Predictability of Interest Rates : An Analysis Using Dynamic Term Structure Models

现有的动态利率期限结构模型在无套利条件下对债券价格的封闭式推导有一定的限制，但对利率水平和波动性的预测能力较差。这项研究消除了这种限制，依靠一个准确的近似方法。通过将波动率指定为利率水平的非线性函数，所提出的模型比现有的限制性模型具有更高的预测能力。此外，通过将所提出的模型的波动性预测与时间序列模型，如GARCH，它被证实，预测能力进一步增加。

Existing dynamic term structure models have restrictions necessary for deriving bond prices in closed-form under non-arbitrage conditions, which, however, deteriorate the predictive power to the level and volatility of interest rates. This study removes such restrictions by relying on an accurate approximation method. By specifying the volatility as nonlinear functions of interest rate levels, the proposed models exhibit higher predictive power than the existing restrictive models. In addition, by combining the volatility forecast of the proposed models with that of time-series models such as GARCH, it is confirmed that the predictive power further increases.