Photo Credit: www.ccr.fr

On Monday, the European Central Bank showed us Eurozone buyers marking down high-risk offices in the sale price itself. Today that repricing arrives as a bill. France charges every property policy a statutory natural catastrophe (CatNat) surcharge, and that surcharge went up. The question for an underwriter isn’t what it costs. It’s what the increase tells you.

Market Signal

Every French property damage policy, household and business alike, carries a compulsory CatNat surcharge based on article A125-2 of the insurance code.

The French Parliament officially established the compulsory CatNat surcharge (or national natural disaster compensation scheme) on July 13, 1982, following disastrous national floods the previous year.

On January 1, 2025, the rate for that mandatory coverage rose from 12 to 20 percent of the property damage premium, applying to all contracts concluded or renewed from that date. This added €1.2 billion per year of coverage capacity.

One rate exists for all perils, including droughts, floods, earthquakes, cyclones, and land movement. France has no subsidence-only surcharge. Because the same 20 percent lands on every policy in the country, it tells you nothing about your building.

But instead of floods, drought is now the primary driver of the bill. CCR, the state-backed reinsurer behind the regime, reports in its Bilan (balance sheets) for 1982 to 2025 that drought accounts for 41.2 percent of cumulative non-motor CatNat losses, second only to flood at 50.4 percent.

Banque des Territoires, reporting on CCR in June 2026, puts drought at 53.1 percent of losses over the last ten years, and records 2022 as the costliest year, close to €3.9 billion of claims, with more than 95 percent of it drought-related.

For 2025, CCR estimates insured drought damage of €770 million to €1 billion. Of €3.71 billion of Cat Nat premiums that year, €1.54 billion came from businesses, representing 44.3 percent once motor is stripped out.

The first strictly uniform, legally binding nationwide RGA (the French-language version of the clay shrink-swell abbreviation) exposure map was created on July 22, 2020. Almost exactly one year after the 20% surcharge increase took effect, the first major overhaul to that clay shrink-swell exposure map since 2020 was imposed on January 9, 2026. It took effect on July 1, 2026.

If you aren’t familiar with the soil type, clay-rich soil shrinks when it dries out. Conversely, when it gets wet, it swells. Think back to pottery class - or Silly Putty that you failed to put away. This weather-driven soil movement is a leading cause of structural damage to buildings in France.

Medium to strong exposure now covers 55 percent of mainland France, up from 48 percent on the 2020 map, based on 446,000 claims from 1989 to 2022. That’s 12.1 million houses in the exposed zones. Centre-Val de Loire went to 90 percent of its territory, up from 82 percent. Remember that data point because we’re about to buy a warehouse there.

Deal Scenario

Let’s look at a modeled scenario to show how you might run your own numbers on a deal.

A 40,000-square-meter single-let logistics warehouse near Orléans, in Centre-Val de Loire. Rent is €50 per square meter per year, anchored on CBRE, which puts new warehouses in the Orléans, Blois and Tours market at €50 to €56 for the fourth quarter of 2025.

BNP Paribas Real Estate’s first-quarter 2026 chart reads €40 prime for Orléans. Assume landlord non-recoverable costs are €100,000. That yields rental income of €2,000,000 and NOI of €1,900,000.

The exit yield in our modeled scenario is 5.50 percent, within the field bounds and not solved for. At 5.50 percent, the value is €34,545,455. Debt is €19 million, interest-only at 4.5 percent, so €855,000 of debt service per year. In our modeled scenario, that’s a 55.0 percent loan-to-value and a debt service coverage ratio of 2.22 times.

JLL puts prime French warehouse yields at 4.90 percent in the second quarter of 2026. Cushman and Wakefield says leased secondary assets trade above 6 percent. A regional box is neither.

The first shock, then, is the premium.

The base property damage premium is assumed at €60,000, since nobody publishes a rate for large warehouses. At the old 12 percent, the surcharge was €7,200. At 20 percent, it’s €12,000. So the statutory step is €4,800 per year, which, capitalized at 5.50 percent, is €87,273 of value, or 0.25 percent of the asset. Noise.

Shock two is the renewal.

Lift the base premium 25 percent after the asset lands in a strong zone, and you pay €90,000 against €67,200 before, up €22,800 per year. Almost all of it is market, not statute.

Shock three is the repair nobody else pays for.

Assume a slab and foundation repair of €1.5 million at a 2 percent annual probability. That 2 percent is a deliberately harsh stress. If a drought recognition order is issued, CatNat pays, but the owner keeps the professional deductible, 10 percent of the damage with a floor of €3,050. In this model, that’s €150,000.

With no order, the owner pays the whole €1.5 million. At a 60 percent chance of recognition, the expected cost is €13,800 per year. At a 0.5 percent probability, the worst-case scenario is 5.67 percent, not 6.19 percent.

The landlord covers that repair cost. Structural work under Article 606 of the Civil Code, such as frame, roof, load-bearing walls, and weatherproofing, can’t be recharged to the tenant, whatever the lease says.

Our model shows three possible outcomes for a warehouse that is remapped into a high-risk climate zone:

Case A. Insurance is recharged to the tenant and the landlord holds the repair reserve. Net operating income is €1,886,200, and value is €34,294,545, down 0.73 percent.

Case B. Insurance is not recharged because the lease is silent or the box is empty. NOI falls to €1,863,400 because the landlord can’t pass a €22,800 annual insurance premium increase on to the tenant. Consequently, the property’s asset value drops by 1.93 percent to €33,880,000 when capitalized at the standard 5.50 percent rate, erasing €665,455 in equity compared to the baseline scenario.

Case C: Case B plus a 25 basis point widening of the exit yield, because buyers have seen the same map. At 5.75 percent, it’s €32,406,957, down €2,138,498, or 6.19 percent, with coverage at 2.18 times and loan-to-value at 58.6 percent.

Summarizing, the statutory surcharge moved value by 0.25 percent. The yield widening moved it by 4.26 percent on its own, which is €1,473,043.

For an insurance bill to do that much damage, it would have to rise €81,017 per year.

The premium is not the cost.

The premium is the evidence you use to set the yield.

And the input that hurts most isn’t clay. Run it at BNP’s €40 of rent instead of CBRE’s €50, and Case C value is €25,450,435, coverage 1.71 times, loan-to-value 74.7 percent.

That’s the covenant conversation.

Underwriting Analysis

So here are five tests to run before you price the exit.

One. Recognition history. CatNat pays for drought damage only after an interministerial order recognizes the municipality, and since January 1, 2023, a professional has 30 days from publication to declare.

Over a long enough window, it is near universal. The brokers’ body PLANETE CSCA reports that 99 percent of local administrative districts, known in France as communes, have been recognized at least once since 1982, for some peril. So the question is how often, and how fast?

The answer? Slow!

For the 2025 drought, CCR counted 345 communes recognized as of May 26, 2026, with 22 percent of requests approved. For 2022, once the campaign had run its course, 5,010 communes were recognized, with 739 refused, representing about 87 percent.

Two. Map zone. Pull the parcel against the 2026 clay map, and be precise about what that map legally does. It binds promises of sale on buildable, undeveloped land and individual house-construction contracts. For a warehouse, it’s free public information. Centre-Val de Loire moved to 90 percent from 82 percent, Bourgogne-Franche-Comté to 74 percent from 58 percent. The state re-rated an asset that changed zone, whether or not its premium has caught up.

Three. Separate the statute from the market. The CatNat layer is fixed by law. The market layer is falling. Marsh reports global commercial rates down 6 percent in the second quarter of 2026, property down 12 percent, and Europe down 9 percent. The French brokers’ body, reporting AMRAE’s 2026 study, says cuts can reach up to 20 percent depending on the account. So if your base premium or deductible rises at renewal while the market is cutting, that is an underwriter telling you something specific about your building.

Four. Read the lease for who pays what. Insurance premiums pass to the tenant only where the lease expressly provides for it. Article 606 structural work never passes through.

Be aware that an anti-negligence penalty is designed to pressure local town halls into taking climate risks seriously. If a town repeatedly suffers natural disasters but its local government refuses to implement an official Risk Prevention Plan (Plan de Prévention des Risques or PPR), the state forces property owners to pay exponentially higher out-of-pocket deductibles for every subsequent disaster claim.

In a commune with no risk prevention plan, that 10 percent deductible doubles at the third order recognizing the catastrophe, triples at the fourth, quadruples after that. In our modeled scenario, 20 percent moves the Case C loss to 6.28 percent and 40 percent takes it to 6.46 percent. While that seems small, it wires Test One straight into cost.

Five. Watch the regime’s own balance sheet. The rate can move again. CCR says 2025 let it begin rebuilding reserves, aiming to withstand a one-in-forty-year event by 2035. Its own 2050 study projects geotechnical drought’s average annual cost rising 83 percent from the hazard alone, €606 million per year, and 103 percent with exposure growth. Those are CCR’s modeled numbers.

The Barusseau bill on adaptation passed the National Assembly at first reading on April 8, 2026. It is PROPOSED, not law, with Senate debate set for October 19, 2026. It would allow, by decree, premium modulation for professional property worth more than €20 million in risk prevention plan zones.

Our modeled warehouse is worth more than that.

Strategic Implications

On September 7, 2025, the reinsurance intermediary Howden Re partnered with the climate modeling firm Reask to combine climate data and subsidence modeling with local meteorological institute data for the United Kingdom and France. Jamie Rodney is Reask’s chief executive, and Tim Edwards is Howden Re’s head of international analytics. Reask’s co-founders, Thomas Loridan and Nicolas Bruneau, presented the work in Paris in December 2025.

The method is the part an investor can copy. They use the Soil Water Index from ERA5, a public monthly reanalysis dataset, which lets them project a drought year as early as October. Their preliminary read said about 11,500 municipalities could meet the drought criteria for 2025, and that recognitions would have run about 32 percent lower had the year followed the 2018 spatial pattern.

Those are their own preliminary modeled numbers. The data is free, monthly, and months ahead of the paperwork. The state hands you the rest for nothing. CCR publishes claims by peril and premiums by payer. Géorisques publishes the clay map.

Meanwhile, Descartes Underwriting, a Paris parametric insurer founded in 2019, writes drought but not subsidence. A parametric trigger on a soil moisture index could fund the deductible and the no-recognition case, independently of any order.

Know, however, that the costliest insured event, $1.3B insured, across Europe, the Middle East and Africa in 2025 wasn’t drought at all. It was a convective storm outbreak on June 25 and 26. French insurers carried that, and rates still fell.

Takeaway

Read the premium before you price the asset.

The 20 percent surcharge is not the risk. It’s the state telling you the drought bill is structural.

The risk sits in the deductible, in the recognition gap, and in article 606. And that risk shows up in the exit yield.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Deal Stress Test™ built for this brief lets you run your own French box through the surcharge step, the renewal stress, the recognized and unrecognized repair cases, and the exit-yield widening, with every input flagged so you can replace it with your own numbers. Free, no signup: Brief 41_CRDF Deal Stress Test™ (xlsx)

New to the framework? The blank master CRDF Signal Tracker™ and Deal Stress Test™ workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S6 Chronic Climate Stress):

• Brief 36 · Land Subsidence & Real Estate: Understanding the Property Appraisal Gap

• Brief 34 · Real Estate Climate Risk: Acute vs. Chronic Threats

Next in sequence:

• Brief 42 · Corpus Christi Drought Downgrades: When Water Moves a City’s Credit Rating

Sources

Every figure above includes the data coverage date, the publication date, and the date I verified it.

The surcharge step — the CatNat rate on property damage contracts raised from 12 to 20 percent, and on motor theft and fire premiums from 6 to 9 percent, from January 1, 2025, adding about €1.2 billion per year of capacity. Article A125-2 of the insurance code (Légifrance) carries the rates.

Ministère de l’Économie et des Finances, arrêtés renforçant le régime des catastrophes naturelles · Data as of applies from Jan 1, 2025 · Published Dec 28, 2023 · Accessed Sep 2026

One rate for all perils. It is the same on every policy and says nothing about any single building.

The regime’s origin — Loi n° 82-600 of July 13, 1982, created the regime after the winter 1981 floods in the Saône, Rhône and Garonne valleys.

Légifrance, Loi n° 82-600; Cohignac & Bidan, Variances (ENSAE Alumni) · Data as of 1981 to 1982 · Published Jul 13, 1982; Variances Nov 30, 2017 · Accessed Sep 2026

The link between the floods and the law comes from secondary histories, not the statute. Drought and subsidence entered the regime in 1989.

The motor line — every property damage contract, for individuals and companies, includes mandatory natural disaster cover, motor vehicles included.

Charpentier, James & Ali, arXiv:2107.07668 · Data as of 2021 · Published Jul 16, 2021 · Accessed Sep 2026

A preprint; used for the scope of the regime, not for loss figures.

Drought’s share of the bill — drought 41.2 percent and flood 50.4 percent of cumulative non-motor CatNat losses 1982 to 2025; 2025 insured drought damage €770 million to €1 billion; 2025 premiums €3.71 billion, €1.54 billion from businesses; 345 communes recognized for the 2025 drought as of May 26, 2026, 22 percent of requests; reserves aimed at a one-in-forty-year event by 2035.

CCR, Les catastrophes naturelles en France, Bilan 1982-2025 · Data as of 1982 to 2025 · Published Jun 2026 · Accessed Sep 2026

The 2025 drought figures are estimates; the recognition campaign was still open.

Over the last ten years, drought accounted for 53.1 percent of losses; 2022 was the costliest year at close to €3.9 billion, with more than 95 percent of losses due to drought.

Banque des Territoires (Anne Lenormand), reporting CCR · Data as of 2016 to 2025 · Published Jun 25, 2026 · Accessed Sep 2026

Secondary report of CCR data.

The clay map — the 2020 map was the first legally binding nationwide shrink-swell map; the 2026 revision, ordered January 9, 2026 and applying from July 1, 2026, puts 55 percent of mainland France in medium to strong exposure against 48 percent, built on 446,000 claims, covering 12.1 million houses; Centre-Val de Loire 90 percent from 82, Bourgogne-Franche-Comté 74 from 58.

Géorisques (BRGM); Banque des Territoires · Data as of 2026 map · Published Feb 2, 2026 (Banque des Territoires) · Accessed Sep 2026

The map binds land sales and house construction contracts, not commercial property.

The deal scenario — 40,000 square meters near Orléans at €50 per square meter per year, €100,000 non-recoverable costs, NOI €1,900,000, 5.50 percent exit yield, value €34,545,455, €19 million interest-only at 4.5 percent; €60,000 base premium; +25 percent renewal; €1.5 million repair at 2 percent with 60 percent recognition; Cases A, B and C, and Case C at €40 rent.

CRREI modeled scenario · Method: three shocks applied in sequence to one asset’s net operating income and exit yield, holding rent, debt, and every other line constant · Modeled — not a specific asset

The base premium is an assumption because no rate for large warehouses is published. The 2 percent repair probability is a deliberately harsh stress, not an observed frequency.

The rent and yield anchors — new warehouse rents of €50 to €56 in Orléans, Blois and Tours; Orléans prime rent of €40; prime French warehouse yields of 4.90 percent in Q2 2026; leased secondary assets above 6 percent.

CBRE; BNP Paribas Real Estate, REview Logistique T1 2026; JLL; Cushman & Wakefield, MarketBeat Logistique France S1 2026 · Data as of Q4 2025 to Q2 2026 · Published Feb 3, 2026 (CBRE) · Accessed Sep 2026

Broker figures. The €40 BNP reading was taken from a chart.

Deductibles and declaration — professional drought deductible of 10 percent with a floor of €3,050, doubled, tripled, and quadrupled at the third, fourth, and later orders where no risk prevention plan exists; 30 days to declare.

France Assureurs · Data as of 2024 · Published Jan 23, 2024 · Accessed Sep 2026

Industry body’s summary of the rules.

Who pays — building insurance premiums pass to a commercial tenant only where the lease expressly provides; Article 606 structural work cannot be recharged.

CG Legal (Grégory Calas) · Data as of leases from Nov 5, 2014 · Published Apr 29, 2026 · Accessed Sep 2026

A law firm’s summary, not legal advice for any specific lease.

Recognition rates — 99 percent of communes recognized at least once since 1982, for some peril.

PLANETE CSCA (Timothée Decaudin) · Data as of 1982 to 2025 · Published Jun 29, 2026 · Accessed Sep 2026

The 99 percent covers any peril, not drought alone.

The 2022 campaign — 5,010 communes recognized for the 2022 drought and 739 refused, about 87 percent.

La Gazette des communes (Lamine Ighil Ameur) · Data as of 2022 campaign · Published Sep 8, 2023 · Accessed Sep 2026

Counts at the close of the 2022 campaign.

The market layer — global commercial insurance rates down 6 percent in Q2 2026, property down 12 percent, Europe down 9 percent; AMRAE 2026 cuts of up to 20 percent.

Marsh, Global Insurance Market Index Q2 2026; PLANETE CSCA, reporting AMRAE · Data as of Q2 2026 · Published Jul 23, 2026 · Accessed Sep 2026

Marsh’s index covers its own placements across all perils.

The regime’s future cost — geotechnical drought average annual cost projected up 83 percent (€606 million per year) from hazard alone and 103 percent with exposure growth by 2050.

CCR, étude impact du changement climatique sur le coût des catastrophes naturelles à horizon 2050 · Data as of 2050 projection · Published Oct 17, 2023 · Accessed Sep 2026

CCR’s own modeled figures.

The flat rate in question — the Barusseau bill would allow, by decree, premium modulation for professional property worth more than €20 million in risk prevention plan zones; passed the National Assembly at first reading on April 8, 2026.

Sénat, dossier législatif ppl25-530 · Data as of 2026 · Published Sep 21, 2026 (page updated) · Accessed Sep 2026

Proposed, not law. Senate debate set for October 19, 2026.

The early read — Howden Re and Reask partnership for UK and France subsidence modeling; about 11,500 municipalities could meet the 2025 criteria; about 32 percent fewer recognitions on the 2018 pattern.

Howden Re · Data as of 2025 season · Published Dec 11, 2025 · Accessed Sep 2026

Preliminary modeled figures from the firms’ own webinar.

The other 2025 bill — the June 25-26, 2025 convective storm outbreak, about $1.3 billion insured, primarily in France, the costliest EMEA event of the year.

Aon, 2026 Climate and Catastrophe Insight · Data as of 2025 · Published Jan 2026 · Accessed Sep 2026

Aon’s own loss estimate.

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you are pricing a French asset that the 2026 map has just re-rated, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

On Wednesday we put a Vancouver rental building through a lender’s flood test, and watched one insurance renewal shrink a loan. Today the water problem runs the other way. In Johor, in southern Malaysia, just across the causeway from Singapore, there isn’t enough of it. And the state has started screening which data centers get approved.

The Moment

On August 28, 2026, Malaysia’s national water regulator, SPAN, reported on the 49 major dams that feed the treatment plants of Peninsular Malaysia and Labuan. Thirty-seven were at normal levels. Nine were at warning level, six in Johor, and one was critical.

SPAN warned that supply disruptions could worsen if levels kept falling. All the treatment plants are still running, and nobody’s taps are dry, so the warning is data-driven, not panic-driven.

Three weeks later, driven by the data center boom, the Malaysian broker RHB Research valued Ranhill Utilities, the listed parent of Johor’s water operator. RHB counts about 8.5 gigawatts of incoming data center capacity in Johor, on top of 1.1 gigawatts already running, which means water is no longer a utility line item in the operating budget. It’s an approval condition.

For investors, lenders and insurers, that turns a water contract into something that looks like a closing condition.

The Story

The state hasn’t frozen data center approvals. It screens them.

In the November 2025 sitting of the Johor state assembly, housing and local government chairman Mohd Jafni Md Shukor said Johor would no longer approve Tier 1 and Tier 2 data centers. That’s how both The Star and Bloomberg reported it.

In March, 2026, the Menteri Besar, the state’s chief minister, put it on the record through Bernama, Malaysia’s official news agency. The committee will only approve Tier 3 and Tier 4 data center developments that use air-cooling technology.

A word for data center readers. Tier in this context refers to the industry’s reliability grade, Tier 1 to Tier 4, which measures redundancy and uptime. It isn’t a water tier. The state is using the reliability label as a screen, and tying it to water used for cooling.

The state’s own written answer to that same assembly sitting sets out the rest. It’s translated from Malay so that some nuance may be lost.

• The approving committee considers first the applications that don’t use the public supply from Ranhill SAJ.

• It makes alternative water a condition of approval.

• It requires water-use efficiency below 1.8.

• It approves small volumes of water, with 2031 as the year of use.

• And it follows SPAN’s water supply guideline, in force since September 1, 2025, which puts reclaimed water for data center cooling ahead of treated drinking water.

There was also talk of a pause. In November 2025, the South China Morning Post reported that investors had been asked to hold off water-cooled expansion for at least 18 months, until mid-2027. But in February, the federal trade ministry, MITI, told Parliament that as of January 2026, no data center investment in Malaysia had been deferred because of water.

Meanwhile, approvals continue.

In May, the state assembly heard that 5,628 megawatts had been approved across 28 data center projects, and that the twenty centers already running were using about 920 megawatts. That’s definitely not a freeze. But it is a screen, and one that’s getting tighter.

In February, households in Benut, Pontian and Simpang Renggam were moved to scheduled water supply. The state’s works chairman said the cause was drought, and that no data center takes water in those areas. But even a correct denial costs something. Every dry spell now puts data centers in the headlines, and the approving committee answers to the same voters.

Structural Forces

Four forces turn this into an underwriting consideration.

First, water now has its own price.

Since August 2025, Johor charges data centers a dedicated rate of RM5.33 per cubic meter. That’s about 50 percent above the old top commercial rate of RM3.55. At the same time, the top rate for every large commercial user rose by 49 percent to RM5.30. So data centers pay three sen more than everyone else at the new rate.

The category is new. The price is mostly the market repricing of water for everyone.

What does that cost an existing plant? In our modeled scenario, take a 100-megawatt data center that still cools with mains water. Mains water is the municipal drinking supply, piped from the local utility and used to cool high-heat servers. The Straits Times figure for that design is about 4.16 million liters per day. At RM5.33, that’s about RM8.1 million per year for water. That’s a legacy design Johor would no longer approve, so it’s the exposure only an existing owner carries.

Second, the approval gate is a compliance test.

By November 2024, officials reported rejecting almost 30 percent of applications for weak sustainability practices. For land held for data centers, the entitlement now depends on a water test. That feeds straight into land and resale value.

Other places are writing similar tests.

For example, New York’s Executive Order 62, signed in July, pauses incomplete state environmental applications for data centers of 50 megawatts or more. Pennsylvania’s order in August requires operators to report annual water use and peak-day demand, and sets strict conservation requirements for developers who want state support.

Third, headroom is thin where the servers are. By headroom, I mean the gap between what the treatment plants can produce and what the state is already using. In Greater Johor Bahru and Kulai, the state told its assembly in September 2025 that design capacity was 1,274.1 million liters per day and average production had reached 1,129.1 million liters per day. That’s a reserve margin of 11.4 percent.

Statewide, the margin is wider, 14.5 percent in the same state data and 12.8 percent in RHB’s estimate. The zone is what matters for data centers.

Against that margin, the federal minister told Parliament that Johor’s data center water demand is projected to reach 384 million liters per day by 2030. RHB’s ceiling may be about 400, based on the old water-cooled figure scaled across every megawatt.

And fourth, public money is years behind. The state has 38 water projects worth RM15.3 billion planned out to 2050. Johor and the federal government are putting about RM5 billion into three reservoirs and three treatment plants. None of that is running yet.

Until it is, a site’s water contract is the scarce thing.

The Next Chapter

So where’s the capital?

This is where the fintech and finance stories come in.

Robin Khuda founded AirTrunk, an Asia-Pacific data center platform, and still runs it as CEO. Blackstone and the Canada Pension Plan Investment Board, CPP Investments, bought it in December 2024, valuing the company at over A$24 billion.

In August 2025, AirTrunk closed a sustainability-linked refinancing round, backed by more than 60 banks and financiers. It covers assets in Australia, Hong Kong, Malaysia and Singapore.

AirTrunk puts the refinancing at about A$15.76 billion excluding Japan, 100 percent of it ESG-linked, and total ESG-linked debt at over A$17.75 billion including Japan. Khuda’s point was that linking the whole global financing structure to sustainability is what a long-term commitment to scaling responsibly looks like.

But let’s look at what Blackstone and CPP Investments actually bought. Not buildings. Growth. And growth here meets the utility before it meets the capital.

In May, 2026, Blackstone used a sophisticated risk-mitigation strategy, bringing in the International Finance Corporation (IFC - part of the World Bank Group) for a US$175 million debt-and-equity injection into AirTrunk’s Johor campuses. IFC’s disclosures verify that at full capacity, the facilities are projected to consume 910 million liters of water annually. Its action plan requires the project to upgrade its monitoring of natural resource consumption to IFC standards, which cover water. That is a monitoring condition, not a cap. The lender bought visibility.

Now, the part that ties this month together, because here the financing writes the engineering spec. AirTrunk published a Green Financing Framework in April 2025. It names three eligible categories for green money, and one of them is water efficiency. For a new data center in a warm climate like Johor, it sets an annual PUE target of 1.4.

PUE stands for Power Usage Effectiveness and measures a data center’s energy efficiency. The global average is 1.52, per the Uptime Institute. Meanwhile, Germany’s Energy Efficiency Act mandates new builds achieve a PUE of 1.2 or lower.

In February 2025, AirTrunk announced JHB2, its second Johor campus, at a design PUE of 1.25 using liquid cooling. Publicly available data from Johor and AirTrunk vary. It seems the actual PUE ranges somewhere between 1.25 (the target) and 1.4 (the green loan threshold).

Here’s the gap a lender should ask about. AirTrunk’s portfolio water usage effectiveness for FY25 is 0.89 liters per kilowatt-hour, comfortably inside Johor’s 1.8. But the first Johor campus is excluded from that figure, because the report counts only sites with twelve months of operation, and the framework reports water at portfolio level. So there is no published Johor water number yet. The first one arrives in the FY26 report.

One more thing worth your attention. AirTrunk’s own FY25 table sorts water withdrawal by water stress level. Melbourne and Sydney sit in the high band. Johor sits in the medium-high or lower band, grouped with Hong Kong, Osaka, Singapore and Tokyo. That is the operator’s classification, while the state rations supply and screens approvals. Underwrite the asset, and you underwrite that disagreement.

And yet outliers still exist, at least on transparency. In August, STT GDC, the Singapore operator, secured up to US$1.37 billion of green financing for a campus of up to 166 megawatts in Iskandar Puteri, Johor. The only water fact in the announcement is a 41.2 percent improvement in company-wide water-use efficiency since 2020. No water condition is disclosed. That doesn’t mean none exists, because the loan agreement isn’t public. But the state screens every approval on water, and the biggest Johor green loan in 2026 says nothing about it.

Meanwhile insurers are watching the soaring cost of risk transfer. The Swiss Re Institute warns that if cities cut water supply in a drought, a data center may have to shift cooling modes or temporarily shut down. It expects global data center insurance premiums to grow from US$10.6 billion to US$24.2 billion by 2030. In other words, it will more than double within the next four years.

One might read the news and think data centers are a tech play with a PR problem. But from an operator’s side, the considerations are a much bigger balancing act before even considering the end product. Site prep and building maintenance, local and regional policy restrictions, lender targets and reporting transparency, and insurance considerations all factor in, while you also face the possibility of not having enough power to run or water to cool.

A data center stops being valued on megawatts and starts being valued on its water contract. Sites with a signed non-drinking, industrial water supply should trade above sites without one, and lenders should treat that contract the way they treat a grid connection, as a condition of drawdown. The upside is the treatment plants landing on time. The downside is a dry year arriving first.

The Strategic Question

So here’s the question for you. If you own, finance, or insure a Johor data center, or land zoned for one, can you show the committee, the lender, and the next buyer four things?

• A signed source of non-drinking water?

• A water usage effectiveness below 1.8?

• A PUE in alignment with local targets?

• And an operating budget that holds at RM5.33 per cubic meter?

If you can’t, the risk isn’t a flood or a fire. It’s an approval you can’t get, or a campus that can’t expand.

And if you invest outside Malaysia, the question travels. In every market where you underwrite data centers, who decides how much water the site gets, and on what test?

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Go Deeper

This is a Story & Future Thinking brief, so there is no companion workbook. The blank master CRDF Signal Tracker™ and Deal Stress Test™ are free and available at climatereadyre.com/tools.

Related briefs

Same signal (S7 Water Stress & Allocation):

• Brief 19 · The Global Water Ledger: Aquifer Depletion and Where Development Slows

• Brief 28 · When a Market Runs Out of Water: Development Moratoria and What They Signal

Next in sequence:

• Brief 40 · Climate Risk & Office Values: Insights from 24K Euro Deals

Sources

Every figure above includes the data coverage date, the publication date, and the date I verified it.

The dam reading — 49 major dams feeding Peninsular Malaysia and Labuan: 37 at normal level, 9 at warning level (6 in Johor), and 1 critical, with all monitored treatment plants still operating normally.

SPAN, Malaysia’s national water regulator, reported via Bernama · Data as of Aug 28, 2026 · Published Aug 28, 2026 · Accessed Sep 2026

The scope is the 49 major dams that feed treatment plants. It is a regulator’s statement relayed by the national news agency, not a published dam-level dataset.

The incoming load — about 8.5 gigawatts of incoming data center capacity in Johor on top of 1.1 gigawatts already operating, with a statewide water reserve margin estimated at 12.8 percent.

RHB Research, as reported by Business Today · Data as of Sep 2026 · Published Sep 19, 2026 · Accessed Sep 2026

This is a press report of a broker note. The note itself is not public, and it gives no basis for RHB’s water ceiling. Nothing in this brief is built on it.

The tier screen — Johor would no longer approve Tier 1 and Tier 2 data centers, per housing and local government chairman Mohd Jafni Md Shukor in the November 2025 state assembly sitting.

The Star, carrying Bloomberg · Data as of Nov 2025 · Published Nov 26, 2025 · Accessed Sep 2026

The tier wording appears in press reports of the sitting, not in any state document located for this brief.

The chief minister on the record — the committee will only approve Tier 3 and Tier 4 data center developments that use air-cooling technology.

Menteri Besar Onn Hafiz Ghazi, via Bernama and Malay Mail · Data as of Mar 10, 2026 · Published Mar 10, 2026 · Accessed Sep 2026

A statement on the record, not a gazetted instrument.

The written approval conditions — non-Ranhill SAJ applications considered first, alternative water as a condition of approval, a water usage effectiveness below 1.8, small volumes with 2031 as the year of use, and SPAN’s water supply guideline in force since September 1, 2025, prioritizing reclaimed water for cooling.

Dewan Negeri Johor, written answer A140, November 2025 sitting · Data as of Nov 20, 2025 · Published Nov 20, 2025 · Accessed Sep 2026

Translated from Malay by CRREI. The planning guideline that the answer refers to for the 1.8 threshold was not opened, and the answer contains no tier wording.

The pause that was reported, and the federal answer — investors reported as asked to hold off on water-cooled expansion for at least 18 months, until mid-2027.

South China Morning Post · Data as of Nov 18, 2025 · Published Nov 18, 2025 · Accessed Sep 2026

A reported request with no official named, and it is disputed by the trade ministry’s own answer to Parliament.

The federal position — as of January 2026, no data center investment in Malaysia had been deferred, was at risk of deferral, or might move abroad because of water constraints.

MITI written parliamentary reply, via Bernama · Data as of Jan 2026 · Published Feb 10, 2026 · Accessed Sep 2026

A federal answer about deferrals. It does not address whether individual applications were refused at the state level.

Approvals still running — 5,628 megawatts approved across 28 data center projects, with the twenty operating centers using about 920 megawatts.

Lee Ting Han in the Johor state assembly, via Media Digital Johor · Data as of May 2026 · Published May 6, 2026 · Accessed Sep 2026

These are electricity and approved capacity figures, not IT load, so they are not comparable to a stated IT megawatt figure.

The February disruptions and the state’s denial — households in Benut, Pontian and Simpang Renggam moved to scheduled water supply, with the works chairman naming drought as the cause and stating that no data center draws water in those areas.

Works exco Mohamad Fazli, via Malaysia Gazette · Data as of Feb 2026 · Published Feb 17, 2026 · Accessed Sep 2026

Translated from Malay. The state denied it, and no independent audit of the affected plants was found.

The tariff — a dedicated data center rate of RM5.33 per cubic meter effective August 1, 2025, with the commercial rate above 35 cubic meters per month rising 49 percent from RM3.55 to RM5.30.

Dewan Negeri Johor, written answer A49 · Data as of Effective Aug 1, 2025 · Published Nov 19, 2025 · Accessed Sep 2026

Cabinet-approved and gazetted by the federal government. The data center rate sits only three sen above the new commercial top rate, so most of the increase is a repricing for all large users.

The water-cooled design figure and the rejection rate — about 4.16 million liters per day for a 100-megawatt water-cooled data center, and officials reporting by November 2024 that almost 30 percent of applications had been rejected for insufficient sustainable practices.

The Straits Times, via Water & Wastewater Asia · Data as of Nov 2024 to Aug 2025 · Published Aug 28, 2025 · Accessed Sep 2026

The 4.16 million liters per day is a generic figure for water-cooled designs, which Johor no longer approves. The base number of applications behind the 30 percent is not stated.

The water bill for a legacy design—a 100-megawatt data center still cooling with mains water at about 4.16 million liters per day — pays about RM8.1 million per year for water at RM5.33 per cubic meter.

CRREI modeled scenario · Method: the published water-cooled consumption figure for a 100-megawatt design multiplied by the gazetted data center tariff over a full year · Modeled — not a specific asset

This applies only to an existing water-cooled plant. Johor would not approve the design today, so it is not the cost of a new campus.

The headroom in the zone — Greater Johor Bahru and Kulai design capacity of 1,274.1 million liters per day against average production of 1,129.1 million liters per day, a reserve margin of 11.4 percent, with a statewide margin of 14.5 percent.

Mohd Firdaus Abdullah, opinion column relaying September 2025 state assembly data, Malay Mail · Data as of Sep 2025 state assembly data · Published Sep 11, 2026 · Accessed Sep 2026

An opinion column relaying state figures. The underlying state document was not located, so the numbers are attributed as reported.

The demand projection — Johor data center water demand projected to reach 384 million liters per day by 2030.

Deputy Prime Minister Fadillah Yusof, written parliamentary reply, via The Edge Malaysia · Data as of Projection to 2030 · Published Nov 19, 2025 · Accessed Sep 2026

The reply does not state the basis for the projection.

The same test elsewhere, part one — New York Executive Order 62 pauses state environmental applications that are not yet complete for data centers of 50 megawatts or more, citing water use.

State of New York, Executive Order No. 62 · Data as of Jul 14, 2026 · Published Jul 14, 2026 · Accessed Sep 2026

In force until the state completes its final generic environmental impact statement. It pauses incomplete applications, not operating facilities.

The same test elsewhere, part two — Pennsylvania Executive Order 2026-05 requires data center operators to report annual water use and maximum day demand, and sets strict water conservation requirements for developers seeking state support.

Commonwealth of Pennsylvania, Governor’s Office · Data as of Aug 18, 2026 · Published Aug 18, 2026 · Accessed Sep 2026

A reporting and conditionality order tied to state support, not a permitting ban.

The public program — 38 water projects worth RM15.3 billion planned for 2025 to 2050, with about RM5 billion from Johor and the federal government going into three reservoirs and three treatment plants.

Bernama, via The Star · Data as of Jul 2026 · Published Jul 23, 2026 · Accessed Sep 2026

Proposed or underway. None of the named plants or reservoirs is operating yet, and the RM5 billion figure combines state and federal funding.

The acquisition — CPP Investments committed to a 12 percent interest in AirTrunk in a transaction implying an enterprise value of over A$24 billion, completed in December 2024.

CPP Investments · Data as of Sep 4, 2024 · Published Sep 4, 2024 · Accessed Sep 2026

Neither the Blackstone nor the CPP release states Blackstone’s percentage, and the founder invested alongside them, so no published figure exists for Blackstone’s stake.

The refinancing — backed by more than 60 banks and financiers, covering assets in Australia, Hong Kong, Malaysia and Singapore, at about A$15.76 billion excluding Japan, 100 percent ESG-linked, with over A$17.75 billion of ESG-linked debt including Japan.

AirTrunk, FY25 Sustainability Report · Data as of FY25 (to Jun 30, 2025) · Published Oct 23, 2025 · Accessed Sep 2026

The KPIs are named, but no thresholds and no margin steps are published, so the size of the incentive cannot be checked. The figures are company-wide, not Johor.

The lender’s condition is a US$175 million debt-and-equity package for AirTrunk’s Johor campuses, whose IFC disclosures project about 910 million liters of water per year at full capacity and require upgrading monitoring of natural resource consumption to IFC standards.

IFC, ESRS 52336 AirTrunk Johor · Data as of Jul 2026 · Published Jul 13, 2026 · Accessed Sep 2026

The US$175 million amount is not stated on IFC’s own disclosure page and comes from the trade press. The action plan item is a monitoring requirement, not a cap on water use.

The financing that writes the spec — AirTrunk’s Green Financing Framework names water efficiency as one of three eligible green categories and sets an annual PUE target of 1.4 for a new data center in a warm climate.

AirTrunk, Green Financing Framework · Data as of Apr 2025 · Published Apr 2025 · Accessed Sep 2026

The framework states a target, not a covenant, and its water impact reporting is portfolio-level only, so you can't read a single campus from it.

The design number — JHB2, AirTrunk’s second Johor campus, announced at a design PUE of 1.25 using liquid cooling.

AirTrunk, JHB2 announcement · Data as of Feb 12, 2025 · Published Feb 12, 2025 · Accessed Sep 2026

AirTrunk’s current campus page shows a different design PUE for the same site, and the company or any Malaysian authority has not published a reason for the change.

The reporting gap — portfolio water usage effectiveness of 0.89 liters per kilowatt-hour for FY25, with the first Johor campus excluded because the report counts only sites operating for at least twelve months, and Johor placed in the medium-high or lower water stress band alongside Hong Kong, Osaka, Singapore and Tokyo.

AirTrunk, FY25 Sustainability Report, Appendix 1 and Appendix 2 · Data as of FY25 (to Jun 30, 2025) · Published Oct 23, 2025 · Accessed Sep 2026

The water stress classification is the operator’s own, drawn from the WRI Aqueduct atlas, and is not the Johor state government’s view. The efficiency metrics carry limited assurance; the classification does not.

The loan that says nothing about water — up to US$1.37 billion of green financing for a campus of up to 166 megawatts in Iskandar Puteri, Johor, with a company-wide water usage effectiveness improvement of 41.2 percent since 2020 as the only water fact released.

STT GDC release, via Bernama · Data as of Aug 2026 · Published Aug 13, 2026 · Accessed Sep 2026

No water condition is disclosed, which is not the same as none existing. The loan agreement is not public, and the efficiency figure is company-wide, not Johor.

The insurance view — if municipalities reduce water supply because of drought or capacity, a data center may have to shift cooling modes or temporarily shut down, and global data center insurance premiums are expected to grow from US$10.6 billion to US$24.2 billion by 2030.

Swiss Re Institute, Sigma Insights 07/2026 · Data as of 2026, projection to 2030 · Published Mar 27, 2026 · Accessed Sep 2026

The study is US-focused with no Malaysia or Johor analysis, and the premium figure is a scenario, not a priced product.

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you own, finance or insure a data center site whose water supply is not yet contracted, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

Canada’s bank regulator ran flood maps across the country’s lenders and published its findings. Over 250 institutions took part, covering CA$904 billion in mortgage and real estate assets and CA$3 trillion in insured property values, about 40 percent of the mortgages in Canada.

Only 12 percent of deposit-taking institutions require or collect information on flood insurance coverage for the properties they lend against. That is the single most useful sentence in the report for a borrower.

It means the file your lender is missing is one you can hand it, before the credit committee prices the gap as unknown.

Market Signal

In September 2025, OSFI, Canada’s bank regulator, and Quebec’s regulator, the AMF, published the results of a standardized climate scenario exercise. The flood module covered eleven urban regions from Vancouver to Fredericton. Toronto was not one of them.

Two findings matter to a borrower. The first is the collection gap: 12 percent of deposit-taking institutions require or collect flood insurance information on the properties they lend against, in a country where only 40 percent of homeowners buy optional flood cover. The second is concentration. Between 10 and 12 percent of in-scope exposures sit in high-risk flood areas, and OSFI projects commercial exposures to stay more concentrated in those zones than residential ones.

Fire is the secondary hazard. OSFI projects the share of lenders’ exposures in high or very-high wildfire zones rising from 16 percent today to 29 percent by 2050.

The regulator also said what comes next. Future work will focus on whether institutions can measure and adequately price catastrophic and climate risk. OSFI’s climate guideline, B-15, is in force now, and it names the channel this brief is about. Physical risk means higher loan-to-value and loss given default because collateral values decline.

The losses are already moving collateral values. Insured severe-weather damage in Canada reached CA$8.55 billion in 2024, the first year above CA$8 billion. The remnants of Hurricane Debby flooded Quebec for about CA$2.7 billion of that, and flash floods in the Toronto area added about CA$990 million.

This is Signal 5, acute climate hazard, flowing into Signal 2, credit and mortgage markets, through Signal 1, insurance.

Deal Scenario

This is a modeled scenario, which means the numbers are not pulled from an actual deal.

A 60-unit older rental building in Vancouver, financed on CMHC-insured terms. Rent is CA$2,100 per unit per month, assumed at about 89 percent of CMHC’s October 2025 average for a purpose-built two-bedroom in Vancouver, which was CA$2,363. Vacancy is 3.7 percent, CMHC’s Vancouver figure. Operating costs are 38 percent of income, including insurance at 5 percent. That 5 percent is a U.S. benchmark from the Federal Reserve, used because no public Canadian series exists for apartment insurance; Canada’s multifamily commercial market runs on private data, broker reports, and specialized corporate tracking indices.

The loan is sized at CMHC’s minimum coverage of 1.20 on a ten-year term, with 40-year amortization and Canadian semi-annual compounding. The interest rate is 4.58 percent, an indicative ten-year CMHC-insured rate from Colliers, about 0.71 points over the Government of Canada ten-year bond. The cap rate is 4.25 percent, the middle of CBRE’s Vancouver range for older low-rise buildings.

On those inputs, net operating income is about CA$902,755, and the building supports a loan of about CA$13.9 million.

Now the renewal. Canadian commercial property insurance is getting cheaper right now. Marsh reports Canadian property rates down 8 percent in the second quarter of 2026, the ninth quarterly decline in a row, and Applied Systems puts real estate property renewals up just 1.02 percent in the first quarter. So a 50 percent jump is not the market. It is what happens to one building when its insurer identifies it as sitting in a high-risk flood zone at renewal. That is the stress being tested here.

• Insurance up 25 percent. The loan this building supports falls by about CA$280,000.

• Insurance up 50 percent. It falls by about CA$559,000. Net operating income drops 4 percent, and the loan drops 4 percent.

• Insurance up 100 percent. It falls by about CA$1.12 million.

Same building, same rent, same tenants. Then put the ladder next to the rate. If the ten-year rate rises half a point, to 5.08 percent, this building loses about CA$920,000 of loan capacity, more than the 50 percent insurance shock. Climate is one input. The rate is the one that gets you.

Underwriting Analysis

Five tests to run before your lender runs them. They are not all Canadian, and they are still valid data points for evaluating a deal or a portfolio at renewal.

First, know which test binds. The title of this brief names DSCR and LTV, and in this modeled scenario only one of them bites. The coverage ratio sets the loan. The loan sits at about 65 percent of value, well under CMHC’s 85 percent ceiling, before and after the shock. That 85 percent limit would only bind above a cap rate of about 5.53 percent, or if a lender cut the building’s value by more than about 23 percent for flood risk. Under insured terms in Vancouver, rising climate costs cut net income, which drops the deal below its permitted covenant ratio, and that is what costs loan proceeds.

Loan-to-value comes into play only with a large collateral haircut. How large can that get? A Bank for International Settlements working paper modeled a loan secured on commercial property in Mobile, Alabama. With a 3 percent annual chance of a major hurricane, the property lost 16 percent of its value, and loss given default rose from 10 percent to 24.8 percent. Even 16 percent stays under this building’s 23 percent threshold. That is an illustration from a working paper, not observed data, and it is a U.S. hurricane case.

Second, find out whose hazard map the lender uses, because the maps disagree. A study for the U.K.’s Climate Financial Risk Forum, written by Jo Paisley and Maxine Nelson at the GARP Risk Institute, compared 13 physical-risk vendors on the same 100 properties. For a one-in-200-year flood, the correlation between the damage estimates of eight of those vendors ran from 0.2 to 0.9. In one geocoding test, a vendor placed a property 1,507 kilometers from the median of the other vendors’ locations. Four vendors gave clients no measure of uncertainty. Which vendor your lender bought is a material fact about your loan.

Third, put insurance inside the coverage ratio, not in a footnote. The Federal Reserve estimates that in U.S. apartment buildings, a dollar increase in insurance costs cuts owners’ net income by about 72 cents. That is U.S. data, and Canada’s rent rules leave even less room to pass costs through. Your lender sizes the loan on the income that remains after that cost.

Fourth, expect the lender to act on the map, not just the price. Since February 2024, Desjardins, the largest mortgage provider in Quebec, has halted financing for property purchases inside the province’s highest-risk floodplains, even though that represents less than 5 percent of its mortgages, per La Presse. National Bank told La Presse it decides flood-zone loans case by case, including whether the property can be insured, how resilient the construction is, and the borrower’s financial strength.

The terms move, not only the price. From January 2026, European regulators expect banks to consider adjusting loan terms, tenor, and pricing based on a borrower’s ESG and climate criteria. Singapore introduced something similar in 2020, allowing banks to consider environmental covenants and to price the risk. In other words, a lender looking at a hazard-exposed property can charge more, shorten the term, or write conditions into the contract.

Fifth, hand the lender the file it does not have. Only 12 percent of lenders collect flood insurance data on their collateral. National Bank is already asking. Its 2025 sustainability report says it sends a climate questionnaire to corporate and commercial real estate borrowers, asking what percentage of their assets sit in regions vulnerable to extreme weather, and it discusses the answers at least once a year, at origination, review, and renewal. In its own table, residential mortgages, a quarter of its loan book, rate Moderate for physical risk. National Bank also says it is integrating an outside climate data solution, without yet saying whose.

So bring a hazard report, the flood coverage you have in force, and any mitigation you have done. Answer the question before it gets priced as an unknown.

Strategic Implications

Watch where the data is going, because it is arriving on the underwriting side first.

• The insurers moved first. In September 2025, Revau, a Canadian agency writing commercial lines, announced it is building the flood and wildfire models of Geosapiens, a Quebec City climate-risk modeling company led by co-founder and CEO Hachem Agili, into its underwriting platform. Revau promises more accurate underwriting from locally calibrated, building-level data.

• Then the carriers. In December 2025, Co-operators announced a long-term flood partnership with Geosapiens to sharpen risk assessment at the property and portfolio level, citing B-15. Co-operators says it has been an investor since the company’s first round.

• And the regulator. OSFI’s flood maps for the scenario exercise came from Riskthinking.AI in Toronto, founded by Ron Dembo. MSCI now owns First Street, the U.S. modeler.

• The caveat. Revau and Co-operators are insurers, not lenders, and neither release says whether the partnership covers commercial buildings or residential property. They show where pricing is heading, not where it has arrived.

What this means is that the same data now sits on both sides of the table. A borrower who buys the data before applying can dispute a bad score, show the mitigation, and argue for loan proceeds instead of accepting a blanket discount. A borrower who does not will be underwritten on somebody else’s map.

So underwrite yourself first. Before you refinance, run your coverage ratio at your insurance cost plus 25, 50, and 100 percent, and at a rate half a point higher. Get a hazard report and a flood insurance quote into the file. And ask your lender which vendor’s map it uses.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Deal Stress Test™ built for this brief lets you run your own building through the insurance ladder, the rate move, and a collateral reduction, with every input flagged so you can replace it with your own numbers. Free, no signup: Brief 38_CRDF Deal Stress Test^TM (xlsx)

New to the framework? The blank master CRDF Signal Tracker™ and Deal Stress Test™ workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S5 Acute Climate Hazard):

• Brief 34 · Real Estate Climate Risk: Acute vs. Chronic Threats

• Brief 11 · Hurricane Helene Aftermath: Western North Carolina Home Insurance Rates Rose 4.4%

Next in sequence:

• Brief 39 · Johor Data Centers: How Tier Bans and Water Limits Impact Investors

Sources

Every figure above includes the data coverage date, the publication date, and the date I verified it.

The regulator’s flood and wildfire findings — the OSFI and AMF standardized climate scenario exercise, covering over 250 institutions, CA$904 billion in mortgage and real estate assets and CA$3 trillion in insured property values, about 40 percent of Canada’s mortgages; 12 percent of deposit-taking institutions require or collect flood insurance information on collateral; 40 percent of homeowners buy optional flood insurance; 10 to 12 percent of in-scope exposures sit in high-risk flood areas, with commercial more concentrated than residential; wildfire exposure in high or very-high zones rising from 16 percent to 29 percent by 2050.

OSFI, Strengthening Climate Risk Financial Resilience: Insights from the Standardized Climate Scenario Exercise · Data as of SCSE 2024, 2050 conditions · Published Sep 11, 2025 · Accessed Sep 2026

The exercise covered eleven urban regions and excluded Toronto. The wildfire and flood shares are projections under the exercise’s scenarios, not observed portfolio losses.

The supervisory channel — Guideline B-15 names physical risk raising loan-to-value and loss given default through reduced collateral value. In force.

OSFI, Guideline B-15, Climate Risk Management · Data as of in force 2025 to 2026 · Published Mar 7, 2025 · Accessed Sep 2026

The guideline states the transmission channel qualitatively in its annex. It sets no loan-to-value or loss-given-default figure.

The loss backdrop — Canadian insured severe-weather losses of CA$8.55 billion in 2024, the first year above CA$8 billion, including about CA$2.7 billion from the remnants of Hurricane Debby in Quebec and about CA$990 million from Toronto-area flash floods.

Insurance Bureau of Canada, using CatIQ data · Data as of 2024 · Published Jan 13, 2025 · Accessed Sep 2026

Insured losses only. Uninsured and public-infrastructure damage are not included in the total, and the largest single 2024 event was Calgary hail, not flood.

The deal scenario — 60 units in Vancouver at CA$2,100 per unit per month, 3.7 percent vacancy, operating costs 38 percent of income with insurance at 5 percent, CMHC terms of 1.20 coverage, 85 percent maximum loan-to-value and 40-year amortization, a 4.58 percent ten-year rate on Canadian semi-annual compounding and a 4.25 percent cap rate; net operating income about CA$902,755 supporting about CA$13.9 million of loan; insurance shocks of 25, 50 and 100 percent cutting loan capacity by about CA$280,000, CA$559,000 and CA$1.12 million; loan-to-value 65 percent before and after, binding only above a 5.53 percent cap rate or a collateral haircut above 23 percent; a 50 basis point rate rise costing about CA$920,000.

CRREI modeled scenario · Method: a single insurance renewal applied to one building’s coverage ratio, holding rent, vacancy, and every other operating line constant, so the loan effect is attributable to the insurance line alone · Modeled — not a specific asset

The rent is an assumption set at about 89 percent of CMHC’s October 2025 Vancouver two-bedroom average of CA$2,363, and the 5 percent insurance share is a U.S. Federal Reserve benchmark because no public Canadian apartment insurance series exists. The 50 percent shock is not the Canadian market; it is a single-building repricing case.

The financing and valuation inputs — CMHC standard rental terms of 1.20 minimum coverage on a ten-year term, 85 percent maximum loan-to-value and 40-year amortization; CMHC’s Vancouver purpose-built two-bedroom average rent of CA$2,363 per month and 3.7 percent vacancy; CBRE’s Vancouver low-rise class B cap rate range, and an indicative ten-year CMHC-insured rate of about 4.58 percent, about 0.71 points over the Government of Canada ten-year bond.

CMHC, Mortgage Loan Insurance for Standard Rental Housing and Rental Market Report Fall 2025; CBRE, Canadian Cap Rates Q2 2026; Colliers, OnMultifamily Bond Yield Tracker · Data as of Oct 2025 to Sep 2026 · Published Dec 11, 2025 · Accessed Sep 2026

The Colliers rate is a broker indication, not a published series, and the spread over the Government of Canada ten-year is not drawn from an official series.

The insurance market backdrop: Canadian commercial property insurance rates fell 8 percent in the second quarter of 2026, a ninth consecutive quarterly decline, while Canadian real estate property renewal premiums rose 1.02 percent in the first quarter of 2026.

Marsh, Global Insurance Market Index (Canada); Applied Systems, Applied Commercial Index Q1 2026 · Data as of Q1 to Q2 2026 · Published Jul 29, 2026 · Accessed Sep 2026

Both are market averages across portfolios. Neither describes what happens to a single building that an insurer re-rates into a high-risk flood zone at renewal.

The insurance pass-through — a dollar increase in insurance costs cuts U.S. apartment owners’ net income by about 72 cents, with insurance running about 5 percent of revenues.

Hughes and Molloy, FEDS Notes, Federal Reserve Board · Data as of 2000 to 2024 panel · Published Sep 19, 2025 · Accessed Sep 2026

U.S. data on U.S. apartment buildings. It is used here as a benchmark, not as a measurement of Canadian pass-through, and Canadian rent rules differ.

The vendor disagreement — 13 physical-risk vendors compared on the same 100 properties; for a one-in-200-year flood, correlation between eight vendors’ damage estimates ran from 0.2 to 0.9; one vendor placed a property 1,507 kilometers from the median of the others; four vendors supplied no measure of uncertainty.

Paisley and Nelson, GARP Risk Institute for the Climate Financial Risk Forum · Data as of 2025 exercise · Published Oct 2025 · Accessed Sep 2026

A benchmarking exercise on a sample portfolio, not an audit of any vendor’s Canadian flood product.

The collateral haircut case — a modeled loan on commercial property in Mobile, Alabama: a 3 percent annual chance of a major hurricane cut value 16 percent and raised loss given default from 10 percent to 24.8 percent.

Pozdyshev, Lobanov and Ilinsky, BIS Working Paper 1274 · Data as of illustrative, 2025 probabilities · Published Jul 7, 2025 · Accessed Sep 2026

An illustration inside a working paper, not observed data, and a U.S. hurricane case rather than a Canadian flood one.

Lenders acting on the map — Desjardins halted financing for purchases in Quebec’s highest-risk floodplains from February 2024, affecting under 5 percent of its mortgages; National Bank decides flood-zone loans case by case, sends a climate questionnaire to corporate and commercial real estate borrowers, reviews the answers at origination, review and renewal, rates residential mortgages, a quarter of its loan book, Moderate for physical risk, and is integrating an unnamed external climate data solution.

La Presse (Charles Lecavalier); National Bank of Canada, 2025 Sustainability Report · Data as of Feb 2024; fiscal 2025 · Published Feb 21, 2024; report published fiscal 2025 (PDF Mar 9, 2026) · Accessed Sep 2026

The Desjardins policy covers purchase financing in the 0-to-20-year flood zone, except for a 35 percent down payment on a flood-proofed property. National Bank’s questionnaire is disclosed in its own report, and it does not name the vendor.

The supervisory terms channel outside Canada — European banks expected to consider adjusting loan terms, tenor, and pricing on ESG criteria from January 2026, and Singapore banks permitted to consider environmental covenants and to price the risk since 2020.

European Banking Authority, EBA/GL/2025/01; Monetary Authority of Singapore, Guidelines on Environmental Risk Management for Banks · Data as of Jan 11, 2026 · Published Jan 8, 2025 · Accessed Sep 2026

Guidelines, not statute, and neither applies to a Canadian lender. The Singapore guidance dates from December 2020 and is a baseline, included to show the direction of travel.

The data arriving on the underwriting side — Revau integrating Geosapiens’ flood and wildfire models into its underwriting platform; Co-operators signing a long-term flood partnership with Geosapiens and disclosing it has invested since the company’s first round; Riskthinking.AI of Toronto supplying the flood data for the OSFI exercise; MSCI completing its acquisition of First Street.

Revau; Co-operators; Riskthinking.AI; MSCI Inc. · Data as of Sep 2025 to Aug 2026 · Published Dec 4, 2025; MSCI Aug 3, 2026 · Accessed Sep 2026

Revau and Co-operators are insurers, not lenders, and neither release states whether the models are applied to commercial buildings or residential property. No volumes or deal terms were disclosed.

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals if you are refinancing a building that an insurer has just re-rated; book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

In the first half of 2026, storm activity ran about 20% above average and global insured catastrophe losses came in at $42 billion, the lowest first half since 2020.

West of Houston, a regional water authority has raised the fee it charges per thousand gallons from $0.50 in 2004 to $3.95 today because the ground beneath its service area is sinking and the regulator has ordered it to stop pumping.

The first number is what an acute year looks like when the storms miss the insured footprint. The second is what a chronic hazard looks like when it finally reaches an operating statement. Neither produced a claim, and only one is in your model.

Market Signal

Physical climate risk arrives on two clocks.

Acute risk is an event. A storm, a fire, a flood, on a single date, producing a single claim and a single repair bill. It is volatile, it is insured, and because it is insured, it is measured, reported, and priced within a year of happening.

Chronic risk is drift. Sinking ground, rising heat, retreating water. It accrues continuously as a slightly higher operating cost, a slightly earlier capital replacement, a slightly lower valuation, and it never produces the moment that forces a number down. It is rarely insured, so it is rarely measured, and what is not measured is not in the comps.

Start with the acute side, because 2025 and 2026 together show how little a loss total tells you about hazard.

Swiss Re Institute puts 2025 global insured natural catastrophe losses at about $107 billion against economic losses of about $220 billion, with the Los Angeles wildfires at roughly $40 billion and severe convective storms at $51 billion. Secondary perils accounted for 92% of insured losses. Aon’s separate tally runs higher, at $127 billion insured, with severe convective storms at $61 billion, the third-highest annual total on record for that peril.

Then the first half of 2026 arrived below trend. Swiss Re counts $42 billion of insured losses against a long-term trend estimate of $66 billion, with severe convective storm at $28 billion, the lowest since 2021. Aon puts first-half economic losses at $111 billion, 25% below the twenty-first-century average, with the United States accounting for 75% of global insured losses in a half-year whose largest single event was an earthquake sequence in Venezuela.

It carries different weight when you look at the two-year combination. Swiss Re’s own explanation for the quiet half is that storm activity ran about 20% above average and yet relatively few of the highest-impact events affected densely populated and highly insured areas. High frequency, low insured loss. That is not a calm year. It is a year whose losses landed where the insurance was not.

Which is the point about acute risk that most underwriting misses.

The loss total says where the storms went and who was insured. The hazard did not move. Swiss Re attributes more than 80% of the long-term rise in weather-related insured losses since 1970 to exposure growth, meaning more development with higher asset values in harm’s way, rather than to the climate signal itself.

So carriers price acute risk annually, based on a loss history that is itself noisy. Nobody prices chronic risk, and it arrives through a different door.

Case Study

Houston’s western suburbs show what that door looks like.

The Harris-Galveston Subsidence District runs a network of more than 190 GPS stations across the metro. Station P029, in Katy, has recorded about 35 centimeters, or about 14 inches, of subsidence since 2007, and from 2021 to 2025 it was still sinking at 2.64 centimeters a year. The fastest station in the network is P111, near Fulshear in Fort Bend County, at 3.31 centimeters a year. Both are on the western side of the metro, where growth has been fastest. The District’s Regulatory Area One, closer to the coast and converted off groundwater decades ago, shows rates below half a centimeter a year, so quoting the fastest stations as a regional rate overstates it several times over.

The peer-reviewed picture agrees on where the problem sits. Space geodetic measurements from 2015 to 2021 find Dallas, Fort Worth, and Houston have the highest proportion of sinking area among the largest US cities, with over 70% of their land subsiding faster than three millimeters a year.

Houston is the fastest of the group, with 42% of its land area subsiding faster than five millimeters a year and 12% faster than ten. San Antonio, Austin, and Houston together hold more than 82% of the buildings that the study rates at very high risk.

The cause is groundwater withdrawal, and that detail turns a geology story into an operating-cost story, because the regulator’s response to the cause is what reaches the property.

The District regulates by area. Regulatory Area One may draw no more than 10% of its annual water demand from groundwater, and Area Two no more than 20%. Area Three, which is the western growth corridor, has been on a conversion schedule that states no more than 70% groundwater through 2025, 40% from 2025, and 20% from 2035 for permittees on a certified reduction plan.

That schedule is why the West Harris County Regional Water Authority exists. It has no taxing authority. It buys and pipes surface water, estimates its conversion program at more than $1.5 billion, and recovers the cost through a fee on every thousand gallons its member districts pump or receive. In 2004, that fee was $0.50 per thousand gallons of groundwater. It rose in almost every year since, to $1.25 in 2010, $2.05 in 2015, $3.20 in 2020, and $3.95 in 2023, where it has stayed through 2026. Surface water is $4.35. The 30% conversion milestone was met in 2010 and the 60% milestone in 2025. The 80% milestone falls in 2035, and the fee schedule is the funding instrument.

Put that on a building. The following is a modeled illustration, not a transaction. A 300-unit garden apartment community in Katy using 100 gallons per unit per day draws about 11 million gallons a year. At the 2004 fee, the authority charged about $5,500 a year for that volume. At the 2023 fee it is about $43,000 on groundwater, or about $48,000 if the district has converted to surface water. That is roughly $40,000 a year of operating cost that did not exist when the asset was underwritten in 2004, before the property’s own utility rate. At a 6% capitalization rate, it is about $630,000 of value.

No storm produced it. No claim was filed. No catastrophe report lists it. A regulator ordered a region to stop sinking, and the cost of that order arrived on a water bill, in increments of twenty-five cents, for nineteen years.

That is chronic risk, and it is dangerous not because of its size. It is that every increment was small enough to absorb, and no one ever re-derived the line.

Strategic Implications

• Model the two tempos in two columns that never merge. Underwrite acute risk for volatility, using the hazard multiplier, the deductible, the carrier count, and the claims history. Underwrite chronic risk for compounding drift, in the utility, insurance, and capital reserve lines, even when nothing hands you the number.

• A compounding hazard modeled as a flat input is the most common error in a long hold. A pro forma carries a wind or fire multiplier because the last catastrophe-model update priced one. It rarely carries a drift for subsidence, cooling load, or water because those arrive as a trend that has to be modeled actively rather than inherited from last year’s template. It is the single most common reason a five-year hold underwrites cleanly and a ten-year hold does not.

• Follow the regulator, not the geology. Subsidence, water stress, and heat reach an operating statement through a rule: a groundwater conversion schedule, an allocation cut, a cooling standard. Physical data usually precedes the rule by years, and the rule reprices the asset. The Harris-Galveston conversion milestones have been public since the 1990s. The 2035 step is already on the calendar.

• Read the loss total as a map of who was insured, not as a measure of hazard. A below-trend year with above-average storm activity is not evidence that exposure fell. It shows the events missed the insured footprint that year. The next year’s renewal will not be priced off that half.

• Date every dataset in the same paragraph. The geodetic study measures 2015 to 2021, the District’s station rates cover 2021 to 2025, the loss figures are full-year 2025 and first-half 2026, and the fee schedule runs 2004 to 2026. Mixing vintages is how a fast station becomes a regional rate and a quiet half becomes a trend.

Future Signal

Watch the modeling industry split its own product line, because that is where the two clocks are being separated for sale.

Cotality, formerly CoreLogic, now sells a chronic-perils product distinct from its catastrophe models. It notes 84 indices with property-level scores on a 1-to-100 scale across four chronic perils: cold wave, heat wave, extreme precipitation, and drought. Acute hazard - meaning flood, hurricane, earthquake, and severe convective storm - sits on a separate line. A client who has bought one has not bought the other, which is a vendor’s way of saying the two tempos are different risks with different data, and the industry has stopped pretending one model covers both.

The same separation is appearing on the insurer side. Moody’s catastrophe-modeling group published its own paper on heat and water stress in August 2026, sampling roughly 159,000 water-intensive US facilities and finding nearly 49,000 facing high or very high water stress in the coming decades, with Texas and California the most concentrated. Its head of catastrophe modeling described heat and water as perils that “are converging, and doing so quickly.”

What follows from that is a sequence, and it runs the same way for every hazard. A vendor scores a chronic peril. The score reaches a lender’s screen. The screen reaches a term sheet. Flood risk took decades and a federal program to run it. Chronic perils now have the vendor scores. They do not yet have the screen.

There are three things to watch.

Whether a lender writes a chronic-peril score into a credit policy, because the first one to do so sets the template.

Whether the Harris-Galveston 2035 milestone produces a fee step in the western authorities’ schedules, because that is the next chronic-cost increment already on a public calendar.

And whether the second half of 2026 returns acute losses to trend, because if it does, the first half will be remembered as what it was, a lucky map rather than a quiet climate.

Brief 5 ran the acute side through a levered Sun Belt deal and found the covenant broke before the return did. Brief 36 takes the chronic side to its conclusion and asks what an appraisal does with a hazard the comps cannot see.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Signal Tracker™ built for this brief lets you log acute-event signals and chronic-drift signals side by side in two columns that never get modeled as one, and score which of them is actually moving your net operating income. Free, no signup: Brief 34_CRDF Signal Tracker™ (xlsx)

New to the framework?

The blank master CRDF Signal Tracker™ and Deal Stress Test™ workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S4 Valuation & Appraisal Gap):

• Brief 5 · Sun Belt Multifamily Insurance and IRR: Climate Risk Behind a 207% Rise

• Brief 1 · Insurance Premium Hikes: Impact on Cap Rates & Property Value

Next in sequence:

• Brief 35 · Colorado Hail Roof Insurance: The New 0.5% Fee

Sources

Every figure above, with the date the data covers, the date it was published, and the date I verified it.

First-half 2026 insured catastrophe losses, Swiss Re — $42 billion, the lowest first half since 2020, against a long-term trend estimate of $66 billion; severe convective storm $28 billion, the lowest since 2021; storm activity about 20% above average, with relatively few of the highest-impact events affecting densely populated and highly insured areas

Swiss Re Institute, first-half 2026 insured catastrophe losses · Data as of Jan to Jun 2026 · Published Aug 11, 2026 · Accessed Sep 2026

A reinsurer’s preliminary estimate of the market it underwrites. Munich Re, Gallagher Re, and Aon published first-half insured estimates of $44 billion, $46 billion, and $47 billion in the same weeks; the spread is narrow, and the direction is unanimous.

First-half 2026 economic losses and concentration, Aon — global economic losses $111 billion, 25% below the twenty-first-century average; the United States accounted for 75% of global insured losses; Venezuela earthquake sequence of June 24, 2026 at $20 to 30 billion economic; US severe convective storms $34 billion economic

Aon, below-average global catastrophe losses masked significant regional impacts, H1 2026 · Data as of Jan to Jun 2026 · Published Jul 22, 2026 · Accessed Sep 2026

A reinsurance broker’s proprietary catastrophe database. Aon frames it as: headline global loss figures can obscure where risk is actually accumulating.

2025 catastrophe losses, Swiss Re — global insured losses of about $107 billion against economic losses of about $220 billion; Los Angeles wildfires about $40 billion; severe convective storms $51 billion; secondary perils 92% of insured losses; more than 80% of the long-term rise in weather-related insured losses since 1970 attributed to exposure growth

Swiss Re Institute, sigma 01/2026 · Data as of full year 2025 · Published Mar 19, 2026 · Accessed Sep 2026

Swiss Re revised economic losses down from $233 billion in its December preliminary release to $220 billion in the final.

2025 catastrophe losses, Aon — total insured losses of $127 billion; severe convective storm insured losses of $61 billion, the third-highest annual total on record for that peril; the Palisades and Eaton fires at $41 billion

Aon, 2026 Climate and Catastrophe Insight · Data as of full year 2025 · Published Jan 2026 · Accessed Sep 2026

2025 was the third-highest severe convective storm year, below 2023 and 2024, so it is not on its own evidence of an accelerating trend.

Houston-area subsidence measurement — station P029 in Katy has recorded approximately 35 centimeters since 2007 and is sinking at 2.64 centimeters per year over 2021 to 2025; station P111 near Fulshear is the fastest in a network of more than 190 at 3.31 centimeters per year; Regulatory Area One rates below half a centimeter per year

Harris-Galveston Subsidence District, 2025 Annual Groundwater Report · Data as of 2021 to 2025 · Published Jun 2026 · Accessed Sep 2026

The District publishes no network-wide average. The fastest stations do not represent the region.

Harris-Galveston groundwater limits by regulatory area — Area One no more than 10% of annual water demand from groundwater; Area Two no more than 20%; Area Three no more than 70% through 2025, 40% from 2025, and 20% from 2035 for permittees with a certified groundwater reduction plan, with a 20% cap otherwise; a disincentive fee applies above the limit

Harris-Galveston Subsidence District, Regulatory Areas · Data as of 2026 · Published date not stated · Accessed Sep 2026

West Harris County Regional Water Authority fee — groundwater fee per 1,000 gallons of $0.50 in 2004, $1.25 in 2010, $2.05 in 2015, $3.20 in 2020, and $3.95 from 2023 through 2026; surface water $4.35 from 2023; conversion milestones of 30% by 2010 and 60% by 2025 met, 80% by 2035 pending; conversion cost estimated at more than $1.5 billion; the authority has no taxing power

West Harris County Regional Water Authority, Pumpage Fee History and Frequently Asked Questions · Data as of 2004 to 2026 · Published date not stated · Accessed Sep 2026

The fee is charged to the authority’s member utility districts, which pass it through inside their own water rates. The authority kept the fee flat in 2024, 2025, and 2026.

Texas subsidence in the peer-reviewed record — Dallas, Fort Worth, and Houston exhibit the highest proportion of sinking area among the 28 most populous US cities, with over 70% of their land subsiding faster than 3 millimeters per year; Houston: 42% of land area subsiding faster than 5 millimeters per year and 12% faster than 10; San Antonio, Austin, and Houston hold more than 82% of buildings rated very high risk.

Ohenhen and colleagues, Land subsidence risk to infrastructure in US metropolises, Nature Cities 2(6) · Data as of 2015 to 2021 · Published May 8, 2025 · Accessed Sep 2026

Peer-reviewed. The paper does not separate Dallas from Fort Worth; per-city detail is in its Supplementary Table 3. The national headline from the same paper is the subject of Brief 36.

Chronic-peril scoring as a separate product — 84 indices across four chronic perils, cold wave, heat wave, extreme precipitation, and drought, with property-level scores on a 1 to 100 scale; acute hazard sold in a separate catastrophe-modeling line

Cotality, Climate Risk Analytics: Chronic Perils · Data as of 2026 · Published date not stated · Accessed Sep 2026

A vendor product page with no research, figures, or methodology behind it. Cited only for the fact that the product line is split, not for any risk estimate.

Moody’s on heat and water stress — roughly 159,000 water-intensive US facilities sampled, nearly 49,000 facing high or very high water stress in the coming decades, concentrated in Texas and California; more than 60% heavy manufacturing and processing, about 25% food and beverage, about 14% mining and quarrying

Claims Journal, Moody’s says heat, water scarcity driving stranded-asset risks · Data as of 2026 · Published Aug 18, 2026 · Accessed Sep 2026

Trade-press coverage of Moody’s own whitepaper, The compound effect: heat, water, and the new risk landscape (August 2026), which describes the vendor’s own model. The quotation is from Mohsen Rahnama, head of catastrophe modeling at Moody’s Insurance Solutions.

Katy water-fee illustration — 300 units at 100 gallons per unit per day, about 11 million gallons a year; authority fee of about $5,500 at $0.50 and about $43,000 at $3.95 per 1,000 gallons; roughly $40,000 of added annual cost, about $630,000 of value at a 6% capitalization rate

CRREI modeled illustration · Method: the authority’s published fee applied to an assumed consumption volume, before the member district’s own rate and before any conservation; the 100 gallons per unit per day figure is an assumption, not a benchmark · Modeled — not a specific property

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you hold Texas assets on a ten-year horizon and your model treats water, ground, and heat as flat lines, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

Arizona’s own groundwater model projects 4.86 million acre-feet of unmet demand in the Phoenix Active Management Area over one hundred years, roughly 4% of projected groundwater demand.

On June 1, 2023, the state stopped approving new determinations of Assured Water Supply based on groundwater across that area. Three years and two court rulings later, the Department of Water Resources has not changed that position.

The binding constraint on America’s fastest-growing desert metro is not capital, and it is not demand. It is whether you can legally deliver water to the door.

Market Signal

Understand the mechanism before the litigation, because the mechanism is what travels.

Under Arizona’s 1980 Groundwater Management Act, a developer inside one of Arizona’s Active Management Areas, eight of them as of September 2026, cannot simply drill a well and build. They must demonstrate a hundred years of assured supply before they can sell lots. Most states have no equivalent test. That is what makes Arizona a leading indicator rather than a local curiosity.

When the state’s updated model said that assurance no longer held for groundwater-only growth, the Department of Water Resources stopped issuing determinations on that basis. Its published position today is unchanged: “the State will not approve new determinations of Assured Water Supply within the Phoenix AMA based on groundwater supplies.”

Existing Certificates and Designations were untouched, and developments inside them may continue to expand. Municipal providers holding a Designation kept serving new development from their approved portfolios. This was never a building moratorium.

The city of Phoenix is the clearest illustration. It draws roughly 60% of its water from the Salt and Verde rivers through the Salt River Project and about 40% from the Colorado River through the Central Arizona Project. Groundwater is about 2%.

So the freeze did not hit Phoenix. It hit the unincorporated groundwater-dependent edge, principally Buckeye and Queen Creek, where land had been bought, entitled, and financed on the assumption that certificates would issue on schedule.

Put yourself on one of those parcels. Overnight, the most important number in the deal is not the cap rate or the rent comp. It is a regulatory finding about an aquifer. Carrying costs keep running, the lender’s clock keeps ticking, and the exit assumes a buyer who can actually build. A freeze like that doesn’t delay a project. It can strand the basis.

Case Study

What happened next created a market, and then a lawsuit.

In November 2024, the Department finalized a rulemaking creating an Alternative Path to Designation of Assured Water Supply, effective November 25, 2024. The intent was to let providers grow incrementally on non-groundwater supplies while reducing groundwater mining.

Under the amended R12-15-710, the hundred-year volume “shall be further reduced by 25 percent of the 100-year volume of each New Alternative Water Supply.” That is a haircut, not a qualifying threshold. A provider had to bring roughly 133% of the demand to be credited with 100%.

Alongside it, the legislature passed Senate Bill 1611, the Ag-to-Urban program, signed June 30, 2025. It converted retired agricultural water rights into development credits using conservation multipliers of 150 in the Phoenix area and 100 in Pinal, capped credits at sixty times mean annual groundwater use, and will repeal itself on January 1, 2036. Senate Bill 1335, effective September 12, 2026, lets applicants keep irrigating for up to two years after applying.

That is the birth of a tradable input. When groundwater alone will not certify a project, the missing supply has to come from somewhere, and retired agricultural water, treated effluent, and recharge credits each acquire a price. Water stops being a site condition and becomes something you assemble, like a capital stack.

The verified scale of it so far is smaller than the headlines. The first Ag-to-Urban approval, on December 9, 2025, covered 825 homes for Forestar Real Estate Group in Buckeye, saving more than 437 million gallons per year against the land’s historic agricultural use, an 80% reduction.

A figure of 60,000 homes circulates widely. It is EPCOR’s own statement about its alternative-path designation covering Buckeye and Surprise, a regulated utility’s estimate of its service-area build-out capacity.

Then the Home Builders Association of Central Arizona sued, and won twice.

On April 21, 2026, in HBACA v. Arizona Department of Water Resources, Maricopa County Superior Court case CV2025-002623, Judge Scott Blaney vacated the two standards the Department had been applying to certificate review. The holding was procedural. The court found these were rules under the Administrative Procedure Act and that the Department had not promulgated them properly. The court did not find the hydrology wrong.

In early June 2026, the same judge voided the 25% alternative-supply provision, holding that “ADWR has in effect attempted to rewrite the governing statute at the agency level” and that the statute “plainly requires that applicants show that water will be continuously available to satisfy the water needs of the proposed use for at least one hundred years.”

Here is why “development is back on” is the wrong read. The Department’s published position has not moved. It has released a further model revision, a Version 3 update in 2026, that re-bases projected existing pumping on the average of the last five years of pumping, and no appellate docket has appeared. More importantly, the June ruling removed the alternative pathway while leaving the underlying determination standing. Local reporting on the day noted it might leave developers no route at all in water-scarce areas.

Notice what the litigation itself does to value. When the rule can flip on a court schedule, the entitlement carries a legal risk premium on top of the hydrology. No institutional buyer underwrites a parcel whose right to build is still being argued.

Let’s look at Cape Town for a different example. In 2017 and 2018, a metropolitan municipality of roughly 4.2 to 4.4 million people came within about eleven weeks of a projected “Day Zero” before rationing to 25 liters per person per day pushed the date away. The taps never ran dry. What changed permanently was how the city prices growth. Markets reprice fastest when a near miss makes the unthinkable concrete.

Strategic Implications

• The certificate is the asset. On a groundwater-only parcel, the binding question is not price or zoning. It is whether the water can be certified at all, and that question now has a litigation history attached.

• Read the ruling before you read the headline. Two trial-court wins on administrative- procedure grounds do not mean the aquifer is fine, and they are not final. Ask specifically whether the Department has changed its published position. As of today, it has not, but by the time you read this, that may have changed.

• Underwrite the alternative-water cost as a line. Effluent, recharge credits, and retired agricultural rights all have prices now, and the Ag-to-Urban multipliers tell you what the state thinks a conserved acre-foot is worth. Where a project needs them, that cost belongs in the basis, not in a footnote.

• Discount vendor build-out estimates. A utility saying its designation “unlocks the potential to support 60,000 new homes” is describing capacity, not permits. The number of homes actually approved under the legislative program to date is in the hundreds.

• Put the Colorado River tier in the model. Reclamation’s August 2025 determination confirms a Level 1 Shortage governing Lake Mead for calendar 2026, reducing Arizona’s apportionment by 512,000 acre-feet, about 18%. On August 21, 2026, Interior finalized a ten-year decision framework for 2027 and 2028 that cuts Lower Basin deliveries by 1.25 million acre-feet in calendar 2027, with Arizona’s share at 760,000 acre-feet under the Lower Basin states’ proposed sharing agreement.

Future Signal

The shift to watch is water moving from a utility question to a priced, tradable input underwritten beside the dirt.

Today the two are appraised separately. Land is a real estate question, and water is a service question. In a market where a parcel without assured supply may never be buildable, the water right and the dirt converge into one asset with one value. The parcel with a designation simply becomes a different asset from the one without.

The pattern generalizes faster than most investors expect. Texas groundwater conservation districts, California’s Sustainable Groundwater Management Act, and stressed basins across the Mountain West are all moving toward the same place at different speeds. Proof of water supply becomes a condition of development.

The investor who learns to read a water designation the way they already read a flood zone will not be surprised when the next desirable submarket quietly stops issuing permits. The one who assumes a favorable court ruling settled it will be.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Signal Tracker™ built for this brief lets you log water, code, and migration signals in your own markets, translate a designation status into financial impact, and score which parcels in your pipeline carry entitlement risk you have not priced. Free, no signup: Brief 28_CRDF Signal Tracker™ (xlsx)

New to the framework? The blank master CRDF Signal Tracker™ and Deal Stress Test™ workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S7 Water Stress & Allocation):

• Brief 19 · Rio Verde Water Cutoff and Scottsdale: 1,000 Residents on Hauled Water

Next in sequence:

• Brief 29 · Wildfire Retrofit Cost vs Insurance Discount: What $23K to $100K Actually Buys

Sources

Every figure above, with the date the data covers, the date it was published, and the date I verified it.

Unmet groundwater demand — 4.86 million acre-feet over one hundred years in the Phoenix Active Management Area, roughly 4% of projected groundwater demand; new determinations based on groundwater halted from June 1, 2023

Arizona Department of Water Resources, Phoenix AMA Groundwater Supply Updates · Data as of Jun 2023, projected to 2121 · Published 2023 · Accessed Aug 2026

A modeled projection, not an observation. Existing Certificates and Designations were unaffected and may continue to expand, so this is not a building moratorium. Arizona has eight Active Management Areas, being Prescott, Phoenix, Pinal, Tucson, Santa Cruz, Douglas, Willcox, and Ranegras Plain, per the Department of Water Resources Active Management Area overview. The Department released a Version 3 model update in 2026. The AMA count is a moving number: the Willcox AMA was designated by Director order on Dec 19, 2024, effective Jan 8, 2025, and Ranegras Plain by Director order on Jan 9, 2026, so the count at the June 2023 freeze was six.

Phoenix water portfolio — roughly 60% from the Salt and Verde rivers via the Salt River Project, about 40% from the Colorado River via the Central Arizona Project, and about 2% groundwater

City of Phoenix Water Services, Drought and Shortage Operations · Data as of 2026 · Published 2026 · Accessed Aug 2026

The alternative pathway — effective November 25, 2024, crediting only 75% of each new alternative water supply, a 25% reduction that requires roughly 133% coverage

Arizona Department of Water Resources, Notice of Final Rulemaking, R12-15-710 · Data as of Nov 2024 · Published Nov 25, 2024 · Accessed Aug 2026

Subsection (H) of this rule was voided in early June 2026.

Ag-to-Urban — Senate Bill 1611, signed June 30, 2025; conservation multipliers of 150 in the Phoenix AMA and 100 in the Pinal AMA; credits capped at sixty times mean annual groundwater use; program repeals January 1, 2036

Arizona State Legislature, SB1611 Senate Fact Sheet · Data as of 2025 · Published Jun 30, 2025 · Accessed Aug 2026

First Ag-to-Urban approval — 825 homes for Forestar Real Estate Group in Buckeye on December 9, 2025, saving more than 437 million gallons per year, an 80% reduction against historic agricultural use

Office of the Arizona Governor · Data as of Dec 2025 · Published Dec 9, 2025 · Accessed Aug 2026

The 60,000 homes figure — a regulated utility’s estimate of its own alternative-path designation covering Buckeye and Surprise

EPCOR, EPCOR Earns Historic ADAWS Designation · Data as of Oct 2025 · Published Nov 6, 2025 · Accessed Aug 2026

A vendor statement about its own service area, describing capacity rather than approvals, and resting on a rule voided in June 2026.

The litigation — HBACA v. Arizona Department of Water Resources, Maricopa County Superior Court CV2025-002623; the two AMA-wide standards vacated April 21, 2026 on Administrative Procedure Act grounds; the 25% provision voided in early June 2026

Gammage & Burnham, legal alert on the summary judgment and Holland & Hart on the second ruling · Data as of Jun 2026 · Published Apr 22 and Jun 24, 2026 · Accessed Aug 2026

Read from counsel of record and the litigating institute rather than from the signed judgments. Both rulings are trial level. No appellate filing had appeared as of August 2026, and the plaintiff coined the two standard names rather than the agency promulgating them.

Colorado River shortage — a Level 1 Shortage Condition governs Lake Mead operations for calendar 2026, reducing Arizona’s apportionment by 512,000 acre-feet, about 18%

US Bureau of Reclamation · Data as of Aug 2025 · Published Aug 15, 2025 · Accessed Aug 2026

Applies to calendar 2026. Reclamation’s August 21, 2026 release on 2027 and 2028 operations sets Lower Basin reductions at 1.25 million acre-feet for 2027, with Arizona at 760,000 acre-feet under the Lower Basin states’ proposed sharing agreement.

Cape Town — a metropolitan municipality of roughly 4.2 to 4.4 million people came within about eleven weeks of a projected Day Zero of April 12, 2018, announced January 23, 2018; rationing held at 25 liters per person per day

City of Cape Town, Census 2022: Cape Town Trends and Changes and NOAA Climate.gov · Data as of 2011 to 2022 · Published Jun 29, 2018 and October 2023 · Accessed Aug 2026

Population interpolated between the 2011 census at 3,740,025 and the 2022 census at 4,772,846. Day Zero moved away from April 12 and never closed the taps.

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you hold entitled land on a groundwater-only edge and want the designation risk priced before your next capital call, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

On June 1, 2023, Arizona stopped approving assured water supply determinations for new Phoenix-area subdivisions based on groundwater.

The model behind that decision projects 4.86 million acre-feet of unmet demand across the next hundred years.

No drought declaration, no emergency, and no disaster. A state agency ran a model, and a category of land stopped being developable.

Market Signal

Water touches real estate differently from every other climate signal in this publication, and the difference is worth understanding before the numbers.

Insurance reprices annually and reversibly. A carrier that withdraws can return. Water rights do not work that way. An aquifer that cannot support a hundred-year assured supply determination does not recover on a policy cycle, and the legal instrument that governs it is closer to a title condition than to an operating expense.

Arizona’s Assured Water Supply program requires a developer to demonstrate a hundred-year water supply before a subdivision can be platted. The 1980 Groundwater Management Act created that requirement, with the operative Assured Water Supply Rules effective in 1995, and for nearly three decades groundwater satisfied it.

In June 2023, the Arizona Department of Water Resources published a new Phoenix Active Management Area groundwater model and stopped approving new groundwater-based determinations. The projection behind it shows 4.86 million acre-feet of unmet demand by 2121.

This is not a building moratorium. Developments within existing Certificates and Designations of Assured Water Supply may continue to expand. What stopped is the approval of new groundwater-based determinations.

Arizona then created an alternative pathway. A developer could qualify by bringing non-groundwater supply to the project, through the Alternative Designation of Assured Water Supply route. In June 2026, the Maricopa County Superior Court voided the 25% supply-reduction provision at the center of that route, which is where Brief 28 picks up the story.

That combination is what makes this a repricing rather than a prohibition. Land with an existing certificate became more valuable. Land without one became worth the cost to serve it with water, plus the transaction cost to acquire it, minus the delay.

In other words, water moved from a utility-connection question to an acquisition line item in a single day.

The scale of the underlying depletion is not in dispute, which is what makes the administrative response predictable rather than surprising. Satellite gravimetry from NASA’s GRACE and GRACE-FO missions has documented sustained groundwater storage loss across the US Southwest and other major aquifer systems since 2002.

That data has been public for two decades. What changed in 2023 was not the hydrology. A regulator finally attached a permitting consequence to it.

For an investor, what’s perhaps most relevant is the sequence of events. Physical data usually precedes regulatory action by years, and regulatory action is what reprices the asset. Watching the hydrology tells you where to look. Watching the agency tells you when.

Case Study

The clearest illustration is not a subdivision. It is a community that already existed.

Rio Verde Foothills is an unincorporated area northeast of Scottsdale. About 1,000 residents, in roughly 500 to 700 homes, many of them expensive, relied on hauled water delivered from a Scottsdale standpipe. The community itself runs to some 2,000 homes.

On January 1, 2023, Scottsdale stopped supplying it.

The city had signaled the decision for eight years, beginning in 2015, to protect supply for its own residents. Rio Verde Foothills lies outside Scottsdale’s municipal boundary and service obligation. There was no contract to breach.

Hauled water prices nearly tripled for some households as tankers had to drive farther to find a source.

No well ran dry. No aquifer collapsed under those specific homes. What failed was an assumption embedded in the value of every one of those parcels, which was that a neighboring city would keep selling water to people it had no obligation to serve.

The physical resource was never the binding constraint. The legal and political arrangement around it was, and that arrangement was never written down anywhere a buyer would have looked.

Additionally, this was not marginal housing on marginal land. Expensive housing carries the same exposure when the entitlement is missing, and in some ways carries it worse, because the buyer pool for a high-value home with an unreliable water supply is considerably thinner than for a modest one. Buyer scarcity compounds water scarcity.

For anyone underwriting land or residential product in the Southwest, that is the transferable lesson. The question is not whether water exists. (Although…!) It is whether an enforceable right to it survives a policy change at an entity you do not control.

Strategic Implications

Water risk enters a real estate decision through three channels, and only the first is commonly subject to due diligence.

• Physical supply. Whether the resource exists is the question everyone asks, and it is the least decisive, because in most Western markets the resource exists in some quantity at some depth at some cost.

• Legal entitlement. Whether you have a right to it that runs with the land and survives a policy change. This is what Rio Verde lacked and what an Assured Water Supply certificate provides.

• Administrative discretion. Whether an agency can change the rule that governs your entitlement, and how fast. Arizona demonstrated the answer is yes, and in a single announcement.

The practical consequence for underwriting is that water belongs in the diligence stack next to title, not next to utilities.

A certificate is an asset. Land inside an existing Designation of Assured Water Supply carries something land next to it does not, and the two parcels can look identical on a survey. That premium is real, legally documented, and rarely in a comp set.

The reverse also holds. A parcel whose development plan assumes a groundwater determination that can no longer be issued is not discounted. It is a parcel with an unpriced acquisition requirement attached, and the price of that requirement is set by whoever currently holds the alternative supply.

Then there is the question of what happens to the buildings already there. Rio Verde eventually reached an arrangement, but the interim taught the market something durable. A home whose water arrives by truck, at a price set by scarcity, along a route that lengthens as sources dry up, is not the same asset it was when the standpipe was open.

A financing consequence follows directly and is easy to miss. A lender underwriting a land loan on a parcel without an assured supply determination is lending against an entitlement that does not yet exist.

That is a different risk from construction risk or lease-up risk. It is closer to a permitting-risk model, and it is binary. The parcel either obtains an alternative designation or is not platted. In the latter case, the loan has no collateral.

Expect that to show up first as tighter land loan terms in affected submarkets rather than as a change in land prices, because credit reprices faster than comps do. It is also the earliest visible signal that the repricing has begun.

Future Signal

Track these four things -

Whether other states copy the Arizona mechanism. The Assured Water Supply framework is unusual in requiring a hundred-year demonstration at platting. Most Western states regulate water rights without explicitly tying them to subdivision approval. If the model spreads, the repricing spreads with it, and it arrives as a permitting change rather than as a shortage.

Whether water rights start appearing in institutional screens. Brief 13 noted sovereign and pension capital concentrating in water-secure Northern European markets while running data centers and logistics. That allocation pattern implies water is already inside proprietary models. When it appears in a published investment policy, entire Sun Belt submarkets reprice at once.

Whether the alternative supply market develops or seizes. Arizona’s alternative pathway only works if non-groundwater supply is actually purchasable at a price development can bear. If it is, this becomes a cost. If it is not, the alternative is theoretical, and the practical effect is closer to a moratorium.

Whether existing certificates start trading separately from the land. If a Designation of Assured Water Supply is the scarce asset, the market will eventually price it independently. Watch for land transactions in which the water entitlement is valued and disclosed as a separate line item rather than absorbed into the per-acre price. That is the moment water becomes a tradeable input rather than a site characteristic, and it changes how every parcel around it is underwritten.

The through-line is that water is the one climate variable where the legal instrument matters more than the physical resource. Insurance can be repriced. Heat can be engineered against. A parcel without an enforceable right to water cannot be developed at any price, and no amount of capital fixes that.

Brief 20 applies the same administrative-risk logic to an Australian office asset, where a rating threshold, rather than a water model, determines who can lease the building.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Signal Tracker™ built for this brief lets you log water entitlement, allocation, and permitting signals in your markets, translate them into financial impact, and score which ones are moving your pricing. Free, no signup: Brief 19 CRDF Signal Tracker™ (xlsx)

New to the framework? The blank master Signal Tracker and Deal Stress Test workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S7 Water Stress & Allocation):

• Brief 13 · GRESB Participation and Real Estate Returns: What $9 Trillion in Capital Screens For

• Brief 28 · Arizona Assured Water Supply Certificate: Subdivisions and a 4.86M Acre-Foot Gap

• Brief 5 · Sun Belt Multifamily Insurance and IRR: Climate Risk Behind a 207% Rise

Next in sequence:

• Brief 20 · NABERS 5.5-Star and Sydney Office Value: When a Rating Decides Your Tenant Pool

Sources

Every figure above, with the date the data covers, the date it was published, and the date I verified it.

Arizona stopped approving groundwater-based determinations — from June 1, 2023, ADWR no longer approves new Assured Water Supply determinations based on groundwater in the Phoenix AMA; the model projects 4.86 million acre-feet of unmet demand by 2121.

Arizona Department of Water Resources — Phoenix AMA Groundwater Supply Updates · Data as of Jun 2023, projected to 2121 · Published 2023 · Accessed Aug 2026

Developments within existing Certificates and Designations may continue to expand. Only new groundwater-based determinations have stopped. This is not a building moratorium. ADWR’s wording is “will not approve,” not “will not process”; it does not state how pending applications were administratively handled after June 1, 2023.

Assured Water Supply program dates — created by the 1980 Groundwater Management Act; ADWR adopted the operative Assured Water Supply Rules in 1995, so groundwater satisfied the enforceable hundred-year test for roughly three decades.

Arizona Department of Water Resources — Arizona Groundwater Code · Data as of 1980–1995 · Published date not stated · Accessed Aug 2026

ADWR says it “adopted new” rules in 1995, which implies a predecessor rule set between 1980 and 1995. The 1995 rules are operative, not the first.

Alternative pathway — the Alternative Designation of Assured Water Supply allows qualification using non-groundwater supplies; the 25% reduction applied to each new alternative supply under R12-15-710(H) was voided by the Maricopa County Superior Court in June 2026 in HBACA v. ADWR.

Arizona Department of Water Resources — Alternative Designation of Assured Water Supply · Data as of 2023–2026 · Published 2024 · Accessed Aug 2026

Rio Verde Foothills water cutoff — Scottsdale ended standpipe deliveries on January 1, 2023; about 1,000 residents, in roughly 500 to 700 homes, relied on hauled water, in a community of some 2,000 homes.

High Country News — What happens when an affluent Arizona suburb’s main water supply is cut off · Data as of Jan 2023 · Published Feb 3, 2023 · Accessed Sep 2026

The 500 to 700 homes and ~1,000 residents pairing traces to High Country News. The ~1,000 figure is haul-water dependents, not the community’s population. Scottsdale had signaled the cutoff since 2015. The community is unincorporated and sits outside Scottsdale’s service obligation. No official population figure exists for Rio Verde Foothills; the census-designated place Rio Verde is a different, adjacent community.

Hauled water prices — prices nearly tripled for some households after the cutoff.

Washington Post — Arizona city cuts off a neighborhood’s water supply amid drought · Data as of Jan 2023 · Published Jan 16, 2023 · Accessed Aug 2026

Multi-decade groundwater depletion — satellite gravimetry documents sustained groundwater storage loss across the US Southwest and other major aquifer systems since 2002.

NASA JPL — GRACE and GRACE-FO · Data as of 2002–2024 · Published 2024 · Accessed Aug 2026

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you are underwriting land or residential product in a water-constrained market and want the entitlement and supply assumptions pressure-tested before you commit, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

Photo Credit: Khadejeh Nikouyeh/The Charlotte Observer/Reuters

Before Hurricane Helene, an estimated 10.2% of commercial properties across western North Carolina carried FEMA-backed flood insurance.

Helene delivered more than 30 inches of rain to the mountains of western North Carolina, approximately 390-400 miles as the crow flies from where it made landfall.

Market Signal

Helene came ashore shortly before midnight on September 26, 2024, as a Category 4 storm near Perry, Florida, with 140-mile-per-hour winds and a 15-foot storm surge. It was the strongest hurricane on record to strike the Big Bend region and the deadliest to hit the US mainland since Katrina, with at least 250 deaths in the National Hurricane Center’s report.

Then it stopped being a Florida story. Over the following 48 hours, Helene tracked through western North Carolina and eastern Tennessee, delivering more than 30 inches of rain into mountain terrain roughly 400 miles from where it made landfall. Catastrophic flooding and landslides isolated entire towns.

Asheville is not a coastal market. People bought into it precisely because of its elevation, distance from the ocean, temperate climate, and long reputation as a climate haven.

Which is why almost nobody had flood insurance. Across western North Carolina, roughly 10.2% of commercial properties held FEMA-backed flood coverage before the storm.

That almost 90% not carrying FEMA coverage doesn’t represent negligence. Flood insurance is severable in the United States, and the mandatory purchase requirement applies only within designated special flood hazard areas. A commercial owner on a hillside in Buncombe County, outside any mapped zone, was not required to buy it, was not advised to buy it, and would have been paying for a risk that the governing map said did not exist.

The map answered a question about coastal storm surge and river flooding. The event was orographic rainfall on saturated mountain soil, a different hydrology that the mapping regime handles poorly.

The insurance market response has been more measured than most coverage suggests, and getting this right matters for anyone modeling the aftermath. North Carolina’s approved homeowners rate settlement produced increases of 4.4% in 2025 and 4.5% in 2026 for western North Carolina counties, against a statewide base rate increase of 7.5% in each year, totaling a cumulative 15% increase over a two-year settlement negotiated with the state. Insurers had originally sought 42% statewide and did not get it.

So the filed, approved, and regulated homeowners number is modest. Individual commercial renewals after a loss event are a different market, negotiated asset by asset with no rate filing behind them, and they have moved considerably more. Both facts are true, and they describe different things.

Deal Scenario

The following is a modeled composite based on a single-asset carrier repricing after a loss event, stated against the filed statewide rate context above. This is not meant to reflect an actual market rate or a specific transaction.

A 48-unit multifamily asset in the Asheville area, built in 2006, acquired months before the storm. The purchase price was $5 million at a projected first-year cap rate of 6.75%, resulting in a projected year-one NOI of $338,000. Insurance ran $65,000. Utilities ran $42,000.

The deal penciled. Nothing about the underwriting was careless. An inland mountain market with stable population growth and no coastal exposure is the kind of asset a conservative sponsor buys deliberately.

The model did not include a flood scenario because the parcel sat outside a mapped flood zone and therefore outside the mandatory purchase requirement. No lender demanded it. No appraisal flagged it. No comp suggested it.

After an event, three things likely move at once, and only the first is visible in the operating statement.

• Insurance. Renewal on a post-event asset in a newly demonstrated hazard area is negotiated, not filed. Expect a step change rather than an escalator, and expect the carrier to ask questions it never asked before.

• Capital expenditure. Stormwater compliance, site drainage, and any structural remediation arrive together. Asheville’s stormwater requirements are published and were tightened after repeated flooding. The five-year cost is deal-specific.

• Exit. The buyer pool contracts to those willing to underwrite a market that has now demonstrated the hazard. That is the largest single value effect and the slowest to appear.

Run those together, and the return does not decline gradually. It steps down once at renewal and again at exit when the comp set has repriced.

The uninsured portion is what makes this different from an insured loss. An asset with flood coverage in place files a claim, absorbs a deductible, and rebuilds. An asset without it funds the entire repair from equity or from a lender willing to advance against impaired collateral. That is a capital structure event, not an operating one.

Underwriting Analysis

The failure here was not the model. It was the question the model was answering.

Every input in that pro forma was defensible. The error was treating “outside the mapped flood zone” as equivalent to “not exposed to flooding,” and those two statements are only loosely related.

So the analysis has to add a question that the standard diligence stack does not ask.

• Get a modeled flood risk score, not only the FEMA designation. First Street’s modeling identifies 14.6 million US properties at substantial flood risk, roughly 70% more than the federal maps designated, on First Street’s June 2020 national assessment. The July 2025 update, High Water, High Stakes, puts it at 17.7 million properties at 1% annual flood risk against 7.9 million in FEMA Special Flood Hazard Areas, roughly 2.2 times. If the two disagree about your parcel, that disagreement is the finding. Since their July 2025 update, major inland flooding causing significant property damage in the US occurred in the Texas Hill Country in July 2026; the Catskills and mid-Hudson Valley (NY/New England) in July 2026; and Indiana, Kentucky, Ohio, and Arizona in August 2026.

• Ask what the site’s upstream watershed looks like. Mountain and hill-country areas flood from above, not from a river rising. The impervious surface and the slope upstream of your parcel determine what arrives during an extreme rainfall event, and neither appears on a flood map. Similarly, it’s worth knowing whether you are downstream of any reservoirs, retention ponds, or even dams that may release water, causing flooding that would not otherwise occur.

• Price flood coverage even when it isn’t required. In a region where 10.2% of commercial properties carry it, the premium is low precisely because the take-up is low. That is the cheapest point at which to buy the protection.

• Ask when the governing maps were last revised. A map built from a rainfall record that predates the last decade of observations is describing a different climate.

A covered asset takes a deductible. An uncovered asset takes the full loss and applies it against equity. That difference is larger than any premium assumption you could make.

Strategic Implications

The portfolio lesson is that the word “safe” does unexamined work.

Investors have spent several years reallocating toward inland, elevated, temperate markets because they thought those places carried less climate risk. The reasoning is sound at the regional level but can be wrong at the parcel level because regional climate averages do not capture the hydrology of a specific valley.

Asheville, Western North Carolina, and the Southern Appalachians appeared on every climate-haven list published between 2018 and 2024. That is not a reason to avoid them. It is a reason to underwrite them with the same rigor applied to a Gulf Coast asset.

The specific discipline that follows is to separate two things that get discussed as one. Regional climate exposure is what you screen for when you select a market. Parcel-level hazard is what you underwrite when you buy a building. Getting the first right doesn't accomplish the second, and the market that felt safest is the one nobody looked at more closely.

The demand-side question decides a five-year hold, and it runs opposite to the damage.

Climate-haven markets grew as people moved to them. A single catastrophic event usually doesn’t reverse that, and rebuilding itself creates near-term demand as displaced households need somewhere to live. Rent can rise in the 18 months after a disaster even as values fall, producing a confusing operating statement.

The longer-term outcome depends on whether the market can still insure and finance itself once rebuilding stops. A place that recovers with coverage intact keeps its buyer pool. A place that recovers into an insurance market that has repriced or withdrawn is holding the same buildings against a smaller set of people who can own them.

That is the variable to track, and it takes several years to resolve. It also means the worst time to judge a post-event market is during reconstruction, when operating numbers are temporarily inflated.

There is also a pricing opportunity here. In markets where insurance take-up is low because the hazard is unpriced, the cost of buying the protection is correspondingly low. An owner who carries flood coverage in a region where nine out of ten owners across western North Carolina do not has bought a very cheap option on being the only financeable asset after an event.

One last point about the map itself. FEMA designations govern mandatory purchase requirements and were primarily developed for riverine and coastal flooding. Orographic rainfall on steep saturated terrain is a hydrology the mapping regime was not designed around, which is why the designation and the actual exposure diverged so widely here. Treat the federal map as a legal boundary rather than a risk assessment.

Brief 9 showed the same failure at market scale in Valencia. Brief 12 looks at how disclosure regimes are trying to force this exposure onto balance sheets before the event rather than after.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Run this on your own deal

The CRDF Deal Stress Test™ built for this brief takes an inland asset and tests it against an unmapped flood scenario, an uninsured loss, and a contracted buyer pool. Free, no signup: Brief 11 - CRDF Deal Stress Test™ (xlxs)

New to the framework? The blank master Signal Tracker and Deal Stress Test workbooks are at climatereadyre.com/tools.

Related briefs

Same signal (S5 Acute Climate Hazard):

• Brief 9 · Valencia Flood 2024 and Property Risk Maps: 491mm in Eight Hours

• Brief 6 · The 30-Year Mortgage and Climate Risk: What the LA Fires Exposed About Loan Duration

• Brief 3 · Hoboken Property Values: The Impact of $230M Post-Sandy Flood Infrastructure

Next in sequence:

• Brief 12 · Climate Disclosure Rules for Real Estate in 2026: CSRD, SB 253, and What Changed

Sources

Every figure from above includes the date the data covers, the date it was published, and the date I verified it.

Commercial flood insurance take-up — approximately 10.2% of commercial properties and 5.2% of homes across western North Carolina held FEMA-backed flood insurance before Helene; NC Newsline attributes the estimates to FEMA and North Carolina Department of Insurance data without naming the dataset or its vintage

NC Newsline — Few were insured against Helene’s floods; those that were faced long claims, frustrating hurdles · Data as of Sep 2024 · Published Apr 22, 2025 · Accessed Aug 2026

Hurricane Helene landfall and impact — Category 4 near Perry, Florida on September 26, 2024, 140 mph winds, 12- to 16-foot storm surge; at least 250 deaths (176 direct); more than 30 inches of rain in western North Carolina, approximately 390 to 400 miles from landfall, measured straight-line

NOAA National Hurricane Center — Tropical Cyclone Report: Hurricane Helene (AL092024) · Data as of Sep 2024 · Published 2025 · Accessed Aug 2026

The larger mileage figure in circulation describes road distance, not distance from landfall. Straight-line distance from the Florida landfall to western North Carolina is approximately 390 to 400 miles.

North Carolina approved homeowners rate changes — western NC counties +4.4% (2025) and +4.5% (2026); statewide base rate +7.5% each year; insurers originally sought 42% statewide.

Blue Ridge Public Radio — North Carolina’s disaster risk is high. How does that affect home insurance premiums? · Data as of 2024–2026 · Published Mar 18, 2026 · Accessed Aug 2026

These are filed homeowners rates for western North Carolina rating territories, not Asheville-specific figures and not commercial rates. Commercial multifamily renewals are negotiated asset by asset with no rate filing behind them.

North Carolina rate settlement, supporting coverage — carries the Asheville and Buncombe territory figures, 4.4% for 2025 and 4.5% for 2026, 8.9% combined against the 21% the Rate Bureau proposed for the territory.y

Asheville Citizen-Times — Asheville homeowners insurance rate increases; future Helene impacts unknown · Data as of Jan 2025 · Published Jan 23, 2025 · Accessed Aug 2026

Publisher and publication date derived from the source URL.

Flood exposure beyond the federal maps — 14.6 million US properties at substantial flood risk, against the 8.7 million FEMA designates in Special Flood Hazard Areas, roughly 70% more

First Street Foundation — The First National Flood Risk Assessment · Data as of 2020 · Published Jun 29, 2020 · Accessed Aug 2026

The 70% is a comparison of two counts, 14.6 million against FEMA’s 8.7 million in Special Flood Hazard Areas, not an increase in risk. The figure is from the June 2020 assessment. It has since been superseded by First Street, High Water, High Stakes, July 2025, which gives 17.7 million properties at 1% annual flood risk versus 7.9 million in FEMA Special Flood Hazard Areas. MSCI acquired First Street in August 2026.

Asheville stormwater requirements — tightened compliance obligations following repeated flooding

City of Asheville — Stormwater, Development Services · Data as of 2025 · Published date not stated · Accessed Aug 2026

Requirements are published. Any five-year cost figure is deal-specific and modeled, not published.

48-unit Asheville asset — $5M basis at a 6.75% projected cap rate, $338,000 projected year-one NOI, $65,000 insurance, $42,000 utilities, with premium and exit stress applied at the asset level

CRREI modeled composite · Method: single-asset carrier repricing after a loss event, stated against the filed statewide rate context above · Data as of 2026 · Published date not stated · Modeled — not a market rate and not a specific transaction

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you are underwriting an inland asset in a market everyone calls a climate haven and want the flood and insurance assumptions pressure-tested before you sign, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC

Photo credit: Kai Försterling for the European Pressphoto Agency (EPA)

On October 29, 2024, the municipality of Chiva received 491.2 mm in 8 hours, including 343 mm in 4 hours, and more than 300 mm fell over the day across the Utiel-Chiva corridor.

That is roughly a year of precipitation, delivered in a single working day.

Valencia was not on anyone’s list of high-risk real estate markets. It had a flood map, a drainage system, and a modern meteorological service that issued its maximum red alert that morning.

The Moment

A DANA is a cut-off low, a pocket of cold air that separates from the jet stream and stalls over warm Mediterranean water. Spain gets them most autumns. This one sat over the Valencian interior and did not move.

The rainfall totals were not marginal. At Chiva, 343 millimeters fell in four hours, and more than 300 millimeters fell across the Utiel-Chiva corridor over the full day. Some areas received more than a full year’s worth of rain in eight hours.

The Spanish meteorological agency raised its red alert for the Valencia interior at 7:36 that morning. It extended it across the whole province by 9:41. The regional emergency alert reached mobile phones at 8:11 that evening. The water arrived in between. Spain still politically contests that sequence.

What matters for an investor is narrower and more transferable than assigning blame. The hazard was forecast. The warning existed. And the built environment could not absorb the gap between the two because it was built on the assumption that the gap would never be tested.

237 people died. Total economic damage is commonly cited at around €16.5 billion. Insured losses are estimated at approximately €4.8 billion, and Spain’s Insurance Compensation Consortium had paid more than €4 billion across 250,663 claims as of November 24, 2025.

The Story

The number that reframes this event is not the damage figure. It is the denominator.

The Valencian Institute of Economic Research put the total value of private and public assets located in the flooded municipalities at €85.9 billion. It estimated that at least 20% of it was destroyed.

A regional flood in a mid-sized European metropolitan area sat atop an asset base approaching €90 billion. The loss was large because the exposure was large, and the exposure was large because nobody had priced the location as exposed.

This pattern makes the phrase “safe market” dangerous. Valencia was safe in the sense that mattered least. It had not recently flooded. It was not safe in the sense that mattered most, in that a great deal of value had accumulated in places water had historically traveled.

Much of the affected development sits on or near ramblas, the dry riverbeds that carry water only during extreme rainfall. For most of a human lifetime, a rambla looks like flat, buildable, well-drained land. It is neither flat by accident nor dry by nature. It is a channel that is empty most of the time.

A flood map drawn from the recent record will show a rambla as low risk because the recent record is a period when it did not rain that hard. The map is not wrong about the past. It is being asked a question about the future that it was never built to answer.

That distinction applies to markets within and beyond Spain.

Valencia had solved this problem once already. After a catastrophic flood in 1957, the city diverted the Turia River into a new channel south of the urban core and turned the old riverbed into a park. It was one of the most ambitious pieces of urban flood engineering in twentieth-century Europe, and it worked.

What it did not do was protect the municipalities inland of the city, where the growth of the following decades actually went. The engineering team defended the flooded area. The development moved to areas that had not yet flooded.

That is a pattern, not a coincidence. Protective infrastructure raises the perceived safety of an entire region, not only the parcels it defends, and the development that follows tends to spread past the edge of the protection. The map showed a protected city. The buildable land sat outside the protection.

Spain covers extraordinary risks through the Consorcio de Compensación de Seguros, a national pool funded by a surcharge on ordinary policies. Flood coverage is therefore close to universal rather than optional, as it is in the US, which is why the estimated insured cost runs to about €4.8 billion in a country where nobody buys standalone flood insurance.

In the United States, flood is severable, federally underwritten, and mostly unpurchased outside mandatory-purchase zones. Same hazard, entirely different distribution of who absorbs the loss. A comparable event in the US would result in a smaller insured loss and a much larger uninsured loss.

Structural Forces

Three forces determine whether a market’s risk map still accurately describes it.

The first is that the underlying probability has moved. World Weather Attribution’s analysis of this event found that one-day rainfall of this intensity is now roughly twice as likely and about 12% more intense than it would have been without climate warming.

If an event has become twice as likely, then a drainage system engineered against a hundred-year rainfall is now defending against something closer to a fifty-year rainfall. The concrete did not change. The number the concrete was poured against did.

A separate analysis from the Barcelona Supercomputing Center found that a warming North Atlantic contributed to the storm’s intensity by loading the atmosphere with additional moisture, adding about 15% to the episode’s intensity. Warm water is fuel. The Mediterranean has been getting warmer.

The second force is that the map and the hydrology have separated. A flood map is a snapshot of modeled expectation at a moment in time, produced on a revision cycle measured in years or decades. The hydrology it describes is now outpacing the revision cycle.

Any investor holding assets in a market whose flood maps have not been recently revised is holding assets priced against a document that is aging out. That is not a Spanish problem. That is why FEMA’s mapping backlog matters in the United States, and why First Street’s modeling identifies 17.7 million properties at 1% annual flood risk against 7.9 million inside FEMA’s designated zones.

The third force is slower and demographic. Places that flood repeatedly eventually lose the population that makes their real estate valuable. That process does not begin with the flood. It begins with the second flood, or with the moment insurance becomes unavailable, whichever arrives first.

Valencia is a large, economically diverse metropolitan region, and it will absorb this. Smaller places do not always do so. The mechanism to watch is whether a market retains the households and employers that make a rent roll durable, because population is upstream of every leasing assumption in a pro forma.

Next Chapter

The practical conclusion is that a “safe market” is not a category. It is a claim about a specific hazard, at a specific parcel, against a specific and dated model.

So the question to ask about any market is not whether it is safe. It is three narrower questions that can actually be answered.

• What is this parcel’s hazard exposure according to a model built this decade, rather than according to the regulatory map, which may be considerably older?

• When were the governing flood maps last revised, and what rainfall record were they built from? A map produced before the last decade of observations is describing a different climate.

• What is the accumulated asset value in the surrounding hazard zone? This is the Valencia lesson. Your loss depends on your exposure, but your market’s liquidity depends on everyone’s.

Often, the worst-hit municipalities sit inland, and the water that destroys the value comes from the sky and down a dry riverbed or mountainside, not always from the sea or storm surge pushing inland up a river.

The transferable point is that the most expensive climate exposures are the ones a market has ignored as exposures. Coastal risk is priced, sometimes badly, but it is priced. Inland flash-flood risk in a Mediterranean metro, a Southern Appalachian mountain town, or a Midwest suburb often isn't priced at all because the last event predates everyone currently underwriting.

Brief 10 takes this into the credit market and looks at how lenders are rewriting mortgage criteria for exactly this reason. Brief 11 puts a deal-level number on the same failure in western North Carolina.

As always, KNOW YOUR SIGNALS and BE CLIMATE READY!

Jamie

Go deeper

This is a Story & Future Thinking brief, so there is no companion workbook. The blank master CRDF Signal Tracker and Deal Stress Test are free and available at climatereadyre.com/tools.

Related briefs

Same signal (S5 Acute Climate Hazard):

• Brief 3 · Hoboken Property Values: The Impact of $230M Post-Sandy Flood Infrastructure

• Brief 6 · The 30-Year Mortgage and Climate Risk: What the LA Fires Exposed About Loan Duration

• Brief 27 · Flood Zoning and Building Permits in Valencia: A EUR 4B Payout

Next in sequence:

• Brief 10 · Bank Lending Criteria and Climate Risk on Property: The New Overlays

Sources

Every figure above, with the date the data covers, the date it was published, and the date I verified it.

Rainfall at Chiva — 491.2 mm in 8 hours, 343 mm in 4 hours, and 160 mm in 1 hour at the Chiva weather station.

European Severe Storms Laboratory — Meteorological analysis of the extreme flash flood situation in the Valencia region; WMO — Devastating rainfall hits Spain in yet another flood-related disaster (Oct 31, 2024), carrying AEMET’s areal totals · Data as of Oct 29, 2024 · Published Nov 5, 2024 · Accessed Aug 2026

ESSL carries the sub-daily station values verbatim. The “more than 300 mm” full-day areal totals and the “a year’s rainfall in eight hours” comparison are AEMET statements carried by the WMO news item.

Death toll — 237 deaths across Spain. 229 were in the Valencia region, 7 in Castilla-La Mancha and 1 in Andalusia.

Plataforma Media — State funeral for 237 victims in Valencia, one year after tragedy · Data as of Oct 2025 · Published Oct 29, 2025 · Accessed Aug 2026

237 is the consolidated national toll used at the October 2025 state funeral. ECMWF Newsletter 183 states “at least 232,” an April 2025 vintage, and cannot carry this figure. The 1957 Turia flood that led to the river’s diversion killed 81 (ESSL, Historical deadly flash floods in Spain, Oct 31, 2024).

Total economic damage — around €16.5 billion.

ECMWF — Extreme precipitation in Spain’s Valencia region (Newsletter 183) · Data as of Oct 29, 2024 · Published Apr 2025 · Accessed Aug 2026

ECMWF carries the €16.5 billion figure verbatim. Damage estimates vary widely by methodology and vintage; wider ranges circulate, but no single named source carries them, so none is stated here.

Insured losses and Consorcio payouts — estimated final insured cost of approximately €4.8 billion; the Consorcio de Compensación de Seguros had paid more than €4 billion across 250,663 claims as of November 24, 2025.

Consorseguros Digital — The 2024 Valencia Floods · Data as of Nov 24, 2025 · Published autumn 2025 (issue 23) · Accessed Aug 2026

Scope: CCS-insured cost only, never total economic damage. These are the source’s own current figures — about €4.8 billion estimated final insured cost, with more than €4 billion paid across 250,663 claims as of November 24, 2025.

Asset value in the flooded municipalities — €85.9 billion of private and public assets located in the affected municipalities, with at least 20% estimated destroyed.

Ivie, Valencian Institute of Economic Research — La DANA puede haber destruido al menos el 20% de los 85.900 millones de euros de activos privados y públicos localizados en los municipios inundados · Data as of Jan 2025 · Published Jan 17, 2025 · Accessed Aug 2026

Attribution — one-day rainfall of this intensity is roughly twice as likely and about 12% more intense in the current climate.

World Weather Attribution — Extreme downpours increasing in southern Spain as fossil fuel emissions heat the climate · Data as of Oct 2024 · Published Nov 4, 2024 · Accessed Aug 2026

The doubling attaches to regional one-day rainfall in an observations-only trend analysis, not to the eight-hour accumulation at Chiva.

Ocean warming contribution — a warming North Atlantic contributed to the storm’s intensity, adding about 15% to the episode’s intensity (Saurral and colleagues, Weather and Climate Extremes, Mar 16, 2026).

PreventionWeb — BSC study reveals warming North Atlantic contributed to intensifying Valencia DANA storm · Data as of Oct 2024 · Published Mar 16, 2026 · Accessed Aug 2026

Warning timeline — AEMET raised its red warning for the Valencia interior at 7:36 on the morning of October 29, 2024, and extended it to the whole province at 9:41; the Generalitat’s ES-Alert mobile warning went out at 20:11.

Maldita.es — Cronología de la DANA: avisos de la AEMET y alertas de Protección Civil, 29 de octubre de 2024 · Data as of Oct 29, 2024 · Published Oct 30, 2024 · Accessed Aug 2026

The timeline is widely reported and remains politically contested in Spain. The timeline comes from Maldita.es’s reconstruction of AEMET’s published warnings and the Generalitat’s alert. The sequence is stated; no causal claim is made about specific deaths.

US flood exposure beyond the federal maps — 17.7 million properties at 1% annual flood risk versus 7.9 million in FEMA Special Flood Hazard Areas, roughly 2.2x. MSCI acquired First Street in August 2026; a commercial index provider rather than an independent foundation now owns the model.

First Street — High Water, High Stakes: FEMA · Data as of Jul 2025 · Published Jul 2025 · Accessed Aug 2026

Commentary and analysis only. Not investment, financial, legal, tax, or professional advice. CRDF tools are illustrative; examples are composites drawn from public data. Do your own due diligence and consult qualified professionals.

I run this analysis on specific deals. If you are underwriting an asset in a market everyone calls safe and want the hazard and insurance assumptions pressure-tested before you sign, book 20 minutes.

Jamie Wolf, MBA — Founder & Publisher, Climate-Ready Real Estate Investing, © 2026, CR REI Holdings LLC