**analysis**and VaR method combined bond

**risk management**factor analysis model, and the model variance - covariance model empirical test and compare a number of bond risk management thinking.

[Keywords] bond yield curve factor analysis VaR

A comprehensive factor analysis model proposed background

People's Bank of China decided that the financial institutions raised the benchmark deposit and lending interest rates since August 19, 2006. Financial institutions, the one-year benchmark deposit rate by 0.27 percentage points from the current 2.25% to 2.52%; year benchmark lending rate by 0.27 percentage points, from the current 5.85% to 6.12%; other deposit and lending interest rates also make a corresponding adjustment. The rate hike is the second rate hike since the rate hike following the October 29, 2004, although the magnitude of the two rate hikes is not too large, but the rate hike impact to the national economy as a whole is not to be ignored, the following only from the point of view of the bond investment analysis of the impact of interest rate increase.

First of all, it will have a greater impact on the bank as the the important bond investments main one. On one hand, the bank's own as bond investment entities, will increase the opportunity cost of the bank; On the other hand, bank financing intermediaries, increase in deposit and lending rates will reduce the competitiveness of the banking facilities.

Secondly, the insurance companies will face greater interest rate risk. The insurance company's sources of funds mainly to long-term funds invest primarily in

**bonds**and notes for the purpose of asset-liability maturity matching. Since the last century, in the 1990s the central bank the previous interest rate cuts, the high interest rate policy the insured before the insurance company to cut interest rates to a historical burden, and the current interest rate insurance companies will undoubtedly worse.

Finally, the bond investment fund management companies and other financial companies will bring greater uncertainty. Such financial company bonds have higher investment demand, and often has a role to enhance the level of activity of the entire market, a result, the impact of this uncertainty will spread to the entire market, thereby increasing the risk of the entire market investment.

Comprehensive factor analysis, compared with the United States and other economically developed countries, the level of risk management of China's bond is lagging Based on the above background, the factor analysis of risk management applied to international mainstream analysis method - VaR method law model, hoping to provide some thoughts on China's bond risk management.

Comprehensive factor analysis model

1, factor analysis of the empirical

Taking into account the long history of China's bond issue, while the bond market is relatively mature, the article uses the bond transaction data as of January 2, 2003 to 2003, 26, 230 trading days of the Shanghai Stock Exchange objects for analysis. In this period of time, the flow of interest-bearing government bonds listed on the Shanghai Stock Exchange Treasury market increased from 17 to 22, discounting 0100112,010210 two floating rate treasury bonds, fixed-rate government bonds for the fitting of the yield curve, including The: 000696,009704,000905,000896,009908,010004,0100010,010103,010107,010110,101112,010115,010213,010214,010215,010301,010303,010307,010308,010311. Fewer As our Exchange Treasuries varieties, thus this article using the Nelson-Siegel method to estimate. This article were calculated for one year, two years, three years, four years, five years, six years, seven years, eight years, nine years, the 10-year Treasury bonds current yield.

First, we use the Bartlett sphericity (Bartlett test of sphericity) and KMO (Kaiser-Meyer-Olkin) statistical test method for one year, two years, three years, four years, five years, six years, seven years eight years, nine years, the 10-year Treasury bonds current yield 10 original variables correlation analysis, the results significantly, indicating a strong relationship between the 10 original variables, and therefore suitable for factor analysis.

Secondly, the use of the ADF test data stationarity. See from the ADF test results (see Table 1), none of the time series of all maturities yield level by 90% confidence level, shows obvious non-stationary characteristics; while all the first-order differential time series were shown good steady state. Comprehensive correlation analysis and the stationarity test result below principal component analysis using a first-order differential (daily change in yield) as an analytical object.

Finally, using the SPSS statistical software diurnal variation of the current yield of the government bonds (first-order differential) and principal component analysis, the results (see Table 2) show that the variance contribution rate of the first three major factors on the rate of return of 65.72 percent, respectively , 29.07%, 5.16%, 99.95% overall cumulative variance explained Therefore, the first three main factors basically have to explain the changes in the Treasury yield curve characteristics. These three factors are the horizontal, tilt and curvature. Characteristic values ??of horizontal and inclined two factors are greater than 1, and the cumulative contribution rate reaches 80% or more, according to the number of principles of determining the common variables, paper extracted from these two factors of these two factors in consolidated form. modeling.

Comprehensive factor analysis model outlined

Horizontal and inclined two factors can explain the vast majority of the changes in value of the bonds, and the changes in bond yields can be expressed as a linear combination of the factors, changes in bond yields can be expressed as:

Comprehensive factor analysis model in our bond risk management and recommendations

In this paper, use the integrated factor analysis and variance - covariance method to calculate VaR value of the the nine bond portfolio of bonds 000696,000896,0009908,010010,10103,010215 these six bonds in April 2003 by the results shown in Table 3 .

From the results of the two methods in Table 3 bond VaR measurement, we can draw the following conclusions:

1, and the variance - covariance method comparison, comprehensive factor analysis of the estimated value of the bond portfolio VaR is low, only the estimated value of the bond portfolio, 3 two-factor method is higher than the variance - covariance method. In addition, the relatively large volatility in the the VaR estimated value of the variance - covariance method, comprehensive factor analysis VaR estimate volatility is smaller. Of bonds subject country, credibility is the highest, and basically there is no risk of default; 2003, China's interest rates basically at a stable level, in this case, the risk of the national debt is basically the bid-ask spread risk Throughout recent years, the spread of the opening and closing of the national debt, there was no violent fluctuations. And insurance institutions, finance companies and other financial institutions to buy government bonds main need is to match assets, less speculative motive. Thus, relatively speaking, the risk of the national debt is relatively low, but there will not be a lot of volatility. So, from this in terms of comprehensive factor analysis of bond VaR estimates are more reasonable.

With the due date approaching, comprehensive factor analysis of the estimated value of the bond VaR is gradually reduced, which means that the risk of the bond portfolio is gradually smaller. From 1 bond portfolio 5 bond portfolio, bond portfolio to the bond portfolio and the bond portfolio to the bond portfolio, their remaining maturity gradually increase their VaR value is gradually increased. In other words, the bond portfolio VaR value is directly proportional to their remaining maturity. For example, the remaining maturity of the bond portfolio between the bond portfolio between 2 and 3, and the VaR value is in the range between 2 and 3 of the VaR of the bond portfolio. This is a the bond risk approximation of the nominal value and reduce price volatility phenomenon ". Bonds VaR values ??calculated by the variance - covariance method this feature, and also contrary to the theoretical and practical phenomenon. For example, bonds 000,896 a bond is about to expire, only six months remaining maturity bonds 000,696, there are more than three years remaining maturity time, regardless of theory, or from the reality, these two portfolio risk posed by the bonds should be higher than the risk of individual bonds 000,696, but the variance - covariance method to calculate the VaR value is just the opposite. Like this, there are many. Thus, the measured values ??of the factor analysis method bonds VaR more accurate, more in line with the theoretical and practical.

[References]

[1] JP Morgan. Risk Metircs-technical document. New York: Morgan Guaranty Trust Company, 1996

[2] Jorion P. Value at risk: The new benchmark for controlling market risk. New York: McGraw-Hill Companies, Inc, 1997

[3] INDUSTRIAL financial market risk management. Tianjin: Tianjin University Press, 2001

[4] Dakai, Yang Yong, "about China's Treasury yield curve," Finance and Economics, 1997, 7, 14-19

## Comment

More aboutComprehensive factor analysis of bond risk managementcomment Total