Estimation in Financial Models - RiskLab
Estimation in Financial Models - RiskLab
Estimation in Financial Models - RiskLab
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Example 2<br />
The solutions of the stochastic dierential equation<br />
dX t =( + X t ) dt + (X t ) dW t ; (3.54)<br />
where X 0 = x 0 and the function takes positive values <strong>in</strong> IR, are called<br />
mean-revert<strong>in</strong>g processes (see also model (1.5)). The unknown parameters<br />
are and . Our aim is to be able to calculate the mart<strong>in</strong>gale estimat<strong>in</strong>g<br />
functions ~G n and G n.<br />
Lemma 1 The function<br />
f(t) E ; (X t jX 0 )<br />
solves<br />
f 0 (t) = + f(t): (3.55)<br />
Proof: Write (3.54) <strong>in</strong> <strong>in</strong>tegral form<br />
X t = X 0 +<br />
Condition<strong>in</strong>g on X 0 we have<br />
and equivalently<br />
Z t<br />
0<br />
( + X s )ds +<br />
Z t<br />
0<br />
(X s )dW s :<br />
Z t<br />
<br />
E ; (X t jX 0 ) = E ; (X 0 jX 0 )+E ; ( + X s )dsjX 0<br />
0<br />
<br />
(X s )dW s jX 0 ;<br />
0<br />
| {z }<br />
=0<br />
+E ;<br />
Z t<br />
E ; (X t jX 0 )=X 0 + t + <br />
Z t<br />
0<br />
E ; (X s jX 0 )ds:<br />
We conclude<br />
dE ; (X t jX 0 )<br />
= + E ; (X t jX 0 );<br />
dt<br />
and the claim follows. Note that the function f r (t) =E ; (X t jX r ), 0 r t,<br />
also solves (3.55), for the proof stays the same apart from<br />
E ;<br />
Z t<br />
0<br />
(X s )dW s jX r<br />
<br />
=<br />
Z r<br />
0<br />
(X s )dW s ;<br />
which is <strong>in</strong>dependent of t and thus plays no role <strong>in</strong> the derivative. 2<br />
37