The Pay-Option ARM sector continues to be at the intersection of all the troubling trends in residential mortgage credit:  rising total delinquencies; a substantial backlog of troubled loans without resolution; severely underwater borrowers; concentrated geographic exposure to boom-and-bust areas.  Risk / reward in this sector going forward will largely be driven by three factors:  (1) when and how the backlog of troubled loans is cleaned up; and (2) whether new delinquencies deteriorate further (from strategic defaults and unemployment) or improve (from credit burnout); and (3) whether loss severities can remain stable despite the enormous number of homes facing liquidation and the troubling age of delinquencies.

While it’s difficult to make blanket relative value statements about the sector, some points are worth stressing based on whether a shelf has a larger or smaller backlog of delinquent loans.  For big-backlog shelves like CWALT and BSMF, many mezz bonds look very attractive to the extent that it will take awhile for a “foreclosure cleanup” to take place.  However, we’d cautious about bonds for which delinquencies are highly aged and returns are vulnerable to a pickup in severities.  In more limited-backlog shelves such as WAMU, the sector has become a more straightforward bet on loss severities – implying that it will pay to keep close tabs on market LTVs.  In the cleaner shelves, we also find that some mezz bonds are priced to very aggressive assumptions, delivering relatively weak yields if severities rise modestly.

Recent Performance Comparison:  Delinquencies

Option ARM credit performance has been varying across shelves in multiple dimensions (pg 3).  Most obvious is the propensity to go seriously delinquent, with the worst recent performers being CWALT and BSMF and the best, by far, WAMU.  The second critical dimension is servicer processing / resolution of troubled loans.  Some shelves show a high degree of aggressiveness in clearing out problem loans (example: DSLA); some have the backlog pile up in foreclosure (WMALT); others have seen the backlog stay in delinquency, without moving loans into foreclosure (BSMF).

New delinquencies in the sector remain well off the peak levels of late-2008 / early-2009 (pg 4).  However, as with most Alt-A sectors, new delinquencies took a troubling turn higher at the end of last year.  While the recession has obviously been a problem, the #1 concern over the next few years is whether deep negative equity provokes a significant wave of strategic defaults.

Anatomy of a Backlog

On the whole, the sector experienced a remarkable slowdown in the pace at which loans move into foreclosure (pg 5-6).  A comparison across shelves reveals a clear case of “Have-Done’s” and “Haven’t-Done’s”:  In quickest 3 shelves about one-quarter of D90s move into foreclosure each month; in the slowest 4, less than 10% are moving to the next stage.  One interesting trend in later stage transition data is that, although there are disparities across shelves in moving loans out of foreclosure and out of REO, the later the stage, the less variation we’re seeing across shelves in processing efficiency.  And the slowest shelves at processing 90+ day delinquencies have been the fastest at liquidating loans out of REO.  This may be a reflection of the extent to which different organizations have ramped up different portions of delinquent processing at different tempos, and are therefore finding capacity constraints at different stages of the process.

Once and Future CDRs

In our August look at Pay-Option ARMs, we observed that the rising backlog of delinquencies meant that CDRs had to rise substantially from the 7-9% CDR pace then being experienced.  We foresaw increases to 20-40 CDR over time, depending on how aggressive servicers became at moving bad loans through the system.  The roughly stable CDRs seen in 2009 H2 reflect a combination of some shelves increasing while others, despite huge delinquency backlogs, are printing flat to falling CDRs (pg 7-8).

Receivables Aging

As a result of the slowdown in problem loan processing, the average securitized Pay-Option loan in the 90+ bucket is 9 payments behind (i.e., a “270”).  The average loan in foreclosure in this sector is 14 payments behind (pg 9-10).  Time increases severities in two ways:  By increasing the costs directly related to time (interest advances, taxes); and by challenging servicers’ ability to get quality sales off of deteriorating properties.  So the ever-lengthening age of delinquencies is putting the recent stabilization of severities at risk.  And the differences across shelves in processing loans translate into substantial differences in the length of time that delinquent loans have been non-current.

Current-to-Delinquent Trends:  Who’s Falling Off the Wagon?

To get a better sense of who’s turned delinquent in recent months, we’ve focused on the subset of borrowers who were current throughout 2009 Q3, but who missed at least one payment in 2009 Q4 (pg 13-14).  Sector-wide, approx 13% of borrowers who kept current in Q3 missed a payment in Q4.

We found three variables to be most powerful in explaining C-to-D incidence.  First – and most surprising – was loan factor.  Even relatively small gradations in cumulative neg am had a major impact; for example, borrowers with 1.08-1.10 factors were 29% more likely to go delinquent in Q4 than borrowers with 1.05-1.08 factors.  We think strategic default might play a role in this, as borrowers may be more likely to walk away from their loans if those loans have gotten significantly larger.  The two other variables with strong predictive power for “C-to-D” trends are Market LTV and original FICO, which implies that it will pay to closely examine the market LTV distribution of still-current loans.

Projected Defaults and Foreclosure Cleanup

In this report we project CDRs for different cohorts, with scenarios primarily differing in what we call Foreclosure Cleanup variables (pg 15-17).  The comparisons across shelves helped illustrate the difference between “Have-Done’s” and “Haven’t-Done’s”.  The “Have-Done’s” such as RALI – where loans have been moved through the system more aggressively – show relatively little sensitivity to foreclosure Cleanup scenarios.  In contrast, the CWALT shelf shows tremendous sensitivity to Foreclosure Cleanup parameters, which can send defaults anywhere from 15 to 38 CDR.  By our view, the RALIs represent a higher-quality cash flow than the CWALTs.  For one thing, there’s much less sensitivity to Foreclosure Cleanup, which is the biggest question mark surrounding non-agency RMBS right now.  Also, RALIs have less aged receivables, on average (pg 10), which probably makes their loss severities more predictable.  The WAMUs and WMALTs are somewhere in between – limited sensitivity to soft foreclosure cleanup, greater exposure to a more aggressive cleanup.

Yield Analysis:  Exposure to Foreclosure Cleanup and Severities

We analyzed yield sensitivities to credit performance scenarios and loss severities for two sample Pay-Option ARM RMBS deals from 2006:  A weaker-performing CWALT deal (69% total delinquencies) and a relatively strong-performing WAMU issue (41% total delinquencies).  Both used super-senior credit enhancement structures (60-25-15 for the CWALT, 60-20-20 for the WAMU).

Some risk / reward observations from this analysis:  (1) The CWALT bonds, which are marked by poor collateral performance and an enormous backlog of distressed loans, are highly sensitive to the form of “foreclosure cleanup” assumption.  And of course, the more aggressive the cleanup assumption (see: “Soft” vs “Hard” cleanup), the greater the sensitivity to severity assumption.  (2) For the CWALT, the senior and junior mezz are highly sensitive to the timing of cleanup, improving with a delayed start to the cleanup in all but the lowest severities (reflecting the term “credit IOs”).  (3) None of the CWALTs is especially sensitive to deterioration in the pace at which still-loans become delinquent.  What stands out to us is that the CWALT mezz bonds look very attractive to the extent that the foreclosure cleanup process doesn’t start ramping up CDRs immediately.

In the WAMU shelf, the super senior tranche of the example structure we analyzed showed relatively limited yield sensitivity to foreclosure cleanup and incremental deterioration of early stage delinquencies.  And while yields are slightly anemic if loss severity rises to the 55-60 range, the bond can handle a sizable increase in severity and still deliver decent yield.  The mezz WAMU tranches are another story.  There, the problem is pricing:  some of these mezz bonds are priced to very aggressive assumptions, delivering relatively weak yields if severities rise modestly.

Option ARM Delinquencies Still Climbing
 
The Pay-Option ARM sector continues to be at the intersection of all the troubling trends in residential mortgage credit:  rising total delinquencies; a substantial backlog of troubled loans without resolution; severely underwater borrowers; concentrated geographic exposure to boom-and-bust areas that saw overbuilding in the boom and heavy foreclosure volume in the bust.

Chart 1 below shows the composition of total delinquencies for the securitized Option ARM study group used in this report.  The group is composed of deals across 9 major shelves totaling $112 billion current face (just over a quarter-million loans).

From an investor’s perspective, 3 aspects of future credit performance will drive bond returns going forward:  (1) loss severities; (2) when and how the backlog of troubled loans is cleaned up; and (3) whether new delinquencies deteriorate further (from strategic defaults and unemployment) or improve (from credit burnout).

Cross-Shelf Performance Comparison

Option ARM credit performance has been varying across shelves in multiple dimensions.  Most obvious is the propensity to go seriously delinquent, with the worst recent performers being CWALT and BSMF and the best, by far, WAMU.  The second critical dimension of credit performance is servicer processing / resolution of troubled loans.  Some shelves show a high degree of aggressiveness in clearing out problem loans (example: DSLA); some have the backlog pile up in foreclosure (WMALT); others have let the backlog stay in delinquencies, without moving the loans into foreclosure.

 Early Stage Delinquencies:  Still a Concern

Charts 3 and 4 below track “current-to-delinquent” rolls (C-to-D) for the Option ARM sector; this is the % the “current” bucket that goes delinquent each month.  The top chart tracks average performance for the sector over time, while the bottom shows most recent 3-month historical average C-to-D across shelves.

New delinquencies in the sector remain well off the peak levels of late-2008 / early-2009.  However, as with most Alt-A sectors, new delinquencies took a troubling turn higher at the end of last year.  This could simply be normal seasonal performance – or it could mark the beginning of another difficult cycle for Option ARM loans.  While the recession has obviously been a problem, the #1 concern over the next few years is whether deep negative equity provokes a significant wave of strategic defaults.

In terms of the level of C-D roll rates, CWALT and BSMF are showing the worst recent performance, though HVMLT, RALI, SAMI and WMALT are clustered closely behind.  However, in terms of the trend change over the second half of 2009, HVMLT and RALI both saw C-D roll rates increase by 15% from Jul to Dec; CWALT, BSMF and WAMU saw increases of 14-15%; DLSA and WMALT saw +9%; AHMA was  –2%.

Anatomy of a Backlog:  Transitions from Seriously Delinquent

Chart 5 below tracks monthly average roll rates out of 90+ day delinquency for the non-agency Option ARM RMBS sector.  On the whole, the sector experienced a remarkable slowdown in the pace at which loans move into foreclosure.  For much of 2008, 35-45% of 90+ loans went into foreclosure, BK or REO each month; by late 2009 that had fallenen to less than 15%.  Theories abound as to the reason for the slowdown, including government intervention (loan mods), a choice by some servicers to avoid moving the loans into foreclosure, and the fact that servicers have been struggling to scale-up their operations to handle the unprecedented volumes of troubled loans.

We lean towards the latter explanation – that it’s proven very difficult for some servicers to deal with the volume.  If that’s true, it implies that there’s no “on” switch for the servicers to hit, so that clearing of the backlog will take some time, probably 2-3 years.

Chart 6 compares the latest 6-month historical average roll rates out of D90+ across shelves.  It’s a clear case of “Have-Done’s” and “Haven’t-Done’s”:  In quickest 3 shelves about one-quarter of D90s move into foreclosure each month; in the slowest 4, less than 10% are moving to the next stage.

Anatomy of a Backlog:  Transitions from Foreclosure

Looking at average roll rates out of foreclosure, we’ve seen a similar trend over the past 18 months:  down.  In 2009 Q4, loans in foreclosure were moving into REO or liquidation at about half of the average pace of 2008.

One interesting aspect of the later stage delinquency transition data is that, although there are disparities across shelves in moving loans out of foreclosure and out of REO, the later the stage, the less variation we’re seeing across shelves in processing efficiency.  For loans in 90+ day delinquency (Chart 6), the fastest three shelves are seeing loans move into foreclosure at more than 4 times the rate of the slowest three shelves.  For loans in foreclosure (Chart 8 below), the fastest shelves are moving loans into REO at about twice the speed of the slowest shelves.

And for loans in REO (not shown) – where transitions into liquidation have gotten faster, rather than slower – the fastest shelves are liquidating at just 1.5 times the pace of the slowest.  Perhaps even odder, the slowest shelves at processing 90+ day delinquencies are the fastest at liquidating loans out of REO.    This last finding may simply reflect the extent to which different organizations’ have ramped up different portions of delinquent processing at different tempos, and are therefore finding capacity constraints at different stages of the process.

Trends:  CDRs and Severities

Chart 9 below tracks sector average historical CDRs over the past two years; Chart 10 tracks average periodic loss severities.  In our August look at Countrywide Pay-Option ARMs, we observed that the rising backlog of delinquencies meant that CDRs had to rise substantially from the 7-9% CDR pace then being experienced.  We foresaw increases to 20-40 CDR over time, depending on how aggressive servicers became at moving bad loans through the system (the sector as a whole averaged 12 CDR in 2009 Q4, about 50% higher than 2008 Q4).  The roughly stable CDRs for the sector in 2009 H2 reflect a combination of some shelves increasing while others, despite huge delinquent loan backups, are printing flat to falling CDRs.

Loss severities finally stopped climbing this summer, and were broadly stable in 2009 H2.  This stabilization accompanied widespread improvement in home price and activity indicators, so we’re inclined to accept it as “real”.  Looking ahead, we remain concerned that the sheer volume of problem properties, combined with the lengthening of times-to-resolution for bad loans, could cause severities to rise again in 2010-2011.

Once and Future CDRs

As noted, some shelves have already begun the inexorable march to 20+ CDR.  Other shelves aren’t moving loans through to REO yet, and so still aren’t getting above 8-10 CDR.

Chart 12 below gives an interesting quantification as to how future CDRs will relate to historical CDRs.  The measure is a ratio of the latest 90+ day delinquencies or worse (including loans in Fcl and REO) to 6-month historical CDRs.  Of course, the lowest ratios are borne by the shelves that have already breached 20 CDR, DSLA and RALI; while those CDRs are likely to continue to rise, it won’t be by a multiple of 3 or more.  At the other end of the spectrum are CWALT, HVMLT and SAMI, where the “D90+ to CDR” ratios are 4.2-5.7.  Each of these shelves has been coming in below 10 CDR.  And as discussed later in this report, each faces extraordinary increases in CDRs going forward.

Receivables Aging

The charts below (13 and 14) attest to the following indelicate arithmetic:  (1) foreclosure and liquidation rates are slowing; and (2) delinquent borrowers aren’t making payments.  Of course, we’re seeing longer and longer times to resolution in this sector.  As a result, the average securitized Pay-Option loan in the 90+ bucket is 9 payments behind (i.e., a “270”).  The average loan in foreclosure in this sector is 14 payments behind.

To us, the ever-lengthening time that borrowers are behind put the recent stabilization of severities at risk (see following page).  Time increases severities in two ways:  By increasing the costs directly related to time (interest advances, taxes); and by challenging servicers’ ability to get quality sales off of deteriorating properties.

Receivables Aging:  Shelf Comparison

Not surprisingly, we’re seeing a fairly wide disparity in the aging of delinquent loans across shelves, with the slowest-to-process shelves having the oldest delinquencies.  CWALT stands out in this regard; in that shelf, about 40% of the D90+ bucket has missed 12 or more payments; over half of the Foreclosure bucket has missed 18 or more payments.

The key question for investors:  Will loss severities on these very, very delinquent loans turn out to be higher or lower than what we’ve experienced in 2009 H2?

Market LTV and Recent Historical Severities

Our “Market LTV” estimation model takes into account cumulative negative or positive amortization on the loans, original LTV and the estimated change in property value since origination, using CBSA- and CBSA-Division-level home price indices.  Chart 17 compares 6-month historical average loss severities across shelves, while Chart 18 compares our estimates of average market LTVs among D60+ loans.  In both charts, the data is expressed as “difference from the sector average”.

What stands out to us about these charts is the lack of variation across shelves in both market LTVs and recent loss severities (with the exception of the WAMU shelf, which is an over-achiever within the sector).  To a large extent, the close clustering of market LTVs reflects the timing and location of the Option ARM mortgage universe; 77% of current face outstanding in this group of MBS is backed by loans from one of the four “sand states” (CA, FL, AZ and NV).  The lack of variation in severities is telling us that market LTV has so far been the main driver of losses and recoveries.  Differences in receivable aging haven’t been a major factor so far.  It’s worth noting that there’s been less cross-shelf variation in REO aging than there has been in foreclosure aging.  However, those difference are likely to re-emerge once firms start clearing out the huge D90+ buckets.

 The State of Neg Am

The good news for the Option ARM sector is that, with 1-year CMT and 12-month MTA both below 0.50%, almost none of the outstanding borrowers who are actually making their required payments are currently generating neg am.  In fact, most borrowers stopped generating neg am in late-2008 / early-2009.  This has helped in two ways:  Most obviously, by reducing neg am’s impact on market LTV and loss severity; more subtly, by delaying the move to a 115% factor, which triggers a severe “payment recast” event for most of these loans.

With the exception of WAMU, we think variations in cumulative neg am across shelves by and large reflect different mixes of loan ages, rather than systematic differences in credit quality or performance.

Current-to-Delinquent Trends:  Who’s Falling Off the Wagon?

In this section, we’ve attempted to develop a profile of those borrowers turning delinquent in recent months.  To do this, we’ve focused on the subset of borrowers who were current throughout 2009 Q3, but who missed at least one payment in 2009 Q4.  Sector-wide, approx 13% of borrowers who kept current in Q3 missed a payment in Q4.  (Of those who missed a payment in Q4, less than 15% were back to current by year-end.)  Variations across shelves by and large weren’t substantial (again with the exception of WAMU).

The following charts (21-24) break out propensity to move from Current to Delinquent by key loan attributes.  We found three variables to be most powerful in explaining C-to-D incidence.  First – and most surprising – was loan factor.  Even relatively small gradations in cumulative neg am had a major impact; for example, borrowers with 1.08-1.10 factors were 29% more likely to go delinquent in Q4 than borrowers with 1.05-1.08 factors.

Why is factor such a strong variable?  Three explanations come to mind: (1) Factor is almost surely correlated with gross margin, which was itself correlated by more traditional credit variables and which accounts for the some of the pace of early neg am accumulation.  (2) Lower factors likely identify borrowers who made at least some greater-than-minimum payments in the early years, a sign of stronger resources overall.  And: (3) the strategic default angle:  it may be that borrowers are more likely to walk away from their loans if those loans have gotten significantly larger.
In other words, it’s not just that you’re home’s gone down in value from $650 K to $500 K– it’s also knowing that you’re loan’s gone from $500 K to $550 K.  Ironically, if this explanation holds true, it’s positive for 2007 production, which didn’t have as much time to accumulate serious neg am before rates started falling.

Charts 23 and 24 highlight two other strong variables in predicting Current-to-Delinquent trends:  Market LTV and original FICO.  It’s fairly obvious that market LTV’s going to be significant in predicting default incidence, including strategic default.  What’s interesting to us is that the variable holds even with a CBSA-/CBSA-Division-level HPI model, for a population of loans that are very concentrated geographically.  To the extent that this behavior continues, it implies (1) that there’s a continued risk of rising strategic defaults; and (2) that it will pay to closely examine the market LTV distribution of still-current loans.

Original FICO also continues to have some explanatory power, especially at the upper reaches of the FICO spectrum.  Interestingly, original LTV was a spotty variable, with strong predictive capacity in some shelves and very little in others.

Projected Defaults:  The Importance of Foreclosure Cleanup

In the following section, we project CDRs for different cohorts with scenarios differing solely based on what we call Foreclosure Cleanup variables.  In all cases, our projections start with the cohorts’ delinquencies as they stood in most recent available data.  In the “Continue H2 Rolls” scenario, the collateral was assumed to continue run at the average transition roll experience of 2009 H2 (Current-to-delinquent, 30-to-60, etc.), though with 25% slower voluntary prepayments.  Over the long run, “Continue H2” is probably the least likely scenario, because it literally implies that the backlog will build up for many years.

Other scenarios increase late stage rolls to clean up the backlog of delinquent loans.  For example, in the “10% D90+ to Fcl / 5% Fcl to REO”, we assume that 10% of D90s go into foreclosure each month, and that 5% of loans in foreclosure go to REO.

The first comparison in Charts 25 and 26 – CWALT 2006 vs RALI 2005-2007 – is instructive to illustrate the difference between “Have-Done’s” and “Haven’t-Done’s” (see also pg 5-6).  The “Have-Done’s” such as RALI – where loans have been moved through the system more aggressively – show relatively little sensitivity to foreclosure Cleanup scenarios.  In contrast, the CWALT shelf shows tremendous sensitivity to Foreclosure Cleanup parameters, which send defaults anywhere from 15 to 38 CDR.

The projections below give a comparison between a relatively clean shelf, WAMU 2006 and the weaker WMALT 2006-07.  Even the WAMUs shows a significant increase in CDRs over time, up to the 18-22 area.  They show limited sensitivity to a “softer” foreclosure cleanup (10% 90+ to Fcl / 5% Fcl to REO), but greater sensitivity to a more aggressive cleanup scenario (20% / 10%).

Some important themes emerge from these projections.  By our view, the RALIs represent a higher-quality cash flow than the CWALTs.  For one thing, there’s much less sensitivity to Foreclosure Cleanup, which is the biggest question mark surrounding non-agency RMBS right now.  Also, the RALIs have younger receivables, on average (pg 10), which probably makes their loss severities more predictable.

The WAMUs and WMALTs are somewhere in between – limited sensitivity to soft foreclosure cleanup, greater exposure to a more aggressive cleanup.

Sensitivity to Early Stage Delinquency Trends

In this chart (Chart 29, below) we illustrate CDR sensitivity to the interaction of two key variables:  Early stage delinquencies and foreclosure cleanup.  For this analysis, we alter the “10%/5%” and “20%/10%” cleanup scenarios by moving early delinquencies – specifically, the Current-to-Delinquent monthly roll rate – up and down 25%.

In the case of the CWALT Option ARM shelf, the early delinquency trend variable is important – but it’s still dwarfed by the potential impact of foreclosure cleanup.  In fact, the overall strength of this variable is limited by the fact that so few loans in the sector are still current.

In our view, the variables stressed in these three pages – foreclosure cleanup and early delinquencies – are two of the three keys to the Option ARM valuation puzzle going forward.  The other key:  Loss severities.  The analysis on the following pages illustrates the interaction of these variables with deal structure and current pricing.

Yield Analysis:  Exposure to Foreclosure Cleanup and Severities

In the following pages we analyze yield sensitivities to credit performance scenarios and loss severities for two sample Pay-Option ARM RMBS deals from 2006:  A weaker-performing CWALT deal (69% total delinquencies) and a relatively strong-performing WAMU issue (41% total delinquencies).  Both used super-senior credit enhancement structures (60-25-15 for the CWALT, 60-20-20 for the WAMU).

The starting point of our credit projections was a collateral group’s average historical transition roll performance for 2009, applied to current delinquencies.  From there we created 4 credit scenarios.  The “Soft Cleanup” scenarios shift monthly “90+ to foreclosure” to a minimum of 10%, and “foreclosure to REO” to a minimum of 5%; “Hard Cleanup” shifts those percentages to 20% and 10%.  “Deteriorate 30%” starts with the “Soft Cleanup / Immediate Start” scenario and increases the monthly current-to-delinquent rolls by 30%.

Some risk / reward observations from this analysis:  (1) The CWALT bonds, which are marked by poor collateral performance and an enormous backlog of distressed loans, are highly sensitive to the form of “foreclosure cleanup” assumption.  And of course, the more aggressive the cleanup assumption (see: “Soft” vs “Hard” cleanup), the greater the sensitivity to severity assumption.

For the CWALT, the senior and junior mezz are highly sensitive to the timing of cleanup, improving with a delayed start to the cleanup in all but the lowest severities (reflecting the term “credit IOs”).  (3) None of the CWALTs is especially sensitive to the “deterioration” scenario, which affects the pace at which still-current loans become delinquent.  With just 31% of the collateral group’s loans current, the sad truth is that the disposition of current loans isn’t very important right now; it’s all about the timing of foreclosures and loss severities.  (4) In assessing the tradeoff between speed and strength of foreclosure cleanup and severities, we think you could make an argument for a kind of “positive convexity” in the curves, in that stronger housing could see a more aggressive cleanup and lower severities, while a weaker housing market with higher severities could witness a slower cleanup.

What stands out to us is that the CWALT mezz bonds look very attractive to the extent that the foreclosure cleanup process doesn’t start ramping up CDRs immediately.

 WAMU Pay-Option ARM Yield Analysis

The super senior tranche of the example structure we analyzed is a fairly high quality cash flow for this sector.  Yield sensitivity to foreclosure cleanup and incremental deterioration of early stage delinquencies is relatively limited.  And while yields are slightly anemic if loss severity rises to the 55-60 range, the bond can handle a sizable increase in severity holds and still deliver decent yield.    It will pay to keep close track of market LTVs in this sector, however, as a warning sign for rising severities over time.

The mezz WAMU tranches are another story.  Sensitivity to foreclosure cleanup scenarios is still reasonable, though it’s markedly higher than for the super-seniors; the bonds also have greater exposure to deterioration of early stage delinquencies.  The real problem is pricing:  some of these mezz bonds are priced to very aggressive assumptions, delivering relatively weak yields if severities rise modestly.