Latent Liability

Study guide: Exam 6U

Latent Liability is liability for injury or damage that surfaces long after the exposure that caused it — occupational disease, toxic contamination, abuse disclosed decades later — so the claims fall on policies written, priced and often long closed years or decades earlier, under wordings that never contemplated them.

\text{Survival ratio} = \frac{\text{Reserves (case + IBNR)}}{\text{Average annual paid losses}}

• The survival ratio is the number of years the carried reserve would last at the recent payment rate; the paid denominator is usually the average of the last three calendar years. It is simple and computable from public data — the annual statement's Note 33 gives five calendar years of asbestos and environmental reserves and payments, direct, assumed and net — but crude: it ignores the shape of future payments and is distorted by one-off payments such as policy buybacks and commutations.
• Why ordinary reserving fails. Claims emerge by calendar year of manifestation, driven by epidemiology, litigation and legal doctrine rather than by an accident-year pattern, so the chain ladder has no stable pattern to project. Practice uses exposure-based (policy-by-policy) projections, claim-count-times-severity models by report year, market-share benchmarks, and the survival ratio as a check, so the IBNR rests heavily on judgement.
• How it affects P/C insurers:
  • repeated adverse development and reserve strengthening, straining surplus, ratings and sometimes solvency;
  • old occurrence policies triggered decades later, with limits potentially stacked across many policy years — one reason claims-made forms and asbestos and absolute pollution exclusions spread in the mid-1980s;
  • coverage litigation over trigger (exposure, manifestation, injury-in-fact, continuous) and allocation (pro rata versus all sums), adding large AO litigation costs;
  • reinsurance disputes and insolvent reinsurers, showing up in Schedule F;
  • run-off and legacy transactions (loss portfolio transfers, adverse development covers) to cap the exposure.
• Examples: Asbestos; environmental cleanup under the federal Superfund law (CERCLA), which imposes strict, joint and several, retroactive liability; silica; lead paint; childhood sexual abuse claims revived when states open filing windows; PFAS "forever chemicals." Many become a Mass Tort.
• Disclosure: SSAP No. 55 requires the reserving methodology and the amounts paid and reserved for asbestos and environmental claims, and the Statement of Actuarial Opinion discloses net asbestos and environmental reserves separately.

Survival Ratio With a One-Off Payment An insurer carries net asbestos reserves (case + IBNR) of \$360M. Net paid in the last three calendar years was \$30M, \$45M and \$45M; the last figure includes a \$20M policy buyback that extinguished that policyholder's future claims. A peer group of insurers with similar exposure carries a survival ratio of 13 years.

Compute the survival ratio with and without the buyback and the reserve implied by the peer benchmark.

Answer As reported:

\begin{align*}
\text{Average paid} &= \frac{30 + 45 + 45}{3} \\
&= \$40.0\text{M} \\[4pt]
\text{Survival ratio} &= \frac{360}{40.0} \\
&= 9.0 \text{ years}
\end{align*}

Excluding the buyback — a one-time payment that also removed future liability:

\begin{align*}
\text{Average paid} &= \frac{30 + 45 + 25}{3} \\
&= \$33.3\text{M} \\[4pt]
\text{Survival ratio} &= \frac{360}{33.3} \\
&= 10.8 \text{ years}
\end{align*}

Peer benchmark:

\begin{align*}
\text{Implied reserve} &= 13 \times \$33.3\text{M} \\
&= \$433\text{M} \\[4pt]
\text{Gap} &= \$433\text{M} - \$360\text{M} \\
&= \$73\text{M}
\end{align*}

The buyback made the reserve look nearly two years thinner than it is, which is why the payment history should be cleaned before the ratio is read. Even after cleaning, 10.8 years against a peer 13 suggests a possible \$73M shortfall — a flag for an exposure-based study, not a reserve estimate, since the peers' mix of policyholders and remaining claim lifetimes may differ.