Wars, cartels and COVID-19: regime switching in commodity prices

ABSTRACT

Commodity prices are extremely volatile, and volatility itself fluctuates over time. Using data from 1959 to 2022, we estimate a 3-state Markov-switching model to identify expansions and contractions in oil and copper price volatility. We found a transition from a low to a medium variance regime for the oil price, in 1979, reflecting changes in the oil market structure. In addition, we identify infrequent and short-lived episodes of unusually high oil price volatility. For copper, there is no transition across regimes, and episodes of high volatility are not synchronized with the periods of high volatility in oil prices. We found that oil prices are much more volatile than copper prices in all states. Oil prices react more strongly to market cartelization, war episodes, and global demand shifts, like the 2008 Great Recession and the COVID-19.

KEYWORDS:

• Commodity price volatility
• Markov-switching
• oil and copper prices

JEL CLASSIFICATIONS:

• C11
• Q02
• Q35

Acknowledgments

We are grateful Alfonso Irarrazabal for helpful comments and discussions. We thank Universidad de Santiago and DYCIT-USACH for funding this research through 03060CG scheme.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

¹ Kilian (2009), Kilian and Murphy (2012), and Lippi and Nobili (2012) found that oil price movements have been driven mainly by a combination of global aggregate demand shocks and precautionary demand shocks, rather than oil supply shocks, as was commonly believed. Caldara et al., (2019), and Christiane and Hamilton, (2019) found a larger impact of supply shocks but also conclude that demand shocks play an important role in explaining oil price fluctuation. In the case of copper and other metals, Jacks and Stuermer (2020) concluded that, as in the case of oil, demand shocks are the main drivers of price fluctuations.

² Dvir and Rogoff (2010) and Dvir and Rogoff (2014), using annual data from 1861 to 2008, found strong evidence of changes in persistence and volatility of oil price across three well-defined periods or, as they called it, three epochs. However, these changes are a one-off event. Episodes of high, medium, and low volatility do not coexist in a given time frame.

³ We restrict our attention to these two important commodities. The analysis, however, is general enough and hence, it can be extended to other commodity prices, which may be relevant for a particular group of countries.

⁴ For a more comprehensive exposition of this method, see Hamilton (1994), Kim and Nelson (1999) and Fruhwirth-Schnatter (2004).

⁵ Multiple logistic regression is a different statistical technique used when the probability of a dichotomous outcome, such as the presence or absence of a consumer choice, needs to be estimated. This approach is used in Boccia and Sarnacchiaro (2020), but is difficult to extend it to computing a non dichotomic outcome, as in this paper.

⁶ See Appendix for details on the estimation procedure and specifications.

⁷ On April 20th, the May 2020 contract futures price for West Texas Intermediate (WTI) plummeted from USundefined to -USundefined a barrel.

⁸ The probability of remaining in the MV and HV regime are similar: close to 0.95.

Additional information

Funding

The work was supported by the DICYT [DICYT 032060CG grant scheme].