Three major publishers report US house prices year-over-year in early-2026 — Case-Shiller +0.7%, FHFA +1.7%, Zillow +0.7%. All are national. None is wrong. The ~1.0pp spread (RAOSCAFF arithmetic) is structural.
FHFA prints +1.7% YoY (Q1 2026); Case-Shiller National and Zillow ZHVI both print +0.7% YoY. The spread is not a contradiction: FHFA tracks conforming, GSE-backed purchase mortgages — capping out the high end — while the other two span all financing or model the entire stock. Same country, different reference periods (Case-Shiller March 2026 / FHFA Q1 2026 = Q1; Zillow April 2026 = Q2), different populations. RAOSCAFF arithmetic: +1.7% − +0.7% = +1.0pp (inputs: §H-2, §H-1/§H-4).
Ask “how much are US house prices rising year-over-year?” in early-2026 and three major publishers answer. All three answers are positive. All three are described as national. They differ by roughly a percentage point — and the spread is structural.
S&P Cotality Case-Shiller U.S. National reads +0.7% YoY (March 2026, NSA). FHFA House Price Index reads +1.7% YoY (Q1 2026). Zillow ZHVI reads +0.7% YoY (April 2026). Each is correct under its own definition. The question a single headline number cannot answer is: which homes, which financing channel, and which window?
The two broad-coverage measures — all financing (Case-Shiller) and the whole modelled stock (Zillow) — converge at +0.7%. The conforming-loan-only measure sits +1.0pp higher. That alignment is the tell: the gap is not noise, not a lag, not a methodology flaw. It is a coverage artifact.
A bold “US house prices YoY +?%” splits into three labelled answers, each filtered through its own population. The FHFA beam registers ~1pp higher (RAOSCAFF arithmetic) — the conforming-only slice and the all-financing / whole-stock measures record different magnitudes; that difference is a coverage observation, not a price-movement claim.
Axis 1 — Which homes are counted (the ~1.0pp gap, RAOSCAFF arithmetic). FHFA tracks only homes behind conforming, GSE-backed purchase mortgages. The conforming-loan ceiling caps out the high end; cash deals and jumbo loans are excluded. Case-Shiller’s national index spans all financing — including jumbo and cash — and is value-weighted. Zillow models the entire stock, including homes that never transact. The two broad-coverage measures land at +0.7%; the conforming-only measure lands at +1.7%. The gap is a coverage artifact of which slice, not a flaw in any index.
Axis 2 — Metro composition moves a number before methodology does. Inside Case-Shiller’s own family: National +0.7%, 20-City Composite +0.8%, 10-City Composite +1.4% — same publisher, same month (March 2026, NSA). The within-publisher span is +0.7pp (10-City minus National, RAOSCAFF arithmetic). Quoting “Case-Shiller” without naming the cut already swings the answer before any cross-publisher comparison is made.
Axis 3 — Reference period is not the same window. Case-Shiller is released on a two-month lag (March 2026 data, released 26 May). FHFA publishes a quarterly cut (Q1 2026) and a monthly cut (March 2026, SA) on the same day. Zillow publishes its April 2026 ZHVI the following month. Any cross-index gap silently blends a calendar difference with a definitional one.
| Index | Which homes | Financing | Method | Period | YoY headline |
|---|---|---|---|---|---|
| S&P Cotality Case-Shiller National | Repeat-sales, value-weighted; national index built from nine Census division indices | All — incl. jumbo + cash | Transaction-based | March 2026 (NSA) | +0.7% |
| FHFA HPI | Conforming, GSE-backed nationwide; equal-weighted | Purchase-only, conforming (caps out high end) | Transaction-based | Q1 2026 / March monthly | +1.7% |
| Zillow ZHVI | Entire housing stock incl. homes never sold | None — modelled valuation | Modelled (Zestimate, 35th–65th pct. [§H-7]) | April 2026 | +0.7% |
Sources: §H-1 (Case-Shiller), §H-2 (FHFA Q1), §H-4 (Zillow). FHFA-vs-others spread +1.0pp = RAOSCAFF arithmetic (§H-2 minus §H-1/§H-4). Within-Case-Shiller composition span +0.7pp (10-City +1.4% minus National +0.7%) = RAOSCAFF arithmetic (§H-1).
If you read one thing: the ~1.0pp gap between FHFA and the others (RAOSCAFF arithmetic) is definitional — conforming-only vs all-financing / whole-stock. It is not a signal that one index is rising faster.
In early-2026, three authoritative US house-price publishers all read “national” and all read positive — yet they sit roughly a percentage point apart (RAOSCAFF arithmetic). That spread is not a measurement error or a turning-market signal; it is the arithmetic consequence of counting different populations over different windows. The conforming-loan ceiling in the FHFA series excludes the high-end and cash segment; the all-financing and whole-stock measures include it. When the broad-coverage measures converge at +0.7% and the narrow-coverage measure prints +1.7%, the honest reading is: FHFA is correct for its population, and its population is narrower. The same logic applies within a single publisher: Case-Shiller National (+0.7%) vs 10-City Composite (+1.4%) — same month, same method, +0.7pp apart (RAOSCAFF arithmetic), purely because of metro composition. A single “US house prices” number has already chosen a population, a method, and a month. This brief makes one claim: name those three choices before reading the spread as a market signal. Nothing here is a view on buying, selling, holding, or any security.
Mirror format — RAOSCAFF anchors on the publishers’ own primary releases and decomposes the spread between them. All three headline figures were fetched live from primary or authoritative sources (S&P Cotality press release; FHFA news release; Zillow Research report corroborated by FRED) during Phase 0 (2026-06-04). The brief treats the spread as a cross-publisher divergence driven by population definition and measure design, not a discrepancy to reconcile. No primary data collection, no analyst estimate, no extrapolation beyond the arithmetic register in FACTS.md.
Every figure traces to a named, dated release in FACTS.md §H. The Case-Shiller figures (National +0.7%, 20-City +0.8%, 10-City +1.4%, MoM NSA/SA) are verbatim from the 26 May 2026 S&P Cotality press release (§H-1) and independently corroborated by Mortgage News Daily (§H-5). The FHFA figures (+1.7% YoY Q1, +0.5% QoQ, +0.1% MoM SA) are verbatim from the 26 May 2026 FHFA news release (§H-2). The Zillow ZHVI (+0.7% YoY, +0.6% MoM, April 2026) was verified during Phase 0 from the Zillow Research report text and FRED corroboration (§H-4); the Zillow page returns HTTP 403 on automated re-fetch (bot protection) — no new figure is introduced beyond the Phase-0-verified value. The FHFA-vs-others spread (+1.0pp) and the within-Case-Shiller composition span (+0.7pp, +0.1pp) are derived and labelled “RAOSCAFF arithmetic.” The brief covers national-level figures only; no metro-level or state-level claims are made beyond the composites the publishers print. Elevated mortgage rates appear only as a publisher-quoted market condition (§H-5), never as a policy judgement. No forecast of the direction of future prices is made.
All headline percentages (+0.7%, +1.7%, +0.7%/+0.6%, +0.8%, +1.4%, +0.5%, +0.1%, +0.72%, −0.22%) are published, traced to §H-1 through §H-5. Derived figures (+1.0pp FHFA spread; +0.7pp and +0.1pp within-Case-Shiller composition spans) are RAOSCAFF arithmetic and labelled as such. Full citations and arithmetic register: FACTS.md §H.