Comparación de un modelo multifactorial de la ETTIparamétrico versus no paramétrico

Abstract

In the financial literature there is no consensus about how many factors are necessary to explain the term structure of interest rates (TSSIR). The one-factor diffusion models are very attractive because their simplicity, however they have important unrealistic properties. In fact there are several authors, such as Canabarro (1995), who show that models whith more stochastic factors describe more accurately the evolution of the term structure. Furthermore the multi-factor models are specially useful to price more complex interest-rate derivatives, such as the so-called yield options, or improve the efficiency of hedging strategies. However these models, specially if nonparametric methods are used, are more complex to estimate and to price. In order to reduce the higher computational cost of the two factor models , Gómez- Valle and Martínez-Rodríguez (2006a) showed a new approach to estimate these models. In this paper, we consider a model with two factors: the interest rate and the instantaneous variance of the changes of the interest rate. Then, we show that the approach of Gómez-Valle and Martínez-Rodríguez (2006a) reduces the computational cost of estimation as well as provides better results than other twofactor parametric models based on the GARCH approach such as Longstaff and Schwartz (1992).