Ordering Multiple Contract Modifications

Why the Order of Claim Modifications Matters

A claim calculation may involve several transformations: first adjusting a loss for inflation, then applying a deductible and coinsurance, capping the policy payment at a limit, and finally calculating reinsurance. Each step acts on an amount produced by the preceding step. Changing the order can therefore change the payment, even when every individual operation is familiar.

The key is to write down what amount each contractual feature acts on. A deductible may apply to the inflated loss or to the original loss; a limit may cap the amount before or after coinsurance; and an excess-of-loss attachment point may apply to a post-quota-share retention or to the original covered amount. Unless the contract specifies otherwise, these are different calculations.

Definition: For a nonnegative amount $x$, define inflation by $I_g(x)=gx$, where $g>0$; coinsurance by $C_c(x)=cx$, where $0\leq c\leq1$; a deductible by $D_d(x)=(x-d)_+=\max(x-d,0)$; and a limit by $L_u(x)=\min(x,u)$, where $d\geq0$ and $u\geq0$. These definitions specify the amount returned by each operation. A contract may instead define its deductible, coinsurance, or limit on a different base amount.

For reinsurance, distinguish the ceded amount from the insurer’s retained amount. A quota share with cession proportion $q$ cedes $q y$ from a covered amount $y$, leaving $(1-q)y$. An excess-of-loss layer attaching at $a$, with upper endpoint $b$, cedes $R_{a,b}(y)=\min((y-a)_+,b-a)$. The insurer’s amount after that layer is $y-R_{a,b}(y)$. These formulas describe per-claim amounts; they do not by themselves determine an aggregate annual payment.

Which Operations Commute?

Two operations commute if applying them in either order gives the same result for every loss in the stated domain. Often the operations can be reordered only by changing a dollar threshold. The following identities make that distinction precise.

$$ C_c(I_g(x))=I_g(C_c(x))=cgx $$

Inflation and coinsurance are both scalings, so they commute. But a fixed deductible does not generally commute with either scaling. For $g>0$,

I_g(D_d(x))=g(x-d)_+=(gx-gd)_+=D_{gd}(I_g(x)).

This identity is a short derivation: multiplying the positive part of $x-d$ by a positive $g$ preserves whether it is zero, and when it is positive the result is $gx-gd$. Thus, applying a deductible $d$ before inflation is equivalent to inflating first and applying a deductible $gd$, not generally to using the same fixed dollar deductible in both orders.

Coinsurance and a deductible have a similar adjusted-threshold relationship:

C_c(D_d(x))=c(x-d)_+=(cx-cd)_+=D_{cd}(C_c(x)),\qquad c\geq0.

When $c>0$, one can also express a deductible applied after coinsurance as an equivalent deductible before coinsurance: $D_d(C_c(x))=C_c(D_{d/c}(x))$. If $c=0$, the post-coinsurance amount is always zero, so the expression involving $d/c$ is not defined and must not be used.

Deductibles and limits also have an adjusted-parameter identity:

L_u(D_d(x))=\min((x-d)_+,u)=D_d(L_{u+d}(x)).

The last equality follows by considering whether $x\leq d$, $d<x<d+u$, or $x\geq d+u$. In each case both expressions are respectively zero, $x-d$, or $u$. The equivalence requires changing the limit from $u$ to $u+d$ when the limit is applied before the deductible.

Other adjusted-parameter identities include $C_c(L_u(x))=L_{cu}(C_c(x))$. They do not mean that applying coinsurance and a limit with the same numerical $u$ in either order is equivalent. With unchanged parameters, many pairs of operations do not commute.

Worked Claim Chains

Worked Example: Inflation, Deductible, Coinsurance, and Limit

A hypothetical policy has claim severity $X$ with $P(X=100)=0.20$, $P(X=300)=0.50$, and $P(X=500)=0.30$, in dollars. The losses are stated at the original price level. For this calculation, apply inflation first with $g=1.20$, then a deductible of $100, then 80% coinsurance, and finally a $250 payment limit. Find the expected policy payment. Compare it with the result of putting the deductible after coinsurance, without changing its $100 amount. No interest or mortality assumption is relevant.

Method and first sequence. Apply the transformations in the specified order: $L_{250}(C_{0.8}(D_{100}(I_{1.20}(X))))$.

For $X=100$, inflation gives $1.20(100)=120$; the amount after the deductible is $120-100=20$; coinsurance gives $0.8(20)=16$; the limit leaves $16.

For $X=300$, inflation gives $360$; after the deductible the amount is $360-100=260$; coinsurance gives $0.8(260)=208$; the limit leaves $208.

For $X=500$, inflation gives $600$; after the deductible the amount is $500$; coinsurance gives $400$; the limit reduces the payment to $250.

The expected payment is $0.20(16)+0.50(208)+0.30(250)=3.20+104.00+75.00=\$182.20$.

Reordered calculation. If coinsurance precedes the same $100 deductible, the outcomes after inflation are $120, $360, and $600. Coinsurance produces $96, $288, and $480. The deductible then leaves $0, $188, and $380; applying the $250 limit gives $0, $188, and $250. The expected payment is $0.20(0)+0.50(188)+0.30(250)=\$169.00$.

Check and common mistake. The first sequence pays $13.20 more in expectation. The difference occurs because the fixed deductible is applied to different bases. The common mistake is to assume the sequence is immaterial because the same four features appear in both calculations. It is not. The adjusted-threshold identity would allow a deductible of $0.8(100)=\$80$ after coinsurance to match a $100 deductible before coinsurance; the second calculation deliberately keeps the deductible at $100.

Worked Example: Deductible and Limit in Either Order

A hypothetical claim severity has distribution $P(X=0)=0.10$, $P(X=100)=0.20$, $P(X=250)=0.40$, and $P(X=400)=0.30$, in dollars. A benefit has deductible $d=\$50$ and limit $u=\$200$. Calculate its expected payment when the deductible is applied before the limit, and when the limit is applied before the deductible using the same $200 limit. Then give the pre-deductible limit that makes the second sequence equivalent to the first. No interest or mortality assumption is relevant.

Deductible, then limit. The payments for losses $0, $100, $250, and $400 are respectively $(0-50)_+$ capped at 200, $(100-50)_+$ capped at 200, $(250-50)_+$ capped at 200, and $(400-50)_+$ capped at 200. Thus the payments are $0, $50, $200, and $200. The expected payment is $0.10(0)+0.20(50)+0.40(200)+0.30(200)=\$150$.

Limit, then deductible. Capping each loss at $200 first and then subtracting the $50 deductible, with a floor at zero, gives payments $0, $50, $150, and $150. The expected payment is $0.10(0)+0.20(50)+0.40(150)+0.30(150)=\$115$.

Equivalent adjusted limit. The identity $L_u(D_d(x))=D_d(L_{u+d}(x))$ says to use a pre-deductible limit of $200+50=\$250$. Capping the four losses at $250 and then applying the deductible gives $0, $50, $200, and $200, so the expected payment is again $150.

Check and common mistake. The first and adjusted sequences agree outcome by outcome. Using the same $200 limit before the deductible does not reproduce the original benefit: it reduces the maximum payment from $200 to $150. A limit’s position relative to a deductible is meaningful only together with the amount of the limit.

Worked Example: Quota Share and an Excess-of-Loss Layer

A hypothetical policy payment $Y$, after all direct policy modifications, takes values $50, $200, and $250 with probabilities 0.20, 0.50, and 0.30. Compare two explicitly different reinsurance arrangements. In Arrangement A, a 20% quota share is applied first; an excess-of-loss layer then attaches at $100 on the post-quota-share retained amount and covers the next $100. In Arrangement B, the same excess-of-loss layer is applied first to $Y$, and a 20% quota share is then applied to the amount remaining after that layer. Find the expected insurer retention under each arrangement. No interest or mortality assumption is relevant.

Arrangement A. For $Y=50$, the quota share cedes $0.20(50)=\$10$, leaving $40; the excess layer cedes zero, so the insurer retains $40. For $Y=200$, quota share cedes $40 and leaves $160; the layer cedes $\min((160-100)_+,100)=\$60$, leaving $100 for the insurer. For $Y=250$, quota share cedes $50 and leaves $200; the layer cedes $\min((200-100)_+,100)=\$100$, leaving $100.

The expected retention in Arrangement A is $0.20(40)+0.50(100)+0.30(100)=8+50+30=\$88$.

Arrangement B. For $Y=50$, the layer cedes zero, leaving $50; quota share then cedes $10, so the insurer retains $40. For $Y=200$, the layer cedes $\min((200-100)_+,100)=\$100$, leaving $100; quota share cedes $20, so the insurer retains $80. For $Y=250$, the layer cedes $100, leaving $150; quota share cedes $30, so the insurer retains $120.

The expected retention in Arrangement B is $0.20(40)+0.50(80)+0.30(120)=8+40+36=\$84$.

Check and common mistake. The expected retentions differ by $4.00. This is not a contradiction: the arrangements define the excess layer on different bases. In Arrangement A its attachment applies after quota share; in Arrangement B it applies to the full policy payment. The common mistake is to move a reinsurance layer across a quota share without changing or restating the treaty’s attachment basis.

A Reliable Calculation Method

For a claim chain involving several modifications, preserve the order in the contract and label every intermediate amount. For a discrete severity distribution, repeat the chain for each possible loss, then use its probability to calculate an expected payment. For a continuous severity, the same pointwise transformations define the covered payment; an expectation must then be calculated using the severity distribution.

Identify the starting loss.
State whether the amount is an original loss, an inflated loss, a policy payment, or a reinsurance retention. Do not treat these as interchangeable bases.

Write each operation as a function.
Record the order and parameters, for example $L_u(C_c(D_d(I_g(x))))$. This makes the input to each step explicit.

Apply the chain to each loss outcome.
Show the amount after each operation. Apply floors, caps, and attachment points to the amount specified by the contract.

Aggregate only after claim payments are determined.
For discrete outcomes $x_j$ with probabilities $p_j$, the expected payment is $\sum_j p_j y_j$, where $y_j$ is the result of the complete chain for outcome $x_j$.

The order matters in aggregate models as well as in an individual claim calculation. A frequency-severity model uses the severity after the specified coverage modifications as its claim amount; changing the claim-level chain changes that severity distribution and can therefore change aggregate results. The earlier material on coverage-modified severities and aggregate models provides the surrounding framework, but the claim transformation itself must still match the contract.

Common Traps and Contract Interpretation

Applying a threshold to the wrong amount. A $100 deductible on the inflated loss is different from a $100 deductible on the original loss. If translating between the two, use the inflation factor to adjust the threshold.
Keeping a limit fixed while moving it across another operation. A deductible followed by a $200 limit is equivalent to a $250 limit followed by a $50 deductible, not to the same $200 limit followed by that deductible.
Confusing a percentage with a dollar layer. Coinsurance scales an amount. A deductible and an excess-of-loss attachment are dollar thresholds. A scale can commute with another scale, but generally not with a fixed dollar threshold unless that threshold is adjusted.
Mixing up ceded and retained amounts. A reinsurance layer’s formula gives the cession. Subtract that cession from the amount entering the layer to obtain the retention; do not report the cession as the insurer’s remaining payment.
Assuming every rearrangement is a contractual equivalence. Algebra can show that two chains agree when parameters are adjusted. It does not authorize changing a contract’s attachment point, deductible, or limit.

Key takeaway: Treat each coverage feature as an operation on a specified amount. Scaling operations can commute with one another; scaling and thresholds, or multiple thresholds, generally require adjusted parameters to produce an equivalent chain. For reinsurance, state explicitly which post-coverage or pre-coverage amount defines each attachment and quota share.

Check Your Understanding

For each problem, preserve the stated order and use the given discrete probabilities where applicable.

A hypothetical loss has $P(X=100)=0.40$ and $P(X=300)=0.60$. Inflate by 10%, apply a $100 deductible, then 50% coinsurance, with no limit. Find the expected payment. (Answer: $71.00)
A benefit has deductible $40 and limit $160. Losses are $40, $120, and $260 with probabilities 0.25, 0.50, and 0.25. Find the expected payment for deductible then limit, and for limit then deductible using the same $160 limit. (Answer: $80.00; $70.00, respectively)
Let $X$ take values $80, $160, and $320 with probabilities 0.20, 0.50, and 0.30. Apply 75% coinsurance and then a $60 deductible. Find the expected payment and the deductible-before-coinsurance amount that gives the same payment for every loss. (Answer: $84.00; $80)
A hypothetical covered payment $Y$ is $80, $180, or $280 with probabilities 0.20, 0.50, or 0.30. Cede a 25% quota share first, then apply an excess-of-loss layer attaching at $100 to the post-quota-share retention and covering the next $100. Find the expected insurer retention. (Answer: $95.00)
A $250 original loss is inflated by 20%, then subject to a $90 deductible and 80% coinsurance, with no binding limit. What is the payment? Verify it by applying an equivalent deductible before inflation. (Answer: $168.00; the equivalent pre-inflation deductible is $75)

What You'll Learn