Uncategorized
Gains in issuance, forbearance and changing callability
admin | November 30, 2018
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
The new non-agency MBS market looks poised to grow significantly next year, with a big boost potentially coming from a flow of loans out of the agency market and into private securitization. Changes at the top of the house at Federal Housing Finance Agency could accelerate the flow. Forbearance stands to get more focus as an underappreciated source of return in legacy RMBS. And the landscape of called legacy RMBS deals stands to shift, too, as interest rates edge higher and loan values fall.
***
Drivers of issuance in 2019
For the first time in a decade, issuance of new private-label RMBS in 2018 will eclipse runoff of the legacy market. And prospects for an encore in 2019 look strong. Prime jumbo issuance has made up 60% of issuance year-to-date. Expanded prime and non-QM transactions have each made up roughly 20% of issuance year to date. Total issuance should come in at roughly $30 billion for this year and could increase 50% or more next year.
Increased issuance will likely hinge on a handful of factors but potentially the most meaningful is the continued flow of GSE-eligible loans into private-label execution. Based on an analysis of one month’s 30-year fixed-rate GSE production, roughly 10% would have had a better private-label execution. Given current trends in issuance, it’s not surprising that jumbo conforming and investor loans made up the overwhelming majority of collateral that would have had a more favorable PLS execution. Private-label execution would have been better on roughly three quarters of investor and 40% of conforming jumbo originations. The improvement in execution potentially is significant. Based on estimated subordination and clearing levels in a private-label deal, private-label execution looks approximately $1-16 better than agency execution on certain jumbo conforming loans (Exhibit 1).
Exhibit 1: Conforming jumbo private-label execution may be 1.5 point better than GSE
Note: We estimate that a 760 FICO, 70 LTV, 30 DTI loan would require roughly 8% credit enhancement to AAA and a 4.0% pass through would trade roughly 1 5/8 to 1/3/4 points back of TBA FNCL 4.0%s. We then size the rest of the capital structure consistent with current published rating agency subordination levels. Assuming a 4.75% note rate we strip off a 25 bp base MSR strip and a AAA IO which we price to a 4 times coupon multiple. We then take that same loan strip a 65 bp guarantee fee, 25 bp LLPA, 25 bp base servicing strip, 10 bp excess servicing strip valued at a 4 times coupon multiple and deliver the loan into FNJMCK 3.5% execution at roughly one point back of TBA. Source: Amherst Pierpont, As of November 27, 2018
A 50% bump to private-label issuance may be conservative if 10% of monthly conventional originations flow into non-agency securitizations. And that 10% is potentially poised to increase further based on pending changes at the Federal Housing Finance Agency. Current FHFA Director Mel Watt will leave office in January next year. Although it’s unclear at this time who the president will nominate, there is significant conjecture that the next director will take measures to reduce the size of the GSEs footprint. A recent Inside Mortgage Finance article identified a leading candidate as administration economist Mark Calabria, who has expressed anti-GSE sentiment in the past. If appointed, it’s plausible that he could increase guarantee fees the enterprises charge. An increase in fees could further skew more agency eligible production into private-label execution.
Forbearance – more than you may think
It’s possible that the equivalent of 10% of the outstanding legacy universe or $30 billion is forbearance inaccurately reported as losses. A growing amount of loans with no forborne principal attached has received principal recoveries in excess of the reported loan balance when the loan leaves the trust. These recoveries are a function of one of two scenarios. In some cases, the forbearance has been misrepresented as an outstanding prior loss, which is reversed in part or whole when the loan leaves the trust through either prepayment or liquidation. In other cases, loans show recoveries on balance reductions never recorded as a loss. Based on capital structure leverage, these recoveries can drive bond prices up 50% or more. Given the scale of the opportunity and potential impact on bond prices, this phenomenon may be the biggest source of alpha in the legacy market or all residential credit for 2019.
To date, an estimated $16 billion in loans with more than $5 billion of potentially unreported forbearance have liquidated. Recoveries rates have averaged 7% on these unreported balances. We estimate that roughly a quarter of the outstanding PLS universe has $31 billion in outstanding unreported forbearance. Model projections predict that 84% of these loans will either prepay or pay in full at maturity suggesting that some or all of $25 billion in unreported forbearance could be recovered when those loans prepay or mature.
There is likely some skepticism among market participants that such a large amount of potential principal recovery could go unnoticed for so long. But unreported forbearance recoveries are a real cash flow event and they continue to occur. In order to estimate the amount of unreported forbearance, we observe abatements on all loans where there is no reported forbearance and therefore no forbearance recovery when the loans leave the pool. We capture both the total recorded loss and balance reductions on loans that abate and then look for any reversal of those losses or balance reductions when the loans leave the pool. In trusts where we see reversals of past losses or balance reductions, we add the outstanding recorded losses and balance reductions as potential unreported forbearance.
An Example: ARSI 2006-M3
One of many specific examples of this phenomenon is ARSI 2006-M3. Over the life of the deal we observe $63 million of loans leave the pool with no reported forbearance. Those loans had $4.2 million in reported outstanding losses and just under $3 million in reported balance reductions. Those loans experienced a 13% recovery on past losses and a 17.8% recovery on previous balance reductions. The deal has an additional $56 million in outstanding losses and $30 million in reported balance reductions, and more than half of those have been recently converted to outstanding reported forbearance in the amount of $45.8 million.
Exhibit 2: Liquidated vs outstanding unreported forebearance – ARSI 2006-M3
Source: Intex, LP, Amherst Pierpont
While there are some discernible patterns in unreported forbearance recoveries, they can vary significantly across shelves and trusts. Additionally, larger pockets of unreported forbearance appear to be localized to certain trustees. While the opportunity looks significant it is by no means uniform across the legacy universe. The ability to identify shelves and bonds that have disproportionate upside to these recoveries may be the biggest potential source of alpha in mortgage credit next year.
What’s the call?
While Fed Chairman Powell this week may have left market participants believing that we are close to the end of the tightening cycle, interest rate risk still exists in the legacy market. Options on collateral also have interest rate duration as the value of the option falls if rising rates reduce the value of the collateral. As a result of this, the profile of deals that get called may shift if interest rates continue to rise, potentially shifting the mix away from higher WAC fixed rate collateral to higher margin post-reset hybrids.
We cut the legacy universe into one year duration buckets by shelf and average collateral price assuming a 4% yield-to-price in an effort to ascertain which shelves have a large population of currently callable collateral and how susceptible that collateral’s price is to higher rates. Based on this methodology it appears that even with a material backup in rates, CWL deals should remain callable given an average collateral price north of 104 and a duration of just over two years. Conversely, the value of the call option should likely be discounted on shelves trading closer to par whose collateral has more rate duration like the BSARM and CWALT shelves.
Exhibit 3: Weighing the value of the call option against higher rates
Source: Amherst Pierpont
Valuing the call option in legacy space is a complex and muti-variable analysis. While the always-performing and re-performing loans will be called at par, the non-performers will be assigned a lower and less transparent valuation. Additional considerations include the value associated with recovering outstanding advances, which are likely currently being financed and the potential value associated with forborne principal. Even with those considerations in mind, it is impossible to know the effective strike price of the call if the owner of the call rights owns discount bonds which would effectively reduce the strike price below par.
This ambiguity was less pronounced against the backdrop of lower rates as seasoned higher WAC, fixed rate subprime deals with relatively small non-performing buckets had a significantly higher likelihood of being called than other profiles. Against higher rates the picture becomes even less clear but the call option will likely remain significantly more valuable for certain shelves and less so for others.