Based On EEMD And BP Neural Network Model RMB Interest Rate Swap Spread And Its Reference Rate Forecast

In terms of preventing interest rate risk,interest rate swaps are very effective tools for commercial banks.However,China's pricing mechanism for interest rate swaps is not yet perfect.Currently,developed countries adopt the method of spread quotation,but due to China's The research on the influencing factors is not thorough enough,so the total price quotation has always been used,and it is still difficult to realize the price difference quotation system in China.This article firstly carried out a descriptive statistical analysis of the four main as-pects of the RMB interest rate swap market from the transaction subject,transaction variety,transaction structure,and transaction volume based on the data of China Cur-rency Network's interest rate swap monthly report.Interest rate swaps The interest rate swaps ran smoothly in January-February,rose from February-February,and de-creased slightly from late April to August.The upward trend and changes after August are due to changes in the central bank's monetary policy in conclusion.When the central bank's monetary policy results in sufficient liquidity in the money market,the short-term reference interest rate of interest rate swaps will tend to decline;when the economy is basically looking good,the long-term reference interest rate will decline;at the same time,the reference interest rate and its swap change in the same direction.Secondly,the swap spread of different reference rates with different maturities is decomposed from the perspective of monetary policy interest rate transmission,and it is decomposed into three parts:swap spread,bank financing cost and interest rate term structure.The results show that whether it is a 1-year or 5-year swap spread decomposition chart shows that shibor3M has better policy guidance effect than FR007.At the same time,compared with the 1-year swap spreads,the 5-year swap spreads clearly reflect the increased uncertainty of the market's medium-and long-term interest rate expectations.Bank financing costs were higher between 2016 and 2017,and lower at other times,and negatively correlated with the maturity of interest rate swaps.The expectation of interest rate fluctuations is just the opposite,which is positively correlated with the maturity period of interest rate swaps,that is,the longer the maturity period,the more intense the fluctuation expectations.For shibor3M,most of the time is positive,for FR007 most of the time is negative.Then use the combination of EEMD and BP neural network algorithm to decompose and predict the 1-year shibor3M swap spread and shibor3M.The prediction results show that the combined time series prediction method of the two methods is more simple than the prediction result of unstable time series.The prediction result of BP neural network is stronger.Finally,this paper summarizes and points out the shortcomings and the corresponding improvement direction.