For today's post mortem, the analysis emphasizes the originally filed Washington Mutual 10-k for December 31, 2007, the 2006 10-k, press reports and press releases from the FDIC and the Office of Thrift Supervision.  Whenever these documents caused a reaction like "Huh? That can't possibly be right," we consulted Mr. T, X-man, Mr. S, Professor S and a mystery ex-Wamu person.  Three of the five of them worked at Washington Mutual in risk management, and the other two are very experienced risk managers who had regular contact with Wamu or Wamu management.  I alone am responsible for any errors that may remain in what follows. As in our Countrywide post mortem on May 18, 2009, we judge Wamu's ability to manage risk by what the 10-k's say about the institution's ability to answer these four critical questions from our April 27 blog post, a pass-fail risk management test for CEOs and members of the Board of Directors:

Question 1: What happens to the market capitalization and net income of the firm if any of these risk factors change: home prices, foreign exchange rates, commercial real estate prices, stock index levels, interest rates, commodity prices?

Question 2: Using an insider's knowledge of the assets and liabilities of the firm, both "on balance sheet" and "off balance sheet," what is the best estimate, monthly for the next ten years, of the probability that the firm will fail in each of these 120 monthly periods?

Question 3: Using only information available to an outsider, what is the best estimate of the probability of the failure of the firm in both the short run and the long run?

Question 4: If the firm is able to answer Questions 1, 2, and 3, what hedging position is necessary to insure that the macro factor sensitivity of the firm and default probability of the firm reach the target levels set by the Board of Directors?

Risk Expertise at Washington Mutual

Looking at the Board of Washington Mutual, one has to conclude that the Board was much less qualified to direct the risk appetite of the organization than the Board of Directors of Countrywide Financial.  The Chairman of the Audit Committee of the Board of Countrywide had a Ph.D. in Economics, 25 plus years of risk experience via service on major bank asset and liability committees, and served as CEO once and Vice Chairman twice for very large financial organizations.  The head of the Finance Committee of the Board of Washington Mutual has a resume that's much less significant.  She worked at Washington Mutual in investments, was hired by CEO Kerry Killinger, and left as a mid-level manager.  She had a bachelors degree from one of the best schools in the United States, but she had no graduate training in economics, finance or risk management.  Among the other board members, there was more knowledge of coffee (the ex-CEO of Starbucks) than there was of risk management.

That's danger sign number one.

Danger sign number two was the head of enterprise risk management who had previously served at CIBC and Bank One.  X-man had worked at Bank One, and Professor S had worked at CIBC.  Both are great risk experts and they had an interesting opinion on the skill set of the head of ERM: no opinion.  That's danger sign number 2.  The risk industry is a small one, many of the risk experts are well known to their peers.  It's a business where you tend to exit quickly if you're not good (like a Chinese restaurant in San Francisco China Town), so it's rare to find someone in a spot that senior who isn't well known even among peers who had worked at the same firms.

Danger sign number 3 was the background of the chief financial officer, who had been in financial services at GE.  X-man said, "I worried that he knew more about financing aircraft engines than mortgages."

The irony of this view of the top of the organization is that the working level was highly skilled.  That was confirmed in 10 e-mails from my friends and hinted at indirectly in what management judged to be important enough to put in the 10-k.  We believe strongly that things would have turned out much differently if senior management had paid more attention to what the working level was doing and to the analysis that it was performing.

With these worries in our pocket, we proceed.

Risk Systems Infrastructure at Washington Mutual

We confirmed with X-man, Mr. T, Mr. S and one other person that web sites identifying the main risk systems vendor to Wamu was the same legacy interest rate risk vendor used by Countrywide.  As we saw there, management was unable to answer the question "What happens to us if home prices go up or down?"  We're up to danger sign number 4.  Let's look at what Wamu management said about the firm in its initial 10-k filing for 2007 and try to draw some conclusions about whether or not that was true at Wamu.  Remember, our perspective is that of a sophisticated outside analyst who only knows what the company has chosen to disclose to investors.  If the internal facts were different, that is a problem caused by the company's disclosure, not by the analyst using what's been disclosed.

Risk Management Comments from the Washington Mutual 10-k for 2007 (Original Version)

Wamu's 10-k described the firm as the seventh largest bank and thrift company in the United States, with 49, 403 employees.  The first key entry in the 10-k is this comment, which parallels Countrywide Financial to a surprising degree:

•  Page 14, 2007: "Effective July 18, 2006, the Company adopted a share repurchase program approved by the Board of Directors (the "2006 Program"). Under the 2006 Program, the Company was authorized to repurchase up to 150 million shares of its common stock as conditions warrant and had repurchased 102,545,978 shares under this program as of December 31, 2007."

The implication of this statement is very very clear: Regardless of the risk analytics and risk systems used by the staff of Wamu, senior management and the Board of Directors reached the conclusion that the firm was so overcapitalized, given the risks it was facing, that it was appropriate to use cash to repurchase common stock of the firm.  With 20/20 hindsight, we know that home prices in Los Angeles peaked in September 2006, just six weeks after this Board decision.

Unlike our Countrywide post mortem, which was based on the 2006 10-k, the 2007 10-k for Wamu acknowledges that the credit crisis is underway and what its implications are:

• Page 17, 2007: "Reflecting the significant credit deterioration, the Company recorded a provision for loan losses of $3.11 billion in 2007, an increase of $2.29 billion from 2006 and about twice the level of 2007 net charge-offs, which totaled $1.62 billion. Adverse trends in key housing market indicators, including growing inventories of unsold homes, rising foreclosure rates and a significant contraction in the availability of credit for nonconforming mortgage products continued to deteriorate throughout 2007 and exerted significant downward pressure on home prices, particularly in areas of the country in which the Company's lending activities have been concentrated….With early indicators in 2008 suggesting that the housing market is continuing to deteriorate, the Company expects that it will experience significantly higher credit costs throughout its single-family residential mortgage portfolios."
• Page 17, 2007: "Because of this disruption, the Company transferred approximately $17 billion of real estate loans to its loan portfolio in the third quarter of 2007, representing substantially all of the Company's nonconforming loans that had been designated as held for sale."

The Company also moved to recognize that it needed more capital, not less, and instead of repurchasing common shares, by the end of 2007 it issued preferred stock and cut the dividend:

• Page 18, 2007:  "To bolster its capital levels and liquidity position, the Company issued a total of $3.9 billion of Tier 1 capital in the fourth quarter of 2007, comprised of $2.9 billion, net, of noncumulative, perpetual convertible preferred stock issued by the Parent and $1 billion of noncumulative, perpetual preferred shares issued by Washington Mutual Preferred Funding LLC, an indirect subsidiary of Washington Mutual Bank. Additionally, commencing in the first quarter of 2008, the Company reduced its quarterly cash dividend rate on the Company's common stock to 15 cents per share"

As we know from hindsight, this ended up being a case of "too little, too late."  The Company mentions the importance of home prices in its discussion of the allowance for loan losses:

•  Pages 21-22, 2007: "The Company allocates a portion of the allowance to the homogeneous loan portfolios and estimates this allocated portion using statistical estimation techniques. Loss estimation techniques used in statistical models are supplemented by qualitative information to assist in estimating the allocated allowance. When housing prices are volatile, lags in data collection and reporting increase the likelihood of adjustments being made to the allowance."
• Page 22: "The Company also estimates an unallocated portion of the allowance that reflects management's assessment of various risk factors that are not fully captured by the statistical estimation techniques used to determine the allocated component of the allowance. The following factors are routinely and regularly reviewed in estimating the appropriateness of the unallocated allowance: national and local economic trends and conditions (such as gross domestic product and unemployment trends); market conditions (such as changes in housing prices); industry and borrower concentrations within portfolio segments (including concentrations by metropolitan statistical area); recent loan portfolio performance (such as changes in the levels and trends in delinquencies and impaired loans); trends in loan growth (including the velocity of change in loan growth); changes in underwriting criteria; and the regulatory and public policy environment."

This is reassuring--it confirms that, at the working level at least, the right things are being focused on.  The Company goes on to describe its risk organization for investors:

• Page 50: "Management's governing risk committee is the Enterprise Risk Management Committee. This committee and its subcommittees include representation from the Company's lines of business and the Enterprise Risk Management function. Subcommittees of the Enterprise Risk Management Committee provide specialized risk governance and include the Credit Risk Management Committee, the Market Risk Committee and the Operational Risk Committee."
• "Members of the Enterprise Risk Management function work with the lines of business to establish appropriate policies, standards and limits designed to maintain risk exposures within the Company's risk tolerance. Significant risk management policies approved by the relevant management committees are also reviewed and approved by the Audit and Finance Committees. Enterprise Risk Management also provides objective oversight of risk elements inherent in the Company's business activities and practices, oversees compliance with laws and regulations, and reports periodically to the Board of Directors."

Only an insider would be able to divine the exact day-to-day operations implied by the second paragraph.  My reading of the first and second sentences in that paragraph is that there is not enough power at the Enterprise Risk Committee to say "Maximum exposure of the firm to a 10% drop in home prices is X."  The sentences seem to imply that the risk limits allowed to the business units were parceled out with a much more vague allocation, contrary to the four questions that we pose above.  We draw this implication because the 10-k contains no language stating the company's mark to market tolerance for changes in any macro factor, including both interest rates and home prices.  Again, at the working level, there may have been a lot done in this regard, but management evidently did not judge it to be useful enough to disclose the results of that effort to the owners of the firm, the shareholders.  Our conclusions seem to be confirmed by this statement in the 10-k:

• Page 51:"The Company's credit risk management process provides for management and accountability to be decentralized through our lines of business."

On pages 52 and 53, the Company acknowledges that home prices are the drivers of credit losses.  The major table showing loan to value ratios, however, shows the loan to value ratios by type and year of origination only on an original loan to value ratio basis.  What we want, of course, is an assessment of the CURRENT loan to value ratio so we, as potential investors in the firm, can make our own judgment of the risk the company faces.  There is a one line "current loan to value ratio" for all loans originated in a given year, which is described as follows:

•  Page 53, 2007: "The average estimated current loan-to-value ratio reflects the UPB [unpaid balance] outstanding at the balance sheet date, divided by the estimated current property value. Current property values are estimated using data from the September 30, 2007 Office of Federal Housing Enterprise Oversight ("OFHEO") home price index."

There is no further explanation of the calculation.  The fact that it was not reported by loan type would lead most observers to conclude that it is a spreadsheet-level adjustment to the aggregated original loan to value ratios reported elsewhere in the table, rather than a loan by loan assessment that investors would prefer.  Page 63 seems to confirm this when it describes "collective impairment" instead of loan by loan impairment analysis:

• Page 63: 2007: "The Company separately evaluates the impairment of the homogeneous and non-homogeneous loan portfolios. The homogeneous portfolio, comprising substantially all loans held in portfolio, is evaluated for collective impairment and consists predominantly of home loans, home equity loans and lines of credit, credit card loans and most commercial business, commercial real estate and multi-family loans"

Page 64 refers both to "pool level" and individual loan level analysis that is done for estimation of the proper allowance for loan losses.  The individual loan level analysis confirms that, at the working level, some very good analysis was being done:

•  Page 64:"The ... model produces an estimate of the cumulative loss over the remaining terms of the loans by analyzing loan-level data, the key attributes of which are static collateral variables including property type, loan type, lien type, original balance, original loan-to-value ratio and documentation type; ARM-specific variables, such as ARM caps, floors and resets; dynamic variables, such as estimated current loan-to-value ratios and the age of the loan; borrower variables, such as original credit score and delinquency history; and market conditions, such as housing price appreciation or depreciation rates by metropolitan statistical area, and interest rates."

Unfortunately, management declined to display in the 10-k a chart showing the kind of stress test with respect to home prices that would be required by U.S. regulators during the February-April 2009 period.

With respect to the silo analysis of interest rate risk reported in the 10-k, we were fairly surprised to see that the disclosure focused solely on the net interest income impact, not the market value impact, of a change in interest rates:

• Page 77: "Net interest income sensitivity declined in the ±100 basis point environments in the January 1, 2008 analysis compared to the January 1, 2007 analysis."

For more than 25 years, regulators, academics and leading risk practitioners have demonstrated that this is the very analysis that led to the S&L crisis of the 1980s, since it fails to include the impact of interest rate movements on the last 28 years of of a 30 year mortgage.  One person and Mr. T confirmed that this mark to market analysis was in fact performed using the legacy interest rate risk system.  Management, however, decided that this macro factor sensitivity analysis of the full balance sheet with respect to interest rates was not important enough to report to investors.  Countrywide, by contrast, did make such interest rate risk disclosure.

Finally, on page 155 of the 195 page 10-k and 5 pages of exhibits, the Company confirms that the repurchases of its common stock continued until May, 2007, well past the point at which home prices could be seen declining:

• Page 155: "The May ASRs were settled in July 2007, and resulted in the Company's receipt of an additional 1.91 million shares of its common stock from the counterparty. At December 31, 2007, there were no outstanding ASR agreements. In addition, during the year ended December 31, 2007, the Company purchased 7.17 million of its shares in open market repurchases."

Post Mortem Summary for Washington Mutual

Our reading between the lines of the 10-k for Washington Mutual contains some real contradictions.  On the one hand, there clearly was some very sophisticated loan level credit risk analytics and prepayment analytics going on that took into account the role of home prices.  On the interest rate risk side, we know from private conversations that more sophisticated analysis was being done than what was revealed to investors and shareholders.  On the other hand, we think that these conclusions are inescapable:

•  If the working level staff prepared a loan by loan total balance sheet analysis of exposure to home prices for management, management chose not to disclose it to investors
• Since the working level did prepare a standard mark to market interest rate sensitivity analysis for management, it is of grave concern that management chose not to make this disclosure either

Given this vacuum of information during the greatest fall in home prices in the last 70 years, it is no wonder that the events in the Epilogue took place. The hardships that these events have brought to the 1,831 shareholders of Wamu and its 49,403 employees are unimaginable.

Epilogue

As 2008 unfolded, the following events occurred after the 10-k was submitted on February 29, 2008:

• Investors demanded the resignation of Mary E. Pugh, head of the Finance Committee of the Board of Directors.  She resigned from the Board in April
• Head of Enterprise Risk Management Ronald J. Cathcart left the Company in 2008, only to be replaced by an executive from Countrywide Financial
• CEO Kerry Killinger was ousted by the Board on September 8, 2008.  He is currently named in 37 federal lawsuits.
• On September 25, 2008, after Wamu lost $16.7 billion of deposits in 10 days, the FDIC seized Wamu and sold its two main operating entities to J.P. Morgan Chase & Co.

With stronger oversight by the Board and senior management, these events should have never happened.  A lot of brilliant analysis at the staff level was misunderstood, ignored, or not disclosed to those people who most deserved to have this analysis: shareholders, investors, and depositors.

Donald R. van Deventer
Kamakura Corporation
Honolulu, May 20, 2009