Taking a cue from David Letterman, we have compiled this list of the top valuation mistakes and in each case give the "daily life analogy" to illustrate the magnitude of the error. There are few institutions that have not made some of the mistakes on this list. It's the job of regulators, management and the Board of Directors to root out these mistakes and eliminate them one by one. This is especially true at the governmental level, where these mistakes have been frequent and large. Given the magnitude of financial losses that can result from these errors, ridding the organization of those who make these mistakes is an important part of the quality control process.
1. The Fake Rolex Watch Mistake: Ask the Investment Bank Which Sold You a Fake Rolex What the Watch is Worth
For many years, buyers of collateralized debt obligations who based their purchase decisions on "ratings and expected loss" turned to the investment banks which sold them these transactions for monthly valuations of the CDOs. From a personal perspective, if you paid $1000 for a fake Rolex watch that you thought was real, would you expect the person who sold you the watch to tell you it's a fake that's only worth $10? The investment bankers selling CDO tranches were simply mini-Madoffs. As long as you continue to believe that the CDO tranches you have already bought from the investment bank are holding their value, you keep buying. Once you realize that the CDO tranche is worth a lot less than you paid, the investment banker's gravy train is over.
2. The Poker Game Mistake: Ask Someone Else Playing in the Same Game How You Should Bet
The second worst valuation mistake one can make is to either (a) reveal the nature of the portfolio you hold OR (b) turn over the portfolio you hold to another market participant who is playing against you in the "game of trading." Is there anyone that naive, one might ask? Yes, governments, government sponsored entities, and hundreds of large financial institutions have revealed their portfolios to fund managers for "valuation and risk management services" and to physically manage the portfolios. This is the daily life equivalent of playing poker for money and then asking someone who's playing in the same game how to bet and play the hand you hold, showing them your cards. Remember Long Term Capital Management? Once everyone else in the market place knew that they had to sell what they owed, bids for every type of security were lowered by other market participants not because all securities were worth less but because everyone knew that LTCM MUST SELL no matter how low the price is. The penalty for revealing your portfolio is that your counterparty now has twice as much information as you do about the market for everything that you own; your counterparty knows their own portfolio, they know their own portfolio, and they know the portfolios of everyone else that's been silly enough to show their cards to them. Your counterparty is now "the house" in a card game that isn't supposed to have a "house." The house always wins. What compounds this error is that many institutions actually pay money to create "the house" that takes their money, paying a competitor of theirs in the market place for valuation and risk management services. If you want advice on how to play cards, ask someone who isn't playing against you.
3. The War of the Worlds Mistake: Believe in a Valuation Technique Because Everyone Else Thinks It's True
In a famous spoof from decades ago, millions of Americans believed that the Earth was being invaded by Martians because they and their friends heard it on the radio and believed it was true. In the same way, hundreds of investors bought collateralized debt obligations based on the "copula method" even though it should have been obvious that the technique was fatally flawed. The August 12, 2005 Wall Street Journal pointed this out in a page 1 story, more than 18 months before Citigroup, Merrill Lynch, UBS and hundreds of other investors took hundreds of billions of dollars of losses on CDOs. The author of this post made a presentation in June, 2007 at the Zurich meeting of the International Association of Credit Portfolio Managers noting that macro factors drive default probabilities up and down, and this valuation technique resulted in simulated losses that were much more severe than the copula simulation technique. What was the reaction? After the meeting two people from one of the largest banks in Germany and one of the largest banks in Switzerland approached the author and said, "That was a nice presentation, but there were 100 people in this room and every single one of them thinks you're wrong." They truly believed that the Earth had been invaded by Martians because everyone else in the believed the same thing. If everyone tells you there are Martians on earth, it's useful to look for a real Martian before taking it for granted.
4. The Cash Card Mistake: Tell an Investment Banker Exactly How Your Firm Evaluates Complex Securities
This is one of the most common and most serious valuation errors in the book. It's the equivalent of giving someone your cash card and telling them your password. The author's former colleague Tatsuo Kishi often talked about how Salomon Brothers was able to make $500 million in profits in the Tokyo market every year for more than a decade in the 1980s and 1990s. The process was very simple. The sales force at Salomon was armed with a story, which of course was not accurate, about the "Salomon approach" to valuing fixed income options. After wining and dining their clients, the Salomon sales force would engage in a collegial discussion "among friends" about how the client evaluated fixed income options. The next day, the Salomon sales people would take what they learned to the rocket scientists in their own firm who used a much better technique. Salomon would then create a package of securities that looked great to the client, but which in fact made tons of money for Salomon. This process would be repeated over and over again until the client finally realized they were the largest buyer in the market, that their own technique was wrong, and that they had been had. Banc One was a huge victim in this regard with amortizing interest rate swaps in the mid-1990s, and other examples number in the hundreds. Like every bank warns its clients, don't write the password for your cash card in the wallet in which you keep the cash card.
5. The Carton of Eggs Mistake: Don't Check to See if the Eggs Are Broken, Just Look at the Egg Carton Before Buying
I've learned the hard way when shopping that it's very important to check the carton of eggs I'm buying to make sure none of the eggs are broken. The same diligence is MUCH more important when buying tranches of securitized assets. Why? In the supermarket, I am willing to assume that the eggs were broken by accident. On Wall Street, one can rest assured that, if Wall Street knows you won't check the egg carton, you'll be buying one dozen broken eggs every time. Even Chris Cox, the former head of the Securities and Exchange Commission, insisted that investors cannot possibly know the value of what they buy without electronic access to the transaction level collateral and underlying borrower details. Whether it's mortgage-backed securities, synthetic CDOs, or CDOs of mortgage-backed or asset-backed securities, one cannot possibly make the most accurate valuation calculation without the loan level or transaction level detail. The credit crisis which began in 2007 has produced a seemingly endless series of stories about "trash" that was securitized, split into tranches, and sold at a considerable profit because investors did not take the time to look at the underlying collateral, piece by piece. In my book with Kenji Imai and Mark Mesler, we tell the story of how Salomon Brothers profited heavily because they looked at the zip codes around closing military basis in the mid-1980s, concluding that home prices would drop and mortgage defaults would rise. Armed with this knowledge, a managing director smiled that they had "ripped the ears off" investors who were either not knowledgeable enough or not diligent enough to pay the same attention to the details of the underlying collateral. A more modern incarnation of this phenomenon is the "NINJA" loan--"no income, no assets" that is the poster boy for the current crisis. What financial institution would make a mortgage loan like this for their own account? No one. So how do you make money from originating NINJA loans? First you package 50 sets of NINJA loans into 50 different mortgage-backed securities. Second, you make these 50 mortgage-backed securities issues the collateral on a CDO with 10 different tranches. If any investors ask hard questions, you just mix enough normal quality mortgage-backed securities issues into the reference collateral until people stop asking questions about the low quality stuff. There is no substitute for checking the collateral, in spite of the best efforts of the structurers, the trustees on each mortgage-backed securities deal, and the trustees on the CDO deal to slow down or block the ultimate owners of the collateral from examining that collateral piece by piece. It's increasingly easy, for the right price, to buy access to the collateral level data and it's also easy, using a powerful risk management system like Kamakura Risk Manager, to simulate the returns on each tranche at each layer of securitization. Exercising this care is not trivial, however, and that has important implications for securitization: if it becomes necessary, more time consuming and more expensive for the investors to insure they're not buying a carton of broken eggs, then the only investors who can afford to do the necessary homework will be "large lot" investors who buy in bulk. That means fewer investors, wider spreads, and lower volumes for securitized assets in general, especially CDOs. This is not the investors' fault--they're not the ones who put the broken eggs in the egg carton!
6. The "My Brother-In-Law Told Me It Was a Good Investment" Mistake: Bernie Madoff and the Rating Agencies
On March 16, Bob Jarrow and I posted on this blog "The Ratings Chernobyl," which was distributed over www.riskcenter.com and www.garp.com on March 9. It talks about the errors of the rating agencies in assessing the credit quality of large financial institutions and discusses the factors which lead to those errors. A more serious set of errors by the rating agencies was to act as a paid agent of the structurers of tranched securities in telling investors that these structured securities were safe. They played a role of the many sincere but misguided people who convinced their friends to invest with Bernie Madoff in spite of the fact that the returns on Bernie's funds were too good to be true. The fact that AAA/Aaa rated tranches were trading at libor plus 300 basis points should have raised a red flag among those looking at securitized assets. Consistent with Point Number 5 on the egg carton risk, one can't rely on the paid agent of the egg company to tell you that none of the eggs in the egg carton are broken. In a very important story by Roger Lowenstein in the New York Times on April 28, 2008 entitled "Triple A Failure," quotes a 20 year veteran of Moody's in charge of asset-backed finance as saying that they don't analyze every loan in rating securitizations. And why should they? They're paid by the egg factory, why check that all of the eggs in the carton are not broken. The egg factor says that they've done the checking. In the Bernie Madoff case, the people claiming that the investment was a good one were undoubtedly those paid out early in the Ponzi scheme by the later-arriving investors.
If you are buying eggs, don't take the word of someone paid by the egg factory that none of the eggs in the carton are broken.
7. The 2+2=5 Mistake: Technical Errors in Valuation
The last category of FAS 157 valuation and stress testing problems are technical in nature. In most cases, the errors are made by those with good intentions, analysts trying to produce accurate valuations. On the road to the correct calculations, the analyst drives the car off the road for a number of reasons--the complexity of the instruments being valued, the "newness" of the valuation concepts to many (including some very experienced finance people), and inconsistency in the views of the experts in both accounting and financial theory. The list below includes a lot of the errors that we've seen often. We hope to elaborate on them in subsequent blog entries:
A. Assuming Normal Distributions when Reality is Not Normal
Every analyst who has to deal with random variables (like macro factors or stock prices) in a valuation grabs the normal distribution first. This can be seriously incorrect. What if we assume future equity returns are normally distributed with their historical means and standard deviation? That implies Bear Stearns would never default because the probability of a -100% return on the stock over a one month time frame is 0.000000%. The bankruptcy of Bear Stearns isn't a "black swan"--the assumption of normally distributed stock returns was just a mistake.
B. Double Counting Credit Risk
Many analysts trying to value collateralized default obligations are double counting defaults. They simulate defaults on the underlying collateral and take the defaults out of promised cash flows, correctly calculating (usually on a monte carlo basis) cash flows in each scenario. The error comes in then adding a "credit spread" to discount these cash flows. Sometimes this is called a liquidity premium instead, but that doesn't make the error less serious. State of the art risk neutral valuation, like the Black-Scholes options formula embedded in GAAP accounting principles, calls for discounting at the risk free rate. This is "the new way" of valuation, using my partner Bob Jarrow's description. The "old way" is bond math--assume that all scheduled payments will be made as promised (not taking out credit losses) and raise the spread over the risk free rate until the net present value of the bond equals its market value. Both the "old way" and the "new way" can lead to correct valuations, but one should not make both adjustments--taking credit losses out of cash flows AND raising the spread for credit risk. It's one or the other. The objective is the highest quality "mid market" valuation, say 68% of par value. If the bid offered spread for this security is so wide--say 78 offered, 58 bid--that no trades take place, that doesn't mean that the proper value is 0 or 58 bid. 68 is still the right price.
C. Not Simulating the Factors Driving Correlated Movements in Defaults, Spreads, and Collateral Values
My book Advanced Financial Risk Management, with colleagues Kenji Imai and Mark Mesler, covered this in detail. Home prices, interest rates, commodity prices, stock indices, foreign exchange rates, and commercial real estate prices are the kinds of common risk factors that drive default probabilities, spreads and collateral values. To do valuation and risk assessment without explicitly taking these into account can lead to serious errors. These errors are most serious in the super senior tranches of CDOs because otherwise the "fat tails" in credit losses are underestimated and the values of super senior tranches are over estimated.
There are many more technical errors that can be made, but these are probably the most important. For more on any of these topics, please contact us at firstname.lastname@example.org and cite this blog in your comments!
Donald R. van Deventer
Honolulu, March 27, 2009. Updated April 1 and April 2, 2009