This paper first develops a new approach, which is based on the Nelson-Siegel term structure factor-augmented model, to compute the VaR of bond portfolios. We then applied the model to examine whether information contained on macroeconomic variables and financial shocks can help to explain the variations of VaR. A principal component analysis is used to incorporate the information contained in different variables. The empirical result shows that, including macroeconomic variables and financial shocks in the Nelson-Siegel term structure factor model, we can observe an obvious tendency towards better VaR forecasting performance. Moreover, the impact of incorporating financial shocks seems to be stronger than that of incorporating macroeconomic variables.
Nelson-Siegel factor model Value-at-risk Encompassing test Backtesting Conditional predictive ability