Defining the Rated Risk Unit

Why the Rated Risk Unit Matters

An experience rating plan needs a defined object whose experience it will evaluate and whose premium it will affect. That object might be an insured, a policy, a location, or a larger account containing several policies. The choice is consequential: it determines which losses are combined, how experience is summarized, and where the resulting premium adjustment is applied.

The rated risk unit is not the same thing as the exposure measure. The unit says which entity is being rated; the exposure measure says how much insured exposure is represented for that entity. Nor does defining the unit, by itself, determine the experience period or the eligibility rules. Those are separate plan-design choices.

Definition: The rated risk unit is the entity to which an experience rating plan assigns experience and applies a resulting premium adjustment. A plan must specify how each relevant loss, expected loss, and premium base is associated with that entity. Examples include one policy, one insured across several policies, one insured location, or one account comprising multiple insured entities.

A sound definition answers two linked questions. First, what is the boundary of the unit: which policies, locations, or insured operations belong to it? Second, how is the experience assigned to that boundary? For example, if an insured has two locations, the plan must make clear whether a loss at the first location affects only that location’s experience or the insured’s combined experience.

The premium application also needs a clear boundary. A modifier calculated for an insured might be applied to that insured’s combined premium, then allocated to its policies for billing. Alternatively, the plan might calculate separate modifiers for individual locations and apply them directly to location-level premiums. Aggregating experience at one level and applying its adjustment at another is possible, but the relationship between those levels must be explicit.

Aggregation and the Experience Ratio

For a basic illustration, let $A_j$ denote actual losses and $E_j$ expected losses for unit $j$, measured on the same basis. A common credibility adjustment uses a credibility weight $Z$, where $0 \leq Z \leq 1$, to blend the unit’s experience ratio with the benchmark ratio of 1:

M_j=1+Z\left(\frac{A_j}{E_j}-1\right), \qquad E_j>0.

Here $M_j$ is the indicated multiplier on the applicable prospective premium base. This formula is used below as an illustrative plan rule, not as a claim that every experience plan uses the same formula. The calculations presume that actual and expected losses cover the same unit and scope, and that the stated value of $Z$ applies to that unit.

If several units are combined, the aggregate actual-to-expected ratio is not generally the simple average of the unit ratios. It is weighted by expected losses. To see why, write $r_j=A_j/E_j$. Then $A_j=E_jr_j$, so:

\frac{\sum_j A_j}{\sum_j E_j} = \frac{\sum_j E_jr_j}{\sum_j E_j} = \sum_j\left(\frac{E_j}{\sum_k E_k}\right)r_j.

Thus, a unit with twice the expected losses receives twice the weight in the aggregate experience ratio. If all units use the same $Z$, applying the formula to the aggregate ratio also gives the expected-loss-weighted average of their indicated multipliers:

1+Z\left(\frac{\sum_j A_j}{\sum_j E_j}-1\right) = \sum_j\left(\frac{E_j}{\sum_k E_k}\right) \left[1+Z\left(\frac{A_j}{E_j}-1\right)\right].

This identity explains an important design consequence: combining units can preserve an expected-loss-weighted indication while changing which entity receives the adjustment. It does not imply that the adjustment to total premium will be identical under every unit definition. Premium bases may not be proportional to expected losses, and plans may use different credibility weights or adjustment rules for different units.

Worked Example: Combining Two Locations Into One Account

A hypothetical insured has two locations. For the same stated experience scope, location A has actual losses of $72,000 and expected losses of $100,000; location B has actual losses of $128,000 and expected losses of $100,000. The plan uses $Z=0.60$. Each location has a $100,000 prospective premium base. Compare rating the account as one unit with rating the locations separately.

Method. Calculate the actual-to-expected ratio for each proposed unit, apply the given credibility formula, and multiply each indicated modifier by its corresponding premium base. All amounts are hypothetical; no interest or mortality assumptions are needed.

Account as the unit. Combine the losses and expected losses:

A=72{,}000+128{,}000=200{,}000,\qquad E=100{,}000+100{,}000=200{,}000.

The account ratio is $200{,}000/200{,}000=1.00$. Therefore, $M=1+0.60(1.00-1)=1.00$. The total premium base is $200,000, so the account-rated premium is $200{,}000(1.00)=\$200{,}000.00$.

Locations as separate units. For location A, $A_A/E_A=72{,}000/100{,}000=0.72$, so $M_A=1+0.60(0.72-1)=1-0.168=0.832$. Its indicated premium is $100{,}000(0.832)=\$83{,}200.00$.

For location B, $A_B/E_B=128{,}000/100{,}000=1.28$, so $M_B=1+0.60(1.28-1)=1+0.168=1.168$. Its indicated premium is $100{,}000(1.168)=\$116{,}800.00$. The two indicated premiums total $200,000.00.

The account-level result leaves the two equal base premiums at their combined benchmark level; the location-level results shift premium from A to B. Their totals happen to agree here because the expected losses and premium bases are equal and the same $Z$ is used. Common mistake: averaging 0.72 and 1.28 without checking the weights. They have equal weights in this example only because both expected losses are $100,000.

Premium Application Can Change the Result

Even when experience is attributed consistently, the premium basis matters. An account modifier might apply to the combined account premium, whereas separate policy modifiers apply to separate policy premiums. Those totals need not match. The difference is not necessarily a calculation error: it can reflect a deliberate choice of rated unit and premium application rule.

Worked Example: One Insured or Two Policy Units

A hypothetical insured has two policies. Policy 1 has actual losses of $30,000, expected losses of $60,000, and a $40,000 premium base. Policy 2 has actual losses of $120,000, expected losses of $60,000, and a $60,000 premium base. Assume $Z=0.50$ for whichever unit is rated. Compare one insured-level unit with two policy-level units.

Method. First combine actual and expected losses for the insured-level calculation. Then, for the policy-level calculation, calculate each policy’s ratio and modifier separately. Apply each result to the matching premium base.

Insured-level unit. Total actual losses are $30{,}000+120{,}000=\$150{,}000$; total expected losses are $60{,}000+60{,}000=\$120{,}000$. The ratio is $150{,}000/120{,}000=1.25$, giving $M=1+0.50(1.25-1)=1.125$. The combined premium base is $40{,}000+60{,}000=\$100{,}000$, so the indicated insured-level premium is $100{,}000(1.125)=\$112{,}500.00$.

If that total is allocated between policies in proportion to their original premium bases, Policy 1 receives $112{,}500(40{,}000/100{,}000)=\$45{,}000.00$, and Policy 2 receives $112{,}500(60{,}000/100{,}000)=\$67{,}500.00$. These allocations sum to $112,500.00.

Policy-level units. Policy 1’s ratio is $30{,}000/60{,}000=0.50$, so its modifier is $1+0.50(0.50-1)=0.75$. Its indicated premium is $40{,}000(0.75)=\$30{,}000.00$. Policy 2’s ratio is $120{,}000/60{,}000=2.00$, so its modifier is $1+0.50(2.00-1)=1.50$. Its indicated premium is $60{,}000(1.50)=\$90{,}000.00$. The policy-level total is $120,000.00.

The insured-level premium and policy-level total differ by $7,500.00. Aggregation changes the experience applied to each premium base; allocating an insured-level adjustment in proportion to premium does not recreate the separate policy-level calculations. Common mistake: calculate the insured’s modifier once, then treat it as if it were the modifier each policy would receive when rated separately.

Aligning Attribution With Application

A useful design check is to trace an experience item from its source to its effect on premium. For each loss, identify the entity to which it is assigned. For each expected loss, identify the same entity and scope. Then identify which premium base receives the resulting adjustment. A boundary mismatch can distort the result even when every arithmetic step is correct.

Define the entity boundary.
State whether the unit is a policy, insured, location, account, or another precisely defined entity. Explain how multiple policies or locations relate to it.

Assign experience once and consistently.
Associate each relevant actual loss and its expected-loss benchmark with the intended unit. Avoid omitting a loss or assigning the same loss to multiple units unless the plan explicitly defines such treatment.

Match the adjustment to the premium base.
Specify whether a unit-level modifier applies to that unit’s premium directly or to a combined premium before allocation. Check that the stated allocation rule preserves the intended total.

Worked Example: Rating Two Locations on One Policy

A hypothetical policy covers two locations. Location A has actual losses of $40,000 and expected losses of $25,000; location B has actual losses of $0 and expected losses of $25,000. The policy’s prospective premium base is $80,000, allocated as $30,000 to A and $50,000 to B. Assume $Z=0.40$. Compare location-level rating with one policy-level unit.

Method. For location-level rating, use each location’s own actual and expected losses, then apply its modifier to its allocated premium. For policy-level rating, combine losses and apply one modifier to the full policy base.

Locations as units. Location A has ratio $40{,}000/25{,}000=1.60$, so $M_A=1+0.40(1.60-1)=1.24$. Its premium is $30{,}000(1.24)=\$37{,}200.00$. Location B has ratio $0/25{,}000=0$, so $M_B=1+0.40(0-1)=0.60$. Its premium is $50{,}000(0.60)=\$30{,}000.00$. The location-rated total is $37{,}200+30{,}000=\$67{,}200.00$.

Policy as one unit. Combined actual losses are $40,000 and combined expected losses are $25{,}000+25{,}000=\$50{,}000$. The ratio is $40{,}000/50{,}000=0.80$, so $M=1+0.40(0.80-1)=0.92$. The policy-level premium is $80{,}000(0.92)=\$73{,}600.00$.

The results differ by $6,400.00 because the location-level modifiers are applied to unequal premium bases, while the policy-level modifier applies to their combined base. The loss at location A is not counted twice; it is included once in the aggregate. Common mistake: apply the account or policy modifier to each location’s premium and also apply location modifiers, which would adjust the same premium base twice.

Common Design Errors and Exam Traps

The unit boundary should be stated before calculating a ratio. Otherwise, it is easy to combine losses at one level while using expected losses or premium bases from another. The following checks are useful in a plan-design question:

Do not confuse a rated unit with an exposure measure. A policy may be the unit while payroll, sales, or another measure quantifies its exposure. Naming one does not define the other.
Do not mix levels in a ratio. If actual losses are aggregated across an account, the expected losses in the denominator must represent that same account and scope.
Do not assume the aggregate ratio is an unweighted average. It is weighted by expected losses, as shown in the derivation.
Do not assume a premium total is preserved by changing the unit. That conclusion requires additional conditions, such as matching bases and common credibility weights; it is not automatic.
Handle zero credibility correctly. If $Z=0$, then $M=1$ regardless of the experience ratio. If $A<E$ and $Z>0$, then $M<1$; if $A<E$ and $Z=0$, the indicated level equals the benchmark.

For the final point, the direction follows directly from the formula. When $A<E$, the difference $A/E-1$ is negative. Multiplying by $Z>0$ keeps it negative, so adding it to 1 gives $M<1$. But with $Z=0$, the product is zero and $M=1$. Therefore, it is incorrect to say that below-benchmark experience always produces a below-benchmark indication without specifying positive credibility.

Key takeaway: Define the rated risk unit by matching three boundaries: the entity whose experience is measured, the entity to which losses and expected losses are attributed, and the premium base to which the adjustment applies. If those boundaries differ, state and justify the aggregation and allocation rules explicitly.

Check Your Understanding

For each problem, use the stated credibility formula $M=1+Z(A/E-1)$ where applicable. Monetary answers are rounded to the nearest cent.

A hypothetical policy is the rated unit. Its actual losses are $72,000, expected losses are $80,000, $Z=0.25$, and premium base is $64,000. Find the indicated modifier and premium. (Answer: $M=0.975$; $62,400.00)
A hypothetical insured has two policies with actual losses of $60,000 and $30,000, and expected losses of $40,000 and $60,000, respectively. If the insured is the unit and $Z=0.40$, find the aggregate ratio and modifier. (Answer: $A/E=0.9000$; $M=0.960$)
Two location units have $(A_1,E_1)=(\$18{,}000,\$30{,}000)$ and $(A_2,E_2)=(\$84{,}000,\$70{,}000)$. With common $Z=0.50$, calculate the aggregate ratio and compare it with the unweighted average of the two location ratios. (Answer: aggregate ratio $=102{,}000/100{,}000=1.0200$; unweighted average $=(0.6000+1.2000)/2=0.9000$)
A unit has actual losses of $45,000, expected losses of $60,000, and $Z=0$. Is its indicated modifier below 1? (Answer: No; $M=1.000$)
A hypothetical account contains two units with expected losses of $40,000 and $60,000, actual losses of $60,000 and $30,000, and a common $Z=0.40$. If the units are combined, calculate the modifier. Then state whether the result equals the expected-loss-weighted average of the two unit modifiers. (Answer: aggregate $M=0.960$; yes, the expected-loss-weighted average is $0.4(1.200)+0.6(0.800)=0.960$)

What You'll Learn