Estimando un Modelo de 2 Factores del Tipo "Exponential-affine" para la Tasa de Interés Chilena

Abstract

In this article we estimate a two-factor model for the risk-free term structure yield in Chile. These factors are the short rate and the central tendency that are not directly estimable. Both factors follow an Ito stochastic process. In the solution of the model we follow the Balduzzi et al. (1996) specification, which provides an estimation procedure that do not depend on the parametric specification of the second factor and an "exponential affine" solution type that allows to estimate the model by mean of only one equation. The data used in this study is the average weekly yields of the bonds "Bonos de Reconocimiento" (BR) of the Chilean stock exchange during May 1993 and December 1997. The results show that when we use a stochastic level for the long term rates the yields adjust better than the case when this level is constant. Also, the speed of the reversion process increases due to the better performance of the short term rate. In addition, using an ARCH specification for the rate volatility we found additional evidence that the variance of the rates.