The US Energy Index is one number for what energy costs an American household. It blends the four prices a household pays directly — gasoline, electricity, piped natural gas, and heating oil — weighted by what households actually spend on each, into a monthly index with the calendar-2000 average set to 100. Everything below is the entire method. There are no adjustments, no smoothing, and no discretionary overrides; the index is recomputed from the public source series on every refresh.
The index is a fixed-weight (Laspeyres-type) relative price index:
I(t) = 100 × Σᵢ wᵢ · pᵢ(t) / p̄ᵢ(2000)
where pᵢ(t) is component i's price in month t, p̄ᵢ(2000) is that component's average price across the twelve months of 2000, and wᵢ is its expenditure weight. A month enters the index only when all four components report a value; in practice the BLS series are continuous, and the index runs monthly from 1978-11 without gaps.
The weights are the average annual expenditures per consumer unit ("household") from the US Bureau of Labor Statistics, Consumer Expenditure Surveys, 2024 — the same survey that weights the CPI — normalized to sum to one. We store the dollars, not the shares: the shares are computed from the dollars at build time, so the weights can never drift from their citation.
| Component | CE line item | $/year | Weight |
|---|---|---|---|
| Gasoline | Gasoline | $2,411 | 49.6% |
| Electricity | Electricity | $1,833 | 37.7% |
| Natural gas | Natural gas | $493 | 10.1% |
| Heating oil | Fuel oil and other fuels | $125 | 2.6% |
The weight is where the index earns its name. Gasoline is nearly half the blend because it's nearly half of what households spend on energy; heating oil is under 3% because most households don't burn it at all. An unweighted average of the four prices would be a statement about arithmetic, not about households. One known simplification: the CE line for fuel oil is "fuel oil and other fuels," which folds in propane and firewood; we map it entirely to the heating-oil price series, the only member of that group with a monthly BLS price. At a 2.6% weight the distortion is small, and we'd rather name it than hide it.
All four components are BLS Average Price Data: actual retail prices, US city average, collected monthly alongside the CPI, not seasonally adjusted. One originating body for the whole spine, so the components are methodologically consistent with each other. We ingest them through FRED's public download endpoint and stamp the originating body.
| Series | Code | Unit | 2000 average |
|---|---|---|---|
| Gasoline, unleaded regular | APU000074714 | $/gal | $1.510 |
| Electricity, residential | APU000072610 | $/kWh | $0.087 |
| Utility (piped) gas | APU000072620 | $/therm | $0.803 |
| Fuel oil #2 (heating oil) | APU000072511 | $/gal | $1.360 |
The base is the calendar-2000 average of each component, so the index reads directly as "energy costs X% more than it did in 2000." Using a full-year average rather than a single month keeps seasonality out of the base. The choice of 2000 is a convention — a round, recent-enough year that predates the 2000s commodity cycles — and changing it would rescale the series without reordering a single month.
The underlying prices are not seasonally adjusted, and energy is strongly seasonal: gasoline rises into summer, heating oil into winter. A month-over-month move mixes the season with the signal. We publish the month-over-month figure because it's true, and we flag it on every page; the year-over-year figure is the one that means what it appears to mean.
The per-state pages sit beside the index rather than inside it; the index itself is a single national figure. They carry two clocks. The freshest reading is residential electricity from EIA Electric Power Monthly Table 5.6.A, updated monthly with about a two-month lag. The full by-category picture — electricity, gasoline, natural gas, heating oil, and total energy — comes from the EIA's State Energy Data System (SEDS), which is annual and runs about a year and a half behind the present. SEDS prices are expressed in dollars per million Btu: one denominator across fuels, so a kilowatt-hour, a therm, and a gallon can be compared per unit of heat delivered. That comparison is made before equipment efficiency — a heat pump extracts more useful heat from a million Btu of electricity than a furnace does from a million Btu of gas — and we state that on each page rather than adjust the numbers.
It is not a total-energy-cost measure — it excludes what households pay for energy indirectly through airfares, shipping, and goods prices. It is not the CPI energy index, though it will usually rhyme with it: the CPI uses finer item strata, seasonal adjustment, and evolving weights, while this index deliberately stays simple enough to verify by hand from the published series. And it is not a forecast. It says what energy cost last month; it does not say what it will cost next month.
The index is recomputed over its full history on every data refresh, so any source revision flows through automatically — nothing is frozen. The weights hold a single CE vintage (currently 2024) between reconstitutions. When a new CE year is published, adopting it is a versioned event: the methodology version increments, the change is dated here, and the full history is recomputed under the new weights so the series stays internally consistent. Version 1.0 is the inaugural method, published 2026-07-11.
Every input is public and keyless: the four price series by their codes above (via FRED or the BLS database), the CE expenditure dollars from the CE annual tables, the state electricity table from EIA Electric Power Monthly Table 5.6.A, and the state history from EIA SEDS. The full index history and every component and state series carry CSV downloads. If your recomputation disagrees with ours, tell us — one of us has a bug, and we'd like to know which.