Derive Cumulative Development Factors From Link Ratios

Why Cumulative Factors Are Products

A link ratio describes development from one valuation age to the next. A cumulative development factor describes development across all the remaining ages to ultimate. The distinction matters because a loss amount develops through a sequence of intervals: the factor for the first interval acts on the amount at the starting age, and each later factor acts on the amount after the preceding development.

For example, a cumulative loss at 12 months may develop to 24 months, then to 36 months, and then to ultimate. If the selected factors for those transitions are 1.40, 1.20, and 1.05, the cumulative factor from 12 months is their product, not their sum. The algebra behind that rule is a telescoping ratio: each intermediate cumulative loss appears once in a numerator and once in a denominator, so the intermediate terms cancel.

The index is important. \(f_j\) takes a cumulative amount at age \(j\) to the next age, while \(F_j\) starts at age \(j\) and includes every selected transition from that age onward. Therefore, \(F_j\) does not include factors for development that has already occurred before age \(j\). If development ages are labeled 12, 24, and 36 months, the factor from 12 to 24 months is the first transition in the product from 12 months; it is not included in the product from 24 months.

Deriving the Product Formula

The product relationship follows directly from the definition of an age-to-age ratio. First consider one accident period with nonzero cumulative losses at each age. Its exact link ratio from age \(k\) to age \(k+1\) is \(C_{i,k+1}/C_{i,k}\). Multiplying the successive ratios from age \(j\) through age \(n-1\) gives:

Every intermediate cumulative value \(C_{i,j+1},\ldots,C_{i,n-1}\) cancels: each appears in a numerator in one ratio and a denominator in the next. Only the final cumulative loss \(C_{i,n}\) and the starting cumulative loss \(C_{i,j}\) remain. Thus, for a single row of exact link ratios, the product from age \(j\) to age \(n\) equals the direct ratio of the row's cumulative loss at age \(n\) to its cumulative loss at age \(j\).

In practice, the selected link ratios \(f_j\) may be chosen from data using a selection procedure rather than being the exact ratios from one accident-period row. The cumulative factor is still defined as the product of the selected factors. However, that selected product need not equal the direct ratio \(C_{i,n}/C_{i,j}\) for any particular row. The telescoping identity explains the structure of the product; it does not imply that separately selected factors reproduce every row's observed development exactly.

A tail factor represents additional development beyond the last explicitly modeled age. It is included once when the destination is ultimate beyond that age. If no tail factor is specified and the last age is treated as ultimate, use a terminal factor of 1.00; do not silently add a tail.

Calculating Factors From Different Starting Ages

A useful way to check the indexing is to write the factor for each starting age separately. Suppose the selected links are \(f_{12}\) from 12 to 24 months, \(f_{24}\) from 24 to 36 months, and \(f_{36}\) from 36 months to ultimate. Then \(F_{12}\) includes all three, \(F_{24}\) includes only the last two, and \(F_{36}\) includes only the final transition. The cumulative factor from a later age is generally shorter because less development remains.

Exact Row Ratios and Selected Factors

When link ratios come from one row of cumulative losses, the product can be checked against the direct ratio between the row's ending and starting cumulative values. This is a useful arithmetic check. When factors are selected across multiple rows, the product is still the cumulative factor implied by those selections, but the direct-ratio check applies only if the ratios being multiplied come from that same row.

Including a Tail Factor

A tail factor is handled as one more multiplicative component after the last modeled development age. The link ratios carry development from the starting age through the last stated age; the tail factor carries development from that last age to ultimate. The starting age determines which link ratios to include, but it does not change whether the tail is needed.

Common Mistakes and a Reliable Procedure

The most frequent arithmetic error is adding link ratios. Link ratios describe successive proportional changes, so they compose by multiplication. Another error is using a factor that begins before the starting age. A cumulative factor from 24 months cannot include the transition from 12 to 24 months, because the 24-month cumulative amount is already measured after that transition.

For the links 1.40, 1.20, and 1.05, an incorrect calculation would be:

The correction is to multiply the sequential factors:

This result is also consistent with the interpretation of the factors: three successive increases of 40%, 20%, and 5% do not combine into a 265% increase from the original amount. Each percentage change applies to the amount after the preceding change. Likewise, using \(1.40(1.20)(1.05)\) as the factor from 24 months would be an indexing error; the correct factor from 24 months is \(1.20(1.05)=1.26\).

Cumulative factors summarize the remaining sequence of development in one number. Their construction is simple once the starting and ending ages are explicit, but that simplicity depends on consistent indexing and a clear endpoint. These checks are aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.