Cat Bond Trigger Families

Why Cat Bond Trigger Families Differ

A catastrophe bond transfers risk only under the conditions set out in its contract. A central design choice is the trigger: the contractual rule that determines whether an event satisfies the condition for protection. Trigger families differ mainly in what they measure. A rule can refer to the sponsor’s own covered losses, losses across an industry, observed physical conditions, or a loss estimate produced by a model.

That distinction matters because the same catastrophe can produce different results under different measures. A sponsor may incur substantial covered losses even when an industry index remains below its threshold. Conversely, a physical measurement or industry index may cross its threshold when the sponsor’s own covered losses are relatively modest. This mismatch is commonly described as basis risk.

This tutorial focuses on identifying the measure and testing whether its contractual condition is met. Meeting a trigger condition is not, by itself, a calculation of how much principal is written down or how much the sponsor receives. Those outcomes depend on additional contract terms.

Definition: A cat bond trigger family is a category of contractual trigger classified by the primary quantity used to decide whether a specified catastrophe condition has been met. An indemnity trigger uses the sponsor’s defined covered losses; an industry-loss trigger uses a defined industry loss index; a parametric trigger uses specified physical measurements; and a modeled-loss trigger uses an estimated loss produced by a specified model and inputs.

The Four Reference Measures

A trigger is never just a label. Its contract must specify the event, the relevant geography and time period, the quantity being measured, and the threshold or other condition that activates it. A trigger test should therefore begin by identifying the exact contractual measure, not by asking whether the catastrophe “seems large.”

Trigger family	Primary reference measure	Key question in a trigger test
Indemnity	The sponsor’s losses that meet the contract’s covered-loss definition	Do the sponsor’s eligible losses meet the contractual condition?
Industry-loss	A defined loss amount or index for the insurance industry	Does the specified industry measure meet its condition?
Parametric	Specified physical observations, such as wind speed or ground motion	Do the observations at the stated locations and times meet the condition?
Modeled-loss	A loss estimate produced using specified model inputs and exposure data	Does the contractually determined modeled estimate meet its condition?

An indemnity trigger is tied most directly to the sponsor’s own defined loss, but determining that loss may require claims development and verification. An industry-loss trigger substitutes an industry-level measure for the sponsor’s individual loss. A parametric trigger relies on observations, rather than claims amounts. A modeled-loss trigger uses a model to translate specified event and exposure information into an estimated loss. These descriptions identify the families; the contract determines the actual calculation and governing data sources.

For a simple threshold trigger, let $Z$ denote the contractually specified measure and let $K$ be its threshold. If the contract says the condition is met when the measure is at least the threshold, the test is $Z \geq K$. The quantity represented by $Z$ depends on the trigger family. It must be expressed in the same units and scope as $K$.

A Common Test for Trigger Conditions

The following sequence keeps the analysis focused on activation. It also prevents a common error: substituting a familiar measure, such as the sponsor’s estimate of total economic damage, for the measure actually named in the contract.

Identify the contract’s event and scope.
Check the event definition, covered territory, measurement period, and any required location or data source.

Name the reference measure.
Determine whether the condition uses sponsor losses, an industry measure, physical observations, or a modeled estimate.

Apply the contract’s inclusion rules.
Use only losses, observations, or model inputs that fall within the stated scope.

Compare the measure with the condition.
Use the stated threshold operator, such as at least, greater than, or within a specified range.

Report activation separately from payment size.
State whether the trigger condition is met. Do not infer a principal write-down or other payment amount unless the contract’s payment terms are also provided.

Worked Example: Four Tests for One Hypothetical Cyclone

A hypothetical cyclone affects a coastal region. A cat bond’s four alternative trigger designs refer to the same event but use different conditions. The indemnity design activates if the sponsor’s eligible covered loss is at least $75 million. The industry-loss design activates if a specified industry index is at least $8.0 billion. The parametric design activates if the qualifying maximum sustained wind observation is at least 150 km/h at a designated station in the covered region. The modeled-loss design activates if the contractually specified modeled loss is at least $100 million. For this event, the sponsor has $82 million of eligible covered loss, the industry index is $7.4 billion, the designated station records 154 km/h, and the modeled loss is $96 million. Determine which conditions are met.

Method. Test each value against its own threshold, preserving its measure and units.

Indemnity. The eligible sponsor loss is $82 million, and the threshold is $75 million. Since $82 \geq 75$, the indemnity condition is met.

Industry-loss. The index is $7.4 billion, compared with a threshold of $8.0 billion. Since $7.4 < 8.0$, the industry-loss condition is not met.

Parametric. The qualifying wind observation is 154 km/h, compared with 150 km/h. Since $154 \geq 150$, the parametric condition is met. The result uses the specified station, not an unlisted observation elsewhere in the region.

Modeled-loss. The modeled loss is $96 million, compared with $100 million. Since $96 < 100$, the modeled-loss condition is not met.

Conclusion and check. The indemnity and parametric conditions are met; the industry-loss and modeled-loss conditions are not. Each comparison used the value named by its own trigger family. This conclusion establishes only which trigger conditions are satisfied, not the amount of any resulting payment or principal reduction.

Why Trigger Measure and Sponsor Loss Can Diverge

The distinction among families can be expressed with two quantities. Let $X$ denote the sponsor’s covered loss, and let $Z$ denote the measure used by a particular trigger. For an indemnity trigger, $Z$ is based on the sponsor’s contractually defined covered loss. For an industry, parametric, or modeled trigger, $Z$ is a different reference measure. Even if a trigger condition is written as $Z \geq K$, it does not follow that the sponsor’s loss $X$ is at least $K$, or even that $X$ is large.

The reason is that the trigger condition compares the threshold to $Z$, not directly to every possible measure of event severity. For example, an industry index summarizes a defined industry population, while the sponsor’s loss depends on the sponsor’s own covered portfolio and terms. A physical observation records a feature of the event, not the amount of insured damage. A modeled estimate depends on its specified model and inputs. These measures may be informative about sponsor loss without being equal to it.

\text{Trigger condition met} \iff Z \geq K, \qquad \text{not necessarily} \quad X \geq K.

This is also why comparing trigger families requires stating the reference measure explicitly. If the question asks whether a parametric threshold was met, comparing the sponsor’s paid claims to that threshold answers a different question. Likewise, a modeled-loss condition is tested using the contractually defined model output, not an informal estimate of likely damage.

Worked Example: Earthquake Index and Physical Reading

A hypothetical earthquake affects a region covered by a cat bond. The sponsor’s eligible earthquake loss is $41 million. One alternative trigger requires that eligible sponsor loss be at least $50 million. A second requires the specified industry-loss index to be at least $3.0 billion; the index is $3.2 billion. A third requires peak ground acceleration of at least 0.40 g at a named station in the covered region; the station’s qualifying reading is 0.42 g. A fourth requires a modeled portfolio loss of at least $45 million; the contractually determined estimate is $47 million. Which conditions are met?

Step 1: Indemnity test. Compare $41 million with $50 million. Since $41 < 50$, the indemnity condition is not met.

Step 2: Industry-loss test. Compare $3.2 billion with $3.0 billion. Since $3.2 \geq 3.0$, the industry-loss condition is met.

Step 3: Parametric test. Compare the named station’s 0.42 g reading with 0.40 g. Since $0.42 \geq 0.40$, the parametric condition is met.

Step 4: Modeled-loss test. Compare the specified modeled loss of $47 million with $45 million. Since $47 \geq 45$, the modeled-loss condition is met.

Conclusion and check. The industry-loss, parametric, and modeled-loss conditions are met; the indemnity condition is not. The $41 million sponsor loss does not override the other three tests because they use different measures. The answer is consistent with every stated comparison.

Thresholds, Scope, and Common Exam Traps

A numerical comparison is only as sound as the quantity being compared. Event scope can change the answer: a parametric contract may name one station rather than the maximum reading anywhere in the region; an industry measure may cover a specified territory; and a modeled-loss condition may prescribe particular exposure data. A value outside the contract’s scope cannot be substituted merely because it appears more representative.

The threshold operator matters too. “At least $50 million” includes exactly $50 million because it means $Z \geq 50$. “Greater than $50 million” excludes exactly $50 million because it means $Z > 50$. Similarly, a contract that requires two conditions joined by “and” requires both; meeting only one does not satisfy the full condition.

Worked Example: Applying a Geographic Condition

A hypothetical wildfire trigger has two requirements: at least 6,000 burned acres must lie within a named protection zone, and a specified industry-loss index must be at least $1.5 billion. The fire burns 7,200 acres in total, of which 5,200 acres are within the named zone. The industry index is $1.62 billion. Determine whether the full trigger condition is met.

Method. Apply each requirement to its stated measure and scope, then combine the results using the contract’s “and” rule.

Geographic area requirement. Only the 5,200 acres within the named zone count. Since $5{,}200 < 6{,}000$, this requirement is not met. The 7,200-acre event total is not the relevant quantity for this condition.

Industry-index requirement. The index is $1.62 billion, which is at least $1.5 billion because $1.62 \geq 1.50$. This requirement is met.

Combined condition. The trigger requires both requirements. The area requirement fails, even though the index requirement passes. Therefore, the full trigger condition is not met.

Check. The conclusion follows from the logical “and”: both conditions must pass. Counting all 7,200 acres would incorrectly remove the stated geographic restriction.

A frequent mistake is to take a number associated with the event and test it against whichever threshold is easiest to identify. For instance, a solver might compare the sponsor’s loss to an industry-loss threshold, or use the region’s maximum wind reading instead of the designated station’s observation. The following deliberately incorrect calculation shows why this fails.

Common mistake and correction: Suppose a contract requires the sponsor’s eligible covered loss to be at least $50 million, and that loss is $41 million. The event also has an industry index of $3.2 billion, while a separate industry trigger has a $3.0 billion threshold. The wrong calculation uses the index for the sponsor-loss condition: $3.2 \geq 3.0$, then concludes the $50 million indemnity condition is met. The correction is to keep each trigger paired with its specified measure: $41 < 50$, so the indemnity condition is not met; separately, $3.2 \geq 3.0$, so the industry-loss condition is met.

The worked examples deliberately stop at trigger activation. To determine a payment or principal reduction, one must also use the relevant attachment, exhaustion, loss-allocation, and settlement terms. A trigger family alone does not specify those quantities.

Why the Distinction Matters

Trigger choice affects how a cat bond responds to a catastrophe and how closely that response may track the sponsor’s own loss. An indemnity design measures the sponsor’s defined losses but requires those losses to be established under the contract. An industry-loss design relies on an external industry measure, which may not match the sponsor’s portfolio. A parametric design relies on specified observations rather than claims amounts. A modeled-loss design relies on a model output based on specified inputs.

None of these descriptions makes one family universally superior. The relevant question is whether the contractual measure and scope fit the exposure and the intended risk transfer. A trigger can be met while sponsor losses are low, or fail to be met while sponsor losses are high, if the chosen measure and the sponsor’s actual exposure diverge. Recognizing that possibility is essential when comparing trigger results; it is not a reason to silently replace the contract’s measure with another one.

Key takeaway: For every trigger test, state the family, identify its contractual reference measure, apply its scope and threshold, and report activation separately from payment size. The four families differ because they do not necessarily measure the same thing.

Check Your Understanding

For each original hypothetical scenario, identify whether the stated trigger condition is met. Apply the stated measure and scope only.

A sponsor-loss trigger requires eligible covered loss of at least $68 million. Eligible loss is $68 million exactly. Is the condition met? (Answer: Yes; “at least” includes equality.)
An industry-loss trigger requires an index greater than $2.4 billion. The index is exactly $2.4 billion, while the sponsor has $90 million of eligible losses. Is the industry condition met? (Answer: No; the index is not greater than $2.4 billion.)
A parametric trigger requires a reading of at least 175 km/h at a named coastal station. That station records 172 km/h; an unlisted station records 181 km/h. Is the condition met? (Answer: No; the named station’s reading is below 175 km/h.)
A modeled-loss trigger requires an estimate of at least $120 million. The contractually specified model output is $118 million, while a separate analyst estimates $126 million using different inputs. Is the condition met? (Answer: No; the specified output is $118 million.)
A hypothetical trigger requires both a qualifying zone loss of at least $24 million and a specified industry index of at least $900 million. The zone loss is $26 million and the index is $880 million. Is the full condition met? (Answer: No; the zone-loss requirement passes, but the industry-index requirement fails.)

What You'll Learn