Dynamic analysis of the relationship between exchange rates and oil prices: a comparison between oil exporting and oil importing countries

The study investigates how crude oil prices and exchange rates are interconnected, focusing on systemic risk transmission across markets. Motivated by evidence that heightened interconnectedness amplified the 2007–2008 crisis and that oil price shocks affect macroeconomic variables and currencies, the paper asks: How do oil prices and exchange rates co-move and transmit shocks over time, and how do these dynamics differ between oil-exporting and oil-importing countries? The authors emphasize that relationships are time-varying and potentially bidirectional, mediated by trade, wealth, and portfolio reallocation channels. Understanding these dynamics is important for policymakers, investors, and risk managers given currency markets’ responsiveness to macroeconomic conditions and oil’s central role in the global economy, especially during episodes like the GFC, European debt crisis, Brexit, and COVID-19.

Prior studies use connectedness/network approaches to examine currency and oil interactions, with mixed findings driven by sample, frequency, and methods. Diebold and Yilmaz (2012, 2014) popularized generalized connectedness, documenting volatility spillovers; subsequent work shows the Euro and USD as key volatility sources (Wen and Wang, 2020), and crisis-sensitive increases in connectedness. Singh et al. (2018) found time-varying links between crude oil and currency implied volatility; earlier periods saw oil driving currencies, later the reverse. Malik and Umar (2019) reported strengthened oil–FX linkages post-GFC. Studies highlight effects of trade policy uncertainty (Huynh et al., 2020), Bayesian time-varying connectedness for G7 (Wan and He, 2021), and commodity–currency spillovers (Nekhili et al., 2021). Evidence on directionality is mixed; some find strong US dollar–oil–financial market co-movements during COVID-19 (Adekoya and Oliyide, 2020), others weaker currency–energy ties (Asadi et al., 2022). Theoretical channels include trade balance effects, wealth effects favoring exporters’ currencies, and portfolio reallocations (Backus and Crucini, 2000; Habib et al., 2016; Beckmann et al., 2020). Overall, literature indicates time-varying, crisis-sensitive, and heterogeneous spillovers across currencies and commodities, with possible asymmetries between oil importers and exporters.

Data: Monthly series constructed from daily data for exchange rates of 11 economies versus USD and crude oil futures. Oil-exporting currencies: Canada, Mexico, Russia, Brazil, Norway. Oil-importing currencies/regions: UK, South Korea, Japan, China, Euro area, India. Exchange rates sourced from FRED (Federal Reserve Economic Data); crude oil proxied by S&P GSCI Crude Oil Index, First Nearby Contract (DataStream code GSCLSPT). Monthly values are averages of daily quotes. Returns computed as log-returns to address unit roots. Sample spans July 1998 to August 2023 (subject to availability). Descriptive and preliminary tests: Return series display leptokurtosis and skewness (D’Agostino, 1970; Anscombe and Glynn, 1983). Jarque–Bera tests reject normality; Fisher–Gallagher weighted portmanteau tests indicate autocorrelation in returns and squared returns. ERS unit root tests indicate stationarity of return series. Econometric approaches: (1) Generalized connectedness (Diebold and Yilmaz, 2012, 2014) using generalized forecast error variance decomposition (GFEVD). (2) Time-varying parameter VAR (TVP-VAR) extended joint connectedness (Balcilar et al., 2021) with refinements by Lastrapes and Wiesen (2021) and Antonakakis et al. (2020). TVP-VAR accommodates time-varying parameters and covariances, is robust to outliers, avoids arbitrary rolling windows, and provides lossless forecast error variance decomposition. Modeling details: TVP-VAR estimated with one lag (BIC). Connectedness metrics derived from TVP-VMA representation and time-varying GFEVD at a 20-step-ahead horizon. Measures reported include total connectedness index (TCI), net total directional connectedness (NTDC), and net pairwise directional connectedness (NPDC). An extended joint scaling/normalization is applied per Lastrapes and Wiesen (2021). Robustness: Asymmetric TVP-VAR estimated separately for positive and negative returns (Appendix 2), yielding qualitatively similar connectedness patterns.

- Average connectedness: The network exhibits substantial cross-market dependence. The Total Connectedness Index (TCI) averages around 45% (approximately 44–45%), indicating that nearly half of the forecast error variance is due to cross-variable spillovers. - Crisis sensitivity: Dynamic TCI spikes above 80% during 2006–2008 (pre-GFC and onset), remains elevated during 2011–2014 (European sovereign debt crisis), declines with the 2014 oil price collapse, resurges around Brexit (2016–2017), and is elevated at COVID-19 onset, evidencing strong crisis amplification. - Roles of countries/assets (TVP-VAR joint connectedness): Canadian dollar emerges as a prominent net transmitter of shocks on average; Russian ruble and Japanese yen are key net receivers. Currency rates of oil-importing nations tend to be net recipients overall (with China an exception in some periods). Crude oil is a moderate net recipient on average but alternates roles over time. - Pairwise patterns: Crude oil has strong influence on Norway and Russia exchange rates; conversely, Norway and Canada exchange rate fluctuations significantly affect crude oil price volatility, evidencing bidirectional linkages. Interconnectedness between crude oil and Norway, Russia, and Canada is consistently higher than with other currencies in the sample. - Time variation in roles: No asset is a perpetual net transmitter/receiver; roles switch over time. Under joint connectedness, crude oil often acts as a net transmitter from 2006 to 2016, whereas under DY it is more often a net receiver. - Exporters vs importers: Connectedness between oil prices and exchange rates is stronger for oil-exporting countries than for oil-importing countries. - Method comparison: TVP-VAR extended joint connectedness yields different role assignments for some currencies relative to DY. For example, Euro is a net receiver under joint connectedness but a transmitter under DY; Norwegian krone shifts from net sender (DY) to net receiver (joint) in many periods. Despite differences, both approaches agree that the Russian ruble and Japanese yen are major net receivers. - Robustness: Asymmetric TVP-VAR for positive and negative returns confirms crude oil as a moderate net receiver in all cases (net values roughly between −2.12 and −0.11) and preserves the main connectedness patterns.

The findings answer the research question by demonstrating that oil–FX connectedness is highly time-varying, bidirectional, and crisis-sensitive, with stronger linkages for oil exporters. The evidence that crude oil and exchange rates alternately transmit and receive shocks shows that causality is not unidirectional and depends on market regimes and events. Elevated TCI during major crises underscores systemic risk propagation across oil and currency markets. Differences between TVP-VAR joint and DY generalized methods highlight the importance of modeling time variation and normalization in capturing dynamic relationships; policy conclusions can change depending on the method. For policymakers and market participants, these results imply that oil shocks can materially affect currency stability (especially in exporter economies), while currency movements can feed back into oil markets, requiring agile hedging, diversified risk controls, and method-aware monitoring frameworks.

The study introduces a TVP-VAR extended joint connectedness framework to analyze the dynamic relationship between crude oil futures and exchange rates for five oil exporters and six oil importers over 1998–2023. Methodologically, the approach avoids arbitrary rolling windows, adapts to time-varying parameters, and refines normalization, providing richer dynamics than the generalized DY method. Empirically, the system exhibits substantial interconnectedness (TCI ≈ 45%), with spikes during global stress episodes. Oil–FX interactions are bidirectional and stronger for exporters; Canada is a key net transmitter while Russia and Japan are net receivers, and crude oil alternates between transmitter and receiver roles over time. These insights matter for investors, risk managers, and policymakers, as heightened oil–FX connectedness signals elevated systemic risk and reduced diversification benefits during turmoil. Future research should broaden country coverage, consider additional connectedness metrics, and examine how the growth of renewable energy reshapes oil–FX linkages.

- Sample composition: Although including both exporters and importers, the set centers on major oil-related economies; broader inclusion of diverse countries could enhance generalizability. - Metric scope: Heavy reliance on the TCI and variance-decomposition-based measures; incorporating additional indices or alternative methodologies (e.g., frequency-domain, high-frequency, or structural identification of shocks) could provide complementary insights. - Method differences: Discrepancies between TVP-VAR joint and DY generalized connectedness suggest sensitivity to modeling choices and normalization; conclusions for specific currencies may vary by method. - Evolving context: Relationships are regime-dependent; results may shift with future macroeconomic changes, policy regimes, or market structures (e.g., energy transition toward renewables).