Workers' Compensation Ratemaking

Workers' compensation (WC) ratemaking has its own exposure base, its own rating bureau infrastructure (NCCI in most U.S. states, independent bureaus elsewhere), and a distinctive individual-risk rating layer (experience rating, schedule rating, retrospective rating) layered atop manual rates. This article walks through each piece with a fully worked experience-mod example.

Exposure base: payroll

WC exposure is payroll per $100 (in NCCI states), classified by job classification code (class code) reflecting hazard level (e.g., clerical vs. roofing). Manual premium is

$\text{Manual Premium} = \sum_c \frac{\text{Payroll}_c}{100} \times \text{Rate}_c,$

before any experience modification, schedule credit/debit, or premium discount is applied.

Loss costs, LCMs, and class relativities

NCCI publishes loss costs by class code — the pure loss cost per $100 of payroll, excluding expense and profit loadings — which each carrier converts to a rate by applying its own loss cost multiplier (LCM): $\text{Rate} = \text{Loss Cost} \times LCM$, where the LCM loads for the carrier's own expense ratio, profit provision, and any company-specific loss experience adjustment. Class relativities are the ratio of a class's loss cost to that of a base class, built from countrywide/state class-level frequency and severity experience and credibility-weighted with the overall state indication (see Credibility Theory).

Benefit-level and law-amendment factors

WC benefits are statutory and legislated (indemnity benefit percentages, maximum/minimum weekly benefits, medical fee schedules). When benefit levels change legislatively, historical losses must be adjusted with a benefit-level factor — the ratio of benefits payable under current law to those payable under the law in effect when the loss occurred — before use in ratemaking or experience rating. A law amendment factor is applied prospectively to reflect a benefit change effective after the experience period but before the future policy period, analogous to trend but driven by statute rather than economic inflation.

Experience rating

The experience rating plan compares an individual employer's actual loss experience over the last three (typically) complete policy years to what is expected for its industry and size, producing an experience modification factor (mod). Two structural features:

1. Split losses: each claim's losses are split into a primary portion (the first $ threshold — e.g., NCCI's "split point," which has evolved to a per-claim value in the tens of thousands) and an excess portion above it, because large claims are more volatile and should get less weight.
2. Credibility weighting via a credibility factor $Z$ (or, in NCCI's actual formula, weights $W$ for primary excess losses and a ballast constant $B$) that limits the influence of small or volatile risks.

The classical (illustrative) mod formula:

$M = \frac{A_p + W A_e + (1-W)E_e + E_p}{E_p + E_e}$

where $A_p$ = actual primary losses, $A_e$ = actual excess losses, $E_p$ = expected primary losses, $E_e$ = expected excess losses, and $W$ is the weight applied to actual excess losses (larger risks get a higher $W$, giving more credibility to their own excess experience).

Worked example

An employer has 3-year actual primary losses A_p = \120{,}000$, actual excess losses$A_e = $180{,}000$, expected primary losses$E_p = $100{,}000$, expected excess losses$E_e = $150{,}000$, and a credibility weight for excess losses$W = 0.20$ (a mid-size risk).

$M = \frac{120{,}000 + 0.20(180{,}000) + 0.80(150{,}000) + 100{,}000}{100{,}000+150{,}000} = \frac{120{,}000+36{,}000+120{,}000+100{,}000}{250{,}000} = \frac{376{,}000}{250{,}000} = 1.504.$

This employer has an experience mod of 1.504 — its manual premium is increased 50.4% because its actual (primary + credibility-weighted excess) losses substantially exceed its size-adjusted expected losses. Manual premium of $500,000 becomes modified premium of $752,000.

Schedule rating and retrospective rating

Schedule rating applies a debit or credit (commonly ±25% or more with company/regulatory approval) for qualitative underwriting factors not captured by the mod — safety programs, management cooperation, medical management. Retrospective rating ties final premium to the insured's own current-term losses, subject to a minimum and maximum:

$R = (B + cL)\times T, \qquad R_{min} \le R \le R_{max},$

where $B$ is basic premium (loads expenses and the cost of the min/max corridor), $c$ is the loss conversion factor (loads ULAE), $L$ is actual incurred losses, and $T$ is a tax multiplier. Large deductible plans shift the risk of small/medium losses to the insured (who reimburses the insurer up to the deductible) while the insurer retains excess layer risk and collateral/credit risk on the deductible reimbursements.

Pitfalls

• Applying a single state's mod formula parameters to another state — split points, $B$ values, and $W$ tables are jurisdiction-specific (NCCI vs. independent bureau states like CA, NY, PA, TX).
• Forgetting to adjust historical losses for benefit-level changes before comparing accident years or computing trend.
• Double counting schedule credit and experience mod effects for the same underwriting characteristic (e.g., crediting for "loss control" both in schedule rating and informally in underwriting judgment).
• Ignoring that the mod is capped/floored in most states (e.g., minimum mod around 0.xx not zero) and that very large risks may be fully credible ($Z=1$, no ballast).

Exam relevance

WC-specific ratemaking, experience rating, schedule and retro rating are core to CAS Exam 8, with foundational loss-cost and classification concepts also appearing on CAS Exam 5.