Why Coverage-Level Records Matter
A policy may provide several types of protection, and those coverages need not share the same dates, limits, or exposure basis. If the data are stored only as one policy-level row, the record can obscure which dates and terms apply to which protection. For ratemaking, that can lead to incorrect exposure calculations or to assigning a coverage’s characteristics to another coverage.
The solution is to identify the level of detail represented by each row. A policy-level record describes the policy as a whole. A coverage-level record describes one coverage under that policy. If that coverage’s terms change during the period being represented, the coverage may need more than one time-bounded segment.
A coverage-level record commonly needs a policy identifier, a coverage identifier, the coverage’s effective and expiration dates, the applicable limit or other relevant term, and the exposure basis and exposure amount being recorded. The fields should describe the same coverage and the same period. Policy-level fields, such as the policyholder identifier or policy number, can be repeated on each coverage record so the records remain associated with their policy.
Repeating a policy identifier across coverage records is intentional; repeating a single policy-wide exposure amount on every coverage row is not. The first practice preserves the relationship between records. The second can duplicate exposure when the coverage rows are later summarized.
Representing Dates and Terms at the Coverage Level
A policy period and a coverage period are related, but they are not automatically identical. A coverage may begin after the policy starts, end before the policy expires, or continue beyond the policy’s expiration date, depending on the hypothetical contract and recorded transaction history. The coverage’s own effective and expiration dates determine the period represented by that coverage record.
The same principle applies to limits and exposure bases. If two coverages have different limits, record each limit with its own coverage. If one coverage has an exposure basis different from another’s, store the basis on the relevant coverage record rather than assuming that the policy-level basis applies to all coverages. Selecting which basis is appropriate for a coverage is a separate decision; here the task is to preserve the basis that applies to the record.
When the terms for a coverage change, a single record covering the entire span may no longer describe the history accurately. For example, if a coverage limit changes on July 1, the portion before July 1 and the portion beginning July 1 have different terms. Representing them as separate segments makes the change visible and lets each segment carry its own dates and limit.
Use the policy identifier to associate related records without assuming the policy has only one coverage.
Give each coverage a distinct identifier or code so its dates and terms are not confused with another coverage’s.
Record the effective and expiration dates that apply to that coverage, rather than copying the policy dates by default.
Use separate time-bounded rows when one row would otherwise assign a single term to periods with different terms.
Record the segment’s exposure basis and amount using a consistent convention, and avoid duplicating a policy-wide amount across coverage rows.
Time-Based Exposure Within a Coverage Record
For the examples below, use a stated hypothetical convention: an annual coverage unit is earned uniformly over 12 covered months, at \(1/12\) of a unit per covered month. Count complete calendar months from the coverage’s effective month through the valuation month, provided coverage is in force for each counted month. This convention is appropriate for these examples because the dates are month-aligned. It is not a universal rule for every exposure basis or every date pattern.
The fraction follows directly from the assumption of uniform earning. If one annual unit is earned evenly over 12 months, the monthly amount is \(1/12\) of a unit. After \(m\) covered months, the accumulated amount is the sum of \(m\) monthly amounts:
The calculation describes exposure for the specified coverage and segment. It does not make exposure units from different coverages interchangeable. In particular, a sum across coverages can be misleading when the coverages differ in terms or basis, even if each exposure amount is calculated correctly.
Worked Example: Different Coverage Dates Under One Policy
A hypothetical policy period is January 1 through December 31, 2025. It has property coverage from January 1 through December 31, 2025, and equipment breakdown coverage from February 1, 2025 through January 31, 2026. Each coverage has one annual coverage unit over its own 12-month term. Using the uniform monthly convention, calculate the earned exposure for each coverage as of October 31, 2025.
Method: calculate by coverage record. The valuation date is October 31, so count covered calendar months from each coverage’s own start month through October. The policy’s January-to-December dates do not replace the equipment breakdown coverage’s dates.
Property coverage. The covered months accrued are January through October: 10 months. Its earned exposure is \(10/12=0.8333\) annual coverage units, to four decimal places.
Equipment breakdown coverage. The covered months accrued are February through October: 9 months. Its earned exposure is \(9/12=0.7500\) annual coverage units, to four decimal places. The full coverage term is February 1, 2025 through January 31, 2026, which is 12 months, so the stated annual unit and the coverage dates are consistent.
Check. The property coverage has 2 months remaining after October, so \(10/12+2/12=1\) annual unit. The equipment breakdown coverage has 3 months remaining, November through January, so \(9/12+3/12=1\) annual unit. The calculations respect each coverage’s own 12-month term.
Common mistake avoided. Applying the policy’s January-to-December period to both coverages would incorrectly count January as an equipment breakdown coverage month and omit January 2026 from its term. Coverage dates, not a convenient policy-level date range, govern this calculation.
Creating Segments When a Coverage Term Changes
A coverage identifier alone may not identify a single set of terms for the entire policy period. If the limit or another recorded coverage term changes, divide the coverage history into segments whose dates do not cross that change. Each segment then has a clear start, end, and applicable term. This is a data representation rule: it preserves the distinct terms in the record. It does not imply that the exposure amounts from unlike segments should always be combined for analysis.
Worked Example: A Midyear Limit Change
A hypothetical commercial policy runs from January 1 through December 31, 2025. Its machinery coverage has a limit of $500,000 from January 1 through June 30 and a limit of $800,000 from July 1 through December 31. For this example, each six-month segment accrues annual coverage-unit exposure at \(1/12\) per covered month. Prepare the coverage segments and calculate earned exposure as of September 30, 2025.
Method: split at the term-change date. The limit changes on July 1, so one full-year row with one limit would misstate at least half of the period. Use one segment for January through June at $500,000 and a second for July through December at $800,000.
| Coverage segment | Dates | Limit | Months accrued by September 30 | Earned exposure |
|---|---|---|---|---|
| Machinery, segment 1 | January 1–June 30, 2025 | $500,000 | 6 | 6/12 = 0.5000 |
| Machinery, segment 2 | July 1–December 31, 2025 | $800,000 | 3 | 3/12 = 0.2500 |
For segment 1, all six covered months have elapsed by September 30, so its earned exposure is \(6/12=0.5000\) annual coverage units. For segment 2, July, August, and September have elapsed, so its earned exposure is \(3/12=0.2500\) annual coverage units. These results are stated to four decimal places.
Check. The first segment’s full six-month term corresponds to \(6/12=0.5000\) units. For the second segment, 3 of 6 covered months have elapsed by the valuation date, so half of its full-segment amount has been earned: \(0.5000/2=0.2500\).
Common mistake avoided. Treating the entire year as one segment at the $800,000 limit would incorrectly assign that limit to January through June. Keeping the two records separate preserves which limit applied to each date range.
Reading and Checking a Coverage-Level Schedule
A useful schedule allows a reviewer to answer, for every row: Which policy does this belong to? Which coverage does it describe? What dates and terms apply to this segment? What exposure basis and amount are recorded? These questions help catch errors that can be hidden in a single policy-level snapshot.
A row should not silently combine two different coverages merely because they are on the same policy. Nor should it combine two periods of one coverage when the record needs to preserve a term change. Conversely, creating extra rows without a reason can be confusing if a segment does not represent a distinct coverage or a meaningful change in the recorded terms. The aim is a clear record grain, not the largest possible number of rows.
Worked Example: Separating Three Coverage Records
A hypothetical policy has a policy period of January 1 through December 31, 2025. Its property coverage runs January 1 through December 31, its equipment coverage runs February 1, 2025 through January 31, 2026, and its liability coverage runs March 1 through August 31, 2025. Each coverage has its own annual coverage unit and earns \(1/12\) of that unit per covered month. At June 30, determine the earned exposure for each coverage and state how many coverage-level records are needed if no terms change within a coverage.
Method: one record per unchanged coverage period. There are three distinct coverages and no within-coverage term changes, so three coverage-level records are sufficient. For each, count only the covered months through June.
Property. January through June gives 6 covered months. Earned exposure is \(6/12=0.5000\) annual coverage units.
Equipment. February through June gives 5 covered months. Earned exposure is \(5/12=0.4167\) annual coverage units, rounded to four decimal places. The coverage continues beyond the policy’s December 31 expiration date and ends on January 31, 2026; its record retains those coverage-specific dates.
Liability. March through June gives 4 covered months. Earned exposure is \(4/12=0.3333\) annual coverage units, rounded to four decimal places. The coverage ends August 31, so it does not receive exposure for September through December.
Check. Each numerator is the count of months in that coverage’s date range through June: 6, 5, and 4. The answers are therefore \(0.5000\), \(0.4167\), and \(0.3333\) units, respectively. Keep these amounts attached to their own coverage records rather than presenting their sum as a single policy exposure with no qualification.
Common mistake avoided. Using the policy period for every coverage would give the equipment and liability records months when those coverages were not in force, and would fail to represent the equipment coverage’s later expiration.
Common Data and Exam Traps
- Copying policy dates onto every coverage. A coverage can have its own start and end dates. Calculate its exposure over those dates.
- Recording one limit for periods with different limits. Segment the coverage at the effective date of the term change so each row has a single applicable limit.
- Confusing repeated identifiers with duplicated exposure. Repeating a policy number across coverage records is useful. Repeating an identical policy-wide exposure amount on every row may inflate a later total.
- Assuming every coverage has the same exposure basis. Record the basis applicable to each coverage rather than allowing one coverage’s basis to stand in for another’s.
- Counting months outside the coverage dates. The policy start or expiration date does not determine the covered months when the coverage’s own dates differ.
- Combining unlike exposure amounts without context. A numerical sum may be computable but not meaningful as a single measure if the rows represent different coverages, limits, or bases.
Check Your Understanding
For each problem, apply the stated record and exposure conventions rather than assuming every coverage follows the policy dates.
- A hypothetical policy runs January 1 through December 31, 2026. Its cyber coverage runs April 1 through December 31 and earns \(1/12\) of an annual coverage unit per covered month. How many coverage-level records are needed if no terms change, and what is the earned exposure by September 30? (Answer: 1 record; 6/12 = 0.5000 annual coverage units.)
- A hypothetical policy has a glass coverage limit of $200,000 from January 1 through March 31, 2026, and $300,000 from April 1 through December 31. Each month earns \(1/12\) of an annual coverage unit. How many segments are needed, and what is earned exposure for each by May 31? (Answer: 2 segments; first segment 3/12 = 0.2500 units, second segment 2/12 = 0.1667 units.)
- A policy runs January 1 through December 31, 2026. A separate flood coverage runs October 1, 2026 through September 30, 2027, with one annual unit earned uniformly over its 12 covered months. What are the coverage dates to record, and what is its earned exposure at December 31, 2026? (Answer: October 1, 2026–September 30, 2027; 3/12 = 0.2500 annual coverage units.)
- A hypothetical policy has two coverages, each with its own unchanged six-month term, from January 1 through June 30, 2026. Each has one annual coverage unit earned at \(1/12\) per covered month. Should the policy be represented by one coverage-level record or two, and what is the earned exposure attached to each on June 30? (Answer: 2 records; 6/12 = 0.5000 annual coverage units for each coverage.)
- A machinery coverage runs January 1 through December 31, 2026. Its limit is $400,000 through August 31 and $600,000 beginning September 1. Each month earns \(1/12\) of an annual coverage unit. As of October 31, how many segments are required, and what is the earned exposure for each? (Answer: 2 segments; January–August is 8/12 = 0.6667 units, and September–October is 2/12 = 0.1667 units.)