Brazilian term structure of interest rate modeling: A Nelson-Siegel approach

Abstract

Forecasting interest rates structures plays a fundamental role in the fixed income and bond markets. The development of dynamic modeling, especially after Nelson and Siegel (1987) work, parsimonious models based in a few parameter shed light over a new path for the market players. Despite the extensive literature on the term structure of interest rates modeling and the existence in the Brazilian market of various yield curves from different traded asset classes, the literature focused only in the fixed rate curve. In this work we expand the existing literature on modeling the term structure of Brazilian interest rates evaluating all the yield curves of Brazilian market using the methodology proposed by Nelson and Siegel. We use Non Linear Least Squares (NLLS) to estimate the model parameters for almost 10 years of monthly data and model these parameters with the traditional VAR/VEC model. The results show that it is possible to estimate the Nelson Siegel model for the Brazilian curves. It remains for future research the modeling of their variances as well as the possibility to develop a global Brazilian model using Kalman Filter using the Diebold. Li. and Yue (2006) approach.