The following paper describes the results of the forecast activity applied on steel prices time series, the method used in this analysis is a rolling window ARIMA regression. The forecast activity is built on prices time series that have been collected by anonymous weekly surveys sent to a sample of steel Italian operators. This pricing collection activity officially started in 2012 thanks to a tight cooperation between DMS stat Lab and Siderweb. After a short introduction, the statistical approach used for the forecast is deeply described and the time series of the four steel product prices considered in this analysis are introduced; then, the result of the forecasting exercise is presented together with final conclusion in the end of this paper.
|