Projecting Ultimate Losses with Chain Ladder

From Latest Cumulative Loss to Ultimate Loss

A loss development triangle contains accident periods at different stages of maturity. Some periods have several years of observed cumulative losses; others have only an early valuation. Chain ladder puts these periods on a comparable basis by projecting each period’s latest cumulative loss to its estimated ultimate loss. The essential calculation is to multiply the latest observed cumulative amount by the cumulative development factor for that observation’s age.

This tutorial treats the selected age-to-age and cumulative development factors as given. Its focus is how to apply the cumulative factor to the correct accident period and development age, and how to check the resulting estimate.

Definition: Let $C_{i,j}$ be the cumulative loss for accident period $i$ observed at development age $j$. The latest cumulative loss for accident period $i$ is the cumulative amount at its most recent observed age, denoted $C_{i,j_i}$. Its projected ultimate loss, denoted $\widehat{U}_i$, is the estimate of that accident period’s cumulative loss at ultimate. If $F_{j_i\to\mathrm{ult}}$ is the selected cumulative development factor from age $j_i$ to ultimate, then $$ \widehat{U}_i=C_{i,j_i}F_{j_i\to\mathrm{ult}}. $$

A cumulative development factor is an age-specific multiplier: it expresses the selected ratio of ultimate loss to cumulative loss at a particular development age. In the notation used here, $F_{j\to\mathrm{ult}}$ means “from age $j$ through to ultimate.” If selected age-to-age factors are $f_{12\to24}$, $f_{24\to36}$, and $f_{36\to\mathrm{ult}}$, for example, the corresponding factor from age 12 to ultimate is their product:

F_{12\to\mathrm{ult}} =f_{12\to24}f_{24\to36}f_{36\to\mathrm{ult}}.

The required cumulative factor depends on the latest age available for the accident period. A 12-month observation must be paired with the 12-month-to-ultimate factor, not with the factor for a more mature period. Once the matching factor is known, the calculation is direct.

Applying the Factor to Each Accident Period

The procedure is the same for each row, even when rows have different latest observed ages. First identify the latest cumulative amount and its development age. Then use the cumulative factor from that age to ultimate. Finally, multiply and retain the units of the original cumulative loss.

Locate the latest observed cell.
For the accident period in question, identify the most recent available cumulative loss and record its development age.

Match the age to its factor.
Choose the cumulative development factor that begins at that exact observed age and ends at ultimate.

Multiply and check.
Calculate latest cumulative loss times the matching factor. If the factor is greater than 1, the projected ultimate should exceed the latest cumulative loss.

Worked Example: Projecting Three Accident Periods at Different Ages

A hypothetical insurer has selected the following factors: 12-to-24 months $1.50$, 24-to-36 months $1.20$, and 36 months to ultimate $1.10$. The latest cumulative losses are: accident period A, $1{,}080$ at 36 months; B, $900$ at 24 months; and C, $600$ at 12 months. Estimate each period’s ultimate loss. Amounts are in thousands of dollars.

Method. Calculate the cumulative factor from each row’s latest age to ultimate, then multiply by that row’s latest cumulative loss.

For accident period A, the latest age is 36 months, so only the 36-month-to-ultimate factor applies:

F_{36\to\mathrm{ult}}=1.10,\qquad \widehat{U}_A=1{,}080(1.10)=1{,}188.

For B, the latest age is 24 months. Its factor includes development from 24 to 36 months and then from 36 months to ultimate:

F_{24\to\mathrm{ult}}=1.20(1.10)=1.32, \qquad \widehat{U}_B=900(1.32)=1{,}188.

For C, the latest age is 12 months. Its factor includes all three stages:

F_{12\to\mathrm{ult}}=1.50(1.20)(1.10)=1.98, \qquad \widehat{U}_C=600(1.98)=1{,}188.

Thus the projected ultimates are $1{,}188$ thousand dollars for each of A, B, and C. As a check, the cumulative factors increase for earlier ages, while all three latest amounts and their respective factors produce the same projected ultimate in this constructed example. The common mistake avoided here is applying the 12-month factor to every row merely because it is the longest factor shown; each row must use the factor beginning at its own latest age.

Why the Multiplication Is Appropriate

The formula follows from what the cumulative development factor represents. If $F_{j\to\mathrm{ult}}$ is defined as the selected ultimate-to-age-$j$ ratio, then it scales a cumulative amount at age $j$ to the corresponding estimated amount at ultimate. In ratio form, the relationship is:

F_{j\to\mathrm{ult}} =\frac{\text{selected ultimate amount}}{\text{cumulative amount at age }j}.

Rearranging gives the projection formula:

\text{selected ultimate amount} =\text{cumulative amount at age }j \times F_{j\to\mathrm{ult}}.

This is also a useful units check. The factor is a ratio, so it has no units. Multiplying a cumulative loss in dollars by that ratio leaves an estimate in dollars; a cumulative loss in thousands remains an estimate in thousands. Addition of the factor to the loss would not preserve the intended meaning or units.

The calculation is a projection under the selected development pattern, not a claim that the ultimate amount is already known. It assumes that the selected cumulative factor is appropriate for the accident period and that the cumulative loss and factor use compatible definitions and valuation bases. For example, paid losses should be paired with factors developed for paid losses, and amounts should be expressed in consistent units. If the factor is not appropriate for the period, multiplication will not fix that model choice.

Worked Example: Applying Factors When Amounts Are in Thousands

A hypothetical warranty portfolio has selected age-to-age factors of $1.40$ from 12 to 24 months, $1.15$ from 24 to 36 months, and $1.08$ from 36 months to ultimate. Accident period A has a latest cumulative loss of $1{,}250$ at 24 months; B has $800$ at 12 months; and C has $2{,}100$ at 36 months. All amounts are in thousands of dollars. Find the projected ultimate for each period.

Method. Build the cumulative factor from each period’s observed age to ultimate. The factors are dimensionless; the projected amounts remain in thousands of dollars.

For A, the 24-month factor includes the 24-to-36 and 36-to-ultimate stages:

F_{24\to\mathrm{ult}}=1.15(1.08)=1.242, \qquad \widehat{U}_A=1{,}250(1.242)=1{,}552.5.

For B, include all three stages:

F_{12\to\mathrm{ult}}=1.40(1.15)(1.08) =1.61(1.08)=1.7388,

\widehat{U}_B=800(1.7388)=1{,}391.04.

For C, only the 36-month-to-ultimate stage remains:

F_{36\to\mathrm{ult}}=1.08, \qquad \widehat{U}_C=2{,}100(1.08)=2{,}268.

The ultimate estimates are $1{,}552.5$, $1{,}391.04$, and $2{,}268$ thousand dollars. Each exceeds its latest cumulative loss because each applicable factor is greater than 1. The common mistake avoided is dropping an intermediate stage: B is at 12 months, so its factor must include both 12-to-24 and 24-to-36 development before the final factor.

Common Exam Trap: Starting at the Wrong Age

A factor may be calculated correctly and still be applied incorrectly. The most common error is to use a factor that starts at a different age from the latest observed cumulative amount. Another frequent error is to add link ratios rather than multiply them when forming a cumulative factor.

Worked Example: Correcting an Age-Matching Error

In a hypothetical liability portfolio, the selected factors are $1.25$ from 18 to 30 months, $1.12$ from 30 to 42 months, and $1.05$ from 42 months to ultimate. One accident period has a latest cumulative loss of $2{,}400{,}000$ at 18 months. Estimate its ultimate loss.

Correct method. Starting at 18 months requires all three factors:

F_{18\to\mathrm{ult}}=1.25(1.12)(1.05) =1.40(1.05)=1.47.

\widehat{U}=2{,}400{,}000(1.47) =3{,}528{,}000.

A wrong calculation would use only the final factor, $1.05$, because it is the factor explicitly labelled “to ultimate”:

$$ 2{,}400{,}000(1.05)=2{,}520{,}000. $$

That calculation starts at 42 months, not 18 months, and therefore omits the 18-to-30 and 30-to-42 development. The correct estimate is $3{,}528{,}000$. The check is that the factor used must begin at the latest observed age; a factor labelled 42 months to ultimate cannot project an 18-month observation by itself.

Assumptions and Checks Before Reporting

Chain-ladder ultimate estimates are conditional on the selected development factors and the observed cumulative losses. Before reporting them, verify the following:

Age alignment: the cumulative factor begins at the latest age for that accident period.
Factor construction: all required stages from that age through ultimate are included, and cumulative factors are formed by multiplication.
Consistent basis: the loss measure, valuation basis, currency, and units agree with those underlying the selected factors.
Arithmetic: recompute the product and final multiplication; a factor above 1 should produce an ultimate above the latest cumulative amount.
Interpretation: the result is an estimate of ultimate loss, not a guarantee of the eventual observed amount.

These checks do not establish that a development pattern is suitable; they help ensure that the chosen pattern has been applied correctly. Estimating the factors and assessing whether the data support them are separate actuarial judgments.

Key takeaway: For each accident period, identify its latest cumulative loss and age, select the cumulative development factor from that age to ultimate, and calculate $\widehat{U}_i=C_{i,j_i}F_{j_i\to\mathrm{ult}}$. Never apply one common factor to rows at different development ages without checking that the factor starts at each row’s age.

Check Your Understanding

For each problem, use the stated selected factors and report the projected ultimate in the same units as the latest cumulative loss.

A period’s latest cumulative loss is $720$ at 24 months. The factors are $1.20$ from 24 to 36 months and $1.10$ from 36 months to ultimate. What is its projected ultimate? (Answer: $950.4$)
A period has cumulative loss $400$ thousand dollars at 12 months. The selected factors are $1.30$ from 12 to 24 months, $1.10$ from 24 to 36 months, and $1.05$ from 36 months to ultimate. Find its projected ultimate. (Answer: $600.6$ thousand dollars)
A period’s latest cumulative loss is $950$ dollars at 24 months. The factors are $1.12$ from 24 to 36 months and $1.08$ from 36 months to ultimate. Find the ultimate, rounding to cents. (Answer: $\$1{,}149.12$)
Period A has $300$ thousand dollars at 12 months, and period B has $500$ thousand dollars at 24 months. The selected factors are $1.40$ from 12 to 24 months and $1.20$ from 24 months to ultimate. Find each period’s ultimate and their combined ultimate estimate. (Answer: A $504$, B $600$, combined $1{,}104$ thousand dollars)
Period A has $1{,}250$ thousand dollars at 36 months; period B has $900$ thousand dollars at 24 months. The factors are $1.15$ from 24 to 36 months and $1.06$ from 36 months to ultimate. Find both ultimates and their combined estimate. (Answer: A $1{,}325$, B $1{,}097.1$, combined $2{,}422.1$ thousand dollars)

What You'll Learn