Development Patterns for Ceded Recoverables

Why Ceded Recoverables Need Their Own Development Assumptions

A ceded recoverable does not necessarily emerge on the same schedule as the underlying gross loss. A claim may be reported to the insurer before the reinsurer has enough information to confirm its share; a recovery may require a separate submission or agreement; and a large loss may take longer to evaluate than routine claims. As a result, gross loss development factors are not automatically suitable for ceded recoverables.

The aim is to estimate the ultimate ceded amount consistently with the treaty and the data being analyzed, then compare that estimate with amounts paid or recognized to date. The key judgment is not simply which factor is largest or most recent. It is whether the selected development pattern plausibly represents how this particular ceded balance emerges.

Definition: A ceded paid triangle records cumulative ceded loss payments by origin period and development age. A ceded incurred triangle records cumulative ceded paid amounts plus ceded case reserves on the same basis. An ultimate ceded recoverable is the estimated total ceded loss amount for an origin period after all relevant development. At a valuation date, total outstanding recoverables equal estimated ultimate ceded recoverables less cumulative ceded paid recoveries. IBNR on an incurred basis is estimated ultimate ceded recoverables less ceded incurred recoverables to date.

Use a consistent measure throughout: for example, do not use a triangle of recoveries actually collected in cash for one calculation and a triangle of ceded amounts recognized as recoverable for another without adjusting for the difference. Also confirm that the data use consistent definitions of loss, origin period, and development age. The earlier material on gross, ceded, and net loss triangles provides the basic triangle framework; here, the emphasis is on choosing assumptions that suit ceded emergence.

Age-to-Age Factors and the Outstanding Balance

Let \(C_{i,j}\) denote cumulative ceded paid recoveries for origin period \(i\), observed at development age \(j\). A simple volume-weighted age-to-age factor from age \(j\) to age \(j+1\) is the sum of observed amounts at age \(j+1\) divided by the sum of the corresponding amounts at age \(j\). The cumulative development factor to ultimate is the product of the selected age-to-age factors from the current age onward.

f_j=\frac{\sum_i C_{i,j+1}}{\sum_i C_{i,j}}, \qquad F_j=\prod_{k=j}^{n-1} f_k, \qquad \widehat{U}_{i}=C_{i,j}F_j

Here, \(f_j\) is the selected factor from age \(j\) to the next age, \(F_j\) is the cumulative factor from age \(j\) to ultimate, and \(\widehat{U}_{i}\) is estimated ultimate ceded paid for origin period \(i\). In practice, selected factors need not be the raw volume-weighted factors: they should be reviewed for relevance, stability, and unusual recoveries.

The outstanding estimate follows by subtracting paid recoveries already recorded. This relationship is a direct reconciliation: ultimate consists of amounts already paid plus amounts still outstanding.

\widehat{\text{Outstanding}}_i=\widehat{U}_i-C_{i,j}

For incurred triangles, use the same logic to estimate ultimate from incurred to date. Then distinguish IBNR from total outstanding: ultimate less incurred is IBNR, while ultimate less paid includes both case reserves and IBNR.

Why the Development Calculation Works

The chain-ladder calculation assumes that, for the portfolio being projected, the relationship between cumulative amounts at successive development ages is reasonably stable. If \(C_{i,j+1}\) is expected to be approximately \(f_j C_{i,j}\), repeating that relationship for each future age gives the product of factors. Multiplying current cumulative amounts by that product therefore projects the amount at ultimate.

C_{i,j+2}\approx f_{j+1}C_{i,j+1} \approx f_{j+1}f_jC_{i,j} \quad\Longrightarrow\quad \widehat{U}_i\approx C_{i,j}\prod_{k=j}^{n-1}f_k

This is a projection assumption, not a guarantee. It is less credible when treaty terms, claim handling, the mix of recoveries, or the definition of the data change. A factor may also be distorted by a single large recovery. In those situations, the analyst should explain why the selected pattern remains appropriate or use a more suitable segmentation or estimate.

Worked Example: Projecting a Ceded Paid Triangle

A hypothetical reinsurer’s recoveries are summarized below in thousands of dollars. The rows are origin years; the columns are cumulative development ages. For this illustration, select the volume-weighted factors shown by the mature diagonal data. Interest rate, mortality assumption, and probability distribution are not applicable: this is a deterministic development calculation, not a discounted or probabilistic model.

Origin year	12 months	24 months	36 months
Year A	400	600	720
Year B	500	750	—
Year C	600	—	—

Method: paid chain ladder. First calculate the 12-to-24-month factor using the two origin years with observations at both ages:

f_{12\to24}=\frac{600+750}{400+500} =\frac{1{,}350}{900}=1.5

Next calculate the 24-to-36-month factor from the origin year observed at both ages:

f_{24\to36}=\frac{720}{600}=1.2

Year A is already observed at 36 months, so its ultimate is 720. Year B’s estimated ultimate at 24 months is \(750(1.2)=900\). Year C’s cumulative factor from 12 months is \(1.5(1.2)=1.8\), so its estimated ultimate is \(600(1.8)=1{,}080\).

\text{Total paid to date}=720+750+600=2{,}070 $$ $$ \text{Total estimated ultimate}=720+900+1{,}080=2{,}700 $$ $$ \text{Total outstanding}=2{,}700-2{,}070=630

Check. Year B has \(900-750=150\) outstanding and Year C has \(1{,}080-600=480\); \(150+480=630\). The result reconciles. The common mistake avoided is applying the 12-to-24-month factor to Year B, which is already at 24 months: that would develop the same period twice.

Incurred Development and the Meaning of Outstanding

Ceded incurred patterns can be useful when case estimates provide information about amounts not yet paid. However, case estimates may be revised, and they may not be equally reliable across claim sizes or stages of investigation. A ceded incurred triangle therefore needs its own review; it should not automatically inherit gross incurred factors.

Worked Example: Separating Case Reserves from IBNR

A hypothetical ceded incurred triangle, in thousands of dollars, has the observations below. At the valuation date, Year B is at 24 months and Year C is at 12 months. For Year C, ceded paid recoveries are 400. For Year B, ceded paid recoveries are 700. Interest rate, mortality assumption, and probability distribution are not applicable; the calculation is deterministic.

Origin year	12-month incurred	24-month incurred	36-month incurred
Year A	800	960	1,000
Year B	900	1,080	—
Year C	1,000	—	—

Method: incurred chain ladder. The 12-to-24-month factor is \((960+1{,}080)/(800+900)=2{,}040/1{,}700=1.2\). The 24-to-36-month factor is \(1{,}000/960=25/24\), or approximately 1.0417. Retain \(25/24\) in the calculation to avoid intermediate rounding.

\widehat{U}_{B}=1{,}080\left(\frac{25}{24}\right)=1{,}125 $$ $$ \widehat{U}_{C}=1{,}000(1.2)\left(\frac{25}{24}\right)=1{,}250

For Year B, estimated IBNR is ultimate less incurred to date: \(1{,}125-1{,}080=45\). Its case reserve is incurred less paid, or \(1{,}080-700=380\). Therefore total outstanding is \(45+380=425\), which also equals \(1{,}125-700\).

For Year C, estimated IBNR is \(1{,}250-1{,}000=250\). Its case reserve is \(1{,}000-400=600\), giving total outstanding of \(250+600=850\). Across both years, total outstanding is \(425+850=1{,}275\).

Check. The total outstanding also equals total estimated ultimate less total paid: \((1{,}125+1{,}250)-(700+400)=1{,}275\). The common mistake avoided is calling ultimate less paid “IBNR”; that amount includes case reserves as well as IBNR.

Diagnosing Differences from Gross Loss Emergence

A ceded pattern may differ from a gross pattern for operational or contractual reasons. The important question is whether the difference is supported by the data and the recovery process, rather than whether the ceded factor happens to be higher or lower.

Separate reporting or review. The insurer may report a loss before it submits a detailed recovery calculation. Review or agreement by the reinsurer can delay recognition or payment of the ceded amount.
Different claim mix. Ceded recoveries may be concentrated in large or complex claims, while gross losses include many smaller claims. Those groups can develop at different speeds.
Layer and attachment effects. A loss must reach the relevant attachment for a layer to respond. Early estimates of gross loss may not yet indicate the ultimate ceded amount reliably.
Case reserve changes. Changes to gross case estimates do not always flow proportionally into ceded incurred estimates, particularly when the ceded share depends on claim size or coverage terms.
Payment and recording practices. A triangle based on cash collected may develop differently from one based on recovery amounts recorded as due. The valuation basis must be understood before selecting a pattern.

A sound review compares ceded factors with gross development as a diagnostic, not as a rule that ceded factors must match gross factors. Investigate large individual recoveries, inconsistent observations, and changes in the mix of recovery types. If distinct components have materially different emergence patterns, separate their projections when the data permit.

Worked Example: Why a Single Ceded Pattern Can Mislead

A hypothetical program has two ceded components, both measured in thousands of dollars. A quota share component has estimated ultimate recoveries of 600, of which 420 have been paid. An excess-of-loss component has estimated ultimate recoveries of 400, of which 80 have been paid. These component ultimate estimates are supplied for this example; the task is to estimate outstanding and assess whether one component’s paid pattern can be applied to the other. Interest rate, mortality assumption, and probability distribution are not applicable.

Method: project and reconcile by component. For the quota share component, the paid proportion is \(420/600=0.70\). Its implied cumulative factor to ultimate is \(600/420=10/7\), approximately 1.4286. Its outstanding amount is \(600-420=180\).

For the excess-of-loss component, the paid proportion is \(80/400=0.20\). Its implied cumulative factor is \(400/80=5\). Its outstanding amount is \(400-80=320\).

\text{Total paid}=420+80=500, \qquad \text{Total ultimate}=600+400=1{,}000 $$ $$ \text{Total outstanding}=180+320=500

If the quota share factor \(10/7\) were applied to all 500 paid, estimated ultimate would be \(500(10/7)=714.2857\), or 714.29 when rounded to two decimals. That would imply outstanding of \(714.29-500=214.29\), rather than the component-based 500. The reason is that the excess-of-loss component has a much smaller paid proportion, so the quota share pattern is not representative of the combined balance.

Check. The component outstanding estimates sum to \(180+320=500\), matching total ultimate less total paid. The common mistake avoided is treating a development factor from one recovery component as universal when the components have different emergence patterns.

Selection Judgments and Exam Traps

A development factor is useful only to the extent that its data and assumptions fit the estimate. Before selecting a ceded pattern, verify that the triangle is cumulative and consistently valued; inspect the underlying observations for large or unusual recoveries; and consider whether the selected factor reflects the current mix of recoveries and claim handling. A factor derived from a small number of observations deserves particular caution.

A frequent exam error is to use a gross factor simply because the gross triangle is more complete. That calculation may be arithmetically correct but conceptually unsupported. Another is confusing a paid development factor with an incurred development factor. Paid triangles project payments; incurred triangles project paid amounts plus case reserves. Their ultimates may be comparable, but the path to ultimate and the interpretation of the outstanding balance differ.

Key takeaway: Estimate ceded recoverables using a development pattern that matches the ceded data and recovery process. For a paid estimate, subtract paid recoveries from estimated ultimate to obtain total outstanding. For an incurred estimate, ultimate less incurred is IBNR, while ultimate less paid is total outstanding, including case reserves.

Check Your Understanding

For each problem, use the stated development assumptions and keep the calculation basis consistent.

A ceded paid amount of 720 at 12 months is subject to selected factors of 1.25 from 12 to 24 months and 1.10 from 24 months to ultimate. Calculate ultimate and total outstanding at 12 months. (Answer: ultimate = 990; outstanding = 270)
A ceded incurred amount is 1,400, and paid recoveries are 900. Apply a selected incurred factor of 1.15 to estimate ultimate. Find IBNR, case reserves, and total outstanding. (Answer: ultimate = 1,610; IBNR = 210; case reserves = 500; total outstanding = 710)
A proportional recovery component has ultimate 800 and paid 600; a separate layer has ultimate 500 and paid 100. Calculate total paid, total ultimate, and total outstanding. (Answer: paid = 700; ultimate = 1,300; outstanding = 600)
A ceded paid triangle has cumulative amounts 300 and 450 at ages 12 and 24 for one origin year, and 500 and 700 for another. Calculate the volume-weighted 12-to-24-month factor. If a third origin year has 12-month paid of 400 and the selected 24-to-36-month factor is 1.20, estimate its ultimate using the factor you calculated. (Answer: 12-to-24 factor = 1.4375; ultimate = 690)
At the valuation date, a ceded incurred estimate is 2,250, paid recoveries are 1,500, and estimated ultimate is 2,700. Calculate case reserves, IBNR, and total outstanding, and check the reconciliation. (Answer: case reserves = 750; IBNR = 450; total outstanding = 1,200; 750 + 450 = 2,700 − 1,500)

What You'll Learn