Abstract (eng)
This thesis presents the applicability of Monte Carlo methods for the valuation of
interest rate derivatives. As the spectrum of interest rate models is wide, I focus
on the Vasicek model, the Ho-Lee model and the Hull-White model. These three
Gaussian short rate models are implemented in C++. In the course of that, I present
improvements of the corresponding simulations. These improvements range from
selecting suitable algorithms for transforming uniform pseudo random numbers to
standard normal pseudo random numbers, to the selection of a suitable programming
language, to the optimization of the program codes and finally to the implementation
of variance reducing techniques. In order to improve the standard deviation of the
simulated interest rate derivative prices, I apply the variance reducing techniques of
antithetic and control variates. For the latter I selected observed bond prices.
In order to implement the Vasicek model, the Ho-Lee model and the Hull-White
model, I calibrate them to observed market prices of an interest rate cap agreement.
The calibration suggests that the Ho-Lee and the Hull-White model fit observed
market prices much better than the Vasicek model. Subsequently, I simulate prices
for the very same cap agreement and compare them with the ones calculated via
closed formulas. The results indicate that improvements in the simulated prices,
due to variance reduction techniques, are always accompanied with an increase in
the computational burden. Thus, it has always got to be accounted for the trade-off
between computational accuracy and computational efficiency.
In conclusion, I implement a periodic cap agreement based on the Hull-White
model as a case study. Exemplary, I show that after the parameters of the short
rate processes are defined, path-dependent interest rate derivatives can be priced
easily. Eventually, I account for the importance of Monte Carlo methods for pricing
path-dependent interest rate derivatives.