risk first.
capital second.

reinsurance is often called insurance for insurers, but the real point isn’t the money. it’s understanding risk.

underwriting means figuring out what can go wrong and pricing it correctly. if that judgment is wrong, capital can get wiped out. having a bigger pool does not automatically make the risk better. it just means there is more money behind the same decision.

this matters for re protocol because onchain systems follow the same rule. you can see a vault balance. you can move a token. but neither tells you whether the insurance risk underneath was priced properly. that part still has to be done right.

risk first. capital second.

what is reinsurance underwriting?

underwriting begins with a practical set of questions: what is covered, what is excluded, how large could the losses become, and how much of that exposure should this balance sheet accept?

historical claims help. so do exposure data, actuarial models, and stress scenarios. but the past is an input, not a promise. a portfolio can look familiar while the cost of settling its claims changes. the underwriter has to decide how much confidence the available evidence deserves.

writing more of the second book would expand the problem. extra capital might provide breathing room, but it would not fix the price. a sound underwriting decision can therefore be a refusal: the risk may be understandable and still not worth accepting on the terms offered.

the structure matters as much as the price.

price is only one part of the contract. the structure determines which losses move to the reinsurer and which remain with the insurer.

facultative reinsurance evaluates an individual risk. treaty reinsurance covers a defined class or portfolio of business under agreed terms. separately, coverage can be proportional or non-proportional. these are different dimensions, rather than a single ladder of categories. [1]

with a quota share, premiums and covered losses are divided at an agreed percentage. if the reinsurer takes 30%, a simplified example gives it $30 of each $100 of premium and $30 of each $100 of covered losses, before commissions and other terms.

excess-of-loss coverage responds above a specified retention, up to a limit. in a simplified $10 million layer above $5 million, a covered $12 million loss would leave $5 million with the insurer and place $7 million in that layer. the actual contract controls how losses are counted and aggregated. [1]

the underwriting questions change with the structure. a proportional deal requires close attention to the economics of the underlying book. an excess layer requires particular attention to the chance that losses reach it and how far they travel through it. neither structure removes risk. each makes a different agreement about sharing it.

premium received does not mean profit earned.

there is another trap: treating cash received today as profit already earned.

insurance obligations develop over time. some claims have been reported but are not settled. others relate to covered events that have happened but have not yet been reported—known as incurred but not reported, or IBNR. reserves are estimates of what those obligations will cost. they can change as evidence improves. [2]

this makes reserve discipline part of underwriting discipline. a portfolio that looks profitable under an optimistic estimate may look very different after claims mature. the meaningful question is not simply how much premium has been written. it is how much remains after the obligations attached to that premium are properly accounted for.

that also changes how a reader should approach performance figures. which period do they cover? how mature are the claims? are they describing premiums written, premiums earned, or profit? a large number without those distinctions can create more confidence than understanding.

underwriting in re protocol

this is where re gets interesting. the capital comes through a different system, but the thing it backs is still an insurance obligation.

according to its documentation, re protocol sources capital through defi, while cover re spc ltd., its reinsurance partner, deploys collateral into real-world reinsurance treaties. the protocol and the reinsurer are distinct entities. the digital interface does not replace the institution making the insurance commitments. [3]

that separation is worth keeping in view. a smart contract can govern a token or route capital. evaluating the underlying insurance exposure still involves claims data, counterparties, contractual terms, and judgment outside the blockchain.

the current documentation also describes yield as a blend of onchain and offchain sources, with a token-specific spread. it would be too simple to describe every unit of token yield as a direct distribution of underwriting profit. [4] for a reader, the useful exercise is to trace each source of return back to the exposure that produces it.

who takes the loss first?

my original article described reUSDe as the first-loss layer. the current documentation makes an important distinction: re’s own capital is the junior buffer; reUSDe is mezzanine; reUSD is senior.

1. re capitalabsorbs underwriting losses first after premiums are insufficient to cover claims and expenses.
2. reUSDeabsorbs further losses if that junior buffer is exhausted.
3. reUSDis affected after the subordinated layers have been depleted.

this is the order described by re, rather than an independent guarantee that the buffers will be sufficient. [5] “senior” describes relative priority. it does not mean a position cannot lose money.

same portfolio. different place in the stack. different exposure. so before asking what a token earns, ask what has to happen before it takes a loss. then ask what is actually sitting underneath it.

onchain doesn’t mean instantly liquid.

insurance capital has a timetable. claims must be settled and collateral obligations satisfied before all of it can be released. tokenization does not shorten those obligations by itself.

re’s redemption documentation says reUSD supports instant redemptions while a liquidity buffer is available, subject to caps, and can switch to quarterly processing when that buffer is exhausted. reUSDe redemption availability is assessed quarterly and depends on collateral requirements and releases. [6]

those conditions are part of the product, not small print to read after a deposit. the ability to transfer a token, find a buyer for it, and redeem it against the underlying arrangement are three different questions. a visible market price does not settle all three.

what transparency actually proves

better reporting is valuable. Chainlink describes proof of reserve as a mechanism for making collateral information available to onchain applications. [7] it can help users examine backing that would otherwise be difficult to observe.

but knowing that an asset exists is different from knowing that it is sufficient for every future obligation. a reserve balance cannot, by itself, tell us whether claim estimates are realistic, whether a contract was priced well, or whether money can be withdrawn when requested.

this is the important part. making the balance visible does not make the underwriting correct. it gives us something to check. we still have to check it.

risk first. still.

this isn’t defi replacing reinsurance. it’s a way to make the capital behind reinsurance easier to inspect. that is useful. but the underwriting still has to hold up.

follow the money. what contract does it back? what claims can hit it? who absorbs those claims first? when can the remaining capital actually come back? if those answers are unclear, a yield number does not clear them up.

capital gives the system capacity. underwriting decides whether that capacity is being used well.

the technology can change how capital arrives.
the risk still has to be priced.