Post date 13 Jul 2026 (original LinkedIn date)
Original LinkedIn version· 1,894 words
‘St Elmo’s Fire (Man in Motion)’ by John Parr, released in 1984, but truly came to fame as the title track of the eponymous film in 1985.
Three interesting papers/reports landed on my desk during June - and inevitably I didn’t have time to read them back then. Read together, to my way of thinking they round out at least one nature-finance argument of interest ... and then opened an even harder one! Fairly standard for our little industry! There is a particular kind of signal you learn to trust = not one person making a point (however well they make it), but rather three people who do not know each other, arriving at broadly the same place in the same month without coordinating. I’m pretty sure that is what happened this June.
The first was from the Grantham Research Institute at the LSE: Prof Nicola Ranger and colleagues’ Too big to fail, which argues that certain ecosystems are macro-critical infrastructure and floats a designation of “systemically important natural systems” modelled deliberately on the too-big-to-fail banks. The second was a Communications Earth & Environment paper by Megan Kelso, Borja Reguero, Michael Beck and co-authors on advancing nature-based insurance. The third was GFANZ’s Investing in Resilience, twenty-two private-finance case studies on what is actually getting adaptation funded on the ground. So three entirely different vantage points: macroprudential governance, the insurance underwriting stack, and the deals.
The debate?
For most of the last decade, a main argument in nature finance was about measurement. Could we value what an ecosystem does? Could we get the data to a standard that credit and investment committees would accept? A great deal of serious effort went into that question, and it was appropriate at the time. These three reports, read together, quietly move us on. Ranger’s team put a figure on forest ‘green water’ (the rainfall that evapotranspiration carries across borders) of more than $5 tln of direct economic dependency across 130-plus countries, with >$1 tln of that landing in China, Europe and the United States alone. Kelso’s starting point is not whether reefs and marshes reduce flood damage; that is now quantified inside the industry catastrophe models insurers already trust; it is whether anyone prices it. GFANZ’s institutions, meanwhile, are translating physical hazard into credit and underwriting metrics almost as a matter of routine practice now. So the measurement problem is being industrialised, even if it is not finished everywhere, but it is no longer the thing standing between capital and nature.
The wall now
So what is the principal blocker now? Look at where each report actually lands, and I think you’ll find the honesty is striking. Ranger’s own diagnosis, underneath all the designation architecture, is a list of market failures: non-excludable benefits, costs borne locally while the value accrues globally, coordination failures across actors who cannot move first alone. Designation is a genuinely powerful act of recognition, but recognition is not a cashflow. Telling a finance ministry that a catchment is systemically important does not tell anyone who pays to keep it that way.
Kelso’s team put the same point as an engineering fact. Insurers’ risk modellers, they report, do not build habitat benefits into their models because they see no client demand for the information. The entire five-frontier structure of that paper is one move repeated - an attempt to manufacture a priced, excludable claim out of a diffuse protective service. Every instrument they describe presupposes a beneficiary who is already contractually on the hook (if only!).
And GFANZ’s central finding is that resilience tends to show up as “avoided losses rather than contractual revenues.” The innovation it identifies as essential (still nascent, it concedes, but decisive, and singled out for nature-based solutions in particular) is the outcome-linked payment structure that converts an avoided loss into a predictable, investable cashflow. Can we therefore all agree that the binding constraint is no longer valuation? Instead it is demand and contractibility. Can I push us to call it the emerging consensus, because that is what it now is?
Excludability issue
Let me pause my stream of thought for a second, as I think it is worth being precise about the distinction, because I suspect that a lot of good work has been aimed at the wrong target. You can price a thing perfectly and still have no market for it, if no one can be made to pay. A restored floodplain attenuates a flood peak downstream - we can measure that attenuation. The reason capital does not flow to it is not that the benefit is unmeasured; it is that the beneficiary cannot be excluded from receiving it, and the claim on them cannot be enforced. That is the excludability problem, and it is arguably a critical layer at which nature finance actually fails.
The power sector
So here is the encouraging part, I think - we have solved a structurally identical problem before, at scale, within living memory. A clean electron is physically indistinguishable from a dirty one, and its principal benefit (a stable climate) is a textbook global public good: non-excludable, diffuse, enjoyed by everyone and owned by no one. By the logic above it should have been unfinanceable. Instead there is now a multi-trillion-dollar market for renewable power. We did not get there by valuing carbon more accurately. We got there by inventing a contract: the Power Purchase Agreement. A fixed price, a long tenor, an identifiable offtaker who agrees to buy a defined unit of output over time. The PPA did not measure the electron better, but it did make the electron bankable.
The Resilience Purchase Agreement (the RPA) can do the same thing for resilience. It would contract a catchment-scale intervention and sells the resulting reduction in risk, as a defined and measured unit, to an offtaker who owns the underlying exposure. The offtaker pays because the avoided loss is theirs: identifiable, attributable, and no longer available to them for free.
The unit
The unit also matters, so let me be specific, because ‘resilience’ in the abstract is exactly the kind of diffuse benefit that has never financed anything (yet!). A catchment is not a single service, rather it is a factory with several offerings, and each line maps to a distinct, separately contractible risk borne by a distinct beneficiary. Take water as the organising frame, because water is where most catchment risk concentrates.
Water quantity - flood: Natural flood management and floodplain reconnection attenuate peak flows. The unit could be the measured reduction in flood exposure; the offtaker is the downstream asset owner, the infrastructure operator, or the insurer carrying that tail.
Water scarcity - drought: The same catchment, managed for infiltration and baseflow, sustains water availability through dry periods. The unit is low-flow resilience; the offtaker is the abstractor, the utility, or the industrial user whose operations halt when a licence is curtailed in a bad summer.
Water quality - pollution: Riparian buffers, wetlands and land-use change reduce nutrient, sediment and contaminant loads. The unit is the measured reduction in pollutant delivery; the offtaker is the water company facing treatment costs and permit exposure, for whom catchment intervention is frequently cheaper than the concrete alternative.
Coastal erosion: At the seaward end, saltmarsh, dune and reef systems attenuate surge and stabilise sediment. The unit is the protection delivered; the offtaker is the port, the coastal infrastructure operator, or the property owner at the shoreline.
Four risks ~ four units ~ a single catchment - each is a measurable delta against a counterfactual& each is attributable to a beneficiary who can be named. That attributability is what converts GFANZ’s avoided loss into contractual revenue, not as an aspiration, but as a line in a signed agreement.
Fitting the three reports together
The RPA is not a rival to natural-systems designation, nor to nature-based insurance. Instead it is the missing middle that each of them, I think, needs.
Ranger’s designation supplies the mandate and the macroprudential tailwind. If a catchment is recognised as macro-critical, the offtaker’s regulator and the host sovereign have reason to treat investment in it preferentially, which lowers the cost of capital for the very interventions an RPA finances. Designation does not write the contract, but it makes the contract cheaper to fund.
Kelso’s nature-based insurance supplies the wrapper. Parametric cover can absorb the tail event that a catchment intervention cannot prevent; the performance insurance her team proposes can stand behind the unit itself, guaranteeing that the resilience an RPA promises is the resilience delivered. That is what makes the offtaker’s counterparty risk bearable and therefore what makes the RPA signable.
GFANZ supplies the capital deployment playbook and, tellingly, names both the instrument class and the aggregation logic. Individual catchment RPAs are small and local, but if you pool them, you have a diversified portfolio at the ticket size and duration a pension pool actually needs. This is how demand-backed institutional capital, the kind of long duration money now looking seriously at natural capital, gets deployed into an asset class that would otherwise be too fragmented to touch.
The strongest objection?
It’s only fair to offer one, right? Ranger’s route may make bilateral contracts unnecessary. If designation flows through into fiduciary duty, capital weights and disclosure, you can in principle move an entire banking book with a rule and never write a single RPA. Why build a contract at the catchment when you can nudge the whole system from the top? For one layer of the problem, I think that is exactly right. For the global-commons, mitigation-flavoured, averaged part of it, regulatory pressure is the correct and more powerful lever. But adaptation value does not ‘average’ as we have explored in my latest two brain dumps. A flood does not hit a global index; it hits a single site. A drought does not curtail the portfolio; it curtails one abstraction licence on one river. The avoided loss is lumpy, local and attributable, and those are precisely the properties that make it contractible, and precisely the properties a portfolio-level rule cannot capture. Remember, ‘the catchment does not average’.
So I don’t believe that the two routes are substitutes. Rather they are complements operating at different scales. Designation and prudential pressure set the mandate and the cost of capital, while the RPA writes the cashflow. Ideally you need both, and each makes the other work harder.
The new challenge
The measurement decade is behind us, and the contracting one is starting. And the unit of account for it will not be the tonne or the hectare, rather it will be the resilience unit, measured at the catchment, priced against a named risk, and (hopefully) sold to the beneficiary who must foot now the bill.
References
1. Ranger, N., Philip, T., Tayler, T. and O’Donnell, E. (2026). Too big to fail: critical natural systems as macro-critical infrastructure. London: Grantham Research Institute on Climate Change and the Environment / Earth Capital Nexus, London School of Economics and Political Science. DOI: 10.21953/researchonline.lse.ac.uk.00138880
2. Kelso, M. A., Taylor-Burns, R., Franco, G., Carter Ingram, J., Reguero, B. G. and Beck, M. W. (2026). “Advancing nature-based insurance to protect people and nature.” Communications Earth & Environment, 7:540. DOI: 10.1038/s43247-026-03753-z
3. Glasgow Financial Alliance for Net Zero (GFANZ) (2026). Investing in Resilience: Lessons from private finance for unlocking investment in adaptation. June 2026.