Why Design Consistency Matters
An experience rating plan uses an insured's past experience to modify a prospective premium. That comparison is meaningful only when the information being compared describes the same risk on compatible terms. A calculation can be arithmetically correct yet produce a misleading adjustment if its losses, exposures, premium, or eligibility test refer to different periods or different units of insurance.
The errors considered here are design errors, not merely calculation slips. A spreadsheet may correctly divide one number by another while the plan has paired the wrong loss period with the exposure, used losses from one coverage to adjust premium for several coverages, or left unclear whether a multi-location account is one risk or several. Such a plan can apply inconsistently even when every user follows the written instructions.
A useful diagnostic is to trace each quantity from its source through the plan calculation. For every loss amount, identify the covered exposure, experience dates, valuation date, and risk unit. For every premium amount, identify its coverage components, policy period, and the portion the plan is authorized to adjust. These checks are more reliable than looking only at the final adjustment.
Match Losses to the Relevant Experience Period
An experience period must be defined in terms that can be applied consistently to both losses and earned exposure. A plan might select claims by date of occurrence, policy effective date, or another stated period basis. It must also specify when losses are valued. If claims are valued at a stated evaluation date, that date is part of the design: claim amounts can change as claims develop.
The basic reason for matching periods can be seen in a loss cost per exposure measure. If \(L_t\) is the loss amount for period \(t\) and \(E_t\) is the corresponding earned exposure, a combined measure across periods is calculated from the matching totals:
This is an exposure-weighted average of the period-specific loss costs, since \(\frac{\sum_t L_t}{\sum_t E_t} =\sum_t \frac{E_t}{\sum_j E_j}\frac{L_t}{E_t}\), when each \(E_t>0\). The weights are the shares of total exposure. If a loss from one period is divided by exposure from another, the result is not this weighted average: it combines outcomes with the wrong denominator and may not describe any observed period.
Worked Example: Losses Paired With the Wrong Exposure Year
A hypothetical commercial auto plan uses accident-year losses valued at December 31, 2025. The analyst selects $270,000 of losses from accidents in 2025 but mistakenly divides by the 2024 earned exposure of 900 vehicle-years. The actual 2025 earned exposure was 600 vehicle-years. Calculate the erroneous and corrected loss costs, and quantify the difference.
Method. Divide the loss amount by earned exposure from the same accident period. The stated valuation date determines the loss amount used; it does not change the accident year to which the losses belong.
Erroneous calculation. The calculation pairs 2025 losses with 2024 exposure: \($270{,}000/900 = $300\) per vehicle-year.
Correct calculation. Match the 2025 losses to the 2025 exposure: \($270{,}000/600 = $450\) per vehicle-year.
The error understates the matched loss cost by \($450-$300=$150\) per vehicle-year. Relative to the correct value, the understatement is \($150/$450=1/3\), or 33.33%. The corrected value checks because \(600 \times $450=$270{,}000\).
Common mistake avoided. A common wrong calculation is to use the most recent available exposure automatically, without checking whether it belongs to the losses selected. The correction is to write down the period definition first, then select the corresponding earned exposure and the loss valuation date separately.
A complete period specification should say whether losses are assigned by occurrence date or another event date, which earned exposure periods correspond to them, and the loss evaluation date. If the plan combines multiple experience periods, each period must be paired before aggregation. A mature loss period should not be paired with an exposure period merely because both are available in a report.
Keep Coverage Bases Consistent
A coverage basis identifies which insurance protection the losses, exposure, and premium represent. For example, a hypothetical account may have property and general liability coverage. Property losses alone cannot be interpreted as experience for the combined property-and-liability premium unless the plan explicitly provides a valid allocation or other defined comparison basis. Similarly, losses from one limit or deductible basis should not silently be compared with premium written on a different basis.
The key correction is not simply to divide by a different number. The plan must define the coverages included in the experience measure and identify the premium components eligible for adjustment. This supports a like-for-like comparison and prevents favorable results for one coverage from unintentionally changing another coverage's price.
Worked Example: Property Losses Used Against Combined Premium
A hypothetical account has $120,000 of property losses in the plan's experience period. Its prospective premium consists of $150,000 for property and $90,000 for liability. An analyst compares the property losses to the combined $240,000 premium, while the plan is intended to adjust property premium only. Calculate the two ratios and identify the appropriate basis.
Method. Compare the property loss amount with the property premium when evaluating the property coverage. Do not include liability premium in that denominator if liability is outside the experience measure and adjustment.
Erroneous calculation. Combined premium is \($150{,}000+$90{,}000=$240{,}000\). The resulting ratio is \($120{,}000/$240{,}000=0.50\), or 50%.
Correct calculation. The relevant property premium is $150,000, so the property-based ratio is \($120{,}000/$150{,}000=0.80\), or 80%.
The 50% result is 30 percentage points below the matched 80% result because it includes $90,000 of liability premium without including liability losses. The correction is to keep the loss and premium coverage bases aligned, or to define and consistently apply a justified allocation if the design intentionally covers both. The calculation checks: \(0.80 \times $150{,}000=$120{,}000\).
Common mistake avoided. Dividing by all premium shown on an account statement can appear conservative or convenient, but it changes the question being answered. The denominator must represent the premium basis that the experience is intended to adjust.
Coverage consistency also matters when an account changes its coverage configuration. If a coverage was added, removed, or materially changed, the plan should explain how the experience and premium are treated so that only comparable coverage is included. Otherwise, the arithmetic may reflect changes in insurance purchased rather than changes in the risk's loss experience.
Define the Risk Unit Explicitly
A plan's risk unit is the entity whose experience is measured and whose premium is adjusted. Depending on the design, that could be an insured account, a location, a vehicle schedule, or another specified unit. No single choice is correct for every plan. The requirement is that the choice be explicit and used consistently in identifying losses, earned exposure, premium, and eligibility.
Ambiguity is especially consequential for accounts with multiple locations or operations. If losses are aggregated across an account but eligibility is assessed separately by location, two users may reach different outcomes while both believe they followed the plan. Account changes also need a defined treatment: for example, whether a new location joins the existing unit or is evaluated separately.
Worked Example: One Account or Two Location-Level Risks?
A hypothetical insured has two locations. During the relevant period, Location A has $55,000 of eligible losses and $70,000 of earned premium; Location B has $25,000 of eligible losses and $50,000 of earned premium. The draft plan says that the “risk” is eligible at $100,000 of earned premium but does not say whether a risk is the account or a location. Calculate the premium and loss totals under both interpretations and show why the ambiguity matters.
Method. Aggregate amounts only after stating the unit of analysis. For account-level treatment, add both locations' amounts. For location-level treatment, evaluate each location separately.
Account-level calculation. Earned premium is \($70{,}000+$50{,}000=$120{,}000\), and eligible losses are \($55{,}000+$25{,}000=$80{,}000\). The account exceeds the $100,000 threshold by $20,000 and qualifies under an account-level rule.
Location-level calculation. Location A has $70,000 of premium, which is $30,000 below the threshold. Location B has $50,000, which is $50,000 below the threshold. Neither location qualifies if each location must independently meet the threshold.
Thus the same underlying data produce an eligible account under one interpretation and no eligible location under the other. A correction could state, for example, that the insured account is the risk unit, that losses and earned premium from all covered locations of that account are aggregated, and that eligibility is tested on their combined premium. If separate location treatment is intended instead, the plan must say so and define how account-level losses and premium are allocated. The account totals check: $120,000 premium and $80,000 losses.
Common mistake avoided. Do not assume that a policy number, named insured, or location is automatically the rating unit. Name the unit in the plan and trace every input to that unit.
Make Eligibility Rules Operational
An eligibility rule should yield the same answer for the same account data regardless of who applies it. To achieve that, specify at least four items: the unit being tested, the quantity or characteristic used, the period over which it is measured, and the date or event at which eligibility is determined. The rule should also state how equality at a threshold is handled.
For example, “accounts over $100,000 qualify” leaves questions unanswered. Does exactly $100,000 qualify? Is the amount written premium, earned premium, or another specified measure? Is it measured for the experience period or the prospective term? Is the test applied to the insured account or to each location? A clear rule answers each question rather than leaving implementation to local judgment.
Worked Example: An Unclear Premium Threshold
A hypothetical draft says, “Accounts with at least $100,000 of premium are experience-rated.” The plan is intended to use earned premium for the most recently completed calendar year, test the named insured as one account, and include accounts exactly at the threshold. Account R has $97,000 of earned premium in that calendar year and $104,000 of written premium. Account S has exactly $100,000 of earned premium. Determine eligibility under the intended rule and identify the error in using written premium.
Method. Apply the specified measure, period, risk unit, and boundary convention in sequence. The stated threshold is inclusive, so an amount equal to $100,000 qualifies.
Account R. The relevant figure is $97,000 of earned premium, which is \($100{,}000-$97{,}000=$3{,}000\) below the threshold. R is not eligible. Comparing its $104,000 of written premium with the threshold would incorrectly make it eligible because it uses the wrong premium measure.
Account S. Its $100,000 of earned premium equals the inclusive threshold, so S is eligible. The boundary check is \(100{,}000 \geq 100{,}000\), which is true.
A usable correction is: “The named insured is eligible when its earned premium for the most recently completed calendar year is at least $100,000; all covered locations are aggregated, and the test is applied on the plan renewal date.” If the design uses a different premium basis or measurement date, those terms should replace the example's terms explicitly.
Common mistake avoided. “Premium” is not a sufficient definition when written and earned amounts can differ. Nor is “over” a safe substitute for a stated comparison operator: “greater than” excludes equality, while “at least” includes it.
A Practical Design Review
A concise review can reveal these failures before a plan is implemented. Work through the following sequence for each proposed adjustment:
Identify which coverage and prospective premium component the experience is intended to affect.
Write the loss-period definition, corresponding earned-exposure period, and loss valuation date.
Specify whether records are combined by account, location, or another defined unit, including the treatment of changes.
State the measure, measurement period, evaluation date, unit, threshold operator, and treatment of boundary values.
Check that every loss, exposure, and premium included in the calculation belongs to the specified period, coverage, and risk unit.
These checks matter because design ambiguity can create both measurement error and inconsistent administration. A period mismatch can distort the apparent loss cost; a coverage mismatch can shift experience across coverages; and a vague unit or threshold can make otherwise identical risks receive different treatment. Clear specifications improve reproducibility and make it possible to explain an adjustment to underwriting, pricing, and the insured.
Check Your Understanding
For each problem, identify the design issue and apply the stated correction.
- A hypothetical fleet has $144,000 of losses from accidents in Year 1. Year 1 earned exposure is 800 vehicle-years and Year 2 earned exposure is 1,200 vehicle-years. An analyst divides by Year 2 exposure. What is the correct Year 1 loss cost per vehicle-year, and how much does the incorrect result understate it? (Answer: Correct $180 per vehicle-year; incorrect $120, understating by $60, or 33.33% of the correct value.)
- A hypothetical package account has $72,000 of property losses, $90,000 of property premium, and $60,000 of liability premium. If the experience is intended to adjust property premium only, what are the incorrect combined-premium ratio and the matched property ratio? (Answer: Incorrect ratio \(72{,}000/150{,}000=48\%\); matched property ratio \(72{,}000/90{,}000=80\%\).)
- A draft plan applies a $75,000 premium threshold but leaves the risk unit unspecified. One insured has two locations with $42,000 and $38,000 of premium. Under an account-level rule, does it qualify? Under a location-level rule, does either location qualify? (Answer: Account premium is $80,000, so it qualifies under account-level treatment; neither location qualifies separately.)
- A plan requires “more than $50,000 written premium per location as of renewal.” A location has exactly $50,000 of written premium at renewal. Does it qualify? How would changing “more than” to “at least” affect the result? (Answer: It does not qualify under “more than”; it qualifies under “at least.”)
- A hypothetical account has $110,000 of earned premium for the calendar year but only $95,000 of written premium for that year. The specified eligibility rule requires at least $100,000 of earned premium, measured at renewal. Is the account eligible, and which tempting comparison is wrong? (Answer: Yes; it qualifies using $110,000 earned premium. Comparing $95,000 written premium with the threshold uses the wrong measure.)