Essent's Provision Doubled for a Reason Nobody Wants to Hear: The Loans Got Bigger
A mortgage insurer near book value tells investors to model a 2–3% claim rate, then explains away a sequential reserve jump with loan-size arithmetic — and quietly stops talking about its P&C 'call option.'
ESNT · Earnings Call · 2026-11-06
The quiet part: reserves nearly doubled, and the stock stalled
Essent Group posted $163 million of net income, or $1.82 per diluted share, for the third quarter of 2026 — a lower print than the second quarter, on an annualized return on average equity of roughly 12%. Book value per share stood at $62.58, up about 9% year over year including dividends. Against a $5.7 billion market cap and $5.6 billion of GAAP equity, the company still trades at something close to its own capital account, which is precisely the setup management keeps pointing at when it explains why it buys back stock instead of writing more business. The number that moved, though, was the provision: $51.7 million in the quarter versus $29.4 million in the second quarter and $44.2 million a year ago, with the portfolio default rate rising to 2.75% from 2.53%. Analysts noticed. Barclays' Terry Ma called it "quite a bit harder than we were expecting" — and what followed was the most instructive exchange on the call.The loan-size alibi — and a new modeling confession
CFO David Weinstock's answer was refreshingly mechanical: “the average loan size is increasing … they have probably grown from an average of like $300 thousand to now to about $400 thousand” — David Bruce Weinstock, Chief Financial Officer · 2026-11-06. Reserving methodology unchanged, loans bigger, dollars of provision up. That is loan size inflation, not credit deterioration — a theme Essent has flagged on prior calls but which now does real work in the P&L. Mark Casale then did something he has not done on the prior five calls in quite this form: he gave the Street an explicit normalization anchor. “we have always said kind of we expect a 2% to 3% claim rate. that is what we price for” — Mark Anthony Casale, Chairman and CEO · 2026-11-06 — translating, with expenses and investment income, into a 35%–45% combined ratio. That is why claim rate sits at the very top of this quarter's keyword set, alongside the new ultimate claim rate framing, with which he flatly separated booked new notices from the number that actually matters: "when we book new notices, that is not what our ultimate … things roll." For a company that spent years boasting of a sub-1% claim rate, publishing a 2–3% target and a 35–45% combined ratio is a maturity statement — an admission that the great benign-credit run is the anomaly, not the base case. Rick Shane of JPMorgan tried to probe where we are cyclically; Casale's reply was that defaults are "close to 22 thousand," new books aren't as good as the pre-2022 vintage, and "we do not see any kind of skeletons in the closet whatsoever." Note this is the same song as May 2026, when he warned “given the seasoning of the portfolio and peak default being 36 to 60 months, you are going to see defaults continue to increase” — Mark Casale, Chairman and CEO · 2026-05-08. Three quarters later, it is happening on schedule.What management stopped talking about
The contrast with recent quarters is the tell. In the March 2026 quarter, Lloyd's and the P&C build-out dominated — a full infrastructure discussion, a retro quota share, diversification-credit arithmetic. This quarter, P&C got one paragraph in the prepared remarks: roughly $320 million of written premiums in 2026, about half earned, a high-90s combined ratio. In Q&A, Casale was explicitly non-committal on expanding appetite, noting some Lloyd's syndicates have "a 100% FAL" and little alignment. The lot of capital that once funded call options is now funding buybacks. Even more striking: artificial intelligence vanished. The August call had a long, animated discussion of token usage, Copilot, Claude, Kiro and EssentEDGE — the kind of thing that gets a specialty insurer invited onto growth screens. It is absent from this transcript entirely. What survives is the concrete version: Vantage scores, which Casale sized as “probably <5% of our originations” — Mark Anthony Casale, Chairman and CEO · 2026-11-06 and which he does not think will "affect the MI industry that much" — a view he has held since at least August, when he called Vantage “a little bit more lenient than FICO score to be sure” — Mark Casale, Chairman and CEO · 2026-08-07. Consistent, but no longer novel.Paid to wait — and the persistency trap underneath
That is the whole thesis in two sentences. Insurance in force ended at $250.1 billion, up just $342 million sequentially and $1.3 billion year over year, with persistency at 83.3% because more than half the book carries a mortgage rate of 6% or lower. Cancellation activity — the subject of the first analyst question, and the basis of MI cancellation entering the keyword set — is mostly just the 2021 vintage amortizing off, peaking around 2028–29 by Casale's own estimate. So the asset base is flat, the capital required to support it is flat, and the free cash flow is enormous: $5.4 billion of effective net cash, down only 6% from a fourth-quarter 2025 peak. Into that gap goes the buyback. $192 million of repurchases in the latest reported quarter, against $193 million of free cash flow — a payout ratio near 100% that management says will exceed 100% for the full year, topped off with the announcement of a fresh $500 million authorization and a $0.35 quarterly dividend. Year to date through October, roughly 8 million shares for about $475 million. At 7.2x price to free cash flow, that is arithmetic, not heroism.we are paid to wait and the way we are kind of managing capital over time. Because remember, insurance in force, it does not grow, the capital needed to support it does not grow.