中文摘要：

本文利用滚动构建VAR模型的方法进行样本外动态预测，估计得出粘性信息假设下的通货膨胀预期π^e，并在此基础上建立非线性的LSTR模型，刻画出通货膨胀率的非对称调整路径。模型估计结果表明，当通货膨胀预期超过某一特定门限值后，对未来通货膨胀的作。用反而减小。另外，通过比较利率和M1这两种货币政策工具对物价的作用，本文发现当通货膨胀预期低于2．8％时，减少M1具有显著的抑制通货膨胀作用；当通货膨胀预期在2．8％-3．9％时，两种货币政策工具对通货膨胀均具有显著的调控作用；而当通货膨胀预期高于3．9％时，利率是抑制未来通货膨胀的有效手段。

英文摘要：

In this paper,the authors iteratively use a VAR-based methodology to construct out-of-sample forecasts of inflation and then estimate the past expectations of current inflation （π^e） based on the sticky-information hypothesis. While the authors construct nonlinearities using logistic smooth transition regression （LSTR） models to describe inflation asymmetry. In the models ,lagged π^eis used as a transition variable and it suggests that if inflation expectations exceeds particular threshold values, its effect on the inflation will be reduced. Furthermore ,by comparing the effects of interest rate and M1 on inflation,it is found that if the irfflation expectations is less than 2. 8% ,M1 can curb inflation,and if the inflation expectations is between 2. 8% mad 3. 9% , both interest rate and M1 are disinflationary ,and if inflation expectations is larger than 3.9% ,thus interest rate has significant influence on inflation.