央行货币政策调控的影响因素识别——基于机器学习方法的实证研究

Identifying the Determinants of Central Bank Monetary Policy Control—An Empirical Study Based on Machine Learning Methods

摘要: 本文使用文本分析方法并结合现有文献，选取了14个可能影响央行货币政策调控的经济变量，并通过随机森林模型和SHAP值法量化了这些变量对2003~2022年央行货币政策调控的影响程度。研究结果表明，在2003~2022年这段时期，以通胀缺口为主的泰勒规则和以中美10年期国债利差为代表的汇率因素在我国央行货币政策调控的参考指标中占据重要地位。有鉴于此，本文建议应通过重视通胀在宏观经济调控中的作用；加强对外汇市场的关注程度；深化利率市场化改革来完善货币政策调控。

Abstract: This paper employs text analysis methods in conjunction with existing literature to select 14 economic variables that may influence the central bank’s monetary policy control. Through the Random Forest model and SHAP value method, it quantifies the impact of these variables on the central bank’s monetary policy control from 2003 to 2022. The results indicate that during this period, the Taylor rule, primarily based on the inflation gap, and exchange rate factors represented by the US-China 10-year government bond yield spread, occupy an important position in the reference indicators for the central bank’s monetary policy control in China. In light of this, the paper suggests that attention should be paid to the role of inflation in macroeconomic control; the importance of the foreign exchange market should be enhanced; and interest rate marketization reforms should be deepened to improve monetary policy control.