Inflation forecasting using a neural network

Abstract This paper evaluates the usefulness of neural networks for inflation forecasting. In a pseudo-out-of-sample forecasting experiment using recent U.S. data, neural networks outperform univariate autoregressive models on average for short horizons of one and two quarters. A simple specification of the neural network model and specialized estimation procedures from the neural networks literature appear to play significant roles in the success of the neural network model.