Federal Funds Rate Prediction Using Robust Radial Basis Function Neural Networks

Since some studies have found that monetary policy influences the financial market, the prediction of effective federal funds rate has been an important issue. In this paper, we construct the M-estimator based robust RBF (MRRBF) neural network and compare the forecasting performances with some other time-series forecasting models for daily U.S effective federal funds rate. We find that the proposed MRRBF network can produce the lowest root mean square errors due to the ability to eliminate the outlier influence.