2

Journal of Accounting Research
Vol, 20 Supplement 1982

Printed in U,SA,
Market-Based Empirical Research

in Accounting: A Review,
Interpretation, and Extension
BARUCH LEV* AND JAMES A. OHLSONf
Introduction
A decade and a half of the most concerted and ambitious research
effort in accounting history is evaluated here. It is the search into the
relationship between pubUcly disclosed accounting information and the
consequences of the use of this information by the major group of users—
equity investors—as such consequences are reflected in characteristics of
common stocks traded in major exchanges. This research effort, known
as "market-based accounting research" {MBAR), obtained its impetus
from major developments in finance theory during the late 1950s and
early 1960s. These were the portfoUo selection theory and its offspring,
the capital asset pricing model {CAPM); and the concept of information-
aUy efficient capital markets {ECM). Subsequent developments in infor-
mation economics, agency theory, and optimal incentive-signaUng models
were quickly incorporated into MBAR. While the impetus to this Une of
research came from other disciplines, accounting was not only a benefi-
ciary but also, to a modest extent, a benefactor. MBAR is relevant to the
study of capital market efficiency, the CAPM, information economics,
and regulation. This contribution distinguishes MBAR from most ac-
counting areas, whose contributions to other disciplines are negUgible. In
* Tel-Aviv University; t University of California, Berkeley, We are grateful to Ray Ball,

Nicholas Dopuch, Robert Freeman, George Foster, Robert Kaplan, Joahua Livnat, and
Katherine Schipper for constructive conunents and suggestions. We have also benefited
from workshop discussions at the University of Alberta (Edmonton), Carnegie-Mellon
University, New York University, and Washington University (St, Louis), Recent survey
articles by Abdel-khalik and Ajinkya [1979], Griffin [1982], and Ricks [1982a] assisted in
the preparation of this paper,
249
Copyright ®, Institute of [Professional Accounting 1983
250 CURRENT RESEARCH METHODOLOGIES IN ACCOUNTING: 1982
both quantity and quaUty, msu'ket-based research is undoubtedly a

unique research endeavor in accounting and therefore deserves periodic
evaluation.
Section 1 of this paper summarizes the expectations raised by MBAR.
As is characteristic in immature discipUnes, the pendulum swung fast:
the early euphoric expectations quickly gave way to disiUusionment or
outright pessimism. Such extreme attitudes, as wiU be seen, are unwar-
ranted. Section 2 reviews and evaluateis the available empirical evidence
classified into four categories: (t) the infonnation content of accounting
data, {ii) the effects of voluntary (discretionary) differences and changes
in accounting techniques on investors, firms, and managers, {Hi) the
effect on the stock market of accounting regulation, and {iv) the impact
of MBAR on other disciplines.
The conclusions of this review and evaluation are as foUows. (i)
Accounting data convey useful and timely information to investors. WhUe
this conclusion definitely holds for eamings data, the margined contribu-
tion of the voluminous noneamings data pubUshed in financial reports is
stiU largely unknown. Given nontrivial costs of information disclosure
and dissemination, this issue obviously deserves more research attention.
Also left open by the "information-content" studies is the precise meaning
of "usefulness of information" and therefore the social role and benefits
of accounting data. This important issue wiU be taken up in the extensions
considered in section 3. {ii) With respect to the market consequences of
cross-sectional differences and time-series changes in discretionary ac-
counting techniques, early studies appearing to indicate investor ration-
ality have given way to discomfiting findings. It is now clear that the
existence of some investor irrationaUty (often termed "functional fixa-
tion" ) cannot be precluded. This is very disturbing because there are no
satisfactory behavioral altematives to investor rationaUty. Furthermore,
the conventional dichotomy between accounting changes having
"cosmetic"and those having real (cash-flow) effects is clearly simplistic.
When the scope of inquiry is extended to include effects on management
compensation and contractural arrangements, it appears that almost aU
accounting changes can have real effects. This broader view leads to the
fundamental question of managers' motives in choosing accounting tech-
niques. Research on aU these important issues is in its infancy, and results
are far from conclusive. Accordingly, various strong poUcy recommen-
dations (e.g., to the Financial Accounting Standards Board) that were
based on a firm belief in investor rationaUty with respect to accounting
information seem premature. {Hi) As for the stock market consequences
of accounting regulation, it is comforting to conclude that, in the few
cases attracting considerable research attention (e.g., replacement cost,
oU and gas accounting, and lines-of-business data), results appear to be
consistent and even conclusive. Furthermore, a continuous improvement
in statistical methodology is evident in this line of research, (iv) FinaUy,
two aspects of the contribution of MBAR to finance theory can be singled
MARKET-BASED RESEARCH IN ACCOUNTING 251
out: (a) that empirical evidence which appears to be inconsistent with

either the equiUbrium risk-return relationship {CAPM) and/or market
efficiency, contributing to a reconsideration of these models, and (6) the
empirical work, started by the Beaver, Kettler, and Scholes [1970] study,
relating fundamental (firm-based) variables to the systematic risk of
common stocks. Unfortunately, Uttle has been done since the early studies
in this area.
Section 3 discusses two areas which, in our view, are potentiaUy
promising extensions of MBAR. The first is a carefuUy structured consid-
eration of the welfare implications of accounting information, and the
second is a retum to the much-neglected issue of capital asset valuation
by fundamental variables, namely, the construction of stock-valuation
models.
L Expectations Raised by Market-Based Accounting

Research
Being utilitarian in nature, the ultimate test of accounting is its
usefulness. However usefulness, Uke the oft-used accounting analogue
"relevance," is an elusive concept raising a host of difficult questions.
How should usefulness be conceptualized and operationally defined and
measured? Is usefulness related to an increase in individuals' weU-being
(expected utiUty) caused by additional accounting information, or does it
mean the efficiency of resource aUocation in the economy with no
reference to individual weU-being? When some individuals are made
better off and others worse off by a particular accounting environment,
how should the individual changes in weU-being induced by the account-
ing environment be aggregated to form a mechanism of social choice?
Should the weights assigned to each individual welfare change be equal
or should specific groups of users (e.g., the "least-informed" or "least-
privUeged") be favored? Can the usefulness of accoimting be fostered by
pubUc intervention and regulation (the case of "market failure")? If so,
what is the optimal structure of accounting institutions (e.g., set in the
pubUc or private sector), and how should accounting choices be made by
poUcymakers and the consequences of such choices be evaluated? These
were the major questions addressed in recent decades by financial ac-
counting scholars, reflecting a predominantly poUcy (normative) orienta-
tion.
Methodological approaches to these issues foUowed by researchers
included deductive (a priori) reasoning, laboratory experiments, and
questionnaire studies. The result was at best a very limited advancement
in the understanding of the basic issues and a negUgible impact on
practitioners and pubUc decision makers. Missing was a widely accepted,
conceptuaUy weU-structured, and empiricaUy testable Unk between ac-
counting (firm-based) information and the specific uses of this informa-
tion. Modem capital market theories, developed in the 1960s, appeared
252 B. LEV AND J. A. OHLSON
to provide the missing link and establish a coherent program for financial
accounting research. The effect of this program on accounting research
cannot be exaggerated; during the past decade and a half, studies based
on capital market theories and data occupied the center stage of account-
ing research.'
The link provided by capital market theories connects the accounting
information system to its function in capital markets. Information has a
dual role in these markets. First, it aids in establishing a set of equiUbrium
security prices that affects the aUocation of "real" resources and the
productive decisions implemented by firms. Second, it enables individuals
to exchange claims to present and future consumption across different
states, thereby attaining both preferred pattems of lifetime consumption
and the sharing of societal risks. This expUcit conceptualization of the
role of information in capital markets appears to provide the elusive
operational framework for the systematic analysis of altemative account-
ing information systems. The outcome of the economic system, as a
function of the information system, can now be anjilyzed. The accounting
research issue therefore becomes less one of identifying the "uniquely
correct" concept of usefulness than of investigating whatever dimension
of usefulness is contextuaUy relevant or interesting. In any event, the
crucial observation is that impUcations of different infonnation systems
rest ultimately on their effects on the uncertain state-contingent multi-
period consumption pattems of individuals, where the pattems are de-
termined and induced by prices which in tum reflect individual beUefs
about the Ukelihood of altemative occurrences.
The three key ingredients of this framework—information, prices and
other market variables, and expected utiUties of contingent consumption
pattems—were, of course, not entirely new even in the eaily 1960s.
However, the lure of capital market theories lay in the possibUity of
obtaining a systematic integration of these three ingredients. Such inte-
gration pointed to a weU-specified and operational agenda for financial
accounting research. The foUowing quotations from the security-price
section of the 1972 Report of the American Accounting Association
Committee on Research Methodology in Accounting, written by Beaver
[1972], capture the essence of the potential research agenda.
First, given capital market efficiency, the information content of
accounting data can be inferred by observing stock-price and vohune
reaction to announcements of these data:
Knowledge of the association between altemative accounting measurements and security
prices is an essential part of knowledge of what information is impounded in security prices,
, , , It is also likely to be important for specifying how prices would change (if at all) as the
[information] system is altered, [1972, p, 427]
' The significant impact of capital market theories was, of course, not restricted to
accounting. It also had a profound effect on economic research which, up to the 1950s,
largely ignored the financial markets, focusing on reed markets. Pioneering research such as
Tobin [1958], Diamond [1967], and Radner [1968] established the link between financial
and real markets in economic theory.
Second, and even more important, capital market efficiency provides

a preference ordering of alternative information systems that can guide
policymakers' decisions:
If the efficient market hypothesis is adopted, then the association with security prices
provides a simplified preference ordering with which altemative measurement methods can
be ranked. That method which is more highly associated with security prices is more
consistent with the underlying infonnation set used in setting equilibrium prices. Hence,
subject to a more complete analysis involving competing sources of information and costs
of altemative methods, the finding provides prima facie evidence that the method which is
more highly impounded ought to be the method reported in the financial statements, [1972,
pp, 427-28] (emphasis added)''
Third, according to the Capital Asset Pricing Model (CAPM), the

relevance of accounting data to the individual investor is restricted to
systematic-risk (P)prediction:
The role of accounting data becomes its predictive ability with respect to fi. Hence, the fi
analysis becomes extremely important as a research method, if the object is to assess the
value of information to the individual investor and if market efficiency is accepted as a fact
oflife. [1972, p, 424]
Fourth, most accounting measurement errors do not induce any rele-

vant errors in assessing future portfolio returns because at the portfolio
level only systematic errors matter:
An examination of portfolio theory suggests that the appropriate context within which to
examine infonnation issues is at the portfolio level, not at the level of individual securities.
Hence, individual security retum errors that are random or independent in nature are not
of concem because they can be diversified away. Only errors that are systematic and still
persist at the portfolio level are of major concem. This context constitutes a major alteration
in the way that measurement errors in accounting have been traditionedly viewed, [1972,
pp, 428-29]
Fifth, the distinction between individuals' behavior and aggregate

market behavior is captured by observing price and volume reactions:
A major difference between price variability analysis and the volume analysis is that the
former deals with effect of an event on the market level (i,e, by examining changes in
equilibrium prices), while the latter deals with the effects of fm event upon the expectations
of individual investors, . . . a related topic is the extent to which accounting data induce
heterogeneous expectations among investors and, hence, an exchange of shares without
changing the equilibrium price of a security. The issue becomes important because non-
zero costs are incurred as a result of exchange of shares, [1972, pp, 414-15]
Sixth, fundamental (intrinsic-value) research using accounting infor-

mation is suspect:
Any research method that relies upon the assumption of market inefficiency in addressing
an information issue should be seriously questioned,,,, In other words, from the point of
view of society, total wealth is unchanged and intrinsic value analysis merely implies a
reallocation of wealth. Hence, the ability to earn abnormal returns can never be a valid
ba8is for "assessing the value of information." [1972, pp, 424-25]
' The potential policy implications of finance theories and empirical evidence for the
FASB were elaborated in Beaver [1973],
Seventh, the accounting research program can be extended to validate

the Efficient Capital Market hypothesis (ECM):
The research methods can be used to explore the issue of market efficiency in greater depth.
Additional elements of the information set can be isolated, and it can be determined
whether the market is efficient with respect to that information element. In particular, cross
sectional as well as time series tests are needed, [1972, p, 429]
Eighth, MBAR is superior to other research methodologies in ac-

counting:
The security price research was chosen because it not only provides a set of specific research
methods, but also possesses both theoretical and empirical support for the relevance of
such research methods for accoimting research, , , , this latter function is lacking in many
applications of research methods from other disciplines, . , . For example, consider some
recent applications of behavioral methods in accounting to investigate the manner in which
individuals process data. The method commits the fallacy of attempting to predict properties
of market behavior based upon the behavior of individuals, [1972, pp, 408, 427]
Thus, a decade ago expectations were high.^ It appeared that modem

finance and general equUibrium theories provided accountants for the
first time with operational definitions of useful (or relevant) information;
with a method of ranking in a social-preference order altemative infor-
mation systems; with a clear-cut objective for accounting data, suggesting
the use of an abundance of reUable data (stock prices); and most impor-
tant, with a weU-structured research methodology purportedly linking
information to individuals' optimal lifetime consumption pattems. This
program seemed to dominate altemative research methodologies.^ How-
ever, as empirical market-based research accumulated, euphoria gave
way to a more sober attitude toward the potential of this line of research
and at times to outright pessimism.
For example, Kaplan [1975, pp. 54-55], in a survey of empirical re-
search, concluded:
What then are my reservations about the future value of empirical research in accounting?
I think that we have now gotten all the easy results and that major new findings are going
to be much harder to obtain,,,, I am much more pessimistic about the ability of empirical
research to give us many insights on a whole variety of important questions now confronting
the accounting profession,,,, it is not clear to me how empirical research in particular can
provide much guidance to current debates on issues which I will describe as involving costly
disclosure.
Such gloomy observations did not deter researchers, who produced

empirical market studies at an ever-increasing rate. This was, at least
' The extent of optinusm is, of course, difficult to evaluate in any precise terms. However,
the optimism can be inferred from the fact that Beaver's [1972] article was influential and
appears to have had a substantial impact on attitudes within the accounting research
community,
* For example, Dyckman, Downes, and Magee [1975, p, 87] note: "Too often theoretical
developments and empirical testing go their independent ways..,, From thiB viewpoint the
efficient market research thread is more complete than most other research methodologies."
partiaUy, rationalized by Dyckman, Downes, and Magee [1975, p. 90] by

ddta accessibiUty: "We have an uneasy feeUng that the attractiveness of
research topics in this [stock market research] area is fostered in sub-
stantial part by the avaUabUity of the data base in the form of stock
market and accounting data tapes."
Of the research agenda items provided by capital market theories, the
one that came under the heaviest attack was the proposition that market
studies can provide accounting policjonakers (e.g., the FASB or SEC)
with a social-preference ordering (choice criteria) of altemative infor-
mation systems. Some (e.g.. May and Sundem [1973]) pointed out that
no direct inference can be madefirommarket studies about the usefulness
of new (hitherto unreported) information items, because market prices
can reflect and react to disclosed information only. Others (e.g., Dyckman
et al. [1975]) noted that equity investors are only a subset of the potential
users of accounting infonnation. These arguments are obviously vaUd.
At a more fundamental level, it has been argued that within the current
institutional setting, where financial information is provided free of direct
charges (thereby possibly leading to the "free rider" phenomenon), se-
curity prices cannot indicate the social desirabUity of information. Go-
nedes and Dopuch [1974, pp. 78-80] noted:
The critical factor is the extent to which nonpurchasers of rights to use information are
excluded from using it. When the rule of excluding nonpurchasers is not enforced then the
prices of firms' ownership shares cannot be used to assess the desirability of the information-
production decisions made by (or imposed upon) these firms. Such prices can, however, be
used to assess the effects of these information-production decisions, . . , Since assertions
about effects are important parts of the justifications offered for recommendations and
prescriptions, we can assess the strength of these justifications by evaluating the theoretical
or empirical support for the assertions about effects In short, assessments of the effects
of altemative accounting procedures and regulations can be useful to accounting policy-
making bodies in making their decisions and to their constituencies in evaluating those
decisions.
This distinction between the ability of market studies to indicate effect

but not desirabiUty appears to have received wide recognition in the
accounting Uterature.* More fundamentaUy, Beaver and Demski [1974],
among others, argued that, given the unavoidabiUty of investor interper-
sonal conflicting preferences, accounting poUcy decisions might lead to
social consequences whose resolution requires ethical criteria, which are
obviously beyond the confines of empirical market research.
It thus came to be widely recognized that making welfare statements
about accounting information issues is practicaUy impossible. As a result,
the task of guiding research on accounting information choices seems to
" But, by no means, unanimity. It has been argued, for example in Lev [1978], that in
efficient markets nonpurchasers of information (i,e,, those who do not invest in information
search and dissemination) will be effectively excluded from reaping the benefits of new
information, since purchasers, by their market actions, will cause prices to quickly impound
the value of information. Market efficiency will thus decrease the seriousness of the "free
rider'' problem, retaining incentives for information search.
have been delegated to poUtical scientists and moral phUosophers.® In a

formal sense this is justified, since strictly speaking, economic analysis
cannot unambiguously resolve welfare issues. Does it then foUow that
the high expectations raised a decade ago by market-based research
methodologies must be restricted to the relatively minor role of assessing
the impact of regulation? In section 3 we take exception to this "aU or
none" view and suggest that, in certain contexts, market studies can
provide poUcjmiakers with the welfare impUcations of accounting infor-
mation issues. To be sure, such welfare analysis is not intended to resolve
questions of policy; this, in our view, would be much too ambitious an
objective. Rather, the impUcations of welfare analysis should be viewed
£is an aid in interpreting empirical findings.
Hence, we suggest considering two links, each represented by a tuple—
information/market variables, and market variables/expected utiUties of
uncertain consumption pattems. The conventional partial link between
information and market variables is not sufficient, because the formula-
tion of accounting poUcy requires a comprehensive perspective connect-
ing information to expected utiUties. It is this final link (market variables/
expected utiUties of consumption pattems) that adds substance and
meaning to empirical findings; but this is also the Unk that has been the
least appreciated in accounting research. The difficulty here appears to
be due substantially to complexities of theory. Capital market theory
does, of course, permit a formal extension such that a variety of infor-
mational issues can be addressed rigorously. However, much of the early
accounting research considered informational issues in an heuristic fash-
ion within standard finance paradigms, such as the CAPM. The final link
was thus more elusive than first anticipated. Our structure of research
evaluation in this paper refiects these observations. Section 2 pertains
principaUy to an evaluation of the first link—information/market vari-
ables. However, attempts to give empirical results precise interpretations
force us at least occasionaUy to consider the second link—market vari-
ables/expected utiUties, and a more complete discussion of this second
Unk is presented in section 3.
1.1 A NOTE ON EXPERIMENTAL DESIGN AND OTHER STATISTICAL
ISSUES
These issues are obviously of considerable importance in any general
research evaluation. We have decided, however, not to elaborate on
methodological problems present in the large number of specific studies
evaluated here. This was done because of space limitations and because
a comprehensive methodological discussion of empirical accounting re-
' Gonedes and Dopuch [1974, p, 117] note: "But, in the end it [the FASB] is a political
institution (nominally under private sector control). Consequently, the actions of this (and
any similar) organization should be viewed as political outcomes, rather than as outcomes
of a process that is supposed to operationalize and implement an accounting theory in a
pure and pristine manner,"
search is presented in another study in this issue (BaU and Foster [1982]).
More important, we have decided against emphasizing experimental
design issues because considerable methodological improvement is evi-
dent in recent MBAR. Most researchers currently appear to be aware of
and account for general experimental design problems, such as those
arising from sample nonrandomization and cross-sectional correlation of
the data. More care is taken in the choice of appropriate statistical
significance tests and in recognizing the limitations of underlying models,
such as the CAPM. New data sets (e.g., intraday stock prices) and a more
exact identification of disclosure timing improve the signal-to-noise ratio
and thereby the probabiUty of detecting a market reaction to information
release. Furthermore, some of the basic findings, such as the stock-price
reaction to eamings announcements (section 2.1.1) or the nonreaction of
stock prices to the release of replacement-cost data (section 2.3.8), appear
to be rather robust across experimental designs. It does therefore seem
superfluous to elaborate on the methodology of the specific studies
mentioned in the foUowing sections.
Similar comments are appUcable to various "research technologies"
that have recently attracted considerable attention in the accounting
Uterature. Among these are the discussions of the API metric: MarshaU
[1975], PateU [1979], Ohlson [1978; 19796], and Livnat [1981]; econometric
issues in market-based studies: Beaver [19816]; certain aspects of the
maintained hypotheses of market efficiency and the CAPM: BaU [1978]
and Beaver [1981a]; pitfalls in regulation research (with particular ref-
erence to the oU and gas studies): Foster [1980]; and studies on the
performance and statistical power of the "market model": Brown and
Warner [1980]. Nevertheless, numerous methodological comments of a
general nature wiU be made in the foUowing sections. In short, we have
chosen to emphasize the evaluation of research objectives, and the
interpretations and impUcations of findings, rather than experimental
design issues, because the former involve subtle issues which are seldom
discussed systematicaUy in the accounting Uterature.
2. The Empirical Evidence: Review and Interpretation

The avaUable evidence on the relationship between stock market data
and accounting information and regulation is classified in this section
into four categories according to the objectives of the studies:
2.1 Information content studies: the marginal information contribu-
tion of accounting signals to the determination of security-returns
behavior.
2.2 Differences in discretionary accounting technigues: the impact on
investors, managers, and firms.
2.3 Conseguences of regulation: the study of the market effects of
accounting regulation.
2.4 Impact on related disciplines: contributions of MBAR to other
fields, especiaUy finance.
2.1 ACCOUNTING DATA CONVEY RELEVANT AND TIMELY
INFORMATION
WhUe not surprising to most astute observers of financial markets, this
conclusion is both reassuring (at least to accountants) and stands in stark
contrast to aUegations suggesting otherwise. For example, there has been
strong a priori criticism of the relevance to users of the accounting
measurement process both by accountants (e.g., claims to the effect that
accounting eamings as measured by generaUy accepted accounting prin-
ciples [GAAP], rather than by "superior" measures, can be of Uttle if any
relevance to rational individuals) and by nonaccountants (e.g., accounting
as "a known untruth" in Boulding's words [1962, p. 55] or as "motions
the accountant goes through" rather than foUow economic principles,
according to Treynor [1972, p. 41]). Furthermore, the ECM hypothesis
has sometimes been incorrectly interpreted as implying that by the time
financial information is released it is completely impounded in security
prices and hence irrelevant and of no social value.' FinaUy, various early
empirical findings suggesting very weak and temporaUy unstable associ-
ations between eamings and stock returns (e.g., Benston [1967] and a
review of these findings in Keenan [1970]) led to considerable misgivings
about the relevance of accounting information to investors.
Most of the studies on the information content of accounting data are
of the "announcement type," examining whether the announcement of
some event results in a change in the characteristics of the stock-return
distributions (e.g., mean or variance). A careful methodological design,
accounting for overaU market effects, randomization of the sample events
over time, exact identification of the announcement dates, and an appro-
priate choice of statistical significance tests, aUows the researcher to infer
from an observed change in the retum distribution or from a contempor-
aneous statistical association the "information content" of the data
released close to the relevant date. The next two subsections review some
of the more compelling evidence.
2.1.1 Earnings Announcements. The weU-known BaU and Brown
[1968] study, in which unexpected annual eamings changes (focusing only
on the sign of change) were correlated with residual stock returns,
initiated research that has accumulated consistent evidence on the rele-
vance and timeUness of accounting earnings.* While the BaU-Brown study
was purportedly a test of the information content of eamings, its intro-
duction contained expressions to the effect that the evidence bears on
the usefulness of accounting (eamings) data. Results revealed the exist-
ence of a statisticaUy significant association between unexpected eamings
and residual returns. The statistical association was measured by the
' For elaboration on such misinterpretations of the ECM hypothesis and a more realistic
concept of informational market efficiency, see Grossman and Stiglitz [1976] and, in the
accounting literature. Beaver [1978],
' In the examination of stock splits effect on the stock market, this association method-
ology had been developed earlier by Fama et al, [1969],
weU-known API metric, which simulates the outcome of a portfolio

trading strategy based on advance knowledge of eamings. Not surpris-
ingly, and of importance, most (about 85 percent) of the stock-price
reaction occurred prior to and thus in anticipation of the eamings
smnouncements.® As is also weU recognized, there appeared to be no
significant association between eamings and returns foUowing the disclo-
sure.
The BaU-Brown statistical findings were based on, among other things,
their eamings expectation model. The simplest model assumed that
expected future eamings are determined by and equal to current eamings;
thus the first difference in eamings equals unexpected eamings. In the
common jargon, eamings are said to foUow a random-walk generating
process. The idea that the relevant variable is unexpected eamings, and
that earnings have to be adjusted to have zero mean, is an important one.
It led to extensive empirical analyses of the behavior of eamings processes
(time-series studies). The emphasis on unexpected eamings points, how-
ever, to an intrinsic limitation of the methodology because of the need to
assume a specific eamings expectation model. Accordingly, there have
been numerous efforts to improve upon the estimation of the unexpected
(new information) element of eamings. For example, PateU [1976] sub-
stituted managements' eamings forecasts for BaU-Brown's naive model.
PateU's findings confirmed the BaU-Brown results in that a trading
strategy based on management forecasts (and assuming prior knowledge
of the actual eamings) was only sUghtly better in providing excess returns
than BaU-Brown's naive expectations model. This permits the foUowing
interpretation. Given that earnings are relevant to the determination of
security prices, the API metric is an indicator of the "quaUty" of the
eamings expectations model, and the relatively naive BaU-Brown eam-
ings behavior model (random walk) appears to be roughly on a par with
more sophisticated schemes, such as PateU's." (The conclusion that no
significant improvement in eamings predictions is atteiined by more
sophisticated prediction models has generaUy been supported by the
eamings forecasting Uterature.) Beaver [1974], extending BaU-Brown's
methodology to incorporate the size rather than just the sign of unex-
pected eamings, found that the most extreme portfoUos, those containing
stocks of firms with the largest positive and largest negative unexpected
eamings, had higher absolute residual returns than portfoUos formed of
firms whose unexpected eamings were moderate in size. Similar findings
are reported by PateU [1976] using, as mentioned above, managements'
earnings forecasts. Thus, both the magnitude and sign of unexpected
eamings add to the statistical association. In a sinular vein. Beaver,
"This might be due to the existence of more timely signals, both accounting (e,g,,
quarterly reports) and nonaccounting, and/or to the speciflc eamings forecast model
employed,
'" This interpretation involves some subtle issues which have not received the attention
they deserve, A complete theoretical analysis is provided in Patell [1979],
Clarke, and Wright [1979] reported that the ranking of 25 NYSE port-
foUos on the basis of percentage change in unexpected EPSs was highly
correlated with the ranking of portfolios on the basis of residual stock
returns.
The BaU-Brown methodology has been used in numerous related
contexts, such as quarterly eamings reports (e.g.. Brown and KenneUy
[1972] and Foster [1977a], the latter using Box-Jenkins expectation
models), and non-NYSE stocks (e.g., Foster [1975] on over-the-counter
(OTC) insurance companies; Brown [1970] on the Australian stock mar-
ket; Firth [1976] on British stocks; and Deakin, Norwood and Smith
[1974] on the Tokyo exchange). AU confirm the existence of a statisticaUy
significant association between unexpected eamings and residual stock
returns.
Whereas the studies so far discussed examined the association between
eamings announcements and the mean of the (unexpected) retum distri-
bution, an altemative methodology focused on another property of the
retum distribution—the variance of residual returns. SpecificaUy, the
residual retum variance during the announcement period was compared
with the average retum variance in the pre- and post(non)annoucement
periods to determine whether they are drawn from the same distribution.
This methodology was first appUed by Beaver [1968] in the examination
of annual eamings announcements. Similar approaches have been em-
ployed in other contexts, for example, by May [1971] in an examination
of AMEX quarterly eamings, Hagerman [1973] in OTC bank stocks, and
McNichols and Manegold [1982] on stock market data from the pre- and
postquarterly report regulation. The variance tests have also been appUed
to more finely partitioned retum data. Morse [1981], using daily data,
obtained results similar to those of Beaver [1968]. PateU and Wolfson
[1981], examining intraday (transaction by transaction) stock prices,
reported a statisticaUy significant change in the retum variance at the
time of quarterly eamings announcements. They further observed that
the regular pattern of negatively seriaUy correlated intraday returns is
"interrupted" at the time of eamings announcements. These and similar
studies found the price variance during the announcement period to be
significantly larger than the average variance in the nonannouncement
period, a finding consistent with the hypothesis that eamings data convey
new information to the market. ^^ More precisely, after (generaUy) con-
troUing for market-wide price movements and averaging over different
announcement periods, a causal link was made between the disclosure
event and the observed shift in the mean or variance of the retum
distribution. This again iUustrates that one can view announcements as
a treatment effect and analyze any dimension of the retum distribution.
A statistical effect is then interpreted as evidence of information content.
" An exception to these findings is reported by Lev and Yahalomi [1972] for the Israeli
stock exchange, where the nonreaction of stock prices to eamings announcements was
attributed to unique capital market imperfections.
Beaver [1968] introduced yet another element to the study of security-

market reaction to eamings announcements by analyzing trading volume.
The findings were consistent with the information content hypothesis,
for above average trading volume was observed around the eamings
announcement dates. As discussed by Beaver in some detaU, trading
volume might reflect individuals' asymmetric perceptions of firms' future
prospects, whUe price reactions capture only the "average" individual's
perception of new information. From a theoretical perspective, one can
observe an above-average reaction in either price or volume without
necessarily having an above-average reaction in the other. EmpiricaUy,
the two variables might often be positively correlated; but this correlation
must not obscure the fact that the variables capture two different aspects
of individuals' beUef-revision processes. It is therefore noteworthy that
relatively few studies have since focused on trading volume. This might
at least partiaUy be due to practical considerations, such as the absence
of readily avaUable high-quaUty volume data. Nevertheless, research
choice between trading-volume and price reactions should not be seen as
arbitrary or determined by practical circumstances. The theoretical/
methodological question of how to interpret price and volume in con-
junction with each other has not been considered in the Uterature. In
view of its importance, a discussion of this issue is provided in section 3.1.
The methodologies used in the information-content research discussed
thus far associated earnings disclosure with realized (ex post) return
distributions. PateU and Wolfson [1979; 1981] recently examined the
anticipated (ex ante) information content of eamings. The methodology
was based on a generalization by Merton [1973] of the Black and Scholes
[1973] option-valuation formula, which expresses the price of a caU option
as a function of various variables, one of which is the average future
variance of the underlying stock returns. Since, as discussed earUer,
eamings announcements were found to induce increases in the stock-
return variance, such increases wiU probably be anticipated by the market
and thereby be reflected in options prices prior to the eamings announce-
ment. This methodology therefore examines the extent to which option
prices anticipate or "predict" increases in stock-return variation to be
triggered by an eamings announcement. PateU and Wolfson's empirical
tests indeed detected in option-price behavior an anticipation of quarterly
earnings announcements. In a sense, therefore, these results and those of
the variance studies are mutuaUy supportive; one cannot occur without
the other, given a market of rational pricing.
Thus, starting our survey with the "good news," it can be concluded
that eamings announcements (both quarterly and annual) affect the
stochastic properties of stock prices (mean, variance, and serial correla-
tion). Moreover, this conclusion appears to be robust across statistical
methodologies, time periods, and stock exchanges in which shares are
traded. These empirical observations are consistent with the beUef that
eamings announcements provide timely and relevant information to
individuals acting in the financial markets. It is probably not unfair to
State that such "information-content" studies are among the most im-
portant in empirical accounting research. In addition to this positive
finding, two important methodological/conceptual innovations emanat-
ing from "information content" research can be included under the
heading of "good news." First, the aforementioned studies have aU
evolved under the conceptual presumption of rational anticipations,
which requires a consideration of what is expected in the financial
markets. Thus, the BaU-Brown study emphasized that one must model
both expected returns and expected eamings. The investigation of eam-
ings expectation models has been extended in various promising direc-
tions (e.g., time-series anfdysis). Second, and this is reaUy an extension of
the previous point, there is no reason to expect an effect of eamings (or
other information) on the retum distribution foUowing the announcement
date. This contention, the informational efficiency hypothesis, is com-
monly used as a maintained hypothesis when experiments of information
content are conducted. The operational presumption is that the market
reaction to an informational item occurs prior to (in anticipation of) and/
or concurrent with its pubUc announcement date. Less appreciated is the
fact that informational efficiency is only a sufficient condition in the
context of information-content analysis. One cannot conclude from the
presence of pre- (or contemporaneous) announcement effects and the
absence of postannouncement effects that the market is informationaUy
efficient. The standard jargon often suggests that such evidence is
"consistent with" informationed efficiency; even so, "consistent with" is
certainly not equivalent to the much stronger "impUes." The test designs
for informational efficiency are more demanding; section 2.4.1 considers
some of these issues. In any event, the central role of rational anticipations
and informational efficiency in the evolution of empirical accounting
research is difficult to overstate.
There is also some "bad news" alongside the "good news." First, and
perhaps most important, the interpretations of the "event-study" findings
are far from obvious. While the term "usefulness" of infonnation was
used rather liberaUy in many early studies, upon reflection it is clear that
to move from the existence of contemporaneous statistical associations
to statements of "usefulness" is nontrivial. This is certainly the case if
"usefulness" is to be related to concepts of social utiUty. The problem
here is related to the difference between necessity and sufficiency. To the
extent one views a positive information-content finding as a necessary
condition for positive social value there are few problems. Information
content is simply regarded as evidence that the examined accounting
data have been used (revised beliefs); otherwise there would have been
no reaction of the market variables. The use of infonnation is clearly a
necessary condition for social value, no matter how the latter is (reason-
ably) defined. Establishing information-content evidence as a sufficient
condition for social value is a much subtler problem. At a minimum, one
must consider carefuUy the precise meaning of usefulness and the mech-
anism by which it occurs. The problem is then to demonstrate that

relatively desirable aUocations of resources and risks occur whenever
appropriate contemporaneous correlations/associations are present. Ac-
cordingly, the entire issue of the role and social value of accounting data
as it relates to information-content studies must be addressed. To the
best of our knowledge, the empirical Uterature has generaUy faUed to do
this, in spite of a substantial body of relevant theoretical knowledge. (See
Ohlson and Buckman reviews [1980; 1981].) This is also tme of the
efficient market hypothesis, where poUcy/welfare impUcations based on
a rigorous definition of informational efficiency are to a large extent
lacking. At a subsequent point (in section 3.1), these problems are
discussed in some detaU.
Second in the list of "bad news" is that although the documented
associations between eamings and market data are without question
present, they are nevertheless quite modest and much of the stock-return
distribution remains to be explained. The problem here is more one of
theory than of statistics and experimental design. The theoretical ques-
tion that needs to be addressed is the development of models that identify
determinants of stock returns in general. Returns are transforms of
consecutive price changes (and dividends); the more basic analysis should
therefore explain asset valuation by accounting variables and other
descriptors of the firm and/or the economy. Few concrete or useful
results exist, and the fundamental theoretical question of "why earnings? "
(rather than other informational items) remains essentiaUy unanswered.
Nevertheless, some statements can be made on this issue, and we shaU
have occasion to further discuss security valuation in section 3.2.^^
Third, there is always the possibiUty of omitted variables. One cannot
be sure whether eamings, rather than variables conelated with eamings,
matter. ^^ This problem is particularly relevant to the retum-variance
methodology, since eamings are rarely if ever disclosed alone. But it is
also obvious that the API methodology cannot prove conclusively that
eamings are more relevant than the numerous other items (e.g., sales)
disclosed in accounting reports that are quite likely to exhibit a high
degree of correlation with eamings. A more conservative interpretation
of the information-content findings is therefore that accounting reports—
rather than eamings in particular—provide timely and relevant infor-
"A related issue which had recently received some initial attention is the observed
cross-sectional differences in the retum reaction to eamings announcements. Why do some
stocks react to eamings disclosures more than others? Is this due to the existence of more
information prior to the eamings announcement? If so, what are the economic circimistances
triggering differential effects? Is it related to the exchange in which the stocks are traded
and/or to insiders' market activities? For initial research on this issue, see Grant [1980],
Atiase [1980], Finnerty [1976a; 1976ft], and Penman [1982a],
" This problem is aggravated in the absence of a well-specified theory relating accounting
data to parameters of security prices. Inferences in this case rely purely on the extent of
observed associations. Also, when research results are evaluated, the impossibility of
"perfectly controlled experiments" in MBAR should constantly be kept in mind.
mation. This, however, would seem to be indisputable. Sharper inferences

can be ascertained only by additional testing, such as that outlined in the
next section.
2.1.2 The Infonnation Content of Other (Noneamings) Financial
Data. Two financial variables were singled out for examination of their
incremental (over eamings) information content—earnings forecasts and
dividends. Foster [1973] reported a market reaction to preaudited EPS
figures announced by management. PateU [1976] found a significant
market reaction during the week that managements' eamings forecasts
were reported in the Wall Street Journal. Gonedes, Dopuch, and Penman
[1976] used analysts' forecasts to form portfoUos and found that these
portfoUos jaelded higher returns than control portfoUos not based on
eamings forecasts. Penman [1980] reported a significant price reaction to
the announcement of managerial eamings forecasts." The existence of a
market reaction to eamings forecasts has thus been established.
Empirical studies generaUy suggest that unexpected earnings and un-
expected dividends do not convey identical information (see Ashley
[1962], Pettit [1972; 1976], Griffin [1976], Laub [1976], Brown, Finn, and
Hancock [1977], Aharony and Swary [1980], and Penman [19826]). Div-
idend data appear therefore to be informative after consideration of
earnings information, and vice versa.^^
The incremental information content of other financial announcements
received relatively Uttle attention. On the surface, this seems somewhat
surprising since, as was mentioned above, unexpected eamings explain
only a relatively smaU part of the unexpected return distribution. Of the
few attempts to investigate this issue, we should mention Benston's
[1967] study. Upon regressing unexpected stock returns on unexpected
eamings and sales data, stock returns were found to be only marginaUy
responsive to changes in both eamings and sales. No study we are aware
of has since focused on the information content of sales data.
Gonedes [1974], using the BaU-Brown methodology, examined a variety
of financial ratios in addition to EPS, reporting that EPS captured most
of the information content of the accounting data examined. Foster and
Vickrey [1978] provided some statisticaUy weak evidence that informa-
tion disclosed in 10-K reports, but not in the annual financial statements,
induced greater-than-average price variability, implying that this incre-
mental 10-K information is used by investors. Foster [1981] examined
the cross-sectional interdependencies in accounting information, that is,
whether the announcement of eamings by one firm provides infonnation
relevant to the valuation of other firms having simUar economic charac-
teristics. For both the NYSE and the AMEX, he found significant
information interdependencies across firms in the same industry.
Of considerable importance is the issue of earnings versus cash flows.
'* For a survey of early earnings forecasts studies, see Abdel-khalik and Thompson
[1978], See also Ball and Foster [1982],
" Exceptions to these findings are reported by Watts [1973; 1976] and Gonedes [1978],
Given strong beliefs, particularly among some practitioners, that cash

flows reflect better than accrual earnings the economic realities of a firm's
performance, it is surprising that so little empirical research has been
devoted to this issue. Some initial attempts using a very crude definition
of cash flows (i.e., earnings plus depreciation) failed to reveal a significant
difference between the two performance measures. Among these are Ball
and Brown [1968] on the stock-price reaction to earnings and cash flows,
and Fama and Babiak [1968] on the relationship between dividend
changes and earnings/cash flows. An investigation of the incremental
(over earnings) information content of properly defined cash flows, in-
cluding all noncash revenues and expenses and changes in working
capital, is clearly called for.
Thus, while the existence of noneamings financial variables that convey
information to investors is both expected and (weakly) empirically sup-
ported, these issues have not been researched adequately. Given the
widespread criticism of historical-cost earnings in general, and related
accruals in particular, it is somewhat surprising that the incremental (and
marginal) information content of the extensive noneamings data in
financial reports has hardly been investigated.^® However, this is not
necessarily a straightforweird matter. Empirical studies have to a large
extent relied on AP7-metric types of methodologies, and these have had
a tendency to generate ambiguous evidence when more than one infor-
mation item is subject to investigation. A review of the earnings versus
dividends studies, for example, illustrates the extant complexities. The
problem is that all such studies must specify models of expected earnings
and expected dividends, and this tends to lead to a bewildering array of
alternatives. Furthermore, competing (or complementary) information
items are likely to be correlated with earnings variables. The confines of
this paper prevent a full discussion of experimental designs that might be
able to handle the problem. We should mention, however, that the
matching of all but one of the information items ("treatment variables")
could serve as a useful experimental design. Another approach, recently
applied by Beaver and Landsman [1982], derives the incremental infor-
mation content of a given item relative to some other data by first
regressing the former (e.g., inflation-adjusted earnings) on the latter (e.g.,
historical-cost earnings) and then examines the association between stock
returns and two independent variables: the original variable (e.g., histor-
ical-cost earnings) and the residuals from the aforementioned "first-
stage" regression; the latter reflecting the incremental information con-
tent of the second variable (e.g., the inflation-adjusted earnings). This
procedure permits the use of two completely uncorrelated independent
variables in explaining returns. On the surface at least, the multicolline-
arity problem is circumvented by this procedure.
" There is indirect evidence on the importance of noneamings data from bankruptcy
and bond-ratings prediction models, (For a review of this work, see Ball and Foster
[1982],)
The marginal information contribution of the various disclosed ac-

counting data is important, for example, for deciding on the degree of
fineness of aggregation in hierarchical reports like 10-K, annual, and
interim financial statements. One example of such a policy issue is the
SEC's current deliberation on "streamlined registration-system" regula-
tion. The existing policy environment therefore suggests that research
could usefully focus on what information beyond earnings matters. As an
added benefit, this research would probably stimulate improvements on
and understanding of available multivariate experimental designs used to
explain residual stock retums.
2.2 VOLUNTARY DIFFERENCES AND CHANGES IN ACCOUNTING
TECHNIQUES: EFFECTS ON INVESTORS, FIRMS, AND MANAGERS
Until recently, research in this area was strictly concemed with inves-
tors' reaction to differences and changes in accounting techniques, al-
though the expected direction and magnitude of reaction was rarely
specified a priori. The issue, stated somewhat superficially, was whether
investors are "misled" by the use of altemative accounting measurement
and disclosure techniques or—the more compelling hypothesis—are able
to "see through" the veil of accounting practices." The basic idea is
straightforward: rational individuals are not concemed with the
"packaging" of information; their beliefs about future states are unaf-
fected by the form of disclosure. Hence, if there are no effects on firms'
cash flows, then it follows that market values should be unperturbed by
firms' choices of (cross-sectional differences) or changes in (time-series
differences) accounting techniques. In a sense, the efficient market hy-
pothesis demands such value invariance. The stipulation of no impact of
accounting changes on cash flows underlies the first class of studies to be
examined. Next, the set of studies analyzing changes in accounting
techniques when cjish-flow impacts are present is examined. Finally,
recent research that extends the issue to examine managerial motives in
selecting accounting techniques and the effects of such choices on the
value of firms through agency-type relationships (e.g., bond covenants)
is discussed. Note that this section deals with voluntary (discretionary)
accounting changes, while the next one (2.3) reviews mandatory changes.
2.2.1 Accounting Cross-Sectional Differences—No Direct Cash-Flow
Effects. Beaver and Dukes [1972] examined the association between stock
retums and earnings based on both the "deferral" and the "flow-through"
methods of accounting for the interperiod tax allocation. Security retums
were found to be more highly eissociated with earnings btised on the
"deferral" method, leading the authors to conclude that this accounting
technique provides more useful information (than the "flow-through"
" This is known as the "naive investor" or "functional fixation" hypothesis. Allegations
about managers' ability to affect stock prices by judicious choices of accounting techniques
appear frequently in both the academic and business press. See, for example, Briloff [1972;
1976].
method) to investors. This was not a direct test of investor abiUty to

adjust for accounting differences, but an application of a methodological
approach (of which additional examples are discussed later) which under
certain circumstances (discussed in Beaver and Dukes [1972]) permits
the ranking of alternative information systems according to their useful-
ness to investors in assessing security returns. However, this study is
noteworthy because it employs investors' "ability to see through account-
ing techniques" as a maintained hypothesis.
A direct test of investor ability to adjust for alternative accounting
methods is provided in the sequel by Beaver and Dukes [1973], who
examined firms that for tax purposes used accelerated depreciation yet
for reporting purposes used either accelerated or straight-line deprecia-
tion. The two samples were matched on systematic risk and earnings
growth, so that if investors could penetrate the veil of depreciation
accounting, differences in P/.E-ratios should reflect the different effects
of depreciation methods on earnings.'® This was found to be the case.
Once the earnings of the straight-line depreciation firms were adjusted to
accelerated depreciation, the P/.E-ratio means of the two samples were
almost identical. Good and Meyer [1973] confirmed that differences
between firms with high and low P/E-r&tios diminish when earnings are
adjusted for differences in accounting methods used for depreciation and
extraordinary items. Eskew [1975] reported that investors in oil and gas
stocks were able to adjust for differences between the "fuU-cost" and
"successful-efforts" methods of accounting for explorations. (More on
this issue in section 2.3.7.) Hong, Kaplan, and Mandelker [1978] reported
no statistically significant price reaction at the merger date for firms
using the "pooling" method to record mergers, despite the higher account-
ing earnings and rates of returns posted by these firms.'^ However, firms
using the "purchase" method showed, on average, significant positive
excess returns during the month of merger. This raises a serious problem
in interpreting results in this area, caused by the possibility of a self-
selection bias. Specifically, if the event of merger or the choice of the
"purchase" method is associated with unexpected earnings increases (e.g.,
firms with unexpectedly large earnings tend to select the income-depress-
ing "purchase" method), the observed market reaction might be due to
the performance of firms rather than to the choice of accounting methods.
This confounding effect due to the possibility of a selection bias will come
up again and again to muddy the interpretation of empirical findings.
Nevertheless, studies in this area generally point to investors' ability to
adjust for differences in a few well-known and clearly disclosed accounting
techniques.
2.2.2 Accounting Changes Having No Direct Cash-Flow (Tax) Effects.
" It has been subsequently established, Beaver and Morse [1978], that cross-sectional
differences in P/E ratios are mainly due to differences in earnings growth. Risk appears to
be a variable of much less, if any, significance in explaining cross-sectional P/E differences.
'* Relative to firms using the "purchase" method.
Archibald [1967; 1972] examined the stock prices of firms that switched
firom accelerated to straight-line depreciation (without changing their
tax-accounting method) and concluded that no significant price reaction
occurred during the month of earnings announcement (even though the
accounting change increased reported earnings by about 10 percent).^
Kaplan and Roll [1972] examined firms switching from accelerated to
straight-line depreciation and firms switching from deferral to flow-
through accounting for the investment tax credit. Both changes improve
reported earnings but have no direct tax consequences. Somewhat sur-
prisingly, they found that the returns of the depreciation-switching firms
fared worse than the market average during the 30 weeks following the
earnings announcements. For the firms switching to the flow-through
method from the deferral method, a temporary increase in price occurred
at the earnings announcement date. However, during the 30 weeks after
the announcement, the switching firms' stocks fared worse than those of
firms that did not switch. Nevertheless, the overall conclusion of Kaplan
and Roll was that investors were not "fooled" by the switches in account-
ing techniques. Cassidy [1976], using a different methodology and data,
confirmed the negative stock performance for the investment tax credit
case.
Ball [1972] examined 430 cases of accounting changes, some having
cash-flow effects and others not, and concluded that investors were able
to distinguish between the real and "cosmetic" effects of accounting
changes on earnings. Harrison [1977; 1978] compared the market perfor-
mance of firms making discretionary as well as nondiscretionary accoxmt-
ing changes with the performance of similar firms that made no account-
ing changes. One striking finding was that firms which made discretionary
changes resulting in earnings (but presumably not cash-flow) increases
experienced returns below those of the control group.^' Moreover, differ-
ential security rates of returns between the two samples persisted beyond
the disclosure date of the accounting change. Note that both of these
findings are similar to those of Kaplan and RoU [1972]. A somewhat
disturbing feature of the Ball and Harrison studies mentioned above is
^ For completion sake, we should mention here several efirly studies on the market
impact of alternative depreciation and investment tax credit techniques: O'Donnell [1965],
Mlynarczyk [1969], and Comiskey [1971], These authors have employed certainty-based
stock valuation models (e,g,, the earnings capitalization model and the P/E formula), rather
than the more popular (and less restrictive) method of observing price reaction to accounting
changes. Although the unanimous conclusion of these studies was that the market did not
react naively to differences in accounting methods, it was not always consistent with the
data. For example, Mlynarczyk's findings appear to indicate that investors did not adjust to
accounting differences in the first two years after the change. The small samples employed
and the certainty-based stock valuation models mitigate against making inferences from
the findings of these studies,
" Again, note that if firms experiencing unexpected profit decreases or even losses tend
to select income-increasing accounting techniques, then such an observed negative market
reaction might be associated with the fortunes of the firms rather than with the accounting
changes. This is a reflection of the possible selection bias discussed above.
the grouping of different types of accounting changes (discretionary and

nondiscretionary, with and without direct cash-flow effects) to test for
market reaction, without an unambiguous preliminary specification of
the expected effects. Consequently, the observed "average" sample re-
action (or nonreaction) to accounting changes might be due to confiicting
and partially offsetting effects of different accounting changes, rendering
interpretation of the findings difficult. While most authors mentioned in
this section generally concluded that investors reacted rationally to
changes in accounting techniques, the existence of somewhat unexpected
findings seemingly inconsistent with investor rationality cannot be ig-
nored. We retum to this issue below.
2.2.3 Accounting Changes Having Substantive (Direct Cash-Flow)
Effects. Direct cash-fiow effects are generally due to the tax implications
of the accounting change. Sunder [1973; 1975], the first to examine this
issue, reported that firms switching to the LIFO inventory method,
thereby decreasing their reported earnings but increasing after-tax cash
fiows during inflationary periods, experienced positive excess returns in
the 12 months preceding the announcement of the change.^^ This finding
seems to confirm the market's rationality, since it suggests a reaction to
the expected cash-fiow increases rather than to the reported-earnings
decreases. However, for a relatively small sample of firms (about 20) that
switched from LIFO to FIFO, Sunder [1975] did not detect the expected
negative market reaction.
Sunder's findings about the "correct" market reaction to the cash-fiow
increase due to the LIFO switch are, of course, consistent with an
altemative hypothesis: firms tend to switch to LIFO and incur the
consequent reported-earnings decrease during unusually good earnings
periods. Positive investor reaction, observed by Sunder for the 12-month
period preceding the switch, might therefore be due to the good fortunes
of firms rather than to the consequences of the LIFO switch. Thus, the
possibility of a self-selection bias must again be considered. Several
studies have attempted to alleviate the self-selection problem and thus
achieve a sharper distinction between the "rational" and "functional
fixation" hypotheses. Results were rather surprising. Abdel-khalik and
McKeown [1978] classified the L/FO-switching firms into two groups
according to the sign of the unexpected earnings change; firms with
positive unexpected eamings that switched to LIFO performed better
than a control group, while the firms with negative unexpected earnings
that switched to LIFO performed worse than a control group (i.e.,
experienced lower stock retums). These findings suggest that investor
reaction was associated with the earnings perfonnance of the switching
firms rather than with (or in addition to) the accounting-method switch.
^ Subsequent to the change the residual retums did not exhibit a discernible pattern,
except for a large negative average residual in the reporting month, which still awaits
adequate explanation. It should also be noted that a systematic risk (P) shift occurred for
the L/FO-switching firms.
Ricks [19826], controlling for the fact that L/FO-adopting firms generally
exhibited unusual earnings increases during the examined period (1974-
75)—a point Sunder did not control for—found that the LIFO-adopting
firms experienced significantly lower residual returns than a control group
during the month of the change announcement. Moreover, Ricks found
that when earnings of LIFO firms were adjusted to a FIFO basis, the
postswitch P/E-r&tios of LIFO firms were lower than those of the control
group. (The FIFO data were derived from the footnotes in the annual
reports; this procedure was not feasible at the time of Sunder's study.)
Similarly, Brown [1980] reported results consistent with Ricks'—an ad-
verse market reaction for the L/FO-switching firms.
The post-Sunder LIFO-switch studies are discomfiting for a number of
reasons. First, the results, taken at face value, are not consistent with the
notion of investor rationality that is traditionally used as a maintained
hypothesis. This is disturbing because there are no behavioral alternatives
to rationality; concepts of "bounded rationality" have never proved to be
particularly useful in economic theory. Hence, one is at a loss to provide
a theoretical explanation of subsequent LIFO results. Second, the impres-
sion left is that much of the early work testing investor rationality simply
was not powerful enough to reject the smooth and attractive rationality
hypothesis. Curiously enough, sufficiently crude econometric techniques
(and crude or incomplete data) are almost sure to accept the expected
conclusion, that is, the null hypothesis of no reaction. Much the same
applies to tests of the efficient market hjrpothesis in general, as evidenced,
for example, in the discussion of market "anomalies" in the June/Septem-
ber 1978 issue of the Journal of Financial Economics (see also section
2.4.1). This suggests that more powerful experimental designs might
reject previously well-established conclusions, such as those in the Beaver
and Dukes [1973] study on the market's rational adjustments to alter-
native methods of depreciation. This seems to be an interesting area for
further work. Indeed, the L/FO-switch issue appears to be still unfolding.
In a recent study, Biddle and Lindahl [1982] sharpened the research
focus by examining the association between residual stock returns and
the realized tax savings due to the LIFO switch, rather than merely
examining market reaction to the accounting switch. In addition, unex-
pected earnings were controlled for, as in the Ricks [19826] study. With
some exceptions, results seem to indicate a positive association between
residual stock returns and LIFO tax savings; this is consistent with
investor rationality.
Finally, is the cash-flow versus non-cash-flow dichotomy underlying
the above research useful beyond a first-order approximation? The LIFO-
adoption studies, for example, have demonstrated that the decision to
switch accounting methods is most likely nonrandom, even beyond the
tax consequences. This suggests that there might always be cash-flow
effects (not necessarily pertaining to taxes) of every managerial decision,
including the choice of accounting method. Accordingly, the somewhat
gloomy observations about investor rationality made above should be

qualified to say that there might be various "hidden" but significant cash-
flow consequences associated with all accounting choices. It is necessary
to examine the motivation of those individuals who decide whether to
change an accounting method or not, namely, the managers. Such a
discussion is provided in the next section.
2.2.4 Management's Motives in Selecting Accounting Techniques. The
observed market reactions to seemingly nonsubstantive accounting
changes raise the question whether there is such a thing as a merely
"cosmetic" change. Although an accounting method change, such as from
accelerated to straight-line depreciation, may have no immediate or
direct effect on cash flows, other subtler and indirect determinants of
cash flows could be affected. For example, management compensation
schemes are generally a function pf reported earnings and hence sensitive
to changes in accounting techniques. There are numerous other contrac-
tual provisions, specified in terms of accounting data, that thus delineate
investment/financing opportunities of the firm. Bond covenants, such as
dividend-payment restrictions, are obvious examples of such investment/
financing restrictions. These observations lead the examination to a
broader, more fundamental question. What are managers' motives and
opportunities in selecting among alternative accounting techniques, and
what is the impact of these choices on future operations of the firms?
Not much is currently known about these issues. An early line of thought
attributed the selection of alternative accounting techniques to managers'
preference (derived from stockholders') for smoothing earnings (e.g.,
Gordon [1964]). Other economic, political, and sociological motives have
been proposed (see, e.g.. Watts [1974; 1977], Gonedes [1976], and Watts
and Zimmerman [1978]), yet only a few are supported by empirical
evidence.^ The theoretical constructs used in these studies are somewhat
incomplete, and they do not typically model rigorously multiperson,
multiperiod equilibria. Especially lacking are strategy considerations and
game-theoretical solution concepts that might contribute to the devel-
opment of formal theory.
Among empirical market studies, Holthausen [1981] attempted to
explain the market reaction to firms' switches fi-om accelerated to
straight-line depreciation by using the provisions of bond-indenture
agreements and management-compensation contracts as possible mana-
gerial motives for the accounting change. The switch from accelerated to
straight-line depreciation increases reported earnings and decreases the
book value of leverage, because it increases both net tangible assets and
^ For example, Hagerman and Zmijewski [1979] reported that the choice of accounting
techniques is related to firm size and management compensation plans. Jarrell [1979]
argued that utility managers use accounting techniques to revalue upward the asset base
used in the rate-setting process, DhaUwal [1980] reported that the choice between the "ftiU
cost" and "'successful efforte" methods for oil and gas exploration is related to the firms'
capital structure, which in turn may affect indentures in firms' loan agreements.
Stockholder equity. Consequently, the restrictiveness of any bond cove-

nants based on leverage and/or eamings is relaxed, possibly enhancing
the firm's ability to issue new debt, pay dividends, and invest in risky
projects.^ Furthermore, the accounting switch might also increase man-
agement compensation tied to eamings. These possible effects of the
accounting change on both stockholders and bondholders might therefore
be reflected in a market reaction to the accounting change. Unfortunately,
Holthausen's empirical results are disappointing. Unexpected stock-price
changes were not related to the existence of management-compensation
plans or to firms' deviation from their dividend constraints. The only
exception was that unexpected price changes were found to be negatively
related to leverage, suggesting that bond covenants in the form of leverage
constraints might have influenced the decision to switch depreciation
method. Collins, Rozeff, and Dhaliwal [1981] on oil and gas accounting
and Leftwich [1981] on accounting for business combinations investigated
the relationship between the observed negative market reaction associ-
ated with the accounting change and various contractual relationships.
Some very weak evidence of a link between the negative market reaction
and possible consequences (costs) of loan contracts was found. These
studies are, of course, only a modest beginning of the search for mana-
gerial motives in selecting accounting techniques and examination of the
indirect consequences of accoimting changes.^^
The statistically weak findings of the studies attempting to explain
accounting switches by contractual motives are disturbing. Is the failure
to detect significant relationships due to the crudeness of methodology
or to an examination of largely irrelevant or immaterial considerations?
Recall that stock-price reactions to an accounting switch were related by
researchers to the existence of contractual arrangements (e.g., manage-
ment compensation). However, given the small size of total management
compensation relative to the market value of the firm, can we really
expect changes in such compensation, resulting from accounting switches,
to be identified by a market reaction? In other words, it is doubtful
whether investors can be expected to react to changes in management
compensation resulting fi'om accounting switches to an extent discemible
by the somewhat crude statistical methodology employed (e.g., residual
analysis). The same argument can be applied to some loan covenants.
For example, will a slight relaxation of a loan restriction on the debt/
equity ratio due to an accounting change be regarded by investors as a
significant economic event inducing them to rearrange their portfolios
(i.e., cause a shift in the demand for the respective securities)? That most
loan covenants and management-compensation systems are not condi-
" Of course, this line of reasoning begs the issue as to why the possibility of accounting
switches is not prohibited in the covenant itself.
^ The obvious motive—increasing future cash flows via tax savings—does not need any
elaboration here. For an empirical study on the choice of inventory methods to maximize
tax savings, see Biddle [1980].
tional upon specific accounting methods (within GAAP) increases our

skepticism. If accounting changes could have a significant effect on
management compensation, would not investors specify in the compen-
sation contracts the precise accounting methods that should be used to
compute compensation? Would not lenders restrict managers in the use
of accounting methods were they concerned with possible manipulation
of the covenants by accounting switches? Given that accounting changes
are generally made infrequently and, once made, are highlighted in
financial reports, it seems doubtful that such changes possess the flexi-
bility and subtlety to serve as a discretionary managerial tool for effecting
contractual arrangements.^^
Empirical researchers in this area have skipped a crucial methodolog-
ical step. Prior to correlating stock-price changes with the existence of
contractual arrangements, they should have estimated the magnitude of
the expected agency costs (i.e., the costs associated with conflicts between
managers and investors/lenders). For example, it would be interesting to
know by how much a change from accelerated to straight-line deprecia-
tion increased management compensation in the sampled firms. Given
such quantitative estimates, the possible market reaction to an account-
ing change can be estimated.^' Furthermore, preliminary estimates of the
expected effects of agency costs on individual firms will enable researchers
to eliminate from the sample cases with negligible effects, thereby in-
creasing the probability of observing a significant relationship between
stock-price changes and contractual agreements. To summarize, the
attempt to explain managerial accounting choices by various contractual
arrangements (or within an agency-cost setting) has an obvious appeal.
However, the empirical validity of this explanation has yet to be substan-
tiated. The empirical identification of managerial motives for making
accounting choices, beyond the obvious one of maximizing after-tax cash
flows, still eludes us.
2.2.5 Evaluation. Can investors "see through" the veil of alternative
accounting techniques? At this stage the evidence does not warrant an
unequivocally affirmative answer. Relatively few cases of cross-sectional
differences in accounting practices have been examined (depreciation,
investment tax credit, and corporate acquisition methods); their impact
on earnings is easily derived and well understood, and yet some anomedies
have been reported (e.g.. Hong, Mendelker, and Kaplan [1978]). Still
^* There are several possibilities here: one, that the effect of accounting changes on
contractual arrangements is negligible for all parties involved. Alternatively, the effect on
managers can be significant from their perspective (e.g., increasing their compensation), yet
the impact on investors and lenders is relatively negligible. In either case market studies
would fail to indicate the motives of accounting changes.
" Note that this comment applies to many other studies reviewed in this paper. In most
cases researchers did not specify in advance the direction and magnitude of the expected
market impact of the event examined. One of the few exceptions is the Biddle and Lindahl
[1982] study mentioned above, in which the tax savings resulting from a LIFO adoption
were determined and then related to residual returns.
open is whether, and at what speed, investors can adjust for subtler
differences in accounting practices that are only partially disclosed, such
as the different assumptions firms can make in estimating pension liabil-
ities, accounting for contractual projects, or the extent of capitalizing
overhead costs (absorption costing).
The evidence on accounting changes, while indicating some ability of
investors to distinguish substantive from nonsubstantive changes, is
marred by surprises. (Strangely enough, some still insist on referring to
them as anomalies.) On the whole, the evidence is consistent with a
multitude of hypotheses, none of which can at this stage be either
reasonably substantiated or entirely ruled out. For example, the market
reaction observed to the disclosure of accounting changes having no
direct cash-flow effects (e.g., Kaplan and Roll [1972]) may indicate that
such changes provide substantive signals about management attitudes
and behavior and about the economic characteristics (e.g., profitability)
of the firm. Here, as in many other cases, self-selection bias is not only a
possibility but is also often consistent with empirical results. For example,
the firms that switched to a less conservative accounting method in the
Kaplan-RoU study had, on average, poor earnings records prior to the
switch (which helped increase earnings). Accordingly, the observed mar-
ket reaction could have been due to the profitability performance of the
firms rather than to the accounting change. Yet another alternative
hypothesis, the observed negative market reaction to accounting changes
that merely increase reported earnings (e.g., Harrison [1977; 1978]), might
suggest that investors are wary of accounting manipulations and are
willing to pay a "premium" for conservative measurement techniques.^
FinsiUy, despite obvious reluctance to admit that the "functional fixation"
hypothesis could be valid, some of the evidence is not inconsistent with
prices that do not always reflect the implications of publicly available
information and with investors having preferences for nominally high
(and/or less volatile) earnings even at the expense of after-tax cash
flows.^^ Particularly disturbing are the post-Sunder findings on the mar-
ket reaction to LIFO adoptions (e.g.. Ricks [19826]). However, the recent
broadening of the scope of analysis to include management motives for
selecting accounting techniques and the examination of various indirect
effects due to contractual arrangements add an important dimension to
the examination of accounting-choice consequences. Despite the initial
mixed results and various methodological difficulties, this Une of research
should be continued as a complement to investigations of the "functional
fixation" hypothesis. As elaborated above, the preferred direction seems
^* More on this argument in Fabozzi [1978] and Hawkins and Campbell [1978].
^ Consider the following comment from the (October 7, 1974) Wall Street Journak
"Because of the negative impact on earnings LIFO conversions haven't always been
welcome news to the stock market Also companies worried about a takeover may be
reluctant to risk the kind of downward pressure on stock prices that has greeted other
companies when they recently converted to LIFO,"
to be a preliminary evaluation of the expected impact of accounting

changes on contractual arrangements and the identification of the ma-
terial arrangements for further market study.
Research in this area, although voluminous, is still in its infancy, and
strong policy recommendations, for example, that the FASB should not
concern itself with uniformity of accounting techniques or with how
financial information is "packaged" (e.g.. Beaver [1973]), are at best
premature. At worst, the potential dysfunctional effects of such a policy
implementation could make investors uncomfortable and cause some to
"withdraw" from the market. The dysfunctional effects might result from
perceived significant informational as5Tnmetries due to the absence of
uniform accounting techniques, and theory suggests that trading is re-
duced whenever informational asymmetries remain in the market. Such
reduced trading eu-gues against efficient risk sharing, and a welfare loss is
incurred.^ (See also section 3.1.)
2.3 THE MARKET IMPACT OF ACCOUNTING REGULATION
We move now to the study of the consequences of accounting regula-
tion. Evaluation of consequences is, or at least should be, an integral part
of any regulatory process. Financial market consequences are of partic-
ular relevance to accounting regulators; many empirical studies have
investigated the market's response to the deliberations surrounding and
enactments of specific kinds of regulation and to the information content
of data that companies were to disclose as a result of enacted regulation.
Given the complexities of pronouncements by accounting standard-set-
ting bodies (the FASB and its predecessors, the SEC, etc.), research on
the consequences of regulation is, as expected, thin. We begin our evalu-
ation of this area with the few regulations that have attracted relatively
concerted research effort (lines-of-business, oil and gas accounting, and
replacement costs) and subsequently consider briefly the less-researched
accounting mandates.
2.3.1 Lines-of-Business (LOB) Reporting (SEC [1969, 1970], FTC
[1974, 1975], and Financial Accounting Standard [FAS] No. 14
[1976]). What is the information content of LOB reports? The first group
of studies investigated the association between LOB information and
mean stock retums. Collins [1975] examined the first SEC-mandated
LOB filings (fiscal 1970), which included comparative data for 1967-69
that were, of course, not disclosed during 1967-69. Using an earnings-
prediction model based on consolidated data only against one that was
based on the LOB data, Collins employed a trading strategy that called
for buying the stocks of firms for which LOB-based earnings forecasts
exceeded the consolidated-based forecasts and selling short the stocks of
firms for which the opposite was true. This trading rule, which exploits
the LOB information, yielded excess returns in 1968 and 1969 but not in
" See, for example, Wilson [1978].

276 B, LEV AND J. A. OHLSON
1970, leading Collins to conclude that segmented data (especially sales)

do provide information beyond aggregated reports. LOB data were also
examined by Kochanek [1975] and by Griffin and Nichols [1976], who
reported that security prices of NYSE firms with extensive segmented
data adjusted to earnings announcements faster than prices of firms with
minimal segmented data. These findings suggest that LOB information
leads to an improved forecasting by the market of future earnings.
Does LOB information affect common stock risk-return relationships?
Kinney [1972] reported a significant association between covariation of
segment earnings and stock systematic risk (/8). Horwitz and Kolodny
[1977] tested whether a shift in systematic risk occurred with the SEC's
LOB disclosure and also compared the stock-price reaction to disclosures
of firms affected by the regulation with that of a control group unaffected
by it. No shift in risk and no price reaction were observed. However,
Collins and Simmonds [1979], using a more advanced methodology,
concluded that LOB disclosure was associated with a statistically signifi-
cant decrease in the systematic risk of the disclosing firms. Similarly,
Ajinkya [1980] concluded that the increased uniformity and greater
fineness of disclosure mandated by the SEC increased investor consensus
on the risk-return assessment of securities at the aggregate level.^^ Despite
some inconsistencies, the weight of evidence lends support to the conten-
tion that the mandated LOB information, particularly sales figures,
affected investor assessment of the return distributions of multiproduct
firms.
2.3.2 Oil and Gas Accounting (FAS No. 19 [1977]). Patz and Boatsman
[1972] examined price behavior of oil and gas stocks surrounding the 1971
Accounting Principles Board (APB) proposal to mandate exclusively the
"successful-efforts" method for unsuccessful oil and gas explorations. No
statistically significant differences were found between the retiims of
firms using the "full-cost" method and those using the "successful-efforts"
method. However, this result was challenged by O'Connor and Collins
[1977], who reported observing a price reaction to the APB proposal.
Patz and Boatsman's nonreaction findings might be due to a "timing"
problem; the market was likely to have reacted earlier than the APB
proposal date, for the APB position was widely known beforehand.
The market reaction to the July 1977 FASB exposure draft on oil and
gas accounting is probably the most hotly contested regulation research
issue. The participants on the academic front were Dyckman and Smith
[1979], who, despite noting a negative impact around the exposure-draft
disclosure date on the residual returns of "full cost" firms, concluded that
their results did not support the existence of a substantial market effect
of the exposure draft. And Haworth, Mathews, and Tuck [1978], who
found that "full-cost" firms were initially adversely affected by the
^' See also Horwitz and Kolodny [1980] for an overall assessment of the SEC's LOB
regulation, and Twombly [1977] for a study suggesting that the SEC's LOB regulation did
not provide new information to capital markets.
exposure-draft release but that there was a subsequent price recovery.

On the other hand, both the Collins and Dent [1979] and Lev [1979]
studies reported a permanent risk-adjusted decline in the stock prices of
"fiill-cost" firms close to the exposure-draft disclosure date. Collins,
Rozeff, and Salatka [1982], in an attempt to resolve this controversy,
tested whether the SEC's August 1978 rejection of FAS No. 19 (i.e., the
SEC allowed the continuation of the "fiill-cost" method) triggered a
market reaction (the "reversal method"). Results indicated an excess
return for "fiill-cost" firms exceeding that of "successful-efforts" firms
during the week of the SEC decision, thus reversing the negative market
reaction sustained by "fiill-cost" firms at the time of the FASB exposure-
draft release (July 1977). A similar market reaction to the SEC reversal
of FAS No. 19 was reported in Smith [1980]. These findings can be
interpreted as corroborating the Collins and Dent [1979] and Lev [1979]
findings with respect to the negative FASB exposure-draft impact. The
weight of evidence therefore suggests that both the FASB and the SEC
regulations on the "full-cost"/"successful-efforts" issue were associated
with statistically significant stock-price reactions of oil and gas stocks.^^
2.3.3 Inflation Accounting (SEC's ASR No. 190 [1976] and FAS No.
33[1979]). With a refreshing uniformity, the empirical studies on the
market impact oiASR No. 190 (Arbel and Jaggi [1978], Beaver, Christie,
and Griffin [1980], Gheyara and Boatsman [1980], Ro [1980; 1981], and
Abdel-khalik and McKeown [1978]^^) were unanimous in their failure to
detect a price or volume reaction to the SEC regulation mandating the
disclosure of replacement-cost data. Specifically, there was no noticeable
market reaction to either the initial proposal of this requirement, to the
date it became effective, or even around the dates when the actual
replacement-cost data were filed with the SEC.** Although the research-
ers mentioned above used basically the same set of data, the methodo-
logical approaches were rather varied, leading to increased confidence in
the finding of no market reaction. One potentially serious methodological
problem shared by most replacement-cost studies is related to the "firm-
size effect," because replacement-cost samples consisted of firms substan-
tially larger than the control samples. Since firm size has been found to
be negatively associated with stock returns and with the volume of trade
(see section 2.4.2), the strength of these consistent empirical findings is
somewhat diminished.^^
Obviously, the nonreaction results are consistent with different hy-
potheses, such as the one that replacement-cost data provide investors
with potentially useful information, but this information has already been
^ See Foster [1980] for a discussion of various methodological issues in these oil and gas
studies.
"^ This study was concerned with Value Line replacement cost estimates.
" A price reaction at the time the regulation became effective might have given some
indication of the market's assessment of the cost of complying with this allegedly expensive
disclosure requirement.
** For elaboration on the firm-size effect in this context, see Freeman [1981].
impounded in stock prices due to private search efforts by analysts.^

This possibility has been examined by Freeman [1982], who found a
correlation between the current year's replacement-cost eamings and the
previous year's stock returns. These correlations are particularly strong
at the industry level, which is consistent with the proposition that
aggregate figures (rather than firm-specific) are easier to forecast and are
thus impounded in prices several months before actual disclosure. An
alternative explanation of the nonreaction findings is that the replace-
ment-cost information may be irrelevant for security-price formation or,
while it is conceptually relevant, the replacement-cost estimation is so
subjective and riddled with measurement errors as to render the pub-
lished data unreliable. Finally, as in all other cases, it cannot be ruled out
that the nonreaction finding is due to methodological limitations.^^
Noreen and Sepe [1981a] applied a version of the residual retum
"reversal method" mentioned earlier to the study of market reaction to
the FASB deliberations on inflation accounting. Suppose that the values
of some firms' stocks were affected by the initial FASB proposal to
disclose inflationary effects. Then the firms which were favorably affected
at the time the initial proposal was contemplated are most likely to be
affected in the same (reversal) fashion by subsequent events that increase
(decrease) the probability of regulation adoption. An analogous scenario
is hypothesized for firms that were unfavorably affected by the proposal.
If such a retum pattem across time exists, it may be possible to detect it
empirically by computing cross-sectional correlations of excess returns
for those events that presumably changed the probabilities of regulation
adoption.^ In the case of the FASB inflation-accounting deliberations,
three events were examined by Noreen and Sepe: the initial proposal of
mandatory disclosure (January 1974), the postponement of a decision
(November 1975), and the announcement of a new exposure draft (Jan-
uary 1979). The patterns of returns correlations supported the hypothesis
that stock prices were affected by the FASB deliberations.^ As to the
disclosure effects of the FAS No, 33 data, initial findings by Beaver and
Landsman [1982] suggested that the cross-sectional correlations between
stock returns and historical costs (unadjusted) eamings were higher than
^ This hypothesis is consistent with the findings of Easman et al, [19791, indicating a
stronger correlation between stock prices and replacement-cost-adjusted eamings than
between stock prices and historical-cost earnings, Baran, Lakonishok, and Ofer [19S0]
provide evidence that association between market betas and general price-level-a^justed
(accounting) betas is significantly higher than those observed between market and histori-
cal-cost betas,
^^ See Watts and Zimmerman [1980] for a detailed discussion of some of these findings
with particular attention to the cost/benefit issue of ASR No, 190,
^ Note that this methodology does not require ex ante specification of the direction and
magnitude of price reaction on individual firms, as is required, for example, in the BaU and
Brown [1968] methodology. This is similar to the variance tesU (Beaver [1968]) that also
circumvent the directional exogenous expectations model requirement,
^ For a methodological criticism of Noreen and Sepe's "correlation-based approach," see
Basu's [1981a] comment and Noreen and Sepe's reply [1981fr],
the correlations between stock returns and the various price-adjusted

estimates of earnings, a finding consistent with the nonreaction to re-
placement costs discussed above.
The main conclusion to be drawn from the research on lines-of-busi-
ness, oil and gas, and inflation-accounting regulation is that a concerted
research effort into the market consequences of regulation yields conver-
gent results. The accounting pronouncements examined below differ from
those above in the paucity of empirical research devoted to them, and
the considerable inconsistencies in the findings make it difficult to draw
clear-cut conclusions.
2.3.4 Sales Disclosure by the 1934 Securities Act. Benston [1973]
examined the impact of the 1934 Securities Act, especially the aspect of
mandated sales disclosure. Two samples were studied: one consisted of
firms that had not disclosed sales figures prior to 1934 (and, of course,
disclosed sales subsequently), and the other consisted of firms which
disclosed sales pre- and post-1934. Benston observed that the distribution
of residual returns had a smaller dispersion in the post-Act period for
both samples, but it was not significantly different for the disclosure and
the nondisclosure samples. Benston interpreted this finding as implying
that the sales-disclosure provisions of the 1934 Act were of no apparent
value to investors. This conclusion was hotly contested on conceptual
grounds by Friend and Westerfield [1975] and by Gonedes and Dopuch
[1974], and on empirical grounds by Deakin [1976]. Deakin found that
the 1934 Act affected the volatility of the Standard and Poor's index of
500 stocks and, in addition, triggered a market reaction to the first filings
under the Act (early 1935). There appears to be little doubt that the 1934
Securities Act had an impact on U.S. Security markets. The question
remains, however, whether one can say anything beyond that. How does
one evaluate the welfare impact of the 1934 Securities Act? The answer
to this question is in part conceptual, and one must consider carefiilly
what market-based variables are relevant. As the criticism of Benston's
[1973] study indicates, there is no obvious reason that the variance of
residual returns shovdd be the object of interest. The question raised is,
of course, quite fiindamental from a methodological viewpoint, and it
deserves more than a cursory discussion; this is provided in section 3.1.
2.3.5 Extraordinary and Special Items (APB Opinions No. 9 [1967]
and No. 30 [1973]). Gonedes [19756] concludedfi-omstock-return behav-
ior that "special items" do have information content but that the classi-
fication of such items by type does not provide incremental information
to the market. Eskew and Wright [1976] provided evidence corroborating
Gonedes' former conclusion but conflicting with his latter one: they
reported that the classification of special items provides investors with
relevant information. In a later study, Gonedes [1978] provided evidence
that conflicts with his previous findings with respect to the information
content of extraordinary items: neither unexpected dividends nor unex-
pected extraordinary items were able to explain security returns beyond
the information reflected in unexpected earnings. Gonedes attributed this
conflict to differences in experimental design, especially in the definition

of extraordinary items.
On the related issue of earnings components, Foster [1975] reported
that investors behave in a way that recognizes the information content of
components of earnings of OTC-listed insurance companies (namely,
underwriting earnings, investment earnings, and capital gains and losses).
Manegold [1981] reported that an unexpected earnings variable devel-
oped from a component-based forecast model (including operating in-
come, depreciation, and interest) was slightly more positively correlated
with unexpected stock returns than were unexpected earnings derived
from an extrapolation of aggregate earnings. Thus, the ambiguity about
the market consequences of "special-items" regulation is consistent with
our meager knowledge of the marginal iirformation content of noneam-
ings accounting data, discussed in section 2.1.2 above.
2.3.6 Lease Capitalization (SEC, ASR No. 147 [1973] and FAS No. 13
[1976]). May, Harkins, and Rice [1978] reported an anticipatory price
adjustment which occurred over the 20 months prior to the public release
of lease information mandated by ASR No. 147 and by APB No. 31
(1973). The actual lease disclosures in financial statements did not induce
any identifiable extraordinary price adjustments. Ro [1978] reported that
the mean returns of firms whose earnings would have been materially (3
percent or more) affected had the leases been capitalized were lower than
the mean returns of a control group of firms. Finnerty, Fitzimmons, and
Oliver [1980] failed to detect a shift in systematic risk caused by infor-
mation mandated by ASR No. 147. Bowman [1980], using regression
analysis, concluded that the lease variable (measured with ASR No. 147
data) does make a statistically significant contribution to the
"explanation" of cross-sectional variation in systematic risk (/8). It should
be noted that the last two findings are not necessarily conflicting: the
relevEint lease information might have been impounded in the market's
risk assessments prior to its disclosure in the financial statements, and
hence the no-shift-in-risk finding.
The mandated SEC lease information thus appears to be associated
with the information set underlying the valuation of both the risk and
return of stocks. This suggests that the market is fairly sophisticated
regarding lease data. As to the subsequent FAS No. 13 requirement of
lease capitalization, Abdel-khalik et al. [1981] failed to observe stock- or
bond-price reaction.*" Pfeiffer [1980] examined market reaction during
seven periods coinciding with major announcements on the FASB lease
deliberations. He reported a negative market reaction near the date of
the FASB public hearing (this period had not been examined by Abdel-
khalik et al. [1981]). Furthermore, the negative reaction was found to be
*° However, see Abdel-khalik, Thompson, and Taylor [1978] for a discussion of the
economic effects of FAS No. 13 on managerial decisions of lessee and lessor companies.
Apparently the regulation affected various financial and investment decisions.
associated with the effects of lease accounting on restrictive debt cove-

nants (related to our discussion in section 2.2.4 above). OveraU, the
evidence appears to suggest that the FASB's more extensive lease-disclo-
sure requirement did not nwiterially improve upon the SEC regulation
mandating disclosure of lease infonnation in footnotes.
2.3.7 Miscellaneous. Dukes [1976] and Vigeland [1981] examined the
market reaction to FAS No. 2 (1974) mandating the expensing of research
and development costs and failed to detect a significant stock-price
reaction.*' The market impact of the highly controversial FAS No. 8 on
foreign currency translation was examined by Dukes [1978] and by
Shank, Dillar, and Murdock [1979], who failed to observe a noticeable
stock-price reaction. It would be interesting to test for a possible market
reaction to the recent modification of this regulation in FAS No. 51
(1981).
Rice [1978], examining the market impact of APB No. 15 (on eamings
per share), suggested that the market had begun assessing the groups of
firms affected and unaffected by the new regulation differently in the
year before the disclosure of fuUy diluted EPS was required.
Fried and Schiff [1981] examined the market reaction to the 1978 SEC
disclosure requirements for companies that change their auditors, in
which these companies were required to enumerate and describe disagree-
ments with auditors. A negative market reaction to a switch of auditors
was initially observed, but when the sample was partitioned by conflict
dimensions (the concem of the SEC), no market reaction could be
detected. Fried and Schiff accordingly concluded that there was no
market evidence supporting the usefulness of the SEC requirement to
disclose disagreements with auditors.
Baskin [1972] and Alderman [1977] examined market reaction to
auditors' consistency exceptions and uncertainty qualifications, respec-
tively. Based on statistically insignificant retum differences between the
"treatment" (qualified reports) and "control" samples, both researchers
concluded that the audit qualifications lacked infonnation content. On
the other hand, based on a standard API methodology. Firth [1978]
reported surprisingly sharp findings to the effect that investors are using
the information provided by the auditor's uncertainty qualification. Bai-
ley [1982] criticized this line of research on the ground that it is very
difficult, if not impossible, to distinguish cross-sectionaUy between audi-
tor-report differences and financial-statement differences (in footnotes,
ratios, etc.), for they might complement each other. Therefore, he argued,
neither market reaction nor nonreaction can be ascribed to the informa-
" For whether this regulation affected managements' R & D decisions, see the conflicting
findings of Dukes, Dyckman, and EUiott [1980] and Horwitz and Kolodny [1980], The
former failed to detect an adverse effect of FAS No, 2 on real R & D outlays by firms, while
the latter, using a questionnaire study, concluded that the regulation caused a relative
decline in R & D outlays of small, high-technology firms. Collins [1978] reported a possibly
weak effect of FAS No, 2 on managerial R & D decisions.
tion content in the auditor's report. This criticism of the difficulties of

adequately controlling for "nontreatment" effects applies, of course, to
many other market-based studies.
Hagerman [1975] evaluated the market effects of requiring banks to
issue financial statements to shareholders (1964 for state banks and 1968
for national banks). No discemible impact on bank-stock prices was
detected.
Ingram [1978] classified social-responsibility disclosures according to
(a) whether they were stated monetarily, and (6) the type of expenditure
(e.g., pollution control, fair business practice, employee health and safety,
etc.). When eamings expectations were controlled for, Ingram observed
a limited price effect of the environmental information for certain indus-
tries in specific years. Anderson and Frankie [1980] also reported evidence
on market reaction for firms that voluntarily reported on social activities
in 1972 and 1973. Griffin [1977] reported a price reaction to disclosure of
"questionable foreign pa3Tnents." However, given that the negative price
reaction reversed itself within two or three weeks, it appears that inves-
tors acknowledged the sensitive-payments disclosure, yet made no per-
manent value changes. Based on this evidence, we conclude that market
reaction to social-responsibility disclosure is at best weak. In general, a
case has not been made for the usefulness of investigating social-respon-
sibility accounting within the MBAR domain.
2.3.8 Evaluation. Where does this research effort leave us? On a
superficial level, the temptation is to describe the evidence on the market
impact of accounting regulation as scanty, contradictory, methodologi-
cally deficient, and wholly inadequate as a guide for public policy.
However, such a conclusion is inappropriate. First, regulations that
attracted comprehensive research resulted in findings that are consistent
and reasonably conclusive. For example, we can be quite confident that
ASR No. 190 (replacement cost) was not associated with a market
reaction but that the FASB's oil and gas exposure draft negatively
affected the stock prices of "full-cost" firms. A concerted and directed
research effort appears to yield convergent results. Second, while the
specific impact of the above-mentioned research on policymakers is
difficult to assess, it is evident that policymakers have become increas-
ingly aware of the relevance of research on the consequences of their
decisions: witness the growing demand by the FASB and the SEC for
market studies and for evaluations of reseeu"ch on consequences.*^ Evi-
dence on social (as opposed to physical) consequences, such as those of
accounting regulation, can never be more than indicative. A case in point
is the constant state of fiux and myriad contradictory results in economic
research on the consequences of various monetary and fiscal policies.
Third, and most important, the quality of regulatory "consequences"
" Some will probably argue that this demand for research is partially derived from PR
considerations and the need to "justify" regulatory decisions. While not entirely unreason-
able, the persistence and growth in demand appear to signal substantive interest.
research has markedly improved during the past decade. Limitations in

the basic tenets of this research—capital market efficiency and the
validity of the CAPM—axe increasingly recognized and attempts are
made to overcome them (e.g., recognition of the possible "size effect" on
returns); the statistical inethodology is constantly improving and new
analytical models are being designed (e.g., the return "reversal method"
for deliberations on regulation); the scope of consequences considered is
expanding (managers, bondholders, etc.); and more timely sources of data
(daily and even intraday prices) are used. While much of the work is in
its infancy, this line of research appears to be on the right track.
Nevertheless, the failure to develop a comprehensive perspective on
the relevance of various methodologies and findings can be criticized.
The cynic is still tempted to say "so what?" to the many identified
consequences of regulations. The key question still open can be stated as
follows: how do we identify a successful or effective regulation? A corol-
lary is: what market variables are potentially relevant and why? Unless
such questions are answered, one must expect confiision of the type that
surrounded the interpretation of Benston's [1973] study. Formal theoret-
ical analysis has the advantage of guiding research toward variables that
bear directly on the welfare impact of regulation. This is not to suggest
that the identification of such variables would lead to a direct or easy
resolution of policy issues—^just that the variables researched would be
more relevant. Thus, we shall have occasion to reconsider these issues in
section 3.1.
2.4 IMPLICATIONS OF ACCOUNTING RESEARCH FOR ECONOMIC

AND FINANCE THEORIES
Empirical MBAR studies are generally based on either or both of the
following assumptions: the existence of efficient capital markets (ECM)
and the descriptive validity ofthe Capital Asset Pricing Model (CAPM).
The ECM assumption justifies the fundamental premise that the impact
of informational events relevant for security valuation (e.g., earnings-data
disclosure or the enactment of a new regulation) could be inferred from
changes in parameters (e.g., mean, variance) of stock-return distributions
around identifiable dates of disclosure. This then makes capital markets
the testing ground for information-content hypotheses. The second as-
sumption—the validity of the CAPM—underlies the widely used tech-
nique of stock-return risk adjustment (or the matching by risk of
"treatment" and "control" samples) and the derived computation of
excess (abnormal) rates of return. In practice, most studies rely only on
the so-called market model and not on the CAPM. As discussed in Ohlson
[1978], this implies that the residuals cannot be interpreted as returns or
the outcome of a portfolio strategy. A sufficient and necessary condition
for the latter is the validity ofthe CAPM, a point not always appreciated.
As a practical matter, this might be of little significance as far as
information-content studies are concerned. Information-content studies
need only specify a statistical model of (expected) returns, and it is not

necessary to postulate the (risk-measure) determinants of equilibrium
expected retums (as in the CAPM). Note that information-content
studies are merely tests that informational items affect the retum distri-
bution; as such they are neutral about what determines the equilibrium
distribution in the first place. However, for tests of the ECM hypothesis,
which involve the performance analysis of portfolio strategies, the matter
cannot be so easily dismissed. Residuals that are not based on the CAPM
have no clear economic meaning, and they certainly cannot be interpreted
as excess retums. Furthermore, given the CAPM, the residuals must have
zero mean, and this implies that the CAPM restricts what portfolio
outcomes are attainable in an efficient market.
Based as it is on two major developments in finance theory in the past
two decades, we expect accounting research to shed some light on the
empirical validity of these theoretical constructs. We now consider the
relevant literature in some detail. The bottom line seems to be that
empirical accounting research has not been particularly supportive of
either the ECM hjrpothesis or the CAPM.
2.4.1 How Efficient Are Capital Markets? During the euphoria of the
1960s (for the ECM hypothesis) every piece of evidence appeared to
corroborate market informational efficiency, especially findings on the
quick and seemingly complete price and volume reaction to eamings
announcements (Ball and Brown [1968], etc.; see section 2.1.1). These
findings were interpreted as evidence in favor of the so-called semi-strong
version of the ECM hypothesis: prices adjust instantaneously to new
information and always refiect all publicly available information.*^ Some
early findings on investor reactions to differences and changes in accoimt-
ing techniques (see section 2.2) were also interpreted as being consistent
with market efficiency. Investors seemed to react rationally to accounting
changes having substantive, cash-fiow effects (e.g., a switch from FIFO
to LIFO) and to ignore accounting changes having no such effects.
However, as pointed out above, more comprehensive tests and carefid
consideration of the evidence generated disquieting or "anomalous" find-
ings at an ever-increasing rate.
Perhaps the evidence most damaging to the naive and unwavering
belief in market efficiency was the accumulation of results indicating the
existence of persistent price adjustments after eamings announcements
were made; this is obviously incompatible with the instantaneous-adjust-
ment property of informationally efficient markets. The more prominent
studies include Jones and Litzenberger [1970], Brown and KenneUy
[1972], Latane and Jones [1977], Joy, Litzenberger, and McEnally
" See Fama's [1970] survey of the ECM hypothesis for a consideration of accounting
information-content studies as corroborating the hypothesis. Compare this to the r«vision-
istic view summarized by Foster [1982] in his recent survey of the ECM literature.
[1977], Watts [1978], Brown [1978], and Joy and Jones [1979.]"" In some
cases, the studies even implied that excess returns could be earned by
using investment strategies based on price-change persistencies subse-
quent to earnings announcements.'*^ Recall that persistent price adjust-
ments after an information release have also been observed in several
studies on the market impact of accounting changes (see sections 2.2.2
and 2.2.3). Curiously and somewhat embarrassingly, even the early Brown
and Kennelly [1972] study, which replicated the Ball and Brown [1968]
study on quarterly earnings data, shows postdisclosure drift in the excess-
returns pattern. Although Brown and Kennelly chose not to discuss and
analyze the matter, (the mean of) postdisclosure drift appears to be
virtually indistinguishable from the predisclosure drift. Protracted price
adjustments to earnings announcements were also reported by Morse
[1981] for both stock exchange and OTC securities. This accumulating
evidence suggests that new information is not impounded in prices
"instantaneously."*®
A related line of research on market efficiency with respect to financial
information focused on the excess returns of portfolios classified by
various firm or stock characteristics. Studies like the well-known Basu
[1977] paper have shown that portfolios comprising low P/.E-ratio stocks,
after adjustment for risk, earn excess returns. And Givoly and Lakonishok
[1979] reported excess returns on portfolios consisting of companies with
a recent upward revision in analysts' earnings forecasts.
Another line of research relevant to market efficiency examined price
reaction to articles in the financial press presumably based on information
in the public domain. For example, Foster [1979] reported on a statisti-
cally significant negative price reaction associated with Briloff's critical
articles in Barrons, and Davies and Canes [1978] reported significant
price reactions to analysts' recommendations reported in the Wall Street
Journal column "Heard on the Street." Oppenheimer and Schlarbaum
[1981] suggested the possibility of earning risk-adjusted excess returns by
following the stock-selection rules provided in the various editions of Ben
Graham's book.
Given such evidence, is the market efficient? As is the case with all
complicated issues, no unequivocal answer can be given. If market effi-
ciency is naively interpreted as prices always reflecting all information in
the public domain (e.g., the entire past sequence of prices, all financial
" It has been suggested by Watts [1978] that some of these seeming market inefficiencies
are period specific. However, Nichols and Brown [1981] do not find a change over time in
this market "inefficiency" with respect to certain unexpected earnings changes,
*^ However, for recent contradicting evidence, see Reinganum [1981],
*" The latter two studies do not, however, suggest that the protracted price adjustments
can be employed practically to develop superior portfolio strategies. Additional "anomalies"
include the "weekend effect" (French [1980]), the "January effect" (RoU [1982]). and the
persistent discounts on closed-ends funds (Thompson [1978]),
and nonfinancial reports emanating from firms and analysts, all industry-
and economy-wide statistics, etc.), then the market is inefficient with
respect to certain information sources and signals.*^ The evidence, par-
ticularly on the persistence of price adjustments after information disclo-
sure and on price reaction to analyses in the media of information in the
public domain, points to different, more realistic concepts of equilibrium.
Such concepts have been addressed formally by Grossman and Stiglitz
[1976], among others. The key attribute of all these models is that
individuals are not presumed to be equally endowed with information. In
equilibrium, security prices transmit some of the information across
individuals and, in that sense, the less-informed individuals become better
informed. In the extreme, as in Grossman [1978], prices transmit all
information and there is a complete convergence of beliefs where the
market becomes trivially informationally efficient. In more realistic
settings, however, prices do not transmit all information and there is only
partial convergence in beliefs (see, e.g.. Diamond and Verrecchia
[1981]). Accordingly, prices never (even in equilibriiun) fiilly reflect the
global information set. The ambiguous concept of "publicly available
information" plays no role in these notions of equilibrium. It is important
to note that prices do not generally reflect all the information held by the
"informed" investors; this accounts for incentives to acquire and process
information. In somewhat heuristic terms, this suggests that the benefits
from acquiring information and forming superior portfolio strategies will
at the margin (or for the marginal investor) equal the cost of gathering
and converting raw data into useful information.''® In such an environ-
ment, an in-depth and time-consuming analysis of data basically in the
public domain (such as that done by Briloff) can be expected to induce
a price reaction. Furthermore, the fiow of information fi-om the more-
informed to the less-informed investors might be reflected in persistent
price adjustments following disclosure that are consistent with those
observed by Patell and Wolfson [1981].'*^
Neither the ambiguity of the empirical evidence nor the inadequacy of
the more primitive concepts of informational efficiency should oversha-
dow the fact that these concepts, even when they are relatively simplistic,
have played and probably will continue to play an important role in
empirical accounting research. In our view, this role is well deserved,
because it is virtually impossible to design an information-content study,
for example, unless one can assume that the market reaction occurs near
•" It cannot, of course, be entirely ruled out that some of the empirical findings supporting
this statement might be due to the invalidity of the CAPM (a point made by Ball [1978]
and many others) and/or to other methodological problems.
*' A formalization of this line of reasoning is complicated by the possibility of increasing
retum to scale of information processing. This is an inherently unstable situation; see
Wilson [1975],
•" The process of price and volume adjustment to new information is currently extensively
researched on both conceptual and empirical levels. See, for example, Hillmer and Yu
[1979].
the announcement date. The usefulness of the market efficiency concept

as a maintained hj^othesis would seem to be indisputable, especiaUy in
view of its reasonableness as a first-order approximation. However, the
evidence has indicated the need to be cautious, and conclusions can no
longer be regarded as foregone. In fact, some of the studies mentioned so
far (e.g., on the market impact of accounting changes, section 2.2) might
weU benefit from replications and refinements. The degree of market
informational efficiency is stiU very much an open question. Neither has
the introduction of more sophisticated concepts been without benefits.
The concept of infonnation aggregation is brought into focus; this further
suggests that the empirical studies of informational efficiency can be
partitioned usefuUy into two distinct sets. On the one hand, there are
studies, such as the one by Kaplan and RoU [1972], that test the market's
basic rationality: the "packaging" of information that is known to ev-
erybody should be of no relevance to security valuation. That is, beliefs
about future cash fiows across states should be independent of these
properties of information. On the other hand, there are instances when
potentiaUy relevant information is (or might be) impounded in security
prices, yet not everybody has access to the information. An example of
this is the Collins [1975] LOB-reporting study. The two scenarios for
informational efficiency are quite different, and their distinctions are
worth maintaining. In the first case, the policy implication is simply that
the issue is a nonissue; that much has long been appreciated (e.g.. Beaver
[1973]). In the second case, however, the relevant policy questions become
more complex. What are the implications of disclosure of nonpublicly
available information already impounded in the structure of prices (e.g.,
replacement-cost data)? Would such a disclosure be of any social useful-
ness? Given a maintained hypothesis of market efficiency, how would one
test for usefulness? Of course, a rigorous and operational definition of
what is meant by an informationaUy efficient financial market is needed.
These important questions are addressed in section 3.1.
2.4.2 The Validity of the CAPM. Disenchantment with the CAPM
is widespread on both conceptual and empirical grounds (e.g., RoU
[1977]). Our concem here is restricted to the accounting research impli-
cations of possible misspecifications of the model, particularly in its
ability to account properly for the systematic factors affecting security
pricing. BaU [1978], among others, has emphasized that some studies
reporting inconsistencies with capital market efficiency (e.g., the excess
retumsfi-omportfolios consisting of Xow-P/E stocks) are actuaUy tests of
a joint hypothesis—informational efficiency and the vaUdity of the
CAPM. Accordingly, apparent inconsistent findings with informational
efficiency might refiect improper risk adjustment by use of the CAPM.^
'" There are two separate issues here: (i) risk is improperly estimated because the return
on the market-portfolio is improperly measured; or (ti) the Ci4PAf itself is not an empirically
valid characterization of equilibrium. In this context, reference should be made to the
Brown and Warner [1980] study which examined, among other issues, the effects of using
Indeed, recent studies provide evidence that the CAPM is misspecified.

Banz [1981] reported on a "size effect" on stock returns. For the period
1936-75, the common stocks of small firms had, on average, higher risk-
adjusted returns than the common stocks of large firms. Reinganum
[1981] reported that the Basu [1977] P/E anomaly discussed above is
actually a surrogate for this "size effect." After the size effect on returns
had been accounted for, there were no longer differential returns among
portfolios of differing P/E-r&tios. However, a recent study by Basu
[19816] challenged Reinganum's findings, reporting the existence of both
a size and a P/E effect on stock returns. In addition to the size effect,
there is some evidence of a dividend effect on stock returns (see Litzen-
berger and Ramaswamy [1979]) and of a "January effect" (see Roll
[1982]). In general, the recent research effort on the empirical verification
of Ross' [1976] "arbitrage theory of capital asset pricing," such as Roll
and Ross [1980], points to the existence of several risk factors priced in
the returns-generating process of assets (as opposed to the one factor—
the market retum-underljong the CAPM). Presently unknown is the
correlation of such effects as firm size and P/E with such risk factors as
they would appear in a more general statement of equilibrium (e.g., Ross'
[1976] JiT-factor model of equilibrium).
The major implication of this research is that the findings of MBAR
derived from risk adjustments by the CAPM may reflect systematic
factors affecting returns such as firm size and dividends, thus blurring
the impact of the examined information or regulation on prices. The
extent of this possible distortion is yet unknown, although the Banz
[1981] and Reinganum [1981] studies, for example, reported that the size
effect on stock returns is substantial. In the meantime, it seems advisable
for accounting researchers using the CAPM to account for the more well-
established systematic effects, such as size, by stratifying their samples
according to these factors. Given, in most cases, the availability of large
samples, such stratification should be quite feasible at negligible cost to
statistical efficiency.
2.4.3 Systematic-Risk Assessment with Accounting Data. As noted
above, the CAPM implies that the systematic risk of a stock is a sufficient
firm-specific characteristic determining a security's expected return. This
central role of systematic risk in the return-equilibrium process naturally
attracted the attention of accounting researchers, who have followed two
main routes: (a) a positive approach, attempting to infer from correlations
with ft values (or from changes in those values) the information content
of financial data and regulation (some of these studies were surveyed in
sections 2.1 through 2.3), and (b) an essentially normative approach,
examining the role of accounting data in explaining and improving the
alternative market indexes and various methods for risk adjustment, A major conclusion
from this study is that ",,, beyond a simple, one-factor market model, there is no evidence
that more complicated methodologies [of risk adjustment] convey any benefit" [1980, p,
249],
assessment of systematic risk. This section considers the latter, normative

approach.
Examination of the relationship between accounting and market-based
risk measures began with the Beaver, Kettler, and Scholes (BKS) [1970]
study that reported significant correlations (particularly at the portfolio
level) between accounting risk surrogates and beta vtdues estimated from
market data. It is gratifjdng that the strongest correlations were found
for the two accounting variables whose choice can be justified on concep-
tual grounds: financial leverage and the covariation of a firm's eamings
with aggregate eamings ("the accounting beta"). The implication of this
finding is that the information provided by some accounting measures is
consistent with the underlying information set used by investors to assess
the riskiness of securities. On the normative level, BKS reported that
accounting risk measures could be used to develop forecasts of future
betas that were somewhat superior to naive beta forecasts (i.e., future
betas are equal to present ones).
BKS's results were essentially confirmed by White [1972] but were
challenged by Gonedes [1973], who was unable to detect a strong asso-
ciation between accounting and market betas. He attributed the differ-
ence in findings to the use by BKS of stock market prices to scale
(defiate) accounting data, a procedure that may have introduced a
spurious correlation with market betas.*' Beaver and Manegold [1975]
attempted to resolve the controversy by investigating a variety of speci-
fications to test the association between accounting- and market-based
betas, confirming the existence of a significant correlation that was not
due to the scaling method.^^ Significant correlation between market- and
accounting-based risk measures was also reported by Hill and Stone
[1980].
Additional evidence on the association between accounting- and mar-
ket-based risk measures was reported by Hamada [1972], establishing
both conceptually and empirically the relationship between financial
leverage and market risk estimates, and by Lev [1974a] for the relation-
ship between operating leverage (the ratio of fixed to variable costs) and
market estimates of risk. A related issue is the possible difference between
financial-leverage measurement based on accounting data and on stock-
market values. Bowman [1980] reported that the degree of association
between leverage based on accounting data and beta values is indistin-
guishablefiromthe association between market-based leverage and beta
values. Furthermore, Bowman found that the two variables—leverage
and the accounting beta—accounted for about 70 percent of the sample
cross-sectional variation in the market estimates of systematic risk.
" See Lev and Sunder [1979] for a discussion of the methodological issues involved in
this argument about spurious correlation due to a common denominator.
'* For continuation of this controversy, see Gonedes [1975a], who also found the associ-
ation between the two risk measures to be statistically significant, though surprisingly small
compared to other studies.
Regarding BKS's prediction results, recent research on the nonstation-

arity of beta (e.g., the Bayesian corrections developed by Vasicek [1973]
and Maier, Peterson, and Vander Weide [1977]) suggests better forecasts
of future betas than the simple extrapolation technique employed by
BKS. Eskew [1979] incorporated some of these refinements in attempts
to improve the naive beta forecasts against which the quality of the
accounting-based forecasts were assessed. He found that the accounting-
based models continue to yield superior predictions of beta values. The
most comprehensive tests of the marginal contribution of accounting
data in improving systematic-risk assessment were conducted by Rosen-
berg and several associates (e.g., Rosenberg and McKibben [1973] and
Rosenberg and Guy [1976]). Their conclusion was that a model incorpo-
rating both accounting £ind market data to predict beta values outper-
forms models based solely on market data. Moreover, the Rosenberg et
al. findings are independently supported by the widespread use of their
risk-estimation service among practitioners. After all, there is something
to be said for the power of the "market test."
2.4.4 Evaluation. What has been the contribution of MBAR to related
disciplines? Admittedly it has been a modest one. An important contri-
bution comes from the research discussed in section 2.4.1 that dispelled
the naive view of the ECM hjrpothesis. Prices do not adjust to all
information instantaneously. Assuming that there are no transaction
costs or costs of processing information, the literature is replete with
studies suggesting that excess returns are attainable. The complexity of
information impoundment, exemplified by Foster [1979] and Patell and
Wolfson [1982], has been documented in several contexts. Future research
must be cognizant of the evidence suggesting (rather strongly) that the
smooth instantaneous-adjustment property can be viewed only as a first-
order approximation. It is also worthwhile to note that at least some of
the evidence can be interpreted as being inconsistent with the equilibrium
risk-return relationship (CAPM), if not as evidence of informational
inefficiency. Without any specification of the precise risk-retvuTi relation-
ship, it can be shown that a sufficient and necessary condition for market
inefficiency is the existence of no-risk arbitrage profits. Of course, to
demand the existence of arbitrage profits as the only valid evidence of
market inefficiency is extremely restrictive; thus, the question of effi-
ciency versus inefficiency is not particularly interesting unless there is
some implied conceptualization of risk-return relationships. This leads to
the conclusion that there is no practical escape from the joint nature of
efficiency testing. As a corollary, the necessity for a benchmark risk-
return relationship has illuminated the difficulty of making claims that
the evidence is consistent with market efficiency. The "joint-hypothesis"
aspect effectively implies that the evidence of a properly identified (and
as yet unknown) risk-return relationship has to be remarkably robust
across many treatment/information effects. This is no small feat to
demonstrate, nor is it likely to occur in the nefu- future. As was indicated
in a previous section, many early tests purporting to find informational

efficiency can now simply be viewed as tests of relatively low power. We
believe that the evolution of accounting research has been consistent
with such a scenario, and much work in finance and economics could
benefit fi-om an appreciation of these developments. The experience of
accounting researchers suggests that "anomalies" are not anomalies once
the tests are of some power and one allows for the inherent difficulty in
identifjdng the "correct" risk-return relationship.
The work having the greatest impact on finance is probably the BKS
study. At least three fundamental determinants of systematic risk can be
identified with some confidence: financial leverage, covariation with
economy-wide earnings (or GNP), and earnings volatility (the last factor
appears to be inconsistent with the CAPM, and its relation with beta
might be due to it being a surrogate for covariation). The ad hoc nature
of the research in this area should be noted, and further progress clearly
awaits the development of a dynamic model of firm earnings behavior,
growth, and market valuation.*^ That systematic risk is related to fun-
damental characteristics of the firm is important and difficult to over-
emphasize. The full significance of this basic observation was appreciated
by finance researchers early on, and it is therefore somewhat disappoint-
ing to note that not much effort since the early studies has been expended
by accountants. The "positive" work has generally been disappointing,
and there have been few, if any, serious "normative" studies. Much of
the "positive" work is best described as "fishing expeditions." Consider-
ation is seldom given to what determines systematic risk and why there
ought to be a change in risk because the accounting and economic
environments have changed.
3. Avenues for Future Market-Based Accounting

Research
Two major areas for future MBAR have been repeatedly alluded to in
the discussion thus far: the first, a study of the implications of welfare
economics for the choice of relevant variables and interpretation of
findings of MBAR; the second, the construction of equity valuation
models to supplement and extend the traditional correlation studies of
MBAR. These two research areas are elaborated thus.
3.1 WELFARE E C O N O M I C S : SOME IMPLICATIONS FOR EMPIRICAL

RESEARCH IN ACCOUNTING
This section opens with observations on some important trends in

accounting research, especially the significant shift fi-om a traditionally
" Steps toward the development of such multiperiod models have been made. See, for
example, Myers [1977], Ohlson [1979a; 1979c], Garman and Ohlson [1980], and section 3.2
of this paper.
strong "normative" orientation (stating which accounting structures

ought to be promulgated) to a more cautious, timid, "positive" orientation
of merely registering consequences of accounting information and regu-
lation. This trend has decreased the relevance of MBAR to policjonakers.
Subsequent sections of this paper attempt to provide a framework for at
least a partial return to the traditional normative approach in accounting
research.
3.1,1 Trends in the Welfare Orientation of Accounting Research. A
major portion of traditional accounting research has been concerned with
making "normative statements" about accounting systems. Such
"normative statements" concerning the preferred (by various criteria)
accounting structures derive their significance by potentially providing
at least partial orderings of sets of alternative accounting systems
(methods of measurement and/or disclosure) in terms of their net societal
benefits.^ Obviously, in a multiperson setting such "normative state-
ments" must mesh with formal concepts of welfare economics if they are
to be accorded scientific status. Traditional "normative" work (such as
that of Edwards and Bell [1961], Sprouse and Moonitz [1962], Chambers
[1966], Ijiri [1967], and Sterling [1970]) was rarely, if ever, formal.
Nevertheless, the ultimate concern of this literature seems to have been
societal welfare, as demonstrated by the liberal use of terms like
"usefulness" and "relevance." Early market-based empirical reseeu-ch was
similarly motivated; the concern for "usefulness" is clearly evident in the
introduction to the influential Ball and Brown [1968] study, and Benston's
[1973] study purports to be a direct test of the social usefulness of the
1934 Securities Act. Similar comments are applicable to the many early
empirical studies that focused on the "information content" of accounting
data. In the extreme, this prescriptive philosophy of empirical research
led to the controversial Beaver and Dukes [1972] contention that the
API measure could be used as a metric ranking the desirability of
accounting alternatives.
Gonedes and Dopuch [1974], however, argued that such a conclusion
is generally inappropriate.^^ Subsequent empirical work has generally
concurred with Gonedes and Dopuch; studies since then have often been
motivated by a more general notion of "impact," namely, that policy-
making bodies "ought to know" about the presence or absence of certain
^ An example of such a normative statement, which postulates an ordering of two
alternative accounting systems—accrual earnings and cash flows—in terms of soci^ useful-
ness is the following statement made by the Study Group of the American Institute of
Certified Public Accountants in its report on Objectives of Financial Statements: "For such
relatively short periods [a month, a quarter or even a year], the accrual basis provides a
more useful measure of enterprise progress than the cash basis" (American Institute of
Certified Public Accountants, Objectives of Financial Statements [New York, 1973], p, 23),
" Note that the question here is not whether the API metric is appropriate or not as an
association measure (as in Marshall [1975]), Gonedes and Dopuch do not concern them-
selves with the choice of an appropriate metric of association, but rather with the relevance
of association analysis for welfare statements.
empirical associations between regulation and market data. A significant

shift is thus evident in the empirical financial accounting literature from
an initial, essentiaUy "normative" approach to a more cautious and
modest research objective emphasizing the "positive" aspects of account-
ing disclosure (e.g., "consequences" of regulation). Apparently awed by
strong statements on the alleged futiUty of addressing welfare issues by
empirical research (e.g., Demski's [1974] "impossibiUty" arguments, Go-
nedes and Dopuch's [1974] "impact but not desirabiUty," and Beaver and
Demski's [1974] argument regarding the need for "ethical judgments"),
accounting researchers have become more and more reluctant to expose
themselves to potential controversy by drawing welfare and/or direct
poUcy inferences from theirfindings.This led to difficulties in interpreting
(and justifying) empirical findings. This problem was apparent in the
previous review sections, and the choice of market-based variables, for
example, has appeared increasingly arbitrary. Clearly, the idea that some
informational item might have an impact on the retum distribution and/
or on trading pattems is much too general to guide researchers substan-
tiaUy in their selection of dependent variables and provide poUcymakers
with useful findings.
Empirical studies focusing on the efficient market hypothesis, as op-
posed to "information-content" studies, were even more ambiguous in
their stated objective to arrive at conclusions of welfare and poUcy.
However, there were at least impUcit suggestions that no societal benefits
can be derived from the disclosure of information already impounded in
the structure of security prices. It has also been argued that, given market
efficiency, how to "package" information is of no importance or relevance.
(See, for example, Kaplan's [1975, sees. 2, 6] review article.) In addition
to the information-content and efficient market Unes of research, some
recent work on accounting regulation derives its motivation from welfare
considerations. Prominent among these are studies motivating the issues
in terms of agency settings, where a change in the accounting environment
is presumed to affect managers' productive decisions. A decrease in
security values associated with a change in accounting regulation is
viewed as detrimental to shareholders. That is, individuals with "long
positions" in those securities are deemed worse off because there has
been a decUne in the value of some of the shares they own. Recent
empirical work investigating the effects on equity values of FASB and
SEC promulgations about the accounting for oil and gas properties
provides examples of such studies.
Unfortunately, the relationships between modem welfare economics
and the three types of accounting research mentioned above—informa-
tion-content studies, efficient market studies, and regulation-value re-
search—have not been articulated rigorously in the empirical Uterature.
The Gonedes-Dopuch paper cited previously argued that association
metrics have no welfare (poUcy) impUcations, but beyond that negative
observation there is Uttle, if any, formal analysis Unking the empirical
research to concepts of welfare economics. The constructive issue is how

and to what extent modern welfare economics can be exploited to enrich
our choice of methodologies and interpretation of empirical findings.
3.1.2 Welfare Implications. Before discussing the specific implications
of more or less well-known welfare economics propositions, some general
comments might prove useful. It is important to appreciate that sweeping
and general results about the net social benefits of changes in the
economic system are rarely, if ever, available. Any conclusions are bound
to be very contextual, emd models with discernible welfare implications
are typically based on rather strong and specific assumptions. In a
sufficiently rich and realistic setting, it goes without saying that some
individuals will be better off and others worse off because of changes in
the economic system. The partial societal preference ordering induced by
the Pareto criterion is therefore, in practice, too restrictive to be helpfiil
in the direct development of accounting policy. Furthermore, moving
beyond the partial Pareto ordering is generally impossible in view of
Arrow's celebrated theorem, which states that, under relatively mild
assumptions, there exists no social preference function aggregating indi-
vidual preferences. Likewise, questions of economic efficiency—where
"efficiency" permits a neglect of distributionsd issues—are complex. For
example. Hart [1975] has demonstrated that very little can be said about
the efficiency characteristics of incomplete markets in pure exchange
economies with two or more rounds of trading in the same set of financial
instruments (rather than just one round of trading). An even higher level
of restrictive assumptions is necessary to obtain insights into economies
with production: the theory of the firm under uncertainty is, to put it
mildly, lean. There is generally no reason to assume that shareholders
are unanimous about the optimal production/financing plan of the firm
(such as value maximization), even if there is only one round of trading.
In fact, value maximization need not imply an efficient allocation of
resources. (See Nielsen [1976] for an excellent review article on this and
related issues.)
However, what makes formal analysis, even under confining assump-
tions, worthwhile is that it can provide reasons that there might be
benefits from a change in the economic system. Thus, if welfare aspects
of alternative disclosure and regulation policies are analyzed, the issue is
to identify reasons for welfare impacts. Why and under what conditions
would more disclosure be socially beneficial? In formal analysis, an
unassailable criterion that can be used to identify socially beneficial
changes in the accounting environment is the Pareto criterion. This is
the dimension which we shall use to evaluate social benefits. While of
limited use in practical policy environments, the Pareto criterion is of
considerable value in the stylized models of equilibrium analysis. To the
extent that Pareto improvement can be demonstrated, important insights
into why changes might be socially beneficial tu-e attained.
Despite the current popularity of discussing (in a rather heuristic
manner) the private and social costs of regulation and information

production, we will not analyze the "direct costs" of alternative pohcies.
The costs of information production are important, but it is hard enough
to identify differential effects of alternative disclosure policies without
imposing cost functions. Models of general equilibrium are currently
much too complex to allow for the analytical derivation of welfare
propositions when there are explicit costs to be borne by the economy as
a whole. While the point about cost-free changes is not particularly deep
or novel, it is instructive to note that this perspective is invoked in the
welfare economics of markets, where the basic premise is that the costs
of operating and transacting in markets are zero. A particularly pertinent
illustration is found in a seminal paper by Arrow [1964]: the major issue
is an explication of the benefits of markets and alternative trading
arrangements, with no consideration given to the problem of finding an
optimal arrangement when there are costs of operating markets.
The above conunents are general; the focus now shifts to particular
concepts of welfare economics as they bear on empirical accounting
research. The first issue is how information affects individual well-being
in pure exchange economies, particularly under the assumption that
individuals have diverse (heterogeneous) information. These settings
relate directly to identifiable implications of the £^CAf hypothesis. Second,
the implications of pure exchange economies with homogeneous infor-
mation for "information-content" studies are considered.*^ Third, how
capital asset values, and changes therein, refiect shareholder welfare is
considered; this aspect is of particular relevance to assessing the impact
of regulation. To capture an important ingredient in the motivation of
this third type of research, the theoretical setting should allow for a
production impact, that is, corporate production and financing plans
change because the environment has changed. However, no results exist
in this third case unless individuals are assumed to have homogeneous
information; this assumption must be retained.*^
(i) Efficient market research. Any analysis of the welfare implications
of ECM as it concerns accounting issues demands a precise definition of
what is meant by "informationally efficient financial markets." The
empirical literature has been vague on this matter. Recently, Beaver
[1981a] discussed concepts of an informationally efficient market; his
major definition is that of price invariance. More precisely, the market is
said to be efficient with respect to an informational item (signal, or
information structure) if security prices are the same whether or not the
Homogeneous beliefs should be distinguished from homogeneous information; neither

implies the other. Homogeneous information is the case when all individuals have identical
partitions on the state space, and they cannot learn anything from each other. Homogeneous
beliefs imply that all individuals have identical probability measures on the (primitive)
state space, and this is unrelated to the particulars of partitions,
" The welfare implications under heterogeneous information are extremely limited. This
is true even if one restricts the economy to be one of pure exchange; see Wilson [1978],
informational item (or output of the information structure) is known by

all individuals in the economy ("publicly known"). Most important, note
that the release of such an informational item might affect some individ-
uals' beliefs without violating the price invariance of efficient markets.
Hence, belief invariance across all individuals is a sufficient but not a
necessary condition for market efficiency. Belief invariance means that
individuals' beliefs are imaffected by an informational item, and this is
clearly sufficient for market efficiency regarding that particular item.
That belief invariance is not a necessary condition is more subtle. The
idea is simply that, some subset of individuals might not have access to
a relevant piece of information, yet the price structure would be unaf-
fected even if they did have access to this information (and therefore had
changed their beliefs). This might occur for a variety of reasons, one
being that relatively "ignorant" individuals might take a very passive role
in the market; for example, they might have inelastic demand functions
and simply hold the market portfolio. Thus, having a large number of
"ignorant" individuals in the economy need not imply that prices reflect
such "ignorance." Note further that under no circumstances need indi-
viduals have homogeneous beUefs. The above definition of informational
market efficiency is precise, intuitively appealing, and therefore poten-
tially useful.^
Suppose now that some empirical research supports the proposition
that the market is efficient with respect to an accounting informational
item (i.e., no price change is associated with the disclosure of this item,
as in the case of replacement costs, section 2.3.3). What are the welfare/
policy implications of such a finding? Informational efficiency does not
necessarily imply that societal risks are efficiently shared. On the con-
trary, the disclosure of the information, which is already impounded in
the existing price structure, would make nobody worse off and possibly
make some individuals better off.®® The reason is simple enough: price
invariance means that the disclosure ofthe information does not eliminate
previously available opportunities/strategies. That is, no individual is
exposed to the possibility that the disclosure of the information might
have a negative wealth-redistribution effect, and under these circum-
stances it is obviously true that all individuals are at least as well off for
having more information. Furthermore, to the extent that there is addi-
tional trading because of the disclosure, the individuals who trade are
strictly better off. This follows directly because individuals would not
involve themselves in additional trading unless they expected an improve-
ment in their well-being (expected utility). This result does not depend
upon the particulars of individual utility functions. (The only exception
^ However, we do not wish to imply that this is the only appropriate definition. The
point is that this is one potentially useful definition. For discussions of other definitions, see
Beaver [1981a] and Foster [1982].
"" A formal discussion of this point is provided in Ohlson [19816]; ideas of this result are
due to Jaffee and Rubinstein [1975].
to additional trading and the implication of strict welfare improvement

occurs wben the market is efficient in the trivial sense of there being
belief invariance, i.e., individual beliefs are unaffected by the disclosure
of the informational item.) The conclusion is of general interest: it shows
that volume of trading is a relevant welfare indicator, provided that an
appropriate hypothesis is imposed on the behavior of prices. As will be
seen below, the simultaneous focus on volume and prices is of relevance
for the remaining discussion as well. Once cast in an appropriate theoret-
ical context, these two variables serve as core ingredients in the interpre-
tation of empirical research studies. The case of ECM should merely be
viewed as one illustration; deeper and less obvious results are considered
in subsections (ii) and {Hi).
It might be useful to consider at this point a (hypothetical) study
illustrating the discussion in the above paragraph. In section 2.3.1,
lines-of-business (LOB) reporting was considered; how could one design
a study that evaluates the social usefulness of these mandated disclo-
sures? One possibility is to view this question in a context in which prices
do not depend on the disclosure of LOB information, because the market
is presumed efficient with respect to this information (even though the
information is not publicly known). This is a presumption (that perhaps
could be tested empirically) and would therefore serve as a maintained
hypothesis in the analysis. The empirical analysis would then focus on
the following statistical question. Did the post-1970 mandated disclosure
of LOB data generate additional (above-average) trading? Presumably,
the statistical analysis of trading data would bear on that question. To
the extent the empirical evidence favored an affirmative answer, one
would then conclude, within the context of the assumptions made, that
social benefits in terms of improved risk sharing were derived from the
mandated disclosure of LOB data. In particular, the conclusion hinges on
the absence of redistributive effects; this is manifest in the maintained
assumption of price invariance. This assumption should not be viewed as
a restriction on the analysis but rather as the "price" one must pay to
arrive at such a sharp social-benefit conclusion. The requirements of
welfare analysis are extensive, a point which cannot be emphasized
enough.
(ii) Information-content studies. Although the previously employed
assumption of price invariance is appropriate for analyzing concepts of
market efficiency related to accounting information, the assumption is
much too strong—if not completely inappropriate—for considering infor-
mation-content, or contemporaneous correlation, studies. Nevertheless,
conclusions consistent with the outlook of such studies are available,
provided suitable characteristics are imposed on the economy. The rele-
vant pricing hypothesis preventing (negative) redistribution effects and
thereby allowing the making of welfare statements now states that prices
in the predisclosure economy must be unbiased estimators of the prices
that would obtain in the postdisclosure economy. In addition,** individ-

uals are assumed to have homogeneous information and identical beUefs
about the realization of specific signals. Furthermore, the economy cannot
be fundamentaUy inefficient; that is, given the avaUable securities and
the information configuration, the risk-sharing arrangement must be
(constrained Pareto-) efficient. Under these conditions the potential
information disclosure yields a weakly superior risk sharing: no individual
is worse off and some individuals might be better off.^^ Again, the reason
is that to the extent additional trading is generated by the infonnation
disclosure, some individuals wiU be strictly better off; the assumptions
used ensure that there can be no negative redistributive effects and
individuals' opportunities are expanded because of the disclosure.
The pricing-hjrpothesis and trading-volume scenario are consistent
with Beaver's [1968] information-content study, which can be viewed as
an empirical analysis of the social usefulness of eamings data. Beaver's
variabiUty test is a direct consequence of the pricing hypothesis, and the
welfare issue is addressed by measuring the trading volume induced by
earnings disclosure. FoUowing an argument developed in Ohlson
[1981c], under appropriate assumptions even the controversial Beaver
and Dukes [1973] contention that there is a relationship between asso-
ciation metrics and the social value of information can be given concep-
tual and theoretical support. The model in this case is the normal-linear
model of the Lintner [1969] CAPM type, under which the pricing hy-
pothesis wiU be (approximately) met. Other mild regularity conditions
ensure additional trading if the incremental information correlates with
security retums above and beyond that of existing information. Note that
the setting is one of pure exchange and the question is to what extent one
can infer improved risk sharing from market variables' reaction to the
information disclosure.
That trading volume can be viewed as a dependent variable indicating
the presence or absence of positive welfare effects should be emphasized.
This observation is expected in a "pure exchange" setting, where markets
serve the purpose of aUocating risks. Trading reflects this use of markets
and is an indicator of the social usefulness of markets. The same per-
spective is applicable to information as long as the economy functions
efficiently: trading is affected by the dissemination of new infonnation,
and thus the trading variable refiects the usefuUiess of the infonnation.
These ideas have been extensively developed in Ohlson and Buckman
[1981], where it is shown that there is generaUy a close relationship
between markets and information in terms of the welfare economics of
risk sharing.
" Price invariance is thus a special case. Note, however, that if the infonnation is relevant
in revising beliefs about the cash flow of the firm, and no individuals have access to the
information prior to disclosure, and all have access postdisclosure, then it is virtually
inconceivable that prices would be unchanged,
^' This result is due to Ohlson [1961c]; it is by no means obvious.
Certainty Economy Uocertainty Economy
Fish Consumption
State 1
~-contract curves ' '
individual l ' s \ /
i n d i f f e r e n c e curve
—-iiiv/
. individual 2's
Bread
i n d i f f e r e n c e curve - ' '
y Consumption
State 2
Trades are in Fish and Trades are in financial instruments

Bread directly. which indirectly determine consump-
tion in states 1 and 2.
FIG. 1
It should be noted that, even in "pure exchange" economies, trading in

securities should not be viewed as a zero-sum game impl5dng no social
benefits from increased trading. Such a perspective is completely inap-
propriate whether one considers economies with homogeneous or heter-
ogeneous information. Trading in financial instruments is a means to an
end—the optimal allocation of consumption across dates and states.
Individuals therefore trade implicitly with one another in consumption
patterns; conceptually this is no different from the trading in fish and
bread conceived by classical economists. In either case, although the
"pie" is fixed in a narrow sense, it is fundamentally incorrect to argue
that this precludes welfare gains. Markets have important roles to fulfill
in both cases for identical reasons; figure 1 shows these similarities using
Edgeworth diagrams. While the Certainty Economy diagram is indeed
familiar, that of the Uncertainty Economy may be less so. The latter
shows an economy in which aggregate consumption is uncertain and
some sharing rules across outcomes are more desirable than others. The
contract curve shows the set of Pareto-efficient sharing rules. Securities
are used to share the risks; thus, given Jiny set of securities, the role of
trading is to generate an allocation that is efficient relative to the set of
securities available. Furthermore, in the uncertainty setting, the potential
welfare gains are greater the richer the market structure is, and the same
occurs as the information available to individuals is refined.^^ Beaver's
'^ Hirshleifer'8 [1971] well-known example of no social value of information should be

viewed as a pathological case; see Hakansson, Kunkel, and Ohlson [1982] and Ohlson and
Buckman [1981]. Similar to Hirshleifer's example one can construct examples in which
there is no social value of markets (whether financial or real).
suggestion—quoted in section 1, point 5—that transaction costs are a net

drain on societal welfare is misleading to the extent that the positive
effect of improved risk sharing is not considered.
The trading perspective would be relevant, for example, in a study such
as Benston's [1973] on the 1934 Securities Act's sales-disclosure require-
ments, because prior beliefs suggest that the disclosure of that informa-
tion is likely to affect equilibrium prices. That is, the special case of price
invariance is too strong. Of course, other assumptions have to be invoked
to ensure that disclosure has no redistributive effects. In particular, it is
assumed that individuals have homogeneous information, markets func-
tion efficiently, and prices in the less-informed (pre-1934) environment
are unbiased estimators of those in the more-informed (post-1934) envi-
ronment. If these assumptions are maintained (and possibly empirically
verified), the stage is set to infer welfare benefits from a comparison of
pre- and post-1934 trading volume.
(Hi) Productive economies: regulation-consequences research. We
now relax the "pure exchange" setting and, further, make assumptions
about the volume of trade to analyze how changes in security prices
reflect individual welfare. The economies are presumed to be productive,
and an exogenous disturbance on firms' production/financing plans is
permitted. This triggers price/volume effects and the issue is to deduce
appropriate welfare statements. In general, without any production im-
pact, there is no relationship between security prices and welfare except
in the trivial case of no change in firms' values (in which case there are
obviously no welfare effects). The relevant theory therefore relates di-
rectly to the neoclassical theory ofthe firm, which addresses the question
of how individuals rank altemative input/output pattems across states
and, specifically, under what conditions shareholders are unanimous in
their evaluation of (potential) changes. The theory also considers
(changes in) the values of firms and their relationship to the desirability
of (changes in) firms' plans. This establishes a link to empirical studies
that analyze the impact of regulations on share values. Such studies
implicitly suggest a welfare impact because of a change in share values
(such as the decrease in stock prices associated with the 1977 oil and gas
exposure draft, section 2.3.2). However, such a conclusion is not generally
valid. The issues are far subtler, and while an increase in share values
might indicate a more efficient allocation of resources, there is no reason
to presume that individuals unanimously agree that the change is an
improvement, because other variables affecting individuals' opportunities
might also change. Stockholders are unanimous about the optimality of
value maximization when (i) the production of new commodity bundles
(consumption pattems) is not contemplated, a condition referred to as
"spanning" ("complete markets" is a special case of spanning), and (ii)
the price structure of basic commodities (but not securities) is unchanged
across production plans. Condition (i) is necesstiry for marginal analysis,
which is essential in the neoclassical theory ofthe firm. Condition (ii) is
also crucial and subsumes condition (i): classical value maximization

prevails because the right-hand side of individuals' budget constraints is
maximized while assuming the absence of an impact on the left-hand side
of the constraints. DeAngelo [1981] has emphasized that, at a technical
level, the argument of stockholder unanimity is that simple.^ However,
the result is limited in that conditions have been imposed on variables
(prices of commodity bundles) that are endogenous in a general equilib-
rium setting. Rubinstein [1978] has demonstrated that condition (ii)
cannot be given the status of an approximation and that individuals are
generally not unanimous, although condition (ii) might be approximately
true. In the same context, Rubinstein showed that crucial to unanimity
is the extent to which there is a need for trading to attain equilibrium. If
there is no need for trading, there is unanimity even when condition (ii)
is not imposed.
Volume of trading in this setting reflects differences in individual
endowments and preferences/beliefs, and these differences mitigate
against unanimity when commodity prices are allowed to respond to
changes in aggregate supplies. Settings in which this occurs are not
difficult to construct. The extreme case of identical individuals, which
allows trivially for unanimity, reflected in no trading, implies a one-to-
one correspondence between the (net) value of firms and societal welfare.
This setting is not particularly interesting in itself, but it does serve to
illustrate that the absence of trading is consistent with stockholder
unanimity.
However, there does exist a positive unanimity result where individuals
have dissimilar preferences/beliefs and prices respond to changes in
aggregate supplies. This subtle result is due to Ekem and Wilson [1974]
and is stated as follows. When (i) the change in firms' plans is small
(relative to social aggregates), (ii) the change in plans does not impart a
shift in the commodity space ("sparming"), and (Hi) the endowments in
the prechange market constitute an equilibrium, then there is unanimity
about the implications of the change in plans; that is, if one individual
thinks s/he is better (worse) off, then all individuals think they are better
(worse) off." Furthermore, the three assumptions imply that there is a
net increase (decrease) in the values assigned to firms' plans. It is implicit
that the effects of trading are of negligible (second-order) magnitude.
Conditions (i) and (IM) are clearly necessary for negligible trading when
there are no strong restrictions on individual preferences/beliefs. Condi-
tion (ii) ("spanning") is quite technical in nature and therefore less
intuitive than the other two, but it is also necessary to prevent trading
beyond a negligible level. The spanning condition is crucial at a more
" The literature generally refers to this result as Fisher's Separation Theorem or "ex
ante" stockholder unanimity,
" Because of condition (HJ) on endowments, the result is generally referred to as "ex
post" unanimity.
general level: it is necessary and sufficient for marginal (local) analysis.

Thus, although the conditions are somewhat stringent, they provide
jointly a setting in which there is a one-to-one correspondence between
firms' values and shareholder well-being. We summarize these conditions
in the requirement that above-average trading be of negligible magnitude;
this is only a necessary condition. (Whether the condition is sufficient is
currently unresolved.)
The studies which examine regulatory effects on production plans, and
therefore on firms' values, faU into the category that must rely on
stockholder-unanimity theory if welfare conclusions are to be derived;
the oil and gas studies reviewed in section 2.3.2 serve as examples.
However, the discussion has revealed that trading variables should be
considered if stockholder unanimity about production effects is of con-
cem: there should be no above-normal trading associated with the ex-
amined regulation. Unfortunately, in contrast to the "pure exchange"
cases (i) and (ii), one cannot conclude that additional trading improves
risk sharing in cases involving production effects. The economy is now
too complex for such a statement.
3.1.3 Summary. It is worthwhile to state some of the more important
welfare implications of MBAR. First, Beaver's concept of market effi-
ciency permits a dichotomy between belief invariance and price invari-
ance across information configurations; the former implies the latter, but
the converse is not true. Market efficiency as belief invariance is partic-
ularly relevant when the market's basic rationality is tested, as in Kaplan
and Roll [1972]. No impact on either prices or trading is expected. The
case of price invariance, without belief invariance, implies a more subtle
concept of efficiency, where social benefits are generated if the disclosure
of information induces above-normal trading. In any specific setting, an
empirical test to that effect is a test of improved risk sharing. These kinds
of tests have not been considered in the literature, although there appear
to be no practical obstacles to their implementation. Second, Beaver's
[1968] information-content study comes close to an empirical analysis of
the social benefits of accounting data. More information is better than
less information, and a properly structured hypothesis about the behavior
of security prices prevents individuals from experiencing (negative) re-
distribution effects. The possibility of a postdisclosvire round of trading
leads to a preferred risk-sharing arrangement. The significance of trading
as an indicator of improved risk sharing is again emphasized, a point
which has not generaUy been appreciated in the empirical literature.
Furthermore, socially useful information correlates with (predicts) future
gross payoffs, and this in tum induces a contemporaneous association
between prices and signsds. This justifies an interest in association met-
rics. This simple point has been lost during the past few yeais. Third, to
the extent that there are production effects due to a change in the
accounting environment, it is under certain conditions legitimate to infer
changes in stockholder well-being from a change in firms' values. The
I
.%
•o -a
rt .2
OQ i
1
1
s
ft;
g
!
key condition is that there be no above-normal trading due to the change:

such trading would reflect the possibility of redistributive effects. Hence,
in studies that consider firm values as a dependent variable in the context
of changes in the accounting environment, a "no-trading effect" should
be assumed or, preferably, be tested for empirically. There are otherwise
no particular reasons to expect a relationship between individual well-
being and firms' values. This is another point that has not been fully
appreciated in the accounting literature. The implications of the welfare
analysis for empirical research in accounting are summarized in table 1.
The above summary remarks stand in contrast to the conclusions
reached by Gonedes and Dopuch [1974]. Although their claim that one
can generally make no statements about the desirability of altemative
accounting environments is indisputable as such, it is also very close to
being vacuous. In economics, no positive statements whatsoever are
possible without assumptions, and nothing is true in general. The sub-
stantive issue then becomes identifying assumptions that can play a
constructive role in the interpretation of empirical findings; this has been
the subject matter of the current section. Gonedes and Dopuch focus on
production economies, and they implicitly suggest that it is only if value
maximization is well defined (optimal for shareholders) that the economy
has a structure adequate to infer welfare from prices. They do not,
however, demonstrate the necessity of their conditions justifjdng MBAR
beyond a general interest in the descriptive consequences of accounting
data and regulations. Particularly noteworthy is the absence of consid-
eration given to the role of information in facilitating improvement in the
risk-sharing arrangement. The potentially useful role of information in
pure exchange is not identified, a matter completely unrelated to the
optimality of value maximization.
These comments should not overshadow the fact that the derivations
of welfare conclusions require stringent assumptions. We thus frame our
discussion in terms of how and why welfare economics can be used
constructively to identify variables which are potentially relevant to the
design and interpretation of MBAR. More complicated, or simply alter-
native, theoretical models might therefore contradict the scenarios con-
sidered. The contextual nature of conclusions is inescapable; and the real
problem, which involves strong elements of value judgment, is how to
choose among two sets of assumptions (neither of which is more general
than the other) when they have different (welfare) implications.
3.2 ASSET VALUATION BY FUNDAMENTAL VARIABLES
Why has MBAR been almost exclusively concemed with the associa-
tion between financial data and stock-price changes (retums), while
ignoring the more basic question of asset valuation by fundamental
(accounting) variables? Empirical evidence suggests that unexpected
accounting information (particularly on eamings) is associated with ex-
cess retums (section 2.1.1), but it is also clear that the explanatory power
of the examined data with respect to the distributions of stock retums is

rather low. There is therefore good reason to consider altemative methods
of relating stock-price characteristics in financial markets to accounting
signals. Furthermore, if the relevance of accounting information to inves-
tors is at issue, surely the extent to which this information accounts for
(explains) the values of stocks, rather than just triggers a change in these
values, should be of major concem.
Consider, for example, the findings of the market nonreaction to the
disclosure of replacement-cost {ASR No. 190) and inflation-adjusted
{FAS No. 33) data, discussed in section 2.3.3. The research technique
employed cannot distinguish between two entirely different explanations
to the nonreaction finding: the replacement-cost data are irrelevant for
security valuation (due to, say, measurement errors), or the data are
relevant but were already impounded in stock prices at the time of
disclosure. Clearly, these two explanations have significantly different
impUcations for poUcymakers. If the replacement-cost or inflation-ad-
justed data are irrelevant, their mandated disclosure should be termi-
nated. On the other hand, if the information is impounded in stock prices,
the relevant issue becomes that of comparing the savings in social costs
from substituting mandated disclosure of price-adjusted data for the
private search effort of this information. Valuation analysis can usefuUy
indicate whether price-adjusted data contribute to the valuation of se-
curities and hence be relevant for poUcymaking.
The issue of valuation by fundamental variables is obviously of consid-
erable importance both from a poUcy point of view (what is the marginal
contribution of accounting information in general, and specific data in
particular, relative to other information sources, to determining capital
values?) and from a strictly "practical" point of view (e.g., to what extent
are accounting data useful in corporate valuation for mergers eind acqui-
sitions?). Why then has this issue been almost completely ignored in
MBAR? Is it because the construction of empirical valuation models in
uncertain and dynamic environments is much more compUcated and
risky (to the researcher) than simply registering a price reaction to an
accounting announcement? Or does misinterpretation of the ECM hy-
pothesis make valuation issues seem redundant, given the market's
aUeged abiUty to correctly (unbiasedly) price equities? Regardless of the
reason, more research on asset valuation' by fundamental variables is
clearly caUed for.
One advantage of fundamental valuation is that it does not generaUy
require an expectational specification of the information item. Consider
LOB reporting; "standard empirical procedures," exempUfied by CoUins*
[1975] study, rely on compUcated expectational models to determine
whether segmented eamings and sales are useful in explaining residual
retums. In our view, an equaUy vaUd approach, which might be more
straightforward and powerful, is to address the foUowing statistical issue.
To what extent are segmented data more useful than consoUdated data
in explaining asset values? The importance of valuation analysis is that

it provides an alternative to "standard empirical procedures." An empir-
ical finding that informational item X is uncorrelated with (residual)
returns does not, of covirse, preclude that X might contribute to a
statistical explanation of values.^ Yet, if the evidence of JTs relevance in
fundamental valuation is compelling, this suggests that X has information
content.
The search for information content does not imply that one a priori
must focus on changes in values rather than on values themselves. From
a theoretical viewpoint, a return relationship is not possible unless there
also is a valuation relationship, and conversely. This point, which is more
fully discussed below, is worth keeping in mind. The unexplored meth-
odological issue, therefore, is the extent to which fundamental valuation
is statistically more powerful than return analysis in detecting the infor-
mation content of accounting data. Successful resolution of this question
requires, of course, that credible stock-price valuation models can be
constructed. This is no easy matter, and a review of available evidence is
of some use.
3.2.1 Available Evidence. Attempts to construct stock-price valuation
models (by fundamental variables) date back to the 1920s and the 1930s.
For example, Meader [1935] formulated a regression model to explain
stock prices where the explanatory (independent) variables were stock
turnover, book value per share, net working capital per share, earnings
per share, and dividends per share. Empirical results were, however,
disappointing: the model's explanatory power was rather weak and pa-
rameter estimates were unstable over time. A later model by Whitbeck
and Kisor [1963], in which P/E ratios were regressed on the expected
growth rate in EPS, the expected dividend payout ratio, and the expected
standard deviation of earnings, did not fare much better. Tested by
Malkiel and Cragg [1970], the cross-sectional explanatory power of this
model was reasonably high (R'^s ranging between .70 and .85), but the
coefficients of the model (particularly those of the EPS growth variable)
were unstable over time. In general, the temporal instability in the
relationship between stock prices and earnings (see Keenan [1970])
appeared to be the major stumbling block in the construction of valuation
models.
In testing their seminal cost-of-capital proposition. Miller and Modi-
gliani [1966] constructed an equity valuation model for electric utilities,
regressing market equity vtdues on current earnings and average total
asset growth rates over the previous four years. For the three years 1954,
1956, and 1957, the Rh ranged between .56 and .77; the earnings coeffi-
^ Recall in this context the evidence presented in section 2.3.3 that the disclosure of
inflation-adjusted earnings {FAS No. 33) was not associated with a stock-price reaction, yet
"accounting betas" based on inflation-adjusted earnings were found to be more highly
associated with market betas than historical-cost-based "accounting betas" (Baran, Lak-
onishok, and Ofer [1980]).
dent w«us highly significant and reasonably stable over time. Brown
[1968] reported similar findings for the railroad industry (1954-63), using
a valuation model similar to that of Miller and Modigliani. His iJ^s varied
firom .73 to .92, and the earnings variable again exhibited the highest
explanatory power with respect to equity values.
Given the practical importance of valuation models, it is not surprising
that advanced constructs were developed by analysts. Prominent among
the published ones are the WeUs Fargo and the Value Line models.^ The
former is based on a dividend-discounting model, where estimates of
future dividends are derived from, among other things, estimates of future
eamings growth rates. While the performance of the WeUs Fargo model
has not been subjected to published empirical examination, the Value
Line model drew attention because of some evidence on its ability to
identify over- and undervalued securities (e.g.. Black [1973]), and the
superior (to extrapolative models) predictive perfonnance of eamings
forecasts provided in the Value Line Investment Survey (Brown and
Rozeff [1978]). The Value Line model places stocks in five different
categories, based on their estimated price perfonnance in the coming 12
months. Among the variables determining this classification are the
rankings of past eamings and prices of a given stock relative to other
stocks evaluated by Value Line, the growth in quarterly EPS of a given
stock relative to other stocks ("eamings momentum"), and the difference
between reported and predicted quarterly EPS ("surprise factor").
The CAPM appeared to provide some of the theoretical underpinning
previously missing in an explicit consideration of uncertainty in valuation.
In an empirical application of this model, Litzenberger and Rao [1971]
estimated the price-to-book values of electric utilities using the following
three variables: the expected rate of retum on (book-value) equity, the
systematic risk of the eamings-to-equity ratio, and the expected rate of
growth of book-value equity. Over the period 1960-66, this model ex-
plained about 55 percent of the cross-sectional variation of observed
price-to-book values; the coefficients ofthe three variables, and especially
the expected eamings coefficient, were statistically significant in most
years. More important, the size of the eamings coefficient was rather
stable throughout the estimation period. In addition to the improved
theoretical specification of this model, the stability of the earnings'
coefficient is probably due to the homogeneity of accounting procedures
within the electric utility industry.®^ Bowen [1981] improved the Litzen-
berger and Rao [1971] model by separating utility eamings into operating
eamings and the credits from the "allowance for funds used during
construction." Although the latter component of eamings appeared to be
of lower quality than operating eamings, as assessed by investors, its
incorporation into the model did improve stock valuation. Beaver and
"* For a more complete description of these models, see Foster [1978].
" For additional empirical investigations using similar valuation models see Litzenbereer
and Budd [1972], Dukes [1976], and Foster [19776].
Morse [1978] regressed E/P ratios on systematic risk and on three

earnings growth measures for the three years subsequent to the E/P
computation. For the period 1956-70, these explanatory variables ac-
counted for about 50 percent of the cross-sectional variation in E/P
ratios. Somewhat disappointing was the inability ofthe beta risk measures
to provide systematic explanatory power. Boatsman and Baskin [1981]
addressed the valuation problem by comparing the predictive ability
(with respect to market values) of value estimates based on the CAPM
with estimates from what they viewed as two competing models: an
earnings capitalization (P/E) model and a model used in determining
current asset values ("indexing model"). Of the three contenders, the
CAPM was found to be superior in predictive ability.
Despite the modest research effort into asset valuation using funda-
mental variables, it appears that GAPAf-derived constructs and some
other models have considerable potential. Some of the parsimonious
models discussed above exhibit considerable explanatory power and
reasonable temporal stability of parameter estimates. Problems, of course,
abound. For example, cross-sectional differences in accounting practices
appear to seriously impair the models' performance, particularly in pre-
dicting P/E ratios. This calls for attempts to adjust the fundamental
variables for various major accounting differences. Although new ideas
and statistical methods to circumvent the temporal instability of coeffi-
cients problem remain largely unexplored, the difficulty of this problem
should not be underestimated. Existing empirical literature suggests that,
at a minimum, firms can be usefully grouped into industries in the
construction of valuation models. The more challenging problem is the
extent to which the time-series behavior of the regression coefficients in
valuation constructs can be modeled explicitly. In the final analysis,
valuation parameters change across time periods because of changes in
macroeconomic variables, such as interest rates.
3.2.2 Security Valuation—Some Conceptual Issues. As was previously
indicated, it is important to appreciate that valuation and value changes
are complementary dimensions in the equilibrium theory of security
behavior. Theoretically, if information content of some data exists, it
should manifest itself in both security-price levels and changes. However,
detecting a price reaction in information-content studies depends cru-
cially on the expectational model used and on an exact identification of
the timing of information disclosure (see section 2.1.1). Valuation studies,
on the other hand, are much less sensitive to these specification issues,
and hence, as a practical matter, the results from applying the two
methodologies might differ. Garman and Ohlson [1980] analyzed the
theory of valuation in multiperiod, uncertain environments, and their
framework can be used to consider the key ingredients in valuation and
return-distribution theory. With some simplification, the relevant steps
can be described as follows.
(i) The value of a security at any given time equals the present value
of expected future dividends adjusted for their riskiness. The risk mea-
sures used across dates depend on standard equilibrium considerations

such as individuals' risk dispositions, beliefs, endowments, and markets
available. The valuation formula, which is a non ad hoc generalization of
the classical discounting formula, is fiilly discussed in Beja [1967], Gar-
man [1978], Ross [1978], and Rubinstein [1976]. The key assumptions
involve little beyond perfect markets and the absence of arbitrage oppor-
tunities, and there is no need for the complete markets assumption.
(ii) Whatever information is useful in forecasting future dividends and
their riskiness affects the current equilibrium value of securities; con-
versely, whatever information affects current equilibrium values should
be useful in predicting future dividends. A mapping from fundamental
descriptors into a security value is therefore the "reduced-form" charac-
terization of a dividends-capitalization prediction.
(Hi) Over time, new information about future dividends becomes
available and values change accordingly. If the relevant information
comprises a vector of fundamental descriptors, then value changes occur
only when the descriptors change, and vice versa. The probability distri-
bution of changes in values, and thus also in retums, is derived from the
mapping of descriptors to prices and the conditional distribution of
changes in the descriptors (given their current state).
To illustrate the above concepts, consider a simple scenario consistent
with the Ball and Brown [1968] study. Assume that current accounting
eamings are parsimonious and efficient predictors of discounted expected
dividends, adjusted for risk. Hence, there will be a reduced-form mapping
from eamings into prices, and, in the extreme, the security value is
characterized by an intertemporally constant P/E multiplier. It now
follows that price changes are related directly to "unexpected eamings,"
as is implied by the BaU-Brown hypothesis. This simple scenario may
capture an aspect of reality, and this is accomplished without resorting to
any notion of "true eamings." The driving force in the analysis is the
information content of eamings as an efficient predictor of future divi-
dends.
The important point, though, is that the information content of any
set of (accounting) variables must manifest itself on a valuation level as
well as that of value changes. The basic question therefore remains. Why
has so much effort been devoted to explaining retums (value changes)
rather than values? Both approaches have advantages and disadvantages,
and the key issue becomes identifying types of questions and circum-
stances under which one approach is preferable to the other. Much
research effort, analytical as well as empirical, is needed to address these
issues. Among the very few to do this were Gonedes and Dopuch [1974],
whose major message was the theoretical superiority of retums analysis
to fundamental valuation. They argued that, in the absence of a well-
developed theory of valuation under uncertainty, the modifications re-
quired in the certainty-capitalization model, especially those dealing with
risk measures, are ad hoc. They also argued that using a discounting
model under uncertainty in a manner analogous to that used under
certainty is ad hoc, because such an extension is not justified by an

explicit theory of valuation under uncertainty. This contention is not
surprising, given the absence of a well-defined theory of dividend dis-
counting under uncertainty at the time of their work. This appears to
have led Gonedes and Dopuch to believe that the CAPM approach to
equilibrium valuation was different from and superior to dividend dis-
counting and fundamental valuation. That this view is inappropriate is
made clear in Rubinstein [1976]. In fact, the discounting-valuation per-
spective is sufficiently flexible to be consistent with all nonarbitrage
concepts of expected retums and risks. This also means that there is
every reason to expect considerable practical difficulties in identifying
and/or operationalizing discounting/valuation relationships. Problems of
identifying sources of uncertainty that are "priced" and the operational
risk measurement are abundant, and the modeling of an expectation
djTiamics (i.e., the stochastic evolution of the descriptors) is obviously a
very complicated task. Additional complexities are present when sto-
chastic macrovariables, such as a changing term structure of interest
rates, are incorporated into the valuation analysis. All these difficulties
call for considerable caution, and undoubtedly much work remains to be
done before valuation analysis can generate credible empirical results.
As a final observation, the conceptual model of expected dividend
discounting points to a third mode of information-content analysis: the
identification of efficient predictors of future dividend streams. This, we
believe, is an empirical research technology worth exploring. Theory
sustains the soundness of this relationship, and the issue becomes empir-
ical practicality. Curiously enough, some of the very early MBAR studies
examined the information content of accounting items from this point of
view (see, for example, Staubus [1965]). The (premature?) death of
dividend-prediction and security-valuation issues is not easy to under-
stand on theoretical grounds.
4. A Final Note
Where does it all leave us? To paraphrase Churchill, did MBAR reach
the end, the beginning of the end, or just the end of the beginning? On
balance, we opt for the third possibility. In many respects MBAR is still
in its infancy, and a number of simultaneous developments seem plausi-
ble. One future reseeu'ch avenue pertains to methodological refinements
in addressing old and familiar questions such as investors' ability to
decode the impact of accounting messages and techniques. The obvious
research reward here is the corroboration/revision of existing findings. A
second line of research will probably raise modestly novel questions using
familiar methodologies, such as tests of information content of noneam-
ings financial data and the evaluation of market consequences of future
accounting regulations. This line of research should generally enrich our
understanding of accounting information and institutions witiiin a capital
market context. However, these two research avenues are more in the
spirit of "the beginning ofthe end" than "the end ofthe beginning." More
innovative and path-breaking research is required for the latter. The
nature of such research can, of course, only be conjectured somewhat
vaguely, but it appears essential that theories of financial information,
rather than just information, be constructed. In any event, we do believe
that this research must be "theory based," more so than the work that
has transpired to date. Only theory can aid us in the development of new
questions and in a more useful (e.g., to policymakers) interpretation of
findings. The greatest challenge to MBAR is the development and
nonsuperfical utilization of theory that transcends the basics of the ECM
and the two-period CAPM. In spite of a rather spotty record in theory
utilization, we remain optimistic that such will indeed be the case.
REFERENCES
ABDEL-KHALIK, A. R., AND B . B . AJINKYA. "Informational Efficiency of the Securities
Markets: A Discussion of the Evidence and Implications for Accounting." ARC Working
paper no. 79-4, University of Florida, April 1979.
ABDEL-KHALIK, A. R., AND J. C. MCKEOWN. "Understanding Accounting Changes in an
Efficient Market: Evidence of Differential Reaction." The Accounting Review (October
1978a): 851-68.
ABDEL-KHALIK, A. R., AND J. C. MCKEOWN. "Disclosure of Estimates of Holding Gains
and the Assessment of Systematic Risk." In Studies in Accounting for Changes in
General and Specific Prices: Empirical and Public Policy Issues, 1978b. Supplement to
Journal of Accounting Research 16: 46-77.
ABDEL-KHALIK, A. R., AND R. B . THOMPSON. "Eamings Forecasts: The State of the Art."
The Accounting Journal (1978): 180-209.
ABDEL-KHALIK, A. R., R. B. THOMPSON, AND R. E . TAYLOR. "The Effect of Reporting
Leases off the Balance Sheet on Bond Risk Premiums: Two Exploratory Studies." In
Economic Consequences of Financial Accounting Standards: Selected Papers, pp. 101-
58. Stamford, Conn.: FASB, 1978.
ABDEL-KHALIK, A. R., ET AL. T?ie Economic Effects on Leases of FASB Statement No. 13,
Accounting for Leases. Stamford, Conn.: FASB, 1981.
AHARONY, J., AND I. SwARY. "Quarterly Dividend and Eamings Announcements and
Stockholders' Retvims: An Empirical Analysis." Journal of Finance (March 1980): 1-12.
AJINKYA, B . B . "Line of Business Reporting and Market Risk Assessment: An Extension.'
Working paper. Accounting Research Center, University of Florida, 1978.
"An Empirical Evaluation of Line of Business Reporting." Journal of Accounting
Research (Autumn 1980): 343-61.
ALDERMAN, C. W. "The Role of Uncertainty Qualifications: Evidence to Support the
Tentative Conclusions of the Cohen Commission." Journal of Accountancy (November
1977): 97-100.
ALEXANDER, G. J. "Applying the Market Model to Long-Term Corporate Bonds." Journal
of Financial and Quantitative Analysis (December 1980): 1063-80.
ANDERSON, J. C, AND A. W. FRANKLE. "Voluntary Social Reporting: An Iso-Bete Portfolio
Analysis." The Accounting Review (July 1980): 467-79.
ARBEL, A., AND B . JAGGI. "Impact of Replacement Cost Disclosures on Investors' Decisions
in the United States." Intemational Journal of Accounting (Fall 1978): 71-82.
ARCHIBALD, T . R. "The Return to Straight-Line Depreciation: An Analysis ofa Change in
Accounting Methods." Empirical Research in Accounting: Selected Studies, 1967. Sup-
plement to Journal of Accounting Research 5: 164-80.
. "Stock Market Reaction to the Depreciation Switch-Back." The Accounting Review
(January 1972): 22-30.
ARROW, K. J. "The Role of Securities in the Optimal Allocation of Risk-Bearing." Review
of Economic Studies (April 1964): 91-96.
ASHLEY, J. W. "Stock Prices and Changes in Eamings and Dividends: Some Empirical
Results." Journal of Political Economy (February 1962): 82-86.
ATIASE, R. "Predisclosure, Informational Asymmetries, Firm Capitalization, Financial
Reports and Security Price Behavior." Ph.D. dissertation. University of California,
Berkeley, 1980.
BALL, R. J. "Changes in Accounting Techniques and Stock Prices." Empirical Research in
Accounting: Selected Studies, 1972. Supplement to Journal of Accounting Research 10:
159-78.
. "Anomalies in Relationships Between Securities' Yields and Yield Surrogates."
Journal of Financial Economics (June/September 1978): 103-126.
, AND P. BROWN. "An Empirical Evaluation of Accoimting Income Numbers.''
Journal of Accounting Research (Autumn 1968): 159-78.
, AND G. FOSTER. "Corporate Financial Statements: A Review of Empirical Research
Methodologies." Journal of Accounting Research (Supplement 1982): 161-234.
BAILEY, W . T . "An Appraisal of Research Design Used to Investigate the Information
Content of Audit Reports." The Accounting Review (January 1982): 141-46.
BANZ, R. W. "The Relationship Between Retum and Market Value of Common Stocks."
Journal of Financial Economics (March 1981): 3-18.
BARAN, A., J. LAKONISHOK, AND A. R. OFER. "The Information Content of General Price
Level Adjusted Eamings: Some Empirical Evidence." The Accounting Review (Jeinuary
1980): 22-35.
BASKIN, E . F. "The Communicative Effectiveness of Consistency Exceptions." 7%e Ac-
counting Review (January 1972): 38-51.
BASU, S . "Investment Performance of Common Stocks in Relation to Their Price-Earnings
Ratios: A Test of Market Efficiency." Journal of Finance (June 1977): 663-«2.
. "Market Reaction to Accounting Policy Deliberations: The Inflation Accounting
Case Revisited." The Accounting Review (October 1981a): 942-54.
"The Relationship Between Earnings' Yield, Market Value and Retum for NYSE
Common Stocks: Further Evidence." Working paper. Faculty of Business, McMaster
University, September 19816.
BEAVER, W. H. "The Information Content of Annual Eamings Annoimcements." Empirical
Research in Accounting: Selected Studies, 1968. Supplement to Journal of Accountitig
Research 6: 67-92.
. "The Behavior of Security Prices and Its Implications for Accounting Research
(Methods)." Committee Reports. Supplement to The Accounting Review (1972): 407-37.
"What Should Be the FASB's Objectives?" Journal of Accountancy (August 1973):
49-56.
. "The Information Content of the Magnitude of Unexpected Elamings." Working
paper, Stanford University, 1974.
. "The Implication of Security Price Research for Disclosure Policy and the Analyst
Community." In Financial Information Requirements for Security Analysts, edited by
A. R. Abdel-khalik and T. C. Keller, pp. 65-81. Durham, N. C: Duke University, 1978.
. "Market Efficiency." The Accounting Review (January 1981a): 23-37.
. "Econometric Properties of Altemative Security Retum Methods." Journal of
Accounting Research (Spring 19816): 163-84.
, A. A. CHRISTIE, AND P. A. GRIFFIN. "The Information Content of SEC Accounting
Series Release No. 190." Journal of Accounting and Economics (August 1980): 127-57.
, R. CLARKE, AND W. WRIGHT. "The Association Between Unsystematic Security
Retums and the Magnitude of the Eamings Forecast Elrror.'' Journal of Accounting
Research (Autumn 1979): 316-40.
, AND J. S. DEMSKI. "The Nature of Financial Accounting Objectives: A Summary
and Synthesis." Studies on Financial Accounting Objectives: 1974. Supplement to
Journal of Accounting Research 12: 170-87.
, AND R. E. DUKES. "Interperiod Tax Allocation, Earnings Expectations, and the
Behavior of Security Prices." The Accounting Review (April 1972): 320-32.
, AND R. E. DUKES. "Delta-Depreciation Methods: Some Empirical Results." The

Accounting Review (July 1973): 549-59.
, P. A. GRIFFIN, AND W. R. LANDSMAN. "The Correlation of Replacement Cost
Eamings with Security Retums." Working paper, Stanford University, Graduate School
of Business, July 1981.
-, P. KETTLER, AND M. SCHOLES. "The Association Between Market Determined and
Accounting Determined Risk Measures." The Accounting Review (October 1970): 654-
82.
, AND W. LANDSMAN. "The Incremental Information Content of FAS 33 Disclosures."
Working paper, Stanford University, July 1982.
, AND J. MANEGOLD. "The Association Between Market-Determined and Accounting
Determined Measures of Systematic Risk." Journal of Financial and Quantitative
Analysis (June 1975): 231-84.
, AND D. MORSE. "What Determines Price-Earnings Ratios." Financial Analysts
Journal (July/August 1978): 65-76.
BEJA, A. "Capital Markets with Delayed 'Learning'." Ph.D. dissertation, Stanford Univer-
sity, 1967.
BENSTON, G. "Published Corporate Accounting Data and Stock Price." Empirical Research
in Accounting: Selected Studies, 1967. Supplement to Journal of Accounting Research
5: 1-14, 22-54.
. "Required Disclosure and the Stock Market." American Economic Review (March
1973): 132-55.
BIDDLE, G. C. "Accounting Methods and Management Decisions: The Case of Inventory
Costing and Inventory Policy." Journal of Accounting Research (Supplement 1980):
235-80.
, AND F. W. LiNDAHL. "Stock Price Reactions to LIFO Adoptions: The Association
Between Excess Retums and LIFO Tax Savings." Journal of Accounting Research
(Autumn 1982, pt. II): 551-88.
BLACK, F. "Yes, Virginia, There is Hope: Tests of the Value Line Ranking System."
.FVnancta/i4na/^ste JburnanSeptember/October 1973): 10-14.
, AND M. SCHOLES. "The Pricing of Options and Corporate Liabilities." Journal of
Political Economy (May/June 1973): 637-54.
BOATSMAN, J. R., AND E . F. BASKIN. "Asset Valuation with Incomplete Markets." The
Accounting Review (January 1981): 38-53.
BouLDiNG, K. "Economics and Accounting: The Uncongenial Twins." In Studies in Modem
Accounting Theory, edited by D. Baxter and S. Davidson, pp. 44-55. Homewood, 111.:
Irwin, 1962.
BouRDEAUX, K, AND S. ZEFF. "A Note on the Measure of Compensation Implicit in
Employee Stock Options." Journal of Accounting Research (Spring 1976): 158-62.
BOWEN, R. M . "Valuation of Earnings Components in the Electric Utility Industry." T?ie
Accounting Review (January 1981): 1-22.
BOWMAN, R. G. "The Debt Equivalence of Leases: An Empirical Investigation.'' The
Accounting Review (April 1980): 237-53.
BRILOFF, A. J. Unaccountable Accounting. New York: Harper & Row, 1972.
. More Debits Than Credits: The Bumt Investor's Guide to Financial Statements.
New York: Harper & Row, 1976.
BROWN, L. D., AND M . S. ROZEFF. "The Superiority of Analysts Forecasts as Measures of
Expectations: Evidence from Eamings." Journal of Finance (March 1978): 1-28.
BROWN, P. "Some Aspects of Valuation in the Railroad Industry." Ph.D. dissertation.
University of Chicago, 1968.
"The Impact of the Annual Net Profit on the Stock Market." The Australian
Accountant (July 1970): 277-82.
, F. J. FINN, AND P. HANCOCK. "Dividend Changes, Eamings Reports and Share
Prices: Some Australia Findings." Australian Journal of Management (October 1977)-
127-47.
, AND J. W. KENNELLV. "The Informational Content of Quarterly Eamings: An

Extension and Some Further Evidence." Journal of Business (July 1972): 403-15.
BROWN, R. M. "Short-Range Market Reaction to Changes in LIFO Accounting Using
Preliminary Eamings Announcement Dates." Journal of Accounting Research (Spring
1980): 38-63.
BROWN, S. L. "Earnings Changes, Stock Prices and Market Efficiency." Journal of Finance
(March 1978): 17-28.
BROWN, S. J., AND J. B. WARNER. "Measuring Security Price Performance." Journal of
Financial Economics (September 1980): 205-58.
CASSIDY, D . B. "Investor Evaluation of Accounting Information: Some Additional Empirical
Evidence.'" Journal of Accounting Research (Autumn 1976): 212-29.
CHAMBERS, R. J. Accounting, Evaluation and Economic Behavior. Englewood ClifEs, N.J.:
Prentice-Hall, 1966.
COLLINS, D . W. "SEC Product-Line Reporting and Market Efficiency." Journal of Finan-
cial Economics (June 1975): 125-64.
, AND W. T. DENT. "The Proposed Elimination of Full Cost Accounting in the
Extractive Petroleum Industry: An Empirical Assessment of the Market Consequences."
Journal of Accounting and Economics (March 1979): 3-44.
, M. S. ROZEFF, AND D . S. DHALIWAL. "The Economic Determinants of Market
Reaction to Proposed Mandatory Accounting Changes in the Oil and Gas Industry: A
Cross-Sectional Analysis." Journal of Accounting and Economics (May 1981): 37-72.
, M. S. ROZEFF, AND W. K. SALATKA. "The SEC Rejection of SFAS No. 19: Tests of
Market Price Reversal." The Accounting Review (January 1982): 1-17.
, AND R. SiMMONDS. "SEC Line-of-Business Disclosures and Market Risk Adjust-
ments." Journal of Accounting Research (Autumn 1979): 352-83.
COLLINS, J. "Financial Reporting Methods and Management Decisions: An Empirical
Exiunination of Information Inductance." Ph.D. dissertation. Northwestern University,
1978.
COMISKEY, E. E. "Market Response to Change in Depreciation Accounting." Ttie Account-
ing Review (April 1971): 279-85.
CORNELL, B., AND R. ROLL. "Strategies for Pairwise Competitions in Markets and Organi-
zations." Bell Journal of Economics (Spring 1981): 201-13.
DAVIES, P. L., AND M . CANES. "Stock Prices and the Publication of Second-Hand Infor-
mation." Journal of Business (January 1978): 43-56.
DEAKIN, E. B . "Accounting Reports, Policy Interventions and the Behavior of Security
Returns." The Accounting Review (July 1976): 599-603.
, G. R. NORWOOD, AND C. H. SMITH. "The Effect of Published Eamings Information
on Tokyo Stock Exchange Trading." International Journal of Accounting (Fall 1974):
124-36.
DEANGELO, H . "Competition and Unanimity." American Economic Review (March 1981):
18-27.
DEMSKI, J. "Choice Among Financial Reporting Alternatives.'' The Accounting Review
(April 1974): 221-32.
DHALIWAL, D . "The Effect of the Firm's Capital Structure on the Choice of Accounting
Methods." The Accounting Review (January 1980): 78-84.
DIAMOND, D . W., AND R. E . VERRECCHIA. "Information Aggregation in a Noisy Rational
Expectations Economy." Journal of Financial Economics (September 1981): 221-36.
DIAMOND, P. A. "The Role of a Stock Market in a General Equilibrium Model with
Technological Uncertainty." American Economic Review (September 1967): 759-76.
DUKES, R. E. "An Investigation of the Effects of Expensing Research and Development
Costs on Security Prices." In Proceedings of the Conference on Topical Research in
Accounting. New York: Ross Institute of Accounting Research, New York University.
1976.
. An Empirical Investigation of the Effects of Statement of Financial Accounting
Standards No. 8 on Security Return Behavior. SUmford, Conn: FASB, 1978.
, T. R. DYCKMAN, AND J. A. ELLIOTT. "Accounting for Research and Development
Costs: The Impact on Research and Development Expenditures.'' Journal of Accounting

Research (Supplement 1980): 1-26.
DYCKMAN, T . R., D . H . DOWNES, AND R . P. MAGEE. Efficient Capital Markets and
Accounting: A Critical Analysis. Englewood Cliffs, N.J.: Prentice-Hall, 1975.
DYCKMAN, T . R., AND A. J. SMITH. "Financial Accounting and Reporting by Oil and Gas
Producing Companies: A Study of Information Effects." Journal of Accounting and
Economics (March 1979): 45-76.
EASMAN, W., ET AL. "The Correlation Between Sustainable Income and Stock Retums."
Financial Analysts Journal (September/October 1979): 44-47.
EDWARDS, E . O., AND P. W. BELL. The Theory and Measurement of Business Income.
Berkeley: University of Califomia Press, 1961.
EKERN, S., AND R. WILSON. "On the Theory of the Firm in an Economy with Incomplete
Markets." Bell Journal of Economics and Management Science (Spring 1974): 171-80.
EsKEW, R. K. "An Examination of the Association Between Accounting and Share Price
Data in the Extractive Petroleum Industry." The Accounting Review (April 1975): 316-
24.
"The Forecasting Ability of Accounting Risk Measures: Some Additional Evidence."
Tfie Accounting Review (January 1979): 107-18.
, AND W. F. WRIGHT. "An Empirical Analysis of Differential Capital Market Reaction
to Extraordinary Items." Journal of Finance (May 1976): 611-28.
FABOZZI, F. J . "Quality of Eamings: A Test of Market Efficiency." Journal of Portfolio
Management (Fall 1978).
FAMA, E . F. "Efficient Capital Markets: A Review of Theory and Empirical Evidence.'
Journal of Finance (May 1970): 383-417.
, AND H. BABIAK. "Dividend Policy: An Empirical Analysis." Journal of American
Statistical Association (December 1968): 1132-61.
, ET AL. "The Adjustment of Stock Prices to New Information.' International
Economic Review (February 1969): 1-21.
FiNNERTY, J. "Insiders' Activity and Inside Information: A Multivariate Analysis.' Journal
of Financial and Quantitative Analysis (June 1976a): 205-14.
. "Insiders and Market Efficiency.'' Journal of Finance (September 19766): 1141-48.
, R. N. FiTZSiMMONS, AND T. W. OLIVER. "Lease Capitalization, and Systematic
Risk." The Accounting Review (October 1980): 631-39.
FIRTH, M . "The Impact of Eamings Announcements on the Share Price Behavior of Similar
Type Firms." Economic Journal (June 1976): 296-306.
. "Qualified Audit Reports: Their Impact on Investment Decisions." The Accounting
Review (July 1978): 642-50.
FOSTER, G. "Stock Market Reaction to Estimates of Eamings Per Share by Company
Officials." Journal of Accounting Research (Spring 1973): 25-37.
. "Accounting Eamings and Stock Prices of Insurance Companies." The Accounting
Review (October 1975): 686-89.
. "Quarterly Accounting Data: Time-Series Properties and Predictive-Ability Re-
sults." The Accounting Review (January 1977a): 1-21.
"Valuation Parameters of Property-Liability Companies." Journal of Finance
(June 19776): 823-35.
. Financial Statement Analysis. Englewood Cliffs, N.J.: Prentice-Hall, 1978.
. "Briloff and the Capital Market." Journal of Accounting Research (Spring 1979)-
262-74.
"Accounting Policy Decisions and Capital Market Research." Journal of Account-
ing and Economics (March 1980): 29-62.
. "Intra-Industry Information Transfers Associated with Eamings Releases." Journal
of Accounting and Economics (December 1981): 201-32.
. "Capital Market Efficiency: Definitions, Testing Issues and Anomalies." Working
paper, Stanford University, February 1982.
FOSTER, T . W., Ill, AND D . VICKREY. "The Information Content of Stock Dividend
Announcements." The Accounting Review (April 1978): 360-70.
FREEMAN, R. N . "The Disclosure of Replacement Cost Accounting Data and Its Effect on
Transaction Volumes: A Comment." The Accounting Review (January 1981): 177-^.
. "Altemative Measures of Profit Margin: An Empirical Study of the Potential
Information Content of Current Cost Accounting." Working paper. University of Califor-
nia, Berkeley, 1982.
FRENCH, K. R. "Stock Retums and the Weekend Effect." Journal of Financial Economics
(March 1980): 55-70.
FRIED, D., AND A. SCHIFF. " C P A Switches and Associated Market Reactions.'' The
Accounting Review (April 1981): 326-41.
FRIEND, I., AND R. WESTERFIELD. "Required Disclosure and the Stock Markets: Comment."
American Economic Review (June 1975): 467-72.
GARMAN, M. "A Synthesis of the Pure Theory of Arbitrage." Working paper. University of
Califomia, Berkeley, 1978.
, AND J. A. OHLSON. "Information and the Sequential Valuation of Assets in
Arbitrage-Free Economies.'' Journal of Accounting Research (Autumn 1980): 420-40.
GHEYARA, K., AND J. BOATSMAN. "Market Reaction to the 1976 Replacement Cost Disclo-
sures." Journal of Accounting and Economics (August 1980): 107-25.
GivoLY, D., AND J. LAKONISHOK. "The Information Content of Financial Analysts' Fore-
casts of Eamings: Some Evidence on Semi-Strong Inefficiency." Journal of Accounting
and Economics (December 1979): 165-85.
GONEDES, N . J. "Evidence on the Information Content of Accounting Messages: Account-
ing-Based and Market-Based Estimates of Systematic Risk." Journal of Financial and
Quantitative Analysis (July 1973): 407-44.
"Capital Market Equilibrium and Annual Accounting Numbers: Empirical Evi-
dence." Journal of Accounting Research (Spring 1974): 26-62.
. "A Note on Accounting-Based and Market-Based Estimates of Systematic Risk."
Journal of Financial and Quantitative Analysis (June 1975a): 355-65.
"Risk, Information, and the Effects of Special Accounting Items on Capital Market
Equilibrium.'' Journal of Accounting Research (Autumn 19756): 220-56.
. "Class Discussion Notes." Manuscript, University of Chicago, 1976.
-. "Corporate Signaling, External Accounting, and Capital Market Equilibrium: Evi-
dence on Dividends, Income, and Extraordinary Items.'' Journal ofAccounting Research
(Spring 1978): 26-79.
AND N. DOPUCH. "Capital Market Equilibrium, Information Production, and
Selecting Accounting Techniques: Theoretical Framework and Review of Empirical
Work." Studies on Financial Accounting Objectives: 1974. Supplement to Journal of
Accounting Research 12: 48-129.
N. DOPUCH, AND S. PENMAN. "Disclosure Rules, Information-Production, and
Capital Market Equilibrium: The Case of Forecast Disclosure Rules." Journal of Ac-
counting Research (Spring 1976): 89-137.
GOOD, W . R., AND J. R. MEYER. "Adjusting the Price-Earnings Ratio Gap." Financial
Analysts Journal (November/December 1973): 42-49, 81-84.
GORDON, M . J. "Postulates, Principles and Research in Accounting." The Accounting
Review (April 1964): 251-63.
GRANT, E. "Market Implications of Differential Amounts of Interim Information.'' Journal
of Accounting Research (Spring 1980): 255-68.
GRIFFIN, P. A. "Competitive Information in the Stock Market: An Empirical Study of
Eamings, Dividends and Analysts Forecasts." Journal of Finance (May 1976): 631-50.
. "Sensitive Foreign Disclosures: The Securities Market Impact" Study conducted
for the SEC Advisory Committee on Corporate Disclosure (November 1977): 694-733.
Usefulness to Investors and Creditors of Information Provided by Financial
Reporting: A Review of Empirical Accounting Research. Stamford, Conn.: FASB, 1982.
-, AND G. G. NICHOLS. "Segmental Disclosure Rules: An Empirical Evaluation."
Working paper, Stanford University, June 1976.
GROSSMAN, S. J. "Further Results on the Informational Efficiency of Competitive Stock
Markets." Journal of Economic Theory (June 1978): 81-101.
, AND J. E. STIGLITZ. "Information and Competitive Price Systems." American

Economic Review (May 1976): 246-53.
HAGERMAN, R . L. "The Efficiency of the Market for Bank Stocks: An Empirical Test."
Journal of Money, Credit and Banking (August 1973): 846-55.
. "A Test of Govemment Regulation of Accounting Principles." The Accounting
-, AND M. ZMIJEWSKI. "Some Economic Determinants of Accounting Policy Choice.''
Journal of Accounting and Economics (August 1979): 141-61.
HAKANSSON, N., G. KUNKEL, AND J. A. OHLSON. "Sufficient and Necessary Conditions for
Information to Have Social Value in Pure Exchange." Journal of Finance (1982).
HAMADA, R. S. "The Effect of the Firm's Capital Structure on the Systematic Risk of
Common Stocks." Journal of Finance (May 1972): 435-52.
HARRISON, T. "Different Market Reactions to Discretionary and Nondiscretionary Account-
ing Changes." Journal of Accounting Research (Spring 1977): 84-107.
HARRISON, W. T., JR. "Accounting Changes in Principles and Estimates: How Different are
They?" In Economic Consequences of Financial Accounting Standards: Selected Stud-
ies, pp. 253-75. Stamford, Conn.: FASB, 1978.
HART, 0. D. "On the Optimality of Equilibrium When the Market Structure Is Incomplete."
Journal of Economic Theory (December 1975): 418-43.
HAWKINS, D . F., AND W . J. CAMPBELL. Equity Valuation: Models, Analysis and Impli-
cations. New York: FEI, 1978.
HAWORTH, H., J. MATHEWS, AND C. TUCK. "Full Cost vs. Successful Efforts: A Study of a
Proposed Accounting Change's Competitive Impact." SEC Directorate of Economic and
Policy Research, Febmary 1978.
HILL, N . C , AND B . K. STONE. "Accounting Betas, Systematic Operating Risk, and Financial
Leverage: A Risk-Composition Approach to the Determinants of Systematic Risk.''
Journal of Financial and Quantitative Analysis (September 1980): 595-637.
HILLMER, S. C , AND P. L. Yu. "The Market Speed of Adjustment to New Information."
Journal of Financial Economics (December 1979): 321-45.
HiRSHLEiFER, J. "The Private and Social Value of Information and the Reward to Inventive
Activity." .American Economic Review (September 1971): 561-74.
HoLTHAUSEN, R. W. "Evidence on the Effect of Bond Covensmts and Management
Compensation Contracts on the Choice of Accounting Techniques.'' Journal of Account-
ing and Economics (March 1981): 73-109.
HONG, H., G. MENDELKEB, AND R . KAPLAN. "Pooling vs. Purchase: The Effects of
Accounting for Mergers on Stock Prices." The Accounting Review (Jsmuary 1978): 31-
47.
HoRWiTZ, B., AND R. KoLODNY. "Line of Business Reporting and Security Prices: An
Analysis of an SEC Disclosure Rule." Bell Journal of Economics (Spring 1977): 243-49.
. "The Elconomic Effects of Involuntary Uniformity in the Financisd Reporting of R
& D Expenditures.'' Journal of Accounting Research (Supplement 1980): 38-74.
"The FASB, the SEC and R & D." Bell Journal of Economics (Spring 1981): 249-
62.
IJIRI, Y . T?ie Foundation of Accounting Measurement: A Mathematical, Economic and
Behavioral Inquiry. Englewood Cliffs, N.J.: Prentice-Hall, 1967.
INGRAM, R. W. "An Investigation of the Information Content of (Certain) Social Respon-
sibility Disclosure.'' Journal of Accounting Research (Autumn 1978): 270-85.
JAFFEE, J. F., AND M. RUBINSTEIN. "The Value of Information in Impersonal and Personal
Markets." Working paper. University of Pennsylvania, 1975.
JARRELL, G. "Pro-Producer Regulation and Accounting for Assets: The Case of Electric
Utilities." Journal of Accounting and Economics (August 1979): 93-116.
JONES, C. P., AND R. H . LITZENBERGER. "Quarterly Eamings Reports and Intermediate
Stock Price Trends." Journal of Finance (March 1970): 143-48.
JOY, O. M., AND C. P. JONES. "Earnings Reports and Market Efficiencies: An Analysis of
the Contrary Evidence." Journal of Financial Research (Spring 1979): 51-63.
JOY, O. M., R. H . LITZENBEHGER, AND R. W. MCENALLY. "The Adjustment of Stock Prices
to Announcements of Unanticipated Changes in Quarterly Eamings." Journal of Ac-

counting Research (Autumn 1977): 207-25.
KAPLAN, R. S. "The Information Content of Financial Accounting Nximbers: A Survey of
Empirical Evidence." Paper presented at a Symposium on the Impact of Research in
Financial Accounting and Disclosure on Accounting Practice, Duke University, December
1975.
, AND R. ROLL. "Investor Evaluation of Accounting Information: Some Empirical
Evidence." Journal of Business (April 1972): 225-57.
KEENAN, M. "Models of Equity Valuation: The Great SERM Bubble." Journal of Finance
(May 1970): 243-73.
KiNNEY, W. "Covariability of Segment Eamings and Multisegment Company Retums."
The Accounting Review (April 1972): 339-45.
KocHANEK, R. "Segmental Financial Disclosure and Security Prices." The Accounting
Review (April 1975): 245-58.
LATANE, H. A., AND C. P. JONES. "Standardized Unexpected Eamings—A Progress Report."
Journal of Finance (December 1977): 1457-66.
LAUB, P . M . "On the Information Content of Dividends.'' Journal of Business (January
1976): 73-80.
LEE, C , AND K. ZUMWALT. "Association Between Altemative Accounting Profitability
Measures and Security Retums." Journal of Financial and Quantitative Analysis
(March 1981): 71-93.
LEFTWICH, R. "Evidence on the Impact of Mandatory Changes in Accounting Principles on
Corporate Loan Agreements." Journal of Accounting and Economics (March 1981): 3 -
36.
LEV, B. "On the Association Between Operating Leverage and Risk.'' Journal of Financial
and Quantitative Analysis (September 1974a): 627-41.
. Financial Statement Analysis: A New Approach. Englewood Cliffs, N.J.: Prentice-
Hall, 19746.
. "On the Adequacy of Publicly Available Financial Information for Security Analy-
sis." In Financial Information Requirements for Security Analysis, edited by A, R.
Abdel-khalik and T. C. Keller. Durham, N.C.: Duke University, 1978.
. "The Impact of Accounting Regulation on the Stock Market: The Case of Oil and
Gas Companies." The Accounting Review (July 1979): 485-503.
-, AND S. KuNiTZKY. "On the Association Between Smoothing Measures and the Risk
of Common Stocks." The Accounting Review (April 1974): 259-70.
-, AND S. SUNDER. "Methodological Issues in the Use of Financial Ratios." Journal
of Accounting and Economics (December 1979): 187-210.
-, AND B. YAHALOMI. "The Effect of Corporate Financial Statements on the Israeli
Stock Exchange." Management International Review 2-3 (1972): 145-53.
LiNTNER, J. "The Aggregation of Investor's Diverse Judgments and I*references in Purely
Competitive Securities Markets." Journal of Financial and Quantitative Analysis
(September 1%9): 347-400.
LITZENBERGER, R. H., AND A. P. BUDD. "Secular Trends in Risk Premiums." Journal of
Finance (September 1972): 857-64.
LITZENBERGER, R. H., AND K. RAMASWAMY. "The Effect of Personal Taxes and Dividends
on Capital Asset Prices: Theory and Empirical Evidence." Journal of Financial Econom-
ics (June 1979): 163-95.
LITZENBEHGER, R. H., AND C. U. RAO. "Estimates of the Marginal Rate of the Preference
and Average Risk Aversion of Investors in Electric Utility Shares: 1960-66." Bell Journal
of Economics and Management Science (Spring 1971): 265-77.
LiVNAT, J. "A Generalization of the API Methodology as a Way of Measuring the
Association Between Income and Stock Prices." Journal of Accounting Research (Au-
tumn 1981): 350-59.
MAIER, S . F., D . W. PETERSON, AND J. H. VANDER WEIDE. "A Practical Theory of fi-
Estimation." Paper presented at Eastem Finance Association Meetings, Boston, April
1977.
MALKIEL, B . G., AND J. C. CRAGG. "Expectations and the Stmcture of Share Prices."
American Economic Review (September 1970): 601-17.
MANEGOLD, J. G. "Time-Series Properties of Eamings: A Comparison of Extrapolative and
Component Models." Journal of Accounting Research (Autumn 1981): 360-73.
MARSHALL, R. M. "Interpreting the API." The Accounting Review (January 1975): 99-111.
MARTIN, D . "Early Warnings of Bank Failure: A Logit Regression Approach." Journal of
Banking and Finance (November 1977): 249-76.
MAY, R. "The Influence of Quarterly Eamings Announcements on Investor Decisions as
Reflected in Common Stock Price Changes." Empirical Research in Accounting: Se-
lected Studies, 1971. Supplement to Journal of Accounting Research 9:119-63.
, J. HARKINS, AND S. RICE. "Lease Disclosure: The Effect of an Accounting Policy
Change on Security Prices." Working paper. University of Washington, March 1978.
-, AND G. SuNDEM. "Cost of Information and Security Prices: Market Association
Tests for Accounting Policy Decisions." The Accounting Review (January 1973): 80-94.
McNiCHOLS, M., AND J. MANEGOLD. "Financial Disclosure and the Behavior of Security
Prices: An Empirical Investigation." Working paper. University of Califomia, Los Angeles,
Febmary 1982.
MEADER, J. W . "A Formula for Determining Basic Values Underlying Common Stock
Prices." The Analyst (November 29,1935): 749.
MEHTA, D . R., ET AL. "The Influence of Dividends, Growth and Leverage on Share Prices
in the Electric Utility Industry: An Econometric Study." Journal of Financial and
Quantitative Analysis (December 1980): 1163-96.
MERTON, R. "The Theory of Rational Option Pricing." Bell Journal of Economics and
Management Science (Spring 1973): 141-83.
MILLER, M. H., AND F. MODIGLIANI. "Some Estimates ofthe Cost of Capital to the Electric
Utility Industry." American Economic Review (June 1966): 334-91.
MLYNARCZYK, F. A. "An Empirical Study of Accounting Methods and Stock Prices."
Empirical Research in Accounting: Selected Studies, 1969. Supplement to Journal of
Accounting Research 7: 63-81.
MORSE, D . "Price and Trading Volume Reaction Surrounding Eamings Announcements: A
Closer Examination." Journal of Accounting Research (Autumn 1981): 374-83.
MYERS, S. C. "The Relation Between Real and Financial Measures of Risk and Retum." In
Risk and Return in Finance, edited by T. Friend, and A. Bicksler, pp. 49-80. Cambridge,
Mass.: Ballinger Publishing Co., 1977.
NICHOLS, W. D., AND S. L. BROWN. "Assimilating Eamings and Split Information.'' Journal
of Financial Economics (September 1981): 309-15.
NiEDERHOFFER, V., AND P. REGAN. "Earnings Changes, Analysts' Forecasts, and Stock
Prices." Financial Analysts Journal (May/June 1972): 65-71.
NIELSEN, N . C. "The Investment Decision ofthe Firm under Uncertainty and the AUocative
Efficiency of Capital Markets." Journal of Finance (May 1976): 587-602.
NoREEN, E., AND J. SEPE. "Market Reactions to Accounting Policy Deliberations: The
Inflation Accounting Case." The Accounting Review (April 1981a): 253-69.
"Market Reaction to Accounting Policy Deliberations: The Inflation Accounting
Case Revisited—A Reply." The Accounting Review (October 19816): 955-58.
NoREEN, E., AND M. WoLFSON. "Equilibrium Warrant Pricing Models and Accounting for
Executive Stock Options." Journal of Accounting Research (Autumn 1981): 384-98.
O'CONNOR, M., AND D . COLLINS. "Full Cost vs. Successful Efforts Accounting in the Oil
and Gas Industries: A Closer Look at the Potential Market Consequences." Working
paper, Michigan State University, May 1977.
O'DoNNELL, J. L. "Relationships Between Reported Eamings and Stock Prices in the
Electric Utility Industry." The Accounting Review (January 1%5): 135-43.
OHLSON, J. A. "On the Theory of Residual Analysis and Abnormal Performance Metrics."
Australian Journal of Management (October 1978): 175-93.
. "Risk, Return, Security-Valuation and the Stochastic Behavior of Accounting
Numbers." Journal of Financial and Quantitative AnalysLs (June 1979a): 317-36.
. "Residual (API) Analysis and the Private Value of Information." Journal of
Accounting Research (Autumn 19796): 506-27.
. "On Financial Disclosure and the Behavior of Security Prices.'" Journal of Account-
ing and Economics (December 1979c): 211-32.
"Efficiency and the Social Value of Public Information in Exchange Economies."
Working paper. University of Califomia, Berkeley, 1981a.
. "Market Efficiency: A Comment." Manuscript note. University of Califomia,
Berkeley, 19816.
. "The Structure of Asset Prices and Socially Useless (Useful) Information." Working
paper. University of Califomia, Berkeley, 1981c.
, AND A. G. BucKMAN. "Toward a Theory of Financial Accounting." Journal of
Finance (May 1980): 537-47.
, AND A. G. BucKMAN. "Toward a Theory of Financial Accounting: Welfare and
Public Information." Journal of Accounting Research (Autumn 1981): 399-433.
OPPENHEIMER, H. R., AND G. G. SCHLARBAUM. "Investing with Ben Graham: An Ex Ante
Test of the Efficient Market Hypothesis.'' Journal of Financial and Quantitative
Analysis (September 1981): 341-60.
PATELL, J. M. "Corporate Forecasts of Eamings Per Share and Stock Price Behavior:
Empirical Tests." Journal of Accounting Research (Autumn 1976): 246-76.
. "The API and the Design of Experiments." Journal of Accounting Research
(Autumn 1979): 528-49.
, AND M. A. WOLFSON. "Anticipated Information Releases Reflected in Call Option
Prices." Journal of Accounting and Economics (August 1979): 117-40.
, AND M. A. WOLFSON. "The Ex Ante and Ex Post Price Effects of Quarterly Eamings
Announcements Reflected in Option and Stock Prices." Journal of Accounting Research
(Autumn 1981): 434-58.
, AND M. A. WOLFSON. "The Intraday Speed of Adjustment of Stock Prices to
Eamings and Dividend Announcements." Working paper. University of Chicago, January
1982.
PATZ, D., AND J. BOATSMAN. "Accounting Principle Formulation in an Efficient Market
Environment." Journal of Accounting Research (Autumn 1972): 392-403.
PENMAN, S. H . "An Empirical Investigation of the Voluntary Disclosure of Corporate
Eamings Forecasts." Journal of Accounting Research (Spring 1980): 132-60.
"Insider Trading and the Dissemination of Firms' Forecast Information." Journal
of Business (October 1982a): 479-503.
. "The Predictive Content of Eamings Forecssts and Dividends." Working paper.
University of Califomia, 19826.
PETTIT, R. R. "Dividend Announcements, Security Perfonnance, and Capital Market
Efficiency." Journal of Finance (December 1972): 993-1007.
PFEIFFER, G. M. "Accounting for Long-Term Financial Commitments: An Investigation
into the Economic Effects of Lease Accounting Changes." Ph.D. dissertation, Cornell
University, 1980.
RADNER, R. "Competitive Equilibrium under Uncertainty.'' Econometrica (January 1968):
31-58.
RAMAN, K. K. "Financial Reporting and Municipal Bond Rating Changes." The Accounting
REINGANUM, M . R. "Misspecification of Capital Asset Pricing." Journal of Financial
Economics (March 1981): 19-46.
RICE, S. J. "The Information Content of Fully Diluted Earnings Per Share." The Accounting
Review (April 1978): 429-38.
RICHARDSON, G. D . "Determinants of the Information Content of Annual E^nings An-
nouncements and the Association Between Information Content and Insider Trading
Activity." Research proposal, Cornell University, January 1982.
RICKS, W. "Market Assessment of Altemative Accounting Methods: A Review of the
Empirical Evidence.'' Journal of Accounting Literature (Spring 1982a): 59-102.
"The Market's Response to the 1974 LIFO Adoptions." Journal of Accounting
Research (Autumn 19826, pt. I): 367-87.
Ro, B. T. "The Disclosure of Capitalized Lease Information and Stock Prices." Journal of
Accounting Research (Autumn 1978): 315-40.
"The Adjustment of Security Retums to the Disclosure of Replacement Cost
Accounting Information." Journal of Accounting and Economics (August 1980): 159-89.
-. "The Disclosure of Replacement Cost Accounting and Its Effect on Transaction
Volumes." The Accounting Review (January 1981): 70-84.
ROLL, R. "A Critique of the Asset Pricing Theory's Tests." Journal of Financial Economics
(May 1977): 129-76.
. "The Tum-of-the-Year Effect and the Retum Premia of Small Firms." Working
paper. University of Califomia, Los Angeles, January 1982.
, AND S. A. Ross. "An Empirical Investigation of the Arbitrage Pricing Theory."
Journal of Finance (December 1980): 1073-1103.
ROSENBERG, B., AND J. GUY. "Prediction of Beta from Investment Fundamentals." Finan-
cial Analysts Joumal {M&y/June 1976): 1-15, and (July/August 1976): l - U .
"Prediction of Beta from Investment Fundamentals: Part Two, Altemative Predic-
tion Methods." Financial Analysts Journal (July/August 1976): 62-70.
ROSENBERG, B., AND W. MCKIBBEN. "Tlie Prediction of Systematic and Specific Risk in
Common Stocks." Journal of Financial and Quantitative Analysis (March 1973): 317-
34.
Ross, S. A. "The Arbitrage Theory of Capital Asset Pricing." Journal of Economic Theory
(December 1976): 341-60.
. "A Simple Approach to the Valuation of Risky Streams." Journal of Business (July
1978): 221-45.
RUBINSTEIN, M . "The Valuation of Uncertain Income Streams and the Pricing of Options.''
Bell Journal of Economics and Management Science (Autumn 1976): 407-25.
. "Competition and Approximation." Bell Journal of Economics (Spring 1978): 280-
86.
SHANK, J. K., J. F. DILLAR, AND R. J. MURDOCK. "FASB No. 8 and Decision Makers."
Financial Executive (February 1980).
SHILLER, R. J. "DO Stock Prices Move Too Much to Be Justified by Subsequent Changes
in Dividends?" American Economic Review (June 1981): 421-36.
SiMMONDS, R., AND D. CoLLiNS. "Line of Business Reporting and Security Prices: An
Analysis of a SEC Disclosure Rule: Comment." Bell Journal of Economics (Autumn
1978): 646-58.
SiNKEY, J. F. "Identifying 'Problem' Banks: How Do the Banking Authorities Measure a
Bank's Risk Exposure?" Journal of Money, Credit and Banking (May 1978): 184-93.
SMITH, A. J. "Stock Price and Risk Related to an Unanticipated Change in Accounting
Standards: The SEC's Reversal of SFAS No. 19." Working paper, Comell University,
January 1980.
SMITH, C , AND J. ZIMMERMAN. "Valuing Employee Stock Option Plans Using Option
Pricing Models." Journal of Accounting Research (Autumn 1976): 357-64.
SPROUSE, R., AND M . MOONITZ. A Tentative Set of Broad Accounting Principles for
Business Enterprises. New York: AICPA, 1962.
STAUBUS, G. "The Association of Financial Accounting Variables with Common Stock
Values." The Accounting Review (January 1965): 119-34.
STERLING, R. R. Theory and Measurement of Enterprise Income. Lawrence: University
Press of Kansas, 1970.
SUNDER, S. "Relationship Between Accounting Changes and Stock Prices: Problems of
Measurement and Some Empirical Evidence." Empirical Research in Accounting:
Selected Studies, 1973. Supplement to Journal of Accounting Research 11: 1-45.
. "Stock Price and Risk Related to Accounting Changes in Inventory Valuation."
The Accounting Review (April 1975): 305-15.
THOMPSON, R. "The Information Content of Discounts and Premiums on Closed-End Fund
Shares." Journal of Financial Economics (June/September 1978): 151-86.
ToBiN, J. "Liquidity Preference as Behavior Toward Risk." Review of Economic Studies
(Febmary 1958): 65-86.
TREYNOR, J. "The Trouble with Eamings." Financial Analysts Joumal (September/

October 1972): 41-46.
TwoMBLY, J. "An Empirical Analysis of the Information Content of Segment Data in
Annual Reports fi-om an FTC Perspective." In Proceedings of the Disclosure Criteria
and Segment Reporting Conference. Gainesville: University of Florida, 1977.
VASICEK, O. A . "A Note on Using Cross-Sectional Information in Bayesian Estimation of
Security Betas." Joumal of Finance (December 1973): 1233-39.
ViGELAND, R. L. "The Market Reaction to Statement of Financial Accounting Standards
No. 2." The Accounting Review (April 1981): 309-25.
WALLACE, W . A. "The Association Between Municipal Market Measures and Selected
Financial Reporting Practices." Joumal of Accounting Research (Autumn 1981): 502-
20.
WATTS, R. L. "The Information Content of Dividends." Joumal of Business (April 1973):
191-211.
. "Accounting Objectives." Working paper no. 7408, University of Rochester, 1974.
. "The Impact of Dividends on Eamings Announcements: A Reconciliation." Joumal
of Business (January 1976): 97-106.
"Corporate Financial Statements: A Product ofthe Market and Political Processes."
Australian Journal of Management (1977): 53-78.
. "Systematic 'Abnormal' Returns After Quarterly Eamings Announcements.'' Jour-
nal of Financial Economics (June/September 1978): 127-50.
, AND J. L. ZIMMERMAN. "Towards a Positive Theory of the Determination of
Accounting Standards.'" The Accounting Review (January 1978): 112-34.
, AND J. L. ZIMMERMAN. "On the Irrelevance of Replacement Cost Disclosures for
Security Prices." Journal of Accounting and Economics (August 1980): 95-106.
WEINSTEIN, M . "The Systematic Risk of Corporate Bonds.'" Journal of Financial and
Quantitative Analysis (September 1981): 257-78.
WEYGANDT, J. "Valuation of Stock Option Contracts." The Accounting Review (January
1977): 40-51.
WHITBECK, V. S., AND M. KisoR. "A New Tool in Investment Decision-Making." Financial
Analysts Journal (May/June 1963): 55-62.
WHITE, M . K. "Public Policy Toward Bankruptcy: Me-First and Other Priority Rules."
Bell Journal of Economics (Autumn 1980): 550-64.
WHITE, R. "On the Measurement of Systematic Risk.'' Ph.D. dissertation, Massachusetts
Institute of Technology, 1972.
WILSON, R. "Informational Economies of Scale." Bell Journal of Economics (Spring 1975):
184-95.
. "Information, Efficiency, and the Core of an Economy." Econometrica (July 1978):
807-16.

2

Uploaded by

Copyright:

Available Formats

2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2

Uploaded by

Copyright:

Available Formats

Journal of Accounting Research

Vol, 20 Supplement 1982

Market-Based Empirical Research

* Tel-Aviv University; t University of California, Berkeley, We are grateful to Ray Ball,

both quantity and quaUty, msu'ket-based research is undoubtedly a

out: (a) that empirical evidence which appears to be inconsistent with

L Expectations Raised by Market-Based Accounting

Second, and even more important, capital market efficiency provides

Third, according to the Capital Asset Pricing Model (CAPM), the

Fourth, most accounting measurement errors do not induce any rele-

Fifth, the distinction between individuals' behavior and aggregate

Sixth, fundamental (intrinsic-value) research using accounting infor-

Seventh, the accounting research program can be extended to validate

Eighth, MBAR is superior to other research methodologies in ac-

Thus, a decade ago expectations were high.^ It appeared that modem

Such gloomy observations did not deter researchers, who produced

partiaUy, rationalized by Dyckman, Downes, and Magee [1975, p. 90] by

This distinction between the ability of market studies to indicate effect

have been delegated to poUtical scientists and moral phUosophers.® In a

2. The Empirical Evidence: Review and Interpretation

weU-known API metric, which simulates the outcome of a portfolio

Beaver [1968] introduced yet another element to the study of security-

anism by which it occurs. The problem is then to demonstrate that

mation. This, however, would seem to be indisputable. Sharper inferences

Given strong beliefs, particularly among some practitioners, that cash

The marginal information contribution of the various disclosed ac-

method) to investors. This was not a direct test of investor abiUty to

the grouping of different types of accounting changes (discretionary and

gloomy observations about investor rationality made above should be

Stockholder equity. Consequently, the restrictiveness of any bond cove-

tional upon specific accounting methods (within GAAP) increases our

to be a preliminary evaluation of the expected impact of accounting

" See, for example, Wilson [1978].

1970, leading Collins to conclude that segmented data (especially sales)

exposure-draft release but that there was a subsequent price recovery.

impounded in stock prices due to private search efforts by analysts.^

the correlations between stock returns and the various price-adjusted

conflict to differences in experimental design, especially in the definition

associated with the effects of lease accounting on restrictive debt cove-

tion content in the auditor's report. This criticism of the difficulties of

research has markedly improved during the past decade. Limitations in

2.4 IMPLICATIONS OF ACCOUNTING RESEARCH FOR ECONOMIC

need only specify a statistical model of (expected) returns, and it is not

the announcement date. The usefulness of the market efficiency concept

Indeed, recent studies provide evidence that the CAPM is misspecified.

assessment of systematic risk. This section considers the latter, normative

Regarding BKS's prediction results, recent research on the nonstation-

in a previous section, many early tests purporting to find informational

3. Avenues for Future Market-Based Accounting

3.1 WELFARE E C O N O M I C S : SOME IMPLICATIONS FOR EMPIRICAL

This section opens with observations on some important trends in

strong "normative" orientation (stating which accounting structures

empirical associations between regulation and market data. A significant

research to concepts of welfare economics. The constructive issue is how

manner) the private and social costs of regulation and information

Homogeneous beliefs should be distinguished from homogeneous information; neither

informational item (or output of the information structure) is known by

to additional trading and the implication of strict welfare improvement

that would obtain in the postdisclosure economy. In addition,** individ-

~-contract curves ' '