Why Exposure Representation Matters
A catastrophe model cannot analyze an insured portfolio as an undifferentiated premium total. It needs a structured description of what is insured, where it is represented, how it is built and used, what it is worth, and which policy terms apply. Those fields form the model’s representation of the exposure: the insured property and related interests that could be affected by an event.
This representation is an abstraction, not a complete description of every physical feature or contractual nuance. The model can only distinguish characteristics that its data structure and classifications support. A portfolio record that combines several buildings into one row may therefore hide differences that matter to the model, even if the total dollar value is correct.
In the catastrophe-model framework, exposure data are the link between the insured portfolio and the modeled event process. The model uses the representation to determine which insured assets are associated with modeled conditions and how their characteristics are passed to later model components. Exposure fields do not, by themselves, calculate damage or insured loss; they establish what the model is being asked to represent.
Core Fields in an Exposure Record
A useful exposure record typically combines several groups of fields. The exact fields and acceptable categories vary by model and line of business, so the model’s data specification must be checked before translating a company’s records.
| Field group | Examples of information | Why it matters to representation |
|---|---|---|
| Location identity | Address, location identifier, site or building identifier, coordinates when available | Distinguishes insured places and gives the model a location reference. |
| Construction | Construction material or class, year built, number of stories, roof type | Describes physical characteristics that may be relevant to the model’s classification of an asset. |
| Occupancy | Building use, such as warehouse, office, or residential; business activity where relevant | Distinguishes how the building or space is used. A building’s use is not the same as how it is constructed. |
| Valuation | Building, contents, and business-interruption values; currency; valuation date; valuation basis | States the amount and type of insured interest represented. |
| Policy and coverage | Covered perils, limits, deductibles, sublimits, exclusions, and policy or coverage identifiers | Describes contractual boundaries that can affect the treatment of modeled losses. |
A location identifier should consistently connect an exposure record to the relevant location and policy information. An address or coordinate may also be included, but a location’s name alone may not be enough to distinguish multiple buildings on one campus. Conversely, one physical building might have separate records for different insured interests or coverage sections. The record design needs to preserve whichever distinctions the model and the coverage data require.
Construction and occupancy should be treated as separate dimensions. Construction describes physical features; occupancy describes use. For example, a masonry building can be used as a warehouse or an office. Entering an occupancy code in a construction field, or assuming that one implies the other, changes what the record says about the exposure.
Valuation and Policy Information Are Different
Valuation fields describe the amount and basis of the insured interest. Common categories include building, contents, and business interruption (BI). The valuation basis should be explicit: for example, replacement cost or another stated basis, along with the currency and valuation date. Two values that use different bases or dates may not be directly comparable, even if both are shown in dollars.
A policy limit is not another asset value. A location can have a total insured value (TIV) larger than a policy limit, because the limit constrains coverage rather than describing the full value of the represented property. Deductibles, sublimits, exclusions, and covered-peril information also belong in policy fields. They should not be silently folded into the building or contents value.
This addition is appropriate only when the values refer to distinct interests that are meant to be included once. A shared policy limit is not added to the asset values. Nor should the same building value be counted twice merely because it appears in both a location schedule and a policy summary. The data must preserve the distinction between the insured property and the contractual terms applying to it.
Worked Example: Building a Location-Level Exposure Record
Worked Example: Building a Location-Level Exposure Record
A hypothetical insured has three separate warehouse buildings at one site. The schedule gives building values of $1,600,000, $1,250,000, and $1,150,000. The combined contents value is $900,000, and BI value for the site is $1,200,000. Assume all values are in U.S. dollars, use the same valuation date and basis, and represent distinct insured interests. The policy schedule identifies the site but does not allocate contents among the buildings. Determine the total represented value and describe what should be recorded at building level versus site level.
Method. Preserve each known building value at the building level, retain the combined values at the site level unless a supported allocation is available, and add distinct values to obtain TIV. This avoids inventing building-level allocations for the contents and BI values.
First, add the building values:
Then add the site-level contents and BI values:
The three building records can carry their respective building values and any known building-specific construction and occupancy attributes. The $900,000 contents value and $1,200,000 BI value should remain attached to the site or coverage schedule unless reliable data support a more detailed allocation. The result is checked by adding the three stated value categories once: $4,000,000 of buildings, $900,000 of contents, and $1,200,000 of BI, totaling $6,100,000.
Common mistake avoided: Dividing the contents and BI values equally among the three buildings would assign $300,000 of contents and $400,000 of BI to each building. Those figures sum correctly, but the schedule does not establish that the values are distributed equally. A numerically balanced allocation is not a factual allocation.
Worked Example: Separating Exposure Values From Contract Terms
Worked Example: Separating Exposure Values From Contract Terms
A hypothetical manufacturer insures two locations. Location A has $5,000,000 of building value, $1,000,000 of contents, and $2,000,000 of BI. Location B has $3,000,000 of building value, $500,000 of contents, and $1,500,000 of BI. A single policy lists a $10,000,000 property limit, a $2,000,000 BI sublimit, and a $250,000 occurrence deductible. Assume all stated exposure values are distinct, use U.S. dollars, and share a common valuation date and basis. Find the portfolio TIV by category, then state how the policy fields should be represented.
Method. Sum the exposure values by category, then record policy terms separately. This makes clear which figures describe the insured interests and which describe contractual limits.
Building value is:
Contents value is:
BI value is:
Therefore:
The exposure records should retain the values by location and category. The $10,000,000 property limit, $2,000,000 BI sublimit, and $250,000 occurrence deductible should be recorded as policy terms with their relevant scope and coverage identifiers. The final check is that the three exposure categories total $13,000,000, while the policy limit and deductible remain separate fields.
Common mistake avoided: Subtracting the $10,000,000 limit or $250,000 deductible from TIV would mix exposure valuation with contract treatment. TIV remains $13,000,000. The policy terms may affect a later loss calculation, but they do not rewrite the stated values of the insured interests.
Why Detail and Classification Affect What Can Be Modeled
The model can use only distinctions that survive the transformation from source records into its exposure fields. Suppose an insurer knows that two buildings have different construction classes but combines them into one record with a single class. The total value may remain unchanged, yet the model can no longer treat the two buildings as separate construction exposures. Likewise, a record that says “commercial” may conceal distinct uses if the model’s occupancy categories make those distinctions relevant.
This is a representational issue, not a claim about the direction or size of any resulting loss difference. The model’s available categories and later calculations determine how represented attributes are used. The exposure-data task is to describe the insured portfolio faithfully within those categories and to avoid implying greater precision than the source information supports.
Worked Example: Preserving Construction Differences in a Portfolio Record
A hypothetical insured reports three separate buildings. Two wood-frame buildings are each valued at $1,000,000, and one masonry building is valued at $2,000,000. All are offices, and their values use the same currency and valuation basis. Compare a record set that preserves each building’s construction class with one that aggregates all three into a single $4,000,000 office record labeled wood-frame.
Method. Sum values by the known construction category before comparing the two representations. This quantifies what is preserved or lost; it does not estimate catastrophe damage.
The wood-frame value is:
The masonry value is $2,000,000, so the overall value is:
With separate records, the model receives $2,000,000 of wood-frame exposure and $2,000,000 of masonry exposure, all classified as office occupancy. With the aggregated record, it receives $4,000,000 classified as wood-frame office exposure. The overall value checks to $4,000,000 in either representation, but the aggregated version has reassigned the masonry building’s $2,000,000 to the wood-frame class. It therefore does not preserve the known construction mix.
Common mistake avoided: Treating equality of total value as proof that two exposure files are equivalent. The totals match, but the attribute-level representation does not.
A Practical Representation Sequence
A disciplined sequence helps keep physical descriptions, values, and policy fields distinct. It also makes explicit which details are known and which have not been supplied.
Decide whether a row represents a site, building, unit, or coverage interest. Use identifiers that preserve distinctions needed by the model and the policy data.
Record construction and occupancy separately, together with other supported attributes such as year built or number of stories. Use the model’s permitted categories rather than silently substituting one field for another.
Keep building, contents, and BI values identifiable. Include currency, valuation date, and valuation basis so that the amounts have interpretable meaning.
Identify the relevant coverage and policy, and record limits, deductibles, sublimits, and other terms separately from insured values.
If a field is unknown, preserve that status rather than assigning a guessed class or allocation. The model’s handling of unknown values should be understood before interpreting results.
The sequence is useful because apparently small choices can alter what the model is capable of distinguishing. A total value may be correct while its location, construction, occupancy, or coverage assignment is not. Conversely, a highly detailed record is not automatically better if the details are unsupported, incompatible with the model’s definitions, or duplicated across records.
Common Representation Traps
- Combining buildings that differ in relevant attributes. Aggregation can hide construction or occupancy variation even when the sum of values remains correct.
- Confusing construction with occupancy. A building’s material and its use describe different characteristics and should not be substituted for one another.
- Mixing values with limits. TIV describes represented insured interests; policy limits and deductibles describe contract terms.
- Leaving the valuation basis implicit. Values without currency, date, or basis may be difficult to interpret consistently.
- Inventing detail to fill a blank. An unsupported class or allocation can give a false impression of precision. Preserve unknown status and follow the model’s documented treatment.
Check Your Understanding
For each problem, distinguish exposure description from policy treatment and show any value additions before checking the answer.
- A hypothetical shop has $1,800,000 of building value, $250,000 of contents, and $450,000 of BI. The values use one currency, valuation date, and basis and represent distinct interests. What is its TIV? (Answer: $2,500,000.)
- A site has three separately identified buildings with building values of $1,250,000, $1,000,000, and $750,000. Its contents value is $300,000 and its BI value is $900,000, both recorded only at site level. Find TIV and state whether the site-level values should be allocated equally to the buildings without supporting information. (Answer: TIV is $4,200,000; no, equal allocation is unsupported.)
- Two locations have exposure values of $6,000,000 and $4,000,000. They share a $7,000,000 policy limit. Find the portfolio TIV and explain whether the limit establishes how much value belongs to each location. (Answer: TIV is $10,000,000; no, a shared limit does not specify location-level values.)
- A portfolio has two wood-frame buildings valued at $1,000,000 each and one masonry building valued at $2,000,000. All are offices. Give the value by construction class and identify what is lost if the entire portfolio is represented as one wood-frame record. (Answer: $2,000,000 wood-frame and $2,000,000 masonry; the aggregated record loses the known masonry distinction and misclassifies its value.)
- A hypothetical policy schedule reports $3,000,000 building, $500,000 contents, and $1,500,000 BI at Location A; and $2,000,000 building, $250,000 contents, and $1,250,000 BI at Location B. A $6,000,000 property limit and a $1,000,000 BI sublimit apply. Find TIV and state whether either policy term should be subtracted when reporting it. (Answer: TIV is $8,500,000; neither term is subtracted from TIV.)