Development Finance: Q3 2026

Four questions before using a market headline

Q3 2026 research briefing | Matt Lenzie, Construction Capital | Published 5 October 2026

A planning approval rate, a count of property sales and a register of company charges describe different stages of the development process. Reading them as one measure of market confidence can produce a misleading appraisal.

This briefing brings four Construction Capital studies together around a practical question: what does each signal establish, and what evidence is still needed for a development finance decision? It is an original synthesis of the published research, with links to the reports, backing tables and reproduction code.

The planning results cover a matched panel of 253 authorities in England and Wales. Sales figures come from a fixed England and Wales extract ending on 28 August. Charge activity uses defined lender cohorts and remains provisional. These differences matter throughout the comparisons below.

1. Is the planning change local, or a change in the mix?

The panel approval rate fell from 71.96% in Q3 2025 to 68.38% in Q3 2026. Of the 3.58 percentage-point decline, 3.29 points are attributed to changes within regions and 0.29 points to changes in the regional mix of decided applications.

The decomposition answers an accounting question. It does not establish why decisions changed, or predict whether a particular scheme will obtain consent. For appraisal, the useful follow-up is to inspect the relevant authority, proposal and planning stage rather than apply the panel rate as a scheme-level probability.

Read the ICPSR planning approval study and backing data. Approval means approved divided by approved plus refused; unknown outcomes are excluded from that denominator and withdrawals are separate.

Figure 1. Additive contributions to the panel decline. Values are rounded; the underlying decomposition totals 3.582 percentage points. Source: ICPSR study 317061, version 1.0.

2. Does the pipeline trend survive a change of definition?

In the 19-authority East of England panel, the year-on-year change in unit mentions moves from −26.6% to +93.4% when approved records whose proposal text matches “reserved matters” are excluded. The full 253-authority panel remains negative under both definitions: −25.4% and −27.1% respectively.

That reversal is a reason to examine the definition before using the headline. It does not mean supply suddenly doubled. Mentions can relate to different application stages, and the filter does not establish that every removed mention is a duplicate. Neither series is a count of deduplicated new homes or a forecast of completions.

Explore the planning-definition study, tables and code on Hugging Face. A useful project question is whether the recorded units relate to initial consent, a later approval stage or a separate proposal.

3. Are the sales records mature enough for the comparison?

The frozen Price Paid extract contains 61,717 Q3 2026 records, with a latest transaction date of 28 August 2026. September is absent, and registration lag also affects earlier transactions. Its ratio to recorded Q3 2025 sales is 29.27%, but that is neither measured completeness nor evidence of a corresponding fall in market activity.

For development appraisal, the distinction is between transactions that happened and transactions available in this extract. Recent counts need a dated source and a coverage explanation. A changing property mix can also affect a median without showing a like-for-like change in value. The study does not produce a house-price index, final Q3 volume or a valuation for an individual scheme.

Open the Kaggle sales study and four analytical tables. Fourteen Q3 transactions lack regional mapping; recent blank new-build shares are deliberately suppressed, not zeros.

4. Does lender breadth describe the distribution of activity?

The curated specialist cohort records 13 active lenders in both Q3 2025 and provisional Q3 2026. Over the same comparison, the top-five share of registered charges rises from 58% to 64%. The latest share is nevertheless below Q1 2026 (68%) and Q2 (66%), so the conclusion depends on the comparison period.

Q3 records 207 charges across 140 borrowers. These are registrations, not lending amounts or necessarily new loans. A stable count of active lenders can coexist with a different distribution of activity, but it does not establish that funding is easier or harder to obtain. Pricing, leverage, available facilities and eligibility need separate evidence.

Read the Harvard Dataverse charge-activity study and replication files. For the wider source discussion, see the Construction Capital Q3 Development Lending Monitor.

Applying the evidence to Development Finance

The four studies support different questions: what happened to recorded planning decisions; how sensitive the pipeline is to its definition; how much recent sales evidence is observable; and how charge activity is distributed within a defined cohort. They cannot be joined at scheme level to trace a development from consent through funding to sale.

For a live proposal, keep the site’s planning position, comparable sales, build costs and funding terms distinct in the appraisal. Construction Capital’s Development Finance guide and service overview explains staged construction funding and the application process. The London and South East Q3 market report and corrected South West Q3 report provide additional regional context.

Methods, backing data and reuse

All four linked releases include a five-page report, selected aggregate inputs, analytical tables, a dictionary, source provenance, validation results and Python reproduction code. The code rebuilds the tables from the supplied aggregates; it does not reproduce the upstream commercial extraction. ICPSR file downloads require sign-in.

The planning panel has seven custom regional groups and only seven Welsh authorities. It is not a complete national census or a probability sample. Panel totals overlap their regional components and must not be added together. The two planning studies share this source panel.

Sales use the fixed extract described above. Charge metadata preserves distinct source dates: snapshotDate 1 October, generatedAt 3 October, chargesMax 2 October and satisfiedMax 3 October 2026. Those fields should not be presented as one settled cutoff. Charge cohorts overlap and must not be summed.

Publisher and interests. Construction Capital is a property-finance brokerage. This site and the four studies share its authorship; they are not four independent validations. Repository hosting does not imply university endorsement or peer review. This descriptive research does not provide scheme-specific financial advice or funding terms.

Attribution. Lenzie, Matt (2026), Development Finance: Q3 2026 Evidence, Construction Capital, 5 October 2026, companion briefing CC-DF-2026Q3-S01. Original commentary and charts: CC BY 4.0. Consult each linked deposit’s licence for its data and code; no licence to underlying proprietary databases is granted.