A very strong El Niño trims about 5% from Northeast heating degree days on the Eastern Pacific analog, worth roughly $100 to $175 a household. Breaking even on a 50% price rise would take a 41.7% cut. And the widely quoted +31.3% forecast already has the warm winter in it.
Diesel closed the week of 14 September at $6.285 a gallon, up $2.41 from a year ago. Heating oil tracks it. Every forecast of a brutal winter for Northeast households is, so far, correct.
Into that arrives a very strong El Niño, and with it a reassuring story: a mild winter is coming, and it will take the edge off. The reassurance is real but small, and one version of it is simply a double count.
Here is the sentence that matters, from the National Energy Assistance Directors Association's winter report of 14 September:
"Across all fuel types, the increase in heating costs is being moderated by expectations for a warmer winter due to El Niño, which typically brings warmer-than-normal winter temperatures to much of the United States, reducing the amount of energy households need to heat their homes."
NEADA's +31.3% for heating oil is the number after El Niño has been applied. The word is "moderated." Without the warm winter, the projection would be higher. So anyone quoting +31.3% and then adding "but El Niño should help" is counting the same benefit twice.
We pulled heating degree days from NOAA station data for twelve cities, against the 1991-2020 normals, for the three strong El Niño winters with usable records. The appropriate analog set needs care. NOAA replaced ONI with RONI as its official ENSO index in February 2026, and under RONI the rankings change: 1982-83 peaks at 2.40, 1997-98 at 2.28, 2015-16 at 2.25. The developing 2026 event is strongly east-weighted, with August Niño 1+2 at +3.4°C against Niño 4 at +0.1°C, and only 1982-83 and 1997-98 are verified Eastern Pacific events. 2015-16 was substantially more westward-weighted than 1982-83 and 1997-98, making it a weaker analog for the presently east-weighted event, so we report it separately rather than folding it in.
| City | 1997-98 | 1982-83 | 2015-16 | EP composite |
|---|---|---|---|---|
| New York | −5.8% | −7.5% | −19.5% | −6.7% |
| Boston | −3.8% | −7.8% | −15.0% | −5.8% |
| Minneapolis | −5.6% | −6.2% | −14.7% | −5.9% |
| Chicago | −5.2% | −3.0% | −11.1% | −4.1% |
| Portland ME | −2.1% | −2.8% | −10.0% | −2.5% |
| Dallas | +8.6% | +10.4% | −26.9% | +9.5% |
| Atlanta | +21.3% | +13.1% | −20.9% | +17.2% |
Two things fall out of that table, and neither is the story you have been told.
The Northeast benefit is modest. Across Boston, Portland and New York the Eastern Pacific composite is −5.0% in heating degree days for October through March. The single best city on that composite is New York at −6.7%. These are not the enormous mild winters of memory.
The South gets colder, not warmer. Atlanta ran 21.3% above normal in 1997-98 and 13.1% above in 1982-83. Dallas ran above normal in both. This is the canonical El Niño southern storm track doing exactly what it is supposed to do, and for a household in Georgia it means a higher bill, not a lower one. El Niño does not abolish heating demand. It moves it.
One honest note on the biggest number in the table. Dallas at −26.9% in 2015-16 is the largest single reduction we measured anywhere. That winter was substantially more westward-weighted than 1982-83 and 1997-98, making it a weaker analog for the presently east-weighted event, and we are not treating it as representative. We report it because it is the maximum, and because even the maximum is not enough.
A 50% price rise needs a 33.3% consumption cut to leave spending unchanged, if the whole bill moves with consumption. Real bills do not work that way. Part of the oil goes to water heating and other non-weather-sensitive uses. Only the space-heating portion responds, so the required cut on that part is larger:
required cut = p / (1 + p) × (H + N) / H
EIA's Residential Energy Consumption Survey puts space heating at 80% of household heating-oil use. Run the numbers and the break-even requirement is:
| Fuel | Space-heat share | Cut needed at +31% | Cut needed at +50% |
|---|---|---|---|
| Heating oil | 80% | 29.6% | 41.7% |
| Propane | 70% | 33.8% | 47.6% |
| Natural gas | 68% | 34.8% | 49.0% |
The Eastern Pacific analog delivers about 5% in the Northeast. The requirement is 41.7%. That is not a near miss, and no plausible winter closes it. Even Dallas in 2015-16, at 26.9%, falls short of the break-even for a fuel it barely uses.
Modelled on 518.8 gallons per household per winter, the RECS 80% space-heat share and an 0.85 degree-day elasticity, here is the Northeast heating-oil household:
| Price scenario | Normal winter | El Niño winter | Saving | Shock offset |
|---|---|---|---|---|
| $5.535 (last official, 30 Mar 2026) | $2,872 | $2,774 | $97 | n/a |
| $5.66 (EIA STEO, Q4 2026) | $2,936 | $2,837 | $100 | n/a |
| +20% on the last official price | $3,446 | $3,329 | $117 | 20.4% |
| +50% | $4,307 | $4,161 | $146 | 10.2% |
| $10.00 a gallon | $5,188 | $5,012 | $176 | 7.6% |
So: roughly $100 to $175, or about 3.4% of the bill. Against a further 50% price rise it absorbs one tenth of the increase. Note the direction of that last column. The worse the shock, the smaller the fraction El Niño covers, because the weather saving stays roughly constant in percentage terms while the price runs away from it.
That is not nothing. On a household already paying $2,900 a winter, $150 is real. It is also not a rescue, and it should not be sold as one.
The relief is wildly unequal, and the reason is how much of each bill is actually weather-sensitive.
For a household whose main heating fuel is electricity, space heating is 24.5% of the annual electricity bill nationally, per RECS: $386 of $1,574. Regionally it runs from 40.4% in the Midwest down to 20.5% in the South. The other three quarters of the bill is water heating, cooling, refrigeration and everything else, and a mild January does nothing to it.
Electricity bills carry a second problem that weather does not reach.
In PJM, the grid operator covering thirteen states, capacity costs are allocated on a five-coincident-peak measure taken in June through September, and transmission on a single-coincident-peak measure. Neither has winter weather as an input. The 2028/29 capacity auction cleared in July 2026 at the price cap for the third consecutive year. Those obligations were fixed before this winter begins, and a warm January cannot reduce them.
How much of that is data centres is genuinely contested, and the honest answer is that both sides of the argument are right about different things. PJM's Independent Market Monitor found that including 17,071 MW of data-centre load in the 2027/28 auction raised its cost by $6.50 billion, or 65.5%, and by $23.1 billion across three auctions. But an independent analysis of the larger 2025/26 price surge attributed roughly half of it to a change in how capacity credit is calculated and another 41% to two Maryland plant retirements, not to data centres. Columbia's review of the literature concludes load growth is not the primary driver of retail increases, pointing instead at transmission and distribution spending, which has risen 160% since 2003.
The distinction that matters here is simpler than the attribution fight. Fuel costs fall in a mild winter. Capacity charges, transmission charges and fixed monthly customer charges do not. Cost allocation varies by grid operator and by state, so this is a PJM-specific mechanism rather than a universal law, but where it applies, the fastest-growing part of an electricity bill is the part El Niño cannot reach.
The 1997-98 winter is the usual reference point. In nominal terms heating oil averaged $0.987 a gallon that October through March. Converted to August 2026 dollars on CPI-U All Items, that is $2.045. The last official residential price, from 30 March 2026, re-based to the same month, is $5.615.
Real increase: +175%. Households are paying roughly 2.7 times what they paid in real terms for the same gallon.
Two cautions. A price comparison is not a bill comparison, because housing efficiency, burner efficiency and gallons burned per household have all changed since 1997. And 1997-98 was itself a warm El Niño winter arriving just before the 1998 crude collapse, so it is a flattering baseline rather than a neutral one. Comparing against it maximises the apparent increase.
Confirms: EIA's residential heating oil survey resumes on 7 October 2026. If the first Northeast observations land near the STEO forecast of $5.66 a gallon, and October through March degree days across Boston, Portland and New York finish within a few points of the −5% composite, then the cushion is worth roughly $100 to $150 and the shock is not offset.
Refutes: a Northeast heating season finishing more than 25% below the 1991-2020 normal, which would approach the break-even threshold and would mean the Eastern Pacific analogs badly understated the teleconnection. Also refuting: heating-oil prices settling back toward $4 a gallon, which would make the weather question largely moot.
Next checkpoints: CPC ENSO Diagnostic Discussion on 8 October 2026, EIA's Winter Fuels Outlook in October, and the first weekly heating oil price of the season on 7 October.