(2017-01-01) Restrepo Tobón, Diego Alexander; Velásquez Giraldo, Mateo; Álvarez Franco, Sara Isabel
In this article we assess the performance of three interest rate dynamic term structure models in order
to estimate the Value at Risk (VaR) of fixed-income portfolios. We find that that the model proposed
by Diebold, Rudebusch and Aruoba performs appropriately in VaR backtesting statistical tests, while the
model from Diebold and Li and a no-arbitrage akin term structure model display serious problems. The
three models assume that the variance-covariance matrix for their underlying factors is constant, which
limits their usefulness in estimating the VaR. Therefore, those models that relax this assumption should
perform better and be more adequate for risk-management of fixed-income portfolios.
