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.
Chapter 1<br />
Stochastic Volatility <strong>Models</strong><br />
1.1 Cont<strong>in</strong>uous time<br />
The value of a stock price S is supposed to follow the process<br />
dS t = S t ( t dt + t dW t ); (1.1)<br />
where W t is a Wiener process and and are functions of t. Consider<strong>in</strong>g<br />
the logarithm of the stock price H ln S and us<strong>in</strong>g It^o's formula we derive<br />
the process followed by H<br />
dH t =( t , 2 t<br />
2 ) dt + t dW t : (1.2)<br />
In the follow<strong>in</strong>g consider the process H t <strong>in</strong>stead of the equivalent process S t .<br />
For many purposes this leads to a more tractable dierential equation. Also<br />
from a statistical po<strong>in</strong>t of view do the <strong>in</strong>crements of H t (i.e. the so-called<br />
log-returns) behave <strong>in</strong> a 'nicer' way as the <strong>in</strong>crements of S t .<br />
In the case t and t , S t follows geometric Brownian motion. This<br />
is assumed <strong>in</strong> the Black-Scholes option pric<strong>in</strong>g model which has been used<br />
as an eective tool for the pric<strong>in</strong>g of options for more than two decades.<br />
When compar<strong>in</strong>g the calculated option values us<strong>in</strong>g the Black-Scholes model<br />
with the option prices there is usually a dierence. Among these biases <strong>in</strong><br />
model prices the well-known \smile"-eect is important: the Black-Scholes<br />
option pric<strong>in</strong>g formula tends to underprice out-of-the-money-options and to<br />
overprice at-the-money-options, that means implied volatility changes with<br />
the strik<strong>in</strong>g price (see Ball [1]). This eect arises from the assumption <strong>in</strong><br />
the Black-Scholes model that volatility is a known constant (\I sometimes<br />
5