Modelling the dynamics of commodity markets and forecasting their prices with time series models

Fluctuations in commodity prices exert a large impact on the global economy, mainly due to energy demand in the industry, as well as the sizeable role of oil in transportation costs. At the household level, commodity prices influence the costs of living both directly (i.e. through petrol prices) and indirectly (through their impact on costs of producing consumer goods). At a firm level, they are an important factor determining profits. For instance, in the airline industry, profits depend heavily on prices of jet fuel, hence oil price forecasts are used to set tariffs and assess aircraft purchasing strategy.

In turn, car manufacturers use oil price forecasts to design new products focused on fuel economy. For the above reasons, understanding commodity price dynamics and the ability to formulate their reliable forecasts are important to many economic agents. For companies, they are helpful in assessing strategic policies or investment decisions with long-term impacts. For countries in which exports of commodities is an important source of revenues, their price forecasts are useful in predicting the budget balance.

Ultimately, for central banks, they would help in assessing the future path of inflation, gross domestic product or external imbalances, and thus would allow for better conduct of the monetary policy. The reference benchmark in commodity price forecasting is a simple no-change forecast, assuming that the future price will stay at the current level.

It turns out that constructing econometric or economic models that can consistently outperform the no-change forecast is a real challenge. In the project, we plan to address this issue and conduct a comprehensive study using a large set of forecasting models and methods to verify whether it is possible to forecast the prices of main commodities (oil, gold, copper) better than with the no-change method. The results of the study will allow us to identify the best performing methods in terms of out-of-sample accuracy, as well as to better understand the dynamics of commodity prices. It could be noted that the results of the study can be exploited in practise in the decision-taking process.

Project director:
Michał Rubaszek, Ph.D., SGH Professor
Financing institution:
National Science Centre
Project duration:
January 2018 - January 2022
Web of science classification category:
Organizational unit (collegium/department/unit):
SGH Warsaw School of Economics » Collegia » Collegium of Economic Analysis