Purpose: Understand how supply and demand dynamics drive the 3-2-1 crack spread over time, with concrete metrics and consensus on projection factors.

1. Crack Spread Over Time

From crack_spread_explanation.txt and the underlying data:

Period Crack Spread (USD/bbl) Key Driver
Early 2020 ~8-12 Normal margins; late Feb spike as crude fell faster than products
Mar-Apr 2020 Negative to ~7 COVID crash: gasoline collapsed faster than crude; 2020-03-23 = -1.27
Mid 2020-2021 ~6-13 Recovery as demand returned
2022 Peak ~62 (Apr 28) Russia/Ukraine; supply tight, demand rebounding
2023-2026 ~17-35 Normalisation; elevated vs pre-COVID

Formula: Crack = ((2*RBOB*42 + 1*HO*42) - 3*Brent) / 3 (USD per barrel)

2. Supply vs Demand & Time Lags

Why the Crack Spread Moves

The crack spread is the margin between: - Input cost: 3 barrels of crude (Brent) - Output revenue: 2 barrels of gasoline + 1 barrel of heating oil

It widens when product prices rise faster than crude (or crude falls faster than products).
It narrows when crude rises faster than products (or products fall faster than crude).

Temporal Differences

Factor Typical Lag Effect on Crack Spread
Crude supply shock Days-weeks Crude moves first -> spread widens if products lag
Product demand shock Days-weeks Products move first -> spread widens if crude lags
Refinery outage 1-4 weeks Less product supply -> products rise -> spread widens
Seasonal demand Months Gasoline (summer), heating oil (winter) drive seasonal peaks
Refinery capacity changes Years Closures -> less product supply -> structurally higher spreads
Inventory draw/build Weeks Draw = tighter product market -> spread widens; build = opposite

COVID example: Demand collapsed almost overnight. Gasoline demand fell faster than refineries could cut runs. Product prices collapsed more than crude -> spread went negative.

2022 example: Sanctions and supply disruption hit crude first; product demand stayed strong. Refineries couldn’t instantly increase output -> product prices rose more than crude -> spread spiked to ~62 USD/bbl.

3. Supply & Demand Metrics

Data Sources (EIA, CME, etc.)

Metric Source Unit Interpretation
U.S. refinery utilization EIA Weekly Petroleum Status % of operable capacity High (>90%) = strong supply; low = outages, maintenance
U.S. crude oil inventories EIA Million barrels High = crude oversupply -> pressure on crude price
U.S. gasoline inventories EIA Million barrels Low = product tightness -> support for gasoline, spread
U.S. distillate inventories EIA Million barrels Low = heating oil/diesel tightness -> support for HO, spread
Days of supply EIA (inventory / consumption) Days Low = tighter market; high = oversupply
Product demand (implied) EIA product supplied Mb/d Demand proxy; rising = support for spread
Refinery runs EIA Mb/d Actual throughput; down = less product supply

Example Ranges (Illustrative)

Metric Typical Tight Loose
Refinery utilization 85-92% <80% >95%
Gasoline days of supply 22-28 <20 >30
Distillate days of supply 25-35 <25 >40

Rule of thumb:
- Utilization down + inventories down -> product scarcity -> spread widens
- Utilization up + inventories up -> product surplus -> spread narrows

How to Use These

  1. Weekly: Compare EIA utilization and inventory changes vs prior week.
  2. Seasonal: Compare current levels to 5-year averages for same week.
  3. Direction: Falling product inventories + stable/rising demand -> support for spread widening.

4. Consensus: Determinative Factors

Primary Factors

  1. Refinery utilization / outages
    • Scheduled and unscheduled maintenance, hurricanes, fires.
    • Lower capacity -> less product supply -> higher product prices -> wider spreads.
  2. Product demand vs crude supply
    • When demand for gasoline/diesel/heating oil rises faster than crude supply or refining capacity, product prices rise relative to crude -> spreads widen.
  3. Seasonality
    • Gasoline: summer driving season.
    • Heating oil/diesel: winter heating and harvest.
    • Spreads tend to peak during peak demand.
  4. Inventory levels
    • Low product inventories -> scarcity premium -> wider spreads.
    • High inventories -> discount -> narrower spreads.

Secondary Factors

  1. Geopolitical / crude supply disruption
    • Affects crude price; can widen or narrow spread depending on product response.
  2. Currency (USD strength)
    • Often inverse to crude; can influence both crude and product prices.
  3. Refinery capacity
    • Structural closures (e.g. 2024-2026) reduce capacity -> structurally higher margins.
  4. Specification / blending constraints
    • Summer gasoline specs, octane, blending components can affect product margins.

Consensus Hierarchy

Rank Factor Typical Use for Projection
1 Refinery utilization Near-term: outages -> spread up
2 Product vs crude inventory balance Near-term: product tight -> spread up
3 Seasonal demand Calendar: summer gas, winter HO
4 Product demand (implied) Trend: growth vs supply
5 Geopolitical / crude supply Event-driven: direction depends on product response

Practical Takeaways

  • No single factor dominates.
    Refiners, traders, and analysts use a combination: utilization, inventories, seasonality, and demand.

  • Crack spreads are volatile.
    Events (COVID, Russia/Ukraine) can override fundamentals in the short run.

  • For insurance:

Refinery clients are more exposed when spreads compress (negative or very low).

Monitor utilization, outages, and product demand for early warning.

5. Hormuz & Refinery Scenario

The Setup

Two factors are currently at play:

  1. Possible extended closure of the Strait of Hormuz
    Roughly 20-25% of global oil passes through the Strait. A sustained closure would cut off Gulf crude (Saudi, UAE, Kuwait, Iraq, Iran) from key markets. Alternative routes (e.g. pipelines, longer sea routes) exist but add cost and delay.

  2. Refiners bringing forward CDU (crude distillation unit) overhauls
    Some refiners are scheduling maintenance earlier than planned, taking crude distillation capacity offline in the near term.

Factor Effects on Spread

Factor Effect on crude Effect on products Net effect on crack spread
Hormuz closure Supply cut -> crude spikes Products lag initially (refiners can’t get crude) Spread narrows initially
CDU overhauls No direct effect Less refining capacity -> less product supply Spread widens
Both together Crude up Product supply doubly constrained Depends on timing and magnitude

Combined Scenario: Two-Way Squeeze

When both occur at once:

  • Phase 1 (crude shock dominates): Crude spikes on Hormuz news. Product prices rise but lag. Refiners with capacity may not have crude to run. Crack spread narrows (crude rises faster than products).

  • Phase 2 (refining bottleneck dominates): If alternative crude flows in (pipelines, other producers, strategic releases) but CDU overhauls have taken capacity offline, the constraint shifts to refining. Less capacity to turn crude into products -> product scarcity. Crack spread widens (products rise more than crude).

  • Phase 3 (extended disruption): If Hormuz stays closed and overhauls extend, both crude and product supply stay tight. The spread direction depends on which side is more constrained. Historically, when refining is the bottleneck, product premiums can exceed crude gains -> spread widens.

Why Refiners Overhaul Now

  • Strategic: Take downtime when crude is disrupted anyway; avoid running at a loss if crude is scarce or uneconomic.
  • Operational: Use a period of expected lower runs to complete maintenance.
  • Risk management: Reduce exposure to volatile crude supply and uncertain feedstock availability.

Implications

  • Short term: Volatility is likely. Initial crude spike can compress spreads; CDU maintenance can later support or widen them.
  • Medium term: If Hormuz disruption persists and refining capacity is reduced, product tightness can dominate -> spread widening.
  • Insurance angle: Refinery clients face both input risk (crude availability, cost) and margin risk (spread compression). A widening spread helps refiners; compression or negative spreads increase loss exposure.

What to Monitor

  • Hormuz shipping status and alternative crude routing
  • EIA refinery utilization (impact of brought-forward maintenance)
  • Product vs crude inventory draws
  • CDU outage schedules and duration

Timing & Transmission

Crude Shock to Product Output

The chain from crude supply shock to lower refinery product output has several stages:

Stage Typical lag Notes
Price impact Hours to days Futures and spot prices react immediately to news
Cargoes in transit 0 lag Tankers already en route continue; no new sailings
Inventory draw 0-4 weeks Refiners run from crude stocks; OECD commercial stocks ~28-30 days
Physical shortfall at refinery 2-6 weeks When inventory + in-transit exhausted; depends on location
Alternative crude arrival 4-8+ weeks Gulf to Asia ~30-40 days; Gulf to Europe ~35 days; West Africa/Brazil to Asia >1 month

Implication: A Hormuz closure does not cut refinery product output on day one. Refiners draw down crude inventories and use cargoes already at sea. Product output falls only when those buffers are exhausted. For Asian refiners dependent on Gulf crude, the physical impact on throughput typically arrives 3-6 weeks after a sustained closure. European and US Gulf refiners with more diverse supply may have longer buffers.

Rule of thumb: Crude supply shock -> reduction in refinery product output = ~2-6 weeks, depending on inventory levels and alternative sourcing.

Demand Shock to Rebalancing

When demand collapses, two adjustments must occur: (1) refiners cut runs; (2) crude producers cut output. The system rebalances when both respond.

Actor Typical lag Historic example (COVID 2020)
Refinery run cuts 1-4 weeks US utilization to 82% by late March; ~5 mbpd global cuts within weeks
Product price / spread Days to 2 weeks Crack spread negative late March 2020
Crude producer cuts (shale) 2-8 weeks North America ~10% cut by mid-2020; price-driven
Crude producer cuts (OPEC+) 4-12 weeks Decisions in days; implementation over weeks

COVID timeline (2020): - Early March: Lockdowns begin; demand starts to fall - Mid-late March: Refinery runs cut sharply; crack spread turns negative (gasoline collapsed faster than crude) - April: ~12.5 mbpd global refining capacity idled; crude producers begin voluntary cuts - Q2 2020: Demand down 15-20%; refiners had cut faster than producers -> crude glut, negative WTI

Implication: Refiners can cut runs within 1-4 weeks. Crude producers respond more slowly (2-12 weeks). In a demand shock, refiners typically adjust first; excess crude builds until producers curtail. The crack spread reflects this asymmetry: it can go negative when product demand collapses faster than refiners cut, or when refiners cut but crude supply is slow to follow.

Rule of thumb: Demand shock -> refinery run cuts = 1-4 weeks. Crude curtailment to rebalance = 4-12 weeks (producer-dependent).

Tanker War (1984-88): Event & Lessons

Event summary

The Tanker War was the anti-shipping campaign during the Iran-Iraq War (1980-1988). Iraq and Iran attacked commercial vessels in the Persian Gulf to weaken each other’s oil exports and war financing. It was not a full closure of the Strait of Hormuz but sporadic attacks on ships loading at or transiting near Gulf ports.

Timeline (partial):

Date Event
May 1981 Iraq declares ships to/from Iranian ports subject to attack; begins air strikes
1981-1983 Iran largely refrains from retaliating at sea
Jan 1984 Iraq escalates with Super-Etendard aircraft; Iran responds; “Tanker War” intensifies
Feb 1985 First tanker sinks (Neptunia, Liberian) after Iraqi Exocet hit
Nov 1986 Kuwait appeals for international protection of its tankers
Mar 1987 US reflags Kuwaiti tankers; Operation Earnest Will begins naval escorts
1987 Iran deploys Silkworm anti-ship missiles; attacks become more effective
Aug 1988 Iran-Iraq ceasefire

Impact on oil flows and prices

  • Shipping: Commercial shipping through the Gulf fell by ~25% at the outset. At its peak, the conflict disrupted less than 2% of ships passing through the Strait; flows continued with higher insurance and war-risk premiums.
  • Iranian exports: Cut by roughly half. Kharg Island (Iran’s main terminal) suffered heavy damage from thousands of raids; original capacity ~7 mbpd was severely reduced.
  • Iraqi exports: Already constrained to ~700 kbpd by 1983 due to Syrian pipeline closure and war damage.
  • Crude prices: Spiked on escalation (e.g. WTI +$0.48 to $19.87/bbl in Aug 1987; peak ~$22.75/bbl mid-July 1987 on US escort fears). Iran cut oil prices to offset higher shipping/insurance costs. OPEC overproduction (~3 mbpd above quotas) offset supply concerns. Real oil prices trended down through the 1980s.
  • Market adaptation: Iran discounted crude to keep buyers; Saudi and others raised output. The system adapted rather than seizing up.

Refinery and crack spread implications

  • Data limitation: Traded crack spread benchmarks (e.g. 3:2:1) were not widely used in the 1980s; direct crack spread series for 1984-1988 are scarce.
  • Inferred dynamics: Crude spikes were short-lived; OPEC oversupply and Iran’s discounts limited sustained crude strength. Refineries continued to receive crude (with higher delivered cost from insurance/premiums). Product supply was not physically cut; the shock was mainly price and cost, not volume.
  • Refinery response: No evidence of widespread run cuts from crude shortage. Refineries adjusted to higher feedstock costs and passed some through to products. Margins would have been squeezed when crude spiked faster than products, then restored as crude eased.

Timing lessons for Hormuz scenarios

Aspect Tanker War (1984-1988) Implication for full closure
Price spike Days to weeks on escalation Full closure: hours to days
Physical flow >98% of ships continued Full closure: 100% cut for Gulf-dependent routes
Refinery throughput Largely unchanged (flows continued) Full closure: 2-6 week lag before throughput falls
Market adaptation Iran discounted; OPEC raised output Full closure: alternatives (pipelines, other producers) take weeks
Duration 4+ years of sporadic attacks Full closure: likely shorter if resolved; longer if protracted

Takeaway: The Tanker War is a partial disruption precedent: attacks raised costs and cut some flows, but did not stop them. A full closure would be more severe. Price impact would be faster and larger; physical shortfall at refineries would follow the 2-6 week inventory/transit lag described earlier.

Transmission Speeds Summary

Shock type First price impact Physical impact on refinery output Full rebalancing
Crude supply (e.g. Hormuz) Hours-days 2-6 weeks (inventory + transit) Weeks-months (alternatives)
Demand collapse Days-weeks 1-4 weeks (refinery cuts) 4-12 weeks (crude curtailment)
Refinery outage Days Immediate (capacity offline) 1-4 weeks (maintenance)

6. Monitoring Checklist

Data referenced: crack_spread.csv, crack_spread_explanation.txt, DATA_GUIDE.txt.
EIA sources: eia.gov/outlooks/steo, eia.gov/petroleum/weekly.