We Tracked 186 Electric Rate Changes Across 31 States
We archive the old rate every time we update a city. Across 186 verified electric rate changes in 31 states, the median household bill rose 5.6% (+$9.31/month)—and Connecticut fell 21%. Full method, including the 177 records we threw out.
Most utility rate coverage tells you a commission approved an increase. Almost nobody goes back afterward to measure what actually landed on the bill—because once a utility publishes a new tariff, the old one quietly disappears from its website.
We keep the old one. This post is built entirely on our own archived rate history—a first-party record we maintain across the 1,068 cities we track. Every time we update a city's rates, the previous values are written to a dated archive file with a reason for the change. Nothing here comes from a third-party index; it comes from our own before-and-after snapshots. This is the first public look at that archive.
Every figure below is electric. We also track water, sewer, and trash, but the archive holds too few non-electric records to support any trend claim, and we make none. See our methodology for how rates are sourced and verified, and about us for who maintains them.
Quick answer: Across 186 comparable electric records spanning 31 states, 77.4% of household bills went up. The median change was +5.6%, or +$9.31/month at 1,000 kWh—moving a median modeled bill from $147.00 to $156.70. The largest verified increase was Youngstown, Ohio at +$28.40/month. The largest verified decrease was Hamden, Connecticut at −$66.25/month.
Read the methodology section before quoting these numbers. We discarded 177 of our 363 electric records—nearly half—because they measured something other than a real price change. That filtering is the whole reason the remaining figures mean anything.
What this dataset is
We track 1,068 U.S. cities across electric, water, sewer, and trash. When a rate changes, we don't overwrite the record—we write the previous version to a dated archive file along with a plain-language reason for the change.
As of this analysis the archive holds:
| Count | |
|---|---|
| Archived change events | 433 |
| Distinct city + utility series | 404 |
| States represented in the archive | 46 |
| Capture window | May 24 – Aug 29, 2026 |
By utility, the archive is overwhelmingly electric: 363 electric records versus 19 water, 18 sewer, and 4 trash.
Two different numbers appear in this post, so here is the difference up front. The archive holds 433 change events across 46 states. The analysis uses the 186 electric records across 31 states that survive the filtering described in the next section. We headline the smaller, stricter number.
One framing caveat we want stated up front: this is a 14-week capture window, not a full calendar year. The underlying rate actions have effective dates reaching back to January 2026, but the archive itself only began accumulating in late May. Treat this as a snapshot of a specific stretch of 2026, not an annual index. It becomes a true year-over-year series the next time we run it.
Why we threw out 177 records
This is the part that matters, and it's where a careless version of this post would go wrong.
A change in our stored number is not the same thing as a change in your bill. Three different things can move a stored rate, and only one of them is a real price change:
1. Seasonal rate structures (125 records excluded). Some utilities publish genuinely different summer and winter schedules. Georgia Power charges a flat 8.0602¢/kWh in winter but a tiered summer rate. Comparing a winter record to a summer record produced an apparent +93.8% jump in Georgia cities. Nothing was approved; the calendar moved. Any record whose source names or notes reference seasonal pricing is excluded.
2. Corrections to our own baseline (50 records excluded outright, plus outliers annotated below). If we previously stored an estimate and replaced it with a confirmed tariff, the delta measures our research improving—not a rate rising. Every record whose prior value was flagged estimated is excluded.
3. Genuine rate actions. A commission order, a supply-rate reset, a board vote. These are what we kept.
The filter that isn't enough
Excluding estimated baselines still isn't sufficient, and we want to show our work on this rather than bury it.
Six Puget Sound Energy cities in Washington show a raw +54.1% increase. But the archive's own reason field records what actually happened: the Washington UTC approved a ~12.2% net residential increase effective January 1, 2026. Our previously stored value—marked confirmed—had used a flat energy rate where PSE's Schedule 7 is tiered, so it understated the baseline bill. Roughly 12 points of that 54% is a real increase, and the rest is us fixing our own math.
The lesson: a "confirmed" label means we cited a real source, not that the modeled bill was right. Where a stored figure and a stated rate action disagree, we trust the stated action. Washington is therefore excluded from the verified rankings below despite technically passing every automated filter.
After all filtering: 186 comparable electric records, one per city, across 31 states.
What actually happened to bills
| Result | Share |
|---|---|
| Bill increased | 144 records · 77.4% |
| Bill decreased | 38 records · 20.4% |
| Essentially unchanged | 4 records |
| Measure | Value |
|---|---|
| Median change | +5.6% |
| Median dollar change | +$9.31/month |
| Mean dollar change | +$8.15/month |
| Median modeled bill, before | $147.00 |
| Median modeled bill, after | $156.70 |
All dollar figures are modeled at a standardized 1,000 kWh/month so cities are comparable. Your bill depends on your actual usage.
The mean sitting below the median is worth noting: a handful of large decreases pulled the average down while the typical household still saw an increase. Most bills rose modestly; a few fell sharply.
Where bills rose most (verified)
Each of these traces to a specific, documented rate action.
Youngstown, Ohio — +$28.40/month (+35.5%)
Ohio Edison's residential Price to Compare rose from 7.99¢/kWh to 10.83¢/kWh effective June 1, 2026. Ohio's Price to Compare is the default supply rate for customers who haven't shopped a competitive supplier—so this hit exactly the households least engaged with the market.
Ohio's statewide median was +$12.80/month (+13.0%) across 19 cities, the steepest of any well-sampled state we can verify.
New Jersey — +$26.60/month (+12.8%) in JCP&L territory
Jersey Central Power & Light's June 2026 Basic Generation Service supply increase moved Trenton, Toms River, and Hamilton Township from roughly $208 to $235/month.
New Jersey is a useful illustration of why statewide averages mislead: the state's overall median was only +$6.60/month (+2.9%), because PSE&G territory ran slightly negative over the same window while JCP&L customers absorbed a double-digit increase. Same state, opposite directions.
Louisiana — +$10.60/month (+7.6%)
Eight Entergy Louisiana cities including Monroe moved together on the LPSC's May 2026 typical-bill benchmark—a clean example of one regulatory action propagating across an entire service territory at once.
An increase we deliberately did not headline
Lansing, Michigan shows +$21.70/month (+21.7%), but the archive reason records a board-approved 6.8% residential increase effective October 1, 2025. The gap is a structural change in how we model BWL's time-of-use schedule, not a 21% price hike. We're listing it here rather than in the ranking because the honest number is 6.8%.
Where bills fell — and Connecticut's real decline
The most interesting finding is that 1 in 5 records went down, concentrated heavily in one region.
Connecticut posted a median decrease of −$54.60/month (−21.2%) across six cities—the only well-sampled state with a decline of that magnitude, and it is thoroughly documented:
- Hamden (United Illuminating): −$66.25/month (−25.3%). A May 1, 2026 Rate Adjustment Mechanism delivery decrease of about 4.88¢/kWh, plus a July 1 Standard Service supply decrease from 13.70¢ to 11.95¢/kWh.
- Bristol, Fairfield, Greenwich, Manchester, and Meriden (Eversource): −$54.60/month (−21.2%). The same pattern—a May RAM delivery decrease near 4.4¢/kWh and a July 1 Standard Service decrease from 12.64¢ to 11.58¢/kWh—bringing the all-in volumetric rate to about 19.34¢/kWh.
Two separate utilities, two separate mechanisms, moving the same direction in the same quarter. Connecticut households went from among the most expensive in our dataset toward the middle of the pack in a single summer.
Helena, Montana fell −$43.99/month (−29.9%) when NorthWestern Energy's quarterly PCCAM supply adjustment dropped the bundled energy rate to 9.9166¢/kWh in May 2026.
The common thread: every large decrease came from a fuel or supply-cost adjustment, not a base-rate cut. Delivery charges are sticky and rise through formal rate cases. Supply charges float with wholesale energy markets and can fall fast. That asymmetry is the single most useful thing in this dataset—the part of your bill most likely to drop is the part nobody votes on.
State-level medians
Ranked by median dollar change, states with at least five comparable records:
| State | Records | Median change | Median % |
|---|---|---|---|
| Maine | 5 | +$18.65 | +6.9% |
| Mississippi | 7 | +$18.00 | +12.2% |
| Vermont | 5 | +$15.96 | +7.3% |
| Ohio | 19 | +$12.80 | +13.0% |
| Michigan | 20 | +$10.81 | +5.4% |
| Louisiana | 10 | +$10.60 | +7.6% |
| Kansas | 11 | +$9.70 | +6.6% |
| New Jersey | 25 | +$6.60 | +2.9% |
| South Carolina | 6 | +$4.68 | +2.8% |
| Oregon | 5 | +$2.95 | +1.9% |
| Pennsylvania | 13 | +$1.26 | +1.0% |
| Tennessee | 6 | $0.00 | 0.0% |
| Nevada | 11 | −$0.28 | −0.2% |
| Connecticut | 6 | −$54.60 | −21.2% |
Washington is omitted; see the baseline-correction discussion above.
Tennessee at exactly $0.00 is not an error. All six sampled cities are TVA distributors, and TVA's wholesale rate held flat through the window—a reminder that public-power structures can behave very differently from investor-owned utilities in the same period.
What this means if you're budgeting
- A ~6% annual drift is a reasonable planning assumption, but it is a median. A quarter of records moved more than 10% in either direction.
- Supply and delivery move independently. Your bill can rise on delivery while supply falls. Both lines deserve attention.
- Territory matters more than state. New Jersey's JCP&L and PSE&G customers had opposite experiences in the same window.
- Check your own city rather than trusting a national average. Start with our city comparison tool or browse utility rates by state.
Methodology and limitations
How the comparison works. For each city and utility we take the earliest archived value as the baseline and the current published value as the endpoint, so a city with multiple updates is never double-counted. Electric bills are modeled as base charge + (energy rate × 1,000 kWh).
Reproducibility. The analysis is a committed script (scripts/analyze-rate-history.py) run against public archive files. The numbers here can be regenerated.
Limitations we want to be explicit about:
- This is a 14-week window, not a calendar year. Rate actions with earlier effective dates are captured, but the sample is not an annual index.
- The sample is not random. We re-verify cities when we have reason to think something moved, which biases the archive toward cities that changed. The 77.4% increase share describes our archive, not all U.S. utilities.
- Electric dominates. With 19 water, 18 sewer, and 4 trash records, we have no basis for claims about non-electric utility trends and are not making any.
- Modeled, not billed. These are standardized comparisons at 1,000 kWh, excluding taxes, local surcharges, and usage-specific tier effects.
- Some contamination likely remains. We caught the Georgia seasonal artifact and the Washington baseline issue by hand. Others may survive automated filtering.
Corrections welcome. If you're a utility or regulator and a figure here misrepresents your rates, tell us and we'll fix it and note the change.
Rates verified against published tariffs, commission orders, and utility rate schedules as of August 29, 2026. Every city page cites its source document and verification date.