In this paper we implement dynamic extensions to the Nelson-Siegel term structure model and investigate hedging performance using the time-varying parameters of these extensions. Using data from brazilian interest rate futures market contracts, we estimate different specifications of the Nelson-Siegel model with special attention to the econometric treatment of decay parameter and common stochastic volatility. Since the model is non-linear, we propose the use of the Kalman Filter combined with GAS-type dynamics (Generalized Autoregressive Score) to estimate the parameters of interest. The great advantage of this methodology is that the likelihood function of the state space model with time-varying parameters is available in closed form, which facilitates the estimation of parameters by maximum likelihood. An innovative feature of this work is the empirical application of Nelson-Siegel models to hedge long-term exposures. We demonstrate how the different estimated specifications perform in the context of immunizing a fixed-income portfolio of long-term credit assets using short-term derivative instruments. Our estimates indicate that considering dynamics in the decay parameter combined with stochastic volatility significantly improves hedge performance. These results are of great interest to financial institutions in the Brazilian market, which do not find sufficient liquidity to hedge long maturities and need to adopt other strategies that increase transaction costs.
Complete tool for submission, evaluation and publication of proceedings for scientific events.
