where for notational convenience we write Φ^ for . Φ^ is called the
pricing kernel. To simplify notation in the remainder the index (к) is omitted.
Furthermore, it is well known from Girsanov’s theorem that the process
W, which is given by
Wt = Wt +
∕'
follows a standard Brownian motion on the space
(ω, 9 ,p).
Thus, technically the transformation is done by multiplying the asset
price under the probability measure P with a suitable P-martingale.5 To get
some economic intuition we apply these results in a simplified economy.
4.2 Interpretation in a simplified economy
Consider the following market with no arbitrage opportunities: On the fil-
tered probability space (Ωz, Q,, Q,t, Pz) with a one-dimensional Brownian mo-
tion Wz defined on it the forward price is given by
dFt = Ftμtdt + FtΣt dWtz , 0 « t « T, (6)
Fo = F > 0,
with μ, Σ deterministic functions depending on t and Ft and Σ = 0. Thus,
this market is complete and there is a unique martingale measure. With
к := d define Φt = exp (— ʃθ κudWu — ɪ ʃθ ∣κw∣2 du) , then applying Itô’s
formula it is easily seen that ΦtFt is a martingale and thus
F = E*' (Fτ ∣Qlt) , 0 « t « T,
1 FJ ∙ 1 r. 1 1
where Pz is defined by
dP' _
dP = ⅛ . 0 «t « t .
The instantaneous covariance, i.e. the quadratic variation, between this
process Φ and the forward price process F is equal to the instantaneous
drift μ of the forward price process.
0For a detailed derivation see for example Karatzas, Shreve [21] and Musiela, Rutkowski
[25].
10