Most heat pump calculators use the wrong gas price

The error is easy to make, hard to see, and it moves the answer in most of the country.

Start with a number that looks wrong. In the first data pull for this site, residential natural gas in Georgia came back at $33.28 per thousand cubic feet. Texas came back at $30.48. Massachusetts at $31.39.

Heating season residential gas in the United States generally runs somewhere between $10 and $18 per thousand cubic feet. Three states sitting at double the top of that range is not an unusual month. It is a systematic problem.

Where the number comes from

The U.S. Energy Information Administration publishes a monthly residential natural gas price for each state. That figure is not a tariff and it is not what any single household pays per unit. It is an average, computed as total residential revenue in the state divided by total residential volume delivered.

For most of the year that division produces a sensible number. In July it does not.

A residential gas bill has two parts. There is a volumetric charge for the gas you actually burn, and there is a fixed monthly customer charge that you pay whether you burn anything or not. The customer charge covers metering, billing, and the cost of keeping a service line connected to your house. It is typically somewhere between $10 and $25 a month depending on the utility.

In January, a household in a cold state might burn 15 or 20 thousand cubic feet of gas. The fixed charge spread across that volume is small, maybe a dollar or two per thousand cubic feet. In July, that same household burns gas only for cooking and hot water, perhaps one or two thousand cubic feet, and some households with electric water heating burn almost nothing at all. The fixed charge has not changed. Divided by a much smaller volume, it now dominates.

The state total behaves the same way. Revenue falls in summer, but it does not fall as fast as volume, because a large share of summer revenue is fixed charges. Revenue divided by volume therefore rises, sometimes to double the winter figure.

What that does to a heat pump comparison

The comparison between a heat pump and a gas furnace turns on the price of electricity relative to the price of gas. Overstate the price of gas and you understate the cost advantage of the furnace.

Here is what the correction did on this site, moving from a single latest month to a December through February mean:

State	Summer month, $/therm	Heating season mean, $/therm	Overstatement
Georgia	3.21	1.50	114%
Texas	2.94	1.75	68%
New York	2.19	1.56	40%
Michigan	1.36	1.01	35%
Massachusetts	3.03	2.49	22%

Under a cold climate heat pump at HSPF2 11.0 against a 95% AFUE furnace, the number of states where the heat pump costs less to run fell from 19 to 4 once the correction was applied. The bug was not a rounding issue. It reversed the answer across a third of the country.

The second bug hiding behind the first

There is a quieter problem with taking the most recent available month, and it survives even if you happen to pull during winter.

States report to EIA on different lags. When this site first pulled data, most states resolved to May while California resolved to March. Two states compared on that basis are not being compared on the same footing, and a table sorted by price ratio is then sorting partly by reporting lag.

The fix for both problems is the same. Pick a fixed window, use identical months for every state, and exclude any state that does not have complete data for that window rather than filling the gap with an estimate. This site currently uses December 2025 through February 2026 and excludes six states on that basis.

Why the error is easy to miss

Nothing about the API response signals the problem. The field is called price, the units are dollars per thousand cubic feet, and the value is a plausible looking number. If you have not held a residential gas bill in your hand and thought about what the fixed charge does at low volume, there is no reason to distrust it.

The error also fails in a direction that most people building these calculators are not motivated to question. It makes gas look expensive and heat pumps look good, which is the answer a large share of the internet already expects. A bug that confirms the expected result gets fewer checks than one that contradicts it.

How to check it yourself

Pull the monthly series for any state that uses a lot of gas and plot it. You will see a sawtooth: low in January and February, rising through spring, peaking in July or August, falling back through autumn. That shape is not the commodity price moving. Wholesale gas is usually higher in winter, when demand is high, so the retail series is moving opposite to the underlying commodity. That inversion is the fixed charge, and once you have seen it in one state you will see it in all of them.

If you want the number that answers a heating question, use the heating months. The rest of the year is measuring something else.

How this was checked

I found this by printing the raw prices before trusting anything derived from them. Georgia came back at $33.28 per thousand cubic feet and Texas at $30.48, against a heating season range I expected to sit between $10 and $18. That gap was too large to be an unusual month.

I then pulled the full monthly series for several states and plotted it. The sawtooth is unmistakable once you look: low in January and February, peaking in July or August. Since wholesale gas is usually higher in winter, the retail series moving the opposite way told me the shape was coming from the fixed customer charge rather than the commodity.

Every figure in the comparison table above is the output of my own fetch script run twice, once with the original latest-month logic and once with the heating season average, against the same EIA series.