There are two potential sources of value in the active bond management The first source is forecasting techniques, it is trying to establish a series of model to predict the various changes of the market by predicting future conditions in the market, managers can be found relatively The investment value of bonds or interest rate risk circumvention to obtain excess returns second potential source of value is the bond market price imbalances determine the value of these two sources are important for bond investments. areas of forecasting techniques is the most important prerequisite for asset allocation, foreign scholars.

This article will focus on research based the term

**structure**prediction active bond

**investment strategy**, and empirical testing of these strategies can be applied in the Chinese market through the the Exchange bond of transaction data.

Predict changes in the term structure of interest rate strategy

The term structure of interest rates reflect the one-to-one relationship between the interest rate and maturity. Term structure of interest rate changes are subject to a number of factors (such as interest rate expectations for the future, the term risk premium and convexity, etc., Zhu Shiwu, Chen Jianheng, 2006 ), and changes in the form is also very complex, but the most important changes in the form of parallel shift, slope changes and changes in the convexity. studies have shown (Zhu Shiwu, Chen Jianheng, 2003), in China's bond market, the yield curve parallel to the moving elements can only explain 52% of the gross movement of the term structure, the rest of the change is the slope changes and camber changes. while in the U.S. market, interest rates parallel the proportion of mobile components explain more than 90%. imagine, if the yield curve is completely parallel mobile , so when interest rates rise, the longer the term of the bonds, the investment rate of return lower; When interest rates decline, the longer the term the higher the rate of return on bond investments if the yield curve is not entirely parallel movement (also mixed with slope changes and camber changes), the performance of each term bonds at different times would not be exactly the same (as shown in Table 1), the following conduct a detailed study of this phenomenon.

Remaining maturity of the bonds and coupon type Exchange bond can be divided into the following six categories: floating rate coupon, fixed-rate coupon of 1-3 years, 3-5 years, 5-7 years, 7-10 years and more than 10 years to August 2003 to March 2005 as the period of sample analysis, statistical the ranking month holding period rate of return of the six types of bonds, such as shown in Table 1

Can be seen from Table 1, the advantages and disadvantages in different points in time, the performance of bonds of different maturities. Summarized as follows: When the operation of the market is relatively stable (eg, August 2003), medium-term bonds (3-7 years), the best rate of return; 2 when the market is in sharp decline (eg, from September 2003 to October, 2004), the best performance of short-term bonds (1-3 years) and floating rate debt; 3. (eg, from May 2004 to July when the market rebounded and continued to rise, from January 2005 to February), the performance of long-term debt (more than 7 years) the most prominent and the performance of each term bonds with the excess returns rate the predictive value of a strong relationship, namely: the performance of the 1-3-year short-term bonds and floating rate debt when the negative predictive value of the excess rate of return is relatively good; When the predictive value of the excess rate of return while a positive number, the better the performance of long-term debt. predictive value based on the rate of return on long-term bonds over the performance of the corresponding national debt of the better years, can than be held for a certain period of government bonds or hold the market portfolio better returns.

Table monthly rate of return for a different period of government bonds Sort

A direct impact on the return on investment for each term bonds due to changes in the interest rate term structure, key research and prediction of interest rate changes in the term structure of the bond investment.

Changes in the interest rate term structure including moving parallel to the slope changes and camber changes, forecast changes in the term structure is also attributed to this form of several changes in the forecast. Them in several changes in the form of parallel shift accounted more important than higher, and thus lift the prediction of the overall level of interest rates. the author (of Zhu Shiwu, Chen Jianheng, 2006) studies predict excess returns on bonds can play a similar role in the prediction because the actual rate of change of the excess returns of long-term bonds more reflects the raising and lowering of the level of interest rates: when interest rate levels, the excess rate of return for long-term bonds will decline, and presents a negative; when the decline in the level of interest rates, the excess rate of return increased short-term money market rates predict changes represent changes in the overall level of interest rates, the negative correlation of the short-term interest rates and long-term bond excess returns can explain excess returns of the level of interest rates movements. findings of the author (of Zhu Shiwu, Chen Jianheng, 2006) describes regression model can roughly predict future changes in interest rates, changes in the term structure prediction only prediction of slope changes and camber changes.

Fitting the term structure model

(A) Nelsen-Siegel model introduced

Nelson-Siegel model is Charles Nelson and Andrew Siegel in 1987 proposed a parameter fitting model is the model through the establishment of the function of the instantaneous forward interest rate, which is derived in the form of a function of the spot rate. One of the biggest benefits of the model is need to estimate the parameters of a relatively small (generally only need to estimate the four parameters), and therefore particularly suitable for a small number of the estimated bond interest rate term structure, and has obvious economic implications of these parameters, making the model itself can easily be understand.

The NS model given moment forward rate,

, ¦³1 is a time constant suitable for this equation, ¦Â0, ¦Â1 and ¦Â2 are the parameters to be estimated. Fixed ¦Â0 When, through different combinations of ¦Â1 and ¦Â2, can produce a variety of shapes forward rate curve, such as monotonous type, horizontal and inverted curve.

Has a clear economic meaning of the parameters in the above equation can be seen from the instantaneous forward rate formula forward rate is essentially three parts by the short-term, medium-term and long-term interest rates.

Which represents the long-term interest rates is the parameter ¦Â0 asymptote that represents the instantaneous forward rate curve f (0, ¦È), ¦È increases with the maturity of f (0, ¦È) curve should tend to ¦Â0 the value.

¦¢1 represents short-term interest rates section is instantaneous forward interest rate curve in the initial position (or short-term) and the departure from the value of the asymptote, it also contains the instantaneous forward rate curve to asymptote approaching speed factors. is a positive number, then the instantaneous forward rate curve is rising with increasing duration of, and vice versa instantaneous forward rate curve decreases with increasing duration of

¦Â2 representative of the mid-part of the interest rate, which determines the nature of the extreme points of the instantaneous forward rate curve and curvature. ¦Â2 is a positive number, the curve on convex and vice versa curve on the concave.

¦Ó1 is a positive number, it is the instantaneous forward rate curve the abscissa (deadline) corresponding to mark the position of the extreme points of the forward rate curve.

(B) fitting results

Nelsen-Siegel model to understand, you can use the model to fit the term structure, but in the process of fitting a few key points to consider, are as follows:

Sample data. Purpose of this article is to predict changes in the interest rate term structure, and the most representative of the domestic bond market term structure of government bonds yield curve. This article will use the the Exchange bond data to fit the term structure of interest rates. order to reflect each term segment of the yield in the fitting process, each term bonds is more evenly distributed (in particular, the need for short-term bonds and long-term bonds), otherwise the fitting out of the curve may be unreasonable, but In 2004, the short-term government bonds listed on the exchange market rarely, affect the fitting of the term structure. Taking this into account, this article from the March 24, 2004 (1-year short-term debt 04 government bonds listed 01), fitting term structure, until March 11, 2005 (1-year short-term debt 04 government bonds delisting 01), about 1 year of data.

Sample data processing in the process of fitting the term structure using the daily trading price of the Exchange bond. Reasonableness of the transaction price for fitting the term structure itself is crucial. Unreasonable prices will lead to unreasonable rate of return, thereby distorting fitted yield curve deformation. Excluding the two types of bonds: First artificial hype bonds; on tax and legal causes some bond yields relatively high or low bonds.

3 constraint of the model parameters. Theoretically speaking, the degree of fit of the model in order to make as high as possible, should not be on the parameters for any of the constraints, but if the parameters are not for any constraints, then the continuity and stability of the parameters sex may not be guaranteed (to see Diehold and Canlin Li, 2002) because the primary purpose of this article is not fitting the the optimum yield curve, but by changes in the model parameters to predict the yield curve, therefore, how to ensure that the model the stability and continuity of the parameter is more critical tasks. Thus, the model parameters are certain constraints of this study, the value of the fixed parameter of ¦Ó1 for the remaining three parameters to be estimated. studies show that 3 lives are most sensitive to the duration of the yield curve movements, the point of maximum change in that crown. parameters ¦Ó1 fixed at 3

## Comment

More aboutEmpirical research based on the forecast interest rate changes in the term structure of the bond investment strategycomment Total