Customer Order Flow And Exchange Rate Volatility

Since its great success in abundant empirical studies for foreign exchange rate fluctuations,market microstructure theory has been considered as necessary supplement to the traditional macro-based exchange rate theories.Under the frame of market microstructure theory,the actual trading process and the information asymmetry among market subjects are the keys to piercing the veil of excess return in asset markets.Customer order flow,as the representative of the real demand and the medium of private information,is one of the most crucial variables in micro-based models and,its correlation with the exchange rate movements is an important entry point to tackle the pricing mechanism problems.The interactions between customer order flow and exchange rate volatility are only realized through the dealers' quotes and their inventory deliveries from the retail market to the inter-bank market.For this reason,we establish a theoretical model for a single dealer's customer order flow and exchange rate volatility under the study frame of the portfolio shift model.A daily-based sample,containing complete customer order flow and bid-ask spread data of one of the biggest market-maker bank in CFETS,including USD/CNY and EUR/CNY currency pairs,has been used to examine the former portfolio shift model and latter in inspections of all other theoretical models in this work.The empirical analysis for our portfolio shift model is carried out by means of structural threshold regression technique,and the results indicate a much stronger explanatory power of customer order flow to exchange rate volatility when the market risk is aggregating.As the information discovery is gradual,the correlation between customer order flow and exchange rate movements needs to be examined under the lagged effect.For this purpose,we use Markov Switching Vector Auto-Regression to model the dynamic relations with the considerable nonlinearity of the regression system.The regression estimation and the pulse simulation show that corporate customer order flow has predictive ability for the exchange rate fluctuations while individual customer order flow exhibits obvious feedback effect and thus fail to forecast the future prices.A closer inspection for the feedback effect is then carried out under the assumption of a negative exponential sensitivity function of the customer order flow to the bid-ask spread.The results further improve that individual customer order flow is largely relied on the dealer's quotes while corporate customer order flow seems independent to them.Finally,to conclude the previous results in this work,we use support vector regression to describe to what extent that the corporate customer order flow can predict the exchange rate movements.Our partially out-of-sample tests show that the predictive power of corporate customer order flow is reliable with determination coefficients above 65%,and thus corporate customer order flow should be taken under careful surveillance for risk management concerns.