公司债信用利差对产出和通胀的预测——基于BMA模型的研究

Forecasting Output and Inflation with Corporate Bond Spread:A Bayesian Model-averaging Approach

本文研究了2008年1月至2016年10月期间中国公司债信用利差对于产出和通胀的预测能力,通过将公司债按照期限和信用级别划分为不同组合得到多个信用利差序列,利用BMA框架下的样本外预测方法对工业增加值指数和居民消费者指数进行预测研究,结果显示信用风险较高以及到期期限较短的信用利差对产出包含显著的预测信息,且预测能力随时间变化而发生变化。这一方面表明了我国公司债信用利差已经包含了显著的经济前瞻性信息,债券定价市场化改革成果得以显现;另一方面也成为我国存在金融加速机制的实证依据,为宏观经济的预测和政策制定提供参考。

In this paper, We forecast output and inflation based on credit spread between corporate bonds and treasuries during 01/2008 and 10/2016. Based on BMA out-of-sample forecast models, we found that credit spreads with high risks or short maturities is significant forecasting output. However, there is little information forecasting inflation. Our result is similar with studies on foreign markets, and shows that credit spreads of Chinese corporate bonds have contained prospective information of economy, which is also evidence of the existence of financial accelerator in China.