How it works

The question

Most comparisons stop at income tax, which is why they so often mislead. “Texas has no income tax” is true and frequently irrelevant: Austin homes cost more than Chicago homes, so the median Austin owner pays more property tax than the median Chicago owner despite a lower rate. A move can also mean going from no car to two, which no cost-of-living index captures.

So this site computes one thing: what you actually have left at the end of the year, in each city, and the difference between them.

in your pocket  =  gross salary
                 −  federal income tax
                 −  state income tax
                 −  local income tax
                 −  Social Security and Medicare
                 −  state disability and paid leave, where the state charges it
                 −  rent + utilities, or mortgage + property tax + utilities
                 −  cars and transport
                 −  food, phone, healthcare, everything else

answer          =  in your pocket THERE  −  in your pocket HERE

The answer can be negative, and often is. That is the point — a cheaper city and a pay cut pull in opposite directions, and only the arithmetic settles it.

The order matters

Each step feeds the next, and getting the sequence wrong is the most common way a relocation calculator produces a confident wrong answer.

1. Living costs, re-priced for the local area. These come first because a renter’s gas and electricity are already inside their rent figure and a buyer’s are not, so the housing step needs to be handed that slice before it can add itself up.
2. Housing — which produces your property tax and first-year mortgage interest.
3. Social Security, Medicare, and state disability where the state charges it.
4. State income tax.
5. Local income tax. Yonkers charges a percentage of your state tax, so it has to come after it.
6. Federal income tax — which uses steps 2 to 5, because state and property tax are deductible. That decides whether itemising beats the standard deduction, which changes what you owe.

Computing federal tax first — the obvious way to write it — would ignore the deduction and overstate federal tax in every high-tax state.

Federal tax and Social Security are the same everywhere

They are federal, so they do not vary by state. If you see them change between two cities in your result, it is because the salary changed, not the location. The detailed table shows a middle column at your current salary precisely so this is visible.

That split is a convention, not a law of nature. We work out the new city at your current salary first, and call whatever is left the pay effect. Doing it the other way round — the pay rise first, then the city — gives different numbers for each half, though the same total. There is no single correct way to divide a change between two causes that happen at once. This is the order we chose, and it is the order the middle column of the table shows.

One real exception: federal tax can differ between cities at the same salary, for people who itemise. State and property tax are deductible, so a high earner with a mortgage in New York pays less federal tax than an identical earner in Texas — but only up to a point. That deduction is capped at $40,400 ($20,200 filing separately), and the cap itself shrinks above $505,000 of income down to a $10,000 floor. Both are applied here. Most households take the standard deduction and never see any of it.

Housing

Every housing field is pre-filled and editable. If you rent, we use your rent. If you own, we amortise a 30-year fixed mortgage at 6.4% — the US average across 2026 Q2, from Freddie Mac’s weekly survey, refreshed with the rest of the data — add property tax at the effective rate — what is actually paid, after assessment ratios, homestead exemptions and caps — and compute first-year interest for the itemisation test.

Buying a house costs more than the mortgage and the tax on it. Roofs, boilers, gutters, a plumber, and the home insurance policy. The national statistics bundle all of that into one figure, and because this site throws away their housing number and rebuilds it from yours, that figure used to be thrown away with it. Buyers were paying nothing at all to keep the house standing.

It is now charged: about $4,600 a year at $100,000 of income and $5,000 at $200,000, reaching roughly $7,300 only above $320,000. (Those were quoted as $4,000 and $7,300 until August 2026 — the first was in 2024 money and the second was the top band’s figure attached to the wrong income.) The published figure averages owners and renters together, and renters pay none of this, so it is divided by the share of households who actually own before it is used. Renters are charged nothing — their landlord pays for the roof.

This is also where home insurance lives. The site used to say insurance was missing and call it the biggest remaining gap. That was the wrong shape: insurance is not a separate thing that was forgotten, it is one ingredient of a figure that was being discarded whole.

Only part of a big mortgage earns a deduction. The interest write-off reaches the first $750,000 of a loan, half that if you are married and filing separately. Borrow more and the extra interest is simply not deductible. This site used to deduct all of it, which in the expensive metros was a large error in the reader’s favour: a single buyer in San Jose on $300,000 borrows about $1.3M at our default home price, and was being shown roughly $10,000 a year too little federal tax. At $500,000 it was around $18,000. Nothing changes for a normal home in a normal city — a Chicago buyer stays under the limit at every salary this site is built for.

The rent we start you with is sized to your household and scaled to your income, and both parts matter more than they sound.

• Size. The obvious figure to use is the metro’s median rent, and it was the wrong one: it is a median across every rented unit in the area, so a single person and a family of four were quoted exactly the same rent. We now use the local median for the number of bedrooms your household implies — one for the adults, another for every two children.
• Income. That median is paid by a household earning roughly the local median income, and renters earn less than that. On a $150,000 salary it worked out at 11% of pay in Chicago, which nobody at that income pays. So the local figure is scaled by a national curve built from what renters in each income band actually spend on rent. The curve crosses 1.0 near $55,000 — about the typical renter’s income — and rises more slowly than income above it, because housing takes a falling share of a rising income.

Buying works the same way. The metro median home value is what the median owner owns, so a high earner was being quoted a cheaper house than they would buy — and, because property tax is charged on the price, a smaller tax bill too. The starting price now scales with income as well.

Both scalings are anchored to the local median owner or renter income, and this detail matters more than it sounds. A single national multiplier looks right until you apply it somewhere expensive: San Francisco’s median home is already owned by high earners, so scaling it by “what a $150,000 buyer purchases nationally” put that household at $1.5m — a third above the local median, while earning below the local median owner. Anchoring locally makes the multiplier exactly 1.0 for the household the median actually describes, which is the only value it can correctly have.

What stays national is the elasticity — how sharply housing spending rises with income, which is a behavioural regularity rather than a local fact. The local price is left completely untouched, so the difference between two cities, which is the one thing this page exists to measure, survives at full strength.

Mortgage principal counts as money out. It builds equity rather than vanishing, but the headline is cash in your pocket, and principal is cash that left it. This makes owning look slightly worse than it is in wealth terms.

Cars

This is the piece most comparisons miss. A cost-of-living index measures prices, not quantities. Moving from Manhattan to Austin does not make your car cheaper — it makes you buy two. Petrol being cheaper in Texas is irrelevant if you went from zero cars to a pair.

So transport is built from a car count, not a price index. The default comes from how many vehicles per adult a place actually has, multiplied by the adults in your household, and you can change it. Each car costs what US households actually spend per vehicle — purchase, fuel, insurance, maintenance and finance charges — at your income level.

State tax is not a small version of federal tax

States write their own rules and they are not always harsher. Three things this site used to get wrong, all of them in the same direction — against the reader:

• Eleven states take a disability or paid-leave contribution straight off the payslip, and none of it was counted. California’s is the big one: 1.3% of every dollar you earn with no ceiling at all, which is $3,900 a year on $300,000. It is now charged, and where the IRS treats it as a state tax it reduces your federal bill too.
• A head of household is not a single person. We used to tax them as one. California publishes its own schedule and gives them the married standard deduction; Maryland puts them on the married table outright; Kansas quietly adds a second exemption on top of the first. We have now read the actual form for every one of the 42 states that tax wages, and 30 of them give a single parent something the single schedule does not. Every one of those was being charged too much — from $75 a year in Alabama to $4,046 in Hawaii, and never the other way.
• Allowances that shrink as you earn more. Several states quietly take back the deduction or exemption they give you, and we used to hand it out at every income. Wisconsin’s disappears completely by $136,000. Connecticut’s is gone above $44,000, and it goes in $1,000 steps rather than smoothly — a dollar over a band edge costs a whole $1,000 of exemption. Colorado’s is the harshest shape of the lot — above $300,000 the standard deduction drops from $16,100 to a $1,000 floor in a single step, worth about $660 a year for a single filer and $1,330 for a couple. South Carolina, Maine, Minnesota, Maryland, Rhode Island and Alabama all do a version of it, and Oregon and Utah do it to a credit rather than to an allowance. These now shrink the way the state says they do — which means the tax we show for higher earners went up, and that is the point: the answer was flattering places it should not have been.
• Seven states changed the law after our tax table was printed. The table comes out once a year, in February, and states pass laws all spring. South Carolina rewrote its income tax outright. Arkansas, West Virginia and Utah cut their rates back to January. Georgia moved three things at once. Every one of them left us charging too much, and every one was found only by opening the state’s own publication and comparing it line by line.
• California lets you itemise on the state return whether or not you did federally, with no cap on property tax and a mortgage limit of $1,000,000 rather than $750,000. For a San Jose buyer that is around $6,800 a year. We only count the two things this site knows — property tax and mortgage interest — so if you also give to charity or have large medical bills, you will do better than this says.
• Fifteen states and Washington DC pay you for having children, on top of the federal credit, and until August 2026 not one of them was counted. New Mexico’s is now: it pays $637 a child below $25,000 of income, $424 up to $50,000, and something at every income above that, and it is refundable — paid out even when you owe nothing. A single parent on $50,000 in Albuquerque with one child was being shown $500 of New Mexico tax where the figure is really $172. The other fourteen are still missing, and each of those states now says so in the list at the bottom of this page.

Every state’s rates and allowances have now been read off that state’s own publication — all 42 of the 42 that tax wages. Twenty-four of them were wrong. 13 are on last year’s published figures because the state has not released this year’s, and those are named below. The data page shows, for every location, the date its state was checked, and that date links to the document it was checked against.

A link to a document is not proof there is anything in it. New Mexico spent a release cited to its own revenue department’s “Personal Income Tax Rates” page, which turns out to be a heading, a menu and a footer with not one figure anywhere on it. Nothing noticed, because everything that checked the citation was really checking the web address. New Mexico’s figures now cite the department’s own rate table and return instructions, and each citation carries the words copied out of the document — which a test compares against the numbers actually shipped, so a citation to the wrong page fails as loudly as no citation at all. The other states are still cited by address alone, and that is being worked through.

14 states now let a homeowner claim mortgage interest and property tax on the state return, and we calculate all of them. Three more do something different with the same idea: New Jersey relieves property tax without itemising at all — the only relief here a renter can claim, at 18% of rent — Wisconsin gives a credit for mortgage interest while ignoring property tax entirely, and Illinois credits 5% of your property tax until your income passes $250,000 — $500,000 for a couple — when it stops dead.

The “What this gets wrong” list below names every state with a rule we know about and do not yet model, and says which way each one runs.

What the state calculation counts, and what it does not

A short version of the two lists, because “state income tax” hides a great deal. Counted: every state’s rate schedule and the schedule a head of household actually files on; standard deductions and personal and dependent exemptions, including the ones that shrink or vanish as income rises; itemised deductions in the 14 states that allow them, for property tax and mortgage interest only; the earned income credit in 23 states; New Mexico’s child credit and its low- and middle-income exemption; disability and paid-leave contributions in eleven states; the states that tax a couple as two people; and city income taxes.

Not counted: almost every other state credit. Credits and rebates that turn on your age, your medical bills, childcare receipts, veteran status, adoption or disability are not calculated at all, because this site never asks about any of them — it asks for a salary, a household and a housing situation. The child credits of fourteen more places are missing for a different reason: they could be calculated and are not yet. Every one of these means the same thing when it applies to you — the tax shown is higher than what you would really pay, so the city that levies it looks worse than it is.

What year the dollars are in

Today’s. Every cost figure behind this site was measured in 2024, the most recent year the federal surveys cover, but the tax rules are 2026 and the salary you type is 2026. Subtracting old costs from a current salary made the money left over look better than it is.

So the 2024 figures are brought forward to current prices. Four separate official measures, because they have not moved together since 2024: rent is up 7.1%, everything you buy 6.1%, and house prices 3.0%. Tax rules are already current and are left alone, and so is your salary.

This matters more than 6% sounds. Money left over is what survives after subtracting a big number from another big number, so an error in the costs lands almost undiluted on the answer: for a Chicago renter on $100,000 it was 13% of the result, and 27% for a buyer.

Where your gas and electricity bill lives

In the housing line, not in the spending one — and this is worth a minute because it was wrong for a long time.

The rent figures come from the Census, and the Census measures gross rent: the rent itself plus the electricity, gas, water, sewer and heating the tenant pays. So a renter’s energy bill is already inside the rent number on this page. This site used to subtract a full national utilities bill on top of that as well, charging renters for the same thing twice — for a couple, around $2,700 a year in Chicago and $4,000 in New York. A family of four is charged more of these and one person less, because the bill follows the household, so the double count was bigger for a family: near $3,700 and $5,700.

A mortgage covers no such thing, so buyers are charged those utilities separately. Either way the line reads rent + utilities or mortgage + utilities, and either way it is counted once.

The phone bill is the exception. It sits in the same national statistic as gas and electricity but it is not part of anybody’s rent, so it stays with the ordinary spending below. It also gets a different price index: what you pay for electricity is intensely local, while a mobile contract costs much the same everywhere.

Everything else you spend

Food, healthcare, the phone bill and the rest come from what US households at your income actually spend, then re-priced for each city using federal regional price levels.

Two adjustments are worth knowing about, because both were bugs before they were features:

• Your basket travels with you. It is chosen once, from your current income, and simply re-priced in the new city. Choosing it separately per city let the survey’s income-band boundaries leak into the answer, producing a phantom five-figure “saving” on food and healthcare that was an artefact of where the statisticians drew a line.
• It is scaled to your household size. Households in the $150k–$200k band average 3.1 people, so a single person was being charged for a family of three. Scaling uses the square-root rule that the OECD uses: needs grow with household size, but not proportionally — two people do not need two fridges.
• It slides smoothly between income bands. The survey publishes nine income bands. Reading your band off the bottom of the one you fall into made spending jump the moment your pay crossed a line: a $1 raise from $199,999 to $200,000 took $14,839 off what you had left, and moved the gap between two cities by $1,119. Each band’s figures are now placed at the average income of the households in it, and anything in between is read off the line joining them. Nothing is invented — the same nine published points, read at the income they were measured at.

Sales tax

There is no sales tax line, and there should not be one. The spending figures above are what households actually handed over at the till, and the government agency that collects them says so plainly: an expenditure is the transaction cost including sales and excise tax. Where somebody reported a price without the tax, it gets added before the figures are published.

This site used to charge sales tax again on top of those figures — once inside the grocery bill, once beside it. That was wrong and it is gone.

What is lost by removing it: the basket carries whatever sales tax the surveyed households paid, which is a national blend. So moving from Oregon, which charges none, to Louisiana, which charges the most at 10.11%, no longer shows any sales tax difference at all. Doing that properly means taking the average out of the basket and putting the local rate back in, and the survey does not publish how much is in there per category. A missing difference of a few hundred dollars is a smaller error than charging the whole thing twice, so this line waits for data that can carry it. The published rates are still in the data browser.

What the form starts you on

The salary box fills in with what a full-time worker is actually paid in the city you picked — and it is one of the least national numbers on this site. The same measure runs from $43,516 in Eagle Pass, TX to $118,050 in San Jose, CA, in today’s money. Each column fills in from its own city, so the two boxes usually start on different figures, and both are yours to change: type your pay on the left and the offer on the right.

Before either city is chosen there is nothing local to use, so the box opens on the US median of $NaN. The published figure is $61,657 — Census ACS 2024, table S2001, median earnings for full-time, year-round workers, the same release as everything else here — and like every other 2024 figure on this site it is brought forward to today’s money, which is where the difference comes from.

Earnings for one worker, not income for a household. The box asks what one person is paid, so a household figure would answer a different question with a much bigger number. Median rather than mean, because earnings have a long tail at the top and the mean describes almost nobody. The Census publishes this for whole metros and for states, but not for the part of a metro inside one state — so the 43 cities that cross a state line start both sides on the same figure, even though their rent and house prices are split by state.

The starting figure matters more than a starting figure usually does, because rent and home price both scale with it: a national salary in a cheap metro quoted a house and a rent that nobody there on that pay would be looking at. It was $150,000 until August 2026 — roughly the 90th percentile of American full-time earnings — then the national median, and now the local one.

Credits, and who counts as an earner

Two federal credits are applied. The Child Tax Credit is $2,200 per child under 17, partly refundable, phasing out above $200,000 (or $400,000 filing jointly). The Earned Income Tax Credit is fully refundable and worth up to $8,231 — for a low-income family with children it is the largest single figure in the whole federal calculation. It was missing entirely until August 2026, which understated what a household on around $18,000 with two children actually has by $7,316 a year.

The children question asks for children under 17 because that is the Child Tax Credit’s own test. The EITC counts a slightly wider group — under 19, or under 24 in full-time education — so a household with an older teenager is credited less here than it is owed, never more. Living costs use the same number, so that household’s grocery bill is understated too.

Married couples are asked how many of them earn. Social Security is capped at the first $184,500 of each person’s wages, so a household on $300,000 owes $18,600 when two people earn it and $11,439 when one does. The engine used to apply that cap once to the household total, understating two-earner couples by $7,161. Where two earn, the split is assumed to be even; a lopsided split owes less.

What is still not asked: your age, whether each child meets the residency and relationship tests, and whether you have investment income. The childless EITC therefore assumes you are 25 to 64 (it is worth at most $664), children are assumed to qualify, and the $12,200 investment-income disqualification never bites. Around thirty states add a credit of their own on top; 23 of them are calculated here, and the list below says which are not and why.

States also give credits for children, and that is a separate thing. Fifteen states and Washington DC have one, and they are not a share of the federal credit — each has its own table, its own age limits and its own income limits. Only New Mexico’s is calculated here. In the other fourteen, a household with children is shown more state tax than it will really pay, and each of those states says so by name in the list below.

Metro areas that cross a state line

43 of the 438 locations here straddle at least one state border. “New York–Newark–Jersey City” is New York and New Jersey. Philadelphia is Pennsylvania, New Jersey, Delaware and Maryland. Washington is DC, Virginia, Maryland and West Virginia. Portland is Oregon and Washington — one of those has an income tax and no sales tax, the other the reverse.

For those metros the calculator asks which state you live in, and that answer drives the state income tax, the sales tax rate, and which city income taxes can reach you. Until August 2026 it did not ask: each metro was reduced to one state, so someone in Newark was quoted New York’s income tax and New York City’s resident tax — about $6,100 a year too much at $150,000, enough to reverse the verdict.

Housing follows the state as well. The Census publishes every table used here at summary level 311 — “metropolitan statistical area › state (or part)” — so a Newark household is quoted New Jersey’s figures rather than the whole metro’s. The difference is not small: median home value is $512,300 on the New Jersey side against $684,700 on the New York side, either side of the $614,200 metro figure that used to be quoted to both. The Indiana part of the Chicago metro is $1,142 rent against $1,453 in Illinois.

Cars move too, and that one surprises people. The New York part of that metro averages 0.486 vehicles per adult against 0.596 across the whole metro, because the five boroughs outweigh the suburbs — so a New York-side household is now started at no car rather than one. Change it if you have one.

A state part is a smaller sample than its metro, so the Census suppresses individual figures in it more often. Where that happens the whole metro’s number is used for that one figure, rather than throwing away the state’s good figures alongside the missing one. Price levels stay metro-wide throughout: BEA publishes those only for whole metros.

Pack, stay, or too close to call

The verdict — stay in one city, pack and move to the other, or neither — is the sign of that final difference and nothing more. It is about money only: this calculator knows nothing about the job, the people, or whether you want to live there.

The figure under it is a difference, not a balance. Each city panel prints its own leftover money; the verdict panel prints the gap between the two, and says which way it points before it shows the number. It used to be labelled “what is left over” over a figure ten times smaller than any real leftover, which is the kind of label that makes a reader distrust everything under it.

There is a third answer, and it is the honest one far more often than you would think. Neither word lights up until the difference is at least 5% of the salary you are paid today — $4,500 a year on $90,000, $7,500 on $150,000 — and never less than $2,500 however small the salary. Below that the page says too close to call and shows you the figure anyway.

Two different reasons sit behind that bar, and they bind at different incomes. One is precision: every cost here is a median for the local area and real households scatter widely around every one of them, so a small difference is inside the scatter rather than a result. The other is that this is a question about whether to move house. A difference has to be worth the move, and $1,000 a year on $90,000 is not — it is about $19 a week, against uprooting a life.

The bar used to be 1.5%, and that was too low for the second reason rather than the first. On $90,000 it called a winner over $1,350 a year. The arithmetic could see that gap; no household would act on it, and a verdict nobody would act on should not be printed as one. At 5% the same household needs $4,500 — a difference that is actually a decision.

A share rather than a flat amount, because the same dollars mean different things at different incomes: $3,000 is a fortnight’s pay on $80,000 and a rounding error on $400,000. A share of gross salary rather than of leftover money, because leftover is at or below zero for a great many real households and a percentage of zero is not a threshold at all. The salary you have now rather than the one on offer, so the bar does not move every time you try a different number in the box you came here to experiment with. And a floor in dollars underneath it all, because below about $50,000 the share shrinks back into the range where the local medians genuinely cannot tell two cities apart.

Share links

A link carries every input and the version of the data it was made with. When the underlying federal figures are refreshed, links already shared keep computing against the data they were created with. Whoever opens your link sees the numbers you saw, not different ones. Nothing is stored on a server — there is no database, and no account.

What this gets wrong

Every model is wrong somewhere. These are the places this one is wrong that we know about.

State by state

Each of these is a rule the state really has that we do not calculate. They are written out here rather than summarised, and each one says which way it runs — whether it means we are charging you too much or too little. This list is generated from the same data the calculator uses, so it cannot quietly fall out of date.

• Alabama. Alabama's itemised deductions here cover property tax, mortgage interest and the Social Security and Medicare tax withheld from your pay. Charitable giving and medical costs are not asked about and are not included, so Alabama tax shown here is higher than the true figure for anyone who has them. Alabama's dependent exemption steps down with Alabama adjusted gross income — $1,000 up to $50,000, $500 to $100,000, $300 above — and never reaches zero. Only the $1,000 figure is used here, so Alabama tax shown is lower than the true figure above $50,000, by the tax on $500 or $700 a dependent. Alabama also allows a deduction for the federal income tax you paid, which is not included, so the figure shown is higher than the truth in the other direction.
• Arizona. Arizona's dependent credit rose to $125 for a child under 17 for 2026 and we still apply $100, so Arizona tax shown here is $25 a year per young child too much. The same credit shrinks above $200,000 of income ($400,000 for a couple), which is not modelled, so above those incomes the figure shown is slightly low.
• Arkansas. Arkansas reduces the benefit of its lower brackets across a narrow band of income between about $94,700 and $97,600. That adjustment is not modelled, so tax shown for incomes inside that band is slightly lower than the true figure.
• California. California's young child tax credit is refundable and worth up to $1,189 for 2025 — once per return rather than once per child — to a household with a child under six and earnings low enough to qualify for California's own earned income credit. It is not calculated here, so California tax shown for such a household is higher than the true figure. California itemised deductions here cover property tax and mortgage interest only — the two figures this site knows. Charitable giving, medical costs and miscellaneous deductions are not asked about and are not included, so California tax shown here is higher than the true figure for anyone who has them. California's exemption credits shrink above $252,203 of income ($504,411 for a couple, $378,310 for a head of household) and that is not modelled, so California tax shown here is lower than the true figure above those incomes.
• Colorado. Colorado's child tax credit is refundable and reaches as much as $3,200 a child at the lowest incomes, tapering as income rises. It is not calculated here, so Colorado tax shown for a family with young children is higher than the true figure. Colorado's add-back is reduced by any state income tax you deducted on your federal return, which this engine does not model — so for someone who itemises federally, Colorado tax shown here above $300,000 is higher than the true figure. Colorado also adds back the federal overtime deduction from 2026, which is not modelled.
• Connecticut. Connecticut allows no itemised deductions and no standard deduction at all — the personal exemption is the only across-the-board subtraction — so a homeowner gets no relief for a mortgage here, and that is Connecticut's rule rather than a gap. Its property tax credit of up to $300 is not modelled, so Connecticut tax shown here is higher than the true figure for anyone who owns a home or a car.
• Georgia. Georgia created a nonrefundable child tax credit of $250 for each child under six, first claimable for 2026. It is not calculated here, so Georgia tax shown for a family with young children is higher than the true figure.
• Idaho. Idaho's grocery credit can instead be claimed on receipts for up to $250 a person rather than the flat $155 used here, so Idaho tax shown is higher than the true figure for anyone who keeps their receipts. It is also reduced for months spent on food stamps, incarcerated or outside Idaho, which is not modelled and runs the other way.
• Maine. Maine's dependent exemption tax credit is refundable and worth $315 a child, $630 for a child under six, up to $165,500 of income. It is not calculated here, so Maine tax shown for a family with children is higher than the true figure. Maine's personal exemption also phases out, from $341,000 of income for a single filer and $409,150 for a couple. That is not modelled, so Maine tax shown above those incomes is slightly lower than the true figure.
• Maryland. Maryland's refundable child tax credit reaches families with low incomes and a child under six or a child with a disability. It is not calculated here, so Maryland tax shown for those families is higher than the true figure. Maryland's itemised deductions here cover property tax and mortgage interest only. Charitable giving and medical costs are not asked about and are not included, so Maryland tax shown here is higher than the true figure for anyone who has them. Maryland's 2026 standard deduction for couples has not been announced — the legislature's own fiscal note says so — so the 2025 figure is used.
• Massachusetts. Massachusetts has no itemised deductions and no standard deduction — a personal exemption and a short list of named deductions are all there is. Its rent deduction of half a year's rent up to $4,000, its commuter deduction and its $440-per-child credit are not calculated, so Massachusetts tax shown here is higher than the true figure for renters, commuters and families.
• Minnesota. Minnesota's child tax credit is refundable and among the largest in the country at up to $1,750 a child, phasing out with income. It is not calculated here, so Minnesota tax shown for a family with children is higher than the true figure. Minnesota's dependent exemption also phases out, by 2 percentage points for each $2,500 of income above $244,500 single and $366,700 joint. That is not modelled, so Minnesota tax shown above those incomes is slightly lower than the true figure.
• Missouri. Missouri lets you deduct a percentage of the federal income tax you paid — 35% at low incomes, falling to nothing above $125,000 of Missouri income, and capped at $5,000. That is not calculated here, so Missouri tax shown is higher than the true figure below that income, by around $62 a year for a single filer on $80,000.
• Nebraska. Nebraska subtracts the full pre-cap state and local income tax from the federal itemised total even when the federal cap already reduced it, so a capped filer loses those dollars twice. That double subtraction is not modelled, so Nebraska tax shown here is lower than the true figure for anyone the federal cap reaches.
• New Jersey. New Jersey's child tax credit is refundable and worth up to $1,000 for each child under six at incomes below $30,000, falling in steps to $200 at $80,000. It is not calculated here, so New Jersey tax shown for such a family is higher than the true figure. New Jersey allows no mortgage interest deduction and no charitable deduction at all, so a homeowner gets relief for property tax only. The extra exemptions for veterans, for filers over 65 or blind, and for a dependent in full-time education are not modelled, so New Jersey tax shown here is higher than the true figure for anyone entitled to them.
• New Mexico. New Mexico's low income comprehensive tax rebate pays up to $839 to a household with modified gross income of $36,000 or less, on a table that widens with the number of exemptions, and it is refundable. It is not calculated here, so New Mexico tax shown for a household under that income is higher than the true figure. New Mexico gives a child day care credit, a property tax rebate, a refundable medical care credit and an extra $8,000 exemption to people aged 65 or over on modest incomes. This site never asks your age or what you spend on care, so none of them is calculated and New Mexico tax shown is higher than the true figure for the households they reach. New Mexico publishes a fourth rate schedule for a married person filing separately, and it is harsher than the single one this engine uses for that status — $136 a year more at $95,850 of taxable income. That schedule is not modelled, so New Mexico tax shown for a separate filer is lower than the true figure. New Mexico's itemised deductions here cover property tax and mortgage interest only, because they are the two figures this site knows. Charitable giving, medical costs and the rest of the federal schedule are not asked about and are not included, so New Mexico tax shown here is higher than the true figure for anyone who has them.
• New York. New York's Empire State child credit is refundable and pays $1,000 for each child under four and $500 for each child from four to sixteen for 2026. It is not calculated here, so New York tax shown for a family with children is higher than the true figure. New York itemised deductions here cover property tax and mortgage interest only. Charitable giving, medical costs above 10% of income and the job expenses New York still allows are not asked about and are not included, so New York tax shown here is higher than the true figure for anyone who has them. New York recaptures the benefit of its lower brackets from higher earners, so that above about $107,650 they pay their top rate on all their income rather than only the part above each threshold. That is not modelled, so New York tax shown here is lower than the true figure above that income — by roughly $481 a year at $150,000. New York City's school tax credit is not modelled either, which pulls the other way for city residents by $221 to $380 a year.
• Ohio. Ohio's personal exemption steps down with income — $2,400 up to $40,000, $2,150 to $80,000, $1,900 to $749,999 and nothing at all from $750,000 — and only the $2,400 is used here, so Ohio tax shown is slightly lower than the true figure above $40,000. Ohio's joint filing credit is measured on income after exemptions and this uses gross, which can land a couple one band low and so overstate their tax slightly.
• Oklahoma. Oklahoma's child tax credit is nonrefundable and available only below $100,000 of income, at 5% of the federal child tax credit or 20% of the federal child care credit, whichever is worth more. It is not calculated here, so Oklahoma tax shown for a family with children is higher than the true figure.
• Oregon. Oregon's kids credit is refundable and worth $1,000 for each child under six, phasing out above $25,000 of income. It is not calculated here, so Oregon tax shown for such a family is higher than the true figure. Oregon's subtraction for federal income tax is measured against a federal bill computed without deducting any state tax, which is exact for anyone taking the federal standard deduction and slightly high for anyone who itemises. Oregon caps the subtraction at $8,500 and a single filer clears that cap by about $70,000 of income, so for almost everyone the approximation never reaches the answer.
• South Carolina. South Carolina's dependent exemption is shown at the 2025 figure of $4,930. The state indexes it every December and the 2026 amount appears only in the 2026 return instructions, which are not published yet. The real figure will be slightly higher, so this errs against the reader.
• Utah. Utah's child tax credit is nonrefundable and worth up to $1,000 for each child aged one to five, tapering as income rises. It is not calculated here, so Utah tax shown for a family with young children is higher than the true figure. Utah's taxpayer tax credit phase-out thresholds and dependent exemption are the 2025 figures — the state has not published 2026 amounts. Utah indexes them upward, so the reduction starts slightly early here and the credit comes out slightly small.
• Vermont. Vermont's child tax credit is refundable and worth $1,000 for each child aged six or younger, in full up to $125,000 of income and in part to $175,000. It is not calculated here, so Vermont tax shown for such a family is higher than the true figure.
• Washington DC. The District's child tax credit is refundable and worth $1,000 for each child under eighteen for households within its income limits. It is not calculated here, so DC tax shown for a family with children is higher than the true figure.
• Wisconsin. Wisconsin's itemised deduction credit is 5% of qualifying deductions above the standard deduction, and it deliberately excludes every state and local tax — so a Wisconsin homeowner's property tax counts for nothing here, which is Wisconsin's rule rather than a gap. Charitable giving, medical costs and casualty losses also qualify and are not asked about, so Wisconsin tax shown here is higher than the true figure for anyone who has them.

Every other state has been read off its own publication with nothing outstanding that we know of — which is a weaker claim than the one this paragraph used to make, and the honest one. A rule nobody has looked for does not appear on a list of rules we know about. State credits for children were exactly that until August 2026: fifteen states and Washington DC have one, none was calculated, and no state said so. Fourteen of them now do. 13 of the 42 are checked against the state’s most recent figures rather than a 2026 document, because those states have published nothing for 2026 — they are named just below.

States still on last year’s numbers

States publish their new brackets and allowances on their own timetable, and many do not until the tax forms come out — which for 2026 means late this year or early next. This calculator has to answer today, so where a state has not published in full we fall back to its last published figures for whatever is missing, and say so here. It is rarely the whole state: Oklahoma’s 2026 rates come from the enacted law and only its allowances are last year’s, Oregon’s brackets come from its own 2026 withholding formulas, and Rhode Island’s figures are 2026 but off a form the state published with a draft watermark. Each entry below says exactly which figures are affected.

Prices rise, so last year’s bands are slightly narrow and last year’s allowances slightly small. That means these figures show a little more tax than you will really owe, not less — the error runs against us, not against you.

• Alabama. Alabama has published nothing for 2026. Its rates, exemptions and deduction charts here are the 2025 figures, all of which are fixed in statute rather than indexed.
• California. California indexes its brackets, standard deduction and exemption credits every autumn and has not published 2026 yet, so every California figure here is the published 2025 one.
• Connecticut. Connecticut's brackets, personal exemption and tax credit here are read off its 2025 Tax Calculation Schedule; the state has published no 2026 one. Connecticut does not index these for inflation, so they should carry unchanged unless the legislature moves them.
• Delaware. Delaware's standard deduction and personal credit here come from its 2025 resident instructions, the most recent published. Its bracket schedule is not in that document at all — Delaware prints the rates in a separate tax table — so the brackets shipped here rest on the annual compilation rather than on a Delaware document, and are the only figures in this state not read off the state's own paper.
• Idaho. Idaho's untaxed band of $4,811 for a single filer and $9,622 for a couple is the published 2025 figure; Idaho publishes the 2026 amount around December.
• Mississippi. Only Mississippi's rate is published for 2026. Its standard deduction and exemptions here are the 2025 figures, which are fixed in statute and were not changed by any 2026 law.
• New Mexico. New Mexico's brackets, deduction and exemptions here are 2026 figures — the state fixed them in statute for 2025 and forward and does not index them. Its child income tax credit amounts are the published 2025 ones, because New Mexico indexes that credit for inflation every year and has not published 2026. The 2026 credit will be slightly larger, so the tax shown here is slightly high.
• Oklahoma. Oklahoma's rates for 2026 come from the enacted law, but its standard deduction and exemptions here are the 2025 figures. They are not indexed, so they should carry unchanged.
• Oregon. Oregon has published no 2026 return forms. Its brackets, standard deduction and exemption credit here come from Oregon's own 2026 withholding formulas, but no 2026 head-of-household standard deduction exists anywhere, so that one figure is the published 2025 amount of $4,560.
• Rhode Island. Rhode Island's 2026 figures here come from a form the state published carrying a draft watermark. They are internally consistent and are the state's own, but they should be re-checked against the final booklet.
• South Carolina. South Carolina's dependent exemption here is the published 2025 figure of $4,930; the state indexes it each December and the 2026 amount appears only in return instructions that are not out yet.
• Utah. Utah's taxpayer tax credit phase-out thresholds and dependent exemption here are the published 2025 figures; the state has not released 2026 amounts.
• Vermont. Vermont has published no 2026 rate schedule, so its brackets, standard deduction and personal exemption here are the published 2025 figures.

• Some local income taxes still use state averages. New York City, Yonkers, Philadelphia, Detroit, Columbus and Cincinnati carry their own published rates, and you are asked whether you live inside the city, because a metro is much larger than the city at its centre. Cleveland, Pittsburgh, Louisville, Kansas City, St. Louis, Baltimore and Portland now carry their own rates too, and every Indiana metro carries its counties’ rates weighted by population. What is left on a state average is the smaller cities, where the average is much closer to the truth.
• Where two people earn, the split is assumed to be even. The form asks for one household salary, so when the maths needs to know what each person earns it halves the total. This matters twice: the Social Security cap is a per-person limit, and a couple who file separately file two returns that are each worked out on their own income. Two equal earners is the case this gets exactly right. A lopsided split puts more of the total under one person’s Social Security cap and owes less there, and climbs one person’s income tax brackets faster and owes more there.
• Filing separately with children puts them all on one return. A child is claimed by one parent or the other, never halved, so the model gives them all to the same person, which is also how it works in life — one parent claims a child, not half of one. Where the child credit is being withdrawn at higher incomes this can work out slightly cheaper than splitting them, by about $1,900 on a combined $500,000. We had that the wrong way round here until August 2026.
• Sales tax differences between states are not shown. The spending figures already contain the sales tax those households paid, so charging it again would double it — but what they contain is a national blend. Two states at opposite ends of the rate table therefore look the same on this line, which understates the gain from moving somewhere that charges less and the cost of moving somewhere that charges more. Worth a few hundred dollars a year at most.
• State low-income credits are modelled in 23 states, not all of them. Around thirty states add their own on top of the federal credit, usually as a share of it. Where two independent sources agreed on the figure it is now counted; where they disagreed, or the state uses its own formula rather than a share — California, Minnesota and Washington all do — it is still missing. New York City’s own credit is not counted either.
• Children are assumed to qualify. The form asks how many are under 17 and takes the answer at face value; it never asks about residency, relationship or a Social Security number, all of which the credits actually require.
• Alabama, Missouri and Oregon let you deduct federal tax from state taxable income. Oregon’s is calculated. Alabama’s runs through its own deduction schedule and Missouri’s is a percentage that reaches zero above $125,000 — neither is calculated, and both states say so on their own line in the list above.
• Income-based phase-outs are calculated in eleven states — the deduction or exemption shrinking in nine, and a credit shrinking in Oregon and Utah. Where a state has one we do not calculate, that state says so in the list above, along with which way it runs.
• Only wage income. No investment income, no self-employment, no rental income, no equity compensation.
• Moving itself is free here. This is a steady-state annual comparison — movers, closing costs and deposits are not counted.
• Indiana’s county tax is charged from day one, and really is not. Indiana fixes which county taxes you on 1 January and does not change it when you move, so somebody moving into Indiana owes no county tax at all in their first year unless they already worked there. That first year is not modelled here, and Indiana’s county rates run from 1.21% to 2.35% — so a move to the Indianapolis area is shown roughly $1,200 a year too expensive at the default salary and about $2,700 at $150,000. From the second year on the figure is right.
• Upkeep does not scale with what the house is worth. Repairs and insurance are now charged to buyers — see below — but from a national figure adjusted for local service prices, not from the price of the house you typed in. A $1.6M house really does cost more to look after than a $500k one, and this does not fully capture that. It leans towards making expensive cities look cheaper than they are.
• The suggested number of cars is a whole number, and it jumps. We start you at the local average vehicles per adult, rounded. The New York side of the New York metro averages 0.486 per adult, so a single person there is offered no car at all; everywhere else in the country is offered at least one. A car costs several thousand a year, so a small difference in that average swings a big number. It is the honest thing to show — you own a car or you do not — and the field is yours to change, but it is worth knowing the edge is there.
• Hotels are priced nationally. A hotel on a trip is bought wherever the trip goes, not where you live, so no local price level is applied to it. Second homes are in the same figure and that reasoning suits them less well.
• The top income band is open-ended. The Census publishes rent burden for “$100,000 or more” as a single group, so the rent curve is anchored at $150,000 and extrapolated above it. Expect it to be roughest for very high earners.
• Bedrooms are inferred, not asked. Two adults are assumed to share a room and children to pair up. If you rent more space than that, or less, the rent field is yours to change.
• Averages are not you. And they are not all the same kind of number either. Housing figures are medians for the local area. Spending and car ownership are averages, which sit higher than the median wherever a few large spenders pull them up. Price levels are index numbers, not dollars. The data page labels every one. Your rent, your car and your grocery bill will differ from all of them, which is why almost every field is editable.

The figures themselves, and where each comes from, are on the data page.

Think something here is wrong?

This list is not finished, and it is not meant to be — every item on it got there because someone noticed. If a number looks off for a place you know, or an assumption on this page does not match how your household actually works, that is worth telling me about even if you are not certain. A model is only corrected by the people it gets wrong.

Two ways to reach me: Email me or open an issue.

This is not financial, tax or legal advice. It is an estimate built from public data to help you think, not a substitute for someone who knows your situation.