MGIC Investment: part of the profit is reserve releases that will not last, and the verdict is hold at $28.64

Six weeks ago this site rated MGIC a Hold at $28.48. It closed at $28.64 on November 3, and the third quarter results released after the close on November 4 do not change the answer. They do sharpen the question. MGIC reserves every new delinquency as if 7.5% will end in a claim; recent groups have ended at 2% to 4%. The difference comes back as profit: $34 million in the third quarter and $108 million this year, about a sixth of pre-tax income.

At $28.64 on the November 3, 2026 close, MGIC trades 11.2% below my Base value of $32.25. The Bear value is $22.17, 22.6% below the price, and the Bull value is $40.46, 41.3% above it. I assign no probabilities to the cases. The verdict is Hold, with new money waiting for $22.57 or less.

Valuation route: a U.S. private mortgage insurer valued from book value and normalized earnings. Normal earnings use a through-the-cycle claims cost instead of this year's reserve releases; the justified price to book follows from that return, the cost of equity and growth. The cross-check capitalizes normalized earnings per share. Values are present fair-value estimates, not 12-month price targets.

What MGIC is

MGIC insures mortgage lenders, and in practice Fannie Mae and Freddie Mac, against losses when a borrower with less than 20% down stops paying. The borrower pays the premium, usually monthly, until the loan is paid down or the home gains enough value for the cover to be cancelled. MGIC founded the modern private mortgage insurance industry in 1957 and is one of six active insurers, with about $307 billion of insurance in force across about 1.1 million loans.

The business has three moving parts. Premiums come in as long as loans stay insured, so persistency, the share of policies still in force a year later, matters as much as new business. Claims arrive only when a delinquent loan ends in a sale for less than the debt, so home prices matter more than unemployment alone. And capital is set by the GSEs' rules, known as PMIERs: MGIC must hold assets above a required minimum, and it uses reinsurance to cut that requirement by about half.

What changed in Washington

On October 22, 2026 the President signed an executive order on housing costs. It allows the FHFA to have Fannie Mae and Freddie Mac build a pilot that identifies borrowers who qualify to cancel their mortgage insurance, using automated home valuations, and requires servicers to tell them. The pilot would apply only to loans made after it starts. Borrowers already have the right to cancel, but many do not ask, so a proactive program would shorten policy lives on future loans. The text was narrower than an earlier draft, and the acting FHA commissioner said in September that the administration is comfortable with today's mortgage insurance structure, which lowered the risk of an FHA premium cut.

Mortgage rates went the other way. The 30-year fixed rate passed 7% in late September and averaged 7.32% in the week to October 22. High rates keep existing borrowers from refinancing, which helps persistency, and slow new business.

The latest results

The delinquency rate rose to 2.52% from 2.32% a year earlier, with 27,814 loans behind on payments, and new notices rose to 14,412 from 13,582 a year earlier. Management expects seasonally weaker credit in the second half, and this is consistent with the slow normalization it has described for three years, but it is the direction that matters for the reserve releases. Insurance in force grew to $307.3 billion, persistency fell to 82.5%, and new insurance written was $17.1 billion.

Book value per share fell slightly in the quarter, to $24.06 from $24.27, despite the profit. The cause is the bond portfolio: unrealized losses rose to $396 million as rates climbed, from $168 million a year earlier. Excluding those marks, book value grew to $25.68 a share, up 8.9% in a year.

Which numbers to trust

• Losses are reported after releases. Losses incurred were 10.0% of premiums in the third quarter and about 10% over the last four. New delinquencies alone cost 24% of premiums at MGIC's own claim assumption. The truth over a cycle lies in between, and I use 20%.
• Book value is understated by the bond marks. The $1.62 a share of unrealized losses will flow back as the bonds mature if they are held, so I value the company on book value excluding them, $25.68.
• GAAP and adjusted earnings agree. Net income of $175.0 million and adjusted operating income of $174.7 million differ only by small investment gains and losses, so there is no hidden adjustment to undo.
• Return on equity is flattered by releases. The reported 14.2% falls to about 12% with my normal claims cost.

Capital, reinsurance and shares

MGIC holds far more capital than the GSEs require and keeps sending it back. Over the past year it bought back about a tenth of its shares at around 1.1 times book. That is sensible while normal returns stay above the cost of capital, and the reinsurance program, already bought out to 2029, limits how much of a downturn would reach the balance sheet.

Primary valuation: book value and normal earnings

I rebuild a normal year from today's run rate: premiums of $954 million and investment income of $244 million, the third quarter times four; claims at 20% of premiums; costs of $182 million, about the third quarter's pace; interest of $36 million; and 21% tax. That gives $623 million of net income, $3.09 a share and a 12.1% return on book value excluding bond marks. With a 10% cost of equity and 2% growth, the justified price is 1.26 times book. The Base value is $32.25.

• Why 20% for claims: it sits between the roughly 10% MGIC actually incurred after releases and the 24% it books on new delinquencies. In a recession the number would be far higher for a year or two; that belongs in the Bear, not in every year.
• Why 10%: the beta is about 0.6, but mortgage insurance is a single bet on U.S. housing credit that nearly broke the industry in 2008. I round up for that.
• Why 2% growth: insurance in force grows 2% to 3% a year and MGIC pays out roughly everything it earns, so book value per share grows mainly through buybacks.

Rules of thumb from the model: each 5 points of claims cost move the value by about $2.30 a share, and each point of cost of equity by about $3.60.

Cross-check: earnings power

The second view ignores book value and growth. Normal earnings of $3.09 a share, paid out in full and capitalized at the 10% cost of equity, are worth $30.94, 4.1% below the Base. The two agree because MGIC earns only a little more than its cost of capital in a normal year, so most of the value is the book itself.

Valuation triangulation

• Primary (book value and normal earnings, Base): $32.25. This is the central value.
• Cross-check (earnings power): $30.94, 4.1% below.
• Nearby assumptions (claims 15% to 25% of premiums, cost of equity 9.5% to 10.5%): $28.16 to $36.90.
• Market: 1.12 times book excluding bond marks, 1.19 times reported book.

What is the market already pricing in?

The market assumes claims of about 28% of premiums in every year, more than MGIC books on new delinquencies today and close to three times what it has actually incurred. That leaves room for a normal credit cycle. It does not leave much room for a deep one, and it is not a big enough discount for a business whose bad years are very bad.

Bear, base and bull scenarios

• Bear: a housing downturn takes $2 a share off book value, then claims settle at 35% of premiums, premiums are 3% lower, costs $190 million, 1.5% growth.
• Base: as described above.
• Bull: credit stays benign, claims run at 12% of premiums, premiums 2% higher, costs $178 million, a 9.5% cost of equity and 2.5% growth.

The Bear combines a book loss with lower normal returns on purpose, because a recession would hurt both at once; I add no further recession discount on top.

How the buy price is set

Confidence is B, which carries a default margin of safety of 30%. I add nothing to it, because the housing cycle already sits in the Bear and the claims assumption. That gives a maximum buy price of $22.57.

With claims between 15% and 25% of premiums and a cost of equity between 9.5% and 10.5%, the same margin gives buy prices from $19.71 to $25.83. The price of $28.64 is above all of them, so the decision does not depend on which is right. The 52-week range is $24.69 to $31.89.

Dividend

MGIC pays $0.17 a quarter, $0.68 a year, a 2.4% yield at $28.64, raised for six straight years. The dividend costs about $137 million a year against normal earnings of about $620 million, so it is well covered; the buyback is the bigger part of the return and the first to shrink if credit turns.

Reasons to own MGIC

• Capital well above requirements: $2.8 billion of PMIERs excess and reinsurance bought out to 2029.
• A shrinking share count: about 11% fewer shares than a year ago, bought near 1.1 times book.
• Conservative reserving: new delinquencies are booked at a 7.5% claim rate while recent ones end at 2% to 4%.
• High mortgage rates keep borrowers in their loans, which keeps premiums coming on the existing book.
• The price needs claims near 28% of premiums to be fair, a cushion against normal credit.

What could go wrong?

• Housing and jobs. A recession with falling home prices turns delinquencies into claims. Today almost none do: MGIC reserves new delinquencies at a 7.5% claim rate and recent groups have ended near 2% to 4%. In a downturn that gap closes from the other side.
• The end of reserve releases. Favorable development has supported earnings for several years. If credit simply normalizes, reported profit falls toward my normalized figure even without a recession.
• Policy. A proactive cancellation program, lower GSE fees that favor FHA loans, or changes to capital rules would all reduce returns. The October order affects only future loans, but its details are not written yet.
• Premium yield. Net premium yield has drifted down as more premium is ceded to reinsurers and older, higher-priced policies run off.
• Buybacks at a premium. MGIC buys back roughly its whole profit at about 1.1 times book. That adds value only while returns stay above the cost of capital.

Management execution

• Expenses are falling: underwriting and other expenses were $93.7 million in the first half of 2026 against $105.2 million a year earlier, and the full year is guided toward the low end of $190 to $200 million.
• The reinsurance program is broad and renewed early, including excess of loss cover already bought for 2027 business, and it cut the capital requirement by about $3.1 billion at mid-year.
• Capital return is consistent: share count down about 11% in a year and six straight years of dividend increases.
• Reserving is conservative, which is why releases keep coming. That is a strength for the balance sheet and a reason to discount the reported profit.
• Growth is limited by the market, not by management: insurance in force grows about 2% to 3% a year.

On my ten-point management check (two points each), MGIC scores 15 of 20: credit for capital discipline, reinsurance, cost control and disclosure, less for growth and for buybacks above book.

Stock Analyza scorecard

The valuation score is 60 of 100.

Return on capital scores low because my normal return of about 12% is only two points above the cost of equity; the reported 14% depends on reserve releases. Business quality and valuation are shown separately and are not averaged.

Confidence score

The grade measures how well the model is identified, not the chance of a loss. MGIC's disclosure is unusually complete; the uncertainty is the cycle, not the data.

What I would watch from here

The earnings call on November 5 should explain the rise in new delinquency notices and confirm the capital plan for 2027. Over the next quarters the reserve release line matters most: as long as older delinquencies cure better than reserved, reported profit stays above my normal level.

The three most important thesis breakers

1. A housing downturn that turns the 7.5% claim assumption from conservative to optimistic.
2. Policy changes, starting with the cancellation pilot, that shorten policy lives on new business.
3. Buybacks that continue at a premium to book into a credit downturn.

Stress test and independent check

I recomputed every scenario with a separate script that values the company as book value plus the discounted excess return instead of a price to book ratio, and the results match to the cent. That checks the arithmetic, not the forecasts.

The weakest point is the claims cost, because five points move the value by about $2.30 and the last decade gives little evidence of what a normal recession costs this book. The strongest bull argument is that the market already assumes claims near 28% of premiums. The strongest bear argument is that mortgage insurance has long quiet periods and short violent ones, and a 30% margin of safety is the minimum for that.

Final verdict: hold at $28.64

MGIC is a well-run insurer with more capital than it needs, and its stock is modestly below what a normal year is worth. Part of today's profit is the reserving of the past being too careful. That is a good sign about the balance sheet, and a reason not to pay for it twice.

Verdict: HOLD at $28.64, watchlist for new money. Base $32.25, 12.6% above the price, Bear $22.17, Bull $40.46, cross-check $30.94, Confidence B (20/25). Required margin of safety 30%, maximum buy price $22.57, range $19.71 to $25.83 under nearby assumptions.

New money buys only at or below $22.57. For an existing holder the original cost is irrelevant, and a Base value above the price supports holding while capital stays strong and credit normalizes slowly. The next tests are the call on November 5 and the delinquency trend through the winter.

Analysis history

What changed: the price barely moved, book value excluding bond marks rose from $25.08 to $25.68, and this analysis values normal earnings with an explicit claims cost instead of a price to earnings multiple, which raises the Base from $29.50 to $32.25. A higher confidence grade sets the margin of safety at 30% instead of 25%, so the buy price rose only slightly, to $22.57. The verdict is unchanged.

Sources

Primary sources are MGIC's third quarter 2026 results release (November 4, 2026), its second quarter 2026 release, quarterly report on Form 10-Q and earnings call (July 29 and 30, 2026), and the third quarter call announcement. The executive order is described in Freddie Mac's quarterly report (October 29, 2026) and reports by HousingWire and CNBC (October 23, 2026). All were accessed through Bigdata.com, and the filings are available on SEC EDGAR.

The reference market price is $28.64, the November 3, 2026 NYSE close, from market data via Bigdata.com. Mortgage rates are from Freddie Mac's weekly survey as reported in October 2026.