Chapter 22: Workers' Compensation Underwriting: The Mandatory Coverage with Unique Rules

"In workers' comp, you are not pricing a contract you wrote. You are pricing a statute you must obey, against a payroll you cannot yet count, on a company whose own three-year history has already decided most of the answer for you." — constructed teaching line, in the voice of a senior commercial underwriter

Overview

Open most policies in this book and the limit of the insurer's promise is printed on the declarations page: \$1 million per occurrence, \$2 million aggregate, a \$20 million building. Open a workers' compensation policy and you will look in vain for that number on the part that matters. The coverage that pays an injured worker's medical bills and lost wages has no dollar limit at all. It pays what the statute of the state where the injury happens says it pays — every covered medical dollar, every covered week of wage replacement, for as long as the law requires, even if that is the rest of a paralyzed worker's life. You did not write those benefits. A legislature did. Your job is to price a promise whose size someone else controls.

That is the first thing that makes workers' compensation — "workers' comp," "WC," or, in the older spelling, "workmen's comp" — unlike anything you have underwritten so far. The second is that you cannot even count the thing you are charging for until after the policy is over. The exposure base is payroll, and a company's payroll a year from now is a forecast, not a fact. So you bind on an estimate, and a premium auditor comes back at the end of the term to count what the payroll actually was and true up the bill. The third unique feature is the lever that makes WC the most personal line in commercial insurance: the experience modification factor, the X-mod, a single number that takes a company's own three-year loss history and turns it directly into a multiplier on its premium. A clean shop pays less than the class; a shop that hurts people pays more. No other common line hands the insured's own safety record back to them as a price quite so bluntly.

This chapter teaches you to underwrite inside those three rules. We start with the statutory bargain that created the coverage and why "no limit" is the right way to think about it. We learn the classification system — NCCI class codes and the governing class — because in WC, more than anywhere, the class is the price. We take apart the X-mod until you can read one cold and know what it is telling you and what it is hiding. We follow the payroll through the premium audit. We find the limits that do exist, on the employer's-liability side. We map the strange corners — the monopolistic state funds where you cannot write at all, the large-deductible and retro programs where the insured keeps the risk. And we end where the best WC underwriters spend their energy: return-to-work, the loss-control lever that bends the X-mod and the loss ratio together.

In this chapter, you will learn to:

• Explain workers' compensation as statutory coverage — a no-fault, no-dollar-limit benefit set by state law, not by the policy — and say why that changes how you price it.
• Use NCCI class codes and identify the governing class, and explain why payroll is the exposure base.
• Read an experience modification (X-mod) factor, interpret a debit or credit mod, and state what it can and cannot tell you.
• Describe the premium audit and why a WC premium is settled after the fact, not at binding.
• Distinguish the unlimited WC benefit from employer's-liability limits, and explain monopolistic state funds and loss-sensitive programs.
• Connect return-to-work and loss control to the X-mod, the loss ratio, and the combined ratio.

Learning Paths

This is a core commercial-lines chapter, and workers' comp shows up on every certification exam, so all readers should work the X-mod and classification sections closely. Here is where each path leans in:

🏠 Personal Lines: WC has no personal-lines twin, but the mechanism — handing an insured their own loss history back as a price — is exactly what telematics and credit-based scores (Chapter 14) try to do. Read §22.3 for the cleanest example in the book of "experience is the rate." 🏢 Commercial Lines: This is your line. The governing class (§22.2), the X-mod (§22.3), the audit (§22.4), and return-to-work (§22.7) are the daily tools of a middle-market underwriter. Harbor Steel's \$11M payroll lives here. 📊 Analytics: The X-mod is a credibility-weighted estimator (Chapter 10) hiding in plain sight — expected losses, actual losses, a primary/excess split, and a credibility factor. §22.3 is where the math of Chapter 10 becomes a real, filed, audited number that prices a real account. 📜 Certification: WC classification, the experience rating plan, statutory vs. employer's-liability coverage (Coverage A vs. Coverage B), and the audit are heavily tested in AINS and CPCU commercial sequences. The key terms here recur on every exam.

22.1 Statutory coverage: benefits defined by law, not the policy

Begin with the decision you are actually being asked to make, because it is stranger than it looks. A metal-fabrication company applies for workers' compensation. You will set a price and decide on terms. But unlike every other submission in this book, you will not decide what the coverage pays when a worker is hurt. That is already decided, by the workers' compensation statute of the state where the worker is employed. Your pricing question is therefore narrower and harder than usual: not "what should we promise and at what price," but "what will this employer's workers cost us under a promise we are required by law to make, and can we charge enough for it?"

Workers' compensation is a system, created by state statute, under which an employer is required to pay defined benefits to employees who suffer work-related injury or illness, regardless of fault, in exchange for the employee giving up the right to sue the employer in tort for that injury. That last clause is the heart of it. Before workers' comp, an injured worker's only remedy was to sue the employer and prove negligence — slow, uncertain, and often defeated by old common-law defenses. Workers' comp replaced that lottery with a bargain, struck across the industrialized world in the early twentieth century and sometimes called the grand bargain: the worker gives up the right to sue and the chance at a large jury verdict; in return the worker gets prompt, certain, no-fault benefits for every covered injury, without having to prove the employer did anything wrong. The employer gives up its common-law defenses; in return it gets a predictable, bounded-per-claim liability and immunity from most injury lawsuits by its own employees.

The phrase that captures the coverage you are pricing is statutory coverage: coverage whose terms, benefits, and obligations are set by statute rather than negotiated in the policy. On a workers' compensation policy, the insuring agreement for the core coverage essentially says the insurer will pay whatever the workers' compensation law requires — it incorporates the statute by reference. You will not find a per-occurrence limit on it because the statute does not impose one. The benefits typically fall into four buckets, and you should know them because they are what your premium has to fund:

• Medical benefits — the full cost of reasonable and necessary medical treatment for the work injury, usually with no dollar cap and no deductible to the worker. A severe burn or a spinal injury can generate medical costs for decades.
• Indemnity (lost-wage) benefits — partial replacement of lost wages, typically a percentage of the worker's average weekly wage up to a state-set maximum, for the time the worker cannot work.
• Disability benefits — scheduled or unscheduled payments for permanent impairment, from the loss of a finger (a "scheduled" benefit in many states) to total permanent disability.
• Death benefits — payments to surviving dependents, plus a burial allowance, when a work injury is fatal.

⚖️ Compliance Corner Workers' compensation is the most state-specific line you will ever underwrite, and the variation is not cosmetic. The benefit levels, the maximum weekly wage, the medical fee schedules, the list of presumed occupational diseases, the rules on which doctor the worker may see, and even whether the coverage is sold by private insurers at all (see §22.6) differ from state to state. The same welder, with the same injury, generates a different cost in two different states. This is why a multi-state employer's WC program is priced state by state, why the application asks for a payroll split by state, and why "what does the statute say?" is a question you will ask in every WC decision. McCarran-Ferguson (Chapter 4) left insurance regulation to the states; nowhere is that more visible than here.

What does "no limit" mean for you, the underwriter? It means the tail risk on a single claim is, in principle, very large. Most WC claims are small — a strained back, a laceration that needs stitches, a few weeks of light duty. But the distribution has a long, heavy tail: the young worker with a catastrophic head or spinal injury whose lifetime medical and indemnity costs can run into the millions. That shape — high frequency of small claims, low frequency of catastrophic ones — is frequency and severity (Chapter 6) in one of its purest commercial forms, and it drives two underwriting consequences you will meet again in this chapter: you cede the catastrophic tail to reinsurance (Chapter 27), and you watch frequency obsessively, because frequency is both a cost in itself and the leading indicator of the next severe claim.

22.2 Classification: NCCI class codes and the governing class

If statutory coverage tells you what the policy pays, classification tells you how much to charge before you have looked at the individual company at all. In workers' compensation, more than in any other line, the class is the price. So the first analytical task on any WC submission is to classify the operation correctly, and the consequences of getting it wrong are larger than almost anywhere else in commercial underwriting.

A class code is a numeric code that groups employers by the type of work their employees do, because the work is what determines the injury risk. A clerical worker, a retail clerk, a long-haul driver, and a structural-steel welder face wildly different odds of a wildly different injury, and the rating system prices each on its own loss experience. In most states this classification system is maintained by the National Council on Compensation Insurance (NCCI), a licensed rating and statistical organization that collects WC loss data across insurers, files class definitions and loss costs with state regulators, and administers the experience rating plan (§22.3). An NCCI class code is one of these four-digit codes — each tied to a published phraseology (the words that define what the class includes) and a loss cost (the expected loss per \$100 of payroll for that class). Some large states — California, New York, and several others — run their own independent rating bureaus rather than using NCCI, but the logic is the same: classify the work, attach a rate, and let each class pay for its own experience.

Here is the move that trips up newcomers and even some experienced commercial underwriters who do not write WC often. A single business does not get a separate price for each individual employee's job. The rules assign one governing class — the basic classification that best describes the business as a whole (in practice, usually the non-clerical, non-standard-exception class that carries the largest share of payroll) — and that governing class anchors how the risk is understood and rated, while a short list of standard exceptions (clerical office employees, outside salespersons, drivers in some states) are split out separately because their work is genuinely different and lower-hazard. You cannot pad the bill by calling half the welders "clerical," and you cannot save the insured money by sweeping the office staff into the welding class. The classification rules, filed and enforced by the bureau, decide which payroll goes where.

WC CLASSIFICATION OF A FABRICATION SHOP — illustrative codes & loss costs   [constructed teaching example]
                                                                  loss cost
  class (illustrative)                          annual payroll    per $100 payroll   pure loss
  ────────────────────────────────────────────────────────────────────────────────────────────
  Structural-steel fabrication (GOVERNING)        $7,200,000         $4.80           $345,600
  Iron/steel erection — outside (if applicable)   $1,300,000         $9.50           $123,500
  Drivers (split out)                             $1,500,000         $3.40            $51,000
  Clerical office (standard exception)            $1,000,000         $0.30             $3,000
  ────────────────────────────────────────────────────────────────────────────────────────────
  TOTAL PAYROLL  $11,000,000                                         (blended)       $523,100
  All figures illustrative. The governing class drives both the rate and how the account is understood.

The diagram makes the central fact visible: the welding/fabrication payroll, at a loss cost many times the clerical rate, is the account. Misclassify a million dollars of fabrication payroll into a lower class and you have quietly under-collected tens of thousands of dollars of premium — and, worse, mis-stated the risk to everyone downstream who relies on your file. Misclassify the office staff up into the governing class and you have over-charged the insured, handed the broker a reason to move the account, and possibly created a refund liability at audit. Classification is not clerical box-checking; it is the foundation the entire price stands on.

📋 At the Desk How you actually classify a risk: you do not guess from the company's name. You read the operations — from the application, the inspection (Chapter 8), the website, and ideally a conversation with the broker — and you match the actual work performed to the bureau's published class phraseology. Three habits keep you out of trouble. First, find the governing class by asking "what is the largest non-standard-exception payroll, and does its phraseology truly describe the bulk of the work?" Second, confirm that any split-out classes (clerical, drivers) really are separately maintained and the payroll is genuinely divisible — bureaus require actual division of payroll by records, not estimates. Third, watch for dual operations: a fabricator that also erects steel at job sites is doing two different things with two different hazards, and the erection class is far more dangerous than the shop. The failure mode here is the "convenient" classification — the one that makes the price competitive — applied to a risk that the audit will reclassify, leaving you to collect the difference from an unhappy insured, or eat it.

⚠️ Underwriting Trap The most expensive classification mistakes are not random errors; they are pressure. A broker working a price-sensitive account will, sometimes innocently and sometimes not, present the operations in the light most favorable to a low class — "mostly assembly, really," when the floor is full of arc welders, or "they don't do erection anymore," when last year's loss run shows a fall from height at a job site. The disciplined underwriter treats classification as a question of fact to be verified, not negotiated. If the governing class is structural-steel fabrication, it is structural-steel fabrication at the audit too, and pricing it as light assembly does not make the welders safer — it just guarantees that the premium you collected was never adequate for the losses that were always coming.

22.3 The experience modification factor (the X-mod)

Now we reach the lever that defines workers' compensation underwriting and the single most important number you will read on a WC submission. Classification tells you what an average employer in this class should pay. But you are not insuring an average employer; you are insuring this one, with its own safety record. The experience modification factor — universally called the X-mod, mod, or EMR (experience modification rate) — is the mechanism that adjusts the manual premium up or down to reflect a specific employer's own loss history relative to others in the same class. It is, in the cleanest form you will see anywhere in this book, the idea from Chapter 10 made operational: credibility-weight the company's own experience against the class, and let the result move the price.

Mechanically, the X-mod is a multiplier applied to the manual premium:

$$\text{Modified premium} = \text{Manual premium} \times \text{X-mod}$$

An X-mod of 1.00 is the class average — this employer's experience is exactly typical, and it pays the manual rate. An X-mod below 1.00 (say 0.85) is a credit mod: better-than-average experience, a 15% discount on the manual premium. An X-mod above 1.00 (say 1.25) is a debit mod: worse-than- average experience, a 25% surcharge. The number is computed by the rating bureau (NCCI or the state bureau) from the employer's own data — typically the most recent three years of payroll and losses, excluding the most recent policy year — using a published formula, and it is then filed and applied uniformly by every insurer. You do not get to invent the mod. You read it, you understand what produced it, and you decide what it means for the risk in front of you.

The detail that separates underwriters who understand the X-mod from those who merely apply it is how the formula treats frequency versus severity. The experience rating plan splits each claim into a primary portion (the first slice of every loss, up to a split point) and an excess portion (everything above it), and it weights the primary portion far more heavily. The reason is profound and worth internalizing: the plan is built to respond to frequency, not to the lottery of severity. Three small claims move your mod more than one large claim of the same total dollars, because three claims signal a pattern — a workplace that keeps hurting people — while one big claim may be bad luck. The plan, in effect, says: we will hold you sharply accountable for how often your workers get hurt (which you control, through safety) and only partly accountable for how badly a given injury turns out (which is often beyond your control). For the underwriter this is a gift, because it means the X-mod is largely a frequency signal, and frequency is the leading indicator of the severe claim that has not happened yet.

ANATOMY OF AN X-MOD — what drives the number          [constructed teaching example]

  EXPECTED LOSSES (from class + payroll) ........ what an average employer THIS SIZE in THIS CLASS should lose
  ACTUAL LOSSES (the company's own 3-yr history)  what this employer DID lose
        │                                          each claim split:  PRIMARY (first slice, weighted heavily)
        │                                                             EXCESS  (the rest, weighted lightly)
        ▼
  X-MOD ≈ credibility-weighted ratio of ACTUAL to EXPECTED
        ├─ actual < expected ........ mod below 1.00 (CREDIT — discount)
        ├─ actual ≈ expected ........ mod near 1.00  (average)
        └─ actual > expected ........ mod above 1.00 (DEBIT — surcharge)

  Bigger employer  → more credibility → mod swings further from 1.00 on the same relative experience.
  Frequency (many small claims) moves the mod MORE than one large claim of equal total dollars.

📊 Model vs. Judgment The X-mod is a filed, formulaic number, and that is exactly why an underwriter has to read behind it rather than just multiplying by it. The mod tells you the what — this employer ran worse than its class over three years — but never the why, and the why is the underwriting. A 1.35 mod produced by one catastrophic claim that has since driven a complete safety overhaul is a very different risk from a 1.35 mod produced by a steady drip of back strains and lacerations that are still happening. The model (the bureau formula) cannot see the new safety director, the cause of each loss, or whether the corrective actions took; you can, by reading the loss runs (Chapter 8) claim by claim. The mod is also backward-looking and lagging — it reflects claims from years ago and will not yet show a risk that is deteriorating right now. So treat the X-mod the way you treat any model output in this book: a powerful summary to be respected, interrogated, and sometimes overridden with documented reasons — never a verdict to be applied without thought.

There is a final, practical reason the X-mod matters beyond your own pricing: it has a life of its own in the marketplace. A company's mod is portable — it follows the company, not the insurer — and it is widely used outside insurance. General contractors routinely require subcontractors to carry a mod below 1.00 to even bid on a job; some owners write a maximum mod into their contracts. This means a debit mod is not just a higher premium; it can cost the insured work, which gives the better employers a powerful incentive to drive frequency down — an incentive you can harness through return-to-work and loss control (§22.7). It also means brokers and insureds care intensely about the mod, scrutinize the data that feeds it, and will (legitimately) dispute errors in the loss data that inflate it. Part of underwriting WC well is knowing that the mod is a number the insured is managing too, and that a sophisticated insured working its mod down is, all else equal, a better risk than one who has never looked at it.

🔍 Check Your Understanding 1. Two welding shops in the same class each have a 1.20 X-mod. Shop A's mod comes from one large head-injury claim three years ago; Shop B's comes from nine small back-and-laceration claims spread across three years. Which shop worries you more going forward, and why does the experience rating plan's frequency emphasis support your answer? 2. A broker tells you their client "deserves" a 0.80 mod because they are "very safe." Where does the 0.80 actually come from, and can you, the underwriter, simply grant it?

22.4 Payroll, premium audit, and the true exposure base

We have a class and a mod. Now we need the quantity they multiply: the exposure base, which in workers' compensation is payroll. (Recall from Chapter 21 that the exposure base is the measurable unit of exposure that the rate is applied to; for WC it is remuneration — payroll — per \$100.) The logic is sound: more payroll means more worker-hours of exposure to injury, and within a class, payroll scales with the number and pay of the people doing the hazardous work. The WC rate is quoted as a loss cost or rate per \$100 of payroll, so the premium build for a single class is simply:

$$\text{Manual premium (class)} = \frac{\text{Payroll}}{100} \times \text{Rate per \$100}$$

Then you sum across classes, apply the X-mod, apply any schedule credits/debits (Chapter 11) and the expense and profit loads, and you have a price. But here is the feature that makes WC operationally unlike property or GL: you are charging for a payroll that has not happened yet. When the policy incepts, next year's payroll is an estimate. So the premium charged at binding is a deposit premium (also called an estimated premium) based on the insured's projected payroll, and the true price is settled only after the term, through the premium audit.

A premium audit is the post-term examination of the insured's actual records — payroll registers, tax filings (such as the federal 941s and state unemployment reports), the general ledger, and overtime and classification detail — to determine the actual exposure that occurred during the policy period, so the final premium can be calculated and the bill trued up. If the company grew and its actual payroll exceeded the estimate, the insured owes additional premium. If it shrank, the insured gets a return. The audit also verifies the classifications — and this is where a "convenient" classification from §22.2 comes home to roost, because the auditor classifies what the records actually show, not what the application optimistically claimed.

WHY A WC PREMIUM IS NEVER FINAL AT BINDING — the audit cycle    [constructed teaching example]

  BIND ───────────────► POLICY TERM (12 months) ───────────────► AUDIT ───► FINAL PREMIUM
   │                          │                                      │
   estimated payroll          actual payroll accrues                auditor counts actual
   → DEPOSIT premium          (growth? layoffs? overtime?)          payroll + verifies class
                                                                     → additional or return premium

  Deposit premium = a PRICE PER THE ESTIMATE.   Final premium = a PRICE PER WHAT ACTUALLY HAPPENED.

📋 At the Desk Underwriting the audit is underwriting the exposure estimate, and it is a real source of profit and leakage. Two habits matter. First, sanity-check the payroll estimate against reality: a fabricator doing \$45M in revenue with \$11M of payroll has a believable labor ratio; an estimate that suddenly drops 30% at renewal with no change in operations is a flag — is the business really shrinking, or is the insured (or broker) low-balling the deposit to win on price, planning to "pay it at audit"? A deposit set deliberately low is a cash-flow trick that leaves you under-collateralized all year. Second, understand audit leakage: uncollected additional premiums on accounts that grew, audits that get waived or disputed, payroll that the insured fails to document by class (which the rules may then charge at the highest-rated applicable class). On a growing account, a disciplined audit is found money; on a shrinking or evasive one, a sloppy audit is a hole in your loss ratio you never see coming.

Two payroll subtleties are worth knowing because they change the number. Overtime is generally included in payroll for WC, but in most jurisdictions only at the straight-time portion — the "extra half" of time-and-a-half is excluded, because the injury exposure of an overtime hour is the same as a regular hour even though the wage is higher. And payroll for certain high-earning individuals — owners, officers, sometimes executives — may be subject to minimums and maximums set by the state, so a single owner's large salary does not distort the premium as if it were ten welders' worth of exposure. These rules are filed and mechanical, but they matter: getting overtime or executive payroll wrong is a common audit dispute and a common source of mispriced deposits.

22.5 Employer's liability and the limits that do exist

Workers' compensation has "no limit," and yet a WC policy does contain dollar limits — just not on the part most people mean. Understanding which coverage is unlimited and which is bounded is essential, because they respond to different claims and the bounded part is where your catastrophe-on-paper actually shows a number. A standard workers' compensation and employers' liability policy contains two coverages, and the distinction is one of the cleanest things you can teach a newcomer:

• Part One — Workers' Compensation (the statutory coverage). This is the no-fault, no-dollar-limit benefit we have been pricing: it pays whatever the state WC statute requires for a covered work injury. No per-claim limit appears because the statute imposes none. This is the coverage the grand bargain created.
• Part Two — Employers' Liability. This is a liability coverage, and it does have dollar limits. It responds to the situations where an injury-related claim against the employer falls outside the no-fault statutory bargain — and so is not paid as a WC benefit but is instead a lawsuit the employer must defend.

Employer's liability coverage protects the employer against legal liability for work-related injury or illness in the circumstances the workers' compensation statute does not exclusively govern. When would an injured-worker situation become a lawsuit rather than a statutory claim? The classic categories include: third-party-over actions (a worker injured on a job site sues a third party — say, a general contractor or a machine manufacturer — who then turns around and sues the employer for contribution); dual-capacity claims (the employer is also, say, the manufacturer of the product that hurt its own employee); injuries to people who fall outside the WC system in that state; and certain consequential claims by family members. These are exactly the situations the no-fault bargain does not neatly cover, and that is why a liability coverage with limits sits alongside the unlimited statutory part.

Part Two limits are typically written as three numbers — for example, an illustrative \$1,000,000 each accident / \$1,000,000 disease–policy limit / \$1,000,000 disease–each employee — and those limits are what your umbrella (Chapter 16) and reinsurance attach over. Note carefully: the umbrella sits over employer's liability (Part Two), not over the statutory WC benefit (Part One), because Part One has no limit for an umbrella to be "excess" of. A newcomer who tells a broker the \$10M umbrella "increases the workers' comp limit" has misunderstood the policy; the umbrella increases the employer's-liability limit and the auto and GL limits, while the statutory benefit remains whatever the statute says, reinsured rather than capped.

⚖️ Compliance Corner The exclusive-remedy protection — the employer's immunity from being sued by its own injured employees — is the consideration the employer received in the grand bargain, and it is robust but not absolute. States carve out exceptions (for an employer's intentional or egregious conduct, for the third-party-over and dual-capacity situations above, and for employers who fail to carry required WC coverage at all). An employer that illegally goes without workers' comp generally loses the exclusive-remedy shield and faces statutory penalties — one more reason WC is, in most states for most employers, mandatory rather than optional. As an underwriter you should also know that some categories of worker (certain agricultural or domestic workers, sole proprietors, some independent contractors) may be exempt or able to opt out under state law, which affects who is actually in the covered payroll — and which the audit will sort out.

22.6 Monopolistic state funds and loss-sensitive programs

Two structural features of the workers' comp market change whether and how you can write a risk at all, and a commercial underwriter has to know both. The first is geographic: in a handful of states, you cannot write workers' compensation as a private insurer, period. The second is about program structure: above a certain size, the most sophisticated insureds stop buying first-dollar coverage and start keeping their own risk, which changes your role from pricing losses to financing them.

A monopolistic state fund is a state-run workers' compensation insurer that is the only lawful source of WC coverage in that state — private carriers are prohibited from writing the statutory coverage there. A small number of U.S. states (historically including Ohio, Washington, Wyoming, and North Dakota) operate this way; employers in those states buy their WC from the state fund, not from you. This matters immediately for a multi-state account: a national employer with operations in a monopolistic state cannot get that state's WC from your policy, so the program is split, and a separate employers'-liability solution (often a "stop-gap" endorsement on the package, since the state fund provides only the statutory benefit and not Part Two) has to fill the liability gap. Do not confuse a monopolistic fund with a competitive state fund: many states (California's is the largest example) have a state-chartered fund that competes alongside private insurers, often as the insurer of last resort, and there you can and do write the business. The underwriting point is procedural but unforgiving — before you quote a multi-state WC account, you check which states are monopolistic and carve them out, because you legally cannot cover them.

The second structural feature is the menu of loss-sensitive programs that large employers use to take back risk in exchange for a lower fixed cost. Recall the deductible and retrospective-rating ideas from Chapters 11 and 12; WC is where they get their fullest commercial expression:

• Large-deductible plans. The employer pays the first slice of every claim (an illustrative \$250,000 or \$500,000 per occurrence), and the insurer pays above it. The insurer still issues the statutory policy and pays every claim from the first dollar (the worker must be paid no matter what), then reimburses itself from the employer for the deductible portion — which means you are now also a credit underwriter, because you are exposed if the employer cannot pay back the deductible losses. Collateral (a letter of credit, a surety bond, cash) secures that exposure.
• Retrospective rating (retro). The final premium is adjusted after the term based on the insured's actual losses during the term, within a negotiated minimum and maximum. A good loss year produces a premium near the minimum; a bad one, near the maximum. The insured is, in effect, paying largely for its own losses plus a fee for the insurer's services and the cap.
• Self-insurance. The largest and most financially strong employers may self-insure WC entirely (where the state permits and they qualify), buying only excess workers' compensation coverage above a high retention to cap the catastrophic single claim. Here you are not pricing the working losses at all — you are selling protection against the tail.

📋 At the Desk The decision of which structure fits a risk is a real underwriting judgment, and it turns on size, loss predictability, and financial strength. The logic mirrors the law of large numbers (Chapter 1): a large employer with a stable, credible loss history has enough of its own experience that its losses are fairly predictable, so it rationally keeps the predictable layer (via a deductible or retro) and pays you mainly to absorb the volatile tail and to handle claims — which is cheaper for it than paying a full guaranteed-cost premium loaded for your uncertainty. A small employer's losses are not credible on their own, so it should buy guaranteed-cost coverage and let your pool absorb the swings. The trap is letting a mid-size employer with a deteriorating book talk you into a deductible plan it cannot collateralize — now you have given up premium and taken on credit risk on a risk that is getting worse. Match the program to the risk's true credibility and balance sheet, not to the broker's preference.

22.7 Return-to-work and the underwriter's loss-control lever

We end where the best workers' comp underwriters and risk-control professionals spend their attention, because it is the one place where the underwriter can actually change the risk rather than merely price it. Most of this chapter has been about reading a fixed history — the statute is fixed, the class is fixed, the X-mod is computed from losses already incurred. Return-to-work is different: it is forward-looking, it is something the insured can do starting tomorrow, and it bends both the loss ratio and the X-mod in the same direction. If you understand only one loss-control lever in WC, understand this one.

The economics are unintuitive until you see them. The largest controllable cost in a workers' comp claim is usually not the medical treatment — it is the indemnity (lost-wage) payments that accrue for every week an injured worker stays off work. A laceration that needs ten stitches costs a little in medical and, if the worker is back on light duty in three days, almost nothing in indemnity. The same laceration, where the worker sits at home for four months because the employer has no light-duty job to offer, can cost many times more in indemnity — and worse, the longer a worker is out, the less likely they are ever to return, as deconditioning, depression, and the claim's own momentum set in. A return-to-work program — a formal arrangement to bring injured workers back to modified or light duty as soon as they are medically able — attacks exactly this. It shortens the indemnity tail on every claim, keeps the worker engaged and recovering, and prevents the moderate claim from drifting into a permanent disability.

Now connect it to the X-mod, and the lever becomes powerful. Recall from §22.3 that the experience rating plan weights the primary (frequency-driven, small-dollar) portion of each claim most heavily. A robust return-to-work program does not just save indemnity dollars; by getting workers back quickly and keeping medical-only claims from becoming lost-time claims, it holds down the very claim costs the X-mod is most sensitive to. Over three years, an employer with disciplined return-to-work and frequency control will see its mod drift below 1.00 — which lowers its premium, which (recall §22.3) can also win it contracts that require a sub-1.00 mod. The underwriter's role is to require, credit, and verify this: to make a return-to-work commitment a condition or a schedule credit (Chapter 11) on a marginal account, and to check at renewal that the program is real, not a binder on a shelf.

⚠️ Underwriting Trap The seductive error is to underwrite WC as a severity line — to fixate on the one big claim and treat the steady stream of small lost-time claims as background noise. It is the reverse. In WC, frequency is the disease and severity is often the symptom: a shop that produces many small injuries is a shop with a safety-culture problem, and that culture is what eventually produces the catastrophic claim. The X-mod is built around this truth (it weights frequency), and so should your judgment. An account with a clean severity record but a rising frequency trend is not a good risk getting lucky on big losses; it is a bad risk that has not had its bad day yet. Price and condition on the frequency trend and the loss-control response, not on the comforting absence of a single large claim.

🤖 Model vs. Judgment A modern WC underwriting model will ingest the class, the payroll, the three-year loss runs, the X-mod, and increasingly external signals (OSHA history, even the granular cause-of-loss coding) and produce a risk score. It is genuinely good at the frequency pattern — better than a human at spotting that this shop's small-claim rate is quietly two standard deviations above its class. What it cannot yet do well is evaluate the response: whether the new safety director is the real thing, whether the return-to-work program described in the broker's submission actually functions on the floor, whether the corrective actions after the 2023 fire (Harbor Steel, below) have changed the culture or just the paperwork. So the division of labor is clear and it is the book's recurring lesson: let the model flag the frequency signal you might have missed, and bring human judgment to the question the model cannot answer — is this risk getting better or worse, and will the controls hold?

🗂️ The Underwriting File

Harbor Steel's workers' compensation — the debit X-mod, the welding governing class, and the return-to-work credit. This is the line where Harbor Steel's people, not its building, are the exposure. The submission from Meridian Risk Partners lists roughly \$11 million of payroll across welders, fabricators, drivers, and office staff — about 180 employees in all. Your task this chapter is to classify the risk, read its mod, and set the WC pricing posture, all on the facts you already have. (Constructed teaching figures throughout; this is the Underwriting File, not a real account.)

Classification. The governing class is structural-steel/metal fabrication — the welders and fabricators on the shop floor carry the bulk of the payroll and the hazard, and the class loss cost is several times the clerical rate. The drivers (the 12-unit flatbed fleet's operators) split out into a driver class, and the office staff into the clerical standard exception. You flag one thing to verify at inspection and audit: does Harbor Steel erect steel at customer job sites, or only fabricate in the shop? Field erection is a far more hazardous class (falls from height), and if any meaningful payroll belongs there, it must be classified — and priced — as such. You will not let it ride as "fabrication."

The X-mod. Harbor Steel's loss history includes several workers'-comp claims over the period — back injuries and a serious laceration near-miss — on top of everything else in the file. Given that frequency, you expect a debit X-mod (above 1.00): the company has run worse than its class, and the experience rating plan, weighting those repeated small-to-moderate claims, pushes the mod up. You read the loss runs claim by claim (Chapter 8) and ask the §22.3 question the mod alone cannot answer — are these back strains and lacerations still happening, or did they cluster before a change in management and safety practice? That answer, not the bare mod, decides how worried you are.

The pricing posture and the lever. You price WC at a debit X-mod, consistent with the losses, and you reach for the one tool that can bend the number: a return-to-work credit, made real as a condition and/or schedule credit. Welding and fabrication are exactly the operations where light-duty work exists in abundance (inspection, layout, fixturing, materials handling), so a credible return-to-work program is both feasible and high-value here — it will shorten the indemnity tail on the back injuries that dominate the loss runs and, over three years, work the mod back toward 1.00.

What this settles and what it doesn't. WC is now priced: a debit X-mod plus a return-to-work credit, with the welding/fabrication governing class confirmed and the erection question flagged for inspection. What remains open is the same thing the mod cannot tell you — whether the safety culture is improving — and the audit exposure on the \$11M payroll estimate. The catastrophic single-claim tail (the young worker with a life-altering injury) is real but, like the property cat, ceded to reinsurance rather than priced into the working layer; we will treat that in Chapter 27. The account moves forward with WC handled; the running disposition is unchanged in shape — a difficult but workable, controllable risk that earns its price if the conditions are met.

Conclusion

Workers' compensation breaks three of the habits the rest of this book builds, and learning to underwrite it is learning to work inside those breaks. The coverage has no dollar limit because a statute, not your policy, defines the benefit — so you price a promise whose size you do not control, and you watch the heavy tail you cannot cap. The exposure base is payroll you cannot yet count, so you bind on an estimate and settle the true price through the premium audit — which makes the deposit a forecast and the audit a source of both found money and quiet leakage. And the X-mod hands the insured's own three-year loss history back as a multiplier on its price, weighting frequency over severity, which makes WC the most personal, most behavior-driven line in commercial insurance — and the one where return-to-work and loss control let the underwriter actually change the risk rather than merely price it.

Two of the book's themes run straight through this chapter. Pricing follows risk here with unusual directness: the X-mod is the company's risk, expressed as a number, and the discipline is to honor it — not to talk yourself into a credit the data does not support, and not to let a "convenient" classification under-collect the premium the losses will demand. And underwriting is judgment: the mod, the class, and the audit are mechanical and filed, but the questions that decide whether Harbor Steel's WC is a good risk — is the frequency trend improving, did the corrective actions take, will the return-to-work program really run on the floor — are exactly the ones no formula answers, and exactly the ones the combined ratio will grade you on two and three years from now.

In the next chapter we move from the people to the vehicles: commercial auto and fleet underwriting, a line in genuine crisis, where nuclear verdicts and severity inflation have made driver selection, telematics, and the management of a fleet's exposure as consequential as anything in commercial lines — including Harbor Steel's twelve-truck fleet, where one poor driving record is already waiting in the file.

Key Terms

• Workers' compensation — a statutory, no-fault system requiring employers to pay defined benefits to employees injured or made ill by their work, in exchange for the employee giving up the right to sue the employer in tort (the "grand bargain").
• Statutory coverage — coverage whose benefits and terms are set by statute rather than by the policy; on a WC policy the insurer agrees to pay whatever the state workers' compensation law requires, with no dollar limit.
• NCCI class code — a numeric classification, maintained by the National Council on Compensation Insurance (or an independent state bureau), that groups employers by the type of work their employees do and carries a published loss cost per \$100 of payroll.
• Governing class — the basic classification that best describes a business as a whole (usually the largest non-standard-exception payroll), which anchors how the risk is rated, with standard exceptions (clerical, drivers) split out separately.
• Experience modification (X-mod) — a filed multiplier applied to manual premium that adjusts price up (debit, >1.00) or down (credit, <1.00) to reflect an employer's own multi-year loss experience relative to its class, weighting frequency over severity.
• Premium audit — the post-term examination of an insured's actual payroll and records to determine the true exposure that occurred and settle the final premium against the estimated (deposit) premium charged at binding.
• Monopolistic state fund — a state-run insurer that is the only lawful source of workers' compensation coverage in that state, where private carriers are prohibited from writing the statutory coverage.
• Employer's liability — the second coverage on a WC policy (Part Two), a liability coverage with dollar limits that responds to work-injury claims falling outside the exclusive no-fault statutory bargain (e.g., third-party-over and dual-capacity actions).

Spaced Review

1. A WC policy has "no limit" on Part One but dollar limits on Part Two. Explain which coverage is which, which one the \$10M umbrella sits over, and why. (§22.1, §22.5)
2. Two welding shops in the same NCCI class submit at the same payroll. Shop A has a 0.82 X-mod, Shop B a 1.28. Before reading a single loss run, what does each mod tell you — and what does it not tell you that you must read the losses to learn? (§22.3)
3. (Back to Chapter 10.) The X-mod is a credibility-weighted estimator. Explain how a larger employer's mod swings further from 1.00 than a small employer's on the same relative experience, and tie that to the credibility idea. (§22.3; Ch. 10)
4. (Back to Chapter 6 / Chapter 21.) Workers' comp uses payroll as its exposure base, and so does much of general liability. Why is payroll a sensible measure of injury exposure, and what is one way it can mislead? (§22.4; Ch. 6, Ch. 21)
5. (The recurring pricing-discipline question.) You are tempted to grant Harbor Steel a credit for its "commitment to safety" and quote a mod-equivalent discount the loss data does not support. Would that help or hurt the combined ratio, and what is the disciplined alternative that still rewards real safety? (§22.3, §22.7; Ch. 3, Ch. 11)