Chapter 21: Commercial General Liability Underwriting: The Coverage That Protects Against Lawsuits

"You can insure the building for what it's worth. You cannot insure the lawsuit for what it's worth, because nobody knows what it's worth until twelve years from now a jury tells you." — A senior casualty underwriter's line, constructed here to make the point this chapter exists to prove: property loss has a ceiling you can see today; liability loss has a tail that runs into a future you are pricing blind.

Overview

Here is the account on your desk and the quiet menace inside it. A metal-fabrication shop wants general liability coverage. The premises hazard is easy to see and easy to price — a visitor slips on the shop floor, a delivery driver is struck by a forklift, the kind of loss that happens, gets reported, gets paid, and closes inside a year or two. You could underwrite that exposure in your sleep. But this shop also fabricates structural brackets that get welded into other people's buildings, and that single fact turns a tidy premises risk into something else entirely: a products-completed operations exposure with a tail that may not surface for a decade. A bracket fabricated this year, installed next year, holding up a mezzanine for eight years, that fails in the ninth and drops a load on someone — that is a claim you are accepting today, at today's price, against a severity and a legal environment you cannot see. The premises hazard is the part of general liability you can underwrite from the application. The products tail is the part that decides, years later, whether the account was a winner or a disaster.

Commercial general liability — CGL, the most-purchased liability policy in commercial insurance — is the workhorse that stands between a business and the lawsuits arising from its premises, its operations, and the products and work it leaves behind. Almost every business buys it; it is the foundation of nearly every commercial program and the platform the umbrella sits on. And it is harder to underwrite than its ubiquity suggests, because liability is a long-tail line: unlike a fire, which announces itself, declares its size, and closes, a liability claim can be reported years after the act that caused it, litigated for years more, and settled for an amount that bears no resemblance to anything the application could have told you. The discipline of CGL underwriting is the discipline of pricing a promise whose cost arrives late and arrives uncertain.

This chapter builds that discipline in seven steps. We open the policy and read what it actually covers — the three insuring agreements and the exclusions that shape them. We draw the single most important distinction in liability coverage, the occurrence versus claims-made trigger, and explain why it governs when a claim is covered at all. We work the exposure base — how payroll, sales, or area become premium through classification. We confront the industry-specific hazards and the loss-sensitive products tail that separate a safe liability account from a dangerous one. We read the additional-insured and contractual-liability demands that modern commerce attaches to every account. We place the umbrella over the CGL. And we preview the management and professional lines the CGL pointedly does not cover, which the next chapters take up.

In this chapter, you will learn to:

• Explain what commercial general liability (CGL) covers across premises/operations, products-completed operations, and personal & advertising injury — and what it excludes.
• Distinguish the occurrence trigger from the claims-made trigger and say which the standard CGL uses and why it matters.
• Choose the right exposure base for a class and explain how classification turns operations into premium.
• Identify the loss-sensitive products tail and the industry-specific hazards that make a liability account riskier than its premium suggests.
• Read an additional insured request and a contractual-liability assumption and price what each does to your exposure.
• Place excess and umbrella liability over the CGL and preview E&O, D&O, and EPL.

Learning Paths

This chapter is the backbone of commercial casualty underwriting, and while every reader should work the trigger and the products tail, here is where each path leans hardest:

🏠 Personal Lines: You write liability too — the personal umbrella (Chapter 16) sits over the same kind of bodily-injury and property-damage exposure. Read §21.1 and §21.2 for the coverage logic, then watch how commercial liability adds the products and operations tail that personal lines mostly lack. 🏢 Commercial Lines: This is your chapter. The exposure base (§21.3), the loss-sensitive tail (§21.4), and the additional-insured machinery (§21.5) are the daily work of a casualty desk, and the Harbor Steel products exposure is the watch-item that runs through the rest of the book. 📊 Analytics: Classification (§21.3) is where rating data lives, and the long tail (§21.2, §21.4) is the reason liability pricing leans so hard on development and trend (Chapter 10). Note why a claims-made book and an occurrence book reserve and model differently. 📜 Certification: §21.1–§21.5 map directly to the CGL coverage, trigger, and exposure-base material tested in the AINS and CPCU commercial-lines sequences. The occurrence/claims-made distinction and the products-completed operations hazard are perennial exam points.

21.1 What CGL covers: premises/operations, products/completed ops, personal & advertising injury

Start with the question the policy answers, because the coverage grant only makes sense as a reply to it. A business, simply by existing and operating, creates the risk that it will injure someone or damage someone's property and be held legally liable for it. A customer slips in the lobby. A scaffold the company erected collapses on a passerby. A component the company manufactured fails and starts a fire. An employee of the company, putting up a sign, defames a competitor. In each case a third party suffers harm, points at the business, and demands money — through a lawsuit, or under threat of one. The CGL is the coverage that responds to that demand: it pays the sums the insured becomes legally obligated to pay as damages, and — critically — it pays to defend the insured against the claim, even a groundless one.

Commercial general liability (CGL) is the standard commercial policy that covers an insured's legal liability to third parties for bodily injury, property damage, and certain personal and advertising injuries arising out of the insured's premises, operations, products, and completed work. Two features of that definition do most of the work and are worth pulling out immediately. First, it is third-party coverage: it protects the insured against claims by others, not against the insured's own losses (that is what property insurance, Chapter 19, does). Second, it carries a duty to defend that is typically broader than the duty to pay — the insurer must defend any claim that could fall within coverage, and those defense costs are usually paid in addition to the policy limits. That defense obligation is not a courtesy; on many liability accounts the defense cost is a large share of the total claim, and a frivolous suit the insured wins still costs real money to win.

The standard CGL (the widely used ISO/Verisk form — see Chapter 5 on bureau forms) organizes its promise into three coverage parts, usually labeled Coverage A, B, and C. You should be able to recite them, because every exposure you assess will sort into one of the three.

THE THREE INSURING AGREEMENTS OF THE CGL                              [constructed teaching example]

  COVERAGE A — Bodily Injury & Property Damage Liability
     │   the core grant: third-party BI and PD the insured is legally liable for
     │   ┌─ PREMISES/OPERATIONS ── the slip-and-fall, the ongoing-work injury (happens AT/DURING the work)
     │   └─ PRODUCTS-COMPLETED OPS ── the product or finished work that fails AFTER it leaves the insured
     │
  COVERAGE B — Personal & Advertising Injury Liability
     │   non-physical "injury" offenses: libel, slander, false arrest, wrongful eviction,
     │   copyright/slogan infringement in the insured's advertising
     │
  COVERAGE C — Medical Payments
         small, no-fault payments for injuries on the premises, regardless of liability —
         a goodwill coverage that heads off small suits

Hold the two halves of Coverage A apart in your mind, because they are underwritten differently and they behave differently in the loss data. Premises/operations is the liability arising from the insured's ongoing activities — the condition of its premises and the work it is currently performing. The harm happens while the insured is operating: the customer falls on the wet floor today, the trench the crew is digging caves in today. This is the near-term, observable, relatively short-tail part of general liability. You can inspect a premises, read the slip-and-fall history, and price it with reasonable confidence.

Products-completed operations is the liability arising from the insured's products and finished work after the insured has parted with them — after the product is sold and in someone's hands, after the construction job is finished and turned over. This is the long-tail half, and it is where the real underwriting risk in CGL usually lives. The harm happens later, sometimes much later, far from the insured's control, in an environment the insured no longer governs. The fabricated bracket fails in year nine. The food product sickens consumers across three states. The contractor's finished roof leaks two winters after the crew left. We give this hazard its own section (§21.4) because it deserves it.

📋 At the Desk The first thing you do with any CGL submission is sort the insured's exposure into these three buckets, because the bucket tells you how hard the account is. A pure premises risk — say, an accounting office — is almost all Coverage A premises/operations, short-tail, easy. A manufacturer or a contractor loads the products-completed operations bucket, and that is the bucket that runs long and surprises you. A media-heavy or marketing-heavy business loads Coverage B, where the exposure is reputational and contractual. The same \$1 million limit means three very different things across these three buckets, because the frequency and severity behind it differ wildly. Read the operations, not just the limit.

Coverage B, personal and advertising injury, is a different animal from Coverage A and one underwriters sometimes wave past. Despite the word "personal," it has nothing to do with bodily injury; it covers a defined list of non-physical offenses — false arrest and detention, malicious prosecution, wrongful eviction, libel and slander, violation of privacy, and the use of another's advertising idea, slogan, or copyrighted material in the insured's own advertising. For most accounts this is a minor exposure thrown in with the package. For a business whose model is advertising and content — a marketing agency, a media company, a retailer with an aggressive ad operation — Coverage B can be the live wire, and you read it carefully or you sublimit it.

A coverage grant is only half a policy; the exclusions define the rest, and the CGL's exclusions are not arbitrary — each one carves out a risk that belongs to a different policy or that is uninsurable as a matter of policy. You do not need to memorize all of them today, but you must know the major ones, because an exclusion is the line between what your CGL pays and what the insured (or another carrier) is left holding:

• Expected or intended injury — the CGL covers accidents, not deliberate harm (the fortuity principle from Chapter 1, written into the contract).
• The auto, aircraft, and watercraft exclusion — bodily injury and property damage arising from these belong to the commercial auto policy (Chapter 23) and aviation/marine forms (Chapter 26), not the CGL.
• Pollution — the broad pollution exclusion pushes environmental liability out of the standard CGL and into specialized environmental coverage; this is one of the most litigated lines in the form's history.
• Professional services — errors in rendering professional advice belong to errors & omissions coverage (Chapter 24), not the CGL.
• Workers' compensation and employer's liability — injury to the insured's own employees is the province of the workers' comp policy (Chapter 22); the CGL covers injury to third parties.
• Damage to the insured's own product or work — the "business risk" exclusions (the ones casual readers call "your product / your work") mean the CGL is not a warranty: it does not pay to repair or replace the insured's own defective product or faulty workmanship, only the resulting third-party damage.

That last exclusion confuses people, so make it precise, because it draws the most important line in products liability. If the fabricated bracket fails and the only loss is that the bracket itself is now worthless — that is the insured's own product, excluded; the CGL is not a money-back guarantee on bad workmanship. But if the bracket fails and the mezzanine it was holding up collapses and injures someone, that resulting bodily injury and property damage to others is exactly what the products-completed operations coverage exists to pay. The CGL does not insure the quality of the insured's work; it insures the consequences to third parties when that work goes wrong.

⚠️ Underwriting Trap Do not let a clean-looking premises lull you on a products account. The most expensive mistake in CGL underwriting is to price the part you can see — the lobby, the parking lot, the slip-and-fall record — and undervalue the part you can't: the product or finished work already out in the world, accumulating exposure with every unit shipped and every job completed. The premises tells you about this year. The products tail tells you about the next ten, and it is the tail that blows up books. When a manufacturer's CGL looks cheap relative to its peers, the first question is always: is the products-completed operations exposure actually being rated, or did someone price a factory like an office?

21.2 The occurrence vs. claims-made trigger

Now the single most important technical distinction in all of liability insurance, and the one most professionals outside the casualty desk get wrong. A liability policy must answer a deceptively simple question: which policy responds to a given claim? Property insurance barely has this problem — the building burns on a date, and the policy in force on that date pays. But liability claims have two relevant dates that can be years apart: the date the injury-causing event happened, and the date the claim is finally made. A child is exposed to a hazard in 2015; the disease manifests and the lawsuit is filed in 2026. Which policy pays — 2015's or 2026's? The answer depends entirely on the policy's trigger, and there are two, and they are not interchangeable.

An occurrence trigger covers injury or damage that occurs during the policy period, regardless of when the claim is reported — even years later. The standard CGL is an occurrence form. If the injury takes place while the policy is in force, that policy responds, no matter how long afterward the claim surfaces. This is enormously valuable to the insured: coverage attaches to the moment of harm and stays attached, so a long-tail injury is covered by the policy that was in force when the harm occurred, even if that policy expired a decade ago and the insurer has long since moved on.

A claims-made trigger covers claims first made against the insured during the policy period, regardless of when the injury occurred (subject to a retroactive date — more on that in Chapter 24). The policy in force when the claim arrives responds, not the policy in force when the harm happened. This is the form used for most professional and management liability — E&O, D&O, EPL, cyber — and it exists to solve a problem occurrence coverage creates for the insurer.

TWO POLICIES, ONE LONG-TAIL CLAIM — injury in 2015, claim filed in 2026     [constructed teaching example]

  TIMELINE:    2015 ─────────────── injury occurs ─────────────── 2026 ─── claim filed
                 │                                                   │
  OCCURRENCE:  the 2015 policy responds ◄────────────────────────────┘
               (coverage follows the DATE OF HARM; the standard CGL works this way)

  CLAIMS-MADE: the 2026 policy responds ◄── (coverage follows the DATE THE CLAIM IS MADE;
               used for E&O / D&O / EPL / cyber — Chapter 24)

  The same facts land on two DIFFERENT policy years depending on the trigger.

Why would the industry ever prefer claims-made? Because the occurrence trigger leaves the insurer with an open-ended liability that is murder to reserve. Under an occurrence form, when you write a policy in 2026 you are accepting responsibility for harms that occur in 2026 but may not be reported for years or decades — the "incurred but not reported" losses, the long gray tail that an actuary has to estimate and hold reserves against (Chapter 10). For slow-developing harms — the textbook examples are asbestos and environmental contamination, where injuries surfaced thirty and forty years after the exposure — occurrence coverage saddled insurers with claims on policies written generations earlier, at premiums set with no inkling of what was coming. Claims-made coverage closes the tail: the insurer's exposure for a given policy year is fixed once that year's claims-reporting window shuts, which makes the liability vastly easier to reserve and price.

📋 At the Desk For your CGL underwriting, internalize this: the standard CGL is occurrence, so when you bind Harbor Steel's general liability this year, you are accepting its products-completed operations exposure for injuries that occur this year — including from products it shipped this year that may not fail and generate a claim until late in the next decade. You will never see most of those claims during the policy term. Your loss runs (Chapter 8) show you reported losses, which on a long-tail line systematically understate the ultimate losses, because the late-reported claims haven't arrived yet. An occurrence liability loss run is a photograph of a parade that is still marching past. Price the parade, not the photograph.

⚖️ Compliance Corner The occurrence-versus-claims-made distinction is not a free choice you make account by account; the form is largely set by line and by market convention, and switching an insured between them creates coverage gaps that are an errors-and-omissions minefield for the broker and a regulatory-complaint risk for you. When a long-tail professional account moves from one claims-made carrier to another, the retroactive date and the tail coverage (extended reporting period) must line up exactly, or a real claim falls into a gap nobody is covering — we work the mechanics in Chapter 24. For the standard CGL you will usually stay on the occurrence form; just never let an insured assume an occurrence mental model when you've placed them on a claims-made policy, or vice versa. Document which trigger applies, in writing, every time.

The historical pivot here is real and worth knowing, because it shaped the modern form. The general liability policy was an occurrence form for most of the twentieth century. Then, in the 1980s, the long-tail catastrophes — above all asbestos and pollution — arrived as claims on policies written decades earlier, and the resulting losses helped drive the mid-1980s liability insurance crisis (the hard market we met in Chapter 3). The industry's response was structural: it revised the standard commercial general liability form (the 1986 revision is the one practitioners still reference) and moved much long-tail professional and pollution coverage toward claims-made triggers, precisely to regain control of the tail. When you hear an old casualty hand talk about "occurrence" versus "claims-made," they are talking about a lesson the industry learned the expensive way.

21.3 Classification and the exposure base (payroll, revenue, area)

Every premium has to be tied to something that measures how much exposure the insured actually presents, and that something is the exposure base — the unit of exposure the rate is applied to. You met the idea implicitly in pricing (Chapter 11); here it becomes concrete, because liability rating lives or dies on choosing the right base and the right class. The premium for a CGL is, at its core, a rate multiplied by the number of exposure units: a manufacturer pays so many dollars per \$1,000 of sales; a contractor pays so many dollars per \$1,000 of payroll; a landlord pays per \$1,000 of square feet. Get the base wrong and no rate is right.

An exposure base is the variable that measures the size of an insured's loss exposure and to which the rate is applied to compute premium — payroll, gross sales (revenue), area (square footage), admissions, or unit count, depending on the class. The base is chosen to track the exposure: it should rise and fall with the actual hazard. The standard liability bases, and the logic behind each, are worth laying out because choosing among them is a real underwriting judgment.

The choice is not cosmetic. Consider a custom manufacturer that both makes products (sold by the dollar) and installs them at customer sites (labor by the payroll dollar). Rate the whole thing on sales and you under-capture the installation injury exposure; rate it all on payroll and you under-capture the products exposure. Real classification splits the operations and applies the right base to each — which is exactly why the application asks for the operational detail it does, and why a vague application (Chapter 8) is a pricing problem, not just a paperwork annoyance.

📋 At the Desk Classification is where casualty underwriting actually happens, and it is more judgment than lookup. The class plan (the ISO general-liability classification system) assigns each kind of operation a class code, an exposure base, and a base rate, the way the workers'-comp class plan does for payroll (Chapter 22). But businesses do not come in tidy single classes. A "machine shop" that also does field installation, runs a retail counter, and rents out part of its building is four exposures wearing one name. Your job is to read the operations and assign each its proper class and base — because a misclassification doesn't just misprice this account, it corrupts the loss experience of the class for everyone, which is how a whole class code goes bad. The discipline is: classify what the insured does, not what it calls itself.

There is a second reason the exposure base matters that beginners miss: most liability premiums are auditable. The premium you quote is an estimate, computed on the insured's estimated sales or payroll for the coming year. At the end of the term, an auditor (the same premium-audit function we develop for workers' comp in Chapter 22) examines the insured's actual records and adjusts the premium up or down to the actual exposure. If Harbor Steel projected \$45 million in revenue and actually did \$60 million, it presented more products and operations exposure than it paid for up front, and the audit collects the difference. This is why the exposure base must be measurable and verifiable — you cannot audit a base you cannot count. It is also why a deliberately understated projection is a red flag worth noting (Chapter 33), and why the auditability of the base is part of what makes a class insurable at all.

🔍 Check Your Understanding 1. A general contractor and a products manufacturer both want \$1 million CGL limits. Which exposure base fits each, and why does the same base not serve both well? 2. An insured projects \$20 million in sales; the audit finds it actually did \$32 million. What happens to the premium, and why does the auditability of the exposure base make that adjustment possible?

21.4 Industry-specific hazards and the loss-sensitive tail

We now go deep on the part of CGL that decides whether an account is a quiet earner or a slow-motion disaster: the way different industries present radically different liability hazards, and the way the products-completed operations tail makes certain accounts loss-sensitive long after you've forgotten you wrote them. This is the section where the practitioner's judgment matters most, because the rate tables and the model can tell you the class average, but only a read of the specific operation tells you whether this account is better or worse than its class — and on a long-tail line, being wrong about that doesn't cost you this year. It costs you in five.

Start with the principle that organizes everything: liability hazard varies enormously by what the business actually does, and the variation is not mainly about frequency — slips and falls happen everywhere — but about severity and latency. Three operational features, more than any others, tell you whether a liability account is dangerous:

1. Does the business make a product or leave behind finished work that can fail catastrophically? A restaurant's worst premises claim is bad; a restaurant's worst food-contamination claim can sicken hundreds. A landlord's worst claim is a fall; a structural fabricator's worst claim is a collapse. The products-completed operations question — what happens when the thing you made or built fails, and who is standing under it? — is the first severity question.
2. How long is the latency between the act and the harm? A product that fails immediately generates a short tail; a product whose failure mode is slow (corrosion, fatigue, cumulative exposure) generates a long one. Long latency means your loss runs are blind to the worst claims for years.
3. How litigious and how regulated is the field? Some industries — construction, healthcare-adjacent products, anything involving children — attract litigation and large verdicts the way a lightning rod attracts strikes. The same physical event produces a different legal loss depending on the field's litigation climate.

The fabrication account on your desk loads all three. It leaves behind structural components that hold weight — failure is catastrophic, not cosmetic. Steel's failure modes (fatigue, weld defects, corrosion) can be slow, so the latency is long. And construction-related products draw aggressive litigation, because when a building component fails the plaintiff's bar names everyone in the chain. That is why a structural fabricator's CGL is a fundamentally different — and more loss-sensitive — risk than a machine shop that makes, say, decorative parts that never bear a load.

📄 Read the Submission

text FIGURE 21.1 — "The bracket that hasn't failed yet" [the Underwriting File] THE SUBMISSION Harbor Steel & Fabrication: CGL within the commercial package; $1M occurrence / $2M general aggregate / $2M products-completed operations aggregate (illustrative). Custom structural steel and fabricated brackets shipped regionally and installed in third-party buildings. THE CONTEXT Welding/cutting/fabrication operations; ~180 employees; ~$45M revenue. ONE PENDING products-liability claim alleging a fabricated bracket failed. Premises history is unremarkable (ordinary shop slips, a forklift near-miss). Occurrence trigger. WHAT IT SHOWS The premises/operations exposure is routine and pricable from the record. The PRODUCTS-COMPLETED OPERATIONS exposure is the real risk: load-bearing components, long latency, litigious field, and a live claim that proves the exposure is real. WHAT IT DOESN'T It does NOT yet tell you whether the bracket actually failed as alleged, whether it was a design or fabrication defect, or whether one claim is a fluke or the first of a pattern — the claim file, the QA records, and the engineering will tell us that. THE DECISION Writable WITH the products exposure properly classified and rated (sales base), the pending claim documented and watched, and the products aggregate set with the tail in mind — not priced like a premises-only risk. THE LESSON On a products account, the premises is the part you can see and the tail is the part that decides the outcome. Rate the tail, document the live claim, and never let the quiet premises record set the price.

The phrase loss-sensitive tail deserves a precise meaning. A liability account is loss-sensitive when its ultimate losses are dominated by claims that develop after the policy period — so the account's true cost is not knowable for years, and the loss runs you underwrite from are, by construction, incomplete. For a products manufacturer or a contractor, the completed operations claims (the failures after the work is done) are exactly these late-developing losses. This has three consequences you must build into your underwriting:

• Your loss runs understate the risk. A three-year loss run on a long-tail account shows you the reported claims, not the ultimate ones. The recent years are the most incomplete — the worst products claims from last year haven't been filed yet. An experienced casualty underwriter mentally "develops" recent years upward, the way an actuary does with loss-development factors (Chapter 10), instead of taking the raw reported numbers at face value.
• The products aggregate is a separate, important limit. The CGL carries a separate products-completed operations aggregate — a cap on all products/completed-ops claims in the policy period, distinct from the general aggregate. On a products account, that limit is doing real work, and setting it (and pricing for it) is a deliberate decision, not a default.
• Severity, not frequency, is the killer. Premises claims are a frequency game — many small losses, manageable, even predictable. Products claims are a severity game — rare, but capable of a single loss that exhausts the limit and reaches into the umbrella. You underwrite the two halves with different instincts: tighten loss control and pricing on frequency, but underwrite survivability on severity.

🤖 Model vs. Judgment A predictive liability model (Chapter 32) is genuinely good at the frequency half of this. Given the class, the revenue, the territory, and the years in business, a well-built model predicts premises and general-liability frequency with real skill, because there is abundant, credible data on how often shops like this generate ordinary claims. Where the model is weakest is exactly where the money is: the rare, severe, long-latency products claim. Those events are too infrequent and too heterogeneous for the model to have seen enough of them, and the features that drive them — the specific failure mode of this product, the quality of this shop's welds, the litigation posture of the buildings this steel went into — are often not in the data at all. The model will price Harbor Steel's CGL off its frequency signal and quietly under-weight the tail. The underwriter's job is to supply what the model can't see: read the pending claim, ask for the QA records, and decide whether the products exposure is being respected. This is the chapter's clearest instance of underwriting is judgment — the model handles the common case; you handle the case that bankrupts the account.

There is one more industry-specific structure to name, because it is the heart of the chapter's argument and one of the book's six themes: pricing follows risk, and on a long-tail line the discipline to price the tail adequately is brutally hard, because the soft-market temptation is overwhelming. In a soft market (Chapter 3), liability is the first place underwriting discipline erodes, precisely because the losses arrive late. You can cut the products rate by 20% this year to win the account, and your loss ratio looks fine this year, and next year, and the year after — because the claims your underpricing failed to fund haven't matured yet. By the time the tail develops and the true cost arrives, the underwriter who cut the rate has been promoted, and the carrier is staring at three accident years of under-reserved liability business. Long-tail underpricing is the most seductive mistake in the industry because the feedback loop is delayed by years. The discipline to charge an adequate products rate in a soft market — to hold the line when the broker has three competing quotes and your boss wants the premium — is the single hardest and most important skill on a casualty desk (Chapter 11 on rate adequacy is the spine of it).

21.5 Additional insureds and contractual liability

Modern commerce runs on contracts, and those contracts routinely require one party to extend its insurance to protect another — which lands on your desk as two of the most common and most underappreciated requests in commercial underwriting: the additional insured endorsement and the assumption of contractual liability. Almost every commercial account you write will carry them, brokers treat them as routine paperwork, and underwriters who treat them as routine paperwork are giving away coverage they didn't price.

An additional insured is a party, other than the named insured, who is added to the policy as an insured for liability arising out of the named insured's operations, products, or premises — typically by endorsement, and typically because a contract demands it. When Harbor Steel signs a contract to supply fabricated steel to a general contractor, that contract almost certainly requires Harbor Steel to name the general contractor as an additional insured on Harbor Steel's CGL. The effect is that Harbor Steel's policy — and Harbor Steel's limits, and Harbor Steel's premium — now also defends and indemnifies the general contractor for claims arising out of Harbor Steel's work. The general contractor gets coverage it didn't buy, paid for by Harbor Steel's premium, on Harbor Steel's policy.

Why does this matter to you as the underwriter? Because every additional insured is an expansion of your exposure that the base premium did not contemplate, in three specific ways:

• More parties can tap the same limit. The policy limit is now shared with every additional insured. A single loss can trigger defense and indemnity obligations to multiple parties, draining the limit faster than a single-insured analysis would predict.
• The coverage can extend to the additional insured's own negligence. Depending on the endorsement wording, an additional-insured grant can pull in liability arising not just from the named insured's work but from the additional insured's own fault — which means Harbor Steel's policy could end up paying for the general contractor's negligence. The specific endorsement edition matters enormously here; the industry has tightened these forms over the years precisely because early-edition wordings gave away far more than anyone intended.
• You lose subrogation against them. An additional insured is now your insured, and you generally cannot subrogate (Chapter 4) against your own insured — so a recovery route you'd otherwise have is closed.

⚠️ Underwriting Trap The lazy way to handle additional-insured requests is to grant them blanket and never look. The disciplined way is to know which edition of the additional-insured endorsement you're attaching and what it actually covers. A blanket additional-insured endorsement on an early, broad edition, attached to a contractor or fabricator that signs dozens of customer contracts, can quietly convert a \$1 million policy into a \$1 million policy shared among twenty parties and reaching their own negligence — for the same premium. When an account's business model involves signing a lot of customer contracts (and a structural fabricator's does), the additional-insured exposure is a rated feature: you note the volume of contracts, you control the endorsement edition, and you price for the multiplication of insureds. The trap is treating a coverage grant to strangers as a clerical stamp.

The cousin of the additional insured is contractual liability, and you must hold the two apart. The CGL covers the insured's liability that the law imposes (tort liability — you injured someone, you're liable). Contractual liability is liability the insured takes on by contract that it would not otherwise have had — most commonly through an indemnification or hold-harmless agreement, in which the insured promises to cover another party's liability. When Harbor Steel signs a supply contract promising to "indemnify and hold harmless" the general contractor against claims arising from the project, Harbor Steel has assumed, by contract, a liability the law would not have placed on it. The standard CGL gives back some of this coverage through an exception for "insured contracts" — a defined category of business agreements where assumed liability is covered — but the scope is specific, and an indemnity that runs beyond the insured-contract definition is an uninsured assumption the insured is carrying naked.

📋 At the Desk When you underwrite an account that signs a lot of commercial contracts — contractors, fabricators, suppliers, anyone in a construction or supply chain — you are really underwriting their contracts as much as their operations, because the contracts decide who their liability ultimately belongs to. You won't read every contract, but you must understand the pattern: does this insured routinely sign broad indemnities running upstream to general contractors and owners? If so, its CGL is doing more work than its operations alone suggest, because it's funding other parties' risk through additional-insured grants and contractual assumptions. The practical move is to require a sample of the insured's standard customer contract, confirm the additional-insured and indemnity language is within what your form supports, and price the account knowing it's carrying contractual exposure on top of its tort exposure. The contract is part of the risk.

🔍 Check Your Understanding 1. Harbor Steel names a general contractor as an additional insured on its CGL. Name two ways this expands the carrier's exposure beyond what a single-insured analysis would show. 2. What is the difference between the liability the CGL covers by default (tort) and the liability an insured assumes through a hold-harmless agreement (contractual)? Why does the "insured contract" exception matter?

21.6 Excess and umbrella liability over the CGL

The CGL has limits — say \$1 million per occurrence and \$2 million in the aggregate — and for many businesses those limits are nowhere near enough to survive a catastrophic liability claim. A single serious bodily-injury verdict, a multi-plaintiff products case, a wrongful-death suit from a structural failure can run into the millions, far past a primary CGL limit. The coverage that sits above the primary policy and responds when the primary limit is exhausted is umbrella (or excess) liability, and it is the top layer of nearly every serious commercial program — including Harbor Steel's \$10 million umbrella.

The terms are close cousins and worth distinguishing. Excess liability provides additional limits over one or more underlying policies, generally following the same terms and conditions as the policy beneath it ("follow form"). Umbrella liability does the same but is broader: it provides excess limits over the underlying policies and can drop down to provide primary coverage for some claims the underlying policies exclude (subject to a self-insured retention — Chapter 12). For Harbor Steel, the \$10 million umbrella sits over the CGL, the commercial auto (Chapter 23), and the employer's liability portion of the workers' comp (Chapter 22), giving the account a single high layer of protection across its major liability exposures.

THE LIABILITY TOWER — Harbor Steel (illustrative limits)          [the Underwriting File]

   $10M  ┌─────────────────────────────────────────┐
         │           UMBRELLA  ($10M)               │  excess over ALL underlying;
         │  sits over CGL + Auto + Employer's Liab. │  can drop down for some gaps
    $1M  ├─────────────┬───────────────┬────────────┤
         │  CGL $1M    │  Auto $1M     │  EL $1M    │  the PRIMARY layer
         │  occ.       │  CSL          │  (per WC)  │  (each its own form)
     $0  └─────────────┴───────────────┴────────────┘
         the umbrella requires each underlying be kept at the REQUIRED LIMIT ("schedule of underlying")

For the umbrella underwriter (often a different desk, sometimes a different carrier), the central concerns are three. First, the underlying limits — the umbrella requires the primary policies to be maintained at specified minimum limits (the "schedule of underlying insurance"), because the umbrella is pricing its layer on the assumption that the primary will absorb the first \$1 million; if the insured lets a primary policy lapse or carries a lower limit than required, the umbrella faces a gap it never priced (the same gap problem we worked for the personal umbrella in Chapter 16). Second, the underlying exposures — because an umbrella sits over auto and employer's liability as well as CGL, the umbrella underwriter must look at all of the insured's liability exposures, not just the general liability; a clean GL account with a terrible fleet (Chapter 23) is a bad umbrella risk. Third, the severity tail — the umbrella is, by design, the layer that pays the catastrophic claim, so umbrella pricing is almost entirely a severity question, and the nuclear-verdict trend (Chapter 23) that is driving auto and product verdicts to once-unthinkable sizes is felt most sharply in the excess layers.

📋 At the Desk When you write the primary CGL knowing an umbrella sits above it, your underwriting still matters to the umbrella — arguably more, because the umbrella is relying on you. If you write a sloppy primary, set the wrong retention, or grant coverage the umbrella assumed you'd exclude, you create a problem that lands on the layer above. On a coordinated program (where your carrier writes both the primary and the umbrella, as on Harbor Steel), the two desks underwrite the account together: the primary controls frequency and the everyday losses, the umbrella backstops severity, and the terms have to mesh — the underlying limits the umbrella requires must match the limits the primary actually carries, with no daylight between them. A program with a gap between the primary limit and the umbrella's attachment point is a coverage hole the insured discovers at the worst possible moment.

21.7 A preview of the management and professional lines (E&O, D&O, EPL)

Finally, draw the boundary of the CGL by naming what it pointedly does not cover, because the gaps are not oversights — they are deliberate carve-outs that belong to other, specialized policies, and a businesses that buys only a CGL has large, modern exposures sitting completely uninsured. Chapter 24 takes these lines up in full; the job here is to recognize the holes the CGL leaves, so you can spot the account that needs more than the workhorse policy.

The CGL covers bodily injury, property damage, and the personal-and-advertising-injury offenses. It does not cover the major categories of liability that arise from decisions, advice, employment, and information rather than from physical harm. Four lines fill those gaps:

• Errors & omissions (E&O) / professional liability — covers liability arising from a professional service performed negligently: the accountant's bad advice, the architect's design error, the consultant's flawed recommendation. The CGL's professional-services exclusion (§21.1) is precisely the hole E&O fills. Any business that sells advice or expertise needs it, and the CGL won't touch it.
• Directors & officers (D&O) — covers the personal liability of a company's directors and officers for wrongful acts in managing the company: breaches of fiduciary duty, mismanagement claims, shareholder suits, regulatory actions against the leadership. The CGL covers the company's liability to third parties for injury; D&O covers the leadership's liability for governance.
• Employment practices liability (EPL) — covers liability arising from employment-related wrongful acts: discrimination, harassment, wrongful termination, retaliation. The CGL excludes injury to employees (that's workers' comp) and doesn't reach these employment torts at all; EPL is the answer, and in a litigious employment environment almost every employer with staff has the exposure.
• Cyber liability — covers liability and first-party costs arising from data breaches and network events: the exposure that barely existed when the CGL form was written and that the modern CGL largely excludes. It is the fastest-growing line in commercial insurance, and Chapter 24 (and the Tindall Stores account introduced there) works it in full.

📋 At the Desk The practical skill this section builds is gap-spotting: looking at an account and seeing not just the CGL exposure in front of you but the exposures the CGL leaves naked. A manufacturer like Harbor Steel is mostly a CGL-and-auto-and-WC risk — its core exposures are physical, and the workhorse lines cover them. But even Harbor Steel has some of these gaps: it holds data (a modest cyber exposure, Chapter 24); if it's a corporation with a board, there's a management-liability question; it has ~180 employees, so the EPL exposure is real. Part of underwriting an account well — and part of serving the insured honestly (the social-function theme) — is naming the coverage the insured doesn't have and should, even when it's not on the submission. The broker who hears "your client's CGL is solid, but they're carrying their employment and cyber exposure uninsured" remembers the underwriter who said it. Coverage you flag is coverage you can write next year.

These are the management and professional lines, and they share a structural feature that ties this chapter together: most of them are written on a claims-made trigger (§21.2), not the occurrence trigger of the CGL. That single difference — and the retroactive-date and tail-coverage machinery it requires — is one of the main reasons Chapter 24 needs a chapter of its own. For now, the lesson is the boundary: the CGL is the foundation, broad and essential, but it is a foundation, not a roof. The modern business needs more, and the underwriter who can see what's missing is worth more than the one who only prices what's submitted.

🗂️ The Underwriting File

Harbor Steel's general liability — the products tail comes into focus. With the property side deep-dived last chapter (Chapter 19), you turn to the general-liability piece of the Harbor Steel program, and the shape of the risk changes character. The property exposure was about a place — a building, a roof, a catastrophe zone. The general-liability exposure is about what Harbor Steel makes and does and the lawsuits that follow it out the door.

Sort the exposure into the three buckets. Premises/operations is routine: a 50,000-square-foot shop with the ordinary slip-and-fall, forklift, and visitor exposures of any fabrication plant — pricable from the record, short-tail, unremarkable. The loss runs show the everyday shop incidents you'd expect and nothing alarming on the premises side. Personal & advertising injury (Coverage B) is minimal — Harbor Steel isn't an advertising-driven business. The risk that matters is the third bucket: products-completed operations. Harbor Steel fabricates structural steel and load-bearing brackets that are installed into other people's buildings, and there is one pending products-liability claim alleging that a fabricated bracket failed. That claim is the watch-item of the entire general-liability analysis.

Apply the chapter. The CGL here is an occurrence form (§21.2), which means binding it accepts the products exposure for harms occurring this term — including from steel shipped this term that may not fail for years; the loss runs you're reading systematically understate the ultimate products losses because the long-tail claims haven't arrived. The exposure base (§21.3) for the products and manufacturing exposure is gross sales (~\$45M, auditable), with the field-installation work split to its proper class. The products-completed operations exposure is loss-sensitive (§21.4): load-bearing components, long latency, a litigious construction field — this is the part of the account that decides whether it was written well, and it must be rated as such, not priced off the quiet premises record. And because Harbor Steel supplies general contractors, expect additional-insured demands (§21.5) naming those contractors on Harbor Steel's CGL, plus contractual indemnities running upstream — coverage extended to third parties that expands the limit-sharing and must be controlled by endorsement edition and priced.

What this chapter settles, and what it doesn't. Settled: the general-liability exposure is assessable and writable, the premises piece is routine, the products piece is the real risk, and the account will carry additional-insured and contractual exposure typical of a fabricator. Not settled: whether that one pending bracket claim is an isolated event or the leading edge of a pattern — that turns on the claim file, the fabrication QA records, and the engineering, none of which the submission resolves; and the final products rate, terms, and the umbrella's interaction with the CGL, which the pricing (Chapter 11), reinsurance (Chapter 27), and capstone (Chapter 40) chapters build. Running disposition: GL assessed; products-completed operations is the watch-item; additional-insured and contractual exposure flagged for control and pricing. The pending claim goes in the file in red ink, and the decision waits for the chapters still to come.

Conclusion

Commercial general liability is the workhorse of casualty insurance — the policy almost every business buys to stand between itself and the lawsuits arising from its premises, its operations, and the products and work it leaves behind. We opened the form and found three insuring agreements: the core bodily-injury-and-property-damage grant of Coverage A, split between the short-tail premises/operations half and the long-tail products-completed operations half; the personal-and-advertising-injury offenses of Coverage B; and the small medical-payments goodwill of Coverage C. We drew the line the whole line turns on — the occurrence trigger of the standard CGL, which follows the date of harm and leaves a long reportable tail, versus the claims-made trigger that the professional and management lines use to close that tail. We worked the exposure base that turns operations into premium through classification, and the audit that trues it up. And we went deep on the loss-sensitive products tail: the half of CGL where severity, latency, and a litigious field combine to make the loss runs lie, where the soft-market temptation to underprice is strongest precisely because the losses arrive late, and where the underwriter's judgment — supplying what the frequency model cannot see — matters most.

We read the additional-insured and contractual-liability demands that modern commerce attaches to every account, and saw that each extends coverage to strangers and assumed liabilities the base premium never contemplated, which is why they are rated features and not clerical stamps. We placed the umbrella over the CGL as the severity backstop of the program. And we named the boundary of the CGL — the E&O, D&O, EPL, and cyber gaps it deliberately leaves — so you can spot the account that needs more than the workhorse policy.

Two of the book's themes ran through all of it. Underwriting is judgment: the model prices the frequency, but only the underwriter, reading the pending claim and the QA records, can respect the products tail the data can't see. And pricing follows risk: on a long-tail line, the discipline to charge an adequate products rate in a soft market — when the losses won't surface to prove you right for years — is the hardest and most consequential discipline on the desk, because the combined ratio it protects (Chapter 3) is the one that arrives late and arrives brutal.

Next, in Chapter 22, we take up workers' compensation — the one liability-adjacent coverage the CGL explicitly excludes, statutory and no-limits and audit-driven, with its own defining lever, the experience modification factor, that turns a company's own loss history directly into its price. Harbor Steel's welders and fabricators are waiting on that desk. The general-liability piece is assessed, the pending bracket claim is in the file, and the products tail is the line we'll be watching all the way to the bind.

Key Terms

• Commercial general liability (CGL) — the standard commercial policy covering an insured's legal liability to third parties for bodily injury, property damage, and personal & advertising injury arising out of its premises, operations, products, and completed work.
• Occurrence vs. claims-made trigger — the two ways a liability policy attaches: an occurrence form covers injury that occurs during the policy period (the standard CGL); a claims-made form covers claims first made during the policy period (most professional/management lines).
• Products-completed operations — liability for bodily injury and property damage caused by the insured's products or completed work after they leave the insured's control; the long-tail, loss-sensitive half of Coverage A.
• Premises/operations — liability arising from the insured's ongoing activities and the condition of its premises; the near-term, short-tail half of Coverage A.
• Personal & advertising injury — Coverage B; liability for a defined list of non-physical offenses (libel, slander, false arrest, wrongful eviction, privacy violation, advertising-idea infringement).
• Additional insured — a party other than the named insured added to the policy (usually by endorsement and usually by contract) as an insured for liability arising out of the named insured's operations, products, or premises.
• Exposure base — the variable measuring the size of an insured's loss exposure, to which the rate is applied to compute premium — payroll, gross sales, area, admissions, or unit count, by class.

Spaced Review

1. The standard CGL is an occurrence form. Explain what that means for a products account whose worst claims may not be reported for a decade, and why the loss runs you underwrite from understate the ultimate losses. (§21.2, §21.4)
2. A manufacturer and a general contractor each want \$1 million CGL limits. Which exposure base fits each, and why is classification "judgment more than lookup"? (§21.3)
3. (Reaching back.) Chapter 1 named adverse selection as the enemy of the pool and moral/morale hazard as the forces that change insured behavior. On a products account, how does a deductible or self-insured retention (Chapter 12) on the CGL push back on morale hazard in the insured's quality control? (§21.4; Ch. 1, Ch. 12)
4. (Reaching back.) In Chapter 6 you learned to think in frequency × severity. Explain why premises/operations is mostly a frequency problem and products-completed operations is mostly a severity problem — and why that changes how you underwrite each. (§21.4; Ch. 6)
5. (The recurring pricing-discipline question.) A broker brings three competing quotes and pushes you to cut Harbor Steel's products rate by 20% to win the account. Your loss ratio would look fine this year and next. Would cutting the rate help or hurt the combined ratio, and why is the answer not visible for years? (§21.4; Ch. 3, Ch. 11)