Analysis of Riskless Short Term Interest Rate GARCH Models with Level Effect

This paper analyzes and compares short term interest rate GARCH models of Korean short term interest rates. It is investigated that short term interest rates are drift-induced stationary, that the drift function of interest rate is nonlinear, and that the conditional volatility of interest rate has asymmetry and level effect. Two kinds of Level-GARCH models, which have the pure adjusting level effect and the recursively additive level effect, are considered. Empirical results show that the nonlinear-asymmetric GARCH model with pure scale adjusting level effect fits best among all the models estimated. While the short term interest rate GARCH models pass through each conventional test, it turns out that all models are rejected by non-parametric omnibus specification tests. The diagnostic results indicate that further model improvements are required for understanding of the dynamics of interest rates.