The Pipeline Decoupling: Normal Inflow, Record Backlog
The three node readings do not move together, and the divergence is the finding. At the front of the pipeline, Building Approvals & Commencements (21510) sits at +0.4936σ — about half a standard deviation above its fifteen-year mean, its highest reading since the September quarter 2021. Behind it, New Housing Supply Forecasts (21520) reads +1.8182σ, the highest in its record, driven by a residential work-yet-to-be-done vector at a series high. Financing the whole structure, Construction Finance & Capacity (21530) reads +2.1333σ on lending capacity and lending volume, with its sector-health vector temporarily absent.
| Node | Terminal Z-score | Period | Read |
|---|---|---|---|
| Building Approvals & Commencements (21510) | +0.4936σ | Q2 2026 | A little above mean — inflow close to its long-run norm |
| New Housing Supply Forecasts (21520) | +1.8182σ | Q2 2026 | Well above mean — committed pipeline at record depth and cost |
| Construction Finance & Capacity (21530) | +2.1333σ | Q4 2025 | Well above mean on two of three vectors — insolvency signal pending |
Note the reference periods. The two physical-pipeline nodes are certified to Q2 2026; the finance node’s most recent certified reading is Q4 2025, because its insolvency input is published annually and lags the lending inputs.
21510 read above 21520 in every quarter from 2011 to 2015, and again through the HomeBuilder surge of late 2020 and 2021. Since the December quarter 2021, 21520 has been the higher of the two in every quarter, 19 in a row. The margin peaked at 2.05 standard deviations in the March quarter 2023, when the inflow contracted, and stands at 1.32 at the June quarter 2026. Source: APN Codex node readings; baseline 2011 to 2025.
The composite view. The series also carries a single 21500 reading of pipeline against need. It takes the two physical-pipeline nodes at a quarter each and Housing Supply Adequacy (21540), population growth less dwelling-stock growth, at a half with its sign inverted, so that a higher reading means a stronger pipeline relative to need. At the March quarter 2026, the latest quarter in which all three are available, the pipeline half contributes +0.4067 and the need half −0.1875, for a composite of +0.2192σ (position 57.4, ranked against 55 quarters). A record pipeline reads as only moderately strong once it is set against population growth that is still running a little ahead of the dwelling stock.
The two halves usually pull in opposite directions. In 2021 the border closure pushed both the same way, and the composite reached its record of +1.5246σ in the June quarter 2021. In 2023 the migration surge took the need half to −1.00 and the composite to about −1.0, while the pipeline half stayed close to zero. Source: APN Codex node readings; baseline 2011 to 2025.
Approvals — a little above mean
53,952 dwellings approved, Q2 2026. Against a fifteen-year mean of 48,927, the inflow of statutory permission is a little firmer than ordinary. Neither a planning boom nor a planning slump is visible in the certified data.
Commencements — a little above mean
52,166 dwellings commenced, Q2 2026. Above the 48,157 mean and up from 45,777 in the March quarter, inside normal variation. Commencements ran at +0.82σ in Q4 2025 — the front of the pipeline is not stalling.
Spread — mildly wide
1,786 approved but not commenced. The node’s high-signal threshold is |1.50σ|; the spread has crossed it in only seven quarters since 2011, most recently in Q1 2023 in the negative direction, when commencements outran approvals; Q4 2025 came close at −1.30σ.
Under construction — record
248,733 dwellings under construction. About 25% above the 198,223 mean and the highest in the record, above the 2022 peak, after easing to +0.41σ at the end of 2024.
Work yet to be done — record
$84,731.4m of residential work outstanding. About double the $41,997.2m mean and the highest reading in the series. A new series high has been set in each of the last eight quarters.
Work done — elevated
$27,867.1m of residential work completed in the quarter. Output is above mean and rising, from +0.71σ in Q1 2025. The sector is producing; it is not keeping pace with the value of what is queued.
Lending limits — far above mean
$32,362.2m in land-development credit limits, Q4 2025. Regulatory lending capacity for development is at its most elevated point in fifteen years, against a $21,426.1m mean.
Construction lending — far above mean
$12,865.6m in business lending to construction, Q4 2025. Nearly double the $6,827.4m mean. Capital is not only available but deployed.
Insolvencies — substituted
NIL RETURN for Q3–Q4 2025. The construction-insolvency vector is sourced annually and the FY2025–26 figures arrive in December 2026. A neutral value stands in for it until then.
The analytically significant relationship in the series is inside New Housing Supply Forecasts (21520), between the count of dwellings under construction and the value of residential work still required to complete them. Both derive from the same source and describe the same committed stock; one measures it in units, the other in dollars of unfinished work. When the dollar measure runs far ahead of the unit measure, the cost of completing each committed dwelling has risen. That is the signature of structural cost inflation embedded in the pipeline, as distinct from a pipeline that is simply larger.
Since the 2020 trough the value of unfinished residential work has risen roughly twice as far, in standardised terms, as the count of dwellings it represents. The count of dwellings under construction has reached a record of its own, about a quarter above its mean; the value of work needed to finish them, about double its mean, has never been higher. Because work done is itself above mean, this is not a stalled sector storing up unfinished projects — it is a producing sector whose backlog is inflating in cost faster than output can clear it. The mechanism is the cost of completion per dwelling: materials, labour and, since 2022, the seven-star energy standard embedded in every new completion.
For valuation practice the implication is a rising cost floor under new supply. Where the cost of completing new dwellings is structurally elevated, existing stock in genuine-demand markets is supported from below, independently of the interest-rate cycle. For development practice the implication is the reverse: any feasibility built on cost assumptions from before this run of series highs will overstate margin.
On the dashed diagonal the two scores are equal, so value and count are equally far from their means. Below it the count leads; above it the value leads. The path ran below the diagonal through the 2015–19 expansion, returned to it in 2020–23, and has climbed away from it since: the June quarter 2026 sits at +1.38 across and +2.82 up, the furthest above the line on record. Source: APN calculation from Australian Bureau of Statistics, Building Activity, Australia, June quarter 2026; baseline 2011 to 2025.
At the rate dwellings were completed over the past year, the stock under construction at June 2026 represents 16.5 months of work, against 10.1 months in the December quarter 2011 and a peak of 17.0 months in the September quarter 2022. This is a physical measure, in dwellings, and is unaffected by prices. Source: APN calculation from Australian Bureau of Statistics, Building Activity, Australia, June quarter 2026 (dwellings under construction; completions, sum of states and territories).
Building Approvals & Commencements (21510) has moved through four epochs across the baseline, and the current quarter is best understood as the end-point of the fourth. The finance node’s history is not tabulated here; its terminal readings above stand on their own.
| Epoch | Years | 21510 configuration | Marker quarters |
|---|---|---|---|
| I · Post-GFC weakness | 2011–2014 | Composite below mean in every quarter to Q2 2013; approvals and commencements both depressed | −1.39σ (Q4 2011) · −1.31σ (Q1 2012) · −1.18σ (Q1 2013) |
| II · Expansion | 2015–2019 | Composite positive from Q4 2014 through mid-2018; approvals peaked at +2.07σ in Q3 2016; both vectors slipped below mean from Q3 2018 | +1.26σ (Q2 2015) · +1.68σ (Q3 2016) · +1.21σ (Q4 2017) |
| III · HomeBuilder distortion | 2020–2022 | Pandemic dip, then a stimulus surge in which approvals led and commencements followed one quarter later; the spread swung from its series high to a deep negative in consecutive quarters | −0.60σ (Q2 2020) · +1.25σ (Q1 2021, spread +2.47σ) · +0.84σ (Q2 2021, commencements +2.39σ, spread −1.60σ) |
| IV · Contraction and recovery to mean | 2023–2026 | Sharp contraction as the stimulus cohort cleared, then a steady recovery through 2025 to the mean and a little above it by mid-2026; the Q4 2025 spread of −1.30σ shows commencements running down earlier approvals, not stalling | −1.38σ (Q1 2023) · −1.08σ (Q1 2024) · +0.24σ (Q3 2025) · −0.03σ (Q1 2026) · +0.49σ (Q2 2026) |
New Housing Supply Forecasts (21520) traces a different path through the same epochs: deeply negative in Epoch I (near −1.8σ through 2011–12), crossing the mean in 2015 and peaking at +1.03σ in Q2 2018, dipping to −0.31σ in 2020, and then rising almost without interruption — +0.44σ at Q4 2023, +0.91σ at Q1 2025, +1.26σ at Q3 2025, +1.51σ at Q4 2025, +1.65σ at Q1 2026, +1.82σ at Q2 2026. Epoch IV is the first period in the baseline in which the inflow node has returned only to around its mean while the pipeline node has continued to climb.
The HomeBuilder sequence is visible in consecutive quarters: in the March quarter 2021 approvals and a spread of +2.47σ carried the node to +1.2454σ; one quarter later commencements at +2.39σ took over while the spread turned to −1.60σ. Because the spread is approvals less commencements, it offsets a commencements surge by construction. Source: APN calculation from Australian Bureau of Statistics, Building Approvals, Australia, and Building Activity, Australia, June quarter 2026.
Construction Finance & Capacity (21530) is a mixed-sign composite: the mean of the two lending vectors, less a sign-inverted construction-insolvency vector, so that rising sector distress pulls the reading down. At Q4 2025 the two lending vectors stand at +2.9454σ and +2.6090σ. The insolvency vector is substituted at zero for Q3 and Q4 2025 under the node’s documented substitution protocol, because the corporate regulator’s construction-insolvency series is annual and the FY2025–26 figures are not published until December 2026.
The reading reflects lending capacity and lending volume fully and sector financial health not at all. Roughly a fifth of the composite’s intended signal is neutralised rather than incorporated. The correct practitioner reading is that institutional capital for development is abundant and deployed, and that the health of the businesses receiving it is currently unmeasured. A materially different composite could emerge once the missing vector is incorporated, in either direction.
Elevated lending limits alongside elevated lending volumes tend to favour operators with existing bank relationships and balance-sheet scale. Whether that capital is reaching the broader sector base, or concentrating in its upper tier, is precisely the question the absent insolvency vector would answer.
| Structurally advantaged | Structurally disadvantaged | |
|---|---|---|
| Who | Holders of existing compliant stock in genuine-demand markets; well-capitalised builders and developers with bank access | Purchasers of new construction; smaller builders on fixed-price contracts; feasibilities priced on pre-2023 costs |
| Why | A record cost floor under new supply supports existing-stock pricing relative to replacement; lending capacity at +2.9454σ is accessible to those banks already lend to | Every committed dwelling costs more to complete than at any point in the record; output is rising but the backlog’s value is rising faster, stretching delivery timelines and margins |
The inflow node is close to its mean and a little above it. The pipeline node is in active escalation, driven by a work-yet-to-be-done vector that has set a new series high in each of the last eight quarters. The finance node is elevated on the two vectors that can currently be measured and undetermined on the third.
- Signal
Work yet to be done — a ninth consecutive series high, or a plateau? The June quarter 2026 release extended the run to eight; the September quarter release is the next opportunity for it to break. A plateau at or near +2.8σ would indicate cost-to-complete pressure stabilising; a further high would extend the decoupling and widen the gap to the dwellings-under-construction vector.
- Signal
December 2026 construction-insolvency data. The single most consequential pending input in the series. It replaces the neutral substitution in the finance composite and determines whether the +2.1333σ reading survives, moderates or reverses.
- Watch
The approvals–commencements spread against its |1.50σ| threshold. Currently +0.33σ. A positive breach would be the first certified evidence of approved projects stalling before construction; a negative breach would repeat the Q1 2023 pattern of commencements running down earlier approvals, which Q4 2025 approached without crossing.
- Monitor
Work done relative to work yet to be done. Output at +1.54σ is rising. If it continues to rise while the backlog plateaus, the pipeline begins to clear; if the backlog keeps outrunning output, delivery timelines lengthen further.
From the September quarter 2020 to the June quarter 2026 the value of work yet to be done rose to 235, dwellings under construction to 135 and work done to 120. Work yet to be done is in current prices and work done is a volume measure, so part of the distance between the gold and brown lines is price change; the distance between the slate and brown lines is not. Source: APN calculation from Australian Bureau of Statistics, Building Activity, Australia, June quarter 2026, and Construction Work Done, Australia.
