Chaining Age-to-Age Factors into Ultimate Multipliers

Why Age-to-Ultimate Multipliers Are Useful

A selected set of age-to-age factors describes development in steps: from one development age to the next. To project a cumulative loss amount observed at an intermediate age to ultimate, it is useful to combine all the remaining steps into one number. That number is the age-to-ultimate multiplier. It depends on the age at which the cumulative amount is observed: a more mature accident year generally has fewer selected development steps left to apply.

This tutorial focuses on chaining selected factors and matching the resulting multiplier to each accident year’s observed age. The selection of link ratios is a separate judgment. Here, treat the stated link ratios and any tail-to-ultimate factor as given selections. A factor is a multiplier, so, for example, 1.08 means a 8% increase from the starting cumulative amount over the specified interval.

Definition: Let \(f_{a,b}\) be the selected age-to-age factor from development age \(a\) to the next age \(b\). Let \(f_{T,U}\) be a selected tail-to-ultimate factor from the last explicit development age \(T\) to ultimate. The age-to-ultimate multiplier \(F_a\) is the product of the selected factors from age \(a\) through ultimate. If the remaining explicit ages are \(a_1=a,a_2,\ldots,a_k=T\), then \(F_a=f_{a_1,a_2}f_{a_2,a_3}\cdots f_{a_{k-1},a_k}f_{T,U}\). If no tail factor is selected, the chain ends at the last age treated as ultimate, and no extra factor is included.

For accident year \(j\), write \(C_{j,a}\) for its cumulative loss amount observed at development age \(a\), and \(U_j\) for its projected ultimate loss. Under the chain ladder projection, the selected age-to-ultimate multiplier for that age gives \(U_j=C_{j,a}F_a\). This multiplication is appropriate only when the cumulative amount and the factors refer to the same loss basis and valuation conventions. The factors must also represent the development path after the observation age, not the path from an earlier age.

How the Chain Is Built

Suppose selected factors connect ages 12, 24, 36, and 48 months, with a tail-to-ultimate factor beginning at 48 months. Then the 12-month multiplier includes all three explicit age-to-age factors and the tail; the 24-month multiplier starts with the 24-to-36 factor; and the 48-month multiplier consists only of the tail. An earlier factor is not applied to an amount already observed at a later age.

Locate the observation age.
Identify the development age of the cumulative loss amount being projected. Use that age as the starting point in the factor chain.

List the remaining factors in order.
Include every selected age-to-age factor after the observation age, followed by the tail-to-ultimate factor if one is selected from the last explicit age.

Multiply to obtain the multiplier.
The product is \(F_a\), the single multiplier from the observation age to ultimate.

Project and check.
Multiply the cumulative amount by \(F_a\). Confirm that no factor before the observation age, or factor already included in the product, has been used again.

Why Multiplication Gives the Remaining Development

The product follows from applying each selected factor to the amount produced by the preceding step. For a simple illustration, suppose the selected factors from age \(a\) to \(b\) and from \(b\) to \(c\) are \(f_{a,b}\) and \(f_{b,c}\). If the starting cumulative amount is \(C_a\), then the projected amount at \(b\) is \(C_a f_{a,b}\). Applying the next factor gives the amount at \(c\):

(C_a f_{a,b})f_{b,c}=C_a(f_{a,b}f_{b,c}).

Thus the two sequential steps can be replaced by one multiplier, \(f_{a,b}f_{b,c}\), without changing the projection. Repeating the same argument for each remaining age gives the full product. The tail-to-ultimate factor is another final step in this chain when it starts at the last explicit age. It is included once, and only once.

This reasoning assumes that the selected factors are multiplicative development factors applicable to the cumulative amounts being projected. It does not establish that the selections are appropriate, nor does it account for a change in the definition of the loss measure between ages. If a factor is already an age-to-ultimate factor, rather than an age-to-age factor, multiplying it by the intervening factors again would double count development.

Worked Examples

Worked Example: Constructing Multipliers at Every Age

A hypothetical liability coverage uses selected factors of 1.20 from 12 to 24 months, 1.10 from 24 to 36 months, and 1.04 from 36 to 48 months. A selected tail-to-ultimate factor of 1.03 begins at 48 months. Calculate the age-to-ultimate multiplier at each listed age, and project ultimate for an accident year with 1,500,000 of cumulative loss at 24 months. Use the chain ladder method and report dollar amounts to the cent.

Method. Starting at each age, multiply only the factors remaining from that age through ultimate. At 12 months, all three age-to-age factors and the tail remain:

F_{12}=1.20(1.10)(1.04)(1.03) =1.3200(1.04)(1.03) =1.3728(1.03) =1.413984.

At 24 months, the 12-to-24 factor has already been passed, so it is excluded:

F_{24}=1.10(1.04)(1.03) =1.144(1.03) =1.17832.

At 36 months, only the 36-to-48 factor and the tail remain. At 48 months, only the tail remains:

F_{36}=1.04(1.03)=1.0712, \qquad F_{48}=1.03.

The accident year’s cumulative amount is 1,500,000 at 24 months, so use \(F_{24}\), not \(F_{12}\):

$$ U=1{,}500{,}000(1.17832)=1{,}767{,}480.00. $$

Check. The projection is greater than the observed cumulative amount, as expected for these factors, all of which exceed 1. The factor 1.20 is not applied to a 24-month amount because it describes development that has already elapsed. The common mistake avoided here is applying the full 12-month multiplier to a later-age observation.

Worked Example: Different Accident Years, Different Remaining Multipliers

For a hypothetical property portfolio, selected factors are 1.18 from 12 to 24 months, 1.08 from 24 to 36 months, and 1.03 from 36 to 48 months. The selected tail-to-ultimate factor begins at 48 months and is 1.02. One accident year has 1,000,000 cumulative loss at 12 months; another has 1,050,000 at 24 months. Calculate each multiplier and projected ultimate. Use the chain ladder method and report dollars to the cent.

Method for the 12-month observation. All explicit factors and the tail remain:

F_{12}=1.18(1.08)(1.03)(1.02) =1.2744(1.03)(1.02) =1.312632(1.02) =1.33888464.

Therefore, the first projected ultimate is:

U_1=1{,}000{,}000(1.33888464)=1{,}338{,}884.64.

Method for the 24-month observation. Start with the 24-to-36 factor. Do not include the 12-to-24 factor:

F_{24}=1.08(1.03)(1.02) =1.1124(1.02) =1.134648.

The second projected ultimate is:

U_2=1{,}050{,}000(1.134648)=1{,}191{,}380.40.

Check. The 24-month multiplier is smaller than the 12-month multiplier because it excludes one factor greater than 1. Both calculations include the tail exactly once. A common mistake is to use \(F_{12}\) for both years merely because they share the same selected factor table; the correct multiplier is determined by each year’s own observed age.

Worked Example: Including a Tail Only From Its Stated Starting Age

A hypothetical specialty insurer has selected factors of 1.06 from 36 to 48 months and 1.025 from 48 to 60 months. A tail-to-ultimate factor of 1.015 begins at 60 months. One accident year has 2,400,000 cumulative loss at 36 months, and a more mature year has 2,500,000 at 48 months. Find both age-to-ultimate multipliers and projected ultimates using the chain ladder method. Report dollars to the cent.

At 36 months. The remaining path includes 36-to-48, 48-to-60, and the tail:

F_{36}=1.06(1.025)(1.015) =1.0865(1.015) =1.1027975.

So the projected ultimate for the 36-month observation is:

U_{36}=2{,}400{,}000(1.1027975) =2{,}646{,}714.00.

At 48 months. The 36-to-48 factor describes development before the observation age and is not included. The remaining factors are 48-to-60 and the tail:

F_{48}=1.025(1.015)=1.040375, \qquad U_{48}=2{,}500{,}000(1.040375)=2{,}600{,}937.50.

Check. Both multipliers include the tail factor 1.015 once because both observation ages precede its starting age of 60 months. The 48-month multiplier is lower because its remaining path has one fewer explicit age-to-age factor. The common mistake avoided is treating the tail as though it begins at every observed age; it begins only at the age specified in the selection.

Reading the Factor Chain Correctly

A convenient way to prevent age-index errors is to write the starting age beside every multiplier before calculating. For example, when factors are selected at 12, 24, 36, and 48 months, a 24-month observation starts with the factor labeled 24-to-36, not the factor labeled 12-to-24. The last factor in the chain must also be read carefully: a tail-to-ultimate factor from 48 months is not an extra 48-to-ultimate factor to apply after an already computed 48-to-ultimate multiplier.

Observation age	First factor included	Factors excluded as already elapsed
12 months	12-to-24	None from the listed chain
24 months	24-to-36	12-to-24
36 months	36-to-48	12-to-24 and 24-to-36
48 months	Tail beginning at 48 months, if selected	All earlier age-to-age factors

If a development age is already treated as ultimate, its age-to-ultimate multiplier is 1. If a tail-to-ultimate factor is selected from that age, however, the multiplier at that age is the selected tail factor instead. These are different conventions about where ultimate is being taken; follow the stated factor chain rather than inserting an unstated factor.

Common Mistakes and Checks

The arithmetic is usually straightforward; most errors come from applying the right number at the wrong age or using a factor twice. Before finalizing a projection, check the following:

Starting too early. A cumulative amount observed at 24 months does not need the 12-to-24 factor. Include only factors after its observation age.
Stopping too soon. If a tail-to-ultimate factor is selected from the last explicit age, it belongs in the multiplier for every observation age that has not yet reached ultimate.
Applying the tail twice. Once the tail has been included in \(F_a\), do not apply it again to the projected ultimate.
Confusing a factor with a percentage increase. A factor of 1.04 means multiply by 1.04; it does not mean multiply by 0.04 or add 1.04 to the cumulative amount.
Rounding intermediate products too soon. Retain sufficient precision through the product and round the final dollar amount to the requested precision.

For instance, in the first worked example, a wrong calculation for the 24-month observation would be \(1{,}500{,}000(1.20)(1.10)(1.04)(1.03)\), which gives 2,120,976.00. That calculation wrongly includes the 12-to-24 factor. The correct remaining multiplier is \(1.10(1.04)(1.03)=1.17832\), so the correct projection is 1,767,480.00. The amount and factor chain should always be paired at the same starting age.

Key takeaway: An age-to-ultimate multiplier is the product of the selected factors that remain after the cumulative loss observation age, including a tail-to-ultimate factor when its starting age is still ahead. Different accident years can therefore use different multipliers from the same selected factor chain.

Check Your Understanding

For each problem, use the stated selected factors and include only the factors remaining from the observation age. Treat all amounts and scenarios as hypothetical.

Selected factors are 1.15 from 12 to 24 months, 1.08 from 24 to 36 months, and a tail-to-ultimate factor of 1.04 from 36 months. Find \(F_{12}\). (Answer: 1.29168)
Selected factors are 1.07 from 24 to 36 months, 1.03 from 36 to 48 months, and a tail-to-ultimate factor of 1.02 from 48 months. Find \(F_{24}\). (Answer: 1.124142)
A cumulative amount of 750,000 is observed at 36 months. The selected 36-to-48 factor is 1.05, and the tail from 48 months is 1.04. Find the projected ultimate to the cent. (Answer: 819,000.00)
Selected factors are 1.20 from 12 to 24 months, 1.10 from 24 to 36 months, and a tail from 36 months of 1.05. One accident year has 500,000 at 12 months and another has 600,000 at 24 months. Find both projected ultimates to the cent. (Answer: 693,000.00 for each accident year)
Selected factors are 1.12 from 12 to 24 months, 1.06 from 24 to 36 months, and 1.03 from 36 to 48 months, followed by a tail from 48 months of 1.02. A cumulative amount of 1,400,000 is observed at 24 months. Find its remaining multiplier and projected ultimate to the cent. (Answer: multiplier 1.113636; ultimate 1,559,090.40)

What You'll Learn