Experience Rating Adjustments

Purpose of an Experience Rating Adjustment

An experience rating adjustment modifies a plan-defined premium base to reflect an insured’s own loss experience. The calculation compares eligible observed losses with expected losses for the same experience period and coverage basis. A credibility factor determines how strongly that comparison affects the adjustment: limited credibility tempers the observed result toward a neutral value.

The calculation is only meaningful when the plan’s rules are followed consistently. A loss limit may apply separately to each claim, expected losses may already incorporate that limit, and the adjustment may be subject to a floor or cap. These are not interchangeable details: changing their order or basis can materially change the result.

This tutorial is aligned with the SOA Exam ASTAM (Advanced Short-Term Actuarial Mathematics) syllabus. The credibility factor and any limitation rules below are treated as prescribed plan inputs; the task is to apply them correctly, not to estimate or select them.

Definition: Let $L_{\mathrm{obs}}$ be observed losses after applying the plan’s prescribed adjustments and limitations, and let $L_{\mathrm{exp}}$ be expected losses for the same period and on a consistent basis. If the plan assigns credibility $Z$, the experience adjustment factor is $$ M_{\mathrm{raw}}=(1-Z)+Z\frac{L_{\mathrm{obs}}}{L_{\mathrm{exp}}} =1+Z\left(\frac{L_{\mathrm{obs}}}{L_{\mathrm{exp}}}-1\right). $$ Here $0\leq Z\leq 1$, and the formula requires $L_{\mathrm{exp}}>0$. The plan applies $M_{\mathrm{raw}}$ to its specified premium base, subject to any stated adjustment bounds.

The ratio $L_{\mathrm{obs}}/L_{\mathrm{exp}}$ is the observed-to-expected loss ratio for the plan’s defined experience. A ratio above 1 indicates observed losses exceed expected losses; a ratio below 1 indicates they fall short. The credibility factor shrinks either departure from 1. For example, $Z=0$ gives a neutral factor of 1, while $Z=1$ gives the full observed-to-expected ratio.

Put Losses on the Plan’s Basis First

Before calculating a ratio, identify what qualifies as observed loss under the plan. A limitation may apply to each claim, to an occurrence, or to an aggregate. The wording controls. For a per-claim cap $u$, a claim with loss $X$ contributes $\min(X,u)$ to limited observed losses. Each claim is limited separately, and the resulting amounts are then summed.

Expected losses must be comparable with the observed losses. If the plan specifies that expected losses are already calculated subject to the same claim limit, use those figures as supplied; do not limit them again. If expected losses are provided on an unmodified basis but the plan requires limited expected losses, the appropriate limited expected amount must be established using the plan’s stated method. A raw actual-to-expected comparison is not valid when the numerator and denominator refer to different coverage, exposure, periods, or loss definitions.

The plan may also specify adjustments to make the experience period comparable with the rating basis, such as exposure or loss adjustments. Use the prescribed adjusted values rather than silently substituting raw historical amounts. The experience adjustment is then applied to the premium base named by the plan. It is not automatically appropriate to multiply every component of a premium by $M$; fixed expenses or other components may be treated separately under the stated rules.

Read the plan rules.
Identify the experience period, eligible losses, credibility, limitation method, premium base, and any adjustment floor or cap.

Calculate plan-basis losses.
Apply claim-level or other required limitations at the level specified, then aggregate. Confirm that expected losses use a consistent basis.

Calculate the raw adjustment.
Divide observed losses by expected losses and blend that ratio with 1 using the prescribed credibility.

Apply bounds and the premium base.
Apply any floor or cap to the raw factor in the order stated by the plan, then multiply the specified base by the final factor.

Why the Credibility Formula Has This Form

The formula is a weighted average of two ratios: the insured’s observed-to-expected ratio and the neutral ratio 1. To see why, suppose the plan’s credibility rule assigns weight $Z$ to the insured’s experience and weight $1-Z$ to the neutral complement. The resulting ratio is

M_{\mathrm{raw}} =Z\left(\frac{L_{\mathrm{obs}}}{L_{\mathrm{exp}}}\right) +(1-Z)(1).

Expanding the right-hand side gives the first formula in the definition. Alternatively, subtracting 1 from the observed-to-expected ratio shows that credibility applies only to the departure from neutral:

M_{\mathrm{raw}} =1+Z\left(\frac{L_{\mathrm{obs}}}{L_{\mathrm{exp}}}-1\right).

Thus, if experience is exactly expected, the factor is 1 for any credibility. If observed losses are 20% above expected and $Z=0.40$, the adjustment is a 8% increase, not a 20% increase. This follows directly from weighting the departure by credibility.

Worked Example: Applying Credibility to an Observed Loss Ratio

A hypothetical rating plan has a premium base of $140,000, expected losses of $60,000, and observed losses of $72,000 for the same experience period and coverage. The prescribed credibility is $Z=0.40$. There are no separate loss limits or adjustment bounds. Calculate the adjustment factor and adjusted premium.

Method. Calculate the observed-to-expected ratio, weight its departure from 1 by credibility, and apply the factor to the plan’s premium base. This avoids applying the full loss ratio when the plan prescribes partial credibility.

First calculate the ratio:

\frac{L_{\mathrm{obs}}}{L_{\mathrm{exp}}} =\frac{\$72{,}000}{\$60{,}000} =1.20.

Observed losses are 20% above expected losses. The credibility-weighted departure is $0.40(1.20-1)=0.40(0.20)=0.08$, so

M_{\mathrm{raw}}=1+0.08=1.08.

Apply the factor to the stated premium base:

\$140{,}000(1.08)=\$151{,}200.

The adjusted premium is $151,200, an increase of $11,200. As a check, $0.08(\$140{,}000)=\$11{,}200$, and adding that increase to $140,000 gives $151,200. The calculation uses the stated plan premium base; it does not assume that every possible premium component is adjusted.

Claim Limits and Adjustment Bounds

A per-claim limit changes the observed loss total before the experience ratio is calculated. It does not generally mean that the entire year’s losses are capped at the per-claim limit. For example, with a $25,000 per-claim cap, three claims of $8,000, $31,000, and $44,000 contribute $8,000, $25,000, and $25,000, respectively, for limited losses of $58,000.

A floor or cap on the adjustment factor is a different operation. First calculate the credibility-weighted raw factor. Then restrict that factor to the range stated in the plan. If $M_{\min}$ and $M_{\max}$ are the prescribed bounds, the final factor is

M=\min\left(M_{\max},\max(M_{\min},M_{\mathrm{raw}})\right).

This expression first raises any factor below the floor to the floor, then lowers any factor above the cap to the cap. Use it only when the plan specifies those bounds. Do not impose a floor or cap simply because the result seems unusually large or small.

Worked Example: Limiting Each Claim Before Credibility

A hypothetical plan applies a $25,000 cap separately to each claim. In one experience period, the insured has four eligible losses of $9,000, $28,000, $41,000, and $16,000. Expected losses are $66,000 and are already calculated subject to the same limitation. The plan assigns $Z=0.60$, and the premium base is $220,000. Calculate limited observed losses, the adjustment factor, and adjusted premium.

Method. Apply the claim limit to each observed loss, sum the limited amounts, and compare that total with expected losses on the same limited basis. This avoids confusing a per-claim cap with a cap on the aggregate.

The claim-by-claim limited amounts are

\min(\$9{,}000,\$25{,}000)=\$9{,}000,\quad \min(\$28{,}000,\$25{,}000)=\$25{,}000,

\min(\$41{,}000,\$25{,}000)=\$25{,}000,\quad \min(\$16{,}000,\$25{,}000)=\$16{,}000.

Therefore,

L_{\mathrm{obs}}=\$9{,}000+\$25{,}000+\$25{,}000+\$16{,}000 =\$75{,}000.

The observed-to-expected ratio is

\frac{\$75{,}000}{\$66{,}000} =\frac{25}{22} \approx 1.1363636.

Credibility is applied to the departure from 1:

M_{\mathrm{raw}} =1+0.60(1.1363636-1) =1+0.60(0.1363636) \approx 1.0818182.

The adjusted premium is $ \$220{,}000(1.0818182)=\$238{,}000$ to the nearest dollar. The increase is $18,000. As a check, the weighted ratio can also be calculated as $0.40+0.60(75/66)=1.0818182$. The common error is to cap the aggregate at $25,000; that would misread a per-claim limit and discard part of the experience.

Worked Example: Enforcing a Maximum Adjustment

A hypothetical plan applies a $30,000 limit per claim. The eligible observed losses are $14,000, $37,000, $29,000, $52,000, $38,000, and $18,000. Expected losses, already on the same limited basis, are $120,000. The plan specifies $Z=0.75$, a maximum adjustment factor of 1.10, and a premium base of $480,000. Calculate the final factor and adjusted premium.

Method. Limit every claim, sum the results, calculate the raw credibility-weighted factor, then apply the maximum. This avoids either applying the cap too early or overlooking it after the raw factor is calculated.

The six limited claim amounts are $14,000, $30,000, $29,000, $30,000, $30,000, and $18,000. Their sum is

L_{\mathrm{obs}} =\$14{,}000+\$30{,}000+\$29{,}000+\$30{,}000+\$30{,}000+\$18{,}000 =\$151{,}000.

The observed-to-expected ratio is $151{,}000/120{,}000=1.2583333$. Before applying the cap, the credibility-weighted factor is

M_{\mathrm{raw}} =1+0.75(1.2583333-1) =1+0.75(0.2583333) =1.19375.

Since $1.19375>1.10$, the plan’s maximum applies, giving $M=1.10$. The adjusted premium is

\$480{,}000(1.10)=\$528{,}000.

The final adjusted premium is $528,000. The raw calculation implies a factor above the allowed maximum, so the maximum—not the raw factor—is used. The result also checks directly: a 10% increase on $480,000 is $48,000.

Common Mistakes and Practical Checks

An experience adjustment combines several rules, so a numerically correct calculation can still answer the wrong plan question. Before accepting a result, verify the loss basis, the credibility calculation, the application of bounds, and the premium base.

Using the full observed-to-expected ratio as the adjustment. If $Z<1$, the plan weights the departure from 1. With a ratio of 1.20 and $Z=0.40$, the correct factor is $1+0.40(0.20)=1.08$, not 1.20.
Applying a claim limit to the total. A per-claim limit is applied to each claim before summing. Limiting an aggregate instead changes the rule.
Comparing losses on different bases. Limited observed losses should not be divided by expected losses that represent an incompatible coverage, period, exposure, or limitation basis.
Applying credibility to the premium rather than the loss ratio. Under the formula here, credibility weights the observed-to-expected loss ratio against 1. Apply the resulting factor to the premium base specified by the plan.
Forgetting the adjustment bounds. Calculate the raw factor first and then apply any stated floor or cap. Do not substitute a bound for the raw calculation.
Using the wrong base. The plan may specify a particular premium component or base. Multiplying an unrelated premium amount by the factor can produce a correct factor but an incorrect adjusted premium.

A useful final check is directional: if observed losses exceed expected losses and $Z>0$, the raw factor must exceed 1; if observed losses are below expected losses, it must be below 1. A bound may change the final factor, but it should not change the raw factor calculation.

Key takeaway: Calculate the plan-defined observed losses and expected losses on a consistent basis, form the observed-to-expected ratio, weight its departure from 1 by the prescribed credibility, and then apply any stated bounds to the factor before adjusting the specified premium base.

Check Your Understanding

For each problem, follow the stated limitation and adjustment rules. Treat amounts as hypothetical.

A plan has expected losses of $80,000, observed losses of $92,000, credibility $Z=0.35$, and a premium base of $175,000. There are no limits or bounds. Find the adjustment factor and adjusted premium. (Answer: factor 1.0525; adjusted premium $184,187.50)
A plan limits each claim to $18,000. Four observed claims are $7,000, $20,000, $15,000, and $31,000. Expected losses are $55,000 on the same limited basis, and $Z=0.50$. Find limited observed losses and the raw adjustment factor. (Answer: limited observed losses $58,000; factor approximately 1.0273)
Observed and expected losses are $54,000 and $60,000, respectively. A plan has $Z=0.70$ and restricts the adjustment factor to the interval from 0.95 to 1.08. What is the final factor? (Answer: 0.95)
A plan’s expected losses are $24,000 per exposure unit. The experience contains three exposure units and $80,000 of observed losses on the same basis. With $Z=0.40$ and a premium base of $90,000, calculate the adjustment factor and adjusted premium. (Answer: factor approximately 1.0444; adjusted premium $94,000)
A plan applies a $20,000 limit per claim. Observed claims are $12,000, $26,000, and $34,000; expected losses are $48,000 on the same limited basis. With $Z=0.60$, the plan caps the adjustment factor at 1.10 and has a premium base of $300,000. Find the final factor and adjusted premium. (Answer: limited observed losses $52,000; raw factor 1.05; final factor 1.05; adjusted premium $315,000)

What You'll Learn