Every driver has felt it. Crude oil jumps on a Monday headline, and the price on the station sign seems to leap the same afternoon. Crude falls the next week, and the sign... barely moves. Then it drifts down a penny at a time, as if gravity had been switched off. Economists have a name for this: rockets and feathers - prices shoot up like rockets and float down like feathers.
Is it real, or is it just selective memory - the psychological trick where price hikes sting and stick in the mind while slow declines go unnoticed? The honest answer is both. Selective memory is real, but the asymmetry is real too: it shows up in the data, it has been measured in peer-reviewed studies for decades, and we can measure it ourselves with the market data this site publishes every day. This guide does exactly that. Using 91 weeks of weekly WTI crude and wholesale gasoline prices - the site's own EIA data export, refreshed October 1, 2026 - we test how much of a crude price move actually reaches gasoline, and whether rises really do pass through faster than falls. Then we explain the machinery behind the asymmetry: the four links in the chain from a barrel of crude to a gallon at the pump, and why each link adds its own delay.
1. What "rockets and feathers" actually means
The phrase describes asymmetric pass-through: the price of gasoline responds more quickly, or more completely, to an increase in the price of crude oil than to a decrease of the same size. Put precisely, if WTI rises 10% and wholesale gasoline rises 8%, but a 10% fall in WTI only pulls gasoline down 6%, that two-percentage-point gap is the feather. Over a year of volatile crude prices, the gap compounds into real money for drivers - and real margin for everyone between the wellhead and the pump.
Three things the phrase does not mean are worth clearing up first. It does not mean every station is gouging - as we will see, several of the mechanisms are ordinary inventory and cost economics. It does not mean the asymmetry is the same everywhere: it is stronger at the retail level in markets with little station competition and weaker where stations fight over every corner. And it does not mean prices never fall - they do, just more slowly and less completely than they rise.
2. The four links in the chain
A barrel of crude does not become a gallon of pump gasoline in one step. There are four links, each with its own pricing logic, and the asymmetry can - and does - arise at every one of them. Understanding the chain is the key to reading pump prices honestly.
| Link | What is priced | Where asymmetry enters |
|---|---|---|
| 1. Crude production | WTI / Brent benchmarks | Geopolitical risk premiums spike instantly on bad news but fade slowly; producers hedge forward, smoothing their own realized prices |
| 2. Refining | RBOB futures, NY Harbor spot gasoline, the crack spread | Refiners widen margins when crude rises (passing cost plus extra); margin compression on the way down is stickier because of maintenance schedules and inventory costs |
| 3. Wholesale distribution | Rack prices at terminals | Distributors hold inventory bought at higher prices and are slow to mark it down; branded vs. unbranded rack differentials move with volume incentives |
| 4. Retail stations | The sign price | Stations price off replacement cost when it rises (fear of selling below tomorrow's cost) but off sunk inventory cost when it falls; local competition sets the speed |
The rest of this guide tests link 2 - crude to wholesale gasoline - with real data, because that is where this site's data is strongest. Then we walk through what happens at links 3 and 4, where the data gets thinner but the economics get more interesting.
3. What the research literature found
The landmark study is Borenstein, Cameron, and Gilbert (1997), published in the Quarterly Journal of Economics, which tested US retail gasoline prices against crude oil price changes and found the response was indeed asymmetric: retail prices adjusted more quickly to crude cost increases than to decreases. The paper is the origin of the "rockets and feathers" framing in economics, and its core finding - that the asymmetry is real and measurable - has been replicated in follow-up work across several countries and decades.
The literature has debated why ever since. The leading explanations fall into three buckets. First, consumer search behavior: drivers search hard for cheap gas when prices are rising (punishing stations that raise prices fastest) but stop searching when prices are falling, which lets stations keep margins wider for longer on the way down. Second, inventory and menu-cost economics: gasoline in the ground at a station was bought at yesterday's wholesale price, and repricing signs, retraining, and repositioning all cost something - costs that make small downward adjustments not worth doing immediately. Third, market power: where a few stations dominate a local market, the incentive to pass savings to drivers quickly is weak. None of these requires a conspiracy; all of them are ordinary profit-seeking in a market with frictions. Keep those three buckets in mind - we will return to them with data.
4. Our test: the crude-to-wholesale link, measured
Here is the honest version of the test. We used this site's own EIA weekly price export (refreshed October 1, 2026): the WTI Cushing spot price and the NY Harbor conventional gasoline spot price. We took every week from January 3, 2025 through September 25, 2026- 91 weekly observations with continuous weekly coverage - and split the weeks by the direction of the crude move. A week counts as an "up-week" if WTI rose more than 1%, and a "down-week" if it fell more than 1%. Then we measured how much wholesale gasoline moved in the same week. The ratio - gasoline's move divided by crude's move - is the pass-through.
| Week type | Weeks (n) | Mean WTI move | Mean gasoline move | Pass-through |
|---|---|---|---|---|
| WTI up more than 1% | 38 | +5.25% | +4.15% | 0.79 |
| WTI down more than 1% | 36 | -4.43% | -2.78% | 0.63 |
The medians tell the same story, which matters because means can be dragged by a few wild weeks: the median up-week saw WTI rise 3.67% with gasoline up 3.07% (pass-through 0.84), while the median down-week saw WTI fall 3.31% with gasoline down only 2.42% (pass-through 0.73). Same-week correlation between the two series is 0.75; the following week's correlation collapses to 0.13, which means at weekly frequency the adjustment happens essentially contemporaneously - there is no hidden "catch-up" the next week that would erase the gap. The feather is not a timing artifact. It is a completeness gap: of every dollar of crude decline, wholesale gasoline gives back about 63 cents in the same week, versus 79 cents of every dollar of crude increase.
5. Cross-check: the daily futures market agrees
Weekly data could hide something, so we ran the same test on a second, independent dataset: this site's daily futures history for WTI crude and RBOB gasoline - 551 matched trading sessions from November 28, 2023 through October 1, 2026. On days WTI rose more than 0.3% (258 days), the average crude gain was 2.27% and RBOB rose 1.67% - a same-day pass-through of 0.74. On days WTI fell more than 0.3% (227 days), the average crude loss was 2.38% and RBOB fell 1.61% - pass-through of 0.68. Same direction, same signature, smaller magnitude: the asymmetry is milder in the hyper-competitive futures market than in the physical weekly market, which is exactly what you would expect if part of the feather comes from physical frictions - inventory, logistics, and local market power - rather than from trading itself.
Two datasets, two frequencies, one conclusion: at the crude-to-wholesale link, gasoline prices absorb roughly three-quarters of a crude increase but only about two-thirds of a crude decrease in the same period. The gap - around 10 percentage points of pass-through - is the measured feather at the wholesale level.
6. Why the asymmetry happens: the six drivers
Measurement tells you the feather exists; economics tells you why. These are the six mechanisms the literature and market structure point to, ordered roughly from the wellhead to the pump.
6.1 Replacement-cost pricing on the way up, sunk-cost pricing on the way down
This is the single most important mechanism at the retail link. A station owner holding 10,000 gallons bought at $3.00 watches wholesale rise to $3.20. If she keeps the sign at the old price, she sells today's inventory at a profit - but she will refill the tanks tomorrow at $3.20, and every gallon sold cheap today is a gallon she must replace at the higher price. Rational station owners therefore price off replacement cost when costs are rising. But when wholesale falls to $2.80, the logic flips: the fuel in the ground cost $3.00, and cutting the sign price immediately means selling at a loss against sunk inventory. Owners wait until the cheaper fuel is actually in their tanks. The result is mechanical: prices track replacement cost upward and inventory cost downward, and inventory turns over in days, not minutes.
6.2 Refinery margin management
Refiners buy crude and sell gasoline; the difference - the crack spread- is their gross margin. When crude spikes, refiners can widen the crack by raising gasoline prices faster than their crude costs rise, padding margins against the risk that crude keeps climbing. When crude falls, they are in no hurry to narrow a fat margin: maintenance schedules, turnaround seasons, and the simple fact that nobody forces a refiner to cut prices all slow the adjustment. Per this site's market data, the gasoline crack spread recently printed $48.33 per barrel - an unusually wide margin that illustrates how much of the pump price is refiner margin rather than crude cost when markets are tight. With US refinery utilization at 92.5% in the latest EIA week, there is little spare capacity to compete those margins away quickly.
6.3 Consumer search asymmetry
Drivers are not equally vigilant in both directions. When prices are climbing, every fill-up hurts, apps get opened, and drivers will cross the street to save three cents - which disciplines stations and forces fast, competitive repricing. When prices are falling, the pain fades, the apps stay closed, and drivers buy at the first station they see. Borenstein's search-cost explanation predicts exactly this: the competitive pressure that forces prices up quickly evaporates on the way down, letting stations hold fatter margins longer. The feather, in this telling, is partly the consumer's own inattention.
6.4 Local market power and edge zones
Gasoline retailing is hyper-local: most drivers will not drive more than a few minutes out of their way. A station at a highway exit with no competitor in sight faces very different incentives than one of four stations on the same intersection. In low- competition "edge zones," the station can let the feather linger for weeks; at contested intersections, a rival's price cut forces a response within hours. This is why the asymmetry varies so much by neighborhood - and why the city-level retail data this site tracks shows such wide spreads between cities (see section 7).
6.5 Menu costs and price stickiness
Changing a price is not free: signs must be updated, point-of-sale systems reprogrammed, and in many jurisdictions prices can only move at certain times. These small frictions - economists call them menu costs - mean stations batch small cost changes rather than tracking them tick by tick. Batching is symmetric in principle, but it interacts with the other five mechanisms: upward batches happen promptly (replacement-cost fear), downward batches wait for the next convenient moment.
6.6 Asymmetric demand response
Finally, demand itself is asymmetric. When gasoline prices spike, some drivers genuinely cut trips, carpool, or postpone travel - demand softens, which eventually pressures prices back down. When prices fall, driving does not surge symmetrically: people do not take extra road trips because gas got ten cents cheaper. This means upward price moves trigger a demand response that eventually corrects them, while downward moves meet no such corrective force - the decline just coasts.
7. The retail layer: why the pump feels even worse
Everything above was measured at the wholesale link. At the retail link - the price on the sign - the feather grows, because links 3 and 4 add the inventory-cost effect and local market power on top of what we measured. A snapshot of this site's city-level retail data (observed September 28, 2026) shows how far the pump price sits above the $3.58-per-gallon NY Harbor wholesale benchmark from the latest EIA week:
| City | Retail regular ($/gal) | Markup over wholesale |
|---|---|---|
| Houston | $3.85 | $0.27 |
| Denver | $4.03 | $0.45 |
| Miami | $4.30 | $0.72 |
| Cleveland | $4.33 | $0.75 |
| Boston | $4.34 | $0.76 |
| New York City | $4.38 | $0.80 |
| Chicago | $4.76 | $1.18 |
| Los Angeles | $6.16 | $2.58 |
The spread - from $0.27 over wholesale in Houston to $2.58 in Los Angeles - is taxes, transport, boutique fuel blends (California's reformulated gasoline costs more to make), and local competition, not crude. Two honest caveats: city retail figures are point-in-time observations, not a time series, so they cannot prove asymmetry by themselves; and wholesale-to-retail markups are not pure profit - they include roughly 50-plus cents of combined federal and state taxes in most states, plus the station's operating costs. The site's city fuel prices page tracks these levels as they update.
8. How to read pump prices: a checklist
If you want to judge whether today's pump price is fair - or predict where it is heading - watch the links in order. Here is the reading order professionals use:
- Start with the crack spread, not crude. The pump price is crude plus the refiner's margin. A $48-per-barrel crack spread (recent reading on this site's market data) means gasoline is priced far above its crude cost - pump relief needs the crack to narrow, not just crude to fall. The markets page tracks the spread daily.
- Check refinery utilization and maintenance season. At 92.5% utilization there is little slack; spring and fall turnaround seasons routinely tighten gasoline supply and widen the feather. Summer driving season does the same on the demand side.
- Watch RBOB futures for the wholesale direction. RBOB leads the physical rack market by days. If RBOB is falling while your local sign is flat, the feather is in its retail phase - declines are coming, slowly.
- Compare your city to wholesale. A markup far above your city's normal range (taxes included) signals weak local competition more than anything about crude. The city table above is your baseline.
- Mind the calendar. The asymmetry is seasonal: it is strongest when demand is strong and refineries are tight (summer), weakest when demand is soft and tanks are full (winter).
- Expect weeks, not days, on the way down. Our weekly data shows the wholesale adjustment is essentially complete within the week - but the retail link adds its own inventory lag. A crude drop on Monday is a pump-price story for late next week at the earliest.
9. Worked example: what the numbers imply (hypothetical)
The following is a labeled hypothetical illustration using the pass-through rates measured above. It is not a prediction - actual weeks vary enormously around these averages, and the retail link adds further delays not captured here.
Suppose WTI crude jumps 10% in a week, from $90 to $99 per barrel. Applying the measured up-week pass-through of 0.79, wholesale gasoline would be expected to rise about 7.9% - so NY Harbor gasoline at $3.58 would move toward roughly $3.86, all else equal. Now suppose the following week crude gives it all back, falling about 9.1% from $99 to $90. Applying the measured down-week pass-through of 0.63, wholesale gasoline would be expected to fall only about 5.7% - from $3.86 toward roughly $3.64. Crude has round-tripped to exactly $90, but wholesale gasoline sits about 1.7% above where it started. That residual - earned by the chain, paid by the driver - is the feather, quantified. In practice the retail sign would show an even larger residual, because the station-level mechanisms of section 6 add their own lag on top.
10. Frequently asked questions
Do gas stations really raise prices faster than they lower them?
Yes - the asymmetry is measurable. In 91 weeks of this site's EIA wholesale data, gasoline captured 79% of crude's upward moves but only 63% of its downward moves in the same week. The effect is well documented in the academic literature going back to the 1990s. That said, your memory exaggerates it: price hikes are salient and memorable, while slow declines are easy to miss.
Is "rockets and feathers" just price gouging?
Not necessarily. Several mechanisms are ordinary economics: pricing off replacement cost when wholesale rises, holding margin while expensive inventory clears when it falls, and the cost of physically repricing. Market power plays a role where competition is weak, and regulators have investigated the pattern repeatedly - but the measured asymmetry exists even in competitive wholesale markets, which points to structural frictions rather than conspiracy as the main driver.
How long does a crude price drop take to reach the pump?
At the wholesale link, the adjustment is mostly complete within the same week (same-week correlation 0.75, next-week correlation near zero in our data). The retail link adds days to weeks depending on the station's inventory turnover and local competition. As a rule of thumb: a crude drop on Monday is a wholesale story that week and a pump-price story late the following week at the earliest.
Why do gas prices vary so much between cities?
State and local taxes differ, transport distances from refineries and pipelines differ, some regions require special boutique fuel blends (notably California), and local station competition varies enormously. In this site's September 28, 2026 snapshot, retail regular ranged from $3.85 per gallon in Houston to $6.16 in Los Angeles against the same $3.58 wholesale benchmark.
What is the crack spread and why does it matter for pump prices?
The crack spread is the difference between the price of refined products (like gasoline) and the crude oil used to make them - roughly, the refiner's gross margin. Pump prices equal crude cost plus the crack plus distribution, taxes, and retail margin. When the crack is wide (recently $48.33 per barrel on this site's data), gasoline can stay expensive even as crude falls. For context on crude benchmarks themselves, see our WTI vs Brent spread explainer.
Can I predict pump prices from WTI crude?
Directionally, yes; precisely, no. WTI explains most of wholesale gasoline's weekly moves (correlation 0.75), but the pass-through is asymmetric and incomplete, refinery margins swing independently, and the retail link adds local delays. Use crude as the direction, the crack spread as the magnitude check, and RBOB futures as the timing signal - the checklist in section 8.
Do prices fall faster where stations compete?
Yes. The search-cost and market-power mechanisms both predict - and local studies confirm - that the feather is thinnest where several stations compete on the same stretch of road and thickest in "edge zones" with a single convenient station. If your neighborhood's prices seem sticky on the way down, a competing cluster a few minutes away is usually falling faster.
A note on what this is - and isn't
This article is educational analysis of publicly available market data, not financial advice. The pass-through statistics are descriptive summaries of past price behavior (91 weeks ending September 25, 2026, plus 551 daily futures sessions ending October 1, 2026) and do not predict future prices. Energy markets are volatile and influenced by geopolitics, weather, policy, and speculation. Nothing here is a recommendation to buy, sell, or hold any commodity, security, or fuel contract. For background on the crude benchmarks referenced, see our WTI vs Brent spread explainer and SPR depletion analysis.
Data: PetroEyes EIA weekly price export and daily futures history, refreshed October 1, 2026. Computation window for wholesale statistics: January 3, 2025 – September 25, 2026 (91 weekly observations); futures cross-check: November 28, 2023 – October 1, 2026 (551 matched sessions). City retail figures observed September 28, 2026. Academic reference: Borenstein, Cameron & Gilbert (1997), "Do Gasoline Prices Respond Asymmetrically to Crude Oil Price Changes?", Quarterly Journal of Economics.