Data analysis
EIA Published August: Residential Electricity Came In Below EIA's Estimate and Stayed Above the Real Baseline
EIA released August 2026 residential electricity prices. We checked the recorded EIA estimate and our own falsifier from post 1.
- Data through
- August 2026 (EIA Electric Power Monthly, released 2026-10-23); CPI through September 2026 (BLS)
Key takeaways
- What is new
- We recorded the EIA estimate and a falsifier for August 2026 before the data came out; this post scores both against the first published EPM value, and says plainly what we got right or wrong.
- What would prove it wrong
- Two checks on the first published September 2026 residential price, in the Electric Power Monthly scheduled for 2026-11-24. Estimate: the October 2026 STEO puts September 2026 at 18.42258 cents per kWh, so 18.32 cents or lower, or 18.53 cents or higher, is a miss outside our 0.10 cent test. Baseline: at the September 2026 CPI-U of 336.940, the 2016 to 2019 real baseline mean equals 17.43249 cents nominal, so 17.43 cents or less would put September below the baseline. That test is set against an all-month baseline: on a 2016 to 2019 September-only baseline the equality point would be 17.68126 cents, and only 2 of the 4 baseline Septembers were above the all-month mean. A revision of August 2026, of the baseline months or of the CPI-U values used would mean recalculating the August comparison and this threshold; a revision is never counted as a falsifier result.
On this page 8 sections
The first published August value came in below EIA's estimate. August 2026 residential was 18.16 cents per kWh, against the recorded STEO estimate of 18.31218 cents, shown as 18.31 in post 1: 0.152 cents lower, outside our 0.10 cent test for a hit (an editorial threshold, not an EIA standard). One print below a recorded estimate is not a trend, and these values are preliminary.
What is new here. In post 1 we put EIA's estimate and one falsifier on the record before the August release. This post scores both.
Recorded estimate versus first published observation#
| Item | Value | Unit | Direction |
|---|---|---|---|
| EIA STEO estimate for August 2026 (October 2026 edition), as recorded | 18.31218 | cents per kWh | not applicable |
| First published August 2026 value, Electric Power Monthly Table 5.3 | 18.16 | cents per kWh | not applicable |
| Size of the difference between the first published value and the recorded estimate | 0.152 | cents per kWh | first published value lower than the estimate |
| August 2025 (EPM, revised from 17.61 in post 1) | 17.57 | cents per kWh | not applicable |
| Year over year, nominal, August 2026 against August 2025 | 3.4 | percent | August 2026 higher than August 2025 |
The comparison is between the recorded STEO estimate and the first published EPM estimate, which remains subject to revision. The October 2026 STEO lists September 2026 as its last historical month, so its August value is an EIA estimate for a month that had already ended, made before the EPM figure was published. In this release EIA marks values for 2026 preliminary and values for 2025 and earlier final. We score the published two-decimal value against the unrounded estimate: a hit means a difference of 0.10 cents or less, boundaries included. Rounding happens only for display, so a displayed difference can look close to the 0.10 cent line without landing on the side you might expect.
Did August trip the registered falsifier?#
In post 1 we tested whether residential electricity was still running above its 2016 to 2019 real baseline. July 2026 was 6.0 percent above the baseline mean. We said a published August price of 17.33 cents per kWh or less would be below the baseline (the registered equality point at the recorded August CPI-U of 334.980 is 17.33109 cents), and that revisions to July 2026 or the baseline months would require recalculating the July comparison.
It did not trip. August came in at 18.16 cents, above the registered nominal threshold of 17.33 cents, and the July 2026 value in this release is unchanged at 18.31 cents. That is the result of the registered nominal test and nothing more; the real comparison, recomputed from the data retrieved for this follow-up, is in the Real versus nominal section. One more month is not proof of a trend; the next signal is in the section on what would prove this wrong next.
Both outcomes are recorded under Falsifier checks in the Corrections log, with the two September checks set out at the end of this post.
Real versus nominal#
Seasonality note: the baseline is a mean of all 48 months of 2016 to 2019, and Augusts across those years averaged above that mean, so August is compared with an all-month baseline. In the post 1 data vintage the all-month real baseline mean was 17.28 cents and the July-only mean was 17.52 cents, in July 2026 dollars. At the recorded August CPI-U of 334.980, the all-month equality point is the registered one given above, and the August-only equality point is 17.71 cents.
With the data retrieved for this follow-up, real August 2026 is 2.55 percent higher than the 2016 to 2019 August-only real average of 17.65 cents. The newly retrieved August CPI-U matches the recorded 334.980, so both equality points above are unchanged.
Real August 2026 residential price: 18.10 cents in July 2026 dollars (August 2025: 18.11). In real terms August 2026 is 0.04 percent lower than August 2025, while CPI-U all items is 3.4 percent higher than a year earlier. Against the 2016 to 2019 real baseline average of 17.28 cents, real August 2026 is 4.78 percent higher than that average. Since January 2022, 41 of 55 months with CPI were above that average.
The window from January 2022 to August 2026 has 56 months. October 2025 is left out: BLS did not publish CPI for it because of the 2025 lapse in appropriations, so its real value is not found and we did not guess it.
EIA versus CPI electricity#
Over the year to August 2026 the EIA residential average price is 3.4 percent higher and the BLS CPI electricity index is 3.8 percent higher, so the two changes differ by 0.4 percentage points. These are different measures: EIA divides utility revenue by kWh sold, and BLS tracks a price index.
The order has flipped since post 1. In July 2026 EIA residential was 5.3 percent higher than a year earlier (4.9 percent in the post 1 data vintage) and CPI electricity was 4.2 percent higher, so EIA was the higher of the two; in August 2026 it is the lower.
What it costs, with a stated assumption#
For a round 1,000 kWh in a month (our assumption, not an average household): $181.60, or 4.81 hours of average private-sector hourly earnings (37.76 dollars, series CES0500000003, preliminary). Average hourly earnings are not household income.
Revisions since post 1#
EIA revised values we use. In this release 67 monthly values across the residential, commercial, industrial and all-sector series differ from the post 1 vintage: 48 in 2025, 19 in 2026 and 0 in 2024. The 2025 changes arrive in the release that marks 2025 values final. The residential changes, in cents per kWh, from the post 1 value to the new value: January 2025 15.94 to 15.92, February 16.43 to 16.40, March 17.09 to 17.03, April 17.55 to 17.20, May 17.37 to 17.33, June 17.47 to 17.44, July 17.45 to 17.39, August 17.61 to 17.57, September 18.08 to 18.05, October 17.97 to 17.95, November 17.78 to 17.77, December 17.24 to 17.11, March 2026 18.56 to 18.60, April 18.83 to 18.86, and June 18.34 to 18.35. The year-over-year figures in this post use the revised August 2025 value of 17.57.
July 2025, the starting point of the July 2025 to July 2026 changes in post 1, moved from 17.45 to 17.39 cents. The year-over-year figures in post 1 describe the earlier data vintage.
Post 1 now carries a dated note listing every figure that moved, with its published value and its value on the new data; the same figures are in the Corrections log.
Baseline history (2016 to 2019 residential prices and CPI-U, plus July 2026 CPI-U) came from freshly retrieved EIA bulk data and BLS files and was compared month by month with the post 1 vintage: no changes found.
July 2026, the month post 1 tested, is unchanged at 18.31 cents. The 2016 to 2019 baseline prices were rechecked against the EIA bulk series updated 2026-10-21 and the CPI-U values against BLS files retrieved 2026-10-24, and neither changed, so the July comparison in post 1 (July 2026 6.0 percent above the baseline mean) does not need recalculating.
What would prove this wrong next#
Both checks use the first published September 2026 value in the Electric Power Monthly scheduled for 2026-11-24, and both were written before that release.
The estimate check. The October 2026 STEO puts September 2026 residential at 18.42258 cents per kWh; it also counts September as a historical month, so this is again an estimate, not a forecast. A hit means the published two-decimal value is within 0.10 cents of that unrounded estimate, boundaries included: 18.33 to 18.52 cents is a hit, 18.32 cents or lower is a miss on the low side, and 18.53 cents or higher is a miss on the high side. A low miss would make two months in a row below the STEO estimate, and we would say so.
The baseline check. We keep the test from post 1 and move it to September. The deflator is CPI-U all items, not seasonally adjusted (CUUR0000SA0), with July 2026 dollars as the base. The 2016 to 2019 all-month real baseline mean, to six decimals, is 17.276143 cents in July 2026 dollars (17.28 at two decimals). Multiplying that six-decimal mean by the published September 2026 CPI-U of 336.940 and dividing by the July 2026 CPI-U of 333.918 gives an equality point of 17.43249 cents per kWh, so a first published September price of 17.43 cents or less would be below the baseline. On a 2016 to 2019 September-only baseline (real average 17.522674 cents to six decimals) the equality point would be 17.68126 cents; only 2 of the 4 baseline Septembers were above the all-month mean. A revision of August 2026, of the 2016 to 2019 baseline months or of the CPI-U values used would mean recalculating the August comparison and this threshold, and we would say so; a revision is never counted as a falsifier result. One month below would describe September, not a trend, and we would be glad to be wrong.
Sources
- U.S. Energy Information Administration, Electric Power Monthly, data for August 2026, released 2026-10-23, Table 5.3. Retrieved 2026-10-24.
- U.S. Energy Information Administration, bulk data file ELEC, series ELEC.PRICE.US-RES.M (series updated 2026-10-21, data through December 2025), used to recheck the 2016 to 2019 baseline months. Retrieved 2026-10-24.
- U.S. Energy Information Administration, Short-Term Energy Outlook, October 2026, series ESRCUUS. Retrieved 2026-10-19.
- U.S. Bureau of Labor Statistics, CPI-U all items CUUR0000SA0, CPI-U electricity CUUR0000SEHF01, average hourly earnings CES0500000003, via the BLS Public Data API v1. Retrieved 2026-10-24.
Definition flags (5)
- Estimate vs first published observation (plan vs operated): the recorded estimate is the EIA Short-Term Energy Outlook of October 2026 (18.31 cents for August 2026 residential). That outlook counts August and September 2026 as historical months, so its values for them are estimates made before the Electric Power Monthly figures came out, not forecasts of future months. The comparison value is the first published Electric Power Monthly figure, released 2026-10-23, still preliminary.
- Preliminary vs final: in the Electric Power Monthly released 2026-10-23, EIA marks values for 2025 and earlier as final and values for 2026 as preliminary. In the earlier release behind post 1, 2025 values were still preliminary.
- Residential vs all sectors: this post uses residential (ELEC.PRICE.US-RES.M and Table 5.3) only.
- Nominal vs real: real values use CPI-U all items (CUUR0000SA0) in July 2026 dollars, the same base as post 1.
- EIA average price vs BLS CPI electricity: different constructions; they need not agree.
How this was made
Written with Grok Bot, an AI. Grok Bot downloaded the files, ran the analysis scripts and drafted the text. GrandPrixx, the reviewer named on the byline, checks the headline numbers against the raw files and reviews the wording before a post is published, and nothing is published without GrandPrixx's explicit go. No other author is implied. The Method page describes the formulas and the review process. The data files come from the public sources named below, and the raw files and scripts are kept in a repository that is currently private. Errors are logged on the Corrections log; to report one, email [email protected].
Sources: U.S. Energy Information Administration, Electric Power Monthly (data for August 2026, released 2026-10-23), Table 5.3, the bulk data series ELEC.PRICE.US-RES.M (updated 2026-10-21, retrieved 2026-10-24) and Short-Term Energy Outlook (October 2026); U.S. Bureau of Labor Statistics series CUUR0000SA0, CUUR0000SEHF01 and CES0500000003, retrieved through the BLS Public Data API on 2026-10-24. Calculations are ours. BLS.gov cannot vouch for the data or analyses derived from these data after the data have been retrieved from BLS.gov. Real Price Desk is not affiliated with or endorsed by EIA or BLS.
This is general information, not advice about your energy contract or finances.