Self-Insured & Captive

Self-Insurance, Captives, Large Deductibles, and SIRs: What the Differences Mean for Your Reserves

Four risk-financing structures look similar from the outside but create different reserve obligations, different regulatory expectations, and different actuarial workflows; here is how to tell them apart and what each one means for the number on your balance sheet.

By Sam · April 2026 · 14 min read

You retain risk. That much is clear from your financial statements. But the legal vehicle through which you retain it determines who must hold an actuarial opinion, what the regulator expects to see, how ceded and net reserves line up, and how sensitive your IBNR estimate is to structural changes over time.

Most risk managers encounter at least two of the four structures described in this article during their careers, and some encounter all four simultaneously (a workers compensation self-insured retention layered under a large-deductible general liability program, with a captive writing the professional liability). When the structures are conflated, buyers make predictable mistakes: they apply a captive’s regulatory rigor to a program that has none, or they assume a large-deductible program carries the same reserve transparency as a qualified self-insurance fund.

This article defines each structure, explains how the reserving problem changes from one to the next, and gives you a checklist for making sure your actuary’s work product matches the structure you actually have.

Four structures, one underlying problem

Friedland defines self-insurance broadly as “a wide range of risk financing arrangements through which organizations pay all or a significant portion of their own losses” (Friedland, p. 6). That definition is deliberately wide. It covers everything from a municipality paying workers compensation claims out of its general fund to a Fortune 500 company routing risk through a Vermont-domiciled captive. The common thread is that the organization, not a commercial insurer, bears the economic cost of losses within a retained layer.

The four structures below are the most common forms that retention takes. Each one changes the reserving problem in a specific, documentable way.

1. Qualified self-insurance

In a qualified self-insurance program, the organization obtains a certificate from a state authority (typically a workers compensation board or an insurance department) allowing it to pay its own claims directly rather than purchasing a guaranteed-cost insurance policy. The organization posts security (a surety bond, a letter of credit, or a cash deposit), hires or contracts with a claims administrator, and funds losses from operating cash flow or a dedicated trust.

Reserve implications. The reserve estimate appears on the organization’s own balance sheet, usually as an accrued liability or a trust-fund balance. Most states require an annual actuarial certification of the self-insured reserve, but the scope and format of that certification vary widely. There is no statutory annual statement in the NAIC sense, no risk-based capital test, and no appointed actuary requirement in the way that applies to a licensed insurer or captive. The actuarial work product is typically a reserve study or a funding study rather than a Statement of Actuarial Opinion, though some states do require a formal opinion.

What changes the estimate. Because the self-insured entity controls its own claims administration (or contracts it to a TPA), operational changes hit the triangle directly. A TPA transition, a new claims director, or a shift in case reserve philosophy will change the development pattern, and the actuary must diagnose and adjust for that change. If you have experienced a TPA change or a case reserve strengthening in your self-insured program, the diagnostic framework in What’s Actually Driving Your IBNR Higher? applies directly.

2. Captive insurance company

A captive is a licensed insurance company whose primary purpose is to insure or reinsure the risks of its owners. Friedland’s definition is precise: a “limited purpose, licensed insurance company, the main business purpose of which is to insure or reinsure the risks of the captive’s owners” (Friedland, p. 6). The key word is “licensed.” Once a captive receives a license from a domicile regulator (Vermont, Utah, Bermuda, Cayman Islands, or one of dozens of other jurisdictions), it accepts the full apparatus of insurance regulation: minimum capital and surplus, statutory financial statements, annual audits, and in most domiciles a Statement of Actuarial Opinion signed by a qualified actuary.

Reserve implications. The captive carries reserves on its own statutory balance sheet, separate from the parent’s GAAP financials. The parent consolidates the captive’s results, but the reserve itself must satisfy two audiences: the domicile regulator (who wants adequacy) and the parent’s external auditor (who wants GAAP compliance). Those two audiences sometimes want different things, and the actuary must document both.

Most captives also operate within a fronting arrangement, meaning a licensed admitted carrier issues the policy, the captive assumes the retained layer via reinsurance, and the fronting carrier retains the excess. That arrangement creates a gross-to-net bridge that must be reconciled. If net development factors exceed gross (which can happen when the captive assumes the working layer where most development occurs), the usual intuition about net reserves being smaller than gross reserves breaks down. The mechanics of that bridge are covered in detail in Fronting, Reinsurance, and Why Your Captive’s Net IBNR Can Exceed Gross.

For a deeper walkthrough of the captive-specific reserving workflow, see IBNR for Single-Parent Captives and IBNR for Group Captives and RRGs.

What changes the estimate. Captives are subject to the same operational changes as self-insured programs (TPA transitions, case reserve shifts), but they also face structural changes that pure self-insureds do not: changes in the fronting carrier, changes in the retained layer (raising or lowering the captive’s attachment point or policy limit), and changes in the reinsurance program above the captive. Each of these changes can alter the development pattern, and the actuary must identify which triangle is affected and whether a Berquist-Sherman adjustment or a data reorganization is needed.

3. Large-deductible program

In a large-deductible program, the insured purchases a policy from a commercial carrier but retains a per-occurrence deductible, often $250,000, $500,000, or $1 million or more. The carrier pays the claim and then seeks reimbursement from the insured for the deductible portion. Because the carrier is liable to the claimant for the full amount (the deductible is a reimbursement obligation, not a policy exclusion), the carrier books a gross reserve and a corresponding receivable from the insured.

Reserve implications. The insured must accrue its deductible obligation on its own balance sheet, usually based on an actuarial estimate of ultimate losses within the deductible layer. The carrier also books reserves, but the carrier’s reserve and the insured’s reserve do not always match: the carrier may use different methods, different data aggregation, or different assumptions about claim development within the retained layer.

This mismatch is a recurring source of confusion. The insured receives a “loss pick” or a “deductible billing estimate” from the carrier, and separately receives an actuarial reserve estimate from its own actuary. The two numbers are answering slightly different questions (the carrier is estimating its reimbursement receivable; the actuary is estimating the insured’s liability), and they are built from different data. If the numbers diverge materially, the buyer needs to understand why.

What changes the estimate. The most consequential structural change in a large-deductible program is a change in the deductible level itself. If the deductible increases from $250,000 to $500,000 between policy years, the development pattern in the retained layer changes because the mix of claims that penetrate the higher deductible is different from the mix that penetrated the lower one. Friedland’s guidance is clear: when deductible levels change between policy years, the actuary should consider reorganizing the data by policy year (rather than accident year) so that each year reflects a consistent retention, or apply a Berquist-Sherman-style adjustment to restate the triangle to a consistent deductible level (Friedland, p. 283).

4. Self-insured retention (SIR)

A self-insured retention is structurally similar to a large deductible, with one critical difference: the insured pays claims within the SIR directly, and the carrier’s obligation does not attach until the SIR is exhausted. The carrier is not liable to the claimant for the SIR layer; it has no duty to defend within the SIR (unless the policy says otherwise); and it does not book a gross reserve on the SIR portion.

Reserve implications. The insured bears full claims-handling responsibility within the SIR layer. That means the insured (or its TPA) sets case reserves, manages litigation, and controls settlements within the retention. The actuarial estimate of the insured’s liability is built entirely from the insured’s own claims data, not from the carrier’s data. This is a meaningful difference from a large-deductible program, where the carrier often provides the claims data because the carrier is paying the claims and seeking reimbursement.

For the actuary, the practical consequence is data quality. In a large-deductible program, the carrier’s claims system typically captures every payment and every case reserve change because the carrier is managing the claim from first report. In an SIR program, the data depends entirely on the insured’s TPA, and the quality varies. Missing report dates, inconsistent case reserve coding, and gaps in payment history are more common in SIR data than in carrier-managed deductible data.

What changes the estimate. The same deductible-change dynamic applies. If the SIR increases, the retained layer broadens and the development pattern shifts. But because the insured controls claims handling within the SIR, operational changes (a new TPA, a new defense counsel panel, a change in settlement authority) hit the triangle more directly and with less delay than they would in a carrier-managed deductible program.

Why the structural difference matters for reserves

The four structures described above converge on the same economic reality: the organization is paying its own losses. But they diverge on three dimensions that directly affect the reserve estimate.

1. Disclosure and regulatory requirements. A captive files statutory financial statements and, in most domiciles, a Statement of Actuarial Opinion. A qualified self-insurer may file an actuarial certification with a state workers compensation board. A large-deductible buyer typically has no filing requirement beyond its own GAAP or governmental accounting standards. The level of regulatory scrutiny shapes the level of actuarial documentation, and buyers who operate in a lightly regulated structure sometimes receive less documentation than they should.

2. Who holds the actuarial opinion. For a captive, the domicile regulator requires a qualified actuary (typically an FCAS or FSA with a reserving credential) to sign an opinion. For a self-insured program, the requirement depends on the state and the line of business. For a large-deductible or SIR program, there may be no regulatory requirement for an actuarial opinion at all; the organization commissions one because its auditor or board requires it. The absence of a regulatory mandate does not mean the organization needs less rigor. It means the organization must impose rigor on itself.

3. How ceded and net reserves line up. In a captive with a fronting arrangement, the gross, ceded, and net reserves must reconcile across the fronting carrier and the captive. In a large-deductible program, the carrier’s gross reserve and the insured’s deductible accrual are calculated independently and often do not reconcile. In a qualified self-insurance program with excess insurance, the self-insured reserve and the excess carrier’s reserve may never be compared at all. The buyer who understands these structural differences can ask the right reconciliation questions; the buyer who does not may be surprised when the numbers do not tie.

When the structure changes

The most dangerous moment for reserves is not when claims develop badly within a stable structure. It is when the structure itself changes: a self-insured program forms a captive, a large-deductible program switches to an SIR, a captive raises its retained limit, or a fronting carrier is replaced.

Each of these changes breaks the continuity of the development triangle. The actuary must decide whether to splice the pre-change and post-change data into a single triangle (with adjustments) or to start a new triangle from the change date. Both approaches have costs. Splicing introduces inhomogeneity; starting fresh sacrifices maturity. The buyer’s job is to make sure the actuary has documented the decision and its consequences.

If you are evaluating whether your actuary handled a structural change correctly, the five leading indicators of adverse development provide a useful cross-check: if any of those indicators moved sharply around the date of the structural change, the actuary should be able to explain why.

What a buyer should ask their actuary

Which structure are you reserving, and does the data match? Confirm that the actuary knows whether the program is a qualified self-insurance, a captive, a large deductible, or an SIR, and that the claims data reflects the correct retained layer. Mismatches happen more often than you would expect, particularly when data comes from a carrier that does not distinguish between gross and net in its extracts.
Has the retained layer changed during the experience period? If the deductible or SIR changed, the actuary should either reorganize the data by policy year or apply an adjustment to restate the triangle to a consistent retention level. Ask which approach was used and why.
Is there a gross-to-net reconciliation? If the program involves a fronting carrier, the gross and net reserves should be reconcilable. If they are not, ask the actuary to explain the gap. If the program is a large deductible, ask whether the carrier’s loss pick and the actuarial reserve estimate have been compared, and what explains any difference.
Who requires the actuarial opinion, and what standard governs it? If the opinion is required by a domicile regulator, it is governed by the NAIC’s requirements and the applicable Actuarial Standards of Practice (ASOPs). If the opinion is commissioned voluntarily, the buyer should still require the actuary to follow ASOP 43 (for the estimate) and ASOP 9 (for documentation). The absence of a regulatory mandate is not a reason for less rigor.
How is the reserve being used on both the statutory and GAAP balance sheets? For captives, the reserve appears on the captive’s statutory statement and on the parent’s consolidated GAAP financials. For self-insured and large-deductible programs, the reserve appears only on the organization’s own financials. Make sure the actuarial estimate specifies which basis it supports and whether a different estimate is needed for the other basis.

What to require in documentation

Regardless of which structure you operate, the actuarial work product should include the following, even if no regulator demands it.

Explicit identification of the structure. The report should state whether it is reserving a captive, a qualified self-insurance program, a large-deductible obligation, or an SIR obligation, and describe the retained layer (per-occurrence retention, aggregate limit, any stop-loss or excess coverage above the retention).
Data source and reconciliation. The report should identify where the claims data came from (TPA, carrier, internal system), whether it was reconciled to financial records, and whether any data limitations were identified.
Retention-level consistency. If the deductible or SIR changed during the experience period, the report should document the change, the adjustment approach (data reorganization, Berquist-Sherman, or other), and the sensitivity of the result to that adjustment.
Gross-to-net bridge (if applicable). For captives with fronting arrangements, the report should show the gross, ceded, and net reserves and explain any case where net exceeds gross at the accident-year level.
Regulatory basis. The report should state whether the estimate is intended for statutory filing, GAAP accrual, funding calculation, or some combination, and note any differences in basis.
Range or sensitivity analysis. Whether the engagement requires a point estimate or a range, the report should include enough sensitivity analysis for the buyer to understand how the result changes under different assumptions about development, trend, or retention-level mix.