Calculating Exact Poisson Claim Probabilities

From a Claim Rate to an Exact Count Probability

A claim rate describes an expected number of claims per unit of exposure; it is not itself the probability of a particular claim count. To calculate the probability of zero, one, or another specified number of claims, first convert the rate to the expected count over the exposure being studied. Then apply the Poisson probability formula.

For a hypothetical portfolio, the exposure unit might be a policy-year, vehicle-year, member-month, or thousand operating hours. The unit must match the one used to state the rate. A common exam trap is to use the rate as the Poisson mean without accounting for how much exposure is present.

Definition: Suppose the number of claims $N$ over a specified period or portfolio follows a Poisson model with mean $\mu$, where $\mu>0$. For an integer $k\geq 0$, $$ P(N=k)=\frac{e^{-\mu}\mu^k}{k!}. $$ If the claim rate is $r$ claims per exposure unit and the total exposure is $E$ units, then $\mu=rE$. The possible claim counts are $0,1,2,\ldots$.

The model assumes that the rate is constant over the exposure being considered and that claim occurrences behave as a Poisson count process. In particular, counts over disjoint exposure periods or groups are independent when the model specifies independent increments or independent groups. If the rate changes over time or across risks, calculate the appropriate mean for each part rather than applying one rate indiscriminately. The formula gives a probability for the count, not for claim amounts or the total dollar cost of claims.

Why the Poisson Formula Has This Form

A Poisson count can be justified as a limit of many independent opportunities for a rare claim. Suppose there are $n$ independent opportunities, each with claim probability $\mu/n$, so the expected number of claims is $\mu$. The count then has a binomial distribution. For a fixed nonnegative integer $k$, its probability is

\begin{aligned} P(N=k) &=\binom{n}{k}\left(\frac{\mu}{n}\right)^k \left(1-\frac{\mu}{n}\right)^{n-k}\\ &\longrightarrow \frac{\mu^k}{k!}e^{-\mu} \qquad\text{as }n\longrightarrow\infty. \end{aligned}

The first factor involving the binomial coefficient and $(\mu/n)^k$ approaches $\mu^k/k!$. The remaining factor approaches $e^{-\mu}$, since $(1-\mu/n)^n$ approaches $e^{-\mu}$, while $(1-\mu/n)^{-k}$ approaches 1 for fixed $k$. This gives the Poisson probability formula. The result explains why the formula is useful for counts of relatively uncommon events across many opportunities; the Poisson model is still an assumption about the count process, not a guarantee that any rare event must follow it.

Calculating Zero, One, and Specified Counts

For zero claims, the factorial and power terms simplify to

P(N=0)=e^{-\mu}.

For one claim, $1!=1$, so $P(N=1)=\mu e^{-\mu}$. For a specified count $k$, use $e^{-\mu}\mu^k/k!$ with that exact value of $k$. These are point probabilities: “exactly three” means $N=3$, not $N\geq 3$ or $N\leq 3$.

Sometimes a question asks for a set of possible counts rather than one count. For example, “at least one claim” is the complement of zero claims:

P(N\geq 1)=1-P(N=0)=1-e^{-\mu}.

For “at most one claim,” add the probabilities of zero and one:

P(N\leq 1)=P(N=0)+P(N=1)=e^{-\mu}(1+\mu).

The method is to translate the wording into counts before calculating. This prevents a frequent error: treating “at least one” as the probability of exactly one.

Worked Example: Zero, One, or Three Claims in a Policy Portfolio

A hypothetical insurer models claims for a group of policies using a Poisson process with a constant rate of $0.4$ claims per policy-year. Five policies are each observed for one year. Find the probabilities of zero claims, exactly one claim, and exactly three claims during the observation period. Give probabilities to four decimal places.

Method. Scale the rate by the total policy-year exposure, then evaluate the Poisson formula separately at each requested count.

The total exposure is $5\times 1=5$ policy-years. Therefore the Poisson mean for the portfolio is

\mu=0.4\times5=2.

For zero claims,

P(N=0)=\frac{e^{-2}2^0}{0!}=e^{-2}=0.135335\ldots\approx 0.1353.

For exactly one claim,

P(N=1)=\frac{e^{-2}2^1}{1!}=2e^{-2} =0.270671\ldots\approx 0.2707.

For exactly three claims, first calculate $3!=6$ and $2^3=8$:

P(N=3)=\frac{e^{-2}2^3}{3!} =\frac{8}{6}e^{-2} =0.180447\ldots\approx 0.1804.

The answers are $0.1353$, $0.2707$, and $0.1804$, respectively. As a check, the three disjoint events have combined probability $0.135335+0.270671+0.180447=0.586453$, which is less than 1; the remaining probability belongs to other counts. Common mistake avoided: using $0.4$ as the mean would ignore the five policy-years. The mean is $2$, not the per-policy-year rate.

Complements and Multiple Exposure Groups

A complement is often the most direct way to calculate a probability involving “at least.” For example, the event of at least one claim excludes only the zero-claim outcome. For “at least two,” subtract both zero and one from 1:

P(N\geq 2)=1-P(N=0)-P(N=1) =1-e^{-\mu}(1+\mu).

When a question distinguishes exposure groups, preserve that distinction if the event depends on which group has claims. If independent groups have Poisson means $\mu_1$ and $\mu_2$, their counts $N_1,N_2$ are independent. Thus a joint event has probability $P(N_1=a,N_2=b)=P(N_1=a)P(N_2=b)$. If only the total count matters, the independent counts can also be treated as a Poisson count with mean $\mu_1+\mu_2$. These facts require the stated independence and Poisson assumptions for the groups.

Worked Example: At Least One Claim in a Vehicle Portfolio

A hypothetical fleet has 10 vehicles, each covered for 2.5 years. Assume a constant Poisson claim rate of $0.06$ claims per vehicle-year. Find (a) the probability of no claims, (b) the probability of exactly two claims, and (c) the probability of at least one claim over the full observation period. Give probabilities to four decimal places.

Method. Convert the fleet to vehicle-year exposure, use the resulting mean for exact counts, and use a complement for the “at least one” event.

The total exposure is $10\times2.5=25$ vehicle-years. The Poisson mean is

\mu=0.06\times25=1.5.

For no claims,

P(N=0)=e^{-1.5}=0.223130\ldots\approx0.2231.

For exactly two claims, $2!=2$ and $1.5^2=2.25$, so

P(N=2)=\frac{e^{-1.5}(1.5)^2}{2!} =1.125e^{-1.5} =0.251021\ldots\approx0.2510.

At least one claim is the complement of no claims:

P(N\geq1)=1-P(N=0)=1-0.223130\ldots =0.776870\ldots\approx0.7769.

As a check, “at least one” and “zero” are complementary, and $0.7769+0.2231=1.0000$ after rounding. Common mistake avoided: the number of vehicles is not the exposure by itself; each vehicle contributes 2.5 vehicle-years, so the total exposure is 25 vehicle-years.

Worked Example: Exactly Two Claims Across Two Operating Groups

A hypothetical equipment program models claims with a rate of $0.8$ claims per 1,000 operating hours. Group A contributes 1,500 hours and Group B contributes 1,000 hours. Assume independent Poisson counts for the two groups. Find (a) the probability of exactly two claims across both groups and (b) the probability of exactly two claims in total with at least one claim in each group. Give probabilities to four decimal places.

Method. Find each group’s mean using exposure in thousands of hours. For the total count, add the independent means. For the event requiring claims in both groups, enumerate the group-count pairs that sum to two.

Group A has $1.5$ thousand hours and Group B has $1.0$ thousand hours. Their means are

\mu_A=0.8(1.5)=1.2,\qquad \mu_B=0.8(1.0)=0.8.

The combined mean is $1.2+0.8=2.0$. Therefore,

P(N_A+N_B=2) =\frac{e^{-2}(2)^2}{2!} =2e^{-2} =0.270671\ldots\approx0.2707.

For exactly two claims in total with at least one in each group, the only possible count pair is $(N_A,N_B)=(1,1)$. By independence,

\begin{aligned} P(N_A=1,N_B=1) &=(1.2e^{-1.2})(0.8e^{-0.8})\\ &=(1.2)(0.8)e^{-2}\\ &=0.96e^{-2}=0.129886\ldots\approx0.1299. \end{aligned}

The second answer is less than the first, as required: it describes a subset of the event of two total claims. Common mistake avoided: using 2.5 hours rather than 2.5 thousand hours would apply the stated rate to the wrong exposure unit. Converting each group’s hours to thousands of hours gives the means $1.2$ and $0.8$.

Calculation Checks and Exam Traps

Before finalizing a result, check the model parameter, the event, and the arithmetic. A short checklist helps catch errors that can otherwise produce plausible-looking probabilities.

Match units. If the rate is per 1,000 hours, express exposure in thousands of hours or convert the rate to a per-hour rate. Do not multiply incompatible units.
Scale once. The Poisson mean is rate times total exposure. Do not multiply by the number of risks again if it is already included in total exposure.
Read the event precisely. Exactly $k$ means one probability, $P(N=k)$; at least or at most requires a sum or complement.
Keep the factorial. For $k\geq2$, omitting $k!$ changes the probability. For example, the factor for exactly two claims is $\mu^2/2$, not $\mu^2$.
Use group detail only when needed. If only the total count matters, combine independent Poisson means. If the event specifies counts by group, calculate the relevant joint probabilities instead.
Check the probability range. Every probability must be between 0 and 1. Complementary events should add to 1 before rounding.

This calculation skill is aligned with the SOA Exam FAM (Fundamentals of Actuarial Mathematics) syllabus. The central discipline is to translate exposure and wording into the correct Poisson mean and event before evaluating the formula.

Key takeaway: For a Poisson claim count, first calculate $\mu=rE$ using consistent exposure units. Then evaluate $e^{-\mu}\mu^k/k!$ for exactly $k$ claims, or use sums and complements for broader events.

Check Your Understanding

For each problem, identify the Poisson mean before calculating. Give probabilities to four decimal places.

A hypothetical insurer uses a rate of $0.25$ claims per quarter for a group with 8 quarters of exposure. Find the probability of zero claims. (Answer: $0.1353$)
A hypothetical portfolio has a rate of $0.12$ claims per policy-year and 10 policy-years of exposure. Find the probability of exactly one claim. (Answer: $0.3614$)
A hypothetical repair program has a rate of $0.5$ claims per 1,000 device-hours and 6,000 device-hours of exposure. Find the probability of exactly three claims. (Answer: $0.2240$)
A hypothetical membership group has a rate of $0.04$ claims per member-month. There are 30 members, each observed for 2 months. Find the probability of at least one claim. (Answer: $0.9093$)
A hypothetical vehicle group has a rate of $0.3$ claims per vehicle-year and 5 vehicle-years of exposure. Find both the probability of exactly two claims and the probability of at most one claim. (Answer: exactly two: $0.2510$; at most one: $0.5578$)

What You'll Learn