Authors
Advisor(s)
Abstract(s)
Na presente investigação são estudadas as diversas extensões do modelo de Nelson e Siegel (1987) com o propósito de analisar quais os modelos que revelam melhores resultados para ajustar e prever a Estrutura Temporal das Taxas de Juro na Zona Euro em contexto de grandes crises económicas e financeiras. Em particular, demonstramos que o modelo de 4 fatores, que inclui um segundo fator inclinação, é o modelo que melhor se ajusta às taxas de juro da amostra, revelando resultados bastante satisfatórios na medida em que consegue capturar as diversas formas que a yield curve assume ao longo do tempo, revelando resultados encorajadores mesmo em contextos de elevada instabilidade nos mercados quando comparado com os modelos de 2 e 3 fatores. Aplicamos, também, os modelos de Nelson e Siegel (1987) para efetuar previsões através de modelos autorregressivos para os seus parâmetros, tal como a literatura indica, e sugerimos um processo alternativo com a mesma finalidade que consiste em modelar diretamente as taxas de juro ajustadas pelos respetivos modelos. Chegamos à conclusão que a qualidade das previsões, geralmente, é superior quando utilizamos o método alternativo e que o modelo de 4 fatores continua a ser válido para efetuar previsões. No entanto, para esse efeito, o modelo de 4 fatores não revela resultados qualitativamente distintos do modelo de 3 fatores como quando avaliamos a sua capacidade para replicar as yield curves dentro da amostra. Por fim, aferimos que a presença de turbulência nos mercados piora significativamente as previsões das taxas de juro de curto prazo mas, regra geral, com o decorrer da maturidade e com o aumento dos desfasamentos para efeitos de previsão, o impacto de crises financeiras sobre a qualidade dos resultados tende a diminuir.
In this research we are studying the various extensions to the Nelson e Siegel (1987) model in order to analyze the models which show better results to adjust and forecast the Term Structure of Interest Rates in the Euro Area in the context of major economic and financial crisis. In particular, we show that the 4 factors model, which includes a second slope factor, is the model that best suits the interest rates of the sample, showing satisfactory results in that it captures the various ways that the yield curve assumes over time, showing encouraging results even in environments of high market instability when compared with the 2 and 3 factors models. We apply also the Nelson e Siegel (1987) model to make forecasts using autoregressive models for their parameters, such as the literature indicates, and we suggest an alternative process for the same purpose which is directly modeling the interest rates adjusted for the respective models. We conclude that the quality of forecasts is generally higher when we use the alternative method and the 4 factors model remains valid for forecasting. However, for this purpose, the 4 factors model does not disclose qualitatively distinct results compared to the 3 factors model when evaluating its ability to replicate yield curves in-sample fit. Finally, we verified that the presence of market turbulence worsens significantly forecasts of interest rates of short term but, generally, over the maturity and with increasing lags for forecasting purposes, the impact of financial crisis on quality of results tends to decrease.
In this research we are studying the various extensions to the Nelson e Siegel (1987) model in order to analyze the models which show better results to adjust and forecast the Term Structure of Interest Rates in the Euro Area in the context of major economic and financial crisis. In particular, we show that the 4 factors model, which includes a second slope factor, is the model that best suits the interest rates of the sample, showing satisfactory results in that it captures the various ways that the yield curve assumes over time, showing encouraging results even in environments of high market instability when compared with the 2 and 3 factors models. We apply also the Nelson e Siegel (1987) model to make forecasts using autoregressive models for their parameters, such as the literature indicates, and we suggest an alternative process for the same purpose which is directly modeling the interest rates adjusted for the respective models. We conclude that the quality of forecasts is generally higher when we use the alternative method and the 4 factors model remains valid for forecasting. However, for this purpose, the 4 factors model does not disclose qualitatively distinct results compared to the 3 factors model when evaluating its ability to replicate yield curves in-sample fit. Finally, we verified that the presence of market turbulence worsens significantly forecasts of interest rates of short term but, generally, over the maturity and with increasing lags for forecasting purposes, the impact of financial crisis on quality of results tends to decrease.
Description
Keywords
ETTJ Yield curve Taxas de juro Nelson e Siegel Subprime Ajustamentos Previsões Term structure of interest rate Interest rate Adjustment Forecasting