has difficulty distinguishing them from operating
cash in cash and cash equivalents. Accordingly interest income (but not expense) is included in
free cash flows.
The analysis was also repeated with growth rates of 1.02 and 1.06, with similar results. A
rate of 1.04 or 1.06 applied to RIM, though reducing errors towards the benchmark might be
considered excessive: they imply a relatively rapid perpetual growth in book return on equity.
For companies that terminated, the liquidating dividend was calculated as the price at
liquidation (see Appendix B). To the extent that distributions in liquidation were stock rather
than cash, calculated values for the DDM are overstated.
The occasional large negative value in this panel arises from capitalizing relatively high
free cash flow in the terminal value calculation with the small capitalization rate that
K1 = 1.04 can produce.
The results in Table 3 were similar when the cost of capital was based on the Fama and French
48
Ou and Penman (1995) document persistent discounts. In the DCF terminal value calculations
we capitalized only positive free cash flows as a perpetual negative free cash flow is not
realistic.
The ex post negative residual earnings could also be due to our specification of the cost of
capital on which the residual earnings calculation is based. The negative amounts were
particularly associated with periods of high interest rates.
This is what Fischer Black had in mind when he advocated calculating earnings as a sufficient
number (with a multiplier) for value. See Black (1980).
Note, however, that the price model errors might reflect the so-called "P/E effect" pricing
inefficiency.
An exception is the accounting for investment in research & development under SFAS No. 2.
Beaver and Morse (1978) document that differences in P/E ratios from the median persist over
time and Ou and Penman (1992) provide similar documentation for premiums.
49
REFERENCES
Horizon (t+T)
t+1 t+2 t+4 t+6 t+8 t+10
Horizon (t+T)
t+1 t+2 t+4 t+6 t+8 t+10
NOTES:
Means are mean over years of means for 20 portfolios to which firms were
randomly assigned in each year, 1973-90. Standard deviations are means of
yearly standard deviations of portfolio values.
Valuation error is actual portfolio price at t minus model price,
deflated by actual price at t. Price model valuation errors are calculated by
setting model price equal to the present value of actual ex-post cum-dividend
price at each horizon, t+T.
DDM refers to the dividend discount model in equation (1) of the text,
DCFM to the discounted cash flow model in equation (8), RIM to the residual
income model in equation (9) with GAAP earnings and book values, and CM to the
capitalized GAAP earnings model in equation (11). TV indicates a terminal
value was calculated for going concerns according to (14a) with the assumed
subsequent growth rate in the terminal payoff indicated within the
parentheses. DDMA is the dividend discount model with a terminal value
calculated according to equation (18).
All calculations include terminal distributions to equity holders for
nonsurviving firms.
Table 2
Mean Mean
(Debt+PS)/P GAAP B/P Price Model DCFM DCFM (TV: 1.04)
1 .000 .640 -0.024 -0.235 -0.388 0.990 0.947 0.877 0.810 0.891 0.532
2 .007 .550 -0.002 -0.025 -0.090 1.005 0.909 0.789 0.714 0.351 0.272
3 .034 .471 -0.002 -0.044 -0.116 1.060 1.009 0.920 0.927 0.575 0.463
4 .072 .546 -0.013 -0.127 -0.205 1.107 1.037 0.956 0.763 0.541 0.718
5 .116 .570 -0.034 -0.163 -0.259 1.197 1.128 1.005 1.018 0.421 0.041
6 .165 .615 -0.034 -0.173 -0.302 1.278 1.167 1.055 0.767 0.450 0.217
7 .221 .669 -0.030 -0.170 -0.331 1.356 1.293 1.140 1.148 0.532 0.498
8 .286 .734 -0.030 -0.183 -0.358 1.470 1.338 1.205 0.915 0.131 0.177
9 .359 .772 -0.034 -0.244 -0.385 1.545 1.409 1.205 0.768 0.570 -0.163
10 .443 .816 -0.044 -0.283 -0.404 1.713 1.530 1.331 1.095 0.694 0.444
11 .538 .896 -0.037 -0.268 -0.399 1.735 1.557 1.281 1.036 0.596 0.085
12 .654 .941 -0.046 -0.294 -0.493 1.957 1.809 1.682 1.205 0.874 0.399
13 .791 .985 -0.036 -0.351 -0.447 2.449 2.227 1.841 1.085 -0.766 -1.711
14 .964 1.047 -0.047 -0.311 -0.511 2.426 2.047 1.922 0.192 1.367 0.639
15 1.176 1.082 -0.043 -0.235 -0.313 2.586 2.313 2.251 1.455 0.967 0.974
16 1.442 1.115 -0.032 -0.265 -0.485 3.036 2.703 2.758 1.000 1.551 1.577
17 1.789 1.162 -0.040 -0.349 -0.497 3.413 3.147 3.219 1.388 1.717 1.412
18 2.302 1.270 -0.042 -0.414 -0.678 4.002 3.622 3.676 1.058 1.787 1.490
19 3.344 1.409 -0.052 -0.604 -1.060 5.380 4.448 4.214 0.110 0.531 2.521
20 10.962 1.432 -0.080 -0.861 -1.336 9.204 7.761 8.307 -3.251 3.542 6.497
Table 2 (continued)
Mean Mean
FCFe /P GAAP E/P RIM RIM (TV: 1.0) CM
1 .096 .075 0.322 0.178 -0.026 0.289 0.009 -0.208 -0.101 0.000 -0.181
2 .037 .071 0.415 0.343 0.272 0.311 0.223 0.150 0.181 0.245 0.210
3 .002 .070 0.515 0.375 0.263 0.353 0.169 0.170 0.286 0.234 0.191
4 .002 .078 0.438 0.292 0.147 0.233 0.134 0.131 0.241 0.154 0.067
5 .008 .082 0.416 0.273 0.162 0.222 0.158 -0.053 0.259 0.156 0.095
6 -.002 .085 0.356 0.219 0.090 0.196 0.049 0.007 0.164 0.098 0.014
7 .008 .090 0.301 0.169 0.023 0.149 0.058 0.029 0.149 0.064 -0.044
8 -.002 .096 0.232 0.114 0.025 0.176 0.116 0.031 0.132 0.050 0.008
9 -.004 .096 0.182 0.042 -0.054 0.089 0.046 -0.046 0.018 -0.025 -0.059
10 .001 .096 0.128 0.002 -0.090 0.110 -0.058 -0.035 0.069 -0.033 -0.071
11 -.006 .103 0.066 -0.072 -0.173 0.000 -0.108 -0.127 0.041 -0.101 -0.139
12 -.003 .099 0.016 -0.074 -0.193 0.103 0.032 0.032 0.071 -0.049 -0.159
13 .011 .097 -0.018 -0.116 -0.181 0.117 -0.069 0.041 0.091 -0.096 -0.112
14 .016 .095 -0.070 -0.141 -0.249 0.080 -0.038 -0.027 0.144 -0.062 -0.171
15 -.012 .096 -0.100 -0.148 -0.184 0.152 0.006 0.039 0.144 -0.033 -0.039
16 .021 .087 -0.119 -0.143 -0.206 0.237 0.008 -0.098 0.262 0.003 -0.083
17 -.032 .085 -0.166 -0.178 -0.276 0.269 -0.129 -0.085 0.183 -0.029 -0.184
18 -.012 .068 -0.247 -0.250 -0.330 0.303 -0.010 -0.038 0.388 -0.029 -0.204
19 .042 .014 -0.356 -0.302 -0.433 0.535 -0.089 -0.037 0.785 0.102 -0.255
20 .539 -.281 -0.248 -0.330 -0.393 1.132 -0.049 0.322 2.330 0.202 -0.098
Notes:
PS is the carrying value of preferred stock and FCFe is free cash flow to common equity. The GAAP E/P
ratio is calculated as Xt /(Pt +dt ) where Xt is GAAP earnings available for common in the portfolio formation
year, t. Pt is the common stock price at the end of year t and dt is the annual dividend for year t.
GAAP B/P is reported book value of common equity to price at t. See notes to Table 1 for descriptions of
valuation techniques and the calculation of the means.
Table 3
Mean Mean
GAAP B/P (Debt+PS)/P Price Model DCFM (TV: 1.04) RIM
1 -.033 .818 0.091 0.002 -0.098 1.231 1.272 1.372 1.149 0.789 0.597
2 .240 .279 0.038 -0.008 -0.059 1.077 0.753 0.663 0.770 0.588 0.490
3 .315 .293 -0.011 -0.080 -0.090 1.197 0.598 0.700 0.657 0.465 0.318
4 .392 .361 -0.015 -0.125 -0.192 1.191 0.910 0.631 0.569 0.374 0.221
5 .459 .471 -0.025 -0.146 -0.218 1.199 0.900 0.373 0.497 0.321 0.169
6 .527 .540 -0.016 -0.150 -0.257 1.274 0.805 0.731 0.424 0.255 0.110
7 .594 .634 -0.013 -0.164 -0.282 1.040 0.848 0.531 0.353 0.180 0.073
8 .672 .780 -0.037 -0.211 -0.431 1.339 0.715 -0.544 0.284 0.111 -0.037
9 .734 .772 -0.018 -0.151 -0.242 1.181 0.552 0.203 0.212 0.068 -0.050
10 .807 .906 -0.028 -0.200 -0.336 1.095 -0.024 0.934 0.146 0.010 -0.125
11 .873 .945 -0.046 -0.221 -0.391 1.307 0.674 1.152 0.073 -0.040 -0.161
12 .946 1.027 -0.036 -0.270 -0.503 0.488 0.839 0.574 -0.004 -0.121 -0.244
13 1.021 1.341 -0.056 -0.326 -0.588 0.815 1.171 0.012 -0.084 -0.198 -0.316
14 1.114 1.475 -0.070 -0.375 -0.605 0.653 0.687 0.368 -0.171 -0.242 -0.327
15 1.229 1.731 -0.075 -0.438 -0.677 0.162 0.891 -0.146 -0.270 -0.318 -0.402
16 1.338 1.839 -0.069 -0.374 -0.648 0.281 0.612 1.865 -0.375 -0.353 -0.416
17 1.530 2.241 -0.067 -0.488 -0.701 -0.584 0.765 -0.101 -0.526 -0.453 -0.502
18 1.744 2.973 -0.086 -0.459 -0.716 -0.898 0.539 1.439 -0.720 -0.550 -0.570
19 2.150 2.925 -0.111 -0.518 -0.803 -0.545 -0.153 1.755 -1.035 -0.698 -0.686
20 3.302 4.290 -0.143 -0.933 -1.282 -2.108 1.189 -3.167 -1.910 -1.073 -0.817
Table 3 (continued)
Mean Mean
FCFe /P GAAP E/P RIM (TV: 1.00) CM
Notes:
See notes to Tables 1 and 2.
Table 4
Mean Mean
FCFe /P GAAP E/P Price Model DCFM (TV: 1.04) RIM
1 -1.851 -.001 -0.016 -0.507 -0.932 0.948 1.978 4.220 -0.049 -0.007 -0.123
2 -.505 .068 -0.026 -0.337 -0.479 1.479 2.123 2.358 0.069 0.051 -0.038
3 -.311 .077 -0.009 -0.220 -0.306 2.364 1.603 1.222 0.132 0.120 0.001
4 -.216 .084 -0.016 -0.118 -0.180 2.500 1.449 1.419 0.184 0.162 0.078
5 -.153 .083 -0.022 -0.143 -0.202 2.105 1.134 1.225 0.242 0.193 0.123
6 -.107 .078 -0.007 -0.128 -0.256 1.814 1.122 1.040 0.289 0.200 0.125
7 -.071 .079 -0.030 -0.157 -0.203 1.654 1.292 0.021 0.322 0.204 0.114
8 -.042 .080 -0.035 -0.139 -0.289 1.300 0.991 0.861 0.327 0.220 0.115
9 -.019 .077 -0.039 -0.085 -0.163 1.087 0.913 0.606 0.346 0.251 0.173
10 -.000 .079 -0.028 -0.089 -0.167 1.081 0.611 0.540 0.384 0.277 0.170
11 .015 .078 -0.034 -0.142 -0.278 1.028 -0.174 0.773 0.360 0.251 0.122
12 .030 .085 -0.056 -0.161 -0.252 0.625 0.592 0.281 0.298 0.171 0.070
13 .047 .089 -0.067 -0.192 -0.345 0.719 0.382 0.371 0.220 0.096 0.002
14 .067 .095 -0.066 -0.231 -0.383 0.773 -0.079 0.060 0.157 0.010 -0.112
15 .094 .100 -0.082 -0.271 -0.417 0.760 -0.207 -0.312 0.053 -0.081 -0.180
16 .128 .104 -0.098 -0.346 -0.580 0.598 0.016 0.141 -0.021 -0.162 -0.270
17 .181 .105 -0.089 -0.422 -0.675 0.087 0.406 -0.060 -0.130 -0.250 -0.358
18 .271 .096 -0.132 -0.481 -0.746 -0.023 -0.419 0.864 -0.181 -0.265 -0.357
19 .484 .065 -0.137 -0.549 -0.807 0.121 0.304 0.967 -0.303 -0.356 -0.431
20 2.697 -.178 -0.109 -0.751 -1.188 -0.596 2.243 3.890 -0.231 -0.320 -0.462
Table 4 (continued)
Mean Mean
(Debt+PS)/P GAAP B/P RIM (TV: 1.0) CM
Notes:
See notes to Tables 1 and 2.
Table 5
Mean Mean
GAAP E/P FCFe /P Price Model DCFM (TV: 1.04) RIM
1 -1.256 1.051 0.050 -1.163 -1.382 -4.165 3.430 3.080 0.491 0.126 0.018
2 -.223 .047 0.030 -0.383 -0.534 -0.602 0.846 1.332 0.063 0.133 0.059
3 -.055 -.041 0.062 -0.179 -0.152 0.407 1.620 2.193 0.233 0.298 0.223
4 .003 -.040 0.066 0.067 0.009 1.082 0.948 1.154 0.415 0.429 0.360
5 .027 -.035 0.048 0.057 -0.001 0.341 0.812 0.959 0.426 0.409 0.350
6 .042 -.056 0.046 -0.051 -0.108 0.881 1.192 0.621 0.369 0.353 0.271
7 .055 -.040 0.044 -0.061 -0.084 1.185 0.944 0.706 0.386 0.317 0.214
8 .062 -.056 0.037 -0.057 -0.146 1.439 1.584 0.314 0.361 0.286 0.194
9 .072 -.043 0.022 -0.078 -0.148 1.631 0.974 0.731 0.337 0.220 0.107
10 .079 -.040 -0.010 -0.113 -0.252 1.362 0.665 0.834 0.290 0.165 0.031
11 .087 -.029 -0.019 -0.166 -0.284 1.184 1.079 0.816 0.252 0.092 -0.040
12 .094 -.030 -0.035 -0.200 -0.381 1.104 0.920 0.981 0.197 0.052 -0.103
13 .102 -.022 -0.042 -0.266 -0.390 0.928 0.567 0.149 0.140 -0.028 -0.187
14 .111 -.019 -0.056 -0.277 -0.478 1.102 0.599 0.043 0.072 -0.069 -0.201
15 .119 -.001 -0.076 -0.350 -0.570 0.571 0.276 -0.143 0.005 -0.142 -0.296
16 .130 -.015 -0.108 -0.417 -0.668 1.120 -0.158 -0.121 -0.080 -0.239 -0.377
17 .144 -.004 -0.103 -0.441 -0.676 0.426 0.523 -0.263 -0.133 -0.268 -0.395
18 .162 -.019 -0.128 -0.527 -0.821 0.867 -0.281 0.313 -0.256 -0.362 -0.477
19 .195 -.035 -0.148 -0.523 -0.896 0.862 0.157 -0.579 -0.348 -0.429 -0.553
20 .358 .148 -0.177 -0.653 -1.091 0.928 -0.058 0.941 -0.532 -0.608 -0.711
Table 5 (continued)
Mean Mean
(Debt+PS)/P GAAP B/P RIM (TV: 1.0) CM
Notes:
See notes to Tables 1 and 2.
Table 6
Mean Ex post Valuation Errors of Valuation Techniques for Selected Horizons, for Portfolios
Formed from a Ranking on the Difference Between GAAP Earnings and Free Cash Flow to Price
Mean Mean
(GAAP E-FCFe )/P FCFe /P Price Model DCFM (TV:1.04) RIM
1 -4.168 2.40881 -0.062 -0.926 -1.348 -3.282 3.755 5.453 0.057 -0.128 -0.289
2 -.568 0.42296 -0.069 -0.508 -0.764 0.233 0.375 2.079 -0.217 -0.229 -0.303
3 -.249 0.24194 -0.074 -0.396 -0.618 0.287 -0.013 0.668 -0.108 -0.133 -0.212
4 -.120 0.16030 -0.066 -0.350 -0.519 0.276 0.124 0.076 0.018 -0.055 -0.136
5 -.050 0.10691 -0.050 -0.186 -0.289 -0.232 0.221 0.365 0.167 0.078 0.011
6 -.009 0.07906 -0.045 -0.177 -0.286 0.839 0.449 -0.033 0.246 0.144 0.077
7 .018 0.06130 -0.034 -0.128 -0.250 0.864 0.173 0.424 0.311 0.198 0.125
8 .039 0.04779 -0.037 -0.130 -0.224 0.633 0.437 0.564 0.321 0.203 0.103
9 .058 0.03818 -0.038 -0.146 -0.234 0.751 0.101 0.396 0.302 0.169 0.079
10 .075 0.02719 -0.051 -0.177 -0.267 0.678 0.430 0.163 0.299 0.173 0.066
11 .093 0.01534 -0.063 -0.175 -0.285 1.069 -0.154 0.644 0.270 0.139 0.058
12 .114 0.00436 -0.061 -0.203 -0.332 1.110 1.050 0.505 0.227 0.117 0.025
13 .139 -0.01991 -0.058 -0.171 -0.265 1.054 0.883 0.737 0.194 0.066 -0.058
14 .169 -0.04539 -0.053 -0.231 -0.396 1.289 0.986 -0.348 0.176 0.082 -0.021
15 .206 -0.08267 -0.059 -0.208 -0.292 1.381 1.082 0.779 0.162 0.049 -0.056
16 .255 -0.12375 -0.058 -0.205 -0.345 1.835 1.452 0.979 0.122 0.033 -0.041
17 .321 -0.17973 -0.050 -0.262 -0.389 2.014 0.846 0.958 0.049 -0.014 -0.137
18 .421 -0.26757 -0.046 -0.247 -0.398 2.513 2.230 1.340 0.017 -0.018 -0.095
19 .617 -0.44429 -0.051 -0.406 -0.608 2.445 1.256 2.324 -0.040 -0.080 -0.199
20 2.028 -1.68530 -0.045 -0.531 -0.971 1.202 2.023 4.150 -0.156 -0.094 -0.196
Table 6 (continued)
Mean Mean
GAAP E/P GAAP B/P RIM (TV: 1.0) CM
Notes:
See notes to Tables 1 and 2.
Table 7
Mean Mean
GAAP B/P GAAP E/P Price Model DCFM (TV: 1.04) RIM
1 .091 -.522 0.091 -0.471 -0.740 0.163 2.001 2.557 1.065 0.626 0.499
2 .319 -.012 0.087 0.059 0.019 1.103 1.235 1.576 0.704 0.651 0.557
3 .361 .028 0.072 0.148 0.107 1.166 1.123 1.014 0.649 0.561 0.489
4 .400 .045 0.061 0.083 0.052 1.385 1.049 0.839 0.602 0.528 0.471
5 .432 .057 0.024 -0.004 -0.068 1.553 1.104 0.548 0.556 0.442 0.332
6 .476 .069 0.004 -0.065 -0.152 1.268 1.139 0.734 0.498 0.364 0.241
7 .523 .081 -0.012 -0.101 -0.179 0.930 0.727 0.698 0.441 0.292 0.173
8 .584 .094 -0.029 -0.160 -0.249 0.923 0.630 0.465 0.382 0.227 0.115
9 .645 .107 -0.038 -0.270 -0.418 0.892 0.809 -0.011 0.304 0.144 0.020
10 .691 .126 -0.089 -0.404 -0.552 0.836 0.582 0.653 0.252 0.042 -0.137
11 .685 .192 -0.118 -0.462 -0.719 1.026 -0.176 0.670 0.233 0.004 -0.165
12 .996 .108 -0.037 -0.245 -0.342 0.776 -0.103 -0.324 0.022 -0.016 -0.083
13 2.215 -.581 -0.017 -0.670 -0.839 -4.410 1.376 1.725 -0.615 -0.163 -0.182
14 1.710 -.054 0.028 -0.277 -0.338 -0.309 1.728 1.244 -0.453 -0.175 -0.149
15 1.492 .056 0.002 -0.252 -0.289 -1.345 0.828 0.813 -0.360 -0.184 -0.200
16 1.449 .097 -0.018 -0.239 -0.425 0.674 0.901 0.916 -0.343 -0.229 -0.285
17 1.438 .125 -0.057 -0.387 -0.667 0.032 -0.319 0.289 -0.371 -0.310 -0.394
18 1.465 .149 -0.100 -0.481 -0.772 -0.052 -0.026 -0.181 -0.437 -0.424 -0.504
19 1.516 .183 -0.144 -0.528 -0.878 0.476 -0.521 -0.108 -0.534 -0.521 -0.559
20 1.833 .328 -0.187 -0.720 -1.091 -0.331 -0.791 -0.405 -0.814 -0.695 -0.766
Table 7 (continued)
Mean Mean
(Debt+PS)/P FCFe /P RIM (TV: 1.0) RIM (TV: 1.04) CM
1 1.846 .457 0.917 0.358 0.240 0.916 0.056 -0.079 1.644 0.332 0.247
2 .454 -.091 0.824 0.421 0.604 0.753 0.169 0.489 1.130 0.661 0.496
3 .455 -.081 0.661 0.504 0.451 0.519 0.330 0.284 0.706 0.507 0.438
4 .408 -.079 0.580 0.467 0.297 0.383 0.266 0.035 0.479 0.450 0.419
5 .481 -.058 0.386 0.262 0.267 0.099 -0.019 0.050 0.398 0.327 0.247
6 .445 -.052 0.306 0.209 0.072 -0.011 -0.084 -0.233 0.296 0.234 0.161
7 .425 -.043 0.188 0.135 0.205 -0.175 -0.170 -0.001 0.190 0.147 0.097
8 .513 -.033 0.129 0.034 0.037 -0.264 -0.315 -0.242 0.150 0.062 0.028
9 .662 -.029 0.068 -0.035 -0.066 -0.353 -0.410 -0.390 -0.041 -0.051 -0.101
10 .833 -.028 -0.096 -0.244 -0.065 -0.604 -0.711 -0.352 -0.173 -0.218 -0.300
11 1.099 .014 -0.266 -0.281 -0.177 -0.892 -0.805 -0.562 -0.442 -0.318 -0.370
12 1.399 .032 0.182 -0.021 -0.139 -0.192 -0.384 -0.486 0.153 -0.029 -0.104
13 3.708 .641 2.128 0.474 0.428 2.689 0.238 0.183 4.355 1.010 0.331
14 2.232 .061 1.161 0.376 0.129 1.257 0.162 -0.138 1.678 0.398 0.183
15 2.188 .029 0.590 0.080 -0.022 0.403 -0.247 -0.327 0.901 0.155 -0.018
16 1.839 -.017 0.487 0.031 -0.072 0.269 -0.290 -0.370 0.580 0.051 -0.150
17 1.833 .012 0.150 -0.131 -0.344 -0.238 -0.527 -0.772 0.226 -0.130 -0.311
18 1.762 .019 -0.092 -0.247 -0.364 -0.604 -0.679 -0.791 -0.056 -0.285 -0.419
19 2.340 -.016 -0.115 -0.177 -0.330 -0.642 -0.593 -0.775 -0.289 -0.373 -0.450
20 3.694 .160 -0.041 -0.665 -0.274 -0.556 -1.412 -0.734 -0.520 -0.394 -0.592
Notes:
See notes to Tables 1 and 2.