. 25
( 39)


cost $50.50”perhaps still worthwhile, but less pro¬table. And purchasing the remaining
19,000 shares may cost you $51 or more.

3. By the time you have shorted the shares in Frankfurt at $51, the price in New York may
have risen to $52. If such execution timing risk exists, this is not pure arbitrage because
¬le=e¬mkts.tex: RP
Section 19·2. Market E¬ciency and Behavioral Finance.

there is a chance of a negative out¬‚ow. The real-world evidence suggests that price dis-
crepancies between markets often disappear within a few seconds.

4. You would also have to account for your ¬xed cost of executing this transaction, such as
setting up your own computer operation to do quick arbitrage-like transactions.

My belief is that in the real world, small arbitrage opportunities can occur from time to time, but
large ¬nancial ¬rms are constantly running automated computer trading programs that search
for even tiny arbitrage opportunities in order to exploit them as soon as they appear”and
thereby make them disappear.
The concept of arbitrage is di¬erent from the concept of a good bet. A good bet would be a The difference between
arbitrage and a good
chance to win $1,000,000 with 99% probability and to lose $1 with 1% probability. But because
there is a chance of losing money in some circumstances, this is “just” a great bet. It is not
an arbitrage. The di¬erence is important: everyone would want to take advantage of arbitrage
opportunities, but someone su¬ciently risk-averse may not like a good bet, even if it is an
absolutely wonderful bet. Conversely, a limited arbitrage need not be better than a good bet.
For example, a single 1 cent arbitrage that cannot be repeated could be a worse bet than the
aforementioned $1 million gain, $1 loss bet.
Unless ¬nancial markets are exceedingly strange, we would not expect to ¬nd either arbitrage There are probably
neither great bets nor
opportunities or great bets. If you agree with this assessment”basically that money does not
arbitrage in very
grow on trees”we can draw some surprisingly strong conclusions about how ¬nancial markets competitive ¬nancial
work. If you disagree, you should not be sitting in class, but somewhere on a beach, ranking markets.
among the richest people in the world. There is little this book can teach you.
Solve Now!
Q 19.1 Explain when and why you would prefer a good bet to an arbitrage opportunity.

19·2. Market Ef¬ciency and Behavioral Finance

Warning: Market E¬ciency is a di¬erent concept from Mean-
Variance E¬ciency. The reuse of the word “e¬ciency” is unfortunate.

19·2.A. Basic De¬nition and Requirements

Formally, ¬nancial economists call a market e¬cient when it uses all available information in Market Ef¬ciency means
that markets use all
its price setting. Thus, market e¬ciency is the degree to which prices re¬‚ect information. In
a fully e¬cient market, you cannot use available information to predict future returns better
than the market can. Unfortunately, this leaves the question vague as to where the market
wants to set expected returns. For example, the CAPM might state that the expected rate of
return on PepsiCo should be 10% (setting a price of $50 for an expected payo¬ of $55), but
you as an investor could determine when the current price of PepsiCo really o¬ered a rate of

Anecdote: Trading Places and Citrus Futures
The 1983 hit comedy Trading Places, starring Dan Akroyd and Eddie Murphy, centers around the trading of
Orange Juice Frozen Concentrate Futures Contracts (securities that promise delivery of oranges) on the New
York Futures Exchange. If it is going to rain or there is a frost, oranges will be scarcer and the futures price
will rise. You can learn more about futures contracts at the website of the New York Mercantile Exchange at
In a 1984 paper in the American Economic Review, Richard Roll found that these citrus futures contracts predict
whether the U.S. Weather Service™s forecast for central Florida temperatures is too high or too low. It is a great
example of how ¬nancial markets help aggregate information better than the best non-¬nancial institution. This
should not be a surprise. After all, there is a lot of money at stake!
¬le=e¬mkts.tex: LP
480 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

return of 20% (an expected payo¬ of $60). You could now draw one of two conclusions: ¬rst,
the CAPM is not the correct model, and the market wanted to set the expected rate of return
for PepsiCo at 20% in the ¬rst place; second, the stock market is not e¬cient. In a sense, the
problem with market e¬ciency is that in many circumstances it is almost a matter of faith: if
you wish to proclaim a belief in market e¬ciency, and if you then ¬nd that “prices are o¬,” you
can still always claim that your model for the appropriate expected returns in ¬nancial markets
was wrong if you want to deny that the market was ine¬cient.
Even though stock market e¬ciency is a fairly modest claim”at least as long as we remain vague
on what the correct model of appropriate expected rates of return is”it is still a surprisingly
useful concept. For example, it is pretty safe to say that no model of ¬nancial markets is likely
to claim that investors can ¬nd great bets “+$1 million with 99% probability” and “’$1 with 1%
probability.” Such an expected return would be way out of line. Even expected rates of return
of 100% per year are surely unreasonable for stocks such as PepsiCo. Few people doubt that
the stock market is, to such a ¬rst approximation, e¬cient. Still, there is a large gray zone:
we do not know the correct model of expected stock returns well enough to know if the stock
market set the price of PepsiCo stock so as to o¬er an expected rate of return on PepsiCo of
10% a year or 12% a year.
Market e¬ciency is intimately related to our perfect markets concept from Chapter 6. It leans
Market Perfection and
Market Ef¬ciency. particularly heavily on the assumptions that there are no transaction costs. That is, even in
the presence of some taxes and opinion di¬erences, if it is just cheap enough to arbitrage
mispricings, someone will end up doing so. Conversely, it is easier to believe that markets are
not (or less) e¬cient if transaction costs are high. If it costs nothing to trade stocks, it would
be easy for any investor to trade on any information that the market has not yet incorporated
in the stock price”and thereby to earn an unusually good expected rate of return or even an
arbitrage. However, the no-free-lunch axiom applies here, too. Low trading costs would make it
less likely that you could expect to ¬nd violations of e¬cient markets. But if it is very expensive
to trade and therefore if the market is not e¬cient and does not respond to news immediately,
it would also be very di¬cult for you to take advantage of such ine¬ciencies. Of course, we
also already know from Chapter 6 that if the market is not perfect, it is not even clear what
“value” means. There would be a whole range of possible values for ¬nancial securities, both
now and in the future. As we learned, no market is perfect or perfectly imperfect”market
perfection is always a shade of gray. Thus, the range of possible valuations is determined by
the extent to which the market is imperfect.
In any case, modern ¬nancial markets for large corporate stocks and index funds in the United
We can assume
reasonably ef¬cient States seem very competitive. There are millions of buyers and sellers, transaction costs are
markets for large
low, and few investors know in advance whether the market will go up or down. It is di¬cult
corporate stocks.
to believe that you or I could outsmart the prices in such markets. After all, thousands of other
traders are likely equally as smart and would ¬‚ock to good bargains and avoid bad bargains
along with us. Of course, the smaller the ¬rm, the less perfect and the less e¬cient the market in
its stock is likely to be. Most stocks on Nasdaq trade only rarely, and can have large transaction
costs. (Not only are the bid-ask spread and commissions often very high [which is the instant
cost of a roundtrip transaction], but it may be particularly di¬cult and expensive to short these
stocks, i.e., speculate that they will decline.) It is unlikely that these stocks will immediately
and fully re¬‚ect all information appropriately. So, market e¬ciency is never white or black, but
always a shade of gray”just as it is for perfect markets.
One conceptual question that vexed academics for a long time was how markets can be e¬cient
Noise Traders.
to begin with. After all, if there is no money to be made, why would anyone bother collecting
information on ¬rms? And if no one bothers to collect information on ¬rms, how can the market
be e¬cient? Eventually, the resolution to this puzzle was that markets can never be 100%
e¬cient”they can only be, say, “99%” e¬cient. In equilibrium, good information collectors
should earn just about enough trading pro¬ts to break even on their costs of information
collecting. They earn this money trading against noise traders, who do not collect information
and who may trade for idiosyncratic reasons (e.g., to pay for a new car).
¬le=e¬mkts.tex: RP
Section 19·2. Market E¬ciency and Behavioral Finance.

We will discuss consequences of market e¬ciency below, but one we can mention right away. In an ef¬cient market,
the announcement stock
The fact that large-¬rm stock markets are pretty e¬cient means that, by and large, you can
reaction should be a
trust these ¬nancial markets to get asset values about right”at least within the limits of the good estimate of the
arbitrage transaction costs listed in the previous section”and to get it right immediately. change in NPV, because
the market should
Would you not rather face an ine¬cient market? If it were ine¬cient, you might be able to accurately re¬‚ect value
at all times.
¬nd some good bets (opportunities that earn unusually high expected rates of returns). But
The advantage of an
it would not all be gravy. In an ine¬cient market, you could not rely on market prices being ef¬cient market.
fair”they could be inappropriately too high or too low. You would never really know whether
you are overpaying or underpaying. Investing would be a very messy business. The advantage
of e¬cient markets is that if you hold a portfolio of many large and liquid stocks, you do not
have to spend a lot of time and money to perform due diligence in order to determine whether
stocks are fairly priced. All you need to do is to make sure you are appropriately diversi¬ed to
meet your risk-reward preference. You can probably accomplish this goal by purchasing just
a few large index-mimicking portfolios.

19·2.B. Classi¬cations Of Market E¬ciency Beliefs

Almost all ¬nancial economists believe in basic market e¬ciency for large markets and liquid Financial markets are
probably close to
securities. No respectable economist believes that it is easy to get very rich trading on easily
available information. Instead, the disagreement is, loosely, about whether stock markets are
“99% e¬cient” or “97% e¬cient.” The school of thought that proposes the 99% view is often
called Classical Finance or Rational Finance; the school of thought that proposes the 97% view
is often called Behavioral Finance. Of course, you can trade millions of dollars in large ¬rm
stocks or market indexes relatively easily and at low transaction costs. Thus, it does not require
huge e¬ciency violations for behavioral ¬nance economists to be right and for classical ¬nance
economists to be wrong. Exploiting just the tiny”say, 3%”violations from market e¬ciency
could make you a star investor. (This is also not coincidentally why so many fund managers
show great interest and publicly proclaim their faith in behavioral ¬nance.) However, don™t take
me too literally here”the 99% vs. 97% is an analogy, and there is really a spectrum of beliefs
in market e¬ciency among economists and fund managers. Let us now look at some such
rough groupings, although you should realize that any classi¬cation schemes really identify
just segments on a continuous line.

19·2.C. The Fundamentals Based Classi¬cation

I like to grade ¬nancial economists into camps based on their degrees of belief in market My preferred taxonomy.

True Believer Financial prices always re¬‚ect the best estimate of net present value of all future
cash ¬‚ows. This means that stock prices should change only if news about fundamentals

Firm Believer Financial prices may sometimes deviate from the appropriate best estimate of
future cash ¬‚ows. However, transaction costs make it practically impossible to ¬nd un-
usually good bets.
A special form of this argument, named after Grossman and Stiglitz, resolves the para-
dox that in an e¬cient market, it would not be worthwhile for any investor to pay for
information”but if no one pays for information, how can the market be e¬cient? This
argument is that markets are so competitive that the expected costs to learning and trad-
ing on more information are exactly equal to the trading pro¬ts.

Mild Believer Financial prices occasionally deviate from the appropriate best estimate of fu-
ture cash ¬‚ows (and the ¬nancial price next period). When they do, the transaction costs
are not high enough to prevent investors from obtaining unusually good bets, although the
pro¬tabilities of these bets generally remain within economically reasonable magnitudes”
maybe a couple of percentage points a year.
¬le=e¬mkts.tex: LP
482 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

Non Believer Financial prices regularly deviate from the appropriate value, and thereby allow
investors to obtain arbitrage opportunities or incredibly great bets.

A ¬rm believer need not be a true believer: ¬nancial price changes may indeed be unpredictable,
but not because of news about fundamentals. (There could be unrelated noise in stock price
changes, especially in the short-run.) Occasionally, we are even handed evidence that refutes the
true believer”but only in certain speci¬c situations. The most dramatic such example occurred
in 2000, when the network company 3COM spun o¬ the PDA company Palm. 3COM retained
95% of Palm™s stock”and announced that each shareholder of 3COM would soon receive 1.525
shares of Palm. After the IPO, Palm closed at $95.06 per share. Therefore, 3COM should have
been worth at least $145”instead, 3COM shares closed at $81.81. (It was impossible to exploit
this discrepancy, because it was impossible to ¬nd Palm shares to short. I know”I tried.) A
mild believer need not be a ¬rm believer: transaction costs may be low enough to permit great
trading strategies based on E-M violation. A non-believer need not be a mild believer: ¬nancial
markets may just beg to be exploited.
In this classi¬cation of market e¬ciency, virtually no academic is a non-believer, and only a
Our best estimate is that
we are somewhere very few remain in the true believer camp. Instead, most ¬nance professors are somewhere
between the mild and
between the “mild believer” camp (the center of behavioral ¬nance) and the “¬rm believer”
¬rm believer camps.
camp (the center of classical ¬nance). The debates between the two more extreme side of these
camps”the “rationalists” and “behavioralists””is intellectually exciting. After all, bringing
new evidence to bear on these disagreements is the process by which we learn more.
Setting the facts aside for a moment, let me tell you my personal views. I sit right in the middle
My opinion.
between the two schools of thought, somewhere in the ¬rm-to-mild camp. In my view, most
investors believe that they have more knowledge and control than they actually have. This is
why I believe that trading in the stock market seems so (inexplicably) active. Investors seem to
believe that they can predict when stocks are going to go up or down. Some pundits like to call
this investor psychology. However, I also believe that an individual investor is unlikely to be
able to ¬nd rate of return patterns in the stock market to earn high excess returns. A very few
sophisticated funds may be able to systematically earn a few extra basis points per year. But
these funds are scarce. Even after decades of academic research to identify better performing
funds, academics usually ¬nd that only about half of all funds outperform the market and half
underperform the market”and even before fund transaction costs.

Anecdote: The Limits of Arbitrage in the Internet Bubble
In May 2005, an Experian-Gallup national survey ¬nds that 65 percent of Americans haven™t heard anything
about a possible “housing bubble.” Another 12 percent have heard “only a little.” Indeed, 70 percent expect
home prices to keep rising, while only 5 percent think they will slip. However, when the facets of a housing
bubble are described to them, about 40 percent go on to say that the scenario is likely to occur in their area in
the next three years.
Source: Business 2.0 at cnn.com.

Even in cases where it is probable that the market mispriced stocks, such as technology stocks during the
famous “Internet bubble” at the turn of the millennium, it was almost impossible for an individual investor to
take advantage of the market ine¬ciency. Believe me, I know. In 1999, I believed Yahoo! (YHOO) was worth
less than what it was trading for. So, I speculated that its stock would go down. After I had lost more than
three times my original investment, I realized that I had to either close my bet or risk personal bankruptcy. So, I
terminated my bet, having lost a lot of money. Yes, I would have been right in the end and made a lot of money
if I had held on longer, but I simply could not a¬ord the risk (and mental anguish) any longer. I learned from
this episode”after 15 years as a ¬nancial economist”that even if the stock market is irrational and even if it
overvalues a stock by three times, it can also be irrational enough to overvalue it by yet another three times.

Anecdote: A Conversation with Eugene Fama
The book website has an impromptu email conversation between myself and Eugene Fama (per-
haps the most famous ¬nance professor alive and a strong defender of market e¬ciency) at
welch.econ.brown.edu/academics/famaconversation.html. This will give you an authentic impression of the
ongoing dialogue among ¬nance professors.
¬le=e¬mkts.tex: RP
Section 19·2. Market E¬ciency and Behavioral Finance.

So why is this debate so tough to settle? The reason is that the signal-to-noise ratio in ¬nancial The low signal-to-noise
ratio causes the dispute.
returns is low. The signal-to-noise description draws on an analogy from physics”the signal
(the appropriate average price change that a smart fund manager could predict) is small com-
pared to the noise (the day-to-day price volatility that clouds our senses). Here it means that
a typical stock may have the signal of an expected rate of return of 0.05% per trading day (14%
per year), but the noise of a typical standard deviation of 2-3% per trading day, which is about
50 times as high. This low signal-to-noise means it is di¬cult for researchers to determine
whether a particular trading strategy has earned high returns [a] because it took on risk, and
the researcher has just not recognized the risk appropriately; [b] because it had a lucky out-
come, which will not repeat; or [c] because the market was ine¬cient. Although these choices
allow us ¬nance professors to continue to write papers to argue one side or the other, most
¬nance professors now agree that when individuals earn an unusual amount of money in a day
or a week, it is more likely due to luck than to ability. The burden of proof is on the side claim-
ing superior ability”and a number of former ¬nance professors have taken up the challenge
and started their own funds.

Important: On a typical day, the typical stock moves up or down by about 10 to
100 times as much as it o¬ers in expected rate of return. Therefore, it is not easy
to attribute past observed stock price performance to investor ability or inability.

Digging Deeper:
Our example was about a model that states that a particular kind of stock should increase by 5 basis points, but
has a 200 basis point volatility. If the noise is uncorrelated (which usually means returns on di¬erent days), how
many trading days would we need to determine whether the true expected rate of return is 5 basis points?
If you have T days, your volatility will decrease with the square-root of T. With 10,000 trading days (about 40
years), the volatility would be roughly 10, 000 · 2% ≈ 2 basis points. This level of volatility would allow you
to determine whether the daily expected rate of return on your stocks is 1 basis point, 5 basis points, or 9 basis
points, but not whether it is 4 basis points or 6 basis points. Of course, 4 basis points per day is a whopping 3%
per year di¬erent from 5 basis points per day. To tell apart the di¬erence between 4 basis points and 5 basis
points, you would want no larger a volatility than about 0.5 basis points”requiring about 160,000 independent
observations. We do not have these 600 years of return history, and even if we did, who would believe that these
daily returns were still drawn from the same distribution? So, we usually have to work with “tricks””primarily
forming portfolios that have less than 200 basis points volatility on an average day, which leads to arguments
about what proper portfolios for testing market e¬ciency are. Can you see now why testing for whether stock
returns follow one or the other model is such a di¬cult and contentious task?

19·2.D. The Traditional Classi¬cation

In contrast to my de¬nition of market e¬ciency above, which focuses on how rational market The traditional
classi¬cation of market
prices re¬‚ect underlying values, the more standard historical de¬nition of market e¬ciency
focuses on information. This distinction is between weak-form, semi-strong-form, and strong-
form market e¬ciency.

• Weak Market E¬ciency presumes that markets are e¬cient enough not to allow the use
of historical stock price information to earn inappropriately high rates of return. This
means that technical analysis (trading based solely on historical price patterns) would
not earn excess returns. Put another way, the weak form assumes that all past prices of
a stock are re¬‚ected in today™s price so that technical analysis cannot be used to beat the

• Semi-Strong Market E¬ciency presumes that markets are e¬cient enough not to allow
the use of any publicly available information to earn inappropriately high rates of return.
This means that fundamental trading (trading based on price and underlying ¬rm infor-
mation) would not earn excess returns. Put another way, the semi-strong form assumes
that all public information is re¬‚ected in today™s stock price, so that neither fundamental
trading nor technical analysis can be used to beat the market.
¬le=e¬mkts.tex: LP
484 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

• Strong Market E¬ciency presumes that the market incorporates even the most private
information held by the deepest insiders in corporations. This means that no trading
would earn excess returns. Put another way, the strong form assumes that all information,
both public and private, is re¬‚ected in today™s stock price, so that nothing”not even
insider information”can be used to beat the market.

In this classi¬cation of market e¬ciency, all ¬nance professors believe that most large ¬nancial
markets are not strong-form e¬cient: insider trading may be illegal, but it works. However,
arguments rage on as to whether markets are semi-strong-form or even weak-form e¬cient, and
even for large and liquid ¬nancial markets (such as the N.Y.S.E., Nasdaq, or the CBOE). Finance
professors regularly publish papers that ¬nd new rules that seem to outperform reasonable
average rates of return by a large margin. Some strategies seem to work, in particular some
forms of momentum (buying stocks that have gone up, selling stocks that have gone down) and
value (buying boring old-economy stocks, selling glamour high-growth new-economy stocks).
Such strategies can o¬er seeming “excess returns” as high as 1-2% per month. Unfortunately,
many strategies disappear almost as quickly as they are discovered”and may have never been
real to begin with. Yet other trading strategies require such high transaction costs that they end
up not being pro¬table in the real world. That is, even though prices may not incorporate all
information and the market may not be e¬cient, the ine¬ciency may be well within the bounds
of transaction costs. Yet some other trading strategies seem to have worked and continue to
work”but why and for how long? Personally, I am not claiming that none of these trading
strategies works. I am just advising caution when real money is at stake.
Solve Now!
Q 19.2 What does it mean for a stock market to be e¬cient?

Q 19.3 Is it more or less likely for a ¬nancial market to be e¬cient when transaction costs are
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

19·3. Ef¬cient Market Consequences

We already mentioned one consequence of e¬cient markets: You can trust prices and don™t
have to waste much time checking that they are appropriate. However, there are a number of
other important consequences that deserve further expounding.

19·3.A. Stock Prices and Random Walks

Here are some trick questions: Look at the various graphs in Figure 19.1. They seem to show Cycles in the market?
what stock market patterns have looked like, do they not? Perhaps. Does it make sense to
think that these are representative for the future? Absolutely not! Graphs A, B, and C display
a strong regular cycling pattern. If this were representative for the future, you would quickly
become a wealthy technical analyst! The patterns would suggest that you should purchase the
stock only when it has “bottomed out””a pattern that you can reasonably detect if you see
a multi-month period of losses followed by about a quarter of stable returns. It need not be
the kind of regular cycles in the ¬gure: any good predictable patterns (such as “every time the
price hits $22, it drops by $2”) would allow you to get rich. Now, if you look hard enough, can
you ¬nd some stocks in the real world that have behaved like these graphs? Yes”because with
over 10,000 stocks currently trading, by pure chance, maybe one or two could show a pattern
that would look remarkably similar to a cycle pattern. But, despite assurances from some stock
analysts that you could have made money if you had just trusted their cycle patterns and that
you should trust them henceforth, the patterns would not be representative of the future”they
would just be historical coincidence.
On the other hand, Graphs D, E, and F could actually be representative. On average, each price Non cycles are more
in the next month is just a tiny bit higher than the previous (i.e., the expected rate of return on
there are ups and downs,
stocks is positive), but the important thing is that there is a lot of noise, up or down”which is too.
by de¬nition unpredictable. Stock prices must largely be unpredictable, or you could outsmart
the stock market. Incidentally, one of these three graphs is a real stock price, while the other
two are simulated. Can you see which one? I cannot! The real-world price series ¬ts right
in with my simulations of patternless day-to-day changes (called random walks and explained
below). In fact, whenever we look at graphical representations of stock prices, they usually look
very much like Graphs D-F and very unlike Graphs A-C. (Solution: Graph E is the actual stock
price series of IBM.)
Let™s look a bit more closely at magnitudes. May 31, 2002 was a decidedly uneventful day for Can we predict stock
prices? The order of
the stock market. The Dow Jones rose 13.56 from 9,911.69 to 9,925.25, a change of 0.14%. On
magnitude of typical
this day, the most actively traded stocks (but not biggest price movers) were MCI WorldCom daily stock price changes
(rate of return of ’1%), Nasdaq100 (’2%), Palm (’30%), Sun (0%), and Oracle (+6%). So, let™s put is tremendous, but the
expected price is not
our statistical and ¬nancial expertise to good use and ask a fundamental question of Finance:
much different from
In a perfect market, if the shares of a company cost $50 today, what do we expect them to cost today™s price.
tomorrow? Put another way, could you reasonably expect someone to be able to have predicted
this day™s stock price movement, e.g., something on the order of ±1% (as was the return of MCI
that day)? Think about it: if you could outpredict the stock price by an average of 1% ($50.00
to $50.50) on a typical day, you would be the world™s most amazing stock picker. Just 1% per
day represents an annual return of

1 + r0,365 = (1 + r0,1 )365 ≈ (1 + 1%)365 = (1 + 3, 678.34%) (19.1)

Too bad. Such magical abilities do not exist in the real world, where you can only expect to
earn rates between 2% and 50% per year, depending on what risk you are willing to take. Let
me put this in perspective: any fund manager who can consistently outperform her peers by
2% per year would be considered a star! Four percent per year makes a super star. In sum,
we can conclude that we do not expect individuals to be able to predict returns by an amount
of 1%/day, a typical daily stock movement”at least without divine guidance or its equivalent
(inside information).
¬le=e¬mkts.tex: LP
486 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

Figure 19.1. Potential Stock Price Patterns

Stock Price

Stock Price




1999 2000 2001 2002 2003 1999 2000 2001 2002 2003

Date Date

(A) (B)


Stock Price

Stock Price




1999 2000 2001 2002 2003 1999 2000 2001 2002 2003

Date Date

(C) (D)

Stock Price

Stock Price



1999 2000 2001 2002 2003 2004 1999 2000 2001 2002 2003

Date Date

(E) (F)

If these patterns were systematic, some of them should make you rich. Which ones?
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

Let us return to MCI. It decreased by 1%. Could the expectation be for MCI to decrease by 1%? Can the price tomorrow
be much lower,
If the expectation were for MCI shares to trade for $49.50 tomorrow, would you not want to
on average? No!
sell the shares for $50.00 today, instead? After one year, with such daily rates of return, you
would be left with only 2.5% of your original investment. If you could cheaply short the stock,
you could even make money from the 1% decline. After all, by selling shares today at $50.00
and repurchasing them tomorrow for $49.50 (i.e., selling short; see Chapter 12), you could reap
a positive 1% return. Every owner would rush to the market to sell, no one would want to buy,
and the price would immediately be lower.
These arguments suggest that we would expect very small daily rates of return, not much above So, what can the price
0.1%/day (with 255 trading days, (1 + 0.1%)255 ≈ 29%/year), which is much less than the day- tomorrow be,
on average?
to-day noise in stock prices. Intuitively, this is what an e¬cient stock market is: we do not
believe anyone can get rich easily, so it must be mostly impossible to predict where stocks are
going, aside from the very small mean (call it m). The best expectation of the price tomorrow
must be roughly the price today. Formally, if time 1 is very close, say just 1 day after time 0,

˜ ˜
E (P1 ) ≈ (1 + m) · P0 ” E (P1 ) ’ P0 ≈ m · P0
E (P1 ) ’ P0
≈m E (˜0,1 ) ≈ m
r ,
” ”

where P is the common notation for “price” and m is just a very small number (and determined
by a model such as the CAPM). This particular process is just the aforementioned random walk
(with drift). Thus, we can conclude that in the absence of easy ways to get rich, stock prices
approximately follow a random walk, at least in the short run.

Important: In the ¬nancial market context, “random walk” refers to a process
in which the expected value tomorrow is (almost) the same as the value today.
Naturally, actual values tomorrow will most likely be di¬erent from the value
The empirical evidence con¬rms this. Stock prices tend to follow roughly a random
walk in the short run. This means that it is not easy to get rich.

It is important that you realize that a random walk is a necessary consequence of an e¬cient Don™t wag the tail.
market, but you cannot conclude that a market is e¬cient just because prices follow roughly
a random walk. In fact, it could be that the true value follows one random walk process for
a long time and the market price follows another. Because market prices follow their own
random walk, merging with the fundamental value based random walk only in the very long
run, transaction costs would prevent you from getting rich, even though we know that market
prices would not always be the best estimate of value, given all information.
¬le=e¬mkts.tex: LP
488 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

Traders have tried all sorts of strategies in their e¬orts to become rich. One such strategy is
Predicting with past
rates of return. the aforementioned technical analysis, which tries to ¬nd patterns in historical stock prices.
For example, it is a popular misperception that stocks that rise one day are more likely to fall
back the next day. Figure 19.2 shows how the daily rate of return on our three stocks (S&P500,
IBM, and Sony) related to the previous day™s rate of return from 1985 to 2003. The graphs show
no pattern that would allow you to get rich quickly. There is de¬nitely not much juice in trying
to predict how a stock will perform tomorrow, given how it performed today. (A small reversal
that we occasionally observe seems to be caused by the bid-ask bounce. This is because if
the stock™s closing price is a bid price, on average it will fall back the next day (when it will be
either a bid or an ask price with roughly equal probability). If the stock™s close price is an ask
price, on average it will gain the next day.) Similar conclusions apply if we extend our use of
historical price information beyond yesterday, although over longer horizons, it appears as if
stocks tend to continue their pattern just a little bit. This is the aforementioned “momentum”
e¬ect and covered in an investments course.
Another variation on attempts to ¬nd market e¬ciency violations tries to predict not which
It is also unlikely that
one stock is expected to stocks systematically go up every day, but just which ones will for the next day: maybe it is
go up today, and
possible to predict that one stock should go up today, and another tomorrow. But, if you re¬‚ect
another stock to go up
on this statement, you realize that you could shift your money from one stock to another to
tomorrow, and so on.
take advantage of di¬erent stocks on di¬erent days. Again, unless the expected daily returns
are tiny, it would be too easy to get rich. And, again, with average rates of return being tiny
compared to the ups and downs, a good stock market pick is more likely to come from a lucky
or unlucky draw than from a systematic ability.
Of course, in the real world, there are ¬nancial transaction costs that would also prevent you
Transaction costs
destroy most hope for from really exploiting misvaluations. You would have to pay money to the broker to buy the
high turnover strategies.
shares, and again to sell shares. (This is why ¬nancial markets are not exactly perfectly competi-
tive, only approximately perfectly competitive.) Even small transaction costs can render trading
strategies with very high turnover unpro¬table. After all, even if the bid-ask spread is only 10 ba-
sis points, if incurred 255 trading days a year, you would only be left with (1’0.1%)255 = 77.5%
of your original investment. So, for a daily trading strategy to earn money, it needs to earn at
least an annual rate of return of 25% before it can overcome the trading frictions”which seems
almost hopelessly large to me.

Anecdote: Great Mathematicians and Gambling: The Origin of the Random Walk
In the 1700s, it was not beneath mathematicians to study how to gamble in order to gamble better. Jacob
Bernoulli (1654“1705) and Abraham DeMoivre (1667“1754) studied the random walk of a gambler™s stake in
fair games.
Later reinventions and applications of the random walk concept abound: Jan Ingenhausz (1730-1799), a physi-
cian and plant physiologist, placed charcoal powder on an alcohol ¬lm and observed that the grains moved
randomly. The botanist Robert Brown (1773-1858) reported erratic dancing of small particles in ¬‚uids at rest.
Albert Einstein (1879-1955) considered such ¬‚uids to be composed of discrete molecules, whose many collisions
with a “Brownian particle” caused the particle to jump in random directions”a random walk. Einstein™s analysis
not only explained Brownian motion, which has itself become a building block of high tech ¬nance nowadays,
but also bolstered the case for the existence of atoms, which was not yet universally accepted. The ¬rst recorded
use of the phrase “random walk” was by Lord Raleigh (1842-1919) in 1899. (Raleigh made a connection between
di¬usive heat ¬‚ow and random scattering and showed that a one-dimensional random walk could provide an
approximate solution to a parabolic di¬erential equation.) The name is believed to have originated with the
description of a drunk who stands on a ladder. The drunk can walk up or down and does so in a random
fashion”just like stocks.
Fortunately, in 1900, Louis Bachelier introduced the random walk theory of ¬nancial market ¬‚uctuations (al-
though Pearson introduced the term “random walk” only later, in 1905), ¬nding that bond prices could di¬use
in the same manner as heat. Unfortunately, this has only pointed out the obvious: it is not easy for an investor
to outperform the market. The ¬rst rigorous and published investigation of the random walk hypothesis was
done by Cowles, an eclectic investor and economist at Yale in the 1930s and 1940s.
Source: Mostly Michael F. Schlesinger, O¬ce of Naval Research, Scienceweek.com, 2001.
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

Figure 19.2. The Relation between Lagged and Current Rates of Return for S&P500, IBM, and


Return Today


’0.03 ’0.02 ’0.01 0.00 0.01 0.02 0.03

Return Yesterday

Return Today


’0.03 ’0.02 ’0.01 0.00 0.01 0.02 0.03

Return Yesterday

Return Today


’0.03 ’0.02 ’0.01 0.00 0.01 0.02 0.03

Return Yesterday

Note: The ¬gures chop o¬ some outliers, especially the crash of 1987 and mini-crash of 1989, but even if they are
included, there is no apparent predictability.
¬le=e¬mkts.tex: LP
490 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

Solve Now!
Q 19.4 From memory, write down the formula for a random walk.

Q 19.5 If stocks follow a random walk, can the price tomorrow be di¬erent from the price today?

Q 19.6 What is the typical movement of a stock on an average day?

Q 19.7 What is the typical expected rate of return on a stock on an average day?

19·3.B. Are Fund Managers Just Monkeys on Typewriters?

So, what about all the televised stock analysts who explain which stocks are undervalued and
What about celebrity
investors? which stocks are overvalued? And what about the aforementioned technical analysis, the art
of seeing shoulders, price barriers, etc., in historical prices and using them to forecast future
prices? (You can try out your own technical analysis at Yahoo!Finance”look up any stock, and
choose “Charts,” then “Technical Analysis”; it is fun, but useless.) And what about famous
investors such as Warren Bu¬ett, George Soros, and many others?
First, as already stated repeatedly, we are now talking about tiny deviations from the random
Even top investors can
have at most mild walk. Even an ability to forecast by 0.01% (yes, 0.0001) better per day yields an annual rate
predictive ability. None
of return that is 3.7% higher than it would otherwise be. So high a superior performance by a
can be expected to have
fund would be widely considered to be stellar performance. It is unlikely that anyone has good
even the ability to
foresee systematically a
day-to-day predictive ability that are larger than transaction costs.
0.1% movement per day.
Second, there are about 10,000 mutual funds today, which invest money on their investors™
Pure chance means that
some investors succeed behalf. How many of them are likely to outperform the stock market overall next year (at least
many years in a row.
before they collect fees)? If they have no ability, about 5,000. How many of these will outper-
form the year thereafter? About 2,500. Even if there is absolutely no ability, pure randomness
will mean that about 10 funds will outperform the market every year for ten years in a row.
What will happen to the funds that underperform several years in a row? They disappear qui-
etly. What will happen to the funds that outperform several years in a row? They proudly
announce their performances, advertise, boast, and collect more investments from outside in-
vestors. They are even better dressed, supported by larger “research teams,” appear more
“professional,” and ¬‚y in executive jets. They are the ones that will be most visible. Indeed, if
you made money ten years in a row in the stock market, would you not believe yourself that you
have the ability to pick stocks? From an investor™s or analyst™s perspective, both the disappear-
ance of funds that have performed poorly and the expansion of funds that have performed
well, will make it appear as if managers with a track record who o¬er to invest your money
indeed tend to have an ability to outperform”and even if they have absolutely no ability. This
is called survivorship bias.

Anecdote: Are women better investors than men?
Analyzing 35,000 households from 1991 to 1997, Terry Odean and Brad Barber found that men trade 45 percent
more than women. Apparently men are too overcon¬dent in their trading prowess. (Men also have a higher
propensity to su¬er from compulsive gambling disorders.) On average, men™s investment rates of returns were
lower than women™s, by a little less than one percent per year. Much, but not all, of the women™s better returns
could be attributed to the higher transaction costs that the men incurred for transactions that did not gain them
higher returns.
Despite strong evidence to the contrary, many investors still believe that stock prices do not follow random
walks, as evidenced by the plethora of ¬nancial talk shows and investment news letters. It would be better for
the general public to watch more sports and cooking shows and fewer investment shows”especially for males
like myself!
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

In truth, maybe there are some individuals who can pick stocks, but the evidence suggests that If you look for future
performance, past
luck is far, far more important than ability. Whenever academics (or the Wall Street Journal)
performance may be
have searched for better forward-looking performance among professional fund managers who your best guide, even if
outperformed in the past, they have found little or no exceptional performance. For example, it is a very, very poor
if managers were truly capable of systematically earning better rates of returns by picking
stocks, you would expect those managers who have picked better in the past to also pick better
in the future. The evidence is that about 54% of mutual funds that have outperformed their
benchmark over the last 1“3 years tend to outperform their benchmark over the following 1“3
years. (This is better than 50%, but not by much. And if we subtract fund fees, the average
performance drops signi¬cantly below 50%.) But, as fund prospectuses aptly note, and as the
evidence suggests, for the most part, past performance is no predictor of future performance.
Even if the market were ine¬cient and even if some fund managers could in fact outperform If there was superior
fund performance, an
the market, these fund managers would charge appropriately high fees to eliminate investors™
investor could not earn
advantages. After all, it is the fund manager who would have the scarce skill”picking stocks” money therefrom. It
and not the typical investor. Investors with money would compete to place money with such would be the fund
managers who would
managers and accept higher and higher fund fees. In the end, it would be highly unlikely that
earn the pro¬ts.
uninformed investors could earn excess returns by investing in actively trading mutual funds.

Important: In an e¬cient market, in which no one can pick stocks better than
anybody else, a large number of investors will beat the market. A small number
of investors will beat the market again and again. In the real world, there is little
evidence that investors who did well picking stocks in the past are better picking
stocks in the future when compared to investors who did poorly.

There are, of course, other ways to make money: Warren Bu¬ett™s fund, Berkshire-Hathaway, Old-fashioned work and
liquidity provision work
for example, runs an insurance and aircraft business. These businesses make money. But it is
better than stock
money earned the old-fashioned way: through hard work and risk taking. Writing insurance is picking.
risky business, and deserves extra return.
Here is my great business idea of the day. I give you stock tips, and I ask for money only if you Funds earn money on
the upside”is this a
make money. In fact, I only want 10 percent of your winnings. So “you have nothing to lose.”
good idea?
I only get something if I help you make money. Sounds like a deal? Now, if I pick a stock
randomly, I have a ¬fty-¬fty chance of making money. If you gain, I get something. If you lose,
I pay nothing. I am in e¬ect arbitraging you! Maybe I should give you the advice to buy a stock,
and your neighbor the advice to sell it. This way, I will surely make money! My only mistake is
that I have told you my plan.
My business model is not as absurd as it sounds. This is exactly how many funds operate: their Many funds are
compensated on the
managers participate in the upside, but not in the downside. (Of course, funds that charge
upside, but the
not only when they make you money, but also when they lose you money are not particularly alternative is not
con¬dence-inspiring, either. What are their incentives?) So, next time someone gives you a palatable, either.
great stock tip, regard it with some skepticism: it probably has a ¬fty-¬fty chance of being
¬le=e¬mkts.tex: LP
492 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

Solve Now!
Q 19.8 If a ¬rm employs 10,000 analysts, how many of them are likely to issue forecasts that
beat the market ten years in a row if none of them has any ability and there are no transaction

Anecdote: The Three Top Investment Books of 1996
The three best-selling investment books of spring 1996 were David and Tom Gardner™s Motley Fool Investment
Guide, based on a popular investment web site; Matt Seto™s The Whiz Kid of Wall Street™s Investment Guide
(Matt Seto was 17 years of age at the time); and the Beardstown Ladies™ Common-Sense Investment Guide,
authored by septuagenarians whose ¬rst book mixed cooking recipes with investment advice. All touted “com-
mon sense methods” to beat the market, earning 30 percent per year or more. Not a week went by without
dozens of prominent radio and TV shows featuring their sound advice. What did I need my Ph.D. in ¬nance for?
It is di¬cult to argue with performance!
Naturally, best-selling books are a great business. However, the stock performance of these three experts was

1. From 1996 to 2002, the Motley Fool investment recommendations of a number of hypothetical portfolios
have been discontinued. In 1997, they launched a real-money portfolio, called DRIP. From 7/28/1997 to
7/31/2002, it lost about 10%, while the S&P500 lost 2.5% and Nasdaq lost 15%. One should not judge a
fund by just 5 years of performance (and certainly not without risk adjustment), but it does appear that
the Motley Fool has not exactly found the Holy Grail of investment opportunities.
2. Matt Seto has stopped publishing his investment performance and has decided to pursue a career as a
3. The Beardstown Ladies, ¬ve books richer, were found to have miscalculated their returns: their returns
were not 30 percent, but 9 percent”signi¬cantly lower than the 15 percent turned in by the S&P500 stock
market index during their investment period.

How disappointing: on average, about one of them should have continued beating the market, one should have
done about the same as the market, and one should have underperformed it. Now, where are my ¬ve minutes
of fame?
Source: Time Magazine.
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

19·3.C. Corporate Consequences

If the Market is E¬cient
If markets are e¬cient, then managers can obtain valuable information from their own market You can learn from your
own market price!
prices. The market price is the conglomerate assessment of many investors, who put their
money where their mouths are. It aggregates a whole lot of information that managers them-
selves may not see so easily. If the stock price is very high, it probably means that the market
sees great opportunities ahead for the ¬rm. Thus, managers should consider growing the busi-
ness. Naturally, a high ¬rm value typically will also allow them to raise more funds from the
¬nancial markets at favorable rates. On the other hand, if the stock price is very low, it probably
means that the ¬nancial market anticipates the business to go down and managers to waste
the remaining money. In this case, managers should think carefully about whether they should
reinvest their money into the business, or into repurchasing their (relatively cheap) stock.
In addition to learning from your own company™s market price, you can also learn from all You can learn from
other market prices.
sorts of other market prices. You can ¬nd out how good your competitor™s opportunities are,
and whether you should get into the fray. Commodity prices are also often very helpful. If
the price of oil in the market is $30/barrel, it probably does not make sense for a ¬rm to plan
ahead based on an oil price of $50/barrel. The market price for oil is indeed fairly e¬cient. A
friend of mine sat on the corporate board for a large multinational oil company when the oil
price was $13/barrel, and the CEO argued that the ¬rm should plan oil exploration for a target
oil price of $20/barrel”the oil price “just had to go up.” Not only did this show tremendous
hubris, it was also outright stupid. The company could purchase oil in the market at rates of
$13/barrel, and thus did not have to do any oil exploration that cost between $13/barrel and
$20/barrel. Indeed, if this CEO could predict where the oil price was going, he could make a
lot more money as an oil trader than as the CEO of the oil company! Why explore for oil if you
can buy oil cheaper in the market?
An e¬cient market also means that it should be impossible to generate value by doing some- Adding value cannot be
done super¬cially.
thing that investors can do themselves. For example, buying another company to diversify does
not add value”investors could buy shares in the target by themselves and thus be themselves
diversi¬ed. They do not need our ¬rm to recognize that the target is undervalued”in fact,
chances are, the target was rightly valued to begin with and it was us who got the target value
wrong. In order for us to pro¬tably take over a target, we must have something extra that
investors cannot do for themselves”synergies, e.g., in the distribution of product or allocation
of overhead.
Market e¬ciency also means that it should not be easy to fool investors. For example, ¬rms Fooling investors cannot
be done easily.
can split their shares”each share trading at $80 would thereby become two shares trading at
$40. Nothing fundamental about the underlying project would have changed. If the market
is e¬cient, investors would not believe these new shares to be worth more than $40/share.
After all, just renaming shares should add no real value to the projects. A similar argument
applies when managers change earnings in ways that investors can see through. For example,
if a ¬rm previously reported a foreign division™s earnings separately, and now consolidates
them into the main earnings, such a change would increase the ¬rm™s consolidated earnings on
paper, but it would not create anything intrinsically valuable. Such changes should not add or
subtract ¬rm value.The same argument applies to dividends. In the absence of taxes or other
complications, a $100 ¬rm that pays $10 in dividends should be worth $90 thereafter”no
value is magically created or destroyed. However, although these arguments are theoretically
appealing, do not believe this too literally. Just because it should be this way does not mean
that it is this way. There is some empirical evidence that paying out money to shareholders and
meeting earnings expectations may indeed raise ¬rm value. After all, markets are not perfectly
¬le=e¬mkts.tex: LP
494 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

If the Market is Not E¬cient
Loosely speaking, ¬nancial markets tend to be fairly, but not always perfectly, e¬cient. Strong
What to do if markets
are not ef¬cient? market e¬ciency is almost surely not a good description of reality. Even in a perfectly rational
market, executives may know the ¬rm value better than the market”for example, they may
know that the company is about to seal a large contract, but this information has not yet
been disclosed. What should an executive do if she knows that the stock price is not equal to
the appropriate market value? (Of course, most executives believe that the ¬nancial markets
do not fully re¬‚ect the value of their companies even if they have no inside information”
as an executive, you should be wary of your own perceptions and biases!) The right way to
conceptualize your problem is to consider what you would do if you were the primary owner of
most shares, so you really care about ¬rm value. In addition, as an executive, you are supposed
to maximize the value of all your shareholders.

If your shares are undervalued, you should recognize that your cost of capital is e¬ectively
too high, given the true characteristics of your project. The reason is that you cannot
raise risky capital at fair prices”especially equity capital. The CAPM clearly is no longer
the right model for the cost of capital.
Assume that you know that your current projects will return $500 tomorrow. Also assume
that you have no cash, and that you can only raise ¬nancing through equity. Now assume
you come across a new project that costs $100 and will return a terri¬c $200 tomorrow.
The problem is that your investors do not believe that the ¬rm will return $700, but
falsely believe that the combined ¬rm will only be worth, say, $200. Thus, to raise $100,
you would have to sell 50% of your ¬rm, and keep only 50% of the true $700 return, for a
true $350 share of it. You would therefore be better o¬ passing up this new project and
just taking the $500 from the old project. Put di¬erently, the opportunity cost of new
capital to fund this project is way too high for you.
You would de¬nitely not want to raise cash at these high prices. Instead, you would want
to do the opposite. The best use of corporate cash may now be to repurchase cheap,
underpriced shares, e.g., from other investors. However, there is an intrinsic paradox
here: as an executive, you are supposed to act on behalf of your shareholders. Therefore,
repurchasing underpriced shares from them at bargain prices would not be what would
make the selling shareholders better o¬.

Shares are overvalued. Now your cost of capital would be very low, so you should be tempted
to take more projects. This is easiest to see if you again consider what you would do if
you were the primary owner of this overpriced ¬rm. You would want to sell more equity
shares at higher prices, and pay the money out in dividends to existing shareholders. (Al-
ternatively, you can just invest in Treasury securities.) Here the paradox is of course that
just one instant later, as CEO, you are now the representative of these new shareholders
that you have just sold overpriced shares to.

You can see that if the ¬rm is misvalued there are no easy recommendations for a CEO acting
on behalf of shareholders. The robust insight, however, is that the CEO of an undervalued ¬rm
should assume a relatively low cost of capital, the CEO of an overvalued ¬rm should assume a
relatively high cost of capital.
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

19·3.D. Event Studies Can Measure Instant Value Impacts

The immediacy of reaction in an e¬cient market o¬ers a surprising application: price reactions Market reactions should
be immediate and re¬‚ect
can allow us to estimate value consequences, using a technique called an event study. The idea
all value changes.
of an event study is that if the public market is valuing projects appropriately, and if the stock
price increases by $1,000,000 on the minute when the ¬rm ¬rst announces the event, then the
value of the new project is likely to be worth about $1,000,000.

Example: The Value Impact of FDA Drug Rejections
For example, say you wanted to know what the value implication of the rejection of a novel drug An Example: Estimating
the value loss when the
application by the Food and Drug Administration (F.D.A.) is. You could compute the sudden
FDA rejects a drug
decline in future expected cash ¬‚ows, discount it properly, and come up with an appropriate application.
valuation estimate. This is not an easy task. But if stock markets value pharmaceutical stocks
appropriately, the stock price reaction to the announcement of the F.D.A. drug application
rejection would be a good indicator of the value loss. After all, if the stock market did not react
immediately, you could on average earn great pro¬ts by shorting the relevant stock and waiting
for the market to catch up.

Figure 19.3. Event Study: Lining up Event Date Information 60

dd d
d dd
ddd d Day 0 for PHA
d d
ddd d

Firm Date Event ppppppppp pp ppp
pp pppp
g gg
d p
DNA 7/9/2001 FDA rejects Xolair g g d ppdd
g g
Stock Price


g dd
nnnnnnnnnnng dddd
g Day 0 for DNA dd
GNLB 6/26/2001 FDA rejects Aslera g nnn
nnnnn nnn
g n

NVS 6/22/2001 FDA rejects Zelnorm Day 0 for NVS
PHA 7/12/2001 FDA rejects Parecoxib gg


Day 0 for GNLB

jun 1 jun 15 jun 29 jul 16 jul 31

Trading Day

(A) (B)


dd dd
d ddd
ddd dd d
dd ddd
d d
d dd
ddd d d
ddd d
d d
d d
ddd d
d p
pp dpd


pppdppppp ppp
ppppppppp pp p ppp
pp p
pp ppp ppp p
pp ppppppppp p
p pha
p g
g gg
gg dpppppp
d p
p g
g g
g d
d d d d dna
d g
gg gg
gg g g
Stock Price

Stock Price



nnnnnn g g dd
g dd
nnnnnn ddd
ddd dd d
nnng d nn
nn dd nnn gg g
gg g g
nnnn g
g nnn
nnn nnn
nnn nnnnn
nnnnn n n n n n n nvs
nnnnnn nnn
nnn nn
nn g


gg jun 1 to jul 31


g g g g g g g g g g g g g g g gnlb

Event Day 0


jun 1 jun 15 jun 29 jul 16 jul 31
’30 ’20 ’10 0 10 20 30

Event Day
Trading Day

(C) (D)

(GNLB™s stock price was scaled by a factor of 11 to ¬t better into this graph.)

Figure 19.3 starts with four events that we have identi¬ed: the FDA rejected Xolair, a Genentech
(ticker DNA) drug, on July 9, 2001; Aslera, a Genelabs (ticker GNLB) drug, on June 26; Zelnorm,
a Novartis (NVS) drug, on June 22; and Parecoxib, a Pharmacia (PHA) drug on July 12. Graph B
plots the price history for these four stocks during June and July 2001. The event day itself
is marked in this graph. To ¬nd out what happens when the FDA rejects a drug, we need to
line up all the returns in event time, as illustrated in graphs C and D. When we compute the
¬le=e¬mkts.tex: LP
496 Chapter 19. E¬cient Markets, Classical Finance, and Behavioral Finance.

rate of return over the three days around our event date (for we are not sure how accurate our
identi¬cation of the announcement day is, and whether announcements happen before opening
or after closing of the stock market), we discover the following:

Firm Date Event “Couple of Days” Return
DNA 7/9/2001 FDA rejects Xolair dropped about 18%
GNLB 6/26/2001 FDA rejects Aslera dropped about 46%
NVS 6/22/2001 FDA rejects Zelnorm stayed put
PHA 7/12/2001 FDA rejects Parecoxib dropped about 7%
Our conclusion FDA rejections are not
always, but often, bad news.

Having lined up everything in event time, we can do more analysis. For example, we can subtract
the rate of return on the market on each event date to eliminate noise induced by the general
movement on the event date. We can compute the average rate of return on the event, which
seems to be around “18%”getting rejected by the FDA is not a good thing for a ¬rm™s market
value. If you know more statistics, you can compute whether this value drop is “statistically
signi¬cant.” (It is.) You can investigate if bigger ¬rms experience a bigger or a smaller drop.
For this, you need to locate the market value at the time:

Firm Event Return Equity Market Value
GNLB $0.1 billion
NVS $2.7 billion
DNA $26 billion
PHA $58 billion

Therefore, the evidence suggests that smaller ¬rms are harder hit, but that the relationship
is not necessarily perfectly monotonic. It would be even more interesting if we knew what
fraction of the drug development portfolio the particular rejected drug would constitute”we
could then determine whether drug developers whose main portfolio drug was rejected su¬er
more. We do not have these data, so no such test! We could test whether it has become worse
or better over time to have one™s drug rejected by the FDA by sorting our events by the event

Firm Event Return Event Day
NVS 6/22/2001
GNLB 6/26/2001
DNA 7/9/2001
PHA 7/12/2001

There does not seem to be a clear relationship here”as any sane analyst would have suspected”
so the evidence does not suggest that the market looked any more favorably or less favorably
upon rejections in June 2001 relative to July 2001. We could investigate returns before or after
the event announcement. If information leaks prior to the FDA announcement, we should see
a drop even before event day 0. An alternative, though very unlikely, reason for such a pre-
announcement price pattern would be if the FDA were more likely to reject a drug if the stock
price had recently gone down. We could also investigate whether we can earn pro¬ts buying or
selling after the event”under market e¬ciency, this should not be the case. (In a larger sample
in Figure 19.4 below, we will examine pre- and post-announcement returns.)

If we had more statistical background, instead of just sorting our events as we did in the
Digging Deeper:
in-text tables above, we should run a regression to predict the announcement rate of return with variables we
deem to be important determinants of the value change upon FDA rejections.
¬le=e¬mkts.tex: RP
Section 19·3. E¬cient Market Consequences.

Table 19.1. Sample Event Study of FDA Drug Rejections

Event Announcement Rate of Return
Symbol Company Drug Day-Day Stock Market Net
’2.82 +5.69
AVE Aventis Re¬‚udan 2.87
12/22/98-12/24/98 ’65.56 ’65.75
AVN Avanir Docosanol 0.19
’35.00 ’36.12
CEPH Cephalon Myotrophin 1.12
’38.32 ’39.40
CRXA Corixa Bexxar 1.08


. 25
( 39)