客户定单流与汇率波动

Customer Order Flow and Exchange Rate Volatility

本文在Evans&Lyons的资产组合变动模型的框架下推导出单一市商的客户定单流与汇率波动的理论方程,并利用我国外汇市场的某主要做市商的日度客户定单流数据对理论模型进行了计量检验,分析了客户定单流对汇率波动的同期影响与二者的动态关系。线性检验结果表明,客户定单流与汇率波动所构成的回归系统具有显著的非线性特征,因此,本文采用结构向量自回归(STR)模型并选取不同的门限变量来刻画客户定单流对汇率波动的贡献;对二者动态关系的研究,则在马尔可夫转移的向量自回归(MSVAR)模型的基础上进行。文章的主要研究结论有:第一,与价格平稳期或交易平抑期相比,当市场处于价格波动期或交易活跃期,定单流对汇率波动具有更高的解释效力;第二,与机构客户相比,个人客户的定单流对汇率波动更加敏感,反馈效应较强;第三,机构客户定单流对汇率变动具有显著的预示作用,在汇率风险的管理中应当重视对定单流变化的监测。

Customer order flow is the core variable in market microstructure frameworks, as well as the key to understand the exchange rate fluctuations. Based on a classic portfolio shift model suggested by Evans and Lyons, we establish a theoretical equation to catch the relation between a single dealer＇s customer order flow and exchange rate volatility. The sample data are from one of the biggest and oldest dealer in CFETS, including USD/CNY and EUR/CNY, considered as two representative currency pairs. The linear fitting shows poor results while the structural threshold regression passes the test of significance. With different threshold variables, we find similar and relatively strong explanatory power of customer order flow and, this power becomes even stronger when the exchange rate volatility or market trading volume is high. With concern about the nonlinearity, we use Markov Switching Vector Auto-Regression model to further explore the dynamics between order flow and exchange rate movements. We find that while individual customer order flow exhibits significant feedback effect, corporate customer order flow is insensitive but predictive to exchange rate volatility. This finding is later supported by impulse response simulations.