CDS Risk Premia Forecasting with Multi-Featured Deep Rnns : An Application on Br[I]Cs Countries

This study tries to predict as accurately as possible CDS risk premia for BRICS countries by utilizing state-of-the-art recurrent neural network architectures. These are ELMAN, NARX, GRU, and LSTM RNNs in the time series setting, taking into account both local and global features. The predictive power of each architecture is compared, and the outcomes differ according to country by country. NARX RNN is found as the best predictor with different settings for Brazil and South Africa. While China gets the more accurate output with ELMAN RNN, Russia’s long short-term memory RNN achieves the best predictors among other countries’ RNNs