A random Walk Down Wall Street: The Time-Tested Strategy for Successful Investing

THE LIMITS TO ARBITRAGE
Thus far we have considered the cognitive biases that
influence investors and, therefore, security prices. The
actions of individual investors are often irrational, or at least
not fully consistent with the economist’s ideal of optimal
decision making. In perhaps the most pathological case,
individuals appear to go mad in herds and bid some categories
of stocks to unreasonable heights. Since the errors of
irrational investors do not cancel out but often reinforce each
other, how can stocks be efficiently priced? Believers in
efficient markets rotely state that “arbitrage” will make the
market efficient even if many individual investors are
irrational. Arbitrageurs, such as professional Wall Street
traders and hedge-fund managers, are expected to take
offsetting positions—such as selling overpriced stocks short
and buying underpriced ones—so that any mispricing caused
by irrational investors is quickly corrected. Rational traders

---

are expected to offset the impact of behavioral traders. Thus,
the second major pillar on which some behavioralists rest
their case against efficient markets is that such arbitrage is
severely constrained. Behavioralists believe that important
limits to arbitrage exist that prevent out-of-whack prices from
being corrected.
Suppose irrational investors cause an oil company security
to become overpriced relative to both its fundamental value
and its peer oil companies. Arbitrageurs can simply sell the
overpriced security short and buy a similar substitute oil
company security. Thus, the arbitrageur is hedged in the
sense that favorable or unfavorable events affecting the oil
industry will influence both companies. A rise in the price of
oil that makes the shorted security rise will make the
arbitrageur’s long position rise as well.
But this kind of arbitrage is extremely risky. Suppose the
“overpriced” security reports some unusually good news,
such as a significant oil strike that was not anticipated. Or
suppose the “fairly valued” security suffers some unforeseen
setback, such as the explosion of a deep-water oil well, which
causes its price to fall. The arbitrageur could conceivably lose
on both sides of the trade. The security that had been sold

---

short could rise, and the security held long could fall. The
kind of arbitrage required to correct perceived mispricings is
extremely risky.
The trader who tries to “correct” perceived mispricings
also runs the risk that investors will become even more
overenthusiastic about the prospects for the “overpriced”
security. Suppose an arbitrageur was convinced during 1999
that Internet stocks were outrageously overpriced. The trader
might sell short the Internet favorites, hoping to buy them
back later at lower prices. But as enthusiasm for the New
Economy continued to grow, the prices of these stocks rose
even further—many of them doubling and then doubling
again. Only in retrospect do we know that the bubble burst
during 2000. In the meantime, many traders lost their shirts.
The market can remain irrational longer than the arbitrageur
can remain solvent. This is especially true when the
arbitrageur is credit constrained. Long Term Capital
Management, a hedge fund in which Nobel laureates devised
the strategies, found itself in an unbearable position when the
prices of its hedges moved against it and it had insufficient
capital to keep them afloat.
The natural players in the game of selling overpriced

---

securities short and buying underpriced ones are global hedge
funds, with trillions of dollars to invest. One might suppose
that these funds would have recognized the unsustainability
of the prices of Internet stocks and exploited the mispricing
by selling short. A study by Markus Brunnermeier and
Stefan Nagel examined hedge-fund behavior during the 1998–
2000 period to see whether these funds restrained the rise in
speculative favorites.
The findings were surprising. Sophisticated speculators
such as hedge funds were not a correcting force during the
bubble period. They actually helped inflate the bubble by
riding it rather than attacking it. Hedge funds were net buyers
of Internet stocks throughout the 1998–early 2000 period.
Their strategy reflected their belief that contagious
enthusiasm and herding of unsophisticated investors would
cause the mispricing to grow. They were playing the game
described earlier in Keynes’ famous newspaper beauty
contest. While a stock selling at $30 might be “worth” only
$15, it would be a good buy if some greater fools would be
willing to pay $60 for the stock at some future time.
It appears that hedge funds also played a destabilizing role
in the oil market during 2005 and 2006. From 2004 to 2006

---

the price of a barrel of crude oil more than doubled. Although
economic forces such as the growth of the world economy
provided some fundamental reasons for the upward price
pressure, it seems that speculative activity, especially by
hedge funds, helped fuel the advance. And the few hedge
funds that went short in the oil futures market experienced
substantial losses. It is clear that arbitrage trades to correct a
perceived price bubble are inherently risky.
And there are also times when short selling is not possible
or at least severely constrained. Typically in selling short, the
security that is shorted is borrowed in order to deliver it to
the buyer. If, for example, I sell short 100 shares of IBM, I
must borrow the securities to be able to deliver them to the
buyer. (I must also pay the buyer any dividends that are
declared on the stock during the period I hold the short
position.) In some cases it may be impossible to find stock to
borrow, and thus it is technically impossible even to execute a
short sale. In some of the most glaring examples of inefficient
pricing, technical constraints on short selling prevented
arbitrageurs from correcting the mispricing.
Arbitrages may also be hard to establish if a close
substitute for the overpriced security is hard to fund. For an

---

arbitrage to be effective, there must be a similar fairly priced
security that can be bought to offset the short position and
that can be expected to rise if some favorable event occurs
that influences the whole market or the sector to which the
security belongs.
One of the best examples used by behavioralists to show
that market prices can be inefficient is the case of two
identical shares that do not trade at identical prices. Royal
Dutch Petroleum and Shell Transport are considered Siamese
twin companies. These companies agreed in 1907 to form an
alliance and to split their after-tax profits 60 percent for
Royal Dutch, 40 percent for Shell. In an efficient market, the
market value of Royal Dutch should always be 1½ times as
great as the market value of Shell. In fact, Royal Dutch has
often traded at a premium to Shell of up to 20 percent over
fair value. In efficient markets, the same cash flows ought to
sell at equivalent valuations.
The problem with this example is that the two securities
trade in different national markets with different rules and
possibly different future restrictions. But even if Royal
Dutch and Shell were considered equivalent in all respects,
the arbitrage between the two securities would be inherently

---

risky. If Royal Dutch sells at a 10 percent premium to Shell,
the appropriate arbitrage is to sell the overpriced Royal
Dutch shares short and buy the cheap Shell shares. The
arbitrage is risky, however. An overpriced security can
always become more overpriced, causing losses for the short
seller. Bargains today can become better bargains tomorrow.
It is clear that one cannot rely completely on arbitrage to
smooth out any deviations of market prices from fundamental
value. Constraints on short selling undoubtedly played a role
in the propagation of the housing bubble during the end of the
first decade of the 2000s. When it is virtually impossible to
short housing in specific areas of the country, only the votes
of the optimists get counted. When the optimists are able to
leverage themselves easily with mortgage loans, it is easy to
see why a housing bubble is unlikely to be constrained by
arbitrage.