Further investigation into Impac Mortgage Part I: Non-recourse trusts
I did more research into Impac Mortgage (IMH) over the weekend, and I plan to share my findings in a series of 3 posts.
In this first post I want to focus on the trust assets. While the trusts are somewhat peripheral to my investment thesis in Impac, an explanation of how they work is central to the following posts I plan to write about earnings, and understanding the trusts helps quantify what potential they might hold for Impac if the housing market recovery becomes robust.
Impac Mortgage was a $250 stock in 2004 (I am including the 10:1 share consolidation that took place). Obviously it has been a long way down. The fall in the stock price from then to now has been entirely because of the mortgage market collapse. This is something to keep in mind while I step through the next few paragraphs.
As I have explained in previous posts, I bought Impac Mortgage because of the growth I anticipate from their origination business. But while the current business model centers around mortgage origination, it hasn’t always been that way. Prior to 2008, in addition to originating mortgages Impac created and ran a number of off-balance sheet trusts. These trusts would buy mortgages, mostly one’s originated by Impac, and pay for those mortgages by selling securitized mortgage obligations to investors. The trusts would pay interest on their obligations from the cash collected on the mortgages.
In each of these trusts, Impac Mortgage holds a residual interest. This was essentially the equity in the trust. It takes the first losses but also stands to benefit from any cash that is collected from the mortgages above and beyond what is needed to pay interest on the mortgage borrowings and meet any of the collateral requirements.
The only liability Impac has is with respect to the residual interest. While the assets and liabilities of the trusts are consolidated because of GAAP rules, they are non-recourse to Impac. As the company stated in the 2011 10-k:
Because the assets and liabilities in the securitizations are nonrecourse to the Company, the bondholders cannot look to the Company for repayment of their bonds in the event of a shortfall.
So while the mark to market on these assets really messes with the earnings, something I will address in my next post, its nothing more than an accounting artifact.
The trusts are invested mainly in At-A mortgages, most of which were originated by Impac. Alt-A mortgages are those whereby the borrower has credit that is usually in-line with a conforming Fannie or Freddie mortgage, but there is one or more aspect of the borrower that makes the loan non-conforming. Such aspects could be:
- credit and income histories of the mortgagor;
- documentation required for approval of the mortgagor; and
- loan balances in excess of maximum Fannie Mae and Freddie Mac lending limits.
While the trusts carry mortgages from throughout the United States, there is a concentration in California and to a lessor extent Florida. I had to go back a bit to find it, but in the 2007 10-K the company disclosed that loans in California made up between 51% of the total loans held within the residential trusts and 61% in the commercial trusts at the end of 2007. This has obviously been a major handicap over the last number of years. I suspect though that based on the inventory numbers I am seeing in California (to give a couple examples, Sacramento has 2 months of inventory right now, while Orange County is down to 39 days of inventory), that may be in the process of changing.
Some other interesting facts to be gleaned from the loan data by vintage (as per the 2007 10-K) are:
- 99% of the residential mortgages were Alt-A mortgages and 99% were first lien
- Original loan to value actually dropped from a high of 82% in 2002 to a low of 76% in 2004 and to 73% by 2007
- A high percent of the 2004-2007 vintages appear to be interest only mortgages. The company described their interest only originations as “adjustable rate and fixed rate interest-only mortgages with 5 to 10 year interest-only periods and terms to maturity of 30 years with six-month to five-year prepayment penalty periods.”
- From within the 2002-2004 vintages, between 75% and 80% were adjustable rate mortgages, with 35%, 48% and 70% of those being hybrid (meaning they start off fixed and become adjustable after a certain period) for each of the years respectively
- In 2002 93% of the mortgages were owner occupied. This fell to 81% by 2004 and to 77% in 2007
The last year that Impac provided a breakdown of the loan amounts outstanding of each vintage was 2007.
In the aggregate, the performance of these trusts appears pretty ugly. At December 31, 2011, the Company’s long-term mortgage portfolio had 21.6% or $2.1 billion of loans that were 60 days or more delinquent. Below is a table illustrating the total principle outstanding and the amount delinquent in the aggregate amount:
But individually, some of the trusts are performing better than others. Impac doesn’t provide a lot of information on the individual performance of the trusts but it does break-down the fair value of the residual interest by year and between single family and multi-family in the 2011 10-K.
Keep in mind there are a little over 7 million shares outstanding, so the fair value of the trusts is not insignificant.
The company notes that “the estimated fair values of residual interests in vintage years 2005 through 2007 is reflective of higher estimated future losses and investor yield requirements compared to earlier vintage years.”
As for how Impac comes up with that fair value, every quarter Impac estimates the net fair value of the residual interest in the trusts. Impac says they determine fair value from a cash flow analysis of each of the trusts projected out to run-off, offset by the expected cash requirements of the bonds. Again, referring to the 2011 10-K:
The Company uses the valuation model to generate the expected cash flows to be collected from the trust assets and the expected required bondholder distribution (trust liabilities). To the extent that the trusts are overcollateralized, the Company may receive the excess interest as the holder of the residual interest.
As shown above, that fair value was estimated at $26.5 million at year end. In Q2, this analysis led to a fair value of $23 million.
Impac has been receiving cash flow from these trusts if the money coming into the trusts (from the mortgages that the trusts hold) exceeds the collateral requirements and the cash needed to pay the senior trust holders. Impac collected $4.4 million from these trusts in Q2. In 2011 net interest income from the trusts were $3.6 million and in 2010 it was $5.7 million.
The cash that Impac receives from the trusts is directly related to LIBOR. In the risk factors that the company stated in its 10-K:
Interest rates on the bonds usually adjust monthly with changes primarily in one-month LIBOR. Derivatives instruments (primarily interest rate swap agreements) inside the securitization trusts initially entered into were designed to offset the risk of movements in LIBOR that created the adverse effect of the interest income collected on the loans being less than interest expense paid to the respective bondholders. However, many of these derivatives agreements have maturities less than the maturities of the loans. Therefore, increases in LIBOR rates could significantly reduce the future cash flows we receive from the retained interests in these securitization trusts.
With short term interest rates seemingly low as far as the eye can see, there should not be much to worry about.
What to make of it?
Clearly the later vintages (2005-2007) of these trusts were full of the sorts of mortgages that ended up defaulting when the housing market collapsed. I don’t think we should expect much from these securities, though with the limited information available, I can’t really say that with certainty. The earlier vintages (2002-2004) appear to hold more promise. They seem to be comprised of more standard mortgages of a better quality. However the amount remaining that is outstanding for these years is far less than the later years, so the upside from these vintages is going to be limited.
But even with all the information I have compiled here, its very difficult to draw conclusions. What we really need to know is how close each of the non-performing trusts are to becoming compliant (and therefore to spinning off cash to Impac’s residual interests). That data isn’t available, or at least is no where I have seen.
One thing I think I can conclude is that I don’t think we should expect any principle return on the residuals. Any cash is going to come from excess interest. I make this deduction simply by noting that the outstanding principle ($9 billion) has been estimated much more than fair value ($5.5 billion), that 20% of the mortgages are 60+ days delinquent, that both of these facts imply that the principle liabilities far exceed the assets and that with senior claims holders being paid off first, it seems very unlikely there will be anything left for the residual equity Impac holds.
The other point that can be made is that the sheer volume of mortgages held is enormous in comparison to the market capitalization of the company. The existing unpaid principle balance is over $9 billion dollars, and the fair value of those assets is estimated at over $5 billion. While its difficult to pin point exactly how changes to delinquencies and cash flow within the trusts will impact the residual cash flow to Impac, its clear that small changes could have a big effect.
I still view this as a nice wild card with the company. With a fair value of $23 million, they are currently worth about $3 per share. If these trust assets turn out to be worth more as housing turns, that would be great. While its not the reason to invest in the company, it’s a nice added bonus.