USA: +1(669)289-8683 
Sign in
The questions that follow and the article
Comparing the Accuracy and Explainability of Dividend, Free Cash Flow, and Abnormal Earnings Equity Value Estimates
will inform your completion of Milestone Three. An understanding of the models in this assignment will assist you in hypothesizing the incremental impact of a new investment project for the company. The understanding of these models will contribute to your ability to look toward the future when considering the direction of an organization. This activity is worth a total of 75 points. See the distribution of points listed before each question.
Prompt: Once you have read the article “Comparing the Accuracy and Explainability of Dividend, Free Cash Flow, and Abnormal Earnings Equity Value Estimates” and, review and complete the questions below. Use the article to inform your responses to the questions below.
Assignment Questions:
1. (20 points) For models 2, 2a, and 2b:
· What is the best way to minimize the weighted average cost of capital?
· What is the effect of the weighted average cost of capital on the market value?
2. (20 points) For models 3, 3a, and 3b:
· What is the relationship between book value of equity and time t-1 and the market value of the equity?
3. (20 points) Discuss model 4 and expand on the importance and the meaning of the market risk premium.
4. (15 points) In your own words, what are the main conclusions for this article, and what could be improved upon in its analysis?
Comparing the Accuracy and Explainability of Dividend, Free Cash Flow, and Abnormal
Earnings
Equity Value Estimates
Author(s): Jennifer Francis, Per Olsson and Dennis R. Oswald
Source: Journal of Accounting Research, Vol. 38, No. 1 (Spring, 2000), pp. 45-70
Published by: Wiley on behalf of Accounting Research Center, Booth School of Business,
University of Chicago
Stable URL: http://www.jstor.org/stable/2672922
Accessed: 10-01-2018 16:52 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
Wiley is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Accounting Research
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
Journal of Accounting Research
Vol. 38 No. 1 Spring 2000
Printed in US.A.
Comparing the Accuracy and
Explainability of Dividend, Free
Cash Flow, and Abnormal Earnings
Equity Value Estimates
JENNIFER FRANCIS,* PER OLSSON,t
AND DENNIS R. OSWALD:
1. Introduction
This study provides empirical evidence on the reliability of intrinsic
value estimates derived from three theoretically equivalent valuation
models: the discounted dividend (DIV) model, the discounted free cash
flow (FCO) model, and the discounted abnormal earnings (AE) model.
We use Value Line (VL) annual forecasts of the elements in these models
to calculate value estimates for a sample of publicly traded firms fol-
lowed by Value Line during 1989-93.1 We contrast the reliability of value
*Duke University; tUniversity of Wisconsin; London Business School. This research
was supported by the Institute of Professional Accounting and the Graduate School of
Business at the University of Chicago, by the Bank Research Institute, Sweden, and Jan
Wallanders och Tom Hedelius Stiftelse for Samhallsvetenskaplig Forskning, Stockholm,
Sweden. We appreciate the comments and suggestions of workshop participants at the
1998 EAA meetings, Berkeley, Harvard, London Business School, London School of Eco-
nomics, NYU, Ohio State, Portland State, Rochester, Stockholm School of Economics,
Tilburg, and Wisconsin, and from Peter Easton, Frank Gigler, Paul Healy, Thomas Hem-
mer, Joakim Levin, Mark Mitchell, Krishna Palepu, Stephen Penman, Richard Ruback,
Linda Vincent, Terry Warfield, and Jerry Zimmerman.
I We collect third-quarter annual forecast data over a five-year forecast horizon for all
December year-end firms followed by VL in each of the years 1989-93. After excluding
firms with missing data, the final sample contains between 554 and 607 firms per year
(2,907 observations in the pooled sample).
45
Copyright ?, Institute of Professional Accounting, 2000
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
46 JOURNAL OF ACCOUNTING RESEARCH, SPRING 2000
estimates in terms of their accuracy (defined as the absolute price scaled
difference between the value estimate and the current security price)
and in terms of their explainability (defined as the ability of value esti-
mates to explain cross-sectional variation in current security prices).
In theory, the models yield identical estimates of intrinsic values; in
practice, they will differ if the forecasted attributes, growth rates, or dis-
count rates are inconsistent.2 Although by documenting significant dif-
ferences across DIM FCF, and AE value estimates our results speak to the
consistency question, our objective is to present a pragmatic exercise
comparing the reliability of these value estimates, recognizing that the
forecasts underlying them may be inconsistent. That is, we try to repli-
cate the typical situation facing an investor using a valuation model to
calculate an estimate of the intrinsic value of a firm. Under this view, the
empirical work addresses which series of forecasts investors seem to use
to value equity securities.
The results show that AE value estimates perform significantly better
than DIV or FCF value estimates. The median absolute prediction error
for the AE model is about three-quarters that of the FCF model (30%
versus 41%) and less than one-half that of the DIVmodel (30% versus
69%). Further, AE value estimates explain 71% of the variation in cur-
rent prices compared to 51% (35%) for DIV (FCF) value estimates. We
conclude that AEvalue estimates dominate value estimates based on free
cash flows or dividends.
Further analyses explore two explanations for the superiority of AE
value estimates. AEvalue estimates may be superior to DIVand FCFvalue
estimates when distortions in book values resulting from accounting pro-
cedures and accounting choices are less severe than forecast errors and
measurement errors in discount rates and growth rates. This effect is
potentially large for our sample, as indicated by the high proportion of
AE value estimates represented by book value of equity (72% on aver-
age) and the high proportion of FCF and DIV value estimates repre-
sented by terminal values (82% and 65%, on average, versus 21% for AE
value estimates).3 Value estimates may also differ when the precision and
the predictability of the fundamental attributes themselves differ. 4 Cet-
eris paribus, more precise and more predictable attributes should result
in more reliable value estimates. Tests of these conjectures suggest that
the greater reliability of AE value estimates is driven by the ability of
2For example, inconsistencies arise if the attributes violate clean surplus, if discount
rates violate the assumptions of no arbitrage, unlimited borrowing, and lending at the rate
of return, or if growth rates are not constant (i.e., the firm is not in steady state).
3We focus on the terminal value calculation because it is likely the noisiest component
of the value estimate, reflecting errors in forecasting the attribute itself, the growth rate,
and the discount factor.
4We define precision as the absolute difference between the predicted value of an attri-
bute and its realization, scaled by share price. We define predictability as the ease with
which market participants can forecast the attribute, and we measure this construct as the
standard deviation of historical year-to-year percentage changes in the attribute.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 47
book value to explain a large portion of intrinsic value and, perhaps, by
the greater precision and predictability of AE forecasts. Moreover, nei-
ther accounting discretion nor accounting conservatism has a significant
impact on the reliability of AEvalue estimates, suggesting that the supe-
riority of the AE measure is robust to differences in firms’ accounting
practices and policies.
To our knowledge, this is the first study to provide large-sample evi-
dence on the relative performance of these models using individual se-
curity value estimates based on forecast data. As discussed in section 2,
Penman and Sougiannis [1998] (henceforth PS) provide empirical eval-
uations of these models for a large sample of firms, for portfolio value
estimates based on realized attributes. The forecast versus realization dis-
tinction is important because realizations contain unpredictable compo-
nents which may confound comparisons of the valuations models (which
are based on expectations).5 PS use a portfolio design to average out the
unpredictable components of the valuation errors, whereas the use of
forecasts avoids this problem entirely and permits a focus on individual
securities’ valuation errors. Another important difference between the
two studies concerns the performance metrics: bias in PS and accuracy
and explainability in our study. PS focus on bias (we believe) because
their portfolio approach is better suited to describing the relation be-
tween value estimates and observed prices for the market as a whole.
Specifically, under a mean bias criterion, positive and negative predic-
tion errors offset within and across portfolios to yield estimates of the net
amount that portfolio value estimates deviate from observed prices. In
our individual security setting, we have no reason to believe that indi-
vidual shareholders care about net prediction errors or care more (or
less) about over- versus undervaluations of the same amount. Thus, we
believe accuracy rather than bias better reflects the loss function of an
investor valuing a given security. Explainability is also an open question
in an individual security setting but is not well motivated in a portfolio
setting where the random assignment of securities to portfolios and the
aggregation of value estimates and observed prices within the portfolio
significantly reduce the variation in these variables.
Our final analysis links the two studies by examining whether, for our
sample, their design yields the same results as our approach. We draw the
same conclusion as PS concerning bias in portfolio prediction errors
based on realizations: AEvalue estimates have smaller (in absolute terms)
5Realizations and forecasts also differ because realizations generally adhere to clean
surplus, but forecasted attributes may not. Over two-thirds of the sample forecasts adhere
to clean surplus in years 0, 1, and 3 but not in years 2, 4, and 5 because of the assumptions
used to construct a series of five-year forecasts (described in section 3). We do not believe
the differences across value estimates documented in this study are driven by violations of
clean surplus both because the violations of clean surplus are modest relative to the docu-
mented absolute prediction errors and because we find similar patterns when we repeat
our analyses using a one-year forecast horizon and include only those securities’ forecasts
which adhere to clean surplus.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
48 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
bias than FCF or DIV value estimates. However, when forecasts rather
than realizations are used to calculate value estimates, this ordering de-
pends on the assumed growth rate: for g= 0% we find the same ranking,
but for g = 4% we find that FCF value estimates have the smallest (abso-
lute) bias, followed by AE and DJVvalue estimates.6 In terms of accuracy,
we find that AE value estimates generally outperform FCF and DIV value
estimates regardless of whether forecasts or realizations are used. Abso-
lute prediction errors are, however, significantly (at the .00 level) smaller
when forecasts rather than realizations are used to calculate value esti-
mates. The forecast versus realization distinction is also important for
comparing DIV and FOF value estimates. While we find that FEF value es-
timates based on realizations are more biased than DIV value estimates
based on realizations (consistent with PS), we also find that FCFvalue es-
timates based on forecasts dominate DIVvalue estimates based on fore-
casts in terms of both bias and accuracy.
Section 2 describes the three valuation models and reviews the results
of prior studies’ investigations of estimates derived from these models.
Section 3 describes the sample and data and presents the formulations
of the DIV, FCF, and AE models we estimate. The empirical tests and re-
sults are reported in section 4, and section 5 reports the results of apply-
ing PS’s design to our sample firms. Section 6 summarizes the results
and concludes.
2. Valuation Methods
2.1 MODELS
The three equity valuation techniques considered in this paper build
on the notion that the market value of a share is the discounted value of
the expected future payoffs generated by the share. Although the three
models differ with respect to the payoff attribute considered, it can be
shown that (under certain conditions) the models yield theoretically
equivalent measures of intrinsic value.
The discounted dividend model, attributed to Williams [1938], equates
the value of a firm’s equity with the sum of the discounted expected div-
idend payments to shareholders over the life of the firm, with the termi-
nal value equal to the liquidating dividend:
T DIV
tF 1 (1+rE)t (1)
where:
VDIV = market value of equity at time F;
F = valuation date;
6For all other growth rates examined (2%, 6%, 8%, and 10%), we find that AEvalue es-
timates dominate FCFand DIVvalue estimates in terms of accuracy and smallest absolute bias.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 49
DJVt = forecasted dividends for year t;
rE = cost of equity capital; and
T = expected end of life of the firm (often T -o).
(For ease of notation, firm subscripts and expectation operators are
suppressed. All variables are to be interpreted as time F expectations
for firm j.)
The discounted free cash flow model substitutes free cash flows for divi-
dends, based on the assumption that free cash flows provide a better
representation of value added over a short horizon. Free cash flows
equal the cash available to the firm’s providers of capital after all re-
quired investments. In this paper, we follow the FCF model specified by
Copeland, Koller, and Murrin [1994]:7
T
C – ECE + ECMSF – DF – PSF (2)
VF t=1 (1+rwAcdt
FCFt = (SALESt – OPEXPt – DEPEXPt) (1-I)
+ DEPEXPt – A WCt – CAPEXPt (2 a)
rWACC = WD(l -)rD + wpSrpS + WErE (2b)
where:
VFCF = market value of equity at time F;
SALESt = sales revenues for year t;
OPEXPt = operating expenses for year t;
DEPEXPt = depreciation expense for year t;
AWCt = change in working capital in year t;
CAPEXPt = capital expenditures in year t;
ECMSt = excess cash and marketable securities at time t;8
Dt = market value of debt at time t;
PSt = market value of preferred stock at time t;
rWACC = weighted average cost of capital;
rD = cost of debt;
rps = cost of preferred stock;
WD = proportion of debt in target capital structure;
WpS = proportion of preferred stock in target capital structure;
WE = proportion of equity in target capital structure; and
X = corporate tax rate.
The discounted abnormal earnings model is based on valuation tech-
niques introduced by Preinreich [1938] and Edwards and Bell [1961],
7The FCF measure specified in equation (2a) is similar to Copeland, Koller, and Mur-
rin’s [1994] specification except we omit the change in deferred taxes because VL does
not forecast this item.
8Excess cash and marketable securities (ECMS) are the short-term cash and invest-
ments that the company holds over and above its target cash balances.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
50 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
and further developed by Ohlson [1995]. The AE model assumes an ac-
counting identity-the clean surplus relation (3b) -to express equity
values as a function of book values and abnormal earnings:9
T AEt
AE BE + (3)
t=1 (1 + rE)t
AEt = Xt- rEBt-I (3a)
Bt = Bt-I+XI-DVt (3b)
where:
VI’E = market value of equity at time F;
AEt = abnormal earnings in year t;
Bt = book value of equity at end of year t; and
Xt = earnings in year t.
2.2 PRIOR RESEARCH COMPARING ESTIMATES OF INTRINSIC VALUES
Several studies investigate the ability of one or more of these valua-
tion methods to generate reasonable estimates of market values. Kaplan
and Ruback [1995] provide evidence on the ability of discounted cash
flow estimates to explain transaction values for a sample of 51 firms en-
gaged in high leverage transactions.10 Their results indicate that the me-
dian cash flow value estimate is within 10O% of the market price, and that
cash flow estimates significantly outperform estimates based on compa-
rables or multiples approaches. Frankel and Lee [1995; 1996] find that
the AE value estimates explain a significantly larger portion of the varia-
tion in security prices than value estimates based on earnings, book val-
ues, or a combination of the two.
In addition to these horse races (which pit theoretically based value
estimates against one or more atheoretically based, but perhaps best prac-
tice, value estimates), there are at least two studies which contrast the el-
ements of, or the value estimates from, the DIV FCF, and/or AE models.
Bernard [1995] compares the ability of forecasted dividends and fore-
casted abnormal earnings to explain variation in current security prices.
Specifically, he regresses current stock price on current year, one-year-
ahead, and the average of the three- to five-year-ahead forecasted divi-
dends and contrasts the explanatory power of this model with the ex-
planatory power of the regression of current stock price on current book
value and current year, one-year-ahead and the average of three- to five-
year-ahead abnormal earnings forecasts. He finds that dividends explain
29% of the variation in stock prices, compared to 68% for the combina-
tion of current book value and abnormal earnings forecasts. Penman
and Sougiannis [1998] also compare dividend, cash flow, and abnormal
9 Clean surplus requires that any change in book value must flow through earnings. The
exception is dividends, which are defined net of capital contributions.
10 Transaction value equals the sum of the market value of common stock and preferred
stock, book value of debt not repaid as part of the transaction, repayment value of debt for
debt repaid, and transaction fees; less cash balances and marketable securities.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 51
earnings-based value estimates using infinite life assumptions. Using re-
alizations of the payoff attributes as proxies for expected values at the
valuation date, they estimate intrinsic values for horizons of T = 1 to T =
10 years, accounting for the value of the firm after time Tusing a termi-
nal value calculation. Regardless of the length of the horizon, PS find
that AE value estimates have significantly smaller (in absolute terms)
mean signed prediction errors than do FCF value estimates, with DIV
value estimates falling in between.
Our study extends previous investigations by comparing individual
securities’ DI, FCF, and AE value estimates calculated using ex ante data
for a large sample of publicly traded firms. In addition to evaluating
value estimates in terms of their accuracy (absolute deviation between
the value estimate and market price at the valuation date, scaled by the
latter), we contrast their ability to explain cross-sectional variation in
current market prices. Both metrics assume that forecasts reflect all avail-
able information and that valuation date securities prices are efficient
with respect to these forecasts. Under the accuracy metric, value estimates
with the smallest absolute forecast errors are the most reliable. The ex-
plainability tests-which compare value estimates in terms of their abil-
ity to explain cross-sectional variation in current market prices-control
for systematic over- or underestimation by the valuation models.”
3. Data and Model Specification
Our analyses require data on historical book values (from Compustat),
market prices (from CRSP), and proxies for the market’s expectations of
the fundamental attributes (from VL). VL data are preferred to other
analyst forecast sources (such as IIBIEIS or Zacks) because VL reports
contain a broader set of variables forecast over longer horizons than the
typical data provided by sell-side analysts. In particular, VL reports divi-
dend, earnings, book value, revenue, operating margin, capital expendi-
ture, working capital, and income tax rate forecasts for the current year
(t = 0), the following year (t = 1), and “3-5 years ahead.””2 Because the
valuation models require projected attributes for each period in the fore-
cast horizon, we assume that three- to five-year forecasts apply to all
years in that interval (results are not sensitive to this assumption). Also,
because VL does not- report two-year-ahead forecasts, we set year 2 fore-
casts equal to the average of the one-year-ahead and the three-year-
ahead forecast. We use data from third-quarter VL reports because this
is the first time data are reported for the complete five-year forecast
” In the OLS regression, bias is captured both by the inclusion of an intercept and by
allowing the coefficient relating the value estimate to current market price to deviate from
a theoretical value of one (bias which is correlated with the value estimate itself). Rank
regressions implicitly control for bias by using the ranks of the variables rather than the
values of the variables.
12 In contrast, IIBIEIS and Zacks contain, at most, analysts’ current-year and one-year-
ahead earnings forecasts (annual and quarterly) and an earnings growth rate.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
52 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
horizon; these reports have calendar dates ranging from F = July 1 to
September 30 (incrementing weekly) for each sample year, 1989-93. Fi-
nally, we restrict our analysis to December year-end firms to simplify
calculations.
VL publishes reports on about 1,700 firms every 13 weeks; 800-900 of
these firms have December year-ends. Because VL does not forecast all
of the inputs to the three valuation models for all firms (e.g., they do
not forecast capital expenditures for retail firms), the sample is reduced
to those firms with a complete set of forecasts. This requirement ex-
cludes about 250-300 firms each year, leaving a pooled sample of 3,085
firm-year observations (a firm appears at most once each year). Missing
Compustat and CRSP data reduce the sample to 2,907 firm-year observa-
tions, ranging from 554 to 607 firms annually. The sample firms are
large, with a mean market capitalization of $2.6 billion and a mean beta
of 0.97. Most of the sample firms are listed on either the NYSE or the
AMEX (82%), with the remainder trading on the NASDAQ
For each valuation model, we discount the forecasted fundamental
attributes to date F We adjust both for the horizon of the forecast (e.g.,
three years for a three-year-ahead forecast) and for a part-year factor,
f (f equals the number of days between F and December 31, divided by
365), to bring the current-year estimate back to the forecast date. We es-
timate discount rates using the following industry cost of equity model:’3
rE = rf + P[E(rm) – rf] (4)
where:
rE = industry-specific discount rate;
rf = intermediate-term Treasury bond yield minus the historical pre-
mium on Treasury bonds over Treasury bills (Ibbotson and
Sinquefield [1993]);
= estimate of the systematic risk for the industry to which firm j
belongs. Industry betas are calculated by averaging the firm-
specific betas of all sample firms in each two-digit SIC code.
Firm-specific betas are calculated using daily returns over fiscal
year t- 1;
E(rm) – rf = market risk premium = 6%.14
For a given firm and valuation date, we assume rE(rwAcc for the FCF
model) is constant across the forecast horizon. The average cost of equity
for the pooled sample is about 13%. The rwAcc calculation requires esti-
mates of rD, rps, capital structure (WD, wps, and WE), and ECMS. The cost
of debt is measured as the ratio of the VL reported interest on long-term
13Fama and French [1997] argue that industry costs of equity are more precise than
firm-specific costs of equity. Results using firm-specific discount rates yield similar infer-
ences and are not
reported.
14 SiX percent is advocated by Stewart [1991] and is similar to the 5-6% geometric
mean risk premium recommended by Copeland, Koller, and Murrin [1994]. We obtain
qualitatively similar results using the arithmetic average market risk premium.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 53
debt to the book value of long-term debt; the cost of preferred stock is
proxied by the VL reported preferred dividends divided by the book
value of preferred stock.’5 We set the pretax upper bound on the cost of
debt and the cost of preferred stock equal to the industry cost of equity,
and we set the pretax lower bound equal to the risk-free rate (results
are not sensitive to these boundary conditions). Following Copeland,
Koller, and Murrin [1994, pp. 241-42] we develop long-term target cap-
ital weights for the rWAcc formula rather than use the weights implied by
the capital structure at the valuation date.’6 For the pooled sample, the
mean cost of debt is 9.3%, the mean cost of preferred stock is 10.3%, and
the mean weighted average cost of capital is 11.8%. Based on Copeland,
Koller, and Murrin’s [1994, p. 161] suggestion that short-term cash and
investments above 0.5-2% of sales revenues are not necessary to support
operations, we define ECMS as cash and marketable securities in excess
of 2% of revenues.
We compute two terminal values for each valuation model, TVFUND,
where FUND = DIV EU’, or A. Both terminal values discount into per-
petuity the stream of forecasted fundamentals after T = 5; the first
specification assumes these fundamentals do not grow; the second as-
sumes they grow at 4%.17 If the forecasted T = 5 fundamental is nega-
tive, we set the terminal value to zero based on the assumption that the
firm will not survive if it continues to generate negative cash flows or
negative abnormal earnings (dividends cannot be less than zero). (The
results are not sensitive to this assumption.) Because we draw similar in-
ferences from the results based on the no growth and the 4% growth as-
sumptions, we discuss only the latter but report both sets of results in
the tables.18
15 VL reports book values of long-term debt and preferred stock as of the end of quarter
1. The results are not affected if we use Compustat data on book values of debt and pre-
ferred stock at the end of quarter 2. In theory, we should use the market values of debt
and preferred stock, but these data are not available.
16 Specifically, we use Value Line’s long-term (three- to five-year-ahead) predictions to
infer the long-term capital structure. We use the long-term price-earnings ratio multiplied
by the long-term earnings prediction to calculate the implied market value of equity five
years hence. For debt, we use VL’s long-term prediction of the book value of debt. For pre-
ferred stock, we assume that it remains unchanged from the valuation date. The equity
weight in the WACC formula, WE, is then given by WE = implied equity value/ (implied equity
value + forecasted debt + current book value of preferred stock). The debt and preferred
stock weights are calculated similarly.
17 The growth rate is often assumed to equal the rate of inflation. Consistent with
Kaplan and Ruback [1995] and Penman and Sougiannis [1998], we use a 4% growth rate.
We draw similar conclusions using growth rates of 2%, 6%, 8%, and 10%.
‘8We also examine a terminal value equal to VL’s long-term price projection (equal to
the VL three- to five-year-ahead price-earnings ratio multiplied by the three- to five-year-
ahead earnings forecast). All models perform extremely well using the inferred price ter-
minal value, with absolute (signed) prediction errors of 16-24% (5-14%) and adjusted
R2s of .77 to .91. Although the magnitudes of the differences are smaller, we find that AE
value estimates dominate FCF value estimates and perform at least as well as DIV value
estimates. Because long-term price forecasts are not available for most firms, we focus on
the more common scenario where terminal values must be calculated.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
54 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
Discounted dividend model specification:
5
VDIV = (1 + rE)-f .5D1V0 + E (1 + rE)( tf )DIVt
t= 1
+ (1 + rE) (5 +f)TVDv (5)
For the pooled sample, the average forecasted dividends for the sec-
ond half of the current year and the next five years are, on average,
$0.36, $0.76, $0.91, $1.05, $1.05, and $1.05. The mean terminal value
estimates for the pooled sample are $8.24 and $12.68 for the no growth
and 4% growth specifications, respectively.’9
Discounted free cash flow specification:
5
VFF F = (1 + rWAcc) 7.5FCF0 + E (1 + rWACC) ‘)FCFit
t=1
+ ( 1 + rWACC<) - (5 +f ) TVFCF + ECMSO - Do - PSO. (6)
The mean estimates of free cash flows for the remaining half of the
current year and the next five years for the pooled sample are $0.57,
$1.56, $0.99, $1.80, $3.98, and $3.98.20 The average terminal values for
the pooled sample are $34.59 (no growth) and $55.86 (4% growth).
Discounted abnormal earnings model specification:2′
VF4E = BQ2 + (1 + rE) f 5 (XO- rE x BQ2)
5
+ I (1 + rE)(t+f) [Xt – rE x Bt+ (1 +, -(5+f) TVAE. (7)
t = 1E
For the pooled sample, the average forecasted abnormal earnings for
the remainder of the current year and the next five years are $-0.05,
$0.33, $0.87, $1.14, $0.66, and $0.66.22 The terminal value estimates for
19 For the 564 (of 2,907) observations where the firm pays no dividends, the value esti-
mates equal zero. We retain these observations in the analysis, unless noted otherwise. Re-
sults excluding these 564 observations are similar to the full sample and are not reported.
20The FCF estimate for year 3 is different from years 4 and 5. For t = 3 the change in
working capital is based on the estimate of working capital in t = 2. For t = 4 and t = 5, the
change in working capital is zero because working capital forecasts are equal across t = 3,
t = 4, and t = 5 (recall that we assume that VL three- to five-year forecasts apply to eaclh.
year in that interval). This causes the FCF forecasts for years 4 and 5 to exceed the FCF
forecast for year 3.
21 We measure book value of equity at the end of Q2, year 0, BQ2. We obtain similar re-
sults using book value at the end of year -1.
22 The abnormal earnings estimate for year 3 is a function of the estimated book value
at the end of year 2. Hence, the estimate of abnormal earnings for year 3 differs from the
estimate of abnormal earnings for years 4 and 5 (which is a function of the constant book
value estimate for years 3-5).
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 55
TABLE 1
Pooled Sample Prediction Errors a
Panel A: Signed Prediction Errors (Bias))b
Mean % a-Level Median % a-Level
Mean Difference Difference = 0 Median Difference Difference =
0
Current Share Price 31.27 n/a n/a 25.12 n/a n/a
Value Estimate
DIV(g= 0%) 7.84 -75.5% 0.00 5.78 -75.8% 0.00
FCF(g= 0%) 18.40 -31.5% 0.00 13.79 -42.7% 0.00
AE (g= 0%) 22.04 -20.0% 0.00 17.91 -28.2% 0.00
DIV(g= 4%) 10.21 -68.0% 0.00 7.44 -68.7% 0.00
FCF (g= 4%) 30.02 18.2% 0.00 22.93 -8.8% 0.07
AE (g = 4%) 24.16 -12.7% 0.00 19.37 -22.9% 0.00
Panel B: Absolute Prediction Errors (Accuracy)c
Central
Value Estimate Median versus FCF versus AE Tendency
DIV(g= 0%) 75.8% 0.00 0.00 0.9%
FCF (g= 0%) 48.5% 0.00 13.2%
AE (g= 0%) 33.1% 20.2%
DIV(g= 4%) 69.1% 0.00 0.00 1.7%
FCF (g= 4%) 41.0% 0.00 18.4%
AE (g= 4%) 30.3% 22.5%
aThe sample securities are for December year-end firms with the following information available for any year
t = 1989-93: third-quarter Value Line forecasts of all fundamental values; Comvustat data on the book value of com-
mon equity for year t – 1; and CRSP security prices. P F = observed share price of security j on the Value Line fore-
cast date; VE-UND = security j’s estimate of intrinsic value based on FUND = dividends (DMV), free cash flows (FCF),
or abnormal earnings (AE). We calculate terminal values based on a no growth assumption and a 4% growth
assumption.
bPanel A reports mean and median signed prediction errors, equal to (ViUND – ,F)PF We also report the
significance level associated with the t-statistics (sign rank statistic) of whether the mean (median) prediction
error equals zero.
cPanel B shows the median absolute prediction error, IV/UND – PJFlIPjF, and the measure of central tendency
(the percentage of observations with value estimates within 15% of observed security pr-ice). The third and fourth
columns report the significance levels for Wilcoxon tests comparing the pooled sample median absolute predic-
tion errors for the noted row-column combination.
the pooled sample are, on average, $6.87 (no growth) and $10.74 (4%
growth).
4. Empirical Work
Panel A of table 1 reports mean and median security prices at the
valuation date and value estimates for the pooled sample.23 For all anal-
yses we set negative value estimates to zero, affecting 16 AE, 80 FCF, and
no DIVvalue estimates. We obtain similar results if we do not set these
estimates to zero. For comparison with Penman and Sougiannis’s results,
panel A shows information on signed prediction errors, (VFUND – P)IP
23 Results for individual years, and using prices five days after the valuation date (to en-
sure investors have fully impounded the information in VL analysts’ forecasts made at time
F), are similar and are not reported.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
56 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
Summary statistics show that all of the models tend to underestimate
security prices, with mean (median) signed prediction errors of -68%
(-69%) for VDIV 18% (-9%) for VFCF, and -13% (-23%) for VAE. The
frequency and magnitude of the underestimation is most severe for DIV
value estimates which are less than price 99% of the time (not reported
in table 1). Tests of the accuracy of the value estimates, reported in
panel B, show median absolute prediction errors along with a measure
of the central tendency of the value estimate distribution. Following
Kaplan and Ruback [1995] we define central tendency as the percent-
age of observations where the value estimate is within 15% of the ob-
served security price. The median accuracy of VAE of 30% is significantly
(at the .00 level) smaller than the median accuracy of VFCF (41%) and of
VDNV (69%). AE value estimates also show more central tendency than
FCF estimates (.22 versus .18); both of these models significantly out-
perform DIV estimates, where fewer than 2% of the observations are
within 15% of observed price.
We also examine the ability of the value estimates to explain cross-
sectional variation in securities prices. Panel A of table 2 reports R2s for
the OLS and rank univariate regressions of market price at time F on
each value estimate,24 and panel B reports the multivariate regressions
of price on VDIV VFCF, and VAE (the full model). The explained vari-
ability of the rank univariate regressions is high for all three valuation
models (between 77% and 90%); however, the OLS results show greater
variation in R2s-between 35% and 71% -with FCFvalue estimates per-
forming substantially worse than AE or DIV value estimates. In particu-
lar, VFCF explains about one-half (two-thirds) of the variation in price
explained by VAE(VDNV).25 Results in panel B calibrate the incremental
importance of each value estimate by decomposing the explanatory
power of the full model into the portion explained by each value esti-
mate controlling for the other two.26 For example, the incremental ex-
24 OLS regressions include an intercept; rank regressions do not. We report OLS results
after deleting observations with studentized residuals in excess of two; this rule eliminates
between 44 and 105 observations for each model. Results based on the full sample are
similar to those reported with one exception: when all observations are retained, the ex-
planatory power of DfVestimates is 22% (versus 51% when outliers are deleted).
25 If the value estimates are unbiased predictors of market security prices, then the in-
tercept (X0) should equal zero and the coefficient relating value estimate to price (X1)
should be one. In all cases, we reject the joint hypothesis that X0 = 0 and XI = 1. Because
these rejections may arise from heteroscedasticity (for the DIVand FCF estimates-but not
the AE estimates-White [1980] tests reject the hypothesis that the variance of the distur-
bance term is constant across observations), we repeat all analyses after transforming the
variables to eliminate the heteroscedasticity. The transformed results (not reported) show
small changes in the parameter estimates; in all cases the results are qualitatively similar to
the untransformed results reported in the tables.
26 For g = 4%, we are unable to reject the joint hypothesis that the intercept equals zero
and the sum of the slope coefficients equals one. For g = 0%, we reject at the .08 level.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 57
TABLE 2
Results of Pooled Sample Regressions of Contemporaneous Stock Prices on Intrinsic Value Estimates a
Panel A: Univariate Regressions of Price on Value Estimateb
Growth Rate = 0% Growth Rate = 4%
DIV FCF AE DIV FCF AE
OLS Coefficient 1.75 0.76 1.23 1.30 0.46 1.09
OLS R2 0.54 0.40 0.73 0.51 0.35 0.71
Rank R2 0.84 0.77 0.90 0.84 0.77 0.90
Panel B: Multivariate Regressions of Price on Value Estimatesc
Growth Rate = 0% Growth Rate = 4%
DIV FCF AE DIV FCF AE
OLS Coefficient 0.16 0.10 1.04 0.04 -0.02 1.06
t-Statistic OLS Coefficient = 0 2.04 4.93 22.18 0.53 -1.14 22.20
t-Statistic Rank Coefficient = 0 10.99 9.67 34.00 11.19 3.95 33.87
Model OLSR2 0.73 0.71
Model Rank R2 0.91 0.91
Incremental OLSR2 0.01 0.00 0.12 0.00 0.00 0.14
Incremental Rank R2 0.00 0.00 0.04 0.00 0.00 0.04
aSee n. a to table 1 for the sample description and the calculations of value estimates.
bPanel A reports results of estimating the following regression: P F = 0 + X1 VfUND + ?j, where =
observed share price of security j on the Value Line forecast date; VPIND = value estimate for security j
for FUND = dividends (DIV), free cash flows (FCF), or abnormal earnings (AE).
cPanel B shows results of estimating the following regression: Pj F= N + p1 VjDI + pt 2ifC + t 3VjAE +
?j The last two rows in panel B show the incremental adjusted R2 provided by the noted value esti-
mate, beyond that provided by the other two value estimates. For the OLS regression, we report White
[1980] adjusted t-statistics. The incremental adjusted R2 is the difference between the adjusted R2 for
the OLS (rank) regression containing all three value estimates and the adjusted R2 for the OLS (rank)
regression which excludes the value estimate in the noted column.
planatory power of VDIV equals the adjusted R2 from the full model
minus the adjusted R2 from the regression of price on VFCF and VAE.
Controlling for VFCF and VDIV VAE adds 14% explanatory power for the
OLS regressions and 4% for the rank regressions. In contrast, neither
VFCF nor VDIV adds much (0-1% incremental adjusted R2) to explaining
variation in security prices.
In summary, the results in tables 1 and 2 indicate that AE value esti-
mates dominate DIV and FCF value estimates in terms of accuracy and
explainability. One explanation for this superiority is that differences in
reliability stem from the AE model containing both a stock component
(BF) and a flow component (AE.), whereas the DIV and FCF models are
pure flow-based models. AE value estimates will dominate DIV and FCF
value estimates when biases in book values resulting from accounting
procedures (such as expensing R&D) or accounting choices (such as a
firm’s accrual practices) are less severe than errors in forecasting at-
tributes and errors in estimating discount rates and growth rates. As an
indication of the potential severity of this issue, we note that book value
of equity represents 72% of the sample mean VAE, and that the terminal
value represents 21%, 65%, and 82% of the mean VAI, VDIV and VFU, re-
spectively. Thus, biases in measuring book values may substantially affect
AE value estimates (but have no effect on DIV or FCF value estimates),
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
58 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
TABLE 3
Results of Pooled Sample Regressions of Contemporaneous Stock Prices
on the Components of Value Estimates a
Panel A: DIVModelb
Growth Rate = 0% Growth Rate = 4%
PV DTV PV DTV
OLS Coefficient 6.47 -2.04 5.64 -0.87
t-Statistic: OLSCoefficient= 1 15.41 -10.66 21.91 -18.45
t-Statistic: Rank Coefficient = 0 9.52 -0.34 12.86 -0.80
Model OLS R2 0.57 0.57
Model Rank R2 0.84 0.84
Incremental OLSR2 0.05 0.00 0.10 0.01
Incremental Rank R2 0.00 0.00 0.01 0.00
Panel B: FCF Model c
Growth Rate = 0% Growth Rate = 4%
NFA PV DTV NFA PV DTV
OLS Coefficient 0.26 0.36 0.71 0.19 0.84 0.21
t-Statistic: OLS Coefficient= 1 -21.30 -6.48 -4.87 -22.97 -1.59 -20.73
t-Statistic: Rank Coefficient = 0 26.94 10.00 15.61 27.03 13.31 12.98
Model OLSR2 0.35 0.32
Model Rank R2 0.82 0.82
Incremental OLSR2 0.04 0.01 0.05 0.04 0.05 0.03
Incremental Rank R2 0.04 0.01 0.12 0.05 0.01 0.12
Panel C: AE Model d
Growth Rate = 0% Growth Rate = 4%
B PV DTV B PV DTV
OLS Coefficient 1.24 2.99 -0.48 1.24 3.05 -0.31
t-Statistic: OLSCoefficient= 1 12.36 19.96 -15.82 12.51 22.73 -28.03
t-Statistic: Rank Coefficient = 0 66.84 22.64 -4.95 67.03 23.75 -5.52
Model OLSR2 0.74 0.74
Model Rank R2 0.91 0.91
Incremental OLSR2 0.45 0.11 0.00 0.45 0.14 0.00
Incremental Rank R2 0.15 0.02 0.00 0.15 0.02 0.00
aSee n. a to table 1 for a description of the sample and the calculations of value estimates and ter-
minal values.
bPanel A reports coefficient estimates and White-adjusted t-statistics for the following regression:
Pj, F = +OO + co I PVPff + co DTVD/v , where PjF = observed share price of security j on the Value Line
forecast date; PVjPJ{ = the present value of the five-year stream of forecasted dividends; DTVDII/ = dis
counted (to time F) value of the terminal value for the noted specification.
cPanel B reports coefficient estimates and White-adjusted t-statistics for the following regression:
Pi, F = Co + AFAf(f + 2PVFFf + 03DTVF7(f + ?j ,, where PjF = observed share price of security j on the
Value Line forecast date; NFAjF = net financial assets at the valuation date (excess cash and marketable
securities – debt – preferred stock); PVF’T the present value of the five-year stream of forecasted free
cash flows; DTVjDWr^ = discounted (to time F) value of the terminal value for the noted specification.
dPanel C reports coefficient estimates and White-adjusted t-statistics for the following regression:
PjFx = CO + (oBjF + (02PVjMjF + o3DT’ +DT j where P. E = observed share price of security j on the Value
Line forecast date; By= book value of equity at the valuation date; PV; – the present value of the five-
year stream of forecasted dividends; DTV’AE, discounted (to time F) value of the terminal value for the
noted specification.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 59
while errors in forecasting flows and estimating discount rates and growth
rates are likely to have a bigger effect on VDIV and VFcF than on V I.
Table 3 provides indirect evidence on the stock versus flow distinction
by examining the incremental explanatory power of the components of
each value estimate:27
= + (0pVDIV + v2DTVJD; + ?, (8)
P’ F = to + wo1FAA + o2PVCf + o3DTV;f + ?j, (9)
P. F = WO + WJBj + PV + 3DTV7F + (10)
where:
PVFUND = the present value of the five-year stream of the fore-
I
casted attribute;
DTV;FUN = the discounted (to time F) value of the terminal value;
and
NFAj = net financial assets at time F = ECMS – D – PS.
For each regression, we report White-adjusted t-statistics of whether
the estimate differs from its theoretical value of one, the adjusted R2 for
each equation and model, and the additional explanatory power added by
each component of the model holding constant the other component(s).
Turning first to the coefficient estimates, we note that for all three mod-
els, we strongly reject the null hypothesis that the coefficient relating
DTVFUND to price equals one, suggesting that none of the terminal values
is well specified. For the AE model, the results also reject the hypothesis
that the coefficient relating price to book value is one, although in all
cases 0h is significantly positive. Comparing OLS [rank] results across the
three panels, we note that the incremental explanatory power provided
by PVDIV of 10% [1 %] and by PVFC-F of 5% [1 %] is substantially less than
the 45% [15%] explanatory power provided by book value alone in the
AE model. Book value also adds more than either PVAE-14% [2%] or
DTVAE o% [1%]. Overall, these results suggest that, despite conserva-
tism in its measurement, book value of equity explains a significant por-
tion of the variation in observed prices.
Our second analysis of the stock versus flow explanation focuses on
situations where we might expect accounting practices to result in book
values that are biased estimates of market value. On the one hand, we
expect that when the current book value of equity does a good job of
recording the intrinsic value of the firm, AE value estimates are more
27 The terminal value component equals the present value, at the valuation date, of the
estimated terminal value five years hence. To be consistent with the FCF model specified by
equation (2), we include net financial assets (NFA), equal to excess cash and marketable
securities minus debt minus preferred stock, as a component in the FCF model.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
60 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
reliable-than FCF or DIV value estimates-because more of intrinsic
value is included in the forecast horizon (and therefore less in the ter-
minal value calculation). On the other hand, even if book values exclude
value-relevant assets, the AE model’s articulation of the balance sheet
and the income statement will link lower book values today with larger
abnormal earnings in future periods. For example, if the net R&D payoff
component of earnings is stable through time (as we expect in equilib-
rium), then the sum of current book value of equity and the discounted
stream of abnormal earnings will result in the same estimate of intrinsic
value if R&D investments were capitalized at their net present values
(Bernard [1995, n. 9] makes a similar point with respect to accounting
distortions which result in overstatements of book values).
We test whether the AE model performs differently for firms with high
R&D spending than for firms with low or no R&D spending. We iden-
tify a sample of High R&D firms by first ranking the sample firms based
on the ratio of Compustat R&D spending in year t – 1 to total assets at
the beginning of year t – 1. About 48% (1,390 firm-year observations) of
the sample disclose no, or immaterial amounts of, R&D expenditures
(the Low R&D sample); the top 25% of firms (the High R&D sample)
have mean annual R&D spending of 7.2% of total assets. Table 4, panel
A reports the results of accuracy comparisons; table 5, panel A shows the
results of the explainability tests. These findings show no evidence that,
for the High R&D sample, AE value estimates are less reliable than DIV
or FCF estimates; in fact, they are significantly more accurate and explain
more of the variation in security prices. Within-model, across-partition
comparisons of accuracy (far right column of table 4) show no differ-
ence in the accuracy of AE value estimates for High versus Low R&D
firms. Differences in R2s between the High and Low R&D samples (panel
A, table 5) indicate that the AE model performs better, not worse, for
High R&D firms.28
We also partition the sample based on the ability of firms to affect the
flow component of the AE model. Unlike free cash flows and dividends,
management can influence the timing of abnormal earnings by exer-
cising more or less discretion in their accrual practices. Whether such
discretion leads to AE value estimates being more or less reliable mea-
sures of market prices depends on whether management uses account-
ing discretion to clarify or obfuscate value-relevant information. Because
we have no a priori reason for believing that one effect dominates the
other in explaining the accrual behavior of the sample firms, we do not
28 Our results concerning the accuracy of AE value estimates for Highl versus Low R&D
firms may be sensitive to our sample of large, relatively stable firms. For their broader sam-
ple, Sougiannis and Yaekura [1997] find that absolute prediction errors for AE value esti-
mates increase with the amount of R&D spending, and Barth, Kasznik, and McNichols
[forthcoming] report a significant negative relation between R&D spending and signed pre-
diction errors (they define the prediction error as price minus value estimate, scaled by price).
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
TABLE 4
Comparison of the Accuracy of Value Estimates Across and Within Sample Partitionsa
Panel A: R&D (as Percentage of Total Assets)
High R&D Sampleb Low R&D Sampleb
versus versus versus versus High versus Low
Value Estimate Median FCF AE Median FCF AE Difference
DIV(g= 0%) 77.5% 0.00 0.00 78.2% 0.00 0.00 0.01
FCF(g= 0%) 46.1% 0.00 49.2% 0.00 0.00
AE (g= 0%) 35.0% 33.9% 0.35
DJV(g= 4%) 71.8% 0.00 0.00 71.9% 0.00 0.00 0.01
FCF(g= 4%) 33.7% 0.00 45.7% 0.00 O?O?
AE (g= 4%) 30.9% 32.0% 0.36
Panel B: Accruals (as Percentage of Total Assets)
High Accrual Samplec Low Accrual Samplec
versus versus versus versus High versus Low
Value Estimate Median FCF AE Median FCF AE Difference
DJV(g= 0%) 78.7% 0.00 0.00 77.0% 0.00 0.00 0.09
FCF (g= 0%) 48.6% 0.00 46.8% 0.00 0.40
AE (g= 0%) 32.9% 34.9% 0.19
DIV(g= 4%) 72.7% 0.00 0.00 71.4% 0.00 0.00 0.34
FCF (g= 4%) 41.9% 0.00 38.3% 0.00 0.17
AE (g= 4%) 29.8% 32.0% 0.38
Panel C: Precision of Attribute
High Precision Sampled Low Precision Sampled
versus versus versus versus High versus Low
Value Estimate Median FCF AE Median FCF AE Difference
DJV(g= 0%) 100.0% 0.00 0.00 67.0% 0.00 0.00 0.00
FCF (g= 0%) 50.5% 0.00 49.3% 0.00 0.67
AE (g= 0%) 43.8% 23.8% 0.00
DIV1l(g= 4%) 100.0% 0.00 0.00 58.5% 0.00 0.00 0.00
FCF (g= 4%) 34.8% 0.36 61.2% 0.00 0.00
AE (g= 4%) 39.3% 24.4% 0.00
Panel D: Predictability of Attribute
High Predictability Samplee Low Predictability Samplee
versus versus versus versus High versus Low
Value Estimate Median FCF AE Median FCF AE Difference
DIV(g= 0%) 71.2% 0.00 0.00 77.1% .0.00 0.00 0.00
FCF (g= 0%) 48.0% 0.00 48.7% 0.00 0.28
AE (g= 0%) 37.4% 31.0% 0.00
DJV(g= 4%) 63.7% 0.00 0.00 72.0% 0.00 0.00 0.00
FCF(g= 4%) 42.0% 0.00 39.0% 0.00 0.40
AE (g= 4%) 32.4% 29.4% 0.08
aSee n. a to table 1 for a description of the sample and the calculations of value estimates and terminal val-
ues. In the columns labeled “versus FCF(AE) ” we report significance levels comparing the median absolute pre-
dliction errors of the noted variables. The column labeled “High versus Low Difference” shows the significance
level for the Wilcoxon test for whether the accuracy of the noted High sample differs from the accuracy of the
Low sample.
bThe High R&D sample consists of the firms in the top quartile of R&D expenses as a percentage of total
assets, measured in year t – 1; the Low R&D sample contains all firms with no disclosed research and develop-
ment expenses in year t – 1.
cThe High (Low) Accrual sample consists of the top (bottom) quartile of firms ranked on the absolute value
of total accruals as a percentage of total assets, measured in year t- 1.
11Precision equals the absolute value of the difference between the forecast attribute and its realization,
scaled by forecast attribute. The Low (High) Precision sample for each FCF(AE) attribute consists of the top
(bottom) quartile of firms ranked on the average precision of the current-year, one-year-ahead, and three-year-
ahead forecasts of that attribute.
ePredictability equals the standard deviation of percentage yearly changes in the historical realized values of
the attribute valued by each model. The Low (High) Predictability sample for each FCF(AE) attribute consists
of the top (bottom) quartile of firms ranked on the predictability of that attribute.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
62 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
TABLE 5
Comparisons of the Explainability of Value Estimates Across and Within Sample Partitionsa
Panel A: R&D (as Percentage of Total Assets)
High R&D Sample Low R&D Sample
Growth Rate = 0% Growth Rate = 4% Growth rate = 0% Growth Rate = 4%
DIV FCF AE DIV FCF AE DIV FCF AE DIV FCF AE
OLS Coefficient 2.13 1.02 1.26 1.69 0.64 1.15 1.34 0.59 1.34 0.90 0.34 1.06
OLS R2 0.75 0.58 0.79 0.74 0.51 0.81 0.32 0.29 0.71 0.28 0.22 0.62
Rank R2 0.87 0.87 0.94 0.87 0.86 0.94 0.81 0.72 0.88 0.81 0.71 0.88
Panel B: Accruals (as Percentage of Total Assets)
High Accrual Sample Low Accrual Sample
Growth Rate = 0% Growth Rate = 4% Growth Rate = 0% Growth Rate = 4%
DIV FCd AE DIV FCF AE DIV FCF AE DIV FCF AE
OLSCoefficient 1.89 0.81 1.26 1.44 0.52 1.13 1.47 0.82 1.21 0.98 0.52 1.04
OLS R2 0.56 0.40 0.75 0.54 0.34 0.72 0.40 0.41 0.66 0.35 0.35 0.63
Rank R2 0.84 0.77 0.91 0.84 0.76 0.90 0.82 0.81 0.89 0.82 0.79 0.89
Panel C: Precision of Attribute
High Precision Sample Low Precision Sample
Growth Rate = 0% Growth Rate = 4% Growth Rate = 0% Growth Rate = 4%
DIV FCF AE DIV FCF AE DVX FCF AE DIV FCF AE
OLSCoefficient 1.82 1.09 1.36 1.34 0.54 1.14 2.11 0.46 0.92 1.55 0.25 0.78
OLS R2 0.32 0.58 0.72 0.30 0.79 0.66 0.70 0.39 0.80 0.66 0.32 0.78
Rank R2 0.80 0.79 0.91 0.80 0.78 0.90 0.92 0.74 0.93 0.92 0.75 0.93
Panel D: Predictability of Attribute
High Predictability Sample Low Predictability Sample
Growth Rate = 0% Growth Rate = 4% Growth Rate = 0% Growth Rate = 4%
DIV FCF AE DIV FCF AE DIV FCF AE DIV FCF AE
OLSCoefficient 1.69 0.60 1.43 1.27 0.33 1.21 0.90 1.21 1.15 0.54 0.49 1.04
OLS R2 0.60 0.42 0.71 0.58 0.34 0.68 0.31 0.77 0.72 0.26 0.31 0.71
Rank R2 0.86 0.77 0.91 0.86 0.78 0.91 0.82 0.77 0.92 0.82 0.76 0.91
aSee n. a to table 1 for a description of the sample and the calculations of value estimates and terminal values. We
report regression results of observed price on the value estimates for each sample partition. See table 4 for a descrip-
tion of the sample partitions.
predict whether AE value estimates perform better or worse for firms
with high accounting discretion. We partition firms based on the level of
accounting discretion available to firms, as proxied by the ratio of the ab-
solute value of the ratio of total accruals to total assets (Healy [1985]).29
Securities in the top (bottom) quartile of the ranked distribution are
assigned to the High (Low) Accruals sample. The mean (median) ratio of
accruals to assets for the High Accruals sample is 14% (12%); for the
Low Accruals sample the mean and the median value is 1.5%. Table 4,
panel B summarizes the accuracy of the value estimates for the accruals
29 Total accruals equal change in current assets (Compustat item #4) – change in cur-
rent liabilities (#5) – change in cash (#1) + change in short-term debt (#34) – depre-
ciation (#16).
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 63
subsamples; table 5, panel B shows similar comparisons for the explain-
ability measure. Within the High Accruals sample, both measures show
that VAE is superior to VDNv and VFCF Comparisons of AE estimates be-
tween the High and Low Accrual samples show no evidence that AE es-
timates perform worse for the High Accrual sample than for the Low
Accrual sample.
In summary, we find no evidence that AE estimates are less reliable
for firms where book values poorly reflect intrinsic values (firms with
high R&D spending) or for firms where there is scope for managing
earnings (firms with high accruals). If anything, the within-sample and
across-sample tests indicate that high R&D spending and high account-
ing discretion are associated with more reliable AE value estimates.30
A second potential explanation for differences in the reliability of VDJ’Y
VFCF, and VAIE is that the precision and predictability of the fundamental
attributes themselves differ. We measure precision as the absolute differ-
ence between the predicted value of an attribute and its realization,
scaled by share price.31 We also examine the bias in the fundamental at-
tributes, measured as the signed difference between the predicted value
and its realization, scaled by share price. We define predictability as the
ease with which market participants can forecast the attribute, measured
as the standard deviation of historical year-to-year percentage changes
in the attribute.32 Ceteris paribus, more precise and more predictable
attributes should result in more accurate value estimates which explain
a greater portion of the variation in observed prices.
We compute bias and precision statistics for each of the current year,
one-year-ahead and three- to five-year-ahead forecasted attributes;33 me-
dian values are reported in table 6, panel A. Bias measures indicate that
30 We also partition the sample securities by capital expenditure spending to investigate
whether FCFvalue estimates outperform AE and DJVvalue estimates when forecasted cap-
ital expenditures are low. (We thank Peter Easton for suggesting this analysis.) Results (not
reported) show no evidence that FCF value estimates are more accurate or explain more
variation in prices than do AE value estimates for the LOW capital expenditure sample.
Across-sample, within-model tests, however, show some evidence that FCF value estimates
are more accurate for the LOW capital expenditure sample than for the HIGH capital ex-
penditure sample.
31 We find similar results if we scale by the absolute value of the predicted attribute.
32 Results based on the coefficient of variation of yearly changes are similar and are not
reported.
33We use the three-year-ahead realization to measure the bias in the three- to five-year-
ahead forecast of each attribute. The realized dividend for year t equals the total amount
of common stock dividends declared in year t (#121). The realized free cash flow per
share in year t equals the net cash flow from operating activities (#308) minus capital
expenditures (#128). The realized abnormal earnings for year t equal earnings per share
after extraordinary items (#53) minus the estimated discount rate multiplied by the book
value of common equity in year t – 1 (#60). To ensure consistency across models, we scale
all variables by the number of shares used to calculate primary earnings per share (#54),
and we delete observations with missing data for dividends, free cash flow, or abnormal
earnings.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
64 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
TABLE 6
Comparison of Selected Properties of the Forecast Attributes a
Panel A: Bias and Precisionb
Bias Precision
Wilcoxon Testsb Wilcoxon Testsb
Median versus FCF versus AE Median versus FCF versus AE
Current-Year
DIV 0.00% 0.00 0.00 0.00% 0.00 0.00
FCF 0.85% 0.00 4.88% 0.00
AE 0.65% 1.32%
One-Year-Ahead
DIV 0.01% 0.00 0.00 0.09% 0.00 0.00
FCF 1.75% 0.00 5.93% 0.00
AE 2.22% 2.85%
Three-Year-Ahead
DIV 0.46% 0.00 0.00 0.57% 0.00 0.00
FCF 4.72% 0.00 7.76% 0.01
AE 4.54% 5.10%
Panel B: Predictabilityb
Wilcoxon Tests
Attribute Median versus FCF versus AE
DIV 0.22 0.00 0.00
FCF 7.72 0.00
AE 3.64
aSee n. a to table 1 for a description of the sample and the calculations of value estimates and ter-
minal values.
bBias (precision) equals the signed (absolute value of the) difference between the forecast attribute
and its realization, scaled by the share price. Predictability equals the standard deviation of percent-
age yearly changes in the historical realized values of the attribute valued by each model. We report
median bias, precision, and predictability measures as well as the significance levels for Wilcoxon tests
comparing these statistics across models.
the median current-year AE (FCF) forecast overstates realized abnormal
earnings by about 0.6% (0.8%) of security price, with current-year DIV
forecasts showing no bias. For all attributes, forecast optimism increases
with the forecast horizon: the median one-year-ahead AE (FCF) forecast
overstates its realization by about 2.2% (1.8%) of price, compared to 4.5
(4.7%) for the three-year-ahead AE (FCF) forecasts. More importantly,
we find that for all horizons, AE forecasts are significantly more accurate
than FCF forecasts, with AE prediction errors ranging from roughly 25%
of FCFprediction errors for current-year forecasts (1.3% versus 4.9%) to
65% for three-year-ahead forecasts (5.1% versus 7.8%). The finding that
DIV forecasts are the most precisely forecasted attribute is to be ex-
pected given firms’ reluctance to alter dividend policies.
For each fundamental attribute, we average the precision of current-
year, one-year-ahead and three-year-ahead forecasts and identify those
observations in the top quartile of average precision (i.e., those with the
largest percentage differences between forecasts and realizations) as the
Low Precision sample and those observations in the bottom quartile of
average precision as the High Precision sample. Comparisons of the accu-
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 65
racy and explainability of the value estimates for precision subsamples
are shown in panel C of tables 4 and 5. With the exception of the accu-
racy of VFCF (4% growth rate), there is no evidence that more precise
forecasts result in more reliable value estimates; if anything, we find the
opposite. For the DIV model, this result is not unexpected, since for a
large portion of the sample firms, VDNV understates intrinsic values (as
shown in table 1);34 in these cases, the slight average optimism in DTV
forecasts (documented in table 6) improves the accuracy of VDIY Simi-
larly, the optimism in AE forecasts compensates for the underestimation
by VAE observed in table 1, resulting in more reliable AE value estimates
for the Low Precision partition than for the High Precision partition.
We also partition the sample based on the standard deviation of the
percentage changes in the attribute; for these calculations, we require
a minimum of ten annual changes in realized dividends, free cash flows,
and abnormal earnings. Table 6, panel B reports median values of the
predictability measure for each model; we also report comparisons of
predictability between each pair of models. Consistent with firms making
few changes in dividend payments and policies, we find that dividends
are highly predictable. Of more interest (we believe) is the finding that
abnormal earnings are significantly (at the .00 level) more predictable
than free cash flows. To assess the importance of predictability on the re-
liability of the value estimates, we rank each set of value estimates based
on the magnitude of the relevant predictability measure and repeat our
tests on the bottom quartile (High Predictability sample) and on the top
quartile (Low Predictability sample). The results in panel D of tables 4
and 5 show no evidence that a more predictable AE or FCF series leads
to significantly more reliable value estimates; however, we do observe
this pattern for the DIV series.
Overall, the results in tables 4-6 provide mixed evidence on whether
the precision and the predictability of the attribute valued by each
model are important determinants of the reliability of the value esti-
mates. Consistent with the AE model’s relative superiority over the FCF
model, we find that AE forecasts are generally more precise and more
predictable than FCF forecasts. However, within-model tests show no
consistent evidence that securities with the most precise or the most
predictable forecasts have more reliable value estimates than do securi-
ties with the least precise or the least predictable forecasts.
5. Comparison to Penman and Sougiannis [1998]
Penman and Sougiannis [1998] compare the signed prediction errors
of DIV FCF, and AE value estimates calculated using realized values of
34This is certainly true for the 20% of firms in our sample which do not pay dividends.
Even for dividend-paying firms, DlVestimates likely understate value because our terminal
value calculations do not include a liquidating dividend.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
66 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
these attributes.35 Although both PS and we conclude that VAE domi-
nates VDNv and VFCF, the studies differ in several respects: PS’s sample is
larger and more diversified than our sample of (median and large) VL
firms; PS’s approach uses realizations and a portfolio averaging process,
while our design examines analysts’ forecasts for individual securities;
PS evaluate bias, while we focus on accuracy and explainability. We link
the two studies by examining, for our sample, the effects of using fore-
casts versus realizations, the portfolio methodology, and other perfor-
mance metrics.
For each firm-year observation with available data, we collect realized
values of DIV FCF, and AE from the 1997 Compustat tape (which includes
fiscal years through 1996). Because we have a five-year forecast horizon,
we are limited to analyzing years 1989, 1990, and 1991. For each sample
year, we randomly assign the sample securities to ten portfolios and cal-
culate the average portfolio value of each attribute for each year of the
horizon. We then discount (at the average discount rate) the mean val-
ues of the attributes to the average valuation date to arrive at a mean
value estimate for each portfolio. We perform this analysis using both
forecasts and realizations, so that we have both an ex ante and an ex
post mean value estimate for each portfolio to compare to the mean
portfolio price. We believe the calculation of the mean value estimates
follows that of PS, except that we have fewer portfolios (30 versus 400)
and fewer firms per portfolio (50-60 versus about 200).
Panel A of table 7 shows the median value estimates and median bias
for the 30 portfolios.36 We report the median signed prediction error
for portfolio value estimates based on realizations (“ex post” value esti-
mates), for portfolio value estimates based on forecasts (“ex ante” value
estimates), and for these same securities’ value estimates calculated us-
ing forecasts and the individual security approach. A comparison of the
ex post portfolio and ex ante portfolio results shows the effects of using
realizations versus forecasts, holding constant the methodology; a com-
parison of ex ante portfolio value estimates with ex ante individual secu-
rity value estimates highlights the effects of the portfolio methodology,
controlling for the use of forecast data. Panel B shows similar compari-
sons for the accuracy metric.37
We draw the following conclusions from the results in table 7. First, ex
post value estimates are considerably smaller than ex ante value estimates,
35 For the specification closest to ours, PS report mean biases of -17% (VDII’), -76%
(VF’C), and 6% (VAE); these compare to our sample mean bias measures of -68% (VDV),
18% (VFCF), and -13% (VAE), reported in table 1.
36We report median statistics for consistency with tables 1 and 4. Results based on
means are similar.
37We do not examine the explainability metric for the portfolio measures because the
point of the portfolio averaging process-to reduce the variability in observed prices and
in value estimates-runs counter to the point of the explainability tests-to assess the ex-
tent to which value estimates explain variation in observed prices.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC
All use subject to http://about.jstor.org/terms
COMPARING EARNINGS EQUITY VALUE ESTIMATES 67
? Cun? C s~
-z Q X m bl*
u u~~~~~~~~~~-
.?~~~~C t 0 Q ~ D=
> r 2 o II 1 0 Q g = 20 < .QO*
>~~ ~~~~~~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > O<
Q Q0
*vi > Q X; O O O O O O > Q X; O O O O O O Q Q Q
C.0 ~ ~ ~ ~ ~
t11 U I AI 2iooooo 1 ???? ? ? I= ‘o – S2S
>~~~~~~~~~L on 11l O 0.000 . o G.I t-c) N c C,
– =gge -e -z nOO ;6.Q
g~~~~~~~~~~~~~~~~~\ 0\ >o “t o El
0 0 C)
.S ~ ~~~ ~ ~~~~~~~~~~~~~~~~~ S .~ 2
u R cL 8 cL E X 0 t X 2 0 ~~~ Q ~ 6 6
W ;4 < U U g g g g g g t < t ON ONt- " on O a tE
~~~~~~ ~~~~~~~~~
t~~~~~~~~~~~~~~~~~~~~e > dI SI;DO :t.om dIXtI t .0 – -0 u Lu u CC) C ) L ) n 0 0 , <
i~~~~~~~~~~0 C) in – C) i ?n 0
b -C’G
~~ 666 666~~~~~a 66 066 14 u S 2
? 00 C) in r C) m > > *- 4> 11 11 0 11 11 t X F ? b.O
0 | ,0; t t 3 t t; 3 tm ; | t ;t 23 t | >3 1 uD 4~ E o; =; b
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 m Q Q C. C
– Q ~~~tk k b k r l
~~~~~~~~~ c~~~~~~~~~~~~~~~~~~~~~~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ j
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC 68 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
regardless of whether the portfolio or individual security approach is
used. This difference is particularly striking for FCFvalue estimates where
the median VFC-based on realizations is zero, reflecting the fact that most
of the sample portfolios (20 of 30, not reported) had negative realized
mean free cash flows. The smaller ex post value estimates are also con-
sistent with the results in table 6 which show that, for our sample, real-
izations fell short of analysts’ expectations by a wide margin. Second,
comparisons of observed prices with both ex ante and ex post value es-
timates indicate that ex ante value estimates dominate ex post value es-
timates: for all models, ex ante values have significantly (at the .00 level)
smaller bias and are more accurate than comparable ex post values.
Third, using either smallest absolute bias or smallest absolute prediction
error as the performance criterion, the realization-portfolio results show
that AE value estimates dominate DIV and FCF value estimates. In con-
trast, the forecast-portfolio results depend on the growth rate: for g= 0%,
VAE dominates VDV and VFC-in terms of bias and accuracy, but for g = 4%, variation in the value estimates has on the performance metric. When the
variability in value estimates is retained-as it is when individual securi-
ties rather than portfolios are valued-the results (far right columns of
table 7) show that AE value estimates consistently dominate FCF and DIV We draw the following inferences from these results. The conclusion robust to the level of aggregation (portfolio versus individual securities),
the type of data (realizations versus forecasts), and the performance
metric (bias versus accuracy). We find, however, that the levels of bias
and accuracy are significantly (at the .00 level) smaller when forecasts
rather than realizations are used to calculate value estimates. The fore-
cast versus realization distinction is also important for conclusions con-
cerning FCF and DfVvalue estimates. Consistent with PS, we find that ex
post FCF value estimates are more biased than ex post DIV value esti- estimates in terms of both bias and accuracy.
6. Summary and Conclusions
This paper compares the reliability of value estimates from the dis-
counted dividend model, the discounted free cash flow model, and the
discounted abnormal earnings model. Using a sample of five-year fore-
casts for nearly 3,000 firm-year observations over 1989-93, we find that
38 See n. 6 for results based on other growth rates.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC COMPARING EARNINGS EQUITY VALUE ESTIMATES 69
the AE value estimates are more accurate and explain more of the varia-
tion in security prices than do FCF or DfVvalue estimates. Our explora- the greater reliability of AE value estimates is likely driven by the suffi-
ciency of book value of equity as a measure of intrinsic value, and per-
haps by the greater precision and predictability of abnormal earnings.
Our analysis of whether accounting discretion enhances or detracts
from the performance of the AE model indicates no difference in the
accuracy of AE estimates between firms exercising high versus low ac-
counting discretion, although there is some evidence that AE value esti-
mates explain more of the variation in current market prices for high
discretion firms than for low discretion firms. We also find no evidence
that AE value estimates are less reliable for firms with high versus low
R&D expenditures. Together these findings indicate no empirical basis
for concerns that accounting practices (such as immediate expensing of
R&D or the flexibility afforded by accruals) result in inferior estimates
of market equity value. Our results are more consistent with the argu-
ment that the articulation of clean surplus financial statements ensures
that value estimates are unaffected by conservatism or accrual practices.
We conclude there is little to gain-and if anything something to
lose-from selecting dividends or free cash flows over abnormal earn-
ings as the fundamental attribute to be valued. Together with the fact
that earnings are by far the most consistently forecasted attribute, our
results suggest little basis for manipulating accounting data (for exam-
ple, to general estimates of free cash flows) when earnings forecasts and
book values are available.
REFERENCES
BARTH, M.; R. KASZNIK; AND M. McNICHOLS. “Analyst Coverage and Intangible Assets.” BERNARD, V “The Feltham-OhIson Framework: Implications for Empiricists.” Contemipo- COPELAND, T.; T. KOLLER; AND J. MURRIN. Valuation: Measuring and Managing the Value of
Companies. 2d ed. New York: Wiley, 1994.
EDWARDS, E., AND P. BELL. The Theory and Measurement of Business Income. Berkeley: Univer-
sity of California Press, 1961.
FAMA, E., AND K. FRENCH. “Industry Costs of Equity.” Journal of Financial Economics 43
(1997): 153-93. Market Effect.” Working paper, University of Michigan and Cornell University, 1995.
. “Accounting Diversity and International Valuation.” Working paper, University of
Michigan and Cornell University, 1996.
HEALY, P “The Effect of Bonus Schemes on Accounting Decisions.” Journal of Accounting
and Economics 7 (1985): 85-107. ville, Va.: Financial Analysts Research Foundation, 1993.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC 70 J. FRANCIS, P. OLSSON, AND D. R. OSWALD
KAPLAN, S., AND R. RUBACK. “The Valuation of Cash Flow Forecasts: An Empirical Analy-
sis.” Journal of Finance (September 1995): 1059-93.
OHLSON, J. “Earnings, Book Values, and Dividends in Equity Valuation.” Contem)porary
Accounting Research (Spring 1995): 661-87.
PENMAN, S., AND T. SOUGIANNIS. “A Comparison of Dividend, Cash Flow, and Earnings
Approaches to Equity Valuation.” Contemporary Accounting Research (Fall 1998): 343-84.
PREINREICH, G. “Annual Survey of Economic Theory: The Theory of Depreciation.” Econo-
metrica (July 1938): 219- 41.
SOUGIANNIS, T, AND T. YAEKURA. “The Accuracy and Bias of Equity Values Inferred from
Analysts’ Earnings Forecasts.” Working paper, University of Illinois, 1997.
STEWART, G. The Quest for Value. New York: HarperCollins, 1991.
WHITE, H. “A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test
for Heteroscedasticity.” Econometrica 48 (1980): 817-38.
WILLIAMS, J. The Theory of Investment Value. Cambridge, Mass.: Harvard University Press,
1938.
This content downloaded from 198.246.186.26 on Wed, 10 Jan 2018 16:52:55 UTC image 1 Journal of Accounting Research, Vol. 38, No. 1, Spring, 2000
>~~~~~~~~~~~~~~~\ on Ln0
u _
QO C ) 0 C 7 e . q m
All use subject to http://about.jstor.org/terms
VFC-dominates.38 This disparity between the bias and the accuracy results
observed for FCF value estimates (for g = 4%) highlights the effect that
estimates in terms of accuracy.
that AE value estimates dominate FCF or DIV value estimates is fairly
mates; however, ex ante FCF value estimates dominate ex ante DIVvalue
All use subject to http://about.jstor.org/terms
tions of the sources of the relative superiority of the AE model show that
Journal of Accounting Research (forthcoming).
rary Accounting Research (Fall 1995): 733- 47.
FRANKEL, R., AND C. LEE. “Accounting Valuation, Market Expectation, and the Book-to-
IBBOTSON, R., AND R. SINQUEFIELD. Stocks, Bonds, Bills and Inflation, 1926-1992. Charlottes-
All use subject to http://about.jstor.org/terms
All use subject to http://about.jstor.org/terms
image 2
image 3
image 4
image 5
image 6
image 7
image 8
image 9
image 10
image 11
image 12
image 13
image 14
image 15
image 16
image 17
image 18
image 19
image 20
image 21
image 22
image 23
image 24
image 25
image 26
Front Matter
Country-Specific Factors Related to Financial Reporting and the Value Relevance of Accounting Data [pp. 1 – 21]
The 1993 Tax Rate Increase and Deferred Tax Adjustments: A Test of Functional Fixation [pp. 23 – 44]
Comparing the Accuracy and Explainability of Dividend, Free Cash Flow, and Abnormal Earnings Equity Value Estimates [pp. 45 – 70]
Stock Returns and Accounting Earnings [pp. 71 – 101]
Research Reports
Financial Packaging of IPO Firms in China [pp. 103 – 126]
Biases and Lags in Book Value and Their Effects on the Ability of the Book-to-Market Ratio to Predict Book Return on Equity [pp. 127 – 148]
Do Stock Prices Fully Reflect the Implications of Current Earnings for Future Earnings for AR1 Firms? [pp. 149 – 164]
Why Do Audits Fail? Evidence from Lincoln Savings and Loan [pp. 165 – 194]
Capsules and Comments
The Effect of the External Accountant’s Review on the Timing of Adjustments to Quarterly Earnings [pp. 195 – 207]
The Incremental Information Content of SAS No. 59 Going-Concern Opinions [pp. 209 – 219]
Back Matter
