The Los Angeles-New York City Gasoline Gap, 2000-2026: Seattle Pays Almost the Same Premium

Net of the state fuel-tax difference, Seattle’s gasoline premium over New York City is larger than Los Angeles’s: 87 cents a gallon against 83.

Los Angeles and New York City gasoline, 2000 to 2026

Two retail markets in the same country, 1370 common survey weeks, and a gap that has not closed since 2015.

Source: U.S. Energy Information Administration, weekly retail regular all formulations. Chart: Eco3min Research.

A weekly dataset of 21 US gasoline markets, the difference series EIA does not publish, and an exact three-term allocation of the California to New York gap.

Eco3min Research · 1875 weekly observations · data through 31 August 2026 · CC BY 4.0

The Los Angeles to New York City gasoline gap reached 1.451 dollars a gallon in the week of 31 August 2026, against a median of 0.415 over the 1370 weeks both cities have been surveyed. The gap has risen in every one of six fixed calendar blocks since 2000, from 0.129 dollars a gallon in 2000 to 2005 to 1.327 in 2023 to 2026. Measured at regional level, where the record runs 34 years, the West Coast to East Coast gap has risen in all seven blocks since 1992. This study publishes the weekly difference series, its percentile history and its decomposition, none of which the U.S. Energy Information Administration distributes as a series.

Latest observation

Executive summary

1875 weekly observations, 1990 to 2026, 21 markets, 34 columns. Licensed CC BY 4.0. Attribution: Eco3min Research.

Key statistics

How the gap grew

The gap widened in every block, and not only in California

Median gap by fixed calendar block, three pairs. The coast measured without California follows the same path as the coast as a whole.

Source: U.S. Energy Information Administration. Chart: Eco3min Research.

Read the chart as three answers to the same question at three levels of aggregation. The terracotta line is the pair the search traffic asks about, Los Angeles against New York City. The steel line replaces the two cities with their entire coasts, which extends the record back to 1992 and removes any city-level idiosyncrasy. The green line removes California from the West Coast altogether. All three rise, and they rise together.

The regional pair is the more demanding test, because it involves no choice of city. Its median has risen in all seven blocks since 1992, from 0.127 dollars a gallon in 1992 to 1999 to 1.089 in 2023 to 2026, a multiple of 8.6. Its latest reading of 1.269 sits at the 97.7th percentile of 1791 weeks, and it has been negative in 18 of them, just over one percent.

The green line is the one that carries the argument, and it is the one that behaves least tidily. The West Coast excluding California premium over the East Coast rises from 0.107 dollars a gallon in the 2000 to 2005 block to 0.746 in 2023 to 2026, a multiple of 7.0, but it does not rise monotonically: it dips by 0.001 between the 2006 to 2010 and 2011 to 2015 blocks. The claim of a strictly rising staircase belongs to the other two series, not to this one.

What the gap is usually attributed to

The dominant reading is specific and, on its own terms, well founded. California levies the highest state gasoline tax in the country, 59.6 cents a gallon effective 1 July 2024 according to the Federal Highway Administration’s Highway Statistics 2024, table MF-121T, against 25.4 cents in New York. It also requires a fuel that is made to its own specification, which narrows the set of refineries able to supply it and is the mechanism Brown, Hastings, Mansur and Villas-Boas identified when they showed that content regulation segments wholesale markets geographically (Journal of Environmental Economics and Management, 2008). Add a refining system with no pipeline connection to the Gulf Coast, and the pieces fit. Refining capacity and pipeline geography belong to the physical side of oil and refined-product markets, the part of a pump price that no tax table explains.

They fit well enough that the arithmetic looks closed before anyone checks it. A 34-cent tax difference and a bespoke fuel are the two facts most readers already hold, and a gap of a dollar and a half feels like the sum of them plus some refinery trouble. The question this dataset can settle is not whether those mechanisms exist, which is not in doubt, but how much of the observed gap sits where they apply.

The premium measured outside California

The test that settles the attribution question needs no tax series at all. EIA publishes a weekly price for the West Coast excluding California, covering Washington, Oregon, Nevada, Arizona, Alaska and Hawaii. Those states share the West Coast’s refining and supply system. They do not share California’s fuel tax, and they do not require California’s fuel specification.

Written as an identity, the state-level gap splits into three terms that sum exactly:

In the week of 31 August 2026 those three terms are 0.697, 0.886 and minus 0.189 dollars a gallon, summing to 1.394. The middle term, the one that contains no California tax and no California fuel rule, is the largest of the three. It accounts for 63.6% of the total, against 50.0% for everything specific to California. The third term is negative because New York State is cheaper than the East Coast average, which is what pushes the two shares above one hundred percent between them.

Take out the state fuel tax and Seattle pays more than Los Angeles

Premium over New York City, median of the 27 weeks from 1 July to 31 December 2024, split at the statutory state gasoline tax difference.

Sources: U.S. Energy Information Administration; Federal Highway Administration, Highway Statistics 2024, table MF-121T. Chart: Eco3min Research.

The city-level version of the same test is blunter. Over the 27 weeks from 1 July to 31 December 2024, chosen because the statutory rates in force are the ones the Federal Highway Administration publishes for that year, Los Angeles ran 1.168 dollars a gallon above New York City and Seattle ran 1.112. Subtract the state fuel-tax difference from each, 34.2 cents for California against New York and 24.0 cents for Washington, and Los Angeles is left with 82.6 cents while Seattle is left with 87.2. The city with California’s tax and California’s fuel rule has the smaller residual of the two.

Two cross-sectional facts from the same tax table make the point without any price data. Pennsylvania taxes gasoline at 57.6 cents a gallon, 2.0 cents below California, and sits on the East Coast. Arizona taxes it at 18.0 cents, 7.4 cents below New York, and sits in the West Coast supply region. The tax ranking and the price ranking do not line up.

Within California, the city effect is close to nothing. San Francisco and Los Angeles face the same state tax and the same specification, and the median difference between them across 1370 weeks is 3.35 cents a gallon, 2.0 cents in the latest week. Whatever separates Los Angeles from New York City is not something about Los Angeles.

Where that reading is weakest

The obvious objection is that “outside California” is not the same as “outside California policy”, and it is a good one. Washington and Oregon both operate clean fuel standards, and Washington’s cap-and-invest programme began in 2023, inside the block where the West Coast excluding California premium rises fastest. Anyone treating the middle term as a clean measure of geography is treating a partly regulatory number as a purely physical one.

Two things survive the objection, and they are narrower than the headline. First, the timing does not fit a 2023 explanation on its own: the West Coast excluding California premium had already risen from 0.107 to 0.462 dollars a gallon between the 2000 to 2005 and 2020 to 2022 blocks, before any of those programmes was in force. Second, Arizona operates neither a clean fuel standard nor a cap-and-invest programme and taxes gasoline below New York, and it sits inside the region carrying the premium. The defensible claim is the narrow one: the gap is not confined to the state whose rules are usually named for it.

A second objection is more fundamental and is conceded outright. The three-term split is an accounting identity, not a causal decomposition. Its terms are correlated with one another, the residual absorbs everything the study does not measure, and calling the middle term a measure of geographic isolation is an interpretation, not a result. The dataset allocates a gap. It does not identify a cause, and no line in it should be read as doing so.

A third is about the tax comparison. It is a cross-section, not a series. Statutory rates for a single year are compared against a retail window chosen to match that year, because the Federal Highway Administration publishes the table annually and the study declined to interpolate between editions. A weekly ex-tax spread would have been the stronger object; it was abandoned rather than estimated.

Explore any pair

The prose above commits to five comparisons. The dataset contains 21 markets, which is 210 unordered pairs, and the one a reader wants is usually not among the five. The figure below draws the weekly difference between any two of them, straight from the published CSV.

Any two of 21 US gasoline markets, weekly since 1990

Pick a market and a reference. The panel reports the latest difference, the median, the number of common weeks and where the latest reading sits in that pair’s own history.

Source: U.S. Energy Information Administration, weekly retail regular all formulations. Every value shown is a difference between two columns of the published CSV.

What follows a wide gap

Because this study has no asset attached to it, the natural forward object is the gap itself. The table reports how the Los Angeles to New York City gap changed over the following 26 and 52 survey weeks, conditioned on which quartile of its own history it started in, across the 1318 weeks with a complete 52-week window.

The only bucket with a clear pattern is the narrowest one, where the gap widened over the following year in 76.1% of weeks. From the top quartile, where the gap sits today, the distribution is close to symmetric: it widened in 54.1% of weeks, and its interquartile range is centred near zero. A wide gap has not historically been followed by a reliable move in either direction.

Two cautions apply to the whole table. The windows overlap, so neighbouring weeks are not independent observations and the effective sample is far smaller than the counts shown. And the buckets are defined on the full-sample distribution, which is only knowable after the fact.

Past distributions are not predictive of future outcomes. Regime-conditional statistics describe historical patterns, not expected returns.

Levels to watch

The series to watch alongside it is the middle term of the identity, the West Coast excluding California premium over the East Coast, currently 0.886 dollars a gallon against a 2023 to 2026 block median of 0.746. If that term fell back toward its 2016 to 2019 median of 0.314 while the California-specific term held, the balance of the gap would shift toward the state-specific explanation, which is the reading the data currently does not support. The next scheduled input is the EIA weekly retail survey, published each Monday.

Data tables

Median gap by fixed calendar block, in US dollars per gallon. Block boundaries are calendar years, fixed before the data was cut, and listed in the filter definitions below.

The 1992-1999 block predates the city series, which begin in June 2000, and the West Coast excluding California series, which begins in May 1998. The 140-cent reading shown for that block covers only its final two years and is reported for completeness rather than comparison.

Statutory state gasoline tax rates used in the tax comparison, in cents per gallon, with the effective dates as published.

Historical turning points

Four weeks are worth looking up line by line, because each is a case where the gap did something the structural reading alone would not predict.

Week of 5 September 2005, the narrowest reading on record. Los Angeles was at 3.004 dollars a gallon and New York City at 3.276, a gap of minus 0.272. This is one of the 54 weeks in 1370 where Los Angeles was the cheaper of the two, and it sits at the 0.4th percentile of the gap’s own history. The direction is the point: an East Coast and Gulf Coast supply shock can invert a gap that structural factors are said to fix.

Week of 8 October 2012. Los Angeles at 4.736, New York City at 3.958, a gap of 0.778, which was the widest reading in the record at that date and stood at the 100th percentile of its own history to that point. It is a reminder that the current level, 1.451, would have looked unthinkable at a time when the record itself was half of it.

Week of 3 October 2022, the widest reading on record. Los Angeles at 6.194, New York City at 3.305, a gap of 2.889. Note the two dates separately: the record gap is not the record Los Angeles price, and neither is the latest week.

Week of 1 July 2024, the start of the tax-aligned window. Los Angeles at 4.480, New York City at 3.391, a gap of 1.089, with Seattle 0.984 above New York City. The two premiums were already within eleven cents of each other before any tax was subtracted.

Against those, the latest week: Los Angeles 5.532, New York City 4.081, a gap of 1.451 at the 94.2nd percentile of 1370 weeks.

Methodology

All prices are U.S. Energy Information Administration weekly retail prices for regular gasoline, all formulations, in US dollars per gallon, taken from the legacy workbooks at eia.gov/dnav/pet/hist_xls/. No series is seasonally adjusted, deflated or smoothed. The survey week is dated to its Monday.

The study’s own columns are differences and one rank:

Filter definitions. Every temporal filter used anywhere on this page is defined here and nowhere else. full_city: the 1370 weeks where both Los Angeles and New York City are published, 5 June 2000 to 31 August 2026. full_region: the 1791 weeks where both PADD 5 and PADD 1 are published, 11 May 1992 to 31 August 2026. full_seattle: the 1215 weeks where both Seattle and New York City are published, 26 May 2003 to 31 August 2026. Calendar blocks: 1992-1999, 2000-2005, 2006-2010, 2011-2015, 2016-2019, 2020-2022 and 2023-2026, each running from 1 January of the first year to 31 December of the last, the final block open-ended at the last published week. tax_window_2024: 1 July to 31 December 2024, 27 weeks, aligned to the statutory rates the Federal Highway Administration publishes for 2024. pre_2020: everything published on or before 31 December 2019.

Scale justification. Price range across the panel is 0.97 to 6.22 dollars a gallon on the two series plotted, a ratio of 6.4, which would ordinarily call for a logarithmic primary axis. The hero chart is nonetheless linear, because the object of the study is an absolute difference in dollars per gallon and a logarithmic axis renders the shaded band between the two lines meaningless. The relative view is supplied instead as a column of the dataset: the gap as a share of the New York City price, whose median is 15.61% over the full sample against 35.55% in the latest week. That column shows the same widening on a ratio scale.

Sensitivity. Three checks, all reported in both directions. Replacing the block median with the block mean moves each block by between 0.003 and 0.077 dollars a gallon and does not change the ordering. Shifting every block boundary one year later gives a first-to-last multiple of 8.7; one year earlier gives 13.5; as published, 10.3. Dropping the 2020 to 2022 block entirely leaves the remaining five still strictly increasing, at 0.129, 0.274, 0.363, 0.703 and 1.327.

Product basket. EIA publishes a reformulated series and an all formulations series for Los Angeles, San Francisco and New York City. They are identical on all 1370 common weeks, maximum absolute difference 0.0000, because all three are wholly reformulated markets. The choice of basket therefore has no effect on any result here, and the assertion is executed in the build script. Note the labelling: EIA calls the reformulated New York series “New York Harbor” and the all formulations series “New York City”. This study uses the all formulations basis throughout, so that city, state and regional series are on the same footing and the three-term identity closes.

Tax rates. Statutory state gasoline tax rates are from the Federal Highway Administration, Highway Statistics 2024, table MF-121T, with the effective dates as published there. They are excise and excise-equivalent rates. They exclude California’s state sales tax on gasoline and New York’s local option, both described in the context box above. The California rate was cross-checked against the issuing agency, the California Department of Tax and Fee Administration, which publishes the same 59.6 cents effective 1 July 2024. No tax series is published in the dataset and none was interpolated.

Reproduction. The dataset is rebuilt from source by the following:

Data sources and references

• DataU.S. Energy Information Administration, weekly retail gasoline and diesel prices, series EMM_EPMR_PTE and EMM_EPMRR_PTE, legacy workbooks. All price columns in the dataset.
• DataFederal Highway Administration, Highway Statistics 2024, table MF-121T, state tax rates on motor fuel. Used for the tax cross-section only.
• DataCalifornia Department of Tax and Fee Administration, motor vehicle fuel tax rates. Used as the cross-check on the California rate.
• PaperBrown, J., Hastings, J., Mansur, E. T. and Villas-Boas, S. B. (2008), “Reformulating competition? Gasoline content regulation and wholesale gasoline prices”, Journal of Environmental Economics and Management 55(1), 1 to 19. On geographic segmentation of wholesale markets by fuel content rules.
• PaperBorenstein, S., Cameron, A. C. and Gilbert, R. (1997), “Do gasoline prices respond asymmetrically to crude oil price changes?”, Quarterly Journal of Economics 112(1), 305 to 339. On the pass-through behaviour of retail gasoline prices.
• Eco3minUS refinery utilization rate, weekly EIA data since 1990, the capacity-side series this study does not use but a reader will want.
• Eco3minEvery US refinery shutdown since 1990, the national closure record.
• Eco3minUS 3-2-1 crack spread, daily refining margin data since 1986.

Limitations

• EIA revises its weekly retail series. Every figure here is as published on 31 August 2026 and a later vintage may differ slightly.
• The three-term split is an accounting identity. Its terms are correlated, and it allocates a gap rather than identifying a cause.
• The tax comparison is a single-year cross-section, not a series. It cannot say how the tax share of the gap has moved over time, and the study declined to estimate one.
• Statutory per-gallon rates exclude California’s ad valorem state sales tax on gasoline and New York’s local option, so the residuals reported are net of excise, not net of all tax.
• The city, state and regional series start on different dates: 1992 for the PADD regions, 1998 for the West Coast excluding California, 2000 for most cities, 2003 for Seattle, Washington, Florida, Massachusetts and Ohio. Any cross-market comparison is limited to the shorter of the two records.
• The forward table uses overlapping windows, so its counts overstate the number of independent observations, and its quartile boundaries are known only after the fact.

FAQ

Related

• US regular gasoline retail price, weekly since 1990
• US premium gasoline retail price, 1994 to 2026
• US refinery closures since 2020, the complete record
• Refining margins, the hidden driver of oil profits
• The WTI to Brent spread, 1987 to 2026
• US electricity prices adjusted for inflation, state by state
• Real crude oil price, inflation-adjusted history
• Commodity regimes and the physical constraints of the energy transition

Eco3min Research · Weekly data from the U.S. Energy Information Administration · Dataset released under CC BY 4.0 · Last updated September 2026