We model firms' choices between alternative means of presenting information, and the effects of different presentations on market prices when investors have limited attention and processing power. In a market equilibrium with partially attentive investors, we examine the effects of alternative: levels of discretion in pro forma earnings disclosure, methods of accounting for employee option compensation, and degrees of aggregation in reporting. We derive empirical implications relating pro forma adjustments, option compensation, the growth, persistence, and informativeness of earnings, short-run managerial incentives, and other firm characteristics to stock price reactions, misvaluation, long-run abnormal returns, and corporate decisions.
The paper is available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=334940
2. 2
Psychological Approaches
to Price Determination
Adam Smith
“The overweening conceit which the greater part
of men have of their own abilities…”
Keynes
“Animal spirits” in stock market
Irving Fisher
Self discipline as a determinant of interest rates
Markowitz
Psychological reference points and insurance
3. 3
Rise of efficient markets hypothesis in
1960s
Dominated accounting and financial research
Starting in early 1980s, appearance of
anomalous evidence about predictability
of stock returns and the processing of
accounting information
5. 5
Traditional asset pricing models failed to
explain the anomalies
Further blows:
Flat relation between beta and expected returns in recent
decades
1987 stock market crash, Japanese stock market behavior,
internet bubble of late 1990’s
Behavioral hypotheses began to proliferate
The “noise trader” theory offered an umbrella
approach to understanding anomalous
patterns as a consequence of exogenous
irrational trades.
DeLong et al (1990)
6. 6
Yet psychological explanations and
theory met with intense, prolonged
skepticism
Not `mainstream’ in finance for about two
decades.
Only recently making headway in accounting.
Despite a rich, compelling body of experimental evidence
on judgment and decision errors of investors,
professional analysts in the processing of accounting
information
Survey: Libby, Bloomfield and Nelson (2001)
Why?
7. 7
Objections to Behavioral Approaches
Much behavioral theory involved telling
stories, no modelling.
Noise trader approach:
Mechanistic investors trade exogenously, often
irrespective of price and other observables
Model in principle infinitely elastic:
Some choice of mechanistic trades can match ex post
any given return pattern.
Verrecchia (2001)
“The major difficulty with substituting some heuristic use of
information for Bayes rule is that potentially it explains
everything, which, in turn, suggests that it explains nothing.”
8. 8
Behavioral stories easy to construct ex
post after peeking at data:
Negative return autocorrelation = overreaction
Positive return autocorrelation = underreaction
PEAD, new issues puzzle = underreaction
Inconsistent, unreconciled stories?
Also, easy to pick out a separate
psychological story for each anomaly:
Model-mining?
9. 9
Some recent modelling approaches
have tried to address these concerns.
Goals:
Explicitly model endogenous choices of imperfectly
rational individuals
Firmly base the assumed biases of individuals in
the model to evidence from psychology
Derive a wide range of empirical implications from
just one/few well-documented biases.
Derive both underreaction and overreaction within
unified setting
Predict circumstances in which one or other should
obtain.
10. 10
Two early examples
Overconfidence/biased self-attribution
Daniel, Hirshleifer and Subrahmanyam (1998, 2001)
Odean (1998)
Book/market effect, size effect, short-run momentum, long-run return
reversal, PEAD, new issues puzzle, weak relation of beta to expected
returns, excess volatility and trading volume
Conservatism, representativeness
Barberis/Shleifer/Vishny (1998)
PEAD, overreaction to long-run earnings trends, short-run momentum,
long-run return reversal, new issues puzzle, excess volatility, regime-
shifting beliefs
Both approaches offer out-of-sample
predictions as well.
Some early wins and misses, but predictive success of these
approaches remains to be seen.
12. 12
Prices set solely by the`marginal’
[= fully rational] investor
Euphonious jargon, conceptual error
In perfect capital market with risk averse
investors, decisions of all investors influence
price.
Every investor is a marginal investor.
Shift in demand curve of any subset of
investors shifts aggregate demand, price.
Prices reflect a weighted average of beliefs.
Rational investors try to arbitrage away inefficiencies
Irrational investors think they are arbitraging away
inefficiencies.
13. 13
Can irrational traders survive?
Irrationally aggressive traders may:
Obtain higher risk premia
DeLong et al 1991
Exploit private information more fully
Kyle/Wang 1997, Fischer/Verrecchia 1999, Hirshleifer/Luo 2001,
Verrcchia 2001
So aggressive/overconfident may survive
If everyone is biased (e.g., limited attention), then
bias cannot be eliminated.
Opportunity cost to devoting more attention to a given issue.
Stock prices noisy, very long time needed for smart
players reliably to beat dumb players
smart players learn to be overconfident?
`sucker born every minute’
15. 15
We base our model upon a pervasive
psychological fact:
Limited processing power and attention
Immense body of evidence that limited attention biases
judgment and choices.
This evidence is a fundamental motivation of
our paper; see Section 2.
Also great deal of evidence that form of
presentation influences investor and analyst
judgments, choices behavior in experimental
settings
16. 16
Evidence of limited attention in
capital markets
Telling anecdotal examples (Rashes (2001)):
MCI Communications versus Massmutual Corporate Investors
(ticker: MCI)
Unrelated industries
Extremely high correlation in returns, volume,
volatility
Cross-effects of news
Ticker TCI vs. Tele-communications Incorporated
ATT takeover announcement of Tele-communications
Incorporated caused a 4.3% jump in price of firm with
TCI ticker symbol.
If people fail in such egregious ways to use publicly
available information, they almost surely make
subtler errors too. Evidence:
17. 17
Market reactions to salient republication
of publicly available news, functional
fixation
Ho/Michaely (1988), Huberman/Regev (2001),
Hand (1990)
Several accounting reporting contexts:
Amir (1993), Aboody (1996), Davis-Friday et al
(1999), Schrand/Walther (2000),
Lougee/Marquardt (2002),
Doyle/Lundholm/Soliman (2002),
Guay/Haushalter/Minton (2002)
18. 18
Our focus is on reporting
Aim to explain a number of puzzling stylized facts
Offer new empirical implications
By choosing a pervasive psychological fact,
our approach can potentially be extended to
a range of issues:
Aggregation
E.g., Segment reporting choice
Time allocation
E.g., Reporting of employee option compensation, pro forma
disclosure adjustments
Strategic reporting choices by management
E.g., pro forma disclosure adjustments
19. 19
General model
Limited attention has two possible
effects:
Signal Neglect
Neglect of publicly available information signals
Example: Ignore the fact that the firm is in a normal state of
the world in which pro forma adjustment is inappropriate
Analytical Failure
Neglect of a publicly known feature of the decision
environment
Example: Ignore manager’s strategic incentive to adjust pro
forma earnings when this increases earnings.
Example: Extrapolate total firm earnings instead of separately
extrapolating each segment according to its known growth
rate.
Example: Ignore the fact that outstanding employee stock
options may eventually be exercised, diluting equity.
20. 20
We model these two effects as the
investor setting a state variable or an
environmental parameter equal to an
arbitrary, incorrect value.
Empirical implications derive from
application-specific parameter
restrictions.
21. 21
Pro Forma Earnings Disclosures
Managers adjust upward pro forma
disclosures
Purportedly to remove non-recurring items.
Manager’s objective:
Higher stock prices
Reputation for fair dealing.
E state: upward adjustment appropriate
N state: inappropriate
Equilibrium:
If the possible upward adjustment is large enough,
managers adjust upward even in N state.
Limited attention: investors think the
manager will adjust if and only if E state.
22. 22
Implications (subset)
Investor valuations upward biased
Average overvaluation increases with excess pro
forma E
Long-run abnormal returns
Untested empirical implications on determinants of the steepness
of this relationship.
But pro forma adjustments can still make stock prices
more accurate on average
Comparative statics on Pr(adjusted pro forma
disclosure). Most counterintuitive:
Increases in the informativeness of earnings for future cash flow
(empirically testable).
When earnings are stronger indicator of value, stronger incentive
to manipulate perceptions of earnings.
23. 23
Managerial Option Compensation
If firms do not expense option costs at issuance,
inattentive investors neglect and overvalue the firm.
Degree of overvaluation depends upon persistence of earnings, as
investors extrapolate the `extra’ earnings based on persistence.
Empirical: long run abnormal returns
If options fully expensed, investors undervalue firm.
Correct investor perceptions can be induced by
amortization of option costs
Expense schedule depends on earnings persistence.
Analysis suggests why firms campaign politically
against required reporting of option expenses
Dechow/Hutton/Sloan (1996)
Affects market valuations
24. 24
Aggregation in Financial Reporting
In projecting future earnings, investors/analysts
should analyze operations of each segment, project
each at its growth rate.
Instead, inattentive investors extrapolate the firm's
earnings at its overall growth rate.
If growth rates constant, firm is undervalued under
aggregate reporting.
Periods of high foreseen general growth.
Extrapolating entire firm at past growth rate ignores the increasing
weight in firm value of faster-growing segments.
Focusing transactions increase valuations.
Effects stronger when greater inequality of growth rates.
Divest firms with divergent growth rates.
However: during low-growth periods, growth rates may revert to a
common mean quickly implications may reverse.
25. 25
Limitations of the Analysis
Early, simplified effort at modeling
limited attention in accounting.
Focuses on the capital market reporting function
of financial statements and disclosures,
We do not consider such issues as optimal
contracting, performance measurement, taxes,
and political constraints
E.g., Watts/Zimmerman (1986), Lambert (2001).
These considerations may affect the conclusions of
the analysis.
26. 26
Further directions for limited
attention in accounting
Aggregation
Does likelihood that a diversified firm disaggregate increase
when its recent earnings came mainly from more-persistent
segments?
Theory and empirics: Hann, Hirshleifer, Teoh (in progress)
Earnings management
Do both overreaction to accruals and underreaction to cash
flows/earnings derive from limited attention?
Hirshleifer and Teoh (in progress)
Derivatives and fair value accounting
Are firms reluctant to mark-to-market because unrealized
fundamental gains and losses are not salient to investors,
whereas a reported derivative gain or loss is salient?
27. 27
Contracting
Why are there “non-contractible” contingencies?
Can there be a reporting role for information aggregation?
Throws away information? Or helps investors focus?
Is redundancy desirable?
Information available to investors on each statement where
they are likely to need it for cognitive processing?
Why do voters let special interest groups manipulate
accounting regulation?
Why does the political process often fixate upon
salient items
E.g. large losses on derivative positions as opposed to
the gains being hedged.