Asymmetric Verification of Business Cycle by Forecasting Turning Points Based on Neural Networks

Abstract

This paper examines the relevance of various financial and economic indicators in forecasting business cycle turning points via neural networks (NN) models. We employ a feed forward neural network model to forecast turning points in the business cycle of China. The NN has as inputs thirteen indicators of economic activity and as output the probability of a recession. The different indicators are ranked in terms of their effectiveness of predicting China recessions. The out-of-sample results show that via the NN model indicators, such as steel output, M2, Pig iron yield and freight volume of whole society are useful in forecasting China recessions. Meanwhile, based on this method, asymmetry of business cycle can be verified.