We all know that markets are efficient, so if you let them, Fannie Mae and Freddie Mac's preferred stock prices will give you a civics lesson.
What is the probability that the Senate bill announced yesterday by Tim Johnson and Mike Crapo will in fact wind down Fannie Mae and Freddie Mac while zeroing out the preferred shareholders? As of Monday, before the bill was announced, the preferred stocks traded at around 45 cents on the dollar. As of around 10 a.m. today, when I checked, they were trading at around 38 cents on the dollar. If you lazily treat "cents on the dollar" as "probability of being repaid at par," then the odds of shareholders getting zeroed went up by about 7 percentage points, from 55 percent to 62 percent. So: kind of likely? But not overwhelmingly likely. And not all that much more likely than it was before the bill was announced.
This is not really surprising, since the announcement's only discussion of Fannie and Freddie is that it would "Wind down and eliminate Fannie Mae and Freddie Mac," which could mean a lot of things. Also because a lot of the bill looks pretty fanciful: The key proposal is to "Mandate 10 percent private capital, up front, and create a mortgage insurance fund for the system to protect taxpayers against future bailouts." The idea would be that private mortgage insurers would be capitalized at 10 percent of the value of mortgages they insured, and would also pay a fee to the government -- in the form of a new Federal Mortgage Insurance Corporation -- to backstop any losses beyond 10 percent.
In thinking about that 10 percent number, it's worth considering:
- That banks are required to have about 4 percent capital against mortgage exposures.
- That Fannie's recent annual credit loss rates, on its single-family guarantee portfolio, have ranged from 0.022 percent in 2006 up to 0.774 percent in 2010, with Freddie's numbers similar.
So 10 percent capital would cushion you against roughly 12 straight years of worst-ever losses.
- That Fannie guarantees $2.86 trillion of single-family mortgages, and Freddie guarantees $1.65 trillion, for a total of $4.5 trillion of guarantees -- and $450 billion of required new capital.
So it seems high? Like, if you are a bank, your choice is to keep your mortgages on balance sheet at the cost of 4 percent capital, or to sell them to a new entity which will have to have 10 percent capital
and pay a guarantee fee. Keeping your mortgages on-balance-sheet seems a lot more cost-efficient than getting the new guarantee.
Of course, maybe it's desirable to push mortgages away from a government-backed securitization process and onto banks' balance sheets, but that is the opposite of what Senators Johnson and Crapo say in their announcement, which pushes securitization and access to the secondary market. (And with good reason: There's a long history of Fannie/Freddie-backed mortgages being cheap, and the government likes cheap mortgages.)
The other question is where you raise the $450 billion of new capital for private mortgage insurers. Now it happens that a lot of people are in fact clamoring to provide new capital to private mortgage insurers. Unfortunately, those people are doing their clamoring in the form of owning Fannie and Freddie stock, and whining/suing/lobbying to try to get paid for that stock. Those people are considerably more likely to be interested in capitalizing the new system if they can, in effect, "credit bid" their existing preferred stock at some value other than zero. And it will be considerably harder to get private capital in the new system if there's a recent precedent of let us just say arbitrarily zeroing the last guys who provided private capital to the system.
So it makes sense that the preferred stock investors are not especially thrown by this latest proposal, and still think there's a good chance that they'll end up in the money. There is an obvious mutual incomprehension between those investors, who think they're entitled to (and will) get back par on their shares, and basically all the people in government, who think they investors are entitled to (and will) get zero.
In a vacuum, you'd expect this mutual incomprehension to work out poorly for investors, since the government controls Fannie and Freddie and more or less gets to decide what the investors get. If you're betting on the government, it's your job to understand the government; the government doesn't have any obligation to understand you.
But here, the government doesn't exercise its control over Fannie and Freddie in a vacuum. It's looking to re-create a mortgage industry supported by private capital, and to do that it will need to go a bit of the way toward understanding what private capital wants -- something that it does not seem to have done very much of, judging from the Johnson/Crapo proposal. But private capital will be very, very happy to say what it wants. It starts with getting paid out on the Fannie and Freddie investments.
(Matt Levine writes about Wall Street and the financial world for Bloomberg View.)
Here is a chart. Just manually looking at HP data for the first seven liquid series of each enterprise at around 10 a.m. today, no guarantees that this is right but the moves are pretty similar -- each series lost between 5 and 9 cents on the dollar, with a mean and median of 6.5 and 6.4 cents:
It also says "Start with S.1217 as the base text and generally maintain its overall architecture," and S.1217 -- the Corker-Warner proposal -- has a more detailed wind-down plan, though one that is still not exactly a clear roadmap to what would happen to Fannie and Freddie. The Wall Street Journal says "The Corker-Warner proposal leaves little for those investors, and Mr. Corker said Tuesday that he believed the forthcoming bill wouldn't be any different."
Mechanism is again not super clear, but e.g. section 214 of S.1217 provides some details. Basically there'd be private mortgage insurance companies -- successor Fannies and Freddies -- capitalized at at least 10 percent.
Banks are essentially required to hold "total capital" of at least 8 percent of risk-weighted assets, though when you add capital conservation buffers and G-SIB surcharges you can get up to around 15.5 percent. (See slide 21 here.) Mortgages are risk-weighted at 50 percent.
See page 74 of the 2010 10-K.
Page 63 of the 2013 10-K brings it up to 2013 (14.7 basis point credit loss ratio); 2010 was the peak. Page 60 of the 2005 10-K brings it back to 2002, when the loss ratio was 0.8 basis points. Yes, 0.008 percent! Freddie's disclosure is more annoying but page 7 of the 2010 10-K shows $14.1 billion in single-family guarantee credit losses, and page 81 shows $1.78 trillion in the single-family guarantee book, so that works out to 79 basis points in 2010.
Fannie: page 87 of the 2013 10-K. Freddie: page 71 of the 2013 10-K.
And I bet its cost of capital is higher than a bank's, because (1) it's a pure monoline with no diversification and (2) it probably has less upside potential than a bank. Just clip fees, be prudentially regulated, etc.
Simplistic way to think about it: Say expected credit losses on "conforming" mortgages are 0.45 percent (i.e. super-high 2009 levels, but not record-high 2010 levels). And say a bank's, or insurer's, cost of capital is 10 percent, and say the FMIC charges a 5 basis point reinsurance fee. So your cost of on-balance sheet mortgages is 0.4 percent for capital (4 percent capital times 10 percent cost of capital) and 0.45 percent for losses, or 0.85 percent per year. And your cost of guaranteed mortgages is 1 percent for the insurer's capital charge (10 percent of 10 percent) and 0.05 percent for reinsurance, or 1.05 percent per year. So you choose to keep on balance sheet. No? Obviously liquidity and other reasons argue against this simple arithmetic, but still.
[Update: Dan Davies points out that I am ignoring the insurer's return on capital, which is fair. If you assume they invest in 10-year Treasuries (~2.7 percent) then the insurance costs only 0.78 percent a year, less than keeping it on balance sheet. If they can invest more aggressively (as Fannie and Freddie did! but, oops) then the insurance gets even cheaper. On the other hand, if you assume the expected credit losses are more like 0.25 percent, still high versus through-the-cycle-averages on conforming loans, then on-balance-sheet goes back to winning again. All this math is simplistic, but "a lot more cost-efficient" was perhaps an overstatement.]
A project with which I have some sympathy, but that's for another time.
To contact the author on this story:
Matthew S Levine at firstname.lastname@example.org
To contact the editor on this story:
Toby Harshaw at email@example.com