Buffett’s “$100 million Japan problem” with Lehman was probably not “Repo 105”

I’ve been reading the report of the Lehman Examiner, Anton Valukas, at the Jenner and Block firm website.

On a project basis, I worked in Lehman Japan in the middle of 2007, helping to determine the nature of a $25 million “hole” between what the senior management there thought income should be, and what they were actually able to show on legal books.

Even though this was haken employment, so to speak, (which meant I should not have had to go through an interview process) there was a small series of interviews, and looking back, I don’t remember being in one where the interviewers were so singularly nervous and trying to hide it at the same time.

What management told me was they thought there was an error in the computer system that accounted for the cash movement of repos, which in most Japanese financial firms is the “I-STAR” system. (Later, I learned that this theory actually came from a rival clique within product control that only knew the Lehman internal management accounting system, whose acronym escapes me.)

Usually, I don’t like to give suggestions during interviews, because I’ve had one or two along the way where the people merely wanted free professional advice on an issue. So to set it up that I should “prove” something about my skills, they basically hand me some work issue of theirs that they know jack shit about.

I get this a lot from accounting headhunters here in Tokyo, too, by the way. They used to make lots of noise about “Sarbanes Oxley” as if internal controls did not exist before 2002. Nowadays, the noise sounds like they are saying “Ay fers”, by which they mean International Financial Reporting Standards. (They imply that US GAAP trained people like CPA’s wouldn’t know these.) But it’s clear a lot of these headhunters, in fact, know very little about it, and so the sound “ay fers” coming out of their mouths now and again sounds like a fart. It might as well be.

But that is all an aside — back to Lehman.

I gave the senior guys (two of whom were definitely younger than me) my best guesses as to what was wrong with their interest expense. I did this because these were “gimmes”–if they hadn’t checked these first, it would be difficult to imagine how they find the office in the morning.

They said interest expense was too high. So:

1) Did your volume of borrowing go up in the last several quarters? (Their answer: Yes, it did, but not enough to create such a wide swing.)


2) Has the interest rate risen on the borrowing? (Again, they said yes, but not enough to create the big number.)

So taking this theory of how the computer system was out of whack, I went through the transactions from mid 2006 to early 2007, and discovered the obvious facts that:

1) when you borrow in yen, and the rate goes from 0.05% to 0.5%, your interest cost is actually ten times higher than it was. Yes, the interest rate itself is still really tiny. But your bill from the lender is going to jump way high; and,

2) if you double your borrowing, you will probably double your interest costs.

Combined, those two factors were creating a 20-fold increase in interest costs to the unit on their borrowing.

I spent only a week on that theory (about the “computer malfunction” causing a problem), which didn’t make the one faction happy, and left the management with no answer. So, thinking about it at home, I concluded something different: it was yen-carry trading. Lehman Japan was using securities as collateral to obtain loans in yen. They would then sell the yen and invest the money overnight or for very short terms of a few days in overseas currencies.

Since the yen could be borrowed for 0.5%, and at the time the U.S. dollar overnight rate was 5% (remember that?), the 4.5% difference in interest would go to Lehman. They would have to manage the currency risk, of course, but as I saw it, yen carry would explain why a company was borrowing money on the short-term and paying it back a few days later.

So the next step would have been to figure out who gets the money. Where does it go?

(But instead the next step appears to have been several days of internal politicking as to whether or not the financial system had a “computer error” that was causing the interest rate paid to be misstated.)

I spent several more weeks digging into the I-STAR general ledger and something Lehman called GFS (Global Financial System), which was a piece of crap. The financial records could clearly show that the yen borrowing was mostly being done for Lehman Brothers International Europe (LBIE or “Libby” as it was called internally). LBIE got the yen funding, and then returned yen back several days later.

To secure the funding, Lehman Japan was either entering into repo trades on its own account, or brokering a repo for an affiliate. These repos were not “Repo 105”, but the vanilla kind that any firm who wants to hold an asset yet have liquidity will do to raise cash. (I’m pretty sure of this!)

What was clear was that the expense to do this, particularly the interest expense of the large borrowings, was sitting on Lehman Japan’s books. But the gain overall Lehman made, or whatever value was being had with the money—or even a charge representing an arm’s length payment for Lehman Japan providing the yen lending facility, was not on the books. So it was a transaction in one direction on the income statement–Japan was getting hit with expense, which wasn’t really it’s own.

This conclusion was equally controversial, and brought yet another internal accounting system into play, the one whose acronym escapes me. It was a management accounting tool that would allocate return on trading or investments. Even though it had a feed into the “GFS” and the “I-STAR”, it seemed to lose some of the data, and sometimes run supplemental formulas that no one could explain.

Each small group had their own little computer system that they claimed was valid. But nobody could make numbers tie or even explain why numbers changed. Around the time I realized this, I also concluded that very few Japanese understand General Ledger, accrual accounting, or other financial reporting concepts. The expat talent around this town was not much better at it, but they had a more flexible style of B.S.

I stayed with that project for about four months, until the time that now we all know is when Lehman started to meet its real capital crunch (August-September 2007) that would inevitably relegate it to history or infamy.

Were Repo 105’s going through the Japan desk? Yes. But as the Examiner’s report states, all the Lehman affiliates were running Repo 105 through the exact system as all the other repo trades (whether or not all the formalities of arms-length were there, or not.)

Like I’ve said, I can’t help but conclude that the real problem in that product control unit was the lack of accounting talent and analytical ability —plain common sense!—to figure out what is going on in a business. I’m still surprised that they didn’t know the fundamentals of the basic, so-called vanilla, trading. Much less which trading would be separated at higher levels and misreported to the public in a sizable fraud.

According to news accounts, Warren Buffett was concerned about some $100 million Japan “problem” that Dick Fuld had not told him about when Lehman was looking for capital in March 2008 (This is page 666 of the Examiner’s report, by the way. Volume two.) Now that Lehman Japan was known to have a problem, I wonder if it wasn’t the legitimate repo financing that nobody seemed to understand. I imagine that at the rate it was going, the $25 million would have become $100 million by March ’08. And that the people there still couldn’t figure it out, but “knew” that “their system is right!” Masters of bullshit and arrogance—MBA.

Valukas’ report is 2200 pages, and it is extremely well footnoted (several thousand footnotes). Maybe I’ll get a chance to focus on other pieces later this weekend.