Fannie Mae, and also Freddie Mac, are exploring new ways to transfer the credit risk of mortgages that they insure.
Unlike their existing risk-sharing programs, which have drawn $12.5 billion of private capital into the mortgage market by referencing $454 billion of mortgages since their July 2013 inception, some of the latest deals transfer the first loss sustained when a homeowner stops making payments.
That goes much further in scaling back the exposure of taxpayers, who are ultimately on the hook for the $6.5 trillion of residential mortgages that Fannie and Freddie insure.
Both Fannie’s Connecticut Avenue Securities (CAS) and Freddie’s Structured Agency Credit Risk (STACR) require losses to reach a threshold before the government sponsored enterprises receive funds to cover them.
It is also closer to the role envisioned by legislation sponsored by Senator Tim Johnson, a Democrat from South Dakota, and Senator Mike Crapo, an Idaho Republican. Their bill would replace Fannie and Freddie with a government reinsurer that would bear losses only after private capital covers the first 10%.
There are yield-hungry investors eager for exposure to housing who are more than happy to take on this first-loss risk, thanks in part to a dearth of issuance private label residential mortgage backed securities.
Roughly $7.0 billion of private label RMBS was issued for the year through November, versus $10.7 billion in risk-sharing deals referencing $369 billion in collateral, according to Barclays.
In October, Fannie and JPMorgan Chase entered into a transaction that offloads the risk of a $989 million reference pool of agency mortgages originated by JPMorgan Chase Bank. The loans in the pool have been delivered to Fannie, which, in turn, has bundled them into a Fannie-guaranteed MBS. As with any MBS guaranteed by the GSE, JPMorgan Chase Bank is required to buy back any loans that don’t meet the agency’s criteria. The bank also acts as the servicer and the provider of reps and warranties.
JPMorgan did not return a request for comment on this article, while Fannie declined to comment.
Unlike CAS, which are general obligations of Fannie Mae, these securities were issued by a special purpose vehicle dubbed J.P. Morgan Madison Avenue Securities Trust, Series 2014-1 (JPMMA).
Fitch was the only agency to rate any of the deal’s tranches. Available to market investors are two M tranches, analogous to ones that have been issued under STACR and CAS. The $19.78 million M-1 tranche, rated BBB-’, priced at 225 basis points over one month Libor. An unrated M-2 tranche for $27.20 million came out with a spread of 425 basis points. Both have a 10-year legal maturity.
By comparison, the M-1 tranche of the first ever risk-sharing deal, a STACR transaction printed in July 2013, came out at 340 basis points over one-month Libor. But yield-hungry investors growing comfortable with the sector have pushed spreads increasingly tighter from late last year through much of 2014. The M-1 notes in Fannie’s most recent deal in November, CAS Series 2014-C04, priced at a spread of 196 basis points, up from 125 basis points in the previous deal, in July, but still below spreads a year ago.
Pricing on the M-2 tranches of risk-sharing deals has had a similar evolution. The first STACR deal came out at 715 basis points over one-month Libor. The M-2 notes of CAS Series 2014-C04 priced at 490 basis points over, from 300 basis points in a July deal.
Another comparison might be to private-label RMBS, since investors in the M tranches of risk-sharing deals are basically taking on credit risk akin to mezzanine and subordinated tranches of such deals.
While the collateral between the two kinds of transactions is different (conforming versus jumbo loans), in Fitch’s view, the payment priority of the M-1 notes of JPMMA will “result in a shorter life and more stable credit enhancement than for mezzanine classes in private label RMBS, providing a relative credit advantage.”
An A-H tranche for $942 million also exists entirely for the purpose of calculating any write downs of the M notes in the case of recourse events. A-H functions as the senior-most tranche, with a 4.75% credit enhancement from the M tranches. As long as the credit enhancement is sustained, the principal payment will be distributed to the M notes. Investors in the M-2 notes only receive payments once the M-1 notes are paid in full.
Interest payments for M-1 noteholders are paid out of an special account that is fed by an interest-only strip of 26.88 basis points on the mortgage pool, paid by JPMC as the servicer, as well as investment earnings on the cash collateral account (CCA) and amounts on deposit in the reserve fund. In Fitch’s view, the IO strip alone should be enough to cover the payments.
As with a traditional mortgage deal, as well as the CAS and STACR transactions, the quality of the pool — in this case referenced — is a key determinant of the deal’s creditworthiness and ultimate performance. The collateral consists of prime, 30-year, fixed-rate mortgages held by borrowers with good credit. The average loan-to-value ratio of the underlying properties is 76, the average FICO score of the borrowers 750 and average debt-to-income ratio 34%.
But there are some key differences between Madison Avenue Securities and Fannie’s CAS deals. While CAS, as well as STACR, are general obligation bonds and remain on balance sheet, JPMMA is issued by an off-balance sheet special purpose vehicle. So the counterparty risk is different: investors are relying on Chase, rated single-A by Fitch, to repurchase loans that Fannie puts back due to violations in representations and warranties. They are also relying on Chase to pay the retained IO strip to the interest account; Fitch has an RPS2+ servicer rating on the bank. By comparison, CAS investors are relying on on Fannie Mae’s AAA’ corporate credit rating.
Another key difference is that Fannie Mae does not retain a first loss class or a vertical slice of the M securities, as it does in a CAS transaction. Fannie will only absorb losses once the 4.75% credit enhancement provided by the M securities has been drawn down. This is a way for Fannie to offload even more credit risk than it does in its CAS deals.
Barclays also argues that the JPMMA-style deals “have an advantage [over CAS] for Fannie in that it can simultaneously buy protection on loans that it guarantees in the agency MBS market.” The STACR/CAS deals, on the other hand, typically reference loans with average seasoning of 6-10 months.
Another difference between JPMMA and CAS, one that may not be as important to investors, is the lack of originator diversity. The JPMMA deal — and others like it — is a pure play on the underwriting of a single bank. But investors might take the view that all mortgages underwritten to Fannie Mae’s standards are equally risky or safe. Even if they view Chase as a marginally better, or worse, underwriter, they are still getting a significant exposure to the bank through CAS. Chase is already one of the most heavily represented underwriters in CAS. In CAS deals 2014-C02 and 2014-C03, the share of JPMorgan Chase mortgages in the reference pools is 8%.
In other ways, however, there is more diversity in the credit metrics of the collateral. For instance, the loans backing JPMMA are less concentrated in California, the state mostly heavily represented in CAS deals. Indeed, nearly a third of mortgages in the pools linked CAS 2014-C02 and -C03 are on California homes.
At any rate, the basic structure of JPMMA could be used to pool together loans and mitigate the risks associated with a single originator.
For lenders and investors there could be other incentives for getting involved with a JPMMA-style deal: a lower fee for the guarantee that Fannie provides. The issuing vehicle retains the interest-only strip of JPMMA, which yields 26.88 basis points over the notional of the reference pool. Observers point out that this IO strip functions as a guarantee fee, or G-fee, since it is effectively paid by the originator.
Fannie and Freddie charge G-fees to originators for servicing conforming mortgages, bundling them into MBS and then selling the securities. The fee is primarily designed to cover the GSEs’ exposure to credit losses from the guaranteed mortgages.
Originators who believe that this fee is too high for the risk being taken on could be incentivized to do such deals. Once the class M securities are paid in full, a class of excess IO receives funds from the IO strip. Originators could sell an excess IO class in the structure to investors, who can pocket the difference between what Fannie is charging - in the case of JPMMA the 26.88 basis points over the reference pool - and what they would pay for that credit risk. “It’s about monetizing the difference between what investors will pay for any excess IO over the [26.88 basis points] in the cash collateral account,” said Clifford Rossi, executive-in-residence at the Robert H. Smith School of Business at the University of Maryland. “It’s an arbitrage on the guarantee fee.”
As result, this sort of risk-sharing deal is an opportunity for investors to take a view on credit risk and G-fees.
Still About Yield
Some players say the spread on JPMMA wasn’t such a good deal for the bank.
“How does that benefit Fannie Mae?” asked Walt Schmidt, manager of mortgage strategies at bank dealer FTN Financial Capital Markets. “It’s worse pricing for them.”
Manish Kapoor, managing principal at West Wheelock Capital, said that, for investors, JPMMA has the same appeal as other risk-sharing deals: These securities are the only new mortgage bonds where one can get a “yield play.”
He added that the secondary market in STACR and CAS also is not particularly active, so investors focus on the primary offerings.
Schmidt said that investors seeking similar prepayment risk to JP Morgan’s JPMAA could purchase agency bonds and strip out other originators.
Barclays pointed out that JPMMA-type risk sharing deals could be eligible for the 75% bucket in REIT portfolios. On an annual basis, at least 75% of a REIT’s gross income must come from real estate-related income such as rents from real property and interest on obligations secured by mortgages. As direct obligations of Fannie and Freddie, STACR and CAS are ineligible for this designation. If other kinds of risk-sharing transactions become eligible, it would open up an enormous investing pool for them. Stock-exchange-listed and nonlisted REITs combined own about $1.7 trillion of commercial real estate assets as of June 30, 2014, according to REIT.com. A lawyer who works with trade association NAREIT did not return an email on the eligibility question.
At any rate, players believe there will be more risk-sharing deals beyond CAS and STACR.
“Based on feedback received from investors, [JPMMA] has generated a lot of interest and more programmatic issuance would likely be supported,” said Suzanne Mistretta, senior director in Fitch’s U.S. RMBS group. “Because there is some reliance on Fannie and Freddie with this program, private issuance will be driven by the participation of the GSEs and their risk sharing mandate and goals set forth by the FHFA [Federal Housing Finance Agency].”
Other players have said the directives of the FHFA, which regulates Fannie and Freddie, would likely act as an incentive to do more of these and other deals that help parcel risk to third parties. The FHFA requires the GSEs to seek new ways of offloading credit risk beyond the STACR/CAS structures. Indeed, for 2014, they were required to do at least one such deal. Freddie had met this requirement purchasing a number of insurance policies covering nearly $555 million of credit losses. So far, Freddie has yet to issue a transaction along the lines of JPPMA.
A Freddie spokesperson declined to comment on this article.
According to Barclays, Fannie did not seem to have fulfilled this requirement until the JPMorgan deal.
“I think there could be more,” said Rossi. “People are still looking for yield [and] there’s nothing there.”
There are also other ways for the GSE’s to offload risk and have a third party take the first loss position.
REIT Redwood Trust has made a private, bilateral deal with Fannie to sell the agency $1.1 billion of conforming loans, on which Redwood will absorb the first 1% of credit loss.
“We saw an opportunity to enter another segment of the market that’s five times larger than the jumbo market to put capital to work,” said Mike McMahon, managing director at Redwood.
Redwood is one of the biggest issuers of private-label mortgage backed securities, which, since the financial crisis, has been confined to loans that are too big to be guaranteed by Fannie or Freddie.
But even these deals have been limited by lingering questions over rep and warranty frameworks and structures and, in a chicken-and-egg scenario, the failure of issuance to reach enough volume for investors to justify devoting manpower to assessing these transactions.
The $7.0 billion of year-to-date issuance of non-agency bonds backed by new mortgages is dwarfed by pre-crisis volumes of $1.2 trillion in each of 2005 and 2006.
Redwood has completed 24 private label jumbo deals post-crisis for a total of $10 billion. This has created over $600 million of credit securities for its investment portfolio.
Unlike JPMMA, the Redwood-Fannie deal does not include a bond component being sold to the market — the lender does not want to dole out the risk in this transaction.
“Why would we want to sell credit risk?” McMahon said. “That’s what we invest in.” He added that Redwood did not simply hold on to the $1.1 billion in mortgages because it does not have the capital or balance-sheet structure to do so. “It would require way too much capital versus investing $11 million for the first 1% of credit risk.”
McMahon added that finding value in deals with credit risk is tough as the competition among investors is fierce right now.
Both Fannie and Freddie have said they are issuing more CAS and STACR deals in 2015. Freddie has said it will expand the program to include an actual loss credit offering. IFR reported that this new product will differ from the current STACR deals because its performance will reflect “precisely what is recouped (or lost) on an individual loan once it is sold or refinanced out of the pool.” Current STACR deals have estimated losses.
Partly to facilitate this future iniative, Freddie announced on Nov. 24 that it will start providing actual loan-level loss data. This will give investors the tools they need to assess an actual loss credit transaction. “Having data openly available in the marketplace allows us to expand the amount of risk transferred to private investors,” said Kevin Palmer, vice president of single-family strategic credit costing and structuring for Freddie.
Laurel Davis, vice president for credit risk transfer at Fannie, said in a November release that the agency plans to issue quarterly next year and is seeeking ways to expand the risk-sharing investor base.
