An approach calculate value at risk and stress test for long term bond

Abstract

This research presents a method of managing and assessing portfolio under risk from interest rates of bonds. The dataset used is interest rates of government zero coupon bonds data from March 2011 to May 2018. The bonds' maturity ranges between 3 months to 45 years. This research started by evaluating the interest rates of the government zero coupon bonds, using principal component analysis. To explain the fluctuations of the interest rates, the research found that the first three principle components consist of level, slope and hump. The research then performed stress test using sample variables at ±1SD and ±2SD, and computed value at risk (VaR). The results from the two methods were then used to compute Net Present Value (NPV) for comparison. This research found that, for the same 95 percent accuracy, the stress test used less calculation than VaR, since there is no need for Yield-Curve scenario simulation. Also, the stress test's result corresponded to that of VaR, which simulated interest rate curve from 100,000 randomizations. In conclusion, the stress test method presented in this research could be adopted by insurance companies, whose main investment is government bonds, to effectively manage risks and assess profits.