Tutorials › CAS Exam 8 - Advanced Ratemaking › Converting Claim Thresholds Into Required Exposure

Home  /  Tutorials  /  CAS Exam 8 / Credibility and Limited Fluctuation

CAS Exam 8 · Credibility and Limited Fluctuation · Tutorial 301 of 989

Converting Claim Thresholds Into Required Exposure

Translate a stated claim-count threshold and expected claim frequency into the exposure needed to meet the count standard.

Beginner 9 min read

What You'll Learn

  • Convert an expected claim-count threshold into required exposure using claim frequency
  • Keep frequency and exposure units consistent when converting rates stated per 100 or 1,000 units
  • Distinguish the expected-count criterion from a probability of observing a minimum number of claims
  • Scale required exposure to a stated portfolio size and duration
  • Combine class frequencies when a portfolio has a fixed mix of exposure types
  • Identify when a single frequency is not appropriate for heterogeneous exposure

From a Claim-Count Threshold to Exposure

A claim-count full-credibility standard can specify how many expected claims are needed to meet a chosen count criterion. For ratemaking applications, that threshold is useful only when it can be translated into the volume of risk needed to produce it. The conversion uses the expected claim frequency per exposure unit: if each exposure unit produces claims at a known average rate, the required exposure is the count threshold divided by that rate.

The calculation is simple, but its interpretation depends on consistent units and a clear understanding of what the threshold means. In particular, an expected count is not the same as a guarantee that a particular number of claims will actually occur.

Definition: Let \(N_0\) be a specified expected claim-count threshold, let \(\lambda\) be the expected number of claims per exposure unit, and let \(E\) be the required exposure measured in those same units. Under a constant-frequency model, the required exposure is $$ E_{\mathrm{req}}=\frac{N_0}{\lambda}. $$ The units of \(E_{\mathrm{req}}\) are the exposure units used to define \(\lambda\), such as policy-years or vehicle-years.

For example, if a standard requires 800 expected claims and the expected frequency is 0.04 claims per policy-year, the required exposure is measured in policy-years. The numerical value of the frequency must be expressed per one policy-year before applying the formula. If it is stated as 4 claims per 100 policy-years, then \(\lambda=4/100=0.04\) claims per policy-year.

Why the Conversion Works

Suppose claim counts follow a Poisson model with constant frequency \(\lambda\) per exposure unit. For \(E\) units of exposure, the total claim count \(N\) has mean \(\lambda E\). The count threshold is met in the expected-count sense when that mean reaches \(N_0\). Setting the expected count equal to the threshold gives the minimum exposure at the boundary:

$$ E[N]=\lambda E,\qquad \lambda E=N_0,\qquad E_{\mathrm{req}}=\frac{N_0}{\lambda}. $$

This is a scaling argument: doubling exposure doubles the expected number of claims when the frequency stays constant. The Poisson assumption supplies a standard count model, but the algebraic conversion itself follows from the specified expected frequency and the expected-count criterion.

This calculation does not say that the observed count will equal \(N_0\), or that it will be at least \(N_0\) with certainty. A random count can be above or below its mean. The conversion answers a narrower question: how much exposure gives an expected claim count equal to the stated threshold?

1
Identify the count threshold.
Use the stated expected claim-count requirement \(N_0\). Do not substitute a probability requirement unless the problem explicitly asks for one.
2
Put frequency on a per-unit basis.
Convert rates stated per 100 or 1,000 exposure units to claims per one exposure unit.
3
Divide threshold by frequency.
Calculate \(E_{\mathrm{req}}=N_0/\lambda\), retaining the exposure unit associated with \(\lambda\).
4
Interpret the result in context.
If the question asks for whole policies, whole years, or another indivisible quantity, round up only after converting to that quantity.

Worked Conversions

Worked Example: Required Policy-Years at a Given Frequency

A hypothetical property portfolio uses an expected claim-count threshold of 1,200 claims. Its expected frequency is 0.08 claims per policy-year. Assume a constant frequency and Poisson claim counts over the portfolio exposure. How many policy-years are required to reach the threshold in expected-count terms?

Method. Convert the threshold to exposure using \(E_{\mathrm{req}}=N_0/\lambda\). Here, \(N_0=1{,}200\) claims and \(\lambda=0.08\) claims per policy-year.

Substitute the values:

$$ E_{\mathrm{req}} =\frac{1{,}200\ \text{claims}}{0.08\ \text{claims per policy-year}} =15{,}000\ \text{policy-years}. $$

Check. At 15,000 policy-years, the expected count is \(0.08(15{,}000)=1{,}200\) claims, exactly the required threshold. The answer is 15,000 policy-years.

Common mistake avoided. The frequency is already per policy-year, so no additional factor of 100 or 1,000 belongs in the calculation.

Worked Example: Frequency Stated per 100 Policy-Years

A hypothetical equipment insurer needs an expected count of 360 claims for a count-based standard. The expected frequency is 2.4 claims per 100 policy-years. If 750 policies are continuously exposed for the same amount of time each year, how many years of exposure are required?

Step 1: Convert the frequency. The rate per one policy-year is

$$ \lambda=\frac{2.4}{100}=0.024 \quad\text{claims per policy-year}. $$

Step 2: Calculate required policy-years.

$$ E_{\mathrm{req}}=\frac{360}{0.024}=15{,}000\ \text{policy-years}. $$

Step 3: Convert policy-years to years for this portfolio. With 750 policies exposed for one full year, the portfolio accumulates 750 policy-years per year. Therefore,

$$ \text{years required} =\frac{15{,}000\ \text{policy-years}}{750\ \text{policy-years per year}} =20\ \text{years}. $$

Check. The expected count per portfolio year is \(750(0.024)=18\) claims. Over 20 years it is \(20(18)=360\) claims. The portfolio requires 15,000 policy-years, or 20 years at the stated size.

Common mistake avoided. Using 2.4 as if it were the rate per policy-year would give \(360/2.4=150\), a value smaller by a factor of 100. The stated denominator, 100 policy-years, must be accounted for first.

Worked Example: Converting Exposure to Whole Portfolio Years

A hypothetical marine portfolio has an expected frequency of 0.15 claims per boat-year. Its count standard requires 750 expected claims. The portfolio contains 320 boats, each insured for a full year, and its size and frequency are assumed unchanged over time. Find the required boat-years and the minimum whole number of portfolio years that reaches the threshold.

Step 1: Calculate required exposure.

$$ E_{\mathrm{req}} =\frac{750\ \text{claims}}{0.15\ \text{claims per boat-year}} =5{,}000\ \text{boat-years}. $$

Step 2: Convert exposure to portfolio years. The portfolio earns 320 boat-years per year, so the unrounded duration is

$$ \frac{5{,}000}{320}=15.625\ \text{years}. $$

Because the question asks for a whole number of full portfolio years, round up to 16 years. Rounding down would leave the expected count below the threshold.

Step 3: Verify the rounded duration. In 16 years the portfolio earns \(320(16)=5{,}120\) boat-years, giving an expected count of \(0.15(5{,}120)=768\) claims. In 15 years it earns \(320(15)=4{,}800\) boat-years, giving \(0.15(4{,}800)=720\) expected claims. Thus 16 is the minimum whole number of years that reaches at least 750 expected claims.

Common mistake avoided. The required exposure is 5,000 boat-years, not 5,000 years. Dividing by the portfolio’s 320 boats is necessary to express the answer as a duration.

Units, Portfolio Mix, and Interpretation

Frequency and exposure must refer to the same risk and time basis. A claim frequency per earned vehicle-year should be paired with vehicle-years, not policy counts unless each policy represents exactly one vehicle-year. Likewise, an annual frequency should not be applied to a month of exposure without converting the time basis. If the problem provides an exposure unit directly, preserve that unit throughout the calculation and convert only when the requested answer uses another unit.

When exposure types have different frequencies, a single frequency is appropriate only if it represents the portfolio mix being studied. For a fixed mix, the portfolio’s expected claims per total exposure unit can be computed from the class-level expected counts. If class \(j\) contributes \(E_j\) units at frequency \(\lambda_j\), its expected count is \(\lambda_jE_j\), and the total expected count is the sum across classes. A changing mix can change the overall frequency, so a rate based on one mix should not automatically be applied to another.

The expected-count threshold also differs from a requirement about the probability of reaching a count. For example, asking for exposure that produces a mean of \(N_0\) uses the conversion above. Asking for exposure such that \(P(N\geq N_0)\) is at least a specified probability is a different calculation, because it depends on the distribution of \(N\), not just its mean. Do not silently replace one criterion with the other.

Key takeaway: For a stated expected claim-count threshold, divide the threshold by the expected frequency expressed per one matching exposure unit. Confirm the units, then convert exposure to policy count or duration if required. The result targets an expected count; it does not guarantee an observed count.

Check Your Understanding

For each problem, calculate exposure in the requested units. Use the stated expected-count interpretation.

  1. A hypothetical insurer requires 500 expected claims. The frequency is 0.05 claims per policy-year. How many policy-years are required? (Answer: 10,000 policy-years)
  2. A hypothetical fleet has an expected count threshold of 270 claims and a frequency of 1.8 claims per 100 vehicle-years. How many vehicle-years are required? (Answer: 15,000 vehicle-years)
  3. A hypothetical portfolio requires 840 expected claims at 0.12 claims per unit-year. It has 350 units exposed for a full year each. What is the required exposure in unit-years and in whole portfolio years? (Answer: 7,000 unit-years and 20 years)
  4. A hypothetical class has a count threshold of 1,100 expected claims and a frequency of 2.2 claims per 100 exposure units. How many exposure units are required? (Answer: 50,000 exposure units)
  5. A hypothetical portfolio has 600 class A policies and 400 class B policies exposed for one full year. Their expected frequencies are 0.03 and 0.08 claims per policy-year, respectively. If the count threshold is 750 expected claims and this mix stays fixed, how many years are required? (Answer: 15 years)

Previous

Next

Back to Tutorials
Clinical Biostats