Two Different Sources of Rating Information
A classification rate and an experience adjustment answer different questions. A classification rate reflects the expected cost for a broader group of risks assigned to the same class. An experience adjustment reflects information in a particular insured’s own past loss record. Keeping those sources separate helps explain why two insureds in the same class can receive different indicated costs, and why two insureds with similar loss histories can still have different indicated costs if they belong to different classes.
Classification rating first assigns a risk to a class using relevant characteristics of the risk. The class rate is then based on the experience or expected cost of the broader group, rather than being calculated solely from that individual risk’s losses. Experience rating uses the insured’s own experience as an additional signal about its prospective cost. The individual record is usually not treated as fully predictive: it can be limited or volatile, so a class-based amount may remain an important benchmark.
The distinction is about the source of information, not whether the risk has been individually examined. Characteristics of an individual risk can determine its classification. What makes a rating classification-based is that its class rate comes from the broader group benchmark, not directly from that insured’s own loss history. Conversely, an experience adjustment is specific to the insured, even when it uses a class rate as its complement.
A Simple Framework for Comparing the Indications
For the calculations below, let \(C_c\) be the classification loss cost for class \(c\), measured in dollars per exposure unit. Let \(\bar L_i\) be insured \(i\)’s adjusted historical loss cost per exposure unit, and let \(Z_i\) be a stated credibility weight between 0 and 1. The illustrative experience-rated cost is
Here, \(Z_i\) determines how much weight the indication gives to the insured’s own record. The remaining weight is assigned to the classification complement. This is a simplified credibility-weighted comparison, not a universal formula for every filed or regulatory experience-rating plan. Actual plans can define experience measures and adjustments differently.
In these examples, exposures are already on a consistent basis, losses are adjusted to the same coverage and cost level, and the supplied \(Z_i\) applies to the indicated insured. The future loss cost is calculated as exposure multiplied by the indicated cost per exposure unit. No interest rate, mortality assumption, or loss probability distribution is used: the examples are deterministic calculations from stated exposure and loss amounts.
Use the broader group’s expected loss cost per exposure unit for the insured’s classification.
Divide the insured’s adjusted historical losses by the corresponding historical exposures.
Combine the insured’s loss cost and the class complement using their respective weights.
Multiply each prospective cost indication by the same prospective exposure to compare total indicated losses.
Why the Credibility-Weighted Formula Has This Form
The formula is a weighted average of two cost indications on the same per-exposure basis. The insured-specific indication is \(\bar L_i\), and the class complement is \(C_c\). If the own-experience weight is \(Z_i\), the remaining weight must be \(1-Z_i\), so the combined indication is \(Z_i\bar L_i+(1-Z_i)C_c\). The weights add to one:
This ensures that if the two indications happen to be equal, the combined result equals that common value. It also gives the intended endpoints: if \(Z_i=0\), then \(R_i=C_c\); if \(Z_i=1\), then \(R_i=\bar L_i\). For an intermediate weight, the result lies between the two indications. In particular, if \(\bar L_i>C_c\) and \(0<Z_i<1\), then \(C_c<R_i<\bar L_i\). These properties help check a calculation.
The formula is appropriate here only under the stated simplified setup: both inputs measure comparable costs per exposure unit, the weights are specified, and no additional plan adjustments are being applied. It should not be used by substituting raw losses for a loss cost, by mixing incompatible exposure bases, or by assuming that every experience-rating plan uses the same calculation.
Paired Rating Examples
Worked Example: One Class Rate, One Insured-Specific Adjustment
A hypothetical service class has $72,000 of adjusted losses across 300 exposure units. An insured in that class had $1,600 of adjusted losses across 8 exposure units. For the simplified comparison, use \(Z=0.40\). The insured is expected to have 10 exposure units prospectively. Calculate the classification-only and experience-rated indicated losses. All amounts are on the same coverage and cost basis; no interest rate, mortality assumption, or probability distribution is used.
Method. Calculate the class loss cost and the insured’s own historical loss cost, combine them using the stated credibility weight, and multiply both indications by the prospective exposure.
1. Classification loss cost. Divide the broader class losses by class exposures:
2. Insured’s historical loss cost.
3. Experience-rated cost. The insured’s experience receives weight 0.40 and the class complement receives weight \(1-0.40=0.60\):
4. Prospective indicated losses. The classification-only indication is \(10(\$240)=\$2{,}400\). The experience-rated indication is \(10(\$224)=\$2{,}240\), which is \(\$160\) lower.
Check. The experience-rated cost, $224, lies between the insured’s $200 historical cost and the class complement of $240. The total difference is \(10(\$240-\$224)=\$160\), agreeing with the calculation. The common mistake this method avoids is treating the class rate as if it already included this insured’s own loss history; it is the group benchmark before the risk-specific adjustment.
Worked Example: Same Individual Experience, Different Class Complements
Two hypothetical insureds each have $960 of adjusted losses over 4 exposure units. One is assigned to Class A and the other to Class B. Class A has $60,000 of adjusted group losses across 200 exposure units; Class B has $36,000 across 200 exposure units. Use \(Z=0.50\) for each insured, and assume each will have 3 prospective exposure units. Calculate the two classification-only and experience-rated indications. The experience and class figures are on a consistent basis; no interest rate, mortality assumption, or loss distribution is used.
Method. First calculate each class’s pooled loss cost. Then use each insured’s common own loss cost with the appropriate class complement.
1. Class complements.
2. Individual loss costs. Each insured has the same historical loss cost:
3. Class A insured.
The classification-only indication is \(3(\$300)=\$900\). The experience-rated indication is \(3(\$270)=\$810\).
4. Class B insured.
The classification-only indication is \(3(\$180)=\$540\). The experience-rated indication is \(3(\$210)=\$630\).
Check. The same own loss cost does not produce the same final indication because the class complements differ. The Class A indication of $270 lies between $240 and $300; the Class B indication of $210 lies between $180 and $240. The method avoids the mistake of assuming that equal individual experience erases the difference between classifications.
Worked Example: Why an Individual Loss Cost Does Not Replace the Class Benchmark
A hypothetical equipment class has $125,000 of adjusted losses across 500 exposure units. An insured in that class has $2,400 of adjusted losses across 6 exposure units. Use \(Z=0.30\), and assume 12 prospective exposure units. Compare the classification-only and experience-rated indicated losses. All losses and exposures are consistently defined; no interest rate, mortality assumption, or loss probability distribution is used.
Method. Calculate the two per-exposure costs, then credibility-weight them. Do not apply the insured’s full historical cost unless the stated weight is 1.
1. Class complement.
2. Insured’s historical cost.
3. Experience-rated cost. The own-experience weight is 0.30, leaving a class-complement weight of \(1-0.30=0.70\):
4. Prospective indications. The classification-only amount is \(12(\$250)=\$3{,}000\). The experience-rated amount is \(12(\$295)=\$3{,}540\), or $540 higher.
Check and trap. Because the own loss cost exceeds the complement and \(Z=0.30\), the answer must be between $250 and $400. It is. A common incorrect calculation is to use the full own loss cost: \(12(\$400)=\$4{,}800\). That calculation silently sets \(Z=1\), rather than using the given \(Z=0.30\). It overstates this simplified experience-rated indication by \(\$4{,}800-\$3{,}540=\$1{,}260\).
Interpreting the Difference
A classification-only indication answers, “What loss cost is assigned to this class?” An experience-rated indication answers, “What cost results when this insured’s own experience is considered alongside the class benchmark?” Neither number should be confused with the other. The class rate is not the insured’s observed loss cost, and the experience-rated result is not necessarily the insured’s raw historical loss cost.
The direction of the change follows directly from the formula:
For \(Z_i\geq 0\), the experience-rated cost moves from the class complement toward the insured’s own loss cost. It moves upward when the insured’s loss cost is higher than the complement, downward when it is lower, and does not move when the two are equal. If \(Z_i=0\), the own experience has no effect in this simplified formula; if \(Z_i=1\), the class complement has no effect. This is a useful arithmetic check, not a claim that every real plan permits either endpoint.
In practice, the experience measure and class benchmark need to be comparable. Differences in exposure definitions, coverage, valuation basis, or cost level can make a direct comparison misleading. A calculation can be arithmetically correct and still be conceptually wrong if, for example, one loss cost includes a coverage that the other does not. The simplified examples isolate the distinction between the two information sources; they do not specify a complete experience-rating plan.
- Classification rating: the rate is associated with the broader class and its expected experience.
- Experience rating: the rate indication responds to the particular insured’s own experience, with the class serving as a benchmark in the illustrated credibility approach.
- Combined interpretation: the risk’s classification and individual experience can both affect the indicated cost, but through different inputs.
Check Your Understanding
For each problem, use the simplified credibility-weighted framework defined above. Assume all losses and exposures are adjusted consistently; no interest rate, mortality assumption, or probability distribution is required.
- A class loss cost is $275 per exposure unit. An insured’s own adjusted loss cost is $175 per exposure unit, \(Z=0.40\), and prospective exposure is 8 units. Find the experience-rated indicated losses. (Answer: $1,880)
- An insured has $720 of adjusted losses over 3 exposure units. Its class complement is $300, \(Z=0.25\), and prospective exposure is 4 units. Find the experience-rated indicated losses. (Answer: $1,140)
- Class X has $48,000 of losses over 160 exposure units; Class Y has $27,000 over 150. Two insureds, one in each class, each have an own loss cost of $240, use \(Z=0.50\), and each has 2 prospective exposure units. Find the two experience-rated indicated losses and their difference. (Answer: Class X $540; Class Y $420; difference $120)
- A class complement is $210 per exposure unit. An insured has $1,080 of adjusted losses over 6 exposure units, \(Z=0.35\), and prospective exposure is 10 units. Find the experience-rated indicated losses. (Answer: $1,995)
- A class complement is $320 per exposure unit. An insured’s own loss cost is $250, \(Z=0.60\), and prospective exposure is 7 units. Find the experience-rated indicated losses. (Answer: $1,946)