A Correlation Study Between The Changes Of The Term Sructure Of Interest Rates And Macroeconomic Variables

As interest rate marketing process of China is accelerating, whether in academic or financial practice, needs of research and interest rate risk management are growing. In order to research interest rate risks effectively and accurately, we must firstly construct a reasonable and accurate term structure of interest rates, and,on the basis of that, study in interest rate risk characteristics and the corresponding interest rate risk management approach.Since the most of financial research subjects are time-varing, therefore the study of fitting the term structure is focused on the dynamic model. But in most of the dynamic model the forecasting ability out of sample is not reliable, and they do not have a concise economic implications. Faced with this situation, Diebold and Li (2006) proposed a dynamic Nelson-Siegel model (DNS model):It retains the simple forms of the static Nelson-Siegel model and clear economic implications (Litterman and Scheinkman,1991), and so that it can explain the term structure of interest rates curve dynamic characteristics, and has a good sample predictive ability (Diebold et al.,2006; Koopman et al.,2010). Therefore, this paper mainly uses the DNS model to fit the term structure curve. However, there are some problems during estimating the model. In order to simplify the model estimation process, making the estimation process feasible, including the two-step estimation which Diebold and Li (2006) used to estimate the dynamic model, most estimation method assumed that DNS model parameters were stationary, ie the residuals of parameter sequence were constant, which limited the fitting of the dynamic model. Therefore, this article hopes to make further explore on this direction. In the estimation of the model, the article does not make that assumption, but assume that the variance of the residuals of the parameter sequence is a stochastic volatility process, and in the actual estimation process, the parameters for the equation of state is assumed to be a diagonal matrix, in order to reduce the complexity of the estimation. The empirical results show that it is feasible.In the basis of a reasonable term structure of interest rates, we can take good use of the resulting interest rate risk characteristics parameters to study. This article mainly studys in the correlations between the changes in the term structure of interest rates and macroeconomic variables. In this area, foreign scholars have done relatively well studied, and domestic scholars in recent years began to have concerns, but the corresponding research are relatively small. On one hand the bond market is not mature enough, on the other hand in the early years our marketing level of the interest rates is relatively low. This makes the policy more powerful in the impact of changes in the term structure of interest rates, and this also makes it difficult to do some effective, convincing empirical research. In addition, there are many divergences in our scholars in this area, and most studies have focused on the formal of equation. In the view of above situation, this article will add the term structure parameters and their volatility, and macroeconomic variables and their volatility into the estimation equation, which attempts to explain from another perspective the correlations.Through theoretical derivation, model updating and ultimately empirical research, this study found that:1. The volatility of the spot rate we got by Fama-Bliss peeling showed a downward trend in the growth as the terms going up. And it partly reflects that as the promotion of the process of marketization of interest rates, our term structure of interest rates are gradually getting close to Western financial markets.2.Through comparing the two DNS models which did not introduce the stochastic volatility model and did introduced,we found that the model which introduced the stochastic volatility did improve the fitting.3. Through the DNS model parameters and their volatility and the volatility of macroeconomic variables and their correlation study, we found that adding the volatility of macroeconomic variables enhanced overall explanatory power in explaining the DNS model parameters, but the effect is not obvious, however it improved the power in explaining the volatility of the DNS model parameters significantly. And the effects in explaining the DNS model parameters volatility is better than the DNS model parameters themselves.