Post date 31 Aug 2026 (original LinkedIn date)
Original LinkedIn version· 1,325 words
“Islands in the stream“ by Dolly Parton & Kenny Rogers, released in 1983 on Rogers’ album, ‘Eyes that see in the dark’, written by the Bee Gees and originally intended for Marvin Gaye in R&B style. RIP Dolly
I love it when you come across a post that picks up on work that you had missed (we can’t see everything) and that piques your interest. Anthony Hobley, formerly of Howden, posted last week about two documents published this year that between them explain a great deal about why adaptation finance is stuck. One is Dr Madhu Acharyya‘s catastrophe risk modelling handbook, and the other is ‘The Legibility Problem Across Real Assets’ by Dr Sophie Taysom and Patrick Schmucki . Anthony’s observation was that sadly neither appears to have been aware of the other and neither cites the other - he called it a Translation Gap.
Full credit to Anthony for picking up on these important contributions, and for sending me off to read the underlying papers, because having now read them (alright, skimmed one, but we’ll get to that!), plus two further working papers Madhu released earlier this year, I think the gap might run even deeper than Anthony’s post suggests.
Let’s start with the handbook: twenty-three chapters of proper apparatus, covering stochastic event catalogues, vulnerability functions, exceedance probability curves, Tail Value at Risk (it’s not a light read - so skimming acceptable!). There’s a whole chapter on measuring value that concentrates in rare, severe events, which happens to be exactly where I think adaptation’s value sits. The book offers eight learning pathways for different readers, from underwriters, to actuaries, modellers, brokers, and regulators, but noticeably and sadly not one for the people who own the assets. No pathway for a pension trustee, a valuer, or an analyst trying to turn catastrophe modelling into a forward view of what a building or a port might be worth, tucked away in an infra sleeve in their portfolio of real assets.
Sophie and Patrick write for precisely those readers, and they’re pretty blunt about what’s missing. Their fifth institutional condition, one to which I keep coming back, is a valuation framework calibrated to where adaptation value actually sits. Conventional expected-loss models understate adaptation because its benefit concentrates in tail events, like the Valencia floods, the LA wildfires, and the recent horrific flooding tragedy in Nepal & Tibet, and average-loss frameworks discount most heavily in exactly that range. They list this absence as one of five conditions blocking adaptation and resilience finance at scale.
Nothing new here, all very consistent with Anthony’s post, but here’s the part that surprised me. Madhu has also recently released an applied case study: his integrated risk framework run over a hypothetical Amazon investment, with a capital structure, WACC, scenario analysis, liquidity stress testing, a 10,000-run Monte Carlo, VaR, real options, etc – all the nuts & bolts one would expect. It’s thorough, and the real options section genuinely adds value (it’s something we’ve been toying with at Rebalance Earth for a little while) - managerial flexibility adds about 19% to expected project value in his model and turns the fifth-percentile outcome from a heavy loss into a gain.
But look at what ‘uncertainty’ means inside that model: revenue growth is a triangular distribution between 8% and 12%, while operating costs sit between 80% and 84% of revenue. WACC gets a normal distribution with a 0.8 point standard deviation, and every input is a gentle riff around the base case, and the model duly reports that all 10,000 simulations produced positive cash flow in every single year. Nothing in it can jump, so nothing does (even tho that doesn’t feel like the real world).
Has the author of the catastrophe modelling handbook, writing a valuation study, left his catastrophe apparatus at home?! I tease! So maybe the translation gap isn’t just between two sets of authors who haven’t read each other, perhaps it resurfaces inside one author’s own body of work, and maybe that’s a pointer to a structural issue. Why structural? Because the insurance sector’s output format is annual. Average Annual Loss, the 1-in-100, the 1-in-200, a table literally called a Year Loss Table. Everything is annualised because the underlying instrument (the policy) is annual. The maths could go further out, but the contract has never asked it to (altho we are starting to see 10-year policies in the world of natural capital).
Meanwhile forward price curves are routine for all commodities, from power to gas, oil, freight, and coffee. But physical climate risk, the one exposure we know for certain is non-stationary, has no forward curve at all. The raw material exists. AAL, Average Annual Loss, or the expected loss on an asset or portfolio per year, averaged over the full probability distribution of possible events is already expressed in currency; it is already a price. Run the same asset through climate-conditioned event catalogues at 2030, 2035, 2040, 2050 and you have a forward physical cost curve. One caveat, though, and arguably quite an important one: a forward curve of averages inherits the very flaw Sophie and Patrick identify, because averages bury the tail. What a valuer actually needs is the tail term-structured ie the 1-in-100 and TVaR at each horizon, not just the mean. Cat models produce those numbers today, but nobody packages them for a valuation.
And valuers now need them. Since 30 April this year, RICS expects members valuing commercial property worldwide to assess and document climate-related factors with evidenced, asset-level substantiation. The profession is required to form a view its own toolkit can’t yet support. Why do I care about this at Rebalance Earth? Because once a forward physical cost curve exists, an adaptation intervention becomes something finance can finally see: the thing that shifts the curve. A restored wetland upstream of a business park, a rewiggled river, an oyster reef protecting a port, each changes the hazard, which changes the event catalogue, which changes the curve and the spread between the with-intervention and without-intervention curves is the counterfactual number that today does not exist. A quantified, term-structured stream of avoided losses now looks a lot like a cash flow, and cash flows can be financed.
It also answers the interdependence problem S&P rightly flag. Flood resilience for one property is near-worthless if the drainage around it fails, so rational owners wait for their neighbours to move first. That’s an argument for financing adaptation at the scale where the physics operates, ie the catchment, the coast, the ‘flowscape’ as I’ve recently heard it called, which is exactly the scale nature-based interventions work at. A catchment can’t ‘free-ride’ on itself.
There’s even a bridge on the instrument side, from within the sector. Acharyya’s other current draft simulates parametric trigger designs (Cat-in-a-Circle, Cat-in-a-Box, hybrid polygons, terms that boggle my little mind and that I’d never previously heard of!) that pay on the physical event rather than the indemnified loss. A trigger’s activation probability is a model-derived number, an intervention that demonstrably attenuates the hazard shifts that probability, and the shift can then be priced. Pointed forward rather than just annually, the sector’s own toolkit gets you most of the way there.
Anthony is right that legibility and translation are complementary diagnoses, but no amount of definitional agreement produces a counterfactual number. However, an instrument does, and our work at Rebalance Earth, treating nature as infrastructure, paid for by the beneficiary businesses and communities it protects, could hugely benefit from this curve pulled from the risk modelling world, and equally the risk modelling world needs asset owners as readers too. Time for us all to speak?
References:
Acharyya, M. (2026), catastrophe risk modelling handbook and RRCL, plus working papers 2026-10 and 2026-11;
Taysom, S. and Schmucki, P. (2026), The Legibility Problem Across Real Assets, Keyah Consulting and Resilion;
RICS (2026), ESG and Sustainability in Commercial Property Valuation, 4th ed.