US Housing Starts Drop 5.2% in September: What It Means for Economic Growth

 

The Census Bureau released September new residential construction data on October 20. Per the official release, housing starts ran at a seasonally adjusted annual rate of 1,221,000, down 5.2% from August and 7.4% from a year earlier, while building permits came in at 1,400,000, down 0.2% monthly and 3.0% annually. Single-family starts were 859,000, down 4.6%, and single-family permits were 845,000, down 4.0%.

 

Before quoting any of those figures, one thing has to be established. The Census Bureau publishes 90% confidence intervals in the same release, and the 5.2% monthly decline in starts carries an interval of plus or minus 11.8%, meaning the decline cannot be statistically distinguished from zero. Only three changes in this release clear significance, all year over year: multifamily starts down 25.9%, total completions down 14.0%, and single-family completions down 11.0%.

 

The comparison against expectations also matters. Per TD Economics' calendar and consensus, starts were expected at 1,305,000 against an actual 1,221,000, a clear miss, while permits were expected at 1,385,000 against an actual 1,400,000, a slight beat. August starts were simultaneously revised up from 1,275,000 to 1,288,000, so the 5.2% decline is measured from a raised base.

 

Placed in the wider picture, the significance becomes clear. With the Federal Reserve having restarted rate hikes in September and the 30-year fixed mortgage rate at 7.40%, the most rate-sensitive sector of the economy is contracting while the composite purchasing managers' index sits at a five-year high.

 

 

 

The headline decline is not statistically significant. Starts fell 5.2% monthly against a 90% confidence interval of plus or minus 11.8%, so on the official basis the change cannot be distinguished from zero.

 

Only three year-over-year changes clear significance. Multifamily starts fell 25.9% against an interval of plus or minus 21.5%, total completions fell 14.0% against plus or minus 11.1%, and single-family completions fell 11.0% against plus or minus 10.0%.

 

Single-family and multifamily moved in opposite directions. Single-family starts rose 2.6% year over year while multifamily starts fell from 471,000 to 349,000, a 25.9% decline that accounts for most of the drop in the total.

 

Starts missed while permits slightly beat. Starts of 1,221,000 compared with a 1,305,000 consensus, permits of 1,400,000 compared with 1,385,000, and August starts were revised upward.

 

Builder confidence hit its lowest since 2012. The National Association of Home Builders index fell to 29 in October with the buyer traffic component at just 18, and 43% of builders cutting prices, the highest share across 2020 to 2026.

 

Rates are the binding constraint. Freddie Mac's October 15 survey put the 30-year fixed rate at 7.40% against 6.27% a year earlier, with the 10-year Treasury yield at 5.117%.

 

 

 

This section addresses a widely overlooked question, namely how much of the monthly movement in housing data is noise.

 

The Census Bureau attaches 90% confidence intervals to every new residential construction release because starts and completions come from a sample survey. Comparing those intervals against the reported changes item by item produces a clear conclusion.

 

Total starts fell 5.2% monthly against an interval of plus or minus 11.8%. Total starts fell 7.4% annually against plus or minus 11.5%. Single-family starts fell 4.6% monthly against plus or minus 10.0%. Single-family starts rose 2.6% annually against plus or minus 9.5%. Multifamily starts fell 7.2% monthly against plus or minus 27.7%. Every one of those five intervals includes zero, so the direction of change is statistically uncertain.

 

The three significant changes are multifamily starts down 25.9% year over year against plus or minus 21.5%, total completions down 14.0% against plus or minus 11.1%, and single-family completions down 11.0% against plus or minus 10.0%.

 

Building permits are a different case. They come from a non-probability sample, so the Census Bureau publishes no sampling error for them. That means permits cannot be assessed with confidence intervals, though their sample coverage is broader and their monthly volatility is usually lower than that of starts.

 

Revisions also belong in the judgement. August starts were revised up from an initial 1,275,000 to 1,288,000, with both July and August totals revised higher, so September's decline is calculated against a raised base. That explains how the same data can be described either as a sharp decline or as a base effect.

 

Taken together, the correct reading is that September's single-month decline is insufficient to support a conclusion on its own, while the genuinely reliable signals sit in the year-over-year completions and multifamily starts, and those two point to contracting completed supply and a clear cooling in multifamily development.

 

 

This data gets misread most often as across-the-board housing weakness, yet the components show two entirely different curves.

 

On single-family, September starts were 859,000, down 4.6% monthly but up 2.6% year over year, permits were 845,000, down 4.0% monthly and 4.8% annually, and completions were 840,000, down 2.4% monthly and 11.0% annually. Single-family starts are therefore still growing against a year earlier, with the pressure sitting in permits and completions.

 

On multifamily, the change is far sharper. September starts were 349,000, down 7.2% monthly and 25.9% annually against 471,000 a year earlier, while completions were 339,000, up 14.9% monthly but down 20.4% annually against 426,000 a year earlier. The only measure rising was permits, at 503,000 in September, up 5.9% monthly and 1.0% annually.

 

Three judgements follow from that structure. First, the 7.4% annual decline in total starts originates mainly in multifamily rather than single-family, so describing it as a collapse across housing misleads. Second, multifamily permits rising 5.9% while starts fell 7.2% says projects are being approved but deferred, which typically points to financing or construction cost constraints rather than vanished demand. Third, multifamily completions falling 20.4% annually means new rental supply will be smaller over the next year or two, which affects rents in the opposite direction from house prices.

 

Construction costs are one concrete reason for the multifamily cooling. Per market coverage, the decline in multifamily starts is tied to rising construction costs, which in turn relate to fuel prices and tariffs. That matches the cost signals visible in corporate earnings, where higher steel, copper and diesel prices feed manufacturing and construction simultaneously.

 

 

The three measures describe different stages of one process, and conflating them produces errors in sequence.

 

Building permits sit furthest forward. They record approved but not yet commenced projects, making them the most leading of the three and reflecting developers' judgement of demand one to two quarters out. Permits fell just 0.2% monthly this time, slightly better than expected, though per market coverage 1,400,000 represents roughly a three-and-a-half-year low, indicating front-end intentions have been running at low levels rather than having just turned down.

 

Housing starts record actual ground-breaking and feed directly into current-period residential investment, giving them the largest immediate effect on gross domestic product. The gap between starts and permits is a useful observation, since permits holding while starts fall means approved projects are being deferred, the classic signature of a cost or financing constraint.

 

Housing completions record finished delivery and determine the actual supply available for sale or rent, making them the most direct influence on prices and rents while carrying the longest lag. Total completions rose 2.2% monthly to 1,192,000 while falling 14.0% year over year, and that annual decline is one of this release's statistically significant changes. Falling completions mean shrinking new supply over the coming quarters.

 

Arranging the three in sequence yields a fuller picture, with front-end permits at low levels, mid-stage starts declining, and back-end completions contracting clearly year over year. That combination describes not a sudden disappearance of demand but a progressive slowdown through the entire supply chain under high rates, one that has already reached the delivery stage.

 

 

Housing is the most rate-sensitive sector of the economy, and the current rate environment is unambiguous.

 

Per Freddie Mac's October 15 primary mortgage market survey, the 30-year fixed rate stood at 7.40%, unchanged from the prior week, against 6.27% a year earlier, while the 15-year fixed rate was 6.71% against 6.73% the prior week and 5.52% a year earlier. That rate had risen for seven consecutive weeks to its highest level in nearly three years. On the long end, the 10-year Treasury yield sat at 5.117%, 19 basis points below the October 5 peak of 5.311%, which was the highest since 2002.

 

Policy direction is where all of this originates. The Federal Reserve raised rates 25 basis points to 3.75% to 4.00% on September 16 with a statement describing inflation as remaining elevated. Rising mortgage rates are therefore the consequence of a policy path rather than a transient market fluctuation. On how rate changes transmit into risk assets, see our analysis of interest rates and Bitcoin.

 

The demand-side response already shows in builder surveys. The National Association of Home Builders index fell to 29 in October from 32 in September, the lowest since June 2012, with all three components declining, current sales conditions at 33 down 2 points, sales expectations for the next six months at 32 down 5 points, and prospective buyer traffic at 18 down 5 points. Chief Economist Robert Dietz said demand persists but that higher mortgage rates, elevated construction costs and weaker consumer confidence are keeping more buyers on the sidelines, noting buyer traffic is at its lowest since April 2012 excluding the April 2020 onset of the pandemic.

 

Pricing behaviour supplies another layer of evidence. In October 43% of builders cut prices, the highest share across 2020 to 2026, with the average cut at 6% and unchanged for a seventh straight month, while 65% used sales incentives, the nineteenth consecutive month at or above 60%. A rising share of price cuts with an unchanged average size says more builders are discounting while the room to discount on any individual project has reached some boundary.

 

Costs are pressing at the same time. NAHB Chairman Bill Owens noted mortgage rates well above 7%, with higher gasoline prices and inflation weakening buyer traffic, alongside rising costs for construction loans, building materials and diesel. Regional three-month moving averages show uneven pressure, with the Northeast at 38 down 1, the Midwest at 40 down 3, the South at 30 down 1, and the West at 26 down 2.

 

 

Placed in the macro picture, this data describes a divergent economy rather than a recessionary one.

 

Housing's direct contribution runs through residential investment. Falling starts mean a weaker contribution to current gross domestic product, and completions down 14.0% year over year mean that weakness persists for several quarters. Goldman Sachs estimates 2027 existing-home sales near 3.9 million, roughly 26% below 2019 levels, indicating that depressed transaction activity is treated as a medium-term rather than short-term condition.

 

Aggregate data over the same period points the other way. The September S&P Global composite output index was 58.4, the strongest since July 2021, with flash services at 58.7 far above the 56.0 expected, and chief business economist Chris Williamson putting the corresponding annualised growth near 5%. While housing sits at post-pandemic lows with builder confidence at a 14-year low, services and overall output are expanding strongly.

 

That divergence is the standard shape of the middle of a hiking cycle. Rate-sensitive sectors come under pressure first while services, a far larger share of the economy, keep expanding on employment and income, so weakness in aggregate measures lags housing. The policy implication is that housing data alone cannot turn the Federal Reserve, since it is neither the main source of inflation nor the main support of aggregate demand.

 

The flash purchasing managers' index on October 23 is the next verification point. Per TD Economics' consensus, manufacturing is expected at 55.5 and services at 57.9, against final September readings of 55.9 and 58.8. If aggregate data keeps strengthening while housing keeps weakening, the divergence widens and December hike pricing consolidates. If aggregate data moderates in step, that means the effect of rates is spreading from housing into the broader economy. The framework sits in our October US flash PMI preview.

 

For risk assets, the transmission path from housing is indirect. Per market coverage, pricing for an October hike stood near 17% when the data arrived, and housing weakness did not change that, because it is insufficient to offset strength in the price components and employment. The sequence still runs from the size of the data surprise into rate expectations, then through the dollar and real rates into valuations. As of 4:16 p.m. Eastern on October 20, bitcoin traded near $82,536, up 0.72% over twenty four hours, for a market capitalisation around $1.66 trillion.

 

Follow BTC spot pricing on MEXC to read US economic data against market prices, where the pair trades as BTC against USDT

 

 

Judgement from here rests on six areas.

 

First is the flash purchasing managers' index on October 23, which determines whether housing weakness is isolated or the leading edge of a broader slowdown.

 

Second is the Census Bureau's October new residential construction release on November 18, when September figures get revised. Given that August was revised up this time, a single month's initial print carries limited signal and the direction of revision is itself information.

 

Third is the path of mortgage rates. The 30-year fixed rate has now held at 7.40% for two weeks and correlates closely with the 10-year Treasury yield, currently 5.117%. Whether rates retreat determines whether the demand-side constraint eases.

 

Fourth is the builder confidence components, particularly buyer traffic. That component now sits at 18, its lowest since April 2012, and it leads actual sales, making it the earliest source of any recovery signal.

 

Fifth is the behaviour of price cuts. With 43% of builders discounting at a stable average of 6%, a widening average would indicate rising inventory pressure, while a falling share would indicate improving demand.

 

Sixth is the gap between multifamily permits and starts. Permits rose 5.9% while starts fell 7.2%, and if that inversion persists, cost and financing constraints remain in place, whereas starts catching up to permits would mean those constraints are easing.

 

 

For James Mitchell, the most valuable element of this data is not the 5.2% decline but the plus or minus 11.8% confidence interval the Census Bureau attaches to it. That interval means the headline cannot be statistically distinguished from zero, yet markets and media will still price and narrate around 5.2%. Only three year-over-year changes pass the significance test, multifamily starts down 25.9%, total completions down 14.0%, and single-family completions down 11.0%. Together those point to supply-side contraction rather than a one-month collapse in demand, and contracting supply pushes prices in the opposite direction from weakening demand.

 

Two misreadings recur. The first equates a falling total with across-the-board housing weakness. Single-family starts actually rose 2.6% year over year, and the 7.4% annual decline in the total comes mainly from multifamily falling from 471,000 to 349,000. Single-family and multifamily have different drivers, the former turning on mortgage rates and household income and the latter on development financing costs and rent expectations, so combining them discards all the useful information. The second ignores the gap between permits and starts. Multifamily permits rising 5.9% while starts fell 7.2% says projects are approved and then deferred, locating the constraint in cost and financing rather than demand, and the conditions for easing that constraint differ entirely from a demand recovery.

 

The macro divergence deserves more attention still. Housing sits at post-pandemic lows with builder confidence at its weakest since 2012, while the composite output index over the same period reads 58.4 and implies roughly 5% annualised growth. That is not a contradiction but the standard mid-cycle pattern, with rate-sensitive sectors leading on the downside while services keep expanding on employment and aggregate weakness lags housing. For investors the implication is that housing data serves as early evidence of rate pressure but not as grounds for expecting a policy turn, because the Federal Reserve's decision inputs are currently prices rather than housing activity.

 

For bitcoin holders, the position in that chain needs to be clear. Housing data does not act on risk assets directly but through how it changes the market's read on growth and the rate path, and after this release pricing for an October hike stood near 17%, essentially unmoved. That indicates weakness in a single sector cannot shift liquidity expectations in the current environment. What can shift pricing is a turn in the inflation components, or housing weakness beginning to surface in employment data. The latter is the milestone worth tracking, because changes in construction employment would convert a sectoral problem into an aggregate demand problem.

 

 

 

Housing starts ran at an annual rate of 1,221,000, down 5.2% monthly and 7.4% annually, building permits were 1,400,000, down 0.2% monthly and 3.0% annually, and completions were 1,192,000, up 2.2% monthly and down 14.0% annually. Single-family starts were 859,000, down 4.6%, and single-family permits were 845,000, down 4.0%. The Census Bureau published the data at 8:30 a.m. Eastern on October 20, with the next release scheduled for November 18.

 

 

It requires caution. The Census Bureau attaches a 90% confidence interval of plus or minus 11.8% to that monthly decline, and because the interval includes zero, the change cannot be statistically distinguished from zero. Only three year-over-year changes in the release clear significance, multifamily starts down 25.9%, total completions down 14.0%, and single-family completions down 11.0%. August starts were also revised up from 1,275,000 to 1,288,000, so the decline is calculated from a higher base.

 

 

They moved in opposite directions. Single-family starts of 859,000 fell 4.6% monthly but rose 2.6% year over year. Multifamily starts of 349,000 fell 7.2% monthly and 25.9% annually against 471,000 a year earlier. The 7.4% annual decline in total starts therefore originates mainly in multifamily rather than single-family. Their drivers differ, with single-family turning on mortgage rates and household income while multifamily turns on development financing costs and rent expectations.

 

 

They are three stages of one process. Building permits record approved but not yet commenced projects, making them the most leading measure and reflecting developer judgement one to two quarters ahead. Housing starts record actual ground-breaking and feed directly into current-period residential investment, carrying the largest immediate effect on gross domestic product. Completions record finished delivery and determine the actual supply available for sale or rent, the most direct influence on prices and rents but with the longest lag.

 

 

Because rates and costs are pressing simultaneously. The National Association of Home Builders index fell to 29 in October, the lowest since June 2012, with buyer traffic at just 18. Freddie Mac's October 15 survey put the 30-year fixed mortgage rate at 7.40% against 6.27% a year earlier. The association's chairman cited rising costs for construction loans, building materials and diesel, while its chief economist noted demand persists but more buyers are waiting. Currently 43% of builders are cutting prices, the highest share across 2020 to 2026.

 

 

Unlikely. The Federal Reserve already raised rates to 3.75% to 4.00% on September 16, and its current decision input is inflation rather than housing activity. Over the same period the September S&P Global composite output index was 58.4, the strongest since July 2021, implying roughly 5% annualised growth. Per market coverage, pricing for an October hike stood near 17% when the data arrived, and housing weakness did not change it. Housing is typically the first sector to come under pressure in a hiking cycle, so its weakness alone is not grounds for a policy turn.

 

 

The effect is indirect. Transmission runs from the size of the data surprise into rate expectations, then through the dollar index and real rates into risk asset valuations and liquidity expectations. This release did not materially change hike pricing, so its direct effect on risk assets is limited. What actually warrants tracking is whether the inflation components turn, and whether housing weakness begins surfacing in construction employment, since the latter would convert a sectoral problem into an aggregate demand problem.

 

 

The information above is provided for general market information and analysis only and does not constitute investment advice, financial advice, legal advice, tax advice or a recommendation to trade. Macroeconomic data and asset prices can both move sharply, and historical relationships, survey indicators and market-implied probabilities do not guarantee future results. The housing data, confidence intervals, survey indices, interest rates and asset prices cited here reflect publicly available information at the time of publication, initial economic data can be materially revised in later releases, building permits derive from a non-probability sample for which no sampling error is published, market-implied probabilities change continuously, and actual policy decisions are determined by the official statements of the Federal Open Market Committee. Statements about statistical significance rest on the officially published 90% confidence intervals and do not assert the true direction of the data. The response of digital assets to macroeconomic data is indirect and unstable, with no reliable fixed relationship. Readers should conduct their own research and make decisions based on their own financial circumstances, investment objectives and risk tolerance, consulting a qualified professional where appropriate. The MEXC Crypto Pulse team accepts no liability for any direct or indirect loss arising from the use of this information.

 

 

James Mitchell specializes in technical analysis, market trends, and trading strategies for both Bitcoin and altcoins. Based in London, he has over 10 years of experience in financial markets. Before joining MEXC Learn, James worked as a senior analyst at a leading European investment firm, where he developed expertise in risk management and quantitative trading. His transition to cryptocurrency markets began in 2017, and he has since become recognized for his data-driven approach. He holds a Master's degree in Financial Economics from the London School of Economics. His analytical approach combines traditional technical analysis with on-chain metrics to provide readers with actionable insights.

 

Areas of Expertise: Technical Analysis, Market Trends and Cycles, Trading Strategies, Bitcoin and Altcoin Analysis, Risk Management.

 

 

 

Want the fastest access to MEXC's latest updates? Join our official Telegram group now!

MEXC Platform: Download App | Zero Fee Trading | Buy USDT Now

Join MEXC Community: X (Twitter) | Telegram | Discord

Get Insights: MEXC Learn | Terms of Service | MEXC Blog | MEXC News

Account Verification: Understand KYC | How to Complete KYC

External Content Platforms: Substack | Medium | Paragraph | LinkedIn | X(News)