A Comparison of Hybrid ARMA-Elman Models with Single Models for Forecasting Interest Rates

Since ANNs model could capture the nonlinearity of time series, it performances well on forecasting time series. Hybrid or combined ANNs with ARMA models are extensively studied and used in financial time series forecasting. But we doubt the necessity to build the hybrid models to forecast time series. Do hybrid models always outperform the single ANNs models? This paper is aimed to answer it. Two kinds of hybrid ARMA-Elman models are built, one with innovations as inputs, another with innovations and original data as inputs. Using the data of benchmark interest rates of China, the empirical results indicate that the hybrid models are superior to the single ARMA model, but perform closely to the Elman recurrent neural network model. The single Elman model is enough on forecasting. But if the correct rate on forecasting change directions is concerned, the hybrid models are preferred.