Somewhere in your pipeline is a deck calling something a hedge. Read it again and check whether it names the exposure. A hedge against what? Over what horizon? Triggered by what? Paying roughly how much when the exposure loses? If the deck can't answer those four questions, the word is decoration.
Hedge, diversifier, third leg, and protection are four different jobs. Most decks use one noun for all four — and nature decks are frequent offenders. This page separates the jobs so you can use the right word on the right line of an IC memo. We apply the same test to ourselves.
four jobs, four tests
| job | what it must do | the test it must pass | what it costs | it fails when |
|---|---|---|---|---|
| hedge | Pay when a named exposure loses | Name the risk, the horizon, the trigger, and the expected payoff | Carry, premium, or drag versus the loss avoided | The payoff is a narrative, or the position falls with the exposure instead of against it |
| diversifier | Move differently enough — for a causal reason — to improve the whole book | Name the driver of its returns and show the marks are honest | Fees, lockup, and the return you gave up for a different engine | The "different movement" is appraisal smoothing or stale marks, not a different cause |
| third leg | Fill the slot after stocks and bonds | A slot is not a test — whatever fills it must still pass the hedge test or the diversifier test | The sleeve you sold or never bought to make room | A third ticker gets counted as a third engine |
| protection | Shrink the underlying loss itself | Name the loss, the mechanism that reduces it, how the reduction is measured, and a horizon measured in years | Multi-year funding cost per unit of loss reduced — not a quarter of carry | It gets sold as a crisis payoff instead of a smaller crisis |
The first three are claims about how a position behaves. The fourth is a claim about the world. That difference is the rest of this page.
most hedges never touch the loss
Consider what the classic hedges actually do. A put option pays you when the index falls — the index still falls. A trend-following sleeve profits from a drawdown — the drawdown still happens. A catastrophe bond pays the sponsor after the hurricane — the hurricane still lands. In every case the loss occurs in full, and the instrument decides whose P&L absorbs it.
That's not a criticism. Rearranging who eats a loss is genuinely valuable — it's what risk transfer is for, and entire markets exist to do it well. But it has a ceiling: when the loss itself is physical, growing, and shared across your holdings, transferring it just moves the bill around a shrinking table. Someone still has to fund the work that makes the loss smaller. We've written elsewhere about why, at the whole-system level, there is no risk transfer — only risk reduction and risk holding.
A hedge rearranges who eats the loss. Protection shrinks the loss.
two claims, two lines of the term sheet
When someone asks whether ensurance is a hedge, the honest answer splits into two claims that belong on different lines.
Line one — financial hedge: a named payoff when a named exposure loses. This is a weak claim for ensurance today. Unless an instrument is parametric — a contract that pays on a measured trigger, like river stage or burned acreage — we do not call ensurance a hedge, and you should hold anyone who does to the four questions in the table. No named trigger, no named payoff, no hedge.
Line two — causal risk reduction: the allocation funds work that lowers the physical loss itself. Watershed function that attenuates the flood peak. Forest condition that moderates fire behavior. Wetlands that buffer storm surge. This is the strong claim, and it is a different kind of claim: it doesn't promise you a payoff in the crisis. It promises a smaller crisis. The horizon is years, not quarters. A single reporting period does not change a floodplain.
Keep the two lines separate — in your memo and in our pitch. Blurring them is how protection gets accidentally sold as crisis convexity, and how allocators end up holding a risk-reduction position while expecting a hedge's behavior.
this is not a cat bond
One section on this, because it's the most common confusion in the room.
Catastrophe bonds and the broader institutional event-risk market do a real job well: they pay a sponsor after an independent, modeled peril, from capital raised in advance, at spreads that have historically moved on reinsurance pricing more than on equity markets. The structure works because the perils are roughly independent events — this season's hurricane doesn't cause next season's — and the world resets after each one.
Ecosystem-tipping losses break both assumptions. They are correlated — a failing water system hits agriculture, property values, insurance availability, and municipal credit in the same region at the same time — and they can be irreversible. There is no reset after a crossed tipping point, which means there is no clean "post-event" state to price the next bond against.
So do not hold ensurance expecting a cat-bond payoff profile, long-volatility convexity, or a crisis lottery ticket. The closest honest crisis claim is regime resilience: a book whose physical dependencies have been funded and maintained enters a bad regime with smaller losses than a book that ignored them. Regime resilience is not the same thing as positive returns in a market drawdown — it's the difference between a bad decade and a broken one. If you've already sized your exposure and are weighing responses, the decision tree is in divest, reprice, or protect.
where nature actually fits today
Run nature itself through the four-job table and it splits in two.
Working nature — farmland, timberland — can be a real-asset diversifier candidate. The causal driver is genuinely different: biological growth plus demand for food and fiber, not the business cycle. But it carries risks the label doesn't advertise: commodity price exposure, leverage in many vehicles, operating risk (weather, disease, management), and illiquidity that flatters the marks. Name those in the memo. And note that working nature's return history belongs to working nature — pure protection does not inherit farmland's track record just because both involve land.
Pure protection — funding an ecosystem to keep doing what it does — is physical risk reduction, full stop, until a contracted stream exists. That means a payor — a utility, an insurer, a municipality, a downstream beneficiary — underwritten to pay for maintained ecological function. Once that contract exists, the flow can be evaluated as a diversifier by the same test as everything else: name the driver, check the marks. Until then, the honest word is protection, and the honest expectation is a smaller loss, not a coupon.
None of this is new work. Land trusts, easement holders, and public agencies have been doing physical risk reduction for over a century, and much of it has worked. What's new is the instrument: certificates make funded protection holdable — a specific claim on a specific natural asset, with proceeds routed to the stewardship that maintains it — rather than a grant that disappears into a fiscal year.
the identity that survives the test
There are two things an allocator can say after making this kind of allocation.
"I bought uncorrelated" is a claim about a statistic — one you don't control, usually measured on marks you have reason to distrust, and one that has a habit of converging to 1 in the exact quarter you needed it not to. Every allocator who said it in 2021 got to test it in 2022.
"I fund the thing that shrinks the loss" is a claim about a mechanism you can verify: this watershed, this function, this measured condition, this beneficiary who depends on it. It doesn't depend on a correlation matrix cooperating. The worst case is still capital at risk: the ecological outcome is not guaranteed, the loss may still land, and you can be down both the funding and the physical damage. That is an investment position with a named mechanism — not a floor, and not philanthropy. The portfolio-level version of this argument — that diversification fails when every holding shares one physical supplier — is in portfolio ensurance.
The second claim is also simply more precise. It names the job.
use the table in your next memo
The four-job table is meant to be lifted. Run every alternative deck through it — cat bonds, royalties, private credit, farmland, and ours:
- If the word is hedge — demand the risk, horizon, trigger, and expected payoff.
- If the word is diversifier — demand the causal driver and evidence the marks are honest.
- If the word is third leg — ask which of the first two tests the position actually passes.
- If the word is protection — demand the loss, the mechanism, and the measurement. And don't expect a crisis payoff; expect a smaller crisis.
The full version of the filter — including the three ways "uncorrelated" claims fail — is in the fine print of the free lunch. To see what a holdable protection claim looks like in practice, browse specific ensurance certificates. And if you're working out which of the four jobs nature could actually do in your book, talk it through with someone who won't call it a hedge unless it is.
When the job is named and you want the claim, not another definition: the position that funds the thing that shrinks the loss.
This is educational content on portfolio construction — not investment advice, and not an offer of any instrument.