Description

I am recommending a short of Countrywide Credit (CCR), the nation’s largest independent mortgage banker, based on aggressive accounting that does not reflect economic reality, a growing debt burden, poor current and historical management of headcount and other expenses at the peak of the mortgage origination cycle, and a looming and under-publicized exposure to sub-prime credit through residual securities kept in securitization transactions.

The Company
In a highly fragmented but consolidating industry made up largely of bank and thrift subsidiaries, Countrywide has a 5-6% market share in both originations and servicing. The company also offers sub-prime mortgages and home equity loans, and is continuing to expand other businesses such as:
· Capital Markets – trading and originating mortgage-backed securities; brokering deals for mortgage servicing rights
· International – principally a JV with Barclay’s to share CCR’s origination and processing expertise with Barclay’s UK mortgage operations
· Banking – acquired in May of 2001, the bank is a start to solving Countrywide’s funding disadvantage vs. traditional banks and thrift
· Insurance – CCR underwrites homeowner’s and other property and casualty policies and reinsures the mortgage insurance industry.

The Stats
Cap - $6B; EV - $23.6B; P/E on 2002 8.4x; P/E on 2003 8x; Short interest - 5.7% of float

The Competition and rate cycle thesis
Countrywide is about to emerge from an enviable environment – peak volumes (resilient purchase market and strong refinance volume), peak margins (the industry is somewhat capacity constrained and fees have been rising), peak earnings. The back end of a rate cycle is characterized by fiercely competitive pricing, as it is difficult for the lenders to rationalize supply as quickly as demand slows. Refinance volume in particular has historically been highly volatile for obvious reasons. If anything, this could be the most difficult cycle in recent memory, as many new players (H&R Block, E*trade, and others) have brought in capacity and this supply along with the rate shopping available through the internet and the operational efficiencies brought on by consolidation in the industry (mostly Washington Mutual buying thrifts) will contribute to a highly competitive pricing environment. Further, the economy is obviously not in the shape that it was in the last rate cycle (1998-99) – whether or not there is a residential real estate bubble I believe that the purchase market is in for a difficult patch. While it is true that the cost of homeownership is relatively low in this interest rate environment, I believe that a large part of the strength has simply been consumers using strong home values to “trade up”. Such a cycle has its limits with the eroding consumer confidence levels, rising consumer debt burdens, and a probable cooling of home price appreciation.

Even in this ideal environment, Countrywide faces a critical disadvantage in its source of funds – it raises capital to fund its investment in generating new loan servicing business from the capital markets while competitors such as Washington Mutual and Wells Fargo can fund this investment from their significantly larger depository operations. CCR alludes to this in its disclosures - “Management believes the amount of equity capital required to finance its MSRs and other retained interests is high relative to its major competitors, and such requirement increases the overall cost of capital for the Company”. It is my view that the combination of the vicious mortgage cycle, fierce price competition and this funding disadvantage have contributed to management’s decisions to use aggressive accounting and to shift business away from traditional mortgage operations and into riskier areas such as sub-prime lending.

The Accounting thesis

Countrywide essentially books a gain in the current year for all of the future profits on servicing the additions to its mortgage portfolio. At the time of the sale CCR capitalizes an asset for the future profit stream (Mortgage Servicing Rights or MSR) from the fees. This is allowed under GAAP’s liberal gain-on-sale accounting – but not worth much of a multiple. The expenses associated with servicing (staffing, facilities, general and administrative) will hit the income statement over the life of the underlying mortgages (on average over the next 6 to 7 years but in some cases as long as 25-30 years) yet the profit is all booked upfront. On average CCR records a gain for approximately 5 years of service fees (which average 40 basis points per year) and thus ends up with an MSR asset that is approximately 2% of the Servicing Portfolio.

Further, without a strong regulatory presence (as pure thrifts and banks have) and with an auditing firm that could not afford to lose the account (CCR uses Grant Thornton, a regional firm – only 4 of the S&P 500 firms use someone other than a Big 5 firm), Countrywide’s management has become very aggressive with an already liberal set of Gain-on-Sale accounting regulations. I have spoken to some industry executives who believe that Countrywide is actually including the benefits of future cross-selling (of home equity loans, etc.) in its estimates of the value of its servicing asset. A simple but telling measure (Capitalized Mortgage Servicing Rights divided by Mortgage Servicing Portfolio) is warranted due to CCR’s less-than-complete disclosure, but it illustrates the aggressive accounting as of 3/31/02):

Capitalized MSR as a % of Primary Servicing Portfolio
Countrywide 1.9%
Washington Mutual (WM) 1.4%
IndyMac Bancorp (NDE) 1.57%

Translation: Countrywide’s recorded Gains on Sale have been 25% higher than industry averages for no reason except that the thrifts, who currently earn exceptional interest margins, have had no reason to be so aggressive. Prior year earnings would be 32% lower if industry averages for securitization gains were used.

It is further troubling that the company does not disclose the amount of loans securitized in any particular period, as many of its peers do. This would enable a more precise comparison, since the gain is really a function of the principal amount of loans sold.

Countrywide also records a gain upon securitization of its sub-prime loans. This gain is a function of the value of the retained interest which is essentially a highly levered residual tranche. In its SEC filings management admits that “the determination of fair values of the retained interests at the initial recording and on an ongoing basis requires considerable management judgment…these assets are not actively traded in stand-alone markets.” Further troubling is the sensitivity of both the gain itself and the company’s earnings to the valuation of these securities. Using the accounting example provided in the company’s 2001 10-K (page 27) illustrates the problem. If the value estimate for the retained interest is off by 25% the reported gain is overstated by 100%! If such an overstatement were to occur systematically throughout the year income would be overstated by 16% using 2001 numbers (and this is only analyzing a small portion of the total gain on sale – the sub-prime portion which is about 27% of the total Gain-on-Sale). Is such an seemingly large overstatement of value possible? Absolutely. The security that Countrywide is typically left with is a first-loss piece on a large pool of sub-prime loans representing a small portion of the total principal (perhaps 2%). Even a small understatement of the credit loss assumption on the entire pool has a significant effect on the value of the retained interest.

So how could Countrywide systematically overstate the value of these residuals (and thus net income) quarter after quarter? Eventually they would have to write down the asset after pressure from auditors and rating agencies. Herein lies the beauty of the scheme – Countrywide invests in derivative securities to hedge its investment in rate sensitive assets such as MSR and residuals. These contracts rise in value when interest rates go down; thus offsetting the losses in the residuals (which lose value as a result of the increase in prepayments associated with the lower interest rates). In its financials CCR nets them out as evidenced by this statement from the 10-K p.39 - “In Fiscal 2001, impairment of retained interests after Servicing Hedge gains was a net loss of $118.8 million”. I believe that this ability to revise the valuation under the offset of a derivative contract gain enables the company to be very aggressive with credit loss assumptions and assumed prepayment speeds on the residual valuations. It is worth noting that the direction of rate movements is inconsequential – either way the company would have an opportunity to revalue their residuals and bury the change under a hedging line item.

In addition, similar to the traditional Gain-on-sale, it is important to remember that the expenses of servicing the sub-prime loans and more importantly collecting delinquencies (a very high cost in sub-prime land) will hit the income statement over the coming years (as salaries, occupancy, and other operating expenses) yet the expected profit on the deal shows up in current year earnings.

Another area worthy of mention - I strongly suspect but cannot say definitively that Countrywide records a gain on sale in the instance of an existing servicing customer who refinances a mortgage (the CFO informed me that the number was “not that large” but was unable to answer definitively) . During my visit to the company I was shown the internal system which essentially mines the servicing portfolio for possible refinance candidates. While this is a worthwhile business strategy to proactively retain customers I believe it to be aggressive to record an accounting gain through income. This gain would essentially be an estimate of the present value of servicing fees in the out years (since through the refi you extend the estimated life of the loan). Thus current and prior year earnings in these years of strong refinance volume would include estimates of probable economic benefits that are years away. Based on statements in the 10-Q and 10-K that “85% of the Consumer Markets Division’s refinance volume was from existing mortgage customers” and “63% of the Consumer Markets Division’s total loan production during the period was source from the Company’s servicing portfolio” I estimate that 16% of total loan production in Q1 2002 and 19% of 2001 loan production was this type of transaction. Assuming that the related gains are proportional, these gains would represent 22% of EPS using 2001 figures (however the actual number could be considerably smaller if the present value was calculated properly).


The residuals thesis

As shown above CCR has the ability to overstate earnings through establishing aggressive valuations for newly created residual securities. I have also shown that the company may in fact be adjusting such values downward after the fact and hiding the losses amongst derivative gains. However Countrywide may not be able to continue this for long. The market risk section of the 10-K notes that “Issues of concern to one or more rating agency in the past have included the Company’s significant investment in retained interests, its involvement in sub-prime lending, as well as its liquidity and capital structure”. I believe it is possible that large write-downs are forthcoming as credit losses continue to rise on sub-prime mortgage pools. At 12/31/01 CCR reported $615 million in sub-prime interest only or residual securities (Note 7). This represents 24% of my adjusted BV of $2.6 billion (I adjust downward to record the MSR at the industry average). While there is clearly not enough disclosure to assess the magnitude, I believe that there is a significant possibility of meaningful write-downs that could affect debt ratings and cause the company to crumble.

Most sell-side analysts are unaware of the magnitude of the exposure (I even have seen a report that suggested that the worst case scenario for credit losses was 5 or 6% of the value of the residuals which shows a gross misunderstanding of the security’s structure). Company management was elusive on the subject. In the Annual Report, the principal value and valuation assumptions related to the residuals are lumped into the same category with home equity residuals, which adds a degree of murkiness. Despite this, I have been able to piece together a few facts over the past few quarters:
· There is some collateral protection for most of the loans. The company disclosed to me that the mean LTV ratio was 72%. However, problems with these pools aren’t at the mean, they are at the tails of the distribution.
· According to company management 70% are Alt A or A-, meaning generally good credit, but low documentation. Unfortunately for CCR this means that 30% are B or C paper.
· In booking the residual, the company used a principal credit loss assumption of “between 4 and 5%” according to management. This is not very high for a sub-prime pool.
· A large majority of the balances are “unseasoned” and have yet to reach peak claims years. You can make this deduction by studying patterns of sub-prime production, recognizing that sub-prime peak loss periods are years 2 to 4.
· The company uses a 20% discount rate to value the cash flows of the security. This was disclosed to give me some valuation comfort, but given the risk, I would expect such a high rate. If anything, this confirms that management sees risk when they look at the loan characteristics that I can’t see.
· Company now uses mortgage insurance on their residuals. This new structure was begun in 2001. It will limit exposure going forward, but doesn’t eliminate the problems already in place.

Conclusion: The details are murky but the risks of meaningful write-downs are very real.

Countering Wall Street’s Diversification Thesis

Both Countrywide management and sell-side analysts have been harping on “diversification” efforts. While the company has never posted an increase in earnings after a refi boom year, “This time is different”, because 32% of earnings come from non-core activities. In actuality, an overwhelmingly large portion of these revenues are also cyclically tied to origination volumes, such as fees for closing credit reports and title services. Also included is income from capital markets activities, where volumes and margins are significantly higher during a refinancing boom (as newly produced mortgages are securitized). Perhaps the analysts predict earnings growth (up 6% in 2003 following a refinance boom year with peak margins and peak volumes) as a way to stay cozy with Countrywide’s management and garner underwriting for CCR’s frequent debt offerings.

Loading Up on Debt

The peak of the cycle is a huge cash drain on the company; CCR added approximately $5 billion in debt in the 10 months ended 12/31/2001. While Countrywide’s management would say that they are investing cash flows in their MSR so they can reap the benefits later, over the same period MSR only increased by $349 million. This does not sound like good economics to me.

At 3/31/02 CCR has a whopping $17.6 billion in notes payable on what I believe is a true equity base of about $2.6 billion (adjust BV for the probable overstatement of MSR). This does not even include the net borrowing through agreements to repurchase securities. So here is a company employing massive leverage to produce mediocre economic returns. Is it sensible to use such leverage when nearly all of your competition is competing for the same business but with a lower cost of capital?

Valuation
CCR trades at only 8x consensus full-year 2002 estimates and 1.3x reported Book Value. However, I believe that both estimates and book value are overstated through aggressive accounting. Further, reported estimates do not reflect the economic reality of the business (see above) and should be discounted heavily even without considering the fact that Countrywide is among the most aggressive of the companies in booking the gains (which would result in another 30% reduction of EPS as described above).

EPS
Adjusting for the aggressive gain-on-sale accounting illustrated above I get to $3.60 for 2002 EPS. You could probably deduce here that one should not pay 13x peak earnings for a slow/no long-term growth business with a forthcoming violent turn in the cycle. However my conviction grows as I examine consensus 2003 estimates. These are too high for the following reasons:
· The extent of the price competition and reduced volume on the back end of the cycle is extremely hard to predict, so the analysts assume a nearly linear and fairly rosy scenario. As one example, the analysts have extrapolated a peak period for the relatively new and equally cyclical capital markets business.
· Salary Expense will likely be higher – it will be tremendously difficult to slash headcount as quickly as most analysts forecast. Most of the analysts simply estimate G&A expense as a % of volume, which works well on the way up, but is questionable once the cycle peaks. Management has increased headcount in each of the last 5 years through both peak and trough earnings. In the spring they added 2,000 more people to handle higher origination volumes. This is a mindset that is tough to break out of. This is further complicated by the current effort to build a banking presence, the increased collection efforts necessary on sub-prime and other delinquent loans, the strategic vision of the company which involves growing the sales force, as well as the added headcount needed to service the growing servicing portfolio.
· Doing a variety of analysis that would be too lengthy to print, I get normalized go-forward, aggressive accounting adjusted earnings in a range of $2.75-$3.00. Applying a slow growth thrift-like long-term average multiple of 10x gives a fair value range of $27.50 to $30. In this range, the risk reward starts to favor covering a short.

Book Value
· Simply booking the securitization gains asset (Mortgage Servicing Rights) at the industry average (see above) would lower Book Value by 40% to 20.60 giving a valuation of more than 2.2x book for a company that typically earns an “aggressive accounting adjusted” 9 or 10% ROE in a normal environment. Since there is a ready market for these Mortgage Servicing Rights, I believe that this Adjusted Book Value of approximately $21 represents the lower end of the likely bottoming out range for the stock even if the market were to catch on to everything in this report.

Risks
The risk with a short of CCR is primarily that the accounting games remain hidden due to their complex nature and the incomplete disclosures that serve as the evidence. Still even in the absence of accounting scandal this short represents a play on the refinance cycle as the market places a multiple on peak earnings power in a highly cyclical business.

Catalyst

· Intensifying scrutiny on accounting and disclosures has not fully run its course.
· The end of the refi cycle will result in downward estimates revisions.