Earlier today I participated in a conference call hosted by Investors Unite titled, “What is Risk Sharing, and How Does it Work?” Below is the text of my prepared remarks for that call. Included at the end is a brief addendum, “Turning the Tables,” which didn’t fit the format or time allotment for the call but I think adds some useful perspective.
The topic I’ve been asked to address this morning is, “What is risk sharing, and how does it work?”
I’d like to start by making a distinction between what I’ll call traditional credit risk sharing and non-traditional credit risk sharing.
Traditional credit risk sharing was what Fannie and Freddie did before they were put into conservatorship, and for all of the 23 years I was at Fannie. When the companies purchased or guaranteed a mortgage, they took on the credit risk of that loan themselves, and backed it with equity capital. They would do risk sharing—typically either by asking a private mortgage insurer to provide more coverage for high loan-to-value ratio loans on the front end, before the loan was acquired, or by asking them to reinsure existing pools of loans after they’d been acquired, on the back end—whenever they thought it made economic sense, both for themselves and for the borrower.
What’s being done today, though, might be called non-traditional credit risk sharing. Fannie and Freddie are in conservatorship, and even though they’re once again making money and are highly profitable, because of the net worth sweep they’re not allowed to retain earnings or build capital. Ostensibly for that reason, Treasury and the Federal Housing Finance Agency, or FHFA, have been requiring them to use different types of risk-sharing mechanisms to transfer as much of their credit risk as possible to “private market” sources. But mandatory risk sharing—as FHFA and Treasury are requiring—removes the normal economic discipline that exists when it’s the company, and not the conservator or regulator, making the choices about which risks to keep and which to share, and on what terms. And that leaves open the question many people are asking: “In the current environment, how are Fannie and Freddie’s risk-sharing decisions being made, and are they good for the borrower?”
As far as I can tell, what FHFA and Treasury are doing now is taking as a given the guaranty fees Fannie and Freddie are charging—which average close to 50 basis points before the extra 10 basis point payroll tax fee—then saying, in effect, “as long as we have enough guaranty fees to cover the cost, any credit risk sharing we do is better than leaving the risk with the companies, because they have no capital.”
There are, however, at least four problems with that approach. First, it penalizes borrowers by enshrining an arbitrary guaranty fee that’s too high for the “squeaky-clean” loans Fannie and Freddie are buying or guaranteeing currently. Second, it doesn’t permit anyone to compare the relative efficiency of the different types of risk sharing being considered. Third, it doesn’t explicitly take into account risk-sharing costs borne by the borrower, such as for deep-cover mortgage insurance. Any finally, it doesn’t at all address the effectiveness issue—whether a particular type of risk sharing really does produce the result it’s alleged to have.
I proposed what I think is a much better way to set risk-sharing standards in my response to FHFA’s request for input on credit risk transfers. And that is, whether you think Fannie and Freddie ever will emerge from conservatorship or not, use their cost of equity capital as the basis for assessing all risk-sharing alternatives—as we used to do, successfully, with traditional risk sharing.
For this approach to work, though, FHFA has to follow through on the directive in the 2008 Housing and Economic Recovery Act and update the companies’ risk-based capital standards to reflect current requirements for taxpayer protection. FHFA should have done that some time ago, but it hasn’t. All of us need to insist that it do so immediately. We won’t be able to assess the value or desirability of any type of risk-sharing mechanism or transaction without a valid reference point, and that reference point is the cost of having the same risk backed by hard equity capital.
Once we know what it costs to back credit risk with shareholders equity, assessing risk-sharing alternatives is easy. You apply what I call a borrower benefit standard, and ask: will the risk sharing techniques or transactions in question result in a lower all-in cost to the borrower, with the same or a better standard of protection for Fannie or Freddie and the U.S taxpayer, compared with the companies retaining the credit risk themselves?
To show how this borrower benefit standard works—and to illustrate why it’s so critical for FHFA to update Fannie’s and Freddie’s capital standards—I’ll briefly address the three main types of risk sharing: “deep-cover” mortgage insurance on the front end, mortgage reinsurance transactions on the back end, and securitized credit-risk transfers.
The cost of front-end mortgage insurance typically is paid by the borrower. Deep-cover MI obviously costs more than standard MI, so for it to be a good deal for the borrower the guaranty fee associated with it has to be reduced by at least enough to offset the higher MI fee. Today, FHFA has no way of determining how much deep-cover MI is worth. If it arbitrarily lowers Fannie or Freddie’s guaranty fees by enough to offset the cost of deep-cover MI, it won’t know if it’s giving away too much fee for the reduction in risk. And if it doesn’t lower guaranty fees enough, deep-cover MI becomes more costly to the borrower. The only way for FHFA to get this right is to give Fannie and Freddie a real cost of capital; otherwise it’s just guessing.
A similar problem exists with back-end reinsurance transactions. Fannie or Freddie already have these loans—and their credit risk. To get reinsurers to take some or all of that risk, the companies have to pay an annual premium. How much should they be willing to pay? Again, without a binding capital standard to serve as a reference point, Fannie and Freddie, or FHFA, have no way of knowing. And if they just pay whatever the MIs ask, the economics quickly will move against them.
There’s one additional piece to the deep-cover or back-end mortgage insurance assessment. That’s counterparty risk. Private mortgage insurers don’t fully back their exposures with capital, so in a severe loss scenario there’s a chance they won’t be able to make good on all of their insurance obligations, as happened during the financial crisis. The way to assess and compensate for counterparty risk is for FHFA to apply the same stress standard to the MIs as it uses in developing Fannie and Freddie’s capital requirements. If the MIs can’t meet that standard, then they’re not providing full protection to the companies, who will be stuck with any losses the MIs can’t pay. To cover this risk, Fannie and Freddie either would need to price for it, limit the amount of risk sharing they do with these counterparties, or do both.
Now, compared with MI risk sharing, securitized credit-risk transfers are in one sense easier to evaluate today, despite their complexity, because their effectiveness and value don’t depend on Fannie and Freddie’s cost of capital; they depend entirely on the structure of the securities themselves, which is known at the outset.
The challenge with risk-transfer securities is determining how to equate a dollar in face value of these securities with a dollar of upfront equity capital. At first blush that would seem to be an impossible task, but in fact the prospectuses for both Fannie’s Connecticut Avenue Securities (or CAS) risk-sharing program and Freddie’s Structured Agency Credit Risk (or STACR) program give you everything you need to make a very good approximation of this equivalency.
I’ll use Fannie’s recent CAS issues to show how this works. In the prospectuses for these deals, Fannie projects credit losses for 64 combinations of annual credit loss and prepayment rates. The average cumulative credit loss for the 64 scenarios, over the life of the securities, is a little over 2 percent of the initial pool balance. The worst loss rate is 10.3 percent, and the loss rate at the 90th percentile is 5.3 percent. All of those are quite severe. The ever-to-date loss rates on Fannie’s loans from the early 2000s have only been about one-half of one percent.
Using the projected loss data, I next did an analysis that anyone with the prospectus, a calculator and 15 minutes of spare time could do. I distributed the projected credit losses from all 64 scenarios into the three categories that replicate the structure of Fannie’s CAS securities: the first 100 basis points of loss, which Fannie takes, the next 300 basis points of loss—from 1 to 4 percent of the initial balance—which the CAS tranche holders are supposed to take, and finally any losses over 4 percent of the initial balance, which Fannie again takes.
The distribution of these credit losses didn’t surprise me: 37 percent of them were at or less than 1 percent of the initial pool balance; 48 percent of the losses fell between 1 and 4 percent of the initial balance, and 15 percent of the losses were in excess of 4 percent. That means that in theory Fannie and its loss-sharing partners would split the credit losses roughly equally, with Fannie taking the first 37 percent, risk-sharing partners taking the next 48 percent, and Fannie the last 15 percent.
Except that’s not what would happen in reality, because Fannie has structured its current CAS risk-transfer securities so that they pay down before most of the credit losses can hit. And Fannie discloses the consequences of that structuring in its prospectuses. Over all 64 scenarios, of the 48 percent of the credit losses that fall between 1 and 4 percent of the initial pool balance—the range the CAS M-2 and M-1 tranches are supposed to absorb—these tranches are shown in the prospectuses as picking up only 4 percent, or less than one-tenth of what they could have taken.
Think about that for a second. In severe credit loss environments, using its current CAS “risk-transfer” securities, Fannie would take the first 37 percent of losses under 100 basis points, also pick up 44 of the next 48 percent of losses, and then take the remaining 15 percent. After the dust had settled, Fannie would have absorbed 96 percent of the credit losses, and holders of the CAS securities…4 percent. That’s it.
Those aren’t just my figures; anyone can calculate them from the data Fannie publishes. But apparently almost nobody does, or if they do they ignore the results.
There are two points I’ll make about them.
First, Fannie’s CAS issues are a complete waste of money. On the loans Fannie covers with these securities, it’s giving up close to a third of their guaranty fees in interest payments on risk-sharing tranches that will absorb at most 4 percent of their credit losses. That makes no sense from any perspective. It’s a waste of Fannie’s money if shareholders win the net worth sweep lawsuits, and it’s a waste of the government’s money if the government wins those lawsuits.
Second, and even more importantly, FHFA, Treasury and the members of the Financial Establishment who tout CAS-like risk-sharing securities as the future of mortgage securitization are pretending that the face value of these securities is the equivalent of equity capital, when they’re worth less than one-tenth of what equity capital is worth because they absorb less than one-tenth of the credit losses.
Saying you’ve transferred credit risk when you actually haven’t is a prescription for disaster. We’ve tried fooling ourselves with securitizations before. In the 2005-2007 bubble years, we had the collateralized debt obligation, or CDO, in which investment bankers gathered up toxic lower-rated tranches of subprime and other high-risk private-label securities and put them into new derivative securities, 80 percent of which the rating agencies gave AAA or AA ratings by assuming that the underlying risky loans couldn’t all go bad at the same time. We know how that turned out.
There’s not much difference between hiding credit risk behind inflated ratings and pretending to have transferred it to investors when you know you really haven’t.
I’m not happy about the fact that FHFA is contemplating doing deep-cover MI or back-end reinsurance transactions without any clear idea of whether they’re good for the borrower or for Fannie or Freddie. But I’m much more worried about the consequences of what FHFA and Treasury are doing with securitized credit risk transfers.
FHFA is making Fannie and Freddie do CAS and STACR securities because Treasury is telling it to, and Treasury is pushing FHFA on this because it wants us to believe it can replace the shareholder-provided equity of Fannie and Freddie—two companies Treasury historically has opposed—with securitized risk-sharing as the foundation of the secondary mortgage market of the future. But that’s a fiction. Fannie and Freddie’s CAS and STACRs, like CDOs, are exercises in deception.
We need to call Treasury and FHFA out on this. People need to write, phone or meet with their contacts at both agencies, and make clear to them that we know the risk-sharing mechanisms they’re forcing Fannie and Freddie to use are deeply flawed, and that they don’t transfer any meaningful amount credit risk. We must insist to Treasury and FHFA that they stop trying to fool us, and instead turn their attention to devising reforms for the system that might actually work.
Turning the tables: a brief addendum
Imagine that using securitized risk sharing as a dollar-for-dollar substitute for equity capital was proposed not by Treasury and FHFA, but by Fannie and Freddie, in response to a change in the minimum capital requirement for their single-family credit guaranty business.
The 2008 Housing and Economic Recovery Act (HERA) permits FHFA to “establish a minimum capital level for the enterprises…that is higher than the level specified in [the 1992 legislation] to the extent needed to ensure that the regulated entities operate in a safe and sound manner.”
Let’s say that Fannie and Freddie were not in conservatorship, and that FHFA exercised its authority under HERA to raise the minimum capital requirement on their single-family credit guarantees from the pre-crisis level of 45 basis points to a full 4 percent. Fannie and Freddie respond with the following proposal: “We’ll use retained earnings and shareholders equity to cover the first one percent of the new capital requirement, but then we’d like to use our CAS and STACR risk-transfer securities to meet the remaining three percent of that requirement.”
Which of these two responses do you believe FHFA and Treasury would be more likely to make?
Response one: “That’s an excellent idea! We applaud you for taking the initiative to bring private investor capital into the mortgage market. Yes, we know some people question the effectiveness of these structures, and think they won’t be available in difficult times, but we’re confident you’ll be able to figure out something before any of that becomes a problem. One percent equity capital should be fine; you can cover the rest of your new 4 percent capital requirement with CAS and STACRs.”
Response two: “You’ve got to be kidding! You can’t possibly use the face value of CAS and STACRs to meet your equity capital requirement. Read your own prospectuses. The way you’ve structured these securities, they hardly ever absorb any losses. It’s your call if you want to issue CAS and STACRs, but the most we’ll give you as a credit against your new equity capital requirement is ten percent of their face value. You’ll have to get the other 2.7 percent in real capital.”
Not much doubt, right? It shouldn’t be the opposite if it’s Treasury and FHFA, rather than Fannie and Freddie, making that same proposal. Yet somehow it is.