Professional Documents
Culture Documents
Behavioral Finance
Ulm University
1 Teaching Goals
1 Teaching Goals
The influence of trading strategies that aim to exploit mispricings can be limited by:
Fundamental risk
Implementation costs
1 Teaching Goals
1 Teaching Goals
1 Teaching Goals
Assumptions (I)
DeLong et al (1990): ”Noise Trader Risk in Financial Markets”
∙ their demand for risky assets is affected by their beliefs and sentiment that are not fully
justified by fundamental news
∙ noise traders tend to overreact/underreact to news
2. Arbitrage is risky
Implications
Assumptions (II)
DeLong et al (1990): ”Noise Trader Risk in Financial Markets”
∙ irrational traders
∙ react on noise rather than on true information
∙ neglect the existence of sophisticated traders
∙ return expectation: Rte = E[Rt ] + ρt , with ρt ∼ N (ρ∗ , σρ2 )
∙ noise traders estimate is above or below the rational expectations benchmark
1 Teaching Goals
1 Teaching Goals
Expected wealth
Expected utility
∂E[U ]
∙ first-order condition: ∂λ
=0
∙ becomes: r + pet+1 − pt (1 + r) − 2γV [pt+1 ] = 0
∙ solution
r + pet+1 − pt (1 + r)
λ= (5)
2γV [pt+1 ]
r + pet+1 − pt (1 + r)
λs = (6)
2γV [pt+1 ]
e
Rt+1
e
Rt+1 =r+ pet+1 − pt (1 + r) (7) λs = (8)
2γV [pt+1 ]
1 Teaching Goals
1 Teaching Goals
Maximizing U (wt ) with respect to λ leads to the optimal demand λn of noise traders
n r + ρt + pet+1 − pt (1 + r)
λ = (9)
2γV [pt+1 ]
Portfolio Return
e e
Rt+1 = r + ρt + pt+1 − pt (1 + r) (10)
1 Teaching Goals
1 Teaching Goals
e
Rt+1 R e + ρt
+ µ t+1
(1 − µ) = 1 (12)
2γV [pt+1 ] 2γV [pt+1
r + pet+1 − pt (1 + r) r + pet+1 + ρt − pt (1 + r)
(1 − µ) +µ = 1 (13)
2γV [pt+1 ] 2γV [pt+1 ]
1 e
pt+1 = r + pt+2 − 2γV [pt+2 ] + µρt+1
1+r
1 e
E[pt+1 ] = E r + pt+2 − 2γV [pt+2 ] + µρt+1
1+r
1
= r + E[pt+2 ] − 2γV [pt+2 ] + µE[ρt+1 ]
1+r
1 ∗
= r + E[pt+1 ] − 2γV [pt+1 ] + µρ
1+r
Rearranging yields
∗
(1 + r)E[pt+1 ] = r + E[pt+1 ] − 2γV [pt+1 ] + µρ
∗
rE[pt+1 ] = r − 2γV [pt+1 ] + µρ
2γV [pt+1 ] µρ∗
E[pt+1 ] = 1− +
r r
1
e
V [pt+1 ] = V r + pt+2 − 2γV [pt+2 ] + µρt+1
1+r
1
= V µρt+1
1+r
µ 2
= [ρt+1 ]
1+r
µ 2 2
= σρ
1+r
1 e
pt = r + pt+1 − 2γV [pt+1 ] + µρt (16)
1+r
with
2γV [pt+1 ] µρ∗
E[pt+1 ] = 1 − + (17)
r r
and
µ 2 2
V [pt+1 ] = σρ (18)
1+r
1 e
pt = r + pt+1 − 2γV [pt+1 ] + µρt
1+r
1 2γV [pt+1 ] µρ∗
= r+1− + − 2γV [pt+1 ] + µρt
1+r r r
µ
2
1 2γ 1+r σρ2 µρ∗ µ 2 2
= r+1− + − 2γ σρ + µρt
1+r r r 1+r
µ
2
1 2γ 1+r σρ2 µρ∗ µ 2 2
= r+1− + − 2γ σρ + µρt
1+r r r 1+r
(19)
µ
2
1 2γ 1+r σρ2 µρ∗ µ 2 2
pt = r+1− + − 2γ σρ + µρt
1+r r r 1+r
µ
2
1 µρ∗ 2γ 1+r σρ2 µ 2 2
= 1+r+ + µρt − − 2γ σρ
1+r r r 1+r
µ
2
1 µρ∗ 2γ 1+r σρ2 µ 2 2
= 1+r+ + µρt − + 2γ σρ
1+r r r 1+r
∗
1 µρ 2γ µ 2 2 2γr µ 2 2
= 1+r+ + µρt − σρ + σρ
1+r r r 1+r r 1+r
∗
1 µρ 2γ(1 + r) µ 2 2
= 1+r+ + µρt − σρ
1+r r r 1+r
1 µρ∗ 2γ(1 + r) µ 2 2
pt = 1+r+ + µρt − σρ
1+r r r 1+r
∗
1 µρ ∗ ∗ 2γ(1 + r) µ 2 2
= 1+r+ + µρt + µρ − µρ ρt − σρ
1+r r r 1+r
1 µρ∗ ∗r ∗ 2γ(1 + r) µ 2 2
= 1+r+ + µρ + µ(ρt − ρ ) − σρ
1+r r r r 1+r
1 (1 + r)µρ∗ ∗ 2γ(1 + r) µ 2 2
= 1+r+ + µ(ρt − ρ ) − σρ
1+r r r 1+r
µρ∗ µ(ρt − ρ∗ ) 2γ µ 2 2
= 1+ + − σρ (20)
r 1+r r 1+r
µρ∗ µ(ρt − ρ∗ ) 2γ µ 2 2
pt = 1+ + − σρ (21)
r 1+r r 1+r
−→ the only source of price variability is noise trader risk (since we abstract from fundamental
risk)
−→ since σρ2 > 0 we observe excess volatility
1 Teaching Goals
1 Teaching Goals
1. No noise traders: µ = 0
pt = 1 (22)
∙ asset is traded at a discount, since noise traders still create risks through
periods of bullish and bearish moods
1 Teaching Goals
1 Teaching Goals
∙ irrational traders can impact asset prices and this impact cannot be
arbitraged away
∙ arbitrage can be risky, risk-averse arbitrageurs lower their demand
∙ hence, mispricings can exist for a long time
∙ return differential
n s e
∆R = (λ − λ )(r + pt+1 − pt (1 + r)) (26)
∙ the difference in share demand is
n s r + pet+1 − pt (1 + r) + ρt r + pet+1 − pt (1 + r)+
λ −λ = − (27)
2γV [pt+1 ] 2γV [pt+1 ]
ρt ρt
= = 2 (28)
2γV [pt+1 ] 2γ µ
σρ2
1+r
∙ λn > λs when ρt > 0: when noise traders overprice the risky asset, they hold more
quantities of the risky asset
Excess returns
e e 1 e
r + pt+1 − pt (1 + r) = r + pt+1 − r + pt+1 − 2γV [pt+1 ] + µρt (1 + (29)
r)
1+r
e e
= r + pt+1 − r − pt+1 + 2γV [pt+1 − µρt (30)
= 2γV [pt+1 − µρt (31)
µ 2 2
= 2γ σρ − µρt (32)
1+r
∙ excess returns increase in the variance of the mispricing σρ2 , but decrease in the overpricing
itself ρt
∙ higher variance in the mispricing leads to a higher variance in noise traders’ demand for the
risky asset which leads to higher price volatility
∙ for the market to clear, the risky asset must trade at a discount which raises its expected
return
∙ overpricing pushes up noise traders demand, which increases the price and lowers the return
n s e ρt µ 2 2
(λ − λ )(r + pt+1 − pt (1 + r)) = 2 2γ σρ − µρt (33)
2γ µ
σr2 ho 1+r
1+r
(ρt )2
= ρt − µ 2
(34)
2γ (1+r)2σ ρ
E[(ρt )2 ] ∗
σρ2 + (ρ∗ )2
E[ρt ] − µ = ρ − µ (35)
2γ (1+r)2 σ 2 ρ 2γ (1+r)2σ ρ
2
∙ if they achieve a large variance in mispricing. This will discourage rational traders from
buying the asset, it drives the price of the risky asset down and raises its expected return
1 Teaching Goals
1 Teaching Goals
Twin Shares
∙ in 1907 Royal Dutch and Shell agreed to merge in al 60:40 basis while
remaining separate entities
∙ Royal Durch share is a claim of 60 percent of the total cash flow
∙ Shell share is a claim of 40 percent of the total cash flow
∙ fundamental value: price of Royal Dutch should be 1.5 times the market
value of Shell
∙ empirically:
∙ strong deviations from 1.5
∙ Royal Dutch is sometimes 35 percent underpriced and sometimes 15 percent
overpriced
Index Exclusions
∙ every few weeks a stock is removed from S&P500 and a new stock is
included
∙ index should be representative for the US economy, no information about
the riskyness of a firm
∙ in an efficient market such an inclusion should not lead to a price change
since fundamental values stay the same
∙ empirically:
∙ average price jump after inclusion is 3.5 percent
∙ after Yahoo was added it price jumped by 24 percent within one day