央行微观调查数据适合作为不良贷款率的预测指标吗——基于MIDAS模型的研究

Is Micro Survey Data from People's Bank of China Suitable as Predictors of Non-performing Loan Ratio—Research Based on MIDAS Model

相比于传统宏观经济指标,央行的微观调查数据是否更加适合作为我国商业银行不良贷款率的预测指标?本文首先提出了利用微观调查数据进行预测的相关理论,接着基于适合解决混频数据预测问题的MIDAS模型进行了实证研究。研究发现:央行微观指标在平均样本外预测误差等多个方面均优于传统的宏观指标,同时人民币名义有效汇率指数、基金及理财投资意愿比例、房价过高难以接受比例等指标尤其适合作为预测指标。本文的结论是:央行的微观调查数据更加适合作为不良贷款率的预测指标。

Compared to the traditional macroeconomic indicators,if micro survey data from People's Bank of China is more suitable as predictors of non-performing loan ratio? In this paper,we first put forward the theory of using the micro survey data to predict and then we carry out an empirical study based on the MIDAS model,which is good at solving prediction problems of mixed frequent data. We find that the micro indicators from People's Bank of China are superior to the traditional macro indicators in aspects of the average outer-sample prediction errors and so on. We also find that the effective RMB nominal exchange rate index,funds and financial investment willingness ratio,unacceptable ratio to high housing price and other indicators are particularly suitable as predictors. The conclusion of this paper is: the micro survey data of People's Bank of China is more suitable as a predictor of non-performing loan ratio.