On the Efficacy of Model Specifications in the Term Structure of Interest Rates

Park, Jeong Min; Kim, Tae Hyung; Cho, Jae Ho

Korean Journal of Financial Studies 2012;41(2):263-308.
Published online April 30, 2012.

Jeong Min Park, Tae Hyung Kim, Jae Ho Cho

이자율기간구조모형의 유효성 분석

박정민, 김태형, 조재호

Abstract

Most of dynamic term structure models (DTSMs) of interest rates are specified in terms of latent state variables to capture the stylized facts observed in the real world such as mean reversion, persistence, and stochastic volatilities of bond yields. In these model specifications, however, the dependency of the conditional second moments on the interest rate, so called the level effect, is implicitly fixed rather than explicitly given. As such, we propose a Bayesian Markov Chain Monte Carlo (MCMC) algorithm to estimate a general stochastic volatility model of the short-term interest rate, and examine the efficacy of the extant DTSMs in capturing conditional second moments which include both stochastic volatilities and the level effects. We find that the posterior distributions of level effect parameters are significantly different across a umber of short-term (weekly and daily frequencies) interest rate series. For example, posterior mean values of the level effect estimated from weekly T-Bill yield data and Korea T-Bill yield data are close to those of ATSM and QTSM respectively. These results imply that when using DTSMs in applied works, we should keep in mind the limitation of each model, especially in terms of the ability to capture conditional second moment.

Key Words: 베이지언추론,수준효과,이자율기간구조모형,확률변동성,Bayesian Inference,Level Effect,Markov Chain Monte Carlo,MCMC,Stochastic Volatility,Term Structure Model of Interest Rate