This was written by someone really smart who manages several billions of dollars.
The RMBS market has been, in our view, the most disrupted and systematically cheap corner of the fixed income universe for the past few years. The buyer base transformation (from ratings-based holders to credit specialists) taking place in the massive (>$1 trillion) Non-Agency market has resulted in securities priced to quite adverse outcomes that appear to be at odds with observable housing market trends over the longer term. High returns (to long average lives) are achievable without improvement in housing.
Since the spring of 2011, the distressed Non-Agency RMBS landscape has become an increasingly broken market where (dramatically lower) prices have become more and more detached from (steady to improving) fundamentals. We are regularly buying securities at 25-35% discounts to trading levels from earlier this year.
The state of disruption in the distressed RMBS market is relatively stark. It isn’t very often that the market presents us with a chance to buy senior debt securities with 20%+ return potential that offer:
- Very low principal risk (IRRs to maturity are slightly positive even in the event of catastrophic housing deterioration from current depressed levels). We believe these securities are priced to withstand more in the way of economic/housing deterioration than just about any other asset category in which we invest.
- Very high (~1,000-1,500+) loss-adjusted credit spreads to long average lives (spreads would be even higher if we used less conservative assumptions with regard to defaults, loss severities, prepayments, etc.)
- Relative value (loss-adjusted yields in distressed RMBS are substantially higher than nominal yields in other asset classes)
- Exposure to an asset class that is naturally (and significantly) improving in credit quality as the worst borrowers continue to default; as these pools become less distressed in credit quality, it is likely that they will also become less distressed in price as they begin to appeal to a broader base of investors and become priced to less severe scenarios and/or lower yields; another way of thinking about this is that credit improvement can occur without housing market improvement)
- Exposure to an asset class that has never endured a distressed cycle, and where traditional investors lack the credit background necessary to underwrite this risk. Much of the highest return profile collateral is found within the SubPrime and Pay Option ARM sectors. These are among the most complicated areas of the market, where credit expertise and security selection are critical, emerging variables such as put-back settlements and modifications take on amplified importance, and collateral is often only available in small size. Due in part to these factors, PPIP managers and mutual fund complexes have been less active in these areas.
- Positive leverage to US dollar debasement / inflation (which would tend to boost homeowner equity, slowing default rates and increasing recoveries)
- High current yields (the static yield on our RMBS portfolio is ~11%, which helps to dampen return volatility)
- Rapid return of cash (many securities, by virtue of seniority, amortize in excess of 20-30% per year; recouping our investment quickly allows us to reduce our dependence on a friendly back-end housing/economic environment)
- Free options on voter-friendly policy initiatives (principal modifications designed to keep people in their homes are increasing in frequency and effectiveness; we are buying securities that should benefit from the lower default, higher recovery, higher prepayment environment that could be expected to result from a scenario in which borrowers have higher home equity)
- Free options on litigation (mortgage put-back settlements). Our portfolio stands to benefit meaningfully if the Bank of America settlement unfolds as planned, and there is a very reasonable chance that this settlement will be revised upward and that other banks will be compelled to offer similar consideration.
In spring 2011, distressed Non-Agency RMBS prices began to fall precipitously. Importantly, the selloff occurred in the absence of unforeseen deterioration in housing fundamentals. Rather, the initial downdraft arose from supply fatigue attributable to the Federal Reserve’s disorderly liquidation of its $31 billion Maiden Lane II portfolio (which consists mostly of SubPrime, Alt-A and Pay Option ARMs) and well-telegraphed additional supply from other players (such as Dexia’s >$7 billion portfolio). Many market participants have relatively short term investment horizons, and were disinclined to buy or hold assets with short term mark-to-market risk when they knew additional supply of those assets was due to hit the market in the near future.
Subsequent to this flood of supply, the severe risk-off climate in August/September weighed further on prices (especially as general de-risking and uncertainty around capital requirements resulted in reduced participation from Wall Street investment banks).
Like most markets, certain parts of the mortgage arena are overbought while others are either under appreciated or thinly sponsored. However, on the whole we believe certain portions of this asset class are poised over the next couple of years to generate higher returns with less risk than virtually any other sector in which we operate. The asymmetry embedded in these assets at current prices (positive yields to maturity even in a severe left tail scenario, with quite high total return potential in the event of reflation or even a perpetuation of the status quo) stands out starkly versus the rest of the RMBS market as well as other asset classes. In this uncertain macro environment, we believe that assets with these kinds of return profiles (where left tails can be cut off but right tail options are substantial) will be increasingly coveted. Scarcity value should also not be ignored – long-dated (5-10+ years), high loss-adjusted spread (>1,000 bps) debt with positive leverage to inflation is hard to find at the moment outside of distressed Non-Agencies (over time, pension funds and insurance companies struggling to meet liabilities should find this asset class difficult to ignore).
The mortgage arena is one in which we have generated excellent returns without leverage across vastly different environments for the housing market and broader economy. Our first foray into the space was in 2005, when we began buying protection on single-name SubPrime ABS. Our short positions peaked at just under $1 billion notional, on which we ended up making north of $700M in profits. Soon after, one of our larger investors requested an RMBS-only separate account, which began investing in January 2008. This account was up (net of fees and without leverage) +17.7% in 2008, +82.8% in 2009, and +20.4% in 2010, and it is up +0.9% year-to-date through November 25th. Our allocation to RMBS within our multi-strategy funds has also grown significantly. At present, we manage about $4.3 billion (market value) in mortgage securities across all funds, about $3.9 billion of which is in Non-Agencies. The recent selloff (again, a function mostly of near-term technicals rather than fundamentals) has created an attractive opportunity to add incremental exposure.
We have been deliberately targeting some of the most distressed mortgage pools that were originated at the peak of the housing market (2005-2007) and the trough of lending standards. The reason we are focused on these poor credit quality pools is that they require a great deal of credit expertise, and most of the mortgage investor base does not employ a credit-intensive approach. Before 2008, the mortgage space was dominated by ratings-based buyers (insurance companies, banks, pensions etc. that needed to buy AAA assets). Credit rating, not credit quality, was the primary criteria for investment. The process of ratings downgrades a few years ago is one of the primary reasons this opportunity exists – the major holders of this debt became forced sellers in droves once it was downgraded. Moreover, they were largely sellers into a vacuum because there was not a well-established distressed mortgage investor base (this is the first nationwide distressed cycle in the mortgage market).
A credit-intensive approach to the space leads us to believe that current prices do not reflect the substantial returns likely to be reaped from these distressed mortgage pools. This is largely a function of the fact that the Non-Agency asset class is huge (over $1 trillion), and the marginal movers/determiners of price tend to be the largest participants (Pimco, Blackrock, TCW, etc.), members of the long-only community that, due to the vast sums of capital they have to deploy, are compelled to take a somewhat macro, less credit-intensive approach to mortgage security analysis. If these large asset managers tore apart every $5M security for sale and stress-tested it for several hundred default, recovery, prepay and modification scenarios (as we do), they would find it difficult to put their money to work). To some degree they need to buy and sell in bulk and apply more blunt methods of security valuation. These dominant methods of mortgage security valuation tend to set prices, and (in our view) fail to appreciate various forms of optionality (on fundamentals, interest rates and policies) embedded in the assets as well as evolving dynamics with regard to default and recovery trajectories. This phenomenon, in conjunction with the more recent supply/demand dislocation in the space resulting from wide scale deleveraging, has resulted in a situation in which these assets are priced to adverse outcomes that seem to be inconsistent with observable housing market trends.
Our ability to perform in-depth analyses into the underlying mortgages helps us discern portfolio attributes (particularly credit quality dispersion) and trends that are generally missed by the bulk of investors. We have exploited numerous of these opportunities in the mortgage market (on both the long and short side) over the last 6 years which have arisen from the market’s tendency to extrapolate current trends into the future. Much of our success in the mortgage space, which has taken place across very different market environments, has rested upon our ability to identify inflection points where those trends break down, creating opportunities for high returns.
Refresher: How Do Defaults Impact Cash Flows in the Securitization?
As with most asset-backed structures, most RMBS have mechanisms whereby liabilities are written down as a trust’s assets become impaired. To illustrate this, let’s assume we purchase the Senior tranche of a $100M securitization consisting of 500 mortgages, and that there is 10% or ($10M) of subordination (tranches that are first to take losses) beneath us. The following will give you a sense of how the cash flow works when a borrower defaults:
- Let’s assume the borrower originally purchased the home for $250k in 2006 (near the peak of the market)
- On this purchase, he received a $200k mortgage (~80% original loan-to-value is typical in our Alt-A pools)
- His home, like everyone else’s, has dropped in value. Now it is worth $150k instead of $250k, so the mark-to-market loan-to-value on his $200k mortgage is 133%.
- He defaults, and the Mortgage Servicer for our securitization forecloses on the home and sells it in the market. Foreclosure/related costs total $70k.
- Recovery for the Trust = $150k value of home MINUS $70k Foreclosure/related costs = $80k
o In mortgage parlance, the term “loss severity” is used more often than “recovery.” Loss severity is merely 100% MINUS Recovery. In this case, the loss severity would be 60% (you recover $80k on a $200k mortgage).
- After this particular foreclosure, the senior tranche holders would receive about $80k on a $200k mortgage).
- After this particular foreclosure, the senior tranche holders would receive about $80k recovery net from the foreclosure process. The $120k loss on the $200k mortgage would be borne by the Junior Tranches, so pro forma for this transaction, there would be $9,880,000 of subordination remaining.
- Credit Trends
We are targeting securities with underlying mortgage pools that can experience significant credit improvement without housing market improvement, i.e., we are targeting pools that are improving in quality not necessarily because home prices are going up, but because the worst borrowers in the pool continue to default. Mortgage market participants have historically used backward-looking models to extrapolate future trends, but we see serious divergences between the borrowers who have defaulted (generally very poor) and the borrowers who will default in the future (in our portfolio, generally less poor). For instance:
- Mark-to-market Loan-to-Values (MTM LTVs) are declining: Delinquent borrowers (who we expect to default) are less out of the money on their mortgages than borrowers who have already defaulted. To give you a sense, recently we bought a SubPrime bond where the MTM LTV of borrowers who defaulted over the past 6 months was 162%, but the MTM LTV of borrowers who are delinquent is 125%.
o This is important because it should (all other things being equal) result in higher recoveries / lower loss severities. For instance, on our hypothetical $200k mortgage above, a 162% MTM LTV would imply a home value of $123k, while a 125% MTM LTV would imply a home value of $160k. This incremental $37k is quite powerful when you look at overall recoveries / loss severities (factoring in the foreclosure-related costs.) In our $200k mortgage example above with $70k in foreclosure-related costs, a 162% MTM LTV home would recover $53k (123-70) while a 125% MTM LTV home would recover $90k (160-70), or 70% more.
- Loan Balance Sizes are increasing: Delinquent borrowers tend to have larger loan balances than borrowers who have already defaulted. For instance, on the same SubPrime bond described above, the currently delinquent borrowers have a loan size $80k larger ($267k) than the borrowers who defaulted over the past 6 months ($187k).o This is important because it should (all other things being equal) result in higher recoveries / lower loss severities. This has to do with the fact that foreclosure costs are somewhat fixed => foreclosure costs will eat up a smaller percentage of a larger loan balance, leading to higher recoveries / lower severities. For instance, assuming like we did above $70k in foreclosure-related costs, a 125% MTM LTV home with a $187k mortgage would recover 43% (187 / 125% – 70 = $80k) while a 125% MTM LTV home with a $267k mortgage would recover 54% (267 / 125% – 70 = $144k), or 26% more.
- Larger Proportion of Borrowers with Positive Equity: as the worst borrowers default, we believe the pools are likely to improve in quality as higher quality borrowers become an increasingly large percentage of the pool. For instance, in the SubPrime bond described above, 39% of the collateral has positive equity (a MTM LTV of less than 100%), with a weighted average MTM LTV of 82% for this subset of borrowers, and 32% of the borrowers have never missed a mortgage payment (the pool is 5.5 years old).
- There are a number of other ways in which the pools appear to be changing in positive ways (less concentration in Florida as those borrowers default, etc.). We would be happy to discuss these trends in greater detail. The broader point is that the market is pricing these securities as if recoveries are going to fall / loss severities are going to increase, while we see patterns in the borrower base that suggest this may not be the case. Even if we see a perpetuation of the status quo in terms of recoveries/severities, these securities should benefit considerably.
Downside Protection / Ability to Withstand Punitive Outcomes
The average price of our purchases is in the high 30s. At these prices, we are creating the underlying homes at ~25-35% of original appraised value (depending on the security). This concept is illustrated on Page 10 of the accompanying presentation. Basically this means that if 100% of the borrowers defaulted tomorrow and we recovered 25-35%, we would break even. This is an unrealistic scenario, given that current recoveries are closer to 40% in our Alt-A pools and large portions of these pools appear unlikely to default, but it is a simple way to convey the ability of these securities to withstand onerous assumptions about housing market trajectories.
Our Base Case Assumptions
While our existing portfolio consists of a diverse array of securities, in general our “base case assumptions” (which we believe are conservative) on our Alt-A book could be summed up as the following:
- Roughly 40% of the borrowers in our pools are 60 days or more delinquent on mortgage payments. We assume they will all default.
- Of the remaining 60% of the borrowers who are still making payments, we assume about half of them default. So we are assuming total defaults of ~70%.
- Current recoveries upon default are 40% (loss severity of 60%). We assume that recoveries decline to 30-32% and remain there indefinitely, implying a steep decline in home prices from already depressed levels with no subsequent recovery. General consensus in the mortgage market is an additional 10-15% decline in home prices, and in general we build in extra room for deterioration in our base case.
- We assume that prepayments (which are a positive for these securities) remain at very low levels (0-2%) for the life of the security.
These are the same base case assumptions we have been using for the past 2 years (despite the fact that housing fundamentals have moderated considerably over that time frame). In general the securities we are buying are priced to deliver 12-20%+ IRRs to long average lives even in the event that these base case assumptions prove correct and housing fundamentals deteriorate considerably from here. However, we’re not buying these securities with the belief that we’ll earn 12-20%+ for 5-10+ years. We’re buying them because we believe these assumptions are likely to prove too conservative and that there is a decent chance spreads will compress to reflect this outlook, creating total return potential in the high teens or more over the next couple of years.
Margin of Safety
Another factor that we believe is likely to drive appetite for distressed RMBS is the unusual benevolence of the left tail scenario. Even in the event of catastrophic housing market deterioration, these securities are priced to deliver positive single-digit IRRs to much shorter average lives. While we are certainly not targeting single-digit IRRs, the fact that these securities can generate positive returns to maturity even under these kinds of market conditions significantly enhances their appeal. In these left tail deflationary scenarios, virtually all other risk asset classes (from equities to high yield to commodities to real estate and beyond) could reasonably be expected to endure substantial negative returns. (Of course, it is important to be
aware that IRRs to maturity are most relevant if an investor has the staying power to hold them to maturity; in the event of an actual left tail scenario of this kind, it is likely that distressed mortgage securities would trade down on a mark-to-market basis, then generate strong returns thereafter as the structural features became increasingly prominent, spurring rapid amortization and rendering the whims of the market less relevant.) Given the chorus of uncertainties weighing on the market at present, we believe that assets with these kinds of return profiles will be increasingly sought after.
This is not an offering of securities for sale in any jurisdiction. Any indication of interest from prospective investors in response to this letter involves no obligation or commitment of any kind.