I first met Emanuel Derman in the late 1990s, when I was starting out as a financial journalist. I felt an immediate affinity. I respected him as a refugee from physics who had progressed much further in the field than I ever had. In person and in his writings, I found him a patient guide to quantitative finance. He downplayed formal mathematics in favour of intuitive explanations of option pricing based on trader decision-making. Derman was also one of my introductions to Goldman Sachs. Just by being the way he was, he served as a glowing reference for the firm. It was impossible to conceive of a bank that employed such a thoughtful humanist being capable of wrongdoing.
Goldman was a different firm then. Underwriting and advisory work dominated, and derivative trading was a relatively obscure business compared to commercial banks such as JP Morgan. In that environment, the conceit that derivatives was an intellectual rather than a commercial pursuit within banking was still defensible. Derman describes his early days at Goldman, “quants were the theorists, traders were the experimentalists, and we collaborated to develop and explore our models”. By 2000, I could tell from the outside that the firm was changing, as it began to industrialise derivatives to catch up with the commercial banks.
Derman left Goldman in 2002 just as the changes were starting to have an impact. Basing himself at Columbia University, in 2004 he published “My Life As a Quant”, an account of his Goldman experiences. Derman’s new book, Models. Behaving. Badly. is his first since the financial crisis. The crisis was “marked” by the failure of models, Derman says (he doesn’t say “caused”—if he did, that would itself be a model) and looking at their uses and abuses is a timely exercise. It’s an opportunity for Derman to recount his Jewish upbringing in South Africa, and the social and political models he witnessed as a child, and to channel his decade in theoretical physics, where models take flight and become fully-fledged theories of nature.
My favourite chapter of the book looks at Jewish theology and Spinoza’s theory of the emotions. Derman has done a service here—when I tried reading Spinoza two decades ago, the formality and obscurity of his language made me give up. Derman persevered, and explains how Spinoza’s Ethics forms a self-consistent psychological model where fundamentals such as pleasure and pain give rise to emotional ‘derivatives’ like humility or devotion. Equally sure-footed is Derman’s account of electromagnetism, going from ancient Greece to Richard Feynman and quantum electrodynamics via Maxwell. Even lapsed physicists who know the punchlines will enjoy his account.
Turning to finance, Derman finds it wanting in comparison, declaring himself to be a disbeliever when it comes to all-embracing financial theories. Focusing on the Efficient Market Model (not hypothesis), Derman shows how simplified assumptions about uncertainty and investor behaviour lead to the idea of excess return as a reward for risk, explaining how flimsy the EMM is when used to value stocks and other investments. Pointing out that the crisis was in some respect a failure of the EMM in that investors failed to demand the right reward for risk, Derman lays down some positive precepts for how models should be used in finance: use them for intuition, interpolation, and be upfront about their flaws and omissions. It’s hard to disagree with any of this.
Does the book expose, as the dust jacket claims, ‘Wall Street’s love affair with models’? Here I am not so sure. Early in the book, Derman lists various types of model: fashion models, artists’ models, weather models, economic models and his favourite, the Black-Scholes model. There is no mention anywhere in the book of the phrase ‘business model’. That surprises me, because outside of academia it is probably the most relevant type of model there is. Most entrepreneurs fail or succeed according to their business models, and the jobs of hundreds of millions of people depend on the business model of their employers (journalists are a good example).
It is not financial models that drive Wall Street, but business models. These might be the transformational business models in a sector like technology or healthcare that power M&A pitches by bankers. In the post-2000 industrialised world of derivatives, business models were built around the contrasting appetite of different types of client for risk-return propositions, from which a trading operation could extract rent or cash flows.
Derman writes: “One uses a model to turn”¦opinions about the future into an estimate of the appropriate price to pay for a security that will be exposed to that imagined future”.
I would say: that’s what the clients do. Inside modern investment banks, one uses a derivatives business model to turn client opinions about the future, and their regulatory constraints, into a structuring franchise with minimal basis risk and low cost of capital. Financial models are one of the mechanisms that shape those client opinions, turning them into reliable buyers of securities (for example if a rating agency uses the model to award a triple-A rating).
I remember the last meeting I had with Derman at Goldman. It took place at the firm’s old headquarters on Broad Street, and we were seated around a table with Derman’s boss at the time, the then-head of firmwide risk, Bob Litzenberger. Litzenberger explained how Goldman risk managers were pushing their bond origination people to start shorting their own markets by using derivatives like credit index default swaps, to avoid filling the bank’s books with credit risk. This was in late 2000, about a year before Litzenberger retired from the firm. In 2007, it turned out that Goldman followed this strategy in subprime, protecting itself from the multi-billion dollar losses faced by its clients, and in some ways benefiting from their losses.
The point is not that the models failed in 2007—they did. The point is that a new investment model will come along, driven by changing opinions and regulation, and large banks will find a business model that industrialises and exploits the clients that believe in it. The challenge is not to improve financial modelling, but rather to catch these business models inside banks before they become too dangerous.
(27 November 2011)
