Your combined ratio held steady last year, but rate increases are slowing and loss costs keep moving. That leaves an underwriting leader with one question: does the price on each policy still match the risk behind it?
A healthy aggregate answers a different question. It shows how the book performed, not whether a single line, class code, or renewal was priced correctly.
This article shows where price and risk separate and how to check each gap, with a three-part framework, a worked example, and a five-step check.
Does a Good Combined Ratio Mean Your Pricing Is Accurate?
Not by itself. Fitch credits the 2025 result partly to a benign hurricane season and outsized favorable reserve development, and neither of those measures pricing accuracy.
What a strong year does not prove
A strong year shows the book covered its costs. It does not show that each line, class code, or renewal carried enough premium for the risk it took on.
How an average hides an underpriced line
AM Best reports 2025 combined ratios above 100 in three commercial lines: commercial auto at 103.5, medical professional liability at 106, and other or products liability at 108. The industry composite was 95.
Anyone who reads the composite alone would miss those lines. Your own book has the same structure: one ratio sitting on top of very different segments.
How Does Underpricing Affect an Insurer's Combined Ratio?
Underpricing raises the loss ratio first and the combined ratio with it, because premium grows more slowly than the cost of the claims it has to fund.
Small gaps compound. The table follows a book that raises rates 3% a year while loss costs rise 5% a year.
Year | Rate index | Loss cost index | Loss ratio |
|---|---|---|---|
Start | 1.00 | 1.00 | 60.0% |
Year 1 | 1.03 | 1.05 | 61.2% |
Year 2 | 1.06 | 1.10 | 62.4% |
Year 3 | 1.09 | 1.16 | 63.6% |
This is an illustration. It assumes a 60% starting loss ratio, a constant expense ratio, and no change in business mix. The figures are arithmetic, not survey data.
By year three the loss ratio sits 3.6 points above its starting level, with the expense ratio held flat. Each renewal starts from the prior year's premium, so an undercharge in one term becomes the base for the next.
Where Does Premium Leakage Start?
Premium leakage happens when the price charged does not reflect the true risk or exposure on a policy, so the insurer collects less premium than the risk warrants.
It starts earlier than most teams look. A 2026 ReSource Pro summit session placed the origin in classification and policy setup, well before premium audit.
Leakage is not always fraud
Many cases involve customers who give inaccurate information without intent to deceive, according to kbs intelligence. Fraud controls catch intent. Leakage controls have to catch honest error, stale data, and gaps between systems.
Where the errors enter
In commercial lines, the weak points are class codes, business descriptions, and exposure bases such as payroll. ReSource Pro described construction risk, where certain class code combinations allowed misclassification and a lower premium than intended.
Adjusters often learn about job duties and operations that never appeared on the application. Data silos between underwriting, claims, and audit let the same leak reopen at every renewal.
The Price-to-Risk Gap Check
Pricing accuracy depends on the data behind the rate: declared exposures, class codes, and rating inputs must reach the pricing system correctly, with a person reviewing every flag.
The Price-to-Risk Gap Check splits the problem into three gaps between the price on a policy and the risk behind it.
Gap | What separates | Where to look | What confirms it |
|---|---|---|---|
Input gap | Declared exposure and actual exposure | Class codes, exposure bases, renewal declarations | Claims or audit findings that contradict the application |
Trend gap | Achieved rate and loss cost trend | Rate change by segment against frequency and severity | A segment loss ratio climbs while the book average holds |
Signal gap | Reported results and pricing quality | Combined ratio without catastrophe and reserve effects | The result shifts sharply once those items are removed |
Gap 1: The input gap
A rating engine prices the inputs it receives, so a wrong class code produces a wrong premium with perfect arithmetic. Compare declared values against claims files, audit results, and outside data at issuance and again at renewal.
Gap 2: The trend gap
The trend gap is the distance between the rate a carrier achieves and the loss cost trend it faces. Soft markets widen it. AM Best reports lower renewal pricing in 2025 than in 2024 for cyber, D&O, commercial property, and workers' compensation.
Fitch describes overall pricing as adequate, while renewal rates are still rising in underperforming segments such as commercial auto and excess and umbrella liability.
Gap 3: The signal gap
The signal gap opens whenever results are read without separating pricing from weather and reserves. Step four below tests for it.
Are My Rate Increases Keeping Up With Loss Costs?
AM Best projects the U.S. property and casualty combined ratio will rise 1.9 points to 96.9 in 2026, as net premium growth slows and repair-driven claims costs climb.
Your own answer has to come by segment, and it takes five steps:
Split the book by line, channel, and geography, so a strong segment cannot cover for a weak one.
Compare the rate change achieved in each segment with its frequency and severity trend, and flag any segment where rate trails loss cost.
Test declared exposure in the flagged segments against what claims and audit have already found.
Rerun the combined ratio without catastrophe losses and reserve releases to see the result pricing produced.
Send each finding to the team that owns the input, whether underwriting, claims, or audit, and record the correction at the source.
Repeat the check each quarter, because loss trend and business mix move between filings.
How InsOps Helps
InsOps builds an insurance-trained AI that assists with the data path feeding your rating and quoting systems. LiLa, our insurance-trained LLM, runs inside your own environment, so PII and PHI never leave controlled infrastructure. A person reviews and validates every mapping before deployment.
Our Integration Gateway provides pre-built connectors for Guidewire PolicyCenter, ClaimCenter, BillingCenter, UnderwritingCenter, PricingCenter, and Quoting Services. LiLa maps and converts payloads from any source into Guidewire structures, so policy, claims, and billing data flow into your workflow without custom engineering.
InsOps does not set your rates. It assists with the data that reaches them, and pricing decisions stay with your actuaries and underwriters.
If you are evaluating how to keep declared exposure, claims, and billing data consistent across systems without adding manual reconciliation, contact us to talk through what this looks like for your operation.
Frequently Asked Questions
What is rate adequacy?
Rate adequacy means the premium charged covers the expected losses and expenses of a policy or segment, with room for a profit provision. It is measured against expected future costs, so a rate that was adequate last year can fall short this year.
Why is accurate pricing so important?
Accurate pricing decides whether a carrier collects enough premium for the risk it accepts. Underpricing leaves claims larger than premium, and overpricing sends customers to competitors.
Is premium leakage always fraud?
No. Many leakage cases come from incomplete declarations, outdated information, and gaps between departments, not intent to deceive.
Are my rate increases keeping up with loss costs?
Compare achieved rate change with loss cost trend inside each segment, not across the whole book. Any segment where rate trails trend is underpriced, and the shortfall carries into every later renewal.
What does a softening market do to pricing?
It lowers the rate increases carriers can obtain, so the trend gap widens in lines where loss costs are still rising. Segment-level checks matter more, because lines can fall behind while the book-wide result looks unchanged.
Does bad data feeding the rating engine matter more than the model itself?
Both matter, but no model corrects an input it never sees as wrong. Insurance-trained AI assists by mapping and validating data before it reaches rating, with a person reviewing every mapping.