Integrated Experience Modification Calculation Workshop

From Experience Data to a Final Modification

A complete experience modification calculation is a sequence of linked steps. First identify the experience years and their exposures. Then apply the plan’s split point to each claim, combine years as specified, and compare the resulting actual and expected primary and excess losses. Credibility and ballast determine how strongly the observed experience affects the result; plan bounds may then limit the factor used to adjust premium.

The exact formula depends on the rating plan. The method below uses one clearly stated hypothetical plan structure. It is not a universal formula: in an actual calculation, use the plan’s definitions of eligible losses, expected losses, year weights, credibility, ballast, and bounds.

Definition: For each claim with loss $X_j$ and split point $s$, the primary loss is $P_j=\min(X_j,s)$, and the excess loss is $Q_j=\max(X_j-s,0)$. Thus $X_j=P_j+Q_j$. For experience year $t$, let $A_t$ and $B_t$ be the sums of primary and excess losses after the plan’s stated adjustments; let $E_{P,t}$ and $E_{Q,t}$ be the corresponding expected losses. A year weight $w_t$ multiplies that year’s actual and expected components when the plan specifies weighted experience.

In the hypothetical plan used here, each component has its own credibility weight: $Z_P$ for primary losses and $Z_Q$ for excess losses. The plan combines credible actual losses with the complementary share of expected losses. Ballast $K$ is added to both the resulting numerator and the expected-loss denominator. The raw modification is

M_{\mathrm{raw}}= \frac{Z_P A+(1-Z_P)E_P+Z_Q B+(1-Z_Q)E_Q+K} {E_P+E_Q+K},

where $A=\sum_t w_t A_t$, $B=\sum_t w_t B_t$, $E_P=\sum_t w_t E_{P,t}$, and $E_Q=\sum_t w_t E_{Q,t}$. The final modification is $M=\min(M_{\max},\max(M_{\min},M_{\mathrm{raw}}))$. In other plan designs, the credibility or ballast structure may differ; do not substitute this formula without checking the specified plan.

Why Credibility and Ballast Have This Form

The credibility blend for primary losses is $Z_P A+(1-Z_P)E_P$. If $Z_P=1$, the plan uses all actual primary losses; if $Z_P=0$, it uses expected primary losses. The excess blend has the same interpretation using $Z_Q$. The formula therefore replaces each expected component by a weighted blend of observed and expected experience.

A useful equivalent expression follows by subtracting the expected-loss denominator from the numerator. Since $E=E_P+E_Q$,

\begin{aligned} M_{\mathrm{raw}} &=\frac{E_P+E_Q+ Z_P(A-E_P)+Z_Q(B-E_Q)+K}{E_P+E_Q+K}\\ &=1+\frac{Z_P(A-E_P)+Z_Q(B-E_Q)}{E_P+E_Q+K}. \end{aligned}

This identity explains the role of ballast. It does not change the direction of the credible experience deviation: the numerator of the adjustment is still $Z_P(A-E_P)+Z_Q(B-E_Q)$. But increasing $K$ increases the denominator, reducing the size of the adjustment toward 1.000. The result is still subject to the plan’s stated minimum and maximum.

A Reliable Calculation Sequence

Check the plan inputs.
List eligible years, exposures, split point, loss adjustments, expected primary and excess losses, year weights, credibility weights, ballast, bounds, and the premium base.

Split each claim.
Apply the per-claim split point before summing claims. Do not cap an annual total unless the plan separately specifies an aggregate cap.

Weight and sum years.
Apply each year’s stated weight consistently to its actual and expected components, then sum the components separately.

Calculate and bound the factor.
Apply the specified credibility and ballast formula, then apply the minimum and maximum in the plan’s required sequence.

Calculate the premium effect.
Multiply the applicable standard premium by the final, not raw, modification. Retain precision until the plan’s stated rounding step.

Worked Example: Two Years, a Per-Claim Split, and Separate Credibilities

A hypothetical plan uses a split point of $25,000 per claim and no year weights other than 1.00. Year 1 has three eligible claims of $10,000, $30,000, and $50,000; expected primary and excess losses are $42,000 and $18,000. Year 2 has claims of $15,000, $28,000, and $40,000; expected primary and excess losses are $67,000 and $28,000. The plan specifies $Z_P=0.70$, $Z_Q=0.25$, $K=\$20,000$, and bounds of 0.800 to 1.200. The standard premium is $240,000. Calculate the final modification and adjusted premium.

Method and claim split. Apply the $25,000 split to each claim. In Year 1, primary losses are $10,000, $25,000, and $25,000, totaling $A_1=\$60,000$. Excess losses are $0, $5,000, and $25,000, totaling $B_1=\$30,000$.

In Year 2, primary losses are $15,000, $25,000, and $25,000, totaling $A_2=\$65,000$. Excess losses are $0, $3,000, and $15,000, totaling $B_2=\$18,000$. The expected components are given by the plan, so they are not inferred by splitting these observed claims.

Combine years. With both year weights equal to 1.00, $A=\$60,000+\$65,000=\$125,000$, $B=\$30,000+\$18,000=\$48,000$, $E_P=\$42,000+\$67,000=\$109,000$, and $E_Q=\$18,000+\$28,000=\$46,000$. Total expected losses are $\$109,000+\$46,000=\$155,000$.

Apply credibility and ballast. The numerator before ballast is $0.70(\$125,000)+0.30(\$109,000)+0.25(\$48,000)+0.75(\$46,000)$. These terms are $87,500, $32,700, $12,000, and $34,500, summing to $166,700. Adding $20,000 ballast gives a numerator of $186,700. The denominator is $\$155,000+\$20,000=\$175,000$. Thus $M_{\mathrm{raw}}=\$186,700/\$175,000=1.066857\ldots$.

Apply bounds and calculate premium. The raw factor is between 0.800 and 1.200, so the final modification is unchanged. Using the unrounded factor, adjusted premium is $\$240,000(1.066857\ldots)=\$256,045.71$, rounded to the nearest cent. The premium increases by $\$256,045.71-\$240,000=\$16,045.71$. As a check, the raw factor is above 1.000 because both actual components exceed their expected counterparts.

Common mistake avoided: The split is per claim. Capping the Year 1 total loss of $90,000 at $25,000 would incorrectly discard eligible experience and would not produce the stated primary and excess totals.

Year Weights and Modification Bounds

A year weight is a plan instruction about how that year contributes to the calculation; it is not automatically a credibility weight. Keep the two roles distinct. For example, if a year has weight 0.5, multiply its actual primary, actual excess, expected primary, and expected excess components by 0.5 before summing, unless the plan explicitly specifies a different treatment. Apply $Z_P$ and $Z_Q$ to the resulting totals only as the plan formula directs.

Worked Example: Weighting Years Before Applying the Bounds

A hypothetical plan supplies the following loss components, already split per claim. Year 1 has weight 0.50, actual primary loss $20,000, expected primary loss $30,000, actual excess loss $10,000, and expected excess loss $10,000. Year 2 has weight 1.00, with corresponding components $50,000, $40,000, $20,000, and $20,000. Let $Z_P=0.60$, $Z_Q=0.40$, $K=\$25,000$, and bounds be 0.900 and 1.050. The standard premium is $180,000. Find the final modification and adjusted premium.

Weight each component. Weighted actual primary is $0.50(\$20,000)+1.00(\$50,000)=\$60,000$. Weighted expected primary is $0.50(\$30,000)+1.00(\$40,000)=\$55,000$. Weighted actual excess is $0.50(\$10,000)+1.00(\$20,000)=\$25,000$, and weighted expected excess is $0.50(\$10,000)+1.00(\$20,000)=\$25,000$.

Calculate the raw factor. The numerator before ballast is $0.60(\$60,000)+0.40(\$55,000)+0.40(\$25,000)+0.60(\$25,000)$, or $\$36,000+\$22,000+\$10,000+\$15,000=\$83,000$. After adding ballast, it is $108,000. The denominator is $\$55,000+\$25,000+\$25,000=\$105,000$. Therefore $M_{\mathrm{raw}}=\$108,000/\$105,000=1.028571\ldots$.

Apply bounds and calculate premium. Since 1.028571 is between 0.900 and 1.050, the final modification is 1.028571…. The adjusted premium is $\$180,000(1.028571\ldots)=\$185,142.86$, rounded to the nearest cent. The final factor is within the stated bounds, and the premium is above the base premium, consistent with the raw factor being greater than 1.000.

Common mistake avoided: Weighting only actual losses would leave the expected benchmark on a different basis. Here the year weights apply to both actual and expected components, as the hypothetical plan specifies.

Floors, Premium Effects, and Common Traps

A bound changes the factor used for premium, not the underlying raw calculation. Calculate $M_{\mathrm{raw}}$ first, compare it with both limits, and then use the bounded $M$. If $M_{\mathrm{raw}}<M_{\min}$, the factor is raised to the minimum; if $M_{\mathrm{raw}}>M_{\max}$, it is reduced to the maximum. When the applicable premium is $P_0$, the adjusted premium is $P_0M$, and the dollar change is $P_0(M-1)$.

Worked Example: A Minimum Modification Controls the Premium

A hypothetical plan has one experience period, a per-claim split point of $10,000, $Z_P=0.80$, $Z_Q=0.50$, ballast of $10,000, and modification bounds of 0.950 and 1.100. Eligible claims are $4,000, $12,000, and $30,000. Expected primary losses are $35,000 and expected excess losses are $15,000. The standard premium is $75,000. Find the plan modification and premium effect.

Split each claim. Primary losses are $4,000, $10,000, and $10,000, so $A=\$24,000$. Excess losses are $0, $2,000, and $20,000, so $B=\$22,000$. Expected total losses are $\$35,000+\$15,000=\$50,000$.

Calculate the raw factor. The credible primary blend is $0.80(\$24,000)+0.20(\$35,000)=\$19,200+\$7,000=\$26,200$. The credible excess blend is $0.50(\$22,000)+0.50(\$15,000)=\$11,000+\$7,500=\$18,500$. After adding $10,000 ballast, the numerator is $54,700. The denominator is $\$50,000+\$10,000=\$60,000$. Thus $M_{\mathrm{raw}}=\$54,700/\$60,000=0.911666\ldots$.

Apply the floor and find the premium. The raw factor is below the 0.950 minimum, so the final modification is 0.950. The adjusted premium is $\$75,000(0.950)=\$71,250.00$, a decrease of $\$3,750.00$ from the standard premium. This result is above the raw-factor premium of $\$68,375$, because the floor limits the credit.

Common mistake avoided: Do not use the raw factor to calculate premium after discovering that a bound applies. The plan’s final factor is the bounded 0.950, not 0.911666….

Several errors recur in integrated cases. Treating a split point as an annual cap changes the loss allocation; applying year weights to actual but not expected losses creates inconsistent comparisons; and adding ballast to the numerator without also adding it to the denominator does not implement the stated ballasted formula. A quick check is to verify that primary plus excess equals each claim’s eligible loss, weighted actual and expected totals use the same year weights, and the final factor respects both bounds.

Key takeaway: Keep claim splitting, year weighting, credibility, ballast, bounds, and premium conversion as separate documented steps. Each step uses a different plan input, and checking the intermediate totals makes the final modification auditable.

Check Your Understanding

Use the stated plan formula and carry out each step in order. Unless noted, amounts are in dollars and split loss components are already given.

A plan has $A=\$80,000$, $E_P=\$100,000$, $B=\$20,000$, $E_Q=\$20,000$, $Z_P=0.50$, $Z_Q=0.25$, and $K=\$10,000$, with bounds 0.800 and 1.200. What is the final modification? (Answer: 0.9231, within bounds)
At a per-claim split point of $15,000, three claims are $8,000, $17,000, and $35,000. Find total primary and excess losses. (Answer: primary $38,000; excess $22,000)
Year 1 has weight 0.50 and components $A_1=\$30,000$, $E_{P,1}=\$40,000$, $B_1=\$10,000$, $E_{Q,1}=\$10,000$. Year 2 has weight 1.50 and components $A_2=\$50,000$, $E_{P,2}=\$40,000$, $B_2=\$20,000$, $E_{Q,2}=\$20,000$. With $Z_P=0.50$, $Z_Q=0.20$, and $K=\$10,000$, find the raw modification. (Answer: 1.0400)
A plan has $A=\$40,000$, $E_P=\$60,000$, $B=\$30,000$, $E_Q=\$20,000$, $Z_P=0.75$, $Z_Q=0.50$, and $K=\$20,000$. Its minimum modification is 0.950. What final modification applies? (Answer: 0.950)
A standard premium of $120,000 is subject to a final modification of 1.080. Find the adjusted premium and dollar change. (Answer: adjusted premium $129,600.00; increase $9,600.00)

What You'll Learn