## Refinining margins in practice: turning “crack spreads” into actual earnings (with filing-based examples) Refining margins behave like an “instant P&L,” but filings show downstream earnings are really a function of **(1) benchmark cracks, (2) crude differentials, (3) throughput/utilization, and (4) costs/turnarounds/outages**—and each can swing results materially quarter-to-quarter. ["Refining segment results are driven"](0001534701-25-000074|0) ["Refining & Marketing segment adjusted EBITDA depends"](0001510295-25-000054|0) --- ## 1) Crack spreads are the headline—**realized margin capture** is the income statement ### Benchmark vs “capture” Marathon Petroleum explicitly uses regional crack spreads as the benchmark starting point, then explains realized performance can differ due to crude mix/costs, yields, RINs, and other factors. ["calculate Gulf Coast, Mid-Continent"](0001510295-25-000012|0) ["differs from market indicators"](0001510295-25-000054|0) PBF formalizes that each system maps to a different benchmark proxy (2-1-1 / 3-2-1 / 4-3-1), which is why one generic “3-2-1” statistic can mislead across regions and configurations. ["2-1-1 crack spread"](0001534504-24-000011|0) ["4-3-1 crack spread"](0001534504-24-000011|0) ### Filing-based “margin bridge” example (how product demand changes flow into dollars) Valero quantifies the drivers of its refining margin change—exactly the mechanism your blog describes (“gasoline vs diesel vs crude differentials” feeding through to earnings). In Q1 2025 (ended March 31, 2025), Valero’s refining margin fell by **$1.044 billion** vs Q1 2024, including roughly **-$680 million from distillate**, **-$300 million from gasoline**, and **-$300 million from crude differentials** (partly offset by +$180 million from other products). ["Refining margin (see note (b))"](0001035002-25-000017|0) ["decrease in distillate"](0001035002-25-000017|0) ["decrease in gasoline"](0001035002-25-000017|0) ["decline in crude oil differentials"](0001035002-25-000017|0) --- ## 2) Utilization and outages: the *volume lever* that turns pricing into earnings ### Throughput multiplies whatever margin environment exists Valero discloses throughput volumes alongside refining margin, which allows a useful per-barrel view (helps compare to crack spread proxies). ["Throughput volumes (thousand barrels per day)"](0001035002-25-000017|0) ["Refining margin is defined as"](0001035002-25-000005|0) **Valero — Refining margin per barrel (computed from filings; USD/bbl)** | Period | Refining margin | Throughput | Margin per bbl | |---|---:|---:|---:| | Q1 2025 | $2,490 mm ["Refining margin"](0001035002-25-000017|0) | 2,828 kbpd ["Throughput volumes"](0001035002-25-000017|0) | $9.78 (computed) | | Q1 2024 | $3,534 mm ["Refining margin"](0001035002-24-000022|0) | 2,760 kbpd ["Throughput volumes"](0001035002-24-000022|0) | $14.07 (computed) | | FY 2024 | $11,325 mm ["Refining margin"](0001035002-25-000005|0) | 2,912 kbpd ["Throughput volumes"](0001035002-25-000005|0) | $10.63 (computed) | | FY 2023 | $19,087 mm ["Refining margin"](0001035002-24-000007|0) | 2,979 kbpd ["Throughput volumes"](0001035002-24-000007|0) | $17.56 (computed) | *(Computed method: margin per bbl ≈ refining margin (USD) ÷ (throughput bpd × days in period).)* ### Planned turnarounds are real “P&L line items,” not just lost volume Marathon isolates **planned turnaround costs** outside segment adjusted EBITDA. In FY2024 it reported **$1.404 billion** of “Refining & Renewable Diesel planned turnaround costs” (vs $1.201 billion in FY2023). ["Refining & Renewable Diesel planned turnaround costs"](0001510295-25-000012|0) It also provides a per-barrel view: **$0.90/bbl** in Q2 2025 vs **$0.66/bbl** in Q2 2024 for refining planned turnaround costs. ["Refining planned turnaround costs per barrel"](0001510295-25-000054|0) ### Unplanned outages: filings emphasize lost margin opportunity + forced purchases PBF states that **unplanned downtime results in “lost margin opportunity” and higher maintenance expense**, while planned downtime is managed against the margin environment and other constraints. ["Unplanned downtime"](0001534504-24-000011|0) Delek notes that when it has outages, it may need to **acquire refined products** to meet needs—turning downtime into a commercial/marketing issue as well as an operational one. ["when we have outages"](0001694426-24-000154|0) PBF’s 10-Q provides a concrete outage disclosure example: the February 1, 2025 Martinez refinery fire, restart timing, and **$250 million** of insurance proceeds received in Q2 2025. ["fire on February 1, 2025"](0001534504-25-000047|0) ["received an unallocated"](0001534504-25-000047|0) --- ## 3) Demand deep-dive by product: where the biggest earnings sensitivity shows up ### Diesel/distillate often dominates the cycle Valero’s FY2024 vs FY2023 refining margin change includes roughly **$4.3 billion** unfavorable from distillate margins (vs **~$2.4 billion** from gasoline), consistent with your point that distillate is highly cyclical and freight/industrial-linked. ["decrease in distillate"](0001035002-25-000005|0) ["decrease in gasoline"](0001035002-25-000005|0) ### Jet fuel: frequently embedded in “middle distillate” competition PBF’s benchmark definitions explicitly include jet in some regional proxies (e.g., 4-3-1 includes jet; its Martinez 3-2-1 includes a jet component), reinforcing that jet tightness can be “felt” inside composite crack spreads. ["4-3-1 crack spread"](0001534504-24-000011|0) ["3-2-1 crack spread"](0001534504-24-000011|0) --- ## 4) The feedback loop: product markets reshape crude differentials and inventory value ### Crude differentials are a separate lever from cracks Valero attributes **~$585 million** of FY2024 vs FY2023 refining margin headwind to a “decline in crude oil differentials,” *in addition* to product margin changes. ["decline in crude oil differentials"](0001035002-25-000005|0) Marathon similarly separates drivers and quantifies that narrower crude differentials and lower crack spreads had a net negative impact of approximately **$500 million** on Refining & Marketing margin in Q2 2025 vs Q2 2024 (then notes mix/yields/RINs can offset). ["net negative impact of approximately $500 million"](0001510295-25-000054|0) Delek is explicit about its WTI-linked positioning: if Midland/Cushing discounts narrow or invert relative to Brent, it can compress their competitive advantage; when discounts widen, their advantage increases. ["WTI Cushing compared to Brent"](0001694426-24-000109|0) ["discounts widen, so does our competitive advantage"](0001694426-24-000109|0) ### Inventory accounting can amplify volatility Marathon flags that inventories are carried at lower-of-cost-or-market and that valuation adjustments can negatively affect earnings if refined product prices fall. ["Inventories are stated at the lower"](0001510295-25-000012|0) --- ## 5) A filings-backed scenario framework (mapping macro to earnings) ### 1) Tight operations + steady demand The “outages stack up” regime tends to show up as (a) lower utilization/throughput and higher turnaround intensity, but (b) potentially resilient realized margins per barrel if the system tightens. MPC’s per-barrel disclosure is a practical way to see whether margin-per-barrel is holding up even when turnaround costs rise. ["Refining & Marketing margin per barrel"](0001510295-25-000054|0) ["Refining planned turnaround costs per barrel"](0001510295-25-000054|0) ### 2) High runs + soft gasoline This regime usually appears as declining benchmark cracks and weaker realized margins, often before the system cuts runs. Phillips 66 provides a clear macro anchor: the **composite 3:2:1 crack spread averaged $16.95/bbl in 2024 vs $28.37/bbl in 2023**, consistent with a materially softer margin environment. ["average of $16.95 per barrel"](0001534701-25-000074|0) ["average of $28.37 per barrel"](0001534701-25-000074|0) ### 3) Industrial slowdown then rebound (distillate leads) Filings often show distillate as the largest swing factor when the cycle turns. Valero’s FY2024 bridge is a template for identifying whether distillate is the dominant headwind/tailwind in realized refining margin. ["decrease in distillate"](0001035002-25-000005|0) --- ## 6) “What to monitor weekly” — aligned to what companies quantify in filings - **Crack spreads / margin proxies:** PSX discloses composite crack spreads; MPC discloses margin-per-barrel frameworks by region. ["composite 3:2:1 market crack spread"](0001534701-25-000074|0) ["calculate Gulf Coast, Mid-Continent"](0001510295-25-000012|0) - **Utilization / throughput:** PSX discloses utilization by region; Valero discloses throughput volumes next to refining margin. ["Capacity utilization (percent)"](0001534701-24-000166|0) ["Throughput volumes"](0001035002-25-000017|0) - **Downtime & turnaround intensity:** MPC discloses planned turnaround costs (total and per barrel); PBF discusses planned vs unplanned downtime economics. ["planned turnaround costs"](0001510295-25-000012|0) ["Unplanned downtime"](0001534504-24-000011|0) - **Regulatory costs (RINs):** Delek highlights RIN price volatility as a material swing factor. ["RINs prices can increase"](0001694426-24-000154|0) - **Policy risk (e.g., California margin caps / maintenance restrictions):** Refiners disclose potential impacts from rules that could constrain maintenance scheduling or cap margins. ["maximum gross gasoline refining margin"](0001510295-25-000012|0) ["authorized the establishment"](0001035002-25-000005|0) --- ## Bottom line (filings-supported) The filings support your blog’s core claim: refining is “product-driven” because **diesel/gasoline/jet pricing and product inventory tightness set the benchmark opportunity**, but **utilization, turnaround/outage timing, crude differential capture, and regulatory costs determine how much of that opportunity converts into realized margin and segment earnings**. ["Refining segment results are driven"](0001534701-25-000074|0) ["Unplanned downtime"](0001534504-24-000011|0) ["decline in crude oil differentials"](0001035002-25-000005|0)

## Refinining margins in practice: turning “crack spreads” into actual earnings (with filing-based examples) Refining margins behave like an “instant P&L,” but filings show downstream earnings are really a function of **(1) benchmark cracks, (2) crude differentials, (3) throughput/utilization, and (4) costs/turnarounds/outages**—and each can swing results materially quarter-to-quarter. ["Refining segment results are driven"](0001534701-25-000074|0) ["Refining & Marketing segment adjusted EBITDA depends"](0001510295-25-000054|0) --- ## 1) Crack spreads are the headline—**realized margin capture** is the income statement ### Benchmark vs “capture” Marathon Petroleum explicitly uses regional crack spreads as the benchmark starting point, then explains realized performance can differ due to crude mix/costs, yields, RINs, and other factors. ["calculate Gulf Coast, Mid-Continent"](0001510295-25-000012|0) ["differs from market indicators"](0001510295-25-000054|0) PBF formalizes that each system maps to a different benchmark proxy (2-1-1 / 3-2-1 / 4-3-1), which is why one generic “3-2-1” statistic can mislead across regions and configurations. ["2-1-1 crack spread"](0001534504-24-000011|0) ["4-3-1 crack spread"](0001534504-24-000011|0) ### Filing-based “margin bridge” example (how product demand changes flow into dollars) Valero quantifies the drivers of its refining margin change—exactly the mechanism your blog describes (“gasoline vs diesel vs crude differentials” feeding through to earnings). In Q1 2025 (ended March 31, 2025), Valero’s refining margin fell by **$1.044 billion** vs Q1 2024, including roughly **-$680 million from distillate**, **-$300 million from gasoline**, and **-$300 million from crude differentials** (partly offset by +$180 million from other products). ["Refining margin (see note (b))"](0001035002-25-000017|0) ["decrease in distillate"](0001035002-25-000017|0) ["decrease in gasoline"](0001035002-25-000017|0) ["decline in crude oil differentials"](0001035002-25-000017|0) --- ## 2) Utilization and outages: the *volume lever* that turns pricing into earnings ### Throughput multiplies whatever margin environment exists Valero discloses throughput volumes alongside refining margin, which allows a useful per-barrel view (helps compare to crack spread proxies). ["Throughput volumes (thousand barrels per day)"](0001035002-25-000017|0) ["Refining margin is defined as"](0001035002-25-000005|0) **Valero — Refining margin per barrel (computed from filings; USD/bbl)** | Period | Refining margin | Throughput | Margin per bbl | |---|---:|---:|---:| | Q1 2025 | $2,490 mm ["Refining margin"](0001035002-25-000017|0) | 2,828 kbpd ["Throughput volumes"](0001035002-25-000017|0) | $9.78 (computed) | | Q1 2024 | $3,534 mm ["Refining margin"](0001035002-24-000022|0) | 2,760 kbpd ["Throughput volumes"](0001035002-24-000022|0) | $14.07 (computed) | | FY 2024 | $11,325 mm ["Refining margin"](0001035002-25-000005|0) | 2,912 kbpd ["Throughput volumes"](0001035002-25-000005|0) | $10.63 (computed) | | FY 2023 | $19,087 mm ["Refining margin"](0001035002-24-000007|0) | 2,979 kbpd ["Throughput volumes"](0001035002-24-000007|0) | $17.56 (computed) | *(Computed method: margin per bbl ≈ refining margin (USD) ÷ (throughput bpd × days in period).)* ### Planned turnarounds are real “P&L line items,” not just lost volume Marathon isolates **planned turnaround costs** outside segment adjusted EBITDA. In FY2024 it reported **$1.404 billion** of “Refining & Renewable Diesel planned turnaround costs” (vs $1.201 billion in FY2023). ["Refining & Renewable Diesel planned turnaround costs"](0001510295-25-000012|0) It also provides a per-barrel view: **$0.90/bbl** in Q2 2025 vs **$0.66/bbl** in Q2 2024 for refining planned turnaround costs. ["Refining planned turnaround costs per barrel"](0001510295-25-000054|0) ### Unplanned outages: filings emphasize lost margin opportunity + forced purchases PBF states that **unplanned downtime results in “lost margin opportunity” and higher maintenance expense**, while planned downtime is managed against the margin environment and other constraints. ["Unplanned downtime"](0001534504-24-000011|0) Delek notes that when it has outages, it may need to **acquire refined products** to meet needs—turning downtime into a commercial/marketing issue as well as an operational one. ["when we have outages"](0001694426-24-000154|0) PBF’s 10-Q provides a concrete outage disclosure example: the February 1, 2025 Martinez refinery fire, restart timing, and **$250 million** of insurance proceeds received in Q2 2025. ["fire on February 1, 2025"](0001534504-25-000047|0) ["received an unallocated"](0001534504-25-000047|0) --- ## 3) Demand deep-dive by product: where the biggest earnings sensitivity shows up ### Diesel/distillate often dominates the cycle Valero’s FY2024 vs FY2023 refining margin change includes roughly **$4.3 billion** unfavorable from distillate margins (vs **~$2.4 billion** from gasoline), consistent with your point that distillate is highly cyclical and freight/industrial-linked. ["decrease in distillate"](0001035002-25-000005|0) ["decrease in gasoline"](0001035002-25-000005|0) ### Jet fuel: frequently embedded in “middle distillate” competition PBF’s benchmark definitions explicitly include jet in some regional proxies (e.g., 4-3-1 includes jet; its Martinez 3-2-1 includes a jet component), reinforcing that jet tightness can be “felt” inside composite crack spreads. ["4-3-1 crack spread"](0001534504-24-000011|0) ["3-2-1 crack spread"](0001534504-24-000011|0) --- ## 4) The feedback loop: product markets reshape crude differentials and inventory value ### Crude differentials are a separate lever from cracks Valero attributes **~$585 million** of FY2024 vs FY2023 refining margin headwind to a “decline in crude oil differentials,” *in addition* to product margin changes. ["decline in crude oil differentials"](0001035002-25-000005|0) Marathon similarly separates drivers and quantifies that narrower crude differentials and lower crack spreads had a net negative impact of approximately **$500 million** on Refining & Marketing margin in Q2 2025 vs Q2 2024 (then notes mix/yields/RINs can offset). ["net negative impact of approximately $500 million"](0001510295-25-000054|0) Delek is explicit about its WTI-linked positioning: if Midland/Cushing discounts narrow or invert relative to Brent, it can compress their competitive advantage; when discounts widen, their advantage increases. ["WTI Cushing compared to Brent"](0001694426-24-000109|0) ["discounts widen, so does our competitive advantage"](0001694426-24-000109|0) ### Inventory accounting can amplify volatility Marathon flags that inventories are carried at lower-of-cost-or-market and that valuation adjustments can negatively affect earnings if refined product prices fall. ["Inventories are stated at the lower"](0001510295-25-000012|0) --- ## 5) A filings-backed scenario framework (mapping macro to earnings) ### 1) Tight operations + steady demand The “outages stack up” regime tends to show up as (a) lower utilization/throughput and higher turnaround intensity, but (b) potentially resilient realized margins per barrel if the system tightens. MPC’s per-barrel disclosure is a practical way to see whether margin-per-barrel is holding up even when turnaround costs rise. ["Refining & Marketing margin per barrel"](0001510295-25-000054|0) ["Refining planned turnaround costs per barrel"](0001510295-25-000054|0) ### 2) High runs + soft gasoline This regime usually appears as declining benchmark cracks and weaker realized margins, often before the system cuts runs. Phillips 66 provides a clear macro anchor: the **composite 3:2:1 crack spread averaged $16.95/bbl in 2024 vs $28.37/bbl in 2023**, consistent with a materially softer margin environment. ["average of $16.95 per barrel"](0001534701-25-000074|0) ["average of $28.37 per barrel"](0001534701-25-000074|0) ### 3) Industrial slowdown then rebound (distillate leads) Filings often show distillate as the largest swing factor when the cycle turns. Valero’s FY2024 bridge is a template for identifying whether distillate is the dominant headwind/tailwind in realized refining margin. ["decrease in distillate"](0001035002-25-000005|0) --- ## 6) “What to monitor weekly” — aligned to what companies quantify in filings - **Crack spreads / margin proxies:** PSX discloses composite crack spreads; MPC discloses margin-per-barrel frameworks by region. ["composite 3:2:1 market crack spread"](0001534701-25-000074|0) ["calculate Gulf Coast, Mid-Continent"](0001510295-25-000012|0) - **Utilization / throughput:** PSX discloses utilization by region; Valero discloses throughput volumes next to refining margin. ["Capacity utilization (percent)"](0001534701-24-000166|0) ["Throughput volumes"](0001035002-25-000017|0) - **Downtime & turnaround intensity:** MPC discloses planned turnaround costs (total and per barrel); PBF discusses planned vs unplanned downtime economics. ["planned turnaround costs"](0001510295-25-000012|0) ["Unplanned downtime"](0001534504-24-000011|0) - **Regulatory costs (RINs):** Delek highlights RIN price volatility as a material swing factor. ["RINs prices can increase"](0001694426-24-000154|0) - **Policy risk (e.g., California margin caps / maintenance restrictions):** Refiners disclose potential impacts from rules that could constrain maintenance scheduling or cap margins. ["maximum gross gasoline refining margin"](0001510295-25-000012|0) ["authorized the establishment"](0001035002-25-000005|0) --- ## Bottom line (filings-supported) The filings support your blog’s core claim: refining is “product-driven” because **diesel/gasoline/jet pricing and product inventory tightness set the benchmark opportunity**, but **utilization, turnaround/outage timing, crude differential capture, and regulatory costs determine how much of that opportunity converts into realized margin and segment earnings**. ["Refining segment results are driven"](0001534701-25-000074|0) ["Unplanned downtime"](0001534504-24-000011|0) ["decline in crude oil differentials"](0001035002-25-000005|0)