The three-year lookback — reading carrier loss runs.

The loss-ratio gate decides whether a book's contingency survives a merger, but a buyer can't test the gate on the paid loss ratio alone — because the paid number lags reality. Claims are reserved before they're paid, so the carrier's forward view (paid plus reserved) can be well above the backward-looking paid ratio. Three years of loss runs, read with the reserved column, is how a buyer sees the real trajectory rather than the flattering snapshot.

§ 01 · The lookback standardWhat to pull, and the two ratios.

The diligence standard is three years of carrier-issued loss runs, requested from every carrier writing 5%+ of the book (a lower threshold for concentrated books, where a smaller carrier can still move the gate). Two ratios matter, and the distinction is the crux. The loss ratio — paid claims divided by written premium — is the conventional standalone metric, but it's backward-looking: it counts only what's been paid. Incurred losses — paid plus reserved — are the carrier's own forward-looking view, and they're what the carrier will use to test the gate as the reserves develop into paid claims. A buyer reading only the paid ratio sees a flattering snapshot; a buyer reading incurred sees what's coming. The gate this feeds is in loss-ratio blending.

§ 02 · Reserved claims tip the gateThe single-claim risk.

The reserved column is the difference between a book that qualifies and one that's about to fail. The worked example is stark: a 45% paid loss ratio looks comfortably under the gate, but $400K of reserved claims on a $2M book pushes the projected total-incurred ratio north of 65% — over the gate, with the forfeiture that follows. The single-claim version is sharper still: one $900K reserved claim on a small book can, by itself, tip the next year's ratio across the disqualification gate, because the small premium base can't dilute a large reserve. That's why the pending-claim review threshold is low — any claim with total incurred (paid plus reserved) over $50K, raised to $100K for larger books, deserves scrutiny: who, what event, expected resolution timing. A buyer who reads only the paid ratio and ignores the reserves is valuing a contingency line that the carrier's own forward view says is gone. The pure-profit contingency this protects is in pure-profit contingency.

§ 03 · Anomaly vs. patternReading the trajectory.

The lookback's deeper purpose is distinguishing a one-year anomaly (priceable, often recoverable) from a multi-year pattern (a structural problem). Four criteria mark an anomaly: a single year elevated with the other two clean, an identifiable root cause no longer in the book, flat claims frequency (a severity-only spike rather than more claims), and a loss ratio that reverts to baseline in the most recent year. Four criteria mark a pattern: a steady multi-year climb, elevated frequency (not just severity), negative carrier commentary, and a target who can't narrate the driver. The distinction matters because an anomaly — a single large claim from a since-departed account — can be priced or worked around, while a pattern signals a book whose loss experience will keep breaching the gate regardless of structure. Reading three years rather than one is what makes the anomaly-versus-pattern call possible; a single year shows neither the climb nor the reversion. The carrier-diligence view of loss ratios is in loss ratios.

§ 04 · The required columnsWhat a loss run must show.

A loss run is only useful if it carries the right columns, so a buyer specifies them in the request: policy number and insured, date of loss and date reported, status, paid, reserved, total incurred, subrogation or recovery, and claim type or coverage line. The reserved and total-incurred columns are the ones that reveal the forward risk the paid ratio hides; the date-reported and frequency data feed the anomaly-versus-pattern read; and the claim-type column surfaces the concentration patterns (a single carrier, line, or producer driving the losses). Pulling these loss runs early — for every 5%+ carrier, across three years, with the full column set — is the diligence that makes the gate test real on the premium map. Without the reserved column, a buyer is testing the gate on a number the carrier has already moved past; with it, the map shows the combined book's true loss trajectory before the codes ever merge. That early read is what turns the lookback from a checkbox into the load-bearing diligence the whole premium-mapping exercise depends on. The agency-code decision the gate test drives is in agency-code architecture.

◆