Why Apply Credibility to Individual Experience?
An account’s past experience can differ from the manual indication, but it may not be appropriate to use that experience in full. Applying assigned credibility provides a structured adjustment: the account’s experience indication receives weight \(Z\), while the manual indication receives weight \(1-Z\). The result is an individual account indication that lies between the two, provided \(0\leq Z\leq1\).
The calculation is only meaningful when the two indications are on a comparable basis. For example, both should apply to the same account, coverage, exposure period, and prospective rating basis. In the examples below, experience indications are assumed to have already been adjusted to the appropriate prospective basis. The credibility \(Z\) is given; determining how to assign credibility is not the task here.
When the indications are expressed as rates rather than total premiums, use the same blend on the rates. If \(r_M\) is the manual rate per exposure and \(r_E\) is the experience rate indication per exposure, then the credible rate is \(r_C=Zr_E+(1-Z)r_M\). Multiplying \(r_C\) by the account’s exposure gives the premium, assuming the rate is charged per unit of that exposure.
From a Credible Premium to a Modification
A premium modification describes the credible premium relative to its manual baseline. If \(M>0\), define the credible modification factor as \(C/M\); the percentage change is \((C/M-1)\times100\%\). A factor above 1 indicates an increase, and a factor below 1 indicates a decrease. The manual premium is the comparison point, not an additional amount to add after blending.
This identity is useful when experience is already summarized as a factor. If \(R=E/M\) is the experience modification factor, the credible modification factor is \(1+Z(R-1)\). Thus credibility applies to the deviation from the manual factor of 1. It does not apply to the whole factor and then replace the manual baseline.
Why the Blending Formula Works
The formula is a weighted average of two indications. Since \(Z+(1-Z)=1\), it preserves the common level of the indications while assigning the specified share to each source. Its position between the two values can also be shown directly. If \(E\geq M\), then \(C-M=Z(E-M)\geq0\), while \(E-C=(1-Z)(E-M)\geq0\). Therefore \(M\leq C\leq E\). If \(E\leq M\), the same differences show \(E\leq C\leq M\).
For the modification-factor form, divide the premium blend by \(M>0\), then substitute \(R=E/M\): $$ \frac{C}{M} =\frac{ZE+(1-Z)M}{M} =Z\frac{E}{M}+(1-Z) =1+Z(R-1). $$ This derivation explains why the manual baseline remains at factor 1 and only the experience deviation is scaled by credibility.
Applying Credibility to Premium Indications
Worked Example: Blending a Premium and Interpreting the Change
A hypothetical small equipment account has a manual annual premium of $24,000 and a prospective experience-based premium indication of $30,000. Its assigned credibility is \(Z=0.40\). Find the credible premium and its percentage change from the manual premium.
Method. Blend the two premiums using their complementary weights. The manual weight is \(1-Z=1-0.40=0.60\).
Credible premium. $$ C=0.40(30{,}000)+0.60(24{,}000) =12{,}000+14{,}400 =\$26{,}400. $$
Modification factor and percentage change. The manual premium is positive, so it is valid to compare by division: $$ \frac{C}{M}=\frac{26{,}400}{24{,}000}=1.10, \qquad (1.10-1)\times100\%=10\%. $$ The credible premium is $2,400 above the manual premium, since \(26{,}400-24{,}000=2{,}400\), and \(2{,}400/24{,}000=0.10\). The 25% experience increase has therefore been reduced to a 10% credible increase. The check agrees with the modification formula: \(1+0.40(30{,}000/24{,}000-1)=1+0.40(0.25)=1.10\).
Common mistake avoided. Multiplying the experience factor \(1.25\) by \(0.40\) gives \(0.50\), but that is not the credible modification: it drops the \(60\%\) manual contribution. The correct factor is \(0.40(1.25)+0.60(1)=1.10\).
Applying Credibility to Rates and Exposure
A rate indication may be more convenient than a premium indication when the account has a stated number of exposure units. First blend the rates on a consistent per-exposure basis. Then multiply by exposure. The premium percentage change is the same as the rate percentage change if the same exposure quantity is used for both the credible and manual premiums.
Worked Example: Credible Rate for a Fixed Exposure Count
A hypothetical commercial account has 200 exposure units. Its manual rate is $680 per unit, its prospective experience rate indication is $748 per unit, and assigned credibility is \(Z=0.50\). Calculate the credible rate, account premium, and percentage change from the manual premium.
Blend the rates. The manual rate weight is \(1-0.50=0.50\): $$ r_C=0.50(748)+0.50(680)=374+340=\$714 $$ per exposure unit.
Calculate both premiums. At the manual rate, the premium is \(200(680)=\$136{,}000\). At the credible rate, it is \(200(714)=\$142{,}800\).
Find the change. The dollar increase is \(142{,}800-136{,}000=\$6{,}800\). The percentage increase is $$ \frac{6{,}800}{136{,}000}\times100\%=5\%. $$ This also follows from the rate change: \((714/680-1)\times100\%=5\%\). As a check, the experience rate was 10% above manual because \((748-680)/680=0.10\); applying 50% credibility gives a 5% credible increase.
Common mistake avoided. The rate indication is per exposure unit, not the account’s total premium. Multiplying $748 by credibility and treating the result as the account premium mixes units and omits the manual contribution. Blend rates first, then multiply by 200 exposure units.
Combining Coverage-Specific Results
If an account has several coverages, calculate each coverage’s credible premium using its own manual premium, experience indication, and assigned credibility. Then add the credible premiums. The account-level percentage change is the change in that total relative to the total manual premium. In general, it is not the simple average of the coverage-specific percentage changes: the premium amounts determine their contribution to the overall result.
Worked Example: Combining Two Credibility-Adjusted Coverages
A hypothetical account has two coverages. Coverage X has a manual premium of $30,000, an experience indication of $36,000, and \(Z_X=0.50\). Coverage Y has a manual premium of $10,000, an experience indication of $8,000, and \(Z_Y=0.25\). Find the credible premium for each coverage, the combined credible premium, and its percentage change from the combined manual premium.
Coverage X. The manual weight is \(1-0.50=0.50\): $$ C_X=0.50(36{,}000)+0.50(30{,}000) =18{,}000+15{,}000 =\$33{,}000. $$
Coverage Y. The manual weight is \(1-0.25=0.75\): $$ C_Y=0.25(8{,}000)+0.75(10{,}000) =2{,}000+7{,}500 =\$9{,}500. $$
Combine and compare. The combined credible premium is \(33{,}000+9{,}500=\$42{,}500\). The combined manual premium is \(30{,}000+10{,}000=\$40{,}000\). Therefore the dollar increase is \(42{,}500-40{,}000=\$2{,}500\), and the percentage increase is $$ \frac{2{,}500}{40{,}000}\times100\%=6.25\%. $$ The result is also checked by combining the components directly: \(0.50(36{,}000)+0.50(30{,}000)+0.25(8{,}000)+0.75(10{,}000)=42{,}500\).
Common mistake avoided. The two credibility values do not give equal coverage weights. In particular, averaging the two credibility values or the two coverage percentage changes would not produce the account’s percentage change. The correct comparison is the total credible premium against the total manual premium.
Practical Checks and Exam Traps
Before accepting an individual experience adjustment, check the inputs and the interpretation. A correct arithmetic calculation can still be wrong if the indications refer to different bases or if the resulting change is described relative to the wrong premium.
- Check the complement. If credibility is \(Z\), the manual weight is \(1-Z\), not another copy of \(Z\).
- Check the endpoints. At \(Z=0\), the result must equal the manual indication; at \(Z=1\), it must equal the experience indication.
- Check the direction. With \(0<Z<1\), the result must lie between the two indications. A result outside that range signals an error unless the inputs or method differ from the stated blend.
- Use a consistent basis. Do not blend an experience premium for one period or exposure base with a manual premium for another without first reconciling them.
- Aggregate dollars before percentages. For multiple coverages, sum the credible and manual premiums separately, then calculate the overall percentage change.
The resulting modification is an adjustment to the manual baseline, not a separate amount to add on top of the credible premium. When the manual premium is zero, the credible premium blend can still be calculated, but a percentage change relative to that manual premium is undefined because it would require division by zero.
Check Your Understanding
For each problem, apply the assigned credibility to the stated indications and compare the result with the manual baseline.
- A hypothetical account has a manual premium of $18,500, an experience indication of $22,200, and \(Z=0.25\). Find the credible premium and percentage change from manual. (Answer: $19,425; 5.00% increase.)
- A hypothetical account has 150 exposure units, a manual rate of $420 per unit, an experience rate indication of $390 per unit, and \(Z=0.60\). Find its credible rate, total premium, and percentage change from manual. (Answer: $402 per unit; $60,300; 4.29% decrease.)
- A hypothetical account’s manual premium is $50,000 and its experience indication is $62,000. If \(Z=0.35\), find the credible premium and credible modification factor. (Answer: $54,200; 1.084, or an 8.40% increase.)
- Coverage A has a manual premium of $12,000, an experience indication of $15,000, and \(Z_A=0.20\). Coverage B has a manual premium of $28,000, an experience indication of $25,000, and \(Z_B=0.40\). Find the combined credible premium and percentage change from combined manual. (Answer: $39,400; 1.50% decrease.)
- A hypothetical account has a manual premium of $40,000 and an experience indication of $52,000. What credibility is needed for a credible premium of $46,000, and what is the resulting percentage modification? (Answer: \(Z=0.50\); 15.00% increase.)