Estimation in Financial Models - RiskLab
Estimation in Financial Models - RiskLab
Estimation in Financial Models - RiskLab
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
The asymptotic properties of the estimator ^ n we obta<strong>in</strong> from the mart<strong>in</strong>gale<br />
estimat<strong>in</strong>g functions (3.48), (3.50) and (3.52), or more generally from the<br />
class of mart<strong>in</strong>gale estimat<strong>in</strong>g functions G n of the form (3.49), are discussed<br />
by Bibby and Srensen [8]. Under natural regularity conditions (see Bibby<br />
and Srensen [8], pp. 7{9) we have<br />
Theorem 4 An estimator ^ n , which solves the equation<br />
G n (^ n )=0;<br />
exists with probability tend<strong>in</strong>g to one as n ,! 1 under P 0 . Moreover, as<br />
n ,! 1,<br />
^ n ,! 0<br />
<strong>in</strong> probability under P 0 and ^ n is asymptotically normal <strong>in</strong> distribution under<br />
P 0 .<br />
For the proof we refer to [8].<br />
As a rst example we consider the Ornste<strong>in</strong>-Uhlenbeck process where the<br />
transition densities are well-known.<br />
Example 1<br />
The Ornste<strong>in</strong>-Uhlenbeck process is the solution of the stochastic dierential<br />
equation<br />
dX t = X t dt + dW t ; (3.53)<br />
with X 0 = x 0 . In this case the drift coecientisb(x; ) =x and the diusion<br />
coecient (x; ) is assumed to be known. The transition probability is<br />
normal with mean F (x; ) =xe and variance () = 2<br />
2 (e2 , 1). Hence<br />
the estimat<strong>in</strong>g function ~G n has the form<br />
~G n () = 1 2 nX<br />
i=1<br />
and we obta<strong>in</strong> ^ n as solution of ~G n () =0:<br />
X (i,1) (X i , X (i,1) e ) ;<br />
^ n = 1 P ni=1<br />
log X (i,1) X i<br />
Pni=1 :<br />
X 2 (i,1)<br />
The estimators ^ n we obta<strong>in</strong> from the mart<strong>in</strong>gale estimat<strong>in</strong>g functions G +<br />
and G are the same because G + and G are proportional to ~G.<br />
In the next example we consider a wider class of stochastic processes.<br />
36