Fannie Mae

IThe business

Fannie Mae does not lend to home buyers. It buys loans from banks and other lenders, puts them in trusts, issues securities backed by them and promises investors that it will pay the money even if a borrower does not.³ For that promise it charges a guaranty fee. In the first nine months of 2026 it provided US$344.1B to the mortgage market, enough to finance about 1.2 million home purchases, refinancings and rental units, and in 2025 it provided US$409.3B.³⁶ At 30 June 2026 it owned or guaranteed an estimated 24% of single-family mortgage debt in the United States and 21% of multifamily debt.³ The Federal Reserve put total residential mortgage debt at about US$17.0 trillion at 30 September 2025, and the company says it owned or guaranteed about 24% of it.⁶ Freddie Mac, its twin, has a mortgage portfolio of US$3.70 trillion, so the two together are about US$7.8 trillion, roughly 46% of that total, which is what people mean by "about half of US home loans" C (calculated).⁷⁶

It is easy to describe and hard to value, because the company is two things at once. One is a utility-like business with a US$4.1 trillion book, 7,000 employees, steady fee income and a loss record that has been low for a decade. The other is a legal arrangement: since September 2008 it has been run by its regulator, the Federal Housing Finance Agency (FHFA), as conservator, and it is owned in all but name by the US Treasury.⁶ Both matter. I describe the first in this section and the second from the next paragraph on, and I value them together in Section VI.

How the money is made

The company earns almost all of its revenue as net interest income: US$28.6B in 2025, of which the guaranty business is the large part.⁴³ Its book of US$4.1 trillion is carried on its balance sheet as US$4.0 trillion of loans owned by trusts and an equal amount of debt owed to the holders of its securities, so the balance sheet is very large and the profit is small beside it.⁵ The average fee on the single-family book was 49.1 basis points in the third quarter of 2026, and 53.8 basis points on loans bought in the quarter, against 48.5 and 56.3 a year earlier.⁵ A basis point is 0.01% a year. One basis point on the US$3.56 trillion single-family book is about US$356M a year before tax C (calculated). That is the unit to keep in mind: a change of 2 basis points moves earnings by about 5%.

Not all of the fee is kept. Ten basis points of every single-family fee are collected for Treasury under a 2011 law and paid on, until 1 October 2032, and the company must also set aside 4.2 basis points of new loans for affordable-housing funds.⁶ These show up as US$2.8B of "legislative assessments" in the first nine months of 2026, 12% of net revenue C (calculated).³ The company also pays for credit risk transfer: US$1.1B of credit enhancement expense in the nine months, which buys protection against losses.³

Pricing is not set only by the company. FHFA must approve changes to the loan-level fees, can direct other changes, sets minimum fees and told the company in January 2025 not to raise base fees broadly until further notice.⁶ On 22 October 2026 the President signed an executive order directing the FHFA Director, under conditions, to have the company and Freddie Mac set up a mortgage-insurance cancellation pilot, widen a title-insurance alternative and consider cutting fees on some refinancings.³ I mention it because it shows who sets the price.

Credit and interest rates

The company is exposed to credit risk, not to the kind of interest-rate risk a bank has. Borrowers on the single-family book have a weighted average credit score of 753 at origination and a mark-to-market loan-to-value of 51%, with a serious delinquency rate of 0.62% at 30 September 2026, up from 0.58% in June.⁵ The multifamily rate was also 0.62%, and the multifamily provision for credit losses was US$446M in the quarter, up from US$69M a year earlier, on weaker property values and higher rates.⁵ Total provision for credit losses was US$1.9B in nine months. The 30-year mortgage rate stood at 7.03% at the end of September against 6.49% in June, and the company says the quarter’s single-family provision reflects higher actual and projected interest rates.³⁵ A hundred-basis-point rise in rates changes the market value of its net portfolio by only US$0.4B, because the guaranty business is left out of that measure by design.⁶ What a higher rate does is raise provisions, cut volume and mark down the Treasury bonds it holds: other comprehensive losses were US$1.0B in the third quarter C (calculated).³

Home prices matter more. The company estimated national prices rose 3.8% in the first nine months of 2026 and expects 2.7% for the year.³ Its credit loss ratio is low because most loans in the book were made at lower prices and with a down payment: the average mark-to-market loan-to-value is half. It is the loans written from 2022 on, at higher prices, that a price fall would hurt.⁵ I take the loss history as good, and I do not take it as a guarantee.

The conservatorship in four dates

September 2008. FHFA put both companies into conservatorship. The board answers to the conservator, not to shareholders; stockholder votes are suspended and the dividends on common and junior preferred stock were stopped.⁶ Treasury agreed to buy senior preferred stock, to cover any shortfall in net worth, and received a warrant to buy 79.9% of the common stock for a nominal US$0.00001 a share.⁶ The company drew US$119.8B in all, the last of it in the first quarter of 2018.³ August 2012. The agreement was amended so that all of the company’s profit was paid to Treasury every quarter, the "net worth sweep". Through the end of 2025 Treasury had received US$181.4B of dividends, which is 1.5 times the US$119.8B it had paid in, without reducing what the company owes it.⁶ Since 2019 to 2021. In September 2019 and January 2021 the terms changed again so that the company may keep its profit until it meets its capital requirement, but each quarter’s gain is added to the preference instead.⁸ Without that the company would be paying out every dollar; with it, the dollar builds the claim.

That is the reason the stock is unusual. The reported profit is real, US$14.4B in 2025, and the net worth rose from US$6.1B to US$109.0B in nine years.⁴ The common stockholders have received none of it, and under the current terms will not.

IIThe moat

The honest test of a durable business is what it did when conditions were bad. For Fannie Mae the test is unusual, because the bad time was 2008, and the answer is that the business survived and the owners did not. The company lost money on loans written in the boom years, had drawn US$116.1B from Treasury by the end of 2016 and US$119.8B in all, and was kept alive because the American housing market could not work without it.⁹³ The product held up. The shareholders were wiped out in everything but name.

What does protect it

• A charter that nobody else has. It is the only company besides Freddie Mac that can guarantee mortgage securities with the status of a government-sponsored enterprise. Banks treat its debt as a high-quality liquid asset, and the securities trade in the Uniform Mortgage-Backed Securities market that it and Freddie Mac share.⁶ A rival would need Congress to create it.
• Scale in a business that is mostly information. It shares a securitization platform with Freddie Mac, and its fixed costs are spread over a US$4.1 trillion book.³ Its administrative expenses were US$3.6B in 2025, about 13% of net interest income, and the administrative expense ratio was 10.25% of net revenue in the third quarter C (calculated).⁴⁵
• A loss record that has stayed low. The single-family serious delinquency rate at the end of 2016 was 1.20%; at 30 September 2026 it was 0.62%, with a weighted credit score of 753.⁹⁵ The company also moves risk to others: about US$39B of risk in force on credit risk transfer deals and US$201.4B on mortgage insurance at the end of 2025.⁶
• The government itself. This is the main edge and I treat it as a mixed blessing. The company says that its status as a government-sponsored enterprise and continued federal support are essential to its access to debt funding, and that Treasury still has US$113.9B of funding commitment to cover a net worth deficit.⁶³ The securities are not backed by the full faith and credit of the United States, but the market behaves as if they were. That belief is the "implicit guarantee", and it is what lets the company borrow at rates near Treasury’s.⁶

Where it is thin

Almost every edge above belongs to the government and not to the company. FHFA sets the minimum guaranty fees and the return targets, and can change the pricing in a week; in January 2025 it told the company not to make broad base-fee increases.⁶ Treasury takes the profit, and Congress and the President can add obligations. The board owes its duty to FHFA and the FHFA Director chairs it.³ A business that cannot set its prices or pay its owners is a franchise that belongs to someone else. I count the edge as real for the business and conditional for the shareholder.

Market share has also slipped. The company estimated that its share of single-family mortgage-related securities issuance was 30% in 2023, and says its higher capital requirement relative to Freddie Mac and the pricing discipline imposed on it have lowered its share in recent years.⁶ Its share of single-family debt outstanding was an estimated 24% in mid-2026.³ Over the same years the Federal Housing Administration (FHA) loans bundled by Ginnie Mae and private securities have competed for the same borrowers.

The guarantee also cuts against the owner. If Treasury’s support were reduced, the company says its borrowing costs could rise sharply; if it became more explicit, the case for a government-backed company paying private owners 10% on its equity gets weaker.⁶ A great business can be a poor purchase at the wrong price, and a franchise built on a public subsidy is a different sort of great.

IIIThe competitive advantage

The nearest comparison is Freddie Mac, and the comparison is useful because the two have the same owner, charter and regulator. Fannie Mae is larger by 11% on mortgage book and earned US$4.2B in the third quarter against Freddie’s US$3.2B.⁵⁷ It also carries the larger weight: a liquidation preference of US$238.0B against US$150.4B, and a higher capital requirement.³⁷⁶ Net worth is 50% of the preference at Fannie Mae and 54% at Freddie, so neither is close to "paying back" Treasury C (calculated). The shares of both traded together: Fannie Mae closed at $5.11 and Freddie Mac at $4.69.¹¹³

The real competitors are the other places a mortgage can go: Ginnie Mae securities for government-insured loans, private-label securities, and banks that keep loans on their own balance sheets. The company says its share of issuance has fallen in recent years partly because capital rules limit what it can charge.⁶ Fannie Mae’s edge over Freddie Mac is size and a larger multifamily book, US$545.4B against US$509.6B for Freddie Mac, built on a loss-sharing model with lenders.⁵⁷ Credit scoring is a place where it moves with the regulator: a first group of lenders could use VantageScore 4.0 from April 2026 and since September 2026 all approved lenders may, which is FHFA policy and not a private advantage.³

Quality and price are separate questions

As a business this looks good: a thin loss rate, a stable fee, a 10.25% administrative expense ratio and an illustrative return of 11.1% on the capital the rules would require.⁵ That return is the company’s own estimate and it assumes capital that the company does not have. As an investment for the common stockholder it is a different question, since the profit does not reach the common stockholder and the price you pay is for a legal outcome. I come back to this in Section VI, with the two questions kept apart.

IVManagement

The people who run Fannie Mae do not set its strategy. The strategy is set by the conservator. Peter Akwaboah, 56, has been Acting Chief Executive since 22 October 2025, when Priscilla Almodovar left; he joined in 2024 as Chief Operating Officer.⁶ Chryssa Halley is Chief Financial Officer, and Brandon Hamara, a Co-President since October 2025 who also sits on the board, has a background in real estate and homebuilding.⁶⁵ William Pulte, the Director of FHFA, has chaired the board since 17 March 2025; the FHFA general counsel, Clinton Jones, is also a director.³⁶ In the third-quarter release Pulte’s title is "Director, U.S. Federal Housing".⁵

Who they work for

In conservatorship the directors "owe their fiduciary duties of care and loyalty solely to the conservator, and not to either the company or our stockholders", and the company says its business is not managed to maximise stockholder value.⁶ That sentence changes how everything else on this page reads. A manager who owes his duty to the regulator will run the company to meet the regulator’s goals: affordable lending, a strong net worth, a steady market. Those goals are not the owner’s.

How they are paid

The chief executive role is limited by law to a base salary of US$600,000, with no bonus and no equity, since the Equity in Government Compensation Act of 2015.⁶ The other senior executives receive base salary and "deferred salary", paid later and partly at risk against a scorecard that FHFA sets as conservator, and no bonus or stock.⁶ The pay therefore points at meeting the scorecard, not at the share price. The company itself says that the cap makes it hard to keep and hire, and that its headcount fell from about 8,200 to 7,000 during 2025.⁶ Nobody in the room gains if the stock goes up.

How much of their own money is in

Almost none. At 29 January 2026 none of the directors or executives owned as much as 1% of the common stock, and none owned any preferred stock. The table shows zero shares for Pulte, Habib, Hamara, Jones, Malik, Sánchez, Stowell, Stucky and Halley, and 23 shares for Akwaboah R (reported).⁶ I treat that as a fact, not a criticism: they are not allowed to take equity pay, and the stock has no vote.

What they did with capital over a full cycle

They did not have discretion. Over 2016 to 2025 the company built its net worth from US$6.1B to US$109.0B by keeping profit, shrank the accumulated deficit from US$124.3B to US$24.3B, grew its credit risk transfer programmes, and added to its retained portfolio of mortgage assets, which rose US$37.6B to US$132.5B in 2025 after FHFA raised the permitted agency securities holding to US$40B in October 2025.⁴⁶ It made no acquisition and bought back no stock. By any measure that can be judged, the capital was kept, not allocated.

Why the controls are rated "not effective"

The 10-Q reports that the disclosure controls were not effective as of 30 September 2026, and calls it a material weakness. The cause is structural and not an accounting error: FHFA is both conservator and regulator, and may take actions that are material to stockholders and known only to FHFA, so management cannot guarantee that it learns of them.³ The company says it does not expect to remediate this while it is in conservatorship, and describes the mitigations: FHFA reads the filings before they are filed and gave written acknowledgment of the 10-Q, and the FHFA Director chairs the board.³ I read it as a warning about the legal arrangement, not about the numbers. It also means that the company can learn of a major policy move at the same time as the market.

VFinancials

Five things in Fannie Mae’s statements do not mean what they first seem to. The net worth of US$119.6B is not the common stock’s equity: the balance sheet carries Treasury’s senior preferred at the US$120.8B it paid for, while its claim is US$238.0B, so the equity left for everyone else is a deficit of US$1.2B on the books C (calculated).⁵ The earnings per share of US$0.21 for the nine months is a quirk, since it divides a leftover by 5.87B shares that include the warrant, and no cash reaches the shareholder.⁵ The US$4.3 trillion balance sheet is mostly trust assets matched by trust debt. The net interest income includes ten basis points that go straight to Treasury. And two of the ten years carry one-off effects: 2017 includes a one-time US$9.9B tax charge after the December 2017 tax law, and 2021 includes a US$5.1B benefit from lower loss reserves.¹⁰¹¹

The size of the earnings

Net income was US$12.3B in 2016 and US$14.4B in 2025, and in the twelve months to September 2026 US$15.4B C (calculated).⁴³ That sounds flat for a decade, because the book grew and the loss record improved but fee pressure and legislative costs absorbed the gains. Net interest income grew from US$21.3B to US$28.6B, 3.3% a year; administrative expenses went from US$2.7B to US$3.6B and in the twelve months to September 2026 were US$3.3B, after the company cut staff from about 8,200 to 7,000 over 2025.⁴⁶ The provision for credit losses swings the most: a benefit of US$5.1B in 2021, a charge of US$6.3B in 2022, US$1.6B in 2025 and US$1.9B in the first nine months of 2026.⁴³ For a single number I use US$15.9B for 2026, which is the nine months plus US$4.0B I assume for the fourth quarter E (estimate), and I grow it 3% a year. In the third quarter the company calls the return on the capital it would be required to carry 11.1%, which is its own estimate.⁵

What an owner can really take out

Today, nothing. Under the Treasury agreement the company may not pay a dividend, buy back shares or issue stock, and every dollar of net worth growth is added to the preference one quarter later.⁶ The twelve months of US$15.4B earnings therefore do not reach the common stock in any form; they raise Treasury’s claim from US$238.0B to US$241.1B on 31 December 2026 and, at my 2026 estimate, to about US$292B by the end of 2029 E (estimate).³

The capital a business needs just to stand still is large here. A bank-style test, the company’s own enterprise capital rule, asks it to carry US$194B of adjusted total capital including buffers, or US$143B of the highest-quality common equity tier 1 (CET1), and it has an available deficit of US$10.9B of adjusted capital.³ The shortfall is US$205B including buffers and US$173B for CET1, which is a 14-year build at the 2025 net income of US$14.4B and about 13 years at my 2026 estimate, before the requirement itself moves C (calculated).³ So the number an owner could take out is zero today and, if the rules did not change, would be zero for roughly 13 years more. Whether the rules or the terms will change is the subject of the rest of the page.

The balance sheet in order of repayment

The stack shows why the common stock has no claim on most of the profit. The senior preferred’s preference of US$238.0B is 1.99 times the net worth of US$119.6B, and each quarter’s gain increases it again. In the 2019 and 2021 amendments Treasury allowed the company to keep earnings, but the price of keeping them was a bigger claim, so what the company retains is not capital for the common.⁸ After the "capital reserve end date" the rules change: the dividend becomes the lesser of the quarter’s gain in net worth and 10% a year on the preference, which at US$241.1B is about US$24B a year, more than the US$14B to US$17B the company earns.⁶ I read this as meaning that if the terms stayed as they are and the capital rules were met, the common would still get nothing.

The junior preferred and the share count

There are 1.158B common shares outstanding, unchanged since 2017.⁴ The count of 5.87B shares that the company uses for earnings per share includes 4.7B shares that Treasury could buy for a nominal US$0.00001 each, and 26M more issuable on conversion of one preferred series.⁵³ So the denominator for any per-share value is 1.184B if the warrant is cancelled and 5.89B if it is exercised, and my values per share are on the 1.184B existing shares, after giving the warrant holders the rest C (calculated). The public shares are worth US$5.9B at $5.11, and US$30B if the warrant is counted.

The sixteen junior preferred series have a stated value of US$19.13B and rank ahead of the common but behind Treasury. They have been paid nothing since 2008. In August 2023 a jury found for the Fannie Mae preferred holders who challenged the 2012 net worth sweep and awarded them US$299.4M, with US$199.7M of interest added by the court; the appeal was decided against the defendants on 24 July 2026 and rehearing was denied on 16 October 2026, so the case may go to the Supreme Court.³ That is a small sum next to US$19B. It matters as a finding, upheld on appeal, that the sweep harmed those holders.

The share price has done three things

From 2018 to 2024, with nothing decided, the shares ranged from $0.35 to $3.28 at year-end, and the lowest close was $0.35 on 30 December 2022 R (reported).¹ On 21 May 2025 the President said he was giving "very serious consideration" to taking both companies public; the shares closed at $7.45 that day and $11.22 the next.¹⁴¹ The highest close was $15.31 on 11 September 2025, and in 2026 the shares have fallen, to $4.21 on 30 September and $5.11 on 2 November, a fall of 67% from the high C (calculated).¹² I do not have a filing that explains the fall, so I describe the dates, not the cause. What I can say is what the filings say changed in the period: the appeals court affirmed the jury award on the net worth sweep, the chief executive left in October 2025, and the 10-Q of 28 October 2026 still says the conservatorship has no end date and that the company’s future is uncertain.³⁶

VIValuation

First, how this goes badly wrong

Before any numbers, I ask what would hurt an owner most. There are three ways. None of them needs the business to fail.

Nothing changes. The conservatorship has no end date. Under the Treasury agreement the company keeps its profit, and the profit is added to Treasury’s claim, which has grown from US$117.1B at the end of 2016 to US$238.0B.⁹³ From 2018 to 2024 nothing was decided and the shares ranged between $0.35 and $3.28 at year-end.¹ My case A, with a 32% chance, puts $1.00 on the shares in 2030 and nothing in between.

The release comes on Treasury’s terms. The government could end the conservatorship, offer shares to the public, and keep its full preference and its warrant. Then the preference, which at my 2026 estimate would be about US$274B at the end of 2028, is larger than the entire value of the company, about US$168B at 10 times earnings, and the common stock is worth about nothing E (estimate). That is outcome B, 22%. The company’s own filing says this directly: the senior preferred and the warrant "reduces our attractiveness to prospective equity investors", and actions by the conservator or a receiver "could substantially dilute or eliminate any value associated with our existing common stock and preferred stock".⁶

I pay for a headline. The shares rose from $0.35 at the end of 2022 to $15.31 on 11 September 2025, forty-four times, and have since fallen 67%.¹ A large part of the price at any time is the market’s guess about the odds of a favourable decision. That is about what an owner pays, not about the business, and it can move in one afternoon, as it did on 21 May 2025 when the shares went from $7.45 to $11.22 in a day.¹¹⁴

What an owner can really take out

Zero. The business earns US$15.4B a year, the owner of the common can take none of it, and the company may not pay a dividend or buy a share.⁶ What the price pays for is not a stream of cash. It is a claim on what remains after a decision that has not been made. I therefore do not discount a stream of cash flows. I list the decisions, give each an estimate, and value the common stock in each.

What I can’t judge

I cannot judge whether the administration will complete a release before January 2029, what Treasury will demand, how the Supreme Court or Congress will act, or what the market will pay for a company that must raise tens of billions of dollars. The only thing the filings say plainly is that the future "is uncertain", that Treasury must consent to an exit, and that the conservatorship "has no specified termination date".⁶ The administration’s own public comments are only partly documented. On 21 May 2025 President Trump said he would decide "in the near future" whether to take both companies public and named the Treasury Secretary, the Commerce Secretary and the FHFA Director as advisers.¹⁴ The company’s 2025 annual report says that "the Administration has made public comments suggesting it is considering various options for the future of Fannie Mae and Freddie Mac", including liquidation or a merger of the two.⁶ I found nothing in a filing that sets a date or terms, and I have not relied on reports I could not open. That is why the odds below are mine and why I show how the answer moves when they change.

Outcomes are decided by politicians and courts, not by the business. The business is the same in every case: US$15.9B of net income in 2026, growing 3%. What differs is who gets it.

What would have to change

The Treasury agreement is a contract, and it shuts every door the common needs.

The simple test is to multiply the steps that a good result needs. I put the odds of Treasury agreeing to end the conservatorship at 55%, of the lawsuits no longer blocking new stock at 80%, of Treasury accepting a smaller claim at 65%, of the capital gap being workable at 80%, and of the market then paying about 10 times earnings at 90%. If these were independent the chance of all five is 21% E (estimate). They are not independent, since one negotiated package would settle several of them at once, so I use 35% for the outcomes C, D and E together. I would be uncomfortable with a number much above that, and I would not be surprised by one far below it.

How the outcomes are built

Net income is US$15.9B in 2026, the nine months reported plus US$4.0B I assume for the fourth quarter, and grows 3% a year.³ In a release the whole company is worth 10 times that year’s net income, which is an earnings yield of 10%, my cost of equity, with no credit for growth because the capital rule would absorb it E (estimate). A release is taken at mid-2028 for B, C and D, and at the start of 2029 for E and F. In B, C and D Treasury exercises the warrant and ends up with 79.9% of the common; the junior preferred is paid its US$19.13B stated value first; the old common holders keep 20.1%. The share count is the 1.184B existing shares, so a 20.1% stake is divided by 1.184B. Case C is Treasury keeping a preference of what it paid in, US$120.8B, which gives 10 times US$16.8B of 2028 earnings, or US$168B, less US$120.8B, less US$19.13B, leaves US$28.5B for the common; 20.1% of that, divided by 1.184B shares, is $4.84. Case D, with no preference left, gives US$149B for the common and $25.35. I discount each at 11%, which is the 4.94% ten-year Treasury yield plus 6.1 points for a stock that pays nothing, has no vote and depends on political choices.¹⁵

The base is the probability-weighted present value, $5.15. For this stock the usual idea of a central case does not work, because no single outcome is typical. The most likely single outcome, A, is worth $0.68 today, and the most valuable that I think plausible, D, $21.32. I call the weighted value the base. My bear is the weighted value of the three outcomes in which the common gets about nothing, A, B and F, $0.34. My bull is outcome D. The weighted value gets 50% of its weight from D and 30% from E, which has a 3% chance. Two-thirds of the odds are worth about a dollar or less.

In plain words: at $5.11, the price needs about a 35% chance that the old holders keep something worth at least about $4.84 a share. That is almost exactly the 35% I gave, and that closeness is not evidence that I am right. It is a sign that my guess and the market’s guess are in the same place. If Treasury ends up keeping a preference of about US$114B with the warrant, the shares are worth the price; if it keeps US$149B or more, they are worth nothing in that structure. The market is not pricing the business. It is pricing the amount of its own claim that Treasury gives up.

What the price implies

The public shares at $5.11 are worth US$5.9B. If the warrant is exercised and the full 5.89B share count is used, the common is worth US$30B. Adding Treasury’s preference at its face value of US$238.0B and the junior preferred at face value of US$19.13B, the market is giving the whole company a value of about US$287B, which is 18.6 times the twelve-month net income of US$15.4B and 2.4 times the net worth of US$119.6B C (calculated). A company like this earning 11.1% on its required capital might be worth 10 times earnings in a free market. So the price already assumes that Treasury will not collect its full preference. Put the other way, if the whole company is worth 10 times earnings, Treasury’s preference alone, US$238.0B today, is bigger than the whole company, US$154B on trailing earnings, and the old holders are in the money only if the preference is reduced.

The margin of safety, and the price against the range

My margin-of-safety price is $3.61. It is 70% of my base, the same rule as on every page, and it is 42% under the price. It is a weaker protection here than on an ordinary company. The cushion protects against being wrong by 30% about a value. It does not protect against the 65% chance that the value is near zero. At $3.61 the shares would still lose about 80% to 100% of their value in outcomes A, B and F. At $5.11 the price is 1% below my base, 24% of my bull, and 15 times my bear.

The price reflects a real debate, and the case for the market being right needs to be said. It says that the President has said he is considering a public offering, that the company earns US$15.4B a year while the estate that owns it needs a way out, that Treasury cannot offer shares at a price that wipes out the people who will buy them, and so the old holders will be paid in some form. If I give that full weight, outcome D is where the price lands. The case that the market is wrong says that Treasury holds a claim of US$238.0B that it has not offered to reduce, that the agreement says the conservator needs its consent, and that the structure has stayed in place for 18 years under different administrations and directors. I give the second side the larger share of the odds, 65% against 35%, and I would not be surprised if the first side were right.

The next best use of the same dollar

Freddie Mac is the same bet. It closed at $4.69 on 2 November with a preference of US$150.4B and 0.650B shares, and the same Treasury and the same warrant.¹³⁷ It is not a second opinion; it will pay off or not on the same decisions. The ten-year Treasury yields 4.94% and pays it.¹⁵ On this site the nearest financial companies show what an owner can take out: American Express gives an owner cash yield of 4.1% in my work and Bank of America 4.9%, both paid out of businesses that decide their own payouts. Fannie Mae gives 0% today. A good business at a price that depends on a political decision is not the same as a good business at a fair price, and I do not give either a grade.

VIIThe case for the price

What most people miss

Three facts cut in the stock’s favour and get little attention. First, Treasury has already been repaid in cash. It put in US$119.8B and received US$181.4B in dividends through 2025, 1.5 times what it paid in, and no dividend is payable now.⁶³ Second, the company has rebuilt its capital faster than most people think: net worth rose from a deficit of US$3.7B at the end of 2017 to US$119.6B, and the deficit in available adjusted capital halved from US$22B to US$11B in nine months of 2026.⁴³ Third, the statutory core capital, which excludes the senior preferred stock, went from a deficit of US$11.8B at the end of 2025 to about zero at 30 September 2026.³

The bull case

In plain words: the business is large, steady and profitable, its owner is the government, and the government has said it is looking at a way out. On 21 May 2025 President Trump said he was giving the idea of a public offering "very serious consideration" and would decide "in the near future", and said the two companies were "doing very well".¹⁴ The FHFA Director and Fannie Mae board chair, William Pulte, said in the third-quarter release on 28 October 2026 that the company "landed another strong performance this quarter" and grew net worth 13% year over year to US$120 billion.⁵ The company reports an illustrative return of 11.1% on the capital the rules would require, up from 10.8% in the second quarter.⁵ If Treasury accepted a smaller claim, the arithmetic is simple: the whole company is worth about US$168B at 10 times 2028 earnings, and the old common would own 20.1% of what is left after the preferred.

I looked for named investors or analysts whose published view on this company I could confirm from a primary source, and I did not find any I could verify this week, so I name none. The bull case here is the President’s statement, the FHFA Director’s statement and the company’s own metrics. The reports that give it more detail were behind pages I could not open, so I did not use them.

Against reality

If the release comes on the terms of C or better, the price is possible and more. At $5.11 the stock needs about a 35% chance of that, against the 35% I give.¹ The distance from the price to the bull is large: the bull, outcome D, is worth $21.32, 4.2 times the price, and it needs Treasury to convert a preference of US$238.0B (about US$274B by the end of 2028) into common while keeping its warrant. Treasury is under no obligation to do that, and it holds every key: the agreement needs its consent for an exit, and a change in the preference needs its signature E (estimate).⁶ The distance to the bear is also large: $0.34 is 93% under the price. The price is a bet that the middle two outcomes, C and D, are both live. Nothing in the business can move that bet. The company can earn US$16B a year in every case and the shares still be worth a dollar.

What a 10x would take. Ten times $5.11 is $51.10. On the 1.184B existing shares, if the warrant were cancelled, that is a market value of US$61B, 3.9 times the twelve-month net income of US$15.4B, which would be possible only in a structure like E or better C (calculated). If the warrant is exercised, so that the count is 5.89B shares, $51.10 is US$301B, 19.5 times the twelve-month net income, and then Treasury’s preference of US$238.0B and the junior preferred of US$19.13B stand ahead, which brings the value of the whole company to about US$558B, 36 times earnings. At 10 times earnings that would need net income of US$56B, 3.9 times the 2025 level. In five years, 10 times is 58% a year. This stock has done it before: from $0.35 on 30 December 2022 to $15.31 on 11 September 2025 is 44 times in 33 months, but it started from a price that assumed nothing.¹ For a company with US$15.4B of earnings and a claim of US$238.0B ahead of the shares, a tenfold rise from here with the warrant in place is close to impossible. Without the warrant it is a matter of a legal settlement, not of the business.

Where I come down

What moved me was the earnings and the capital: US$15.4B in twelve months, a net worth of US$119.6B, and a shrinking deficit in available capital. What did not move me is that none of it reaches the common stock under the current terms, and that the filing says the company is "significantly undercapitalized" with a US$205B shortfall.³⁶ The price sits where my weighted value sits, which tells me the market and I agree on the odds and not that either of us is right. What would raise my values is a signed change in the senior preferred terms, a Supreme Court decision that ends the litigation, and a capital rule that closes the gap without new stock. What would lower them is a fee cut ordered by FHFA, a rise in serious delinquencies above 1%, or any plan that treats the common as the last claim to be paid.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest, and a test of each.

• Policy. The price of every loan is set by FHFA and now by an executive order of 22 October 2026 that asks it to consider cutting fees on some refinancings.³ A cut of 2 basis points is about US$0.57B of net income a year, 3.6% of earnings C (calculated). The company says the change in FHFA leadership or the Administration could bring changes that affect it.³
• Receivership or merger. The 2025 annual report says proposals could "result in Fannie Mae’s liquidation or dissolution, or its merger or consolidation under common ownership with Freddie Mac". A merger would combine two companies with a preference of US$388.4B between them C (calculated).⁶⁷³
• Credit. Multifamily provisions were US$446M in the third quarter on weaker values and higher rates, and the 30-year rate was 7.03%.⁵³ Single-family serious delinquency is 0.62% and rising slowly.
• Governance and people. The company has had an acting chief executive since October 2025, directors who own no stock, and a board chaired by its regulator. It reports a material weakness in disclosure controls that it does not expect to fix in conservatorship.³
• Liquidity of the shares. The common trades on the over-the-counter market, where the daily volume was 6.9M shares on 2 November and 17.5M on 12 October, and where moves of 10% in a day have been common recently.² A holder may not get the displayed price in size.

What a bad year looks like in the model

The surprise is how little a credit cycle costs and how much a delay or a bigger Treasury claim costs. A 20% fall in earnings takes $1.74 off the value; a two-year delay $0.50; but halving the odds of a good outcome takes $2.41, about half the value. The risk is not a bad year of earnings. It is a bad decision.