Econ 138: Financial and Behavioral Economics

Noise-Trader Risk in Financial Markets

February 8 & 13, 2017

Reading:
J.B. DeLong, A. Shleifer, L.H. Summers and R.J. Waldmann “Noise
Trader Risk in Financial Markets.”
N. Barberis and R. Thaler, “A Survey of Behavioral Finance,”
Sections 2.1 & 2.2.
R.J. Hawkins “Maximizing Expected Utility: CARA Utility and
Mean-Variance Optimization,” in “Supplementary Material” bCourses
folder.
Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 1/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Overview:
A primary context of the model is Friedman:
“that speculation is . . . destabilizing . . . is largely equivalent to
saying that speculators lose money, since speculation can be
destabilizing in general only if speculators on . . . average sell
. . . low . . . and buy . . . high.”
This model focusses on the limits of arbitrage:
1 Fundamental risk.
2 Noise-trader risk.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 2/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Overview:
Fundamental risk is the risk that new information about a security
validates a (mis)perceived overpricing or underpricing.

Imagine a stock is overpriced.

You (an arbitrageur) sell the stock short.

Then either realized dividends and/or new news about

dividends validates the price at the current or even a higher
level: You lose!

To manage this fundamental risk you will limit the size of any
one bet.

This limits the effectiveness of arbitrage.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 3/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Overview:
Noise-trader risk is the risk that investor sentiment exacerbates a
perceived overpricing or underpricing.

Imagine a stock is underpriced.

You (an arbitrageur) buy on the dip in anticipation of a

reversion to the rational price.

The crowd gets even more pessimistic and drives the stock to
an even lower level: You lose!

To manage this noise-trader risk you will limit the size of any
one bet.

This limits the effectiveness of arbitrage.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 4/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Overview:
“This risk of a further change of noise traders’ opinion away from
its mean – which we refer to as “noise trader risk” – must be
borne by any arbitrageur with a short time horizon and must limit
his willingness to bet against noise traders.

“All the main results of [this] paper come from the observation
that arbitrage does not eliminate the effects of noise because noise
itself creates risk.”

“The risk resulting from stochastic changes in noise traders’

opinions raises the possibility that noise traders who are on average
bullish earn a higher expected return than rational, sophisticated
investors engaged in arbitrage against noise trading.”

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 5/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Overview:
Noise traders can earn higher expected returns:
1 Solely by bearing more of the risk that they themselves create.
2 From their own destabilizing influence, not because they
perform the useful social function of bearing fundamental risk.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 6/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The Model:
Overview:

The model contains noise traders and sophisticated investors.

Noise traders falsely believe that they have special information

about the future price of the risky asset.

Noise traders select their portfolios on the basis of such

incorrect beliefs.

Sophisticated traders buy when noise traders depress prices

and sell when noise traders push prices up.

Such active contrarian investment strategies push prices

toward fundamentals, but not all the way.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 7/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The Model:
The Agents & Assets:

Agents live for two periods.

The only decision agents make is to choose a portfolio when

young.
The economy contains two assets that pay identical dividends:
1 A safe asset s that pays a fixed dividend r .
2 An unsafe asset u that pays a fixed dividend r .

The price of the safe asset is fixed at one.

The quantity of the unsafe asset is fixed at one.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 8/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The Model:
The Agents:
There are two types of agents:
1 Sophisticated investors (arbitrageurs A) with rational
expectations.
2 Noise traders (N).

The fraction of noise traders is µ.

The fraction of arbitrageurs is 1 − µ.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 9/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The Model:
The Agents:

Choose their portfolio when young.

Are focussed on the ex ante mean of the distribution of the
risky-asset price at time t + 1.
The sophisticated investors get the mean correct.
The noise traders misperceive the mean by ρt ∼ N ρ∗ , σρ2 .

Form portfolios of u and s with quantities of u

( λN A
t and λt ) purchased at price pt .

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 10/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Constant absolute risk aversion (CARA) utility

The CARA utility wealth assumption implies

U(w ) = −e −2γw
where
γ = the rate of absolute risk aversion.
w = wealth in dollars.

If the return on the risky asset is normally distributed, then w is

also normally distributed

1 (w −µ)2
p(w ) = √ e − 2σ2 .
2πσ

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 11/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Constant absolute risk aversion (CARA) utility

With this assumption we can write the expected utility as

Z ∞
E [U (w )] = U (w ) p(w )dw
−∞
Z ∞
1 (w −µ)2
= U (w ) √ e − 2σ2 dw .
−∞ 2πσ
And, with a bit of calculusa , it can be shown that maximizing the
expected utility is equivalent to maximizing

E [w ] − γVar [w ]
a
R.J. Hawkins “Expected Utility Supplement.”

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 12/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Rational Traders:

Arbitrageurs determine the number of shares of the risky asset λA

t
for their portfolio by maximizing
h i h i
A A
E [U] = E wt+1 − γVar wt+1 .

To this end, the wealth in the next period is calculated:

 
A
wt+1 = λA (r + p t+1 ) + w A
− λ A
t t (1 + r )
p
|t {z } |
t
{z }
wealth from risky asset
wealth from risk-free asset

= λA A
t [r + pt+1 − pt (1 + r )] + wt (1 + r )

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 13/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Rational Traders:
 A   A 
Arbitrageurs maximize E wt+1 − γVar wt+1 with
A
wt+1 = λA A
t [r + pt+1 − pt (1 + r )] + wt (1 + r )

Taking the expectation and variance we maximize

E [U] = λA A
t [r + E [pt+1 ] − pt (1 + r )] + wt (1 + r )
 2
− γ λA t Var [pt+1 ] .
Setting
∂ E [U ]/∂λA
t = r + E [pt+1 ] − pt (1 + r ) − 2γλA
t Var [pt+1 ] = 0
we get
r + E [pt+1 ] − pt (1 + r )
λA
t =
2γVar [pt+1 ]
Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 14/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Noise Traders:

Noise traders determine the number of shares of the risky asset λN

t
for their portfolio by maximizing
h i h i
N N
E [U] = E wt+1 − γVar wt+1 .

To this end, the wealth in the next period is calculated:

 
N
wt+1 = λN (r + pt+1 + ρ t ) + w N
− λ N
t t (1 + r )
p
|t {z } | t {z }
wealth from risky asset
wealth from risk-free asset

= λN N
t [r + pt+1 − pt (1 + r ) + ρt ] + wt (1 + r )

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 15/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Noise Traders:
 N   N 
Noise traders maximize E wt+1 − γVar wt+1 with
N
wt+1 = λN N
t [r + pt+1 − pt (1 + r ) + ρt ] + wt (1 + r )

Taking the expectation and variance we maximize

E [U] = λN N
t [r + E [pt+1 ] − pt (1 + r ) + ρt ] + wt (1 + r )
 2
− γ λNt Var [pt+1 ] .
Setting
∂ E [U ]/∂λN
t = r + E [pt+1 ] − pt (1 + r ) + ρt − 2γλN
t Var [pt+1 ] = 0
we get
r + E [pt+1 ] − pt (1 + r ) + ρt
λN
t =
2γVar [pt+1 ]
Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 16/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The market price for the risky asset:

The market price pt follows from the market-clearing requirement:

µλN A
t + (1 − µ) λt = 1 .

Substituting our expression for λN A

t and λt we find that

1  
pt = r + E [pt+1 ] − 2γVar [pt+1 ] + µρt
(1 + r )

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 17/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

In the steady state the risky asset price is characterized by:

A stationary price distribution function.

No change in expected price:

E [pt+2 ] = E [pt+1 ]

No change in price variance:

Var [pt+2 ] = Var [pt+1 ]

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 18/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

Recalling our expression for pt we can write pt as

1  
pt+1 = r + E [pt+2 ] − 2γVar [pt+2 ] + µρt+1
(1 + r )

and
 
1  
E [pt+1 ] = E r + E [pt+2 ] − 2γVar [pt+2 ] + µρt+1
(1 + r )
or
1  
E [pt+1 ] = r + E [pt+2 ] − 2γVar [pt+2 ] + µE [ρt+1 ]
(1 + r )

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 19/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

With our result so far that
1  
E [pt+1 ] = r + E [pt+2 ] − 2γVar [pt+2 ] + µE [ρt+1 ]
(1 + r )

we can substitute the steady-state relations

E [ρt+1 ] = ρ∗ , E [pt+2 ] = E [pt+1 ], and Var [pt+2 ] = Var [pt+1 ]

to obtain
1  
E [pt+1 ] = r + E [pt+1 ] − 2γVar [pt+1 ] + µρ∗ .
(1 + r )
or
2γVar [pt+1 ] µρ∗
E [pt+1 ] = 1 − +
r r
Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 20/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

Returning to our expression for pt

1  
pt+1 = r + E [pt+2 ] − 2γVar [pt+2 ] + µρt+1 ,
(1 + r )

calculating
 
1  
Var [pt+1 ] = Var r + E [pt+2 ] − 2γVar [pt+2 ] + µρt+1
(1 + r )

is simplified considerably by recalling that Var [constant] = 0.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 21/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

Since Var [constant] = 0:

1  0
 
:0 :0

[pt+2 ] − 2γ

Var [pt+1 ] = Var r +
 E  
Var
[p
t+2 ] + µρt+1

(1 + r )

reduces to
 
µρt+1
Var [pt+1 ] = Var
1+r
and
 2
µ
Var [pt+1 ] = σρ2
1+r

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 22/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The steady-state market price for the risky asset:

Pulling it all together we have

1  
pt = r + E [pt+1 ] − 2γVar [pt+1 ] + µρt
(1 + r )
and
2γVar [pt+1 ] µρ∗
E [pt+1 ] = 1 − +
r r
and
 2
µσρ
Var [pt+1 ] =
1+r

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 23/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Substituting:
 
 
1  
pt = r + E [pt+1 ] −2γ Var [pt+1 ] +µρt 
(1 + r ) 
 | {z } | {z } 

µσρ 2 µσρ 2
2γ ( 1+r ) ∗ ( 1+r )
1− r
+ µρ r

we obtain
  2 
µσρ
1 2γ 1+r µρ∗

µσρ
2
pt = r + 1 − + − 2γ + µρt 
 
(1 + r ) r r 1+r

and, the price of the risky asset is

2
µ (ρt − ρ∗ ) µρ∗ 2γ

µσρ
pt = 1 + + −
1+r r r 1+r

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 24/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

2
µ (ρt − ρ∗ ) µρ∗ 2γ

µσρ
pt = 1 + + −
1+r r r 1+r

Interpretation of terms:
1 The fundamental value.
2 Fluctuations due to variation in noise-traders’ misperceptions.
3 Deviation due to non-zero average noise-trader misperception.
4 Additional price risk due to uncertainty in pt+1 :
noise traders “create their own space.”
Prices are driven down and returns are driven up.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 25/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

Do noise traders do worse than rational traders? (EMH)

The difference in expected return E [∆RN−A ] can be written:
 
E [∆RN−A ] = λN A

t − λt r + E [pt+1 ] − pt (1 + r )
| {z }
expected return
where
r + E [pt+1 ] − pt (1 + r )
λA
t =
2γVar [pt+1 ]
and
r + E [pt+1 ] − pt (1 + r ) + ρt
λN
t =
2γVar [pt+1 ]

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 26/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

Solving ∆RN−A = λN A


t − λt [r + E [pt+1 ] − pt (1 + r )]:

Recall that
r + E [pt+1 ] − pt (1 + r )
λA
t =
2γVar [pt+1 ]
and
r + E [pt+1 ] − pt (1 + r ) + ρt
λN
t =
2γVar [pt+1 ]
most terms cancel leaving:
ρt 1 + r 2
 

N A
 ρt
λt − λt = =
2γVar [pt+1 ] 2γ µσρ

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 27/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

Solving ∆RN−A = λN A


t − λt [r + E [pt+1 ] − pt (1 + r )]:

Recall that
1  
pt = r + E [pt+1 ] − 2γVar [pt+1 ] + µρt
(1 + r )

so r + E [pt+1 ] − pt (1 + r ) becomes

1  
r + E [pt+1 ] − r + E [pt+1 ] − 2γVar [pt+1 ] + µρt (1 + r )
(1 + r )

and most terms cancel leaving:

 2
µσρ
r + E [pt+1 ] − pt (1 + r ) = 2γVar [pt+1 ] − µρt = 2γ − µρt
1+r

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 28/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

Bringing it all together:

 h i
∆RN−A = λN t − λ A
t r + E [p t+1 ] − pt (1 + r )

where  2
  ρt 1+r
λN
t − λA
t =
2γ µσρ
and
µσρ 2
h i  
r + E [pt+1 ] − pt (1 + r ) = 2γ − µρt
1+r

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 29/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

Bringing it all together:

 h i
∆RN−A = λN t − λ A
t r + E [pt+1 ] − p t (1 + r )

2 ! "  #
µσρ 2
 
ρt 1+r
= 2γ − µρt
2γ µσρ 1+r
2
ρ2

1+r
= ρt − t
2γµ σρ

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 30/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

The expected returns of rational and noise traders:

The expected difference in return favors the noise traders!

"  #
ρ2t 1+r 2

E [∆RN−A ] = E ρt −
2γµ σρ

or
2 + ρ∗ 2 (1 + r )2


σρ
E [∆RN−A ] = ρ∗ −
2γµσρ2

because

Var ρ2t ≡ E ρ2t − E [ρt ]2 −→ E ρ2t = σρ2 + ρ∗ 2

     

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 31/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

σρ2 + ρ∗ 2 (1 + r )2


∗
E [∆RN−A ] = ρ −
2γµσρ2

Interpretation of terms:

ρ∗ : “Hold more” effect.

σρ2 term: the buy high-sell low or “Friedman” effect.

ρ∗ 2 term: “price pressure” effect.

The denominator: “create space” effect.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 32/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

σρ2 + ρ∗ 2 (1 + r )2


∗
E [∆RN−A ] = ρ −
2γµσρ2

Interpretation of terms:

Hold more and create space tend to raise noise traders’

relative expected returns.

The “Friedman” and price pressure effects-tend to lower noise

traders’ relative expected returns.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 33/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Summary:

Risk created by the unpredictability of noise-traders’ opinions

significantly reduces the attractiveness of arbitrage.

As long as arbitrageurs have short horizons and so must worry

about liquidating their investment in a mispriced asset, their
aggressiveness will be limited even in the absence of
fundamental risk.

Noise traders may be compensated for bearing the risk that

they themselves create and so earn higher returns than
sophisticated investors even though they distort prices.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 34/ 35
The DSSW Noise-Trader Model
DeLong, Shleifer, Summers, & Waldmann, “Noise Trader Risk . . . ,” (1990).

Summary:

The essential assumption used is that the opinions of noise

traders are unpredictable and arbitrage requires bearing the
risk that their misperceptions become even more extreme
tomorrow than they are today.

This model suggests that much of the behavior of professional

arbitrageurs can be seen as a response to noise trading rather
than as trading on fundamentals.

Lecture 7 – Noise-Trader Risk: R. J. Hawkins Econ 138: Financial and Behavioral Economics 35/ 35