Utility maximization problem with transaction costs - Shadow price approach Jin Hyuk Choi joint work with Mihai Sîrbu and Gordan Žitkovi´ c Department of Mathematics University of Texas at Austin Carnegie Mellon University Pittsburgh, Sep 12 th , 2011 1 / 25
25
Embed
Utility maximization problem with transaction costs ...s_slides.pdf · Utility maximization problem with transaction costs - Shadow price approach Jin Hyuk Choi joint work with Mihai
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Utility maximization problem with transaction costs- Shadow price approach
Jin Hyuk Choijoint work with Mihai Sîrbu and Gordan !itkovic
Department of MathematicsUniversity of Texas at Austin
Problem settings and related work! Merton problem! Shadow price approach
Heuristic derivation of the free boundary ODE
Our result (for power utility)
Value function < !"# $ Shadow price process"# Explicit condition for market parameters"# Existence of C2 solution to the free boundary ODE
Sketch of the proof! Existence of C2 solution! Construction and Verification of the Shadow process.
2 / 25
Market Model : Davis and Norman (’90), Shreve and Soner (’94)
Stock : dSt = St(µdt + !dWt), µ > 0, ! > 0.
Bond : B % 1.
Transaction costs : " > 0, " & (0, 1),
Bid St " (1 + ")St, Ask St " (1 ' ")St
Portfolio :
!"""#"""$
#0t : # of shares of B#t : # of shares of Sct : consumption rate
.
Initial position : (#00,#0)=($B, $S)
Admissibility : (#0,#, c) & A (S,",") i!
Self-financing d#0t = Std#
(t ' Std#)t ' ctdt
No-bankrupcy #0t + St#
+t ' St#'t * 0
where # = #)t ' #(t is the pathwide minimal decomposition of # into adi!erence of two non-decreasing processes. (#)t = Lt ,#
(t =Mt in Shreve and Soner (’94))
3 / 25
Investor : Davis and Norman (’90), Shreve and Soner (’94)
Utility functions (CRRA) : Up(x) =%
ln x, p = 0xp
p , p & ('!, 1) \ {0} , for x > 0.
Investor’s goal : to maximize expected utility by consumption
u(S,",") " sup(#0,#,c)&A (S,",")
E& ' !
0e'%tUp(ct)dt
(,
where % > 0 is a discount factor.
4 / 25
Related Work - no transaction cost, Melton(’71)
Proof : There exists an explicit solution to the HJB equation, we can writedown optimal strategy and do the verification.
Remark : For 0 < p < 1, u(S, 0, 0) < ! "# µ <)
2(1'p)!2%p .
5 / 25
Related Work - transaction cost
Magill and Constantinides (’76) : Optimal behavior (heuristic)Davis and Norman (’90) : Analytic proof, several technical conditions.Shreve and Soner (’94) : Rigorous proof, only assumed u(S,",") < !.
6 / 25
Shadow price approach
Motivation : In the frictionless market, we can use duality. Can weconstruct a stock price process which somehow absorbs transactioncosts? (S,",")! (S+, 0, 0).
Consistent price processes : S " {S : St , St , St , t * 0}Shadow price process : We call S+ & S a shadow price if
u(S,",") = u(S+, 0, 0) < !
Observations :
For S & S , u(S,",") , u(S, 0, 0)
u(S,",") = u(S+, 0, 0) = infS&S
u(S, 0, 0)
7 / 25
Related work - Construction of Shadow price
Kallsen and Muhle-Karbe (’10) showed that if p = 0 (log utility) andµ < !2, the shadow price process can be constructed.
They derived a free boundary ODE based on the following observation -the optimal strategy does not trade when St < S+t < St.
In log utility case (p=0), their methods work, becasue the explicitexpression for the optimal strategy exists.
But their methods only work for p = 0 (log utility).
How can we do for p " 0 ?
8 / 25
Our approach - derivation of free boundary problem
If the Shadow price process S+ exists,
u(S+, 0, 0) = infS&S
u(S, 0, 0)
Let’s parametrize S & S with an Ito-process Y :
St = eYt St,
dYt = *µtdt +*!tdWt,
with y , Yt , y, y " ln (1 ' "), y " ln (1 + ")
In the complete market, by convex duality,
u(S, 0, 0) = ($B+$SS0)p
p
+E& ' !
0E ('& ·W)
pp'1
t e%
p'1 tdt(,1'p,
where &t "µ+*µs+!*!s+
12*!
2s
*!s+!.
9 / 25
Our approach - derivation of free boundary problem
u(S+, 0, 0) = infS&S
u(S, 0, 0)
= infY
u(S, 0, 0)
= infy&[y,y]
inf*µ,*!
u(S, 0, 0)
= infy&[y,y]
($B+$SeyS0)p
p w(y)1'p
where w(y) " inf*µ,*!E& ' !
0E ('& ·W)
pp'1
t e%
p'1 tdt(.
We set dP = E ( p1'p& ·W)dP, then,
w(y) = inf*µ,*!EP& ' !
0e%
p'1 tep
2(1'p)2
- t0 &
2udu
dt(.
We can write down the HJB-equation for this optimal stochastic control.
10 / 25
Our approach - derivation of free boundary problem
HJB equation : For y & (y, y),
inf*µ,*!
.( p
2(1'p)2&2 ' %
1'p )w(y) + (*µ + p1'p*!&)w-(y) + 1
2*!w--(y) + 1
/= 0
Boundary condition : At y and y, turn o! the di!usion, which correspond to
w--(y) = w--(y) = !
Order reduction : Set w(y) = g(w-(y)) for g : [x, x] ./ R, (x = w-(y), x = w-(y)),
inf*µ,*!
.( p
2(1'p)2&2 ' %
1'p )g(x) + (*µ + p1'p*!&)x + 1
2*!x
g-(x) + 1/= 0,
with g-(x) = g-(x) = 0,' x
x
g-(x)x dx = y ' y.
(boundary conditions from g-(x) = xI-(x) = xw--(I(x)) , where I(x) = (w-)'1(x).)
11 / 25
Free boundary ODE
Finally, after simple change of variable, we obtain a free boundary problem :!""#""$
By using the series expension of g, we may find a fractional tailor expansionfor any order.
22 / 25
Summary
We get a direct proof to the original control problem, by constructing theshadow price process.
We get an explicit necessary and su"cient condition for finiteness of thevalue function.
We may do sensitivity analysis for small transaction cost, any order.
We may extract more financial interpretations from the structure of theODE.
23 / 25
References
A. Skorokhod. Stochastic equations for di!usion processes in a boundedregion. Theory of Probability and its Applications, 6:264-274, 1961.
J. Kallsen and J. Muhle-Karbe. On using shadow prices in portfoliooptimization with transaction costs. Annals of Applied Probability,20:1341âAS1358, 2010.
M. Davis and A. Norman. Portfolio selection with transaction costs.Mathematics of Operations Research, 15:676-713, 1990.
S. Shreve and M. Soner. Optimal investment and consumption withtransaction costs. The Annals of Applied Probability, 4:609-692, 1994.