Why It Matters
When the news reports "existing home sales ran at a 4.0 million SAAR in March," that doesn't mean 4 million homes sold in March. It means if every month ran at March's seasonally-adjusted pace, the full year would produce 4 million sales. SAAR is the federal government's standard way to report monthly housing data — existing home sales, new home sales, housing starts all come out this way. The reason: raw monthly housing data is wildly seasonal. March sells about 10% more homes than January. June sells about 40% more than November. Without seasonal adjustment, month-to-month comparisons would be useless. SAAR strips out the seasonal pattern, then annualizes so readers can compare a single month's pace to any other month's pace on equal footing.
At a Glance
- What it is: Monthly activity, seasonally adjusted, then annualized by multiplying by 12.
- Where you see it: NAR existing home sales, Census new home sales, Census housing starts, pending home sales index — all standard federal housing releases.
- What it means in practice: "4M SAAR" means the monthly pace, if sustained year-round with no seasonality, would yield 4 million annual units.
- Why it matters for investors: Lets you compare any month's pace to any other month's without getting fooled by spring/summer vs winter patterns.
- What it does NOT mean: 4M SAAR does NOT mean 4 million homes sold in that single month. The annualization is a pace estimate, not an actual count.
How It Works
Two transformations in one step. SAAR does two things to the raw monthly number. First, seasonal adjustment: the BLS and Census use a statistical method called X-13ARIMA-SEATS that identifies the typical seasonal pattern in a time series (March always sells more homes than November) and removes it, leaving the underlying trend. Second, annualization: the seasonally adjusted monthly figure gets multiplied by 12 to convert to an annual pace. The math is simple — it's the underlying statistical purpose that matters. SAAR lets you ask "if the entire year ran at this month's pace, how many sales would we see?" without distortion from calendar effects.
Why raw monthly data is misleading. Pull raw NAR existing home sales by month. November 2024 shows maybe 350K sales. June 2024 shows maybe 460K. That's a 31% difference — but it's almost entirely seasonal. Buyers want to close before school starts, inventory peaks in spring, summer is the active window. Compare raw November to raw June and you'd think the market collapsed by 31%. In SAAR terms, November 2024 at a 3.96M pace and June 2024 at a 3.89M pace would actually show the market was marginally weaker in June. The SAAR framing strips the seasonal noise so you see what actually changed.
SAAR vs YoY vs MoM — when to use each. SAAR is what the federal release gives you. Year-over-Year compares this month to the same month a year ago — also strips seasonality, but uses a different method (same-month comparison instead of statistical adjustment). Month-over-Month compares this SAAR to last SAAR, giving you the short-term direction. For investor analysis: SAAR tells you the pace level; MoM SAAR tells you if the pace is accelerating or decelerating; YoY tells you where you stand against the prior year. All three answer different questions — don't pick one and ignore the others.
Not every data series uses SAAR. Many non-housing federal series report raw monthly numbers (unemployment rate, CPI). The housing industry is peculiar in its SAAR convention because housing is so heavily seasonal. When you see a monthly number for mortgage rates or unemployment, it's usually a level, not a SAAR. When you see monthly numbers for home sales, starts, or permits, it's almost always SAAR. Read the data release footnotes — the agency will say "seasonally adjusted annual rate" explicitly when that's what they're publishing.
Real-World Example
Diego Ramírez reads the monthly NAR release correctly.
Diego is tracking the Phoenix metro for an acquisition. March's NAR existing home sales release drops. The headline says "U.S. existing home sales rose to 4.02 million SAAR, up 0.8% from February."
He pulls three readings from the release:
- Current month SAAR (March): 4.02 million
- Prior month SAAR (February): 3.99 million — a 0.8% month-over-month gain
- Year-ago month SAAR (March last year): 3.85 million — a 4.4% year-over-year gain
Three conclusions. First, the level is 4.02M — that's the pace. Sustained for 12 months with no seasonality, the U.S. would produce 4.02M existing home sales. Second, the direction (MoM SAAR of +0.8%) says the market is accelerating slightly. Third, the YoY (+4.4%) says the market is in better shape than a year ago. Three different questions, three different answers — all from the same release.
Diego does NOT make the mistake of multiplying March's SAAR of 4.02M by the actual March seasonal pattern to estimate raw March sales. The SAAR already absorbed that — it's a pace number, not a count. He just reads it as "the U.S. housing market is currently running at a 4M annual rate of existing home sales."
Pros & Cons
- Makes monthly housing data directly comparable across months — no seasonal distortion
- Standard reporting convention across NAR, Census, NAHB — no translation needed between sources
- Lets you track acceleration and deceleration on a monthly cadence
- The 12× annualization is simple arithmetic — no complex weighting
- Seasonal adjustment is done by professional statisticians via X-13ARIMA-SEATS, not rule-of-thumb
- Easy to misread as "number of sales that month" instead of annualized pace
- The seasonal adjustment model can be revised, which retroactively changes historical SAARs
- Useless for truly non-seasonal series — applying SAAR to mortgage rates or unemployment would be misleading
- Month-to-month SAAR changes can be noisy — small statistical revisions produce 0.5-1% swings that aren't real signal
- Doesn't help with long-horizon planning — SAAR is about pace NOW, not cumulative YTD or annual total
Watch Out
- SAAR ≠ monthly sales: A 4M SAAR release does not mean 4M homes sold that month. It's a pace — one month annualized. Actual monthly sales are roughly SAAR/12 adjusted for the seasonal pattern.
- Revisions happen: Census and NAR revise prior months' SAARs when new data improves the seasonal model. A March release might revise February's SAAR along with publishing March's. Check the footnotes for revision notes.
- Small MoM SAAR changes are mostly noise: A 0.3% MoM SAAR move is within the statistical noise of the seasonal adjustment. Don't read too much into single-month moves under 1%. Look at 3-month trends.
- Watch for formula confusion: Some series are reported as monthly counts (raw), some as SAAR, some as cumulative year-to-date. The units matter. A 400K monthly count is NOT comparable to a 4M SAAR.
- Seasonal patterns shift: The pandemic disrupted housing seasonality (buyers didn't follow typical patterns in 2020-2022). Seasonal adjustment models have adjusted, but older SAARs calculated with pre-pandemic seasonal factors may be biased. This is documented in BLS and Census methodology notes.
Ask an Investor
The Takeaway
SAAR is the federal convention for making monthly housing data comparable across months. When you see "existing home sales at 4M SAAR" or "housing starts at 1.4M SAAR," that's one month's activity converted to an annual pace after stripping seasonality. Read the level (the SAAR itself), the direction (month-over-month change), and the context (year-over-year change). All three answer different questions. If you're cross-referencing NAR, Census, and NAHB monthly releases, they all use SAAR — no translation needed. Distributed through FRED and the publishing agencies' direct feeds. For historical analysis, FRED's graphs default to SAAR for most housing series.