What an Incurred Chain Ladder Estimates
An incurred chain ladder uses the development history of cumulative incurred losses to project each accident year’s ultimate loss. Incurred losses include amounts paid to date and case estimates for reported claims. The method is especially useful when a paid triangle alone does not reflect the current claim estimates, but the result depends on whether the selected development pattern is appropriate for the portfolio.
The calculation has two distinct parts. First, apply selected incurred development factors to the latest cumulative incurred value for each accident year to estimate ultimate incurred loss. Second, to estimate the total unpaid claims liability, subtract cumulative paid loss from that ultimate. The difference between ultimate incurred and current incurred is not the total unpaid liability; it is the projected increase in incurred loss after the current valuation.
A conventional chain ladder calculation assumes that historical development factors are informative about future development, that the cumulative incurred amounts are measured consistently, and that changes in claim handling, case reserving, coverage, and the mix of business do not invalidate the selected pattern. These are substantive assumptions, not guarantees. If the data are inconsistent or the pattern is unstable, mechanically applying factors can produce misleading estimates.
Calculation Framework
Let \(C_{i,j}\) be cumulative incurred losses for accident year \(i\) at development age \(j\). For an observed transition from age \(j\) to \(j+1\), the link ratio is \(C_{i,j+1}/C_{i,j}\), provided \(C_{i,j}\ne 0\). A selected factor \(f_j\) is applied to cumulative values at age \(j\). The selected factor need not equal any one observed link ratio; it is an actuarial selection informed by the available experience.
Here \(m\) is the final development age in the selected pattern, \(j_i\) is the latest observed age for accident year \(i\), and \(\widehat{U}_i\) is its projected ultimate incurred loss. If \(P_i\) is cumulative paid loss for that year at the same valuation date, the total unpaid claims estimate is \(\widehat{U}_i-P_i\). This requires paid data as well as the incurred triangle.
For each accident year, identify its latest observed cumulative incurred value and development age.
Multiply the factors for all remaining development ages to obtain that age’s age-to-ultimate factor.
Multiply the latest cumulative incurred value by its age-to-ultimate factor.
Subtract cumulative paid loss, not current incurred loss, from projected ultimate incurred loss.
Worked Incurred Triangle
Worked Example: Projecting a Four-Year Incurred Triangle
A hypothetical insurer has the following cumulative incurred triangle, in thousands of dollars. The latest cumulative paid losses at the valuation date are also shown. For this exercise, the selected factors are 12-to-24 months \(1.46\), 24-to-36 months \(1.20\), and 36-to-48 months \(1.10\). Assume no further development after 48 months, so the 48-month-to-ultimate factor is \(1.00\). Calculate each ultimate and total unpaid claims liability.
| Accident year | 12 months incurred | 24 months incurred | 36 months incurred | 48 months incurred | Paid to date |
|---|---|---|---|---|---|
| Year 1 | 1,000 | 1,500 | 1,800 | 1,980 | 1,850 |
| Year 2 | 1,100 | 1,595 | 1,914 | — | 1,650 |
| Year 3 | 1,200 | 1,740 | — | — | 1,300 |
| Year 4 | 1,300 | — | — | — | 700 |
Method. This is an incurred chain ladder using the supplied selected factors. The observed link ratios are consistent with the selections: Year 1 has \(1{,}500/1{,}000=1.50\), \(1{,}800/1{,}500=1.20\), and \(1{,}980/1{,}800=1.10\); the observed 12-to-24 ratios for Years 2 and 3 are \(1{,}595/1{,}100=1.45\) and \(1{,}740/1{,}200=1.45\). The exercise specifies the selections; it does not require choosing a factor-selection method.
Step 1: Find the age-to-ultimate factors. At 12 months the remaining factors are \(1.46\), \(1.20\), and \(1.10\), giving \(1.46\times1.20\times1.10=1.9272\). At 24 months the factor is \(1.20\times1.10=1.32\). At 36 months it is \(1.10\). At 48 months it is \(1.00\).
Step 2: Project each ultimate. Year 1 is already at 48 months, so its projected ultimate is \(1{,}980\times1.00=1{,}980\). Year 2 is at 36 months, giving \(1{,}914\times1.10=2{,}105.40\). Year 3 is at 24 months, giving \(1{,}740\times1.32=2{,}296.80\). Year 4 is at 12 months, giving \(1{,}300\times1.9272=2{,}505.36\). All amounts remain in thousands of dollars.
Step 3: Calculate total unpaid claims. Subtract paid-to-date amounts from projected ultimate incurred losses: Year 1, \(1{,}980-1{,}850=130\); Year 2, \(2{,}105.40-1{,}650=455.40\); Year 3, \(2{,}296.80-1{,}300=996.80\); and Year 4, \(2{,}505.36-700=1{,}805.36\).
The total projected ultimate is \(1{,}980+2{,}105.40+2{,}296.80+2{,}505.36=8{,}887.56\). Total paid is \(1{,}850+1{,}650+1{,}300+700=5{,}500\). Therefore, total unpaid claims are \(8{,}887.56-5{,}500=3{,}387.56\), matching the sum \(130+455.40+996.80+1{,}805.36=3{,}387.56\). The reconciliation checks the accident-year calculations.
Common mistake avoided: Subtracting current incurred from ultimate would give Year 2 \(2{,}105.40-1{,}914=191.40\), not its total unpaid claims of \(455.40\). The difference is the projected increase beyond current incurred; the existing gap between incurred and paid is \(1{,}914-1{,}650=264\). Together, \(191.40+264=455.40\).
Why the Factor Product Works
The age-to-ultimate multiplier follows directly from successive projections. Suppose a cumulative incurred amount \(C_{i,j}\) develops first to age \(j+1\) by factor \(f_j\), then to age \(j+2\) by factor \(f_{j+1}\). The first projection is \(C_{i,j}f_j\). Applying the next factor gives \(C_{i,j}f_jf_{j+1}\). Repeating this step through the final age gives the product of every remaining selected factor.
This is a chained projection: each factor applies to the cumulative amount produced by the prior step. Adding the factors or applying only the next factor would not represent successive percentage development.
More Applications and Checks
Worked Example: Calculating Liabilities for a Partially Mature Triangle
A hypothetical specialty insurer has three accident years, with cumulative incurred losses in thousands: Year A has \(600\) at 12 months, \(900\) at 24 months, and \(1{,}080\) at 36 months; Year B has \(700\) at 12 months and \(1{,}015\) at 24 months; Year C has \(800\) at 12 months. Paid losses at the valuation date are \(950\), \(800\), and \(500\), respectively. Use selected factors \(1.48\) from 12 to 24 months and \(1.20\) from 24 to 36 months, with no further development after 36 months.
Method and age-to-ultimate factors. At 12 months, the multiplier is \(1.48\times1.20=1.776\). At 24 months it is \(1.20\); at 36 months it is \(1.00\).
Year A. Its latest incurred value is \(1{,}080\) at 36 months, so ultimate is \(1{,}080\times1.00=1{,}080\). Unpaid claims are \(1{,}080-950=130\).
Year B. Its latest incurred value is \(1{,}015\) at 24 months, so ultimate is \(1{,}015\times1.20=1{,}218\). Unpaid claims are \(1{,}218-800=418\).
Year C. Its latest incurred value is \(800\) at 12 months, so ultimate is \(800\times1.776=1{,}420.80\). Unpaid claims are \(1{,}420.80-500=920.80\).
The total ultimate is \(1{,}080+1{,}218+1{,}420.80=3{,}718.80\). The total paid amount is \(950+800+500=2{,}250\), so the aggregate unpaid estimate is \(3{,}718.80-2{,}250=1{,}468.80\). The accident-year amounts also sum to \(130+418+920.80=1{,}468.80\). The common error avoided is applying the 12-to-24 factor to Year B, which is already observed at 24 months; only its 24-to-36 factor remains.
Worked Example: Separating Case Outstanding from Future Development
For a hypothetical accident year, cumulative incurred loss at 24 months is \(2{,}400\) thousand dollars and cumulative paid loss is \(1{,}700\) thousand dollars. The selected factors are \(1.15\) from 24 to 36 months and \(1.08\) from 36 months to ultimate. Calculate ultimate incurred loss, total unpaid claims, and the projected increase in incurred loss after 24 months.
Method. Chain the remaining factors to get the 24-month-to-ultimate multiplier: \(1.15\times1.08=1.242\). Thus projected ultimate incurred loss is \(2{,}400\times1.242=2{,}980.80\) thousand dollars.
Total unpaid claims are ultimate less paid: \(2{,}980.80-1{,}700=1{,}280.80\) thousand dollars. The projected increase in incurred loss after 24 months is ultimate less current incurred: \(2{,}980.80-2{,}400=580.80\) thousand dollars. Current case outstanding is incurred less paid: \(2{,}400-1{,}700=700\) thousand dollars. These pieces reconcile: \(700+580.80=1{,}280.80\).
The wrong calculation would report \(580.80\) as total unpaid claims because it subtracts current incurred from ultimate. That omits the existing \(700\) thousand dollars of incurred-but-unpaid case estimates. The corrected total is \(1{,}280.80\) thousand dollars, equal to ultimate less paid.
Interpretation and Exam Traps
An incurred triangle contains cumulative amounts, so a link factor projects one cumulative value to another. Do not add a projected cumulative value to the current cumulative value: the factor has already scaled the full cumulative amount. Similarly, do not confuse the age of a row with the development age of its latest observed cell. The correct age determines which remaining factors enter the product.
A final check is to verify that projected ultimate is at least current incurred and that total unpaid claims are nonnegative, assuming the data and selected factors support those conditions. Also verify the aggregate identity: total unpaid equals total projected ultimate less total paid. If this identity fails, check for a missing accident year, a mismatched valuation date, or subtraction of incurred rather than paid.
Check Your Understanding
For each original problem, use the stated selected factors and report monetary answers in the units given.
- A hypothetical accident year has cumulative incurred loss of \(1{,}250\) at 12 months and cumulative paid loss of \(600\). Selected factors are \(1.40\) from 12 to 24 months and \(1.10\) from 24 months to ultimate. Find the ultimate and unpaid claims. (Answer: ultimate \(1{,}925\); unpaid claims \(1{,}325\))
- A year has cumulative incurred loss \(2{,}160\) at 24 months and paid loss \(1{,}500\). Selected factors are \(1.25\) from 24 to 36 months and \(1.04\) from 36 months to ultimate. Find its age-to-ultimate factor and unpaid claims. (Answer: factor \(1.30\); unpaid claims \(1{,}308\))
- A triangle has Year R incurred \(900\) at 12 months and paid \(400\); Year S incurred \(1{,}050\) at 24 months and paid \(700\). Selected factors are \(1.50\) from 12 to 24 months and \(1.20\) from 24 months to ultimate. Calculate both ultimates and total unpaid claims. (Answer: Year R ultimate \(1{,}620\), Year S ultimate \(1{,}260\); total unpaid claims \(1{,}780\))
- At a valuation date, a hypothetical year has incurred \(3{,}000\) and paid \(2{,}100\). Selected remaining factors are \(1.10\) and \(1.05\). Calculate ultimate, total unpaid claims, and projected increase in incurred after the valuation date. (Answer: ultimate \(3{,}465\); unpaid claims \(1{,}365\); future incurred increase \(465\))
- Two accident years have latest incurred values \(1{,}600\) at 24 months and \(2{,}200\) at 12 months, with paid amounts \(1{,}200\) and \(900\), respectively. Selected factors are \(1.25\) from 12 to 24 months and \(1.20\) from 24 months to ultimate. Find each ultimate and aggregate unpaid claims. (Answer: ultimates \(1{,}920\) and \(3{,}300\); aggregate unpaid claims \(3{,}120\))