GMM을 이용한 Vasicek형태의 일요인 이자율 모형의 추정

Abstract

There has been many researches of the term structure of interest rates in theory and empirical studies. No researches has been performed in the forecasting comparison of models. This paper first attempts to compare various Vasicek types of interest rate models under the risk neutral measur. Our researchis different from Chan et al (1992). Chan er al (1992) compare models under the physical measure. Our paper compares four types of Vasicek models; pure Vasicek, damped Levy Vasicek, damped Levy diffusion Vasicek and finally jump diffusion Vasicek (Das and Foresi). We use GMM for estimation and forecast. We propose new orthogonal moment conditions which are different from Longstaff and Schwartz (1992), and Chen (1994). From our empirical results, we find that short interest rate and yield data in KTB(Korea Treasury Bond) have fat tails and skewness, but the small jump Levy models have more explanatory power than the large jump models. In the view of performance, damped Levy diffusion Vasicek model and Das and Foresi model have good forecasting abilities.