Why All Risk Models Understate Risk?
There are three types of reasons: mechanical, psychological and market.
Parameter Risk – all of the parameters of risk models are uncertain. That fact is usually ignored.
Residual Risk – there are two parts to this one. Within the range of the data and outside the range. Within the range, the process of modeling always produces smoother results than the actual observed results. This understates risk. Outside the range, the method might over state or understate risk. Possibly by orders of magnitude.
Randomness – many of the risks that we model with random variables are not at all random. They are causal, but we do not know how to follow the casual chain to its conclusion. Reality of these risks will involve much more discontinuities that are usually included in our continuous risk models.
Humans are hard wired to have a better memory of good times than bad times. This manifests itself in many of the biases chronicled by psychologists.
Many of those biases boil down to the fact that we all tend to see the world as we want it to be, rather than as it is.
Because of the above, the market tends to underprice risk. You often do not get paid enough for the real risk, you get paid for the risk in the model. Those few who look past the models and come the closest to understanding the real risk will simply not play. So the markets are dominated by folks with models that understate risk.
The place to play is identified clearly above. Did you notice?
Tags: RiskYou can comment below, or link to this permanent URL from your own site.