The most common mistake in reading cost of living data is treating a national average as a description of a household. No family pays the national average price for anything. They pay the price in their county, for the specific goods they actually buy, out of their specific income. Almost every misleading claim about affordability, in either direction, comes from collapsing that distance and hoping nobody checks. Here is what each of the major series actually measures, and where each one goes wrong when it is quoted carelessly.
The CPI measures a basket, not your basket
The Consumer Price Index published by the Bureau of Labor Statistics tracks the price of a fixed basket of goods and services, weighted by what urban consumers buy on average. That weighting is the whole story. Housing carries roughly a third of the index. If your rent jumped 15 percent and gas fell 15 percent, the index may barely move while your budget moves a great deal.
The index is also a rate, not a level. A falling inflation rate means prices are rising more slowly, not that anything got cheaper. When people say the data contradicts their experience, this is usually why. Both things are true at once: the rate decelerated and the level never came back down.
Averages and medians are not interchangeable
A mean can be dragged anywhere by a small number of extreme values. A median cannot. The gap between the two is often the most informative number in a dataset, and it almost never gets quoted.
The Federal Reserve’s Survey of Consumer Finances gives a clean demonstration. In 2022 the median retirement account balance among families that had one was $86,900. The mean was $334,000. Same families, same year, same question. One number describes the middle of the distribution and the other describes what happens when you let a small number of very large accounts into the arithmetic. Quote the second and you get a country that is comfortably prepared for retirement. Quote the first and you get something else.
Watch for the word “conditional”
That $86,900 figure carries a qualifier most write-ups drop: it counts only families who have a retirement account at all. The same survey found 54.3 percent of families held one in 2022, up from 50.5 percent in 2019. So the median describes a little over half the country and says nothing about the rest.
Conditional statistics appear everywhere in affordability reporting. Average student loan balance per borrower excludes everyone who never borrowed. Median home sale price excludes every home that did not sell. Average employer premium contribution excludes workers with no offer of coverage. Each exclusion moves the number in a predictable direction, and each one is usually stripped out somewhere between the source table and the headline.
National prices, local budgets
The Census Bureau and the National Association of Realtors put the median U.S. home sale price in the range of $400,000 to $420,000 in 2024. That single figure covers markets where the median is half of it and markets where it is triple. Housing is the largest line in most budgets and the most geographically variable, which means a national housing number is the least transferable statistic in the whole set.
The MIT Living Wage Calculator handles this differently. It builds a required income from the ground up for a specific county and a specific household composition, then reports what that household needs to cover basic costs. It answers a different question from the CPI and the two should never be substituted for each other. One tracks change over time. The other tracks adequacy at a point in time.
Nominal figures without a year are worthless
The federal minimum wage has been $7.25 an hour since 2009, according to the U.S. Department of Labor. The number has not changed. Its purchasing power has, every year, by whatever the inflation rate was. A nominal figure quoted without its year and without an adjustment is a claim about arithmetic dressed up as a claim about the world.
This cuts in both directions. Wage growth reported in nominal terms during a period of rapid price increases overstates gains. Wage growth reported in real terms during a period of slow price increases can understate how people experience their paychecks, because real adjustments use the average basket rather than theirs.
Percent change versus level
A 3 percent increase on a small base is a small number. A 3 percent increase on a large base is not. Childcare and health coverage are both large enough that ordinary-sounding percentage increases translate into hundreds of dollars a month, while a dramatic-sounding percentage move in a small category changes nothing anyone notices.
The reverse trick is also common. Quoting the level of a total, such as aggregate household debt, without a denominator makes any number sound alarming, because populations and price levels both grow. Aggregate totals are useful for establishing that something is not a niche problem. They are nearly useless for establishing whether it got better or worse.
A short checklist
Before repeating any affordability figure, five questions settle most of it. What year is it from, and is it nominal or inflation adjusted? Is it a mean or a median? Who is excluded from the denominator? Is it a national figure being applied to a local decision? And does the source publish the underlying table, or only the press release?
That last one matters more than it sounds. A figure you can trace to a source table can be checked. A figure that only exists in a summary usually cannot, and it tends to acquire precision it never had as it gets repeated. Organizations that collect the underlying affordability series in one place with their sources attached make that tracing faster, which is most of the work.
Why this matters beyond accuracy
Bad numbers do not just mislead, they misdirect. If the affordability problem gets described entirely through one series, the policy conversation narrows to whatever that series measures. Wages are one input. Housing prices, health premiums, childcare costs, transport and education costs are other inputs, and they moved on their own schedules for their own reasons.
The Federal Reserve’s Survey of Household Economics and Decisionmaking, fielded in October 2025, found that 63 percent of adults would cover a $400 emergency expense with cash or its equivalent, and that 73 percent described themselves as doing okay or living comfortably financially. Neither figure is a wage statistic. Both describe affordability more directly than any single price index does. Reading the data honestly means holding several series at once and resisting the one that happens to support the point you already wanted to make.