E-Prints Complutense

Abstract

Over the past decade, no other tool in financial risk management has been used as much as Value at Risk (VaR). VaR is an estimate to determine how much a specific portfolio can lose within a given time period at a given confidence level. Nowadays, in order to improve the performance of VaR methodologies, researchers have suggested numerous modifications of traditional techniques. Following this tendency, this paper explores the use of the model proposed by Nelson and Siegel (with the aim to estimate the term structure of interest rate, TSIR) to implement a simulation to calculate the VaR of a fixed income portfolio. In this approach the dimension of the problem is reduced as the price of the portfolio depends on a vector of four parameters. Subsequently, we can use Monte Carlo simulation techniques to generate future scenarios in these parameters and use them to reevaluate the portfolio. The resulting changes in portfolio value are arranged and the appropriate percentile is determined to provide the VaR estimate. Despite the fact that this approach theoretically facilitates the calculation of VaR on fixed income portfolios, we show that the PROBLEM in practise ignores price sensitivities. So this method cannot therefore be used to calculate VaR on fixed income portfolios.