Did Homes Beat Inflation? 71% of Measured Home Resales Did

Homes are sold as an inflation hedge, and at the market level the deed record agrees. Across 3,196 qualifying postal markets with enough repeat-sale deed pairs, 99.78% have a median real return above zero, and the median of those market medians is +2.37% per year after inflation. Read that way, beating inflation looks close to universal.

But a market median hides who actually won. A postal market can post a positive median while a large share of its individual owners still lost ground to CPI. So we went underneath the market medians, down to the ownership windows themselves: every repeat-sale pair, deflated over its own hold by the inflation it actually lived through.

Measured that way, the picture is far less comfortable. Of 5,958,008 individual ownership windows, only about 71% beat inflation. Roughly 29% lost to CPI. And whether an owner cleared the bar varied sharply by purchase decade.

The scoreboard

The two levels disagree in tone, and both are correct. At the market level almost every place clears inflation. At the ownership level nearly three in ten resales did not. The market median is a fact about places; the 71% is a fact about resales.

The real finding: about 29% of measured ownership windows lost to inflation

The pooled median ownership window returned +2.38% per year in real terms, close to the market-level figure. But the spread around it is wide and lopsided.

A quarter of all ownership windows came in at or below -0.49%/yr real (the 25th percentile), and the bottom decile lost 4.05%/yr in purchasing power. At the other end, the top decile gained 14.86%/yr real. Beating inflation was the majority outcome, not a guaranteed one.

It varied sharply by purchase decade

Split the same ownership windows by the decade of purchase and the story sharpens.

Homes bought in the 2010s beat inflation 86.8% of the time, at a median of +4.08%/yr real, the strongest cohort in the record. Homes bought in the 2000s were close to a coin flip: only 52.8% beat inflation, at a median of just +0.25%/yr real, and in 493 of 1,266 qualifying markets the 2000s cohort's median real return was actually negative. The 1990s cohort beat inflation 77.0% of the time (+2.29%/yr), and the 2020s so far 69.7% (+2.60%/yr). The 2020s cohort necessarily contains only homes that have already resold, at shorter holding periods, so it is a somewhat different population from the earlier decades.

In this realized record, real outcomes varied dramatically by purchase decade, even within the same postal market. A 2000s buyer who caught the pre-crisis peak faced a very different real outcome than a 2010s buyer who bought into the recovery, even in the same postal market.

Why the market median hides who won

This is why the reassuring market-level figure, 99.78% of markets positive, is not the whole truth. A market median can sit comfortably above zero while the owners who bought at the wrong moment sit well below it. The +2.37%/yr median of market medians is real and worth knowing, but it describes the typical place, not the typical owner, and it is silent on the roughly 29% of ownership windows that lost to CPI.

How we measured it

Appreciation. Realized annualized appreciation from repeat-sale deed pairs, the same home selling twice, over holds of 1 to 30 years.

Inflation. Each pair is deflated over its own hold window using the BLS CPI-U, all items, US city average, series CUUR0000SA0, via the Fisher relation: real = (1 + nominal) / (1 + CPI) - 1. Unlike a single long-run average, this matches every sale to the inflation it actually lived through: a 2004-to-2009 hold is deflated by 2004-to-2009 CPI, a 2019-to-2024 hold by 2019-to-2024 CPI.

"Beat inflation" means beat the national CPI-U, not a local cost of living. CPI-U is a single national series; we did not apply market-level inflation, so a market where local prices rose faster than the national average was measured against a bar that, for its own residents, ran low.

Gate. Market-level shares use the 3,196 postal markets with enough qualifying pairs; the 2000s decade view uses the 1,266 markets with enough 2000s-purchase pairs.

What beating inflation does not mean

Four cautions keep this honest.

Gross, not net. These are price-only figures. They ignore transaction costs, carrying costs such as property taxes, insurance, and maintenance, and renovation spending, and repeat-sale pairs can embed renovation-driven gains. Real returns net of those costs are lower than the +2.38%/yr price-only median, and the 71% share would be lower too.

Survivorship. We can only measure homes that resold. Homes bought and never sold again, including those lost to foreclosure or held through a downturn, are absent, so the 71% is best read as an upper bound.

Realized, not a forecast. Every figure here is historical and realized. It describes what happened to homes bought between the 1990s and now; it is not a projection of future real returns.

Postal geography. Our market keys are postal-city groupings, not municipal boundaries. "New York, NY," for example, is the Manhattan postal grouping, not all five boroughs.

Related studies

This study takes the real, inflation-adjusted side of the ledger; its companions cover the rest. Our real home appreciation by city study reports the nominal side of the same deed-pair scan, a 5.19%/yr median at a stricter pairs_n >= 800 gate. The house rent premium study measures how much house leases sign above the citywide median lease, and cash flow or growth asks whether the markets with high gross rental yields, gross of expenses, are the ones where prices rose fastest. And how much will my house be worth in 10 years turns the same nominal appreciation rates, a 5.27%/yr median of market medians, into dollar scenarios, realized history rather than a forecast.