US Macro

Housing Starts

Housing starts — the count of new privately-owned housing units on which construction was begun in a given month — is the most direct measure of US homebuilding activity. Because residential investment is interest-rate sensitive and leads the broader economic cycle by 6-12 months, housing starts is one of the most consistently useful leading indicators for recession-watching.

May 14, 2026HOUSThousing · us-macro · leading-indicator · construction

HOUSING_STARTS

—

The single most reliable leading indicator of US economic cycles. Housing starts respond to interest-rate changes faster than almost any other real-economy variable; when Fed tightening begins working its way through the economy, housing is the first sector to feel it. The 2007-2008 housing collapse preceded the GFC recession by roughly 12 months; the 2022-2024 housing slowdown preceded the (eventually averted) recession concerns of 2024-2025 by a similar lead time.

HOUSING_STARTS

—

What it measures

Housing starts is the number of new privately-owned housing units on which construction began during a given month, in seasonally adjusted annual rate (SAAR) form:

The series we track — FRED HOUST — is the seasonally adjusted annual rate, published monthly by the US Census Bureau in thousands of units. A reading of "1,400" means 1.4 million units at an annual pace. The release schedule: approximately the 17th of each month, at 8:30 AM ET, as part of Census's "New Residential Construction" report.

The Census report also includes:

Housing permits (leading indicator, with a 30-60 day lead over starts)
Housing completions (lagging indicator, when units are finished)
Single-family vs multi-family breakdown
Regional breakdowns (Northeast, Midwest, South, West)

Historical range: peak of approximately 2.3 million units in January 2006 (pre-bubble-burst); trough of approximately 478,000 in April 2009 (GFC). Recent readings (2024-2025) have been in the 1,250-1,400 thousand-units range — moderate but below the 1.5+ million pace that would be needed to satisfy long-term household formation demand.

Why it matters

Two angles.

The leading-indicator angle. Housing starts is one of the most consistently useful leading indicators for the US economy. The 2007-2008 collapse preceded the GFC recession by roughly 12 months. The 2022-2024 housing slowdown (single-family starts fell from a March 2022 peak of ~1.8M to roughly 950K in early 2023) was an early signal of Fed-tightening transmission to the real economy. Whenever the Conference Board's Leading Economic Index turns negative, housing-starts data is one of the most important inputs. The mechanism: housing is the most interest-rate-sensitive sector of the economy, so it's the first to feel Fed tightening; the slowdown then ripples through construction employment, building-materials demand, and broader consumer spending with multi-quarter lags.

The household-formation-and-supply-balance angle. US household formation runs at approximately 1.5-1.7 million per year over long horizons (driven by population growth, immigration, and demographic transitions). For the housing market to be in long-run balance, housing starts should match this pace. Sustained shortfalls (as in the 2010s, when starts averaged ~1 million per year) produce housing-supply shortages that show up as rising home prices, declining affordability, and political pressure for housing-policy reform. Sustained excess (as in 2004-2006) produces overbuilding that can collapse when conditions tighten. Current housing-starts levels of 1.3-1.4M suggest a modest structural supply deficit that's contributing to ongoing affordability challenges.

What moves it, and what it moves

Moves housing starts:

Mortgage rates (via UST10Y, ultimately via Fed policy). The dominant single driver — when mortgage rates rise sharply, buyer demand falls, builders pull back on speculative construction, and starts decline within 3-6 months.
Homebuilder confidence. The NAHB Housing Market Index (a separate monthly survey) tracks builder sentiment and is a useful leading indicator within the leading indicator.
Construction material costs. Lumber, steel, copper, and concrete prices affect builder margins; when costs surge (as in 2021), builders moderate starts.
Construction labor availability. Post-2017 immigration policies and post-COVID labor-supply shifts have affected construction labor; tighter labor markets can constrain how many starts builders can actually undertake.
Land-use regulation. State and local zoning, environmental review, and permitting timelines affect the volume of feasible projects. The substantial regional variation in housing starts (Texas/Florida vs. California/New York) is largely explained by these regulatory differences.

Housing starts moves:

Construction-sector employment (lagged by 1-3 months).
Building materials demand (lumber prices, steel demand).
Homebuilder equity prices (XHB ETF, individual homebuilder stocks).
Furniture, appliances, and other "new home" retail categories (with 6-12 month lag — first you start a house, then you furnish it).
GDP composition (residential investment is roughly 4% of total GDP; large swings produce visible GDP-growth-rate effects).
Federal Reserve policy expectations (housing-start weakness is read as a signal that tightening is working, supporting cut expectations).

A worked example: the 2007-2009 collapse

US housing starts peaked at approximately 2.27 million units (SAAR) in January 2006, during what was widely recognized at the time as a housing bubble. Single-family starts specifically had peaked at ~1.8 million units; multi-family added another ~470K.

The decline began gradually:

2006: 2.27M → 1.5M (year-end) — declining throughout the year as house prices peaked and the marginal buyer disappeared
Throughout 2007: 1.5M → 1.0M — the subprime crisis emerged in mid-2007 (Bear Stearns hedge funds failed in July); housing starts declined through the year
Throughout 2008: 1.0M → 550K — Lehman failed in September; financial conditions tightened dramatically; new construction collapsed
April 2009: 478,000 units — the trough. A 79% decline from the January 2006 peak over 39 months.

The 2009 trough level was the lowest in the housing-starts series' history (Census has data back to 1959). For comparison, the prior low was 798K during the 1981 Volcker recession.

The recovery was extraordinarily slow:

2009-2014: housing starts gradually recovered from 478K to ~1.0M — five years of construction running well below the long-run household-formation rate
2015-2019: continued recovery to 1.3-1.4M
2020-2022: surge to 1.8M (driven by post-COVID demand shift to housing, low mortgage rates)
2023: declined to ~1.3M as rates surged
2024-2025: stabilized at 1.3-1.4M

The full recovery to the 2006 peak took 16 years — single-family starts didn't sustainably return to the pre-bubble peak until 2022.

The episode produced two lasting insights: (1) housing-starts collapses are durable; even after a financial-crisis-induced collapse stabilizes, the recovery can take a decade or more; (2) sustained shortfalls in housing supply translate to multi-decade affordability problems, which is the central housing-policy issue of the 2020s.

The current cycle, and the open question

The structural debates:

Affordability ceiling. Current mortgage rates (6-7%) and home prices have made entry-level home purchases dramatically more expensive than at any time since the 1980s. Whether housing starts can sustain 1.3-1.4M annual pace at these conditions, or whether they'll decline further until affordability improves, is the central near-term question.
Demographic demand for housing. Long-run household formation has been running at ~1.5M per year, but with significant variation. Aging demographics, smaller household sizes, and immigration trends all affect this. Some economists argue household formation has structurally slowed to ~1.2M per year; others argue immigration plus delayed millennial household formation will produce a 1.7M+ pace through the 2020s.
Multi-family vs single-family divergence. Multi-family starts were soft in 2024 (apartment-construction pipeline retrenchment) while single-family was more stable. Whether these patterns persist or rebalance is meaningful for housing-supply composition.
Policy interventions. Federal, state, and local housing-policy debates around zoning reform, federal mortgage support, and direct construction subsidies are increasingly active. Any major federal housing-policy initiative could materially affect starts.

What you watch: the monthly Census release; the permits-to-starts spread (when permits exceed starts, builder caution is rising); the single-family vs multi-family breakdown; the NAHB Housing Market Index (builder confidence, monthly); homebuilder equity prices (a real-time read on market expectations); and regional breakouts (Texas/Florida vs California/Northeast).

FAQ

What counts as a 'housing start'?

A housing start is the beginning of construction on a new privately-owned residential structure. The Census Bureau counts a start when 'excavation for the footings or foundation is complete' (single-family) or 'when work has begun on the structure' (multi-family). The unit is the dwelling unit, not the structure — a 100-apartment building counts as 100 housing starts. The series excludes manufactured housing (mobile homes), publicly-owned units (subsidized public housing developments), and major renovations. The data is reported in 'thousands of units, seasonally adjusted annual rate' — so a reading of '1,400' means 1.4 million units at an annual pace, derived from the monthly count multiplied by 12 and seasonally adjusted.

What's the difference between housing starts and housing permits?

Permits are the leading indicator; starts follow with a 30-60 day lag. A builder obtains a building permit from the local government before construction can begin, then begins construction (the 'start') sometime in the following weeks or months. Permits are typically a stronger leading indicator because they capture the builder's planning decisions earlier in the cycle. The Census Bureau publishes both series in the same release ('New Residential Construction'). Most professional analysts watch both, and the gap between permits and starts can be a useful signal — when permits rise but starts don't follow, it usually means builders are getting permits but waiting for better market conditions before breaking ground.

Why are housing starts such a strong leading indicator?

Because residential construction is highly interest-rate-sensitive and ripples through multiple sectors of the economy. The transmission channels: (a) Mortgage rates affect home affordability and thus buyer demand — when rates rise, builders pull back on speculative construction; (b) Construction lending rates affect the cost of carrying inventory of unsold homes; (c) Material costs (lumber, steel, concrete) affect builder margins; (d) Labor availability affects how fast projects can complete. All four channels are sensitive to Fed policy and broader cyclical conditions. When housing starts decline meaningfully, it signals that interest-rate transmission to the real economy is working — and recessions typically follow within 6-12 months. The 2007-2008 housing start collapse preceded the GFC recession by roughly 12 months.

What's the difference between single-family and multi-family starts?

The Census Bureau breaks housing starts into 'single-family' (detached homes) and 'multi-family' (buildings with 2+ units, which is dominated by 5+ unit apartment buildings). The two cycles often diverge. Single-family starts are driven by individual buyer demand, mortgage affordability, and homebuilder confidence. Multi-family starts are driven by institutional investor demand for rental properties, local rent growth, and developer financing availability. In recent cycles, multi-family has been more volatile than single-family — the 2020-2022 apartment-construction boom (driven by low rates and strong urban rent growth) gave way to a 2023-2024 multi-family collapse (rates too high, vacancy rising in some metros). The aggregate housing-starts figure can mask these subsector dynamics.

What was the 2007-2008 housing start collapse?

The most dramatic single-cycle decline in housing starts in US history. From a January 2006 peak of approximately 2.3 million units (SAAR), housing starts fell to roughly 478,000 units by April 2009 — a 79% decline over 39 months. The collapse was the canonical signal of the housing bubble bursting and the broader financial crisis approaching. By the time NBER officially dated the recession start to December 2007, housing starts had already declined by approximately 40% from their peak. The episode produced the canonical 'housing leads the economy' observation that remains a cornerstone of recession-watching methodology. The full recovery took until 2022 — single-family housing starts didn't return to the 2006 peak for 16 years.