Housing Starts MoM

Housing Starts MoM is the monthly % change in new privately owned housing units started across the United States. It is 1 of 3 headline series in New Residential Construction, jointly announced by the Census Bureau and HUD, the federal housing department. A start is recorded when excavation begins for footings or foundation, and each apartment counts as 1 unit. Coverage excludes HUD-code manufactured, or mobile, homes, existing homes and public construction, while sister series track permits, or authorizations, and completions, or finished homes.

Coverage spans the nation and 4 Census Regions: Northeast, Midwest, South and West. The sample supports national and regional estimates only, with no state or local detail. New Residential Sales covers only the built-for-sale subset, while starts include contractor-built, owner-built and built-for-rent units that never enter sales. The annualized rate is neither a forecast nor a projection.

The report appears on workday 12 after the survey month at 8:30 in the morning Eastern Time. A workday excludes federal holidays, so holiday periods shift timing by rule. Each release covers permits authorized in the prior month and follows sampled permits month by month until start or completion. Non-permit areas are canvassed by driving roads at least once every 3 months.

Each release also reports permits, authorized but unstarted units, units under construction and completions, with full tables on the Census site. Each release revises data for the prior 2 months, and once each year rates for the prior 63 months are revised with new seasonal factors. Annual estimates are finalized with the release covering the second month of the following year. A single monthly move is often noise, since permits need about 3 months and starts and completions need about 6 months to confirm trend.

The headline total sums 4 reporting cuts, and each cut captures a different slice of new supply. Single-family homes form the bulk, while multifamily swings drive volatility and regional splits shape estimation.

• Single-family: 1-unit homes, including qualifying townhouses with separate walls, heating and meters and no unit above or below, which form the bulk of starts.
• 2-to-4-unit: small multi-unit buildings, a small residual slice of starts.
• 5-or-more-unit: apartment buildings where each apartment counts as 1 unit at excavation, the volatile swing piece.
• By Census Region: Northeast, Midwest, South and West splits in 8 sets by type and region for estimation.
• Backlog and pipeline: authorized but unstarted units sit between permits and starts, and units under construction sit between starts and completions. Quarterly splits cover built-for-sale, contractor-built, owner-built and built-for-rent.

1. Permit universe: monthly estimates count units authorized by permits across all permit places.
2. Authorization ratios: statisticians compare permit-place estimates with sampled offices across 14 ratios in 8 sets by type and region based on 5 years of data.
3. Builder follow-up: field staff ask owners or builders for actual start months. Permits for buildings with 1 to 4 units sampled at 1 in 50 and all permits for buildings with 5 or more units included.
4. Coverage add-ons: historical patterns adjust for pre-permit starts and late reports. No adjustment covers unpermitted building in permit areas, then weighted starts from 80 non-permit areas join starts from about 900 permit offices for total starts.
5. Seasonal adjustment and annualizing: X-13ARIMA-SEATS software removes seasonal patterns. For example, 100000 divided by 0.80 gives 125000 and 48000 times 12 gives 576000.
6. Monthly change: MoM % equals the SAAR difference over the prior SAAR times 100 using unrounded data. For example, (576000 minus 560000) over 560000 gives 2.9%. Each change carries a 90% confidence interval, and it counts only when the interval excludes zero.