Planning evidence · Sensitivity explorer 01
Could unknown planning outcomes reverse the Q3 result?
The panel's approval rate fell. That direction holds even under the most favourable assignment of recorded unknown outcomes. One regional result is less clear.
01 / The finding
A small missing category can matter to a small change.
The matched panel's approval rate fell from 71.96% in Q3 2025 to 68.38% in Q3 2026. Those rates use approvals divided by approvals plus refusals. They leave out 110 recorded unknown outcomes in 2025 and 131 in 2026.
Assign all 2026 unknowns to approvals and all 2025 unknowns to refusals: the panel still falls by 2.31 percentage points. Reverse those assignments and the fall widens to 4.91 points.
Extending the ICPSR planning study's panel sensitivity test to each custom regional group finds one exception: the East of England's small decline can become a small increase.
02 / Try the assignments
Give the unknowns an outcome.
Choose a geography, then assign some of its recorded unknowns to approvals. The remainder are treated as refusals. Each period can use a different assignment. Withdrawals stay outside the calculation.
Q3 2025 versus Q3 2026
Starting scenario: all recorded unknowns assigned to refusals.
These scenario rates include every recorded unknown. The known-outcome comparison above excludes them. Bar lengths use a fixed 0% to 100% scale.
| Geography | Authorities | Known outcomes | Assignment bounds | Direction across assignments |
|---|---|---|---|---|
| Panel | 253 | -3.58 | -4.91 to -2.31 | Always falls |
| East of England | 19 | -0.53 | -1.64 to +0.77 | Can change direction |
| London & South East | 102 | -2.15 | -3.26 to -0.89 | Always falls |
| Midlands | 55 | -9.34 | -11.97 to -7.49 | Always falls |
| North East & Yorkshire | 20 | -2.56 | -3.89 to -1.76 | Always falls |
| North West | 27 | -9.10 | -10.43 to -7.52 | Always falls |
| South West | 23 | +3.25 | +2.54 to +4.11 | Always rises |
| Wales | 7 | -14.62 | -15.59 to -13.76 | Always falls |
Seven custom regional groups, not the standard UK regional classification. Wales contains only seven authorities. The panel row aggregates the groups; it is not an eighth region.
03 / How to interpret it
Direction is one check. Coverage is another.
Five regional declines remain negative at both extremes. The South West remains positive. The East of England ranges from −1.64 to +0.77 percentage points, around a known-outcome change of −0.53 points.
For the East of England, describing a firm deterioration without this qualification would overstate the evidence. For the panel, the unknown category alone cannot explain away the decline. Neither conclusion resolves selection bias or errors in the underlying capture.
Using this in a development finance review
Keep the planning comparison separate from a scheme's consent position, conditions, cost plan and funding terms. These aggregates do not measure the probability that a particular project receives consent or a loan.
The Construction Capital London & South East Q3 report supplies broader regional context. Here, that group's decline survives every recorded assignment, with a change between −3.26 and −0.89 points. Construction Capital's Development Finance guide and service overview explains the funding process alongside which this evidence may be considered.
The calculation
For a period, let A be approvals, R refusals and U recorded unknown outcomes. Assign x of those unknowns to approvals, with x an integer from 0 to U.
Scenario rate = 100 × (A + x) / (A + R + U)
- Find each period's minimum by setting x = 0 and maximum by setting x = U.
- Lowest change = 2026 minimum minus 2025 maximum.
- Highest change = 2026 maximum minus 2025 minimum.
- A positive lower bound implies a rise under every assignment. A negative upper bound implies a fall under every assignment.
These are deterministic bounds, not confidence intervals. They do not assign probabilities. Different allocations in the two years deliberately test the full possible range. Rates and differences use unrounded counts; displayed results are rounded to two decimals.
04 / The audio companion
Planning approvals fell. Does that change your development finance case?
Listen to Georgina work through three checks before a planning headline enters a funding case: the denominator, recorded unknown outcomes and the gap between regional context and a specific project.
Open the episode and source notes · Read the full transcript
Episode 14 · 5 minutes 49 seconds · 8 October 2026. Written by Matt Lenzie / Construction Capital. This is an audio interpretation of the same evidence, not a separate study or independent corroboration.
05 / Watch the calculation
When a fall becomes a rise.
Follow the East of England counts from approvals and refusals to both extreme assignments of recorded unknown outcomes. The visual walkthrough then compares the regional interval with the wider panel.
Watch the video on YouTube. Use the source counts and Python calculation below to reproduce the result.
2 minutes 57 seconds · 11 October 2026. Written by Matt Lenzie / Construction Capital; narrated by Georgina. A visual adaptation of the same frozen evidence, with captions. No new observations or independent study.
06 / Open the evidence
Counts, code and provenance.
This is an interactive extension of study CC-DF-2026Q3-01, using its frozen aggregate edition of 5 October 2026. It adds regional sensitivity calculations, not new observations or independent corroboration.
Trace the source
The versioned ICPSR record contains the original report, definitions and backing data. It draws on Construction Capital's Landstack-derived aggregates. The same 253-authority panel has monthly tallies in the year-earlier comparison.
The separate planning unit-definition study on Hugging Face tests what happens when a different inclusion rule changes the headline. That is a different sensitivity question using the same publisher's evidence.
Reuse and affiliation
Matt Lenzie / Construction Capital produced this analysis. Construction Capital is a property-finance brokerage. The linked research and website reports share its authorship; repository hosting does not imply peer review or endorsement.
Original text and analytical explanations are CC BY 4.0; original code is MIT licensed. Underlying planning-data rights remain with their holders. No blanket open-data licence is asserted. See the component-specific licence and source manifest.
Suggested citation: Lenzie, Matt (2026). Could unknown planning outcomes reverse the Q3 result? Construction Capital, version 1.0, 8 October 2026. CC-DF-2026Q3-C01. Public research tool. Cite ICPSR317061.V1 for the underlying study.