Freddie Mac

IThe business

Freddie Mac is a company that Congress created in 1970 to make sure that money keeps flowing into home loans.³ It does not lend to homeowners. A bank or other lender makes a mortgage, sells it to Freddie Mac, and Freddie Mac packages many of them into a security, promises investors that the principal and interest will be paid even if a borrower stops paying, and sells that security into the world’s bond markets. For that promise it charges a guarantee fee, which on the loans it bought this year averages 53 basis points a year on a new single-family loan, about half of one percent.³ It also passes on some of the risk of loss to private investors through credit risk transfer deals, and it holds a smaller portfolio of mortgages and securities for itself.³ That is the whole machine. In the third quarter of 2026 it provided $114B of mortgage funding, which the company says financed 420,000 home purchases, refinancings and rental units.³

Its portfolio is $3.70 trillion of mortgages, of which $3.19 trillion are single-family and $510B multifamily, meaning apartment buildings.³ The single-family part is about a fifth of the $15.0 trillion of single-family mortgage debt that the company cites for the end of the second quarter of 2026, and another large share belongs to its twin, Fannie Mae C (calculated).³ Freddie Mac has 7,334 full-time employees and a head office in McLean, Virginia.⁵

Single-family: the big business

Nearly all the money is made guaranteeing single-family loans. In the first nine months of 2026 the Single-Family segment earned $9.0B of net income, up 34%, from $16.1B of net revenues, and Multifamily earned $1.7B.³ The segment bought $311B of new loans in the nine months, 15% more than a year earlier, helped by refinancing, which rose from $52B to $97B.³ Its mortgage rates tell the story of the year. The 30-year rate was 6.15% at the end of 2025, 6.49% at the end of June and 7.03% at the end of September 2026, which reduces new loans and refinancing but also locks in the old low-rate loans that Freddie Mac already guarantees.³ I read the earnings this way: most of the profit comes from the fee on the book that is already there, not from new loans in any one year, which is why earnings were steady through the swing in rates.

Two rules are worth knowing. First, Freddie Mac does not keep all the credit risk. Of the single-family portfolio, 60% by balance is covered by some credit enhancement: mortgage insurance on 22%, and credit risk transfer on about half, with a maximum coverage of $230B across the book.³ In the nine months, it sold risk on $141B of new loans.³ Second, a part of every guarantee fee is not Freddie Mac’s at all. A law of 2011, extended in 2021, added 10 basis points to the fee, and the company remits that to Treasury: $3.0B in 2025 and $2.28B in the first nine months of 2026, which are in its expenses.⁵³ At the same time it paid $1.37B for credit enhancement.³ Both come before profit.

Multifamily: the second engine

Multifamily finances apartment buildings, and its model changed this year. For years the company bought the loans and sold most of the risk in so-called senior-subordinate deals. In 2026 it has moved to fully guaranteed securities, on which it keeps all the credit risk and earns a higher fee: $60.8B of the $62.8B securitised in the nine months, against $26.7B of $44.9B a year before.³ The average fee on the multifamily guarantee rose to 60 basis points from 54.³ The book grew 6% to $510B, 66% of the new lending was classed as mission-driven affordable housing against the 50% the regulator asks, and the 2026 cap on purchases was set at $88B, up from $73B.³⁵ Credit is the thing to watch: the delinquency rate on the multifamily book rose to 0.58% from 0.44% at the end of 2025, with higher interest rates and a 1.6% fall in property prices in the quarter.³ Multifamily was 14% of 2025 net income.⁵ I do not think the shift to holding more of the risk has been tested by a bad cycle yet.

How it got here

The company has been in conservatorship since September 2008, which makes the current arrangement 18 years old. The terms have been rewritten five times, and each rewrite changed what holders could hope for. The 2012 rewrite swept every dollar of profit to Treasury. The 2019 and 2021 rewrites let the company keep its profit and build capital, at the price of adding the same amount to Treasury’s claim. The 2025 rewrite made Treasury’s consent the only condition for an exit.⁵⁸

The state of play now

The third quarter of 2026 was a good one for earnings and a hard one for the stock. Net income was $3.2B, up 17%, but part of the rise came from $0.6B of investment gains and the quarter included a $0.8B provision for credit losses, against $0.2B a year before.³ Over the nine months, the company released credit reserves of $0.4B where it had built $1.2B a year earlier, which flatters the comparison.³ Net worth rose to $81.0B.³ Nothing in these numbers reaches the shareholders: the company’s own report says its maximum payout ratio under the capital rules is 0.0%, and that every quarterly increase in net worth is added to Treasury’s liquidation preference.³ The shares closed at $4.69 on 2 November 2026, down from $10.14 at the end of 2025, after a September in which they fell from $5.57 to $3.91.¹² I have not been able to verify what caused the September fall from a primary source, and I do not claim to know it.

IIThe moat

As a business, Freddie Mac has one of the strongest positions I have looked at. As an investment, the question is who owns the strength. These are two different questions, and I answer them separately.

What protects the business

• A charter and a backstop. Freddie Mac and Fannie Mae are the only two companies that Congress chartered to do this, and investors buy their securities because of the belief in US government support. The company’s own filing says its credit ratings follow the US government’s, AA+ from S&P and Fitch and Aa1 from Moody’s for senior debt, and that its preferred stock is rated D, C and Ca.⁵ A rival would need a charter, not just capital.
• Scale in a fungible market. The single-family securities of the two companies are blended into one product, the Uniform Mortgage-Backed Security, one security that either company can issue, so a buyer of the bond does not need to know which one guaranteed it.⁵⁷ The deepest market in mortgage bonds depends on that.
• Credit data and experience. The serious delinquency rate on its single-family book is 0.61%, and it has decades of data on which loans fail.³ The company’s credit enhancement and loan-level pricing are possible only with that history.
• Loans already on the book. About three in five of the single-family loans carry credit enhancement the company has already bought, and every loan that is already guaranteed keeps paying the fee until it is repaid.³

Where the strength goes

The rights that give the business its strength are also rights the company cannot use for its owners. Freddie Mac is, in its own words, “not managed to maximize stockholder returns,” and the conservator, the Federal Housing Finance Agency, decides the strategy.⁵ The Purchase Agreement with Treasury says that the company cannot pay dividends, buy back its shares, issue new equity, dispose of assets outside the ordinary course or leave conservatorship without Treasury’s consent.⁵ The Federal Housing Finance Agency can change what the company does at any time. This year Freddie Mac began accepting a second credit score, VantageScore 4.0, from all sellers on 9 September 2026, moved to a single price grid for both scores in October, and published its own Freddie Score; the company says this could lower the guarantee fees it receives.³ A business with this kind of customer, product and regulator is not setting its own price.

The test of any edge is what it earns through bad times. Freddie Mac’s test was 2008, when it lost so much that it needed $71.6B from Treasury to stay solvent, most of it from 2008 to 2011.³⁵ In the ten years from 2016 to 2025 the machine earned between $7.2B and $12.1B a year, except in 2017, when it earned $5.6B after a one-off tax charge, and net worth fell below zero only once, at the end of 2017.⁴ The business is good. What the common holder has lost is not the business, but the claim on it.

IIIThe competitive advantage

The competitor that matters is the one that wears the same chain. Fannie Mae is also in conservatorship since September 2008, also owned in effect by Treasury, and also run by the same regulator, whose director, William Pulte, is the chairman of both boards.⁵ Anything that happens to one is likely to happen to the other, and the market treats them as a pair. The business difference is size, and the stock difference is almost entirely about the number of shares.

Fannie Mae’s book is $4.30 trillion against Freddie Mac’s $3.70 trillion, which is 16% larger, and it earned $14.4B in 2025 against $10.7B.⁷¹⁰⁴ Freddie Mac has a slightly larger weight in multifamily, 13.8% of the book against 12.8%, and in the nine months of 2026 Multifamily made 15.7% of its net income C (calculated).³⁷ Fannie Mae has a bigger hole: a shortfall of $205B against the capital rules, with $119.8B drawn from Treasury in the crisis against $71.6B for Freddie Mac.⁷³ And Fannie Mae’s Treasury claim is $238.0B, 58% bigger than Freddie Mac’s $150.4B C (calculated).⁷³

The difference that matters most to a shareholder is not in the business at all. Freddie Mac has 650M common shares and Fannie Mae 1,158M. If Treasury were to use its warrant, Freddie Mac would have 3.23 billion shares and Fannie Mae 5.76 billion C (calculated). At the 2 November closes, Freddie Mac’s market value is $15.2B, which is 1.4 times its 2025 net income, and Fannie Mae’s is about $29.4B, or 2.0 times C (calculated). In plain terms, the market pays less per dollar of earnings and per dollar of net worth for Freddie Mac, and it also has a smaller shortfall in comparison to its size. I do not think that makes it a better business. It says that the price of the stock is mostly a comment on how likely the shares are to receive something.

Where either could lose ground

The traditional threats are Ginnie Mae and the Federal Housing Administration, which guarantee government-backed loans, and banks that keep loans on their own balance sheets. The bigger changes of the past year have come from inside the system: the new credit score rules, an executive order of 22 October 2026 that allows the regulator to direct a pilot for cancelling mortgage insurance and to expand a programme under which the companies may accept a title-risk review in place of a title insurance policy, and the regulator’s control of how the two companies price risk.³ I cannot say how these change profit, and neither can the company, which says it is evaluating them.³ What they show is that a competitor to Freddie Mac in the next five years is more likely to be its own regulator’s policy than another firm.

IVManagement

Who controls it

The managers of Freddie Mac do not control it. The Federal Housing Finance Agency has been the conservator since September 2008 and holds the powers of the company’s shareholders, officers and directors.⁵ The board of directors is chosen by the conservator, and the company’s own report states that while in conservatorship the directors “have no fiduciary duties to the company or its stockholders.”⁵ The chair is William Pulte, who has been the Director of the Federal Housing Finance Agency since March 2025 and is also chairman of Fannie Mae’s board.⁵ Clinton Jones, the agency’s general counsel, is also on the board.⁵ Five directors, Farbman, Jones, Kittle, Parrott and Pulte, joined in March 2025, and on 17 March 2025 the conservator removed one director from the board.⁵

The chief executive is Kenny Smith, a retired Deloitte consulting partner who has led financial-services work for more than 40 years, and who took office on 17 December 2025.⁵ He was the fourth chief executive appointment since March 2024. Michael Hutchins, the president, served as interim chief executive from March to September 2024 and again from March to December 2025, and Diana Reid held the job from September 2024 to March 2025.⁵ The turnover tells me that the job does not carry the tools of a chief executive, since the strategy comes from the conservator and is written into annual scorecards.

How they are paid, and what that pays them to do

By law the chief executive can be paid a base salary of $600,000 and nothing else: the Equity in Government Compensation Act of 2015 freezes it at the level of 1 January 2015, and the STOCK Act bars bonuses for senior executives in conservatorship.⁵ The other executives are paid through a plan in which most of the money is a deferred salary; for the president the 2025 target was $5.1M in total, of which $1.5M was at risk.⁵ No part of it is paid in stock. The company has no equity plan with shares available.⁵ So the incentive is to meet the conservator’s scorecard and to keep the machine running safely, not to raise the stock price. The company says the cap on pay has made retention and succession planning difficult.⁵

The ownership says the same thing. The 19 directors and executive officers together own 3,093 shares, and the report notes that those shares were owned before the conservatorship.⁵ The only holder over 5% is Treasury, with the warrant for 79.9%, which it has not exercised and may do until 7 September 2028.⁵ The last time Pershing Square, run by William Ackman, reported its holding on a Schedule 13D was 31 March 2014, when it reported 9.78%, and the company’s report says nothing newer.⁵ I do not state anyone’s current holding.

What they did with the money

Over a full cycle the managers have done what the conservator allowed. Freddie Mac took $71.6B from Treasury and has paid it $119.7B in dividends, which is $48.0B more than it took C (calculated).³ From 2019 it was allowed to keep its profit, and net worth went from negative $0.3B at the end of 2017 to $81.0B at 30 September 2026.⁴³ It paid nothing to common holders and bought back nothing. Over the same ten years, net revenues rose 56%, from $14.9B to $23.3B, but non-interest expense rose 113%, from $4.0B to $8.6B C (calculated).⁴ Much of that is not the managers’ choice: $3.0B a year is the legislated guarantee fee that must be remitted to Treasury, and a further $1.4B in nine months is credit enhancement purchased from outside investors.⁵³ Salaries and benefits were $1.17B in the nine months of 2026, down 10% on lower headcount, which is the one cost line the managers control.³

The conclusion is not about their skill. A manager in this company is paid to be careful and loyal to the conservator, and the evidence is that they have been both. What I cannot judge is how they would behave with a free hand and an equity incentive, because there is no history of it since 2008.

VFinancials

Freddie Mac reports in US dollars and the figures here are in US$ billions. The company’s year is the calendar year, and the latest quarter is the third quarter of 2026, filed on 29 October 2026.³ Net income is what the company earns after tax, and net worth is its total equity on the balance sheet.

Revenues have grown, earnings have not kept pace, and net worth is mostly new. Net revenues, which are net interest income plus non-interest income, rose from $14.9B in 2016 to $23.3B in 2025, a 5.1% yearly rate, and net income rose from $7.8B to $10.7B, 3.6% a year C (calculated).⁴ The 2025 net interest income of $21.4B was the highest on record in the table, up 8% on 2024, but non-interest income fell from $4.2B to $1.9B, mostly from changes in the value of derivatives and securities.⁴ Net income can swing on market gains and on tax: in 2017 income tax expense was $11.2B, against $3.8B a year earlier.⁴

What an owner can really take out

For every owner of an ordinary company I ask what is left after the business has paid for staying where it is. For Freddie Mac the answer today is nothing. A common shareholder receives no dividend, because the Purchase Agreement and the conservator both bar one, and the net income all goes into Treasury’s claim.⁵ This is why the company’s earnings per common share, after the amount attributed to Treasury, are about zero in every year in the table: $0.15 in 2021, $0.11 in 2022, negative $0.05 in 2023, nil in 2024 and negative $0.02 in 2025.⁴ The reported earnings of $10.7B are therefore not owner cash. In the eventual case where the company is released, an owner would take out only what is left after growing capital with the book: my model has $7.8B of distributable cash in the first year after a recapitalisation against $13.1B of reported earnings, or about 59% E (estimate). The difference is the capital that must be added each year to keep it at 12.7% of risk-weighted assets, because the book grows about 3% a year. Even then the owner’s share of that cash is as set by the political settlement, which I discuss in Valuation.

The balance sheet

The balance sheet shows $3.53 trillion of assets and a net worth of $81.0B, 2.3% of assets C (calculated).³ Nearly all of it is the mortgages in the trusts that the company guarantees, which the accounts consolidate, so loans and trust debt are each over $3.2 trillion.³ Freddie Mac’s own borrowing is only $166.5B, down from $207.0B at the end of 2025.³ Liquidity is not the problem; it holds $132B in short-term and other investments, and Treasury’s remaining commitment is $140.2B.³ The question is capital. The rules ask for $162B of adjusted total capital including buffers, the company has $4B, and the gap is $158B.³ The $14.1B of junior preferred stock counts as Tier 1 capital; the $72.6B of senior preferred stock does not.³ If the senior preferred were converted into common equity, the gap would fall to about $85B before any further profit C (calculated). That is the number the whole valuation turns on.

Share structure

There are three kinds of stock, and they are not equal. The senior preferred stock belongs to Treasury and is first in line. The junior preferred, 24 series with a $14.1B par value, ranks behind it and ahead of the common, and the conservator has barred its dividends.⁵ The common stock is last: 650,059,553 shares are outstanding, and if Treasury uses its warrant, it would hold 79.9% of a company with 3.234 billion shares, leaving today’s holders with 20.1% C (calculated). Fannie Mae has 1,158M common shares, which is why a Freddie Mac share is a larger slice of a smaller company.⁷

Accounting to watch

• Reserves. The credit loss allowance is $8.0B on a $3.4 trillion book. In the first nine months of 2026 the company released $0.4B where it had added $1.2B in 2025, which it ascribes to a change in how it generates house-price scenarios, and the allowance is sensitive to house prices: a 3-point lower forecast would add $1.2B.³
• Market gains. Non-interest income swings with interest rates and hedges; it was $0.9B in the third quarter and $1.4B in nine months.³
• Net worth is not capital. The ratios on the first page use net worth, but the rules count only $4B of it.³
• Deferred tax assets. Net deferred tax assets are $4.0B and part of them is deducted from regulatory capital.³

VIValuation

First, how this goes badly wrong

Before any numbers I ask what would most likely hurt an owner of the common stock, and work backward. I see three ways.

• The government keeps its claim in place. Treasury’s liquidation preference is $150.4B, rises to $153.6B at the end of this year and grows by about the amount of net income each year, which is $10B to $13B.³ Under the Purchase Agreement, the company cannot repay it for the foreseeable future, and a release that leaves it as it stands leaves the common with nothing but an option. That is my S2. The Treasury’s consent is needed for any exit other than a receivership since the January 2025 amendment, so the holder of the claim sets the price of the release.⁸⁵
• The fix needs a lot of new capital at a poor return. Even in a friendly release, the company has to raise about $61B to meet the ERCF, and the return on that capital is not good: net income of $13.1B on required capital of $178B is a 7.4% return, less than the 10.0% I require. A business that earns less than its cost of capital is worth less than the capital in it, and the more the rules ask for, the worse the answer: a 10% higher requirement cuts my base from $3.93 to $1.20.
• Nothing is left for the common. The company’s own 10-K says that after a change in its structure, the shares “could diminish in value, or cease to have any value.”⁵ That is my S6, at 15%.

What an owner can really take out

Reported earnings are $10.7B for 2025. What an owner can take out today is nothing, since the payout ratio under the capital rules is 0.0% and every dollar of net worth growth is added to Treasury’s claim.³ After a recapitalisation, an owner can take out only the earnings less the capital that must be added as the book grows. That is $7.8B in the first year, 59% of the $13.1B of reported earnings. The gap between those two numbers is why a price-to-earnings ratio is the wrong measure of this stock. The market value if the warrant is exercised is $15.2B, 1.4 times 2025 net income, which sounds cheap, but nothing is paid to the common and 79.9% of any future payment is Treasury’s C (calculated).

What I can’t judge

I cannot judge whether the government will let go of its claim, how much it will want in return, or when. Nobody can, and the people who claim to are guessing. The company’s own report says the Administration has made comments about a public offering, in or out of conservatorship, and about exit, and that it cannot predict whether or when any of this will happen or on what terms.⁵³ What I can verify is narrow. Freddie Mac’s last filing of a new agreement with Treasury, an Item 1.01 report, was dated 8 January 2025, and the Federal Housing Finance Agency’s ten latest news releases, from 31 March to 27 October 2026, do not announce a change in the conservatorship or an offering.⁹¹² The 22 October 2026 executive order concerns mortgage insurance and title review, not the ownership of the company.³ So as of 3 November 2026 I can say that there is talk, and no signed plan. My probabilities are therefore a statement of belief and I show how they move the answer instead of pretending they are known.

Two other inputs I cannot know: how much Treasury will accept, and what the junior preferred holders will get. For the first I use the warrant as the limit. For the second I use par in S3 and half of par in S4, and the difference is $1.61 a share.

How the cases are built

The common stock is the last claim on the company, so I value it by working down from the whole. Step one is a fully capitalised company at the release date. I assume the release on 31 December 2029, which gives time for the court case, the warrant, and the capital raise; the warrant expires on 7 September 2028 unless it is exercised or extended, so that part would happen sooner.⁵ Step two takes away the new capital that has to be raised, and step three the junior preferred. Step four splits what remains by outcome.

Earnings. I use $12.0B of normal net income in 2027, growing 3% a year. The first nine months of 2026 were $10.6B, the third quarter alone was $3.2B, or $12.9B a year, and 2025 was $10.7B.³⁴ I start below the third-quarter rate because 2026 includes a $0.4B credit reserve release and investment gains, and because the company says the new credit-score pricing could lower the guarantee fees it receives.³ It is my judgement, and I show the answer from $10B to $14B.

Capital. The rules ask for 12.7% of risk-weighted assets, including buffers, which was $162B on $1,272B at 30 September 2026.³ I let both grow 3% a year. I keep the company at that level after the release, so each year it keeps 3% of the requirement for growth and pays out the rest.

Discount rate. The ten-year Treasury yield was 4.91% on 2 November 2026.⁶ I add 5.1 points, which is about 4.5 points for the general risk of owning equity and about 0.6 for the policy risk that this particular company carries, both my judgement. That gives 10.0%. The same rate is used for each outcome, though S5 and S6 plainly carry more risk than S1, and I do not pretend otherwise.

The step that surprises me most is how little of the whole is left for today’s holders. The fully capitalised company is worth $111B. Of this, $61B must be put in by new investors, $14B belongs to the junior preferred, and the remaining $35B is split 79.9% to Treasury and 20.1% to the people who own the 650M shares today, which is $7.1B, or $10.92 a share at release. That is the whole of S3. Everything else in the table is a variation.

My three cases

I need a bear, a base and a bull, but the outcomes do not fall on one line, so I define them this way. Bear is the weighted value of the three outcomes in which holders get little or nothing, S1, S2 and S6, which together carry 72% of my odds: $0.83. Bull is the weighted value of the three in which a release gives holders a stake, S3, S4 and S5, which carry 28%: $11.92. Base is the weighted value of all six, $3.93. The bull is therefore not a happy ending but the average of three, and 36% of it comes from S5, which I give a 3% chance. I do not think any one of these is the price the stock ought to have. The range is the answer.

What the price needs

At $4.69 the market value of the 650M shares is $3.05B. If the warrant is exercised and the other 2.58 billion shares are counted, the whole is worth $15.2B. Run my model backwards, and the price needs one of four things, holding the rest. The odds of a release that gives holders a stake would have to be 35% instead of 28%. Or the discount rate would have to be 9.6% instead of 10.0%. Or normal net income would have to be $12.4B instead of $12.0B, which is about the third-quarter run rate of $12.9B. Or the capital requirement would have to be 12.4% of risk-weighted assets instead of 12.7% C (calculated). No single one of those is extreme, and that is the point: the price is within the error of my own inputs, and the answer is a matter of how far a reader trusts each of them.

I multiply the odds out to see how hard the good outcomes are. I give a release of some kind by the end of 2030 a 40% chance: S2, S3, S4 and S5 together. Given a release, I think there is a 70% chance that Treasury converts or cancels its claim, since a release that leaves $150B-plus ahead of the common is not a recapitalisation. That is 40% times 70%, or 28%, for S3, S4 and S5. For S3 alone the steps are a release, a conversion and a capital raise that completes on the terms I assumed, and each step has to come off. The litigation that the 2021 agreement ties to the $70B of new equity is nearer an end than it was: the appeals court affirmed the $313M judgment on 24 July 2026 and denied rehearing on 16 October 2026, which as I read it leaves only a petition to the Supreme Court.³ I count that as a step done, not as a reason for the price.

The two tables say the same thing in different ways. The base value is most sensitive to the discount rate and to the amount of capital the rules ask for, because both multiply a small residual: the common gets a sliver of a large number minus another large number. A one-point change in the discount rate moves the base by 40% to 56%. An interesting oddity is that lower growth after the release raises the value, since a business that earns less than its cost of capital destroys value as it grows.

The margin of safety, and the price against the range

My margin-of-safety price is $2.75, which is 70% of the base. I use the same 30% cushion as on the other pages, and for this page I think it is thin, not generous, because the inputs are less knowable here than anywhere else on the site and a change of one point in the discount rate takes more than that off the base. The price of $4.69 is 19% above my base value of $3.93 and below my bull of $11.92. It is above the base.

Why the market may be right: I could be wrong on the odds, since the Administration has talked about an offering and the stock has already shown what a hint does, with a rise of about 33-fold from December 2022 to September 2025.¹ The shares are also few, 650M, so a modest dollar sum is needed to change the price, and there is a view among investors that a release would happen sooner than 2029. The price is 54% below the $10.14 of the end of 2025, and part of that is a lower chance of a quick release, which is also consistent with my numbers. I do not know what the market knows; I know what the filings say, and what they say is that nothing has been signed.

The next best use of the same dollar

The simplest alternative is a ten-year Treasury note at 4.91%, certain, with payments in cash.⁶ A dollar in Freddie Mac’s stock pays nothing for years. If S3 is the outcome the return is about 31% a year to the end of 2029, and if it is S1, S2 or S6 the dollar is mostly lost. For a comparison on this site, I look at American Express, where I value the stock at $324 against a price of $323.71 and the earnings yield is 5.3% against a ten-year yield of 4.98% when I wrote it. The difference is not the yield. Amex pays a dividend and owners can see the cash. Freddie Mac’s edge in earnings is real, but none of it is reachable by owners until someone else decides.

Quality and price are separate questions

The business is of high quality. It earned between $5.6B and $12.1B in each of the last ten years, serious delinquencies on single-family loans are 0.61% and net charge-offs 0.01% of loans.⁴³ The common stock is a different question, and today looks like a high-quality business at a price that depends on a decision of someone else. I would not call the price a bargain, and I would not call it a trap. It is a price for an option, and I value the option below what is being paid.

VIIThe case for the price

What most people miss

Three facts cut in the holders’ favour and get little attention. First, Treasury has already been paid more than it lent. Freddie Mac drew $71.6B and has paid Treasury $119.7B in cash dividends, $48.0B more, and the undrawn commitment is still $140.2B.³ Whatever one thinks of the bargain, a claim of $150.4B on top of that is a choice about how much more to take, not a debt that has to be repaid. Second, the 2025 amendment changed what stands in the way. Before 2 January 2025, exit without Treasury’s consent required a resolved lawsuit and enough capital; since then the Purchase Agreement requires Treasury’s consent for any exit other than a receivership, so the question is a decision, not a test.⁸ Third, the lawsuits that the 2021 agreement made a condition for new common stock are nearer to their end: the judgment of $313M was affirmed on 24 July 2026 and rehearing was denied on 16 October.³ The common holders’ share of that judgment is $31M, about 5 cents a share C (calculated).⁵

The bull case

In plain words: Freddie Mac earns more than $10B a year, its net worth is $81B and rising by about $3B a quarter, the credit is clean, and the only things between the stock and a normal price are two decisions that the government has not made. The first is whether Treasury lets its claim be converted into common stock, and keeps the 79.9% stake through the warrant instead; the second is whether the company raises the roughly $61B of new capital it needs. If both go the way holders hope, my model has the stock worth $10.92 at the end of 2029, 31% a year from today’s price, and more if the junior preferred take a haircut. The company’s own reports say the Administration has made comments about a public offering of its shares and an exit from conservatorship.⁵³ The stock is also small: 650M shares means that a few billion dollars of value is a large percentage change.

I looked for named investors and analysts whose published views on Freddie Mac I could verify from a primary source, and found none that I can attribute with a date and a document. William Ackman’s Pershing Square reported a 9.78% stake in a Schedule 13D dated 31 March 2014, which the company’s report still lists as its latest, and that is a filing, not a current view.⁵ Press reports of price targets and of statements by the Treasury Secretary exist; I have not opened a primary source for them, so I leave them out.

Against reality

If Treasury converts its claim, the company is released, and the capital is raised on the terms of my base, the price is possible: the common would be worth $10.92 a share at the end of 2029 against $4.69 today. Today’s numbers are a long way from that. Nothing is signed, and the report of 29 October 2026 says the company cannot predict whether or when any transaction will happen.³ The capital counted is $4B against $162B. Treasury’s claim is $150.4B and will be $153.6B in two months.³ The conservatorship has run for 18 years and, in the words of the 10-Q, has “no specified termination date.”³ I count the odds of a release of any kind by the end of 2030 at 40%, and of one that converts Treasury’s claim at 28%, and at 28% my weighted value is $3.93. The price needs about 35%.

What a 10x would take. Ten times $4.69 is $46.90 a share, a market value of $30.5B on today’s shares and $152B if the warrant is exercised. In my S5, where Treasury has no stake, the share is worth $54 at the end of 2029, so a tenfold move needs that outcome, which I give a 3% chance. In any outcome where Treasury holds 79.9%, a tenfold move needs a market value of $152B, which is 1.4 times what I think the fully capitalised company is worth ($111B) and 14 times its 2025 net income C (calculated). For most large companies a tenfold rise in a few years is close to impossible; for a company with $3.7 trillion of mortgages it would be unthinkable if the shares were an ordinary claim. They are not. This stock rose from $0.35 in December 2022 to $3.27 in December 2024, 9.3 times, and to $10.14 by December 2025, 29 times C (calculated).¹ It did so because of a change in what people thought the government would do, and it fell back by more than half this year for the same reason. The 10x is a bet on a policy outcome, with a small probability, and I do not count it in my base.

Where I come down

What moved me was the earnings power, $10.7B in 2025 and $10.6B in nine months of 2026, and the fact that a conversion of Treasury’s claim would cut the capital gap from $158B to about $85B. What did not move me was the price: at $4.69 the market is paying for more than my 28% chance of the good outcomes, and it is within the error of my discount rate. What would raise my values is a signed agreement that converts the senior preferred, an exercise of the warrant, and a capital plan with a raise near the $61B I assume. Each of those would move my odds by more than any change in the earnings.

VIIIRisks, and what would change my view

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

• Housing and interest rates. The 30-year mortgage rate was 7.03% at the end of September 2026, up from 6.15% at the end of 2025, and the forecast of house price growth for the next 12 months is 1.3%.³ A 3-point lower forecast would add $1.2B to the credit loss allowance.³ The company holds $8.0B of reserves against $18.7B of non-accrual loans.³
• Multifamily credit. Delinquency rose to 0.58% from 0.44% in nine months, with higher rates, and property prices fell 1.6% in the third quarter. Because the company changed to fully guaranteed deals, it keeps more of the risk than before.³
• The regulator’s hand. The Federal Housing Finance Agency has changed credit-score rules and pricing in 2026, and can change fees, volumes and capital.³ It sets the multifamily cap, which is $88B for 2026.⁵
• The sovereign. The company’s ratings follow the United States’; Moody’s cut the sovereign to Aa1 on 16 May 2025 and the company three days later.⁵
• Legal. The securities class action from 2008 was revived in part by the Sixth Circuit on 21 August 2026 and sent back to the district court.³
• Dilution and capital structure. The warrant for 79.9% is nominally priced at $0.00001 a share, so any exercise cuts today’s holders to 20.1% of the company.⁵ New capital would cut it again.

What a bad year looks like in the model

The surprise is that earnings matter less than politics. A drop in normal earnings from $12.0B to $9.0B, a housing downturn, costs $3.90 of the $3.93 base, while a shift of ten points of odds from a good outcome to a bad one costs about as much. The risk is not a bad year. It is a bad decade in which the claim keeps growing and nothing is decided.