Implementation and performance of various stochastic models for interest rate derivatives (Q2722296)

Some of the formal and practical aspects of a few stochastic models for the valuation of interest rate financial derivatives are discussed in this paper. All of them satisfy the no-arbitrage requirement, since they all fall within the \textit{D. Heath, R. Jarrow} and \textit{A. Morton} [Econometrica 60, 77-105 (1992; Zbl 0751.90009)] framework; \textit{J. Hull} and \textit{A. White} [Rev. Financial Stud. 3, 573 ff. (1990)] has the advantage of being analytically tractable, but it allows interest rates to become negative. \textit{F. Black} and \textit{P. Karasinski} [Financial Anal. J. 47, 52 ff. (1991)] instead does not suffer from this problem, but pays the price of being purely a numerical in implementation. A third model, recently proposed, is due to \textit{B. Flesaker} and \textit{L. P. Hughston} [Risk 9, 46 ff. (1996)]. It encloses all the nice aspects of the other two: analytical tractability and non-negativity of interest rates. Some results for the three models are presented, with performance comparison for calibration on market data. These data represent two standard situations on actual markets: a ``normal'' and a volatile market. It is shown that the \textit{Flesaker-Hughston} model performs better than the former two in all cases considered, both in terms of better calibration on average and of better reproducing the price structure for a number of market instruments.