External Adjustment in Commodity Exporting Economies During Energy Price Downturns

ABSTRACT

Recent examples of energy price downturns did not alter trade surpluses in commodity exporting countries. Economic theory contends that such developments may indicate that the changes observed in energy prices are perceived as permanent by the market. We perform a trend-cycle decomposition of oil prices and find that permanent shocks account for about 90% of oil price variation. The observed variability of oil prices attributable to permanent shocks may be explained by the increased financialization of oil markets. This finding is in line with the empirical evidence that the current account surplus is generally unaffected by permanent terms-of-trade shocks.

KEYWORDS:

• Current account
• oil price
• terms of trade
• trend and cycle decomposition

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/08853908.2024.2366240.

Acknowledgments

The views expressed in this article are solely those of the authors and do not necessarily reflect the official position of the Bank of Russia. We are grateful to Kseniya Yakovleva for her valuable research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The dataset analyzed during the current study is available from the corresponding author upon reasonable request.

Notes

¹ Terms of trade represent the ratio between a country’s export prices and its import prices. Thus, an adverse shock occurs if the former fall or the latter rise to decrease the ratio as a whole.

² Close to our study is the research of Cashin, Liang, and McDermott (1999), who decompose 44 individual commodity series into permanent and temporary components. The authors find that temporary shocks influence commodity prices.

³ For a comprehensive discussion on the role of financial factors in the oil market and the expectations of financial investors, see the work of Fueki et al. (2018).

⁴ Figures 1 and 2 show the distribution (median and 25th and 75th percentiles) of 27 trajectories centered around the zero period (local minimum export values for 2009 and 2015; see Appendix 1 online for details).

⁵ İşcan (2000), Cashin and McDermott (2002), Bouakez and Kano (2008), and Adler, Magud, and Werner (2017) do not find a stable empirical link between current account fluctuations and terms-of-trade shocks.

⁶ See Harberger (1950) and Laursen and Metzler (1950).

⁷ For a more comprehensive survey, see Singh (2007).

⁸ Source: https://fred.stlouisfed.org/series/DCOILWTICO.

⁹ Source: https://www.dallasfed.org/research/igrea.

¹⁰ See Kilian (2009), Kilian and Zhou (2018), and Kilian (2019).

¹¹ Source: https://fred.stlouisfed.org/series/CUUR0000SA0R.

¹² The methodology of choice is similar to Leduc, Moran, and Vigfusson (2023) who use a very similar technique to decompose the oil price into trend and cycle. However, we use different specifications for these components and, in contrast to Leduc, Moran, and Vigfusson (2023), employ the extended Kalman filter to allow for time-varying coefficients.

¹³ To bootstrap the k-th difference of the oil price, we use the standard moving block bootstrap with block-length equal to 36 (Künsch 1989). (We utilize the fact that the k-th difference is stationary for k > 0.) We draw 10,000 bootstrap samples, calculate $S_k$ for each of them, and find the standard deviation of $S_k$ over all 10,000 samples.

¹⁴ Although this statistic is not “well-defined” since both the raw data and the trend component are non-stationary so their theoretical variances are infinite, their respective sample variances are finite (for a finite sample) and should have the same order of magnitude for a given sample size (compare Murray and Papanyan 2004).

¹⁵ To check the robustness of our results to the choice of exogenous variable D, we ran the model with Kilian’s (2009) business-cycle index replaced by the OECD + 6 industrial production index proposed in Baumeister and Hamilton (2019). More precisely, in one experiment, we used its growth rate, and in the other one, its stationary component as estimated via the HP filter (see Kilian and Zhou (2018) on extracting trend from this index). In both cases, the cycle in oil price appeared to be even less pronounced than in the baseline model, though the optimal likelihood was substantially lower.

¹⁶ The value of the smoothing parameter used in the experiments is not explicitly given in the article.

¹⁷ Source: https://fred.stlouisfed.org/series/PMETAINDEXM.

¹⁸ In 2019, iron ore accounted for 23.8% of exports. See https://oec.world/en/profile/country/aus.

¹⁹ In 2019, copper ore accounted for 26% of exports. See https://oec.world/en/profile/country/per.