SG Group · October 21, 2026
Can the equity recovery reach housing and households?
US trading date: October 20, 2026. Information cutoff: October 21, 05:25 JST. Read index divergence, September US housing and state employment, August European construction and September Canadian prices with their periods and uncertainty kept distinct. Use the free section’s housing table to compare permits, starts, completions and the uncertainty intervals shown for monthly changes in starts and completions, separating reported values from evidence still needed before extending one equity session into a conclusion about the whole economy.
1．US equity gains were uneven, while housing weakness requires careful interpretation
This October 21 edition covers regular US trading on October 20, local time. After the cash equity session ended, the S&P 500 stood at 7,712.57, up 21.84 points; the Dow Jones Industrial Average was 52,495.88, up 529.85; and the Nasdaq Composite was 26,424.82, down 6.07. Expressed to two decimal places, the respective changes were +0.28%, +1.02% and -0.02%. Together, the three indexes show that US equities did not rise with uniform strength.[6][7][8][9]
The dates need to be distinguished. These are morning observations on October 21 in Japan, but the New York trading date is October 20. The regular session covered here ended at 4 p.m. locally, or 5 a.m. the following morning in Japan. The reported levels are post-session cash equity index values, not futures or contracts for difference. Prices of commodities or currencies trading overnight are not being grouped with them as if all were finalized at the same moment.
Reporting on the session highlighted corporate profit news and easing fuel-cost pressure, while weakness in some large technology stocks contributed to differences among indexes. This edition does not evaluate individual companies or securities. At the market-wide level, it examines how expectations for profits can coexist with concerns about interest rates and household burdens. Rising share prices should not be expanded into evidence that every part of the economy has improved.[6]
US housing starts released on the same day were 1,221,000 at a seasonally adjusted annual rate for September. The point estimate was down 5.2% from the previous month, but the confidence interval for that change includes zero. Noting a figure that suggests housing weakness is different from asserting that the direction of the decline is statistically certain. Equally, this uncertainty does not establish that financing or affordability problems do not exist. The interpretation should neither extend nor excessively narrow what the data can support.[1]
The free section proceeds through the stages from housing authorization to starts and completion, the breadth of changes in state employment, and European and Canadian releases covering different periods. The purpose is not to grade equities and the real economy on one scale. It is to explain where differences and delays arise as financing conditions reach construction activity, employment reaches household spending, and energy reaches costs.
The S&P 500, Dow and Nasdaq Composite do not have identical constituents or weighting methods. A larger Dow gain therefore does not show that every US industry improved to the same degree. A small Nasdaq decline does not show a comprehensive retreat in demand for technology or capital investment. Differences among indexes establish that price formation was uneven that day, but do not automatically explain the composition of real economic activity.
Higher prices also need to be separated from the quantity of financing. A rising equity index does not mean companies received an equivalent amount of new funding that day. Trading existing shares is distinct from raising funds through new equity issuance or borrowing. Market valuations can influence financing conditions, but contracts, credit decisions, issuance and loan execution lie in between. This distinction is fundamental when connecting equities with housing or capital expenditure.
The relationship between profits and rates cannot be assigned one fixed direction either. Expectations for larger future profits can support valuations, while more demanding interest-rate or risk assumptions used to convert those profits into present value can work in the opposite direction. A business requiring borrowing also differs from one able to use internal funds. An aggregate index combines these different circumstances.
Housing makes those differences easier to understand. Buyers face long-term repayment obligations, while builders face financing burdens during construction. Even if rates and prices move in the same direction, existing contracts, refinancing, new purchases and projects already under construction respond at different times. A one-day change in equity valuations cannot immediately be translated into an improvement in households' ability to acquire a home.
The approach here is first to establish the completed trading result and then identify the activity and period measured by each statistic. Differences need not be dismissed as anomalies: different parts of the economy may be moving at different speeds. The United States remains the focus, with European and Canadian evidence used for comparison and links to Central and South America distinguished as trade, income and financing channels. The regional picture is built from supported evidence rather than filling gaps with unverified same-day news.
2．Housing permits, starts and completions measure different stages
Scroll within this region to read the table.
| Stage | Units | Monthly change and caveat |
|---|---|---|
| Permits | 1,400,000 | -0.2% |
| Starts | 1,221,000 | -5.2% ±11.8 percentage points; 90% interval includes zero |
| Completions | 1,192,000 | +2.2% ±10.8 percentage points; 90% interval includes zero |
The starting point for reading US housing statistics is not to combine permits, starts and completions into a single interchangeable activity measure. Permits concern authorization, starts concern the beginning of construction, and completions concern movement toward usable supply. These stages involve time lags, and the three figures released for one month do not track the same group of homes. Falling permits and rising completions are therefore not necessarily inconsistent. Previously started projects can finish while new plans become more cautious.
September housing starts, released by the Census Bureau and Department of Housing and Urban Development on October 20, were 1,221,000 at a seasonally adjusted annual rate. The estimate was 5.2% below the revised August rate of 1,288,000. However, the 90% confidence interval around that monthly change is plus or minus 11.8 percentage points and crosses zero. A lower published point estimate must therefore be distinguished from statistical confirmation that the actual change was a decline. A large reported decline does not make its direction certain.[1]
Permits were at an annual rate of 1,400,000, down 0.2% from the revised August rate of 1,403,000. Completions were 1,192,000, estimated to be 2.2% above August's 1,166,000. The monthly completions change also has an interval, plus or minus 10.8 percentage points, that includes zero. By contrast, the year-over-year completions decline was 14.0%, plus or minus 11.1 percentage points, and that interval excludes zero. Monthly and annual comparisons have different reference periods and uncertainty; selecting whichever supports a preferred view is not an adequate assessment of activity.[1]
The annualized unit also matters. The 1,221,000 figure does not mean that many homes were started during September alone. It expresses the seasonally adjusted monthly pace in annual terms to facilitate comparison. It is neither a finalized annual total nor a guarantee that the same number will be built over the following year. Actual full-year starts or unadjusted regional counts require their corresponding series. Confusing these units can exaggerate the scale of a supply shortage or construction activity.
Single-family and multifamily housing also differ beneath the aggregate. Single-family starts were at an annual rate of 859,000, estimated to be 4.6% below August's 900,000, with an interval that also includes zero. Starts in buildings containing at least five units were 349,000. That category is not all multifamily housing: including buildings with two to four units changes the coverage. A difference between a news report's multifamily total and an official five-or-more-unit figure is not automatically an error.[1]
Industry assessments and statistical measurements have different roles as well. The National Association of Home Builders cited financing costs, building-material costs, labor shortages and affordability constraints in discussing the release. These are relevant descriptions of conditions perceived by the industry, not a causal estimate from this release of how many starts each factor removed. Rates should not be made to explain the entire movement: land, permitting, labor and sales expectations can overlap.[2]
Mortgage rates affect buyers' monthly payments, while construction-loan rates affect builders' costs during the project. They are not the same rate, and terms differ with maturity, collateral, underwriting and fixed versus floating arrangements. A reported policy-rate change does not mean every mortgage or construction loan changes on the same day by the same amount. Differences between existing and new contracts introduce delays as monetary conditions pass through renewals.
Following construction stages helps illuminate those delays. A permit may not proceed to a start until sales expectations and financing are adequate. After a start, expenditure increases as work progresses; after completion, delivery or sale can move the project toward cash recovery. Housing counts alone, however, do not reveal whether financing came from borrowing or internal funds, or which contracts were renewed. Connecting construction counts to cash flow requires information outside the statistics' scope.
Nor can the difference between permits and starts be treated directly as the change in unstarted homes. The reported figures are annualized activity flows, unlike the stock of unstarted authorized units at month-end. Cancellations, expirations and starts of projects permitted at different times also intervene. Distinguishing flows from stocks avoids the mistaken conclusion that every home represented by the difference between permits and starts is waiting in a backlog.
Housing connects financing, employment, materials and regional supply, but a single starts release is not a complete picture of that system. What can be established here is September's point estimates and the uncertainty of the comparisons. Separately following the movement from permits to starts, trends across several months and supply after completion provides a stronger explanation than one month's fluctuation. One-day equity changes and months-long construction progress should not be read at the same speed.
3．Read employment breadth through both counts and statistical confidence
Employment is an important link between housing and households. Labor income supports consumption and repayment, while construction and related services also create jobs. Yet employment need not move in the same direction and proportion in a month when housing starts decline. Existing projects continue, and employers can adjust hiring, hours, outsourcing and deployment in response to orders. Headcounts do not capture every form of adjustment.
The September US state employment release, published on October 20, reported statistically significant monthly nonfarm payroll increases in two states and a decrease in one. Employment was described as essentially unchanged in the other 47 states and the District of Columbia. This does not mean that not a single job changed in those 47 states. Estimates can move without being judged to exceed the uncertainty of the survey. A numerical movement and a clearly established change are different things.[3]
The significant increases were 14,900 jobs in New Jersey and 12,500 in South Carolina. Wisconsin recorded a decline of 13,700. This does not mean those three states explain the national change. An aggregate across states and a list of statistically significant state changes answer different questions. Adding only those three changes does not reconstruct national employment, and activity elsewhere must not be assigned a value of zero.[3]
Unemployment rates declined significantly from the previous month in five states; 45 states and the District of Columbia showed no significant difference. Again, the absence of a significant increase is not equivalent to strength in every local labor market. The unemployment rate is the proportion of the labor force that is unemployed, and it can change because of labor-force participation as well as employment. Both numerator and denominator are needed to understand why a rate falls.[3]
This distinction also helps when comparing establishment employment with household employment status. The former counts jobs and the latter describes people's status, so their coverage and aggregation are not identical. Multiple jobholding, survey scope and revisions can produce different short-run impressions. Rather than retaining one favorable result and ignoring another, the first task is to establish what each measures.
State data reveal regional variation that a national average can obscure. The same numerical change has a different proportional meaning in large and small states. Industrial composition also affects sensitivity to rates, energy, trade and public spending. But the employment changes in the named states cannot be attributed solely to one policy or industry from these totals. State aggregates are a starting point, followed where necessary by industry detail or evidence over a longer period.
Several stages connect hiring decisions with household expenditure. More jobs can expand income opportunities, but spendable resources also depend on hours, wages, taxes, living costs and debt service. If necessary expenses rise faster than nominal income, discretionary room may not expand. Employment growth alone therefore does not establish a comprehensive improvement in consumption or housing affordability.
Conversely, households' circumstances can change through hours and real purchasing power even if headcounts are flat. Firms adjusting through shorter hours affect the employment count at a different time from firms reducing staff. If existing jobs are retained while hiring is restricted, jobseekers and incumbent workers can experience different conditions. The level of employment and opportunities to move into new work are distinct dimensions.
Matching reference periods is useful when comparing housing and labor data, but does not establish causality. Both releases include September observations, yet their methods and units differ. A weak housing point estimate and limited statistically significant state employment changes can be described as simultaneous observations. Calling that evidence that housing worsened national employment would require additional support. Co-occurrence and causation must remain separate claims.
The purpose of placing monthly statistics in a daily market article is not to reduce equity performance to one cause. It is to distinguish traded expectations from the newly established limits of evidence about the real economy. Regional employment and household burdens do not change everywhere at once on a day when profit expectations shift. State data help assess breadth, but do not independently establish that the whole country moved in one direction.
4．Align publication dates and reference months for Europe and Canada
When connecting a US session with Europe or Canada, numbers visible on the same day need not measure the same period. October 20 equity prices describe that day's trading, but US housing data released that day cover September and European construction production covers August. Canada's September consumer prices were released the previous day, October 19. Separating when markets received information from when activity or prices were observed makes the chronology clearer.
Eurostat's October 20 release estimated that seasonally adjusted construction production rose 1.8% in the euro area and 2.0% in the European Union in August from July. The annual increases were 1.4% and 2.6%, respectively. Although the direction contrasts with US housing starts, European construction production and US new residential starts are not the same indicator. Coverage, ongoing activity versus new initiation, and the reference month all differ. These two observations alone cannot establish that Europe's economy outperformed America's.[4]
A production index compares activity over a period. Its unit differs from the number of homes started, and construction includes nonresidential work. Checking coverage before comparing growth rates clarifies what the comparison means. The common feature is sensitivity to financing and construction costs, not that these are identical markets responding at the same speed.
The August increase establishes an initial estimate of greater European activity in that month. It does not directly demonstrate better October orders or future lending conditions. Previously contracted work may have progressed, or new demand may have contributed. Orders, permits and credit conditions are needed to distinguish these possibilities. Evidence about completed past activity should be used separately from evidence about future activity.
Canada's September consumer price index increased 3.1% year over year, accelerating from August's 3.0%. Excluding gasoline, however, the increase slowed to 2.2% from 2.4%. Acceleration in the headline and deceleration in an exclusion series can coexist. The headline reflects the overall prices households face, while exclusion measures can help assess breadth. Neither needs to be treated as the sole correct measure.[5]
Month over month, Canada's headline index rose 0.1% before seasonal adjustment and 0.4% after it. The difference reflects whether recurring seasonal movements are adjusted. Comparing the annual 3.1% directly with the monthly 0.4% does not demonstrate a sharp disinflation: the periods and adjustment methods differ. A number's unit, monthly or annual comparison, and seasonal-adjustment status belong together.[5]
Energy is one shared channel linking the United States, Canada and Europe. Yet a one-day crude-price movement does not pass unchanged into household prices in each country. Exchange rates, taxes, refining, distribution, contracts and subsidies intervene. The lag from import costs to consumer prices is not constant either. Temporary fuel-cost relief and a lasting moderation in overall household living costs should not be treated as one event.
The same energy-price change can also work differently in producing and consuming regions. Households and businesses using fuel experience a cost change, while producers experience changes in revenue or investment plans. Both groups exist within one country. Labeling an oil-producing country as an unqualified beneficiary of higher prices is therefore inadequate. Producer income and consumer burdens need separate examination.
Publication order matters too. Canada's September prices were already public on October 19, not a new simultaneous release for the October 20 US session. European August construction and US September housing were same-day releases, but arrived in their respective local time sequences. Information released after the close should not be projected backward as a cause of regular-session price changes. Availability must be established before causality is discussed.
European construction, US housing starts and Canadian prices are therefore not competing answers to one question. They describe different stages, regions, periods and dimensions of costs. Even when considering shared rates or energy changes, results need not converge simultaneously. Separating what has been observed from what remains to be checked allows comparisons to retain their meaning. Across countries, aligning definitions before comparing magnitudes is especially valuable.
5．A rising session is not the same conclusion as broad economic improvement
The central reading question is what a rising equity index establishes and what it leaves unproven. Prices incorporate future profits, discount rates and perceptions of risk, while housing and labor statistics measure past activity. An equity gain on a day with a weak housing point estimate is therefore not inherently contradictory. Participants may emphasize other information or may already have expected weakness. A single statistic cannot fully explain one day's index change.
First, improving profits and improving demand overlap but are not identical. Profits depend on input costs, fixed costs, interest expense, exchange rates and business mix as well as revenue. Lower costs can improve profits without a large increase in sales volume. Conversely, revenues can grow while faster cost growth compresses profits. A market-wide explanation should distinguish the breadth of sector profits from measures of household and business demand.
Second, an index gain and the breadth of its constituents' gains need not coincide. A few large constituents can dominate a capitalization-weighted index. Price-weighted and capitalization-weighted measures also assign different importance to the same market. Divergence between the Dow and Nasdaq does not directly establish an equal-sized flow of money from one to the other. An index summarizes prices, not the amount of capital flowing in.
Third, financial conditions are not determined by policy rates alone. Long-term yields, credit premiums, underwriting, collateral valuations and financing maturities all shape actual burdens. A lower rate does not necessarily constitute comprehensive easing if available borrowing shrinks. Funding a new housing project also differs from refinancing existing debt. The direction of rates and whether financing can be obtained should be examined separately.
Fourth, slower price growth does not mean the price level has returned to where it was. Annual inflation can slow while prices remain higher than a year earlier. Households pay expenditure amounts, not inflation rates alone. Purchasing power cannot be established without considering income growth and committed expenses. Canada's divergence between headline and gasoline-excluded inflation illustrates why one number is insufficient.
The free section's purpose is to keep these questions separate. A future increase in starts would provide evidence of more new construction initiation. That one observation would not prove a permanent improvement in lending conditions. More employment supports income, but real purchasing power and housing affordability require their own checks. Identifying what an indicator is meant to answer prevents one favorable release from becoming a solution to every problem.
The same caution applies in reverse. Fewer starts do not establish that every household cut spending. Existing homes, rentals, renovation and relocation involve activities distinct from new starts. No significant employment change in many states also does not prove an absence of household difficulty. Stating what statistics do not measure is not avoiding judgment; it keeps explanations within the evidence's legitimate scope.
Connections beyond North America require the same sequence. Changes in US demand can affect trading partners through orders and logistics, but export composition, currencies and contract duration differ. Employment income and remittances can provide links to Central American households, while commodity demand and export revenue provide other links to South American producing regions. These are channels to investigate, not claims that fresh same-day statistics were released everywhere.
Facts and outlooks should also remain distinct. The completed index values, September housing estimates and state employment, August European construction, and September Canadian prices are observations with specified periods. How future demand, credit, costs and policy responses combine belongs to conditional analysis. Rather than blending a forecast into a reported fact, identifying what further information would change the explanation lets readers compare subsequent releases within the same framework.
The main takeaway from the free section is to read market outcomes, economic activity and household burdens on their respective time scales. Index differences illuminate market composition; construction stages illuminate project progression; statistical significance illuminates confidence in labor changes; and inflation components illuminate the breadth of costs. Preserving those distinctions and limits, rather than forcing one strength-or-weakness verdict, provides a foundation for understanding the next release.
A consistent sequence also helps compare different interpretations. An explanation emphasizing profits can differ from one emphasizing household burdens despite using the same numbers. Before choosing between them, check whether they address the same period, economic actor and transmission channel. Short-term prices and long-term supply answer different questions and should not automatically be used to refute one another.
Today’s Market Takeaways
US cash indexes did not rise uniformly. The weak housing point estimate needs to be read alongside its confidence interval and construction stage. State employment significance, national reference periods and inflation components provide the starting point for assessing real-economy breadth.
References for the free section
- Census/HUD: New Residential Construction, September 2026 (October 20)
- NAHB: September housing starts (October 20, 2026)
- BLS: State Employment and Unemployment, September 2026 (October 20)
- Eurostat: August 2026 construction production (October 20)
- Statistics Canada: September 2026 CPI (October 19)
- AP: October 20 US market close (updated October 21, 05:22 JST)
- Yahoo Finance: S&P 500, October 20 cash close
- Yahoo Finance: Dow Jones Industrial Average, October 20 cash close
- Yahoo Finance: Nasdaq Composite, October 20 cash close
This SG Group publication provides general information and education. It is not investment advice or a solicitation to buy, sell, hold or allocate financial products. Statistics may be revised, and conditional analysis does not guarantee future outcomes.
Related reading
Analysis & Commentary
The rest of this article is paid content
Read only this article
Unlock the paid section of this article with a one-time purchase.
One-time payment Pay Per Article l Daily Europe Americas Market Analysis
$1.00
Payment is completed on Stripe's secure checkout page.
Subscriptions renew automatically and can be canceled anytime. Cancellation terms are described in the Legal Notice (Commercial Transactions).
Verify your access without leaving this page.
This browser does not support passkeys. Please use another method.
Subscriptions and purchased articles
Check your subscription, purchased articles and payment method without leaving this page.
JavaScript is required to manage subscriptions and purchases here.
Subscriptions and purchased articles
Review and manage your Reader Account here. You stay on this article the whole time.
Sign in to your Reader Account to see your subscriptions and purchased articles.