Analysis of the Brazilian yield curve: a no-arbitrage factor-augmented vector autoregression approach

Abstract

This dissertation: ANALYSIS OF THE BRAZILIAN YIELD CURVE: A NO-ARBITRAGE FACTOR-AUGMENTED VECTOR AUTOREGRESSION APPROACH, applies a parsimonious method to analyse the Brazilian term structure exploiting a vast number of macroeconomic variables. The procedure, developed in Moench (2008), combines the short-term interest rate with the principal components extracted from a large macroeconomic dataset. The short-term dynamics are described by a factor-augmented vector autoregression. Subsequently, the term structure is obtained by the no-arbitrage method. The results in-sample and out-of-sample of the so called No-arbitrage Factor Augmented Vector Autoregression (NAFAVAR) model is compared with the model in Diebold and Li (2006), since this model delivers both in-sample fitting and out-of-sample forecasts. The results of the NAFAVAR model outperforms the competitor model in some maturities of the term structure, which could be helpful for out-of-sample forecasts. The NA-FAVAR model seems to adapt well to the Brazilian interest rate market, which could help financial agents to evaluate and forecast securities using a model with macroeconomic interpretation.