Skyward Specialty's Loss Pick Rose 1.3 Points as Margins Improved

Skyward Specialty's core segment loss ratio rose to 62.6% in the second quarter of 2026 from 61.3% a year earlier, and the same segment's combined ratio still improved, to 86.9% from 88.3% (Form 10-Q, August 7, 2026). The expense ratio fell 2.7 points over the same stretch. Growth in accident and health and global agriculture raised the loss pick, and paid for it twice over.

Key Takeaways

• Segment underwriting income reached $49.6 million against $34.5 million, a 43.8% increase on a book whose loss ratio deteriorated. The expense side gave back 2.7 points against 1.3 points of loss.
• Accident and health plus global agriculture were 31.1% of segment gross written premium, against 20.1% a year earlier, while global property plus captives fell from 27.5% to 20.3%.
• Two unrelated 1.3-point figures appear in the quarter, and much of the secondary coverage ran them together: the loss ratio rose 1.3 points, and catastrophes accounted for 1.3 points of the ratio itself.
• Net retention rose to 65.6% from 58.0% in a quarter when global property catastrophe reinsurance rate on line fell about 16%, the steepest annual decline since the late 1990s.

Two Different 1.3-Point Figures

The quarter produced two unrelated 1.3s. The segment's loss and loss adjustment expense ratio rose 1.3 points year over year. Separately, catastrophe losses accounted for 1.3 points of that 62.6% ratio. They are not the same number and they point in opposite directions.

The entire adverse move sits in the non-catastrophe line, and the filing is unambiguous about the cause: the increases "were due to shifts in business mix, primarily from growth in the accident & health and global agriculture divisions, both of which generally have higher loss ratios." CFO Mark Haushill put the same point to analysts: "The non-CAT Loss Ratio of 61.3 was up 1.4 points year-over-year, driven by business mix" (earnings call, August 5, 2026).

That sits beneath the release's headline, a 46% jump in diluted operating earnings per share. A carrier deliberately moved premium weight toward two divisions it knows carry higher loss ratios, absorbed the loss-cost consequence, and came out with more underwriting margin than before.

The Expense Ratio Paid for the Loss Ratio

The weights show a larger rebalance than the growth rates suggest. Accident and health plus global agriculture together were 31.1% of segment gross written premium, against 20.1% a year earlier, an 11-point swing in twelve months. Global property plus captives fell the other way, from 27.5% to 20.3%. Skyward did not simply grow into new lines. It substituted, and it substituted out of the two divisions carrying the most catastrophe and severity tail in the book.

The property side was priced for it. Global commercial property rates fell 12% in the quarter and casualty rose 2%, the eighth consecutive quarter of composite decline. CEO Andrew Robinson was blunt on the call: "there's just too much capital knocking around in the insurance industry."

The size of the shift lets you back into how far apart the two halves of the book run. A segment loss ratio is a premium-weighted average of division picks, so a mix shift of known size implies a differential of calculable size. One adjustment first: segment net earned premium grew 28.0% while gross written grew 14.2%, so the earned mix is still catching up and the weight shift in earned premium is smaller than the 11-point written one.

Attributing the full 1.4-point non-catastrophe rise to mix, and assuming the displaced divisions held their picks flat, puts accident and health and global agriculture in the mid-to-high 70s on loss ratio against divisions running in the mid-to-high 50s. That is exactly where a medical stop-loss book and an assumed crop quota share belong. Nothing in the arithmetic requires any division to be deteriorating.

The expense side gave back twice what the loss side cost: 2.7 points against 1.3. Net policy acquisition costs fell to 14.0% of earned premium from 15.1%, and other operating expenses to 10.9% from 12.8%. The mechanism is structural rather than a one-quarter accident. Global agriculture growth came through what Robinson described as "effectively a quota share reinsurance solution amongst the major AIPs" on the US dairy livestock price protection program, a federally reinsured product. Assumed quota share on a government-rated program carries almost no origination expense, and its loss ratio is set by the federal rating structure rather than by Skyward's underwriting.

Medical stop-loss behaves the same way: high loss ratio, thin acquisition load, and a premium base that grows with medical trend. Aegis Risk put 2024-to-2025 stop-loss premium increases at 8.8% to roughly 10.5% depending on deductible. For anyone benchmarking a specialty carrier the consequence is that the target loss ratio is not a company-level constant. It is a weighted average that moves whenever the mix does, and comparing a carrier's loss ratio against its own prior year is close to meaningless when a tenth of the book changes divisions inside four quarters.

The Retention Decision

The clearest evidence that Skyward considers the retained book calmer sits on the reinsurance line. Net retention on a written basis reached 65.6% in the quarter against 58.0% a year earlier, and 65.2% for the first half against 60.9%. Segment net written premium grew 25.9% while gross written grew 14.2%.

Had the segment ceded at last year's rate, its $667.8 million of gross premium would have produced about $387 million net rather than the $427.0 million it actually retained, a difference of roughly $40 million in a single quarter.

That decision runs against the price signal. Global property catastrophe rate on line fell about 16% across the 2026 renewals through July 1, the steepest annual decline since the late 1990s. Reinsurance has rarely been cheaper in real terms, and Skyward bought proportionally less of it. A carrier does that when the volatility it was buying protection against has left the portfolio, not when it is economizing.

The complication is what that leaves the combined ratio measuring. The improvement to 86.9% is real, and it is the only comparison that survives a rebalance of this size, because it captures the acquisition-cost offset the loss ratio alone discards. But it now rests on two things that are not underwriting: a federally set rating structure Skyward does not control on the agriculture side, and a retention decision that keeps more of a book whose catastrophe exposure has been deliberately reduced.

Both are defensible and both are reversible by someone else. A change in the federal program's rating, or a hurricane season that reaches the reduced-but-not-eliminated property book, would move the loss ratio without any change in Skyward's own selection. The quarter shows a carrier that has traded catastrophe volatility for rate-structure dependence, which is a different risk rather than less of one.