For Yves Smith
Updated -- Below is a note from Alan Boyce at Absalon regarding inverse floaters, a subject that goes to the heart of the debate over whether Fannie Mae and Freddie Mac are preventing millions of lower income home owners from refinancing. You can see the Boyce-Mayer-Hubbard proposal for streamlined mortgage refinancing here:
Not only is the large bank-GSE cartel preventing millions of Americans from refinancing, but these same cartel players are also thwarting Fed monetary policy and hurting all our economic prospects. Housing is the primary conduit for Fed monetary policy. As Laurie Goodman of Amherst Securities has said over and over, there is a cost to doing nothing.
Last week when ProPublica reported that Fannie Mae was managing its portfolio using specific trades to further prey upon low income borrowers in precisely the way that Boyce has described for the past three years, http://www.zerohedge.com/contributed/fannie-mae-freddie-mac-shown-mortgage-market-predators, the Friends of Freddie sprang into action. We first wrote on this on the IRA web site in 2009 thanks to Boyce and others in the Berlin-DC-LA mortgage thread...
Rebuttals of the Freddie predator thesis came flying. Even the good folks at American Banker were hornswoggled into believing that Freddie was innocent of predatory behavior in managing its portfolio. But then FHFA head Ed DeMarco publicly responded. < http://www.propublica.org/article/freddie-macs-regulator-says-trades-were-shut-down-because-they-were-risky>
Of note, Freddie Mac also shut down their investor website Wednesday. Was not up as of this morning (but was put back later in the day <http://www.freddiemac.com/mbs/html/sd_remic_lookup.html>. That is where Boyce looked at 4 years of REMICs and found the 29 deals where Freddie provided the MBS collateral and took back the IIOs. Don’t worry, we have copies of all the PDFs that were up on the FHFA site.
It may be that the defensive reaction by FHFA regarding the web site came about due to Ed DeMarco’s public rebuke of the ProPublica report, a response that was a tad too informative say sources close to the matter. DeMarco essentially admitted to the predatory portfolio trades by Freddie, then indicated that the bad acts were not taking place any longer. To paraphrase the Bard, me thinks DeMarco protests overmuch.
One additional line of inquiry that could be pursued in more detail by Congress is a somewhat more specific issue surrounding who had exactly what information about the policy decisions when. Did any of the investment team at Freddie have any non public information about policy decisions or strategies when they made these investments?
FHFA talked about a Chinese wall in the original story, if we remember correctly. Can we look more at exactly how that wall worked in both directions? More interestingly, how can FHFA oversee both sides of the proverbial Chinese wall without becoming a source of information exchange?
Option one. Investment team knew about policy in non-public ways
Option two: Policy team knew about investment strategy
Option three: Neither team knew about the other, but FHFA knew about both
Key passage from FHFA response to Senator Casey:
“It strikes me that we still don’t know some important things:
1. When did the FHFA’s concerns arise?
2. How did it come to the FHFA’s attention?
3. Did the FHFA – did Mr. DeMarco himself – approve the transactions in the first place?
4. Freddie has ceased these but the company hasn’t wound them down. So they still require specialized risk management. So how have the FHFA’s concerns been alleviated?
5. Did the FHFA think that Freddie was trying to sell down its portfolio while keeping risk on its balance sheet?
6. Why was Freddie doing these in the first place?
Can’t wait to hear the answers to these and other questions when DeMarco next appears before Congress. Tomorrow PM will post a Q&A with Alan on the Freddie Mac story on ZH -- Chris
Inverse Floaters Explained
Inverse floaters (a.k.a. Inverse IOs or IIOs) are complex securities. The discussion below is intended to provide enough detail to understand the transactions, so a reader who is not an expert in structured finance may still find this challenging. Agency MBS have no credit risk, as that is borne by the GSE guarantor, but they do have a myriad of interest rate risks. These liquid and transparent securities are sliced and diced to create highly complex tranches. Complicated securities have a relatively long history in mortgages, often for quite compelling reasons. Securities are designed to meet specific investor objectives, such as allowing investors to take specific risks or hedge existing businesses or portfolios, raising proceeds from the sale of securities. Other tranches are designed to appeal to investors who want to invest in agency MBS but cannot for regulatory or other reasons. Nonetheless, the complicated nature of securitizations makes it hard for a casual observer to understand the transactions and value them using standard accounting and risk metrics.
To understand the difference between an IO (interest only) and PO (principal only) security, consider the cash flows from a fixed rate mortgage. Each month, the borrower makes a payment that involves principal and interest. In the simplest form, the PO tranche receives the principal portion of the payment, while the IO tranche receives the interest component. If the borrower decides to pay off the mortgage, the principal portion receives the entire portion of the payoff, and the IO will receive no future payments. In many bond deals, principal position also receives some interest to make it more attractive to borrowers. This security is called a “floater,” as it receive some interest in addition to principal, usually based on a fixed spread to a short-term interest rate like LIBOR. LIBOR is the “London Inter-Bank Offer Rate”, which is the rate at which large banks are willing to pay for short-term deposits. In these deals, the IO piece receives the fixed rate of interest from the mortgage, less the short-term interest payment that is made to the LIBOR floater tranche, which contains most of the principal. These IO securities are referred to as IO Inverse Floaters.
The LIBOR floater securities (composed of principal payments plus LIBOR-based interest) are attractive for investors because of their low sensitivity to interest rate changes. While prepayment speeds may vary, the bonds are originated and trade close to par, so the economic value to the investor is roughly the same in different prepayment environments. When a borrower pays off his or her mortgage, the LIBOR floater investor expects to be able to re-invest the proceeds in another LIBOR floater security that also pays interest based on short-term rates. Such prepayments might be due to a borrower refinancing the mortgage, defaulting on a mortgage, or selling and moving to a new home.
Structuring securities does not reduce the risk inherent in a pool of fixed rate mortgages. When one class of securities, the floater tranche, is designed to bear little interest rate or prepayment risk, the remaining tranche, the IO tranche, will bear all the remaining risk. In these cases, the IO securities are valuable when borrowers do not refinance at the same time that LIBOR interest rates are low relative to the interest rate on the underlying mortgages. The IO securities are even more valuable when borrowers continue to pay very high interest rates without refinancing, as has been the case with many of the Freddie Mac IO securities.
To understand the risks facing IO and PO classes, it is useful to consider to a relatively simple numerical example. Suppose we begin with $100 million of outstanding Freddie Mac-guaranteed mortgages that are paying an interest rate of 6.5%. First, let us consider an investment in these loans as a whole. These mortgages would be packaged into a bond paying a coupon of 6%, with the other half a percent going to pay for mortgage servicing and Freddie Mac’s guarantee fee. A Freddie Mac 6% coupon security currently trades at a price of about 110 cents on the dollar, so the $100 million in mortgages would have a market value of $110 million. As mentioned earlier, these securities do not present any credit risk, as Freddie Mac guarantees them and will pay an investor in the case of default. However, these bonds still face prepayment risk: if all of the borrowers were to simultaneously refinance their mortgages today, the bondholders would receive their principal payments, but the future stream of interest payments would be completely wiped out. Thus, the $110 million investment would drop in value to $100 million, a loss of $10 million.
Now, let us consider what happens when we split the payments into two types of classes: floaters and inverse floaters. Floaters get most of their value from principal payments and are structured to receive just enough interest so that they trade on the market at par (ie, the market value and face value are equal). For our example, suppose that we create $90 million in face value of floaters that will pay an interest rate equal to the 1-month LIBOR, currently about 30 basis points. The investors holding these floaters will then receive the principal payments from the $90 million in collateral, as well as an interest rate of 0.3%. We then take the other $10 million in principal payments and all of the remaining interest payments and package them into an inverse floater class. The owner of the inverse floater will receive the principal payments and full interest payments from the $10 million in collateral, and it will also receive the remaining 5.7% (6% - 0.3%) interest on the $90 million of mortgages placed into the floaters. This works out to an initial coupon rate of 59.7% for the $10 million (face value) inverse floater. However, this coupon can drop quickly if LIBOR increases.
While these securities are relatively complex, we can actually understand their value just through the use of algebra. First, in terms of pricing, we know that before they are split up, all of the mortgages together are worth $110 million. We also know that the floaters are specifically designed to trade at par, so the $90 million in face value of floaters should trade at very close to $90 million on the market. This makes intuitive sense, given that the floaters have no default risk and that their floating coupon payment provides insulation from any interest rate risk. This therefore implies a $20 million valuation for the inverse floaters, since the two classes together must be worth the total $110 million. Turning to risk, we observe that the floaters have no prepayment risk: they are currently trading at par, and if all borrowers simultaneously prepay their mortgages, investors will receive $90 million. This means that the entirety of the prepayment risk is borne by the inverse floater class. Indeed, if all borrowers were to prepay, the $20 million investment would drop to the $10 million in face value, a loss of $10 million. Thus, the investors who hold the inverse floaters effectively retain the risk on the entire $100 million of mortgages, even if it does not appear that way on their balance sheets.
A Real World Example: Freddie Mac Series 3807
To better understand how inverse floaters work, it is useful to look an actual security issued by Freddie Mac in which the agency retained a position. One particularly striking example of this is Series 3807 of Freddie Mac’s multiclass certificates, which was closed on February 28, 2011. This deal involved a total sale of $5.65 billion of mortgage-backed securities (MBS). This analysis focuses on one class in Group 4, which consists principal and interest payments on $1.88 billion of mortgage-backed securities. As of January 2011, the underlying mortgages had a weighted average loan age of 41 months (meaning most were originated about 3.5 years earlier in mid-2007) and an above-market interest rate of 7.05%. The MBS have a pass-through coupon of 6.50%, which goes to the bond investor. According to disclosures in the prospectus filed with the SEC, Freddie Mac purchased all of the IO classes of the bonds and a PO bond that is based on about $134 million of the principal balances. The remaining cash flows were sold into 7 tranches of LIBOR floaters.
Table 1, shown below from the front page of the offering circular, shows the various bond classes created in Group 4 of this deal. There are 8 bond classes based on principal payments (FC-FL plus PO) and one bond class based on interest payments (SI). The first 7 bond classes, FC-FL, totaling $1,746,143,354, were apparently sold to outside investors. Freddie Mac retained the entire balances of PO ($134,318,720) and SI ($1,746,143,354). For structuring reasons, PO and SI were combined into a new class, SB, based on the PO notional principal of $134,318,720. In other words, investing $1 in SB is exactly identical to holding balances of $1 of PO and $13 of SI, an effective leverage ratio of 13:1.
Table 1 (from front page, S-1, of the Offering Circular Supplement for Multiclass Certificates, Series 3807, dated January 20, 2011)
Next the analysis turns to the interest payments for each class of securities, as shown in the Table 2, below. There are 7 floating rate classes, FC through FL, that have an initial coupon of 0.75%, as compared to 6.25% for SI and an enormous 81.25% for SB. PO is not listed because the class does not receive interest. To fully understand this deal, an analyst must look at the characteristics of the underlying collateral. A table on page 9 of the offering circular lists Group 4 as backed by $1,880,462,074 of Freddie Mac guaranteed fixed-rate mortgages with an weighted-average interest rate of 7.05%, and a pass-through coupon of 6.50%.
Table 2 (from the Term Sheet, P. S-4, of the Offering Circular Supplement for Multiclass Certificates, Series 3807, dated January 20, 2011)
To create the securities in Group 4, Freddie Mac took $1.88 billion of principal from their portfolio and split it into the 8 classes that receive principal payments (FC through FL, and PO). As principal is repaid through amortization and prepayments, each of these eight securities will receive its pro rata share of the cash flows. The division of the interest payments is somewhat more complicated: PO receives no interest, each of the floating rate classes (FC through FL) has a coupon of LIBOR + 0.49%, which initially comes out to 0.75%, and SI has a coupon of 6.51% – LIBOR, or 6.25% to start.
Creating such floating rate classes from a pool of fixed-rate mortgages is made possible through the deal’s structuring. To see this point, consider the interest payments on classes FC through FL, PO, and SI, which account for the entirety of the Group 4. PO receives no interest, classes FC through FL receive a total of (LIBOR + .49%) * $1.746 billion in interest payments, and the SI class receives (6.51% – LIBOR) * $1.746 billion. Added together, these account for interest payments of 7% of 1.746 billion, which also equals an interest rate of 6.5% on $1.88 billion. This is exactly equal to the pass-through coupon of 6.5% on $1.88 billion of MBS in Group 4.
Understanding the structure of this deal, one can now refocus on SB, which has an initial coupon of a whopping 81.25%. Freddie Mac held the entirety of SB (which is the combination of SI plus PO). This high coupon is possible because the SB class amounts to a highly levered position on the interest payments of the agency MBS collateral: $1 of SB represents $1 of PO and $13 of SI. At origination, more than 89% of the total interest payments from the collateral group go to SI, while PO only accounts for 7.1% of the total underlying principal. The SB class is effectively a highly levered bet that LIBOR will remain historically low and that borrowers will not prepay.
Impact on Freddie Mac’s Portfolio Risk
To understand the how this deal impacts Freddie Mac’s exposure to prepayment risk, this document now attempts to calculate the market price of SB, which is unknown because the security is not publicly traded. The method is based on computing the price of the underlying bonds with a 6.5% coupon and subtracting off the value of the LIBOR floaters. This calculation is easier because the floating rate principal classes (FC-FL) typically trade close to par. All of the risk is concentrated in the interest-only classes.
The bonds are backed by securities that are the equivalent to a Freddie Mac MBS with a 6.5 coupon, which traded at roughly 111 in February 2011 (today, it is just under 112). Thus the total market value of the bonds with a face value of $1.88 billion would have been $2.09 billion (1.88*1.11). Freddie Mac sold off LIBOR floaters with a total market value of $1.75 billion, so its position is worth about $340 million. Using mark-to-market accounting, it would appear that Freddie Mac sold off 84% of its positions and held the remaining 16%. However, consider Freddie Mac’s potential exposure to losses before and after this transaction. Before this deal, if all of the underlying mortgages had simultaneously prepaid, the agency’s $2.09 billion worth of MBS would have been reduced to the $1.88 billion of principal, a loss of $210 million. Now, full prepayment would bring the SB class’s value from $340 million to its principal balance of just $134 million, a loss of $206 million. Despite selling $1.75 billion worth of bonds, Freddie Mac has retained almost all of the risk of these mortgages. In fact, as shown below, by holding a short position in LIBOR, Freddie Mac might have actually increased its portfolio exposure. Thus this transaction appears to violate the intent of the mandate given by Congress to reduce its portfolio risk.
Beyond prepayment risk, deals such as those described here also give Freddie Mac exposure to LIBOR that it did not previously have. As discussed above, the coupon on SB is given by 84.63 – (LIBOR x 13), so an increase in LIBOR mechanically reduces the coupon on SB by a factor of 13. Such LIBOR exposure presents additional risk relative to the Fed Funds rate or other borrowing costs for U.S. institutions. Today, the 1-month LIBOR is currently 0.27%, only 10 basis points higher than its historical low reached in June of 2011. Before dropping dramatically during the financial crisis, the 1-month LIBOR was consistently over 5% and was above 3.8% as recently as October 2008. If LIBOR rates rise, the value of the Inverse IO securities will fall rapidly given the 13:1 leverage. That is, for every 25 basis point increase in LIBOR, the interest rate earned on class SB will fall by about 3.25 percent. Although a large increase may initially seem unlikely given today’s low rates, the price of Eurodollar futures imply that LIBOR will be above 3% by 2017. An increase in LIBOR rates to 3% would effectively wipe out half of the cash flows associated with Class SB.
Beyond fluctuations in LIBOR itself, Freddie Mac has placed itself at risk to a growth in the spread between LIBOR and the Federal Funds rate, which effectively measures the agency’s borrowing costs. Although the two rates tracked each other closely during the months that deals were made, the graph below illustrates the spread’s growth over the past 6 months from less than 10 basis points to more than 27 basis points. The spread has widened as European banking and sovereign debt problems have grown.
Further increases in this difference would be quite costly to Freddie Mac, as it borrows at rates close to the federal funds rate but has agreed to make payments based on LIBOR. Freddie Mac’s exposure to LIBOR could have easily been avoided, but in order to create these heavily levered securities, the agency had to create floating interest rate classes with coupons directly tied to LIBOR.
Finally, at the height of the financial crisis during the fourth quarter of 2008, the spread between Fed Funds and LIBOR grew rapidly, reaching more than 200 basis points (2%). The current European banking crisis might well lead to a spike in LIBOR relative to Fed Funds, exposing taxpayers to additional risk associated with Europe. Once again, Congress’ mandate to reduce the risk of Freddie Mac’s retained portfolio appears to have been evaded.
 According to documents, Freddie Mac purchased all of the SI and PO class certificates. According to Appendix A, these securities were put into a new class of securities, SB, that were equivalent to, and could be exchanged for, the entire balance of SI and PO. The discussion will focus on the SB securities, but both classes represent equivalent positions on the same collateral The SB class in this deal is what is known as a Modifiable and Combinable REMIC (MACR) class, a type of security that exists primarily to provide for efficient combination and splitting of cash flows in the future. MACR classes can be exchanged for certain proportions of REMIC classes, and these combinations are created at the time of origination and are publicly available in the offering circular (OC). Even though the SI class lists interest on a notional principal of only $1.746 billion, it effectively retains the risk on the entire $1.88 billion of notional principal, as explained below.
 Investors holding the original collateral in Group 4 of this securitization would receive initial payments of 6.50%. The remaining 0.55% is paid to other parties in the original MBS, including the servicer of the mortgage and Freddie Mac in the form of a guarantee fee (Gfee). These Gfees are negotiated in private and are not public information. Large originators have been able to negotiate very low Gfees, to the detriment of the taxpayer backed mortgage guarantor business within the GSEs.