Faust, Economics, Psychology, and Models
I propose that we review this Faustian bargain.
Here are the basics for financial models. No financial model has the status of a physical law (regardless of who is pushing the model, regardless of whatever mathematical framework is being used, regardless of how sophisticated it is, and regardless of whether fat tails are in it or not). Any financial model is really only a phenomenological construct. The parameters that are needed to specify the model are “implied”. In practice this means that a model, intended to price some security, first prices other securities or quantities in accord with market prices. Other parameters are determined by fits to historical data, and/or using “fundamentals”.
But this procedure, which the finance industry has used since the beginning of the “Quant Age” in the 1980’s, has now broken down for many securities. Models that are constructed using the best information, using the best attempts at consistency, and using conservative assumptions on the fundamentals, cannot describe the insanely low market prices for many securities. This is a huge problem. It cannot be said too strongly – the implied parameter approach used to specify model parameters is now dysfunctional in some markets.
Whose fault is this? Does the fault lie with the Faustian bargain?
Let’s start with economics. It is clear now that market prices do reflect economics – e.g. the recession. For example, long-term debt clearly has a component that is ruled by economics (and not just default). Spreads for bonds are now so high that they imply unbelievable default rates. The Fed stress tests are macro-economically driven. So why aren’t economic variables found in models for bonds? Answer: The Faustian bargain.
Actually, there is at least one model that started to address the issue of economics and financial modeling, the “Macro-Micro” model. Also, factor models provide a framework for including many variables. I believe such models should be taken more seriously and made more explicit. This is a daunting challenge, but I believe we can no longer avoid it. We need to get away from the Faustian bargain.
What about psychology? Market prices are now at least partially ruled by psychology, involving e.g. fear and damage control. The so-called “investor rationality” has disappeared (actually this statement is just a tautology that re-expresses the breakdown of the financial axioms; investors individually under the circumstances are actually being very rational). Psychology has always been instrumental in the markets, but now psychology has broken the markets. However, we have no “psychology” parameters in models. Is this a fundamental error? Yes, I think so. Can we get away from the Faustian bargain here? I don’t know. It is a humbling experience.
In this space of fog, there is at least one clear point. Can’t we just blame the quants? Here the answer is clear: “Nope”. Quants just implement the general religious dictum provided by the academics (naturally with bells and whistles). For the record, the recent catastrophic mortgage meltdown is not the first time quants have been blamed. I remember an incident once reported to me where a particularly obnoxious mortgage trader got up in front of a group meeting and demanded that the head quant apologize for his model, which the trader accused of causing mortgage-trading losses! Ugh.
What does the future hold? The banks have argued that we cannot mark-to-market for securities if there is no market for those securities. They certainly have a point. So they argue that we should mark-to-model. Uh-oh. No more constraints on the models from the market? What is going to constrain the models now? Certainly not “model validation”, which doesn’t address the fundamental problem - as I said no model is “valid” in the sense of physics. What about “parameter reasonableness” criteria? The answer would be “Yes” for some parameters, but “No” for other parameters that depend on the broken model vs. market consistency. We are in uncharted waters. I can foresee the emergence of two classes of models: #1 for traders that need to trade on the market, and #2 for capital determination and regulatory reporting. Confusion will reign.
Now by no means do I want to imply that models are useless. We need models. Nor do I want to imply that we abandon implied parameters. We want to have models that describe market prices, if possible. But current models may be too narrowly focused. At least with present horrible market conditions, and maybe in general, I believe that we need, somehow, to include economics and psychological constructs in the models.
Maybe someday with a return of confidence, the blue skies will re-emerge, investors will again become “rational”, and the current models relying on the Faustian bargain will again have reasonable implied parameters that do describe the traded market prices. And maybe not.
In any case, we should keep in mind that even if the good times return, the entire financial system could well exhibit systemic fragility again – that is, it could break again.
Faust was a winner for a long time. However, Faust had his problems. Where did he wind up? For the quant modelers reading this, I recommend that for background you get a CD or DVD of the opera “La Damnation de Faust” by Berlioz, and listen closely. Then shut off the CD player, go to your local University, look up a friendly Economics professor, and then find out where the Psychology Department is located. That’s where I’m headed.
------------
Refs:
1. Fed SCAP stress tests:
http://online.wsj.com/public/resources/documents/scap2009424.pdf
2. Market vs. Model: Andrew Davidson Industry Insight - Proposed FSP FAS 157-e:
http://www.riskcenter.com/story.php?id=18201
3. Bond spreads and defaults:
http://online.wsj.com/article/SB123803103548043611.html#mod=loomia?loomia_si=t0:a16:g12:r2:c0.25105:b23576652
4. Macro-Micro Model: Chapters 47 – 52, my book:
http://books.global-investor.com/books/19880.htm?ginPtrCode=10202
5. Faust:
http://en.wikipedia.org/wiki/Faust
----------
© 2009 Jan W. Dash. All rights reserved.
