One of the pleasures of reading the American Scientist in the last decade or so has been the writing of Prof. Henry Petroski, the engineer and historian of technology currently on faculty at Duke University. Petroski used his columns in serialized fashion to lay out his since-published theory of “success through failure.”
By this term, Petroski means that as engineers innovate — for example, using new materials or methods to push the envelope of what it is possible to design and build — they inevitably make mistakes by relying on models (tangible or numerical) of previous technology to assess their designs’ reliablity and safety. Often only after catastrophe strikes — perhaps a structural failure of some kind like a Tacoma Narrows Bridge or a Kansas City Hyatt walkway — does the profession look backward and ensure that its design and test tools are updated to match the reality of leading-edge technique. There are certainly analogues outside civil engineering, as anyone can verify who has followed Intel’s search for insanely subtle chip bugs. In any case, Petroski suggests that anticipation and avoidance of catastrophic failure is possible, but only if one understands the hubristic risks of conventional design and thinking.
Well, now. We are presently in the midst of a catastrophic failure of financial engineering. Financial engineering is a construct invented in academia to recognize that the term “finance,” used historically for an essentially analytical discipline, practically a branch of applied economics, was no longer adequate to describe the activities of Wall Street in designing and delivering to the marketplace new financial instruments. Engineers design things. They have quantitative skills (e.g., great attention to detail, facility with simulation modelng, capacity to solve systems of differential equations, etc.) that are lacking in the general population, even those with strong business skills. Ergo, designing new financial instruments was an activity of financial engineers, who must be trained and socialized as such.
Pretty much any university that has an engineering school and a business school has therefore joined the two together in a new interdisciplinary department, center or program. Here in New York City, where I live, there are significant financial engineering programs so-named at both Polytechnic (now a unit of NYU and therefore the de facto partner of NYU’s Stern School of Business, which has so far avoided the terminology although it teaches the topics) and at Columbia. It is clear to me from interaction with these institutions that as recently as earlier this year (even after the great unraveling began), these programs constituted one of their great hopes for institutional growth in prestige and funding. At the time, I wondered how those plans would play out if it turned out that financial engineering developed the bad odor that it now distinctly has, of the date of this posting. Now we shall find out.
In a recent column in The New York Times, financial journalist Roger Lowenstein observed that it was the “conceit” of believing that risk could be modeled in ways that have always worked historically that prevented us from seeing the lessons of Long Term Capital Management in 1998 and anticipating the failures of the housing and credit markets of 2008. In essence, Mr. Lowenstein points to the peril inherent in much social science: extremely rigorous and precise analytical methods used without sufficient thought given to their applicability court what statisticians might call “model specification” error. It does no good to be precise if you are completely wrong about what you are trying to model and how it works. As hard as it is to get engineering right in the physical world where we know Newton’s laws aren’t changing any time soon, it’s even harder in domains where what we know we know only by observing patterns of historical behavior, and rarely from first principles or well tested hypothesis.
Frank confrontation of this weakness may be emerging as the professional consensus on how to find the way forward in financial engineering. In the most recent online newsletter of Poly’s program, department chair Prof. Charles Tapiero calls for a “new finance” that will reexamine the ways in which global markets assess risk. I further suggest that it is only by taking a Petroskian, intellectually humble, “forensic” approach to the current catastrophe that academic institutions with programs in financial engineering can recapture some of the high ground that has been lost in a month when we have seen securitization declared “dead” and the search for fall guys begin. Absent an intellectual reinvention, sterile quantitative exercises have no future in the financial world that will unfold in the months ahead.