Asset Pricing

under Asym.
Information

Rational
Expectation
Equilibria

Classification
of Models
Asset Pricing under Asymmetric Information
Static
Uniform Price
Discr. Price
Rational Expectations Equilibrium
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade Markus K. Brunnermeier
Herding
1987-crash
Princeton University

November 16, 2015

Asset Pricing
under Asym.
Information
A Classification of Market
Rational
Expectation
Equilibria
Microstructure Models
Classification
of Models
• simultaneous submission of demand schedules
Static
Uniform Price • competitive rational expectation models
Discr. Price
(Limit Order • strategic share auctions
Book)
Contrast
• sequential move models
Dynamic
Sequential Trade
• screening models in which the market maker submits a
Herding supply schedule first
1987-crash
• static
 uniform price setting
 limit order book analysis
• dynamic sequential trade models with multiple trading
rounds
• strategic market order models where the market maker sets
prices ex-post
Asset Pricing
under Asym.
Information
Screening Models à la Glosten
Rational
Expectation
Equilibria

Classification
of Models

Static 1 Uninformed (risk-neutral) market maker sets whole supply

Uniform Price
Discr. Price schedule
(Limit Order
Book) • market making sector is competitive
Contrast

Dynamic
• oligopolistic market making sector
Sequential Trade • market maker is monopolist
Herding
1987-crash 2 Possibly informed trader submits
• a single order which is executed at uniform price
• many little orders in order to “walk along the limit order
book” (discriminatory prices)
Asset Pricing
under Asym.
Information
Uniform Price Setting - Glosten 1989
Rational
Expectation
Equilibria

Classification
of Models
• Contrast competitive market maker sector with
Static
Uniform Price monopolistic market maker (specialist system NYSE).
Discr. Price
(Limit Order
Book)
• Model setup
Contrast
• market maker(s) set price (supply) schedule
Dynamic
• single trader submits order
Sequential Trade
Herding • risk-averse with CARA utility function
1987-crash
• endowment shock of u
• private signal S i = v + 
• two-dimensional screening problem
Glosten (1989) reduces it to a one-dimensional problem
(see later)
Asset Pricing
under Asym.
Information
Uniform Price Setting - Glosten 1989
Rational
Expectation • Competitive price schedule: P co = E [v |x]
Equilibria
• Perfect competition
Classification • ⇒ expected profit for any order size x is ZERO.
of Models • prevents market makers from effectively screening orders
Static • ⇒ leads to instability
Uniform Price
Discr. Price formally, existence problem for certain parameters
(Limit Order
Book) (Hellwig JET 1994 shows that this is due to unbounded
Contrast
support of type sapce and it existence problem is different
Dynamic to the one in Rothschild & Stiglitz)
Sequential Trade
Herding
1987-crash
• Monopolistic price schedule:

P mo = arg max E [[P mo (x ∗ (·)) − v ]x ∗ (·)],

where x ∗ (·) is the optimal order size.
• principal-agent problem
• principal sets menu of contracts (x, P mo (x))
• Cross-subsidization: large profit from small trades
small (-ve) profit from large trades
• market with monopolistic setting stays open for larger
trade sizes than a market with multiple market markers
Asset Pricing
under Asym.
Information Discrim. Pricing (Limit Order Book)
Rational
Expectation Glosten 1994 - BRM 2000
Equilibria

Classification
of Models
• “upper tail” conditional expectations for next marginal
Static
Uniform Price
order y
Discr. Price
(Limit Order P co (y ) = E [v |x ≥ y ]
Book)
Contrast
• trader who buy only a tiny marginal quantity have to pay
Dynamic
Sequential Trade a higher (ask) price ⇒ small trade spread
Herding
1987-crash
• competitive market makers do not know whether trader
only buys first marginal unit or continues to buy further
units.
• cross-subsidization from small orders to large orders
• limit order book is immune to “cream skimming” of orders
by competing exchanges (no advantage of order splitting).
Asset Pricing
under Asym.
Information Discrim. Pricing - Biais, Rochet & Martimo
Rational
Expectation Oligopolistic Market Makers
Equilibria

Classification
of Models • oligopolistic screening game (special cases I = 1, I = ∞)
Static
Uniform Price
• Stage 1: risk-neutral market maker(s)
Discr. Price
(Limit Order set supply schedule p(x) (limit order book)
Book)

• Stage 2: informed trader buy x = Ii x i shares

Contrast
P
Dynamic
Sequential Trade • x i for market maker i R xi
Herding
transfer to mm i: t i (x i ) = 0 p(q)dq, T (x) = i t i (x i )
P
1987-crash •
• trader’s endowment shock u
• trader’s signal
S, where v = S + ε.
ε ∼ N 0, σ 2
u and S have bounded support.
• trader’s final wealth W = v (u + x) − i t i (x i )
P
(conditional on u, S, wealth W is normally distributed
with E [v |S] = S, Var [v |S] = Var [ε])
Asset Pricing
under Asym.
Information
BRM: One Dimensional Screening
Rational
Expectation
Equilibria
• Stage 2: (ctd.) - “Glosten (1989)-trick”
• with CARA utility function
Classification
of Models ρ
Static
E [W |u, S] − V [W |u, S]
Uniform Price
2
Discr. Price
ρ 2
(Limit Order = (x + u) S − T (x) − (x + u) Var [v |S]
Book) 2 | {z }
Contrast
=σ 2
Dynamic  
Sequential Trade
Herding
ρσ 2 2 ρσ 2 2
   
1987-crash
= uS − u +  xS − ρσ 2 xu − x − T (x)
 
2 | {z } 2 
| {z } θx
independent of x i.e. θ:=S−ρσ 2 u
| {z }
depends on x
=⇒Info-Rent

• This reduces it to a one-dimensional screening problem

• function v (θ) = E [v |θ] of (one-dimensional) type θ
·
1 ≥ v (θ) ≥ 0
Asset Pricing
under Asym.
Information
BRM: First Best Benchmark
Rational
Expectation
Equilibria
• ex-ante  
transfers trading volume
 ↓
 ↓ 
Classification
of Models optimal trading mechanism τ (θ) , x (θ)
 
Static
 
Uniform Price
Discr. Price
(Limit Order
θ
ρσ 2
Book)
Z  
2
Contrast
max θx (θ) − x (θ) − τ (θ) f (θ) dθ
Dynamic {τ (θ),x(θ)} θ 2
Sequential Trade
Herding
1987-crash
Z θ
s.t. (τ (θ) − v (θ) x (θ)) f (θ) dθ = Π
θ
• Π determines how surplus is distributed between P and A

θ
ρσ 2
Z
=⇒ max (θx (θ) − x (θ)2 − v (θ) x (θ) −Π)f (θ) dθ
θ | 2 {z }
surplus
Asset Pricing
under Asym.
Information
BRM: First Best Benchmark
Rational
Expectation
Equilibria
for a given θ
Classification
of Models

Static
θ − ρσ 2 x (θ) − v (θ) = 0
Uniform Price
θ − v (θ)
Discr. Price
(Limit Order x ∗ (θ) =
Book) ρσ 2
Contrast

Dynamic
θ−S
= E [−u|θ] , since u = −
Sequential Trade
Herding
ρσ 2
1987-crash

• Assume x ∗ (θ) < 0 < x ∗ θ

=⇒ ∃θ0 s.t. x ∗ (θ0 ) = 0

• almost all θ-types trade
(see later that ∀θ > θ0 =⇒ buy
∀θ < θ0 =⇒ sell )
Asset Pricing
under Asym.
Information BRM: Monopolistic Screening
Rational
Expectation x ∗ (θ) and xm (θ)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: xxx. xx
Asset Pricing
under Asym.
Information
BRM: Implementable Allocation
Rational
Expectation
Equilibria
under Adverse Selection
• social planner must elicit information
Classification
of Models • Revelation Principle
Static
Uniform Price
Any allocation that can be achieved with non-linear
Discr. Price
(Limit Order
schedules T (x) can also be achieved with a truthful direct
Book)
Contrast mechanism {τ (·) , x (·)}.
Dynamic • Incentive compatibility
Sequential Trade
Herding

ρσ 2  b2
1987-crash
    
θ ∈ arg max θx θb − x θ − τ θb
θb 2
 
   ρσ 2  2  
=⇒ U (θ) = max  θx b −
θ x b − τ θb 
θ
θb

| 2 {z }

informational rent
{τ (·) , x (·)} transfers and allocation
Asset Pricing
under Asym.
Information
BRM: Dual (Mirrlees) Approach
Rational
Expectation {U (·) , x (·)} informational rent (see Fudenberg & Tirole Ch.
Equilibria
7)
Classification
of Models Lemma 1:
Static
Uniform Price
A pair {U (·) , x (·)} is implementable iff U (·) is convex on
Discr. Price   ·
(Limit Order
Book) θ, θ , and for a.e.θ, U (θ) = x (θ) ,
Contrast
dU (θ,θ(θ)
b )
∂U
Dynamic
dθ = ∂θ = x (θ) .
Sequential Trade ↑
Herding envelope theorem
1987-crash
Asset Pricing
under Asym.
Information
BRM: Monopolistic Screening
Rational
Expectation Rθ
Equilibria
m.m.(principal) gets θ τ (x(θ)) − v (θ)x(θ) replacing τ
ρσ 2 2
Classification
of Models
from information rent U(θ) = θx(θ) − 2 x (θ) − τ (x(θ)), the
Static m.m.’s objective becomes
Uniform Price
Discr. Price
(Limit Order
Book)
θ
ρσ 2
Contrast
Z
Dynamic max {[θ − v (θ)]x(θ) − [x(θ)]2 − U(θ)}f (θ)dθ
Sequential Trade {U(·),x(·)} θ 2
Herding
1987-crash
subject to
  
U (·) is convex on θ, θ
IC
U̇ (θ) = x (θ) ∀θ (almost everywhere)

ex-post PC U (θ) ≥ 0 ex-post participation constraints

(ex-post: since traders decide after knowing θ whether to participate)
Asset Pricing
under Asym.
Information BRM: Monopolistic Screening
Rational
Expectation Dual Approach
Equilibria ·
(replace x(θ) with U(θ) )
Classification
of Models  
Static
max Bm U (·) , U̇ (·)
Uniform Price
U(·)
Discr. Price
θ
ρσ 2
(Limit Order
Z  
Book) 2
Contrast := [θ − v (θ)] U̇ (θ) − U̇ (θ) − U (θ) f (θ) dθ
Dynamic θ 2
Sequential Trade
Herding
1987-crash U (·) convex
s.t.
U (θ) ≥ 0
Temporarily ignore convexity constraint and check ex-post.
(Sufficient condition: U (·) is convex if
 
d 1−F (θ)
∀θ > θ0 dθ <0 (18)
 f (θ)
d F (θ)
∀θ < θ0 dθ f (θ) > 0 (19)
Asset Pricing
under Asym.
Information
BRM: Monopolistic Screening
Rational
Expectation
Equilibria

Classification
of Models
    Z θ
Static
L U, U̇ = Bm U, U̇ + U (θ) dΛ (θ)
Uniform Price θ ↑
Discr. Price
(Limit Order
Book)
∞ many Lagrange multipliers
Contrast
different from type to type
Dynamic
Sequential Trade
(ex-post constraint)
Herding
1987-crash
By complementary slackness condition, support of Λ be
constrained in (Um )−1 (0), (θ-types which get zero info ret)
view Λ (θ) as c.d.f., i.e., ∃ a measure Λ
Z θ
dΛ (s)
Λ (θ) = Rθ (slight abuse of notation)
θ dΛ (s)
θ
Asset Pricing
under Asym.
Information
BRM: Monopolistic Screening
Rational
Expectation Aside: Integrating by parts
Equilibria
Z θ Z θ

Classification
of Models
U (θ) d [Λ (θ) − F (θ)] = − U̇ (θ) (Λ (θ) − F (θ)) dθ+U θ
θ θ
Static
Uniform Price
 
Discr. Price
(Limit Order
Consequently, max L U, U̇ =
Book)
Contrast
R θ  F (θ)−Λ(θ)

ρσ 2

Dynamic
Sequential Trade
= θ θ − v (θ) + f (θ) U̇ (θ) − 2 U̇ (θ)2 f (θ) dθ
Herding

1987-crash
+U θ (Λ (θ) − 1)


max only if Λ (θ) = 1 (since U θ is arbitrary)

pointwise maximization over U̇ (θ)

  θ − v (θ) F (θ) − Λ (θ)
∀θ ∈ θ, θ , xm (θ) = +
ρσ 2 f (θ) ρσ 2
| {z }
x ∗ (θ)
Asset Pricing
under Asym.
Information
BRM: Monopolistic Screening
Rational
Expectation
Equilibria

Classification Complementary slackness condition (dΛ = 0 for some θ)

of Models

Static ∀θ ∈ [θ, θbm ] Λ (θ) = 0

Uniform Price
Discr. Price
(Limit Order
∀θ ∈ θam , θ Λ (θ) = 1
Book)
Contrast

Dynamic =⇒ given (18) & (19), U (·) is convex and

Sequential Trade
Herding
1987-crash Proposition 2
∃θam > θ0 and θbm < θ s.t.
(i) for all θ ∈ [θ, θbm ) , xm (θ) = x ∗ (θ) + ρσF2(θ)
f (θ)
(ii) for all θ ∈ [θbm , θam ] , xm (θ) = 0 (no info rent)
1−F (θ)
(iii) for all θ ∈ θam , θ , xm (θ) = x ∗ (θ) − ρσ

2 f (θ)
Asset Pricing
under Asym.
Information BRM: Monopolistic Screening
Rational
Expectation x ∗ (θ) and xm (θ)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: xxx. xx
Asset Pricing
under Asym.
Information BRM: Monopolistic Screening
Rational
Expectation Price Schedule
Equilibria
for θ > θam we know
Classification Rθ Rθ
of Models (1) θ ≥ θam U (θ) = 0 + θm U̇ 0 (s) ds = θm x (s) ds
a a
Static ρσ 2 xm (θ)2
Uniform Price (2) U (θ) = θxm (θ) − 2 − T (x (θ))
Discr. Price
(Limit Order ρσ 2 xm (θ)2 Rθ
Book) (1)=(2) T (x (θ)) = θxm (θ) − 2 − θm x (s) ds
Contrast a
Dynamic
Differentiate w.r.t. θ
Sequential Trade
Herding ∂T ∂x ∂x ∂xm
1987-crash = xm (θ) + θ − ρσ 2 xm (θ) − xm (θ)
∂x ∂θ ∂θ ∂θ
∂T
= θ − ρσ 2 xm (θ)
∂x
We have
F (θ)
xm (θ) = x ∗ (θ) + 2
| {z } ρσ f (θ)
θ−v (θ)
ρσ 2
Asset Pricing
under Asym.
Information BRM: Monopolistic Screening
Rational
Expectation Price Schedule
Equilibria

Classification
of Models
∂T F (θ)
Static =⇒ = θ − θ + v (θ) −
Uniform Price ∂x f (θ)
Discr. Price
(Limit Order ∂T F (θ)
Book)
Contrast tm (x) = = v (θ) −
∂x f (θ)
Dynamic
Sequential Trade
Herding
1987-crash
for θ < θbm similar steps

∂T 1 − F (θ)
= v (θ) +
∂x f (θ)

Note that

tm x = 0+ = θam > θbm = tm x = 0−

“small trade spread”

Asset Pricing
under Asym.
Information
BRM: Oligopolistic Screening
Rational
Expectation
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic ...
Sequential Trade
Herding
1987-crash
Asset Pricing
under Asym.
Information Limit Order Book vs. Uniform Pricing
Rational
Expectation Röell (1998)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order • Model setup
Book)
Contrast • order size of trader is exogenous
Dynamic • is double exponentially distributed f (x) = 12 ae −a|x|
Sequential Trade
Herding • conditional expectations
1987-crash
• E [·|x ≥ y ] ⇒ linear schedule in limit order book
• E [v |x] = v0 + γx assumed ⇒ linear uniform price schedule
• p u (x) = v0 + II −1 d I γ
−2 γx versus p (x) = v0 + I −1 a + γx
Asset Pricing
under Asym.
Information Limit Order Book vs. Uniform Pricing
Rational
Expectation Röell (1998)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: Limit Order Book.

Asset Pricing
under Asym.
Information Limit Order Book vs. Uniform Pricing
Rational
Expectation Röell (1998)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: Limit Order Book and Uniform Pricing.

Asset Pricing
under Asym.
Information
A Classification of Market
Rational
Expectation
Equilibria
Microstructure Models
Classification
of Models
• simultaneous submission of demand schedules
Static
Uniform Price • competitive rational expectation models
Discr. Price
(Limit Order • strategic share auctions
Book)
Contrast
• sequential move models
Dynamic
Sequential Trade
• screening models in which the market maker submits a
Herding supply schedule first
1987-crash
• static
 uniform price setting
 limit order book analysis
• dynamic sequential trade models with multiple trading
rounds
• strategic market order models where the market maker sets
prices ex-post
Asset Pricing
under Asym.
Information
Sequential Trade Models à la
Rational
Expectation
Equilibria
Glosten & Milgrom (1985)
Classification
• order size is restricted to x ∈ {−1, +1}
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: Bid-Ask Spread.

Asset Pricing
under Asym.
Information
Sequential Trade Models à la
Rational
Expectation
Equilibria
Glosten & Milgrom (1985)
Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
• Monopolistic Market Maker - Copeland & Galai (1983)
Contrast
• bid-ask spread is partially due to monopoly power
Dynamic
Sequential Trade
partially due to adverse selection
Herding • difficult to handle in multi-period setting
1987-crash

• Competitive Market Makers - Glosten & Milgrom (1985)

• bid-ask spread is only due to adverse selection
• multi-period setting
Asset Pricing
under Asym.
Information
Glosten & Milgrom (1985)
Rational
Expectation
Equilibria

Classification
of Models
• Model Setup
Static • value of the stock v and v
Uniform Price • with probability α an informed trader shows up
Discr. Price
(Limit Order
Book)
• with probability (1 − α) an uninformed trader shows up
Contrast • all traders are chosen from a pool of a continuum of
Dynamic
Sequential Trade
traders, i.e., the probability that they will trade a second
Herding time is zero (rule out strategic considerations as in Kyle’s)
1987-crash
• informed traders know true ve → buys if v > a sells if
v < b.
• uninformed traders buy with probability µ and sell with
probability 1 − µ.
• Note: Traders can only buy or sell 1 unit (No-Trade is also
not allowed!)
Asset Pricing
under Asym.
Information
Glosten & Milgrom (1985)
Rational
Expectation
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: Tree.
Asset Pricing
under Asym.
Information Glosten & Milgrom (1985)
Rational
Expectation Calculating Bid-Ask Spread
Equilibria

Classification
of Models

Static
• Buy order
Uniform Price
Discr. Price
(Limit Order
Book)
P (v ) = θ
Contrast
P (buy|v ) = α + (1 − α) µ
Dynamic
Sequential Trade
Herding
P (buy|v ) = (1 − α) µ
1987-crash

Bayes’ Rule

(α + (1 − α) µ) θ
P (v |buy) =
(α + (1 − α) µ) θ + (1 − α) µ (1 − θ)
P (v |buy) = 1 − P (v |buy)
Asset Pricing
under Asym.
Information Glosten & Milgrom (1985)
Rational
Expectation Calculating Bid-Ask Spread
Equilibria

Classification
of Models
• Sell order P (v |sell) =
Static
Uniform Price
Discr. Price (1−α)(1−µ)θ
(Limit Order
Book)
= (1−α)(1−µ)θ+[α+(1−α)(1−µ)](1−θ)
Contrast

Dynamic
P (v |buy) > P (v ) > P (v |sell)
Sequential Trade
Herding
P (v |buy) < P (v ) < P (v |sell)
1987-crash

• Market Maker makes zero expected profit

(potential Bertrand competition)

b = bid = E [v |sell] = v P (v |sell) + v P (v |sell)

a = ask = E [v |buy] = v P (v |buy) + v P (v |buy)
Asset Pricing
under Asym.
Information
Remarks to Glosten & Milgrom (1985)
Rational
Expectation
Equilibria

Classification 1 quotes are regret free

of Models

Static 2 v< b < a < v

Uniform Price
Discr. Price 3 (a − b) → gain from liquidity traders = loss to insider
(Limit Order
Book)
Contrast 4 bid-ask spread (a − b) increases with α
Dynamic
Sequential Trade
5 over time price converge to true value
Herding
prices follow a martingale Et pt+1 |Itl = pt
 
1987-crash 6
(changes in prices are uncorrelated)
7 Simple setting price at t depends only on # buy orders −
# sell orders (sequence of trades does not matter)
a+b
8 mid point of bid ask spread 2 is not current market
maker’s expectation.
Asset Pricing
under Asym.
Information
Extensions
Rational
Expectation
Equilibria

Classification
• Easley and O’Hara (1987)
of Models • ‘small and large’ order size
Static • noise traders submit randomly a small or a large sized
Uniform Price
Discr. Price order
(Limit Order
Book) • informed traders always prefer large order size (if bid and
Contrast
ask is the same for both order sizes)
Dynamic
Sequential Trade
⇒ m.m. will set larger spread for large orders
Herding • Separating equilibrium
1987-crash
• Informed traders’ order size is 2
• Uninformed traders’ order size is 1 and 2 (exogenously given)
⇒ Spread for small orders = 0
• Pooling equilibrium
• Informed traders’ order size is 1 and 2
• Uninformed traders’ order size is 1 and 2 exogenously given)
⇒ Larger spread for larger orders
Asset Pricing
under Asym.
Information
Extensions
Rational
Expectation
Equilibria

Classification
of Models • “event uncertainty” (also in Easley & O’Hara (1992))
Static • with prob γ info is like in Glosten & Milgrom
Uniform Price
Discr. Price • with prob (1 − γ) no news event occurs
(Limit Order
Book) (nobody receives a signal)
Contrast

Dynamic • No-Trade → signals that nothing has occurred!

Sequential Trade 1
Herding ⇒ quotes will pull towards 2
1987-crash
updating
1 whether event has occurred AND
2 about true value of the stock
• transaction price is still a Martingale
but no longer Markov!
Asset Pricing
under Asym.
Information
Herding - Avery & Zemsky (1998)
Rational
Expectation
Equilibria

Classification
of Models

Static • Relates Glosten-Milgrom model to herding models (BHW

Uniform Price
Discr. Price
1992)
(Limit Order
Book)
Contrast
• Price adjustment eliminates herding and informational
Dynamic cascades if market maker learns at the same speed as
Sequential Trade
Herding other informed traders.
1987-crash
• Herding can still arise in a more general setting with event
uncertainty and a more complicated information structure
which guarantees that the market maker learns at a slower
speed compared to other traders.
Asset Pricing
under Asym.
Information 1987-Crash
Rational
Expectation Jacklin, Kleiden & Pfleiderer (1992)
Equilibria

Classification
of Models

Static
Uniform Price
Discr. Price
(Limit Order
Book)
Contrast

Dynamic
Sequential Trade
Herding
1987-crash

Figure: Underestimating portfolio insurance traders θ.