NCCI’s July 8 research brief on the relationship between worker wages and WC loss experience confirmed a pattern that underwriters have discussed informally for years: loss ratios fall consistently as worker wages rise, and the primary driver is claim frequency, not per-claim cost. For self-insured employers concentrated in higher-wage occupations, the reserve implication is direct and has received almost no attention.
The reserve mechanism
When an actuary sets IBNR under the Bornhuetter-Ferguson method, expected losses typically derive from payroll multiplied by an industry pure premium rate. That rate encodes the average-industry claim frequency across all wage tiers in the classification. For an employer whose workforce earns substantially above classification averages, the embedded frequency assumption is too high. The a priori expected losses overstate what the program will actually produce, and IBNR is inflated as a result.
NCCI’s brief quantified the severity side: moving up one wage tier (roughly 25 percentile points) corresponds to about a 15% increase in combined claim severity, with indemnity severity rising approximately 21% per tier and medical severity around 10%. But frequency declines enough across wage tiers that the net effect on aggregate loss cost per dollar of payroll is negative. Payroll grows linearly with wages; claim counts do not follow the same slope.
A structural explanation is tenure. NCCI noted that short-tenured workers are nearly twice as likely to be injured as full-tenured workers. Higher-wage employees tend to stay in their roles longer, which mechanically reduces injury rates independent of job hazard. When payroll expands because existing employees are earning more, the claim frequency baseline does not move with it.
The reserve error is not in the development factors or the tail. It sits one level upstream, in the expected loss calibration that anchors the credibility-weighted development method.
Who it affects
Self-insured employers in financial services, technology and software, healthcare management and administration, and senior or supervisory roles in construction face the largest potential overstatement. The common thread is a workforce where average wages substantially exceed the classification benchmark embedded in NCCI loss costs. NCCI’s brief noted that benefit maximums begin affecting indemnity calculations above 150% of state average weekly wage, a threshold only about 7% of claimants reach, so the frequency dynamics dominate the math well before statutory caps enter the picture.
Employers in these segments may also notice that their own development is running below expected across multiple accident years on the WC development triangle. That pattern is not random favorable experience; it may reflect a structural overstatement baked into the a priori loss rate.
What this means for your next review
Ask your actuary what expected pure premium rate anchors the a priori in your WC IBNR, and whether it was calibrated against your own workforce’s wage distribution or an all-industry NCCI benchmark. If your program carries three or more accident years of credible data, compare your actual claim frequency rate per unit of payroll against the frequency implied by the expected loss rate. A persistent gap, your actual frequency running below expected year after year, is a signal worth addressing formally rather than absorbing into favorable development.
NCCI identified limited payroll (payroll caps by class code) as a potential system-level fix, noting California already applies caps for select codes. No broader state filing has been proposed. For self-insured programs, a recalibration of the expected pure premium in the actuarial opinion is the available correction today. Once the a priori is reset to reflect below-average frequency history, IBNR decreases across every accident year still in development, not just the most recent.