Common Errors in Coverage Modification Calculations

Why Coverage Calculations Go Wrong

Coverage-modification errors often begin before any arithmetic: a ground-up loss is mistaken for an insurer payment, or a threshold is interpreted as strict when the contract includes equality. These errors can change the payment distribution, its atoms, and its moments. The reliable starting point is to translate the wording into a payment function, then evaluate that function at each possible loss.

Throughout this tutorial, $X$ denotes a nonnegative ground-up loss, and $Y^L$ denotes the insurer’s payment per loss. A deductible $d$ and payment limit $u$ are amounts in the same monetary units as $X$. Unless the contract states otherwise, examples use hypothetical policies and losses.

Definition: Under an ordinary deductible $d$ and a maximum insurer payment $u>0$, the payment per loss is $$ Y^L=\min\big((X-d)^+,u\big), \qquad (z)^+=\max(z,0). $$ Thus, losses at or below $d$ produce no payment; losses above $d$ produce payment up to $u$; and a loss of at least $d+u$ produces the maximum payment $u$. The ground-up loss $X$ and payment $Y^L$ are different random variables.

Atoms at Zero and at the Payment Limit

A modified payment need not have the same distributional form as the ground-up loss. Even when $X$ has no probability at any individual value, a deductible can concentrate probability at zero and a limit can concentrate probability at $u$. For the payment function above,

P(Y^L=0)=P(X\le d), \qquad P(Y^L=u)=P(X\ge d+u).

The equality signs matter. A loss exactly equal to the deductible has zero payment, while a loss exactly equal to the amount needed to exhaust the limit has payment $u$. For $0<y<u$, the event that the payment is at most $y$ is $X\le d+y$. Therefore,

F_{Y^L}(y)=F_X(d+y), \qquad 0\le y<u,

and $F_{Y^L}(y)=1$ for $y\ge u$. This CDF includes the mass at zero; the cap atom is the remaining probability that has not accumulated below $u$.

Derivation: Since $X\ge0$, the payment is zero precisely when $(X-d)^+=0$, which is equivalent to $X\le d$. The payment reaches $u$ precisely when $(X-d)^+\ge u$, or $X\ge d+u$. For $0\le y<u$, the payment is at most $y$ exactly when $X\le d+y$: losses at or below $d$ pay zero, and losses above $d$ pay $X-d$ until the cap is reached. Hence $F_{Y^L}(y)=F_X(d+y)$ on that range.

Worked Example: Separating Ground-Up Loss From Insurer Payment

A hypothetical policy has deductible $d=\$500$ and maximum insurer payment $u=\$1{,}000$. The ground-up loss distribution is: $X=\$0$ with probability $0.20$, $X=\$500$ with probability $0.30$, $X=\$1{,}200$ with probability $0.30$, and $X=\$3{,}000$ with probability $0.20$. Find the payment distribution, its mean, and the probability of a maximum payment.

Method. Apply $Y^L=\min((X-500)^+,1{,}000)$ separately to each ground-up loss.

If $X=0$, then $Y^L=\min(0,1{,}000)=\$0$, contributing probability $0.20$ at zero. If $X=500$, then $Y^L=\min(0,1{,}000)=\$0$, contributing another $0.30$. Thus $P(Y^L=0)=0.20+0.30=0.50$.

If $X=1{,}200$, then $Y^L=\min(1{,}200-500,1{,}000)=\$700$, with probability $0.30$. If $X=3{,}000$, then $Y^L=\min(2{,}500,1{,}000)=\$1{,}000$, with probability $0.20$. The payment distribution is therefore $\$0$ with probability $0.50$, $\$700$ with probability $0.30$, and $\$1{,}000$ with probability $0.20$.

The expected payment is $$ E[Y^L]=0(0.50)+700(0.30)+1{,}000(0.20)=210+200=\$410. $$ The maximum payment occurs when $X\ge d+u=\$1{,}500$. Only the $\$3{,}000$ outcome meets this condition, so $P(Y^L=\$1{,}000)=0.20$.

Check. All payment outcomes lie between $\$0$ and the $\$1{,}000$ limit, and their probabilities sum to $0.50+0.30+0.20=1.00$. The common mistake this method avoids is subtracting the deductible from every loss or treating $X$ itself as the insurer’s payment.

Coinsurance Must Be Placed at the Contracted Stage

Coinsurance is a percentage, but that does not by itself specify what amount the percentage applies to. A contract may apply coinsurance to the covered amount after the deductible and before a stated coverage-layer cap. Another contract may instead state a maximum insurer payment after coinsurance. These are different payment functions; the phrase “coinsurance and a limit” is not enough to determine the order.

For a contract that applies coinsurance rate $c$, $0\le c\le1$, to the amount remaining after the deductible and then caps that pre-coinsurance covered amount at $u$, the function is

Y^L=c\min\big((X-d)^+,u\big).

Here the maximum payment is $cu$. By contrast, if $u$ is explicitly the maximum insurer payment after coinsurance, the function is

Y^L=\min\big(c(X-d)^+,u\big).

The formulas agree below the point where either cap matters, but may differ above it. Always identify what the limit caps before calculating.

Worked Example: Coinsurance Before or After the Stated Cap

A hypothetical contract has deductible $d=\$400$, coinsurance rate $c=0.80$, and a $\$1{,}500$ cap on the covered amount before coinsurance. Ground-up loss $X$ equals $\$400$, $\$1{,}400$, or $\$3{,}400$, with respective probabilities $0.25$, $0.50$, and $0.25$. Find the expected insurer payment.

Method. Because the wording places the cap before coinsurance, use $Y^L=0.80\min((X-400)^+,1{,}500)$.

For $X=\$400$, the amount after the deductible is $\max(400-400,0)=\$0$, so the payment is $0.80\min(0,1{,}500)=\$0$. For $X=\$1{,}400$, the amount after the deductible is $\$1{,}000$, below the cap, so the payment is $0.80(1{,}000)=\$800$. For $X=\$3{,}400$, the amount after the deductible is $\$3{,}000$, which is capped at $\$1{,}500$, so the payment is $0.80(1{,}500)=\$1{,}200$.

Weighting by the stated probabilities gives $$ E[Y^L]=0(0.25)+800(0.50)+1{,}200(0.25)=400+300=\$700. $$

Check and trap. The maximum payment is $0.80(1{,}500)=\$1{,}200$, as the contract’s cap is pre-coinsurance. A mistaken calculation, $\min(0.80(X-400)^+,1{,}500)$, gives payments $\$0,\$800,\$1{,}500$ and mean $0+400+375=\$775$. That calculation incorrectly treats $\$1{,}500$ as an after-coinsurance insurer-payment limit.

Thresholds and Layer Boundaries

A layer attaching at $a$ and ending at $b$, where $0\le a<b$, pays the part of the loss in that layer:

Y^L=\min\big((X-a)^+,b-a\big).

At $X=a$, the payment is zero. At $X=b$, the payment has reached the layer maximum $b-a$, just as it does for every $X>b$. Accordingly, the payment has a zero atom of size $P(X\le a)$ and a maximum-payment atom of size $P(X\ge b)$. The amount of the payment at either boundary follows from the function, not from an informal reading of “above,” “up to,” or “excess of.”

Worked Example: Counting a Loss Exactly at the Layer Limit

A hypothetical excess layer attaches at $\$1{,}000$ and ends at $\$3{,}000$. Ground-up losses are $\$1{,}000$, $\$2{,}500$, $\$3{,}000$, and $\$4{,}500$, with probabilities $0.20$, $0.30$, $0.20$, and $0.30$, respectively. Find the payment distribution, expected payment, and probability of a full layer payment.

Method. The layer width is $3{,}000-1{,}000=\$2{,}000$, so use $Y^L=\min((X-1{,}000)^+,2{,}000)$.

At $X=\$1{,}000$, $Y^L=\$0$. At $X=\$2{,}500$, $Y^L=\min(1{,}500,2{,}000)=\$1{,}500$. At $X=\$3{,}000$, $Y^L=\min(2{,}000,2{,}000)=\$2{,}000$. At $X=\$4{,}500$, the payment remains $\$2{,}000$, because it cannot exceed the layer width. Thus the payment probabilities are $0.20$ at zero, $0.30$ at $\$1{,}500$, and $0.50=0.20+0.30$ at $\$2{,}000$.

The expected payment is $$ E[Y^L]=0(0.20)+1{,}500(0.30)+2{,}000(0.20)+2{,}000(0.30) =450+400+600=\$1{,}450. $$ A full layer payment occurs for $X\ge \$3{,}000$, including the outcome exactly at $\$3{,}000$. Therefore its probability is $0.20+0.30=0.50$.

Check and trap. The expected payment is between zero and the layer maximum of $\$2{,}000$, and the payment probabilities total one. Counting only losses strictly greater than $\$3{,}000$ would give $0.30$, incorrectly omitting the $0.20$ probability at the boundary.

A Practical Error-Checking Routine

Use the contract wording to settle the payment function before substituting loss values. Then check its range and boundaries. These checks are especially useful when the loss distribution is discrete or has mass at a threshold.

Name the variable being calculated.
Keep ground-up loss $X$ separate from insurer payment $Y^L$. If the question asks for a payment probability or expectation, do not substitute a moment of $X$ without transforming the loss.

Translate every contract term.
Write the deductible, layer, coinsurance rate, and limit in a payment function. State whether a limit applies before or after coinsurance and whether it is a maximum insurer payment.

Test equality at each threshold.
Evaluate the function at the deductible or attachment point and at the point where the cap is exhausted. These values determine whether boundary probability belongs in the zero or maximum-payment atom.

Check the result against the contract.
Payments must be nonnegative and cannot exceed the applicable maximum. For a discrete distribution, verify that all probabilities, including atoms, sum to one before taking an expectation.

A further distinction is between an expectation per loss and an expectation conditional on a positive payment. If the latter is requested and $P(Y^L>0)>0$, then

E[Y^L\mid Y^L>0]=\frac{E[Y^L]}{P(Y^L>0)}.

This follows because $Y^L=0$ whenever $Y^L=0$, so $E[Y^L]=E[Y^L\mathbf{1}_{\{Y^L>0\}}]=P(Y^L>0)E[Y^L\mid Y^L>0]$. Dividing by the probability of a positive payment is appropriate only for that conditional question; it is not the mean payment per ground-up loss.

Common Errors to Avoid

Using ground-up losses as payments. Apply the payment function to each loss first. A deductible does not reduce every loss by a fixed amount; losses at or below it have zero payment.
Dropping the zero atom. Under an ordinary deductible, $P(Y^L=0)=P(X\le d)$, not just the probability that $X<d$.
Dropping the cap atom. Losses at the cap-exhaustion threshold and above all produce the maximum payment. For a standard deductible and limit, that event is $X\ge d+u$.
Moving coinsurance across a cap without checking the wording. A pre-coinsurance covered-amount limit $u$ gives maximum payment $cu$; a post-coinsurance payment limit $u$ gives maximum payment $u$.
Treating a layer boundary as a strict inequality. The payment function determines the boundary value. In the standard layer function, $X=b$ already produces the full layer payment.
Reporting a conditional mean as an unconditional mean. Divide by $P(Y^L>0)$ only when the question explicitly conditions on a positive payment.

Key takeaway: A coverage calculation is only as reliable as its payment function. Identify what each threshold applies to, preserve equality at boundary values, and account for probability atoms at zero and at the cap.

Check Your Understanding

For each hypothetical contract, write the payment for each relevant loss before calculating the requested quantity.

A policy has deductible $d=\$250$, maximum payment $u=\$750$, and ground-up losses $\$200,\$250,\$600,\$1{,}500$ with probabilities $0.20,0.10,0.40,0.30$. Find $E[Y^L]$, $P(Y^L=0)$, and $P(Y^L=\$750)$. (Answer: $E[Y^L]=\$365$, $P(Y^L=0)=0.3000$, and $P(Y^L=\$750)=0.3000$.)
A layer attaches at $\$500$ and ends at $\$1{,}800$. Losses $\$500,\$1{,}000,\$1{,}800,\$2{,}200$ are equally likely. Find the expected layer payment and the probability of a full layer payment. (Answer: $E[Y^L]=\$775$ and $P(Y^L=\$1{,}300)=0.5000$.)
A contract has $d=\$300$, coinsurance $c=0.70$, and a maximum insurer payment of $\$900$ after coinsurance: $Y^L=\min(0.70(X-300)^+,900)$. Losses are $\$300,\$1{,}000,\$2{,}000$ with probabilities $0.20,0.50,0.30$. Find the expected payment. (Answer: $E[Y^L]=\$515$.)
A policy applies coinsurance $c=0.60$ to the covered amount after a $\$500$ deductible, with a pre-coinsurance cap $u=\$1{,}000$. Losses $\$400,\$800,\$1{,}500,\$2{,}500$ have probabilities $0.10,0.20,0.40,0.30$. Find the expected insurer payment and the maximum possible payment. (Answer: $E[Y^L]=\$456$ and the maximum payment is $\$600$.)
A layer attaches at $\$1{,}000$ and ends at $\$3{,}000$. Losses $\$400,\$1{,}000,\$2{,}000,\$4{,}000$ have probabilities $0.20,0.20,0.30,0.30$. Find $P(Y^L=0)$, $P(Y^L=\$2{,}000)$, and $E[Y^L]$. (Answer: $P(Y^L=0)=0.4000$, $P(Y^L=\$2{,}000)=0.3000$, and $E[Y^L]=\$900$.)

What You'll Learn