摘要
防控信用债的流动性风险关系到防范系统性金融风险和维护国家安全,其核心在于对风险进行有效测度和预警。本文采用2009年1月—2020年12月的中国信用债月度数据,通过尾部相关性测度信用债流动性风险,从融资约束、信用风险和噪声交易角度构建预警风险因子体系,并采用神经网络等包含11种设定的机器学习模型,预警信用债流动性风险并识别重要风险因子的作用机理。研究表明:第一,包含一层隐含层的神经网络对信用债流动性风险的预警能力最强,在不同类型债券和不同外部环境的样本中预警能力的稳定性较强,能准确预警市场层面的流动性枯竭事件;第二,券龄的重要性最高,新发行的债券会通过引发噪声交易的方式形成流动性风险,但随着券龄增加,流动性风险减少的程度是递减的;第三,流动性风险是由多类风险因子协同运动生成的,其中,经济状况变动、货币政策改变或跨市场冲击与券龄的非线性联动对于驱动流动性风险最为重要。
Preventing and controlling the credit bond liquidity risk is crucial for preventing systemic financial risks and safeguarding national security,making it the top priority in financial operations.The purpose of this paper is to develop effective measures for measuring and early warning credit bond liquidity risk.These measures can provide assistances in completing the important task of improving the financial regulatory system and guarding against systemic financial risks.Using the monthly data of China's credit bonds from January 2009 to December 2020,this paper measures the liquidity risk of credit bonds through tail correlation,constructs an early-warning risk factor system from the perspective of financing constraint,credit risk and noise trading.Using 11 kinds of machine learning models including neural network to early-warn liquidity risk of credit bonds,and identify the mechanism of key risk factors.The findings of this paper are as follows.Firstly,the neural network with one hidden layer has better early warning ability on the liquidity risk of credit bonds.It has strong early warning stability in different types of bonds and different external environment samples,being able to accurately warn the liquidity depletion events at the market level.Secondly,bond age is the core risk factor driving liquidity risk.Newly issued bonds will form liquidity risk by causing noise trading.With the increase of bond age,liquidity risk will only decline in a decreasing manner.Thirdly,liquidity risk is generated by the coordinated movement of multiple types of risk factors.Among them,the nonlinear linkage of changes in economic conditions,changes in monetary policy,cross market impact and bond age serve as the most important factors that may drive liquidity risk.
作者
张宗新
周聪
Zhang Zongxin;Zhou Cong(School of Economics,Fudan University;China Southern Power Grid Energy Development Research Institute Co.,Ltd.)
出处
《国际金融研究》
北大核心
2024年第5期74-85,共12页
Studies of International Finance
基金
国家自然科学基金面上项目“基于机器学习算法优化的中国资本市场系统性风险监测、预警与管控研究”(72073035)资助。
关键词
信用债
流动性风险
机器学习
风险测度
风险预警
Credit Bond
Liquidity Risk
Machine Learning
Risk Measurement
Risk Warning