AIG's $145 Million Reserve Release Comes With an Excess Casualty Catch

AIG booked $145 million of net favorable prior-year reserve development in the second quarter of 2026, up from $112 million a year earlier, pushing its General Insurance accident-year combined ratio to 88.1% (AIG, SEC Form 8-K, August 6, 2026). The release came from workers' compensation and property lines; U.S. excess casualty moved the other way, with slight reserve strengthening in the same quarter.

That single line, favorable in the short tail and adverse in the long tail, is the same seam actuary.info has now traced through five large-cap commercial writers this reporting season. AIG's second quarter under new CEO Eric Andersen produced a genuinely strong headline: $686 million of General Insurance underwriting income, up 10% year over year, on $7.5 billion of net premiums written, up 9% (AIG, August 6, 2026). But the segment tables tell a more divided story than the 88.1% suggests, and the division runs along exactly the fault line this site has been tracking since Travelers and Chubb released reserves in mid-July while Everest and CNA built them.

The Print: A Clean Beat With a Corebridge Asterisk

AIG's headline combined ratio of 89.0% improved 30 basis points from 89.3% a year earlier, "largely due to higher favorable PYD and an improved expense ratio, partially offset by higher catastrophe-related charges," the company said in its release (AIG, August 6, 2026). Catastrophe charges rose to $210 million, or 3.4 loss ratio points, from $170 million a year earlier, including $75 million of net losses tied to the Middle East conflict. Adjusted after-tax income per diluted share reached $2.00, up 10%, while net income per diluted share fell to $1.78 from $1.98, a decline the company attributed to swings in the fair value of its remaining Corebridge Financial stake and other equity securities rather than to underwriting. AIG sold that remaining Corebridge interest on May 7 for roughly $710 million in proceeds, closing out the multi-year unwind of its former life and retirement subsidiary (AIG, August 6, 2026).

CEO Eric Andersen, who succeeded Peter Zaffino as president and CEO effective June 1, 2026 after Zaffino moved to executive chair (Insurance Journal, January 6, 2026), used the release to frame the quarter against a shifting market rather than a triumphant one. "Our strong quarterly results demonstrate our ability to perform well in the current market, which has transitioned from an extended phase of broad positive pricing into a more selective environment," Andersen said (AIG, August 6, 2026). Net investment income fell to $1.127 billion from $1.466 billion, reflecting the Corebridge exit, while book value per share rose to $77.39 from $74.14 and AIG returned $904 million to shareholders through $641 million of buybacks and $263 million of dividends. Return on equity slipped to 9.4% from 11.0%, though core operating ROE, which strips out the Corebridge and investment-portfolio noise, held at 11.1%.

Three Segments, Three Different Reserve Stories

AIG's General Insurance segment splits into North America Commercial, International Commercial, and Global Personal, and the second-quarter combined ratio moved in a different direction and for a different reason in each one.

Segment	Combined ratio	Change	AYCR, as adjusted	AYCR change	Underwriting income
North America Commercial	84.0%	−1.9 pts	86.7%	+0.5 pts	$372M (+24%)
International Commercial	91.3%	+5.4 pts	87.3%	+2.3 pts	$200M (−33%)
Global Personal	92.9%	−5.6 pts	91.2%	−4.9 pts	$114M (+356%)

Source: AIG second-quarter 2026 earnings release, SEC Form 8-K, August 6, 2026. Change columns are year over year; negative combined ratio change is an improvement.

North America Commercial, the segment that houses AIG's U.S. excess casualty book, is where the release-versus-reserve tension is sharpest. Its combined ratio improved 190 basis points to 84.0%, which the company attributed to "higher favorable PYD, lower catastrophe-related charges and lower general operating expense ratio" (AIG, August 6, 2026). But its accident-year combined ratio, the metric that strips out both catastrophes and prior-year development to isolate the current period's own loss pick, moved the other way, rising 50 basis points to 86.7%, driven by "higher acquisition ratio and AYLR due to changes in business mix, in addition to rate pressure, particularly in Property" (AIG, August 6, 2026). Underwriting income in the segment still climbed 24% to $372 million on 9% premium growth to $3.1 billion, so nothing here reads as distress. It reads as a headline ratio flattered by reserve release and a lighter catastrophe quarter while the segment's own current-year loss ratio quietly got worse, the identical mechanism this site found in Chubb's North America Commercial segment, where the combined ratio worsened to 85.4% even as the consolidated number improved on releases concentrated outside U.S. commercial casualty.

International Commercial had the weaker quarter, and reserves are not the reason. Underwriting income fell 33% to $200 million from $300 million even as net premiums written grew 11% to $2.6 billion, and the segment's combined ratio worsened 540 basis points to 91.3%. AIG attributed the move to "higher catastrophe-related charges, primarily due to losses related to the Middle East conflict, higher AYLR, reflecting rate pressure, and higher acquisition ratio" (AIG, August 6, 2026). The AYCR moved from 85.0% to 87.3%, a 2.3-point deterioration that predates and is separate from the catastrophe load, concentrated in financial lines where the company has flagged continued rate pressure. A segment that grows premium by double digits and still loses a third of its underwriting income is not a reserving problem; it is a pricing-and-catastrophe problem showing up before the reserve actuaries even get involved.

Global Personal is the one segment where the improvement is real and not reserve-driven. Its combined ratio improved 560 basis points to 92.9% and its AYCR improved 490 basis points to 91.2%, which AIG credited to "lower AYLR and acquisition ratio reflecting earn-in of improved High Net Worth business commission terms" and a lower expense ratio (AIG, August 6, 2026). Unlike the two commercial segments, personal lines carried no material prior-year development in the quarter, so a genuine underwriting improvement, renegotiated distribution economics earning into the book, is doing the work the reserve line did elsewhere.

Where AIG Lands in the Q2 2026 Casualty Divergence

Four large-cap carriers split visibly on casualty reserves this reporting season: Travelers converted $578 million of prior-year reserves into profit and Chubb added $283 million, while Everest booked roughly $200 million of North America casualty treaty strengthening and CNA took a $77 million after-tax legacy mass-tort charge, its second consecutive year of a similar-sized action (see actuary.info's read on the Chubb, Everest, and CNA split). AIG's $145 million net favorable print puts it in the release column in aggregate, but the composition looks nothing like Travelers' or Chubb's.

Carrier	Net prior-year development	Primary driver
Travelers	+$578M favorable	Workers' comp and property; no pressure reported in umbrella or commercial auto
Chubb	+$283M favorable	Concentrated in personal lines and international, not North America commercial casualty
AIG	+$145M favorable	Workers' comp and property/special risks; slight strengthening in U.S. excess casualty
Everest	−$200M adverse	North America casualty treaty, older accident years
CNA	−$77M adverse (after-tax)	Legacy mass tort, second consecutive year

Figures as reported by each carrier for the second quarter of 2026. Sources listed below.

What AIG's print actually resembles most is not Travelers or Chubb's clean release, but Hartford's Q2 2026 result, where a four-point sequential combined-ratio improvement masked $46 million of adverse general liability development and $26 million of adverse commercial auto liability development, both concentrated in the 2023 and 2024 accident years, funded by more than $110 million of favorable workers' compensation development. AIG's aggregate is smaller relative to its $7.5 billion of quarterly premium than Hartford's swing was to its own book, and the company used the word "slight" for the excess casualty move rather than disclosing a dollar figure, which is itself informative: a genuine build worth naming in the release, the way Everest and CNA did, gets a number attached; a soft signal that management is watching but not yet quantifying gets an adjective. That is a real distinction, not a rhetorical one, and it puts AIG's excess casualty book in the same "early signal, not yet a charge" category that Everest's CFO described choosing to act on ahead of its scheduled annual review, one quarter and one materiality threshold behind.

Book mix explains part of why AIG's excess casualty signal stayed small while Everest's treaty book moved by nine figures. Everest's strengthening sits in North America casualty reinsurance, which aggregates losses across many primary cedents and is mechanically more exposed to a handful of adverse jurisdictions or large-account claims than a single primary carrier's diversified excess casualty layer. AIG writes excess casualty as a primary or lead-umbrella insurer across a broad portfolio of accounts, which dilutes any single adverse claim or venue far more than a reinsurance treaty does. The same underlying loss-cost signal, whatever is driving Everest's older accident years to develop adversely, would be expected to show up smaller and later in a primary excess book like AIG's, which is consistent with "slight strengthening" rather than a nine-figure charge in the same quarter.

What a Small Signal in a Large Book Means for Reserving

The actuarial question AIG's print raises is not whether $145 million of net favorable development is adequate; against $7.5 billion of quarterly premium and a General Insurance book with tens of billions of dollars in earned premium annually, it plainly is a modest number in either direction. The question is what a reserving actuary should infer from a favorable aggregate that specifically excludes the one long-tail casualty line named in the release. Excess casualty loss picks are set years before claims mature, so a signal small enough to describe only in adjectives this quarter is exactly the kind of data point that either resolves to noise by the time the annual long-tail review runs, or becomes the seed of next year's larger action, the pattern CNA's second consecutive mass-tort charge already illustrates on a different exposure.

North America Commercial's own AYCR deterioration, 50 basis points on higher acquisition ratio and accident-year loss ratio tied to business mix and property rate pressure, is the more immediate signal for pricing actuaries, because it sits in the current accident year rather than in a reserve estimate for older ones. A segment where the headline combined ratio improves on release and lower catastrophes while the current-year pick simultaneously gets worse is telling the same story Chubb's North America Commercial segment told a month earlier: property retrenchment and casualty mix shift are moving the accident-year loss ratio in a direction the aggregate combined ratio does not show, and the aggregate is the number most readers will remember from this release.

For actuaries pricing or reserving adjacent excess casualty and umbrella business, AIG's quarter adds a fifth large-cap data point to a pattern that is now consistent across the reporting season: wherever a carrier discloses reserve movement by line, the long-tail casualty line is the one moving unfavorably or, at best, moving less favorably than the aggregate. Travelers was the exception that proves the rule, explicitly stating no pressure in umbrella or commercial auto; every other carrier that has reported this quarter, AIG included, found at least one long-tail casualty line developing the wrong way even while its consolidated number looked fine.

Capital Return Against an Unresolved Casualty Signal

AIG returned $904 million to shareholders in the quarter, split between $641 million of buybacks and $263 million of dividends, while diluted shares outstanding fell to 533.5 million from 577.9 million a year earlier (AIG, August 6, 2026). That pace of repurchase, funded in part by the roughly $710 million of Corebridge proceeds that closed in May, is the kind of capital action a company takes from a position of confidence rather than one bracing for a reserve charge, the same framing Everest's management used when it paired a $200 million casualty strengthening with $585 million of operating income and continued buybacks in the same quarter. A carrier that believed its excess casualty signal was about to become a nine-figure problem would be more likely to slow repurchases and preserve capital ahead of the finding than to keep buying at scale while calling the movement "slight."

Andersen's own framing of the market as having "transitioned from an extended phase of broad positive pricing into a more selective environment" lines up with the pricing data actuary.info has tracked from Marsh's global insurance market index, which put U.S. casualty rates excluding workers' compensation up 11% in the second quarter even as the global commercial composite fell 6% for an eighth consecutive quarterly decline. AIG is writing into a casualty market where rate is still running well above the broader cycle, which is the same condition that let Chubb's and Travelers' current-year loss picks run conservative enough to produce releases elsewhere in their books. Whether AIG's own excess casualty pick has been running similarly conservative, or whether the "slight strengthening" this quarter is the first crack in that conservatism, is not resolvable from a single quarter's disclosure. It is the specific question AIG's next long-tail reserve review, and the level of detail the company chooses to disclose about it, will answer.

Further Reading on actuary.info

• P&C Q2 2026 Casualty Reserves: Chubb and Travelers Release While Everest and CNA Build - The four-carrier split this AIG print now extends to five.
• Chubb's 83.8% Combined Ratio Meets a Softening Casualty Book - The same headline-improvement-masks-commercial-segment pattern AIG's North America Commercial segment repeats.
• Hartford Q2 2026: Short-Tail Releases Fund a Casualty Reserve Build - The closest structural match to AIG's own workers' comp release funding a quiet excess casualty signal.
• Everest's Net-Zero Reserve Development Hides a $200M Casualty Build - What a treaty reinsurer's version of the same signal looks like at nine-figure scale.
• CNA's Flat P&C Reserves Hide a $77 Million Legacy Mass-Tort Charge - How a small, adjective-scale signal in one year can become a named charge the next.
• Social Inflation and the Casualty Loss Development Factor Adjustment - The loss-trend mechanics behind why excess casualty picks lag rate and loss-cost shifts by several accident years.