本研究利用金融業拆款利率與財務短期利率模型來探討台灣短期利率之動態擴散效果，試圖找出利率模型估計之最佳實證模型。本研究發現估計擴散模型時一律採用線性之漂浮項並不恰當，因為實證結果發現不同的擴散模型應該配適其特有之漂浮項型式，例如：NARCH模型應該採用非線性型式。此外，本文實證結果顯示GARCH模型高估短期利率的波動性，然而「漂浮項不對稱之非線性NARCH模型」不但在解釋波動性上優於其他擴散模型，此外更具備了反應正負衝擊所帶來的不對稱現象，因此可視為短期利率之最佳模型。 This paper estimates a dynamics model of short-term interest rate and allows sensitivity of the volatility process to interest rate levels. The empirical results that GARCH effect, NARCH effect and stochastic volatility effect in the diffusion function fit the date well and superior to the single factor model of the level effect. If we use GARCH model to estimate the diffusion process will cause over-evaluate the fluctuation, so this paper especially make use of NARCH model to describe the true dynamic fluctuation reacts of short interest rate. The empirical research points out that modeling of the linear drift GARCH model and the nonlinear drift NARCH model in the short interest rate are the best. In addition, I develop some asymmetric framework in mean and diffusion function. When estimating the nonlinear drift NARCH model, the asymmetric response in the drift function is the best model.
