Assessing the extent and persistence of major crisis events in the crude oil market and economy: evidence from the past 30 years

The paper addresses how long and how strongly major crisis events affect the crude oil market and broader economies, and whether different event types exhibit distinct magnitudes and timings of impact. Motivated by recent episodes (e.g., COVID-19, Russia–Ukraine conflict, financial turmoil) that triggered sharp oil price swings and economic disruptions, the study emphasizes the importance of long-term effects over immediate reactions, as long-run consequences can alter market efficiency and equilibria. Given oil market institutional features (e.g., OPEC+ quota adjustments typically annual) and financial market conventions, the authors define short-term as up to one year and long-term as beyond one year. They identify key gaps: difficulty quantifying diverse events for long-horizon econometric analysis, lack of unified treatment across event types, and endogeneity concerns. The study proposes to quantify events via oil futures price changes and to use a Proxy-SVAR to uncover general rules governing the long-term impacts of crises on oil prices, production, stock markets, industrial production (global, US, China), and inflation (US, China).

Prior work has struggled to quantify major crisis events for econometric analysis. Four main approaches are reviewed: (1) Dummy variables around event dates (0/1), used to study volatility/spillovers (e.g., Karali and Ramirez 2014; Wen et al. 2021). This provides only rough estimates, cannot differentiate intensities across many events, nor depict time-varying effects. (2) Event study methods for specific events (e.g., Gulf War, Russia–Ukraine war) that capture short-term impacts but are unsuitable for long-term horizons where multiple shocks confound effects. (3) Before–after comparisons via segmented regressions or indicators (e.g., VAR/HP filters, Hurst exponent analyses), which may mix effects of concurrent events and limit generalization. (4) Constructed indices for specific event types, such as Geopolitical Risk (GPR) or COVID-19 indicators (cases, deaths, news), effective within a type but not for integrating all crises simultaneously. The literature rarely provides a systematic, unified, and endogenous-shock-aware assessment across diverse crisis types. Endogeneity of events (e.g., oil-related geopolitical conflicts) is typically unaddressed, except for studies on OPEC announcements. This motivates a Proxy-SVAR approach using suitable external instruments to identify structural shocks from major crises and obtain generalizable long-term patterns.

Design: A Proxy-SVAR (instrumented SVAR) identifies the structural shock associated with major crisis events (MCEs) using external instruments correlated with the crisis shock and orthogonal to other shocks. Following Stock and Watson (2012), Mertens and Ravn (2013), and applications in oil markets (Känzig 2021), the model estimates impulse responses without directly imposing strong structural parameter restrictions. Shock scaling follows Känzig (2021), normalizing to a 10% positive oil price shock for interpretation. Event set and proxy construction: The study covers January 1990–December 2022 and selects 50 landmark crises across four categories: (1) wars/geopolitical conflicts (21), (2) natural disasters (16), (3) economic and financial crises (7), and (4) public health emergencies (6). Selection used multiple sources: systemicpeace war list (magnitude ≥52), EM-DAT disasters (≥10 million affected and ≥$10B economic loss, adjusted), Wikipedia lists for financial crises and epidemics, and prior literature. To quantify events, the instrumental variable is the percentage change in WTI futures closing prices from the last trading day before to the first trading day after each event; daily changes are aggregated to monthly to form the MCE proxy series. Data: Monthly variables include WTI spot price (deflated by US CPI to 2015 base), world oil production (Datastream, MBBL/DAY), S&P 500 (Investing.com), OECD+6 industrial production index (Baumeister and Hamilton dataset), US industrial production (FRED), US CPI (FRED, 2015 base), China industrial production (constructed fixed-base index using industrial added value from WIND to extend back to 1989), and China CPI (WIND, 2015 base). Stationarity: Unit root tests (ADF, PP) show MCE is I(0); other series become stationary after log and first differencing. The VAR is estimated with the original MCE and first-differenced logs of other variables. Identification leverages the correlation of the instrument with the target crisis shock and orthogonality to other shocks; first-stage F-statistics assess instrument strength by category. Estimation and validation: Impulse responses with bootstrapped confidence intervals (10,000 draws) are reported. Instrument strength: overall robust F≈19.7 (no weak IV issue); wars/geopolitics robust F≈13.5; public health robust F≈77.9; natural disasters robust F≈0.52 and financial crises robust F≈0.09 (weak IVs, thus results treated as preliminary for these categories). Historical decomposition of real oil price evaluates cumulative contributions of crises to major swings (e.g., 1990 Gulf War spike; 2008 GFC and 2020 COVID-19 collapses; 2022 Russia–Ukraine surge).

General impacts (all events): - Oil market: Major crises cause an immediate surge in real oil prices that subsequently weakens, turns negative, oscillates, and converges to zero by around 20 months. World oil production falls on impact and recovers after about one period. - Macroeconomy: S&P 500 initially rises modestly then turns negative by period 2 before converging; world industrial production peaks at about 0.358% in period 1 then declines to zero; US industrial production shows a short-run positive response; China’s industrial production is negative on impact, turns positive (peaks near period 3 at about 1.61%), then negative again before converging. CPI rises in both US and China, with the US response larger early (through period 5) and China’s later response exceeding the US, reflecting domestic stimulus and spillovers from US monetary policy. - Durations: Oil market effects ≈2 years; macroeconomic effects ≈2.5 years. Historical decomposition: Major crises explain large jumps/drops in real oil prices: sharp rise during the 1990 Gulf War; modest rise around 1999 NATO bombing; limited effect from 9/11; a ~70% fall during the 2008 GFC; a >71% decline during COVID-19 onset; and a surge to a new peak by June 2022 amid the Russia–Ukraine war and sanctions. By crisis type: - Wars/geopolitical conflicts: Strong positive oil price impact and initial oil production decline; S&P 500 drops about 2.99% on impact with a V-shaped rebound; world, US, and China industrial production indices decline; CPI rises in both countries. Duration: oil market ≈3 years; US macro ≈2.5 years; China macro ≈4 years. Instrument strong (robust F≈13.5). - Natural disasters: Oil price spikes then retreats; world oil production declines then rebounds; S&P 500 shows a short-lived positive response; world and US industrial production respond positively initially (driven by export/reconstruction demand, as most severe events occur in Asia), while China’s IP declines on impact then temporarily rises. CPI initially declines in both countries (deflationary demand shock), then turns positive as policy eases; China’s CPI response later exceeds the US. Duration: ≈1.5 years globally; ≈2 years in China. Instrument weak (robust F≈0.52). - Economic/financial crises: Oil prices fall quickly; world oil production initially rises (demand-side slump); China’s industrial production is highly volatile due to aggressive infrastructure stimulus and subsequent overcapacity, while US IP reacts less. CPI rises in both countries amid monetary/fiscal stimulus. Duration: oil market >3 years; global macro >4 years. Instrument weak (robust F≈0.09). - Public health emergencies: Oil prices decline from periods 2–10 and converge thereafter; early positive oil production response reflects 2020 events (Vienna pact breakdown, Saudi price war) before turning negative after period 5; S&P 500 overreacts positively at first then turns negative quickly; world and US IP positive in first two periods then negative by period 3; China’s IP shows stronger and more persistent negative effects due to stringent containment and supply chain disruptions. CPI: US initially positive then declines; China fluctuates around zero with diminishing amplitude. Duration: oil market impact >3 years (COVID-19 case); macro ≈2.5 years globally; ≈3 years in China. Instrument strong (robust F≈77.9).

The study identifies general rules governing how major crises transmit to the oil market and real economy. Supply-side shocks (wars/geopolitical conflicts, natural disasters) reduce production and raise oil prices, depressing industrial production (notably in China), with geopolitical conflicts causing stronger and longer-lasting effects than natural disasters. Demand-side shocks (financial crises, public health emergencies) lower oil prices and can initially increase measured oil production relative to demand, with financial crises exhibiting the most persistent and widespread macro impacts. Across all categories, CPI pressures are notable: wars/geopolitics and financial crises are inflationary in both countries, while natural disasters initially depress inflation before policy easing lifts CPI; health emergencies induce significant US inflation via aggressive monetary expansion, with spillovers affecting China’s inflation in later periods. These findings address the research questions by quantifying the magnitude and persistence of crisis impacts, distinguishing mechanisms by event type, and highlighting cross-country asymmetries (more prolonged and intense effects in China relative to the US). The results underscore the need for tailored policy responses: rapid risk containment and monitoring, diversified and resilient energy sourcing, strong financial system safeguards, inflation spillover management, and firm-level risk hedging.

The paper constructs a unified, endogeneity-aware framework to quantify and assess the long-term effects of diverse major crisis events on the crude oil market and the economy. Using an oil-futures-based proxy in a Proxy-SVAR with monthly data (1990–2022) across 50 events, the study finds: (1) crises hit oil markets more severely but with shorter duration (≈2 years), while macroeconomic effects are milder but longer (≈2.5 years); (2) supply-side events (wars, natural disasters) raise oil prices and reduce industrial production, whereas demand-side events (financial crises, health emergencies) depress oil prices and can raise measured production relative to demand; (3) financial crises have the most persistent effects (oil market >3 years; macro >4 years), while public health emergencies like COVID-19 depress oil prices for over three years and weigh on activity for roughly 2.5–3 years; (4) crises trigger inflation with international spillovers, especially from US policy responses to China’s CPI. Policy recommendations include establishing crisis monitoring and early-warning systems; diversifying energy imports and accelerating renewable energy to mitigate supply shocks; strengthening financial system governance to mitigate contagion; managing inflation spillovers via exchange rate and trade policies; and promoting firm-level risk management and hedging. Potential future research could refine event quantification beyond futures-based proxies, develop stronger instruments for categories with weak first-stage strength, and extend analysis to additional countries, sectoral dynamics, and distributional impacts.

- Instrument strength varies by event type: robust for overall events, wars/geopolitics, and public health emergencies, but weak for natural disasters and especially financial crises (robust F≈0.52 and ≈0.09, respectively). Results for weak-instrument categories should be interpreted cautiously. - Event classification and selection thresholds (e.g., EM-DAT intensity, systemic peace magnitude) may omit relevant events or introduce selection bias. - China’s industrial production series required constructing a fixed-base index from industrial added value, which may introduce measurement error. - The proxy uses immediate futures price changes around event dates, which captures market reactions but may reflect concurrent news or expectations unrelated to fundamentals. - Monthly frequency may smooth short-lived dynamics; alternative frequencies or model specifications could reveal additional channels. - External validity beyond the US and China, and OECD+6, may be limited; cross-country heterogeneity warrants further study.