Uncategorized
Screening for prepayment protection in non-QM MBS
admin | June 21, 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.
Investors in non-QM private MBS have seen plenty of fast prepayments. Certain loan features may give investors protection from fast speeds and subsequent reinvestment at today’s current low interest rates. Geography and limited documentation and, to some extent, loan balance appear to offer significant prepayment protection while FICO has less impact.
The credit box in non-QM lending has grown and stretched fairly substantially since the inception of these deals. At the same time, competition among lenders has increased, driving risk premiums on loans lower. The confluence of an expanding credit box, lower risk premiums and a substantial rally in interest rates could leave non-QM susceptible to fast prepayments. It’s a good time to look at how sensitive different types of non-QM collateral are to different amounts of refinancing incentive using a traditional S-curve analysis.
Screening for prepay protection in non-QM is somewhat akin to taking a test with the answers provided, or at least good indications. The agency MBS market is extraordinarily efficient in pooling collateral stories that may afford investors with more stable cash flows. It is a good roadmap to possible similar protection in non-QM space. The key difference in analyzing these stories in credit space is that there can often be both a rate and credit spread tightening component to refi incentive in private lending. And while the rate component is relatively easy to measure, the credit component is less so given the significant variance across different programs underwriting standards.
New York, New York
One collateral story that appears to offer substantial prepayment protection in non-QM is geography. New York loans tend to be less sensitive to refinancing incentive given the friction to refinancing associated with the state’s mortgage recording tax. Loans in New York, New Jersey and Florida show very little response to 50 bp of refinancing incentive, with all three states prepaying at less than 10 CRR. As these loans go deeper in the money, prepayment rates on New York loans remain relatively muted. Given an additional 100 bp of refi incentive, New York loans still pay at less than 8 VPR while New Jersey and Florida loans prepay nearly twice as fast as New York. Loans in California pay nearly three times as fast as those in New York given 150 bp of total refi incentive. (Exhibit 1)
Exhibit 1: Non-QM S-curves by state
Source: Amherst Insight Labs, Amherst Pierpont Securities
There are a host of variables that could suppress or elevate prepayment rates. However the most significant variable is generally loan size. The average loan size on New York loans observed was just approximately $480,000 significantly larger than New Jersey and Florida Loans that averaged $350,000 and $390,000 respectively. The average loan size of California loans which paid three times faster than their New York counterparts were just $50,000 larger, averaging $530,000.
Looking at the effect of loan size in isolation suggests that lower loan balances may offer limited protection given small changes in refinancing incentive but materially greater ones as loans go deeper in-the-money. At 50 bp in-the-money prepayment differences between low and higher loan balance loans tend to be fairly muted. Loans with balances less than $200,000 and greater than $1 million both prepay at less than 10 VPR. However that difference becomes far more pronounced given an additional 100 bp of refi incentive. Loans with balances $1 million or greater prepay at just under 25 CPR given 100 bp of refi incentive while low loan balance loans prepay at roughly half that speed. Loans with balances between $600,000 and $1 million prepay roughly in line with loans that are greater than $1 million suggesting that there is little prepay protection once the loan breaches the conforming jumbo limit.
Exhibit 2: Loan size matters…up to a point
Source: Amherst Insight Labs, Amherst Pierpont Securities
Other traditional frictions to refinancing like those underwritten using less than full documentation and lower credit score borrowers seem to offer mixed sources amounts of prepay protection. Given a small amount of refinancing incentive full and limited documentation loans prepay at roughly the same speeds. As refinancing incentive increases, the disparity in prepayment rates widens materially and it appears limited doc loans experience prepay burnout at 200 bp in-the-money or more. At 50 bp in-the-money both types of loans prepay at less than 10 VPR but given an additional 100 bp of refi incentive full doc loans prepay at nearly 30 VPR while limited doc ones pay at just beloe 17 VPR. Both sets of loans have average balances of $550,000. Prepayment rates on limited documentation loans plateau at roughly 25 VPR given 200 bp of moneyness or more. Deep in-the-money speeds on full doc loans are greater than 45 VPR. (Exhibit 3)
Exhibit 3: Deep in-the-money limited doc loans tend to exhibit burnout
Source: Amherst Insight Labs, Amherst Pierpont Securities
Prepayment rates across a wide distribution of credit scores are relatively benign assuming 50 bp of refi incentive. As loans travel deeper in-the-money, lower FICO loans tend to pay in-line with better credit borrowers. At 150 bp in-the-money, loans with a 640 to 660 FICO prepay at just over 22 VPR while loans with credit scores 100 points higher prepay significantly slower, at just over 14 VPR. This is likely a result of S-curve analysis only capturing pure rate incentive as lower FICO may have had additional refinancing incentive as a result of credit spread compression. Somewhat counterintuitively, higher FICO loans in non-QM trusts may offer better overall convexity given fast out-of-the-money speeds and less susceptibility from refinancing risk as a function of credit spread compression. (Exhibit 4)
Exhibit 4: Low FICO loans offer limited prepay protection in non-QM trusts
Source: Amherst Insight Labs, Amherst Pierpont Securities