Seven Numbers That Decide If a Move Pays for Itself

For most of a year my move sheet had one row on it. Rent. Two towns, one number each, a gap of a bit over $700 a month, and I caught myself describing that gap to people as though it were the answer.

It was not even the question. The cheaper town charged more per kilowatt-hour, sat in a county that had just refreshed its assessed values, and fell in a different insurance rating territory. None of that was hidden. All of it is published somewhere. I just had not built a row for any of it.

So here is the frame I use now: seven numbers, each an annual dollar figure, each pulled from the agency that publishes it. Seven is not a magic count. It is how many lines move enough between two American towns to change a decision.

Why pay is not one of the seven

If your employer sets salary by location you already know that delta — ask, get it in writing, and put it at the top of the sheet as a given. It is not something you research. It is something you are told.

What you are not told is the spending side, and that is where a move is won or lost. The Bureau of Labor Statistics' Consumer Expenditure Survey put average annual expenditures for 2024 at $78,535 per consumer unit, against average income before taxes of $104,207. I pulled both through the BLS public API rather than the PDF tables, because an endpoint is quotable and a PDF is not:

api.bls.gov/publicAPI/v1/timeseries/data/CXUTOTALEXPLB1101M
api.bls.gov/publicAPI/v1/timeseries/data/CXUINCBEFTXLB1101M

The series pattern is CXU + an item code + LB1101M for all consumer units. Housing is CXUHOUSINGLB1101M, transportation CXUTRANSLB1101M, healthcare CXUHEALTHLB1101M, utilities and public services CXUUTILSLB1101M. Read on 19 August 2026, 2024 was still the newest year available. Two things worth knowing before you try it. The keyless v1 endpoint has a small daily quota per IP address, and once it is spent every call returns the same sentence instead of data — the daily threshold for total number of requests allocated to the user with registration key has been reached — so register for a free key at the BLS developer page. And when the quota is gone, the same survey sits in flat files at download.bls.gov/pub/time.series/cx/ with no key at all; cx.data.1.AllData is a 121 MB text file holding every series and year, which you filter yourself.

Where the money actually moves when you move

Take those 2024 averages apart and the geography shows up immediately.

Housing plus transportation plus healthcare is $45,781 — 58.3% of everything spent. Food is a large number and an almost immovable one. A gallon of milk does not cost triple in Boston.

The Bureau of Economic Analysis measures precisely that difference. Its regional price parities, current release 19 February 2026 covering 2024, put the highest state overall at California, 110.7, and the lowest at Arkansas, 86.9 — a spread of about 27%. Now the housing-rents component of the same estimate: 154.3 in California, 54.2 in West Virginia, with the District of Columbia at 155.0. Close to a factor of three, inside a national index whose overall range is a bit over a quarter.

That is the whole argument for building the sheet around shelter and letting groceries go. The next RPP release is 10 December 2026.

The seven lines, and who publishes each one

Lines 1 through 3 carry the spread. Lines 4 through 7 carry the surprises, and two of them get forgotten entirely. Health coverage is priced by rating area rather than by state — each state draws its own, usually as groups of counties, so a two-hour drive can change the premium; the healthcare.gov preview will price a county for you without an account. Getting back is the row nobody budgets: count the trips you genuinely cannot skip in a year and price each one in both directions.

One line swamps the other six

Shelter needs two different numbers, answering different questions.

The American Community Survey gives you the structure of a housing market: what the typical unit costs against local income, how much of the stock is owner-occupied, how old it is. But local ACS data is a five-year average. The current file, 2020-2024, released 29 January 2026, has its center of gravity around 2022 — I worked through that lag and its geography traps in four Census tables worth pulling. A town whose rents jumped in 2025 will not show it there.

The ACS gives you the shape; live listings give you the level. Pull ten current listings matching the unit you would actually rent — same bedroom count, same rough part of town — and take the median of the asking rents, not the mean. One relisted luxury build drags a mean and barely touches a median. Then confirm the listings are real before you trust their prices, which is its own set of five lookups.

One caution: api.census.gov now rejects keyless requests. I sent one on 19 August 2026 and got back a page headed "Missing Key." Signup is free; until you have a key, work in the browser at data.census.gov.

Electricity, worked out loud

Line 4 is the easiest of the seven to compute exactly, so it is a good place to watch the method.

EIA's Electric Power Monthly, Table 5.6.A, publishes average residential price by state. Data for May 2026, released 23 July 2026, read on 19 August: Ohio 19.52 cents per kWh, Tennessee 14.47, Texas 16.44, Idaho 12.35, against a US average of 18.44. Twelve months earlier the US average was 17.37, so the national residential price is up 6.2%.

The arithmetic, in the open, for a household using 12,000 kWh a year:

• Ohio: 12,000 × $0.1952 = $2,342
• Tennessee: 12,000 × $0.1447 = $1,736
• Difference: $606 a year

And now the correction that half the value of this sheet comes from. The 12,000 kWh is an assumption I supplied. Only the prices came from EIA, and usage is the half nobody checks. EIA puts the average U.S. residential customer at 10,791 kWh a year, with Louisiana's average customer buying 14,774 and Hawaii's 6,178 (How much electricity does an American home use?, 2022 figures). Pick the wrong end of that range and the annual answer is off by hundreds of dollars before the two towns are even compared. A Tennessee house runs air conditioning more hours than an Ohio house; an Ohio house may burn gas for heat, a separate bill that never appears in cents per kWh at all. So the honest row is price from EIA, usage from a benchmark such as the Residential Energy Consumption Survey, and a note on the cell recording which half you actually know.

Better still, ask the seller or the landlord for twelve months of bills for that meter. One real year of kilowatt-hours settles the usage half outright, and asking costs nothing before you have signed anything.

Deltas, not levels — plus a column for how far you trust each

The sheet is two columns of annual dollars and a third that subtracts them. Everything annual. Never set a monthly rent against an annual tax bill — that is the error behind those spectacular "I'll save $60,000" spreadsheets.

Then add two columns most people skip.

Confidence. Mark every cell quoted (an insurer or utility gave you this figure for this address), published (an agency prints it, but for a geography wider than your address), or estimated (you built it from a benchmark). A $4,000 delta made of quoted cells is a decision. The same $4,000 made of estimated cells is a research plan.

Date pulled. All of these sources revise. BEA republishes in December, EIA refreshes monthly, an insurance quote expires in weeks. Write the date beside the number or you will not be able to reconstruct your own sheet in three months — the same discipline FBI crime figures forced on me, since those get revised quietly as late submissions arrive.

The cell I copied into both columns

The most expensive habit on that early sheet was not a wrong number. It was a placeholder. I had exactly one insurance figure, the premium I was already paying, and it went into both columns because no quotes existed yet and an identical number felt neutral.

It is not neutral. Any number identical in both columns subtracts to zero and disappears out of the delta, so the sheet was quietly asserting that insurance costs the same in both places. That is the one thing insurance reliably does not do. A premium is built from rate filings a company makes state by state, from a rating territory inside that state, and from a deductible that may not even be the same kind of thing in the two places — Texas's insurance regulator puts it plainly, that a deductible "may be a specific dollar amount or a percentage" (TDI, what to know about deductibles). A percentage-of-dwelling wind or hail deductible and a flat $1,000 are not two versions of one row.

If you cannot get a real figure for a line yet, leave the cell empty and put a question mark in the delta. A blank cell nags at you. A placeholder lies to you.

What the delta has to pay back before the move is free

The annual delta is half the calculation. Against it sits everything the move costs once: the truck, the deposits, the overlap month of two rents, plates and licenses at the far end, and the working days you do not work.

Payback in months = one-time costs ÷ (annual delta ÷ 12).

Total your own; what follows is arithmetic, not a benchmark, since a one-time total swings with distance, with how much you own and with whether a vehicle has to be shipped or re-registered. A $9,000 annual delta absorbs a $12,000 move in sixteen months. A $2,000 delta takes six years — longer than most remote contracts last — and at that point you are not making a financial decision. You are making a life decision that happens to cost $2,000 less a year. Both are fine. Mistaking one for the other is not.

Fill two cells tonight

Take your two leading candidates. Open EIA Table 5.6.A, find the residential price for each state, multiply both by the same usage figure, and write the difference down with today's date beside it.

One lookup and one multiplication, and it is the first cell on the sheet that is measured rather than assumed. Then go get an insurance quote on both addresses, because that is the row most likely to reorder your list.

Sources, read 19 August 2026: BLS Consumer Expenditure Survey 2024 tables, pulled through its public data API; BEA regional price parities, 19 February 2026 release; EIA Electric Power Monthly Table 5.6.A, data for May 2026, released 23 July 2026; data.census.gov for ACS tables B25064, B25077, B25103, B25044 and B08301; healthcare.gov plan preview and the CMS marketplace public use files; Texas Department of Insurance on deductibles.

The last two steps are not on the sheet and cannot be put there. Which of these lines your household actually spends against — a two-car household and a one-car household are not reading the same table — and how long a payback you are willing to sit inside. Neither is arithmetic, and neither gets answered here; pricing your return or writing your policy is work for someone licensed to do it, once the address is real. What the sheet is good for is smaller and duller: every cell names the agency the number came from, and a cell that cannot is not finished.