Deutsche Bank’s $12 Billion in Hidden Losses: Why Whistleblower Charges Have Merit and Why They Matter

I’ve read the Department of Labor complaint of former Deutsche Bank employee Eric Ben-Artzi that, among other things, alleges that the German bank violated SEC, GAAP and IFRS (International Financial Reporting Standards) requirements and probably also Sarbanes Oxley, and have had follow up conversations with him and his counsel. The charges are specific and sufficiently well supported to merit serious investigation. In their efforts to dismiss his charges, defenders of the bank’s position overlook the public reporting requirements for complex financial transactions like the one in question, a leveraged super senior (LSS) trade. Nor does their Panglossian “everything worked out for the best” argument stand up to scrutiny.

Deutsche’s CFO issued a statement on Thursday justifying the bank’s actions; we’ll address that after providing some context.

Overview of Allegations

Lost in the fulminating about these allegations, and the likely reason that the Financial Times gave this story such prominent play is that not one, but three whistleblowers came forward with overlapping concerns. All these individuals had to recognize that going this route would be at best a career-limiting move at Deutsche and would likely result in being fired.

Deals that did not have three (or any) whistle blowers:

AIG CDS
Timberwolf CDO
Hudson CDO
Magnetar CDOs
Dead President CDOs
Taylor Bean and Whitaker fraudulent conveyances
London Whale
LIBOR rigging
MF Global Euro trade scam

A former financial firm exec said via e-mail:

Much (though not all) of what we define as the financial crisis happened without any whistleblowing at all. When asked about their questionable behavior in front of committees or judges, the bankers and industry insiders say that it’s just the way the world works – immoral, but not illegal. It’s part of why bankers react so strongly against Obama and OWS – they feel they are now being persecuted for stuff that everybody was doing and nobody – that they knew – thought it was wrong or illegal.

For 3 employees in a bank, 2 of whom were pretty senior, to speak up about the same deal, something unusual was happening.

In fairness, some of the pushback is due to the fact that the Financial Times’ lengthy and otherwise impressive discussion of the transactions and the whistleblower charges failed to discuss how Deutsche Bank’s alleged misbehavior fell afoul of accounting and regulatory requirements. We’ll start with Ben-Artzi’s allegations, not simply because we have the most detail on them, but also because they constitute the overwhelming majority of the alleged hidden losses, $10.4 billion of the $12 billion at issue.

To give a very brief and simplified overview (see this post by Lisa Pollack for more details), the so-called LSS trade involved the creation of a special purpose vehicle. The asset side of the SPV was a portfolio of corporate credit exposures, achieved through credit default swaps.* The SPV was leveraged because it was not fully funded.

Normally, in a CDO (and this structure was a CDO), even in a fully synthetic deal (all the assets are CDS) you have someone (usually multiple someones) lined up to take the losses if the thing blows up completely. And usually, the investors who take the riskiest exposures put up cash so that the SPV has collateral and does not have to chase investors to fund the CDS payouts until the losses exceed the level of the posted collateral. But in a LSS trade, the investors put up only 10% of the notional amount.** This was non-recourse***; if the deal lost more money than that, too bad for the protection buyers.**** Now these deals did have certain triggers that would allow them to be unwound to contain losses, but it’s not hard to imagine that if one LSS deal needed to be unwound, probably other did too, and the pressure of selling would exacerbate losses.

On top of that, there was a funding mismatch. The 10% collateral was provided not through the issuance of bonds or notes that matched the maturity of the credit default swaps (5 to 10 years) but commercial paper sold to Canadian investors that had to be rolled every 30 to 90 days.

Deutsche was the protection buyer in this deal. Even though it might turn around and lay that risk off to customers, it “faced” the SPV. It would not sell the CDS, it would enter into offsetting trades (and the offsets were often not exact, but that’s not germane to the regulatory discussion). So Deutsche was exposed if anything went wrong with this scheme.

It isn’t hard to see that the biggest risk in this trade is that the losses exceed the 10% that the investors ponied up. That risk was called the “gap option” or gap risk. That is also the risk that Deutsche undervalued significantly or simply failed to include at all in its financial reports (more on that shortly). Because this trade was $130 billion notional and was the largest exposure in the bank’s trading book during the 2007-2009 period, the undervaluation was significant, to the point where incorporating those costs would have raised questions about the bank’s stability during the crisis. Ben-Artzi pegged the value as $10.4 billion lower at the worst point of the crisis, when Deutsche’s equity was €32 billion ($44.5 billion) at the end of 2008.

The main argument of defenders of Deutsche’s case is that this trade was properly classified as a Level 3 asset. The Level 3 classification was called “mark to model” or “mark to make believe”, in that modelers could use “unobservable inputs” in coming up with their valuation. But as the filing details, Deutsche went far outside the permissible boundaries for Level 3, and not in one way, but several.

In its public filings, Deutsche told investors it complied with GAAP and IFRS, so material violations of these rules would be securities fraud, or more specifically a violation of Rule 10 b5 and Section 17a of the Securities Exchange Act. It’s hard to see how disappearing an exposure that is over 20% of your equity isn’t material. And it proved to be a real risk, in that the entire Canadian asset backed commercial paper market froze, leaving various foreign banks, including Deustche, that had set up these SPVs arguing with investors that they weren’t obligated to honor the “liquidity provider” provisions of these deals. The refusal to repay the Canadian investors led to what we called “the mother of all Canadian restructurings.”

According to IAS 39, compound derivatives transactions like the LSS trade are to be accounted for as a single transaction. Thus failing to include the gap option at all in its financials (as Deutsche did at the outset and during considerable periods of 2008 and 2009) was improper; you can’t cherry pick certain flattering pieces of a complex transaction and omit ones you don’t like.

And even though a transaction is Level 3 (mark to model), that does not mean you can just slap numbers on it willy nilly. Level 3 assets are still a form of mark to market; that means the model has to comply with specific requirements.

Here is how the filing says Deutsche accounted for the gap option in LSS trade:

1. Deutsche tarted doing LSS trade in 2005 without a methodology or model.***** Initially they just haircut the trade to allow for the gap risk

2. They knew the haircut to be wrong, so they developed several models and put one into production. That one was yanked after a short period because senior management saw that it wouldn’t let them do the business in any size. As Colin Lokey pointed out (hat tip Francine McKenna):

This action was apparently taken after the firm determined that modeling the gap risk was “economically unfeasible.” In other words: “Yeah we modeled it and it turned out the losses would be huge, so we scrapped that model.”

3. Deutsche then established a reserve instead

4. The reserve was still in use when Ben-Artzi joined. Even then (late 2010-early 2011, well past the acute phase of the crisis), he estimated that the reserve undervalued the actual exposure by several hundred million dollars, a material amount

So consider: strategy 1, haircutting the trade by 15%, was recognized by the bank as being not kosher. Both GAAP and IFRS prohibit taking a finger-in-the-air discounts; IAS 39 requires firms to establish fair value even when no active market exists and to make the most use of observable inputs and minimize reliance on entity-specific inputs. It also stipulates that the model be benchmarked against market data, and any available information from historical or comparable transactions, should be used in developing the fair value estimate. The bank later tried pretending the gap risk was included in a reserve for its entire derivatives portfolio (point 3 and 4), which Ben-Artzi believes was only $1.5 billion at its maximum versus the $10.4 billion he calculated as the maximum exposure. The use of a global reserve would also fall afoul of the requirement both IAS 39 and GAAP to account for complex derivatives on a fair value basis.

Deutsche also included its collateral pool in its banking book and thus reported it on a hold to maturity basis, rather than including it in its trading book. Again, that’s impermissible; it violates the “form over substance” rule and enabled the German bank to hide the leverage of its trades.

Now if you think this is all caviling, have a look at paragraphs 386 through 392 of this suit. In short form, Barclays enter into two LSS tranches as a protection buyer. The notional amount was $6 billion and the trades were levered ten times, like the typical Deutsche bank transaction. That meant the investors had put up $600 million in collateral. Barclays tried to restructure privately and when that failed, the unwinding wound up in court.

On January 13. 2009, Barclays valued the tranches at $1.02 billion and also calculated $185 million for unwind/rehedge costs, producing an overall loss of $1.2 billion.

In paragraph 473 the judge confirms that all Barclays gets is the initial $600 million of collateral, and not any of the premium or liquidity payments made later.

The value of the gap option is equal to the exercise value as of this date plus the time value to maturity (more or less the value of future exercise). It is therefore AT LEAST the difference between the value of the tranche on January 13, 2009 and the collateral, which is $600 million.

That $10.4 billion figure used buy the Financial Times was based on valuing the gap option at 8%, which was the high end of the Ben-Artzi’s estimate for the value of the gap option as 3% to 8% of notional. That appears to be conservative, since the value in the Barclays case in January 2009 was 10%. Thus that means that the hidden losses for the gap option alone were as high as $12 billion.

Note that this is far from the total of the whistleblower charges. Matthew Simpson, a senior credit correlation trader, charged Deutsche with mismarking the value of trades, exaggerating the margin between where protection was bought and where it was sold, which in turn led to bigger reported profits and hence trader bonuses, along with other securities law and accounting violations, including the failure to value the gap option properly. An even earlier former complainant charges the bank with an even longer list of violations and included the gap option among them.

The CFO’s Defense

It’s pretty surprising that CFO Stefan Krause could duck away from a criminal probe of carbon credit-related tax evasion to address the whistleblower charges. Krause’s defense boils down to:

1. When an earlier whistleblower brought this matter up (Matthew Simpson) we hired Fried Frank, a law firm, and had a former SEC prosecutor representing us. We did a really big investigation, spent a lot of time with the auditor KPMG and then presented our findings and millions of documents to the SEC and the New York Fed. They seem fine with what we did

2. We haven’t realized any losses

The problem is that some parts of the discussion undermine his claims, at least as far as Ben-Artzi is concerned.

One troubling tell is the “millions of documents” to the SEC and Fed. Providing a monster data dump is often a strategy of obfuscation.

Second is that the most detailed substantive (as opposed to procedural) discussion is of a modeling issue with Simpson that had nothing to do with the gap option:

The trader made several allegations about the methodology used to value position in the bespoke book and the sufficiency of collateral end reserves…

In addition, shortly after the market risk employee raised this allegation and was fully debrief by Fried Frank, he was given the opportunity to meet with senior finance and market risk management personnel to explain his concerns. He was also given the opportunity to present an alternative valuation approach to the one taken by the bank which he declined. The market risk employee has also given the name and contact information for one of the SEC attorneys responsible for the ongoing investigation.

Based on this extensive investigation, Fried Frank came to the following conclusions. Regarding the valuation model, Fried Frank investigations found that the model the bank used to value the bespoke book was a customized, proprietary model design to value the unique composition of that book. This model has served no limitation which were exacerbated as the market began to deteriorate during the economic crisis. These limitations and limitations of other models were known and openly discussed at the bank. While the bank developed, tested and implemented alternative models that would overcome the productions model limitations, the financial impact of these limitations was addressed through the taking of reserves.

Moreover, KPMG, the bank’s external auditor was aware of the model’s limitations. KPMG was involved in and appraised of the bank’s continuing efforts that ultimately led to a switch to a different model in early 2010.

If you parse this carefully, there are two issues, the valuation of the “bespoke book” and the collateral reserves. If you remember, the latter was one of the not-kosher ways for dealing with the gap risk. And the discussion excerpted above is about the former, not the latter.

Ben-Artzi also believes the model changes mentioned in 2010 relate to a correlation model, specifically the model for the underlying bespoke tranche. Apparently it was valued using an old model that couldn’t match the market price of senior CDX tranches (a corporate CDS index) during the crisis. It was called the “5 Factor Copula Model”, and was replaced by the “Base Correlation Random Recovery” that matched market prices. Note that it very well be that the old model inflated the values of Deutsche’s underlying tranches, but there is no way of verifying this independently. Ben-Artzi believes that the transition to better models that could calibrate to the market took place a couple of years earlier at most other banks And most important, Ben-Artzi did not include this issue in his complaint.

Here is the part that does relate to Ben-Artzi’s issue, the gap option. Bear in mind that every accountant I have spoken to has said the use of reserves as a fudge for valuing the gap option is unambiguously impermissible under GAAP:

Regarding reserves, Fried Frank found that the reserves for the business were determined using agreed upon methodologies that were modified and adjusted as market conditions deteriorated. The reserve was sizeable and included amounts for each of the specific valuation issues identified by the trader. The issues surrounding the reserves for the book were widely known and debated among the relevant control groups and the trading desks. And the actual reserve numbers went through several layers of review before being finalized and booked.

Notice how weasley this is. “Agreed upon”. By whom? “Sizeable” is not necessarily adequate, particularly when you are talking about the biggest single trading book risk in the bank. The valuation issues were identified by Simpson, not Ben-Artzi, hence there is no assurance Ben-Artzi’s issues were addressed. Contrary to the impression of openness (“the issues….were widely known and debated”), Ben-Artzi had to do considerable digging to get the information he needed and reports that he got inconsistent answers from differing groups in the bank.

And consider: if the bank had done a thorough investigation and was confident of the correctness of its actions, why was it so secretive with Ben-Artzi? As part of his job, he needed to have access to historical information, including the reporting of the position. A compliance officer at one of the other five banks that was in the LSS trade in a meaningful way said that Ben-Artzi operated in exactly the way he’d expect someone hired in that sort of role to behave.

Yet it took Ben-Artzi months to find out that stress testing was used in place of valuation, and he had no idea an investigation had taken place prior to his arrival at the bank until he called the employee complaint line after getting conflicting answers from different groups. It isn’t at if Simpson’s concerns about the trade were a state secret; his firing, as well as the opening of internal and SEC investigations regarding his complaint about trading irregularities, were reported in Reuters. You’d think if the provided provided the ringing exoneration that Krause said it did that the bank would have made that clear internally.

As for the “no losses have been realized,” this is real double-speak. Remember, we told you earlier that the collateral loans are being carried inappropriately in a hold to maturity book. There is still about $2.4 billion in the Montreal Accord pool, and any losses would occur there first. My understanding is they’d only be impaired (written down) in case of credit deterioration, when the LSS trade underlying falls in value for other reasons. Super senior tranches (and the gap option, which is close in value to a super senior tranche attaching at the collateral level) are very time sensitive, and decay quickly when losses in the portfolio aren’t realized. While the bank did unwind some trades in 2010, it’s almost certain that they’d unwind positions that had no or only small losses first. Thus successful unwinds to date don’t prove the rest of the positions are healthy; in fact, the failure to have unwound the entire book by now suggests the reverse.

Why It Would Have Been Better if Deutsche Bank had Been Forced to Mark Its Books Correctly

The Financial Times pointed out that if Deutsche had been required to mark down its positions to the degree that Ben-Artzi and the other whistleblowers suggest, it would have lowered Tier 1 capital by more than €8 billion, which would have also put the bank’s capital below the 8% minimum required by German regulators. Felix Salmon has argued that we were all much better off to have swept this problem under the rug than risk a destablizing bailout. That logic is the justification for any and every big-bank-coddling measure, from refusing to indict HSBC or any employees for criminal money laundering, to using borrowers to foam the runway and attenuate housing-downturn-related losses.

In fact, as Francine McKenna points out, Deutsche Bank was bailed out on the sly by the Fed through the use of special lending facilities.

In addition, Bank of America was bailed out in January 2009 to the tune of $20 billion more from the TARP than its initial borrowings in October 2008, plus $118 billion in asset guarantees and the world did not end. UBS got a large and complex bailout (and remember, Switzerland is not a very big economy) and the world did not end. Deutsche would not have slipped much under the minimum capital ratio. A comparatively modest $5 billion to $10 billion capital infusion would have been adequate.

Moreover, the effects would have been salutary. On Thursday, FDIC vice chairman Tom Hoenig in a Financial Time op-ed highlighted the capitalization ratios of the global systemically important banks. Deutsche is the weakest, with a capital level less than half the average. And German regulators would be more likely to impose the UBS rather than the Citigroup treatment on Deutsche. UBS served as a wake-up call that banks were running risks that were outsized relative to what the real economy could rescue in a worse-case scenario. The Swiss National Bank was the only banking regulator to require a bailed-out bank to conduct an independent investigation of why it screwed up and publish the finding. It is also imposing tough capital ratios on its banks, an elsewhere unheard of 19%.

And let’s get to the basic issue. The LSS trade, if properly modeled and accounted for, would never have been done in the volume that it was and Canadian investors would have been spared at least some of the disruption and losses that they suffered. But the traders and their managers were able to pull out more in bonuses by booking the trade in a way that exaggerated its profits. This is looting, pure and simple. Tough accounting and controls won’t catch all bad or sloppy behavior, but it would have stopped this trade, and you can be sure it would have prevented others.

––––––
*These CDS were not 0-100%, but specific attachment points, typically 10% to 70%, making this deal a single tranche CDO.
** This was funded by commercial paper, so the mechanism was that the SPV would issue commercial paper in an amount equal to 10% of the notional value of the LSS. So if the LSS had $150 million of exposures, the CP issued would total $15 million. Even though I used the term “cash”, the SPV would buy notes.
*** The transactions had certain triggers that in theory would get investors to pony up more money. Deutsche had no illusions on this front. Several internal presentations (included as exhibits to the Department of Labor complaint) describe the transaction as “non-recourse” with “non-recourse” boldfaced.
**** You can see why this is a dumbass trade. It’s hard to imagine it would make sense, with the risks properly estimated, in anything other than unusual circumstances, when risk pricing was out of whack.
***** This is pretty stunning. It’s one thing to kludge a trade or two because you’ve done something exotic and the customer was so dumb you are pretty confident you loaded enough margin into the pricing that you’ll be OK under most circumstances. It’s quite another to move a strategy into production without having sorted out how you will hedge the trade (the hedging strategy determines how you price it). And remarkably, they tried hedging a credit position with….equities! We’ll turn the mike over to Colin Lokey:

Even more astonishing is what the bank apparently did to hedge against this faulty hedge — yes, that’s right, this is yet another example of a hedge of a hedge gone wrong. From the Financial Times:

“A person close to Deutsche Bank says the bank hedged its exposure by betting that the S&P 500 stock index would drop.”

That’s it?! You have tens of billions in theoretical losses on a complex credit derivatives book to protect against and you bought yourself some S&P 500 puts?!