Post date 18 May 2026 (original LinkedIn date)
Original LinkedIn version· 3,949 words
“The Ocean and the Butterfly“ by the Dave Matthews Band, released in 2023 on the album ‘Walk around the moon’
The Dave Matthews Band is an all-time favourite, and I can listen to Crash on a loop quite happily. I’ve been lucky enough to hear them in concert 3 times and each time the song is played subtlely differently! Bit of a different direction today, giving discount rates, IRRs, option pricing theory a rest, altho I have no doubt we’ll return to them all in due course - after all, there’s so much more to dig into for each at the nexus of nature and investing.
Nature credits of one flavour or another (carbon, biodiversity, etc) have been the principal route to monetisation of Nature. There is a quiet conversation happening in UK natural capital right now between people who have been building credit-based products for the last several years and people whose models depend on those credits clearing. But I wouldn’t say that it’s a particularly comfortable conversation. In my opinion, credits are not delivering the cashflows that funds underwrote against, and the reasons are structural rather than fixable by waiting longer or working harder.
I want to set out where I think this leaves us, drawing on four strands of recent work that have been sitting on my desk this past couple of month: Sarah Conway‘s Risk & Resilience Ledger pieces on the Rhino and Spekboom outcome bonds, the Robertson, Lave and Doyle papers on the US stream credit market (these are older but I still enjoy rereading them), the LIFE Biodiv CrEW report on biodiversity credit schemes for European wetlands, and the recent ocean finance work from WWF, Katapult Ocean, Builders Vision and the Green Finance Institute.
Read together, these tell a fairly clean story. Credits are a useful instrument for some things and a poor instrument for the much bigger problem they are increasingly being asked to solve. So I want to argue that the financeable backbone for UK natural capital sits elsewhere, in long-duration contracted resilience cashflows from regulated and statutory counterparties. Credits can stack on top as upside, when they work at all.
Extending the Conway thinking
Sarah’s central claim is that nature finance keeps running aground because three concepts get conflated: economic value, financial cashflow, and investibility. A trillion-dollar valuation of an ecosystem service tells you nothing about whether any portion of that value can be captured as contracted, predictable, attributable revenue. Most of nature’s value sits in public goods and avoided losses, which markets cannot finance.
She uses the Rhino and Spekboom bonds as worked examples. Rhino is donor-funded outcome payments wrapped around an IBRD bond, with investor returns ranging from 1.1% to 2.9% IRR depending on rhino population growth; Spekboom moves to contracted carbon revenue (a 14-year, $120m structure with a 2.41% floor and 5.078% potential, anchored by Amazon’s ERPA and Verra VM0047 ARR). The shift from Rhino to Spekboom is meaningful: from socialised donor risk to a contracted revenue stream. But her own honest reading lands squarely on the limits. Six years of design phase and roughly $6.7m of GEF and co-financing went into making rhino outcomes legible enough to bond. Spekboom only works because three decades of carbon market institution-building sit underneath it. Outside carbon, that infrastructure does not exist, and her conclusion is that in 2026, the only financeable route is still essentially carbon.
I think that is correct as far as it goes. What it underweights is that even carbon, as a UK route-to-market, has cashflow problems serious enough to disqualify it as the backbone of an institutional natural capital product, and that finding ought to accelerate the case for an alternative.
Carbon’s specific failure modes for UK fund duration
Look at what UK woodland carbon actually delivers as a cashflow profile. The Woodland Carbon Code requires verification at year 5, then year 15, then every 10 years thereafter, with carbon units released over a project life that runs to 100 years for native broadleaf. A natural capital fund with a 15 to 20 year life cannot hold those cashflows to maturity. Either it exits at year 12 to 15 into a thin secondary market at a meaningful discount (because the buyer is taking residual delivery risk on units not yet verified), or it has to securitise forward, which requires the offtake architecture Conway describes Spekboom as having engineered. UK Woodland Carbon Units do not sit in that kind of architecture. Tom Gegg has written extensively about these challenges.
The lumpiness effectively compounds the duration problem. A native broadleaf scheme generates near-zero verified units for the first 5 to 15 years, then significant releases at validation events, then long quiet periods. That is the worst possible profile for a fund needing distribution yield to LPs on something resembling a normal infrastructure schedule. The discount required to clear that into a secondary market is what makes the all-in IRR for UK woodland carbon look much worse than the headline £/tCO2 price suggests.
Peatland is worse on a further different dimension. The Peatland Code is treated as avoidance rather than removal, which makes it ineligible under SBTi’s mitigation hierarchy. Demand collapses to a narrower buyer pool: companies acting on brand or beyond-value-chain mitigation grounds, not companies meeting science-based targets. Most large UK peatland is in this category, so a sizeable share of the natural capital that ought to be financeable runs into a structural demand ceiling. This should and in my view must change - but there is a long way to go to win this particular argument.
The fungibility problem is arguably the deepest of the three. UK Woodland Carbon Units sit in a domestic registry with no convertibility into Verra, Gold Standard, or Article 6 markets. A Microsoft or a Shell looking for a tonne of CO2 removal can buy in Africa, Latin America, or Southeast Asia at materially lower cost. The premium UK buyers are paying is essentially a brand premium for British nature, which is real but discretionary, fragile, and politically exposed. Recent HM Treasury and DEFRA signalling on the natural capital subsidy envelope makes the demand side look much less stable than the methodology layer suggests.
So Sarah’s “carbon is mature” line is technically true at the methodology layer and substantively false at the financeable cashflow layer for UK projects. The architecture exists, but the bankable revenue does not, at least not in the form a UK 15 year fund can hold.
BNG inherits all the credit-form structural problems and adds British ones
BNG is the closest thing the UK has to a compliance market. Statutory demand, regulator-defined units, 10% net gain requirement, 30-year management, etc. On paper it should be different to voluntary carbon. The Robertson, Lave and Doyle work on US stream credits is the framework I want to use to test that.
After 15 years of federal Clean Water Act regulation, billions of dollars of stream restoration spending, and statutory buyer obligation, the active US stream credit market in 2019 covered 7.18 million linear feet, which translates to roughly 33,000 acres of real estate. Supply was concentrated in 25 of 37 Corps districts, with 97% of credits and 77% of stream banks in just 10 districts in a coastal arc between Delaware and Texas. Compare this to the relatively stable wetland market, which had 561,000 acres of credits available in 2019 across every Corps district. Statutory demand was not sufficient to scale stream credits. The market grew slowly, concentrated geographically, and stayed contested at the metric level.
The reasons that market underperformed are structural to the credit form, not transitional, and Robertson, Lave and Doyle kindly document them carefully. The credit unit is constructed from algorithms, concept curves, ratio multipliers and Excel lookup tables that are scavenged from non-capitalist scientific traditions and combined in ways that violate basic mathematical principles (adding ordinal scores, multiplying dimensionless indices). The ratios that regulators impose, 2:1, 3:1, sometimes 10:1, exist precisely because nobody trusts the underlying measurement. The geographic problem is intractable: scientists insist on watershed-scale integration, while credit producers insist on parcel-level responsibility. The system “fails at the boundaries”, as one of their interviewees put it. Arguably BNG inherits all of this and adds three “British”-specific issues.
The first is avoidance behaviour. Stream banking in the US works (to the extent it actually does) partly because Clean Water Act §404 permitting makes avoidance genuinely difficult. UK BNG has the same stated mitigation hierarchy on paper, but in practice developers and consultants have learned to redesign schemes to minimise BNG units required, score onsite delivery aggressively, and treat offsite credit purchase as the absolute last resort. DEFRA policy direction since the late 2025 consultation has if anything signalled modest uncertainty from the original ambition. The “compelled buyer” Conway and Robertson both identify as the precondition for credit market function is not actually compelled in BNG in the way the regulation reads.
The second is the ecological critique. The Statutory Biodiversity Metric scores habitat distinctiveness and condition uplift, with multipliers (often 2:1 or 8:1) arguably doing exactly the political work Robertson describes: managing the gap between what science can defensibly say about biodiversity and what regulators need to be able to enforce. Many ecologists would insist that this is not a measurement of biodiversity gain, rather it is a regulatory accommodation that serves political feasibility. That accommodation is honest about what credits can do; it is perhaps also a permanent ceiling on what they can become.
The third is the 30-year enforcement problem. The management obligation runs longer than most UK fund lives but shorter than the ecological timescales the credits claim to address. Section 106 and conservation covenant enforcement depends on local planning authorities that are not (yet?) resourced for it. Funders are pricing in real tail risk that the obligation will not be enforced, which could undermine additionality and therefore unit integrity.
The aggregate effect is that BNG, like US stream credits, will probably produce a small, geographically concentrated, contested market, but not the institutional asset class some of the early commentary suggested and many still hope for.
The European wetlands report says the same thing in different language
The LIFE Biodiv CrEW work is worth engaging with. They screened more than 90 voluntary biodiversity credit schemes globally, narrowed to 8 mature candidates, applied a 60+ parameter assessment framework grounded in BCA, IAPB and WEF principles, and selected two for pilot testing in EU wetlands: Verra’s SD VISta Nature Framework and the Wallacea Trust methodology. Both use multimetric basket-of-metrics approaches. Both are flexible, scientifically rigorous, and well-supported.
The conclusions section is the part that matters for this piece. Their key learnings include that most existing schemes are implicitly designed for large, intact landscapes in biodiversity-rich regions outside Europe and require adaptation to Europe’s fragmented land ownership and regulatory environment. They flag that voluntary markets face structural constraints in meeting high-integrity requirements, naming the voluntary nature of the market itself, the absence of stable and predictable financing pathways, limited integration with existing regulatory frameworks, and uncertainty around long-term stewardship obligations. Their explicit conclusion is that achieving high-integrity outcomes depends on credit scheme design alongside complementary policy frameworks, long-term financing mechanisms, and supporting institutional arrangements.
A careful European Commission-funded report, written by people who clearly genuinely want voluntary biodiversity credits to work, has reached the conclusion that voluntary biodiversity credits cannot themselves carry the financing weight, and that they sit at best as one input into a larger architecture that has to be built around them.
Marine adds complexity that exposes the limits even more sharply
The marine world inherits every stream and wetland problem and amplifies them. The boundary problem becomes structural rather than awkward. There is no parcel boundary in marine because the seabed is held by the Crown Estate, aquaculture leases are 25 to 30+ years rather than freehold, and currents, tides, larval drift and nutrient flows operate at scales of tens to hundreds of kilometres with no analogue to even a HUC-12 watershed. And sadly Marine Net Gain feels likely beyond the life of our current parliament.
Property rights are weaker than terrestrial, which hits financeability directly. A 30-year debt instrument anchored in a 25-year seabed lease has a duration mismatch that any structuring has to engineer around. The metric problem is genuinely harder rather than just less mature. The deck describes OSL’s 13-year Plymouth dataset as exceptional, which is the relevant signal: if the best-evidenced site in the UK is 13 years of monitoring at one location, the rest of the sector is at T0 or pre-T0, operating on the kind of conceptual curves and educated guesswork Robertson identifies as the dominant feature of US stream metrics.
Buyer archetypes might also be weaker than they appear when stress-tested against Sarah Conway’s “defined payer” criterion. Water companies are the strongest, because OFWAT creates structural demand through AMP-allocated capital, but marine pathways are not yet recognised in PR24 outcome delivery incentives. Infrastructure developers (offshore wind, ports) are project-by-project demand, bespoke and not yet a market. FMCG buyers remain voluntary, brand-driven, and subject to exactly the same SBTi-validation, consumer sentiment and CFO-budget risks that have caused voluntary carbon offset purchasers to retreat.
The wider ocean finance picture arguably sharpens the point. The WEF ‘subprime ocean’ framing from Bergenas and Singh argues that ocean-dependent businesses are running 2008-style mispricing, with $7.3 trillion flowing to nature-negative activities in 2023 against $220bn supporting nature-based solutions. The Katapult Ocean/Builders Vision/Dealroom Ocean Gap report quantifies the institutional firepower at roughly $100trn against publicly disclosed blue mandates of $69bn, less than 0.1%. Impact capital sits at $1.57trn with under $16bn in ocean deals. The capital is not absent, it’s just that the contractible cashflow architecture seemingly is.
The GFI Revenues for Nature marine guidebook is doing useful work cataloguing what does work: user fees, payments for ecosystem services, blue bonds, debt-for-nature swaps. But the honest reading of that catalogue is that almost none of these are credit-based instruments. The successful models are contracted revenue arrangements with regulated or statutory counterparties, blended with public capital. That convergence on contracts (not credits) is the part of the picture that should be loudest.
The architecture that actually fits institutional capital
Pulling these threads together, the structurally different proposition is contracted resilience cashflow. I want to be specific about what that means and why it solves the problems credits cannot. The product is a Resilience Purchase Agreement: a 10 to 15 year contracted, inflation-linked payment from a regulated, statutory, or creditworthy corporate counterparty, in exchange for delivery of a defined environmental outcome (water quality at a defined catchment monitoring point; flood attenuation at a defined gauge; sediment load reduction; biodiversity uplift on a defined site). The deliverer carries delivery risk and prices the contract against a counterfactual cost (a treatment works the water company would otherwise build; an engineered flood defence the RMA would otherwise commission; a mitigation cost the developer would otherwise pay). The contract is then aggregated, packaged, and held as an institutional-grade infrastructure asset.
The change in payer is structural. Sarah Conway’s “defined payer” criterion is satisfied weakly by voluntary corporate carbon buyers and BNG developers, but is satisfied strongly by a regulated water company funding upstream interventions through AMP-allocated capital, because Ofwat’s outcome delivery incentives make the spend non-discretionary. The same logic applies to a flood RMA discharging a statutory duty, an offshore wind developer meeting habitat regulations conditions, or a local authority delivering an Environment Act duty. The buyer is not choosing to pay; they are choosing how to pay, and the contract is one option versus a more expensive engineered alternative.
The metric problem becomes tractable because the contract prices the avoided cost rather than the ecosystem service. Robertson at al.’s central observation is that environmental credit markets fail because the unit measures something the science cannot defensibly support (wetland function, stream condition, biodiversity uplift). An RPA can sidestep this by pricing the counterfactual. The water company is not buying ‘X units of nutrient removal verified to a recognised methodology’, it is buying ‘Y kg/yr of phosphorus reduction verified at a defined catchment monitoring location, paid as £/kg-P-removed against a reference baseline’. That is closer to a performance-based service contract than to a credit. The verification problem becomes an outcome-monitoring problem, still hard, but bounded to a measurable in-stream, in-aquifer, or at-asset metric rather than to a constructed habitat unit.
The duration matches institutional capital. Pension funds and insurers have 20 to 30+ year liabilities they need to match with long-duration, inflation-linked, low-correlation assets. AMP cycles run on five-year regulatory periods but provide visibility on 15 to 25 year capital programmes. Flood RMA duties are statutory and indefinite. Offshore wind operational lives are 25 to 35 years. The cashflow profile available from contracted resilience is closer to a regulated utility revenue stream than to a credit issuance.
The asset is off-balance-sheet for the counterparty. A water company buying upstream nutrient services through an RPA records opex, not capex. Compared to building a new tertiary treatment works at £50 to 100m capex with 25-year asset life, the RPA is a smaller, more flexible commitment in which the counterparty (the originator-deliverer) carries delivery risk. That asymmetry is what makes the RPA compete on price even before any ecosystem co-benefits are counted.
What the worked example looks like
The Evenlode Landscape Recovery project is the cleanest live example I can point to of this type of architecture. Its total economic value is built up from carbon credits, BNG units, water quality contracts, and water quantity contracts. Read in the framing above:
The water quality and water quantity contracts are the institutional core. Long duration, regulated counterparty, non-discretionary demand, measurable at the monitoring point.
Carbon and BNG are credit layers stacked on top. Smaller, lumpier, politically exposed, but still real.
The investor or DEFRA underwrites the contracted resilience cashflow as the base case. Carbon and BNG add yield rather than determining viability. That decouples the project from the fragility Conway and Robertson et al. document in credit markets. A fund holding Evenlode is not betting on the Woodland Carbon Code clearing in 2031 or on BNG demand surviving the next Defra consultation; instead it is holding a long-duration regulated payment with optional upside. The validation signal will be that institutional capital can commit to this shape.
Where this leaves the bigger, strategic question
The strategic question for UK natural capital doesn’t therefore have to be ‘how do we mature the credit markets?’ The credit markets will mature on whatever timescale they will, and the LIFE Biodiv CrEW conclusion (that voluntary credits remain early-stage and require complementary policy frameworks and long-term financing mechanisms) tells you that timescale is likely to be long and the ceiling rather low.
The right strategic question is ‘where else can we find non-discretionary regulated demand that an RPA can price against?’ The answer is larger than the discussion sometimes acknowledges, and it includes the obvious candidates (water utilities through AMP, flood RMAs through statutory duty, BNG when developers cannot avoid offsite delivery, etc), but also several that have not had attention they deserve:
Marine Net Gain, if it materialises with statutory teeth, would create a marine analogue to BNG with the same RPA logic.
Nutrient neutrality with proper enforcement creates regulated catchment-level demand for phosphorus and nitrogen reduction.
Ofgem-incentivised network resilience spend on flood and storm protection at substations and grid assets.
MOD, NHS, HE/FE estate climate adaptation and water management, where statutory duty and budget visibility combine.
Crown Estate and TCE Scotland leasing conditions for offshore wind and aquaculture, which can create developer-side mitigation obligations.
Each has the same shape, namely a regulated or statutory payer, a defined exposure, a measurable outcome, and a counterfactual cost the natural capital intervention can beat. Where all three line up (creditworthy counterparty, measurable outcome, cost-competitive intervention) the architecture works now, not in five to seven to ten years.
What to do with the credits, then
I am not arguing for abandoning credits, but I am arguing that we should discuss putting them in their right place in the overall structure.
Credits work best as upside layers stacked on a contracted base. Woodland Carbon Code and Peatland Code units issued from a Landscape Recovery scheme that already has water quality and water quantity contracts running can be sold as additional yield, with the fund’s IRR not depending on them clearing. BNG units issued from sites already under conservation covenant for water or flood reasons sit in the same bucket. The credit layer becomes icing rather than the cake.
Credits also work as a development-finance tool, sitting alongside grant capital, to bridge from project inception to the point where a contracted RPA can be priced. That is what the LIFE Biodiv CrEW pilots are functionally testing: whether voluntary biodiversity credits can finance the early restoration work that proves out the catchment-scale outcomes a regulated buyer might later contract for.
Where credits struggle to work, on the above reading, is as the underlying for an institutional fund. The duration mismatch, the metric instability, the demand fragility and the boundary problem all argue against treating them as the financeable backbone. Sarah Conway’s translation diagnostic was almost certainly right. The next step is to be honest about where the translation actually lands.
For UK natural capital practitioners trying to mobilise institutional capital at scale, the work for the next 24 months is identifying counterparties, structuring contracts, and packaging cashflows in a form pension funds and insurers can hold. Credits will be useful where they are useful, but the financeable architecture sits underneath them.
References
1. The Rhino Bond Sarah Conway, Risk & Resilience Ledger (Substack article), 31 Mar 2026.
2. Nature Is Not a Trillion-Dollar Asset Class Sarah Conway, Risk & Resilience Ledger (Substack article), Apr 2026.
3. The Spekboom Deal Is Not Just a Bond Sarah Conway, Risk & Resilience Ledger (Substack article), 28 Apr 2026.
4. Making a market in environmental credits I: Streams of value Robertson, Lave & Doyle. Peer-reviewed paper, Environment and Planning E: Nature and Space, Dec 2023. https://doi.org/10.1177/25148486231151810
5. Making a market in environmental credits II: Watershed moments Robertson, Lave & Doyle. Peer-reviewed paper, Environment and Planning E: Nature and Space, Dec 2023. https://doi.org/10.1177/25148486231151807
6. Biodiversity Certification and Credit Systems for European Wetlands LIFE Biodiv CrEW (NABU, ELO, Eurosite, aeco, Sylva, bloomlabs), EU LIFE Programme. EU-funded benchmarking report, Feb 2026.
https://www.bloomlabs.earth/
7. Finance Deep Dive: Overview & Opportunities (Regenerative Aquaculture Forum) Rebalance Earth. Industry forum deck, 2026. rebalance.earth
8. The ocean is now a subprime asset. Only a sustainable blue economy will calm the waters Johan Bergenas & Shashank Singh, WWF / World Economic Forum. Opinion article, 22 Apr 2026. https://www.weforum.org/stories/2026/04/ocean-sustainable-blue-economy/
9. The Ocean Gap Katapult Ocean, Builders Vision & Dealroom. Investment research report, 2025. https://oceanreturns.com/the-ocean-gap/
10. Revenues for Nature Guidebook: Revenue Models and Financing Mechanisms for Protecting and Restoring Coastal and Marine Ecosystems Green Finance Institute Hive (R4N programme). Practitioner guidebook, Apr 2026. https://www.greenfinanceinstitute.com/wp-content/uploads/2026/04/R4N-GUIDEBOOKS-MARINE.pdf
11. Navigating Ocean Risk: Value at Risk in the Global Blue Economy Kennedy, Xu et al. Research report, 2021 (cited in The Ocean Gap). https://www.researchgate.net/publication/355844065
12. State of Finance for Nature 2026 UNEP. Policy report, 2026 (cited in WEF article for the $7.3trn / $220bn figures). https://www.unep.org/resources/state-finance-nature-2026
13. World Bank prices 14-year Spekboom Restoration Outcome Bond in South Africa World Bank / IBRD. Press release, 23 Apr 2026. https://www.worldbank.org/en/news/press-release/2026/04/23/world-bank-prices-14-year-spekboom-restoration-outcome-bond-in-south-africa
14. Imperative closes $91m Phase 2 blended financing Imperative Inc. Company announcement, Apr 2026. https://www.imperativeinc.com/media/imperative-closes-91m-phase-2-blended-financing/
15. Five rules for scientifically credible nature markets zu Ermgassen, Swinfield, Bull et al. Peer-reviewed paper, Nature Ecology & Evolution, 12 Jan 2026 (cited in EU wetlands report). https://doi.org/10.1038/s41559-025-02932-z