Why Cat Bond Loss Layers Matter
A catastrophe bond does not necessarily respond to every dollar of catastrophe loss. Its terms can specify that a particular tranche absorbs losses only within a defined band: above an attachment point and up to an exhaustion point. Those boundaries determine when the tranche begins to incur loss and when its capacity has been used up.
For a layered bond structure, the key task is to take the loss measure specified by the contract and allocate it to the relevant bands. That allocation is not automatically the same as the total catastrophe loss. It depends on the contract-defined loss input and on each tranche’s position in the structure. The examples below assume that this input has already been determined; they focus on interpreting and applying the layers.
All amounts in the examples are in millions of dollars unless stated otherwise. Let \(Z\) denote the contract-defined covered loss used for the allocation. It is important to identify this input before applying layer boundaries: a gross catastrophe estimate and a contract-defined covered loss are not interchangeable unless the contract makes them so.
Reading Attachment, Exhaustion, and Limit
A layer described as “40 to 100” attaches at 40, exhausts at 100, and has a limit of 60. The phrase describes a band of losses, not a guarantee that the tranche pays 100 when loss reaches 100. It can absorb at most the 60 lying between its attachment and exhaustion points. If it is represented by a bond tranche with principal of 60, the layer limit and the tranche principal match under the usual fully collateralized structure.
The loss allocated to one layer can be written compactly as:
This expression takes only the part of \(Z\) above attachment, then caps that amount at the layer’s limit. Equivalently, it is zero when \(Z\le A\), equals \(Z-A\) when \(A<Z<E\), and equals \(E-A\) when \(Z\ge E\). At exactly \(A\), the layer has no loss; at exactly \(E\), it is fully exhausted.
How the Layer Formula Works
The formula follows directly from the meaning of a band. First, losses below \(A\) cannot reach the layer, so the amount entering the calculation must be zero when \(Z-A\) is negative. This is achieved by taking \(\max(Z-A,0)\). Second, no layer can absorb more than its width, \(E-A\), so the amount must be capped at that width. Taking the minimum of the amount above attachment and the limit applies that cap.
This also explains why subtracting the attachment point alone is insufficient for a large loss: that subtraction can exceed the tranche’s limit. The cap is essential.
Worked Example: Applying One Tranche’s Boundaries
A hypothetical catastrophe bond tranche covers the layer from 40 to 100 and has principal of 60. For each of three contract-defined covered losses, calculate the allocated layer loss and remaining principal: \(Z=25\), \(Z=72\), and \(Z=118\).
Method. The attachment point is \(A=40\), the exhaustion point is \(E=100\), and the limit is \(E-A=100-40=60\). For each loss, apply \(L(Z;40,100)=\min\{\max(Z-40,0),60\}\). Under the stated assumption that allocated layer loss reduces principal dollar for dollar, remaining principal equals \(60-L\).
When \(Z=25\): \(Z-A=25-40=-15\), so \(\max(-15,0)=0\). The allocated loss is \(\min(0,60)=0\), and remaining principal is \(60-0=60\).
When \(Z=72\): \(Z-A=72-40=32\). This is below the limit of 60, so the allocated loss is \(\min(32,60)=32\). Remaining principal is \(60-32=28\).
When \(Z=118\): \(Z-A=118-40=78\), which exceeds the limit. The allocated loss is \(\min(78,60)=60\), and remaining principal is \(60-60=0\). The 18 of covered loss above the exhaustion point does not create additional loss in this tranche.
Check. Allocations are never negative and never exceed the 60 limit. The tranche has no loss below 40, a partial loss between 40 and 100, and is exhausted above 100. A common mistake is to assign \(118-40=78\) to the tranche; the correction is to cap the allocation at 60.
Allocating Loss Across a Stacked Structure
In a sequential stack, one tranche covers the first band, the next covers the band above it, and additional tranches cover higher bands. Each layer is evaluated against the same contract-defined covered loss \(Z\), using its own attachment and exhaustion points. Losses allocated to separate, nonoverlapping layers can then be added to obtain the total allocation to the stack.
For a stack with contiguous layers, the sum of allocations cannot exceed the amount of covered loss within the stack’s overall range. Loss below the first attachment does not enter the stack. Loss above the top exhaustion point does not increase any tranche’s allocation under these assumptions.
Use the contract-defined covered loss supplied for the calculation, not an unstated or substituted gross-loss measure.
For each tranche, record \(A\), \(E\), and its limit \(E-A\). Check that the bands are ordered and that their widths match the stated limits.
Apply the layer formula separately to each band. Do not subtract a prior tranche’s allocation from \(Z\) and then treat the remainder as a new total-loss input.
Sum tranche allocations and compare the result with the covered loss lying within the stack’s total range. Check that no tranche allocation exceeds its limit.
Worked Example: Allocating Loss Across Three Tranches
A hypothetical bond structure has three contiguous tranches: 50 to 100, 100 to 180, and 180 to 250. Their principal amounts equal their respective layer limits. A modeled contract-defined covered loss is \(Z=215\). Calculate each tranche’s allocated loss and remaining principal.
Method. Calculate each band’s limit, apply the layer formula to \(Z=215\), and then subtract the allocated loss from the corresponding principal.
First tranche, 50 to 100: its limit and principal are \(100-50=50\). The amount above attachment is \(215-50=165\), so the allocation is \(\min(165,50)=50\). Remaining principal is \(50-50=0\).
Second tranche, 100 to 180: its limit and principal are \(180-100=80\). The amount above attachment is \(215-100=115\), so the allocation is \(\min(115,80)=80\). Remaining principal is \(80-80=0\).
Third tranche, 180 to 250: its limit and principal are \(250-180=70\). The amount above attachment is \(215-180=35\), so the allocation is \(\min(35,70)=35\). Remaining principal is \(70-35=35\).
Check. Total allocated loss is \(50+80+35=165\). The stack spans 50 to 250, and the covered loss within that range is \(215-50=165\). The allocations reconcile, and none exceeds its tranche limit. A common error is to calculate the second tranche’s amount as \(215-50-100\); attachment points are loss levels, not amounts to subtract cumulatively.
From Allocated Layer Loss to Principal
An allocated layer loss describes the amount of the contract-defined loss assigned to that band. It is not automatically identical to a principal reduction under every possible transaction. For the examples here, assume a standard structure in which each tranche has principal equal to its layer limit and allocated loss reduces that principal dollar for dollar, subject to a floor of zero. This assumption makes the layer allocation directly interpretable as the principal reduction.
A tranche can also have a different principal amount from the stated loss limit, or its terms can specify a different write-down mechanism. In those cases, the attachment and exhaustion points still identify the covered loss band, but the principal consequence must be determined from the transaction terms. The formula below is therefore a conditional mapping under the stated dollar-for-dollar assumption, not a universal rule for every bond.
where \(P=E-A\) under the one-to-one principal-and-limit assumption. The reduction cannot exceed \(P\), because the layer allocation cannot exceed its limit. If the principal is not equal to the limit, the contractual method for translating allocated loss into a write-down must be stated before calculating remaining principal.
Worked Example: Partial Loss in a Higher Layer
A hypothetical tranche has attachment point 70, exhaustion point 110, and principal of 40. A modeled covered loss of 92 applies to this layer. Assuming allocated loss reduces principal dollar for dollar, determine the allocation, reduction as a percentage of initial principal, and remaining principal.
Method. First find the layer limit: \(110-70=40\), which matches the tranche principal. Then calculate the amount above attachment and cap it at the limit.
The amount above attachment is \(92-70=22\). Since \(22<40\), the allocated layer loss is 22. The principal reduction is therefore 22, and the remaining principal is \(40-22=18\).
The proportion of initial principal reduced is \(22/40=0.55\), or 55%. The remaining proportion is \(18/40=0.45\), or 45%.
Check. The covered loss of 92 lies between 70 and 110, so the tranche should be partially, not fully, exhausted. The 22 reduction is less than the 40 initial principal, and \(22+18=40\). A common mistake is to use \(92/110\) as the write-down percentage; the relevant loss for this layer is the amount above attachment, not the entire covered loss.
Common Allocation Errors
Layer calculations are compact, but a small interpretation error can materially change the result. Check the following before accepting an allocation:
- Ignoring the attachment point. Applying the whole \(Z\) to the tranche overstates loss when the layer begins above zero. The relevant amount before the cap is \(Z-A\), not \(Z\).
- Failing to cap at exhaustion. If \(Z\) exceeds \(E\), the tranche still absorbs only \(E-A\). Loss beyond \(E\) belongs outside that layer.
- Confusing attachment with a deductible on every tranche. An attachment point is a loss level in the structure. In a stack, each band is evaluated by its own boundaries; the values of the lower attachments are not all subtracted again.
- Using the wrong loss measure. The allocation is only as appropriate as its input. Confirm that \(Z\) is the covered loss specified for the layer rather than assuming any reported catastrophe total is eligible.
- Equating limit and principal without support. They match in the standard assumption used here, but transaction terms may specify another relationship or write-down method.
A useful final check is to compare each tranche’s allocation with its limit and to reconcile the sum across the stack. For contiguous layers spanning \(A_1\) to \(E_n\), the total allocation should be zero below \(A_1\), rise with covered loss while it falls within the stack, and be capped at the sum of tranche limits once the loss reaches \(E_n\).
Check Your Understanding
For each problem, treat the stated loss as the contract-defined covered loss. Unless a problem says otherwise, assume principal equals layer limit and allocated loss reduces principal dollar for dollar.
- A tranche attaches at 25 and exhausts at 75. If covered loss is 61 and initial principal is 50, find the allocated loss and remaining principal. (Answer: allocated loss 36; remaining principal 14.)
- A three-tranche stack covers 0 to 40, 40 to 90, and 90 to 150. For covered loss \(Z=112\), find the allocation to each tranche and the total allocation. (Answer: 40, 50, and 22; total allocation 112.)
- A tranche covers 120 to 200. Calculate its allocated loss for covered losses of 95, 154, and 245. (Answer: 0, 34, and 80, respectively.)
- A tranche covers 35 to 85 and has principal of 50. Covered loss is 68. Find the principal reduction, remaining principal, and percentage of initial principal reduced. (Answer: reduction 33; remaining principal 17; 66% reduced.)
- A hypothetical structure has tranches covering 50 to 100, 100 to 160, and 160 to 210. Gross modeled catastrophe loss is 230, but the contract-defined covered loss supplied for this calculation is 174. Allocate the covered loss across the tranches and state the total allocation. (Answer: 50, 60, and 14; total allocation 124.)