With recent prices for typical senior private label residential mortgage backed securities (subprime, Alt-A, and prime jumbo) now 25%-40% higher than two months ago, it would be a stretch to call them an opportunity of a lifetime. They are not even the opportunity of the past two months. Nevertheless, even with the increases, the prices of many of these securities remain at levels that should generate annualized returns in the 20%-25%+ range so long as we do not see another much more severe leg to the recent recession. Class A-2D of GSAMP 2006-HE5 (one of the securities referenced in Markit's ABX 07-1 AAA index), for which Reuters has provided recent price quotes in the ballpark of 20 cents on the dollar, provides an example.
There are differences between GSAMP 2006-HE5 and other subprime pools, but the risk-reward trade-offs on these securities are more similar than different. Therefore, examining a single security from this pool provides useful insights into the return potential for other subprime securities, and to a lesser extent, Alt-A and prime jumbo securities.
As of August 2009, the outstanding principal balance of the mortgages in GSAMP 2006-HE5 was $482 million, 46.5% of the original principal balance when the deal was originated in 2006. The other 53.5% has been repaid or charged off. Of the remaining mortgages, $252 million, or 52% of the outstanding balance, is at least one payment past due, is in the process of foreclosure or bankruptcy, or has been foreclosed.
GSAMP 2006-HE5 is divided into two groups. Losses on either group can erode the subordinate M-1 through M-5 certificates (losses have already wiped out the M-6 and junior certificates). Principal and interest payments on the Group 2 certificates (including A-2D) depend on underlying principal and interest payments on the Group 2 loans.
Class A-2D is the “last-to-pay” Group 2 senior security. This means that class A-2D will receive principal before any of the subordinate certificates, but after the A-2B and A-2C certificates are paid in full, unless the subordinate certificates are wiped out. If the subordinate certificates are completely written down, principal payments will be made to the senior certificates (including A-2D) in proportion to their outstanding principal balances. This means that the cash flows on the remaining Group 2 senior certificates (including class A-2D) should mirror the cash flows on the remaining Group 2 loans outstanding at the time, after adjusting for administrative expenses and hedges. Delinquency rates for each group are currently within 1% of the overall average for both pools combined, so the future performance of Group 2 should be relatively similar to the overall pool.
The table below (a slightly modified version of a table from an August 25, 2009 Wells Fargo monthly report for this pool), provides a snapshot of the certificates, excluding classes X, B-1, B-2, M-9, M-8, M-7, and M-6. The $102 million original principal of these certificates has been wiped out by losses.
Class | Pass-Through Rate | Beginning Certificate Balance | Interest Distrib. | Principal Distribution | Current Realized Loss | Ending Certificate Balance |
A-1 | 0.425% | 91,808,763 | 31,432 | 1,583,203 | 0 | 90,225,560 |
A-2A | 0.325% | 0 | 0 | 0 | 0 | 0 |
A-2B | 0.385% | 91,206,448 | 28,287 | 4,877,032 | 0 | 86,329,416 |
A-2C | 0.435% | 115,439,000 | 40,452 | 0 | 0 | 115,439,000 |
A-2D | 0.525% | 48,975,000 | 20,712 | 0 | 0 | 48,975,000 |
M-1 | 0.585% | 41,491,000 | 19,553 | 0 | 0 | 41,491,000 |
M-2 | 0.595% | 38,379,000 | 18,395 | 0 | 0 | 38,379,000 |
M-3 | 0.605% | 22,820,000 | 11,122 | 0 | 0 | 22,820,000 |
M-4 | 0.655% | 19,190,000 | 10,125 | 0 | 0 | 19,190,000 |
M-5 | 0.685% | 19,190,000 | 10,589 | 0 | 5,297 | 19,184,703 |
M-6 | 0.765% | 4,391,550 | 2,706 | 0 | 4,391,550 | 0 |
Total | 492,890,762 | 193,373 | 6,460,235 | 4,396,847 | 482,033,679 |
Since origination, the pool has experienced $151 million of loan losses, 14.6% of the original principal balance of the pool, including $58 million since January 2009, or 10.2% of the outstanding principal balance as of January 2009. The $58 million of losses on the mortgages has resulted in $42 million of losses on the subordinated certificates. This $42 million has accounted for about 50% of the total principal reduction since January 2009. The $42 million of certificate losses is lower than the $58 million of mortgage loan losses because the average interest rate on the underlying mortgages is higher than the average interest rate on the securities, and this excess interest offsets losses.
click to enlarge
The chart below illustrates certificate losses expressed in dollars and as a percentage of monthly principal reductions since January. The difference between the purple and blue bars is the amount of total certificate principal repaid each month. Recent certificate interest payments (not shown in the graph) have averaged around 5% of principal payments.
The amount that investors ultimately recover depends on the future path of the yellow line in the chart above. If this line remains at levels similar to the 50% average since January, then investors will ultimately be repaid about 50 cents per dollar of principal. (Actually, the path of the yellow line will begin to diverge from investor cash flows if the subordinate certificates are wiped out. If the loans continue to perform as they have for the past six months, the overall performance of the certificates will be similar to the recent past, but class A-2D will receive significantly less than 50 cents on the dollar of principal. This is because the principal balance of the senior certificates cannot be written down, and interest will continue to accrue on the full principal balance on the notes. Therefore, investors will receive more interest but less principal than if the certificate principal were written down, but the net cash flows would be about the same.)
The yellow line peaked at almost 62% in March, and has decreased to just over 40% in August. On the surface, this is encouraging. However, the graph below shows that total delinquencies plus bankruptcies, foreclosures, and real estate owned have increased as a percentage of the outstanding principal balance over the past few months. This suggests that losses should grow in relation to principal and interest payments on the loans, unless foreclosures result in better net recoveries than they have so far in 2009.
It is illustrative to consider investor returns in three different scenarios assuming the class A-2D certificates are purchased at 20 cents on the dollar.
1. The relationship between loan losses and net cash flows on the certificates remains similar to the performance so far this year. In this case, the certificates pay interest at an effective rate of five times one month LIBOR plus 1.2% (interest accrues at LIBOR plus 0.24% of the face amount, but this is multiplied by five if the securities are purchased at one-fifth of their face amount). Principal is likely to begin being repaid in about three years when the principal of the subordinate certificates is written down to zero. The subsequent cash flows would be roughly equivalent to principal repayments at 50 cents on the dollar (250% of the original investment), plus interest at an average rate around LIBOR plus 2%. Overall, the investment should generate an average annualized return around 25%.
2. Loan performance improves enough that the class A-2D certificates are repaid in full. In this case, it will probably take about four or five years until any principal is repaid to class A-2D, after classes A-2B and A-2C are repaid in full. Nominal returns in this scenario would probably be about twice the previous scenario, but over a longer duration, so average annualized returns should also be in the rough ball park of 25%.
3. Loan performance worsens to the point where the mortgage loans experience average principal losses of 80%-85%. This would imply that almost all loans foreclose and recovery rates on foreclosed loans are low (less than 15-20 cents on the dollar). In this case, the investment should pay interest at five times one month LIBOR plus 1.2% for the next two or three years until losses eliminate the subordinate certificates. After this, investors would receive cash flows roughly equal to the original principal investment (20% of par).
The first of the three scenarios above would correspond to lifetime loan losses of 42%-45% of the original principal balance of this pool, roughly 15% higher than recent projections by Standard and Poors for similar pools. Loan losses for 2009 and 2010 would be around 30% of the January 2009 pool balance, higher than the 21%-28% range for the “More Adverse” scenario from the Fed’s Supervisory Capital Assessment Program (SCAP) stress tests. In addition, losses on legacy residential mortgages in 2011and 2012 would be similar to losses for 2009 and 2010. In this case, banks would probably require another round of large capital raises.
The second (most favorable) scenario would result in lifetime loan losses of around 32%, roughly 15% lower than Standard and Poors’ projections for similar pools. Loan losses for 2009 and 2010 would be around 25%, the middle of the “More Adverse” SCAP stress test scenario.
In the third (worst) scenario, lifetime loan losses would exceed 50%, much worse than Standard and Poors’ and SCAP stress test projections. This scenario is consistent with a very severe environment where most banks (including large banks like Citigroup (C), Bank of America (BAC), and Wells Fargo (WFC)) could only survive as independent entities via massive capital raises.
If you believe that an outcome similar to the first two scenarios is much more likely than the third, then class A-2D at 20 cents on the dollar represents an excellent value. Unfortunately, it is not possible for the average retail investor to purchase this type of security, and there are problems with the few publicly-traded funds that have launched recently to purchase distressed mortgage assets. Here are a couple of examples.
1. Invesco Mortgage Capital (IVR) has purchased Alt-A, prime jumbo, and commercial mortgage-backed securities (presumably at very attractive prices), but about half of the company’s equity capital has been allocated to leveraged purchases of agency securities supported by short-term repurchase agreements. About half of the interest rate risk is hedged via interest rate swaps, but the remaining cash flows remain exposed to interest rate risk. This would hurt returns and possibly require securities to be sold at a capital loss if interest rates spike. Given the massive recent expansion of the Fed’s balance sheet, and the potential for inflation if a recovery takes hold, betting that interest rates will remain near or below current levels seems like a poor strategy.
2. PennyMac Mortgage Investment Trust (PMT) was established to purchase distressed mortgage loans (rather than mortgage-backed securities), and maximize investor value by handling the mortgages efficiently, modifying mortgages when appropriate. Unless prices increase precipitously compared with recent lows, purchasing distressed whole loans is a good strategy, but it would be better to also have the option to purchase mortgage-backed securities if they offer better returns.
It would be ideal if retail investors could take direct positions in mortgage-backed securities or indices of mortgage-backed securities. Taxpayers are being asked to fund potential losses on legacy mortgage-backed securities, so it seems fair that they should also be allowed to make direct investments, without paying high investment management fees. This could be achieved by creating versions of Markit’s ABX and CMBX indices (as well indices for Alt-A and prime jumbo mortgage-backed securities) to be traded on a public exchange. Hopefully, Markit or another entity will move forward with this strategy, perhaps at the encouragement of banks and other market participants who would benefit from increased liquidity, before prices increase to less attractive levels.
Disclosure: Long PMT
