Professional Documents
Culture Documents
1 / 13
Outline
What is a pricing kernel? Liu, Shackleton, Taylor, and Xu (2009) Kuo, Liu, and Coakley (2010)
2 / 13
Option pricing
in general we have c (K ) = e rT E Q [(ST K )+ ] = e rT
0
(x K )+ fQ (x )dx (x K )+ fQ (x ) fP (x )dx fP (x )
= e rT
0
=
0
= E P [m(ST )(ST K )+ ] the pricing kernel, also called the stochastic discount factor, is the random variable m(ST ) fQ (x ) m(x ) = e rT fP (x )
Pricing Kernel (USTC) 3 / 13
A representative agent
In a general equilibrium, assume a representative agent faces the problem of maximizing expected utility over consumption, given budget constraint and market clearing conditions,
max Et [
j =0
j u (ct +j )]
4 / 13
subject to ct = et pt ct +1 = et +1 + xt +1 where : e: p: x: the holding of the asset by the agent the original consumption level if nothing is traded price of the asset payo of the asset
5 / 13
Forming a Lagrangian L = u (ct ) + Et [u (ct +1 )] + t (et ct pt ) + t +1 (et +1 ct +1 + xt +1 ) taking the rst order conditions, L = 0 u (ct ) t = 0 ct L = 0 Et [u (ct +1 )] t +1 = 0 ct +1 L = 0 t pt + t +1 xt +1 = 0
6 / 13
7 / 13
the basic pricing formula for all assets is pt = Et [mt +1 xt +1 ] or simply E [mR ] = 1 the pricing kernel must be positive it must be downward sloping
8 / 13
the pricing kernel is estimated as the ratio of RND and physical distributions RND is assumed to follow (1) a mixture of 2 lognormal distributions (2) the generalized beta distribution of the 2nd kind (GB2) (3) the cubic spline physical distribution is estimated using the GJR-GARCH model the geometric mean of the ratios over time are relatively smooth; its generally downward sloping but there is a small hump in the middle
9 / 13
specify functional form for the pricing kernel for LIBOR futures options follows the intertemporal CAPM of Merton (1973) and makes use of state variables, which are economic variables whose innovations can forecast future returns sample period is from 2000.01 to 2008.2, monthly frequency
10 / 13
Test the following two functional forms with GMM A power function of returns on wealth and an exponential function of the sv m = c0 (1 + RW ) e c1 r +c2 +c3 with moment condition i ,t +1 )|It ] = 1 E [c0 (1 + RW ) e c1 r +c2 +c3 (1 + R
11 / 13
A sum of orthogonal Chebyshev polynomials in wealth and an exponential function of the sv (Chapman (1997) and Rosenberg and Engle (2002)) m = (RW )e c1 r +c2 +c3
n
(RW ) = 0 C0 (1 + RW ) +
k =1
k Ck (1 + RW )
12 / 13
To Recap
the pricing kernel contains information essential for asset pricing E (mR ) = 1 it can be derived as the ratio between RND and physical distribution, hence intimate relation between the pricing kernel, RND, and physical distribution alternatively, we can assume specic functional forms that contains a measure for investor risk aversion using econometric method, we can estimate a pricing kernel that conforms to economic theory by being everywhere positive and monotonically downward slopingh
13 / 13
Empirical Pricing Kernels for FTSE 100 Index Options from 1993.07 to 2003.12
3
10
41
1.2 1.15 1.1 1.05 1 0.95 0.9 0.85 0.8 0.9 0.95 1 1.05 Market return 1.1
0 0 5 5 10 1 sv 2 sv 3 sv 0.95 1 1.05 Market return 1.1 10 15 20 0.9 1 sv 2 sv 3 sv 0.95 1 1.05 Market return 1.1
15
20 0.9
FIGURE 2. Market-related component of the pricing kernels estimated by minimizing the second HJ distance
The pricing kernels are either with real interest rate (1 sv), with real interest rate and maximum Sharpe ratio (2 sv), or with real interest rate, maximum Sharpe ratio, and volatility (3 sv), where sv stands for state variables.
42
1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.95 1 Market return 1.05
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.95
1 Market return
1.05
0.95
1 Market return
1.05