Why Consistency Matters
Inflation and loss elimination ratio (LER) calculations can look straightforward while quietly mixing amounts from different years or different populations. A loss may be measured at historical prices, a deductible may be stated in current dollars, and the data used to estimate an average may include only losses above a deductible. Combining these quantities without adjustment can produce an answer that is arithmetically tidy but conceptually wrong.
The safest approach is to make the basis explicit before calculating: identify the valuation dates, decide whether each coverage term is historical or current, and state which expected loss belongs in the LER denominator. These checks are aligned with the SOA Exam FAM (Fundamentals of Actuarial Mathematics) syllabus.
An inflation adjustment changes the monetary basis of a loss. If \(I_0\) is the relevant price index at the loss date and \(I_1\) is the index at the valuation date, the trend factor is \(I_1/I_0\). A historical loss \(x_0\) becomes \(x_1=x_0(I_1/I_0)\). The ratio is later index divided by earlier index: the adjusted amount should move in the direction indicated by the index change.
A deductible is not automatically adjusted just because a loss is trended. If a historical loss is being brought to current dollars and the policy deductible is stated in current dollars, trend the loss first and then apply that current deductible. If instead both the loss and deductible are expressed in historical dollars, apply the historical deductible on that basis. Either approach can be valid when the terms are converted consistently; mixing them is not.
A Short Consistency Check for LER
The LER identity follows directly from splitting each ground-up loss into the part paid by the insurer and the part eliminated by the deductible. For every \(x\geq 0\),
If \(x\leq d\), the insurer payment is zero and the eliminated amount is \(x\). If \(x>d\), the insurer payment is \(x-d\) and the eliminated amount is \(d\). Taking expectations gives \(E[X]=E[(X-d)_+]+E[\min(X,d)]\). Dividing by \(E[X]\) gives the two equivalent LER expressions. This also explains why an expected payment can be used to find LER, but only when it is divided by the expected ground-up loss on the same basis.
Worked Calculations and the Errors They Avoid
Worked Example: Reversing the Index Ratio
A hypothetical book of losses has the following 2022 ground-up severity distribution: $500 with probability 0.20, $1,000 with probability 0.50, and $2,000 with probability 0.30. The relevant price index is 125 in 2022 and 150 in 2025. Calculate the 2025 expected loss and the LER for a $900 deductible stated in 2025 dollars.
Method. Trend each loss using the 2025 index divided by the 2022 index, then apply the deductible in 2025 dollars. First, the trend factor is \(150/125=1.20\). Thus the three 2025 losses are \(500(1.20)=600\), \(1{,}000(1.20)=1{,}200\), and \(2{,}000(1.20)=2{,}400\).
The expected 2025 ground-up loss is \(0.20(600)+0.50(1{,}200)+0.30(2{,}400)=120+600+720=\$1{,}440\). The amounts eliminated by the $900 deductible are $600, $900, and $900. Their expected value is \(0.20(600)+0.50(900)+0.30(900)=120+450+270=\$840\).
Therefore, \(\operatorname{LER}=840/1{,}440=0.5833\), or 58.33%. As a check, expected insurer payment is \(0.20(0)+0.50(300)+0.30(1{,}500)=\$600\), and \(1-600/1{,}440=0.5833\).
Common mistake and correction. Reversing the index ratio gives \(125/150=0.8333\), which would reduce historical losses even though the index rose. The correction is to use later index divided by earlier index; the resulting trend factor of 1.20 increases each loss.
Worked Example: Compounding Annual Inflation Rates
A hypothetical severity distribution has 2021 losses of $1,000, $2,000, and $4,000 with probabilities 0.25, 0.50, and 0.25. Annual inflation is 6% in 2022, 4% in 2023, and 5% in 2024. Find the 2024 LER for a $2,000 deductible stated in 2024 dollars.
Method. Successive annual changes multiply. The cumulative factor is \(1.06(1.04)(1.05)=1.15752\). The 2024 losses are therefore \(1{,}000(1.15752)=\$1{,}157.52\), \(2{,}000(1.15752)=\$2{,}315.04\), and \(4{,}000(1.15752)=\$4{,}630.08\).
The expected ground-up loss is \(0.25(1{,}157.52)+0.50(2{,}315.04)+0.25(4{,}630.08)\) \(=289.38+1{,}157.52+1{,}157.52=\$2{,}604.42\). The deductible eliminates $1,157.52, $2,000, and $2,000 in the three cases. Thus the expected eliminated loss is \(0.25(1{,}157.52)+0.50(2{,}000)+0.25(2{,}000)\) \(=289.38+1{,}000+500=\$1{,}789.38\).
The LER is \(1{,}789.38/2{,}604.42=0.6871\), or 68.71%, to four decimal places. As a check, expected insurer payment is \(2{,}604.42-1{,}789.38=\$815.04\), and \(1-815.04/2{,}604.42=0.6871\).
Common mistake and correction. Adding the annual rates gives 15% and a trend factor of 1.15. That is not the compounded change: each year's rate applies to the already adjusted amount. Use the product of the annual factors, 1.15752, for the three-year trend.
Worked Example: Applying a Current Deductible After Trending
A hypothetical 2020 loss is $1,600. The relevant index rises from 100 in 2020 to 125 in 2024. A policy in 2024 has a $1,800 deductible. Calculate the insurer payment for this loss, and compare it with the result of deducting first and then trending the remaining payment.
Method. The 2024 trend factor is \(125/100=1.25\), so the loss at 2024 values is \(1{,}600(1.25)=\$2{,}000\). Apply the 2024 deductible to that trended loss: the payment is \((2{,}000-1{,}800)_+=\$200\).
The incorrect order first applies the $1,800 current deductible to the $1,600 historical loss, producing zero, and then trends zero; that gives a payment of $0. The deductible and loss were on different bases in that calculation. The corrected calculation puts the loss in 2024 dollars before applying the 2024 term.
There is a valid alternative if the deductible is converted too: $1,800 in 2024 dollars is \(1{,}800/1.25=\$1{,}440\) in 2020 dollars. The historical payment is then \(1{,}600-1{,}440=\$160\), which trends to \(160(1.25)=\$200\). Both consistent methods agree.
Check. The 2024 loss of $2,000 is $200 above the current deductible, so the correct payment is positive and equals $200. Applying a current deductible directly to an unadjusted historical loss cannot answer the stated question.
Worked Example: Keeping the LER Denominator Consistent
A hypothetical ground-up loss \(X\) is $100 with probability 0.50, $200 with probability 0.30, and $500 with probability 0.20. For a $150 deductible, calculate the LER. Then identify why a ratio calculated only among losses above the deductible is not the same LER.
Method. First calculate the unconditional ground-up mean: \(E[X]=0.50(100)+0.30(200)+0.20(500)=50+60+100=\$210\). The deductible eliminates $100, $150, and $150 in the three cases, so \(E[\min(X,150)]=0.50(100)+0.30(150)+0.20(150)=50+45+30=\$125\). The LER is \(125/210=0.5952\), or 59.52%, to four decimal places.
For comparison, \(P(X>150)=0.30+0.20=0.50\). Conditional on exceeding the deductible, the probabilities of $200 and $500 are \(0.30/0.50=0.60\) and \(0.20/0.50=0.40\). The conditional mean ground-up loss is \(0.60(200)+0.40(500)=\$320\), while conditional mean insurer payment is \(0.60(50)+0.40(350)=\$170\). The ratio \(170/320=0.5313\) is a ratio conditional on exceeding $150, not the unconditional insurer-payment ratio used in the LER identity.
Unconditionally, expected insurer payment is \(0.50(0)+0.30(50)+0.20(350)=\$85\). Hence \(1-85/210=0.5952\), agreeing with the LER calculated from eliminated loss. Dividing the conditional payment by the unconditional ground-up mean, or dividing unconditional eliminated loss by a conditional mean, would mix populations and give no valid LER.
Common mistake and correction. A deductible-truncated data set does not, by itself, represent all ground-up losses. Use the unconditional ground-up denominator and account for losses at or below the deductible when the target is the portfolio LER. If those losses or their probability are unknown, the unconditional LER cannot generally be recovered from the truncated observations alone.
A Practical Error-Checking Sequence
Before accepting an inflation-adjusted LER, check the calculation in a fixed order. This is particularly useful when a question includes several dates, annual trend rates, or loss data subject to a deductible.
Write down the loss date, valuation date, relevant index values, and the date basis of each deductible or other term.
For index values, divide the later index by the earlier index. For successive annual rates, multiply the annual factors rather than adding the rates.
Trend the ground-up loss to the term's basis, or convert the term to the loss's basis. Do not combine a historical loss with a current deductible without conversion.
Use expected eliminated loss over expected ground-up loss for the same population and valuation basis. Alternatively, subtract the expected insurer payment divided by that same expected ground-up loss from one.
For a nonnegative loss and ordinary deductible, \(0\leq \operatorname{LER}(d)\leq 1\). Verify that expected ground-up loss equals expected eliminated loss plus expected insurer payment.
The final checks catch different errors. A reversed index ratio may make a rising index reduce losses. Added annual rates miss compounding. Applying coverage before trending may silently treat a current term as historical. A mismatched LER denominator changes the population being compared. Writing down the basis at each step makes these mistakes easier to detect than relying on a plausible-looking final percentage.
Check Your Understanding
For each problem, keep the loss, coverage term, and denominator on the stated basis. Answers are provided for checking.
- A hypothetical loss of $800 occurred when the relevant index was 200; the valuation-date index is 250. What is the trended loss? (Answer: $1,000.)
- Annual inflation rates over three successive years are 3%, 4%, and 2%. Calculate the cumulative trend factor and trend a $500 loss. (Answer: factor 1.092624; trended loss $546.31.)
- A hypothetical historical loss of $1,600 is trended by a factor of 1.25. The current deductible is $1,800. What is the insurer payment? (Answer: $200.)
- A ground-up severity is $100 with probability 0.40 and $300 with probability 0.60. Find the LER for a $200 deductible. (Answer: \(160/220=0.7273\), or 72.73%.)
- A loss is $100 or $400, each with probability 0.50. For a $150 deductible, calculate the LER and check it using expected insurer payment. (Answer: LER \(=125/250=0.5000\); expected payment is $125, and \(1-125/250=0.5000\).)