IN ONE SENTENCE — Real estate finance lends against buildings whose rents repay the debt, so the banker underwrites a property and its leases before a company and its balance sheet.
Where it sits. At Meridian Bank, Real Estate Finance (REF) is a product line of Corporate Banking, next to project finance and asset finance in what many banks call “structured” or “specialised” lending. It lends to professional property investors: listed property companies (in France, SIIC), unlisted funds (OPCI, SCPI managers, pan-European core funds), insurers’ property arms, private equity real estate funds, developers and family offices. The borrower is almost always a single-purpose vehicle (SPV) that owns one building or one portfolio and nothing else. Retail mortgages belong to the retail bank; they appear here only through the covered bond pool, which Treasury fills with both.
How it earns money. Three lines. First, the margin on drawn loans: in the euro area in 2026, roughly 130 to 200 bp for a well-let prime office at 50 to 55% loan-to-value, 220 to 350 bp for a building under redevelopment. Second, fees: an arrangement fee of 80 to 150 bp of the facility paid at signing, a commitment fee on undrawn capital expenditure tranches, an agency fee of a few tens of thousands of euros a year, waiver and extension fees. Third, ancillary business: the interest-rate hedge that the facility agreement makes compulsory (swap or cap sold by the markets desk, see D02), accounts and cash management for the rents, sometimes the advisory mandate when the building is sold. These ranges are market orders of magnitude, not quotations; the case in section 5 fixes its own numbers. A front-office banker in REF carries a portfolio of perhaps EUR 1bn to 2bn of commitments, so a team of ten can represent a book as large as that of a mid-sized corporate coverage group, with far fewer clients.
The team and its neighbours.
| Function | What it does | Document it produces |
|---|---|---|
| Origination (REF bankers) | Owns the sponsor relationship, wins the mandate | Pitch, indicative term sheet |
| Structuring and execution | Models cash flows, negotiates documentation | Credit application, financial model |
| Credit risk, real estate team | Independent opinion, rating, slotting or PD | Risk opinion, rating sheet |
| Collateral valuation unit | Instructs and challenges external valuers | Valuation review memo |
| Loan agency (middle office) | Drawdowns, covenant tests, waivers | Compliance certificates log |
Around them: the syndication desk (sell-down to other banks, insurers and debt funds), the markets desk (hedging), legal, the technical adviser hired for development loans (a firm of building engineers that certifies each drawdown), the environmental and social risk team, and Treasury, which decides whether the loan can be refinanced through the covered bond issuer.
INTUITION — A corporate loan is a bet on a management team. A real estate loan is a bet on an address, a lease schedule and a valuer’s yield. If the tenant leaves, the SPV has no other business to fall back on; if the borrower fails, the bank owns a building. Everything on the desk follows from those two facts.
A realistic week. Monday: pipeline meeting; three live requests, of which one is a refinancing where the valuer has just lowered the value by 8% and the loan-to-value covenant is close. Tuesday: site visit of a tower to be refurbished, with the sponsor’s asset manager and the technical adviser; the banker looks at floor plates, lifts and plant rooms, and asks where the electrical substations are located relative to the flood level. Wednesday: the structurer updates the model with the new letting assumptions and prepares the credit application; the valuation unit sends its comments on the draft valuation report (the yield retained for the refurbished building looks 25 bp too tight against recent transactions). Thursday: pre-screening with credit risk; discussion of the slotting category and of whether the works phase must be treated as a development exposure. Friday: the agent’s team processes a drawdown request on another project, checks the technical adviser’s certificate and the cost-to-complete test, and releases funds. In between: quarterly covenant certificates, a tenant’s request to assign its lease, a hedge to roll.
WHY IT MATTERS ON THE DESK — In real estate finance the sustainable layer is not a label added at the end. Energy performance changes the rent a building can command, the yield at which it is valued and the date at which it may become unlettable: three inputs of the credit analysis itself. This is the business line where “green” is closest to plain credit risk.
An investment loan finances a building that already produces rent. The bank lends a share of the value, receives interest quarterly out of the rents, takes little or no amortisation, and is repaid at maturity (five to seven years) by a sale or a refinancing. The final repayment is called the balloon. The risk is therefore double: can the rents pay the interest every quarter (income risk), and will the building be worth enough at maturity for someone else to refinance the balloon (value and refinancing risk)?
A development loan finances a building that does not yet exist in its final form: a new construction or a heavy refurbishment. There is no rent during the works. Interest is either paid by the sponsor from equity or capitalised, that is, added to the loan and financed by an interest reserve included in the budget. The loan is drawn progressively, each drawdown being certified by the technical adviser against the works actually completed. The risks are of a different nature: cost overrun, delay, contractor default, and letting risk at delivery. Margins are higher, loan-to-cost ratios lower, and banks ask for completion guarantees from the sponsor.
Many real transactions, including the case of this module, are hybrids: an acquisition or refinancing tranche on the existing building, a capital expenditure (capex) tranche for the works, and conversion into an investment loan once the building is delivered and let (“stabilised”).
| Feature | Investment loan | Development loan |
|---|---|---|
| Source of repayment | Rents, then sale or refinancing | Sale or refinancing after letting |
| Key ratio | LTV, ICR, debt yield | LTC, LTGDV, cost to complete |
| Drawdown | One shot at closing | Monthly, against certificates |
| Interest | Paid from rents | Capitalised or paid from equity |
| Typical margin (2026) | 130 to 200 bp | 220 to 350 bp |
The debt is sized on a value, so the banker must understand where the value comes from. Valuers of income-producing property use two families of methods: income capitalisation and discounted cash flow. Both rest on the same idea.
IN ONE SENTENCE — A building is worth its net rent divided by the yield that investors require, exactly as a perpetual bond is worth its coupon divided by its yield.
FORMULA
is the market value (in practice net of transfer taxes); the net operating income, that is the annual rent actually received less the costs the landlord cannot recharge to tenants (vacancy, unrecovered service charges, management); the capitalisation rate, called “cap rate” or “yield”, observed on recent transactions of comparable buildings.
The yield is the price of the market: it moves with long-term interest rates, with the risk premium investors ask for the asset class and with the quality of the specific building. Because is the ratio of two numbers, a small change in the yield moves the value a lot. From 5.00% to 5.50%, value falls by 9.1% () with the rent unchanged. Between mid-2022 and 2024 prime office yields in the euro area widened by more than 100 bp and values fell by 20 to 30% in many markets with no tenant leaving: that is the mechanism.
PITFALL — The yield is not a return and not a discount rate. A low yield means an expensive building, usually because rents are expected to grow or because the lease is long and the tenant strong. Comparing a 4.5% yield with a 5.1% cost of debt and concluding that “leverage is negative” is correct arithmetic; concluding that the building is a bad investment ignores rental growth and reversion.
FORMULA
is the outstanding debt; the latest market value; the annual interest (all-in rate, after hedging, times ); the scheduled annual amortisation; the debt yield. For development loans, two more: (loan-to-cost) and (loan-to-gross-development-value).
Term by term. The loan-to-value measures the cushion of equity that absorbs a fall in value before the lender loses: at 60%, the value can fall by 40%. The interest cover ratio measures the cushion of rent before interest goes unpaid: at 2.0x, half the rent can disappear. The debt service cover ratio adds amortisation to interest; it matters for amortising loans and long leases. The debt yield is the lender’s own ratio: it ignores both the valuer’s yield and the interest rate, and asks one question: if I had to take the building, what income would I earn on my loan? A debt yield of 9% means that the lender would be made whole at any market yield below 9%. It became standard after 2008 precisely because LTV is only as good as the valuation and ICR only as good as the interest rate of the day.
INTUITION — LTV is how far the water can rise before it reaches your floor; ICR is how long you can hold your breath; debt yield is whether you could live in the house yourself if the owner handed you the keys.
Take a building valued at EUR 100m with a net operating income of EUR 5.5m, financed by a EUR 60m loan at an all-in rate of 4.50% amortising 1% a year. LTV is 60.0%, interest is EUR 2.7m, ICR is 2.04x, DSCR is 1.67x and debt yield is 9.17%. If the rate rises to 6.50% on an unhedged loan, ICR falls to 1.41x although nothing has happened to the building: this is why facility agreements impose hedging.
INTERACTIVE LAB — The four ratios of a real estate loan
Compute LTV, ICR, DSCR and debt yield for an investment loan and see how the interest cover reacts to the all-in rate.
| Parameter you can move | Range | Default |
|---|---|---|
| Market value of the building | 20 to 500 EUR m | 100 EUR m |
| Net operating income | 1 to 30 EUR m | 5.5 EUR m |
| Loan outstanding | 10 to 400 EUR m | 60 EUR m |
| All-in interest rate (after hedging) | 1 to 9 % | 4.5 % |
| Annual amortisation | 0 to 5 % of loan | 1 % of loan |
| Result at default settings | Value |
|---|---|
| Loan-to-value | 60.0 % |
| Annual interest | 2.70 EUR m |
| Interest cover ratio | 2.04 x |
| Debt service cover ratio | 1.67 x |
| Debt yield | 9.17 % |
| Implied capitalisation rate | 5.50 % |
| Fall in value that wipes out the equity | 40.0 % |
Try it. Raise the loan to EUR 70m: LTV moves by 10 points but the debt yield falls below 8%, the level under which many lenders stop.
Pricing. The margin is quoted over three-month EURIBOR and rises with LTV, with letting risk and with the quality of the location. Lenders use internal grids (LTV bucket by asset quality) checked against what competitors, including debt funds and insurers, offer for the same deal. The arrangement fee is paid at signing; in a club deal each lender receives a participation fee out of it.
Hedging. Rents are fixed (indexed on a cost or rent index); EURIBOR is not. The facility agreement requires that 70 to 100% of the loan be hedged for its life, by a swap (the borrower pays a fixed rate and receives EURIBOR) or a cap. The hedge counterparty ranks alongside the lenders on the security. The all-in rate used in the ICR is the swap rate plus the margin.
Security package. A first-ranking mortgage on the building (in France a hypothèque or a lender’s privilege), a pledge over the shares of the SPV, an assignment of rents by way of security (in France a cession Dailly), pledges over bank accounts, assignment of insurance proceeds and of the hedge. For a development: assignment of the construction contracts and a sponsor completion guarantee.
Covenants. Tested quarterly or semi-annually and certified by the borrower: maximum LTV (on an annual external valuation commissioned by the agent), minimum ICR (historical and projected over twelve months), sometimes minimum debt yield. Two levels are common: a cash trap level, at which surplus cash is retained in a blocked account instead of being paid to the sponsor, and a default level, a few points beyond. A breach can be cured by the sponsor through a partial prepayment or a cash deposit.
Documentation. The LMA publishes a facility agreement for real estate finance (investment transactions and, separately, development transactions); the text is reserved to members and this module relies on public summaries. Its logic follows the life of a building: representations on title and leases, undertakings on insurance, valuation, letting and disposal, account waterfall, and events of default.
WHO DOES WHAT — The structurer builds the cash-flow model and the covenant headroom table. Credit risk challenges the letting and exit assumptions. The valuation unit, independent from the front office, selects the valuer from a panel and reviews the report: under the EBA Guidelines on loan origination and monitoring, the valuer must be independent from the lending decision and rotated. The agent tests covenants and holds the accounts.
Losses in commercial real estate lending are rare and lumpy. They concentrate at the end of cycles and at loan maturity, when a building worth less than expected must be refinanced in a market with fewer lenders. Four typical paths: (1) the main tenant leaves and the building needs capex nobody budgeted; (2) yields widen and the balloon exceeds what a new lender will advance; (3) a development is delivered late or over budget; (4) the building becomes obsolete, technically or by regulation. The fourth path is where energy performance enters, and it combines with the other three: an obsolete building loses its tenant, needs capex and suffers a wider yield at the same time.
Recovery is slow but usually high: the lender enforces the share pledge or the mortgage, or more often agrees with the sponsor on an orderly sale. The loss given default depends on the fall in value since origination, on the time to sell and on the costs. This is why regulation gives real estate collateral a specific treatment (section 6) and why the valuation is the single most sensitive document in the file.
Buildings account for about 40% of final energy consumption and about 36% of energy-related greenhouse gas emissions in the European Union, according to the European Commission. Policy therefore targets them directly, and it does so through rules attached to the building itself: the right to let it, to sell it without a discount, to keep it in operation. For a lender whose only recourse is the building, a rule that makes the collateral unlettable at a known future date is not an “ESG factor”; it is a maturity-dated credit event.
Three channels carry energy performance into the numbers of section 2. Income: large tenants have their own climate commitments and reporting duties and increasingly refuse inefficient space, so void periods lengthen and rents fall for poor buildings. Yield: investors capitalise the risk of obsolescence and the cost of the works they will have to pay; the difference in value between otherwise comparable efficient and inefficient buildings is called green premium when measured from the average and brown discount when measured from the top. Capex: to remain lettable the owner must spend, and that spending ranks, economically, ahead of the lender’s balloon.
PITFALL — Green premium and brown discount are the same price difference seen from two reference points, and neither is a free gift of the market. A large part of the measured premium pays for real differences: newer plant, lower service charges, better location and tenants. Section 8 comes back to what the evidence does and does not show.
The energy performance certificate (EPC) rates a building on a letter scale from its calculated or measured primary energy use per square metre. In France it is the diagnostic de performance énergétique (DPE), with a double threshold on energy and on greenhouse gas emissions. EPCs are national: a class B in one Member State is not a class B in another, which is a recurrent problem for pan-European lenders and for the Taxonomy.
The recast Energy Performance of Buildings Directive, Directive (EU) 2024/1275, entered into force on 28 May 2024. What matters to the desk:
The transposition deadline was 29 May 2026. In mid-July 2026 the Commission opened infringement procedures by sending letters of formal notice to all 27 Member States for failing to notify full transposition. In November 2026 the directive’s dates therefore stand, but the national thresholds that tell an owner whether a given building is in the “worst 16%” are, in most countries, not yet in force. A banker must underwrite the 2030 and 2033 dates without knowing the exact threshold: the prudent reading is that an EPC class F or G office is at risk.
France already has its own instrument for existing buildings, the dispositif Éco Énergie Tertiaire (known as the décret tertiaire): tertiary buildings of 1,000 m² or more must cut final energy consumption by 40% by 2030, 50% by 2040 and 60% by 2050 against a reference year no earlier than 2010, or alternatively reach an absolute consumption threshold set by type of activity. Consumption is declared each year on the OPERAT platform run by ADEME; the declaration of 2025 consumption was due by 30 September 2026. Sanctions are modest (publication of non-compliant entities, limited fines), so the constraint works mainly through tenants and buyers who read the certificate.
IN ONE SENTENCE — The stranding year is the first year in which a building emits more per square metre than a decarbonisation pathway allows for its type and country.
The Carbon Risk Real Estate Monitor (CRREM) publishes, by country and property type, pathways of carbon intensity (kgCO2e per m² per year) and of energy intensity (kWh per m² per year) compatible with a 1.5°C carbon budget, derived by downscaling a global budget to the building sector and then to each market. The pathways are free to download and are aligned with the buildings method of the Science Based Targets initiative. Investors, valuers and banks use them as a common yardstick: plot the building’s intensity against the pathway; where the two lines cross is the stranding year; the area between them after that year is the excess emissions, which can be priced.
INTUITION — The pathway is a speed limit that falls every year. A building that does nothing keeps driving at the same speed; the stranding year is when the limit drops below it. A retrofit slows the car, which buys years before the limit catches up again.
Two properties of the tool matter in practice. First, a building’s carbon line is not flat even if nothing is done, because the electricity grid decarbonises: CRREM applies projected grid emission factors. In a country with low-carbon electricity such as France, an all-electric building strands late on the carbon pathway and the binding line is the energy pathway. Second, “stranded” in CRREM’s sense means misaligned with a 1.5°C budget; it does not mean illegal or unlettable. The legal constraint comes from MEPS and national rules, the economic one from tenants and buyers. Good credit files keep the three notions apart.
The Climate Delegated Act, Delegated Regulation (EU) 2021/2139, defines when a building activity makes a substantial contribution to climate change mitigation. Three activities matter to a lender (criteria as in force in November 2026, simplified):
| Activity | Substantial contribution test |
|---|---|
| 7.1 Construction of new buildings | Primary energy demand at least 10% below the national nearly zero-energy building threshold; airtightness and life-cycle carbon tests above 5,000 m² |
| 7.2 Renovation of existing buildings | Complies with the national requirements for major renovation, or reduces primary energy demand by at least 30% |
| 7.7 Acquisition and ownership | Built before 31 December 2020: EPC class A, or within the top 15% of the national or regional stock by primary energy demand. Built after: meets the 7.1 energy criterion |
To be aligned, the activity must also do no significant harm to the other objectives (notably a climate risk and vulnerability assessment for adaptation) and respect minimum safeguards. Two consequences are counter-intuitive. A deep retrofit that takes a building from F to B is aligned under 7.2 as a renovation (the works), but the finished building is not automatically aligned under 7.7 as an asset, since class B is neither class A nor proof of belonging to the top 15%. And because EPC scales are national, the top 15% test relies on studies of the building stock commissioned by the industry, country by country.
The criteria are being rewritten. On 17 March 2026 the Commission published draft amendments to the Climate and Environmental Delegated Acts, with a consultation that closed on 14 April 2026. According to public summaries of the draft, new buildings would have to meet the zero-emission building standard of the EPBD, and an additional route would be opened for existing buildings whose primary energy demand has been reduced by at least 60% over the previous ten years. Adoption was announced for the third quarter of 2026, with application intended from 1 January 2027. This module could not confirm from a primary source that the final text had been published in the Official Journal by November 2026: check before using the new criteria in a term sheet.
Green loans. A real estate loan is labelled green under the Green Loan Principles (LMA, APLMA and LSTA; latest edition March 2025) when its proceeds finance or refinance eligible green projects, with four components: use of proceeds, process for project evaluation and selection, management of proceeds, and reporting. Real estate is the natural home of the use-of-proceeds format because the asset is identified and its performance is certified by third parties (EPC, certifications such as HQE, BREEAM or LEED). Eligibility criteria are set by each bank’s green financing framework; most refer to the Taxonomy thresholds above, to an EPC level, or to a certification level. Sustainability-linked structures (see B03) also exist in real estate, with key performance indicators on energy intensity or on the share of certified assets; they suit portfolio facilities to property companies more than single-asset loans.
Green mortgages and renovation loans. On the retail side, banks offer mortgages with a rate discount for efficient homes and renovation loans; in France the regulated zero-rate eco-loan (éco-PTZ) is distributed by banks. The Energy Efficient Mortgage Label, run by the European Mortgage Federation and European Covered Bond Council, is a market quality label for such products. In December 2023 the EBA delivered a report to the Commission on green loans and mortgages, proposing a voluntary EU green loan label based on the Taxonomy and the integration of a “green loan” concept in the Mortgage Credit Directive; no such label exists in law in November 2026.
Covered bonds. Mortgage loans, residential and commercial, are refinanced through covered bonds issued by a specialised entity of the bank (in France a société de crédit foncier or société de financement de l’habitat). Under Article 129 of the CRR, commercial mortgage loans are eligible for the preferential treatment of covered bonds up to 60% of the property value. A green covered bond is a covered bond whose proceeds are allocated to a sub-portfolio of loans on efficient buildings, under the ICMA Green Bond Principles or, since 21 December 2024, under the European Green Bond Standard (Regulation (EU) 2023/2631), which requires Taxonomy alignment of the allocated assets. For the REF desk this is the funding link: a loan that qualifies for the green pool can be refinanced by Treasury at a slightly lower cost (the greenium of covered bonds, a few basis points at most, of the order of what C01 measures for the Helios bond), and Treasury will ask the desk for the EPC and the Taxonomy assessment of every loan.
The second climate channel is physical. For a building the relevant hazards are acute (river and coastal flood, storm, wildfire, hail) and chronic (heat, drought-induced subsidence of clay soils, sea-level rise). Two tools are used on the desk: hazard maps (in France, the flood risk prevention plans, PPRI, and the Géorisques database; commercial vendors offer scores by address under climate scenarios) and the technical adviser’s review of the building’s vulnerability (where the plant rooms, substations and server rooms are; cooling capacity in a heat wave).
Insurance absorbs most of the direct loss, and the facility agreement requires it, with the lender named as loss payee. In France natural catastrophes are covered by the public-private Cat Nat regime: every property damage policy carries a compulsory extension financed by a uniform surcharge, raised from 12% to 20% of the property premium on 1 January 2025, with state-guaranteed reinsurance by CCR. Because the surcharge is uniform, the price of insurance does not reflect the exposure of a specific address; the signal that a building is in a flood zone reaches the owner through deductibles, the requirements of the PPRI and, increasingly, through valuers and lenders.
PITFALL — “It is insured” closes the discussion too early. Insurance is renewed every year and the loan lasts seven; business interruption cover is capped in time; deductibles are not negligible for large assets; and no policy pays for the lower rent and wider yield of a building known to flood.
Banks are also pushed from above. The EBA Guidelines on the management of ESG risks (EBA/GL/2025/01), applicable to large institutions since 11 January 2026, require that environmental risks be reflected in collateral valuation, in the monitoring of real estate portfolios and in plans required by CRD VI. The EBA Guidelines on loan origination and monitoring (EBA/GL/2020/06) already asked lenders to consider the energy efficiency of collateral. CRR3 itself touches valuation: Article 229 caps the prudential value of a property, but allows an upward revaluation where works “unequivocally” increase the value, energy-performance improvements being cited as an example. The Pillar 3 ESG templates require large banks to publish their real estate collateral by EPC class and energy-consumption bucket (section 6.5). The supervisory message is consistent: the bank must know the energy performance of what it holds as collateral, building by building.
The lifecycle below is that of a refurbishment financing with a green capex tranche. Durations are typical: three to four months from mandate to signing, two years of works, then five years of investment loan.
INTERACTIVE LAB — Lifecycle of a green retrofit financing
Follow the transaction from the first call to the last covenant test, with the role and the document at each step.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | REF banker (origination) | Meets the sponsor’s finance director; collects the business plan, lease schedule, energy audit and works programme | Indicative term sheet and pitch |
| 2 | Structurer | Builds the cash-flow model: works budget, drawdown curve, letting assumptions, covenants | Financial model and sizing memo |
| 3 | Collateral valuation unit | Instructs an external valuer from the panel for an as-is value and a value on completion; reviews the draft | Valuation report and internal review memo |
| 4 | Technical adviser | Audits the works programme, budget, planning, contractor and the energy design against the EPC target | Technical due diligence report with cost and delay contingencies |
| 5 | E&S risk and sustainable finance team | Tests eligibility against the bank’s green financing framework and the Taxonomy criteria; reviews the climate risk assessment | Green eligibility opinion |
| 6 | Credit risk, real estate team | Rates the transaction, assigns the prudential category, issues an independent opinion | Risk opinion and rating sheet |
| 7 | Credit committee | Approves amount, hold, pricing floor and conditions precedent | Credit approval with conditions |
| 8 | Legal counsel and structurer | Negotiate the facility agreement, security, hedging and green provisions | Signed facility agreement and security package |
| 9 | Syndication desk | Sells down part of the underwritten amount to club lenders | Transfer certificates; final hold |
| 10 | Loan agent with technical adviser | Processes each monthly drawdown against the certificate of works and the cost-to-complete test | Drawdown notices and monitoring reports |
| 11 | Sponsor and certified assessor | Deliver the building; obtain the EPC and the environmental certification; sign leases | Practical completion certificate, EPC class B, lease agreements |
| 12 | Loan agent and credit risk | Convert to investment terms; test covenants quarterly; commission the annual valuation; collect energy data | Compliance certificates, annual review, green loan report |
| 13 | Treasury and covered bond team | Check eligibility of the stabilised loan for the cover pool and the green sub-portfolio | Pool inclusion and allocation report |
Origination and structuring (weeks 1 to 4). The banker’s first question is the business plan: what rent, from whom, when. The second is the works: is the budget fixed-price, who bears overruns, how much contingency. The structurer turns the answers into a model with a monthly timeline during the works and quarterly afterwards. The output is a sizing memo that shows the debt capacity under each constraint and the covenant headroom in a downside case.
Valuation (weeks 2 to 6). Two values are needed: the market value of the building as it stands, and the market value on the special assumption that the works are completed and the building let at market rent. Valuers in Europe work under the RICS Red Book or the European Valuation Standards; both now require the valuer to comment on sustainability characteristics and on the capex needed to meet regulation. The valuation unit challenges the yield and rent retained and, for a retrofit, checks that the as-is value does not already embed the upside.
Technical and green due diligence (weeks 2 to 8). The technical adviser’s report is to a development loan what the valuation is to an investment loan. For a green loan it carries an additional duty: confirming that the design delivers the energy target, with the thermal study and the projected EPC. The sustainable finance team checks the criteria of the bank’s framework, drafts the green clauses (use of proceeds, information undertakings, declassification) and asks for the climate risk and vulnerability assessment that the Taxonomy requires.
Credit approval and documentation (weeks 6 to 14). Credit risk writes an opinion separate from the front office’s application. The committee approves a structure and often adds conditions: a pre-let threshold before the second half of the capex tranche can be drawn, a sponsor guarantee of cost overruns, an interest reserve account. Lawyers then negotiate; the points that take time are the completion guarantee, the cure rights and the letting undertakings.
Works phase (two years). Every month the borrower sends a drawdown request with the contractor’s invoices. The technical adviser certifies the progress and the cost to complete: at each drawdown the undrawn tranche plus committed equity must cover what remains to be spent. This test is the main control of a development loan. The agent also follows the letting: a signed pre-let changes the risk more than any ratio.
Stabilisation and investment phase. On practical completion, with the EPC delivered and a minimum occupancy reached, the margin steps down and the investment covenants start. From then on the agent tests LTV, ICR and debt yield, commissions a valuation each year and collects the building’s actual energy consumption, which feeds the bank’s own reporting (section 6.5).
WHO DOES WHAT — Three signatures protect the bank from greenwashing in this process, and none belongs to the front office: the certified assessor who issues the EPC, the technical adviser who confirms the design, and the sustainable finance team that gives the eligibility opinion. The banker’s job is to make sure the facility agreement obliges the borrower to deliver the documents on which those three rely.
All figures in this section are case assumptions for a fictional sponsor and a fictional building. Regulations, standards and methods are real. Amounts are in EUR million unless stated.
Question. What is Tour Galilée worth today, what would it be worth after the works, and what is Lutetia asking Meridian for?
Data. Tour Galilée is a 28-storey office tower delivered in 1994 in the 13th arrondissement of Paris, 150 metres from the Seine. It has 30,000 m² of lettable area, a single-glazed curtain wall, gas boilers and its original air handling. Its DPE is class F. The single tenant, a public-sector company, leaves on 31 December 2026 for a new building. The owner is an SPV, SCI Galilée 13, held by Lutetia Property, an unlisted property company owned by two institutional investors. Meridian is the existing lender: a EUR 90m loan matures in March 2027.
The external valuer retains, for the building as it stands, a market rent of EUR 380 per m², 8% of income lost to structural vacancy and non-recoverable costs, and a yield of 6.50%. For the refurbished building (new façade, all-electric heat pumps, new ventilation and building management system, DPE class B targeted, environmental certification), it retains a rent of EUR 520 per m², 5% of leakage and a yield of 5.25%. For reference, the prime yield of the Paris central business district stood at 4.35% in mid-2026 according to a broker’s quarterly publication (section 8); the 13th arrondissement trades wider.
Calculation.
Result. The building is worth EUR 161.35m today (EUR 5,378 per m²) and EUR 282.29m once refurbished and let (EUR 9,410 per m²): an uplift of EUR 120.93m, to be compared with a works budget of EUR 75m (EUR 2,500 per m², all professional fees included).
Reading. The uplift has two sources that the formula separates. Income rises by 41% (from 10.488 to 14.82). The yield tightens by 125 bp, which on its own multiplies value by 6.50/5.25 = 1.238. The product, 1.413 times 1.238, gives the factor 1.75 between the two values. More than a third of the value creation therefore rests on the valuer’s yield, a market price that the sponsor does not control. The break-even is easy to compute: the project destroys value if, on completion, the market capitalises the new income at more than 14.82 divided by the EUR 236.35m of as-is value plus works, that is 6.27%.
INTERACTIVE LAB — Tour Galilée: value as it stands and value after retrofit
Capitalise net operating income before and after the works and see how much of the uplift depends on the exit yield.
| Parameter you can move | Range | Default |
|---|---|---|
| Lettable area | 10000 to 60000 m² | 30000 m² |
| Market rent as it stands | 250 to 600 EUR/m² | 380 EUR/m² |
| Leakage as it stands (vacancy, non-recoverables) | 0 to 20 % | 8 % |
| Yield as it stands | 4 to 9 % | 6.5 % |
| Market rent after retrofit | 350 to 900 EUR/m² | 520 EUR/m² |
| Leakage after retrofit | 0 to 20 % | 5 % |
| Yield after retrofit | 3.5 to 8 % | 5.25 % |
| Works budget | 500 to 5000 EUR/m² | 2500 EUR/m² |
| Result at default settings | Value |
|---|---|
| NOI as it stands | 10.488 EUR m |
| Value as it stands | 161.35 EUR m |
| Value as it stands per m² | 5,378 EUR |
| NOI after retrofit | 14.820 EUR m |
| Value after retrofit | 282.29 EUR m |
| Value after retrofit per m² | 9,410 EUR |
| Value uplift | 120.93 EUR m |
| Works budget | 75.0 EUR m |
| Income factor (NOI after / before) | 1.413 x |
| Yield factor (yield before / after) | 1.238 x |
| Break-even yield on as-is value plus works | 6.27 % |
Try it. Set the yield after retrofit to 6.00%: three quarters of a point of yield removes EUR 35m of value, almost half the works budget.
The request. Lutetia asks Meridian to arrange a EUR 150m seven-year facility (December 2026 to December 2033): Tranche A of EUR 90m to refinance the existing loan, Tranche B of EUR 60m to finance 80% of the works, drawn monthly during 2027 and 2028 after Lutetia has spent its own EUR 15m. Lutetia wants the loan labelled green.
Question. How far is the building from a decarbonisation pathway today and after the works, and does the renovation pass the Taxonomy test?
Data. The energy audit gives a final energy consumption of 280 kWh per m² per year: 120 of gas and 160 of electricity. The design study for the retrofit gives 95 kWh per m², all electric. Case emission factors: 0.23 kgCO2e per kWh of gas and 0.05 kgCO2e per kWh of French grid electricity, held constant for simplicity. Primary energy: coefficient 1 for gas and 1.9 for electricity, the conversion factor assumed here for the French DPE since 1 January 2026 (it was 2.3 before; verify the factor applicable to the calculation method used). The works have an embodied carbon footprint estimated by the design team at 250 kgCO2e per m².
The pathway is a stylised one, built for teaching: 60 kgCO2e per m² in 2020, falling by 8% a year. It has the shape of a CRREM pathway but it is not the CRREM pathway for French offices; on a real file the analyst downloads the current pathway file from crrem.org and uses the country and property-type series.
Calculation.
The stranding year solves , so .
Result. Before the works the lines cross at 2026.3: the building is stranded now. After the works they cross at 2050.4. Annual emissions fall from 1,068 tCO2e to 142.5 tCO2e, a saving of 925.5 tCO2e a year. The works emit 7,500 tCO2e, repaid by the operational saving in 8.1 years. Primary energy demand falls from 424 to 180.5 kWh per m², a reduction of 57.4%; final energy falls by 66.1%.
Reading. Four lessons. (1) The renovation passes the current Taxonomy test for activity 7.2 with a wide margin (57.4% against 30%), and the 60% reduction in final energy required by the French tertiary rules for 2050. (2) It would miss by a short distance the additional route discussed in the 2026 draft (60% reduction in primary energy over ten years): final consumption would have to be 89.3 kWh per m² or lower, about 5.7 kWh per m² less than designed. A lender who wants the finished building, and not only the works, to qualify should raise this with the design team before the budget is frozen, when a few more kilowatt-hours cost little. (3) Gas is 43% of the energy but 78% of the emissions (27.6 out of 35.6): leaving gas, not insulation alone, moves the stranding year, which is specific to a low-carbon grid such as the French one. (4) The carbon payback of 8.1 years is longer than the loan. A retrofit is not free in carbon terms, though it compares well with demolition and reconstruction, whose embodied emissions are several times higher.
INTERACTIVE LAB — Stranding year against a stylised decarbonisation pathway
Compute the carbon intensity of the building before and after retrofit, find where each line crosses the pathway, and test the reduction in primary energy demand.
| Parameter you can move | Range | Default |
|---|---|---|
| Gas before works | 0 to 250 kWh/m² | 120 kWh/m² |
| Electricity before works | 50 to 300 kWh/m² | 160 kWh/m² |
| Gas after works | 0 to 150 kWh/m² | 0 kWh/m² |
| Electricity after works | 40 to 200 kWh/m² | 95 kWh/m² |
| Emission factor, gas | 0.18 to 0.26 kgCO2e/kWh | 0.23 kgCO2e/kWh |
| Emission factor, grid electricity | 0.02 to 0.5 kgCO2e/kWh | 0.05 kgCO2e/kWh |
| Pathway intensity in 2020 | 30 to 100 kgCO2e/m² | 60 kgCO2e/m² |
| Annual decline of the pathway | 3 to 12 % | 8 % |
| Primary energy factor of electricity | 1.9 / 2.3 | 1.9 |
| Embodied carbon of the works | 50 to 800 kgCO2e/m² | 250 kgCO2e/m² |
| Area | 10000 to 60000 m² | 30000 m² |
| Result at default settings | Value |
|---|---|
| Carbon intensity before | 35.60 kgCO2e/m² |
| Carbon intensity after | 4.75 kgCO2e/m² |
| Stranding year before works | 2,026.3 year |
| Stranding year after works | 2,050.4 year |
| Annual emissions before | 1,068.0 tCO2e |
| Annual emissions after | 142.5 tCO2e |
| Annual saving | 925.5 tCO2e |
| Share of gas in emissions before | 77.5 % |
| Embodied emissions of the works | 7,500 tCO2e |
| Carbon payback | 8.1 years |
| Primary energy demand before | 424.0 kWh/m² |
| Primary energy demand after | 180.5 kWh/m² |
| Reduction in primary energy demand | 57.4 % |
| Reduction in final energy | 66.1 % |
| Electricity after works needed for a 60% PED reduction (no gas) | 89.3 kWh/m² |
Try it. Set the grid emission factor to 0.35, the order of magnitude of a coal-heavy grid: the all-electric retrofit no longer escapes stranding, which is why CRREM pathways are country-specific.
Question. If Meridian simply refinanced the EUR 90m on the unrenovated building, what would its position be at the 2033 deadline of the EPBD?
Data. Scenario for 2033: the building is re-let as it stands, but market rent for class F space is 10% lower than today and the yield required by buyers for a building above the regulatory threshold has widened by 100 bp, to 7.50%. The debt is unchanged at EUR 90m (investment loans amortise little).
Calculation.
Result. The value falls by 22.0%, of which 13.3% from the yield alone. LTV rises from 55.8% today to 71.5%, above the 65% covenant, with no new lending.
Reading. A 100 bp brown discount on the yield is a scenario, not a forecast; section 8 discusses the evidence. What the calculation shows is the leverage of the mechanism: the lender who “does nothing” on a 56% LTV loan ends above 70% through a moderate combination of rent and yield, and at that point the refinancing market for a non-compliant building is thin. Doing nothing is a credit decision with its own risk profile, and usually the worse one. This comparison belongs in the credit application: the retrofit loan is larger (EUR 150m against 90m) but secured on a building that will still be lettable at maturity.
INTERACTIVE LAB — Brown discount: from yield and rent to LTV
Translate a wider yield and a lower rent on the unrenovated building into value and loan-to-value on the existing EUR 90m loan.
| Parameter you can move | Range | Default |
|---|---|---|
| NOI as it stands | 5 to 20 EUR m | 10.488 EUR m |
| Yield today | 4 to 9 % | 6.5 % |
| Widening of the yield | 0 to 300 bp | 100 bp |
| Fall in rent | 0 to 40 % | 10 % |
| Loan outstanding | 50 to 150 EUR m | 90 EUR m |
| Result at default settings | Value |
|---|---|
| Value today | 161.35 EUR m |
| LTV today | 55.8 % |
| Value in the scenario | 125.86 EUR m |
| Change in value | -22.0 % |
| of which yield alone | -13.3 % |
| LTV in the scenario | 71.5 % |
Try it. Find the yield widening that breaches the 65% covenant with rent unchanged: about 110 bp. With a 10% fall in rent, about 30 bp is enough.
Question. What is the total cost of the project, including financing, and what margin does it leave Lutetia?
Data. Terms proposed by Meridian for the works phase: margin 260 bp, loan fully swapped at 2.50%, so an all-in rate of 5.10%; commitment fee of 40% of the margin (104 bp) on the undrawn part of Tranche B; arrangement fee of 125 bp on EUR 150m. Tranche A is drawn for the 24 months of the works; Tranche B is drawn evenly, so it is on average half drawn. Interest and fees are paid by Lutetia from equity.
Calculation.
| Item | Computation | EUR m |
|---|---|---|
| Interest on Tranche A | 90 x 5.10% x 2 years | 9.180 |
| Interest on Tranche B | 30 x 5.10% x 2 years | 3.060 |
| Commitment fee | 30 x 1.04% x 2 years | 0.624 |
| Arrangement fee | 150 x 1.25% | 1.875 |
| Financing costs | 14.739 |
Total development cost is the as-is value (the opportunity cost of the building), the works and the financing costs: 161.35 + 75 + 14.74 = EUR 251.09m.
Result. Profit is 282.29 - 251.09 = EUR 31.19m, or 12.4% of cost. The yield on cost is 14.82 / 251.09 = 5.90%, against an exit yield of 5.25%: a development spread of 65 bp. Lutetia’s equity in the project is EUR 101.09m: the EUR 71.35m of equity already in the building and EUR 29.74m of new cash.
Reading. A profit on cost of 12.4% is thinner than the 15 to 20% developers usually ask for speculative projects, and it excludes letting fees and rent-free periods. The sponsor’s logic is defensive as much as offensive: section 5.3 shows what the alternative is worth. For the lender the important figure is the cushion: value on completion can fall by 11% (31.19 / 282.29) before the project is worth less than it cost, and by 47% before the EUR 150m of debt is uncovered.
INTERACTIVE LAB — Development economics of the retrofit
Add up as-is value, works and financing costs, and compute profit on cost, yield on cost and the sponsor’s equity.
| Parameter you can move | Range | Default |
|---|---|---|
| As-is value | 80 to 300 EUR m | 161.35 EUR m |
| Works budget | 30 to 150 EUR m | 75 EUR m |
| Cost overrun on works | 0 to 30 % | 0 % |
| Duration of the works | 12 to 42 months | 24 months |
| Tranche A (drawn throughout) | 0 to 150 EUR m | 90 EUR m |
| Tranche B (drawn evenly) | 0 to 120 EUR m | 60 EUR m |
| All-in rate during works | 3 to 8 % | 5.1 % |
| Commitment fee on undrawn Tranche B | 0 to 200 bp | 104 bp |
| Arrangement fee | 0 to 200 bp | 125 bp |
| NOI on completion | 8 to 25 EUR m | 14.82 EUR m |
| Exit yield | 4 to 7.5 % | 5.25 % |
| Result at default settings | Value |
|---|---|
| Interest on Tranche A | 9.180 EUR m |
| Interest on Tranche B | 3.060 EUR m |
| Commitment fee | 0.624 EUR m |
| Arrangement fee | 1.875 EUR m |
| Financing costs | 14.739 EUR m |
| Total development cost | 251.09 EUR m |
| Value on completion | 282.29 EUR m |
| Profit | 31.20 EUR m |
| Profit on cost | 12.4 % |
| Yield on cost | 5.90 % |
| Development spread (yield on cost less exit yield) | 65 bp |
| Sponsor equity in the project | 101.09 EUR m |
| of which new cash | 29.74 EUR m |
| Fall in completed value before value is below cost | 11.1 % |
| Fall in completed value before debt is uncovered | 46.9 % |
Try it. Add a 20% cost overrun and six months of delay: profit on cost falls below 5%. This is the scenario the completion guarantee and the contingency are sized for.
Question. How much senior debt can the project carry, and which ratio decides?
Data. Meridian’s credit policy for a refurbishment of this quality (case assumptions): LTV on value on completion at most 55%; loan-to-cost at most 65% of as-is value plus works; debt yield on stabilised NOI at least 9.5%; ICR on stabilised NOI at least 2.0x at the hedged all-in rate of 4.40% (swap 2.50% plus the post-completion margin of 190 bp).
Calculation and result.
| Constraint | Computation | Debt capacity (EUR m) |
|---|---|---|
| LTV 55% | 55% x 282.29 | 155.26 |
| LTC 65% | 65% x 236.35 | 153.63 |
| Debt yield 9.5% | 14.82 / 9.5% | 156.00 |
| ICR 2.0x | 14.82 / (2.0 x 4.40%) | 168.41 |
The loan-to-cost constraint binds at EUR 153.63m; the request of EUR 150m fits under all four. At EUR 150m: LTV on completion 53.1%, LTC 63.5%, debt yield 9.88%, ICR 2.25x, and DSCR 1.67x once the loan amortises 1.5% a year (EUR 2.25m) from 2029.
Reading. The three “value-side” constraints land within EUR 3m of each other, which is a sign of a coherent policy: each protects against a different error (the valuer’s yield, the cost base, the income). The ICR is slack because the swap rate is low relative to the property yield; it would bind first if rates rose before the hedge is traded, which is why the hedge is executed at signing. Note what the LTV constraint does not say: during the works the loan is secured on a building site. On the as-is value, Tranche A alone is at 55.8%, and at mid-works the lender relies on the cost-to-complete test, not on LTV.
INTERACTIVE LAB — Debt capacity under four constraints
Compute the maximum loan allowed by LTV, loan-to-cost, debt yield and interest cover, identify the binding constraint and read the ratios at the requested amount.
| Parameter you can move | Range | Default |
|---|---|---|
| Value on completion | 150 to 450 EUR m | 282.29 EUR m |
| As-is value plus works | 120 to 400 EUR m | 236.35 EUR m |
| Stabilised NOI | 8 to 25 EUR m | 14.82 EUR m |
| Maximum LTV | 40 to 70 % | 55 % |
| Maximum loan-to-cost | 50 to 80 % | 65 % |
| Minimum debt yield | 7 to 12 % | 9.5 % |
| Minimum ICR | 1.25 to 3 x | 2 x |
| Hedged all-in rate after completion | 3 to 8 % | 4.4 % |
| Requested loan | 80 to 220 EUR m | 150 EUR m |
| Amortisation after completion | 0 to 4 % a year | 1.5 % a year |
| Result at default settings | Value |
|---|---|
| Capacity under LTV | 155.26 EUR m |
| Capacity under loan-to-cost | 153.63 EUR m |
| Capacity under debt yield | 156.00 EUR m |
| Capacity under ICR | 168.41 EUR m |
| Binding capacity | 153.63 EUR m |
| LTV on completion at requested loan | 53.1 % |
| Loan-to-cost at requested loan | 63.5 % |
| Debt yield at requested loan | 9.88 % |
| ICR at requested loan | 2.25 x |
| DSCR at requested loan | 1.67 x |
Try it. Raise the all-in rate to 6%: the ICR becomes the binding constraint and capacity falls to EUR 123.5m. Hedging at signing freezes the constraint where it is.
Question. On what basis is the facility green, what does the facility agreement say, and what are the green terms worth to Lutetia?
Structure retained. Meridian’s green financing framework accepts renovations that meet the Taxonomy criterion of activity 7.2. The label is therefore anchored on the works:
Data for the valuation of the green terms. The 10 bp step-down applies to the outstanding amount over the five years 2029 to 2033: EUR 150m at the start of 2029, reduced by EUR 2.25m each year. Energy price assumed for the tenant: EUR 0.15 per kWh, blended.
Calculation and result. The balances at the start of each of the five years add up to EUR 727.5m-years; at 10 bp the step-down is worth EUR 0.7275m to Lutetia over the life of the loan, EUR 0.15m in the first year. That is 0.97% of the works budget and 0.60% of the value uplift. The energy saved, 185 kWh per m², is worth EUR 27.75 per m² a year at the assumed price, EUR 0.83m a year for the building: more than five times the margin step-down, and it accrues to the tenant through lower charges. It explains 19.8% of the EUR 140 per m² rent uplift assumed by the valuer.
Reading. The honest hierarchy is the following. The retrofit is decided by the value at stake (EUR 121m of uplift, or the avoidance of section 5.3). The energy bill explains a fifth of the rent uplift; the rest is the price tenants pay for a renovated building and for compliance with their own commitments. The green margin incentive is two orders of magnitude smaller than the uplift. As with the greenium of the Helios bond (C01), the pricing benefit of the label is real, small and not the reason for the project. What the green loan format brings is discipline: a defined scope of works, an independent confirmation of the energy target, and data delivered every year, all of which the credit analysis needs anyway.
A second point for the sustainable finance team: after completion, can the whole loan be reported as financing a Taxonomy-aligned asset under activity 7.7? Only if Lutetia shows that the building is in class A or within the top 15% of the national or regional stock. Class B does not prove it. The banker should ask for the stock study with the valuation, and should not promise Treasury an inclusion in the green covered bond pool before seeing it.
INTERACTIVE LAB — What the green terms are worth, in proportion
Value the margin step-down over the investment phase and compare it with the works budget, the value uplift and the tenant’s energy saving.
| Parameter you can move | Range | Default |
|---|---|---|
| Loan at stabilisation | 50 to 250 EUR m | 150 EUR m |
| Annual amortisation | 0 to 4 % of initial loan | 1.5 % of initial loan |
| Green margin step-down | 0 to 30 bp | 10 bp |
| Years of step-down | 1 to 7 years | 5 years |
| Works budget | 30 to 150 EUR m | 75 EUR m |
| Value uplift | 20 to 250 EUR m | 120.93 EUR m |
| Final energy saved | 0 to 300 kWh/m² | 185 kWh/m² |
| Energy price | 0.05 to 0.4 EUR/kWh | 0.15 EUR/kWh |
| Area | 10000 to 60000 m² | 30000 m² |
| Rent uplift | 20 to 400 EUR/m² | 140 EUR/m² |
| Result at default settings | Value |
|---|---|
| Sum of annual balances | 727.5 EUR m-years |
| Step-down, first year | 0.150 EUR m |
| Step-down over the period | 0.7275 EUR m |
| as a share of the works budget | 0.97 % |
| as a share of the value uplift | 0.60 % |
| Energy saving for the tenant | 27.75 EUR/m² a year |
| Energy saving for the building | 0.8325 EUR m a year |
| Share of the rent uplift explained by energy | 19.8 % |
| Energy saving divided by first-year step-down | 5.55 x |
Try it. Even a 25 bp step-down, far above market practice, is worth 2.4% of the works budget. The label does not finance the retrofit.
Question. What is the expected annual loss from flooding, who bears it, and is adaptation capex worth requiring?
Data. The tower stands in the flood zone of the Seine mapped by the Paris flood risk prevention plan; the reference event is the flood of January 1910. The technical adviser notes that the electrical substation, the chillers’ pumps and the lift machinery pits are in the second basement. Case assumptions: reinstatement value EUR 180m (EUR 6,000 per m²); a major flood with an annual probability of 1% damages 12% of the reinstatement value and closes the building for nine months; a moderate flood with an annual probability of 3% damages 2% and closes it for one month. Lost rent is counted on the stabilised NOI of EUR 14.82m. Insurance: the Cat Nat extension of the property policy, with a deductible assumed at 10% of direct damage for a commercial property, and a business interruption policy that covers lost rent. Adaptation option: relocating the substation and pumps to the podium levels and installing removable flood barriers, for EUR 3m, which the engineer estimates would cut damage and closure time by 60%.
Calculation.
| Event | Probability | Damage | Lost rent | Total loss |
|---|---|---|---|---|
| Major flood | 1% | 12% x 180 = 21.60 | 9/12 x 14.82 = 11.115 | 32.715 |
| Moderate flood | 3% | 2% x 180 = 3.60 | 1/12 x 14.82 = 1.235 | 4.835 |
Expected annual loss: 1% x 32.715 + 3% x 4.835 = EUR 0.472m, or 3.2% of NOI. Retained by the owner after insurance (the deductible on direct damage): 1% x 2.16 + 3% x 0.36 = EUR 0.032m a year.
Result. With adaptation, the gross expected annual loss falls by 60%, to EUR 0.189m: EUR 0.283m avoided each year. Capitalised at the building’s yield of 5.25%, the avoided loss is worth EUR 5.40m, against a cost of EUR 3m. But the owner’s own retained loss falls by only EUR 0.019m a year, worth EUR 0.37m capitalised.
Reading. Here is the real difficulty of physical risk in a market with mutualised insurance. For society and for the insurance pool the adaptation is worth nearly twice its cost; for the owner, who pays a uniform Cat Nat surcharge whatever the address, it is worth an eighth of its cost in avoided deductibles. The private incentive is weak, and it is the lender who can restore it, for three reasons of its own. First, the loan lasts seven years and the policy one: terms can change. Second, nine months without rent is a covenant event even if insurance eventually pays, and business interruption cover has limits. Third, the Taxonomy’s do-no-significant-harm test for adaptation requires that material physical risks identified in the assessment be addressed by an adaptation plan: no adaptation, no aligned green loan. Meridian therefore makes the relocation of the technical rooms part of the works programme, a EUR 3m item inside the EUR 75m budget, and requires business interruption cover of at least 24 months.
Heat is the second hazard. The thermal study must show that the refurbished building keeps acceptable indoor temperatures in a heat-wave design scenario without oversizing the cooling plant, which would damage the energy target; the technical adviser reviews it. No number is attached here because the effect goes through rent and tenant retention more than through a loss event.
INTERACTIVE LAB — Flood risk: expected annual loss, insurance and adaptation
Compute the expected annual loss from two flood events, the part retained after insurance, and the value of adaptation works for society and for the owner.
| Parameter you can move | Range | Default |
|---|---|---|
| Reinstatement value | 50 to 400 EUR m | 180 EUR m |
| Annual NOI | 5 to 25 EUR m | 14.82 EUR m |
| Annual probability, major flood | 0.2 to 5 % | 1 % |
| Damage, major flood | 2 to 40 % of reinstatement value | 12 % of reinstatement value |
| Closure, major flood | 0 to 24 months | 9 months |
| Annual probability, moderate flood | 0 to 10 % | 3 % |
| Damage, moderate flood | 0 to 10 % of reinstatement value | 2 % of reinstatement value |
| Closure, moderate flood | 0 to 6 months | 1 months |
| Deductible on direct damage | 0 to 30 % | 10 % |
| Reduction of loss by adaptation works | 0 to 90 % | 60 % |
| Cost of adaptation works | 0.5 to 10 EUR m | 3 EUR m |
| Capitalisation rate | 3 to 9 % | 5.25 % |
| Result at default settings | Value |
|---|---|
| Total loss, major flood | 32.715 EUR m |
| Total loss, moderate flood | 4.835 EUR m |
| Expected annual loss, gross | 0.472 EUR m |
| as a share of NOI | 3.2 % |
| Expected annual loss retained by the owner | 0.0324 EUR m |
| Gross expected annual loss with adaptation | 0.189 EUR m |
| Gross loss avoided each year | 0.283 EUR m |
| Avoided loss, capitalised | 5.40 EUR m |
| Owner’s retained loss avoided each year | 0.0194 EUR m |
| Owner’s avoided loss, capitalised | 0.37 EUR m |
Try it. Double both probabilities, a crude way to represent a wetter climate scenario: the social value of adaptation doubles and the owner’s private value still does not cover the cost.
Question. In December 2033, can the balloon be refinanced in a bad market, and how does this compare with the do-nothing loan?
Data. Balloon: EUR 150m less five annual repayments of EUR 2.25m, that is EUR 138.75m. Stress for the refurbished building: NOI 5% below plan and exit yield at 6.00% (75 bp wider than the valuation on completion). A new lender advances 55% of value and tests ICR at an all-in rate of 5.00%. For the unrenovated building under a severe version of section 5.3: NOI 15% below today’s, yield at 8.00%, debt of EUR 90m.
Calculation and result.
Debt yield at exit is 10.15% and the refinancing ICR 2.03x. A new lender at 55% LTV would advance EUR 129.06m, leaving a gap of EUR 9.69m for Lutetia to fill with equity. For the unrenovated building: value EUR 111.44m, LTV 80.8%, new loan EUR 61.29m and a gap of EUR 28.71m.
Reading. Under stress, the retrofit loan stays inside its 65% covenant and needs an equity contribution of 6.5% of the original loan at refinancing: uncomfortable for the sponsor, not a loss for the bank. The do-nothing loan, although 40% smaller, shows a gap three times larger and an LTV at which enforcement becomes a real scenario. The larger loan is the safer loan. This is the central credit argument of the file, and it holds only if the retrofit is delivered: hence the weight placed on the technical adviser, the cost-to-complete test and the completion guarantee.
INTERACTIVE LAB — Refinancing stress at maturity: retrofit against do-nothing
Stress NOI and exit yield at maturity, compute LTV, debt yield and the refinancing gap for the refurbished building and for the unrenovated one.
| Parameter you can move | Range | Default |
|---|---|---|
| Retrofit loan at stabilisation | 80 to 220 EUR m | 150 EUR m |
| Annual amortisation over 5 years | 0 to 4 % of loan | 1.5 % of loan |
| Stabilised NOI, plan | 8 to 25 EUR m | 14.82 EUR m |
| NOI shortfall at exit | 0 to 30 % | 5 % |
| Exit yield | 4.5 to 8 % | 6 % |
| LTV offered by a new lender | 40 to 65 % | 55 % |
| All-in rate at refinancing | 3 to 8 % | 5 % |
| NOI of the unrenovated building today | 5 to 20 EUR m | 10.488 EUR m |
| NOI shortfall, unrenovated | 0 to 50 % | 15 % |
| Exit yield, unrenovated | 6 to 10 % | 8 % |
| Do-nothing loan | 50 to 120 EUR m | 90 EUR m |
| Result at default settings | Value |
|---|---|
| Balloon of the retrofit loan | 138.75 EUR m |
| Exit value, refurbished | 234.65 EUR m |
| LTV at exit, refurbished | 59.1 % |
| Debt yield at exit | 10.15 % |
| ICR at the refinancing rate | 2.03 x |
| New loan available, refurbished | 129.06 EUR m |
| Refinancing gap, refurbished | 9.69 EUR m |
| Gap as a share of the original loan | 6.5 % |
| Exit value, unrenovated | 111.43 EUR m |
| LTV at exit, unrenovated | 80.8 % |
| New loan available, unrenovated | 61.29 EUR m |
| Refinancing gap, unrenovated | 28.71 EUR m |
Try it. Find the exit yield at which the retrofit loan breaches 65% with a 5% NOI shortfall: about 6.6%, some 135 bp wider than the valuation on completion.
Question. The building is delivered in December 2028, on time and on budget. In June 2029 two things differ from the plan: the certified assessor issues a DPE in class C, and the leases signed average EUR 500 per m² instead of 520. What happens under the facility agreement, and what should Meridian do?
Data. The as-built energy study gives a final consumption of 110 kWh per m² instead of 95: the tenant of the upper floors has installed a trading floor with its own cooling, and the heat-recovery unit performs below its specification. Leakage and yield are as in the valuation on completion.
Calculation and result. With 110 kWh per m², carbon intensity is 5.5 kgCO2e per m² and the stranding year on the stylised pathway moves from 2050.4 to 2048.7: almost nothing. Primary energy demand is 209 kWh per m², a reduction of 50.7% against the pre-works level, and final energy is down 60.7%. NOI is 30,000 x 500 x 0.95 = EUR 14.25m, the value EUR 271.43m; at EUR 150m of debt, LTV is 55.3%, debt yield 9.50% and ICR 2.16x at the all-in rate of 4.40%.
Reading. Sort the consequences by nature, because the agreement does.
Credit. Nothing is breached. LTV is ten points inside the covenant, interest cover is comfortable, and the debt yield sits exactly on the level Meridian’s policy required at origination. The loan converts to investment terms at 190 bp. The annual review records a rent 4% under plan and no change of rating.
Label. The renovation still exceeds the 30% test of activity 7.2, so the works financed by Tranche B remain what they were described to be. But the contractual condition for the 180 bp margin was the class B certificate by the long-stop date, and it is not met. The step-down does not apply; the borrower must present a remediation plan; and unless the agreement gives a cure period, the loan is declassified. A well-drafted clause gives one: here, twelve months to recommission the heat recovery, separate the tenant’s process cooling in the calculation where the method allows it, and apply for a new certificate.
Bank reporting. The sustainable finance team updates the allocation register: the loan stays in the green portfolio during the cure period with a flag, and leaves it if the cure fails. The collateral moves in the Pillar 3 table to the bucket that corresponds to 209 kWh per m², and the EPC label recorded is C. Treasury is told that the loan is not a candidate for the green cover pool.
Two lessons for the desk. First, the gap between design and measured performance is the normal state of buildings, not an accident; a green loan that tests only a projected certificate will sooner or later report something untrue. This is why the agreement asks for actual consumption every year and why the long-stop date is placed six months after delivery, once the systems have run through a winter. Second, the credit and the label are separate risks with separate remedies. A banker who treats a missed EPC as a default overreacts and damages the relationship; one who quietly keeps the label damages the bank. The declassification clause exists so that neither happens.
| Item | Value | Where |
|---|---|---|
| NOI as it stands / after retrofit | EUR 10.488m / 14.82m | 5.1 |
| Value as it stands / after retrofit | EUR 161.35m / 282.29m | 5.1 |
| Works budget; value uplift | EUR 75m; EUR 120.93m | 5.1 |
| Break-even exit yield | 6.27% | 5.1 |
| Carbon intensity before / after | 35.6 / 4.75 kgCO2e per m² | 5.2 |
| Stranding year before / after (stylised pathway) | 2026 / 2050 | 5.2 |
| Reduction in primary energy demand | 57.4% (test: 30%) | 5.2 |
| Carbon payback of the works | 8.1 years | 5.2 |
| Do-nothing 2033: value change; LTV | -22.0%; 71.5% | 5.3 |
| Financing costs; total development cost | EUR 14.74m; EUR 251.09m | 5.4 |
| Profit on cost; yield on cost | 12.4%; 5.90% | 5.4 |
| Binding debt capacity (loan-to-cost) | EUR 153.63m | 5.5 |
| At EUR 150m: LTV, LTC, debt yield, ICR, DSCR | 53.1%, 63.5%, 9.88%, 2.25x, 1.67x | 5.5 |
| Green step-down over five years | EUR 0.7275m (0.97% of works) | 5.6 |
| Tenant energy saving | EUR 27.75 per m², EUR 0.83m a year | 5.6 |
| Flood: expected annual loss, gross / retained | EUR 0.472m / 0.032m | 5.7 |
| Adaptation: cost; capitalised gross benefit | EUR 3m; EUR 5.40m | 5.7 |
| Exit stress: balloon; LTV; refinancing gap | EUR 138.75m; 59.1%; EUR 9.69m | 5.8 |
| Do-nothing exit stress: LTV; gap | 80.8%; EUR 28.71m | 5.8 |
Meridian underwrites the EUR 150m, keeps the agency and sells EUR 75m to two club lenders shortly after signing. Its final hold is EUR 75m, half of every drawing.
IN ONE SENTENCE — A loan repaid by the rents of the financed building is “income-producing real estate” (IPRE), and CRR3 gives it risk weights driven by the ratio of the loan to the property value, by a supervisory category or by the bank’s own PD.
Three treatments coexist, and a REF banker meets all three.
Standardised approach. Article 126 of the CRR as amended by Regulation (EU) 2024/1623 weights IPRE exposures on commercial property by the exposure-to-value ratio: 70% up to 60%, 90% between 60% and 80%, 110% above 80%. (Where repayment does not depend on the property’s cash flows, the part of the loan up to 55% of the value is weighted 60% and the rest as an unsecured exposure to the borrower.) Land acquisition, development and construction (ADC) exposures fall under Article 126a and are weighted 150%. Whether a heavy refurbishment of an emptied tower is an ADC exposure during the works is a classification question that credit risk must settle and document; the prudent answer is yes until completion.
Slotting. Under the IRB approach, IPRE is a class of specialised lending. A bank that cannot estimate PD for it assigns each loan to one of four supervisory categories (Article 153(5)): strong (70%, or 50% under 2.5 years of remaining maturity), good (90%, or 70%), satisfactory (115%) and weak (250%), with regulatory expected losses of 0.4%, 0.8%, 2.8% and 8%. The EBA consulted from May to August 2026 on revised technical standards for the assignment to categories, in which, according to public commentary, ESG factors become a risk driver that can improve as well as worsen the category.
F-IRB with a PD model. Case assumption: Meridian has supervisory approval to use its own PD estimates for IPRE, with regulatory LGDs. For a loan secured by commercial property, CRR3 sets an LGD of 20% on the part covered by the collateral after a 40% haircut and 40% on the rest.
FORMULA
is the prudential value of the property; its value after the 40% haircut; the exposure of all lenders ranking equally on the mortgage. The risk weight then follows the corporate IRB formula given in B03, with and the PD.
CASE — Stabilised phase: = 282.29, = 169.37, above the exposure of 150, so = 20%. With a PD of 1.00% the risk weight is 41.03%; on Meridian’s EUR 75m, RWA are EUR 30.77m and capital EUR 4.00m. Standardised: exposure-to-value 53.1%, weight 70%, capital EUR 6.83m. Slotting “good”: 90%, capital EUR 8.78m. A bank bound by the output floor counts 72.5% of the standardised weight at the margin, 50.75%, capital EUR 4.95m (fully phased in 2030; the floor is 55% in 2026). Works phase, at mid-works: exposure of all lenders 132 (120 drawn plus 40% of 30 undrawn), = 161.35, = 96.81, secured share 73.3%, = 25.33%; with a PD of 2.00% the risk weight is 64.65%.
NOTE — Two caveats. Eligibility of the property as collateral under F-IRB is subject to conditions on legal certainty, valuation and monitoring, and, for loans repaid by the property’s own cash flows, to a test on national loss rates; if it fails, the LGD is 40% and the risk weight doubles to 82.06%. And the prudential value is not the latest market value: Article 229 limits upward revaluations, with an exception for works that improve energy performance, so the valuation unit must document why the post-retrofit value can be used.
INTERACTIVE LAB — Capital for an IPRE loan: standardised, slotting, F-IRB
Compute the collateral-adjusted LGD and compare RWA and capital across the three CRR3 treatments and the output floor.
| Parameter you can move | Range | Default |
|---|---|---|
| Meridian’s exposure at default | 10 to 150 EUR m | 75 EUR m |
| Prudential value of the property | 80 to 400 EUR m | 282.29 EUR m |
| Exposure of all pari passu lenders | 20 to 300 EUR m | 150 EUR m |
| One-year PD | 0.1 to 5 % | 1 % |
| Slotting risk weight | 70 / 90 / 115 / 250 | 90 |
| Treated as ADC under the standardised approach | no / yes | no |
| Result at default settings | Value |
|---|---|
| Collateral value after 40% haircut | 169.37 EUR m |
| Secured share of the exposure | 100.0 % |
| Collateral-adjusted LGD | 20.00 % |
| F-IRB risk weight | 41.03 % |
| F-IRB risk weight if collateral not eligible (LGD 40%) | 82.06 % |
| Exposure-to-value | 53.1 % |
| Standardised risk weight | 70 % |
| RWA, F-IRB | 30.77 EUR m |
| Capital, F-IRB | 4.00 EUR m |
| Capital, standardised | 6.83 EUR m |
| Capital, slotting | 8.78 EUR m |
| Capital at 72.5% of standardised RWA | 4.95 EUR m |
Try it. Switch the works phase to ADC: the standardised weight goes to 150%, which is why floor-bound banks price development loans so much higher.
Assumptions stated once: FTP liquidity premium of 42 bp for seven-year funding (6 bp per year of tenor); cost-income ratio of 40% applied to revenues before any green step-down; economic expected loss equal to PD times the collateral-adjusted LGD times EAD; arrangement fee of 125 bp on EUR 150m, of which 80 bp is passed on to the club lenders on their EUR 75m, the net EUR 1.275m being spread over seven years; contingent liquidity charge of 4 bp on undrawn amounts (HQLA carry of 40 bp on an assumed 10% outflow rate).
Stabilised year, Meridian’s EUR 75m, margin 190 bp:
| Line | Computation | EUR m |
|---|---|---|
| Net margin after FTP | 75 x (190 - 42) bp | 1.110 |
| Net arrangement fee, spread | 1.275 / 7 | 0.182 |
| Revenues | 1.292 | |
| Operating costs | 40% | -0.517 |
| Expected loss | 1.00% x 20% x 75 | -0.150 |
| Pre-tax result | 0.625 |
The loan is just under the 12% hurdle before any ancillary revenue. Three variations tell the story of the green feature.
During the works the return is lower despite the higher margin: revenues EUR 1.640m at mid-works (60 drawn at 218 bp net, 15 undrawn at 100 bp net, plus the fee), costs EUR 0.656m, expected loss EUR 0.334m, capital EUR 5.55m, RAROC 8.78%. The development phase is accepted because it creates the stabilised loan; committees look at the return over the life.
Under other capital regimes the same stabilised loan returns 9.48% for a floor-bound bank, 6.87% under the standardised approach and 5.34% under slotting “good”. This spread explains who lends: banks with real estate PD models, insurers and debt funds outside the bank capital framework compete on the best assets, and standardised banks retreat to higher-margin situations.
INTERACTIVE LAB — RAROC of Meridian’s hold in Tour Galilée
Rebuild revenues, costs, expected loss, capital and RAROC of the EUR 75m hold, and test the green step-down, a rating upgrade, the funding benefit and the capital approach.
| Parameter you can move | Range | Default |
|---|---|---|
| Drawn amount held | 0 to 150 EUR m | 75 EUR m |
| Undrawn commitment held | 0 to 60 EUR m | 0 EUR m |
| Margin | 100 to 350 bp | 190 bp |
| Green step-down | 0 to 30 bp | 0 bp |
| FTP liquidity premium | 0 to 80 bp | 42 bp |
| Net arrangement fee kept | 0 to 3 EUR m | 1.275 EUR m |
| One-year PD | 0.1 to 5 % | 1 % |
| Collateral-adjusted LGD | 20 to 40 % | 20 % |
| Capital approach | F-IRB / Output floor (50.75%) / Standardised 70% / Slotting good 90% | F-IRB |
| Result at default settings | Value |
|---|---|
| Net margin after FTP and step-down | 1.110 EUR m |
| Commitment fee net of liquidity charge | 0.000 EUR m |
| Net arrangement fee spread over 7 years | 0.182 EUR m |
| Revenues | 1.292 EUR m |
| Operating costs (40% of revenues before step-down) | 0.517 EUR m |
| Expected loss | 0.1500 EUR m |
| Pre-tax result | 0.625 EUR m |
| Risk weight | 41.03 % |
| RWA | 30.77 EUR m |
| Allocated capital | 4.000 EUR m |
| RAROC after tax | 11.72 % |
Try it. With the 10 bp step-down, find the margin that restores 12% under F-IRB: about 210 bp. Then apply the PD of 0.75%: 190 bp is enough.
A seven-year bullet-like loan is expensive to fund: the 42 bp premium absorbs more than a fifth of the margin. Two things soften it. Undrawn capex commitments attract only the contingent liquidity charge until drawn. And once stabilised, the loan is eligible for the covered bond issuer’s cover pool up to 60% of the property value (EUR 169.37m, above the loan): covered bonds are the cheapest term funding a bank has, and Treasury passes part of the benefit to the desk. A green covered bond adds little in price but widens the investor base; its allocation report needs the EPC and, under the European Green Bond Standard, a Taxonomy assessment that class B alone does not satisfy (section 5.6).
Under IFRS 9 the loan is held at amortised cost if it passes the “solely payments of principal and interest” test. A margin that varies with a sustainability condition raised doubts until the IASB’s amendments of May 2024, effective for periods beginning on or after 1 January 2026, clarified how contingent features of this kind are assessed and added disclosures; a 10 bp step-down linked to the EPC of the financed building is consistent with a basic lending arrangement. The net arrangement fee kept for the retained share enters the effective interest rate; the part earned for arranging the shares sold is recognised at signing. Expected credit losses are booked from day one (stage 1, twelve-month loss); a covenant breach, a watch-list entry or a rating downgrade moves the loan to stage 2 and lifetime loss. Forward-looking information must include, where material, the regulatory deadlines of the collateral: an unrenovated class F building approaching 2030 is a legitimate stage 2 indicator in an internal policy.
Pillar 3 ESG. Large banks disclose loans secured by immovable property by EPC label and by energy-consumption bucket of the collateral. At 424 kWh per m² of primary energy the tower sits in the 400 to 500 bucket; at 180.5 it moves to the 100 to 200 bucket. The framework is being simplified: the EBA’s no-action letter of 6 August 2025 asked supervisors not to prioritise enforcement of the Taxonomy-related templates, including the Green Asset Ratio, and its final draft technical standards of June 2026 (not yet adopted by the Commission) discontinue the Taxonomy-alignment disclosures in Pillar 3 for large institutions while keeping the climate-risk templates. The collateral-by-energy-performance view is what remains, and it is the one supervisors use.
Taxonomy reporting. Where Meridian still reports alignment under Article 8 of the Taxonomy Regulation, the capex tranche counts as aligned under activity 7.2 during the works; the whole loan would count under 7.7 only with class A or top-15% evidence.
Financed emissions. PCAF’s standard (Part A, third version, December 2025) attributes a building’s emissions to the lender in proportion to the outstanding amount over the property value at origination.
CASE — Before the works Meridian’s share of Tranche A is EUR 45m on a value of 161.35: attribution 27.9%, financed emissions 297.9 tCO2e out of 1,068. After the works, EUR 75m on the origination value: attribution 46.5% of 142.5 tCO2e, that is 66.2 tCO2e, a fall of 78%. Had the denominator been updated to the completed value, the figure would be 37.9 tCO2e. The convention matters: with a fixed denominator, the bank’s attribution share rises because it lent more, even though the building is worth more.
INTERACTIVE LAB — Financed emissions of the loan before and after retrofit (PCAF)
Attribute the building’s emissions to Meridian with the PCAF factor and see the effect of the property value convention.
| Parameter you can move | Range | Default |
|---|---|---|
| Building emissions before | 100 to 3000 tCO2e | 1068 tCO2e |
| Building emissions after | 0 to 1500 tCO2e | 142.5 tCO2e |
| Meridian outstanding before works | 10 to 150 EUR m | 45 EUR m |
| Meridian outstanding after works | 10 to 150 EUR m | 75 EUR m |
| Property value at origination | 80 to 300 EUR m | 161.35 EUR m |
| Property value on completion | 100 to 450 EUR m | 282.29 EUR m |
| Result at default settings | Value |
|---|---|
| Attribution factor before | 27.9 % |
| Financed emissions before | 297.9 tCO2e |
| Attribution factor after, origination value | 46.5 % |
| Financed emissions after, origination value | 66.2 tCO2e |
| Change in financed emissions | -77.8 % |
| Financed emissions after, completion value | 37.9 tCO2e |
| Intensity before | 6.62 tCO2e per EUR m lent |
| Intensity after | 0.88 tCO2e per EUR m lent |
Try it. Keep emissions unchanged after the works: financed emissions rise by two thirds simply because the loan is larger. Financed emissions reward decarbonisation, not lending.
| Risk | Control | Owner |
|---|---|---|
| Cost overrun, delay | Fixed-price contract, contingency, cost-to-complete test, completion guarantee | Technical adviser, agent |
| Letting risk at delivery | Pre-let condition, conservative rent, interest reserve | Structurer, credit risk |
| Valuation too optimistic | Independent panel valuer, internal review, rotation | Valuation unit |
| EPC target missed | Design confirmed before signing, long-stop date, declassification clause | Sustainable finance team |
| Regulatory stranding | EPC and pathway analysis in every file; portfolio heat map by maturity | Credit risk |
| Flood, heat | Hazard screening, adaptation in works, insurance covenants | E&S risk, agent |
| Refinancing at maturity | Debt yield floor, amortisation, exit stress test | Credit committee |
| Rate risk | Compulsory hedge executed at signing | Markets desk |
| Mislabelling | Eligibility opinion, annual green report, allocation register | Sustainable finance team |
| Data gaps | EPC and consumption undertakings in the agreement | Agent, data office |
Red flags.
The Paris office market in 2026. According to a broker’s press release of 6 October 2026, take-up in Île-de-France reached 1.05 million m² over the first nine months of 2026, 14% less than a year earlier; immediate supply stood at 6.6 million m² and the vacancy rate at 11.8% for the region, against 9% in Paris. Another broker’s mid-year publication put the prime office yield of the Paris central business district at 4.35% in June 2026 and French commercial property investment at EUR 6.8bn in the first half, of which about EUR 2bn in offices. A market with a tenth of its stock vacant is one where tenants choose, and where the gap between good and obsolete buildings is a letting question before it is a valuation theory.
Evidence on the green premium. The founding academic study, Eichholtz, Kok and Quigley (2010), found on United States offices that certified buildings commanded rents a few per cent higher and sale prices markedly higher than comparable buildings nearby. Later studies in Europe find premia of the same sign and smaller or less stable size. Three cautions for a credit file: certified buildings are newer and better located, and controls are imperfect; results depend on the cycle; and what a lender needs is the discount on the worst buildings, which transaction data capture poorly because such buildings trade rarely. This module therefore treats the brown discount as a scenario parameter (section 5.3), not as an established number.
Regulatory calendar. EPBD recast: transposition deadline passed on 29 May 2026, infringement procedures opened against all Member States in July 2026. Taxonomy: draft revised criteria of 17 March 2026, adoption announced for late 2026 and application intended for 2027. CRR3: the EBA must report by 31 December 2027 on the treatment of exposures secured by commercial property (Article 126(4)), so the IPRE weights may move. Slotting: revised technical standards under consultation in 2026. Pillar 3 ESG: final draft standards of June 2026; in November 2026 the EBA indicated that the simplified supervisory reporting of ESG data would start only with a late-2028 reference date.
Open debates. Should prudential rules favour loans on efficient buildings (a “green supporting factor”)? Supervisors have so far answered that capital must follow risk, and that if efficient buildings are less risky, PD and LGD models should show it; the 2026 slotting consultation moves in that direction. Is the Taxonomy’s top-15% test usable without harmonised EPCs? Should financed emissions include the embodied carbon of financed works? And who pays for adaptation when insurance is mutualised?
INTERACTIVE LAB — Self-assessment: real estate finance
Check the mechanics, the standards and the bank-side economics of the module.
Q1. A building yields NOI of EUR 6m and is valued at a 5% yield. The yield widens to 6%. The value falls by:
Answer: C. Value goes from 120 to 100: minus 16.7%, that is 1 - 5/6.
Q2. Which ratio is independent of both the valuer’s yield and the interest rate?
Answer: A. Debt yield is NOI over debt; it uses neither the valuation nor the cost of debt.
Q3. In a development loan, the main monthly control before a drawdown is:
Answer: D. There is no rent and no stabilised value during works; the lender checks that remaining funds cover remaining costs.
Q4. Under the recast EPBD, non-residential buildings must be brought below the threshold set at the 26% worst-performing of the stock by:
Answer: B. 16% threshold by 2030 and 26% threshold by 2033.
Q5. Under the Taxonomy criteria in force in November 2026, a renovation (activity 7.2) qualifies if it reduces primary energy demand by at least:
Answer: C. 30%, or compliance with national major-renovation requirements. The 60% figure belongs to a 2026 draft for existing buildings.
Q6. A refurbished office with EPC class B, built in 1994, is Taxonomy-aligned under activity 7.7:
Answer: B. For buildings built before 31 December 2020 the test is class A or top 15% by primary energy demand.
Q7. In CRREM’s sense, the stranding year is:
Answer: D. It measures misalignment with a carbon budget, not a legal ban.
Q8. Under the CRR3 standardised approach, an IPRE loan on a commercial property with an exposure-to-value of 53% is weighted:
Answer: A. 70% up to 60%, 90% from 60 to 80%, 110% above.
Q9. Under F-IRB, the regulatory LGD for the part of a loan covered by commercial property collateral after haircut is:
Answer: C. 20% on the secured part after a 40% haircut on the property value; 40% on the unsecured part.
Q10. In the case, the 10 bp green step-down is worth to the borrower over five years about:
Answer: B. EUR 727.5m-years of balance at 10 bp: less than 1% of the works budget.
Q11. PCAF attributes a building’s emissions to a lender in proportion to:
Answer: D. The denominator is the value at origination, which is why a larger loan raises the attribution share.
Q12. Why does a uniform natural-catastrophe insurance surcharge weaken the owner’s incentive to adapt a building?
Answer: A. The owner captures only the avoided deductible, a small share of the avoided loss.
Q13. In the case, the green step-down leaves RAROC above the hurdle only if:
Answer: C. The discount alone costs 1.4 points; a PD of 0.75% brings it back above 12%, and a 10 bp lower FTP recovers most of it.
A logistics building is valued at EUR 80m with NOI of EUR 4.4m. The loan is EUR 48m at an all-in rate of 4.00%, amortising 2% a year. Compute LTV, ICR, DSCR and debt yield. The yield then widens by 100 bp: what is the new LTV?
Solution. LTV = 48 / 80 = 60.0%. Interest = 48 x 4.00% = 1.92; ICR = 4.4 / 1.92 = 2.29x. Amortisation = 0.96; DSCR = 4.4 / 2.88 = 1.53x. Debt yield = 4.4 / 48 = 9.17%. The initial yield is 4.4 / 80 = 5.50%; at 6.50% the value is 4.4 / 0.065 = EUR 67.69m, a fall of 15.4%, and LTV is 48 / 67.69 = 70.9%. ICR and debt yield are unchanged: only the value-based ratio moves.
A hotel emits 50 kgCO2e per m². The pathway starts at 80 in 2020 and falls 6% a year. A retrofit would bring emissions to 15 and primary energy demand from 300 to 190 kWh per m². Find both stranding years and test activity 7.2.
Solution. . After retrofit: . The retrofit buys 19.5 years. Primary energy demand falls by 110 / 300 = 36.7%, above 30%: the renovation meets the substantial-contribution criterion; alignment still requires the do-no-significant-harm tests.
Meridian holds EUR 40m of a stabilised six-year loan: margin 170 bp, FTP 36 bp, amortised fee EUR 0.06m a year, PD 0.75%, LGD 20%, F-IRB risk weight 36.79%. Compute RAROC, the RAROC under the standardised weight of 70%, and the margin that reaches 12% under F-IRB.
Solution. Revenues = 40 x 1.34% + 0.06 = 0.596. Costs = 40% = 0.2384. Expected loss = 0.75% x 20% x 40 = 0.06. Pre-tax = 0.2976; after tax 0.2232. RWA = 14.716; capital = 1.913. RAROC = 0.2232 / 1.913 = 11.67%. Standardised: RWA 28, capital 3.64, RAROC 6.13%. Break-even: the pre-tax result must be 12% x 1.913 / 0.75 = 0.3061, so revenues x 0.6 = 0.3661 and revenues = 0.6102; the margin income is 0.5502, that is 137.5 bp net on 40, or 173.5 bp gross of FTP.
1. Why do real estate lenders look at debt yield as well as LTV? LTV depends on a valuation, which depends on the market yield of the day; in a boom, tight yields inflate values and make any loan look safe. Debt yield compares income with debt and ignores the valuation and the interest rate. It tells the lender what return it would earn on its loan if it owned the building. Used together, the two ratios catch different errors.
2. How does energy performance enter a credit analysis? Through three inputs: the rent the building can achieve and how long it stays vacant, the yield at which it will be valued at maturity, and the capex needed to stay lettable. I would map the EPC against the regulatory dates falling within the loan’s life, 2030 and 2033 under the EPBD, and run an exit stress with a wider yield and lower rent. If the building needs works, the question becomes whether the business plan and the sponsor can deliver them.
3. What is a stranding year, and what are its limits? It is the year in which a building’s carbon or energy intensity crosses above a science-based pathway such as CRREM’s. It is a useful common yardstick and a good early-warning signal. Its limits: it measures alignment with a budget, not legality or lettability; it depends on grid assumptions; and in low-carbon grids the energy pathway matters more than the carbon one.
4. Is a retrofit loan riskier than refinancing the building as it is? During the works, yes: there is construction and letting risk and no income. Over the life of the loan, often no: the unrenovated building faces falling rent, a wider yield and a refinancing gap at maturity. In the case, the EUR 150m retrofit loan ends at a lower stressed LTV than the EUR 90m do-nothing loan. The comparison must be explicit in the credit paper.
5. What makes a real estate loan “green”, and what is the weak point? Under the Green Loan Principles, an identified use of proceeds meeting eligibility criteria, a selection process, tracking of proceeds and reporting. In real estate the criteria usually refer to an EPC level, a certification or the Taxonomy thresholds. The weak point is the difference between green works and a green asset: a renovation can qualify while the finished building does not meet the ownership criterion. A second weak point is projected performance that is never checked against actual consumption.
6. How is an IPRE loan treated for capital under CRR3? Under the standardised approach, by exposure-to-value: 70%, 90% or 110%, and 150% for development exposures. Under IRB, either by slotting into four supervisory categories with fixed weights, or with the bank’s own PD and a regulatory LGD of 20% on the collateralised part. The resulting capital can differ by a factor of two for the same loan, which shapes who can lend at what margin.
7. The sponsor asks for a 15 bp green discount. What do you answer? I would first show what it costs: on our hold, each 10 bp is about 1.4 points of RAROC. Then I would look for what pays for it: a better rating once the EPC is delivered, cheaper funding through the green cover pool, ancillary business. If none exists, the discount is a commercial gesture and should be sized as such. I would also remind the sponsor that the discount is worth about 1% of the works budget; the structure and certainty of funds matter more.
8. A building is in a flood zone and insured. Why still worry? Insurance is annual and the loan is not; cover for lost rent is limited in time; deductibles remain. A long closure can breach the interest cover even if a claim is paid later. And the market may discount the value of an exposed building whatever the policy says. I would ask where the technical rooms are and require adaptation works and extended business interruption cover.
Regulation
Supervisors and standard setters
Market data
Academic and reference