What the Loss Elimination Ratio Measures
An ordinary deductible leaves the policyholder responsible for the first part of each covered loss. The loss elimination ratio (LER) measures the expected amount of ground-up loss removed by that deductible as a proportion of the expected ground-up loss. It is a per-loss severity measure: its numerator and denominator are expectations for the same ground-up loss variable.
The LER is not the probability that a loss is below the deductible, and it is not automatically the fraction of an insurer’s total payments that has been eliminated. Its value depends on the severity distribution and the deductible, as well as on consistent definitions of the losses being compared. This treatment is aligned with the SOA Exam FAM (Fundamentals of Actuarial Mathematics) syllabus.
The condition \(E[X]>0\) is needed because the expected ground-up loss is the denominator. A zero expected loss would make the ratio undefined. The finite-mean condition ensures that the denominator is an ordinary finite expectation. For the definition to describe the intended coverage, \(X\) and \(d\) must refer to the same loss basis and currency, and the deductible must be an ordinary per-loss deductible rather than a franchise deductible or an annual aggregate deductible.
Why the Eliminated Amount Is \(\min(X,d)\)
For one loss, compare its size with the deductible. If \(X\leq d\), the deductible removes the whole loss, so the eliminated amount is \(X\). If \(X>d\), the deductible removes \(d\), and the amount above \(d\) remains eligible for payment before any other coverage terms. In both cases, the eliminated amount is the smaller of \(X\) and \(d\).
Taking the expectation of this per-loss amount gives the expected loss eliminated. Dividing by \(E[X]\) expresses that amount relative to the expected ground-up loss. This is the defining construction of the LER; it does not require a particular parametric severity distribution. For a discrete distribution, the numerator is found by weighting \(\min(x,d)\) at each possible loss amount \(x\). For a continuous distribution, it is still the expectation of the same capped amount.
Identify the ground-up loss variable \(X\), including the currency and price level in which it is measured.
For every possible loss amount \(x\), calculate the eliminated amount \(\min(x,d)\).
Calculate \(E[\min(X,d)]\) and \(E[X]\) under the same severity model, then divide the first by the second.
Assumptions and Interpretation
The definition is simple, but its interpretation depends on what \(X\) represents. In an ordinary severity model, \(X\) is the amount of one ground-up claim before the deductible is applied. If the model is conditional on a claim occurring, both expectations are taken under that conditional claim severity distribution. If zero-loss outcomes are included instead, they must be included consistently in both expectations. Mixing a claim-conditional numerator with a distribution that includes non-claim periods in the denominator would not give the stated per-claim LER.
The coverage assumed here has a fixed ordinary deductible \(d\) applied to each loss. Other terms—such as a policy limit, coinsurance, exclusions, or an aggregate deductible—can change the amount eliminated or the relevant loss basis. They should not be silently folded into this definition. In particular, a franchise deductible does not generally remove \(\min(X,d)\): its payment rule differs at the deductible boundary.
The deductible and severity also need a consistent price basis. If historical losses have been trended to current dollars, a fixed current-dollar deductible must be compared with those trended losses. If both severity and deductible are expressed in an earlier price basis, that can also be consistent, provided both are genuinely on that same basis. Applying a current fixed deductible to untrended historical losses can distort the ratio because the comparison then uses mismatched dollar values.
| Item | Required interpretation for this definition |
|---|---|
| \(X\) | Nonnegative ground-up loss per claim, on a specified and consistent basis |
| \(d\) | Fixed ordinary deductible applied separately to each loss, in the same units as \(X\) |
| Numerator | Expected ground-up loss eliminated per loss, \(E[\min(X,d)]\) |
| Denominator | Expected ground-up loss per loss, \(E[X]\), finite and positive |
| Interpretation | Expected eliminated ground-up loss as a proportion of expected ground-up loss |
Worked Calculations
Worked Example: A Deductible Applied to a Discrete Severity Distribution
A hypothetical insurer models one claim’s ground-up loss \(X\) in dollars with the following distribution. The ordinary deductible is \(d=\$1{,}000\). Calculate the LER.
| Loss \(x\) | Probability | Eliminated amount \(\min(x,1{,}000)\) |
|---|---|---|
| \$0 | 0.20 | \$0 |
| \$500 | 0.30 | \$500 |
| \$2,000 | 0.40 | \$1,000 |
| \$8,000 | 0.10 | \$1,000 |
Method. Calculate the expected ground-up loss and the expected amount eliminated by applying the deductible to each possible loss. The probabilities sum to \(0.20+0.30+0.40+0.10=1.00\).
Expected ground-up loss.
Expected loss eliminated.
Therefore,
Check and common mistake. The result is between 0% and 100%, as expected. A frequent error is to count every loss above the deductible as having its full amount eliminated. For the \$2,000 loss, the deductible eliminates \$1,000, not \$2,000; the remaining \$1,000 is above the deductible.
Worked Example: Trending Losses Before Applying a Fixed Deductible
A hypothetical portfolio has historical ground-up claim severities of \$1,000 with probability 0.50 and \$5,000 with probability 0.50. Claims are trended by a multiplicative factor of 1.20 to a current price basis. The deductible is fixed at \$3,000 in current dollars. Find the current-basis LER.
Method. Trend each loss first, then apply the deductible that is stated in current dollars. The current-basis severities are \(1.20(1{,}000)=\$1{,}200\) and \(1.20(5{,}000)=\$6{,}000\).
Expected current-basis ground-up loss.
Expected amount eliminated. The deductible eliminates \$1,200 on the first loss and \$3,000 on the second:
Thus,
Check and common mistake. The expected eliminated amount is less than the expected ground-up loss, and the ratio is less than 1. Applying the \$3,000 current-dollar deductible directly to the historical \$1,000 and \$5,000 severities would mix price bases; the correct comparison trends the losses before applying the fixed current-dollar deductible.
Worked Example: A Loss Exactly Equal to the Deductible
For a hypothetical product, one claim has the following ground-up severity distribution: \$0 with probability 0.10, \$800 with probability 0.20, \$2,500 with probability 0.40, and \$6,000 with probability 0.30. The ordinary deductible is \(d=\$2{,}500\). Calculate the LER, paying particular attention to the loss exactly equal to the deductible.
Method. Apply \(\min(X,2{,}500)\) to each outcome. A loss equal to the deductible has its entire amount eliminated, just as a loss below the deductible does.
Expected ground-up loss.
Expected amount eliminated.
Therefore,
Check and common mistake. The loss of exactly \$2,500 contributes \$2,500 to the eliminated amount. It is incorrect to treat the deductible as eliminating nothing at equality: under an ordinary deductible, the amount above the deductible is zero when the loss equals \(d\), so the entire ground-up loss is retained by the policyholder.
Common Interpretive Errors
- Using the wrong amount in the numerator. The eliminated amount on a loss \(x\) is \(\min(x,d)\), not \(x\) for every loss and not \(d\) for every loss.
- Treating the LER as a probability. It is a ratio of expected dollar amounts. A value of 40% does not mean that 40% of claims fall below the deductible.
- Mixing loss populations. The numerator and denominator must come from the same severity basis, including the same treatment of zero-loss outcomes.
- Mixing price bases. A deductible and the severities to which it is applied must use compatible currency and inflation bases.
- Ignoring the coverage form. The formula here is for an ordinary per-loss deductible; a franchise or aggregate deductible has a different operation.
Check Your Understanding
For each problem, use the ordinary per-loss deductible and the stated ground-up severity distribution.
- A hypothetical claim has severity \$500 with probability 0.40, \$1,500 with probability 0.40, and \$3,500 with probability 0.20. With a \$1,000 deductible, calculate the LER. (Answer: 0.5333, or 53.33%)
- A severity is \$0 with probability 0.10, \$1,200 with probability 0.30, \$3,000 with probability 0.40, and \$7,000 with probability 0.20. With a \$3,000 deductible, calculate the expected ground-up loss and the LER. (Answer: \(E[X]=\$2{,}960\); LER \(=0.7297\), or 72.97%)
- A hypothetical severity is \$2,000 with probability 0.25, \$5,000 with probability 0.50, and \$12,000 with probability 0.25. Find the LER for a \$4,000 deductible. (Answer: 0.5833, or 58.33%)
- Historical losses are \$1,000 and \$5,000, each with probability 0.50. They are trended by a factor of 1.20, and the current fixed deductible is \$3,000. Calculate the current-basis LER. (Answer: 0.5833, or 58.33%)
- A severity is \$1,000 with probability 0.50 and \$5,000 with probability 0.50. Find the deductible that gives an LER of 0.5000, assuming the deductible is between \$1,000 and \$5,000. (Answer: \$2,000)