afaanz 0677
TRANSCRIPT
-
8/8/2019 AFAANZ 0677
1/74
1
Explanations:
Why Post-Earnings-Announcement Drift Occurs,
Why Stock Prices Tend to Rise After a Stock Split,
and Some Reasons Why Managements Smooth Earnings
-
8/8/2019 AFAANZ 0677
2/74
-
8/8/2019 AFAANZ 0677
3/74
3
also explains (1) why the size of the drift varies by size of
the company, (2) that the market is not efficient, (3) why stock
prices tend to rise after a stock split, and (4) some of the
incentives for managements to smooth earnings.
Keywords: post-earnings-announcement drift; models; capital
asset pricing model; efficient market hypothesis;
market microstructure; stock splits; income smoothing;
earnings management; economic analysis; field study
-
8/8/2019 AFAANZ 0677
4/74
4
I. INTRODUCTION
Ball and Brown (1968) identified post-earnings-announcement
drift (PEAD) decades ago. Since the drift is incompatible with
our current theoretical underpinnings, then so long as it
remains unexplained, we cannot be assured that we have a correct
understanding of the capital markets. Research subsequent to
Ball and Brown (1968) has repeatedly and overwhelmingly
confirmed that PEAD does exist, but has not explained why PEAD
exists. The purpose of this paper is to explain why post-
earnings-announcement drift occurs.
The Importance of the Drift
This work has already been exposed for comment and review.
Earlier reviewers have criticized the inclusion of information
concerning what the phenomenon is and why the phenomenon is
important, in essence claiming that everybody knows that anyway.
However, not all possible readers are well-trained in capital
markets research (e.g., those whose training is in behavioral
research), so it is important to include the information for
such readers. Further, inclusion of this information is
important because capital-markets researchers who will become
well trained in the future may well have had this method of
-
8/8/2019 AFAANZ 0677
5/74
5
analysis incorporated into their training. Accordingly,
information on the importance of the drift is included in the
Appendix.
Is the Drift a Real Phenomenon?
Ball and Brown (1968, 170-173) found abnormal returns both
before and after the quarterly announcement of earnings. While
the abnormal returns before the earnings announcement are
relatively easy to theorize away as leakage of information into
the market, such is not the case with respect to the abnormal
returns after the earnings announcement (i.e., PEAD). Fama
(1998, 283) stated that most long-term return anomalies tend to
disappear with reasonable changes in technique or when (288)
alternative approaches are used to measure them, but Fama
(1998, 304) stated this about PEAD specifically:
Which anomalies are above suspicion? The post-earnings-announcement drift first reported by Ball and Brown (1968)has survived robustness checks, including extension to morerecent data (Bernard and Thomas, 1990; Chan et al., 1996).
-
8/8/2019 AFAANZ 0677
6/74
6
Liu, Strong, and Xu (2003, 90) said that "Fama (1998) refers to
PAD 1 as 'the granddaddy of underreaction events' and as the only
established anomaly above suspicion."
How Prior Research Was Done
The traditional methodology of PEAD research can be
summarized as follows. First, portfolios (varying in number)
were formed based on the magnitude and direction of earnings
surprises for quarterly earnings announcements. Second, the
actual market prices of securities were compared to prices
derived from a normative model (i.e., CAPM). Third, a
divergence from the normatively predicted price was found.
Fourth, portfolios with positive earnings surprises were found
to have had positive drift above the normatively predicted
price, while portfolios with negative earnings surprises were
found to have had negative drift below the normatively predicted
price. Fifth, the amount of drift was found to be monotonically
positive for increasing positive earnings surprise portfolios,
and monotonically negative for increasing negative earnings
1that is, the drift, which some authors call post-announcement
drift, or PAD
-
8/8/2019 AFAANZ 0677
7/74
7
surprise portfolios. Sixth, no satisfactory explanation for the
existence of PEAD was found. As Ball (1978, 105) stated, "the
notable feature of the anomaly is its consistency across the
studies."
Organization of the Work
Part II reviews some literature relevant to PEAD. Part III
discusses CAPM as a pricing model. Part IV uses a different
pricing model than CAPM to analyze the post-earnings-
announcement drift, plus the results of a field study to explain
why the drift occurs. Part V contains tests of hypotheses; the
results are consistent with the results in Part IV, but they
would not be consistent if the analysis in Part IV were
incorrect. Part VI discusses the implications of the analysis.
The Appendix contains information concerning why the drift is
important.
II. SOME BACKGROUND RELEVANT TO POST-EARNINGS-ANNOUNCEMENT DRIFT
Securities Markets
The world value of publicly traded securities (whether
equity or debt) plus the derivatives based on those securities
is a very large number. As a result, both academics and
-
8/8/2019 AFAANZ 0677
8/74
8
practitioners have devoted a great deal of effort to the theory
of capital markets, including the "proper" valuation of capital
assets. The PEAD anomaly exhibits a divergence of actual from
theoretical market prices, which means that present theory is
not correct.
The Post-Earnings-Announcement Drift Anomaly
Expressed simply, (1) stock prices adjust to a new level
upon an announcement of earnings, but (2) after the
announcement, prices begin to drift relative to the market for a
period of approximately sixty trading days, or about three
calendar months, then (3) make another significant adjustment
around the announcement of the following quarter's earnings.
Over a period of decades, there have been many published studies
by numerous researchers, typically resulting in a finding of
post-earnings-announcement drift for post-earnings-announcement
periods of months or quarters. Table 1 contains a partial list.
------------------------------
Insert Table 1 about here
------------------------------
-
8/8/2019 AFAANZ 0677
9/74
-
8/8/2019 AFAANZ 0677
10/74
10
adjustments that occur at those dates (320) (Bernard andThomas 1990, 305-306 and 320).
! Ball and Brown (1968) first documented post-earningsannouncement drift. Subsequent researchers sought eitherto examine the robustness of Ball and Brown's findings orto investigate the extent to which drift was observedbecause of the research methods used rather than because ofa market inefficiency. After 30 years of research, drifthas been shown to be very robust, and it remains anapparent market inefficiency (Soffer and Lys 1999, 308).
! Post-earnings announcement drift is among the mostpersistent market anomalies. ... Post-earningsannouncement drift ... is the predictability of abnormalreturns after the earnings have been announced. Ball andBrown (1968) initially observed the drift. Severalsubsequent studies have examined the phenomenon in greatdetail. This anomaly has persisted for over 33 yearsdespite extensive research to explain it (Asthana 2003, 1).
Present Theory of How Capital Markets Value Assets
The present paradigm of how capital markets value assets
consists of a combination of the capital asset pricing model
(CAPM) plus the efficient market hypothesis (EMH). The capital
asset pricing model is a normative model for the valuation of an
asset. The efficient market hypothesis addresses the speed with
which prices change.
-
8/8/2019 AFAANZ 0677
11/74
11
The Capital Asset Pricing Model (CAPM)
The capital asset pricing model (CAPM) (Copeland and Weston
1979, 169) is
The expected return on any security equals the risk-free rate
plus [(the expected market rate less the risk-free rate) times
(the covariance of the security with the market divided by the
variance of the market return)].
Observe that the risk-free rate, the expected market rate,
and the variance of the market return do not differ across
securities. Given an initial price observed in the market, the
expected return during a period can be translated into the
equilibrium price of the risky asset (Copeland and Weston 1979,
170). Thus, the CAPM becomes a normative pricing model. In the
capital asset pricing model (CAPM) formula, there is only one
term which varies by security, so--at least, according to the
formula-- the price of an asset is determined by its covariance
(which is measured by beta)--how it moves relative to the market
as a whole; nothing else matters, and nothing else can change
the price relative to the market.
) RVAR(
) R , RCOV(] R-) R[E(+ R=) R E(m
m j f m f j
-
8/8/2019 AFAANZ 0677
12/74
12
The Efficient Market Hypothesis (EMH)
The concept of the efficient market is one in which the
marketplace uses information as soon as the information becomes
available, and immediately evaluates all the effects of that
information in setting asset prices. The information and its
effects are not limited to factual information; the effect may
exist only on the expectations of market participants. Because
the efficient market hypothesis does not explicitly state a
formula for valuing assets, the concept that changes in
expectations can affect securities' values is compatible with
the efficient market hypothesis.
! In an efficient capital market, security prices fully
reflect available information in a rapid and unbiasedfashion and thus provide unbiased estimates of theunderlying values (Basu 1977, 663).
! A market in which prices always fully reflect availableinformation is called efficient." (Fama 1970, 383)
! I take the market efficiency hypothesis to be the simplestatement that security prices fully reflect all available
information (Fama 1991, 1575).
-
8/8/2019 AFAANZ 0677
13/74
13
The Effect of Combining the Capital Asset Pricing Model and theEfficient Market Hypothesis
A problem arises from the interaction of the EMH with a
mechanical application of the CAPM. If CAPM and EMH are both
correct simultaneously, then CAPM is how the market sets the
price, and everybody knows the correct price. Any movement away
from this price must be interpreted as mispricing. If this
mispricing occurs in an efficient market, then sophisticated
market participants can be expected to buy or sell as necessary
to force the security's price back to its theoretically
"correct" price. Therefore, PEAD cannot occur if CAPM and EMH
are both correct simultaneously.
Further, manipulation of any security's market price is
impossible. No buyer would pay more than the publicly known
CAPM price, and no seller would accept less than the publicly
known CAPM price. That is not what is observed to happen in
reality. Market manipulation occurs to such an extent that the
Securities and Exchange Commission does not even go after the
small cases.
-
8/8/2019 AFAANZ 0677
14/74
-
8/8/2019 AFAANZ 0677
15/74
15
! One argument supported by the EMH contends that if anysystematic method of obtaining abnormal excess returnsexists and if that method becomes known to the public, thenthe mechanics of an efficient market will negate therealization of any further benefits derived from the use ofthat method (Bidwell and Riddle 1981, 211).
! Perhaps the evidence most damaging to the naive andunwavering belief in market efficiency was the accumulationof results indicating the existence of persistent priceadjustments after earnings announcements were made; this isobviously incompatible with the instantaneous-adjustmentproperty of informationally efficient markets. ... Insome cases, the studies even implied that excess returnscould be earned by using investment strategies based onprice-change persistencies subsequent to earningsannouncements (Lev and Ohlson 1982, 284-285).
! Bernard and Thomas (1990, 308) called post-earnings-
announcement drift
. . . the most direct evidence . . . that a market-efficiency anomaly is rooted in a failure of information to
flow completely into price. The evidence also emphasizesthat even in a market where prices fail to reflect allavailable information, one can still observe unusual stock-price activity concentrated around news releases. Thepuzzling question is, if a portion of the "news" becamepredictable months earlier, why did the associated pricemovements not occur then?
! ...it is puzzling that, although so many years have gone bysince the post-announcement drift was first documented by
Ball and Brown (1968), arbitrageurs still have not fullytraded on this phenomenon and thereby eliminated it (Bartov1992, 622).
-
8/8/2019 AFAANZ 0677
16/74
16
! Hew, Skerratt, Strong, and Walker (1996, 283) stated:
Over the last quarter of a century a number of US researchstudies have reported an empirical phenomenon that has cometo be known as post-earnings-announcement drift (henceforthreferred to as PAD). Loosely speaking, the PAD phenomenoncan be described as a tendency for abnormal stock returnsto be predictable on the basis of past earnings news. Inother words, share prices fully react to earnings news onlywith a considerable lag.
The PAD phenomenon directly challenges the semi-strong formof the efficient markets hypothesis (EMH), which statesthat the market reacts fully and instantaneously to allpublicly available information. Of all the various sourcesof public information about a company, earnings isgenerally regarded as the single most important, regularlyreported item (see, for example, Beaver, 1981a, p. 135).Thus, if it can be shown that market prices fail to fullyimpound earnings information, this will seriously underminethe EMH. It is primarily for this reason that many leadingacademics regard PAD as one of the most serious empiricalchallenges to the EMH (see Ball, 1992 and Lev and Ohlson,
1982).
! Among the documented departures from market efficiency thepost-earnings-announcement drift (PEAD) is arguably themost puzzling. ... While the PEAD is well documented inthe literature, the reasons for the persistent under-reaction to earnings announcements are not well understood(Jacob, Lys, and Sabino 2000, 330).
! Ball and Brown (1968) document that a systematicrelationship between current unexpected earnings and stockreturns continues even after earnings are announced. Thisrelation implies that earnings information already publiclyavailable can be used to predict abnormal stock returns, aphenomenon termed post-earnings announcement (PEA) drift,
-
8/8/2019 AFAANZ 0677
17/74
17
which is a violation of the (semi-strong) efficient markethypothesis (Kim and Kim 2003, 383).
Importance of the Belief in Market Efficiency
The reigning paradigm in capital markets research assumes
market efficiency. The post-earnings-announcement drift is
evidence which contradicts the assumption because the market,
having known since at least 1968 that the drift exists,
nevertheless does not adjust for it.
! Finance theory is grounded on the assumption that securityprices adjust to new information as it becomes available(Latan and Jones 1977, 1457).
! The proposition that securities markets are efficient formsthe basis for most research in financial economics. Avoluminous literature has developed supporting thishypothesis. Jensen (1978) calls it the best established
empirical fact in economics. Indeed, apparent anomalies .. . are treated as indications of the failures of models .. . rather than as evidence against the hypothesis ofmarket efficiency. ... Despite the widespread allegianceto the notion of market efficiency, a number of authorshave suggested that certain asset prices are not rationallyrelated to economic realities. ... Arrow (1982) hassuggested that psychological models of "irrational decisionmaking" of the type suggested by Tversky and Kahneman(1981) can help to explain behavior in speculative markets.These types of claims are frequently dismissed because they
are premised on inefficiencies and hence imply the presenceof exploitable excess profit opportunities. This paperargues that existing evidence does not establish thatfinancial markets are efficient. . . . (592) ... Aclassic example is the discounts on closed end funds. Eventhough the underlying assets are easily valued, market
-
8/8/2019 AFAANZ 0677
18/74
18
values do not accurately reflect fundamentals (600). ...The evident difficulty economists have in explaining anysignificant amount of the variations in speculative priceson the basis of "news" about fundamentals also suggeststhat valuation errors are being made continuously (600)(Summers 1986, 591-592).
Get the Picture
The post-earnings-announcement drift phenomenon consists of
a drift, with a predictable pattern, in the prices of individual
securities relative to the market. Those prices of individual
securities increasingly diverge over time from the normative
CAPM price. Table 2 lists a number of studies which contain
pictorial representations of the drift.
------------------------------
Insert Table 2 about here
------------------------------
The variety of pictorial representations and variety of
accumulation periods indicate that there is no standard method
of pictorially representing the post-earnings-announcement drift
phenomenon. Consequently, Figure 1 displays a generalized
pictorial representation of post-earnings-announcement drift for
a positive earnings surprise, based on the text of earlier
-
8/8/2019 AFAANZ 0677
19/74
-
8/8/2019 AFAANZ 0677
20/74
20
that a firm announces positive (negative) unexpectedearnings for quarter t , the market tends to be positively(negatively) surprised in the days surrounding theannouncement for quarter t + 1.
Booth, Kalunki and Martikainen (1996, 1197) summarized
Bernard and Thomas (1989 and 1990) as suggesting "that at least
a portion of the price response to new information around
earnings announcements is delayed. The delay might occur, for
instance, because traders fail to assimilate available
information, or because of costly information processing."
III. PREVIOUS TESTS OF CAPM AS A PRICING MODEL
Previous researchers accepted CAPM as being the way the
capital market prices an asset, and when they found PEAD, they
concluded that they had found an anomaly. Unfortunately, they
did not start from the beginning; they assumed that CAPM is
correct rather than tested whether CAPM is correct. As Foster,
Olsen, and Shevlin (1984, 577) stated:
Assumptions typically made in these tests include that thetwo-parameter asset pricing model...--CAPM--is
descriptively valid....
If the early researchers would have taken the position that
CAPM was a hypothesis (null hypothesis: the market values
-
8/8/2019 AFAANZ 0677
21/74
21
securities according to CAPM) rather than a fact, then their
work which showed a market-price drift from the CAPM price would
have led them to conclude that the market does not price
securities in accordance with the capital asset pricing model.
Instead of saying that they found an anomaly where the actual
market price differs from the correct price per CAPM, they would
have said that CAPM does not accurately model the correct market
price .
The Nails in the Coffin
------------------------------
Insert Figure 2 about here
------------------------------
Consider Figure 2. Take a time line during which post-
earnings-announcement drift has continuously been found by using
CAPM as a normative price. (Recall that Ball and Brown 1968,
165 and 167, used data from 1946 to 1966.) Take any Point A
significantly after the start of the entire time period, assume
that this is the starting point of a study which finds PEAD, and
assume that the CAPM price is correct at this starting point.
-
8/8/2019 AFAANZ 0677
22/74
22
Thus, at any intermediate Point B, and at any final Point C,
because of PEAD, we know that the CAPM price is not the correct
price. However, Point A is itself an intermediate and/or final
point during a period when there was drift, so Point A is itself
a Point B or a Point C with regard to some earlier starting
point. Thus, Point A illustrates that the CAPM price is correct
only by assumption, but wrong by proof.
Beaver (1974, 568) said, "to say that a given model is
deficient implies that there is a 'better' model." The
preceding information indicates CAPM is deficient. This leads
to the question: What pricing model should be used instead of
CAPM? To answer this, we must first understand that the CAPM is
a model, and just a model. As a model, there are assumptions
which underlie the model.
! The CAPM is developed in a hypothetical world where thefollowing assumptions are made about investors and theopportunity set:
5. Asset markets are frictionless and information iscostless and simultaneously available to all
investors.
(Copeland and Weston 1979, 160-161)
-
8/8/2019 AFAANZ 0677
23/74
23
! A perfect capital market is a key assumption in recenttheories of security pricing. 2 It is assumed that the costsof transactions, information-gathering, and portfoliomanagement are all zero.... (Mao 1971, 1105)
So far, however, no one has examined the implications ofnonzero costs of transaction, information, and management.... ...the question arises of how the theory of securitypricing is affected by this form of market imperfection(Mao 1971, 1106).
! The asset pricing models rest upon the twin assumptions ofrational behaviour and perfect markets. In this context, aperfect market is one in which there are no transactionscosts and all individuals are price-takers (Ball, Brown,and Officer 1976, 2).
CAPM was developed as a model for a frictionless (costless)
market, but it is being applied to a market which is not
costless. Prior researchers have assumed a costless market
because they used the capital asset pricing model. However, if
no investor is paying any costs, that also means that no
brokerage firm or broker is receiving any revenue. Observe that
brokerage firms are in business to make a profit, which requires
that they receive revenue, and further observe that individual
brokers have self-selected into an occupation in which they are
paid on the transactions completed by their customers. Since
2Maos footnote at this point explicitly mentioned Lintner,Sharpe, Mossin, and Fama.
-
8/8/2019 AFAANZ 0677
24/74
24
brokers want to earn more money rather than less, they have an
incentive to cause their customers to engage in transactions.
IV. A PRICING MODEL THAT WORKS
Precisely because nobody is selling an "overpriced"
security to force it down to the CAPM price, we can conclude
that all those who wanted to sell have sold, and as much as they
wanted to sell. Because the market price is not even higher, as
would be the case if more people would have bought, we can
conclude that all those who wanted to buy have bought as much as
they wanted and could afford to buy. Thus, by definition, the
security price is a market price.
According to standard economic theory, when there is no
interference with the market--as in the case of a large,
centralized, liquid auction market with many buyers and sellers-
-the price of everything is set by supply and demand. Supply
and demand (possibly the most fundamental model in all of
economics) works in explaining not only prices of individual
securities, but also the price level of the entire market.
According to the model of supply and demand which is taught to
undergraduate business students as part of their required core
-
8/8/2019 AFAANZ 0677
25/74
25
business curriculum, this market price is set by the
intersection of a supply curve and a demand curve, as shown in
Figure 3.
------------------------------
Insert Figure 3 about here
------------------------------
Now let us return to Figure 1. This pictorial
representation of post-earnings-announcement drift for a
positive earnings surprise represents a portfolio. It can also
be defined as representing what happens to the average security
in the portfolio. However, the line representing the price
relative to the market is merely a transformation of a market
price, which can be analyzed in the following manner.
1. Let us take three points J, K, and L on the PEAD, as
shown in Figure 4.
------------------------------
Insert Figure 4 about here
------------------------------
-
8/8/2019 AFAANZ 0677
26/74
26
2. Let us show only the three points J, K, and L, as
shown in Figure 5.
------------------------------
Insert Figure 5 about here
------------------------------
3. Since these three points represent price points (even
if in the pictorial representation of the drift they
are transformations of price points), let us rename
the axes as shown in Figure 6. (Alternatively,
subtract out the effects of the market, and one
obtains the same Figure 6.)
------------------------------
Insert Figure 6 about here
------------------------------
4. Let us draw in the supply curve S 0 and demand curve D 0
which intersect to create price point J, as shown in
Figure 7.
-
8/8/2019 AFAANZ 0677
27/74
-
8/8/2019 AFAANZ 0677
28/74
28
one shift of the demand curve plus at least one shift of the
supply curve. (In order to simplify the analysis, possible
changes in slopes and shapes of the curves are ignored.)
By analogy, but without repeating the detailed analysis,
the post-earnings-announcement drift for a security with a
negative earnings surprise consists of two components, at least
one shift of the supply curve plus at least one shift of the
demand curve. Next I will explain why there are shifts in both
demand and supply.
Not All Investors Are Market Experts
The capital asset pricing model and the efficient market
hypothesis each assume that all investors are experts at
investing. This is false. Consider the following quotations.
! A national survey of investor knowledge concludes thatfewer than one in five U.S. investors have a basicunderstanding of financial terms and the ways varioussecurities work (staff reporter for The Wall Street Journal 1996, A6).
! Charles Schwab Corp., the giant pioneer of discountbrokers, is preparing to roil the financial-services worldwith a bold move that could make it the McDonald's offinancial advice.
-
8/8/2019 AFAANZ 0677
29/74
29
Departing from its no-frills roots, the San Francisco firmlate this year plans to launch what amounts to a fast-foodversion of money management....
Schwab says it has no choice but to cater to its customers,who now want a dab of hand-holding along with their low-cost services. "The nature of people coming to Schwab haschanged dramatically," says ... Schwab's president. "Weare going into the advice business because our customersare pulling us into it." (Schultz and O'Brian 1996, C1)
Many people rely on someone else, such as securities
brokers, to tell them what to buy or sell.
! Until recently, little was known in a formal way about thedecision processes of investors (American AccountingAssociation 1972, 409).
! ...on what basis and from what sources do private investorssecure information upon which to make decisions about theirinvestment or disinvestment in corporate stocks?
Stockbrokers were rated most important by 46.8 percent
(Parker 1981, 38). When I had my own certified public
accountant (CPA) firm, I met with a number of securities brokers
and discussed (among other things) their training, their
licensing, regulation, and their actual practices whereby they
get customers to buy securities. In other words, I conducted a
field study, and as Abdel-khalik and Ajinkya (1979, 45) state
-
8/8/2019 AFAANZ 0677
30/74
30
about field studies, "external validity and the practical
significance of the results are high."
Securities Brokers Sell Securities
The capital asset pricing model is deficient because it
implicitly assumes that securities brokers have no influence
over who buys or sells what, nor at what price. Brokers are
implicitly assumed to be mere order takers rather than
instigators of transactions. This is a fallacy, as the assumed
mathematically rational investors would all flock to discount
brokers for cheap execution of orders rather than pay high-
priced brokers. Brokers have self-selected into an occupation
in which they do not get paid unless they get clients, and those
clients engage in transactions. One of the responsibilities of
the brokerage firm's securities analysts is to provide a "story"
which the brokers use to "pitch" (i.e., sell) that security to a
client.
The economic analysis in this paper explains what happens.
This discussion of the process by which transactions occur
explains why it happens, and is based on a field study in which
I spoke with branch managers and brokers in multiple securities
-
8/8/2019 AFAANZ 0677
31/74
31
firms large, medium, and small. Some of the brokers had been so
successful over a multi-year period at selling securities that
they had been asked to help train new brokers. An assumption
inherent in this article is that brokers who have proven so
successful at selling securities over a multi-year period that
they have been asked to train other brokers, actually know how
to sell securities. Instead of using models which fail to
explain why PEAD occurs, this article uses the behavior of real
humans who are experts at how the market really works to explain
why at least some things happened the way they did, without
claiming to explain everything which has ever been observed.
This is an accepted approach in economics, and some certifying
examinations (e.g., the Uniform CPA Examination) directly test
knowledge of economics. Henderson (2005, A16) said of
Schelling, a winner of the Nobel Prize in Economics:
Mr. Schelling did it as a true social scientist, withspectacular results. ... His specialty was understandingthe behavior of real humans.... ...Mr. Schelling firsttold illustrative stories and then, using words, showed whythings happened the way they did.
-
8/8/2019 AFAANZ 0677
32/74
32
What Securities Brokers (i.e., Experts in Market Microstructure)and Published Sources Told Me About Why Market TransactionsOccur
Not all clients buy right away the stock being pushed by
the broker. Sometimes the broker must get back with them,
whether once or repeatedly. Many investors like predictability
in their investments. While academics refer to risk and to
stockholders as agents who specialize in risk-bearing, at least
some of the investors want safety, which is proxied by
predictability. 3
Consider the following items.
! Securities are not bought in this country; theyre sold.Somebodys broker calls them up and says buy XYZ stock, andthey just buy it, says John Fedders, a Washington, D.C.,lawyer who is the former director of the SECs Division ofEnforcement (France 1996, 83).
3This is why companies smooth earnings. Companies do not
sell their stocks; securities brokers sell their stocks.
Brokers would rather have an easy sell than a hard sell.
If company managements do not smooth their earnings to make
it easier for the brokers to sell their stock, then the
brokers can just sell some other company's stock.
-
8/8/2019 AFAANZ 0677
33/74
33
! What is a broker? Hes a salesman. (Wynter, quoting abrokerage firm founder who recruits securities brokers,1998, B1)
! Smith Barney Inc. is raising the bar for its brokers toearn bonuses. ...
The Smith Barney plan is part of a broader trend of higherhurdles for brokers. In the past several years, somebrokerage firms have increased the annual production quotasfor brokers to meet in order to keep their jobs or earnmore than bare-bones commissions.
The bulk of pay for most of the nation's about 93,000stockbrokers is tied directly to the trading commissionsthey generate (Siconolfi 1995, C22).
Why Post-Earnings Announcement Drift Occurs, Why CompaniesSmooth Earnings, and Why the Market Is Not Efficient
Brokers pitch the same stock to different clients. This
makes sense to brokers because they can learn one story,
supplied by one of the brokerage firms analysts, and repeat it
to different clients. This cuts down on the work brokers have
to do, but it means that the brokers are in danger of alienating
multiple customers if the story is wrong.
To help assure that the story is right, instead of issuing
reports shortly before new earnings information becomes
available, analysts delay issuing their reports until after the
-
8/8/2019 AFAANZ 0677
34/74
34
company has announced its earnings. After all, as brokers
pointed out to me, why should customers believe the story when,
right after you pitch a stock to them, the company announces
earnings which are down from what you said they would be? The
need for time to physically accomplish all the tasks involved in
issuing an analysts report accounts for the delay after the
earnings announcement before the drift begins. (Recall that
during the period of the early PEAD studies, not even the
professional market participants had instant electronic
communications, so they relied instead on slower means.)
Analysts do not issue a report on the same security every
quarter, so when a report is issued, it is something new instead
of something stale.
Once the brokers have the analysts report available,
different brokers react differently. Some brokers are always
rushing to get the latest stock to push. Others are more
comfortable continuing to push the stocks for which they already
know the story and how their customers are likely to react; they
are slower to start pushing the different stock mentioned in the
new analysts report. But by pushing the stock, the brokers are
-
8/8/2019 AFAANZ 0677
35/74
35
causing an outward shift in the demand curve for that particular
stock. (See Figures 7 and 8.)
Not all customers want to buy the stock right away. Some
want to wait and see what happens to the stock and its price.
Brokers pitch the reliability of earnings--a reason for
companies to smooth their earnings--yet are sometimes told that
maybe the customers will buy it later. (For example, the
substance of Ip [1997, C1 and C2] is that investors look for
certainty about their stocks earnings.)
What do brokers do now?
Brokers develop a sense of whether they can sell a
particular customer or not. After all, they are salesmen, and
they make their living by getting people to do something. It
makes no sense to most successful brokers to pitch something
else to a customer who will not buy what is being pitched today,
because (a) the broker will just confuse the customer, and (b)
if the customer did not buy the first stock, he is not likely to
buy the second stock. Consequently, the broker makes some note
-
8/8/2019 AFAANZ 0677
36/74
36
to contact the customer again around the time the company is
getting ready to release its next earnings report.
Some customers, if they indeed will buy, want to buy before
the next earnings announcement. Other customers want to
actually see that the company has indeed reported particular
earnings before they buy. In either case, the customers want to
see that the reported earnings were what the broker told them to
expect, which is another reason why market professionals like
smoothed earnings. Some customers want to wait for yet another
period before they buy, so they can be sure that the earnings
the broker tells them to expect actually are the reported
earnings. All this points to market professionals liking
smoothed earnings.
Verecchia (1981, 271) said that at an intuitive level,
there is a belief that observed abnormal volume around the
release of earnings reports must mean something, but he offered
no explanation of why the abnormal volume occurs. When brokers
all across a brokerage firm contact their wavering customers
around the time of the next earnings announcement, they reach
some customers who will buy, some who want to wait and see what
-
8/8/2019 AFAANZ 0677
37/74
37
the announced earnings are and then buy, some who either cannot
be reached or who are out of town for a short time, and some who
want to wait and talk it over with someone (typically a spouse)
before they buy the stock. This physical process, caused by a
change in the amount of broker sales activity in the stock
around the time of the next earnings announcement, accounts for
the observed nonrandom drift around the ( t + 1) earnings
announcement and later announcements. It has been observed that
a disproportionately large fraction of postannouncementdrift is concentrated in the few days preceding andincluding the next quarter's earnings announcement (Bernardand Thomas 1989, 29-30).
Abarbanell and Bernard (1992, 1190) found a "concentration of
the post-earnings-announcement abnormal returns around
subsequent earnings announcements . . ." above what drift would
be expected if the drift occurred randomly. This simply results
from brokers following up with customers, trying to sell the
same stocks they tried to sell before. Brokers are not
necessarily following up with every customer and concerning
every security, but only those customers who the broker thinks
can be sold particular securities. Further, because this
security is again coming to the attention of the broker, the
-
8/8/2019 AFAANZ 0677
38/74
-
8/8/2019 AFAANZ 0677
39/74
39
This first hypothesis is expected to be impossible to
satisfy unless this work contains the correct solution to what
causes post-earnings-announcement drift.
If A, then B
H2: On average, companies experiencing post-earnings-
announcement drift have earnings more extreme than expected. In
other words, on average, their standardized unexpected earnings
(SUEs) are positive.
This hypothesis number two is related to the way securities
brokers sell stocks to their customers.
If not A, then not B
H3: On average, once the SUEs of companies experiencing
post-earnings-announcement drift turn negative, the drift either
disappears or turns negative.
-
8/8/2019 AFAANZ 0677
40/74
40
This hypothesis number three is related to the fact that
once the earnings story is no longer there for the securities
brokers to use to get their customers to engage in transactions
on those stocks, the extra transactions caused by the brokers
efforts cease, and so do the effects on prices of those extra
transactions.
The Evidence Which Tests the Hypotheses
Bernard and Thomas (1989 and 1990) and Ball and Bartov
(1996) used the same data set. Ball and Bartov (1996, 325)
state that data from their studies were kindly supplied by
Bernard and Thomas. Therefore, the evidence all comes from the
same data set. Concerning hypothesis two, Bernard and Thomas
(1990, 310, Table 1) report that lagged standardized unexpected
earnings are positive in the first quarter for drift firms, and
remain positive until quarter four. Ball and Bartov (1996, 326)
state that their own Table 1 reports pooled regressions of SUE
on its four lagged values. Their signs follow the familiar
(+,+,+,-) pattern. Thus, research hypothesis two is supported.
Concerning hypothesis three, Bernard and Thomas (1990, 327,
Table 5) report that abnormal returns (ARs) using lagged
-
8/8/2019 AFAANZ 0677
41/74
41
quarterly earnings information are negative during the three-day
period [-2,0] at quarter 4the same time that SUE turned
negative. One well-known phenomenon from the original Ball and
Brown (1968) paper is that prices change even before earnings
are announced. Ball and Bartov (1996, 327, Table 2) also report
that cumulative abnormal return (CAR) for lagged quarterly
earnings information becomes statistically significantly
negative at quarter 4. Thus, research hypothesis three is
supported.
Because both research hypothesis two and research
hypothesis three are supported by reference to multiple
previously published works which all used the same data set,
research hypothesis one is also supported. Therefore, all three
of the research hypotheses are supported.
VI. IMPLICATIONS OF THE ANALYSIS
Why Smaller Firms Have Larger Drifts
Foster, Olsen, and Shevlin (1984, 598) reported that "the
smaller the firm size, the larger the absolute magnitude of CAR j
in the [+1, +60] period...." Bernard and Thomas (1989, 9)
-
8/8/2019 AFAANZ 0677
42/74
42
stated, "we replicate those results and demonstrate that they
persist over a longer time period."
PEAD is more pronounced for small firms than for large
firms because large firms have much more ongoing transaction
volume than do small firms. Consequently, when brokers start to
push a large firms stock, the additional volume generated is
not as big a percentage difference as it is for a small firms
stock. Brokers push the demand curve out more for a small firm
with low volume than for a big firm with high volume. More than
one broker has told me personally that, by himself, he can move
the price of all but the biggest stocks if he really believes in
it and starts selling it to all his customers.
Why the Market is not Efficient
The market is not efficient because not everybody receives
the same information at the same time. Companies guide
analysts toward reasonable earnings expectations. The best big
customers of the analysts firm get the news first. Then word
gets out to the retail brokers, who in turn call their own best
customers first. The information is still being used months
later to pitch the stock to customers who wavered earlier.
-
8/8/2019 AFAANZ 0677
43/74
43
Consider this real-life example. I was working in South
Carolina one day when I heard from a reliable source that not a
single truckload of freight had moved in Pittsburgh,
Pennsylvania the previous day. Some small shippers had been
crying for weeks for someone to come pick up a load, but had not
been able to get anyone to come because the truckers were so
busy with bigger shippers. Now, even those single loads had
been picked up, and nobody had any freight to be moved.
Obviously this meant that economic activity in Pittsburgh had
slowed from what it had been previously, so it had predictive
value for earnings of companies whose activities were
concentrated in Pittsburgh. Anyone who had access to that
information about lack of freight could conduct transactions
during quarter t based on it. However, other people would never
receive the information and so would be forced to wait for
earnings guidance about quarter t ahead of the earnings
announcement; such guidance could occur during or after the end
of quarter t . Still others would not be informed of the
information or its effects until after the earnings
announcement--maybe months later--in quarter ( t + 1) or later.
Yet others would be presented with a story after earnings for
-
8/8/2019 AFAANZ 0677
44/74
44
quarter t had already been announced in quarter ( t + 1). Then,
some investors would not buy until sometime around the
announcement of earnings for quarter ( t + 1), to see whether the
broker's story is reliable. Thus, by definition, the market is
not efficient because people receive and act on the same
information at different times . Researchers have failed to
realize this because they have learned so well the concept of
the efficient market hypothesis (EMH). Concerning the EMH, a
quick reaction has been mistaken for a complete reaction.
This section has explained positive post-earnings-
announcement drift. Negative post-earnings-announcement drift
occurs when brokers try to get customers to sell securities
which report earnings lower than the expectation. For
unsophisticated, unknowledgeable investors, the expectation is
indeed naive: how this report compares to the previous one
(that of last year or last quarter). The broker's incentive is
to get the customer to sell the security and then use that money
to buy something else. This pitch that the customer should sell
a bad stock and buy something else results in two commissions,
not just one.
-
8/8/2019 AFAANZ 0677
45/74
45
Stock Splits la Fama, Fisher, Jensen, and Roll (1969)
Numerous individual investors believe that buying a stockthat splits is profitable, but nobody can really explainwhy that should be. ... Microsoft splits in half. Whatchanged? To an economist its absolutely bogus. Samecompany, double the number of shares at half the price.Yet it seems to work.... (Ip 1998, B1)
Ball, Brown, and Finn (1977) replicated and extended Fama,
Fisher, Jensen, and Roll's (1969) study of stock splits.
Companies which split their stocks have no intrinsic difference
in value, yet Fama, Fisher, Jensen, and Roll (1969) found a
price increase in the United States market, and Ball, Brown, and
Finn (1977) found a price increase in the Australian market.
The price increase has not been accounted for, although
Ball, Brown, and Finn (1977, 106-107) observed that
share splits sometimes are regarded as means of alteringmarket prices of shares to bring them into a more "popular"price level and so to "broaden" the market for the share,which presumably implies a higher price.
Nobody has yet explained why this occurs, but it can be
easily explained by using supply and demand. There is a market
for shares of stock in the company. After the split, brokers
can sell the stock more easily to unknowledgeable investors.
-
8/8/2019 AFAANZ 0677
46/74
46
Because unknowledgeable investors do not understand stock
splits, at least some brokers tell customers that all the
customers need to know is that before, they would have had to
pay some price (say, sixty dollars per share), but now they can
buy the same number of shares in the same company for less
money. Second, whether or not the broker takes that approach
with a customer, it is easier for a broker (read salesman) to
sell a stock for thirty dollars than it is to sell a stock for
sixty dollars because the investor has less total money at risk
at the lower price. The result is that brokers actions to sell
post-split shares to customers move the post-split shares
demand curve outward, thereby increasing the price.
-
8/8/2019 AFAANZ 0677
47/74
47
REFERENCES
Abarbanell, J. S. and V. L. Bernard. 1992. Tests of analysts'overreaction/underreaction to earnings information as anexplanation for anomalous stock price behavior. TheJournal of Finance 45:3 (July): 1181-1207.
Abdel-khalik, A. R. and B. P. Ajninkya. 1979. Empirical
Research in Accounting: A Methodological Viewpoint .Sarasota, FL: American Accounting Association. This waspublished as Accounting Education Series, Volume No. 4.
Affleck-Graves, J. and R. R. Mendenhall. 1992. The relationbetween the Value Line enigma and post-earnings-announcement drift. Journal of Financial Economics 31:1(February): 75-96.
American Accounting Association. 1972. The behavior ofsecurity prices and its implications for accountingresearch (methods). The Accounting Review , CommitteeReports, Supplement to Volume 47: 407-437.
Arrow, K. J. 1982. Risk perception in psychology andeconomics. Economic Inquiry 20 (January): 1-9. Cited inSummers, L. H. 1986. Does the stock market rationallyreflect fundamental values? The Journal of Finance 61:3(July): 591-601.
Asthana, S. 2003. Impact of information technology on post-earnings announcement drift. Journal of InformationSystems 17:1 (Spring): 1-17.
Ball, R. 1978. Anomalies in relationships between securities'yields and yield-surrogates. Journal of FinancialEconomics 6: 103-126.
-
8/8/2019 AFAANZ 0677
48/74
48
Ball, R. 1992. The earnings-price anomaly. Journal of Accounting and Economics 15:2/3 (June/September): 319-345.
Ball, R. and E. Bartov. 1996. How nave is the stock marketsuse of earnings information? Journal of Accounting and Economics 21:3: 319-337.
Ball, R. and P. Brown. 1968. An empirical evaluation ofaccounting income numbers. Journal of Accounting Research 6:2 (Autumn): 159-178.
Ball, R. and R. Watts. 1972. Some time series properties ofaccounting income. Journal of Finance (June): 663-681.
Ball, R., P. Brown, and F. J. Finn. 1977. Sharecapitalization changes, information, and the Australianequity market. Australian Journal of Management 2:2(October): 105-125.
Ball, R., P. Brown, and R. R. Officer. 1976. Assetpricing in the Australian industrial equity market.Australian Journal of Management 1:1 (April): 1-32.
Ball, R., S. P. Kothari, and R. L. Watts. 1993. Economicdeterminants of the relation between earnings changes andstock returns. The Accounting Review 68:3 (July): 622-638.
Bartov, E. 1992. Patterns in unexpected earnings as anexplanation for post-announcement drift. The Accounting Review 67:3 (July): 610-622.
Basu, S. 1977. Investment performance of common stocks inrelation to their price-earnings ratios: a test of theefficient market hypothesis. The Journal of Finance 32:3(June): 663-682.
Basu, S. 1978. The effect of earnings yield on assessments of
the association between annual accounting income numbersand security prices. The Accounting Review 53:3 (July):599-625. This paper won the American AccountingAssociation 1977 competitive manuscript award.
-
8/8/2019 AFAANZ 0677
49/74
49
Beaver, W. H. 1968. Information content of annual earningsannouncements. Empirical Research in Accounting: Selected Studies, 1968 , Supplement to Journal of Accounting Research , 67-92.
Beaver, W. H. 1974. Implications of security price researchfor accounting: a reply to Bierman. The Accounting Review 49:3 (July): 563-571.
Beaver, W. 1981a in source. Financial reporting: anaccounting revolution . Englewood Cliffs, NJ: Prentice-Hall. Cited in Hew, D., L. Skerratt, N. Strong, and M.Walker. 1996. Post-earnings-announcement drift: somepreliminary evidence for the UK. Accounting and BusinessResearch 26:4 (Autumn): 283-293.
Beaver, W. H. and W. R. Landsman. 1981. Note on the behaviorof residual security returns for winner and loserportfolios. Journal of Accounting and Economics 3:3(December): 233-241.
Beaver, W., P. Kettler, and M. Scholes. 1970. The associationbetween market determined and accounting determined riskmeasures. The Accounting Review 45:4 (October): 654-682.
Bernard, V. L. 1993. Stock price reactions to earningsannouncements: a summary of recent anomalous evidence and
possible explanations. Advances in Behavioral Finance ,Richard H. Thaler, Editor, 303-340.
Bernard, V. L. and J. K. Thomas. 1989. Post-earnings-announcement drift: delayed price response or riskpremium? Journal of Accounting Research 27 (Supplement):1-36.
Bernard, V. L. and J. K. Thomas. 1990. Evidence that stockprices do not fully reflect the implications of currentearnings for future earnings. Journal of Accounting and
Economics 13: 305-340.
Bernard, V. L., J. K. Thomas, and J. Wahlen. 1997. Accounting-based stock price anomalies: separating marketinefficiencies from risk. Contemporary Accounting Research (Summer): 89-136.
-
8/8/2019 AFAANZ 0677
50/74
-
8/8/2019 AFAANZ 0677
51/74
51
Brown, S. L. 1978. Earnings changes, stock prices, and marketefficiency. The Journal of Finance 33:1 (March): 17-28.
Brown, S. L. 1979. Earnings announcements and auto-correlation: an empirical test. The Journal of FinancialResearch 2:2 (Fall): 171-183.
Calegari, M. and N. L. Fargher. 1997. Evidence that prices donot fully reflect the implications of current earnings forfuture earnings: an experimental markets approach.Contemporary Accounting Research 14:3 (Fall): 398-433.
Chambers, A. E. and S. H. Penman. 1984. Timeliness ofreporting and the stock price reaction to earningsannouncements. Journal of Accounting Research 22:1: 21-47.
Chan, L., N. Jegadeesh and J. Lakonishok. 1996. Momentumstrategies. Journal of Finance 51: 1681-1713. Cited inFama, E. F. 1998. Market efficiency, long-term returns,and behavioral finance. Journal of Financial Economics 49:3 (September): 283-306.
Chari, V. V., R. Jagannathan, and A. R. Ofer. 1988.Seasonalities in security returns: the case of earningsannouncements. Journal of Financial Economics 21:1 (May):101-121.
Collins, D. W. and P. Hribar. 2000. Earnings-based andaccrual-based market anomalies: one effect or two?Journal of Accounting and Economics 29: 101-123.
Copeland, T. E. and J. F. Weston. 1979. Financial Theory and Corporate Policy . Reading, MA: Addison-Wesley PublishingCompany.
Deakin, E. B., G. R. Norwood, and C. H. Smith. 1974. The effectof published earnings information on Tokyo Stock Exchangetrading. The International Journal of Accounting Education
and Research 10:1 (Fall): 123-136.Easton, P. D. 1985. Accounting earnings and security
valuation: empirical evidence of the fundamental links.Journal of Accounting Research 23 (Supplement): 54-77.
-
8/8/2019 AFAANZ 0677
52/74
-
8/8/2019 AFAANZ 0677
53/74
53
Freeman, R. and S. Tse. 1989. The multiperiod informationcontent of accounting earnings: confirmations andcontradictions of previous earnings reports. Journal of Accounting Research 27 (Supplement): 49-79.
Freeman, R. and S. Tse. 1989. The multi-period informationcontent of earnings announcements: rational delayedreactions to earnings news. Journal of Accounting Research 27 (Supplement): 49-79. This is the title under which thearticle was cited in Bernard and Thomas. 1990. Evidencethat stock prices do not fully reflect the implications ofcurrent earnings for future earnings. Journal of Accounting and Economics 13: 305-340. On page 49 of thearticle as published (see entry immediately preceding), theauthors definitely state that this is the title used forearlier versions, including a conference draft, of thepublished paper.
Henderson, David R. 2005. The great game. The Wall StreetJournal (October 11): A16.
Hew, D., L. Skerratt, N. Strong, and M. Walker. 1996. Post-earnings-announcement drift: some preliminary evidence forthe UK. Accounting and Business Research 26:4 (Autumn):283-293.
Ip, G. 1997. Investors now search for predictable earnings.The Wall Street Journal (March 31): C1 and C2.
Ip, G. 1998. Bull market in stocks is trampling someinvestment rules of the past. The Wall Street Journal (April 6): B1 and B2.
Jacob, J., T. Lys, and J. Sabino. 2000. Autocorrelationstructure of forecast errors from time-series models:Alternative assessments of the causes of post-earningsannouncement drift. Journal of Accounting and Economics 28: 329-358.
-
8/8/2019 AFAANZ 0677
54/74
54
Jensen, M. C. et al. 1978. Symposium on some anomalousevidence regarding market efficiency. Journal of FinancialEconomics 6 (June/September): 93-330. Cited in Summers,L. H. 1986. Does the stock market rationally reflectfundamental values? The Journal of Finance 61:3 (July):591-601.
Jones, C. P. 1973. Earnings trends and investment selection.Financial Analysts Journal 29:2 (March-April): 79-83.
Jones, C. P. and R. H. Litzenberger. 1970. Quarterly earningsreports and intermediate stock price trends. The Journalof Finance 25:1 (March): 143-148.
Jones, C. P., R. J. Rendleman, Jr., and H. A. Latan. 1984.Stock returns and SUEs during the 1970's. The Journal of Portfolio Management 10:2 (Winter): 18-22.
Jones, C. P., R. J. Rendleman, Jr., and H. A. Latan. 1985.Earnings announcements: Pre- and -post responses. TheJournal of Portfolio Management 11:3 (Spring): 28-32.
Joy, O. M. and C. P. Jones. 1979. Earnings reports and marketefficiencies: an analysis of the contrary evidence. TheJournal of Financial Research 2:1 (Spring): 51-63.
Joy, O. M., R. H. Litzenberger, and R. W. McEnally. 1977. The
adjustment of stock prices to announcements ofunanticipated changes in quarterly earnings. Journal of Accounting Research 15:2 (Autumn): 207-225.
Kim, D. and M. Kim. 2003. A multifactor explanation of post-earnings announcement drift. Journal of Financial and Quantitative Analysis 38:2 (June): 383-398.
Kormendi, R. and R. Lipe. 1987. Earnings innovations, earningspersistence, and stock returns. Journal of Business 60:3(July): 323-345.
Latan, H. A. and C. P. Jones. 1977. Standardized unexpectedearnings--a progress report. The Journal of Finance 32:5(December): 1457-1465.
-
8/8/2019 AFAANZ 0677
55/74
55
Latan, H. A. and C. P. Jones. 1979. Standardized unexpectedearnings--1971-77. The Journal of Finance 34:3 (June):717-724.
Latan, H. A., C. P. Jones, and R. D. Rieke. 1974. Quarterlyearnings reports and subsequent holding period returns.Journal of Business Research 2:2 (April): 119-132.
Latan, H. A., O. M. Joy, and C. P. Jones. 1970. Quarterlydata, sort-rank routines, and security evaluation. TheJournal of Business 43:4 (October): 427-438.
Latan, H. A., D. L. Tuttle, and C. P. Jones. 1969. E/P ratiosv. changes in earnings in forecasting future price changes.Financial Analysts Journal (January-February): 117-120,and continued on 123.
Lev, B. 1989. On the usefulness of earnings and earningsresearch: lessons and directions from two decades ofempirical research. Journal of Accounting Research 27(Supplement): 153-192.
Lev, B. and J. A. Ohlson. 1982. Market-based empiricalresearch in accounting: a review, interpretation, andextension. Journal of Accounting Research 20 (Supplement):249-322.
Litzenberger, R. H., O. M. Joy, and C. P. Jones. 1971. Ordinalpredictions and the selection of common stocks. Journal of Financial and Quantitative Analysis 6:4 (September): 1059-1068.
Liu, W., N. Strong, and X. Xu. 2003. Post-earnings-announcement drift in the UK. European FinancialManagement 9:1: 89-116.
McWilliams, J. D. 1966. Prices, earnings, and P-E ratios.Financial Analysts Journal 22:3 (May-June): 137-142.
Mao, J. C. T. 1971. Security pricing in an imperfect capitalmarket. Journal of Financial and Quantitative Analysis 6:4(September): 1105-1116.
-
8/8/2019 AFAANZ 0677
56/74
56
Marais, M. L. 1989. Discussion of Post-earnings-announcementdrift: delayed price response or risk premium? Journal of Accounting Research 27 (Supplement): 37-48.
Mendenhall, R. R. 2002. How nave is the market's use of firm-specific earnings information? Journal of Accounting Research 40:3 (June): 841-863.
Nerlove, M. 1968. Factors affecting differences among rates ofreturn on investments in individual common stocks. Review of Economics and Statistics 50: 312-331.
Nichols, D. C. and J. M. Wahlen. 2004. How do earnings numbersrelate to stock returns? A review of classic accountingresearch with updated evidence. Accounting Horizons Volume18, Number 4 (December): 263-286.
Nicholson, S. F. 1968. Price ratios in relation to investmentresults. Financial Analysts Journal 24 (January-February):105-109.
Parker, L. D. 1981. Corporate annual reports: a failure tocommunicate. International Journal of Accounting Educationand Research 16:2 (Spring): 35-48.
Qualls, H. 1972. Buy high or low? The investment experience.Australian Stock Exchange Journal 1 (December): 33 and 36-
37.
Reinganum, M. R. 1981. Misspecification of capital assetpricing: empirical anomalies based on earnings yields andmarket values. Journal of Financial Economics 9 (March):19-46.
Rendleman, Jr., R. J., C. P. Jones, and H. A. Latan. 1982.Empirical anomalies based on unexpected earnings and theimportance of risk adjustments. Journal of FinancialEconomics 10: 269-287.
Rendleman, Jr., R. J., C. P. Jones, and H. A. Latan. 1987.Further insight into the standardized unexpected earningsanomaly: size and serial correlation effects. TheFinancial Review . 22:1 (February): 131-144.
-
8/8/2019 AFAANZ 0677
57/74
57
Riahi-Belkaoui, A. 2002. Level of multinationality as anexplanation for post-announcement drift. The InternationalJournal of Accounting 37: 413-419.
Rosenberg, B., K. Reid, and R. Lanstein. 1985. Persuasiveevidence of market inefficiency. The Journal of PortfolioManagement 11:3 (Spring): 9-16.
Schultz, E. E. and B. O'Brian. 1996. Schwab to offercustomized investment advice. The Wall Street Journal (May30): C1 and C25.
Siconolfi, M. 1995. Smith Barney raises the bonus hurdlesconfronting brokers. The Wall Street Journal (December19): C22.
Soffer, L. C. and T. Lys. 1999. Post-earnings announcementdrift and the dissemination of predictable information.Contemporary Accounting Research 16:2 (Summer): 305-331.
Staff reporter for The Wall Street Journal . 1996. Surveyconcludes investors likely to wind up suckers. The WallStreet Journal (May 15): A6.
Summers, L. H. 1986. Does the stock market rationally reflectfundamental values? The Journal of Finance 61:3 (July):591-601.
Tversky, A. and D. Kahneman. 1981. The framing of decisionsand the psychology of choice. Science 211 (January): 453-458. Cited in Summers, L. H. 1986. Does the stock marketrationally reflect fundamental values? The Journal of Finance 61:3 (July): 591-601.
van Huffel, G., P. Joos and H. Ooghe. 1996. Semi-annualearnings announcements and market reaction: some recentfindings for a small capital market. The EuropeanAccounting Review 5:4: 693-713.
Verecchia, R. E. 1981. On the relationship between volumereactions and consensus of investors: implications forinterpreting tests of information content. Journal of Accounting Research 19:1 (Spring): 271-283.
-
8/8/2019 AFAANZ 0677
58/74
58
Watts, R. L. 1978. Systematic "abnormal" returns afterquarterly earnings announcements. Journal of FinancialEconomics 6 (June-September): 127-150.
Wynter, L. E. 1998. Rare black firm on Wall Street learns howto make its way. The Wall Street Journal (March 4): B1.
-
8/8/2019 AFAANZ 0677
59/74
59
Appendix
The Importance of the Drift
Several authors have given reasons why the drift is
important.
! Explanation of the anomaly is of interest for threereasons. First, it could resolve the conflicting resultsamong studies of earnings announcements. Second,explanation of the anomaly could provide evidence on theperformance of the two-parameter model 4 as a vehicle fordescribing the determination of securities' expectedreturns or on the appropriateness of the market portfolioused as a proxy for aggregate wealth. Third, it suggestscaution in designing experiments involving earnings ordividend yields (i.e., ratios of earnings and dividends pershare to share price) as independent variables and excessreturn as the dependent variable, because these experimentsare sensitive to model misspecification and can be expectedto produce anomalous evidence (Ball 1978, 104-105).
! Whatever is its intrinsic importance, investigation of thepost-earnings-announcement drift anomaly gains interestfrom the fact that it combines in one problem severalvexing conceptual, institutional, and econometricquestions. These include such issues as computing"correct" abnormal returns, the size effect, thedifferences between measured returns and "true" returns(based on underlying equilibrium prices), anomaliesresulting from different compounding methods, alternativeearnings expectation models, and the specification and
interpretation of tests of market efficiency (Marais 1989,46-47).
-
8/8/2019 AFAANZ 0677
60/74
-
8/8/2019 AFAANZ 0677
61/74
61
challenges to the EMH (see Ball, 1992 and Lev and Ohlson,1982). (Hew, Skerratt, Strong, and Walker 1996, 283).
! Post-earnings announcement drift, the phenomenon wherestock prices continue to drift in the direction of theinitial price response to an earnings announcement, is oneof the most prominent and perplexing market anomaliesdocumented in the accounting literature (Collins and Hribar2000, 102).
The reason why post-earnings announcement drift is soanomalous is the lack of a coherent theory to fit theempirical facts (Collins and Hribar 2000, 120).
! ... post-earnings-announcement drift ... remains one of themost puzzling anomalies in accounting- and financial-economics-based tests of capital market efficiency withrespect to earnings information. Ongoing research studiescontinue the search for explanations, but thus far thepost-earnings-announcement drift anomaly has defied allattempts at rational explanation (Nichols and Wahlen 2004,284).
Since the drift anomaly is so important, it has attracted a
great deal of research effort. Table 1 contains a partial list
of the published studies which address post-earnings-
announcement drift in some fashion, although not uniformly
calling it by that terminology.
Ball (1992, 319) stated:
Based on the number of published papers which deal with thephenomenon, and on the implications of the phenomenon, it
-
8/8/2019 AFAANZ 0677
62/74
62
appears that gaining an understanding of the cause of thephenomenon is worthwhile.
-
8/8/2019 AFAANZ 0677
63/74
63
Table 1
List of Some Studies Which AddressPost-Earnings-Announcement Drift (PEAD)
Ball and Brown (1968)Beaver (1968)Breen (1968)Nerlove (1968)Nicholson (1968)Latan, Tuttle, and Jones (1969)Beaver, Kettler, and Scholes (1970)Jones and Litzenberger (1970)Latan, Joy, and Jones (1970)Litzenberger, Joy, and Jones (1971)Ball and Watts (1972)Brown and Kennelly (1972)Qualls (1972)Black (1973)Jones (1973)Black and Scholes (1974)Deakin, Norwood, and Smith (1974)Latan, Jones, and Rieke (1974)Foster (1975)Basu (1977)Brown, Finn, and Hancock (1977)
Foster (1977)Joy, Litzenberger, and McEnally (1977)Latan and Jones (1977)Ball (1978)Basu (1978)Brown (1978)Watts (1978)Brown (1979)Joy and Jones (1979)Latan and Jones (1979)Beaver and Landsman (1981)
Bidwell and Riddle (1981)Reinganum (1981)Lev and Ohlson (1982)Rendleman, Jones, and Latan (1982)Chambers and Penman (1984)Foster, Olsen and Shevlin (1984)
-
8/8/2019 AFAANZ 0677
64/74
64
Jones, Rendleman, and Latan (1984)Easton (1985)Jones, Rendleman, and Latan (1985)Rosenberg, Reid, and Lanstein (1985)Kormendi and Lipe (1987)Rendleman, Jones, and Latan (1987)Chari, Jagannathan, and Ofer (1988)Bernard and Thomas (1989)Easton and Zmijewski (1989)Freeman and Tse (1989)Lev (1989)Bernard and Thomas (1990)Abarbanell and Bernard (1992)Affleck-Graves and Mendenhall (1992)Ball (1992)Bartov (1992)Ball, Kothari, and Watts (1993)Bhushan (1994)Booth, Kallunki, and Martikainen (1996)Hew, Skerratt, Strong, and Walker (1996)van Huffel, Joos, and Ooghe (1996)Bernard, Thomas, and Wahlen (1997)Brown (1997)Calegari and Fargher (1997)Soffer and Lys (1999)Mendenhall (2002)Riahi-Belkaoui (2002)
Asthana (2003)Liu, Strong, and Xu (2003)
-
8/8/2019 AFAANZ 0677
65/74
-
8/8/2019 AFAANZ 0677
66/74
66
Figure 1
A Pictorial Representation of Post-Earnings-Announcement Drift
t=0 t=1
Pricerelativeto themarket
Time
-
8/8/2019 AFAANZ 0677
67/74
67
Figure 2Pictorial Proof that the CAPM Price is Wrong
PriceRelativeto theMarket
t = 0Point A Point B Point C
Timet > 0
-
8/8/2019 AFAANZ 0677
68/74
68
Figure 3A Model That Works
Price
Quantity
MarketPrice
S
D
-
8/8/2019 AFAANZ 0677
69/74
-
8/8/2019 AFAANZ 0677
70/74
70
Figure 5
Isolation of Points Selected in Figure 4
t=0 t=1
Pricerelativeto themarket
Time
JK
L
-
8/8/2019 AFAANZ 0677
71/74
71
Figure 6
Undoing Transformation of Points in Figure 5
t=0 t=1
Price
Quantity
JK
L
-
8/8/2019 AFAANZ 0677
72/74
72
Figure 7
Price Determination of First Point on PEAD
t=0 t=1
Price
Quantity
J
KL
D0
S0
-
8/8/2019 AFAANZ 0677
73/74
73
Figure 8
Price Determination of Second Point on PEAD
t=0 t=1
Price
Quantity
J
KL
D0
S0
D1
-
8/8/2019 AFAANZ 0677
74/74
Figure 9
Price Determination of Third Point on PEAD
t=0 t=1
Price
Quantity
J
KL
D0
S0
D1
S1
D2