Why Audit Compliance With Filed Rating Rules?
A production premium can be calculated correctly by a system and still be noncompliant if the system used the wrong approved rule version, applied a factor to an ineligible risk, or included an adjustment the filing does not permit. A compliance audit asks a focused question: does the transaction as rated agree with the approved rules that govern it?
The audit needs more than a premium total. It needs the governing filed rules, the transaction facts used to rate the policy, the rule version selected, the factors applied, any adjustments and their supporting records, and the charged premium. The result is a traceable comparison between the amount permitted by the rules and the amount produced in the transaction.
Establish the Audit Benchmark
Start by identifying exactly which approved rules govern the transaction. A filing may define applicability by policy effective date, endorsement effective date, or another specified date. Do not assume that issue date, quote date, entry date, or processing date controls. Read the rule that establishes the effective-date basis, then apply it consistently.
The benchmark should be the approved version of the rating rules, not a copy of the factors currently in a production table unless that table has been verified against the approval. Record the filing or version identifier, effective-date rule, relevant manual pages or tables, and the calculation and rounding conventions. If a transaction crosses an effective date, determine from the approved rules whether the change applies to the whole policy, only a changed coverage, or another defined unit. Do not invent a transition rule where the filing does not provide one.
Next, reconstruct the rating inputs from source records. Examples include territory, classification, coverage limit, deductible, and the characteristics used to determine eligibility for a credit. A production record may show what the system used, but an audit also needs evidence of what the verified transaction facts support. A wrong input can produce a mathematically accurate calculation that nevertheless fails to comply with the approved rules.
Apply the filing's stated effective-date basis to the policy or transaction. Preserve the reason for selecting that version.
Compare the values used in production with reliable source records and the definitions in the approved rules.
Use the applicable filed factors, eligibility conditions, permitted adjustments, and required rounding sequence.
Calculate charged premium minus expected premium, then identify the rule or transaction fact that explains any difference.
Keep the transaction identifiers, rule version, calculation, source evidence, and exception disposition so the conclusion can be reproduced.
Recalculation and the Meaning of a Deviation
For a simple multiplicative rating plan, the premium before permitted adjustments is the base premium multiplied by the applicable rating factors. If the approved rules allow a specific adjustment, apply it only when its conditions are met. The full calculation must follow the filing's order and rounding instructions; the expression below assumes no intermediate rounding and a final rounding step.
Here, \(r_j\) denotes an applicable filed rating factor, while \(a_m\) denotes an adjustment factor that is permitted and supported for the transaction. An adjustment factor is not automatically valid merely because the system offers it: the auditor must verify its eligibility conditions, authorization, and any limit or cap. In a plan with additive adjustments, minimum premiums, or a different sequence, use that plan's actual formula instead of forcing it into this multiplicative form.
The comparison is
A positive deviation means production charged more than the independent calculation; a negative deviation means it charged less. A zero deviation shows agreement in amount, but it does not necessarily prove every rule was followed: two errors could offset one another. Therefore, an audit checks the factors and conditions as well as the final premium.
Worked Applications
Worked Example: Selecting the Correct Effective-Date Version
A hypothetical property policy has a base premium of $800. Its policy effective date is October 1, although it was issued on September 28. The approved rule specifies that the policy effective date selects the rating version. The prior version has territory, class, and limit factors of 1.05, 0.90, and 1.20. The version effective October 1 has corresponding factors of 1.10, 0.90, and 1.25. A documented 5% credit is allowed under either version, and the premium is rounded to the nearest whole dollar at the end. Production used the prior version. Find the expected premium and the deviation.
Method. Choose the version using the governing date rule, then independently calculate with that version's factors. The policy effective date is October 1, so the October 1 version applies even though the issue date is September 28.
Before the credit, the correct calculation is \(800 \times 1.10 = 880\), then \(880 \times 0.90 = 792\), then \(792 \times 1.25 = 990\). Applying the documented credit gives \(990 \times 0.95 = 940.50\). Rounding once, at the required final step, gives an expected premium of $941.
For comparison, production's prior-version calculation is \(800 \times 1.05 = 840\), \(840 \times 0.90 = 756\), \(756 \times 1.20 = 907.20\), and \(907.20 \times 0.95 = 861.84\), which rounds to $862. The deviation is \(862 - 941 = -79\), so production charged $79 less than the expected premium.
Check. The error is attributable to the version selection: the verified credit is allowed, and both calculations use the stated final-dollar rounding. The common mistake is to use issue date to select the prior version; the filing explicitly makes policy effective date controlling.
Worked Example: Testing Eligibility and a Permitted Credit
A hypothetical commercial account has a base premium of $1,200, a territory factor of 0.95, a class factor of 1.20, and a limit factor of 1.10. The filed rules allow an 8% fleet credit only when an account has at least five eligible vehicles. A separate 5% safety credit is allowed with specified documentation, which the account has supplied. The account has four eligible vehicles. Production applied both credits. Find the expected premium, the production premium, and the deviation. Round the final premium to cents.
Method. Verify the rating factors, test each adjustment against its own condition, and apply only the adjustments supported by the facts. The fleet credit fails its five-vehicle threshold; the documented safety credit qualifies.
First calculate the premium before credits: \(1{,}200 \times 0.95 = 1{,}140\); \(1{,}140 \times 1.20 = 1{,}368\); and \(1{,}368 \times 1.10 = 1{,}504.80\). The allowed safety credit gives \(1{,}504.80 \times 0.95 = 1{,}429.56\). Thus the expected premium is $1,429.56.
Production applied both credits: \(1{,}504.80 \times 0.92 = 1{,}384.416\), then \(1{,}384.416 \times 0.95 = 1{,}315.1952\). Rounded to cents, production charged $1,315.20. The deviation is \(1{,}315.20 - 1{,}429.56 = -114.36\): production charged $114.36 less than expected.
Check. The difference arises from the ineligible fleet credit, not from the documented safety credit. The common mistake is to treat the presence of a fleet-credit field, or the account's general description as a fleet, as sufficient evidence of eligibility. The filed threshold is based on eligible vehicle count, which is four here.
Worked Example: Summarizing an Exception Report
An auditor selects four hypothetical transactions from a defined review sample and independently verifies each expected premium. The report shows the following expected and charged amounts: Policy A, $640.00 and $640.00; Policy B, $875.50 and $900.50; Policy C, $1,120.00 and $1,064.00; Policy D, $450.75 and $455.75. Calculate each deviation, the exception count, the total expected and charged premium, the net deviation, and the sum of absolute deviations.
Method. For each transaction, subtract expected premium from charged premium. Then aggregate signed deviations for the net result and absolute deviations for the gross amount of disagreement.
Policy A: \(640.00 - 640.00 = \$0.00\). Policy B: \(900.50 - 875.50 = +\$25.00\). Policy C: \(1{,}064.00 - 1{,}120.00 = -\$56.00\). Policy D: \(455.75 - 450.75 = +\$5.00\). Three of the four sampled transactions have nonzero deviations, so the sample exception count is 3.
Total expected premium is \(640.00 + 875.50 + 1{,}120.00 + 450.75 = \$3{,}086.25\). Total charged premium is \(640.00 + 900.50 + 1{,}064.00 + 455.75 = \$3{,}060.25\). The net deviation is \(3{,}060.25 - 3{,}086.25 = -\$26.00\), which agrees with \(0 + 25 - 56 + 5 = -26\). The sum of absolute deviations is \(0 + 25 + 56 + 5 = \$86.00\).
Check. The net amount is not the same as the total amount of error: the positive and negative deviations partly offset. The common mistake is to report only the net $26 shortfall and overlook $86 of absolute differences across the sample. These sample results describe the reviewed transactions; they do not, without an appropriate sampling design and analysis, establish the error rate or total financial impact for all policies.
Interpreting and Documenting Exceptions
An exception report is a starting point for investigation, not a complete audit conclusion. A useful record identifies the transaction, the rule version selected, the expected and charged premiums, the signed deviation, the relevant rating inputs, and the suspected rule or data issue. The auditor should then verify the explanation against source evidence and the approved rules.
| Exception pattern | Audit question | Evidence to check |
|---|---|---|
| Different factor from the approved table | Was the correct version and factor cell selected? | Filing, effective date, territory or class support, production calculation |
| Credit or surcharge applied | Were all eligibility conditions satisfied, and was the adjustment permitted? | Risk records, required documentation, authorization, applicable cap |
| Premium differs despite matching factors | Was the calculation sequence or rounding rule followed? | Filed calculation instructions, intermediate values, transaction conventions |
| Production record does not match source facts | Did the rating system use an incorrect or incomplete input? | Application, verified records, change history, transaction-level data |
Separate confirmed rule deviations from items that remain unresolved, and distinguish rating-rule errors from missing evidence or data-quality issues. If an exception report contains duplicate records, reversals, or multiple entries for the same policy transaction, reconcile those records before summing premiums. Otherwise, the report may overstate the number or dollar amount of exceptions.
For a sample review, state the population and selection method, the number of transactions reviewed, and the number and types of exceptions found. Report both counts and dollars where useful, while making clear whether the dollar measure is net or absolute. Do not extrapolate a sample total to the full portfolio unless the sample design and estimation method support that conclusion. This keeps the audit's claims aligned with the evidence.
Common Audit Traps
- Choosing a version by convenience. The most recently deployed table is not necessarily the version that governs a transaction. Follow the effective-date rule in the approved filing.
- Accepting production inputs as verified facts. A system's stored territory or eligibility flag may itself be wrong. Compare it with appropriate source evidence.
- Allowing an adjustment because it is available. A permitted credit still requires its stated eligibility conditions and any required documentation.
- Relying only on premium agreement. Offsetting factor errors may leave the final amount unchanged. Inspect the calculation path as well as the result.
- Confusing net and gross impact. Positive and negative deviations can cancel in a net total. Preserve both signed and absolute summaries.
Check Your Understanding
For each problem, identify the governing rule or condition before calculating.
- A hypothetical policy has a $500 base premium and one territory factor. The prior version's factor is 0.90; the new version's factor is 1.08. The filing says the policy effective date controls, and the policy is effective January 1 but issued December 20. What is the expected premium before any other adjustment? (Answer: $540.00; use the new version.)
- A hypothetical account has a $2,000 base premium, territory factor 1.10, and class factor 0.80. A 10% credit requires at least six eligible units; the account has five, but production applied the credit. What is expected premium before rounding, and what is the deviation if production charged the discounted amount? (Answer: Expected $1,760.00; charged $1,584.00; deviation \(-\$176.00\).)
- A hypothetical policy's premium after filed rating factors is $1,500. A documented adjustment is permitted up to 7%, but production applies 9%. If adjustment is a multiplicative credit and amounts are rounded to cents, find the expected premium and deviation. (Answer: Expected $1,395.00; charged $1,365.00; deviation \(-\$30.00\).)
- Two hypothetical transactions have expected and charged premiums of $720 and $750, and $1,050 and $1,010, respectively. Find the exception count, net deviation, and sum of absolute deviations. (Answer: 2 exceptions; net deviation \(-\$10\); absolute deviations total $70.)
- A transaction has an expected premium of $980.40 under the applicable filed version. Production charges $1,000.40. The exception report also shows a $20.00 difference between one production factor and the filed factor. What is the signed premium deviation, and what should the auditor verify before concluding that the factor explains it? (Answer: \(+\$20.00\); verify the underlying rating input, applicable version, other factors and adjustments, calculation sequence, and rounding.)