Your Personal Inflation Rate Is Not the CPI, and You Can Calculate It

Every month a single number leads the news, and every month some share of the people reading it conclude the number is wrong. Their rent went up 9%. Their electricity bill doubled over two summers. The figure on the screen said 3.7%.

The figure is not wrong. It is answering a different question than the one people hear. The Consumer Price Index measures the price change of one specific basket, weighted by what urban households bought, and the agency that publishes it is unusually blunt about the limits of that. The CPI, in the Bureau of Labor Statistics’ own words, “does not necessarily measure your own experience with price change.”^[6]

Which raises the useful question. If the published rate is not yours, what is yours, and can you work it out? You can. It takes the published category changes, which are free, and your own spending shares, which are the part most people do not have.

What the CPI actually measures

The index tracks what it costs to buy a fixed basket of goods and services over time. Government staff collect roughly the same items from roughly the same stores month after month, so that what changes in the index is price rather than quantity or quality.

That design decides what the number can and cannot tell you. It is built on “the experience of the relevant average household, not of any specific family or individual,” and the Bureau says it does not produce official estimates of inflation for subgroups of the population at all.^[6] It is also not a complete cost-of-living measure. The Bureau calls it a conditional cost-of-living index, because things that plainly affect how expensive life feels, including changes in income taxes and in social and environmental conditions, sit outside the definition on purpose.^[6]

Two more boundaries are worth knowing before you compare the index to your bank statement. The CPI-U covers urban consumers, which the Bureau puts at over 90% of the U.S. population, so rural and farm households are outside it.^[6] And the weights lag: the Bureau offers its own example, that CPI data in 2023 rested on Consumer Expenditure Survey data collected in 2021.^[6] Today’s index runs on the spending habits of two years ago.

The weights belong to an average household

A price index is two things multiplied together: a set of price changes and a set of weights. The price changes are measured in the market and are the same for everyone shopping in the same place. The weights come from a survey of what households spend, and they are where your number and the published one part company.

In a quiet month the weights barely matter, because every category moves at roughly the same pace. September 2026 was not a quiet month. These are the published 12-month changes, next to the share of the basket each category carries:

The weights in that middle column are the ones the 2026 index is actually computed with: the Bureau publishes relative importance once a year from December data, and the December 2025 set rests on 2024 spending.^[5] They reconcile, which is a useful check that the arithmetic below is the same arithmetic the index does. Food at 13.7% of the basket rising 2.5%, energy at 6.4% rising 20.4%, and the remaining 79.9% rising 2.6% come to 3.7%.

Multiply a weight by a change and you get that category’s contribution to the headline. Gasoline, at 2.9% of the basket and up 34.2%, contributed about 1.0 percentage point of the 3.7. Take gasoline out of the basket entirely and renormalize, and the rest of it rose about 2.8%. Give gasoline 8% of a budget instead, which is ordinary for a long commute in a thirsty car, and the same prices produce about 5.3%.

The Bureau demonstrates this itself. Its research index for Americans 62 and older, the R-CPI-E, prices the same areas, the same outlets, and the same items as the CPI-U. The only substantive change is the weights, drawn from the subset of surveyed households whose reference person or spouse is 62 or older. It produces a different number, which is the entire point.^[10]

A third of the index is not your mortgage

Shelter is 35.6% of the basket and owners’ equivalent rent is 26.2% of it on its own, making OER the largest single line in American inflation by a wide margin.^[5] Almost nobody writes a check for it.

OER is the Bureau’s estimate of “the implicit rent that owner occupants would have to pay” for their own homes.^[7] It does not price owned homes at all, treating them as “capital (or investment) goods distinct from the shelter service they provide,” which puts house prices and mortgage interest outside the index as investment costs rather than consumption.^[7] Owners are asked what their home would rent for, and that answer sets the weight, but the month-to-month price change is measured from the rental sample instead. Owners’ own responses are not used in estimating price change.^[7]

Follow that through for two readers. If you rent, the index is measuring something close to your reality, though it is measuring the average of all leases rather than yours. If you own with a fixed-rate mortgage, more than a quarter of the index is tracking a hypothetical rent you will never pay, while the housing costs that genuinely move on you are the ones the CPI set aside: property taxes, insurance, and repairs. Those arrive through the escrow account, and an escrow analysis can lift a fixed mortgage payment by a few hundred dollars a month without any index registering it as shelter inflation.

There is also a timing quirk. The Bureau divides the housing sample into “six panels, with the housing units in each panel priced twice per year.”^[8] A given unit is revisited roughly every six months, and most of the units in it are occupied by tenants in the middle of a lease, whose rent does not change until renewal. The result is an index that turns slowly in both directions. The Bureau runs a research series precisely to see the faster signal: the New Tenant Rent Index “measures prices renters would face if they changed housing units every period,” using only the first observations after new tenants move in, published beside an all-tenant companion built the same way for comparison.^[9]

One consequence of shelter’s size is counterintuitive right now. At 3.2%, shelter came in below the 3.7% headline, so its weight is pulling the index down rather than up. The all items less shelter index rose 4.0% over the same 12 months.^[3] A household whose rent is locked and whose fuel bill is not has every reason to feel a number well above the published one.

Your city had its own number

The national figure is an average of places as well as of households, and the Bureau publishes the pieces. Over the 12 months to September 2026, the all-items index rose 4.2% in both the Northeast and the Midwest and 3.4% in the South and the West. Metro areas spread further.^[4]

A reader in Tampa and a reader in Honolulu both saw 3.7% on the news. Their own official local numbers were 1.9% and 5.2%. Before you conclude that your city is an outlier, though, check the obvious alternative: local price levels and local spending patterns are entangled. Housing is the largest expense in the national survey at 33.4% of average household spending,^[11] and where housing is dearest it crowds out everything else, which changes your weights as well as your prices.

Three indexes, one month, three answers

September 2026 produced three official inflation rates from the same month of price collection. The all-items CPI-U rose 3.7% over 12 months. The CPI-W, which weights the basket by the spending of urban wage earners and clerical workers, rose 3.9%. The chained CPI-U, which allows for substitution between categories as relative prices shift, rose 3.5%.^[1]

Nothing disagrees here. One set of prices, three sets of assumptions about whose basket, and whether shoppers substitute when relative prices move. The gap matters in practice, because contracts and benefits name a specific series rather than “inflation.” The Social Security cost-of-living adjustment, for instance, is the percent increase in the third-quarter average of the CPI-W, not the headline CPI-U figure that leads the news.^[6] The Bureau is careful to add that it publishes the series and does not decide how other agencies use them.^[6]

So if a number in your life is indexed, the index it names is the one that governs, and it will not be the one you heard on the radio. Two tenths of a point a year, compounded across a retirement, is real money.

Which is the honest frame for all of this. There is no single true inflation rate waiting to be found. There are baskets, and an index is a decision about which one to price.

Calculate your own rate

The arithmetic is a weighted average and nothing more. For each category, multiply your share of total spending by that category’s published 12-month change, then add the results up.

1. 1
   Total your spending for the last 12 months

   Use one figure for the whole period, excluding transfers between your own accounts, credit card payments, and anything you saved or invested. Those are not consumption, and the CPI excludes them too.
2. 2
   Split it into eight or ten categories

   Housing, food at home, food away from home, electricity and gas, gasoline, insurance, health care, and everything else is enough. Precision beyond that changes the answer by a rounding error.
3. 3
   Turn each category into a share

   Divide each category by the total. The shares must sum to 100%, which is the step that catches a missing category or a double-counted transfer.
4. 4
   Look up each category's 12-month change

   Table 1 of the monthly CPI release carries them, unadjusted, next to the national weights. Use the same month for every category.
5. 5
   Multiply, then add

   Share times change, category by category, summed. The result is the rate that applied to your basket rather than the average one.

Worked through with the September 2026 figures, two households look like this. The renter without a car: 34% of spending on rent at +2.8%, 11% on groceries at +1.8%, 8% on restaurants at +3.5%, 4% on electricity at +4.4%, 5% on medical services at +2.5%, 2% on airfare at +22.4%, and the remaining 36% at the core rate of +2.6%. That comes to 3.1%.

The homeowner with two cars and a long commute: 24% on shelter at +3.3%, 12% on groceries at +1.8%, 5% on restaurants at +3.5%, 4% on electricity at +4.4%, 2% on piped gas at +5.0%, 9% on gasoline at +34.2%, 4% on auto insurance at minus 5.0%, 6% on medical services at +2.5%, and 34% at +2.6%. That comes to 5.4%.

Same month, same published prices, 2.3 percentage points apart, with the national 3.7% sitting between them. The fuel-heavy household is not imagining things, and the renter is not being ungrateful.

What the number is good for

Step three is the one that stops people, because almost nobody knows their category shares to the nearest percentage point. Guessing defeats the exercise: the whole result turns on the weights, and estimates of your own spending run low.

That part your transaction history already answers. In LumaLedger, the Income & spending page totals every month by category with transfers and card payments stripped out, so the shares are a division rather than a reconstruction. The annual view in Reports gives you the same breakdown over a full twelve months, which is the window the 12-month index changes are measured over.

LumaLedger

Income & spending

Past 6 months

Apr

May

Jun

Jul

Aug

Sep

IncomeSpending

Spending · Sep$6,178

Housing$2,450

Groceries$780

Transportation$610

Dining$540

Shopping$520

Utilities$390

Everything else$888

Once you have the number, be clear about what it does. It will not change any price, and it is not a forecast. What it gives you is a correct denominator. A 6% rise in a category that the index says rose 2% is a decision you made and can revisit; a 6% rise in a category that rose 6% is weather, and cutting the category will feel like a punishment for nothing.

It also tells you where your exposure sits. A household with 9% of its budget in gasoline has made a large, undiversified bet on fuel prices without deciding to, and the fix is a different commute or a different car rather than a tighter grocery budget. Pricing a vehicle per mile turns that exposure into a number you can compare against alternatives. A household with 34% in rent has made the same kind of bet on the local rental market.

One last reframing. The published rate is useful for exactly what it was built for: a consistent national measure that contracts, benefits, and monetary policy can reference without arguing about whose basket counts. For your household, the figure you want is the one weighted by your life, and the shares a budget framework assigns to needs, wants, and saving are the same shares this calculation runs on. Work them out once and both jobs are done.