When a Climate Shock Hits the Balance Sheet of a Bank

The source we were pointed to never loaded. Instead of a paper, we got a Cloudflare error page, which means we have no study of our own to quote. What we do have is a title, "From risk to resilience: Climate change," and a set of well-documented ideas about how climate risk reaches bank balance sheets. So this piece sticks to those, and we flag where figures are illustrative rather than measured.

Start with a simple scenario. A coastal county in Florida gets hit by a hurricane that damages 15% of the homes in a mortgage pool. Insurance covers some of it, some owners are underinsured, and a few just walk away from homes worth less than the loan.

That chain of events is why large lenders such as Citigroup (C), Morgan Stanley (MS) and Goldman Sachs (GS) now spend real effort on climate scenario work. Their exposures differ a lot, but each one touches climate risk through lending, trading, underwriting or advisory work.

Why does this matter in late October 2026? Insurance pricing has been moving for several years in the most exposed US states, and some insurers have pulled back from writing new policies altogether. When coverage gets scarce or costly, collateral values come under pressure, and banks feel it on the credit side.

Regulators have been circling the issue for a while. The Network for Greening the Financial System, formed by central banks in 2017, built scenario sets that many large institutions still reference. US policy has been less steady than European policy, and that unevenness is part of the story we want to tell.

Physical Risk and Transition Risk Are Different Animals

Climate finance people split the problem into two buckets. Physical risk covers the direct damage from floods, fires, storms and heat. Transition risk covers what happens to asset values when policy, technology or consumer taste shifts away from carbon-heavy activity.

The two behave differently over time. Physical losses arrive in lumps and are often local. Transition losses build slowly, then can reprice an industry in a few quarters once a policy change or a cheap new technology arrives.

Consider a coal-fired utility with a 30-year loan. If a carbon price or cheap renewables make its plants uneconomic by year 12, the lender holds a loan backed by assets worth far less than planned. Nothing physical happened to the plant. The collateral just stopped being worth what the model said.

A common misconception is that banks can sell risky assets whenever they see trouble coming. Loans are not like listed shares. Many are illiquid, relationship-driven and held to maturity, so the exit is slow and sometimes costly.

Another trap is treating the two risks as separate. A region hit hard by storms may also pass tougher building codes, which raises costs for developers the bank has lent to. Physical events can accelerate transition policy, and the combined effect can exceed either one alone.

How Scenario Analysis Turns Weather Into Loss Rates

The mechanics are less mysterious than they sound. A bank takes a climate pathway, such as warming of 2 degrees versus 3 degrees by 2100, and translates it into variables it already models: property values, default probabilities, commodity prices, GDP by region.

Take an illustrative example, and we stress that the numbers are ours for demonstration. Suppose a $10 billion commercial real estate book has 20% of its value in flood-prone zones. If a severe scenario cuts those property values by 25% and the loans there carry an average loan-to-value ratio of 65%, the cushion shrinks fast.

A 25% price drop on a property financed at 65% LTV pushes the effective LTV to about 87%. That is still under 100%, so the loan stays above water on average, though the margin for further shocks is thin. In a 35% drop, the effective LTV passes 100%, and loss given default becomes a real number rather than a rounding error.

Here is the cause-and-effect worth following. Higher insurance premiums reduce net operating income. Lower income raises the cap rate investors demand. A higher cap rate lowers value. Each step feeds the next, which is why simple linear estimates tend to understate the damage.

Timing matters too. Most bank climate scenarios run 30 years, but loan books turn over in 5 to 7 years. Critics fairly point out that a long horizon can make the risk look distant, while the portfolio that faces it is a much younger one. Both views have merit, and good analysis reports both.

Data quality is the weak point. Property-level flood maps are patchy, corporate emissions disclosures vary in quality, and models trained on past weather struggle with a climate that is changing. Treat any single loss estimate as a range, not a point.

What C, MS and GS Bring to the Table

Citigroup is the most exposed of the three on the traditional lending side. It has a large global corporate loan book, heavy emerging market presence and a sizeable card and consumer business. Emerging market sovereigns and corporates can face acute physical risk, which makes country-level climate analysis relevant to Citi's credit costs.

Morgan Stanley sits differently. Its wealth management arm earns fees on client assets, so its climate exposure runs through market values and client behavior more than through its own loan losses. A prolonged equity drawdown tied to a climate event would hit fee revenue directly.

Goldman Sachs leans on trading, advisory and underwriting. Energy transition creates deal flow: financing for grid upgrades, renewables, carbon capture, and the restructuring of carbon-heavy firms. It also creates market risk if commodity and credit spreads swing sharply.

Risks first. For all three, a correlated shock is the worry: a severe event that hits credit, markets and client wealth at the same time. Diversified banks hold up better against single-sector stress but less well against one that spreads across everything.

Policy risk is real as well. If US rules loosen while European rules tighten, global banks face compliance work on two tracks and may see competitive effects that are hard to predict. Reputational risk can cut either way, with some clients penalizing banks for financing fossil fuels and others penalizing them for refusing.

Opportunities are just as concrete. Global investment in clean energy has run well above $1 trillion a year in recent years by most industry estimates, and that capital needs arrangers, lenders and market makers. Banks with strong project finance and structuring skills are positioned to earn fees from it.

Here is a scenario with numbers, again illustrative. Say a bank's advisory and underwriting arm earns a 1.5% fee on a $4 billion transition-related financing. That is $60 million in revenue from a single deal. Ten such deals in a year would be $600 million, which matters even for a firm of Goldman's scale.

Now flip it. Suppose that same bank carries $2 billion of loans to a sector facing a 20% impairment in a transition shock. That is a $400 million hit. The fee income and the credit loss can live in the same institution, so net exposure depends on how well the two are balanced.

Putting This Into a Real Portfolio

For an individual investor, the first question is whether climate risk is already priced into bank shares. Our view is that markets have priced it only partly. Markets tend to react to events, such as a bank taking a large provision after a disaster, but they are slower to price slow-building exposure.

Start by reading the disclosures. Each of these firms publishes climate-related information in annual reports and sustainability filings. We look for sector concentrations in energy, utilities and real estate, the share of lending in high-risk geographies, and whether the firm reports scenario results in any detail.

Next, think about position sizing rather than a yes-or-no call. If financials make up 20% of an equity portfolio, and a third of that sits in these three names, a single climate-driven repricing could move the whole portfolio more than expected. A cap on any single bank at, say, 3% to 5% of the portfolio is a reasonable discipline.

Consider how these holdings correlate with the rest of your exposures. An investor who already owns coastal real estate, regional insurers and large banks may be stacking the same risk three times. We find that diversifying across sectors and regions works better than trying to guess which bank is best prepared.

Entry timing is tricky because climate risk rarely announces itself. A reasonable approach is to add on valuation weakness unrelated to climate, then monitor a few indicators: insurance availability in major lending regions, regulatory stress test results, and commentary from the banks on credit quality in real estate and energy.

Common implementation problems include relying on ESG ratings that disagree with each other, assuming disclosure equals risk management, and overreacting to headlines. A bank that publishes a thorough report may still have large exposures, and one that says little may be fine. Read the numbers, not the tone.

Building Resilience Into Your Own Approach

Pull the threads together. Climate change reaches banks through credit, markets and fees, and the three firms we discussed sit at different points along that chain. Citigroup carries more direct credit and country risk, Morgan Stanley more market-linked fee risk, and Goldman Sachs a mix of deal opportunity and trading exposure.

None of this argues for avoiding bank stocks. It argues for knowing which channel you are exposed to and sizing accordingly. Resilience here means a portfolio that can absorb a regional shock without forcing a sale at a bad time.

A few concrete steps follow. First, review how much of your portfolio sits in financials and in real estate-linked assets, and tally the overlap. Second, read the latest climate disclosures for any bank you own and note the sector and region concentrations.

Third, set a position limit for each name and stick to it, rebalancing when a holding drifts past it. Fourth, keep some of your allocation in assets with low sensitivity to physical events, such as short-duration government bonds, so you have funds available if prices dislocate.

Finally, revisit the question once a year. Regulation, insurance markets and technology costs all move, and a view formed in 2026 may need updating by 2027. We would also encourage readers to find the original study behind this topic, since we could not access it, and compare its figures with the illustrative ones used here.