Not all oil types are alike in trade substitution

Following Russia’s invasion of Ukraine, the European Union and other countries implemented trade-related sanctions, with particular focus on oil and gas given the EU’s historical reliance on Russian energy. The feasibility and economic costs of such sanctions hinge on how easily lost Russian imports can be replaced. A widely overlooked aspect is that crude oil differs markedly in quality—primarily density (API gravity) and sulfur content—which affects refinery yields, costs, and process compatibility. Light, sweeter crudes tend to yield more premium products and require less desulfurization, while heavier, sour crudes yield more low-value fractions and impose higher upgrading and desulfurization costs. Refineries’ installed equipment and configuration constrain the feasible ‘crude slate.’ This study asks how substitutable different crude oil types truly are for European refineries and how that substitutability affects the welfare and production impacts of sanctions that remove access to specific crude oil types (not merely sources).

The paper builds on literature recognizing crude oil types as imperfect substitutes. Çakır Melek et al. estimate substitution elasticities between light and heavy crude to study effects of the U.S. shale boom and the export ban, with much of the evidence centered on the U.S. market. Prior work has explored price differentials across crude qualities and the economics of crude heterogeneity, but there remains a gap for Europe. The work also connects to the Feenstra (1994) framework and subsequent refinements by Broda and Weinstein for estimating elasticities and exact price indices accounting for variety changes, commonly used to gauge gains from variety. Traditional Armington-type approaches identify varieties by country of origin; this paper departs by defining varieties by chemical composition (API gravity and sulfur), aligning more closely with refinery technology and costs. Related contributions discuss welfare gains from increased variety, the ‘gains from variety,’ and the measurement of aggregate price indices with taste shocks (Redding and Weinstein).

Data: The study uses the European Commission’s Crude Oil Import Register (COIR), which reports EU member states’ crude oil imports by field of origin with quality attributes (API gravity, sulfur). Monthly data for 2013–2019 are aggregated to annual frequency. Prices are CIF, embedding transport and geographic factors. Missing quality data are supplemented (e.g., McKinsey Energy Insight, EIA). For Russian ‘Other Crude Oil,’ the authors assume a 70% Urals (medium-sour) and 30% Siberian light sweet split based on production shares.

Hedonic pricing: A hedonic model regresses CIF import prices on quality characteristics and trade frictions (time, importer, exporter fixed effects; volume; distance; pipeline connection). API gravity and sulfur enter (with API squared) to capture nonlinear price–quality relationships.

Elasticity estimation: The paper applies Feenstra’s GMM-based approach (as refined by Broda–Weinstein) to estimate micro-elasticities of substitution. Unlike standard Armington where varieties are by country, the authors define varieties at three aggregation levels: (i) field of origin, (ii) country of origin, and (iii) oil type based on chemical composition (e.g., light sweet, medium sour, etc.). This permits identification of within-type vs. between-type substitutability. The Feenstra framework leverages heteroskedasticity across varieties and independence of demand and supply shocks to identify σ, correcting for heteroskedasticity in panel estimations of first-differenced import shares and prices.

Nested CES production: To approximate production disruptions under sanctions, the authors use a nested CES production function for refinery input aggregation. Within-type elasticities capture substitution across varieties in the same oil type; a lower between-type elasticity (γ) captures limited substitutability across oil types. They compare outcomes under a standard single-σ CES versus the nested CES where γ < α (within-type elasticity), simulating scenarios: removal of medium-sour imports; removal of all Russian imports; and observed changes in crude baskets 2021 vs. 2023.

Exact price index: They compute exact CES import price indices accounting for entry/exit of varieties following Feenstra. For the nested case, they compute within-type indices and aggregate across types allowing for changing expenditure shares. A robustness version holds oil type-specific tastes fixed at pre-COVID levels (2013–2019) to reflect sanctions as exogenous deviations from optimal crude slates. Elasticities used are the Euro Area estimates from Table 2 and within-type elasticities from supplementary material.

• Hedonic pricing (Table 1): API gravity positively and significantly affects price (coef 1.051), with a negative squared term (-0.014), indicating diminishing marginal price premia for higher API. Sulfur content reduces price (-0.952). Pipeline connection is associated with a lower CIF price (-1.764). R^2 = 0.982. Example: Brent (38° API, 0.4% S) vs. Basrah Heavy (23.7° API, 4.12% S) implies an average discount of about $6.22/bbl for Basrah Heavy in-sample.
• Elasticities of substitution (Table 2): Elasticities decline with aggregation that recognizes quality. For the Euro Area: by oil field, σ ≈ 36.0; by country of origin, σ ≈ 36.2; by oil type (quality-based), σ ≈ 4.19. Country-specific oil-type elasticities are notably low: Austria 2.77, Belgium 9.64, France 3.94, Germany 5.83, Italy 6.13. This indicates much lower substitutability between oil types than suggested by country-based estimates (and lower than U.S. estimates such as 17.1 in Broda–Weinstein for U.S. crude imports).
• Production disruptions under sanctions (Table 3): Accounting for nested CES (γ < α) implies larger production losses relative to a standard CES. Differential production loss (nested minus standard CES) in percentage points: Euro Area: removing medium-sour imports −4.74 pp; removing Russia −1.56 pp; observed 2021→2023 basket change −0.61 pp. Germany shows up to −7.89 pp in the observed 2021→2023 comparison.
• Replacement patterns and quality mismatch: EU could not compensate lost medium-sour volumes post-sanctions (Fig. 4), with overall medium-sour imports falling. Replacement shifted toward U.S. light sweet grades with higher API and lower sulfur, creating a mismatch with European diesel-oriented product demand and refinery configurations.
• Import price indices (Table 4): 2021→2022: CES exact index +43.2%; nested CES +42.8% (within-type +47.5%, between-type −3.19%); observed import price +42.9%. 2022→2023: CES −16.8%; nested CES −6.12% (within −12.6%, between +7.40%); observed −13.6%. With fixed tastes: nested CES shows +47.8% (2021→2022) and −5.64% (2022→2023), indicating muted price declines post-sanctions when accounting for constrained crude slates.
• Additional price differentials: Hedonic fitted values imply U.S. vs. Russian crude quality differences yield about a $3.23/bbl price premium for U.S. grades. Diesel vs. gasoline prices: diesel rose relative to gasoline by 8.22% from 2022 to 2023 on average across EU members, consistent with higher costs and diesel yield shortfalls when replacing medium-sour with lighter crude.

Recognizing crude as a heterogeneous input fundamentally changes the assessment of sanctions and supply shocks. The estimated between-type elasticity is much lower than within-type substitution, implying that losing access to an entire oil type (e.g., medium-sour Urals) disrupts refinery production more than suggested by Armington-style or country-based CES models. The nested CES framework captures larger production losses and a smaller decline in exact import prices post-2022 compared to a homogeneous CES, reflecting the constrained ability to substitute across types. Empirical evidence shows that EU refineries shifted toward U.S. light sweet crude, which, despite higher quality in abstract terms, yields less diesel and requires adjustments to align with European product demand, raising input costs and altering product yields. This mismatch propagated to consumers via higher relative diesel prices, consistent with pass-through from higher refinery production costs. The analysis corroborates that ignoring crude quality understates both disruption risks and price effects when imports from key suppliers are curtailed.

The study provides European evidence that the elasticity of substitution across crude oil types—defined by API gravity and sulfur content—is substantially lower than traditional country-based measures. Accounting for this quality heterogeneity through a nested CES model reveals that sanctions or supply losses that eliminate access to specific oil types cause larger production disruptions and attenuate price declines relative to homogeneous CES assumptions. Policy evaluations of energy sanctions and security must incorporate crude quality to credibly assess refinery operations, welfare impacts, and consumer prices. The framework and estimates can inform strategic stockpiling, refinery investment, and diversification strategies. Future work could integrate refinery-specific configurations at plant level, explore dynamic adjustment and reconfiguration costs, and extend the approach to other heterogeneous commodities where input quality shapes production technology and substitution possibilities.

• Data coverage and timing: Core COIR panel spans 2013–2019; post-2021 analyses rely on observed changes and indices rather than fully integrated panel estimations with identical metadata.
• Measurement and classification: Some imports are labeled as ‘Other Crude Oil,’ notably for Russia; the 70/30 Urals/Siberian Light assignment, while grounded in production shares, introduces classification uncertainty.
• Price measure: CIF prices embed transport costs and geography/infrastructure (e.g., pipelines), conflating pure quality premia with logistics; pipeline connection status and distance are controlled but may not fully capture time-varying transport frictions.
• Aggregation of types: Grouping into five oil types simplifies a continuum of qualities; within-type homogeneity is an approximation, though supported by high within-type elasticities.
• Model structure: Nested CES imposes constant elasticities within and across types and abstracts from plant-level constraints, dynamic reconfiguration costs, and capacity bottlenecks.
• External validity: Estimates are EU-focused and may not generalize to markets with different refinery configurations, product demand mixes, or infrastructure.