Forecasting latin-american yield curves: an artificial neural network approach

Abstract

This document explores the predictive power of the yield curves in Latin America (Colombia, Mexico, Peru and Chile) taking into account the factors set by the specifications of Nelson & Siegel and Svensson. Several forecasting methodologies are contrasted: an autoregressive model, a vector autoregressive model, artificial neural networks on each individual factor, and artificial neural networks on all factors that explain the yield curve. The out-of-sample performance of the fitting models improves with the neural networks in the one-month-ahead forecast along all studied yield curves. Moreover, the three factor model developed by Nelson & Siegel proves to be the best choice for out-of-sample forecasting. Finally, the success of the cross variable interaction strongly depends on the selected yield curve.

Forecasting the Term Structure of Crude Oil Futures Prices with Neural Networks

“…β 0t , β 1t , β 2t and λ t for all t. The main problem of such an approach is, that λ t may be unstable due to unexpected jumps. While the model will fit the data very well, its predictive power deteriorates (Vela, 2013). We find the optimal values of λ t by minimizing sum of squared errors of Nelson-Siegel approximations of WTI futures term structure for each observed point in time.…”

An Efficient Deep Learning Based Model to Predict Interest Rate Using Twitter Sentiment

“…In addition, these models cope with complexities like non-linearity, structural breaks, and seasonality issues laying in variables. Most of the research focuses on the prediction of interest rates considering the time dynamics [46][47][48][49][50]. There are several studies which follow machine learning models to predict the interest rate.…”

Modelling and forecasting interest rates during stages of the economic cycle: A knowledge-discovery approach

“…The main motivation behind this approach is to address the limitations of previous streams to handle complex non-linear relationships between variables, in addition to seasonality and the presence of structural breaks. The majority of research in this sub-stream relates to predicting interest rates, that is, modelling the time dynamics of one specific point on the yield curve (Jacovides, 2008;Kim & Noh, 1997;Oh & Han, 2000;Vela, 2013;Zimmermann et al, 2002). Work has also been published on analysing the characteristics of the economic cycles on the basis of the slope of yield curves attempting to forecast recessions using machine-learning techniques, such as Support Vector Machines (SVM) (Gogas et al, 2014).…”

“…It is shown that neural network models outperform regression models, as evidenced by the R-squared and mean square error performance metrics. In Vela (2013) the predictive power of the yield curves for different countries in Latin America is analysed based on variables from previous research such as Nelson and Siegel (1987) and Svensson (1994).. In comparative terms, the neural networks had a higher yield than the other models, but for some curves, the neural networks did not exceed the random walk model.…”

“…The empirical stream deploys the use of different econometric methods and knowledge-discovery techniques in order to model the structure and dynamics of the yield curve (Enke & Thawornwong, 2005;Gogas, Chionis & Pragkidis, 2009;Gogas, Papadimitriou, Matthaiou & Chrysanthidou, 2014;Jacovides, 2008;Ju, Kim & Shim, 1997;Kim & Noh, 1997;Oh & Han, 2000;Vela, 2013;Zimmermann, Tietz & Grothmann, 2002). Recently, the use of knowledge-discovery techniques has gained popularity, mainly due to the fact that these are able to handle complex, non-linear relationships between variables, seasonality and the presence of structural breaks (Gogas et al, 2014;Jacovides, 2008;Ju et al, 1997;Kim & Noh, 1997;Oh & Han, 2000;Vela, 2013;Zimmermann et al, 2002). However, these are not exempt from limitations.…”