If your captive buys reinsurance, your net IBNR should be smaller than your gross IBNR. That is the whole point of ceding risk. Except sometimes it is not, and if you do not understand why, you will misread every reserve report your actuary gives you.
This article answers one specific question: why can net IBNR exceed gross IBNR, and when is that a problem versus just a feature of the structure? If you need the foundational concept first, the plain-English IBNR explainer covers what the number means. If you want the single-parent captive or group captive context, the single-parent captive guide and the group captive guide are the right starting points. This article is the deeper technical treatment those guides point toward.
The three layers every captive owner should know by name
Your actuary’s reserve study should present the estimate in three views: gross, ceded, and net. Each one answers a different question, and if you are only seeing one of them, you are missing two-thirds of the picture.
Gross is the total estimated ultimate loss before any risk transfer. It is the number the program would owe if no reinsurance existed and no fronting carrier retained any layer. Gross includes every dollar of expected loss, whether the captive ultimately pays it or not.
Ceded is the portion of that gross loss that has been transferred to someone else through a reinsurance agreement or retained by the fronting carrier under the fronting arrangement. In a simple structure, ceded is the excess layer above the captive’s retention. In a more complex structure, it can include quota share participations, aggregate stop-loss corridors, or fronting carrier retained layers.
Net is what the captive keeps: gross minus ceded. This is the number that hits the captive’s balance sheet, the number the captive’s auditor cares about, and the number the captive’s board should be signing off on.
Friedland notes that net claims and net premium should generally be less than or equal to their gross equivalents (Friedland, p. 331). When they are not, something in the structure needs explaining. That does not necessarily mean the number is wrong. It means the bridge between gross and net is doing something the reader needs to understand.
If your actuary only gives you net, that is a red flag. You cannot evaluate the reasonableness of a net reserve without seeing the gross number it was derived from and the ceded number that was subtracted from it. An actuary who only delivers net is asking you to trust the bridge you cannot see.
What fronting actually is
A fronting carrier is an admitted insurer that issues the policy to the insured, typically because the captive is not licensed to write admitted business in the jurisdiction where the insured operates. The fronting carrier puts its paper on the risk, its name on the certificate of insurance, and its financial rating behind the policy. Then it cedes most or all of the risk to the captive through a reinsurance agreement.
The captive is the real risk-bearer. The fronting carrier is the legal risk-bearer. Friedland identifies captives as licensed insurers (Friedland, p. 6), but the fronting arrangement exists precisely because the captive’s license does not extend to every state or jurisdiction where the parent company operates. The fronting carrier solves the licensing gap, and the captive pays a fronting fee for the service.
Two common flavors matter for reserving:
100% cession. The fronting carrier issues the policy and cedes the entire risk to the captive. Gross and net in the captive’s books are identical because the captive assumes everything. The fronting carrier’s gross includes the entire policy, but its net is zero (or nearly zero, depending on commission structure). This is the simpler case.
Fronting carrier retains a layer. The fronting carrier issues the policy, cedes the working layer (say, the first $250,000 per occurrence) to the captive, and retains the excess layer above that. The captive’s “gross” in its own books is already limited to the working-layer retention. This is the structure that creates the anomaly described in the next section, and it is the more common arrangement in practice because the fronting carrier wants some skin in the game above the captive’s retention to manage its own regulatory capital requirements.
Why the fronting-retains-excess structure creates the net-exceeds-gross anomaly
This is the single most misunderstood piece of captive reserving, and it trips up experienced practitioners, not just first-time captive owners.
Start with a concrete example. Suppose the captive’s working-layer retention is $250,000 per occurrence. The fronting carrier retains everything above $250,000. The captive’s actuary builds a development triangle of losses limited to $250,000 per claim. That is the correct triangle for the captive’s exposure.
Now suppose a claim is sitting at $180,000 in incurred losses at twelve months of development. In the captive’s working-layer triangle, this claim is fully within the retention. The development factor applied to project this claim forward is based on the historical pattern of how working-layer-limited claims develop.
At twenty-four months, the same claim has grown to $350,000 in incurred losses. But the captive’s triangle sees only $250,000 of it, because the claim breached the retention. The remaining $100,000 sits in the fronting carrier’s excess layer. In the captive’s triangle, this claim went from $180,000 to $250,000, a 39% development factor. In the unlimited (gross-of-all-layers) triangle, the same claim went from $180,000 to $350,000, a 94% development factor.
Friedland flags this exact pattern (Friedland, p. 331). The limited development factor is smaller than the unlimited factor for this individual claim, but the aggregate effect across the triangle can go in either direction depending on the mix of claims near the retention boundary. When several claims cluster just below the retention at early maturities and then breach it as they develop, the captive’s net development factors can temporarily exceed what you would expect from the unlimited gross triangle. The captive’s net IBNR, calculated from those limited development factors, can exceed a naive expectation of “gross minus the excess layer.”
This is not an error. It is a feature of how loss limitation interacts with development. But it requires explanation in the reserve report. If your actuary shows you a net IBNR that exceeds what you expected and cannot walk you through which claims are driving it and how the limitation is affecting the development pattern, the report is incomplete.
The anomaly is most pronounced in two situations. First, in early-maturity accident years where a large share of open claims are still developing toward the retention boundary. These claims are fully within the captive’s layer today but may breach it tomorrow. The development factors at these maturities carry the most uncertainty. Second, in programs where the retention has not changed in several years but claim severity has been trending upward. In that scenario, more claims reach the retention than the historical pattern anticipated, and the limited development factors from older accident years understate what the current book will produce.
The CAS has addressed this complexity directly. A CLRS presentation on ceded reserving notes that ceded development patterns are typically slower than gross patterns, and that estimating ceded reserves is not as simple as subtracting net from gross. The development dynamics differ by layer, and treating them as a mechanical subtraction produces the wrong answer.
The per-occurrence-first-then-stop-loss sequence
For captives and group captives with both per-occurrence excess coverage and aggregate stop-loss protection, the order in which the actuary applies the limitations matters. The correct sequence, per Friedland (p. 332), is to estimate ultimate claims limited to the per-occurrence retention first, then apply the aggregate stop-loss as a final adjustment to the per-occurrence-limited ultimates.
The group captive guide covers the per-occurrence and aggregate stop-loss structure in the context of member allocation. Here, the point is narrower: the sequencing of limitations.
Why the order matters. If you apply the aggregate stop-loss to unlimited gross ultimates first and then layer in the per-occurrence limitation, you get a different and incorrect answer. The aggregate stop-loss is designed to cap the captive’s retained losses after per-occurrence limits have already been applied. Reversing the sequence double-counts the excess coverage on large claims and understates the captive’s net retained reserves.
A concrete example: suppose the captive has a $250,000 per-occurrence retention and a $2 million aggregate stop-loss. A single claim develops to $500,000. In the correct sequence, the captive sees $250,000 of this claim (per-occurrence limit), and that $250,000 counts toward the $2 million aggregate. In the incorrect sequence, the $500,000 counts toward the aggregate first, potentially triggering the stop-loss prematurely and making the captive’s retained position look smaller than it actually is.
Ask your actuary explicitly: in what order are the per-occurrence and aggregate limitations being applied? If the answer is not immediately clear from the report, the documentation needs improvement.
The six questions to ask about the gross/ceded/net bridge
This is the section to screenshot and bring to your next actuarial review. Each question has a short description of what a good answer looks like.
1. Are all three views in the report? Good answer: separate exhibits for gross, ceded, and net, each with their own development triangles, selected factors, and point estimates. If only net is shown, ask why.
2. Does ceded plus net reconcile to gross? Good answer: yes, within rounding. If they do not reconcile, there is either an allocation issue or a layer that is being estimated independently and not constrained to the gross total. Either way, the gap needs explaining.
3. Are development factors shown separately for gross and net? Good answer: yes, and the report explains any accident years where net factors exceed gross. If the factors are only shown on a net basis, you cannot evaluate how the layer limitation is affecting the development pattern.
4. Which layers are being reserved separately versus in aggregate? Good answer: the report explicitly names each layer (working layer, per-occurrence excess, aggregate stop-loss) and describes whether it was estimated using a separate triangle, a simulation, or a deterministic application to the layer below. Layers blended into a single triangle without explanation should raise questions.
5. How is a single large claim that breaches the per-occurrence retention handled during development? Good answer: the actuary identifies claims near the retention boundary, describes how they are capped in the working-layer triangle, and discloses whether any excess development above the cap is being allocated to the ceded layer. If the answer is “we just cap at the retention,” ask whether the cap is current or whether it was applied at inception.
6. If there is aggregate stop-loss coverage, what is the sequence of limitations? Good answer: per-occurrence limitation is applied first, then aggregate stop-loss is applied to the per-occurrence-limited ultimates (Friedland, p. 332). If the answer describes the reverse sequence, the reserve may be understated.
What breaks when the fronting or reinsurance structure changes
A mid-experience-period change to retentions, limits, or cession percentages invalidates the historical triangle. If the captive’s per-occurrence retention moved from $150,000 to $250,000 at the last renewal, every accident year in the triangle prior to the change was developed on a $150,000 limit. The development factors from those years do not describe how $250,000-limited claims develop. Using them without adjustment will produce the wrong net IBNR.
Friedland references Berquist and Sherman’s general principle that the actuary should reorganize data when conditions change: policy year for accident year when limits change (Friedland, p. 283). For a retention change, the actuary has two options. Restate the historical triangle on the new retention basis (if individual claim data is available), or use a loss development triangle on an unlimited basis and apply the new retention limit to the projected ultimates using a severity distribution. Either approach requires the actuary to disclose that the adjustment was made and how.
The operational signal: if the captive renegotiated a fronting arrangement, changed its per-occurrence retention, added or removed an aggregate stop-loss layer, or switched fronting carriers in the last three years, the triangle needs re-basing. Ask whether the historical data has been adjusted and how. If the answer is that the old retention data is being used as-is, the estimate has a structural flaw that will surface as adverse or favorable development on the next diagonal.
This problem is more common than most captive owners realize. Retentions in the captive market move at nearly every renewal cycle as loss trends, reinsurance pricing, and regulatory capital requirements shift. A captive that has held the same retention for five consecutive years is the exception, not the norm. Every retention change creates a discontinuity in the triangle that the actuary needs to handle explicitly.
Uncollectible reinsurance and disputes
Friedland notes a specific exception to the general rule that net should not exceed gross: uncollectible reinsurance (Friedland, p. 331). If the captive’s reinsurer is financially impaired, or if there is an active coverage dispute over a ceded claim, the net IBNR the captive should carry is higher than the mechanical gross-minus-ceded calculation would suggest.
This is not hypothetical. In runoff books, especially those with environmental or asbestos exposure, reinsurance disputes and reinsurer insolvencies have produced situations where the ceding company’s net reserves exceeded gross because recoveries that were theoretically owed were practically uncollectible. Captives are smaller and less likely to face asbestos-era disputes, but the principle applies to any situation where ceded recoverables are doubtful.
The governance implication is straightforward: the captive’s board should review reinsurance recoverables aging alongside the reserve estimate. How old are the outstanding recoverables? Is the reinsurer paying on time? Is there any correspondence suggesting a coverage dispute? A receivable that is twelve months past due is not a recoverable; it is a credit risk. If the actuary’s net estimate assumes full collectibility and the board knows the reinsurer is slow-paying, the two views need to be reconciled before the number gets booked.
For fronted captive programs specifically, the fronting carrier’s financial strength matters too. The fronting carrier is on the policy paper, and if the fronting carrier enters receivership, the insured looks to the fronting carrier, not the captive, for payment. But the captive’s reinsurance obligation to the fronting carrier does not disappear. Understanding who owes what to whom, and what happens if one party in the chain cannot pay, is the kind of structural question the captive board should be asking annually.
What good documentation looks like
A reserve report on a fronted captive should include, at a minimum, the following: separate development triangles for gross, ceded, and net losses; an explicit statement of which layers are being reserved on a limited versus unlimited basis; documentation of any structural change to retentions, limits, or cession percentages in the historical period and how the actuary handled it in the analysis; a reconciliation showing that gross equals ceded plus net within rounding for each accident year; and, if applicable, a separate discussion of uncollectible reinsurance or disputed recoverables with a reserve recommendation that is not netted against the main estimate.
If your report does not contain all five elements, the documentation is incomplete. The actuary may have done the work correctly and simply not documented it well enough for a non-actuary to follow. But documentation that you cannot audit is documentation that you cannot defend to a regulator, an auditor, or a board. For a fronted captive, the structural complexity makes clear documentation more important, not less.
Three concrete actions for your next review
1. Request the gross, ceded, and net views as three separate exhibits. Verify that they reconcile. If your current report only shows net, ask the actuary to add gross and ceded as separate exhibits starting with the next valuation. The incremental effort is modest because the actuary almost certainly produces all three views internally; the question is whether they are delivered to you.
2. Ask about the working-layer-versus-excess split. Specifically: are there accident years where the net development factors exceed gross? If so, ask the actuary to explain which claims are driving the anomaly and whether the current retention is properly reflected in the limited triangle. If you changed retentions in the last three years, ask how the historical triangle was re-based.
3. Review your reinsurance recoverables aging. Ask the fronting carrier or reinsurer for an aging report of outstanding recoverables. If any recoverable is more than ninety days past due, raise it with the actuary and ask whether the net reserve estimate reflects the collectibility risk.
For the single-parent captive context behind these structures, see the single-parent captive guide. For the group captive and RRG perspective on per-occurrence and stop-loss layering, see the group captive guide. And if you want a second set of eyes on the gross-to-net bridge, the independent review note explains why an independent actuarial assessment adds particular value for fronted captive programs where the bridge is complex.