,

124 CHAPTER 5 THE EVlDENCE

WU.Oo. O. P. "Tbo RolaúveiDformatioo ContoDt of Accnla1 aod Cub Flowa: Combiuod Bv-
ideoce at tbe Bamlnp Announcement anel Annual R.elease Date." Studles on Alui'TUI-
tlw M~.,ru o.f AccOWIIing. Incomc. Suppl.ement to tbe JoMmal of Accollllting Re-
search (1986), 165-200.
- - . "Tbe Incrementai Infonnation Contcnt of the Accrual and Funds Components of
6
Bamings After Controlling for Eamings." Accounling Review (April1987), 293-322.

Market Efficiency

Market efficiency is a central feature of the capital IIlllikets and deals with the rela-
tion between security prices and information. In ao efficient market, prices "fuuly
reflect'' information. More precisely, a securities m.arket is defined to be efficieot
with respect to a particular infonnation system if and only if the security prices act
as ü everyone observes the signals from tbat information system. Market efficiency
deals with bow the capital markets process infonnation in general and financiai r;e-
porting information specifically.
Marlcet efficíency is important to financiai reporting for several reasons. h in-
fluences whether public disclosure is expected to have significant effects on seco-
rity prices. lt also affects whether disclosure of ao item in the footnotes va:sus
recognition of the item in the body of the financiai statements is expected to IDIIke a
difference in security prices. It influences how important financiai reporting dala
are relative to the total mix of information.
In ao inefficient capital market, financiai statement data are not necessarily
reflected in prices, and hence ao individual investor is not necessarily a "price
taker" with respect to that information. An inefficient capital market affords oppor-
tunities for investors to earn abnormal retums from investment strategies keycd on
financiai reporting data tbat are not fully reflected in prices.
Each of the major constituencies has a vital interest in the prices of a firm's
securlties and the effect of infonnalion on prices. The interest in security prices
arises because of the economic consequences associated with security prices.. For
example, changes in prices of a security alter the market value of the investor's
wealth. The price of the security also affects the terms on which the firm obtaiDS ad-
ditional financing. This in tum can affect management's perceived cost of capital
126 CHAPTER 6 MARKET EFF1CJENCY CHA.PT'ER 6 MARKET EFFICJENCY 127

and alter the nature of the projects undertaken. At an eoonomy-wide levei, capital Market Efficiency and Universal Access to lnformation
fonnatioo and resource allocatioo can be affected. Although prices may be of im-
mediate interest, tbe ultimate coocem is with the atteodant consequences. The eco- Marlcet efficieocy is defined in terms of universal access to the iofonnation system
nomic consequences of marlret efficiency are essentially the same as the coose- of interest. Specifically,
quences of financiai reporting discussed in Chapter 2.
The purpose of this chapter is not to take a position in favor of or against mar- ibe m.arlcet is efficient with respect to some specified infOIIIlatioo system, if and only
if security prices act as if everyone observes the information system.2
ket efficieocy with respect to a particular ioformatioo system. Nor is it to provide a
detailed, in-depth technical evaluation of the research. The summary of the research
will be brief and nontechoical. Instead, this cbapter focuses on the origins, defini- If prices have this property, they are said to fully reflect that ioformation. This defi-
tioos, various forms, evideoce, research design theories, aod implicatioos of market nition can be illustrated by ao eveot that is the subject of several empirical studies
efficieocy and its importance to the financiai reporting constitueocies. of market efficiency. The eveot is a firm's annouocement that it switched from an
accelerated method of depreciation to straight-line depreciation in reporting to
shareholders in the annual report. However, both before and after tbe change the
firms used ao accelerated depreciation method for tax purposes.
6-l ORIGIN OF THE MARKET EFFlCIENCY CONCEPT
In this example, marlcet efficiency means that prices act as if every investor
The issue of market efficieocy with respect to financiai statement ioformation origi- knows that the firms changed depreciation methods, knows what those methods are
oated in the practice of security analysis, which is defined as the process of findiog and what impact they have on reported earnings, knows that there was no change in
mispriced securities. In this context, market inefficiency was expressed as a depar- depreciation method for tax purposes, and knows the potential implications for
ture of a security's price from its "intrinsic value." Using financiai statement data to managements' motivation to make such a change. Obviously, such universal knowl-
find mispriced securities is exemplified by Graham and Dodd (1934), who fonnal- edge does not literally exist. However, the definition of market efficiency states
ized the art of security analysis as it is practiced by the professional investmeot only that prices act as if such a condition holds. In other words, the prices that pre-
community. Much of the professional invêstmeot industry offers services to clients vail with limiled knowledge of such information among investors are the same as
based oo the cooteotioo that they can successfully find mispriced securities. prices that would prevail if everyone literally knew sucb iofonnation. This may ex-
The concept o f the inlrinsic value o f a security is an intriguing aspectof security plain in part why maoy find the concept of market efficiency difficult to accept.
analysis. A security, such as a common stock, possesses value because it is perceived Much empirical research in market efficiency focuses on a major implication
to possess attributes desired by individuais. Commoo stocks are thought to possess of market efficiency known as the "fair garoe" property.3 If the market is efficie~t
value because they represeot claims to future, uncertain cash flows in the form of fu- with respect to some ioformation, the investor is playing a fair game with respect to
ture divideods. In elemeotary discussions of the demand for a commodity, prices are . that iofonnation. Uoder the fair game property of market efficiency, the abnormal
viewed as retlecting individuais' wealth, tastes, and beliefs. These factors are typically expecled returns from trading strategies based oo that information are zero. The in-
viewed as being personal or subjective and different across individuais. lndeed, for ability to eam abnormal returns is a particularly important implicatioo of marlcet ef-
most commodities, the intluence of endowments, preferences, and beliefs oo prices ficieocy to the professional investment community.•
and the subjective nature of the "value" of a commodity are readily acknowledged. The fair game approacb requires that abnormal returns must be defined cela-
Yet security analysis has introduced the ootioo o f the intrinsic v alue o f a security. The tive to some benchmark of normal expected returns. In many early empirical stud-
use of the term connotes an objective concept, independent of subjective influeoces. 1 ies, the retums on a market portfolio were used as the bencbmark against which to
compare the retum performance of portfolios arising from a trading strategy based

6-2 DEFINJTION OF MARKET EFF1CIENCY

2
A mote formal statement of this definition is: 1be lllllrket ia efficient with reapect to some infonnation
Subsequent notions of market efficiency did not explicitly rely on concepts of in- system if and ooly if the pricet of tbe lleCIIrities .-e tbe wnc u they woold be in m OChetwise idenlical
trinsic value. For example, Fama (1970) states that a securities market is efficient if economy (with the aame COIIfiluntiOD of p r e f - md enciowme1u md ~ ayatema) exc:ept
security prices "fully retlect'' the infonnation available. As Fama points out, the tha! evuy individuallw acceu to the information system as weU. See Beawr (1981) foca furtta dis-
cusalon of thia definitioo.
termsfully rejlect and biformation available are ambiguous. 3An analo&Y is a "faiz coin" in which the probabillty ofa head la .50. If the coin la "falr," the probabllity
remainl .50, repnfleu of lhe frequeacy md eequeDCe of beada aad talb of put U...
4
Retum la defiDed u tbe casb divideDd plua tbe c:hanp iD price owr the deflned pmod divided by the
1
For example., Gralwn and Dodd (1934, p. 17) defined intrin.si.c value u "'n aenenl terml, it ia under- price at the bepDina of the period. Retum c:an be viewed u a penllllàCe c:hanp iD price adjlllted foc
stood to be that value which is justified by the facts." dividend. Hence, the tenns nt1U11 aad pria change will be used interchanaeably tbrouJbout.
,.

128 CHAPTER 6 MARKET EF'FICIENCY CHAPTER 6 MARKET EFFlCIENCY 129

oo informatioo. Howevcz-, some felt tbat such a benchmarlr; might be inappropriate plies wealc form efficiency. However, the implicatioo does not worlc in the revcne
if tbe atrateiY teodcd to tc~ult in portfolios whoae "riak" was differcot from that of order.
the uwket portfolio. Portfolio theory and the capital asset pricing model permitted
the specificatioo of a rislc-adjusted retum benchmark and was perceived to be a con-
tribution to empirically testing marlret efficiency. However, tests of market effi- 6~4 EVlDENCE REGARDING MARKET EFFICIENCY
cieocy are a joint test of mark:et efficiency and the assumptions made about how to
1bis cbapter does not cootain an in~th discussioo of the evidence regarding mar-
measure the appropriate bencbmark, which meaos the issue is a particularly difficult
ooe to test empirically. ket efficiency for several reasons. (I) Some of this research is summarized in Chap-
ter 5. (2) Excelleot, detailed reviews of this researcb are available in Dyck:man aod
Morse (1986) and Fama (1991). (3) The nature of the evidence and the strength of
the findings either supporting or rejecting market efficiency are continually chaog-
6-3 FORMS OF MARKET EFF1CIENCY ing and are the subject of curreot controversy and debate. However, the coocept.
A discussion of market efficiency must specify the information systems for whicb implications, and theory of market efficiency are lik:ely to be of cootinuing interest
the market efficiency coodition is being defined. This is criticai because the market and concern. This cbapter focuses on these aspects o f mark.et efficiency. This sec-
may be efficient with respect to some information systems but not others. In gen- tion briefly summarizes a sample of the major studies that provided evidence on
eral, statements such as "the market is efficient" or "the market is not efficieot" are market efficiency.
ambiguous and incomplete. Empirical research characteriz.ed as efficient market research bas been the
Fama (1970) delineated three major forms of market efficiency: weak, semi- subject of considerable attentioo in recent years. Critics of financiai accountiog
strong, and strong: standards implied that securities are mispriced because of accounting practices. Tbe
empirical research (for example, on changes in accounting method) arose in m-
1. The market is efficient in tbe wealc form if prices fully reflect infonnation regard- sponse to these conteotions and were offered as tests of these contentions. 'lbe m-
ing lhe past sequence of prices. 1bis fonn of m.arket efficieocy bas obvious impli- searcb represents an important methodological contribution to financiai accotmtiog
cations for tecbnical analysis, aod it includes lhe raodom walk lheory of stock
prices. research because it subjects the allegations to empirical testing. It constitutes a more
2. The market is efficient in tbe semistrong form if prices fully reflect ali publicly systematic and more rigorous use of evidence than bad occwred previously aod is
available information, including financiai statement data. Tradiog strategies based one of the major areas of researcb that contributed to the introduction of empirical
on published finaocial statement data will not lead to abnonnal retums. testing in accounting researcb. As such, the research is a considerable revolutioo in
3. The market is efficient in tbe strong fonn if prices fully reflect all information, io- financiai accounting research. Following Fama's (1991) classification, three clasSes
cluding inside information. Hence, even baving aécess to privately bcld informa- of empirical tests of market efficiency are discussed: tests for return predictability,
tion will not lead to strategies promising abnonnal expected retums. event studies, and tests for private information.
In a 20-year retrospective, Fama (1991) suggested that empirical tests for
market efficiency be classi.fied as tests for retum predictability, event studies, and Tests for Return Predictability
tests for private informatioo. 1bese classifications are best thought of as coarse par- The earliest forms o f these tests are tests of wealc form efficiency. Is the past se-
titioning of all information systems into three broad categories in whicb the bound- quence of retums hetpful in predicting future abnormal returns? H current prices
aries between them are not precisely defined. However, tbe distinctioo is useful for fully reflect the past sequence of prices and price cbanges, the answer is no. Ooe
classifying empirical researcb oo market efficiency. form of the tests consists of examining the serial correlation of security retwn.
For many purposes, the information systems of interest may need to be more Under some simplifying assumptions, in an efficient market the serial correlatioo of
finely partitioned. For example, market efficiency with respect to the prices of firms successive price changes (or retums) is expected to be zero. This is also known as
that switched accounting methods is a finer partition o f semistrong form efficiency. the random walk: theory of stoclc prices. If this is the case, then investors are playing
The relationsbip among the three types of market efficiency must be distin- a "fair game" with respect to the past sequence of price cbanges.
guisbed. The informatioo systems described in weak form efficiency are a proper Prices respond only to the unexpected portioo of any new information and do
subset o f the informatioo systems described in semistrong form efficiency, wbich in so in ao unbiased fashion. Because price changes reflect the reaction to the unex-
tum are a proper subset of the information systems described in stroog form effi- pected portion of ao announcement, they will be uncorrelated with past price
ciency. As a result, strong form efficiency implies semistrong form efficiency im- cbanges. News, by definitioo, is unanticipated and cannot be pr:edicted by past price
130 CHAPTER 6 MARKET EFF1CJENCY CHAPTER 6 MARKET EmctENCY 131

cbanges. In contrast, consider a market in whicb prices are "slow'' to react to infor- Cumulative average residual prlce change
mation. This is a fonn of underreaction to information and would lead to positive ~t)
correlation in price changes (price increases tend to be followed by price increases).
30
By contrast, consider a market in which prices "overreact" to information. Good
news is associated with an excessive price increase and with a later reversal by the
20
market. In this setting, price changes would exhibit negative serial correlation. Price
increases tend to be followed by price declines, and price declines tend to be fol-
lowed by price increases. 10
Now consider a market in which prices respond quickly and in an unbiased
fashion to news. The expected serial correlation of successive price cbanges is zero.
Price increases are no more likely to follow price increases than they are to follow
price declines. Justas a fair coin has no memory, neither do price changes. The ex- -lO
pected v alue o f future change in price conditional on past price cbanges is the sarne
regardless of the past sequence of price changes. This is also known as the random -20
walk theory of stock price changes. The empirical tests of serial correlation have
been collected since 1900 andare summarized in Cootner (1964). More recent tests
-30
are summarized in Fama (1970, 1991). The evidence suggests tbat the behavior of
successive stock price changes is well approximated by zero serial correlation. ex-
cept over extremely sbort time intervals sucb as within a given trading day. FIGURE 6-1 Cumulative residual pric:e dumge for ''winner" anel "loser''
An illustration is provided in a study by Beaver and Landsman (1981). Start- portfolios; vertical uk = cumulative average residual pric:e
ing in a given month (month 0), ali securities on the New York Stock Excbange dumge (CAR), horizoatal uis =
time in montbs relative to
(NYSE) are placed into one of two portfolios based on their abnormal retum bebav- portfollo seledion (month O is month of portfollo seledion).
ior over the past 20 months. The "winner" portfolio consists of those stocks wbo~ Sowec: Bcaw:r md LaDdamall (1981). Figure I, p. 23S.
price changes outperform the changes in the market index during the previous 20
months. The "loser" portfolio consists of those stocks that underperformed the mar- portfolios are formed. In other words, is there abnormal póce change behavior from
ket o ver the sarne 20 months.s Figure 6-1 shows the behavior of the two portfolios tbe point of selection onward? In months 1 througb 20, "wisdom by hindsight" is no
during those 20 months (month -20 to month 0). longer operating. If stock: prices are slow to react to information. the "winners"
The winner portfolio outperformed the market as measured by the cumulative should continue to be "winners," and the cumulative residual price change should
residual price change by over 30% during the time period, whereas the loser port- continue to rise. Similarly, tbe losers portfolio should continue to decline. If stock:
folio underperformed by over 30%. The result is not surprising, because "wisdom prices overreact systematically, then the winners sbould turn into losers and the cu-
by hindsigbt" was used to select the two portfolios. The magnitude of the difference mulative residual price cbange should decline, wbereas the loser portfolio should
in the cumulative average residual (CAR) performance (+30% versus -30%) is a tum into winners.6 If the stock prices behave as a "fair garoe" with respect to past
function of how many months of "wisdom by hindsight" is exercised. The differ- price cbanges, the cumulalive residual price changes should remain flat, because the
7
ence would be smaJler if a 5-month rule had been used or larger if a 30-month rule average residual price change being tuúkd to the cumulated sum is zero.
bad been used. lt is a reflection of how much information can affect prices within a
6t'bo sum of tbc two portfolioa constitutcs tbc cn.t irc nw:lcec. which C8IIDOt outpedonn itself. Hcocc, 8l
given period.
any point in time thc sum of tbc two cumulativo residual pricc cbanp must bc zero by construction.
As far as market efficiency is concemed, the issue is the expected behavior of Hencc, if ooc portfolio•s cumulativo residual pricc changes iDaease, thc odlcr' 1 cumulalivo residual
the cumulative residual price change in the months after month O, the month the price chaJJ&e must decline. . . . .
'By analO&)' to tbc fair coin, suppose you start witb $100 and toes a fau c:oin wberc $3 11 woo if a bcad 1S
tosled anel $3 1s loat lf tail appean. Bven witb a fair coin, tbc toaca will not bc evenly aplit llftu 20
toascs. If 15 beadl anel 5 tail.s occur after 20 u - , tbc wealtb bas oow powD to Sl30.1flhe coin la fair,
'More preciacly, lhe vllriablo l.s tbc IIIISystematic retum, whicb l.s thc " tcnn defined in Olapter 2 foot- tbc cxpccced wcaltb of continuina to play tbc &ameis $130. 1bial.s aoaloaou to lhe winDer portfoUo.
notc 9. Thc price c l!anao l.s adjusted for div idcnda (bcncc, rctu.ml) and for eedl firm'a ayatcmatic rWt Sitnllarl.y, auppoec 011 lhe firat 20 totaM, 5 1-s. aad 15 tail.s had occuned. At tbe eDd of lhe tweatieth
(beta) u woU as matket-wlde condltions (bcncc, rcaidual retum). The 11115yatc.QWic retUm l.s tbcn cumu- tosa, tbc wealtb l.s $70, which la ~0&0111 to tbc to.- ponlolio. Wballa tbc expec:ted wea1dl ot OQIIIIIDu-
lated over thc 20 months and labcled lhe cumulativo reaidual price c:bangc. Thl.s l.s thc aamc vllriablo u log to play tbc game? If tbc coln la fair, tbe cxpccced wealtb la $70. 11 you- to paph tb1a bebavior, it
discussed in footnotc 5 of Chaptcr 5. would look csscntially tbc samc u Figure 6-1 .
.,

CHAPTER 6 MARKET EFF10ENCY 1~3

I 32 CHAPTER 6 MARKET EFF1CJENCY

Figure 6-1 reporta the behavior of tbe portfolios for the 20 months after the Event Studles
selection. The cumulative residual price change ia tlat for both portfolios, which is Event studies examine the residual price change of a sample of finos for a ~
conaiatent with the market ooither aystematically overrcacting nor underreacting to of time on either side of ao identifiablc "event," sucb as announcements of ~
infonnation but on average reacting in ao unbiased maoner. The major exceptioo to stock splits aod divideods, cash dividends, ~gs f~recasts, or chml:g.e s in .c-
tbe fair game behavior is for extremely short time intervals, such as transactioo-to- counting methods. The influeoce of economy-Wide and. if necessary, addi~onal ~
transaction price changes. This is generally considered to be due to marlret "mi- tors such as industry-wide infonnatioo on stock prices is extracted to ~~~ a resid-
crostructure" reasons, such as nonsynchrooous trading aod bid-ask phenomeoon. ual return. 1be expected value of the residual price changes, not coodittoDlDg ~n tbe
1be microstructure literanue is reviewed by Patell and Wolfson (1984) aod Fama event, is zero. Events studies have examined several aspects of market eflicte.:y
(1991). Nonsynchronous trading tends to induce the appearance of correlation in with respect to accounting data.
price changes over short intervals. The bid-ask pbenomenon induces the appearance
of negative serial correlatioo in observed price changes (reversals) over extremely
short intervals. In both cases, the appearance of serial correlatioo is induced by a Speed of Adjustment to Eamings Announcements
fonn of m.easurement enor in the observed prices, not because of market ineffi- The evideoce indicates that weekly residual price changes respond quicldy to m -
ciency, and in any event. persists over only short time intervals. nouncemeots of annual and quarterly earnings (see Figure 5-2), and Morse (1911)
aJso finds a rapid speed of reaction on daily data (see Figure 5-4). In both gr3Jãs..
the dotted line indicates the magnitude of the average price cbange during nooeam-
Prediction of Retums-Other Conditioning Variables ings announcement period. As discussed in Chapter 5, the large sp~ oo da~ (wcck)
Past price changes is only one variable oo which to conditioo the prediction of fu- O reflects the fact that price changes are more likely to occur dunng ~gs an-
ture returns. Some of the variables that have some ability to predict future returns nouncements. The average residual price change returns to normal very qmckly.
are systematic risk. fino size (as measured by the market value of common stock Patell and Wolfsoo (1984) inspect transaction-to-transactioo price chanr.es
outstanding), aod the book-to-mark:et (B1M) ratio (Fama aod French 1992). The based on intraday data. Their study finds similar results, whi~h are repo~ed in ~g­
capital asset pricing model (CAPM) suggests that systematic risk (beta) is the sole ure 6-2. Here, price changes are measured on an hourly baslS. The maJOr ~
determinant of differential expected returns at any point in time. Both size and occurs during the bour the earnings announcement is reported on the newsw.e.
BTM appear to have predictive power eveo after controlling for di.fferences in sys- There is a small, but statistically significant, reaction for as much as four hours af-
tematic risk. In particular, the average returns for small firms is higher than for large terward. Beyond four bours, price change has returned to normal. Often the e.n-
firms, and the average return for firms with high BTM ratios is higher than firms ings announcement is closely followed by other informatioo, such as other news te-
with lower ratios. Jeases by management, teleconference calls between managemeot and ~ysts, and
The tests are testing a joint bypothesis of market efficieocy and a model of reports by analysts. Thus, the subsequeot significant, albeit small, reactton coul~ be
market equilibrium that reflect the factors that affect expected returns. As a result, due to the tendeocy of other announcements to cluster around or follow the ~s
the interpretation of these empirical results has beeo the subject of coosiderable announcements. Even four bours is a relatively short time interval and lS u.:b
controversy. Is the predictive ability of size or BTM evidence of JlllU'iret ineffi- sborter than the interval over whicb abnormal retums are observed in the predicôon
ciency, or does it merely reflect the fact that the CAPM is not a correct model asset o f returns studies discussed earlier.9
pricing equilibrium? This is difficult to resolve, and the evidence is opeo to either The Patell and Wolfson (1984) study of intraday data suggests an extremdy
interpretatioo. rapid reactioo to earnings announcements, where rapid is defined in terms o~ bours
The BTM evidence is particularly interesting because the accounting book after the announcement. Security prices may react significantly yet only parttally to
value of equity is a component of this ratio.8 1be market inefficiency interpretation the information when it is released. Even if this were the case, there is no reasa. to
is that a high ratio implies that market value is small relative to book value and is ao see wby the slowness to respond is particular to earning~ ~o~tion..Becam~e ~f
indication that the market value is currently underpriced. Eventually, the underpric- the attention given to earnings by the investment commuruty, 1t ts an unlikely c*i-
ing wíll correct itself and bence future returns are expected to be higher during this date to be subject to slowness to respond.
interval (Lakonisbok, Shleifer, and Vishoy, 1994).

'In fact, addressing thís issue becomes inc:reasiogly difficult as the wiodow is shorteDed because of ..ar-
-otber studies tbat docnmenr a relatiooslúp between CWTCDt accounting meas\U'e and fulure retums are k:et microstructure issues, sucb as the bid-ask pbenomenon. In fact, Patdl and Wolfson use a - of
Ou md Pemnan (1989), and Lcv and ~ID (1993). price reversal as one of lhe indicators of spced of respoose.
134 CHAPTER 6 MARKET EFFlCIENCY CHAPl'ER 6 MARICET EFFICIENCY 135

changes lead eamings cbanges and lead the book value of equity for as much as six
years. Thus far, the picture is one of a mark:et tbat responds quickly and in an aotici-
patory manner to eamings. It may not say mucb for the ~liness of eamings, but it
is a remarkable testimony to the efficiency of the market with respect to eamings
announcements.

Evidence on Pricing of Disclosures

As discussed in Chapter 5, the research documents a number of instances in wbicb
data that is disclosed in the footnotes to the annual report is priced. Tbe examples
included disclosures about pension assets and obligations, fair value of investment
securities, fair value of financiai instruments, oonperforming loans, and interest-rate
sensitivity schedules of banlcs. Many of these disclosures, sucb as the pension dis-
closures, are complex and difficult to inteipret. In this sense, one would think they
would be a natural candidate for being ignored by a naive, inefficient lll8lket Yet
this information appears to be priced by the marlcet. Similar statements could be
made regarding the other disclosures. This evidence is consistent with a market in
-2 o 2 4 6 8 10 which prices reflect a fairly high degree of sophistication in the interpretation of fi-
nanciai reporting informatioo and in wbich footnote information is incarporated
Number of hours before and after eamings announcement
into the total mix of infoonation affecti.ng price.

FIGURE 6-2 An index of tbe number of extreme prlce changes (ig- Evidence on Changes in Accounting Method
norlng sigo) in the hours surrounding the earnlngs an- and Differences in Accounting Method
nouncements. The actual number is divided by the ex-
pected number of extreme prlce changes based on a As discussed earlier, diversity in accoonting methods under GAAP and the ability
nonreport period. H the actual number is eqoal to the
to change accounting methods are two of the major features of the financiai report-
expected number, the index is 1.0.
Sourc~: Patell and Wolfson ( 1984).
ing system. As a result. the rcsean:b on marlcet efficiency has focused on the switch ·
in depreciation methods from accelerated to straight-line depreciation for reporting
purposes (Archibaid 1972; Ball 1972). These studies assume that if the mark:et were
efficient with respect to this information, the cbanges in prices, ceteris paribus,
would be zero. 10 Tbese studies found that there were no abnonnal residual price
Prices Lead Eamings-The Information
cbanges associated with a firm. changing accounting method even when the effect of
Content of Prices
the change was to m.ake eamiogs higber than it otherwise would bave beeo if no
Studies fmd that security prices anticipate eamings prior to the earnings announce- cbange had taken place.
ment For example, Figure 5-l from the Ball and Brown ( 1968) study indicates that Alternatively, prices act as if investors look beyood accounting numbers and
for the fmns with good eamings news the cumulative residual price changes are take into account the fact that eamings are being generated by a d.ifferent account-
positive well before the announcement month. which means prices are anticipating ing method. Further evidence supponing this contention is provided in the analysis
the earnings news for several months prior to the annuai announcement Part of this of price-earnings ratios of firms using different depreciation methods. As discossed
anticipation is simply dueto the release of quarterly earnings announcements. How-
ever, this anticipatory effect also occurs because of the availability of other infor-
mation that permits investors to revise expectations about eamings. 1ot'ests of maitet efflcíeocy have uaumed tbat tbe informalion effect would be zero in a madret tbat

Furthermore, this anticipation could be because prices are responding to more fully rel1ectecl tbo inf011X18tion. A con~extual ara-t (In odlec worda, DO ditec:t impM:t 011 c:ub flowl)
wu typically oft"ered ID juatlfy such 1D UIUIDpliCCI. AJtboQab tbo empirical evideDce ls viewed M IUp-
timely noneamings information. For example, the researcb on the information con- portlve of madret efficleocy, an &rJIIIDOill could have beell lldvanced tbat tbe cballp In medlod itaelf
tent of prices and on delayed recognition discussed in Chapter 5 indicates that price conveyed informalioo, coo.si.stent with tbe sipaliDa litendure diac:ussed in 0\apiK 2.
136 CHAPTER 6 MARKET EFflOENCY CHAPTER 6 MARKET EFFlCJENCY 137

in Cbapter S, Beaver and Dukes's (1973) study of accelerated versus strai.ght-line Abarbanell and Bemard (1992) suggest that the reasoo for market oaive pro-
depreclation indic:atca tbat tbe pri<:c-rcported eamings ratios of thesc firms do sig- cessing of eamings informatioo may be the financiai analysts. The study concludes
Dificaody ctitfer. However, Wbeo reported eamingt adjuated to compute pro forma that analysts' earnlngs forecasts do not fully reflect the time-series behavior of eam-
fNII'DiDp buod on a COIDIDOD mecbod of compudng deprociation, the prico-oamiDgs inge. 1bere ia serial dependeDce in tbe forecast eJTOrS wben analysts' forecasts are
ratios are not significantly different. This finding is consistent with a market that used as the benchmark. A necessary condition for a forecast to fully reflect avail-
adjusts for accounting method differences, in contrast to a marlcet in wbich there is able eamings information is that the forecast errors (unexpected earnings) are seri-
myopic reliance placed on reported eamings. ally uncorrelated. The study then examines the extent to wbich tbis may explain the
postannouncement drift and concludes that it may explain part but not ali of the
drift.
Postannouncement Drift in Earnings With the view that analysts play ao important role in the price fonnation
However, the story is not clear~t. A series of studies have documented abnormal process, Dechow, Sloan, and Sweeney (1996) address the new-issue puzzle by ex-
retums for several months subsequent to the eamings announcements. The eartier amining long-term earoings forecasts by analysts. The new-issue puzzle, which is
studies in this series were critici.zed for ha~g omitted potentially important factors summarized by Loughran and Ritter (1995), suggests that after an initial price rise
that determine expected retnms and having incorrecdy misspecified the asset pric- shortly after stock in an initial public offering is publicly traded there is a negative
ing model. However, later studies addressed these concerns and the postannounce- abnormal retum that persists for five years after the initial public offering (IPO).
ment drift in residual retum still remained. Dechow, Sloan, and Sweeney find that analysts' forecasts are optimistic for IPO
Foster, Olsen, and Sbevlin (1984) documenta significant residual retums after firms relative to a sample of non-IPO finns. As the u post earnings performance is
the eamings announcement.ln particular, the positive drift in the cu.mulative residual revealed, the average forecast errar is negative, and the stock price reacts nega-
price change continues for the good earnings news portfolio, with a negative drift tively, leading to the relative underperformance of the IPO firms.
continuing for the bad eamings oews portfolio. In defining residual retums, Foster, Contrary to conventional wisdom in which the analysts are thought to be part
Olsen, and Sbevlin use the retums on fums of comparable size as the benchmark.. of the institutional frameworlc that makes the market efficient, these studies suggest
They find that the magnitude of the postannouncement drift is greater for smaller that analysts may in fact be a contributing factor. Of course, interpreting the future
firms. lnterpreted as evidence of marlcet inefficiency, the evidence suggests that the abnormal retums as evidence of market inefficiency is subject to the caveats already
market is slow to react to the full implications of an eamings announcement, although discussed.
the research indicates that it clearly reacts significantly at the time of announcement. Chapter 5 discussed the empirical evidence on the pricing of accruals. Al-
Moreover, it suggests that this inefficiency is greater for smaller fums. Presumably, though accruals are priced by the market, research by Sloan ( 1996) suggests that the
there are systematic differences in the financiai reporting environments of large and market may be mispricing the accrual component of earnings. In particular, Sloan
small firms, such as a smaller oumber o f analysts follow smaller fi.rms. provides evidence consistent with the contention that the market treats both the ac-
Subsequent research by Bemard and Thomas (1989, 1990) represents an crual and cash flow components of earnings as having the same degree of persis-
econometric tour de force in attempting to control for other factors and for remov- tence. However, the time-series behavior indicates that the accrual component actu-
ing sources of misspecification. 1be postannouncement drift was observed in their ally has less persistence than the cash flow component Sloao then forros a trading
study as well, and tbe effect was persistent. They also document the fact that the ef- strategy based on the accrual component and demonstrates that the future residual
fect is larger for small firms than for large firms. Their research suggests that the retums are negative (positive) for those firms with a relatively high (low) amount of
market prices act as if they do not fully reflect the time-series behavior of eamings. accruals. Again, why the market participants would systematically ignore a differ-
The interpretation of these results are the subject of much debate. In addition ence that apparent from ao aoalysis of past time series is a mystery.
to the research design issues of model misspecification and risk mismeasurement,
the findings are anomalous in at least two respects. The evidence appears to be in
conflict with much of the otber- research discussed eartier, such as dramatic and Summary of Evidence
swift response when intraday retum data are inspected and the evidence that sug- The results taken as a whole are indeed a paradox. Foster (1986, p. 399) observes
gests prices lead earnings by severa! years. Can prices both lead and lag eamings?
The evidence is also anomalous because eamings are the least likely candi-
date for inefficiency. Earnings are widely analyzed by the investment community. Notwithstanding this evidence, tbe efficient market model continues to play an Íl'Ilp«-
tant role in tbe literature. One reasoo is that compcting models are oot well aniculatcd.
No other single firm-specific variable receives more attention by the aoalysts and A second reasoo is that non-m.arlcet inefficiency explanatioos for the anomalous evi-
other capital marlcet participants than eamings. The result is paradoxical. dence exist.
138 CHAPTER 6 MARKET EFflCIENCY CHAPTER6 MARKETEFFJCIENCY 139

Similarly, Fama (1991) concludes For the first issue, the empirical studies do not directly test marlcet efficiency.
The security prices that would prevail if there were universal knowledge of the in-
In sbort, some event studies suggest tbat stock prices do not respond quicldy to some formation are not observed, because not evecy investor is literally aware of the
specific infonnation. Given tbe event-study boom of tbe last 20 years, bowever, some change in accounting methods and its potential implications. Instead, tbe empirical
anomalies, spurious and real. 8I'e iocvitable. Moreover it is important to emphasize tbe studies attempt to infer that the security prices bebave as if such a condilion held.
main point. Event studies 8I'e tbe cleanest evideoce we have on efficiency (at least un- The research on cbange in depreciation metbods provides an example.
encwnbered by tbe joint-hypotbesis problem). Witb few exceptions, the evidence is
supportive. (p.1602) The empirical evidence is interpreted ac; showing no price effects associated
with the change in method and is further interpreted as evidence consistent with an
efficient market Prior to these studies, advocates of market inefficiency argued that
Retums to Private Information a change in depreciation methods would induce a price change because of a myopic
Ignoring information costs, there are potential retums to private information. The reliance on the earnings number (in other words, market inefficiency with respect to
events studies on earnings announcements clearly show that, if an investor were to this information). Hence, the evidence, which indicates no average residual price
trade on advance earnings information. abnormal returns are possible. However, be- change, is interpreted as consistent with market efliciency.
cause there are costs to information. there is no economic reason to believe that the For the second issue, marlcet efficiency is typically formulated in terms of the
market would be efficient in the strong form (with respect to ali infonnation, in- null hypothesis of no price change. Market efficiency is then not rejected, unless it
cluding privately held information). The costs of obtaining and trading on privately can be rejected at "publishable" leveis of significance. In this context, the power of
held information can be sufficiently high to preclude abnormal retoms, after costs. the test (the ability to reject the oull when it is false) is criticai. Failure to control for
The costs include not only search and analysis costs, but also transactions costs of other factors influencing price changes (even when they are uncotrelated with the
trading and legal liability associated with trading on privately held infonnation. event of interest) can reduce the power of the test. An altemative approach would
Bernard and Thomas (1990) claim that the postannouncement drift is within the be to adopt "marlcet inefficiency" as the null 1be purpose of the analysis would be
range of reasonable estimates of transactions costs. to see whether extant research methods produce results that are capable of rejecting
Moreover, if some of these strategies from apparently readily available data the null o f inefficiency. This, of course, requires a specification o f what "ineffi-
lead to such large abnormal retums, one might expect this to be reflected in the in- ciency'' means in terms of prices. In many cases this is unclear. However, in the
vestment performance of portfolios managed by professionals (for example, mutual cbange in depreciation methods example, positing a constant price-earnings multi-
funds, pension funds, and endowment funds). However, the evidence of superior pie provides an estimate of the expected price change and constitutes one possible
performance by managed portfolios is quite weak, with many studies finding the ab- specification o f the inefficiency. The altemative was rejected by the evidence.
normal performance to be negative once fund expenses, including management fees The third issoe of research design concems interstudy comparability in gen-
and trading costs, are included. This literature is reviewed by Fama (1991), who eral and specification of the security retum metric in particular. Since Ball and
concludes that if these are the superior, informed individuais, they are trading be- Brown's (1968) study the literature has extended this early wodc by expanding the
yond the point where the marginal benefits to investors equal marginal cost. set of signals examined, by expanding the technology for measuring the infonna-
tional variable (for example, choice of earnings forecasting model), and by expand-
ing the technology for deriving a security return metric. As a result, we have experi-
enced a contemporaneous expansion of both the topics studied and the methods
6-5 RESEARCH DESIGN ISSUES used to study them. In many cases, a study will alter both topic and method simulta-
As with empirical research in many other areas, efficient market research is the sub- neously, making comparison with previous studies difficult. When a difference in
ject of controversy and the evidence is subject to di verse interpretation. One reason results is observed, it is difficult to identify the source of the difference, because
for diverse interpretations is that some features in the research design reduce the more than one dial on the research machine has been tumed. Because of the ten-
power of the tests. This section focuses on six research design issues: (1) the rela- dency to turn more than one dial ata time in each additional study, it is difficult to
tionship between the definition of market efficiency and the tests of market effi- infer the reason for the different results.
ciency, (2) the power of the test and the choice of the null hypothesis, (3) interstudy A fourth research design issue is self-selection bias. Evidence soggests that
comparability and the variety of topics and methods choices available to the re- many of the firms studied differ from other firms in a systematic fashion. Their
searcher, (4) self-selection bias, (5) the markets studied, and (6) joint bypothesis residual price cbange behavior in the month before the change is very moch like the
testing and model misspecification. "loser" portfolio in Figure 6-1. Fmns that switch depteciation methods do so for a
140 CHAPTER 6 MARKET EFf'ICIENCY CHAPTER 6 MARXET EFflCIENCY 141

rcason and differ in systcmatic ways in addilion to pdce bebavior, sucb as earnings
6~6 THEORIES OF MARKET EFf1CIENCY
perfOI'IJWJCe. Simllady, firma that uso- tbo poolinJ v~ pW'Clbuo ~tment ~or
busiocas combinalions differ from cach other. In parucular, tbe restdual pnce Market efticiency involves a tbeory of the process by which information becomes
cbange bebavior of firm.s tbat use tbe pW'Clbase treatment is similar to the ~'winner" reflccted in prices. Unless individuais are cbaracterized as throwing away some-
portfolio in Figure 6-1. If firms differ in a systematic way in tbe months pnor to the thing of value, information is not used because it is costly. 'The basic argument for
event of intcrest, there is a confounding effect, which can increase ambiguity in tbe market efficiency is one of competitive markets. Large numbers of buyers and sell-
intcrpretation of tbe results. ers will compete with one another for tbe interpretation of financiai reporting inf~­
A fiftb research design issue is the securities mark.ets studied. For the most mation. Competitive forces will cause prices to reflect unbiased assessments of tbe
part, samples have been drawn from firms Iisted on tbe excbanges. Most frequently, implications of the information for security prices. In tbis setting, investors woold
tbe New York Stock Exchange bas been the population from whicb tbe samples are not process information in a biased or inefficient fashion., because it would be
drawn. Although listed securities represent tbe bulk of tbe market value of common against their own economic self-interest do so. A related argument is eveo if indi-
stocks traded, it is difficult to determine tbe extcot to whicb tbe evidence of listed vidual investors might be prone to such inefficieocies prices will not, because inef-
securities can be generalized to tbe over-tbe-counter (OTC) market. For example, ficient prioes would offer "arbitrage" opportunities (riskless aboormal retums) and
those who believe that analysts are tbe "force" tbat makes tbe market efficieot often competition amoog arbitrageurs would cause prices to fully reflect information and
argue tbat tbe OTC m.arket may be inefficient witb respect to many items of infor- remove opportunities for arbitrage. In this setting, why would one expect prioes oot
mation. Evidence indicates tbat sucb companies receive less analyst following. Is it to "fully reflect" publicly available informatioo? 'The answer, of course, lies in tbe
then appropriatl( to infer less market efficieocy for sucb securities? Researcb by possibility that such data are not universally available at zero cost to all individuais.
Sbores (1990) indicates that eamings announcements are relatively more important Knowledge of tbe cbaoge in metbods may not be universal. Moreover, tbere
for such finos, but tbat the OTC stock prices bebave similar to listed securities as may not be universal (costless) access to other information on the implications of
far as tbe speed of reaction to earnings announcements are concemed. the change. In other words, obtaioing the training (for example, knowledge of de-
There may be altemative mechanisms. For example, for OTC securities, tbe preciation methods, understanding that a furo can bave a different set of books for
broker-dealers who "make" the market in that particular stock may fulfill this role. tax and for annual report purposes) may be costly aod bence such knowledge is oot
They stand ready to buy and sell shares at announced bid and ask prices. Such indi- universal. However, ioterpretation of the change goes beyond this and involves tbe
viduais would appear to bave natural incentives to ensure tbat no one "knows" more assessmeot of management's motivations for cbangiog depreciation metbods (for
than tbey do. As a result, the bid and ask prices may reflcct tbe information in the example, retlecting management's expectations about future earniogs or plans for
possession of tbe broker-dealer, and sucb an information system may be extcnsive. additional asset acquisitions). Sucb analysis (information) is provided by the finan-·
Moreover, as will be shown, tbeories of market efficiency need not rely oo the exis- cial and accounting community, but perbaps not oostlessly. Hence, market prices
tence of "experts" such as analysts or broker-dealers. might not retlect tbis potentially costly ioformation. This constitutes a simplified
The sixth research design issue is the joint bypothesis issue. A test of mark.et explanation of why a nonzero probability of mark.et inefficieocy witb respect to
efficiency is ajoint test of efficiency, an equilibrium model of capital asset pricing, "publicly available" information might be assessed. However, despite the cost of
and the measurement of the relevant factors. For example, in tbe CAPM, systematic the information, security prices might act as if tbe information were costlessly avail-
risk (beta) was assumed to be the only factor determining differential expected re- able to all investors. This provides one intexpretation of the empirical research on
turn. A finding o f aboormal retums could occur because of market inefficiency, the semistrong form efficiency. In particular, from this perspective the empirical studies
incorrect cboice of an asset pricing model, or measurement error in systematic risk . of change in accounting methods are viewed as testing mark.et effic ieocy witb n>
It can be difficult to disentangle the iofluence of eacb of these factors. spect to more informalioo tban merely the knowledge that a change took place (in
These research design issues tbat may lead to difficulty in intexpreting inarket other words, "sopbistication" of tbe use of the datais also an issue).
efficiency research are a common basis used by tbe researcb community in judging Grossman (1976) and Grossman and Stiglitz (1980) provide analyses of con-
tbe power of tbe empirical evidence. 11 Such issues can explain wby market effi- ditions under which market efficiency would or would not be obtained. In tbese
ciency is a oontroversial topic among the research community. models, individuais "extract" infOIJDation from prices. In tbe Grossman and Stigtitz
model, individuais choose to become informed or uninformed, and at equihõriom
each individual is indifferent, because eitber action (after deducting informatioo
11Wbile lhe di.scussion bas focused on lhe research design issues of efficient market studies, many of costs) offers the same expected utility. For tbere to be incentives to purcbase iofor-
tbese samc issues ariae in otbez .-eu of security p]ce researcll, such as tbose discussed in Chaplcr S. mation., prices canoot "fully reflect'' tbe information obtained. Heoce, the martet
142 CHAPTER 6 MARKET EFFICIENCY
CHAPTER 6 MARXET Ef'FlCIENCY 143

must be inefficient with respect to that information. In the Grossman model every ~ to. ~e uogarbled inf0110ation system. Moreover, assume that this is true for
individual is equally uninfonned in that each receives a garbled signal, but prices every m<;li.vt~ who ~mprises the market. However, the idiosyncratic bebavior,
act as an aggregation of everyone's information, such that the price ''reflects" infor- b~ defini?on, ts esse~tially uncorrelated amoog individuais. As a result, security
mation that is superior to that held by each and every individual. However, individ- p~ce, ~hich can be VIewed as a "consensus" across investors, is effectively able to
uais extract this superior information from prices and the price ''fully reflects" that diversify away ~ large i~~c component, such that only the knowledge
superior information. This model a1so has issues of private incentives for informa- (the_uo~~led ~tgnal) pers~ m terms of explaining the security price. By analogy,
tion production, because each individual is assumed to costlessly observe price (and the mdiVIdual mvestor beliefs can ~ viewed akin to individual securities and the
extract infonnation from it), which eliminates the incentive to privately seek infor- security price can be viewed as an aggregate akin to a portfolio.
mation if it is costly (in other words, a free-rider phenomenon occurs). The small amouot of koowledge is the systematic component across in-
In a related veio, Verrecchia (1979) has constructed a model in which price vestors. Although it is dwarled by the idiosyncratic behavior at the individual in-
acts as an aggregator of beliefs (as distinct from an aggregator of information). As vestor levei, it is the only portion that persists at the secwi.ty price level This does
the number of individuais increases, prices behave as if everyone observed the un- not req~ the_ exis~ o~ any "experts." Moreover, the quality of the knowledge
garbled signal. The major difference is that there is no explicit learning from prices reflected m pnces ts constderably higber than the average quality of knowledge
involved. Hence, while prices reflect the ungarbled signal, individual beliefs or across !he individuais who comprise the market. Analogously, in portfolio theory
portfolios may not reflect this information. Neither Grossman nor Verrecchia re- the v~ of retum of any portfolio is strictly less than the average variance of
quires the existence of a subset of "initially" more informed individuais (for exam- the sectmties' returns that comprise the portfolio, if securities' retums are~
ple, superior analysts). These analyses deal with private information production. lated with one another. 12
Ooe such collective choice would be centralized production of signal, thereby
avoiding the costs of redundant production. Forecasting FootbaH Games
Bagehot (1971) and Treynor (1974) introduce a model of superior informed
individuais (or at least perceive themselves to have superior information). As are- ~e fact th_at a co~nsus can reflect "greater than average" knowledge or insigbt is
sult, they are the major "active" participants in the market, because the uninformed illustrated m a seemmgly uorelated context-the prediction of outcomes of football
engage in buy-and-hold strategies that minimize trading. Although each informed games. Each week during the football season, the sports staff of the Boston Globe
individual is informed relative to the uninformed, no one individual observes a predicts the outcomes of the NFL professional football games for the coming Suo-
noiseless signal. When an active trader comes to the market, the person most likely day or Mooday. They are to pick which team will do better than predicted by the
to be on the other side of the transaction is another active trader. In this setting, they Las V~gas "tine." The Las Vegas tine is determined by the market for legalized_
trade because they canoot perfectly extract the noise from their own private signal. gambling and represents a market consensus of which team is favored and the
The price reflects some average of the two private sigoals and hence reflects their "spread," the expected differeoce in the scores. Members of the sports staff can be
information, but retums to each trader are normal, because no one active trader has viewed as experts potentially possessing superior insigbt or superior information.
consistently superior information to any other active trader. Treynor characterized However, the market has at least two factors operating in its favor. It represents a
this model of market efficiency with a quotation from the Pogo comic strip, ''We ~nsensus a~ has the power of diversification relative to the forecasts of any indi-
bave met the enemy, and he is us." VIdual. Also m the market, money, as well as pride, is at stake.
Individual sports staff members forecast whicb team will do better relative to
the spread. The cumulative results for the last week of the NFL season are summa-
Illustration of a Theory ~ze~ in_ Table 6- 1 for 1987 through 1996. If individual staff members bave no supe-
The crux of a theory of market efficiency that does not rely on the existence of a set nor 1ns~gb~ t~ the market, by purely guessing they can expect 50% accuracy. Over-
of "experts" is that the levei of koowledge reflected in prices is greater than merely all, the mdiVIdual forecasters were rigbt 50.2% of the time and wrong 49.8% (5,064
versus 5,023}, whicb is not significantly different from 50%. 1be best forecaster
the "average" levei of knowledge among investors in the market. Some sirnple
analogies illustrate this point.. Consider eacb individual containing a "small"
amount of knowledge and a considerable amount of idiosyncratic behavior. This
can be modeled as each individual receiving a &arbled signal from an inform.ation
system that provides an ungarbled signal disaufsed by a ''noise" component. The
garbliog is so large that any inspection of that individual'$ behavior provides llttle
indication that such an individual is contributing to the efficiency of the market with
 CHAPTER 6 MAR.KET EFflOENCY 145
144 CHAPl"ER 6 MARKET EFFlClENCV

1be purpose of thc example is not to demonstratc that no superior forccastc:n

TAJILE 6-1 FoteQUtina Foocba11 aame.• ot tootball pmee exiat. Tbe Wuatration ia c:ertainly not lntroduccd to abow that no
superior analyata exiats. However, the example does illuatrate that thc conseDSUS
YBAR
forecast imbcddcd in madcct prices can be very diffi.cult to beat evcn by thoee wbo
FORCASTER 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 TOTAL plausibly are viewed as experts with superior insight, lmowledge, or ski11.13

Will McDonough
Beattbeüne 115 110 91 112 114 97 96 115 101 69 1020 lnterim Summary
110 111 113 93 93 104 111 91 105 82 1013
Lost to lhe Une The empirical research arose in response to claims in the financiai and accounting
Michlel Madden community that market inefficiencies existed. As a result, the empirical findings
128 114 100 112 109 99 101 95 107 84 1049
Beattbeüne 984
Lost to tbc Une 97 107 104 93 98 102 106 111 99 67 have largely preceded a formal, conceptual developmeot A variety of theories of
RonBoraes market efficiency have been explored, some of which compare informed with unin-
108 110 107 109 103 106 93 102 73 1031
Beattbeüne 120 formed investors, some focus oo competition among informed individuais, and stilJ
lOS 113 94 98 98 98 101 113 104 78 1002
Lost to tbe line
others focus on the consensus property of security prices that permit prices in prin-
JohnCamey 833
Beat tbe Une 116 104 97 102 103 107 107 97 cipie to be better informed than most, if not all, the individuais that comprise tbe
109 117 107 103 104 94 100 109 843 consensus. These theories help explain how privately beld information may (or may
Lost to tbe line
Nick Cafardo 108 not) become fully reflected in security prices, bow costs are a potential baaier to
Beat tbe Une 108 mark:et efficiency, and how market efficiency need not rely oo universal sophisti-
117 117
Lost to lhe Une
Jim Greenridge cated or even a subset of superior informed investors in order to function.
94 103 197
Beat tbe line 228
Lost to lhe Une 127 101
Frank Dell' Apa 376 6-7 IMPLICATIONS OF MARKET EFFICIENCY
Beat lhe line 98 104 97 77
107 103 104 52 366
Lost to tbe Une This sectioo is devoted to a discussioo o f the implications o f assuming that the III3Ib:t
Mark Blaudschun is efficieot with respect to publicly availab~ information (semistrong form effi-
108 !OS 65 278
Beat tbe line 285
Lost to tbe Une 98 101 86 ciency). Discussing the implications is important but difficult A difficulty arises bc-
VinceDoria cause the potential implications o f market efficiency cannot be discussed in isolation.
104 68 172
Beat tbe Une
102 83 185 but rather in the contextof other judgments of fact and value judgments. To the extent
Lost to tbe line possible, the discussion will be explicit about the other judgments, perceptions, and
Oveull
Beat lhe line
5064 assumptions being introduced. However, an element of subjectivity and personal
5023 opinion inevitably creeps into a discussion such as this. The statements that follow il-
Lost to lhe line
lustrate why many constituencies in the financiai reporting environment have shown
•ra.tDtfrom •a~o~>e SlaJfs Pro ~i«fions ~ oftire 8051011 Olobe for tire jiNIJ wuk oftire ngUÚlr NFL season
an interest in and perceive that they are affected by market efficiency. However, as
for tire yean 1987 to 1996. nu are omilted.
will become apparent, many o f the statements made have not been derived rigorously
and rely oo intuition. Moreover, empirical evidence is often laclcing on criticallinks,
which establisb the importance of market efficiency to these various constituencies.
With this disclaimer, the following implications are offered.
bad a accuracy of 51.6%. This "superior'' performance could be due to either luck
or to skill. However, even this difference is not significantly different from 50% at 1. The efficient market implies that tbe substance rather than lhe form of disclosure
conventionallevels of significance. Moreover, with nine forecasters, ex post s~me may be tbe more important policy issue. A market tbat is effícient with respect to
would be expected to do better by chance. Consistent with the l~ck interpretatJ.~n. public informalion may not be efficient with respect to private infoonation.
when the group as a whole is examined. four did better bu~ five did worse. Co~u:­
tency of superior forecasting ability across years is also an unportant factor. lf mdi- 13
Libby (1981) di.scusses severa! contcxts wbece a COIISCIJ.SWI outper!OilllS lhe avecage of those co~
vidual years are examined. the best forecaster overall outperformed the line in only ing tbe consensus. 1b.is is an extremely simple model of investor behavior mainly designed to illiJSIDie
six years but fell short four years. Similar behavior was observed for the other tbe diveoification cffecU that may be opcraling on price. Bagebot (1971) and ~yoor (1974) quCSiioa
tbe lack of i.ndcpendmce across investors.
forecasters.
 ..
146 CHAPTER 6 MARKET EFflQENCY
CHAPTER 6 MARICET Ef'f1CJENCY 147

Hence, bringing an item of information into the public domain (disclosure) is an quire the ability of tbe "aven,e prudeot investor" to conduct such m -.lylis
issue of substance. However, once in the public domain, wbether the format used However, along other dimensions, tbe malysts' activities are 11Jbsútutes f«~
to display the data is a substantive issue is unclear, at least in its impact on secu- ci~ ~g inf~on. To the extent that analysts provide altemative swrces
rity prices. For example, although it may be important tbat an item appear some- of information to mvestors tbey are in competition with publicly available finan-
where in the annual report. it may not make any difference in tenns of the price of ciai information.
the socuri~ whetber the item is reportcd in the footnotes or in the body of the . . However, ~ infOimation is a per:isbable commodity aod the analyst <Xlllllllu-
statement 4 Similarly, it would make little d.ifference whether the income state- ruty ~ a co~ve one, it may be difficult for any one analyst to dcmooetratc
ment effects of an item are explicitly reported if they could be inferred from bal- CODSlStently supenor performance. However, from a broader perspective, supa:ior
ance sheet disclosure. 15 ~ormance may not be tbe appropriate metric by which to measure the CODID"bu-
2. To believe that merely because an item does not appear in the financiai statements tion of the analyst _community. Every analyst may be playing approximaldy a
it is not reflected in prices may be naive. Security prices reflect a rich information ~sum game against the rest of the community. But, this does not necessarily
system of whicb the financiai statements may be only a part. Consider this impli- unply that the ~ efff:ct of ~ activitiC:S is valueless. The process of competing
cation in the context of fair vaiue disclosures of financiai instrumellts, pension as- ~sts .searchin~ for information may m part result in secwity prioes tbat reflect
sets and obligations, and nonperforming loans, ali of which tbe evidence indicates a n~ informatíon ~atem. In this eense, the activitiea of the analyst OOIDIDIIDity
are priced by the IDllik.et From an informational perspective, footnote disclosure may m part be a public good. From this perspective, evaluation of perfODDMK:e in
may be substantively equivaient to recognition of the item in the financiai tenns of ability .to generate abnonnal muros may be inappropriate.
statements. Tbe public good upecta of analysts' activities raises the issue of JOVem-
Wby might a disclosure versus recognition make a difference in socurity mental regulation of their activities. The financiai reporting regulatioos em alter
prices? There are severa! possible reasons even in an efficient market. 1be loca- tbe nature and lev~l of such activities (for example, its powers over c:lisclosure).
tion of the item might affect management' s behavior. For example, it has been Presumably, reduction of resoun:es spent on private-sector infonnation search and
contended that requiring fair vaiue investment socurities to be recognized will tbe prevention of wealth mdistribution because of information 8S}'1Iliiidries would
cause bank management to change its investment policy. In particular, it has been be two of the ~- In particular, tbe SEC can affect the analyst COIIUIJUDi.ty
claimed that bank management will shift to shorter-term investment socurities, not only by altecing the nature of mandated disclo.<mre but a1so by regulldiug the
whose IDllik.et values are less volatile. 1be location of tbe item could also affect manner or process by which infonnation flows from corporations to analysts.
the vaiue of the firm depending on how the contracts of the firm are written in 4. In ~ efficient_IIlllrket.. the vast majority of "avenge prudent investors" may not be
tenns of accounting numbers. Recontracting costs would be reflected in the socu- active~y and li~y mvolved ~ trying to interpret tbe implications of finaociai
rity prices. Location of the item might affect how regulators view tbe item. For ~g for the ~ce of stock. 1 The average investor who has no expenise in fi-
example, bank regulators may regulate a bank d.ifferently depending on whether nanctal ~ accounttng matters may be defening the analysis to those who bave a
the fair vaiue of financiai instruments are merely disclosed or recognized in the comparaüve advantage (sucb as the information intermediaries). Hence, the ~
accounts. The form of financiai statement presentation could affect the probability maod for financial reporting infonnation is an indirect one. 1be interest of the av-
of cbanges in future tax legislation. Tbe location of the item influences bow mucb erage prudent investor may be protected by ensuring that a given data item is <fis..
resources management and auditors place in estimating and auditing the item. ~losc:<l f? a num~ of analysts so that they compete with one another fOI" the
However, the magnitude of these other effects is not clear. unp~ons and mterpretations of that item for the cash flows aod price of
3. The research, together with the informal evidence cited above, can affect the way secunty.
in which we view the role of information intermediaries. Analysts constitute part 5. Regulators ~y wish to reconsider the nature of the concem for the naive in-
of the process by which information becomes reflected in prices. Along some di- vestor. Ifthe ~vestor~ no~~ naive, is in effect facing a fair game wilh re-
mensions, the activities of the analysts are complementary to the financiai report- spect to publicly available informatJon, can the investor still be barmed? If so
ing information. By their analysis of financiai statements, complex information, how? If an investor holds a well-diversified portfolio, is a portion or a11 of tbe risk
such as pension footnote information, becomes reflected in prices. It does not re- of nondisclosure diversifiable?
~ can occur when firms are foUowing policies of less than ful1 disclo-
~ which gives some individuais the opportunity to trade on tbe privatc inf«-
matlon. 1bere may be otber consequences as weU. Portfolio positions beld by in-
14In principie, tbe fonnat could coovey information regarding maoagcment's expcctatioos. Howevec, vestors may differ even though tbe prices are the same. For examp1e
sucb m effect remains uodocumented empirically. 'The discussion also assumes that there are no second- "'improperly'' diversified portfolios is one of the con.sequences of "~
order effects that would altec tbe cash tlows of tbe fum. Foc example, if tbe J.ending opportunities of a
finn are influenced by the location of fioaocial leases in the balaDce aheet, secuóty prices could be al-
tered by format choices. Mor-e generally, it assumes oo effect that would altec lhe produc:tioo or financing
decisions of the finn. 1
"Por example, tbe dlsclosure of tbe LIPO may be substantively equ1valeot to boo1dn& LIPO in the body 'The tecm awrate prutútrt invuror relen to lhe inveatoc wbo does not bave prof-aaJ akilla in pro-
of tbe statements ~ an infonnational penpective. Henco, a finn is able to 111e LIPO for tax, md the CC41ing aod lolerpretina tho data. 1be lDdlrect role ot the averace prudent investoc 1.s aot a .nx:ent obler-
federal taxatioo constraint that LIPO must theo also be llled for reportlng purposes may not be a substan- vatioo. Kópb (1973) dilcusses the rnyth of lhe informed Jayman, md Doug1u (1933) ootea lhe ori&iDaJ
tive constraint. In otber words, tbe disclosure permits investon to adjust the financiai atatements to a intentloa ~lhe Securities Acts was that lhe professlonal would act as IID interpret« aod di~....- of
non-LIFO basis. tbe SEC disclosurea.
148 CHAPTER 6 MARKET EFFIOENCY CHAPTER 6 MARKET EFACIENCY 149

perceiving tbe market to be inefficieot. Otber consequeoces include the direct 30% had beco expected. Infonnatioo has arrivcd since tbe publication of last
costa of reponina, disacmi.oatioa, c:ertifyina, aod proceasina tbe informalion; the year' s eaminp, -s tbil Information may bave led to a substantial revision in
costs of priva~cly acelán& tbe infonnation if it is oot publicly available; thc costs ear:ninp CXJICC'Itloos since tbe last year's earninp announcements. At a mlni-
pe.l.d to aoalysta and ocben to pedorm that function for the lnvestor; and transac- mum, quarterly eamlnga and dividend announcemcota are capable of altering ex-
tions costa sucb u brolcenge fees. pectations. Forecucl by Dllllagemeot and by analysts are also a source of expecta-
tions changes. More qualitative information on cootract awards, aoquisitioos,
litigatioo, and product development may also play an important role. In this rc-
spect, it is not stuprisiog that lhe empirical evidence cited in <llapter 5 indicates
Nonimplications that expectation models of earoiogs that use last year's eamings as tbeir most rc-
A number of unwarranted interpretations may be drawn from the assumption of cent input have limited ability to explain price movements in thc one or two
market efficiency. In some cases, the interpretations cannot be drawn unless addi- months prior to tbe announcement of tbe subsequeot year's eamings.
In sum, understanding prices aod price changes without k:nowledge of lhe
tional structure is assumed. In other cases, the interpretations do not fully appreciate information set re6ected in prices and without k:nowledge of the expectatioos em-
the processes by wbich security prices arise and reflect information. Severa! spe- bedded in prices may be difficult. This inference process is particularly diflicult
cific illustrations are provided in ao attempt to document these assertions. bccause prices re6ect a composite across investors, wbo differ with respect to in-
fonnation and expectations as well as other factors such as wealth and altitudes
toward risk:.. It is DOt surprising to find particular cases in whicb the price reaction
1. Market efliciency does not connote social desirability. Market efficiency is con- to infonnation scc:ms anomalous. Howevec, one must be careful to avoid labeling
oemed with thc relationsbip between informatioo aod security prices. No value our inability to explain such pbenomenon as evideoce of mark.et inefficiency,
laden or normative conootatioos are implied. However, while the link canoot be marlcet irrationality, or tbat all-absorbing residual, ''market psycbology ."
explicitly made, many believe that greater market efficiency will indirectly lead to The first three nonimplications of the efficient maricet deal with common
better resource allocatioo, lowering of risks of investing with a increased willing- misunderstandings about the meaning of an e.fficieot mark:et or tbe process by
ness to invest, greater capital formation, aod a "fairer" system of more equitable whicb infonnatioo becomes retlected in prices. A second class of nonimplications
dissemination of informatioo. deals witb using tbe concept of mark:et efficiency to evaluate adequacies or defi-
2. Mark:ef efficiency does not imply clairvo~ance. The future is uncertain, the p~ces ciencies in tbe cwrent iustitutional structure. The concept of market efficiency by
reflect investors' el(pectati~xpectations that may or may oot be realized. itself is not sufficiently ricb to warrant infereoces regarding institutional structure.
Hindsi~t can often suggest that security prices retlected an expectation that dif- Two illustrations 1re provided. For purposes of continuity, these will be numbered
fered grCatty from tbe realization. However, this is not evidence of market ineffi- consecutively witb respect to lhe previous nonimplications.
ciency but ratber of uncertainty. Bveo an eflicient market does oot bave a crystal 4. The efficieot maitet does not imply that information intennediaries, sucb as ana-
bàll. A oumber of large "surprises," causing substantial revisions in security lysts, are useless. TilC efficieot market concept is silent on lhe value of their ser-
prices, is consistent with an efficient market. It is not valid to evaluate the effi- vices. 1bere is nodring in eitber the definition or the evidence coocerning market .
ciency of a set of security prices in the ligbt of information that ooly became efficiency that specifies the process or institutional mecbanism by which informa-
available after those prices were fonned. tion becomes reflccted in prices. Analysts may play an important role in tbe
A somewbat more subtle variatioo is to suggest that the mark:et did bave tbe process of generating and disseminating information to tbe investment commu-
infOJJillltion available but incorrectly interpreted subsets of tbe ioformation. For nity. Howevec, altbough the efficieot market evidence, in aod of itself, does not
example, certain signals were ignored, wbicb with tbe benefit of bindsigbt we indict analysts, it does not support them either.
"know" should bave received mucb greater attention. At any poiot in time a num-
S. The effi.cient maitet does not imply that no basis can be provided for govemmental
ber of signals will be available and it is uolikely that each signal, considere<~ indi-
intervention in tbe disclosure process. To draw inferences regarding tbe role of gov-
vidually, leads to the same prediction ofthe future. To examine the evidence ret-
eromental iotecveution in disclosure decisions, additional analysis of tbe nature of
rospectively aod determine wbat would have led to tbe best forecast is not valid
the institutional mecbanisro by wbich infonnation becomes retlected in prices is re-
and relies oo bindsigbt just as mucb as the earlier scbeme.
quired. Barlier it was suggested that the analyst community may play ao important
3. Ao efficient madr.et does not imply tbat investors will necessarily perceive the role in sucb a process. However, little is known in a formal way about tbe process.
marlcet to be efficieot. Perceived marlcet inefficieocies may be widespread even Corporatioos bave incentives to provide infonnatioo voluntarily to tbe aoalyst com-
thougb tbe security prices fully retlect published information for two reasons. munity, aod lhe analyst community bas incentives to search for information from
First, any one investor may be aware of only a portion of the informatioo that is corporations aod other souroes. Given this tlow o f information in the private sector,
re.flected in prices. As a result, from this myopic perspective, securities may ap- the additional iofonnation content of tbe mandated disclosures of tbe SEC may be
pear to be mispriced because prices reflect informatioo not available to that indi- minuscule. Tbe basis for governmental interveotion must stem from some aspect of
vidual. Second, interpreti.og price responses to information requires a model of the the private-sector incentives that cause them to be imperfectly aligned with the so-
expectations undedying tbose prices. We may be tempted to interpret the mark.et's cial incentives. Cbapter 7 discusses regulation in depth.
response to prices in tbe ligbt of a simplistic, naive expectations model. For eum- Tbe cboice among differeot financiai reporting systems involves choosing
ple, annual earnings per sbare may be 30% above the previous year's figure; yet arnong di.ffering c:onsequences, whicb a.ffect individnals differently. Heoce, some
the ptices migbt react adversely to the announcemeot if ao increase of more than individuais may be better off under one method, and otben may be betta off
150 CHAPTER 6 MARKET EFFICIENCV
CHAPI"ER 6 MARKET EF1'10ENCV 151
under an alternativo method. The issue is one of social choice, which involves
making interpersonal welfare comparisons. to concem over infonnation asymmetry oould have sevcnl e:ffects. (1) It could .lead
to an attempt to extract the information by observing tbe behavior of tbose per-
ceiv~ to bave ~ore information (for example, by obscrving the prices at whicb tbe
more informed mvestors would be willing to trade). 17 (2) It could lead to a ~11-
6~8 IMPORTANCE TO FINANCIAL di~ersified, passive portfolio strategy dcsigned to insulate the inves~ from IIadioB
REPORTING CONSTITUENCJES w1th others wbo migbt bave superior information. (3) It could cause the investor to
This section attempts to summariz.e the importance ofthe implications from the per- retreat from sucb markets and invcst in markets in which information 8S}'IDDletly is
spective of the various constituencies in the financiai reporti.ng environment: in- of less concern. (4) It could also induce the investor to reduce the total 8JJlOUDt of
vestors, regulators, management, auditors, and information intermediaries. This dis- investment
cussion faces the same di.fficulties cite<!. at the outset of the previous section. As a The potential effects of information asymmetries on investor bebavior were
result, the same disclaimer that was tendered there applies here as well. discussed at greater length in Cbapter 2. The list of potential effects is o:ffered as il-
lustrative and is not intended to be complete. Moreover, little is known about bow
lnvestors important these effects are empirically, and it is not unreasonable to believe that tbe
e:ffects may vary across investors. The effects on investor behavior is of concem not
The relation between financiai information and security prices can influence in- only to investors themselves but also to the other constituencies. As a result, madet
vestors' demand for financiai information. As previously discussed, in an efficient efficiency is important to these other groops in part because of its potentiai effects
market much of the demand for financiai information may be an indirect demand. on investor bebavior.
Similarly, the consequences are indirect, aibeit important, via the effect o f financiai
information on security prices. For example, changes in security prices aiter the
Financiai Reporting Regulators
wealth of the investor and in general aiter the consumption-investment opportuni-
ties available to each investor. Security price changes can be viewed as inducing a The primary concern of the financiai reporting regulators is investors (broadly ttrr
redistribution of wealth among investors. fined to include creditors). The investor orientation is partially motivated by ooo-
Investor behavior can also be affected by what information the investor pos- ce~ over the welfare of the investors and the •'fai.mess" of the security mart.ets in
sessos, by what information is reflected in prices, and by what information is held wbich they trade. Perceived adversities and inequities may happen to investors be-
by others, including other investors and management These factors can poteotially cause of informational deficiencies (for example, failure to disclose). However-, tbe
influence investor behavior with respect to consumptioo, portfolio selection, and regulators (SEC and FASB) also share a concem over the effects on resource aJlo..
costs incurred. including information costs and transactions costs. cation and capital formation. For example, fuller disclosure will tend to lead to á
The perception that prices fully reflect a rich, comprehensive information sys- more efficient allocation of resources because investors will be in a more informed
tem may reduce an individual's direct demand for financiai information. In sucb position to judge wbere their funds can be used most produotively and profitably
cases, the individuais act as "price takers" and are willing to adopt, as their own, the given the risks involved. Some also argue that a more favorable climate for caprt.aj
consensus beliefs reflected in prices. In one sense, they can be "free riders" with re- formation is provided by fuller disclosure because of the effect on the perc::eived
spect to the information reflected in prices, and it may be optimal for them to pursue "faimess" of the marlcet lnvestors are said to be more willing to invest funds in tbe
relatively simple, costless portfolio strategies. ne~-issue marlcet if there is greater disclosure and less risk of fraud or m.isrepresen-
Conversely, the perception that security prices do not fully reflect some infor- tation about the productive opportunities of the furo issuing the securities. Mom-
mation can lead to adopting portfolio strategies designed to reap abnormai expected over, the subsequent "marketability" of the securities is said to be a function of tbe
returns by exploiting the informationai inefficiency. (1) The perceived inefficiency perceived ''faimess" of the excbange marlcets. In other words, if the excbange mar-
can lead the investor to expend time and energy in searcbing for and interpreting ket is efficient witb respect to a óch. comprehensive information system, investors
such informatioo, to pay othecs (such as information intermediaries) to provide such ba~e less concern over information asymmetries at the time they eventually seU
informatioo, or to incur transactions costs pursuing active portfolio management. therr shares and bence are more willing to invest in the new-issue market
(2) It could also cause the investor to adopt less diversified portfolios by investing These arguments have not been developed rigorously and rely on intuitioo.
disproportionately in perceived "underpriced" securities. Moreover, there has been little empirical evidence of the effect of disclosure ou re-
The perception that some information is not fully reflected in prices may
make the investor reluctant to trade in securities if it is felt that the information sys-
I~ Htentwe 011 "lumina from prlces" cifed iD lhe aec:tioo 011 tbooriea of IJI.Irbt ef6ciency U - u-
tem is possessed by others, such as management or other investors. This approacb ample of extrllctin& information from tbe bebavior of ott- uwbt plllticipaum.
 1
I '2 CHAP'reR 6 MARXET EFFIClENCY CHAPTER 6 MARKET EFFlCIENCY 153

source allocation or capital fonnation. A simple relationshíp between the level of The above statement contains two aucial premises: (1) Market efficiency implies
dUclosuro and the rate of capital formation may not exist. For example, increued tbat fundamental research is uaelcu, ud (2) thoe SBC disclosurc syatem requirea tbe
dilclown may oot nooeuarily 1ead to incrouod capital fOI'IJWion. 1bo discloaure anUIDption that such research is u.seful Given these two premises, rccicence by tbe
(tbe particular signal reportcd) may lead investors to reduce tbeir expectations re- SEC Advisory Committee, wbich supports the SBC, to accept tbe efficient marlcet
garding future returns or to increase tbeir assessments of risk. As a result, capital hypothesis is not surprising. Howev~. neither premise is correct. Market efficiency
formation may be lower than it would have been bad tbe information not been pro- in the semistrong form could in fact provide a "friendly" climate for mandated dis-
vided and reflected in security prices. closure. However, given that tbe puception that such premises are valid and given
1be information that is fully reflected in security prices will partially deter- that acceptance of market efficiency is equated with tbe demise of the SEC, market
mine the price effects of any given financial reporting regulation. For example, if efficiency would probably not be accepted by supporters of the SEC.
prices fully reflect ao information system, requiring disclosure of signals from that
system will have no impact on prices, assuming the disclosure is expected to cause
no change in tbe production and financing decisions of tbe finn (and, as a result, no Management
change in beliefs regarding future cash flows). Of course, a price effect could occur Management a1so has ao interest in mark:et efficiency and its effects on the investor.
if the disclosure is expected to alter tbe firm' s future cash flows for tbe reasons The maximization of shareho1derwealth is a commonly cited criterioo for manager-
stated earlier. ial choice behavior. The stewardsbip view implies tbat management bas a responsi-
What are tbe implications of marlret efficiency for the SEC under the assump- bility to act in the interest of the investor, and this is reinforced by the legailiability
tion that the market is not efficient in the strong form but is efficient in the semi- of management under legislation such as the Securities Acts of 1933 and 1934.
strong form? The motivation for requiring disclosure would be essentially to bring Management has a vital intcrest in the price of the firm' s securities. Managt>
private information into the public domain. In a market that is not efficient in the ment compensation in tbe form of stock options and stock ownership is directly re..
strong form, this is a potentially substantive issue, because privately held informa- lated to the performance of stock prices. lndirectly, other forms of compensation
tion is not reflected in prices. Once the data are placed in the public domain, semi- may aiso be affected by stock price performance. The maximization of shareholder
strong form market efficiency provides tbe assurance that such data will be fully re- wealth is often operationally taken to be equivalent to, or at least highly related to,
flected in prices. In other words, requiring public disclosure is ao effective remedy the maximization of share price. The financing decisions of the fino are a prime re-
(at least in terms of security prices) for any perceived undesirable effects associated sponsibility of management The stock price received on a new issue determines ei-
with tbe presence of nonpublicly available data. ther how much proceeds are available for additionai investment (assuming a fixed
In many respects, this pair of assumptions offers the most friendly or most fa- number of shares issued) or how much additionai shares must be issued (assuming a
vorable climate for disclosure regulation from tbe perspective of market efficiency. fixed amount of desired net proceeds). In either case, the fortunes of the current
This does not deny that there are many other considerations as well, some of which shareholders are affected by the price of the new securities.
are discussed in Chapters 2 and 7. Moreover, this pair of assumptions appears to be Financiai reporting is considered to be a prime responsibility of management,
a reasonable description of reality based on ao overall evaluation of the empirical and market efficiency is related to the effect on security prices of management
research discussed earlier. From tbe above anaiysis, one might expect that the SEC choices among different financiai reporting systems. Moreovec, managements,
has considerable incentive to adopt this pair of assumptions. In fact, by its policy of competing among one another for investors' funds, have incentives to provide in-
active disclosure regulation, the SEC can be viewed as acting as if it believes in formation to investors and information intermediaries, among others. Hence. man-
semistrong form efficiency but not strong form efficiency. However, the stated pol- agement has incentive to voluntarily reveal private information (Healy and Palepu
icy position of the SEC makes no explicit reference to this concept. 1993), which can be influenced by costs, such as out-of-pocket costs, competitivo
Moreover, the Report of the SEC Atlvisory Committee on Corporate Disclo- disadvantage to unilateral disclosure of operations, and legailiability.
sure (SEC 1977, p. XXXVIll) discusses market efficiency in essentially negative Market efficiency has poteotial implications for (1) certain management deci-
tones. sions, such as a change in accoonting methods, (2) the legal liabil:ity imposed on
management for nondisclosure, (3) the choice of the levei of complexity of the fi-
nanciai reporting system, (4) tbe concem of investors over moral hazard (for exam-
1be Committee believes that notwithst.anding the interestin.g and clearly signi.ficant ple, fraud) by management, their willingness to invest funds, and hence the price
wodt done by economists and others in developing the efficient market hypothesis, the
evideoces tbat fundamental research is essentially useless are not yet, and may never that they are willing to pay for the firm's securities, (5) the concem of investors
be, sufficiently telling to justify the elimination of a disclosure system premised on the over marketability of the securities at some future time and hence the price they are
proposition tbat such research is useful and necessary. willing to pay now, (6) the choice of format or form of disclosure (in other words,
 ..,
154 CHAPTER 6 MARKET EFFICIENCY
CHAPTER 6 MARKET EFFKlENCY 155
'
supplemental disclosure versus inclusion in tbe financiai statements), and (7) tbe in- preter, and analyzer of infonnation for prediction (prospective analysis), and (3) •
formation content of accruai accounting and eamings determination, relative to interpreter of events after-the-fact (retrospective analysis). Infonnation intermcdi-
other forms of financiai disclosure. aries compete with one another in the gatheri:ng and interp.retation of infonnatioa.
Wyatt (1983) reports that management behaves as if it does not believe the ef- Moreover, information intermediaries compete with reported financiai informaiÍOil
ficient market hypothesis. Wyatt documents several transactions in which manage- in supplying information to tbe capital markets. As discussed earlier, infonnalioo
ment appears to care about the accounting method and is willing to incur costs in intermediaries can be ao important part of tbe process by which information be-
order to structure a transaction to have a desired effect on reported results. Follow- comes reflected in prices. On the other hand, market efficiency may make it more
ing are at least two explanations. difficult for individual intermediaries to reap returns from tbeir analysis activities.
First, in many cases management does not believe security market.s are effi- In the first role, these intermediaries have incentives to seek out and to dis-
cient with respect to publicly available financiai reporting data. Market efficiency seminate information. In the second and third roles, the intermediaries are producers
does not necessarily require a majority of corporate management to believe in il of information. Here more primitive information, such as financiai statement data,
Corporate managements do not tend to be strong advocates of market efficiency. In is an input or factor of production. The conclusions of the analysis, interpretalioo,
many cases, the vaiue of management expertise arises precisely because of some and processing are the output or product of the analysis and are a fonn of
imperfection in the product or factors markets. Given that imperfections are the rai- information.
son d'être of management. it is an easy leap for management to perceive capital Disclosure of an item can have different effects depending on tbe roles of tbe
markets to be inefficient as well. information intermediary. (1) To the extent that the information intermediary is a
Second, management may be motivated to structure a transaction in a particu- seeker of information not fully reflected in prices, publicly available information
lar way for reasons other than perceptions of market inefficiency. As discussed in can constitute a competing source of information. In this sense, information inter-
Chapter 5, many of the firm's contracts, such as management compensation and mediaries may oppose regulations for greater disclosure if they feel that such cJis...
bond covenants, are defmed in terms of accounting numbers. These contracting ef- closures would intrude on their domain and effectively compete with the infoona-
fects can affect management behavior. tion they provide. (2) To the extent that the information intermediary is a proclucer
of information, either of prospective or retrospective analysis, financiai disclosure
Auditors constitutes a factor of production. As a result. the information intermediary mipt
A demand for auditing arises because of a potentiai asymmetry of information be- favor a regulation to increase the disclosure of financiai infonnation because it is
tween management and investors. Auditing financiai information potentially re- tantamount to providing more of a factor of production for the infonnalion
duces that asymmetry and in this sense facilitates market efficiency. The auditor has intermediary.
a responsibility to investors as the independent certifier of the financiai statements Market efficiency can also influence the orientation of information searcb or
prepared by management. Hence, the auditor is concemed about the effects on analysis. (1) As a seeker of information, the information intermediary makes a deci-
investors. sion on whicb sources of information to pursue. Market efficiency can influence tbe
More specifically, market efficiency has severa! potential implications for the sources of information soughl For example, if the market is perceived to be effi-
auditor, including (1) the legailiability associated with nondisclosure, (2) the infor- cient with respect to publicly available information, ao information intennec~Ury
mation content of the accrual accounting system, which is the heart of the account- would concentrate on gathering infonnation thal is not publicly available. 1bc in-
ing system being audited, (3) the resources of the auditor allocated to issues of form formation intermediary will seek out sources thal are not fully reflected in póa:s.
versus substance, (4) the advice given to management with respect to its financiai (2) As a producer of information, the information intermediary can have diverse ob-
reporting decisions, such as a change in accounting method, (5) the assessment of jectives. For example, the analysis could be directed at detecting mispriced securi-
the potential effects of a change in reporting systems and whether it is a preferable ties or it could be directed at assessing tbe risk of different securities. If the infor-
one, and (6) the use of security prices as a means against which to evaluate the thor- mation intennediary felt that the market was efficient with respect to a ócb,
oughness of an audit. comprehensive information system, the information intennediary may choose to di-
rect the anaiysis toward risk assessment rather than towanl finding mi.spric:ed
securities.
lnformation lntermediaries
As indicated previously, the services of information intennediaries are com-
The information intermediary aiso can be affected by market efficiency. The infor- monly sold on the basis that they can select mispriced securities. Clearly, marbtef-
mation intermediary plays many information-related roles, including that of (1) a ficiency can be viewed as a threat to such clai.ms. Market efficiency of the eani-
seeke r of lnfonnation not already fully reflected in prlces, (2) a processor, inter- strong form implies that mispriced securlties do not exist with respect to publicly
.
156 CHAPTER 6 MARKET EFflOENCY CHAPTER 6 MARKET EF'FICIENCY )57

availablc data. An information intermcdiary wbo offers the services on the premise Archíbald, T. "Stock Market Reaction to DqRciation Switch-Baclc." Accounting Review
that misprlced securlties can be selected based on publloly available infonnation is (1anuary 1972), 22-30.
unlikely to embrace semistrong fonn efficiency.18 For example, the anaiyst commu- Bagebot, W. wrbe Only Game in Town." Financiai Analyst.r Journal (Mareh-Aprill971 ).
12--22.
nity bas typically expressed strong opposition to market efficiency.
Ball, R "Changes in Accountiog Tecbniques and Stock Prices." EmpiriaJI RestUJTch in Ac-
coiUiling: Sekcted Studhs. Supplement to lhe Joumal ofAccoUIIIÜlg Rescarch (1972).
1-38.
6-9 CONCLUDING
I
REMARKS - - and P. Brown. "An Empirical Evaluation of Accounting lncome Nombers." Joumol
Tbis chapter explores tqe definition, theory, evidence, implications, and acceptatlce ofAccoiUiling Research (Autumn 1968), 159--178.
of mark.et efficiency. The market is efficient with respect to an infonnation system Beaver, W. '7be Information Content of Annual Eamings Announcements." Empirical Re-
search inAccounting: Selected Studies. Supplement to the Journal ofA ccoiUiling Re-
if prices act as if everyone bas access to that information system. In this sense, search (1968), 67-92.
prices are said to "fully reflect" tbe infonnation system. The origin of market effi- - -. "MarketEfficiency." Accounting Review (January 1981), 23-37.
ciency with respect to financiai information is security analysis. The empirical eVi- - - and R Dukes. "Interperiod Tax Allocation and Delta-Depreciation Metbods: Some
dence arose in response to contentions in tbe financiai and accounting communities Empirical Results." Accounting Review (July 1973), 549--559.
that the market is inefficient With res~ to certain financiai statement infqpnàtion. - - and W . Landsman. "Note on lhe Behavior of Residual Secwi.ty Retums for WlllDCI"
The early evidence was interpreted to be c;onsistent with the·contention that securlty and Loser Portfolios." Joumal of Accounting and Economics (December 1981).
prices respond quicldy and in a sopbisticated manner to financiai statement data. 233-241.
Although the bul.k: of the evidence is consistent with market efficiency with respect Bemard, V., and J. Thomas. "Post-Eamings Announcement Drift: Delayed Price Response or
to publicly available information, some evidence is not so clear-cut. Tbis ~searcb Risk Premium.?" Journal ofAccounling Research (Supplement 1989), 1-48.
bas been labeled anomalous evidence by some, open to a variety of interpretations, - -. "Evidence That Stock Prices Do Not Fully Retlect the Implications of Cuueut Eam-
ings for Future Eamings." Journal of Accounling and Economics (December 1990),
one of wbicb is market inefficiency. 305-340.
Researcb provides various theories of market efficiency. The theoretical and re- Cootner, P., ed. The Random Character of Stock Market Prices. Cambridge: MIT Prcss,
searcb design issues influence tbe acceptance o f various forros of market efficiency by 1964.
tbe researcb comm~ty . Conditional upoo market efficiency o f the semistroog fon:n, Decbow, P., R. SI_o an, andA. Sweeney. "1be Relation Between Affiliated Analysts' Long-
severa! potentiai implications and oonimplications are presented, and their importance Tenn Eamings Forecasts and the Overpricing of Equity Offerings." Unpublisbed
to various constituencies is discussed. The cbapter also examines the 39ceptaoce by the working paper, Wharton, 1996.
financiai reporting constituencies.lt is suggested that the degree of acceptance may not Douglas, W. "Protecting lhe Investor." Yale Review (1933), 523-524.
rest primarily on the tbeoretical and researcb design issues previously discussed but Dyckman. T., and D. Morse. Efficient Capital Mark.et.r and Accounting: A Criticai Analysis.
rather whether tbe aoceptance is viewed to be in the self-interest of that particular 2d ed. Englewood Cliffs, NJ.: Prentice--Hall, 1986.
group. However; disclosure regulation is consistent with market efficiency with re- Fama, E. "Efficient Capital Markets: A Review of Theory and Empirical Work." Joumol of
Finonce (May 1970), 383-417.
spect to publicly available data but marlret inefficiency witb respect to nonpublicly
- -. "Efficient Capital Marlrets: li." Journal of Finance (December 1991), 1575-1617.
available data. Whether recognition veLSUS disclosure is a substantive issue is anotber
- - and K. French. '"'be Cross-Sectioo of Expected Stock Retums." Journal of Finance
matter, aod one that is stilllargely unresolved. Severa! important implications of mar- (July 1992), 427--465.
ket efficiency cut across severa! constituencies. Moreover, there are constituency- Foster, G. Financial Statement Analysis. 2d ed. Englewood Cliffs, N.J.: Prentice-Hall, 1986.
specific implications of market efficiency. - - , C . Olsen, and T. Shevlin_"Eamings Releases, Anomalies and the Behavior of Secu-
rity Returns." A ccounling Review (October 1984), 574-{i()3.
Graham, B., and D . Dodd. Security Analysis. New York: McGraw-Hill, 1934.
BIBLIOGRAPHY Grossman, S. "On the Efficiency of Competitive Stock Markets Wbere Traders Have Diverse
Abaibanell, J., and V. Bemard. "Tests of Aoalysts' Overreactioo!Underreaction to Eamings Information." Journal of FiNJnce (May 1976), 573-585.
Information as an Bxplanation for Anomalous Stock: Price Behavior." Journal of Fi- - - and J. Stiglitz. "On the Impossibility of Infonnationally Efficient Markets." Ameri-
nance (July 1992), 1181-1207. ClUI & onomic Review (June 1980), 393-408.

Healy, P., and K. Palepu. "''he Effect of FlJlliS' Financiai Disclosure Strategies on Stock
11
0f course, the informatioo intmnediary claims an ability to select mispriced secupties based oo oon- Prices." Accounting Horir.ons (March 1993), 1- 11.
publicly available infonnatioo. However, sucb claims potcotially cxposc tbc analyst to legalliability. Kripke, H. "The Myth of the InfoiiilCd Layman." Business IAwyer (January 1973), 631 ~ 8.
 ...

I 58 CHAPTER 6 MARKET Ef'FIOENCV

Lakonisbok, I., A. Shleifer, and R. Vishny. "Contrarian Investment, Bxtrapolation, and

Risk." Journal of Firuuu:e (December 1994), 1541-1578.
Lev, B., and R. Thiagarajan. "FUndamental Infonnation Analysis." Journal of Accounting
Research (Autumn 1993), 109-215.
Libby, R. Accounting and Human lnformation Processing TMory and Applications. Eogle-
7
wood Cliffs, NJ.: PJentice-Hall, 1981.
Loughran. T ., and J. Ritter. "Tbe New lssues Puzzle." Journal of Firuuu:e (March 1995),
23-51.
Morse, D. "Price and Trading Volume Reaction Surrounding Barnings Announcements: A
Closer Ex.amination." Journal ofAccounting Research (Autumn 1981), 374-384.
Regulation
Patell, J., and M. Wolfson. 'The lntraday Speed of Adjustment of Stock Prices to Barnings
and Dividend Announcements." Joumal of Accounting and Economics (June 1984),
223-252.
Securities and Bxchange Commission. Report of the SEC Advisory Committee on Corporate
Disclosure. Washington, D.C.: U.S. Government Printing Office, 1977.
Shores, D. 'The Association Between Interim Information and Security Retums Surrounding
Barnings Announoements." Joumal ofAccounting Research (Spring 1990), 164-181.
Sloan, R. "Do Stock Prices Reflect Information in Accruals and Cash Flows About Future
Barnings?" Accounting Review (July 1996), 289-315.
Treynor, J. "Bfficient Markets and Fundamental Analysis." Firuuu:ial Analysts Joumal
(Marcb-April1974), 14.
Verrecchia. R. "On the Theory of Market Information Efficiency." Joumal of Accounting
and Economics (March 1979), 77-90. Fmanciai reporting takes place in a regulated environment. Given that regulation is
Wyatt, A. "Efficient Madcet Theory: Its lmpact on Accounting." Joumal of Accounuuu:y an important part of the financiai reporting environment. this chapter explores sucb
(February 1983), 56-65. issues as: Why regulate7 How should regulation be conducted7 Who sbould regu-
lare (the SEC or the FASB)? 1 In doing so, tbe discussion will serve to synthesize the
major concepts of the framework developed in the previous chapters.

7~ 1 MANDATED FINANCIAL REPORTING

The regulation of financiai reporting receives its impetus from the Securities Acts
of 1933 and 1934, whlch gave the SEC statutory power to ensure "full and fair dis-
closure" by corporations issuing securities on an interstate basis. The Acts specifi-
cally grant the SEC tbe power to determine the accounting standards for reports
filed with the SEC. The SEC delegates autbority to the FASB to determine gener-
ally accepted accounting principies with respect to statements filed with the SEC.l
Often the illustrations refer only to the SEC. However, in general. the discussion of
the rationale for regulation applies to the FASB as well as to the SEC.

1
This chaptec draws beavily on Cbapter XX of tbe SEC ÀIIVISory Committee Report on Corporau Di.r-
clo.swre (SEC 1977}. An excellent review of finaocial reporting re&Watioo alao appean in Cbaptec 7 of
Watts anel Zimmennan (1986).
lwbile lhe SBC bu delepted auch authority Ul lhe PASB,Ihe SEC bu ofteo cboeen to exerdle ~over­
sight" witb respect to FASB decisions (aee Homgxen 1972. 1973; Armstrong 1977; amona otbus).