THE LIMITS OF

ARBITRAGE
ANDREI SHLEIFER AND
ROBERT W.VISHNY
 IN REALITY ， ALMOST ALL ARBITRAGE
REQUIRES CAPITAL, AND IS TYPICALLY
RISKY.
PROFESSIONAL ARBITRAGE IS
CONDUCTED BY A RELATIVELY SMALL
NUMBER OF HIGHLY SPECIALIZED
INVESTORS USING OTHER’S CAPITAL.
ARBITRAGE BECOMES INEFFECTIVE IN
EXTREME CIRCUMSTANCES. (PRICES
DIVERGE FAR FROM FUNDAMENTAL
VALUES)

ABSTRACT ANOMALIES IN FINANCIAL MARKETS

CANNOT BE ELIMINATED BY ARBITRAGE
 FUNDAMENTAL CONCEPTS-ARBITRAGE
“THE SIMULTANEOUS PURCHASE AND
SALE OF THE SAME, OR ESSENTIALLY
SIMILAR, SECURITY IN TWO DIFFERENT
MARKETS FOR ADVANTAGEOUSLY
DIFFERENT PRICES.”
NO RISK AND NO CAPITAL
(SHARPE AND ALEXANDER 1990)
 REALISTIC ARBITRAGE ARE MORE
COMPLEX
• REQUIRE CAPITAL - GOOD FAITH MONEY
• IN SHORT RUN, ONE MAY LOSE MONEY
• DIFFERENT TRADING HOURS, SETTLEMENT DATES, AND DELIVERY TERMS.
• IF PRICES ARE MOVING RAPIDLY, THE VALUE MAY DIFFER (RISK
ARBITRAGE)
 PBA( PERFORMANCE-BASED ARBITRAGE)
• MORE COMMONLY, CONDUCTED BY RELATIVELY FEW PROFESSIONAL,
HIGHLY SPECIALIZED INVESTORS;
• OUTSIDE INVESTORS ( PREFERABLY WEALTHY INDIVIDUALS, BANKS,
ENDOWMENTS)
• BRAINS AND RESOURCES ARE SEPARATED BY AGENCY RELATIONSHIP
• ALLOCATE FUNDS BASED ON PAST RETURNS OF ARBITRAGEURS
• PBA IS A PHENOMENON OF RESPONSIVENESS OF FUNDS UNDER
MANAGEMENT TO PAST RETURNS.
• AUTHORS BRING IN REFERENCES OF MANY RESEARCH PAPERS ON
‘DELEGATED PORTFOLIO MANAGEMENT’
1. AN AGENCY MODEL OF LIMITED ARBITRAGE

① THREE TYPES OF PARTICIPANTS: NOISE TRADERS, ARBITRAGEURS,

AND INVESTORS IN ARBITRAGE FUNDS WHO DO NOT TRADE ON
THEIR OWN.
② FUNDAMENTAL VALUE IS V;(ONLY ARBITRAGEURS KNOW AND THEIR
INVESTORS DO NOT KNOW)
③ THREE TIME PERIODS: 1, 2 AND 3
④ AT TIME 3, VALUE V BECOMES KNOWN TO ARBITRAGEURS AND
NOISE TRADERS.
I. AN AGENCY MODEL OF LIMITED ARBITRAGE
BOTH ARBITRAGUERS AND INVESTORS ARE RATIONAL
Arbitrageurs Cumulative resources under management including their borrowing
capacity-Ft
I. AN AGENCY MODEL OF LIMITED ARBITRAGE
• COMPETE IN THE PRICE THE CHARGE;
• ASSUME MARGINAL COST CONSTANT, SO COMPETITION DRIVES PRICE
TO MARGINAL COST
• BAYESIANS, ALLOCATE FUNDS ACCORDING TO PAST
PERFORMANCE( PBA );
• AN INCREASING FUNCTION:
• IF G IS THE INVESTORS SUPPLY OF AGGREGATE FUNDS TO THE
ARBITRAGEURS AND RETURN ON ASSET IS P2/P1

F2  F1 * G{( D1 F1 ) * ( p2 p1 )  ( F1  D1 ) F1}
 I. AN AGENCY MODEL OF LIMITED
ARBITRAGE
• BENCHMARK: ZERO RETURN
• A LINEAR FUNCTION: G ( x)  ax  1  a a 1
• x IS THE ARBITRAGEUR’S GROSS RETURN
F2  a{D1 * ( p2 p1 )  ( F1  D1 )}  (1  a)F1  F1  aD1 (1  p2 p1 )
• IF P2<P1 , GAIN FUNDS; OR ,LOSE FUNDS

• THE HIGHER IS a THE MORE SENSITIVE TO PAST PERFORMANCE

II. PERFORMANCE-BASED ARBITRAGE AND MARKET EFFICIENCY

The first bench mark is efficient markets, p1=p2=V.

Alternative benchmark is arbitrageurs resources are
limited and can always raise F1.
The case of a=1(arbitrageur not getting any more
money when he loses some)
first order condition:

Inequality： : D1=F1
Equality： : D1<F1
The initial displacement is large and will recover
with a high probability; if they fall, it can’t be large
and fully invested at 1
 II. PERFORMANCE-BASED ARBITRAGE AND
MARKET EFFICIENCY
UNCERTAINTY OF THE EFFECT:

 A HIGHER a COULD MAKE MARKET LESS EFFICIENT, BY

WITHDRAWING FUNDS;

 A HIGHER a WILL MAKE PRICES ADJUST QUICKLY BY GIVING MORE

FUNDS AFTER A PARTIAL REVERSAL OF THE NOISE SHOCK.
III. DISCUSSION OF PERFORMANCE-BASED ARBITRAGE
• PBA SUPPOSES THAT ALL ARBITRAGEURS HAVE
THE SAME SENSITIVITY, AND WILL INVEST ALL
FUNDS WHEN ASSET MISPRICED.
• IN REALITY, THEY DIFFER. RESOURCES
INDEPENDENT AND INVEST MORE WHEN PRICE
DIVERGE FURTHER; NEED NOT TO LIQUIDATE.
• BIG SHOCK NEED MORE TO ELIMINATE, IF NOT,
MISPRICING GETS DEEPER.
• LITTLE FRESH CAPITAL AVAILABLE TO STABILIZE.
SO PBA IS LIKELY TO BE IMPORTANT .
 IV. EMPIRICAL IMPLICATIONS
• A. WHICH MARKETS ATTRACT ARBITRAGE RESOURCES?
LARGE FUNDS CONCENTRATED IN A FEW MARKETS, BOND
MARKETS & FOREIGN EXCHANGE MARKET;
THE ABILITY TO ASCERTAIN VALUE;
SPECIALIZED ARBITRAGEURS AVOID VOLATILE;
SHORT HORIZONS MAY BE MORE RELEVANT
 IV. EMPIRICAL IMPLICATIONS
B. ANOMALIES
• HIGHER HISTORICAL RETURNS.
• EMH: COMPENSATION FOR HIGHER RISK IMPLAUSIBLE: LARGE
NUMBER OF DIVERSIFIED ARBITRAGEURS.
• FEW SPECIALIZED ARBITRAGEURS CARE ABOUT TOTAL RISK
FUNDAMENTAL OR IDIOSYNCRATIC.
• FAILING TO RECOGNIZE PRICE-REVISAL .
 IV. EMPIRICAL IMPLICATIONS

• IN EXTREME CIRCUMSTANCES, LOSE ENOUGH MONEY

AND LIQUIDATE;
• INVESTORS BECOME KNOWLEDGEABLE ABOUT THE
STRATEGIES, DIMINISH WITHDRAWALS; BUT IT WILL BE
SLOWLY FOR INVESTORS TAKE ACTION.
 V. CONCLUSION

• PBA MAY NOT BE FULLY EFFECTIVE IN BRINGING PRICES TO

FUNDAMENTAL VALUES;
• SPECIALIZED PROFESSIONAL ARBITRAGEURS MAY AVOID EXTREMELY
VOLATILITY;
• THE AVOIDANCE SUGGESTS A DIFFERENT APPROACH TO
UNDERSTANDING PERSISTENT EXCESS RETURNS.