基于央行调查数据的通胀预期转化：算法基础与理解分歧

Central Bank Survey Data-based Inflation Expectations Conversion

基于中国人民银行调查数据计算出的通胀预期,被广泛应用于通胀预期管理和货币政策分析等重要研究中。但是调查数据转化为通胀预期需要使用现实通胀率作为计算基础,进而通过统计算法获得通胀预期序列。因此,通胀预期的转化与计算过程是否科学,对相关研究结论是否可信至关重要。本文基于调查数据转化为通胀预期的算法过程,运用2000年1季度至2014年3季度的相关数据,得到三点结论:第一,通胀预期的计算必须使用同比通胀率,如果使用环比数据计算,获得的预期通胀率预测误差是同比数据的4倍;第二,基于全国或城市CPI的转化结果没有显著差异,不存在必须使用后者的问题;第三,通胀预期调研数据的计算基础与人民银行现行调查问卷内容的设计存在不匹配之处,问卷内容存在理解歧义,改进调研问题的具体内容可以消除理解歧义并解决不匹配问题。

This paper investigates a seemingly technical problem when using the survey data of People 's Bank of China for the calculation of inflation expectation, which involves the choice of the real inflation statistics. We adopt both the balance approach and probability method to calculate inflation expectation series from 2000Q1 to 2014Q3 in China and prove that this problem is far from trivial since inflation expectation series derived from the application of quarter-over-quarter statistics is neither informative to real series nor in accord with the nature of survey data. Only if the base period of the statistics is set to be the same quarter of previous year instead of preceding quarter can we expect to get a meaningful inflation expectation series. Our findings provide important implications for related studies and policymaking in China.