国债回购市场利率价差预测能力的再检验

Retesting Predictive Power of Repo Bond Interest Rate Price Spread

本文在给出利率预期假说理论模型和相关推论的基础上,在误差修正模型框架下采用因子分解技术,将两个利率序列分解成长期记忆成分和短暂成分,并通过将短暂成分对利率价差进行回归,以此来检验回购市场长短期利率价差的预测能力。结果表明,利率价差对去除长期记忆成分后未来利率变化的短暂成分的预测能力显著增强,而对于短期利率序列的纯长期记忆成分的预测能力则很差。

On the basis of theoretical models and associated deductions of expectation hypothesis of interest rate term structure, under the framework of weak version of interest rate expectation hypothesis and error correction model, the paper decomposes the two interest rate series into long-memory components and transitory component by using factor decomposition procedure, regresses the transitory components to interest rate spread to test the predictive power of long-short interest rate spread in repo market. The results show that, for the two sub-samples, interest rate spread can provide significant predictive power for future short term interest rate when the permanent component is removed from short term interest rate series, but get weak predictive power for the permanent component. Thus, it proves that the variation of short term interest rate mainly results from long-memory component or permanent component.