Logic:

1

• Logic
• Search costless information permits investors to
  continually search for superior return investment
  opportunities
• Allocation and Action costless trading permits
  investor to capitalize on discovery of such
  opportunities, leading to instantaneous changes
  in supply or demand from the investor who hopes
  to profit from his/her insight
• Aggregation all investors changes in supply and
  demand are summed at the trading post by the
  market maker to determine whether aggregate
  supply and demand are equal at the current
  bid/ask prices. If not,
• New Price Quoted by MM in a continuous market,
  and without misrepresentation of the changes in
  supply and demand (without bias) the market maker
  establishes a new price to clear the market given
  the offers to buy and sell that have been
  submitted.
• New Allocation, Action and Aggregation investors
  reconsider their supply and demand offers in
  relation to the new price set and revise their
  offers to buy and sell. These new offers come to
  the market maker, who once again attempts to
  establish a market clearing price
• New Price Quoted The process continues until a
  price is set that is consistent with all
  investors optimal investment allocations (e.g. no
  further excess supply or excess demand at the now
  current price). That is the
• Equilibrium Price or the market clearing price
  that establishes excess supply equals excess
  demand equals zero. All trades are executed.

2

• Conclusions from the economic logic
• 1. Instantaneous and Unbiased Price Changes
  Prices will change instantaneously and unbiasedly
  in response to new publicly available
  information, moving from one equilibrium level to
  another.
• 2. Prices Fully Reflect Information Price
  levels, therefore, will tend to reflect market
  "consensus expectations" formed from all existing
  publicly available information.
• 3. Statistically Independent Price Changes and
  Returns Because of the two points above,
  sequential price changes will be statistically
  independent over time and will appear as a
  "random walk" or as "white noise." They are
  independent random drawings from the urn.
• 4. Uselessness of Public Information Some
  investors may earn "superior" returns in the
  short run using public information, but, on
  average, no investors using publicly available
  information could generate above average (risk
  adjusted) returns.

3

• Practical Implications of the Theory for
  Interpreting Actual Market Price Changes
• Does this mean that no one should be able to make
  above average returns -- what about Peter Lynch
  and all the other market gurus?
• How much competition (how many investors) is
  enough to move the market efficiently?
• With no excess return opportunities the incentive
  to search public files for information all but
  disappears -- can we all be dart throwers?
• Does the theory mean that all investment
  opportunities are, in some sense, correctly
  valued by the market?
• If security markets are efficient, why do we
  presume that the markets for real goods and
  services (e.g., factors of production) are
  inefficient in a manner that allows firms to
  capture the value added from investment
  opportunities with a positive net present value?

4

• Tests to see if the theory conforms to the
  realities of the marketplace
• Do prices change instantly and unbiasedly when
  new information about an event is signaled to the
  market? (TEST 1 are abnormal returns different
  from zero only when new informational signals are
  given to the market)
• Do prices incorporate all available public
  information? (TEST 2 are abnormal returns
  unrelated to past informational signals or past
  events)
• Do price changes (or returns) follow a pattern
  that looks like "white noise" or a "random walk"?
  (TEST 3 are abnormal returns unrelated to past
  returns and to other unimportant trading
  characteristics (volume, trading day)
• Do some investors generate superior returns, and
  why? (TEST 4 are abnormal near zero for large
  sub-groups of investors -- pension fund managers,
  individual investors, etc..)

5

Some market observations Are they evidence of
EMH?

• Shares of Western Digital, an Irvine, Calif.,
  maker of computer disk drives, fell sharply
  yesterday. On Wednesday, after the market
  closed, Western said that its gross margins in
  the second quarter had declines to 20 percent
  from 21 percent in the first quarter. Fell 2 3/4
  to 16.50 has been as high as 19 1/4 in recent
  days.
• Three computer companies fell on a downgrade from
  Smith Barney. Apple slipped 3/4, almost 2, to
  38 1/4 after being lowered to "outperform" from
  "buy," while Compaq stumbled 7/8, or 2.5, to
  333/4 and Dell Computer slid 2 1/8, or 5, to 41
  1/2 after being downgraded to "underperform" from
  "neutral."
• Kmart fell 7/8, or 6.3, to 12 7/8 after posting
  fourth-quarter profit of 27 cents, sharply lower
  than the 48-cent mean estimate of a team of
  analysts surveyed by First Call. Industry
  analysts reportedly believe that Kmart will
  continue to struggle with competition from retail
  rivals.

6

• Amgen, also on Nasdaq, advanced 33/4, or 5.7, to
  69 after a positive story in this newspaper said
  that the Thousand Oaks, Calif., biotechnology
  company had had strong drug sales in the past
  year. The stock was included in S.G. Warburg's
  research report sent out to clients, which rates
  the stock an "add" and notes that its target
  price by year's end is 72.
• Altera rallied sharply, 33/8, or 6.3, to 56 7/8.
  The San Jose, Calif., concern said late Monday
  that February month-to-date order rates for its
  integrated circuits has exceeded company
  expectations. Needham raised 1995 earnings
  estimates for Altera by 25 cents a share to
  2.50.
• Walt Disney Co. shares gave up 1 1/4 to 55 3/4.
  The Burbank, Calif., entertainment concern is
  reportedly losing another top executive. The New
  York Times said Richard Frank, head of Disney's
  television programming business, intends to leave
  soon. Frank has denied recent rumors that he
  will depart.
• Shares of Celex Group, a specialty retailer and
  catagogue company, dropped sharply in Nasdaq
  trading yesterday. The company, based in
  Lombard, Ill., announced that November and
  December sales were below Wall Street
  projections. Fell 4 1/4 to 10 1/4.

7

• One technology stock that performed well even
  though its better-than-expected earnings were
  announced after the market closed was Microsoft.
  The software giant added 4 3/8 to 91 1/8. After
  trading, it announced fiscal first-quarter
  earnings of 78 cents a share, or eight cents more
  than expected. It was trading 1 1/2 higher in
  after market trading on Reuters Instinet.
• Sun Microsystems (Nasdaq), jumped 9 1/4 to 67
  7/8. The workstations maker reported
  surprisingly strong fiscal first-quarter
  earnings. Its 85 cents a share in net income,
  compared with 40 cents for the year-earlier
  period, beat estimates of 65 cents.
• Cypress Semiconductor, surged 3 3/4 to 36. The
  chip maker posted third-quarter net of 67 cents a
  diluted share, versus 32 cents for the
  year-earlier period. Wall Street estimates were
  for 59 cents.
• Foundation Health, moved ahead 3 to 39 3/4. An
  analyst at Salmon Brothers initiated the firm's
  coverage of the managed-care services provider
  with a "buy" rating.
• Marriott advanced 13/8 to 32 5/8. The company
  confirmed it agreed to buy a minority stake in
  Ritz-Carlton Hotel.

8

• Apple Computer edged higher despite denials from
  the companys chairman that the computer maker
  was on the verge of being acquired. Apple
  (Nasdaq) improved 5/8 to 32 1/4. Apples suitor,
  Sun Microsystems, (NASDAQ) slipped 5/8 to 43 1/2.
  The computer makers former suitors all gained in
  the session, with Oracle (NASDAQ) climbing 2 7/8
  to a 52-week high of 49 1/8. Hewlett-Packard
  shot up 3 ½ to 84. International Business
  Machines jumped 4 to 107.
• Boeing couldnt scratch out a gain on what, at
  first blush, looked like a stellar fourth-quarter
  performance. The raw numbers showed earnings
  doubled analysts forecasts, but a closer look
  revealed that the performance owed a lot to tax
  credits and lower-than-expected research and
  development costs. Boeing finished flat at 77.
• R.P. Scherer fell 3 7/8 to 38. The company that
  makes oral drug-delivery systems, posted fiscal
  third-quarter net income of 54 cents a share,
  compared with 47 cents a share in the
  year-earlier quarter. Analysts were expecting
  the company to earn 55 cents a share.

9

• Tests to see if the theory conforms to the
  realities of the marketplace
• Do prices change instantly and unbiasedly when
  new information about an event is signaled to the
  market? (TEST 1 are abnormal returns different
  from zero only when new informational signals are
  given to the market)
• Do prices incorporate all available public
  information? (TEST 2 are abnormal returns
  unrelated to past informational signals or past
  events)
• Do price changes (or returns) follow a pattern
  that looks like "white noise" or a "random walk"?
  (TEST 3 are abnormal returns unrelated to past
  returns and to other unimportant trading
  characteristics (volume, trading day)
• Do some investors generate superior returns, and
  why? (TEST 4 are abnormal near zero for large
  sub-groups of investors -- pension fund managers,
  individual investors, etc..)

10
EMH TEST 1are returns related to informational
signals?

• abnormal returns Ut rt - E(rt)
• E(rt) are returns predicted by CAPM or some other
  models
• tendency for abnormal returns to occur at the
  time of the signal conveying new information to
  the market
• no tendency for abnormal returns to be different
  from zero when no new information is signaled
• no tendency for abnormal returns ( or -) to
  continue occurring subsequent to the signal

11
Difficulties in studying price impact of
events/signals
12
CARs before and after signals of dividend changes
13
EMH TEST 3 are returns related to past returns
or to non-events?

• no tendency for next periods abnormal returns to
  be related to past abnormal returns. Ut1 not
  related to Ut or Ut-1
• no tendency for abnormal returns to be related to
  non-value related events -- day of week, month of
  year, end of month days, sun spots, hemlines.
  Ut1 not related to these events
• remember in all price changes there is an
  expected component, E(r), and an unexpected
  component, U. The theory focuses on U, but
  sometimes measures use r and change in P.

14
(No Transcript)
15
(No Transcript)
16
abnormal returns in day t compared with those in
day t1
17
abnormal returns after large price declines
experienced by a security during a days trading
(2800 NYSE, 1400 Nasdaq -- 1963-91)
18
serial correlation of weekly returns (the
comovement of returns in t with those in ts)
is Ut related to Uts?
19
returns, rt, by month-of-year mean monthly
return 1887-1986
20
returns, rt, by day-of-the-week annualized by
days in year
21
Cumulative intra-day returns
22
The Lesson are markets efficient?

• TEST 3 based on past returns -- mostly efficient
• TEST 3 based on calendar events -- some
  anomalies, especially day-of-week, and
  turn-of-year
• TEST 1 based on mergers, dividends, management
  changes, repurchases, splits -- mostly efficient
  
• TEST 1 based on earnings surprises, analysts
  earnings forecasts, NYSE listing, insider trading
  -- some anomalies
• TEST 4 based on mutual fund and pension fund
  performance -- mostly efficient