The Federal Housing Finance Agency released the results of the annual stress tests for Fannie Mae and Freddie Mac on Monday, and the report is a nice little milestone. For the first time, the crisis-era government bailouts of the government-sponsored entities,combined with their hypothetical future bailouts, are profitable for the government. Here’s the accounting:
Now I am cheating here a bit. Strictly speaking, the stress tests found that “Fannie Mae and Freddie Mac could need as much as $125.8 billion in bailout money from taxpayers in a severe economic downturn.” That number is bigger than my $49.2 billion. The difference is a $76.7 billion “impact of re-establishing valuation allowance on deferred tax assets.” Basically, Fannie and Freddie lost a lot of money in the past, which reduces their income taxes in the future, unless they don’t have income in the future, in which case it doesn’t. And they count the future tax reductions as an (enormous) asset on their balance sheets, unless they don’t expect to have income in the future, in which case the asset goes away, reducing their equity and requiring — perhaps — another bailout. The FHFA ran the stress tests two ways, both assuming that the deferred tax asset went away and required more bailout funds, and not. The Fannie/Freddie deferred tax assets got quite a workout in the last bailout, and I can only assume that any future bailout would involve lots of contortions around them, but really who cares. The disappearance and reappearance of the deferred tax assets are not a real cash expense, and if everyone just agreed to ignore them, nothing in the world would be any different. So let’s just ignore them.
The FHFA might be cheating here a bit, too, in the boring sense that any time a government agency releases stress test results, you can complain that the test wasn’t stressful enough. The 2016 stress test scenario “is based upon a severe global recession which is accompanied by a period of elevated corporate financial stress and negative yields on short-term U.S. Treasury securities,” which sounds bad-ish, but which seems to leave Fannie and Freddie curiously untouched. The scenario results in credit losses equal to 0.56 percent of their combined portfolio, meaning that for every $100 of loans that they own, Fannie and Freddie would get back $99.44 even in a “severely adverse” stress scenario. That seems … maybe … a little generous? For “severely adverse”?
But I don’t make the stress test rules, and this stress test says another crisis would cost taxpayers just $49.2 billion in new Fannie and Freddie bailout money, less than the taxpayers have made in profits on the last bailout. And then, of course, after the new bailout, taxpayers would presumably go back to raking in money from Fannie and Freddie, and end up even further ahead. Until the next bailout, world without end.
There is a weird … idea? meme? accounting error? … that says all of this is somehow a risk to the taxpayers that can be solved by giving Fannie and Freddie more money now. The way Fannie and Freddie work now is that, in round numbers, they operate with zero capital: Any profits that they make each quarter go to the Treasury, and if they have a loss in a quarter, the Treasury makes up the difference. In recent years they’ve had profits and returned money to Treasury. (Thus the $59.2 billion of government profits.) But it’s always possible that in the future they will lose some money, and Treasury will have to come up with the extra money. For instance, if the stress test scenario occurred, Treasury would have to come up with an extra $49.2 billion. It would still be ahead, but less ahead.
That’s true: The government, in every practical sense, “owns” Fannie and Freddie, so it’s at risk if they lose money. The weird part is the idea that the government can avoid this risk by just letting Fannie and Freddie keep all the money that they make. This is called “building capital,” and the idea is basically that if, instead of paying that $59.2 billion to the Treasury, Fannie and Freddie had kept it, there would be much less risk that Treasury would have to come up with $49.2 billion to bail them out in the future.
Which, again, is true. It’s just that then Treasury wouldn’t have the $59.2 billion. Getting $59.2 billion and giving back $49.2 billion of it still leaves you with $10 billion. Just never getting the $59.2 billion in the first place is strictly worse.
But it is easy for me to say that, because I think of Fannie Mae and Freddie Mac as wholly owned subsidiaries of the U.S. Treasury. This makes the analysis easier. For instance, the stress tests aren’t about whether Fannie and Freddie “could need as much as $125.8 billion in bailout money from taxpayers,” or whatever; they’re just about how much that division of the government could lose in a recession. (In the same way that, when the U.S. Postal Service loses money, no one calls it a “bailout.” It’s just a loss.) It’s of no interest whether the division has a lot of capital, or a little capital, or no capital, or negative capital: The U.S. government has plenty of money, and that’s the relevant unit of account.
Not everyone agrees. I mean, Fannie and Freddie aren’t wholly owned subsidiaries of the government: They have common and preferred stock outstanding, and while the current terms of the bailouts prevent them from ever making payments on that stock, the shareholders have been suing for ages to try to reverse those terms and get their money back. They argue, with some justification, that the bailout terms are unfair and illegal, and that they should get back their rights as owners.
The government is fighting those lawsuits, but it too can’t quite bring itself to say that Fannie and Freddie are fully controlled agencies of the government. Partly this is an accounting issue — by not consolidating Fannie and Freddie, the government doesn’t have to count their debt toward the debt ceiling — but it is also sort of a political-philosophical issue. Nobody ever exactly made the choice to permanently nationalize the entire U.S. mortgage market. Fannie and Freddie are in “conservatorship,” a supposedly temporary state of government protection that happens to have lasted for the better part of a decade. There’s no mandate or plan to keep them nationalized forever, and everyone has to talk piously about the importance of returning private capital to the mortgage markets.
Everyone kind of knows the basic mechanics of how to do that: Create some privately owned mortgage companies, either from scratch or from the bones of Fannie and Freddie, to buy or guarantee mortgages. Raise capital for those companies from investors, perhaps giving existing Fannie/Freddie investors some credit for their existing holdings, perhaps not. Regulate those companies, and make sure they’re well enough capitalized to absorb any reasonable credit risk in the mortgage market. Let those companies buy some sort of backstop from the government, at a fair price, to further socialize the credit risk of mortgages.
And yet people have been talking about this for ages, and nothing much has happened. One reason for this is pretty obvious: For the government, keeping Fannie and Freddie as a branch of the government is pretty great. They’re easy to control, available as an instrument of housing policy, and, crucially, profitable. Perhaps things will change if Fannie and Freddie run into trouble again. But the latest stress tests suggest that even that may not change anything.
- The initial bailout and repayment figures come from ProPublica’s great bailout trackers for Fannie Mae and Freddie Mac. The stress test numbers come from the “Treasury draws required” lines in the “Results without re-establishing valuation allowance on deferred tax assets” column in the FHFA report, pages 6 and 7.Incidentally, the last time we talked about the Fannie/Freddie stress tests, two years ago, things looked much worse, with an $84.4 billion draw required even without accounting for deferred tax assets.
- It seems especially silly to get worked up about Fannie and Freddie’s potential tax liabilities when they are just divisions of the U.S. government anyway, but that’s the sort of thing that you’re not supposed to say, so I’m putting it in a footnote.
- It’s not really zero, though it will be by 2018. From the Fannie Mae 10-Q:
In addition, as a result of our agreements with Treasury and dividend directives from our conservator, we are not permitted to retain our net worth (other than a limited amount that will decrease to zero by 2018), rebuild our capital position or pay dividends or other distributions to stockholders other than Treasury.
Right now the “capital reserve amount” is $1.2 billion for each entity (see page 5 of the Fannie 10-Q and page 18 of the Freddie one), which is close enough to zero for our purposes.
- Of course even if Treasury had to come up with more — say, the $125.8 billion number that the stress tests also mention — getting the $59.2 billion first is still a good thing. Let’s say the government had left that $59.2 billion in the entities, and then they found themselves $125.8 billion short. The $59.2 billion would be wiped out, and the entities would still be $66.6 billion short, even after building all that capital. Then what? Well, maybe the government would just allow them to fail, and private creditors — holders of Fannie Mae and Freddie Mac securities — would eat the loss. But … um … probably not? One, because the government has funding commitments to provide up to $258.1 billion more in funding to the entities (see page 3 of the stress test results), but two, because those funding commitments are in place for a reason. Fannie and Freddie are not just random private companies that the government can cheerfully allow to fail if they eat through their capital: They are the underpinnings of the mortgage system. So Treasury rescued them in their last crisis, and without some sort of structural reform, it will rescue them in their next crisis.
- Ugh, I’m sure someone does, but please don’t tell me about it.
- So for instance:
FHFA Director Melvin Watt in a February speech warned that the companies’ falling capital buffers could one day cause investors to doubt their guarantees of mortgage-backed securities. Such uncertainty would cause mortgage rates to go up.
“The most serious risk and the one that has the most potential for escalating in the future is the enterprises’ lack of capital,” Watt said.
This makes no sense if you think that investors think that the companies are government agencies, and that the government is on the hook for their guarantees. I think that, because (1) the government was effectively on the hook for their guarantees in 2008, and (2) since then the relationship between the government and the entities has gotten even closer. But the government can’t … just … say that. So the government has to pretend that it doesn’t back Fannie and Freddie, and that investors don’t think it backs Fannie and Freddie.
- And even do something about it, like the credit risk sharing deals that allow Fannieand Freddie to offload a portion of their credit risk to the private market — while still keeping the entities themselves in conservatorship.
- This paragraph, in broad outline, describes the Johnson-Crapo plan (which zeros the existing preferred shareholders), the Fairholme Capital Management plan (which gives those existing holders equity in the new entities), and even the Parrott-Ranieri-Sperling-Zandi-Zigas plan, which is a little different (involving fixed-dividend securities in a new national corporation) but has similar basic features.
- I wrote this two and a half years ago:
My assumption has long been that the status quo is great for the government. Right now, Fannie and Freddie are completely controlled by the federal government, while keeping the technical trappings of private ownership (20.1 percent private shareholders, no explicit government guarantee) and thus staying off the Treasury’s balance sheet. They can be used as a tool of housing policy (particularly, they can keep guaranteeing cheap mortgages) without being explicitly federalized.
Still true! Still the same status quo.