Do the Official U.S. Inflation Numbers Tell the Truth About the Cost of Living?

Two Numbers That Are Both True

Ask an economist how bad inflation has been and you get a strange answer: prices are up about 25% since January 2020, and the pace of price increases has cooled to something close to normal ^[2]. Both statements are accurate. Neither one is the whole story.

Here's the distinction that trips people up. The inflation "rate" you hear on the news measures how fast prices are climbing right now. It says nothing about how high they already climbed and stayed. A rate near 3% just means prices are rising slowly this year — not that anything got cheaper, and not that the jump from a few years ago went away ^[2]. That gap between the rate and the level is where most of the argument over trust in the numbers actually lives.

The two official scorekeepers are the Consumer Price Index, run by the Bureau of Labor Statistics, and the Personal Consumption Expenditures index, which the Federal Reserve actually targets. CPI comes from about 80,000 price checks collected every month at stores and rental units nationwide, weighted by what a survey says households actually buy ^[1][4]. PCE counts some things CPI doesn't, like the part of your health insurance your employer pays for, and it updates its weighting faster. That's part of why PCE usually reads a few tenths of a point lower than CPI ^[5][6]. Shelter alone is about a third of CPI but only around 16% of PCE — one reason the two indexes can tell slightly different stories about the same economy ^[6].

The Formula Nobody Agrees On

Every method choice behind these indexes has a defender and a critic, and both sides are arguing from the same facts. Take housing. Since 1983, CPI hasn't measured what it costs to buy a home. It measures what a homeowner's house would rent for, called owners' equivalent rent. The BLS says this isolates the cost of living in a home from the separate question of investing in one ^[9]. Critics say the estimate lags behind what's really happening in rental markets, so it can understate a housing squeeze while renters are living through it ^[10].

Then there's substitution. If steak gets expensive and people buy more chicken, should the index assume they kept buying steak anyway? Modern CPI leans toward "no" — it adjusts for how people actually shift their spending when prices move. Statisticians call that a more honest cost-of-living measure. Critics call it a built-in way to shrink the number, since any formula that assumes people downgrade what they buy will always show less pain than one that doesn't ^[5][8].

The single biggest turning point here was the 1996 Boskin Commission. It found the old CPI was overstating true inflation by about 1.1 percentage points a year, with a plausible range of 0.8 to 1.6 points ^[7]. That finding pushed CPI toward the substitution-aware formulas and "hedonic" adjustments used today — the idea that a $1,000 phone now does far more than a $1,000 phone did in 2010, so part of that price is really a quality gain, not inflation ^[11].

Notice the pattern: the agencies present each of these changes as a technical fix, a step toward accuracy. What they rarely do upfront is put a number on how much each change lowered the reported rate. That silence is exactly where distrust grows, even among people who find the underlying logic sound.

Four Ways to Read the Same Chart

The first camp says the index is basically right, and that these changes are real corrections, not tricks. This is the view of the BLS itself, most academic economists, the Fed, and centrist research groups like the Peterson Institute ^[28]. Their best evidence isn't the government vouching for itself — it's outside confirmation. Independent trackers like MIT's Billion Prices Project, the private tool Truflation, and the Cleveland Fed's median CPI all move in step with the official number ^[14][15][16]. Economist Alberto Cavallo showed what real manipulation looks like by comparing online prices to government data in Argentina, where they diverged sharply — while in the U.S., they matched closely ^[15]. This camp's blind spot, by its own critics' account: it can undersell how much strain the agency is now under. In 2025, budget cuts forced fewer price collections and the commissioner was fired, real risks even CPI's defenders admit ^[28][29].

The second camp says the index materially understates the real cost of living. This includes ShadowStats' John Williams, the Chapwood Index, and some Austrian-leaning economists. Their case: it's not a coincidence that formula changes keep pushing the number down, especially since CPI directly controls trillions of dollars in Social Security payments and tax brackets ^[7][22]. That's a genuine structural incentive, whether or not anyone is actually gaming it. But the loudest versions of this argument fall apart under scrutiny. ShadowStats doesn't recompute inflation using old methods — it just adds a fixed number on top of the official rate. The Chapwood Index relies on informal, unverifiable price surveys. Both produce figures wildly out of line with every independent tracker, including private ones with no reason to go easy on the government ^[12][13][30].

The third camp calls the index honest, but too narrow to answer the question people actually ask. This includes conservative writer Oren Cass, whose Cost-of-Thriving Index argues CPI can be arithmetically correct and still miss the point: can a family afford the life it expected? A single earner in 1985 could cover a house, health insurance, a car, and college on far fewer weeks of pay than today, because big lumpy costs — housing, health insurance, childcare, college — outran both the broad price basket and wages ^[17]. Progressive researchers make a similar working-families argument. Critics at the American Enterprise Institute push back hard, arguing Cass's math counts the full sticker price of health insurance while ignoring what employers cover, and leans on an outdated single-earner household that overstates how much ground families really lost ^[18].

The fourth camp, often called "greedflation," mostly agrees the number is accurate and argues the real story is who kept the money. Groundwork Collaborative found corporate profits accounted for 53% of price growth in one stretch of 2023, compared with about 11% across the four decades before the pandemic ^[21]. Real wages fell hardest for lower earners during the same period ^[19]. The pushback, made across the spectrum, is that fat profit margins more likely followed high demand and pandemic supply shocks than caused them — and those margins later shrank back down ^[28]. Because this camp mostly accepts CPI's accuracy, it sidesteps the trust question rather than answering it.

When a Raise Isn't Really a Raise

Here's a mechanism that matters as much as any formula debate: the inflation rate measures prices, not paychecks. A "nominal" wage is the dollar number on your paycheck. A "real" wage adjusts that number for inflation, showing what it can actually buy. These two can move in opposite directions at once, and for a stretch of 2021 into 2022, they did.

Real average hourly earnings fell about 1.7% over the year ending in December 2022 ^[19][20]. That means many workers got raises on paper and still lost buying power — the same hours of work, for less in real terms, even as headlines reported wage growth. Real earnings turned positive again in 2023 through 2025, as inflation cooled faster than pay ^[19]. The squeeze wasn't shared evenly either. Lower-wage workers were hit hardest during the worst stretch, based on Atlanta Fed wage tracking, though some saw faster catch-up growth afterward ^[26][27].

None of this means CPI itself was miscounted. A correctly measured 8% inflation rate paired with a 6% raise still adds up to a real pay cut. That's a wages problem stacked on top of a pricing problem — not proof the pricing problem was faked.

The Same Country, Very Different Prices

A single national number also hides where the pain is worst, and housing is the biggest reason why. Federal data put California's overall price level about 10.7% above the national average as of 2024 ^[24]. In the San Francisco area in 2023, $100 bought only about $84.58 worth of goods — compared with roughly $124 worth in the cheapest metro areas in the country ^[25].

That gap matters because a bigger paycheck in an expensive city doesn't always cover the cost of living there. Once you subtract higher housing costs, workers in pricey coastal metros can end up with less real buying power than workers earning less somewhere cheaper ^[24][25]. Some of the recent migration toward lower-cost states looks like people acting on exactly that math ^[24]. A national CPI can be too gentle a read for someone in San Francisco and too harsh a read for someone in a low-cost metro, in the very same month. It isn't wrong — it's an average, doing what averages do.

What Would Actually Settle This

Some parts of this fight aren't just contested opinion — they're genuinely unresolved. Nobody agrees on the exact size of the drag from decades of methodology changes. Mainstream economists put it at a fraction of a point per year; the Boskin-era analysis implied more; there's no single settled figure ^[8][13]. Whether the BLS's 2025 budget cuts are already hurting data quality is unknown. An inspector general audit is underway, and even the agency's defenders say the risk is real, even though no evidence of deliberate rigging has turned up so far ^[28][29].

The greedflation debate is unsettled too — whether high profit margins caused the price surge or simply rode along with it remains a live argument ^[21][28]. And the effect cuts differently depending on who you are. The CPI-W version used for Social Security may run about 0.3 points higher than a more precise alternative, while an experimental measure built for seniors suggests their heavier healthcare spending might mean the official number runs low for them instead ^[22][23].

Underneath it all sits one fact nobody disputes: there's no single mathematically correct way to build a price index, only defensible choices that read differently depending on which side of the number you're standing on ^[5][8]. That's also why coverage of this fight splits so predictably. The BLS and the Cleveland Fed treat it as ongoing technical refinement. ShadowStats and Chapwood treat it as a cover-up. American Compass reframes it as an affordability story rather than a measurement one. Groundwork Collaborative and EPI point past the index entirely, toward corporate profits. None of that disagreement is really about the arithmetic — it's about which question people think the number is supposed to answer. Until the pending audit of BLS's own data collection reports back, that question stays open.