Hacker News new | past | comments | ask | show | jobs | submit login
The math that killed Lehman Brothers (maths.org)
70 points by vaksel on June 30, 2009 | hide | past | favorite | 31 comments



Note: In a prior life I was a part of the CDO research group supporting the clients at Lehman's trading desk in NYC. I'm familiar with what happened there.

The article contains a few errors. One, the article claims Lehman owned lots of senior tranches, which lost value when default rates unexpectedly rose. That's incorrect. We could sell that stuff to pension funds and make money on it, so we did, and losses in that tranche didn't hurt us; it hurt the retirees.

Lehman got hit hard because they could not sell the highest risk tranches (referred to as "toxic waste"). But they made enormous fees on originating the CDOs, and the toxic waste paid high interest rates (20-30%), so Lehman reluctantly kept it on their balance sheet and constantly tried to sell it off to other people (most of whom were too smart to take it).

We tried to hedge the risk by shorting the mezzanine tranches (those were between the toxic waste and the senior AA or AAA tranches). If default rates went up, and we hedged correctly, the gains from the synthetic shorts would make up for the losses on the toxic waste subs.

But for a variety of reasons this wound up not working as well as expected. Particularly, the correlation of default rates between subprime and quality debt decreased substantially and unexpectedly, which means these shorts did not generate as much cash as was lost when the equity tranche tanked.

Also, for an amusing tangent, we had bomb-proof windows to the outside world on the trading floors. A fertilizer bomb could go off in Times Square and we'd still be trading. Unfortunately we were not so well protected from the bomb that went off inside the trading floor.


For an amusing tangent to your amusing tangent, Blizzard's offices have bullet-proof windows and doors to prevent disgruntled World of Warcraft players from going on shooting rampages.

(http://pc.ign.com/articles/662/662143p1.html)


I have a hard time reading an article about why CDS instruments are to blame when the author doesn't know what they are. CDS = Credit Default Swaps, not Credit Derivative Swaps. All it takes is putting CDS into Google, five seconds tops.

I also take issue with the fact that the unconcern for subprime risk was the insurance--that ignores the risk model they had that showed real estate value going up. They didn't care if subprime defaulted because they'd get the property which would be worth more than the note and they'd make even more money. They knew the gig would be up if the CDS had to be executed en masse, they just didn't think that would ever need to happen. It's not that they drove drunk because they had insurance, they didn't know they were intoxicated (only had one drink, I swear!).

So much for child prodigies.



No, that just says that a credit default swap is one type a credit derivative contract. Derivatives can have many triggers, default is the one talked about here.


this article leaves more questions than it answers. how does the author know the contents of an "investment plan" that was on the CEOs desk? where does the 3% number come from that is the bulk of the assumptions about the housing market.. and more than a few others..

even worse though, its flat out wrong in other areas.. lehman absolutely knew the risk of its CDOs and in fact was trying to offload them as quickly as possible. in contrast to goldman, lehman was famous for making money packaging and selling its bonds on a fee based model. whereas goldman's model relied more on proprietary trading - actually making bets on the direction of bond moves, lehman would theoretically be profitable regardless because they were making fees from the beneficiary of the bond and taking a cut from placing the bond with customers. the problem that killed them was the turnaround time from when lehman received the assets and "commission" to create the bond and when they could offload it to customers. during this time the bond was actually on lehman's books. basically, wall street caught lehman holding a bad hand of these bonds before they could unload them to the rest of wall street. rumors were started just like they were for bear sterns and financing got pulled like a house of cards..


'math' killed? or stupid assumptions made by the analysts ?


I'd say "stupid assumptions" assuming that the analysts cared about the company at least as much as about their own pockets.


In English, however, the noun mathematics takes singular verb forms. It is often shortened to math in English-speaking North America and maths elsewhere.

http://en.wikipedia.org/wiki/Mathematics#Etymology

The author goes to Oxford.


I was not talking about the usage 'math' or 'maths'. I was saying, it is not the fault of math/maths when we make wrong assumptions like 'house prices can't go down' or 'prices will always increase' etc. GIGO.


In the early 1980s in California certain prominent second mortgage lenders filed bankruptcy after advertising long and loud that "you can't lose on California real estate."

In theory, when you invested, you loaned your money directly to a California homeowner secured by a deed of trust.

In reality, given that the borrower was paying 15 points on the loan, plus 24% interest, the loan hardly fell into a typical "secured" category. These borrowers were in fact taking out 3rd, 4th, and 5th loans on properties whose only "equity" consisted of paper value generated by mass phony appraisals done by or in coordination with a whole set of players who had a stake in keeping the game going.

When it all collapsed, people were observing how naive all those investors were to invest in something that sounded too good to be true.

It looks like our Wall Street wizards fell victim to the same impulse. This led to the bad assumptions that fueled their greed and to their ultimate demise.

Nothing new under the sun, said Solomon. It is all too true.


Large gains always means you are assuming a large risk. No one has yet to come up with a way of removing (or greatly reducing) risk without also greatly reducing the gains.

Ignore large gains at your own peril.


Duly noted, and I agree.


Thanks, and here I had been thinking that "the maths" was a neologism like "The Interwebs".


Glad I'm not alone on that one.


Blind reliance on models killed Lehman Brothers. Math was just the tool they used to inflict lethal injury on themselves.

I have also seen strong arguments (but no concrete evidence yet) which suggest the government officials involved were pro-Goldman and anti-Lehman.


Dick Fuld was never 'one of the boys': Lehman had a tortured past, and grew up after its spinout from American Express as a bond house, not an elite investment bank, and only began rising in the M&A league tables in the 2000s. Until then they made money and were good but never had the social status of the GS execs.

To the other commenter: Lehman was too terrified of its own capital base to participate in the LTCM bailout, but yes they didn't win any friends with that.


I heard that too (again without proof) - that Lehman (& Bear Stearns) refusal to participate in the LTCM rescue still rankled with other market participants.


The article does little to show hows maths brought down Lehman Brothers. Rather, it shows how important assumptions about the future (and your uncertainty surrounding those assumptions) are for structuring financial contracts.


"Therefore, investing on CDO is a riskier choice than betting for Manchester United."

Article is obviously watered down to make it more digestible, but how he came up with such flawed logic on this example is beyond me.


I think the point he didn't make very well was that CDOs and the direction of the stock market were correlated and most CDO participants also were stock market participants. Manchester United's chance for victory is presumably uncorrelated with either of those, so if you were a stock market participant who also bet on Manchester, you were better off than if you were a stock market participant who bet on CDOs.


I agree.

I think his point was pretty clear, but his subsequent conclusion regarding the risk factor just seems completely nonsensical to me -- I mean, you could bet on any not-correlated-to-the-stock-market-event, regardless of its probability or odds factor, and it would have yielded less risk than holding CDOs as per his assessment. He's just talking about diversifying, but with the assumption that a stock market crash / economic crisis is imminent.


Ironically, he made the same assumptions that killed many CDO players- that past data was sufficient to predict future data.


"Plus" is a magazine for school pupils, largely.


Rather minimal "maths". There's no mention of the pricing models used for CDS and CDOs. That would have been more interesting.


I agree. And allow me to suggest this paper: http://arxiv.org/abs/0809.1393

It's not everyday that you see an algebraic geometer writing on CDOs ;-)


I love the candid bio at the end of the article.


The same thing happened to Long-Term Capital Management. You can read up on it here:

http://en.wikipedia.org/wiki/Long-Term_Capital_Management

They were bailed out when they failed too so that set the precedent for the bailouts we had recently.


An excellent book on LTCM is 'When Genius Failed' (not an affil link):

http://www.amazon.com/When-Genius-Failed-Long-Term-Managemen...


to be more specific, LTCM set the precedent for Fed organized bailouts.. private bailouts have a long history, going back at least to JP Morgan


Excellent article, but the notion that we should believe that idiocy was at the heart of this seems quite wrong to me. At best, willful idiocy.

It seems more likely to me to fall under the Wall Street koan, "You'll be gone, I'll be gone", rather than bad math.




Consider applying for YC's Summer 2025 batch! Applications are open till May 13

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: