1.04.2012

Non-Agency RMBS: Private Label MBS - Quick Commentary and Valuation Overview

A few months ago I wrote an introductory post on Maiden Lane and Non-Agency RMBS.  Since then, I've wanted to bring on someone to help write posts on the various flavors of structured/securitized finance.  If you remember, in November, I reached out to readers to see if people would like to contribute.  Luckily, I got a perfect match on the structured finance front.  In 2012, we will try to do at least one post a month discussing everything from CMBS to Prime-X to CLO liabilities.  As we add new writers throughout the next few months, I will add them to a contributor section in the right side-bar so you can see all of the relevant posts to a topic.

With all that said, here is an introductory post on the valuation on private label / non-agency RMBS.  We will be doing a few posts explaining some of the introductory materials in the coming weeks.  Enjoy!

Private Label MBS - Quick Commentary and Valuation Overview

As a subset of the structured products world, non-agency RMBS are essentially claims to cashflows from pools of non-agency guaranteed mortgages. These securities are backed by mortgages that were not qualified to be securitized by the GSEs due to their large balances, lack of full documentation, low credit scores or non-standard terms such as negative amortization schedules. As a result, non-agencies are very sensitive to the credit performance of their underlying collateral.

Current Landscape - 2011 performance:

Non-agencies experienced continuous price appreciation throughout 2010 and into March of 2011, driven primarily by supply constraints and relative cheapness of the sector. Since then excess supply from Maiden Lane II and European Banks, rally of the forward curve and heightened macro-economic volatility have brought the sector down 30% from their March peak. Nomura Securities has recently published that the spread between the ABX 06-2 AAA and CDX HY 5yr is trading at 1 year wides – the relative value is further highlighted by the fact that spreads in the ABX are loss adjusted. In light of the opportunities provided by this sector, I have provided an overview of the valuation process of non-agency bonds below.

Valuation Overview

1) Identifying Characteristics of the Bond 
  • Collateral type- borrower type (SP, AA, OA, Jumbo), Fixed vs ARM, loan count, geographic concentration
  • Structure- seniority, credit enhancement (OC/XS), sequential vs pro rata, delinquency and loss triggers
This information is readily available on both Intex and Bloomberg. The classification is useful as different collateral types/seniorities will result in drastically different cashflow projections and discount rates.

2) Projecting Cash Flows 

The 3 primary drivers of non-agency RMBS cashflows are CPR (voluntary prepayments), CDR (defaults) and SEV (severities). These assumptions take the form of time denominated vectors and should reflect both the current collateral performance and the investor's macroeconomic views. Close to the entire universe of non-agency RMBS waterfalls is modeled on Intex. Intex can take in the user's assumed cpr/cdr/sev vectors and output the expected cashflow profile of the bond (which will vary depending on its position in the deal's capital structure). Thus, producing an accurate cash flow profile is dependent on the investor's cpr/cdr/sev assumptions. (Dealers will often aid investors in running the bonds based on their assumptions of the collateral.)

In general, non-agency collateral types have exhibited the following lifetime cpr/cdr/sev:


While these numbers can be used as general guidance, cashflows must be projected at the deal level. Below are some more general points on cpr/cdr/sev to use for further guidance:

  • Historic 1-month, 3-month, 6-month, and 1-year cpr/cdr/sev data for a deal can be found on Bloomberg by typing “SEV”. This can also be found on Intex by expanding the history in the collateral section. In absence of special situations that can cause drastic spikes in immediate prepay, default or severities (ie. servicer takeover or resolution of servicing lawsuits), the most recent 1-month or 3-month numbers can be used to proxy for the deals’ near term cpr/cdr/sev’s.
  • Longer term prepayments will generally experience burnout and trend downwards as borrowers with the means to prepay will have already done so during 2010-2011, when refi rates hit historic lows.
  • Medium term default levels will likely exceed those of short term as cdr’s have been relatively muted in 2011 due to servicer lawsuits from improper foreclosures and modification initiatives. Cdr’s are expected to pick up again in the upcoming year as the foreclosure moratorium continues to lift and as servicers run out of loan mod options. Long term defaults will likely also exhibit a slight drop as the collateral pool will also experience credit burnout. Note that differences in cdr curves can be seen across servicers (Countrywide serviced bonds have muted cdrs while Carrington bonds have recently been liquidating above 20 cdr)
  • 2 main ways to project long term default:
    • Collateral multiplier: Look through Bloomberg, Intex or LP’s collateral stratifications and identifying the % size of the 1 year current and 60+ (inclusive of BK, Foreclosure and REO) buckets. One can then run various default vectors on intex until they reach remaining liquidations to around 1.5-2x size of 60+ delinquency bucket and/or 100 minus 1-1.25x size of the 1-year current bucket. 
    • PD by Collateral Bucket: Assign a probability of default to every permutation of CLTV, FICO, and Loan Balance bucket within the remaining collateral and take a sumproduct of the probability with the relative size of each bucket to arrive at the remaining liquidations number.
  • Given a large portion of cashflows for distressed non-agency sectors come from liquidation proceeds, projecting severities is perhaps the most important part of RMBS investing/trading. Base adjustments to cohort level severities are made based on loan size, where smaller loans generally receive lower recoveries,  geographic distribution of loans, where higher concentrations of loans in judicial states imply longer liquidation timelines and higher severities, general timeline for liquidations of delinquent loans and HPA adjusted for LTV. In general, dealers are expecting a slight increase in medium term severities due to 2011’s extended liquidation timeline. Long term severity will likely decrease below current levels due to generally better collateral composition (LTV of current collateral is low from a historical perspective), increased concentrations of modified loans (modified loans exhibit 7-8% lower severities) and will also incorporate the investor’s view on hpa (higher hpa implies lower severity).
  • Additional considerations: For subprime front pay bonds, investors pay very close attention to servicing trends. Recent consolidation of the servicing industry (acquisition of Litton, HomeEq and Saxon by Ocwen) has resulted in significant servicer recaptures and disruptions to cashflows to front pay subprime bonds (Ocwen is known to be one of the most aggressive servicers in terms of recaps). One can sanity check payments streams for these bonds by looking at projected interest versus actual interest received in recent periods.
3) Discounting/Valuation

Finally, one must discount the obtained cashflows with an appropriate risk adjusted yield/oas to arrive at an intrinsic price (yields are most commonly used). These yields generally vary with macroeconomic fluctuations, housing price uncertainty and housing policy developments.  Given the uncertainty of the collaterals’ cashflows, investors/dealers will generally run yields of non-agencies at different cashflow scenarios (stress/optimistic) and arrive at a narrow range of potential prices.

From recent BWIC’s, the range of non-agency collateral types have exhibited the yields presented below:


4 comments:

  1. tinyinvestor1/04/2012

    this is amazing, keep it up. very interesting stuff

    ReplyDelete
  2. Anonymous1/05/2012

    I wonder if you have created a xls worksheet (or direct us to one) which replicates the Index assumption to valuation screen.

    ReplyDelete
  3. Wow very accurate details. This is a great stuff. Thanks!

    ReplyDelete
  4. fjm2221/09/2012

    Can anyone explain how the HARP proposals have affected ginnie, fannie and freddie pre-pmt rates? Gunlach was interviewed saying thet Ginnie are yielding 5% and have explicit gov backing. Why is the spread so large?

    ReplyDelete