Machine Learning Approaches to Macroeconomic Forecasting

Abstract

Forecasting national unemployment is one of the most important problems of modern economies, and most researchers have relied upon statistical techniques with their stringent data assumptions and low accuracy rates to predict changes in this macroeconomic data. This paper describes how a neural network using leading economic indicator data can help to pre­ dict civilian unemployment rates. Results show that the neural network provides superior esti­ mates of rates one month into the future compared to multi-linear regression and two naive forecasting techniques.