INSTITUTIONAL RISK INTELLIGENCE / ENTERPRISE
Acadia Healthcare — Institutional Risk Demonstration
Acadia’s disclosed PLGL development shows that claim frequency, prior-year severity and insurance terms can move enterprise earnings materially; IRI’s commercial value is to identify facility-level mechanisms early enough to test corrective conditions before they mature into reserve development.
DECISION BRIEF
The decision this evidence supports now.
WHAT IRI MAKES VISIBLE
Intelligence not contained in any single source.
The 2025 PLGL reserve increased from $87.5M to $181.8M while estimated insurance recoverables increased from $9.3M to $28.8M.
Acadia recorded a $52.7M unfavorable 2025 adjustment tied to prior-year settlement costs, increased claim frequency and less favorable insurance terms.
After August 31, 2025 the captive disclosed $15M per claim and $25M for certain batched claims, with unnamed third-party reinsurance at $75M aggregate subject to exclusions.
Historical litigation verifies TDC reinsured a $15M excess $60M layer and Marsh USA acted as Acadia’s reinsurance broker for the 2017 placement; this is not represented as current.
A further $28.6M PLGL reserve adjustment was disclosed in Q2 2026.
DECISION IMPLICATIONS
Link facility mechanisms prospectively to claim frequency/severity.
Use findings as testable hypotheses for underwriting, captive governance and reinsurance.
Do not infer current external carrier identities.
Track corrective controls against recurrence and claims.
CONVERGENCE
Evidence → mechanism → consequence.
observedhigh confidence
PLGL development is an earnings-level variable.
The reserve nearly doubled year over year and the 2025 adjustment was attributed to settlement costs, increased frequency and less favorable insurance terms.
Why it matters. Reducing uncertainty around recurrence mechanisms can matter when reserve estimates are sensitive to frequency and severity.
CRF · insufficient evidenceCMDS · insufficient evidenceCOVE/F · appliedCOMPASS · not applicable
Framework analysis and evidence requirements
CRF
Enterprise PLGL development establishes financial context but does not identify which continuity stage, if any, generated the claims.
CMDS
Reserve and frequency disclosures do not establish a CMDS mechanism or stage.
COVE/F
The financial finding can be converted into an evidence requirement and accountable recurrence-control test without attributing a violence or extraction code.
Harm · observed PLGL reserve development and earnings adjustments are observed enterprise consequences.
Reinforcement Loops · hypothesis Repeated facility mechanisms may contribute to claim frequency or severity, subject to prospective linkage.
Verify: Finding prevalence versus later claims by facility and policy year · 90-day pilot, then quarterly
COMPASS
The enterprise PLGL finding contains no person-level longitudinal differential assessment.
Evidence still required: Facility-level event chronology; Claim-to-facility linkage; Stage-specific corrective-action evidence; Institution-specific policy, reporting, authority and corrective-action evidence; Facility-level recurrence data; Corrective-action closure evidence
Test this finding
Validate: Reconcile later filings and claims data.
Falsify: Later development is unrelated to detectable facility mechanisms.
Evidence: A-10K, A-Q2, A-PLGL
derivedhigh confidence
Risk transfer does not eliminate enterprise exposure.
At year-end 2025 Acadia disclosed $181.8M of PLGL reserves and $28.8M of estimated insurance recoverables.
Why it matters. The same mechanism can affect captive retention, reinsurance, earnings and capital differently.
CRF · not applicableCMDS · not applicableCOVE/F · appliedCOMPASS · not applicable
Framework analysis and evidence requirements
CRF
This finding concerns risk financing allocation rather than a clinical continuity stage.
CMDS
Risk-transfer structure alone does not establish a CMDS mechanism or stage.
COVE/F
The finding needs ownership and verification across captive, reinsurance and enterprise consequences.
Structural Drivers · observed Captive retention and external reinsurance govern how covered losses propagate.
Actors · observed Acadia and its captive are identified; the current external reinsurer remains unnamed.
Harm · observed Enterprise reserves materially exceed estimated recoverables at the evidence cutoff.
Verify: Share of modeled loss paths reconciled to current coverage · Before the next underwriting cycle
COMPASS
The risk-financing finding contains no person-level longitudinal assessment.
Evidence still required: Current coverage schedules; Policy-specific exclusions and attachment points
Test this finding
Validate: Obtain policy-year loss runs and coverage schedules.
Falsify: Policy evidence materially changes the public allocation.
Evidence: A-10K
observedhigh confidence
Historical tower architecture is verified; current external identities are not.
TDC is verified historically for a $15M excess $60M layer and Marsh for the 2017 placement; current SEC disclosure names only an unnamed third-party reinsurer.
Why it matters. Propagation can be shown without fabricating a current carrier.
CRF · not applicableCMDS · not applicableCOVE/F · appliedCOMPASS · not applicable
Framework analysis and evidence requirements
CRF
Historical tower identity is a provenance and temporal-validity finding.
CMDS
Broker and reinsurer identity does not establish a CMDS mechanism or stage.
COVE/F
The finding requires a verification gate that prevents historical counterparties from being represented as current.
Actors · conflicting Historical external actors are verified while current external identities remain unavailable.
Verify: Temporal identity validation for every named external counterparty · Before outreach or external attribution
COMPASS
Counterparty provenance has no COMPASS person-level assessment use.
Evidence still required: Current policy or authoritative counterparty confirmation
Test this finding
Validate: Obtain current policy or direct confirmation.
Falsify: A current authoritative source establishes the historical placement remains current.
Evidence: A-TDC-GA, A-TDC-NY-SEAL, A-10K
ECONOMIC EXPOSURE
Known money and modeled exposure are separated.
Booked / actual
$52,700,0002025 unfavorable PLGL reserve adjustment2025 Form 10-K
$28,600,000Q2 2026 PLGL reserve adjustmentQ2 2026 results
$49,000,0002025 EBITDA guidance reduction attributed to incremental PLGL expenseDecember 2025 PLGL update
Disclosed reserve
$181,800,000PLGL reserve at Dec. 31, 20252025 Form 10-K
$28,800,000Estimated insurance recoverable at Dec. 31, 20252025 Form 10-K
Modeled exposure
No governed value available.
Counterfactual value
No governed value available.
PROPAGATION
Where the exposure travels next.
Facility recurrence
Repeated mechanisms generate incidents and complaints.
Claim frequency / severity
Maturing events increase defense, settlement and reserves.
Captive layer
Professional-liability risk is financed through the captive up to disclosed limits.
Reinsurance recovery
Covered losses can propagate subject to terms and exclusions.
Earnings / capital
Reserve adjustments affect operating results.
STAKEHOLDER LEVERAGE
Exposure becomes actionable when the consequence chain is visible.
Select a stakeholder to trace exposure → newly visible intelligence → available action → downstream systemic effect. Every relationship remains bound to evidence state and explicit unknowns.
SELECTED STAKEHOLDER
Facility operations
- Exposure carried
- Operational recurrence, patient-safety events, complaints and control failures.
- IRI makes visible
- Governed recurrence and convergence patterns across facility evidence rather than isolated incidents.
- Available action
- Test recurrent mechanisms, attach corrective conditions and document closure evidence.
- Downstream systemic effect
- Fewer repeated mechanisms can reduce the upstream event stream that later matures into claims and enterprise exposure.
Evidence: A-10KClass: hypothesis
Evidence boundary
Facility-specific causation and StoneCrest-specific PLGL dollars are not established by enterprise disclosures.
INTERVENTION / VERIFICATION
Change the mechanism, then prove it changed.
Bind facility findings to policy-year loss runs and claim taxonomy.
Test whether governed recurrence findings lead claim frequency or severity.
Decision: Authorize the data join and define the pilot facility set.
Implementation: One governed policy-year data join using existing loss-run, RMIS and corrective-action records.
Economic effect: Tests an upstream control against publicly disclosed reserve and earnings volatility.
Target: Disconnect between facility evidence and actuarial development · Verify: Finding prevalence versus later claims by facility and policy yearAttach corrective conditions to recurrent mechanisms and monitor closure.
Create an auditable risk-control loop.
Decision: Approve required closure evidence and escalation thresholds.
Implementation: Add evidence requirements to the existing corrective-action workflow.
Economic effect: Reduces uncertainty about whether funded remediation changes recurrence before renewal.
Target: Repeated process failure · Verify: Recurrence and claim trend after verified closureUNCERTAINTY
What would change the conclusion.
Unknowns
Current third-party reinsurer identity.
Current broker of record.
Current policy-specific attachment points/exclusions beyond aggregate disclosure.
No public filing attributes enterprise reserve development specifically to StoneCrest.
Sensitivities
Reserves/recoverables change as claims develop.
Predictive value requires prospective validation.
Limits
Do not represent enterprise PLGL as StoneCrest-specific.
Do not represent historical TDC/Marsh as current.
METHOD / PROVENANCE
Evidence → CRF → CMDS → COVE/F → MAM + PHC → Action → Monitor/Verify
Enterprise disclosures establish context, not facility causation.
Observed money remains separate from modeled exposure.
Unknown external identities remain unknown.
| ID | Source | Scope | Class |
|---|---|---|---|
| A-10K | Acadia Healthcare 2025 Form 10-K | PLGL insurance, reserves and reinsurance | observed |
| A-Q2 | Acadia Healthcare Q2 2026 results | Q2 2026 PLGL adjustment | observed |
| A-PLGL | Acadia PLGL actuarial update | PLGL expense and guidance | observed |
| A-TDC-GA | TDC National Assurance Company v. Marsh USA, LLC | Historical TDC layer and Marsh broker role | observed |
| A-TDC-NY-SEAL | TDC National Assurance Company v. Marsh USA LLC — sealing orders | Sealing boundary | observed |