Why Event Definitions Matter
A catastrophe sequence may involve several locations, perils, and days of damage. Whether those losses constitute one event or several is not determined simply by their proximity in time or their connection to the same broad weather system. The governing contract language sets the boundary. That boundary can affect how a catastrophe bond’s trigger is evaluated, so an analyst must read the event definition before grouping losses.
An event definition typically specifies the covered cause or peril, the eligible territory, a time period, and sometimes a required causal connection. It may also say how the time period starts and whether the party making a claim can select its start. Those details matter: two clauses that both mention “96 hours” can classify the same sequence differently if one starts at the first covered loss and the other permits a selected window.
Keep two questions separate. First, is a loss eligible under the contract—for example, is it caused by a covered peril in the covered territory? Second, if it is eligible, should it be grouped with other eligible losses as one event or treated as part of another event? An hours clause helps define the second boundary; it does not make an otherwise excluded peril or location covered.
A Systematic Reading of the Clause
Read the operative wording in a fixed order. Start with the peril and territory, then identify the causal test and time rule. Finally, check whether the clause allows a party to select the period, whether periods may overlap, and how losses outside a period are treated. If the wording does not resolve one of these issues, do not silently supply a standard rule: identify the ambiguity and state the assumption needed to classify the sequence.
Decide whether the loss arises from a cause named in the clause. A hurricane-related flood is not necessarily covered by wording that covers hurricane wind only.
Check the specified countries, regions, counties, or other geographic limits for each loss location.
Determine whether the wording requires losses to arise from the same storm, occurrence, or other specified cause.
Identify the period length, its starting point, and whether the clause permits a selected start or requires a start anchored to a particular loss.
Classify eligible losses inside the defined period together only to the extent the contract permits. Do not infer that a loss outside the period is covered by a separate event unless it independently satisfies the wording.
A useful mathematical statement is possible when the clause specifically defines an event as starting at the first covered loss and running for \(H\) hours. Let \(t_0\) be that first loss time and \(t_i\) a later loss time. A later loss can fall within that event’s time window only if \(0 \leq t_i-t_0 \leq H\), with the endpoint treated according to the contract’s wording. This time test is necessary, not sufficient: the later loss must also meet the peril, territory, and causal conditions.
Worked Applications
Worked Example: Applying a Start-Anchored 168-Hour Clause
A hypothetical bond covers windstorm damage in two named coastal counties. Its clause defines an event as covered damage from the same windstorm beginning at the time of the first covered damage and continuing for 168 consecutive hours. A new event period begins with the first covered damage after that period. The clause includes the stated endpoint. Damage is reported at 03:00 Tuesday, 15:00 Friday, and 12:00 the following Tuesday. All three reports concern the same windstorm and are in the named counties. Classify the reports.
Method. Use the first covered damage as time zero, convert each later time difference into hours, and compare it with 168 hours. The peril, location, and common-storm conditions are given as satisfied.
From Tuesday 03:00 to Friday 15:00 is 3 days and 12 hours: \(3 \times 24 + 12 = 84\) hours. The Friday report is within the first event period because \(84 \leq 168\).
From Tuesday 03:00 to the following Tuesday 12:00 is 7 days and 9 hours: \(7 \times 24 + 9 = 177\) hours. The final report is outside the first event period because \(177 > 168\). The first period ends the following Tuesday at 03:00, so the report at 12:00 is 9 hours beyond its endpoint.
Classification. The Tuesday 03:00 and Friday 15:00 reports fall in the first event. The following Tuesday 12:00 report does not. Under the stated new-period rule, it begins a second event period, subject to the clause’s other requirements.
Check. The two event periods start 177 hours apart, and the later report is 9 hours after the first period’s endpoint. The common mistake is to call all three reports one event because they arise from the same windstorm; the 168-hour boundary prevents that grouping under this wording.
Worked Example: Separating Peril and Geographic Eligibility
A hypothetical contract covers hurricane-generated wind damage in Counties A and B. It defines an event as covered damage from the same hurricane within 72 hours after the first covered damage. It excludes flood damage and damage outside the named counties. A hurricane produces these reports: wind damage in A at 08:00 Monday; wind damage in B at 20:00 Monday; flood damage in A at 22:00 Monday; and wind damage in adjacent County C at 23:00 Monday. Which reports can be grouped in the covered event?
Step 1: Test peril. The first, second, and fourth reports are wind damage. The third is flood damage, which the clause expressly excludes. Its closeness in time does not turn it into covered wind damage.
Step 2: Test geography. Counties A and B are named. County C is not. Therefore, the report in C fails the geographic test even though it is wind damage from the same hurricane.
Step 3: Test time and causal connection for eligible reports. The first two reports are 12 hours apart: from 08:00 to 20:00 is \(12\) hours, which is within 72 hours. Both are attributed to the same hurricane and occur in named counties.
Classification. The wind reports in A and B can be grouped in the covered event. The flood report fails the peril test; the County C report fails the geographic test. Neither excluded report becomes part of the covered event merely because it occurred within 72 hours.
Check. The only grouped reports pass all three relevant tests: covered peril, covered location, and event time, with the stated common cause. A common error is to apply the hours clause first and include every loss in the window. The correction is to screen for coverage conditions before grouping.
Worked Example: Avoiding Overlapping-Window Chaining
A hypothetical contract covers earthquake damage in a specified region. It defines each event as beginning with the earliest not-yet-assigned covered damage and includes subsequent covered damage from the same earthquake sequence during the next 96 consecutive hours, inclusive of the endpoint. Damage outside that period begins a new event if it is otherwise eligible. The clause does not allow overlapping periods. Covered damage from the same sequence occurs at hours 0, 80, 100, and 190, measured from the first report. Determine the event assignment.
Method. Start the first period at hour 0. Its endpoint is \(0+96=96\). The report at hour 80 is inside because \(80 \leq 96\); the report at hour 100 is outside because \(100>96\). Thus the first event contains the reports at hours 0 and 80.
The hour-100 report is the earliest eligible damage not yet assigned, so it starts the second period. That period ends at \(100+96=196\). The report at hour 190 is inside because \(190-100=90\) hours, and \(90 \leq 96\). Therefore, the second event contains the reports at hours 100 and 190.
Classification. There are two events: the reports at hours 0 and 80 form the first; the reports at hours 100 and 190 form the second. The periods are 96 hours each and do not overlap.
Check. The hour-100 report is 100 hours after the first period’s start, so it cannot be pulled into the first event. The common mistake is to chain gaps: 0 to 80 is 80 hours and 80 to 100 is 20 hours, so a reader may put the first three reports together. That calculation is not the contract’s test. The clause measures each event from its own start, not from the most recent report.
Why the Boundary Rule Works—and Where It Stops
For a start-anchored clause, the time test follows directly from the definition of a consecutive-hour window. If the first covered loss occurs at \(t_0\), a window of length \(H\) ends at \(t_0+H\). A later loss at \(t_i\) is inside exactly when it is no earlier than \(t_0\) and no later than the endpoint. Subtracting \(t_0\) from each side gives \(0 \leq t_i-t_0 \leq H\). This derivation explains why a chain of short gaps is not enough: each loss is compared with the event’s start, unless the clause expressly defines a different rule.
This result applies only to a clause with a start anchored at the first covered loss and a fixed duration \(H\). It does not decide whether the peril, location, or causal condition is met, nor does it determine how a contract permitting selection of a period should be applied. If the clause permits a party to choose a consecutive-hours window, classification may depend on the permitted election rules and on which eligible losses the selected period captures. State those rules before doing the time comparison.
Common Traps and Practical Significance
- Treating an hours clause as a coverage grant. A loss outside the covered territory or from an excluded peril remains outside coverage, even when it occurs during the event window.
- Measuring from the wrong time. A clause may start at first covered damage, first manifestation, or a selected time. Use the stated starting point, not automatically the first date in a news report.
- Chaining successive gaps. Several individually short intervals do not establish that every loss is within \(H\) hours of the event start.
- Assuming every contract uses the same rule. Duration, start time, endpoint treatment, selection rights, and treatment of out-of-window losses are wording-dependent.
- Confusing causal association with event identity. A broad storm system may produce wind and flood losses, but the contract can cover only one peril or can define the event more narrowly.
Event boundaries matter because they determine which losses are considered together when applying the contract’s event-based provisions. The analyst should document both the classification and its assumptions: identify the eligible losses, quote or summarize the operative boundary rule, show the time calculation, and flag any ambiguity in the wording. This disciplined approach is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus and avoids substituting intuition for the contract definition.
Check Your Understanding
For each hypothetical clause, apply the wording as stated and distinguish eligibility from event grouping.
- A clause covers hail damage in Districts North and East, with a 48-hour period starting at first covered damage. Hail losses occur in North at 10:00 Wednesday and East at 04:00 Friday. Are they in the same event by time, assuming the same storm and no other restrictions? (Answer: Yes; the elapsed time is 42 hours, within 48 hours.)
- A clause covers wildfire damage in three named counties but excludes smoke damage. A covered fire damages a named county at noon Monday; smoke causes damage in that county at 14:00 Monday. Does the smoke loss join the covered event? (Answer: No; it fails the peril/coverage test, regardless of the short time gap.)
- A start-anchored 120-hour clause begins with the first covered loss at 06:00 Saturday. Another eligible loss occurs at 06:00 Thursday. Is the second loss within the first event period? (Answer: Yes; exactly 5 days, or 120 hours, after the start, and the stated endpoint is included.)
- A clause assigns the earliest unassigned eligible loss to a new 72-hour period. Eligible losses from the same storm occur at hours 0, 60, and 130. How many events result? (Answer: Two; hours 0 and 60 are in the first period, and hour 130 begins the second.)
- A clause covers hurricane wind in County M only, for 96 hours from first covered damage. At hour 0, wind damage occurs in M; at hour 40, wind damage occurs in County N; at hour 50, flood damage occurs in M; at hour 90, wind damage occurs in M from the same hurricane. Which reports can be grouped in the covered event? (Answer: The hour-0 and hour-90 wind reports in M; the County N report fails geography and the flood report fails the covered-peril test.)