Post date 4 May 2026 (original LinkedIn date)
Original LinkedIn version· 2,274 words
“Ain’t no mountain high enough” written by Nickolas Ashford and Valerie Simpson in 1966, released in 1967 by Marvin Gaye and Tammi Terrell, and made famous by Diana Ross of Supremes fame in 1970.
This is a story about how we attribute value to things in the future or things that will run into the future. For something like fifteen years now, anyone who has wandered into a serious conversation about the economics of nature has ended up in the same place - the discount rate. Stern versus Nordhaus; whether 1.4% is defensible; whether 5% is defensible; whether either number can really be doing all the work being asked of it. Much ink has been spilled - the literature is enormous and the conferences are exhausting. The progress, if you measure it by what has actually changed in how capital is allocated to nature, is sadly modest at best.
I want to suggest that this is partly because the discount rate fight, although intellectually rich, has been the wrong fight. Or rather, it has been the right fight conducted at the wrong level of depth. The issue the debate has been circling, that standard finance machinery does something strange when applied to nature, is clearly real. But the discount rate isn’t where the strangeness lives - it’s a symptom, not a cause, and treating it as the cause has produced a literature that argues itself in circles while the underlying problem goes unaddressed.
There’s been a quiet shift over the last few years that, I think, finally puts the argument on a more useful footing. I think it’s worth walking through.
The standard story, briefly
The Ramsey formula that underpins almost all long-horizon benefit-cost analysis decomposes the discount rate into pure time preference and a wealth effect: r = δ + ηg. Pure time preference is how much you discount future welfare just because it’s in the future. The wealth effect captures the fact that if future generations are richer than us, a pound to them is worth less than a pound to us. It played out first in climate circles. The whole Stern-Nordhaus argument was essentially a fight about the right values for δ and η, with Stern picking values that produced a low rate and Nordhaus picking values that produced a higher one.
This formulation has one very large hidden assumption. It treats consumption as a single composite good. It assumes that whatever future generations are consuming, more of it is more, and the relative prices of the things inside that bundle don’t shift in any way the modeller needs to worry about. Stern and Nordhaus disagreed about a lot of things, but they agreed on this and so did almost everyone else.
This assumption is fine when the goods inside the bundle are reasonably substitutable. Steel and concrete, computing power and labour, or heating and air conditioning. Within some range, you can trade one for another, and the relative prices don’t move dramatically over time. The single-rate Ramsey framework copes.
It is not fine when one of the goods inside the bundle is non-substitutable, becoming scarcer, and serves as an input to the production of the other goods. Think of ecosystem services. The framework was never built for them, and the discount rate debate has been a long, frustrating attempt to make a single-good formula handle a two-good problem by adjusting the parameter values rather than fixing the formula.
What Gollier actually did
Christian Gollier’s contribution, going back to a 2010 paper called Ecological Discounting and developed in subsequent work, was to point out that you can fix the formula directly, and that doing so produces a result the older debate had been struggling towards.
The fix is straightforward in its logic, even if the maths takes some work. If you have two goods that aren’t perfect substitutes, call them produced capital and ecological services, then you have two growth rates. Produced capital is becoming abundant and its growth rate is positive and substantial, roughly tracking real GDP per capita at perhaps 2% per year. Ecological services are stagnant or declining and their growth rate is zero or negative. Run those two growth rates through the Ramsey logic separately and you get two discount rates, not one. Gollier’s calibration suggested something around 3.2% for consumption and 1.5% for nature.
This is not a rejection of standard welfare economics, rather it’s standard welfare economics applied properly to a two-good world. The earlier debate had been doing it wrong, not because anyone was being careless, but because the simplifying assumption of a single composite good had been baked in so deeply that nobody bothered to question it.
France has actually adopted differential rates for public investment appraisal partly on this basis. The UK Treasury Green Book’s declining long-run rate schedule is a cousin of the same idea, although less explicit about why. The theoretical case is, by this point, settled and the remaining questions are really about calibration and implementation.
The implementation step
The piece of the argument that, for me, properly closes the loop arrived in 2024. Drupp, Hänsel, Gollier, Sterner and a dozen other economists published a short Science paper translating Gollier’s theoretical insight into a policy-ready rule that governments could actually adopt without rewriting their entire cost-benefit framework.
The trick was in many ways political as much as economic. They observed that you can either change the discount rate for ecosystem services (theoretically clean, but politically explosive because it opens up a thousand other questions about discounting practice) or you can keep a single discount rate and uplift the willingness-to-pay for ecosystem services over time. These two operations are mathematically equivalent as they’re the same adjustment, just relocated to different sides of the equation. But the second is far easier for a Treasury function to swallow, because it slots into existing guidance the way agencies already adjust the value of a statistical life or the value of travel time as incomes rise.
The rule is simple. Future willingness-to-pay for ecosystem services should rise at a rate of ξ × (g_C − g_E), where g_C is the growth rate of consumption, g_E is the growth rate of the ecosystem service, and ξ is the income elasticity of willingness-to-pay. Their proposed default sets ξ at 1, which means WTP for ecosystem services rises one-for-one with income, faster when ecosystems are declining.
The numbers this produces are not particularly subtle. Under their default, holding ecosystem services constant, the present value of those services over a century rises by 131% relative to current government guidance. On a threatened-species trajectory, by 188%. On a Living Planet Index trajectory of biodiversity decline, by over 1,200%. Those aren’t sensitivity analyses, instead they’re a statement that current guidance is, by the authors’ own calibration, undervaluing scarce ecosystem services by something between a factor of two and a factor of thirteen.
The social cost of carbon estimates that drive climate policy in dozens of jurisdictions are, on this analysis, undervalued by more than 50% just from the failure to account for relative price changes.
What this means for the rhetorical version of the argument
There’s a more tricksy version of all this that has been doing the rounds, most recently in an excellent piece by Dr Roger Iles arguing that we shouldn’t apply a discount rate to nature at all because nature is an appreciating asset. As provocation, I personally think this is spot on. But sadly I fear that as implementation guidance, it risks getting stuck, because finance does need some time-value framework, and ‘zero’ is arguably just another arbitrary number.
What Gollier and Drupp give us can be thought of as a more durable version of the same intuition. Nature is, in real terms, an appreciating asset, but the appreciation shows up in the cash flow projection, not in the discount rate. Standard practice projects flat real cash flows for nature-related assets and then argues about what rate to apply. My contention is that we can project rising real cash flows that reflect scarcity, and apply the same discount rate you’d apply to anything else. Mathematically equivalent, institutionally far easier, and it stops the conversation getting stuck in a debate about ethics-versus-economics that risks going nowhere.
Where this still doesn’t reach
But I want to be honest about the limits of this framework, because I think the limits matter, and they matter in a way that has implications I’ll want to come back to.
The Drupp rule is built on smooth substitution. It handles “pollination services are getting scarcer at minus zero point four percent per year” beautifully. It does not handle the case where pollination services in a given region collapse to something close to zero because a tipping point has been crossed. It does not handle ecosystem state changes, regime shifts, or the sort of non-linear loss the IFoA Planetary Solvency report has been quantifying for the last few years.
This isn’t a fatal flaw, but it does mean the framework is right for most nature-related decisions and wrong for the most consequential ones. For tipping-point risk, you need a different approach, one built on precautionary targets, safe minimum standards, cost-effectiveness analysis bounded by tolerable worst-case outcomes. The Drupp rule handles the smooth zone, but we need something else to handle the cliff edge. Both are needed, and combining them properly is, I think, where the next decade of progress actually sits.
What investors should take from this
Is this just theoretical niceties? I would argue not. For a working investor or asset owner, the discount rate debate has been frustrating because it never really translated into anything actionable. Pick a number, defend it, sensitivity-test around it, and the answer doesn’t change much.
The Gollier-Drupp framing changes that, if you’re willing to take it seriously. It says: stop arguing about the rate, start interrogating the cash flow projection. If you’re modelling an asset whose underlying economics depend on ecosystem services, which, depending on how broadly you define it, is most of the real economy, then projecting flat real cash flows is the same as assuming nature is substitutable, which is the same as assuming the underlying scarcity dynamics aren’t real. That’s a position, and it might even be the right position for some assets, but it should be defended explicitly, not assumed by default.
For long-horizon allocators, the implication is far sharper. If your liabilities run thirty or fifty years into the future and the assets meant to discharge them have nature exposure, then the cash flow projection is doing the work that matters most for matching duration to obligation. Getting that projection wrong, by assuming away scarcity premia that are coming whether you model them or not, produces an asset that looks well-matched to liabilities on the surface and is poorly matched in substance.
I think that this is the part the discount rate debate kept missing. The argument was always supposed to be about how to value the long-term properly. By focusing on the rate rather than the projection, it ended up missing where the nub of the long-term value actually lives.
Where this leaves the conversation
I think/hope that the discount rate debate is, for practical purposes at least, over. Gollier won the theoretical argument and Drupp et al. wrote the implementation guide. Governments are slowly adopting it, much too slowly. So the remaining contention is really about the speed of institutional adoption, not substance.
But there’s a second, larger question hiding underneath the discount rate one, and it’s the question I’d rather investors and asset owners were arguing about now. And that’s the discussion I will take up next.
References
Core sources for the argument
Drupp, M.A., Hänsel, M.C., Fenichel, E.P., Freeman, M., Gollier, C., Groom, B., Heal, G.M., Howard, P.H., Millner, A., Moore, F.C., Nesje, F., Quaas, M.F., Smulders, S., Sterner, T., Traeger, C., and Venmans, F. (2024). “Accounting for the increasing benefits from scarce ecosystems.” Science, 383(6687), 1062–1064.
Gollier, C. (2010). “Ecological discounting.” Journal of Economic Theory, 145(2), 812–829.
Gollier, C. (2019). “Valuation of natural capital under uncertain substitutability.” Journal of Environmental Economics and Management, 94, 54–66.
The Stern-Nordhaus debate and its lineage
Stern, N. (2006). The Economics of Climate Change: The Stern Review. HM Treasury, UK Government.
Nordhaus, W. (2007). “A Review of the Stern Review on the Economics of Climate Change.” Journal of Economic Literature, 45(3), 686–702.
Weitzman, M.L. (2007). “A Review of the Stern Review on the Economics of Climate Change.” Journal of Economic Literature, 45(3), 703–724.
Ackerman, F. (2007). Debating Climate Economics: The Stern Review vs. Its Critics. Report to Friends of the Earth UK, Global Development and Environment Institute, Tufts University.
Dasgupta, P. (2021). The Economics of Biodiversity: The Dasgupta Review. HM Treasury, UK Government.
Supporting context on planetary boundaries and tipping points
Jones, A., Bedenham, G., Goldman, M., Ranchin, A., Spencer, N., and Trim, I. (2026). Planetary Solvency: Tipping into the wild unknown — Global nature risk management. Institute and Faculty of Actuaries and Anglia Ruskin University.
Lenton, T.M., Milkoreit, M., Willcock, S., et al. (2025). The Global Tipping Points Report 2025. University of Exeter.
Richardson, K., Steffen, W., Lucht, W., et al. (2023). “Earth beyond six of nine Planetary Boundaries.” Science Advances, 9(37).
The critique that motivated the piece
Iles, R. (2026). “Are we discounting the future out of existence? A calling for finance.” LinkedIn article.
Policy and institutional context
HM Treasury (2021). Environmental Discount Rate Review: Conclusion. HM Treasury, UK Government.
NGFS (2023). Recommendations toward the development of scenarios for assessing nature-related economic and financial risk. Network for Greening the Financial System.