Spero Consulting

Archive for November 3rd, 2008

Behind AIG’s Fall, Risk Mod­els Failed to Pass Real-​World Test.

You Don’t Say! Our sub­ti­tle is the title of today’s Wall Street Jour­nal front-​page arti­cle about AIG (obviously).

As always we’ll point inter­ested read­ers to our essay, Uncer­tainty Man­age­ment, which empha­sizes the broader notion of unmea­sur­able uncer­tainty over the nar­rower notion of (mea­sur­able) risk, and there­fore per­mits really bad things to happen.

We mean bad things out­side the scope of someone’s purely math­e­mat­i­cal model, which, as an abstrac­tion of real­ity, may ignore imag­in­able and unimag­in­able bad things. (We’re all for math – when it is thought­fully and con­sci­en­tiously applied. In fact, we think such appli­ca­tion is one of the things that we do best.)

In that regard, we’ll once again note the sub­ti­tle of the above-​referenced essay, Or How Trad­ing is Like Play­ing in a Cul­vert on a Hot, Sunny, Summer Day. See dear reader, once one con­sid­ers that one could drown from a flash flood – even on a pre­sum­ably and locally Sunny day – the allure of such adven­ture dulls greatly – at least for the rea­son­able among us.

In other words, your mother may have been a scold, but there was prob­a­bly a good rea­son for her to warn you about play­ing in cul­verts and drainage ditches (pro­vided that she loved you, of course). She may not have dis­cussed it in prob­a­bilis­tic terms, but that doesn’t mean she can’t recall read­ing about such drown­ings, say, forty years ago, or even before you were born.

More­over, the fact that you didn’t read about any such cases in, say, the past ten years, doesn’t mean they don’t exist, and there, of course, lies the Prob­lem of Induc­tion, and the over-​reliance on infer­ences from rel­a­tively short-​duration, historical, data sets. (See our beau­ti­ful excerpt from St. James’ only Epis­tle on our Quotes page.)

The prob­lem, dear reader, is that few senior man­agers (and almost no board members) understand the val­u­a­tion and risk mod­els used for secu­ri­ti­za­tions, and many of the traders, con­sul­tants, and ana­lysts who wield such tools often suf­fer from, what one may call, “fram­ing” issues; we don’t mean that aspect of home con­struc­tion despite its recent relevance.

We mean that if one’s only tool is a ham­mer, then lots of things look like nails. The metaphoric ham­mer may be an intan­gi­ble Visual Basic or “C” pro­gram­ming algo­rithm, but the point remains the same; it’s just harder for senior man­age­ment to see what one is pound­ing in their cubi­cle, office, or trading-​floor seat.

To be sure, if any­one within most of the larger firms would have com­plained of the sys­tem­atic risk – and how every­thing could go bad all at once – and the inap­plic­a­bil­ity of the stan­dard mod­els, which gen­er­ally don’t per­mit such events, then that per­son most cer­tainly would have been told that they don’t know what they’re talk­ing about. Pos­si­bly, that they are unso­phis­ti­cated or too negative.

Per­haps we just don’t pay enough atten­tion to what hap­pens in all of the large firms, but if the reader dis­agrees with our pre­ced­ing para­graph, please note that there have been few recent suc­cess sto­ries within major firms like the gains enjoyed by Nas­sim Nicholas Taleb, John Paul­son, or Andrew Lahde–all inde­pen­dent fund man­agers. (If the new reader has read this far, then it is highly likely that they’ll like the link under Andrew Lahde’s name and his con­dem­na­tion of many things in one fell swoop.) We know that our exam­ples form a very small data set, but mostly what we’ve heard is how the more suc­cess­ful large firms haven’t lost as much as their brethren. We don’t recall any of them actu­ally do well this year.

Also, we’ll prob­a­bly have more to say about our boy, Taleb. We very much like his trad­ing style, as it reminds us of the value of the Sec­ond Amend­ment and laws that per­mit con­cealed carry. See, dear reader, carrying a pis­tol is very much like buy­ing deep-​out-​of-​the-​money puts. There’s a small, ongo­ing cost and a minor irri­ta­tion, but when cer­tain bad things hap­pen, there is an option to exer­cise to pro­tect ones self, and that value can­not be underestimated.

We haven’t said any­thing about CDS – the source of AIG’s prob­lems – in this post but plan to do so shortly.