Long memory in oil and refined products markets

Kyongwook Choi, Shawkat Hammoudeh

Research output: Contribution to specialist publicationArticle

42 Scopus citations


We test for the presence of long memory in daily oil and refined products prices' absolute return, squared return and conditional volatility, using several parametric and semiparametric methods. This study finds strong evidence of long memory (LM) in the daily absolute and squared spot and futures returns for crude oil, gasoline and heating oil but at different degrees. The FIGARCH model also demonstrates strong evidence of LM for volatility for most of oil and products prices' returns, with also different resilience levels. Structural breaks have only the partial effects of slightly reducing persistence for just absolute and squared returns. Examining the forecasting behavior of two competing models, the less parsimonious ARFIMA which satisfies the LM property, and the parsimonious ARMA with short-term processes, the ARFIMA model provides significantly better out-of-sample forecasts at all forecasting horizons for all three petroleum types.

Original languageEnglish
Number of pages20
Specialist publicationEnergy Journal
StatePublished - 2009


Dive into the research topics of 'Long memory in oil and refined products markets'. Together they form a unique fingerprint.

Cite this