Why Claim-Free Experience Does Not Always Mean a Zero Modification
A claim-free experience period has no included actual losses, but an experience modification is generally not calculated by setting the factor to zero. The plan gives some weight to the insured’s observed experience and some weight to the expected experience. With no observed losses, the credibility-weighted estimate still retains the portion based on expected losses. A plan’s minimum factor can also limit how far the final modification falls.
The calculation therefore has two distinct stages: first calculate the raw credibility-weighted modification from the plan’s loss basis; then apply the stated minimum-factor rule. Keep the expected-loss denominator, the credibility weight, and the floor in their specified roles. This tutorial is aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus.
Deriving the Claim-Free Result
The plan’s credibility-weighted estimate combines actual and expected losses. The estimate gives weight \(Z\) to actual losses and weight \(1-Z\) to expected losses. To express that estimate as a modification, divide it by the expected losses used as the plan’s benchmark.
This also shows why the claim-free result is \(1-Z\): setting \(A=0\) makes the actual-loss part of the credibility-weighted estimate zero, but leaves \((1-Z)E\). Dividing by \(E\) gives \(1-Z\). The result depends on the stated model; it is not a universal rule for every experience-rating plan. For example, if the plan specifies a different formula, use that formula rather than substituting this one.
A minimum factor changes the final answer only when the raw factor is below the stated minimum. The operation is a maximum, not an addition: the minimum factor replaces a lower raw value. It does not alter \(Z\), \(A\), or \(E\), and it is applied after the raw modification has been calculated.
Determine which actual and expected losses are included, and confirm that the supplied credibility applies to that same basis.
If there are no included claims or losses, set \(A=0\). Do not treat a claim as zero merely because it is below a separate threshold unless the stated plan basis says to do so.
For claim-free experience, calculate \(1-Z\). If the plan provides a minimum, compare this raw value with the minimum only after the calculation.
Use the larger of the raw modification and the minimum factor, if one is specified. Preserve unrounded values during the calculation when a rounding rule has not been supplied.
Worked Claim-Free Calculations
The examples use hypothetical experience-rating plans. In each one, the values and plan rules are stipulated for the problem; no interest rate, mortality assumption, or probability distribution is used. These are deterministic calculations from supplied inputs, not projections from a loss distribution.
Worked Example: One Expected-Loss Total With a Binding Minimum
A hypothetical plan has expected losses of $150,000, credibility \(Z=0.42\), and no actual losses during the experience period. Its minimum modification is 0.62. Calculate the raw and final modifications. Assume all amounts are on the same loss basis and the supplied credibility applies to the full period. No interest, mortality, or loss distribution is assumed.
Step 1: Identify the inputs. The expected losses are \(E=\$150{,}000\), the included actual losses are \(A=\$0\), credibility is \(Z=0.42\), and the minimum factor is \(F=0.62\). Since \(E>0\), the loss-ratio formula is defined.
Step 2: Calculate the raw factor.
As a check using the weighted-loss form, \(ZA=0.42(\$0)=\$0\), while \((1-Z)E=0.58(\$150{,}000)=\$87{,}000\). The weighted losses are therefore \(\$87{,}000\), and \(\$87{,}000/\$150{,}000=0.58\).
Step 3: Apply the minimum. The final modification is
The minimum binds, so the final factor is 0.62. The common mistake here is to report 0.58 because it is the claim-free raw result; the plan’s minimum requires the larger value.
Worked Example: Combining Expected Losses Across Experience Years
A hypothetical plan evaluates three claim-free experience years together. Expected losses are $72,000, $96,000, and $132,000, and the plan supplies one credibility of \(Z=0.55\) for the combined period. The minimum modification is 0.40. Calculate the final modification. The same loss basis applies in every year; no interest, mortality, or loss distribution is assumed.
Step 1: Combine the expected losses. Because the supplied credibility applies to the combined period, add the expected losses before using the formula:
Step 2: Confirm combined actual losses. Each year is claim-free on the stated basis, so \(A=\$0+\$0+\$0=\$0\).
Step 3: Calculate the raw modification.
A direct check gives a weighted loss estimate of \(0.55(\$0)+0.45(\$300{,}000)=\$135{,}000\). Dividing by total expected losses gives \(\$135{,}000/\$300{,}000=0.45\).
Step 4: Apply the minimum.
The final modification is 0.45; the minimum does not bind. A common error is to apply the supplied credibility to each year and then add the resulting factors. A factor is not an amount to sum: with one credibility for the combined period, aggregate the expected losses and calculate one modification.
Worked Example: Claim-Free Components With Different Credibilities
A hypothetical plan separates expected losses into two components and assigns a separate credibility to each. Component A has expected losses of $240,000 and credibility \(Z_A=0.30\); Component B has expected losses of $160,000 and credibility \(Z_B=0.50\). Both components are claim-free, and the plan’s minimum overall modification is 0.60. Calculate the overall raw and final modifications. No interest, mortality, or loss distribution is assumed.
Step 1: Calculate each component’s weighted losses. For a claim-free component \(j\), the weighted loss estimate is \((1-Z_j)E_j\).
Step 2: Add the weighted losses and expected losses.
Step 3: Calculate the overall raw modification.
Step 4: Apply the minimum. Since \(0.62>0.60\), the minimum does not bind, and the final modification is 0.62.
This aggregation weights each component’s factor by its expected losses. It is not the unweighted average of the two component factors: that average would be \((0.70+0.50)/2=0.60\), which incorrectly gives the $240,000 component the same weight as the $160,000 component. The weighted-loss calculation checks the answer against the overall expected-loss denominator.
Conditions, Interpretation, and Common Traps
The formula is useful only when its inputs match the plan definition. In particular, \(A\) and \(E\) must cover the same experience period and loss basis, and \(Z\) must be the credibility assigned to that basis. If losses are split into components with different credibilities, calculate each component’s weighted losses and aggregate those amounts, as in the third example. Do not apply one credibility to the total unless the plan supplies one for the total.
The denominator must be positive. If \(E=0\), the ratio \(A/E\) and the modification formula are undefined; do not conclude that the factor is zero or use \(1-Z\) without an applicable plan rule. The earlier material on zero expected losses and small denominators treats that issue. Also, claim-free means no losses counted under the specified experience basis. It does not mean that expenses or every other premium component are zero.
| Situation | Correct treatment | Common error |
|---|---|---|
| Claim-free experience, \(E>0\) | Calculate \(M_{\mathrm{raw}}=1-Z\) | Set the modification equal to zero |
| A minimum factor \(F\) is specified | Calculate \(M=\max(M_{\mathrm{raw}},F)\) | Add \(F\) to the raw modification |
| One credibility applies to several years | Combine the expected and actual losses, then calculate one factor | Add annual modification factors |
| Components have different credibilities | Calculate and aggregate their weighted loss amounts | Take an unweighted average of component factors |
| Expected losses equal zero | Recognize that the ratio formula is undefined | Divide by zero or assume the claim-free rule still applies |
A useful check is to consider the endpoints of credibility. If \(Z=0\), the raw factor is 1, because none of the observed experience is given weight. If \(Z=1\) and the period is claim-free, the raw factor is 0 under this particular formula, because the weighted estimate gives full weight to zero actual losses. A plan minimum may raise that result. These endpoint checks can reveal a reversed credibility weight or a sign error.
Check Your Understanding
For each hypothetical plan, calculate the requested factor from the stated inputs. Unless specified otherwise, assume the loss basis is consistent, expected losses are positive, and no interest, mortality, or loss distribution is used.
- A claim-free account has expected losses of $90,000, credibility \(Z=0.25\), and a minimum factor of 0.70. What are the raw and final modifications? (Answer: raw \(0.75\); final \(0.75\).)
- A claim-free two-year period has expected losses of $110,000 and $140,000, one supplied credibility \(Z=0.48\), and a minimum factor of 0.55. Calculate the combined expected losses and final modification. (Answer: combined expected losses \(\$250{,}000\); raw \(0.52\); final \(0.55\).)
- A claim-free plan has one expected-loss total of $360,000, credibility \(Z=0.37\), and minimum factor 0.58. Calculate the raw factor, identify whether the minimum binds, and give the final factor. (Answer: raw \(0.63\); the minimum does not bind; final \(0.63\).)
- A claim-free account has component A expected losses of $100,000 with credibility \(0.20\), and component B expected losses of $300,000 with credibility \(0.60\). The plan minimum is 0.50. Calculate the overall raw factor using weighted losses and then the final factor. (Answer: weighted losses \(\$80{,}000+\$120{,}000=\$200{,}000\); total expected losses \(\$400{,}000\); raw \(0.50\); final \(0.50\).)
- A claim-free account has expected losses of $180,000, credibility \(Z=0.65\), and minimum factor 0.38. Calculate the raw factor and final factor, and state whether the minimum changes the answer. (Answer: raw \(0.35\); final \(0.38\); the minimum binds.)