Guide
Treaty reinsurance, explained
What a treaty is, the two families it comes in, and every metric used to run one — from quota share and surplus share, through the reinsurance tower, to how a large loss and its reinstatement actually get paid. Thirty-two terms, one page, each linked to its full definition.
What a treaty is
A treaty, or treaty reinsurance contract, is one arrangement covering a whole class of business — all fire risks in a territory, say, or an entire motor account — for a defined period, usually twelve months. It is obligatory: every qualifying policy is ceded automatically the moment it is written, and the reinsurer must accept it on the terms agreed at placement, with no risk-by-risk underwriting. That is what makes it different from facultative reinsurance, which is negotiated risk by risk — the reinsurer can inspect, price, amend or decline any single submission. A cedent with a treaty covering most of its book will still go facultative for the one risk that is too large for the treaty, excluded from it, or unusual enough to want a second opinion on.
Underneath either arrangement sits retention: the share of a risk the cedent keeps for its own account, with the rest ceded. Retention can be set per risk, per event, or in the aggregate over a year, and it is the number almost everything else in this guide is defined relative to — it is where a non-proportional layer starts to respond, and how much of a proportional treaty the cedent still carries itself.
Automatic
Treaty
Covers a whole class of business for a period. No individual underwriting; new business is protected the instant it is written.
Negotiated
Facultative
Covers one named risk. The reinsurer can inspect, reprice, amend or decline it before binding.
How a treaty is placed and shared
A single treaty is rarely held by one reinsurer alone, and a cedent's own reinsurance programme is rarely just one contract — both facts create a small vocabulary of their own.
RetrocessionReinsurance a reinsurer buys for itself, protecting the book it has already assumed from its own cedents — the same mechanism, one link further up the chain. A reinsurer holding $2.0B of aggregate hurricane exposure across dozens of cedents might buy $500M of retro excess of $500M, so no single season can consume more than a fifth of its capital. Line slip & subscription marketThe market convention that lets many reinsurers each take a small percentage “line” of the same risk instead of one carrier taking it all, via a single slip that circulates for signatures. If a risk is offered at 100% and reinsurers write lines totalling 140%, every line gets signed down proportionally so the total comes back to 100%. Facultative-obligatory (fac/oblig)A hybrid: the cedent decides, risk by risk, whether to cede a qualifying exposure into the facility, but once it does, the reinsurer must accept it on pre-agreed terms — no individual underwriting, no right to decline. It trades the reinsurer's selection rights for the cedent's speed. Inuring reinsuranceThe pecking order among a cedent's own covers: which contract applies and is netted down first, before another is even asked to respond. A facultative recovery that reduces a $50M loss to $30M net means the treaty sitting above it only ever sees the $30M — get the order wrong in the wording and two reinsurers can each assume the other pays first. Proportional treaties
In a proportional treaty the reinsurer shares every risk, every premium dollar and every loss dollar in a fixed ratio — it is on risk from the first dollar of loss, not just the large ones. There are two ways to set that ratio.
Fixed percentage
The reinsurer takes the same fixed percentage of every risk in the treaty, regardless of size. A 40% quota share on a $50M book cedes $20M of premium, and every $1M claim produces a $400,000 recovery — the ratio never moves.
Moves with risk size
The cedent keeps one fixed amount per risk — its “line” — and cedes only the surplus above it, so the split moves with each risk's size. With a $2M line and 9 lines of capacity, a $10M risk is retained 20% and ceded 80%, while a $2M risk is kept entirely net.
Proportional treaty economics
Because premium and losses move together in a fixed ratio, the money on a proportional treaty is really negotiated somewhere else — in the commission structure and the administrative machinery that keeps both sides honest.
Ceding commissionThe percentage of ceded premium the reinsurer pays back to the cedent to cover its acquisition and administration costs — the main lever a proportional treaty is actually negotiated on. A treaty might pay 30% provisional commission, sliding down toward 22.5% if the loss ratio deteriorates. Profit commissionAn extra payment to the cedent out of whatever profit the treaty actually made, calculated after losses, ceding commission and the reinsurer's expense allowance are all deducted. $3.5M of calculated treaty profit at a 25% profit commission rate pays the cedent $875,000, on top of its ceding commission. Sliding scale commissionA ceding commission that moves inversely with the loss ratio between an agreed floor and ceiling, instead of staying fixed — sharing result volatility without the full complexity of a separate profit commission. A loss ratio ten points better than planned might lift commission from a 30% base toward a 35% cap. Bordereau & OGPIA bordereau is the periodic schedule listing every risk, premium or claim ceded under the treaty; Original Gross Premium Income (OGPI) is the subject premium the treaty's rate is applied to. A treaty rated at 35% of OGPI simply applies that percentage to whatever gross premium the quarterly bordereau reports. Deposit & minimum premiumA deposit premium is an estimated instalment paid during the year, before the actual subject premium is known; it is trued up against the adjusted premium at year end, subject to a minimum premium floor. A treaty depositing $2.0M against an estimate might owe a further $350,000 once the real book comes in larger than planned. Portfolio transferThe payment that puts an incoming reinsurer on risk for business already written before it joined a treaty (a portfolio entry), or releases an outgoing one from remaining exposure when it leaves (a portfolio exit). A new reinsurer taking over a renewal typically receives a share of the unearned premium reserve to compensate for exposure it never priced from day one. Non-proportional treaties: the tower
Where proportional treaties share every risk in fixed proportion, non-proportional treaties are stacked as a tower — a vertical run of excess-of-loss layers above the cedent's retention, each its own contract with its own price, limit and panel of reinsurers. A programme of $15M xs $10M, $25M xs $25M and $50M xs $50M gives $90M of cover above a $10M retention; a loss burns through the retention first, then each layer in turn, until it is exhausted or the tower runs out. Layers are stacked, rather than sold as one large contract, because appetite is not uniform — the market willing to write the frequency-exposed bottom layer is rarely the one that wants the remote top layer.
Excess of loss (XoL)The reinsurer pays only the slice of a loss between an attachment point and a limit, priced from the cedent's own loss experience rather than a fixed share of premium. Written “$20M xs $5M”: a $12M loss recovers $7M, and a $30M loss recovers the full $20M limit, leaving $5M unprotected above it. Attachment point & limitThe attachment point is where the reinsurer's liability begins; the limit is the most it will pay for one loss; their sum is the exhaustion point, above which the cedent is bare again unless a higher layer picks it up. A $40M xs $10M layer attaches at $10M and exhausts at $50M. Large-loss and catastrophe metrics
A handful of clauses exist specifically because one event, or one very large claim, can behave very differently from an ordinary loss — either by touching thousands of policies at once, or by taking years to fully develop.
Catastrophe XoLResponds to the combined loss from a single insured event across many policies, rather than to any one claim — a hurricane producing thousands of individually small claims can still exceed a large catastrophe layer in total. 4,000 claims averaging $45,000 add up to $180M, comfortably exhausting a $150M xs $20M cat layer even though no single claim came close to $20M on its own. Hours clauseDefines how long a catastrophe can run and still be treated as a single occurrence for a layer — commonly 72 hours for windstorm, 168 hours for earthquake including aftershocks. A second storm cell striking 80 hours after a 72-hour clause's window has already closed must be declared as a separate occurrence, with its own retention. Aggregate stop lossProtects the cedent's entire annual result rather than any single loss or event, attaching once the year's net incurred losses cross an agreed amount — usually stated as a loss ratio. A 25% xs 80% stop loss caps the effective net loss ratio at 80% until the cover itself exhausts at a 105% loss ratio. Clash coverExcess-of-loss protection for the accumulation that happens when a single event or claimant triggers several policies at once — one aviation accident hitting an airline's, an airport's and a contractor's liability policies together, say. It responds to what all three retained losses add up to combined, not to any one policy's retention alone. Loss corridorA band of loss, expressed in loss-ratio points or dollars, that the cedent must bear entirely alone even though cover applies both below and above it — a pricing tool that narrows the reinsurer's exposure over one stretch in exchange for a lower premium. A treaty might cover up to an 80% loss ratio and again above 100%, leaving the cedent to carry the 80%-to-100% corridor by itself. Index / stability clauseAdjusts a long-tail layer's attachment point and limit for inflation between placement and final settlement, so a retention fixed in nominal dollars does not quietly shrink in real terms over years of claims development. An inflation index moving from 100 to 135 over eight years lifts a $10M xs $5M layer to $13.5M xs $6.75M, protecting the same real slice of the loss. Metrics that test every layer
Whatever shape a treaty takes, the same handful of figures decide whether — and how much — it actually pays.
ReinstatementRestores an excess-of-loss layer's limit after a loss has eaten into it, for an additional premium — “one reinstatement at 100%” effectively doubles the layer's real aggregate exposure over the year. A $20M layer costing $4M with one 100% reinstatement, hit by a $12M recovery, triggers $2.4M of reinstatement premium for the 60% of the limit used. Ultimate net loss (UNL)The contractual definition of the loss figure that actually gets tested against a layer's attachment point: sums paid in settlement plus allowed expenses, less salvage, subrogation and any inuring reinsurance recoveries. A $28M paid loss can net down to a $22M UNL once defence costs, salvage, subrogation and a facultative recovery are all applied — it is the UNL, not the gross claim, that a layer is measured against. Rate on line & paybackRate on line is premium divided by limit, expressed as a percentage; the payback period, its reciprocal, is how many loss-free years of premium it takes to fund one total loss to the layer. A $50M xs $50M layer priced at $3M runs a 6.0% rate on line and a 16.7-year payback — the market's fastest sanity check on whether a layer is cheap. IBNRThe reserve held for losses that have already happened but have not yet been reported, plus IBNER (further development expected on claims already known) — it matters more in reinsurance than anywhere else because a claim can take years to travel from insured, to broker, to cedent, to reinsurer. A book with $61M reported against a $106.9M expected ultimate is carrying $45.9M of IBNR. Loss ratio & combined ratioLoss ratio is incurred losses divided by earned premium; add the expense ratio (expenses as a percentage of premium) and the total is the combined ratio — the standard read on underwriting profit before investment income, where under 100% is a profit. $37M of incurred losses on $50M of earned premium is a 74% loss ratio; stack a 25% ceding commission on top of that and the combined ratio comes to 99%. Claims control clauseGives the reinsurer the right to direct — or at minimum approve — how a claim is handled once its reserve crosses an agreed trigger, well before the loss actually reaches the layer's attachment point. A trigger set at 60% of a $2M retention activates once a claim's reserve is revised past $1.2M, long before it threatens the layer above. Funds withheldThe cedent keeps the reinsurer's share of premium and reserves on its own balance sheet instead of remitting cash across, crediting a negotiated interest rate in its place — collateral that removes most of the cedent's counterparty credit risk to the reinsurer. $7.2M of net premium due to a reinsurer might never leave the cedent's books under this structure, earning credited interest instead of moving as cash. Experience vs. exposure ratingTwo ways to price a layer: experience rating trends the account's own historical losses forward; exposure rating largely sets that history aside and builds a price from a severity curve independent of it. Actuaries typically blend both, weighted by credibility — a stable account with ten clean years might be priced 70% from experience, while a catastrophe layer with no retained losses is priced almost entirely from exposure. Worked example: a large loss through a real programme
Harborline Mutual retains the first $10M of any loss, then buys a tower of $15M xs $10M (with one reinstatement at 100%, costing $2.5M) and $75M xs $25M above that. A single hurricane produces a $46M ultimate net loss.
Cedent’s retentionThe first $10M of the loss stays with Harborline Mutual, before any layer responds.
First layer: $15M xs $10MCovers the loss from $10M to $25M. The $46M loss blows straight through this band, so the layer pays its full $15M limit and is completely exhausted.
Reinstatement premium owedBecause the first layer was used in full, the one-reinstatement clause requires Harborline Mutual to pay 100% of the original premium again — $2.5M — to put the $15M limit back for the rest of the year.
Second layer: $75M xs $25MCovers the loss from $25M up to $100M. Only $21M of the $46M loss falls in this band ($46M − $25M), so the second layer pays $21M and is nowhere near exhausted.
Total reinsurance recovery$15M from the first layer plus $21M from the second: $36M recovered against a $46M loss.
Net cost to the cedent$10M retention plus the $2.5M reinstatement premium: $12.5M — the true cost of the loss once the programme has done its work and been paid for.
So what
A $46M hurricane became a $12.5M net event for Harborline Mutual — but $2.5M of that was reinstatement premium, not loss. A programme is only as good as its reinstatements: buying a $15M limit is one decision, and buying enough reinstatements to survive a bad year is a second one, easy to skip when the first premium invoice already feels expensive.
Read this before you quote anything
Every figure on this page is illustrative and deliberately round — not a market rate, a quotation, or evidence of what any real layer should cost. Real treaty wordings vary enormously, and the version in your own contract beats the version described here every time.
Where to go from here
Every term on this page has its own full page in the Atlas — definition, worked example, quiz and word problem. If you would rather test yourself first, try the interview questions, or work through the 30-minute tour if treaty structures are new to you.