Tutorials › CAS Exam 5 - Basic Ratemaking and Estimated Claim Liabilities › Deriving the Chain Ladder Projection Formula

CAS Exam 5 · Chain Ladder Estimation · Tutorial 680 of 989

Deriving the Chain Ladder Projection Formula

Learn why chain ladder projections use a product of age-to-age factors and how to apply that product to estimate ultimate losses and reserves.

Intermediate 10 min read

What You'll Learn

  • Define cumulative losses, selected age-to-age factors, and the ages used in a projection
  • Derive the cumulative development factor as a product of successive selected factors
  • Project an accident year from its latest observed age to ultimate
  • Incorporate a tail factor without applying already-passed development ages
  • Calculate an estimated reserve as projected ultimate less the latest cumulative loss
  • Identify common errors in factor order, factor multiplication, and starting age

Why the Projection Formula Is a Product

A chain ladder projection starts with a cumulative loss amount observed at a particular development age and carries it forward using selected age-to-age factors. The key calculation is not a new selection method: it is the algebra that combines the selected factors into one multiplier from the current age to ultimate. That algebra also clarifies how to handle accident years that are observed at different ages.

The method can be applied to cumulative paid losses or cumulative incurred losses, provided the selected factors and the cumulative amounts are on a consistent basis. The examples use hypothetical data. This tutorial is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.

Definition: Let \(C_{i,j}\) be cumulative losses for accident year \(i\) evaluated at development age \(j\). Let \(f_j\) be the selected age-to-age factor from age \(j\) to the next age \(j+1\). A factor is applied by multiplication: a projection from age \(j\) to age \(j+1\) is \(C_{i,j+1}=C_{i,j}f_j\). A selected tail factor is a multiplier from the last stated development age to ultimate.

The projection formula is a deterministic procedure conditional on the selected factors. For it to be meaningful, the selected factors must correspond to the development ages being traversed, the cumulative amounts must use a consistent loss basis, and the factors must be applicable to the accident year being projected. The standard calculation also presumes usable cumulative values; zero or negative cumulative amounts can make ratio-based factors or their interpretation problematic. The algebra below does not establish that the factors are appropriate or that future development will follow the selected pattern.

Deriving the Cumulative Development Factor

Suppose an accident year is currently observed at age \(k\), and the projection proceeds through ages \(k+1,\ldots,n\). Each successive factor acts on the amount produced by the prior step. Therefore, the factors compound rather than add.

1
Project one age.
At age \(k+1\), the projected cumulative loss is \(C_{i,k}f_k\).
2
Project the next age.
At age \(k+2\), multiply the age-\(k+1\) projection by \(f_{k+1}\), giving \(C_{i,k}f_kf_{k+1}\).
3
Continue to the final age.
Each new factor multiplies the amount already projected, so the full multiplier is the product of every factor from the current age through the final projection age.
$$ \begin{aligned} C_{i,k+1} &= C_{i,k}f_k,\\ C_{i,k+2} &= C_{i,k+1}f_{k+1} = C_{i,k}f_kf_{k+1},\\ C_{i,n} &= C_{i,k}\prod_{j=k}^{n-1}f_j. \end{aligned} $$

If the final step is a selected tail factor \(f_{\text{tail}}\), append it once to the product. If the latest observed age is already the age immediately before the tail, do not include earlier age-to-age factors.

Theorem: Under the selected-factor projection procedure, the projected ultimate for accident year \(i\), observed at age \(k\), is its latest cumulative loss multiplied by the product of selected factors for all development steps still to be projected. If the ultimate projection includes a tail factor, that factor is included once, after the factors for the stated development ages.

The reserve estimate on the same cumulative basis is the projected ultimate less the latest observed cumulative amount. For paid losses, this difference estimates unpaid losses. For incurred losses, it is the difference between projected ultimate and current incurred losses; it should not be described as unpaid paid losses without a separate reconciliation.

$$ \widehat{C}_{i,\text{ult}} = C_{i,k}\left(\prod_{j=k}^{n-1}f_j\right)f_{\text{tail}}, \qquad \widehat{R}_i = \widehat{C}_{i,\text{ult}}-C_{i,k}. $$

When no tail factor is used, omit \(f_{\text{tail}}\). When the selected factors already include a final factor to ultimate, do not append a second tail factor. The notation \(\widehat{C}_{i,\text{ult}}\) denotes an estimated ultimate, not an observed value.

Worked Projections

Worked Example: Projecting from an Intermediate Age

A hypothetical portfolio has cumulative incurred losses of $2,400 thousand for one accident year at 24 months. The selected factors are 1.20 from 24 to 36 months, 1.10 from 36 to 48 months, and a tail factor of 1.05 from 48 months to ultimate. Calculate the projected cumulative losses at each subsequent age, the ultimate, and the estimated reserve. Use thousands of dollars throughout.

Method and starting point. This is a selected-factor chain ladder projection. The latest observation is at 24 months, so the 12-to-24-month factor, if one existed, would not be used.

Project to 36 months. Multiply the observed amount by the 24-to-36-month factor:

$$ 2{,}400(1.20)=2{,}880. $$

Project to 48 months. Apply the next factor to the 36-month projection:

$$ 2{,}880(1.10)=3{,}168. $$

Project to ultimate. Apply the tail factor once:

$$ 3{,}168(1.05)=3{,}326.4. $$

Equivalently, the cumulative multiplier from 24 months to ultimate is \(1.20(1.10)(1.05)=1.386\), and \(2{,}400(1.386)=3{,}326.4\). The estimated reserve is \(3{,}326.4-2{,}400=926.4\) thousand dollars. Check: the amount increases at each step because all three multipliers exceed 1, and the reserve plus observed cumulative loss equals the projected ultimate: \(926.4+2{,}400=3{,}326.4\). The common mistake avoided is applying a factor for development before the latest observed age.

Worked Example: Projection with a Tail Factor

A hypothetical accident year has cumulative paid losses of $1,750 thousand at 36 months. Selected factors are 1.08 from 36 to 48 months, 1.04 from 48 to 60 months, and 1.03 from 60 months to ultimate. Find the cumulative development factor from 36 months to ultimate, the projected ultimate, and the estimated unpaid reserve.

Method. Multiply the factors for the two remaining stated development intervals and then the tail. The starting age is 36 months, so the calculation begins with the 36-to-48-month factor.

Combine the factors.

$$ 1.08(1.04)=1.1232, \qquad 1.1232(1.03)=1.156896. $$

Thus, the cumulative development factor from 36 months to ultimate is 1.156896. The successive projections provide a check on the product:

$$ \begin{aligned} C_{48} &= 1{,}750(1.08)=1{,}890,\\ C_{60} &= 1{,}890(1.04)=1{,}965.6,\\ \widehat{C}_{\text{ult}} &= 1{,}965.6(1.03)=2{,}024.568. \end{aligned} $$

The estimated unpaid reserve is \(2{,}024.568-1{,}750=274.568\) thousand dollars. In dollars, that is $274,568. Check: \(1{,}750(1.156896)=2{,}024.568\), matching the sequential calculation. The common mistake avoided is adding the factors or treating the tail as a replacement for the 60-to-ultimate step already specified; here the tail is precisely that final step.

Worked Example: Accident Years at Different Development Ages

A hypothetical paid-loss triangle has three accident years with the following latest cumulative observations: Accident Year A is at 24 months with $1,400 thousand; Accident Year B is at 36 months with $1,512 thousand; Accident Year C is at 48 months with $1,478.4 thousand. The selected factors are 1.20 from 24 to 36 months, 1.10 from 36 to 48 months, and 1.05 from 48 months to ultimate. Project each ultimate, then calculate the combined estimated reserve relative to the sum of the three latest observed amounts.

Method. Apply only factors for ages later than each accident year's latest observation. Accident Year A needs all three factors; B needs the 36-to-48-month factor and tail; C needs only the tail.

Accident Year A, observed at 24 months:

$$ \begin{aligned} C_{A,36} &= 1{,}400(1.20)=1{,}680,\\ C_{A,48} &= 1{,}680(1.10)=1{,}848,\\ \widehat{C}_{A,\text{ult}} &= 1{,}848(1.05)=1{,}940.4. \end{aligned} $$

Accident Year B, observed at 36 months:

$$ \begin{aligned} C_{B,48} &= 1{,}512(1.10)=1{,}663.2,\\ \widehat{C}_{B,\text{ult}} &= 1{,}663.2(1.05)=1{,}746.36. \end{aligned} $$

Accident Year C, observed at 48 months:

$$ \widehat{C}_{C,\text{ult}}=1{,}478.4(1.05)=1{,}552.32. $$

The combined projected ultimate is \(1{,}940.4+1{,}746.36+1{,}552.32=5{,}239.08\) thousand dollars. The latest observed amounts total \(1{,}400+1{,}512+1{,}478.4=4{,}390.4\) thousand dollars. Therefore, the combined estimated reserve is \(5{,}239.08-4{,}390.4=848.68\) thousand dollars. Check: the accident-year reserve estimates are \(540.4\), \(234.36\), and \(73.92\) thousand dollars; their sum is \(848.68\). The common mistake avoided is applying the same full factor product to every accident year, regardless of its latest observed age.

Why the Product Form Matters

The product follows directly from successive conditional steps in the projection procedure. If an amount is increased by 20% and then increased by 10%, the second increase applies to the already increased amount. Algebraically, the total multiplier is \(1.20(1.10)=1.32\), not \(1+0.20+0.10=1.30\). Chain ladder factors work the same way: each age-to-age factor acts on the cumulative loss resulting from the prior age.

This also explains why the selected factors must be aligned to the accident year's maturity. An accident year observed at 36 months has already passed through earlier development intervals. Applying those earlier factors again would project from the wrong starting point and overstate the ultimate whenever those factors exceed 1.

Common exam trap: Do not add age-to-age factors, multiply by factors from ages already observed, or apply a tail factor twice. For a current amount \(C_{i,k}\), write down the sequence of remaining development ages first; then multiply exactly those selected factors, in order.

Check Your Understanding

For each hypothetical projection, identify the applicable factor sequence before calculating. Amounts stated in thousands are in thousands of dollars.

  1. A cumulative incurred amount is $640 thousand at 24 months. The selected factors are 1.25 from 24 to 36 months and 1.08 from 36 months to ultimate. Find the projected ultimate. (Answer: $864 thousand)
  2. A cumulative paid amount is $500 thousand at 12 months. The factors are 1.30 from 12 to 24 months, 1.15 from 24 to 36 months, and 1.04 from 36 months to ultimate. Find the cumulative factor to ultimate and the projected ultimate. (Answer: factor 1.5548; ultimate $777.40 thousand)
  3. A hypothetical accident year has $2.4 million of cumulative losses at 36 months. Factors are 1.12 from 36 to 48 months, 1.06 from 48 to 60 months, and 1.03 from 60 months to ultimate. Find the projected ultimate and estimated reserve. (Answer: ultimate $2,934,758.40; reserve $534,758.40)
  4. Accident Year A has cumulative losses of $720 thousand at 24 months, while Accident Year B has $880 thousand at 36 months. Selected factors are 1.20 from 24 to 36 months and 1.10 from 36 months to ultimate. Find each ultimate and the combined reserve. (Answer: A ultimate $950.40 thousand; B ultimate $968 thousand; combined reserve $318.40 thousand)
  5. A cumulative paid amount is $1,275,000 at 48 months. The factor from 48 to 60 months is 1.04, and the tail from 60 months to ultimate is 1.025. Find the total factor to ultimate and the estimated reserve. (Answer: factor 1.066; reserve $84,150)
Clinical Biostats