GOR AG online webinar on "Commodity Price Forecasting in Procurement"

The German OR Society working group in Forecasting is launching a series of webinars leading up to our in-person meeting next year in Munich.

Our 1st speaker is Nico Beck, PhD candidate at the Friedrich-Alexander-Universität Erlangen-Nürnberg, and Scientific Researcher in Data Science at the Fraunhofer-working group for Supply Chain Services SCS. Nico will talk about his research topic as outlined below.

The webinar is open to members of the GOR and all other interrested parties, no membership required!

Please RSVP in the Meetup page!

Commodity Price Forecasting in Procurement
Commodity price forecasting has gained attention in forecasting, since it promises a huge benefit in procurement. Therefore, forecasters work on improving their models to enhance the accuracy. However, in areas like stock or price forecasting, which are very volatile, even small improvements compared to the naïve (no-change) forecast can be seen as a success. In general, it is unclear whether the information gained from the forecast induces a benefit in the actual procurement.
To assess the actual value of commodity price forecasting, we predict prices of four steel types, that are highly relevant for large industry sectors, on a weekly basis for a horizon of twelve weeks. Afterwards, we simulate eight half year periods of procurement for each of the steel types, where the procurement decisions are fully based on the forecasts. We compare ARIMA, ARIMAX, ETS and Historical Consistent Neural Networks regarding their forecast accuracy and the costs they produce in the procurement simulation.
We show that a good model forecast accuracy does not necessarily produce a low cost in the procurement simulation. Nevertheless, all forecast approaches can reduce costs compared to a baseline procurement policy, where the best approach outperforms the latter on 20 of 32 sub-periods.