5.3 Loss Run Analysis and Historical Loss Trending
Key Takeaways
- Total incurred equals paid losses plus outstanding reserves, and every loss run must carry a valuation date because the same policy year reports different numbers as reserves develop.
- Frequency and severity move independently: rising frequency points to a process or reporting-culture problem, while rising severity points to catastrophic exposure and a limits adequacy question.
- Immature policy years always understate losses because claims are still being reported and open reserves develop upward, so years must be compared at the same development age.
- Lag time, the interval between date of loss and date reported, is a reporting-culture indicator, and long lag correlates with higher severity because the investigation window is lost.
- Claim counts must be normalized against an exposure base such as adjusted patient days, occupied beds, encounters, or provider FTEs so that organizational growth is not misread as deterioration.
Quick Answer: A loss run is the carrier or third-party administrator's claim-by-claim report, valued as of a stated date. Total incurred = paid + outstanding reserve. Frequency and severity move independently and demand different responses: rising frequency points to a process failure, rising severity points to a catastrophic exposure and a limits question. Immature policy years always understate losses, and raw claim counts must be normalized against an exposure base before any trend is reported.
Why the Exam Tests Loss Run Analysis
Domain 2 task C asks the risk manager to "analyze historical loss experience" and "address trends." These items sit at the Analysis cognitive level. You are shown a set of numbers and asked what they actually mean and what to do about it. The risk manager is the person who reads the loss run and converts it into a defensible recommendation on retention, limits, and targeted loss control.
The authority boundary applies here too. The risk manager analyzes and escalates; the claims professional sets case reserves and the actuary projects ultimate losses. An answer choice in which the risk manager unilaterally adjusts reserves on the loss run to make a trend look better is wrong for reasons that go well beyond bad practice.
What a Loss Run Is and What It Carries
A loss run is a report produced by the insurer or the TPA listing every reported claim and occurrence in a policy period, valued as of a specific date. A loss run without an "as of" valuation date is unusable, because the same policy year will show different numbers every month as reserves develop and claims close.
| Field | What it means | Why the risk manager reads it |
|---|---|---|
| Claim number | Unique file identifier | Ties the loss run to the claim file and the event report |
| Policy year / coverage line | Which program period and which policy responds | Prevents mixing GL, PL, auto, and workers' compensation in one trend |
| Date of loss (DOL) | When the incident occurred | Assigns the claim to the correct policy year on an occurrence basis |
| Date reported (DOR) | When the organization learned of it | With DOL, produces lag time |
| Status | Open, closed, or reopened | Open claims can still develop; closed ones generally cannot |
| Paid indemnity | Dollars paid to or on behalf of the claimant | The realized cost of the harm |
| Paid expense / ALAE | Allocated loss adjustment expense — defense counsel, experts, court costs, records | Often 30 to 50 percent of a defended professional liability file's total |
| Outstanding reserve | The adjuster's current estimate of what remains to be paid | The forward-looking half of the number |
| Total incurred | Paid + outstanding reserve | The single figure used for trending |
| Cause / allegation code | Retained foreign object, medication error, fall, diagnosis delay | Enables the analysis that drives loss control |
| Location / department / service line | Where the exposure sits | Directs the intervention to an accountable owner |
| Recovery / subrogation | Amounts recovered from third parties | Net cost differs from gross cost |
The most common analytic error in this table is trending paid losses only. Paid dollars understate a young year severely, because the largest claims are still open and their value lives entirely in the outstanding reserve. Trend on total incurred.
Frequency and Severity Are Different Problems
Frequency is the number of claims per unit of exposure. Severity is the dollars per claim. They move independently, and confusing them produces the wrong intervention.
- Rising frequency usually signals a process, system, staffing, or training defect — or, importantly, a change in reporting culture rather than a change in actual harm. The response is operational: workflow redesign, competency validation, equipment standardization, environmental fixes. Because predictable frequency is cheap to fund, a stable high-frequency pattern often supports a higher retention, not more limit.
- Rising severity signals catastrophic exposure, a shift in the litigation environment, or inadequate early resolution. The response is financial and legal: verify tower adequacy, review the attachment points, examine reserve adequacy, and consider early resolution on the outlier files. Buying loss-control training does nothing about a single $18,000,000 verdict.
| Low severity | High severity | |
|---|---|---|
| Low frequency | Retain and monitor; not a program priority; watch for emerging cause codes | Verify limits and tower adequacy; catastrophic modeling; early resolution strategy; reserve adequacy review |
| High frequency | Loss control and process redesign; predictable, so a candidate for a higher retention | Systemic failure — escalate to the board; pursue loss control and limits simultaneously; consider service-line intervention |
Maturity: Why You Cannot Compare a Fresh Year to an Old One
A policy year gets more expensive as it ages. Claims that occurred in the year have not all been reported yet — that unreported inventory is incurred but not reported (IBNR) — and the claims that are open develop upward as discovery proceeds and reserves are strengthened. Loss triangles and loss development factors are the actuarial machinery for projecting that development, and they are taught in the actuarial section; what matters for reading a loss run is the discipline that follows from them.
Compare years at the same development age. A policy year valued at 12 months is compared to prior years also valued at 12 months, not to their current mature values. A director who looks at a six-month-old year showing $410,000 and a five-year-old year showing $6,800,000 and concludes the safety program is working has made the single most common error in the domain.
Lag time — the interval between date of loss and date reported — is a reporting-culture indicator and an early-warning metric. Long lag means the organization learned late, which means the scene was gone, memories had faded, staff had turned over, and the defense started from behind. Track median lag by department and by cause code, not just the average. Lag also matters directly for risk financing: on a claims-made program, a claim discovered late may fall outside the reporting period entirely.
Severity distribution. Losses are not normally distributed; a handful of claims drive most of the dollars. Report the median alongside the mean, and always list the top ten claims by total incurred. If two open files carry 79 percent of a program's incurred, the "trend" is really a statement about those two files.
A loss run valued as of 12/31/2025 shows policy year 2025 with 9 claims and $410,000 total incurred, against policy year 2020 with 31 claims and $6,800,000. A service line director tells the quality committee that the safety program is clearly working. The best analytic response is to:
A Worked Loss Run
General liability and professional liability, obstetrics service line, valued as of 12/31/2025:
| Claim | Date of loss | Date reported | Status | Paid indemnity | Paid ALAE | Outstanding | Total incurred |
|---|---|---|---|---|---|---|---|
| C-1001 | 2023-02-14 | 2023-03-02 | Closed | $145,000 | $38,000 | $0 | $183,000 |
| C-1002 | 2023-06-30 | 2024-11-18 | Open | $0 | $62,000 | $900,000 | $962,000 |
| C-1003 | 2023-09-05 | 2023-09-06 | Closed | $0 | $9,500 | $0 | $9,500 |
| C-1004 | 2024-01-22 | 2024-02-10 | Open | $50,000 | $74,000 | $1,250,000 | $1,374,000 |
| C-1005 | 2024-08-11 | 2024-08-15 | Closed | $22,000 | $6,000 | $0 | $28,000 |
| C-1006 | 2025-03-19 | 2026-01-27 | Open | $0 | $15,000 | $400,000 | $415,000 |
Four readings a CPHRM should produce from this table:
- Concentration. Total incurred across all six files is $2,971,500. Two open files, C-1002 and C-1004, carry $2,336,000 of that — roughly 79 percent. Any statement about "the obstetrics trend" is really a statement about two claims.
- Lag. C-1002 was reported almost 17 months after the date of loss, and C-1006 more than 10 months. Both are the largest or fastest-developing files in their years. That is the classic pattern: late reporting and high severity travel together, because the organization loses its investigation window.
- Maturity. Policy year 2025 shows a single claim. It is six to nine months old at valuation. It will grow. Reporting 2025 as an improvement over 2023 would be indefensible.
- Expense load. ALAE is $204,500 across the six files, and two of them have paid expense with zero indemnity. Defense cost is a real cost of risk and belongs in the total, not in a footnote.
Normalizing Against Exposure
Raw counts are meaningless in a growing organization. Divide by an exposure base that scales with the activity generating the loss:
| Exposure being measured | Appropriate base |
|---|---|
| Hospital professional liability | Adjusted patient days; occupied beds; inpatient admissions |
| Physician professional liability | Employed provider full-time equivalents (FTEs) |
| Ambulatory and clinic liability | Encounters or visits |
| Surgical liability | Surgical cases or operating-room minutes |
| General liability | Total visitors, square footage, or adjusted patient days |
| Workers' compensation | Employee FTEs or payroll |
Report the rate, such as claims per 1,000 adjusted patient days or incurred dollars per employed provider FTE. A system that grew 42 percent through acquisition and saw claim counts rise 40 percent did not deteriorate; its rate improved slightly. Presenting the raw count to a board without normalization invites the wrong decision.
Benchmarking compares that normalized rate against peers of similar bed size, teaching status, service mix, and geography, using industry and society benchmark studies. Treat external benchmarks as directional context. The organization's own trend line, measured consistently at the same maturity and against the same exposure base, is far more actionable, because it controls for the differences no benchmark can.
Turning the Analysis Into a Risk Financing Decision
The point of the analysis is a recommendation:
- Stable, credible, high-frequency, low-severity experience supports raising the retention, because the losses are predictable and the premium charged to insure them exceeds their expected cost plus a reasonable margin.
- Volatile severity or a widening tail supports buying more limit and re-examining attachment points.
- A concentrated cause code — three retained foreign objects in one operating room, or eight falls on one unit — supports targeted loss control with a named owner, a deadline, and a re-measurement date, reported through the risk committee to the board.
- Deteriorating lag supports a reporting-culture intervention, not a coverage change.
Exam Traps on Loss Runs
- Accepting a loss run with no valuation date. The number is meaningless without an "as of."
- Trending paid losses only. Total incurred includes the outstanding reserve, which is where the young, large claims live.
- Comparing an immature year to a mature year. Compare at the same development age.
- Reacting to raw counts. Normalize against an exposure base first.
- Calling one shock loss a trend. One catastrophic verdict is a severity event, not a pattern.
- Reading falling incident reports as improved safety. It can equally mean reporting fatigue or fear of blame, which is the opposite conclusion.
- Buying limits to fix a frequency problem or ordering training to fix a severity problem.
- Adjusting reserves on the loss run. Reserve setting belongs to the claims professional and the actuary; the risk manager analyzes and escalates.
Over three years a health system's patient-fall liability claims rise from 14 to 38 per year while the average incurred per fall claim stays close to $9,000. Which response does the data indicate?
A health system's reported general liability claims rose from 120 to 168 over two years. During the same period it acquired two hospitals and adjusted patient days rose 42 percent. Before presenting a deteriorating trend to the board, the risk manager should: