General Liability Insurance Pricing Guide

What determines price for General Liability?

General Liability sits at the intersection of three structurally distinct coverage forms — premises & operations, products liability, and completed operations — each with its own exposure base, tail length, and severity profile. They're bundled on one policy form but demand separate actuarial treatment. The core pricing challenge isn't complexity per se; it's that class ambiguity forces judgment at the point of base rate selection, products tail risk undermines frequency-based credibility, and jurisdiction — not geography — drives severity assumptions. This guide covers the exposure bases unique to GL, the rating factors that matter most across coverage parts, the methods that suit GL's specific data challenges, and the market forces currently reshaping loss costs.

• Products/completed operations severity (about $120,130 on a five-year average) exceeds premises/operations severity (about $70,232), yet they share a single policy — mispricing one coverage part can subsidise the other.

• At 45 months of development, only 50% of products/completed ops losses come from known claims versus 78.7% for P&O — the IBNYR composition is fundamentally different.

• ISO's 5-digit classification code remains a foundational rating variable, and Verisk reported filing over 55 new General Liability classifications in 2024 to address emerging class gaps.

• Increasing inflation added $83.4B–$103.3B to Other Liability–Occurrence losses over 2015–2024, representing 27.4–34.0% of booked losses and DCC.

• The standard-to-E&S boundary is itself a pricing variable, reflecting ongoing risk migration and more continuous pricing across markets.

Exposure measures unique to General Liability

GL is unusual among commercial lines in requiring multiple exposure bases within a single policy. P&O uses payroll (contractors, per $1,000), area (premises-only risks, per 1,000 sq ft), or gross sales (mercantile/manufacturing). Products liability rates on gross sales. Completed operations rates on contract revenue or receipts. By contrast, workers' compensation uses payroll universally, commercial auto uses vehicle count, and property uses TIV — all single-base lines.

The actuarial rationale: GL covers third-party liability to the public, and no single activity measure captures hazard across all industry classes. A retailer's customer-contact exposure scales with sales, not payroll; a contractor's hazard scales with labour hours, not revenue. The variable bases (payroll, sales, receipts) require premium audit after expiry — a structural cost absent in fixed-base lines like property. Products liability introduces a specific temporal mismatch: sales revenue is collected in the policy period, but the products generating losses may remain in service for years. Current-year receipts for completed operations can include payments on unfinished work, further decoupling premium from loss-generating exposure.

Rating factors that shape General Liability premiums

Classification code and class group

The ISO 5-digit classification code is the dominant predictor. It simultaneously encodes industry, hazard type, and operations profile. The interaction between class group and coverage part produces a large severity spread in the dataset: manufacturers' products severity ($196,337) versus contractors' P&O severity (about $85,351) — roughly a 2.3× differential. Mixed operations create class ambiguity that resists automated assignment. A shoe manufacturer with a retail outlet must allocate wholesale sales to manufacturing codes and retail sales to retail codes; failure to segregate produces incorrect premium. Verisk's 2024 filing of 55+ new classification codes for a wider range of risks indicates that classification granularity remains a key lever for pricing accuracy.

Territory and jurisdiction

Territory is one of three standard ISO rating variables, but for GL the operative dimension is jurisdiction, not geography. US jury-based civil trials, contingency fees, and punitive damages create severity expectations fundamentally different from UK or EU exposures. Swiss Re's country-level comparison rates the US "High" on all six social inflation drivers; France and Germany rate "Low" on contingency fees and jury-based systems. Florida produces more nuclear verdicts per capita than California. Models must encode the litigation environment — plaintiff attorney density, litigation funding penetration, venue-shopping exposure — not just zip code.

Loss history and experience rating

ISO's experience rating plan requires separate credibility calculations for P&O and products/completed operations sublines before blending. The maximum single loss (MSL) parameter caps the impact of any single catastrophic claim on the experience modification. For products liability, the ISO Credibility Subcommittee recommended an empirical Bayes credibility procedure for classification ratemaking. Five to ten years of loss history is standard, but for products the relevant exposure window extends well beyond the experience period due to latent defect emergence.

Limits, structure, and ILFs

Increased limits factors vary by class group and coverage part. The ILF structure must satisfy monotonicity (non-decreasing) and concavity (increasing at a decreasing rate). Products/completed operations ILFs saw no changes in the 2024 Verisk review, while P&O ILFs increased 3–7% — reflecting divergent severity trend trajectories. Published excess loss factors underestimate per-occurrence excess loss potential for GL and should not be used directly for excess pricing without adjustment for clash loading.

Size of risk

Verisk's size-of-risk supplement, built on 18.5 million GL policies representing $43 billion in premium, demonstrates that after controlling for class code, risk size carries independent predictive power. This is a significant publicly documented GL finding: exposure base volume alone does not fully capture risk-size effects.

Binary eligibility versus continuous adjustment

High-hazard classes (fireworks manufacturing, junkyards, amusement parks) function as binary decline triggers in admitted markets — no rate is calculated. Novel exposures such as PFAS often begin as policy-form exclusions, especially in GL, amid uncertainty and limited loss data. As data accumulates, binary exclusions migrate to continuous rating factors. At Lloyd's, underwriting appetites and authorities are reflected in underwriting systems and controls, including automated checks and escalation of control failures or breaches, rather than relying solely on discretionary calls. The standard-to-E&S boundary is dynamic.

How actuaries price with GL's long tail and thin data

• Bornhuetter-Ferguson: Suits GL because it blends reported experience with prior expected loss ratios, stabilising estimates when development is immature — critical for products/completed ops where only 33% of losses are from known claims at 21 months.

• Bayesian severity estimation: Addresses excess layer pricing when observed claim counts are insufficient; applies the ILF to a posterior severity estimate rather than raw experience, reducing volatility.

• Separate large/attritional development: A single structural defect claim can dwarf years of premium; isolating large claims into a distinct development triangle prevents one outlier from distorting attritional LDFs.

• Exposure rating via ILFs: When individual account data is too thin for experience rating, ILF-based exposure rating becomes the primary method — particularly for higher excess layers where credibility to observed experience approaches zero.

• Credibility blending with industry benchmarks: ASOP No. 25 requires that subject experience may be subdivided (primary vs. excess losses) with different credibility levels for each.

• Tail factor estimation: Tail factor estimation typically combines actuarial judgment with a range of methods, including curve fitting, benchmark data, and paid-incurred approaches.

What's shaping General Liability pricing now

The NAIC reported the Other Liability–Occurrence combined ratio at 110.1% in 2024, with adverse prior-year reserve development of about $10.0 billion. Nuclear verdicts (≥$10M) totalled $31.3 billion across 135 cases in 2024 — up 116% in value and 52% in frequency year-over-year. Product Liability–Occurrence has experienced notable loss pressure in recent years. GL claim frequency has remained broadly flat, while severity appears to be the dominant cost driver, outpacing inflation. Third-party litigation funding can increase litigation costs and prolong case timelines, contributing to higher claims severity and social inflation. GL rates generally increased in 2025, with increases varying by risk profile, and Lloyd's has placed US General Liability under enhanced oversight alongside Cyber.

How hx supports General Liability insurance pricing

Configurable pricing logic for complex rating structures

General Liability's unique challenges require pricing logic that standard raters struggle to express. The hx Decision Engine lets actuaries implement these rules in native Python—including knockout criteria, coverage-specific calculations, and control interactions—then deploy changes with full governance and version control.

GL mixed-operation risks (manufacturer with retail outlet) require split-class logic that legacy raters handle via manual offline workarounds. hx's Decision Engine implements multi-class allocation rules in native Python with real-time validation against appetite guardrails.

Submission triage aligned to appetite

General Liability submissions arrive with documentation that determines both insurability and pricing tier. hx Submission Triage extracts this data from unstructured broker submissions and surfaces it alongside appetite checks and indicative pricing, so underwriters can identify gaps before investing time in full analysis.

Products liability submissions flagged for high-hazard SIC codes (e.g., medical device manufacturing) trigger jurisdictional severity checks before routing. hx Submission Triage surfaces appetite checks and rule-based triage during submission processing.

Portfolio intelligence for aggregation management

General Liability's systemic risk requires portfolio-level visibility that policy-by-policy pricing can't provide. hx Portfolio Intelligence enables batch rating and what-if analysis, with portfolio reporting and insights to support testing and analysis.

Completed operations aggregate exposure accumulates across policy years as prior work remains in-force, creating hidden concentration risk standard dashboards miss. hx Portfolio Intelligence provides portfolio reporting, scenario analysis, configurable dashboards, and the ability to re-run historic portfolios on new models.

Audit trails for evolving regulatory requirements

With increasing regulatory scrutiny, actuaries need documented lineage from model assumptions to individual policy pricing decisions. hx captures every action automatically, creating the governance trail General Liability's regulatory environment demands.

Mid-year ISO class rate updates require synchronized model changes across P&O and Products sublines with full traceability for Lloyd's actuarial sign-off. hx includes version control features intended to support governance and regulatory validation requirements in insurance pricing workflows.

Explore hx for General Liability insurance →

This guide is part of Hyperexponential's insurance pricing resource library. For more information on how hx supports General Liability pricing, contact us.