Risk models and tools at Baruch College
February 11, 2012
Emanuel Derman gave the last presentation of the day on mathematical models and their role in financial markets. His presentation seemed to build on some of his earlier ideas with Paul Wilmott on the "Modeller's Manifesto".
Emanuel said that there was a "scandal based on models" is wrong; models did (and do) have their faults but they were not a root cause of the crisis. He started his presentation (somewhat "tongue in cheek") by putting forward a "Theory of Deliciousness" to see how one might arrived at the value of something being more or less delicious. This involved discussion of "realised deliciousness" and "expected or implied deliciousness", plus definitions around equally (relatively) delicious things and absolute deliciousness. See post on FT Alphaville for more background, but fundamentally by analogy Emanuel was putting across that there is no "fundamental theory of finance" and that finance is not physics.
He said that economists do not know the difference between theorems and laws. He seemed to be critical of some recent work from Andrew Lo (see recent post) on putting together a "Complete Theory of Human Behaviour" for once again attempting to codify something that it is uncodifiable.
Emanuel described how economists should be more aware of what is and isn't a:
- Metaphor – using something physical/tangible to represent a less tangible concept or idea. See this link for his interesting example on sleep/life and debt interest
- Model – extending the behaviour of one thing to another. A model aircraft is a very useful model of a full-size aircraft with know inputs and useful outputs of interest. We can try to model the weather but here the inputs are known (temperature, wind etc) but the model is hard to define. In finance it is hard to really see what both the inputs are and what the outputs are too.
- Theory – the ultimate non-metaphor. Here he gave the example of Moses asking the burning bush who shall I say sent me to which God replies "I am what I am". Put another way, you can't ask why on a theory, it just is.
- Intuition – a premise put forward based neither on logical progression nor on experimentation.
Emanuel said that in Finance there is no absolute value theory, and the majority of models are relative value in nature. From a common sense point of view, the world is not a model. Things change dynamically and in this way effectively all models are wrong to some degree. In summary all financial models are short volatility.
He ended his presentation by saying that nature cares more about principles than regulations (prescriptive regulators beware I guess). His parting quote was by Edward Lucas who said "If you believe that capitalism is a system in which money matters more than freedom, you are doomed when people who don’t believe in freedom attack using money."
- Bruno Dupire of Bloomberg said that it was important that a financial product was aligned with the needs of the customer, and cited certain complex products (with triggers) as being more in the interests of the vendor not the customer.
- Bruno also said that the hedgeability of a product was also key to a more stable financial system (presumably pointing at products like CDO^3 etc). He said that residual risk (that left after hedging with simpler products) should be measured and costed for. Bruno also mention the problems with assessing long term volatility where traders will try to set this input to what best suits their own P&L
- Leo Tilman said that risk management needs to be a decision-support discipline and not a policing function. He later suggested that risk managers should have to work as consultants for a while to understand that they get paid for serving the needs of the customer, not just stopping all activity/risks (in fairness to risk managers, I guess they might ask who is my customer? the trader? the CEO? the firm?).
- Dilip Madan added to the models debate by saying "what is not in the assumptions will not show up in the conclusions".
- Emanuel likes the old GS partner model for banking, and mentioned the example of Brazilian banks where banks/banking staff(?) did not enjoy limited liability. Dilip said he understood the advantage of this but no limited liability would stifle entrepreneurship.
- Leon Tatevossian said that post-crisis the relationship between risk managers and traders is better than before, and that there was also greater co-operation between empiricists and modelers. Leo add that risk managers and traders need to speak the same language and understand what each other means by "risk".
- Bruno said that models were much less of a problem than leverage.
- All seemed to agree that the tools were not invalidated by the crisis, but the framework in which they are used was the important thing.