D&O's 54.5 Loss Ratio Sits on a Premium Base That Shrank by a Third

AM Best's July 2026 special report puts the U.S. monoline D&O direct loss ratio at 54.5 for 2025, up from 49.0 a year earlier, against direct premium that has fallen to roughly $10 billion from nearly $15 billion in 2021. Reserves for accident years 2023 and 2024 already proved inadequate, and claims from those two years are closing more slowly than the 2018-2019 cohort whose own slow development triggered the last D&O hard market.

The Premium Base Behind the Loss Ratio

AM Best's own arithmetic makes the denominator problem explicit. Direct premium written for the monoline D&O line fell for a fourth consecutive year in 2025 to roughly $10 billion, down from nearly $15 billion in 2021, a one-third contraction in four years (AM Best, cited in Insurance Journal, July 13, 2026). Pricing declined in 10 of the past 11 quarters, though the pace has been easing: the fourth quarter of 2025 showed a decline of roughly 4%, smaller than the double-digit cuts that characterized 2023 and 2024 (Risk & Insurance, July 2026). A 54.5 loss ratio calculated against a premium base that has shrunk by a third is not the same 54.5 that would print against 2021's volume; on-leveled to that earlier rate environment, the same claim costs would sit against a materially larger denominator and the reported ratio would read lower. TransRe frames the line's expected breakeven at roughly a 70-point ultimate loss ratio once typical acquisition and internal expense loads are applied (TransRe, 2026), which puts 54.5 comfortably under that threshold on paper. That comparison only holds if the reserves behind the ratio are adequate, and AM Best's own report says the two most recent accident years are not.

Reserves That Missed on 2023 and 2024

AM Best is specific about which vintages are behind schedule. Reserve levels for accident years 2023 and 2024 already proved inadequate in the 2025 calendar year, and claims from those two years are closing more slowly than claims from 2018 and 2019, the cohort whose own slow-motion adverse development touched off the last D&O hard market (AM Best, July 2026, via Insurance Business). Associate director David Blades put the read directly: "This might indicate an underlying deficiency that could lead to a downturn in D&O liability underwriting results over the near term" (David Blades, AM Best, quoted in Insurance Business, July 2026). A calendar-year ratio that looks acceptable in 2025 is, by construction, borrowing comfort from accident years that have not finished developing, and a slower closure pattern than 2018-2019 means that borrowing has further to run than a simple two-year-old vintage would otherwise suggest. If the correction plays out the way 2018-2019 did, it will not arrive as a single-quarter charge but as a multi-year reserve strengthening spread across successive calendar years.

AM Best senior industry analyst Christopher Graham was more direct about the forward-looking pricing implication: "Despite generating solid direct underwriting results during the past few years, the competitive D&O marketplace is expected to become a little tighter in 2026, with underwriting margins likely to shrink" (Christopher Graham, AM Best, quoted in Risk & Insurance, July 2026). That is AM Best itself calling the floor on the soft cycle, not a pricing actuary's forecast, and it is worth separating from the loss-ratio headline: a rating agency naming margin compression as the expected 2026 outcome is a different signal than a single quarter's favorable or unfavorable print.

Scale matters here too. D&O makes up roughly a quarter of the broader Other Liability, Claims-Made line that carries it in most statutory filings (Risk & Insurance, July 2026), which means a reserve correction in D&O specifically will not always announce itself in the aggregate casualty figures a reserving actuary reviews line-by-line at the segment level; it can sit inside a blended Other Liability, Claims-Made result long enough to obscure which sub-line is actually driving the movement. A carrier reviewing its own Schedule P triangles for that broader line has a direct reason to isolate the monoline D&O component specifically rather than relying on the blended casualty result to signal the correction AM Best is already flagging.

Where Severity Is Migrating Inside the Tower

The reserve deficiency is not evenly distributed across the tower, and neither is the mispricing. TransRe's 2026 market analysis found that low and mid-excess layers carry the highest sensitivity to rising claim severity yet have captured the least price appreciation of any segment since 2013.

Those are precisely the layers a soft market prices off relative distance from the primary rather than off the severity actually accumulating inside them, and they are the layers where a 44% cumulative rise in legal-services fees since 2020 eats through the working layer and pushes cost upward into the layer above it (TransRe, 2026). Severity itself has a fresh data point behind it: Cornerstone Research's 2025 securities class action report put the median settlement at $17.3 million, up 20% from 2024 and the highest figure since 1997, even as the number of settlements fell to 74 from 88 a year earlier (Insurance Journal, citing Cornerstone Research, March 2026). Fewer cases settling for more money is precisely the frequency-down, severity-up pattern AM Best flagged for the line as a whole, and it erodes a low or mid-excess layer's adequacy faster than a stable frequency count would suggest, because excess layers price off the tail of a claims distribution, not its average.

A newer severity driver sits underneath those aggregate settlement figures rather than replacing them. AI-related securities class actions have outnumbered data breach securities suits since early 2025 and roughly doubled from 2023 to 2024, with average D&O settlement values in that category up 27% to approximately $56 million (Insurance Business, July 2026). That is a materially higher average than the line's overall median, and it is arriving as a new allegation category rather than a variant of an existing one, which means underwriters cannot simply lean on historical loss triangles built before generative-AI disclosure claims existed. A tower priced on a pre-2024 severity curve for technology and disclosure risk is, by definition, pricing a claims environment that no longer describes the accounts renewing today.

The De-SPAC Tail and a Fresh Wave of IPO Paper

Two transactional-exposure stories are running in opposite directions inside the same book. New SPAC-related securities filings have all but stopped, with only one filed in the first half of 2026 against five for all of 2025 (The D&O Diary, July 2026), but the claims generated by the 2021-2022 SPAC boom are still working through their development pattern, a cohort-specific reserve risk that will not show up in any aggregate frequency count because the underlying filings already happened years ago. At the same time, the IPO market that had gone quiet from 2022 through 2024, averaging roughly 110 completed deals a year against 397 in 2021, rebounded to 202 completed IPOs in 2025 before leveling off amid economic uncertainty tied to the Iran conflict (Insurance Business; AM Best, July 2026). Each new IPO is fresh transactional D&O exposure written into a market that has cut rate in 10 of its last 11 quarters, which means the incremental business arriving just as the cycle turns is being priced on soft-market comparables rather than helping rebuild the adequacy the reserve deficiency has already eaten into. That dynamic works against the intuitive read that new premium volume from a rebounding IPO market should offset four years of shrinkage: volume added at soft-market terms during the exact quarters a rating agency is flagging reserve inadequacy dilutes the book's average adequacy rather than restoring it, since the new accounts are underwritten against the same pre-2024 severity assumptions the rest of the tower is already behind on.

The Renewal Decision

AM Best has effectively already made the call the market itself has been slow to make. Christopher Graham's own language, margins "likely to shrink" in 2026, is the rating agency telling underwriters the floor has arrived, not an actuary's forward-looking assumption. The desk question is not whether the cycle turns; the report says it is turning. The question is whether a given carrier moves ahead of that turn or waits for the loss ratio itself to confirm it, at which point the low and mid-excess layers, sitting 13 to 17 points below their 2013 peak and absorbing the most severity, are already several renewal cycles further underpriced than the primary layer.

For the next renewal, that argues for three specific moves rather than a general rate call. Push any increase into the low and mid-excess layers rather than spreading it evenly across the tower, since that is where the gap between price and severity is widest. Ask for the accident-year 2023 and 2024 loss triangles broken out by claim-closure status rather than incurred totals alone, since AM Best's flag is about the pace of closure, not just the reported reserve level. And price new transactional business, IPO and de-SPAC alike, off current severity trend rather than the soft-market comparables still sitting in most underwriting files. The next public data points to watch are AM Best's follow-up special report, typically issued around mid-year, and the year-end 2026 statutory filings and first-quarter 2027 10-Qs, where any further accident-year 2023-2024 strengthening will show up as a specific, disclosed reserve charge rather than a qualitative warning.