本文以Gray(1996)的模型為基礎,建立一個台灣短期利率的實證模型,其中允許條件期望值和條件變異數均隨狀態而轉變。實證結果發現在不同狀態下,台灣短期利率都呈現回歸長期水準的現象,但卻有兩種不同的動態行為。在政經事件頻傳的高波動期,利率的長期水準較高,條件期望值的收斂速度較快,而條件變異數較大,但為常數。所以利率波動性沒有水準效果,也沒有波動聚集的現象。在相對穩定的低波動期,利率的長期水準較低,條件期望值的收斂速度較慢,條件變異數仍沒有水準效果,但有波動聚集的現象,惟波動的持續性低。我們發現,這兩種相異的行為可能是由中央銀行對市場的不同干預所造成的。
In this paper we follow Gray (1996) to analyze Taiwan’s short-term interest rate using the regime switching model. We find that the model that allows both the conditional expectation and conditional variance to change with regimes is an appropriate model. Both the level and volatility of the interest rate are higher in one regime, while they are lower in the other. The estimation results suggest that the short-term interest rate exhibits mean reversion in both regimes, but has different dynamic patterns. In the high-volatility regime, the interest rate converges to the long-run level more quickly, but its volatility has no GARCH effect. In the low-volatility regime, the interest rate converges more slowly to the long-run level, but exhibits volatility clustering. We also find that these two patterns may result from the Central Bank’s different interventions.