Understanding Fluctuations Through Multivariate Circulant Singular Spectrum Analysis
First published: 2020
Brief summary
Applies multivariate circulant singular spectrum analysis to a panel of energy commodity prices (oil, natural gas, coal, propane), identifying a dominant 96-month (8-year) common cycle explaining over 20% of shared variability across commodities.
Article
Understanding Fluctuations Through Multivariate Circulant Singular Spectrum Analysis is a preprint published by arXiv in 2020. It applies multivariate circulant singular spectrum analysis to a panel of energy commodity prices (oil, natural gas, coal, propane), identifying a dominant 96-month (8-year) common cycle explaining over 20% of shared variability across commodities.
The analysis focuses on 96 months (8 years), explaining 20.3% of variability. The data source is panel of oil (Brent, Dubai, WTI), natural gas (Europe, US, Japan), coal (Australia, South Africa) and propane price series. This gives the cycle claim a specific numerical and evidential setting rather than presenting periodicity only as a visual impression.
The article reports the following result: The second component in terms of relevance is the 96-month, 8-year, cycle. Within the 20.3% of variability explained by this cycle, it is mainly described by co-movement of oil prices with European natural gas, while coal and propane show distinct phase relationships. The interpretation is strongest when it survives detrending, structural-break tests, changing market composition and out-of-sample validation.
For cycles researchers, the article brings together singular spectrum analysis, energy commodities, 8-year cycle, commodity co-movement. It is relevant to commodity-cycle research because long waves, business-cycle components and structural breaks can overlap in price records.
Because it is a preprint, the work should be read alongside later peer-reviewed publications and independent replications. It remains useful because the proposed cycle, dataset and analytical approach are stated clearly enough to be scrutinised.
Source details and credits
- Source / publisher: arXiv
- Source type: Preprint
- URL type: PDF
- Credits: arXiv
- URL: https://arxiv.org/pdf/2007.07561