Private MI Loss Ratios Climb as Cure-Driven Reserve Releases Fade

Essent Group's mortgage insurance loss ratio climbed to 13.6% in the second quarter of 2026 from 7.0% a year earlier, and its provision for losses and loss adjustment expenses rose to $29.4 million from $15.3 million (Essent Group, Q2 2026 results, StockTitan, August 7, 2026). MGIC's loss ratio moved from a net recovery of -1.2% in the second quarter of 2025 to a positive 4.6% this quarter, even after booking $43 million of favorable reserve development (MGIC Investment Corporation, Q2 2026 results, August 2026). Neither shift reflects a credit event; both reflect a reserving mechanism running out of the cure-rate cushion that has held private mortgage insurance loss ratios near zero since 2023.

That mechanism is worth naming precisely, because trade coverage of the July and August 2026 print season has largely treated the higher provisions as an early warning on credit quality. It is closer to arithmetic. Private mortgage insurers reserve for a delinquency the moment a servicer reports it, using an assumed probability that the loan eventually produces a paid claim. For three years, actual cure rates, the share of delinquent loans that return to current status without a claim, ran well above what insurers assumed when they first booked the reserve. Every quarter that a cohort of delinquencies cured better than assumed, insurers released the excess reserve as favorable prior-year development, and that release flowed straight into a suppressed loss ratio. What Essent, MGIC and NMI Holdings reported between July 30 and August 14, 2026 is that cushion thinning across the board, at different speeds and for different reasons at each company.

The Reserving Arithmetic Behind Three Years of Near-Zero Loss Ratios

Private mortgage insurance reserving runs on two separate assumptions layered on top of each other. The first applies to loans that are currently performing: an insurer estimates how many will newly become 60-plus days delinquent in a future period, the new-notice frequency, and holds no case reserve against them until a servicer actually reports the delinquency. The second applies to loans already reported delinquent: the insurer applies a claim rate, sometimes called the delinquency-to-claim rate, to each open notice, representing the probability that the specific delinquency ultimately produces a paid claim rather than a cure. MGIC disclosed that it held its initial claim rate assumption for new delinquency notices received in the second quarter of 2026 at 7.5%, unchanged from prior quarters, even as actual claim rates on notices from two to three years ago ran in the 3% to 4% range (MGIC Investment Corporation, Q2 2026 earnings call, The Globe and Mail transcript, July 30, 2026). That gap, a 7.5% initial assumption against a 3% to 4% actual outcome, is the entire source of favorable development: every seasoned delinquency notice that cures instead of claiming releases the difference between the conservative initial pick and the smaller amount actually needed.

MGIC's own disclosure puts a number on how fast that cure process runs: post-COVID vintages have cured at roughly 90% within four quarters of first becoming delinquent. That is unusually fast by pre-pandemic standards, driven by the same forces that have supported the broader mortgage credit book since 2022, high borrower equity cushions built up through years of home price appreciation, a tight labor market that limited the duration of individual borrower hardship, and loss mitigation programs that resolve delinquencies faster than the foreclosure-driven claim process common before 2020. As long as new delinquency notices kept arriving at a modest pace and each vintage kept curing at close to a 90% rate, insurers could keep releasing reserve from the prior year's over-conservative claim rate assumption faster than new notices required fresh reserve. That is the mechanical reason loss ratios across the sector sat near zero, and in MGIC's case briefly went negative, for most of 2024 and 2025.

Three Carriers, Three Different Prints

The second quarter of 2026 broke that pattern unevenly. Essent's loss ratio nearly doubled year over year even as its default rate stayed essentially flat at 2.53%, against 2.54% in the first quarter (Essent Group, Q2 2026 earnings call, The Motley Fool transcript, August 14, 2026). A flat default rate with a rising loss ratio is itself a reserving-mechanics signature: the number of delinquent loans on the books did not change much, but the reserve held against each one grew, consistent with a smaller favorable adjustment being applied against the same claim rate assumption. NMI Holdings moved the opposite direction on a sequential basis, its loss ratio improving to 8.3% in the second quarter from 13.3% in the first, though it still landed slightly below the 9.0% posted a year earlier (NMI Holdings, Q2 2026 results, GlobeNewswire, July 30, 2026). NMI's claims expense fell to $13.1 million from $20.7 million in the first quarter, and its default inventory was essentially unchanged at 8,020 loans versus 8,044 three months earlier, so the first-quarter spike reads as a one-off reserve item rather than the start of a trend, at least so far.

Insurer	Q2 2026 loss ratio	Prior-year quarter	Sequential quarter
Essent Group	13.6%	7.0% (Q2 2025)	n/a
MGIC Investment	4.6%	(1.2%) (Q2 2025)	14.1% (Q1 2026)
NMI Holdings	8.3%	9.0% (Q2 2025)	13.3% (Q1 2026)

MGIC's own sequence tells the sharpest version of the story. Its first-quarter 2026 loss ratio of 14.1% was itself a jump, and the $43 million of favorable development booked in the second quarter, tied explicitly to "better-than-expected cure activity" on delinquency notices received in 2025 (MGIC Investment Corporation, Q2 2026 earnings release, August 2026), pulled the ratio back down to 4.6% rather than eliminating it outright the way similar releases did through 2024 and early 2025. The favorable development is real and recurring. It is simply no longer large enough to fully offset the reserve requirement generated by new notices, which is a different claim than the reserve being wrong.

Why NMI's Combined Ratio Still Looks Comfortable

NMI's headline combined ratio improved 110 basis points to 27.7% in the second quarter, and that number alone would suggest nothing in the sector has changed (NMI Holdings, Q2 2026 results, GlobeNewswire, July 30, 2026). The reason the combined ratio moves so little even as the loss ratio swings by several points a quarter is structural: NMI's expense ratio ran 19.4% in the second quarter, more than double its 8.3% loss ratio, so the loss ratio has to move a large number of points before it registers meaningfully in the combined ratio at all. A private mortgage insurer's expense base, sales and underwriting staff, technology, and ceding commissions on reinsurance, is largely fixed relative to premium and does not fluctuate with the delinquency cycle the way loss costs do. That expense-ratio floor is doing real work to keep MI combined ratios in the high 20s and low 30s across the sector even as the loss-ratio numerator underneath quietly drifts upward, which means combined ratio alone is a poor instrument for spotting the reserving shift happening inside these results.

Management at NMI is not projecting the current quarter's improvement forward without caveat. "We'd expect our default population to trend a bit higher from here," the company said on its July 30 call, citing normal seasonal patterns into the second half of 2026, even while characterizing the underlying portfolio as "exceptionally high-quality" (NMI Holdings, Q2 2026 earnings call, The Motley Fool transcript, July 30, 2026). That guidance, combined with named regional pressure in Florida, Texas, and parts of the Sun Belt and Mountain West, suggests NMI's second-quarter improvement is a comping effect against an elevated first quarter rather than evidence the sector-wide reserving pressure has bypassed the company.

What Happens When New Notices Outpace Cures

The forward risk in this mechanism is not that cure rates collapse; it is that the rate of new delinquency notices grows faster than the rate at which existing notices cure, even if the cure rate itself stays elevated by historical standards. Every carrier still books a claim rate assumption meaningfully above the actual claim rate being realized on seasoned vintages, which means every carrier is still generating some favorable development each quarter. But favorable development is a flow measured against a shrinking base: if the delinquency inventory that can still produce a release keeps aging out and closing while the inflow of new notices holds steady or rises with normal seasonality, the arithmetic mechanically compresses the release relative to the new-notice reserve requirement, exactly what Essent's flat default rate paired with a rising provision, and MGIC's persistence of a positive loss ratio despite continued releases, are showing in real time. None of the three insurers has reported a change to its underlying claim rate assumption on new notices; MGIC explicitly held its 7.5% initial pick steady. The loss ratio move is a base-effect story, not a re-estimation of ultimate credit risk on the current book.

Reinsurance, PMIERs and Where Ceded Losses Land

Rising gross provisions do not translate one-for-one into rising capital pressure, because private mortgage insurers cede a large share of new risk through quota share and excess-of-loss reinsurance structures approved by Fannie Mae and Freddie Mac under the Private Mortgage Insurer Eligibility Requirements, the capital framework known as PMIERs. Essent reported 97% of its insurance in force subject to reinsurance protection, with its principal operating subsidiary running a PMIERs sufficiency ratio of 172% and $1.5 billion of excess available assets above the required minimum (Essent Group, Q2 2026 earnings call, The Motley Fool transcript, August 14, 2026). NMI reported $3.7 billion of PMIERs available assets against $2.1 billion of risk-based required assets, a $1.6 billion excess, with capital relief flowing through its stack of quota-share, excess-of-loss, and insurance-linked note transactions running from 2018 through 2026 (NMI Holdings, Form 10-Q, Q2 2026). Because ceded loss reserves move in the same direction as gross reserves under these treaties, higher gross provisioning this quarter also increases the losses insurers pass to reinsurance partners, which is a large part of why PMIERs sufficiency ratios above 170% at both companies have not moved in a way that signals stress. That capital cushion, and the mechanics of how it built up, is covered in detail in actuary.info's July 2026 analysis of PMIERs capital relief across the sector, which found reinsurance and insurance-linked notes covering roughly half the six major insurers' combined required assets.

What a Normalizing Loss Ratio Means for Pricing and Buybacks Into 2027

The practical question for actuaries pricing new mortgage insurance business, and for investors evaluating capital return programs, is what a genuinely normalizing loss ratio does to margin once the cure-rate tailwind is fully spent rather than partially spent. Private MI premium rates were set, largely through rate cards filed in 2023 and 2024, against loss-cost assumptions that already priced in some reversion from the unusually favorable 2021 through 2023 cure experience; the question this earnings season raises is whether the pace of that reversion, several points of loss ratio per year rather than a slow multi-year drift, is faster than those rate cards assumed. If new-notice volume continues at a steady pace into the back half of 2026, as NMI's own guidance suggests, and the pool of seasoned delinquencies still capable of producing outsized releases keeps shrinking, loss ratios in the mid-teens to low-20s, still comfortably profitable by historical MI standards but multiples of the 2024 base, become a more reasonable planning assumption for 2027 than a further return toward zero. That has direct consequences for capital return: Essent's net income of $189.7 million and MGIC's $182.1 million both still comfortably funded dividends and buybacks this quarter, but a book of business priced on the assumption that reserve releases would remain a persistent tailwind rather than a fading one carries less margin for error than the headline profitability suggests, particularly if new insurance written continues to grow, as it did at NMI, where new insurance written rose 29% year over year to $16.05 billion in the quarter.