Downey Financial Corp. (NYSE:DSL) released earnings for Q1 2007 yesterday and the stock enjoyed a nice jump up. In fact, during the past few days, DSL has been charging ahead, perhaps due to some shorts hoping to cover ahead of earnings and possibly due to investors getting more comfortable with the banking sector.
Nonetheless, the uptick in shares of DSL is quite surprising considering today’s valuation is essentially the same as last year’s, if not more expensive on a forward looking basis, despite a balance sheet that continues to deteriorate and earnings quality that has become progressively weaker. I’ve covered DSL in previous posts on this website and given today’s 8-K filing I’ve updated my analysis on the Company to incorporate Q1 2007 results.
DSL 1-yr chart:
The key basis for shorting DSL is that the Company’s earnings model is being increasingly driven by capitalized interest income from negative amortization (“CINA”) and that this cannot be sustained over time. Recognizing interest income derived from negative amortization loans is completely acceptable and proper. However, the bank’s customer concentration, loan class concentration, and the current economic/housing situation facing the region it operates in cast significant doubt as to how tangible the Company’s CINA truly is. Even during excellent real estate time periods, a shareholder should prefer to see much more cash interest income as opposed to CINA. Instead, conditions in Southern California have considerably deteriorated compared to last year, DSL’s operating metrics are worse, yet the stock’s valuation has remained virtually unchanged.
Table I provides an update to the overall impact CINA is having on DSL and compares Q1 2007 to Q1 2006 (highlighted). What is striking is that interest income from loans barely grew on a quarter to quarter basis yet CINA grew by 28%. In addition, CINA accounts for 31% of total interest income for Q1 2007 compared to 25% in Q1 2006. On a sequential basis, CINA for Q4 2006 was 29% according to the 8-K (not shown in Table I) so CINA continues to account for a growing portion of interest income. What’s of even more consequence is the continuing trend of CINA accounting for a huge portion of DSL’s net interest income and pretax operating income. For Q1 2007, CINA accounted for 111% of the Company’s pretax operating earnings which actually came in below Q1 2006 figures.
Comparing Q1 2006 to fiscal year 2006 results shows that this higher contribution of CINA to EBT does not appear seasonal as CINA accounted for 87% of EBT in 2006, a slight increase from Q1 2006 levels. As a result, it wouldn’t be surprising if this trend continues on in 2007 where operating earnings are wholly driven by DSL’s CINA recognition.
The trend of balance sheet deterioration has also continued into Q1 2007. Table II updates DSL’s delinquent loans and non-performing assets information. Table II compares Q1 2007 to Q1 2006 as well as to fiscal year 2006 to get an idea of the sequential change on the balance sheet.
Delinquent loans and non-performing assets have grown by 20% and 30% from the end of 2006. In addition, the absolute dollar level of these accounts is starting to be significant. While DSL’s total loan portfolio is about $13 billion, having 1%+ of loans in these two categories shows a dramatic spike from just two years ago. In addition, a significant portion of option ARMs have started to reset so it’s entirely possible that these figures, which have grown rapidly, continue to grow at a brisk rate in the coming year.
As stated in a previous post, DSL’s financial statements and valuation assume that CINA is 100% money-good with a low likelihood for impairment. At $68 per share, DSL is valued at about 10.0x trailing EPS. This may seem reasonable until one considers the potential for CINA impairment. Table III is an updated sensitivity analysis of implied EPS based on the level of CINA impairment
As Table III demonstrates, EPS is being increasingly driven by CINA (no surprise as we established that EBT was as well). What’s even more “impressive” is the potential overstatement in DSL’s equity book value as a larger portion of the Company’s retained earnings are from the flow through of CINA. Even a slight dent to the certainty of the CINA results in a questionable valuation of DSL and not just on an EPS basis. Table III is really just a basic illustration to show that EPS could be modestly to significantly overstated if the negative amortization loans the Company is basing its capitalized interest income from are impaired.
It does not show the full impact to the Company because any identification of loan impairment would be applied on a book value basis, meaning that if DSL identifies a class of impaired negative amortization loans, those assets will be immediately written down along with any future corresponding CINA from those specific loans. This is essentially what a short is looking for because it would “break” the current operating and accounting mechanism that’s in place. Any problems related to the currently-viewed safe, prime, option ARMs secured by residential real estate in Southern California would result in a book equity write down, corresponding to a more expensive DSL on a price to book basis. In addition, the second-order effect would be that DSL would lose its credibility to some extent as any future CINA recognition would not be valued as 100% “money-good” by the market as it currently is. This would result in a revaluation of DSL based on the adjusted book value as well as its “true” EPS.
To further illustrate this point, going back to Table I, if one were to eliminate CINA from Q1 2007 operating results the Company would have about $189MM in interest income against $149MM in interest expense, resulting in $40MM of net interest income. Adding in non-interest income of $18MM and operating expenses results of $65MM results in a pretax operating loss of $7MM as opposed to the $77MM in pretax operating earnings booked. In contrast, applying the same method to Q1 2006 would have resulted in pretax operating income of $21MM, a big difference compared to the $78MM in reported EBT for Q1 2006, but at least earnings as opposed to a loss. To further illustrate the continued impact of CINA on earnings, the same method applied to 2003 figures results in $167MM in pretax earnings compared to the $176MM in pretax earnings reported, hardly significant when compared to more recent periods.
While DSL is not a subprime lender, the market does not seem to factor in the geographic exposure of the Company and more importantly the indication from other banks that subprime-like problems are spreading to prime portfolios typical of what the Company carries on its books. Some subprime lenders have been able to “get out” of their problems by selling large blocks of their portfolios which may give the market comfort in that there are buyers for even the most radioactive assets if the price is right (or appears to be). However, the stocks of companies like Fremont General (“FMT”) are still below their pre-disaster level and DSL, despite having less subprime exposure, is still at risk based on regional conditions in a historically volatile market. The recent bounce in the broader financial services stocks seems to have currently put the market at ease in regards to the actual “hard” issues occurring on the ground.
For example, on April 17, 2007, the Los Angeles Times ran an article by David Streitfeld entitled “The Mortgage Meltdown” which offered some statistics that should be quite relevant to any mortgage lender focused on Southern California. Some key takeaways were that the first quarter of 2007 had an 800% increase in foreclosures (11,033) compared to the prior comparable period while 46,760 homeowners were sent default notices. In addition, while there’s been more pressure in low-income areas, the trend is expected to reach affluent neighborhoods as well. The article also mentions that the current rise in foreclosures is confusing because it is occurring during a seemingly healthy economy and this level of foreclosures has been historically driven by job loss/economic contractions.
Another important takeaway from the article is that most of the defaulted loans in the past quarter were from loans made during 2005. The article does not specify why but many following the sector would probably expect this to occur as many of those loans reset to much higher rates. One could expect this trend to impact a broader swath of option ARM borrowers as the year progresses. The burden of higher payments might “knock out” lower income borrowers early on but eventually the newly reset rates could become too burdensome for more middle class earners as well, especially when a larger portion of prime loans were comprised of no-doc loans. Consequently it will not be surprising to see this trend of 2005 vintage loans defaulting continue throughout 2007. Table IV offers some statistics provided in the Los Angeles Times article (original source is DataQuick Information Systems).
Disclosure: Author manages a hedge fund that is short DSL