Self-Insured & Captive

Hospital and Health System Professional Liability IBNR

How actuaries estimate unpaid medical professional liability claims for hospitals and health systems, why the captive structure creates a net-exceeds-gross wrinkle, and what to require in the report.

By Sam · April 2026 · 13 min read

If your hospital, health system, or academic medical center retains professional liability risk through a captive insurer, your balance sheet carries a liability for claims that have already occurred but have not yet been fully paid. The most uncertain piece of that liability is IBNR: incurred but not reported, which in actuarial practice means both the claims nobody has filed yet (pure IBNR) and the future development on claims your adjuster already knows about (IBNER). Together, those two components are what actuaries call broad IBNR, and for hospital professional liability they present a specific challenge: the tail is long, the claims are high-severity and low-frequency, and the captive structure that most health systems use introduces a gross-to-net relationship that can behave in ways that look wrong until you understand the mechanics.

This article explains how actuaries estimate unpaid professional liability claims for hospitals and health systems, what makes the line different from other long-tail casualty lines, where the standard methods break down, and what you should require in documentation. It is written for the hospital CFO, risk manager, captive board member, or finance director who receives a reserve study and needs to evaluate it, not for the actuary who produces it.

Why hospital professional liability is its own reserving problem

Hospital professional liability shares the basic reserving machinery with other long-tail casualty lines: loss development triangles, cumulative development factors, Bornhuetter-Ferguson anchoring for recent years. But the line has characteristics that change how those tools behave and where they fail.

The tail is long. Medical malpractice claims against hospitals routinely take five to ten years to close. Obstetric injury claims (particularly cerebral palsy cases involving minors) can remain open far longer, because the statute of limitations in most states does not begin to run until the minor reaches the age of majority. A delivery complication today may not produce a filed claim for eighteen years. Surgical complications, missed diagnoses, and medication errors follow shorter but still substantial timelines, with litigation, expert discovery, and settlement negotiations stretching resolution across multiple accident years. That tail length means the actuary must select development factors and tail factors for maturities where the triangle has little or no data, and those selections carry a disproportionate share of the judgment in the review.

The frequency is low and the severity is high. Unlike workers compensation, where a self-insured employer may see hundreds or thousands of claims per year, a hospital professional liability program typically produces a small number of claims per accident year, sometimes fewer than twenty in the retained layer. Each claim can be very large. A single birth injury verdict or settlement can represent a material fraction of the entire accident year’s ultimate. That combination of low frequency and high severity makes the loss development triangle noisy: a single large payment in one development period can swing a column of age-to-age factors, and historical patterns become unreliable predictors of future development.

The exposure base is clinical volume. Friedland identifies occupied beds and outpatient visits as the standard exposure base for hospital professional liability (Friedland, p. 132). That means the expected-claims anchor in a Bornhuetter-Ferguson or expected-claims analysis is built from clinical activity times an expected claim rate per unit of exposure. If your hospital has added a new surgical pavilion, acquired an outpatient network, closed an obstetric unit, or expanded into telehealth, the exposure base has shifted, and the expected-claims anchor must be adjusted. A flat expected-claims ratio (what the industry calls a loss ratio; Friedland uses “claim ratio” because she reserves “loss” for non-claims economic concepts) applied to last year’s bed count without accounting for changes in clinical mix, procedure volume, or patient acuity is a silent error that propagates through every method that relies on the expected-claims calculation.

The structure is usually a captive. Most hospital systems that retain professional liability do so through a captive insurer rather than a pure self-insurance trust or a large-deductible policy. That captive is a licensed insurance company with regulatory reporting obligations, capital requirements, and an annual Statement of Actuarial Opinion. The captive structure also creates a layered risk architecture: the captive typically assumes a working layer (for example, the first $1 million or $2 million per occurrence), a fronting carrier issues the policy, and excess or reinsurance covers layers above the captive’s retention. That layered structure has consequences for how the reserve is estimated and how the gross, ceded, and net numbers relate to each other. (See Self-Insurance, Captives, Large Deductibles, and SIRs for the structural distinctions.)

The arithmetic in plain language

The core equation is the same one that applies to every casualty line:

Ultimate claims = paid to date + case outstanding + IBNR

Case outstanding (also called case reserves) is the adjuster’s or defense counsel’s current estimate of what each known claim will cost to resolve. IBNR is everything else: claims not yet reported, plus development on known claims above what case outstanding already reflects.

For hospital professional liability, a simplified example anchors the intuition. Suppose your captive’s 2022 accident year shows the following as of December 31, 2025 (36 months of development):

Paid claims: $4.2 million
Case outstanding: $6.8 million
Reported claims (paid + case outstanding): $11.0 million
Cumulative development factor (reported, 36-to-ultimate): 1.85

If the actuary applies the chain ladder method, the projected ultimate is $11.0 million times 1.85, which equals $20.35 million. IBNR is projected ultimate minus reported: $20.35 million minus $11.0 million, or $9.35 million. That IBNR is 85% of the reported claims to date, which may look alarming but is typical for a long-tail professional liability line at 36 months of maturity.

Now apply the Bornhuetter-Ferguson method. Suppose the actuary selects an expected claim ratio of 65% applied to earned premium of $28 million, producing expected claims of $18.2 million. The percent unreported is 1 minus (1 divided by 1.85), or approximately 46%. The BF ultimate is $11.0 million (actual reported) plus $18.2 million times 46%, which equals $11.0 million plus $8.37 million, or $19.37 million. BF IBNR is $8.37 million.

The difference between the chain ladder ($9.35 million) and BF ($8.37 million) is roughly $1 million, about 10% of the IBNR. That gap is driven entirely by how much weight each method gives to the actual reported claims versus the expected-claims anchor. For a 36-month accident year on a long-tail line, BF gives roughly 54% credibility to the actual data and 46% to the prior expectation. Chain ladder gives 100% credibility to the actual data. If the 2022 accident year happens to include an unusually large or unusually small claim, the chain ladder will amplify that anomaly; BF will dampen it. That is why BF is the natural default for recent accident years on hospital professional liability.

When the standard methods break down

The methods themselves are sound. What breaks them is when their assumptions no longer hold. For hospital professional liability, the most common failure modes are:

Low claim counts make development factors unreliable. With only a handful of claims per accident year in the retained layer, a single large settlement or verdict in one development period can produce an age-to-age factor that is an outlier, not a pattern. The chain ladder amplifies that instability because it projects from the actual reported or paid amount; if the base is distorted, the projection is distorted by the same multiplicative factor. Friedland’s example of a paid bodily injury triangle shows a CDF of 90.00 on the latest accident year (Friedland, p. 134), illustrating the extreme leverage that thin data creates. A single unexpected $500,000 payment at that leverage multiplies to a $45 million swing in projected ultimate.

Case reserve practices vary by defense counsel. Hospital professional liability claims are typically managed by specialized defense firms, and those firms set case reserves based on different philosophies. Some reserve to expected value; some reserve to a likely outcome; some reserve high to avoid surprises. When a hospital changes defense counsel, merges with another system that uses a different firm, or centralizes claims management under a new risk management structure, case reserve adequacy shifts. That shift inflates or deflates the reported development triangle in ways that have nothing to do with the actual trajectory of the claims. The diagnostic signal is the same one described in the case reserve strengthening article: age-to-age factors for the reported triangle change in recent calendar periods relative to pre-change history.

Service mix changes alter the exposure profile. Adding a trauma center, expanding neonatal intensive care, or closing an obstetric unit changes the frequency and severity profile of claims. Obstetric claims are the dominant driver of severity in hospital professional liability; a system that closes its labor and delivery service will see a decline in expected severity that takes years to fully emerge in the triangle. An actuary who applies a historical development pattern without adjusting for the service mix shift will overstate the reserve for the post-change accident years.

Tort reform and venue changes. State-level tort reform (damages caps, certificate-of-merit requirements, screening panels) and changes in litigation venue (shifts to plaintiff-friendly jurisdictions) alter severity trend. A damages cap that was enacted five years ago is now embedded in the historical pattern; a cap that was lifted last year is not. The actuary must identify which trend environment produced the historical data and whether that environment still applies to the projection period.

The net-exceeds-gross wrinkle

This is the structural feature that most often confuses hospital captive boards and one that Friedland addresses directly.

In a typical fronted captive structure, the fronting carrier issues the policy and retains certain layers (often the excess above the captive’s retention), while the captive assumes the working layer through a reinsurance agreement. Friedland notes that in this arrangement, net development factors can exceed gross development factors (Friedland, p. 331), which violates the intuitive expectation that net should always be smaller than gross.

The mechanics are straightforward. Gross development includes all layers: the captive’s working layer plus the excess layers retained by the fronting carrier or transferred to reinsurers. Net development includes only the captive’s working layer. For hospital professional liability, large claims that ultimately pierce the captive’s retention develop differently than claims that stay within it. In the early development periods, case reserves on large claims may be set within the captive’s layer; as the claim develops and the reserve increases beyond the retention, the excess portion is ceded. The gross triangle captures the full development (including the ceded portion), while the net triangle captures only the portion within the retention.

What happens in practice is that small and mid-size claims, the ones that stay entirely within the captive’s layer, can develop faster (as a percentage of their ultimate) than large claims that ultimately pierce the retention. The net triangle, which is dominated by these working-layer claims, shows a development pattern that requires higher age-to-age factors at certain maturities than the gross triangle. The result is a net CDF that exceeds the gross CDF for some development intervals.

This is not an error. It is a structural feature of how layers interact with development timing. But if you see net IBNR exceeding gross IBNR on your captive’s reserve study and nobody has explained why, that is a gap in communication. The fronting and reinsurance article walks through the gross-to-net bridge in detail.

What to ask your actuary

These questions are specific to hospital professional liability. They supplement (not replace) the general questions in the documentation checklist.

What is the exposure base, and does it reflect changes in clinical volume and service mix? The actuary should be using occupied beds, adjusted patient days, outpatient visits, or a composite metric that captures the scope of clinical operations. If your system has acquired or divested facilities, opened or closed service lines, or seen material shifts in surgical volume or obstetric deliveries, the exposure base for each accident year must reflect those changes. A flat bed count applied across years of changing clinical scope undermines the expected-claims anchor.
How are obstetric claims handled in the development pattern? Obstetric claims, particularly birth injury cases, drive severity and extend the tail. Some actuaries segregate obstetric claims into a separate triangle or apply a separate development pattern; others include them in a combined triangle with a blended pattern. Neither approach is wrong, but you need to know which one your actuary chose and why, because the treatment determines how sensitive the estimate is to a single large obstetric claim.
Which method drives the most recent two accident years, and why? For a low-frequency, high-severity line at early maturities, the chain ladder is rarely appropriate as the primary method. The leveraged CDF problem is acute. BF or expected claims should anchor the recent years, with chain ladder used as a reasonableness check. If the chain ladder is the primary method for any year with fewer than 36 months of development, ask what justifies trusting the actual reported data at that maturity.
How does the net development pattern relate to the gross pattern, and is the gross-to-net bridge documented? If your captive’s reserve study shows net CDFs exceeding gross CDFs at certain maturities, the actuary should explain why and demonstrate that the relationship is consistent with the layer structure. An unexplained divergence between gross and net is a red flag.
What severity trend is selected, and how is it supported? Hospital professional liability severity trend is driven by verdict and settlement escalation, medical inflation in damage awards, and changes in the tort environment. The actuary should document the selected trend, its source (program experience, industry benchmark, or blend), and whether the projection period reflects any changes in tort reform, damages caps, or venue.
Is the estimate a central estimate, a management best estimate, or something else? ASOP 43 defines the actuarial central estimate as an expected value over reasonably possible outcomes. Know which standard your number meets and how the point estimate sits within the indicated range.

What to require in documentation

The documentation checklist consolidates professional-standards requirements into a single reference. For hospital professional liability specifically, the report should include:

Separate triangles by claim type or coverage at a minimum distinguishing obstetric from non-obstetric claims, or physicians from facility, depending on the program structure. The report should present paid and reported triangles for each segmentation with enough history to observe the full development pattern through settlement.
Method selection by accident year and segment with stated rationale, including a comparison of chain ladder, Bornhuetter-Ferguson, and (where data supports it) frequency-severity results.
Exposure base documentation confirming occupied beds, adjusted patient days, outpatient visits, or the selected composite metric by accident year, with adjustments for facility acquisitions, divestitures, and service line changes.
Gross-to-net bridge showing gross, ceded, and net ultimate by accident year, with an explanation of any accident year where net IBNR exceeds gross IBNR.
Severity trend selection documented with source and basis, reflecting the tort environment applicable to the projection period, including disclosure of any damages caps, certificate-of-merit requirements, or venue changes.
Large-loss treatment disclosing the threshold, the number and amount of large losses by accident year, how they are handled in each method (included, excluded, capped, or developed separately), and whether any individual claim materially influences the selected ultimate.
Defense counsel and case reserve practices disclosing whether case reserve adequacy has been reviewed, whether any changes in defense counsel or reserving philosophy occurred during the experience period, and how the actuary addressed any identified adequacy shift. (See Berquist-Sherman in Plain English for the standard correction.)
Sensitivity analysis showing how the reserve changes under alternative severity trend, CDF, or expected-claims assumptions for the two or three most recent accident years, which is where the estimate is most uncertain.

If the report does not include these elements, send the list back and ask for the missing sections. A report that combines obstetric and non-obstetric claims into a single undifferentiated triangle, shows only net results without a gross-to-net bridge, and provides no sensitivity analysis is not giving you the information you need to evaluate the number on your balance sheet.