Using artificial neural networks to forecast producer price index for New Zealand

Abstract

Trend in the producer price is of much value to the central bank authorities in identifying the cost-push inflation that can improve their understanding of future directions of inflation in the aggregate economy and informulating sound policies and macroeconomic plans. Forecasting of the producer price movement is complex; the popular use of conventional methods is fraught with inaccuracies which often produces misleading results. This study explored the reliability and accuracy of the use of artificial neural networks (ANNs) for modelling and predicting producer price index (PPI) trend in New Zealand. The study also compared ANNs results with those produced by the autoregressive integrated moving average (ARIMA) as an alternative. Results showed that the ANNs model outperformed the ARIMA model as a more reliable and accurate tool for time series data prediction. The method developed could guide economists and macroeconomic policymakers in making more accurate forecasts.