Collateral trends in the new non-agency
admin | January 18, 2019
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 private label market is not only growing, it is evolving. An analysis of different shelves over time shows even within a given issuer’s program, the composition of collateral can change fairly meaningfully. Not surprisingly, shelves the market considers prime tend to show a relatively stable collateral profile. The others, not so much.
While it seems the new non-agency market has a myriad of euphemisms to describe loans that are being securitized, the reality is they are distinguished by a few things: their loan credit quality, their loan size and eligibility for agency MBS securitization, and by features that determine if the loan is a Qualifying Mortgage. These criteria help filter the current universe of new non-agency loans into a handful of categories, and those categories help distinguish the most active non-agency MBS issuers from each other. While the market often brands a deal, shelf or issuer as simply ‘prime’ or ‘non-QM’ the reality is that the new non-agency is rarely homogeneous and that investors may be exposed to multiple different types of loans in a single transaction. Given this, we offer an alternative to these often inaccurate monikers.
The universe of new non-agency loans falls into a handful of categories:
- Prime agency conforming
- Prime jumbo agency conforming
- Prime jumbo non-agency
- Prime non-QM
- Non-prime non-QM
Applying this taxonomy allows us to compare shelves within a given market convention like prime or non-QM as well as observe how deals within a given shelf might be changing over time. Starting with prime shelves the major noticeable trend is the pronounced increase in jumbo conforming loans being securitized in private label deals. However this phenomenon is far more pronounced in JP Morgan’s JPMMT shelf than in Redwood’s SEMT shelf. (Exhibit 1)
Exhibit 1: Loan level taxonomy across prime major shelves
The JPMMT shelf is also seeing a growing concentration of conforming balance loans being securitized – a phenomenon completely absent from Redwood securitizations. Additionally, based on this taxonomy, it appears there is modestly greater risk layering in the JPMMT shelf where non-prime loans, which we define as below a 700 FICO, non-QM loans make up nearly 5% of the pool in some more recent transactions.
Turning to expanded prime, we look at the Flagstar, Galton and Redwood’s ‘CHoice’ program. Looking across the three shelves illustrates how different shelves that are characterized by the same broad designation can be. While all three shelves contain a meaningful amount of prime jumbo non-conforming loans, those percentages can range anywhere from 16% to 75% even within a given shelf. Flagstar’s FSMT shelf appears to have the most diverse collateral mix of the three issuers. The Flagstar program is really more of a hybrid program, securitizing everything from hyper-prime jumbo loans to 100% investor pools. Flagstar stands out from other issuers given their large and steady population of conforming balance loans in their deals. They also have the smallest, albeit growing, population of non-prime non-QM loans. Looking at collateral trends across the other issuers, the population of prime non-QM loans in Galton’s GFMT shelf has grown steadily over time as the population of non-prime loans has decreased, an apparent credit positive. The population of non-prime non-QM loans has decreased in Redwood’s SEMT CH shelf as well, with those loans being replaced by a combination of agency conforming and jumbo conforming loans. (Exhibit 2)
Exhibit 2: Loan level taxonomy across major expanded prime shelves
Looking at non-QM issuers shows that label to be a pretty stark misnomer albeit in varying degrees. The population of non-prime, non-QM trusts tends to vary substantially across issuers and trusts with the population of those loans decreasing across some issuers and increasing across others over time. One caveat to this is that the taxonomy is derived from Loan Performance data and this data is unavailable for certain later 2018 vintage transactions, making it more difficult to ascertain the more recent trajectory of collateral trends across affected shelves.
One shelf where we do have a complete picture of collateral trends is Angel Oak’s AOMT shelf. Over time the population of non-prime, non-QM loans has steadily decreased, generally being replaced by higher balance prime loans. This trend should be a credit positive for later vintage Angel Oak deals, although given the deleveraging and benign credit performance of earlier vintage deals this change may only be modestly impactful. Caliber’s COLT shelf tends to have fairly limited variability in the composition of its collateral over the life of issuance. Non-prime, non-QM loans tend to vary somewhat and increased into the latter half of 2018. Those loans made up roughly 45% of the issuer’s 2016-3 transaction and 60% of their 2018-3 deal. The increase in non-prime, non-QM loans has been at the expense of smaller concentrations of prime conforming and jumbo conforming loans. Both the Angel Oak and Caliber shelves contain smaller populations of the Deephaven and Verus non-QM shelves which are generally marked by consistently larger populations of non-prime loans.
 For a detailed description of the taxonomy see APS Portfolio Strategy: The emerging non-agency MBS market landscape, September 21, 2018