A factory can own its machinery and still depend on an asset absent from its balance sheet: a reliable supply of clean water. If that supply deteriorates, the machinery may produce less income while the loan used to buy it still needs servicing. This is the credit mechanism behind Frank Elderson’s call to treat nature as economic infrastructure.
In his October 26 COP17 speech, the ECB executive board member argued that preserving natural capital supports productive capacity and resilience. For lenders, the useful question is how that broad proposition changes an assessment of a particular borrower. Environmental dependence identifies where to investigate; it does not by itself establish a loan loss or an investment return.
A water dependency becomes a credit question at the borrower
Consider a hypothetical processor that needs water for cleaning and cooling. Restrictions could reduce operating hours; alternative supplies could increase costs. Either route would weaken cash available for interest and principal. A bank therefore needs to connect the physical constraint to the customer’s margins, liquidity and ability to adapt, rather than attach a generic environmental label to its sector.
Location changes that calculation. A second supplier is less useful if both suppliers draw on the same stressed catchment. Water storage might bridge a short interruption but offer little protection against a persistent shortage. A project that lowers water consumption could improve resilience, but its financing bill and operating costs also belong in the credit assessment. These are analytical examples, not reported outcomes at a named company.
The NGFS conceptual framework separates the sources of physical and transition risks from their economic effects and subsequent financial transmission. That sequence matters: a dependency becomes financially material through the severity of a disruption, available substitutes and the borrower’s capacity to absorb it. A vulnerability map supplies an opening question, not the final answer.
The large portfolio percentages describe a screening universe
The research behind the headline scale is substantial but needs its date and denominator. A 2023 ECB occasional paper estimated that approximately 72% of euro-area non-financial companies were highly dependent on at least one ecosystem service, and that almost 75% of corporate bank loans went to such firms. Its AnaCredit sample used December 2021 observations, covering around 4.2 million companies and €4.3 trillion in loans.
Those estimates describe dependencies in a research sample. They are neither an October 2026 portfolio census nor a claim that three-quarters of lending will default. A company may be highly dependent on water while operating where supplies remain dependable. Conversely, moderate direct exposure may conceal dependence through suppliers. The paper includes indirect supply-chain links, helping explain why a narrow view of a borrower’s own premises can miss risk.
For a bank investor, the screening result suggests that nature risk can cross conventional sector boundaries. It does not provide a justified percentage haircut to earnings or equity. That would require assumptions about hazards, adaptation, borrower finances and recoveries which the dependency percentage alone cannot supply.
A risk register needs a response that changes exposure
The implementation gap is more specific than a lack of awareness. In a May supervisory blog, Elderson said around two-thirds of banks did not systematically connect nature-risk materiality assessments with management action. Few had defined nature-related key risk indicators, and existing indicators often lacked thresholds that would trigger a response.
An indicator becomes useful when someone owns the decision it informs. A lender could seek site-level data, examine a client’s contingency arrangements or make financing conditional on a credible adaptation plan. Those possibilities illustrate the mechanism; they are not a new universal lending requirement. The same supervisory publication explicitly says its good-practice examples have no legally binding effect and do not establish new regulatory obligations.
There is also evidence of practical progress. The NGFS implementation compendium, updated on October 13, reviews 35 initiatives and six case studies. The variety is useful precisely because different institutions face different exposures. A sophisticated model should be judged by whether it improves decisions, not by the number of environmental variables it contains.
Restoration needs a payer as well as a public benefit
Elderson’s speech also raises a financing problem: some projects protect public goods without producing direct revenue. Cleaner water or reduced flood exposure can benefit many users, but a lender still requires a defined repayment source. Economic value to society and cash available to a project company are different quantities.
A plausible financing structure would need to identify who pays for resilience, who maintains the asset and who bears performance risk. It would also need to avoid counting the same benefit twice across beneficiaries. These are underwriting questions, not evidence that any particular restoration project offers attractive returns.
The strongest qualification is that firms can adapt, substitute inputs and change locations. Better borrower-level evidence showing those responses would reduce estimated vulnerability. Evidence that substitutes fail together, insurance is unavailable or interruptions persist would strengthen it. The financial implication of the COP17 intervention therefore rests on measurable changes in cash flow and protection, not on translating a large dependency statistic directly into a loss forecast.