At a glance: 934,000 non-domestic properties, 598 million m² of floorspace, 75.9% rated below EPC Band B. The proposed 2031 MEES EPC B standard targets 67,000 buildings holding 36% of all certificated floorspace. Offices are the worst asset class. Gas heating concentrates in the lowest-rated stock. 389,000 viable roofs have no solar PV.
Executive summary
This report characterises the non-domestic building stock of England and Wales through the lens of energy performance, regulation and decarbonisation readiness. It draws on 1.30 million Energy Performance Certificates lodged between January 2012 and June 2026, deduplicated to 934,000 unique properties, and 15.5 million associated retrofit recommendation records, read alongside the DESNZ Non-Domestic Building Stock study.
Key findings:
- New certificates improved from 20% F–G in 2012 to 1.3% in 2026. Part of the gain is methodology, not retrofit, and 17.8% of latest certificates have already expired.
- Offices are the worst asset class: 81% below B, 25% in E–G, largest floorspace base.
- Big means better rated. 80% of sub-100 m² units are below B versus 47% above 10,000 m². Properties over 2,500 m² are 4.6% of the stock and half the floorspace.
- Heating is 57% electric, 36% gas. Gas concentrates in the worst-rated sectors.
- 389,000 properties (41.7% of the stock) have a viable roof and no solar PV. Installed PV and batteries are invisible to EPC data.
- E–G share runs from 9.5% in Newham to 23.4% in Leicester. The worst stock sits in the weakest rental markets, compounding the brown discount.
- All exposure figures are upper bounds. MEES covers rented stock only and the register does not record tenure. Title matching tightens the bound: 54% of matched sub-B floorspace sits on leasehold titles, and 88% is held by private companies rather than public bodies.
01 The stock in one view
The certificated non-domestic stock covers 934,000 unique properties and 598 million m² of floorspace. That floorspace figure sits almost exactly on the Valuation Office Agency (VOA) total of 594 million m² reported by DESNZ, although the two universes are not identical.
The stock is dominated by small premises. DESNZ finds 94% of premises are under 1,000 m², and this register shows the same shape: 88% of certificated properties are under 1,000 m², yet properties over 1,000 m² hold 68% of all floorspace.
| Metric | Figure |
|---|---|
| Unique certificated properties | 934,000 |
| Total certificated floorspace | 598 million m² |
| Properties rated below Band B | 75.9% |
| Properties rated E, F or G | 16.9% |
[CHART:fig1]
The distribution is mid-heavy. C and D together hold 59% of properties. The tails matter for different reasons. The 38,000 F and G properties are already unlettable without an exemption. The 551,000 C and D properties are the underwriting exposure of the next decade: most are too small to be caught by the proposed 2031 EPC B standard, but investors and lenders benchmarking portfolios to B will price them as if they were.
02 The regulatory position
Three regimes do most of the work in this market.
MEES (Minimum Energy Efficiency Standards)
Since 1 April 2023 it has been unlawful to continue letting a non-domestic property rated F or G in England and Wales without a valid exemption. The government's 2021 consultation proposed raising the floor to C by 2027 and B by 2030. The interim government response has since reshaped that trajectory:
- From 2031, private rented buildings over 1,000 m² are proposed to reach EPC B where cost effective
- Buildings of 1,000 m² or under are intended to stay on the current E floor
- The 2027 interim C milestone will not be taken forward
- The 7-year payback test and exemptions remain
[CHART:fig2]
The headline exposure: the proposed 2031 EPC B standard concentrates the obligation on 67,000 sub-B properties over 1,000 m², which hold 216M m² — 36% of all certificated floorspace. A further 642,000 smaller sub-B properties face no new requirement beyond the E floor, but 709,000 properties below B remains the relevant number for investors and lenders who underwrite to B portfolio-wide regardless of the legal minimum.
One scope limit applies to every exposure figure in this report. MEES covers the private rented sector only. Owner-occupied buildings are out of scope, and the EPC register does not record tenure, so all counts here include owner-occupied stock the regulations do not touch.
EPC reform
EPCs are valid for ten years. 17.8% of the latest certificates in this snapshot are already more than ten years old, meaning a large share of the "known" stock is flying on expired data. Government consulted in 2025 on reforming EPC metrics and requiring valid certificates throughout a lease term rather than only at transaction points. Either change would force a re-certification wave and reveal band slippage that current data hides.
Ownership and tenure: what title data adds
Linking the register to HM Land Registry's Commercial and Corporate Ownership Data (CCOD) partially closes the tenure gap. Address matching against 4.4 million corporately held titles resolves 52% of the certificated stock, rising to 60% for properties over 1,000 m².
Two findings:
Tenure. 54% of matched sub-B floorspace sits on leasehold titles, rising to 59% for the sub-B cohort over 1,000 m² that faces the proposed 2031 EPC B standard. A registered leasehold title is direct evidence of a landlord and tenant structure — this is a floor on the rented share, not an estimate.
Who holds the problem. 88% of matched sub-B floorspace is held by private limited companies and LLPs, with public sector bodies at under 4%. The compliance burden of the proposed standard lands almost entirely on private corporate balance sheets.
[CHART:fig3]
03 The stock is improving, but from the flow, not the base
Certificates lodged in 2012 were 20% F or G. Certificates lodged in 2026 are 1.3% F or G, and 43% reach A or B. The median asset rating of new lodgements has improved from 88 to 55.
Two forces drive this: genuine improvement in the buildings being assessed, and a 2022 methodology update (SBEM revisions reflecting a decarbonising electricity grid) that re-rated electrically heated buildings more favourably. The step change visible in 2022–2023 is partly reclassification, not just retrofit.
Reading this correctly: these charts describe the flow of newly assessed properties, not the standing stock. Properties only re-certify on sale, letting or construction. The standing stock includes hundreds of thousands of properties last assessed under older, harsher-on-electricity conventions.
[CHART:fig4]
[CHART:fig5]
04 Asset class: offices carry the highest risk concentration
Offices are the worst-performing sector by band distribution. 81% sit below Band B; 25% are rated E, F or G. No other major asset class combines that depth of sub-standard stock with that volume of floorspace.
Retail is the largest sector by count (33% of properties) but distributes more evenly across bands. Industrial and storage assets rate better on average: large modern sheds built to Part L 2021 standards regularly reach A or B. Hospitality (hotels, restaurants, pubs) is another concentration of poor-rated stock, where operational complexity and listed building constraints slow retrofit.
The practical read for investors: sector diversification does not insulate a portfolio from EPC risk. Offices overweight any fund relative to their market share, and they are the sector where the 2031 compliance problem is largest.
[CHART:fig6]
[CHART:fig7]
[CHART:fig8]
05 Size and rating: the small-building paradox
Building size and EPC rating are positively correlated across the non-domestic stock. 80% of sub-100 m² units sit below Band B. For properties above 10,000 m², that figure falls to 47%.
This creates what might be called the small-building paradox for MEES. The buildings with the worst ratings are mostly too small to be caught by the proposed 2031 EPC B standard (which applies above 1,000 m²). The buildings the standard does catch — larger, institutionally held — are, on average, better rated already. Compliance effort concentrates on assets that are nearer to B anyway; the tail of very poor-rated small stock falls outside the mandatory programme entirely.
Properties over 2,500 m² are just 4.6% of the certificated stock by count but hold approximately half of all certificated floorspace. This is the cohort where the regulatory and capital markets pressure is sharpest, and where the operational data infrastructure to manage it is also most developed.
[CHART:fig9]
06 Heating fuel: gas concentrates in the bottom half
The heating fuel split across the non-domestic stock is:
| Fuel | Share |
|---|---|
| Electric | 57% |
| Gas | 36% |
| Oil | 3% |
| Other / community | 4% |
Gas heating concentrates in the worst-rated sectors and building types. Among F and G-rated properties, gas is the dominant fuel in a disproportionate share of cases. This pattern matters for Scope 1 carbon accounting: gas-heated assets carry direct fuel combustion exposure that electric assets do not, and as MEES tightens, the retrofit path for gas-heated buildings involves either heat pump replacement or deep fabric improvement — both more capital-intensive than the incremental improvements available to electrically heated stock that SBEM already rates favourably.
The electrification of heating is already visible in new-build and recent-refurbishment stock. Among certificates lodged after 2022, electric heating dominates at rates well above the stock average. The standing stock of gas-heated D, E and F properties represents the durable retrofit challenge.
[CHART:fig10]
[CHART:fig11]
07 Solar: 389,000 viable roofs, no PV
389,000 non-domestic properties — 41.7% of the certificated stock — have a viable roof geometry and no recorded solar PV installation. This is the addressable rooftop solar opportunity visible from EPC data.
Two caveats apply. First, EPC assessments do not record installed PV or battery storage: a property with a 500 kWp rooftop array looks identical in the register to one with a bare roof, unless the assessor explicitly noted the system. The true installed base is higher than the register shows. Second, "viable roof" is a geometric assessment from the DESNZ building stock model, not a structural or planning survey. Not every viable roof is a deployable roof.
The EPBD solar mandate provides context: non-residential buildings over 500 m² must include solar where technically and economically feasible by 2028 under the directive as transposed. Of the 67,000 sub-B properties over 1,000 m² facing the 2031 MEES standard, a significant share will face concurrent solar obligations.
[CHART:fig12]
[CHART:fig13]
[CHART:fig14]
08 Geography: the worst stock sits in the weakest markets
The E–G share of non-domestic stock varies materially by local authority area: from 9.5% in Newham to 23.4% in Leicester. The geographic pattern has a consistent economic character. The worst-rated stock concentrates in secondary and tertiary markets with lower average rents, weaker tenant demand, and thinner investment liquidity.
This matters for the brown discount. The valuation penalty on poorly rated stock is already documented in transaction data across UK commercial markets. In locations where the discount compounds with structural rental weakness and limited comparable evidence, the commercial case for pre-emptive retrofit is harder to build. The capital is needed most where the economics of providing it are least attractive.
For investors with geographic concentrations in secondary markets, the EPC tail is a double exposure: regulatory cost and market liquidity risk moving in the same direction.
[CHART:fig15]
09 What EPC data cannot tell you
This analysis rests on the EPC register, and the register has four structural gaps that matter for anyone using it to price risk or allocate capital.
Ratings are modelled, not metered. The assessed rating reflects the design intent of the building's fabric and fixed services. It does not measure how the building actually performs. The gap between rated and operational performance — well-documented in domestic stock and increasingly visible in commercial data — means a D-rated building can consume less energy than a C-rated one in practice, or significantly more.
Tenure is not recorded. Every MEES exposure figure in this report is an upper bound. The register cannot distinguish rented from owner-occupied stock, and MEES applies only to the rented population.
Renewable generation is invisible. Installed PV, battery storage, and heat pump systems are not reliably captured in EPC data. The register understates the proportion of the stock already generating or storing energy on site.
17.8% of certificates are expired. Buildings are only re-assessed on sale, letting, or new construction. A property last certified in 2013 under different SBEM conventions may be better or worse rated under current methodology.
[CHART:fig16]
10 Implications for investors, asset managers and lenders
For investors and lenders underwriting to EPC B portfolio-wide: 709,000 sub-B properties is the relevant number regardless of the legal minimum. The proposed 2031 standard creates a stepped obligation; capital markets are likely to price to the higher bar before the legislation requires it.
For asset managers sequencing retrofit capital: size is the primary sorting variable. The 67,000 properties over 1,000 m² facing the 2031 standard hold 36% of all floorspace and should anchor sequencing decisions. Within that cohort, gas-heated stock carries the steepest improvement cost and the longest lead time.
For energy service providers: the 389,000 viable-roof, no-PV properties are a directly addressable pipeline. The solar mandate under the EPBD adds a concurrent compliance driver for larger assets in that group.
For policymakers: the tenure gap remains the central measurement problem. Without reliable tenure data in the register, MEES exposure estimates carry irreducible uncertainty. EPC reform proposals that require valid certificates throughout a lease term would close part of this gap.
Data sources and methodology
- EPC register for non-domestic properties, England and Wales: 1.30 million certificates lodged January 2012 to June 2026, deduplicated to 934,000 unique properties by UPRN and address matching.
- DESNZ Non-Domestic Building Stock Study (2023 edition).
- HM Land Registry Commercial and Corporate Ownership Data (CCOD), matched by address and postcode to 52% of the certificated stock.
- Valuation Office Agency rating list totals.
- EPBD (EU/2024/1275) solar readiness provisions.
- MEES interim government response, 2024.
All figures describe the certificated stock only. Properties without a valid EPC are not included. Ratings are modelled assessments, not metered performance. Tenure is not recorded in the EPC register; all MEES exposure figures are upper bounds.