Earnings Management with the help of historical cost accounting:
Not for managers but for investors
Aoyama Gakuin University
Graduate School of International Management
4-4-25 Shibuya, Shibuya-ku
Tokyo 150-8366, Japan
Revised October 2003
An accounting earnings figure under clean surplus is a change-of-value proxy
under current value accounting as generally expected, but is transformed into a
level-of-value proxy under historical cost accounting. The apparent informativeness of
the accounting earnings figure on the stock price is not just a self-fulfilling prophecy but
also a logical conclusion brought about by a much maligned principle of accounting.
However, it is not the end of the story. A particular accounting structure, i.e.,
historical cost accounting, catalyzes the process of self-fulfillment by inducing
managers to manage the earnings figure to convey more value relevant information to
investors under the Law of Conservation of Income.
Earnings management with the help of historical cost accounting:
Not for managers but for investors
Earnings management has been a hot topic in accounting research for years. The basic
premise of the research is that managers seek personal gains by smoothing or manipulating earnings
figures to the disadvantage of misled investors.
However, as Sunder (1997, p. 78) summarizes, “evidence in favor of the
income-smoothing hypothesis is weak and inconsistent.” In spite of thorough scrutiny by able
researchers, we do not have conclusive evidence yet. It poses a natural question: can we reach any
persuasive conclusion on this topic? First of all, can we detect it with publicly available data even if
managers actually manipulate earnings? Again Sunder (1997, p. 78) points out, “if managers leave
obvious tracks of their income management activity in publicly available data, their contract would
be modified to discourage them from doing so. It is not surprising that they cover their tracks
reasonably well.” If we assume the rationality of investors in the sense that managers cannot
systematically fool investors, earnings management should be endogenous as Demski (1996) and
Arya et al. (1998) claim. My analysis indeed belongs to this “endogenous” camp, though, unlike
them, I do not use the principal agent framework.
In addition, though real income management such as timing capital spending and
maintenance expenditure must be substantial, most of the literature concentrates on accounting
earnings management such as discretionary accruals. This strategy is understandable because it is not
feasible to estimate normal real income, i.e., income without real discretion, unless we are managers
ourselves. But a lost key is not necessarily found under the lamppost.
I shall first pose some questions about the premises of the current research and offer a
different perspective. I claim historical cost accounting and the random walk nature of accounting
earnings are intertwined with earnings management. As usual, an accounting structure shapes the
behavior of economic agents. Conclusions of the analysis follow. 2. Is the real income management minor?
Schipper (1989, p. 92) claims the inclusion of real earnings management is a “minor”
extension. However, it is not unreasonable to argue the opposite, that is, the amount affected by real
earnings management is quantitatively substantial and conceptually significant.
First, in a modern manufacturing firm, a substantial part of its expenditures is fixed in the
short-run. Therefore, managers have a great deal of discretion in choosing the timing of capital
spending, and thus manage their earnings. The timing of discretionary expenses such as R & D,
advertising and maintenance can be shifted a year or two without any substantial effects on the
long-run performance of the firm.
Second, unlike accounting earnings management, real income management is beyond the
control of accounting standard setters and external auditors. Even if we succeed in implementing and
enforcing rigid standards which make accounting earnings management impossible, a firm can use
real, instead of accounting, decisions to manage earnings. Plugging the “loophole” of earnings
management may be impossible unless the control of the firm’s expenditure itself is taken out of the
hands of its managers. This “road to serfdom” is hardly a desirable policy alternative.
Consequently, real income management is likely to be more important than accounting one
in actual practices because managers have a great deal of discretion in real income management due
to the lack of constraint imposed by external auditors.1 Indeed, one of the important functions of
budgetary control is supposed to be the adjustment of expenditures and other activities to ever
changing environments in order to meet the “target” earnings. As is shown in Figure 1, managers
have two types of means in managing earnings that in turn have two types of influence on results. In
brief, I want to emphasize the importance of Cell 4 as well as Cell 3 in Figure 1 in the research of
earnings management. 3. Is the self-serving manipulation a matter of fact?
The inherent difficulty of estimating unobservable “true” earnings notwithstanding, the
premise that managers manipulate earnings for their self-serving purpose is not so self-evident. Why
would investors enter into contracts in which they are systematically ripped off by managers? It is
very difficult to accept such an argument that investors are fooled all the time. Above all, investors
can substantially influence the welfare of managers in an indirect but definitive way, i.e., not buying
shares in the first place and selling any shares they hold.
Though being often treated as a 10 billion-dollar worth Robinson Crusoe in the literature, a
typical CEO is only one of the players in the game called a firm, and plays it not as a dictator but as
a corporate bureaucrat.2 The objectives of individual employees are naturally quite different from
those of their CEOs. The latter cannot arbitrarily direct their subordinates to do their bidding,
especially if they seek to advance their personal interests at the expense of other core employees.
Surveillance of CEOs by junior colleagues must be more pronounced if the welfare of junior
colleagues is closely tied to the future of the firm. Indeed, labor market is internalized3 and
promotion is a kind of tournament in large U.S. as well as Japanese firms.4 Most U.S. CEOs are
core employees of the firm, having served the firm for 20 years or more (Vancil 1987, Cosh and
1 Dechow and Sloan (1991, p. 78) also point out this aspect though Murphy and Zimmerman (1993)
challenge the robustness of their empirical evidence.
2 Even in Business Week
, which often depicts corporate leaders as superhuman heroes or villains, Byrne
(1987, p. 33) claims that “Today’s CEO rarely runs everything. An imperial facade may hide surprisingly
3 Within the boundary of one specific corporate group, but not necessarily one legal entity.
4 See Rosenbaum (1979) for the Unites States and Hanada (1993) for Japan.
Hughes 1987 and Kato and Rockel 1992).5 IBM’s Louis Gerstner is a conspicuous exception, not a
norm; routine advancements within the firm do not make news headlines.
Also, accounting standards are constructed outside the control of corporate managers on
the basis of user-primacy, and evidence on capture of standard setting by managers is hardly
persuasive.6 Financial statements are also audited by independent auditors who have different
objectives from managers’. Even if motivation for manipulation of earnings for self-serving
purposes exists, its magnitude, at least that of accounting earnings management, is constrained by
these governance structures.
Instead, we have some plausible counter-arguments against the assumption that managers
single-handedly pursue their personal pecuniary gains with seemingly unbounded rationality. First of
all, a successful liar needs a perfect memory. Also, in order to manipulate earnings for their personal
benefit, CEOs would need to persuade their junior colleagues not to follow their own personal goals
in often not so friendly environments. Considering the limited human information processing
abilities and discretionary powers, honesty pays in the long run, as Benjamin Franklin pointed out
Yes, in the long run, we are all dead. But, the modern public corporation is a going concern.
There are many devices to mitigate opportunistic behavior. For example, Vancil (1987) shows that
most large U.S. firms adopt relay process, that is, the joint management with the next CEO-candidate
for an extended period to mitigate the short-sighted management before the incumbent CEO turnover.
In the sample of 56 R & D intensive firms, Dechow and Sloan (1991) find more than 80%8 of them
adopted the relay process. In addition, 50 out of 56 retiring CEOs remained in the firm after their
retirement. In macroeconomics, models based on infinitely living (representative) agents are very
popular, and implications based on those models are often not far from the reality. For example, the
dynasty assumption is crucial for the validity of the famous Ricardian equivalence theorem,9 which
captures the real economy fairly well. I believe a large public corporation as an economic decision
entity is the closest analog to the infinitely living representative agent in the real economy.
Moreover, though CEOs are routinely assumed not to care about personal reputation after
they retire, their marginal utility of being respected as an elderly statesman may well exceed the
utility of a few extra millions in their saving account.10 I suggest not that CEOs are altruists but
5 The CEO market does not seem to be a contestable market (Baumol et al. 1988) because it needs low
entry barrier with small, if any, sunk cost which does not hold in the CEO market.
6 The universal application of rational choice theory is self-defeating. Its logical consequence is that we
should not take what researchers at business schools assert at face value because they are surely captured
by their “clients,” Corporate America. Though there may be a grain of truth in this reasoning, researchers
as well as managers seem to have other motivations than material well-being.
7 See also Akerlof (1983).
8 Even though they adopt a relatively narrow definition of relay process.
9 See Barro (1974).
10 “The outstanding discovery of recent historical and anthropological research is that man’s economy, as
a rule, is submerged in his social relationships. He does not act so as to safeguard his individual interest in
rather that monetary rewards are not the only arguments in their utility function.11 Purely
self-interested people can enhance their utility by doing the “right” thing. As Hume (1978, p. 415)
points out, “Reason is, and ought only to be the slave of the passion, and can never pretend to any
other office than to serve and obey them.” Rationality as understood in economic analysis is
instrumental and silent on substance. It is not unreasonable to conjecture that already well-to-do
CEOs care a great deal about reputation of a rather abstract kind. Many of them enter public service
where they earn a mere fraction of their private sector earnings.12
Time consistency argument raises another question about the validity of self-serving
manipulation argument. Real world contracts are not complete, having much room for discretion.
Therefore correspondence between reported earnings and compensation may not be well-defined ex
post, even if it seems ex ante. For example, Dechow et al. (1994) argue that compensation
committees adjust earnings-based incentive compensation ex post, though evidence is
inconclusive.13 Jensen and Murphy (1990) also point out the insensitivity of compensation to
performance.14 Moreover, the most (and only15) robust empirical regularity established so far is the
high correlation between compensation and firm size, which supports the assertion of Simon (1957)
that hierarchical structure, not performance, determines the salaries of managers.16
The case of Robert Allen of AT&T implies another serious doubt on the validity of
performance-related pay in general. In spite of the disastrous performance of the company in the
recent years, he is reported to have received a huge bonus.17 Of course, Allen could give us many
pieces of evidence on how vital his efforts are to save the company from a more serious financial
trouble that might have happened without his leadership.18 Taking also into account the fact that
most board members of a particular company are executives of other companies, we cannot expect a
the possession of material goods; he acts so as to safeguard his social standing, his social claims, his
social assets. He values material goods only in so far as they serve this end.” (Polanyi 1945, p. 53, quoted
in Hargreaves Heap and Varoufakis 1995, p. 161).
11 This argument is more akin to Becker (1976) than to Polanyi (1945), though.
12 Even President Clinton’s annual salary is only two hundred thousand dollars; even some university
professors earn more.
13 They claim that the compensation is shielded from the negative effect of restructuring charge at a
statistically significant level. But the estimated increase of the compensation through the committees is
about 50,000 dollars or six percent of the annual cash compensation, which is substantial for a lowly
doctoral student but seems too low to affect the decision of Fortune 500 CEOs on whether restructuring
on which their positions themselves are surely at stake be implemented. As Goldberger (1991) and
McCloskey (1985, 1996) rightly point out, statistical significance is not the same as economic
significance. Its abuse in the literature is pervasive and many researchers equate statistical significance
with economic one without careful examination. See Black (1982) and Leamer (1983) for more
comprehensive criticism on econometric practices in general.
14 But, their much-publicized finding seems to be an order-of-magnitude underestimate. See Cyert et at.
15 Some may count the fact that the older CEOs become, the more the resignation is likely as another
16 As Simon points out in his Nobel lecture (1979, p. 497), the neo-classical explanation of Lucas (1978)
needs many ad hoc auxiliary assumptions as complex as what is to be explained.
17 Allen does not seem to be an exception.
18 This story is suggested by Shyam Sunder.
high-powered compensation scheme sensitive to minor earnings changes, even though formally
disclosed formulas seem to be so.19
Actually, the causality of seemingly related performance pay may be in the opposite
direction: after the amount of compensation is determined exogenously, the “formula” is set up to
lead to that amount. This is said to be a common practice in feasibility studies in which assumptions
are “adjusted” until the desired results are obtained. Furthermore, many (most?) social scientific
studies, of course including this piece, may have precisely this nature.
In summary, managers are expected to have a fairly strong incentive to behave in
accordance with investors’ interests. At least the premise that managers manipulate reported income
substantially for their
self-serving purposes should not be taken as a fact to be explained. Anyway,
no conclusive evidence has ever been offered. Besides, I want to emphasize that I am a
methodological individualist and my framework is in line with this stance. A firm is a bundle of
activities and the interest of participants must be maximally coordinated if participants are assumed
not to be fooled systematically.20 On the other hand, most researchers in the literature seem to
discard methodological individualism by equating the firm with the CEO. Also, I want to emphasize
the role of monitoring by other managers, which seems to be underestimated, rather than the CEO’s
reputation, which has a danger of explaining everything.
However, I do not claim managers keep themselves from managing earnings. I rather want
to investigate another possibility: managers manage earnings for their self-interest defined broadly,
not narrowly, and by acting for the benefit of investors, they manage to achieve their own personal
goals in the long-run.
Before discussing earnings management of a different kind, I want to clarify the
oft-neglected and tricky relation among the value, stock price and accounting number in the next
section. 4. Is value beneath the surface or just out there?
relevance21 of accounting numbers including earnings figures is no less debated
topic than earnings management in empirical accounting research. However, despite the use of
advanced statistical techniques, researchers often fail to specify what the value means. Black (1993,
19 As for compensation with stock options, the more serious mystery to be resolved is why stock options
themselves are used instead of just linking compensation to stock prices, which could enhance the value
of the firm because expensed compensation is tax deductible for the firm in the United States. Tax
consideration for managers, usually asserted, is not
sufficient to justify the practice (Stiglitz 1994, pp.
20 I do not consider an inefficient equilibrium brought about by informational cascades discussed in
Chapter 4, because investors can exit easily and anonymously from the scene just by selling shares.
21 Though I only consider the relevance for the level of stock price, the earnings figure is expected to
have the information content for the level and
change of stock price because some investors surely try to
extract information on the change of value from the earnings figure. I am indebted to Shyam Sunder for
the clarification of the point. See also Zhang (1998) for an argument from an Ohlsonian point of view.
p. 2) address the issue with his usual clarity:
I think of “value” and “price” as distinct. Every firm and every security has a value, even
though we may have only imprecise estimates of it. Only a security that trades has a price,
and its price may differ from its value (though investors who think they see this difference
are often wrong).
Among accountants, the Ohlson camp is equally explicit saying “‘the market’ was wrong” (Bernard
1995, p. 735).22
According to these researchers, the value is related to the stock price and accounting
number as in Figure 2. In their framework, the accounting number and stock price are competing for
predicting the value that is unobservable. Because they assume the value is a kind of substance
beneath the surface different from a superficial market price, I call this approach the substance
theory of value. Actually most researchers seem to adopt this approach and to try to capture the
value relevance of the accounting number which is not
captured by the market price, though they are
not so explicit as Black (1993) and Bernard (1995).
However, there is sleight of hand in their approach: they assume
the market price gravitate
toward the fundamental value in the long run, though the former deviates from the latter in the short
run. Without this hidden assumption, the correlation between the stock price and accounting number,
which is all what they can and do find, does not tell anything about the unobservable value.
Moreover, if the market price deviates from the fundamental value for a long time, as they seem to
assume, the fundamental value cannot be well defined. If the market price is “wrong,” the
investment decision is “wrong.” Then, the consumption decision is “wrong,” which leads to the
“wrong” demand for and supply of goods. Then, we cannot know the fundamental value, which is
based on the economy being at equilibrium. Since Walras introduced auctioneer
in his general
equilibrium model, no one has ever given us any satisfactory out-of-equilibrium model.23 If
transaction is implemented off equilibrium, the fundamental value cannot be calculated.
In any case, this substance theory of value is not new at all, though Bernard (1995, p.734)
believes it to be. Rather, this is an old refrain from classical economists including Adam Smith and
Karl Marx, who distinguished the market price from the natural price, but was allegedly buried by
neo-classical subjectivism in the late nineteenth century. What we see is what we value: the market
price is the value. Figure 3 describes this approach.
I call this approach the relation (market) theory of value because the value is the market
relation. The relative price decided by the supply and demand is the value. The value of a security is
22 In their case, the book value is the
23 The disequilibrium analysis initiated by Robert Clower and others in the 1960s seems to have get
aborted before reaching any consensus among economists.
nothing but its posted market price as the value of a government liability called twenty-dollar bill is
twenty-dollar as posted. In this approach, if we find the correlation between the stock price/value
and accounting number, we can claim the accounting number has value relevance. I suspect many
practitioners are engaged in this kind of analysis.24 But this astrology-like25 approach does not seem
to be interesting enough to sustain our intellectual curiosity.
Actually, we have a more sophisticated relation approach that synthesizes the subjective
element of the relation theory and the objective element of the substance theory as described in
Figure 4. I call this approach the equilibrium relation theory of value because the market price
coincides with the fundamental value at equilibrium.26 If we want to maintain the fundamental value
framework, this approach is the only logical option. Continual equilibrium is not a choice but an
imposition. Therefore, it is more precise to claim the fundamental value is adjusted to coincide with
the market value. Isn’t it a tautology? Yes, any research program must start from some tautological
premises as the conservation of energy is imposed
on physical theories.
Also, “objective” used here must be understood as inter-subjective. At equilibrium, each
subjective belief is completely aligned and consequently become objective. This metamorphosis
from subjective into objective is an example which shows that inherently subjective social reality
becomes objective in human society (Searle 1995, pp. 7-9).
Although I myself have chosen the equilibrium relation theory of value as my starting
point, the following analysis is applicable to those who do not agree with my position. 5. What does the accounting earnings figure stand for?
In the United States and other developed economies, market participants decompose the
stock price into the earnings and the price-earnings ratio (P ? E·PER), though this formula is not an
equation but identity like the famous monetary identity, MV ? PY. However, as V (velocity) in the
monetary identity, the price-earnings ratio is thought to be relatively stable. With this (refutable)
assumption, the formula becomes equation.27 If the PER is stable, the earnings figure becomes a
summary figure for the value of the firm. Considering limited human computational abilities, using a
summary figure is a reasonable strategy for financial decision. Moreover, it is preferable from
managers’ point of view “because they can inform investors and creditors without informing
competitors. Many firms feel they already give too much information to competitors in their
financial statements.” (Black 1993, p. 3) Here, we can see the subtle interaction between what kind
of information be revealed and how managers and investors behave in response to it.
24 Ou and Penman (1989) seem to pursue this line of analysis.
25 Because I am not a demarcationist, being astrology-like should not be taken as derogatory or
unscientific, whatever it means. Many studies on cancer are also astrology-like in the sense that what they
find is only statistical correlation between cancer and something else.
26 Because the stock does not have a terminal value, we need the transversality condition.
27 Of course, no one claims the relation is exact. What is debated is how close the relation is, though the
judgment on closeness is inherently subjective.
Black (1993, p. 10) actually compared the variation in relation to price with the following
four summary numbers: option-adjusted primary earnings before extraordinary items (OEBE),
earnings before extraordinary items, interest, taxes, and depreciation and amortization (EBITD),
option-adjusted free cash flow (OFCF), and option-adjusted book value of equity (OBE). To
measure variation, he ranks the ratios and divide the first quartile point by the third. The results are
striking: for every year between 1980 and 1991, OEBE, an earnings figure, outperforms the three
rivals and OBE, a book value based number, is the worst.28 Black (1993, p. 11) also estimates the
timeliness of these four number and again OEBE outperforms the others in every lag up to 10 years.
I say “striking” because the earnings figure is in general based on historical cost and
criticized as lacking relevance for decision making. Another indirect evidence for the
informativeness of conventional accounting earnings is an apparent and puzzling lack of
informativeness of current value based earnings: “The major findings are simple and dramatic: (1)
Once historical cost earnings are known, the [FASB] Statement 33 earnings variables provide no
additional explanatory power with respect to differences across firms in yearly stock price changes.
(2) Even after any one of the Statement 33 earnings variables is known, knowledge of historical cost
earnings still provides additional explanatory power. In this sense, historical cost earnings strictly
dominate the Statement 33 earnings variables.” (Beaver and Landsman 1983, p. 10, quoted in
Mattessich 1995, p. 123)29
If summary figures were not sufficiently informative, managers would be required to
disclose more information, which may be Pareto-inferior to both investors and managers. Decision
usefulness is an important objective of accounting information, though it is not necessarily the
objective. From this point of view, the existence of some summary accounting figures that closely
track the economic income is desirable. In particular, if we have a proxy for the permanent income,
which investors want to know most, it could enhance the efficiency of the market greatly. Among
some candidates, the market has singled out the earnings figure, and empirical evidence suggests that
it does a good job. Therefore, it is hardly surprising that the correlation between the stock price and
the earnings figure is more stable than other plausible candidates such as cash flow and book value
However, if we equate the value with the stock price, isn’t the accounting number
irrelevant and redundant for the valuation of the stock because we know the value by definition? I
believe it is not necessarily so. First, “the accounting numbers are finding their way into the market
as the firm is preparing its accounting statements. The accounting process itself is informing the
28 Reported figures are OEBE (2.0), EBITD (2.8), OFCF (3.8), and OBE (4.7).
29 Mattessich seems to conclude practitioners lack expertise to use new information, which implies the
severely limited rationality of practitioners (p. 98). However, if historical cost earnings are already
constructed taking account of price changes, it is no wonder that historical earnings are more informative
than current value based ones. From a somewhat different angle, Lim and Sunder (1991) show that
historical cost accounting may provide the closest approximation of the current economic value of the
firm under the existence of measurement errors.
- Not for managers but for investors
- Yoshitaka Fukui