Spectral analysis of oil price fluctuations in the Philippines

Abstract

Spectral analysis was conducted on available time-series data for different oil prices to characterize underlying periodicities and market drivers. Specifically, the discrete Fourier transform was used as a spectral analysis tool to obtain their power spectrum distributions and reveal prevailing patterns on the oil price fluctuations. Results show that the multi-decade time series is non-stationary, and the magnitudes of the lowest-frequency peaks in the power spectrum distribution reflect the overarching inflationary price trend instead of stationary periodic cycles occurring around a fixed mean. The dynamics of the refined products derived from Dubai crude were also found to have high temporal correlation with Dubai crude. Meanwhile the local oil prices showed no shared spectral peaks with the USD/PHP exchange rate, indicating that any currency-driven effects are not expressed as common periodic cycles detectable by PSD peak matching. Lastly, a resolution-dependent spectral gap was identified as the higher-resolution weekly datasets for RON95 and RON97 revealed near-annual cycles that are otherwise aliased in monthly-sampled datasets.