Risk Management = Risky Business

I came across an interesting video on the McKinsey website, in which the recently deceased Peter Bernstein gives a thirteen minute talk on risk in the real world and in the financial markets.

http://www.mckinseyquarterly.com/Organization/Strategic_Organization/Peter_L_Bernstein_on_risk_2211

Risk, in modern portfolio theory, is essentially the volatility of an asset price. This definition of risk is debatable, as Bernstein and most successful investors believe is that risk is uncertainty, in the sense that we do not know what the future holds. Realistically, there are a range of outcomes to any investment situation, and as investors, we do not know where the eventual outcome will lie in that range, what the actual range truly is, and when that outcome will occur. MPT fails in the sense that it says that risk occurs is a nice smooth normal curve. Our experiences in the real world tell us otherwise. We are continually getting 3 and 4 standard deviation events every couple of years - something that should not occur often according to MPT. The reason why there is this discrepancy is because the risk model is wrong. In fact, most risk models and therefore risk management tools are wrong because they attempt to do the impossible - predict the future and insure against it.

Bernstein goes on to say mistakes will be part of the investment process, and that the successful risk mitigation does not depend on determining when a particular risk will come to fruition, nor what type of risk will occur, but in how well we are prepared to deal it. Most risk management tools fail because they ignore their own fallibility, create a sense of security in their users, which eventually leads to these massive failures we read about in the WSJ each morning. Take LTCM for example. LTCM used sophisticated techniques such as fixed income arbitrage, statistical arbitrage, and pairs trading, and it did so in obscure markets such as Russian bonds. While this can, and did lead to good returns for LTCM initially, LTCM relied heavily on its risk models. Models that did not foresee the liquidity crisis in Russia, and did not foresee how LTCM's simple involvement in the game changed other players actions, and therefore the usefulness of their risk model. In short, they were ill prepared to deal with this outcome because their risk models did not, and in fact, could not tell them that this outcome would occur and that it would occur at that point in time. Another prime example is the crash of 1987, in which portfolio insurance was the main culprit. With markets in the midst of crashing, the portfolio insurance (puts) of most large institutional investors kicked in en-masse. This caused major investment funds to aggravate the situation by essentially selling large amounts of stock all at the same time, further pushing the indices down and causing a feedback loop. In this situation, risk management's new creation (portfolio insurance), misled investors once again because it simply could not foresee the actions of other investors, and it did not have the option to correct itself.

What is the lesson from all of this? The lesson is that risk models and risk management tools simply cannot work in every scenario. First off, it is important to understand that you and your decision making process are fallible, and that you will never be 100% correct, nor be able to see all of the potential outcomes - both positive and negative. Therefore, it is important to focus on the worst-case scenarios, and then protect yourself as much as possible in these scenarios, because the worst-case scenarios are the ones that matter the most to your long-term investment success. This involves having a good understanding of probability theory, as pertains to the magnitude of potential losses. Most investors protect against loss by hedging, which is a good option, although it involves you taking a position in another instrument with potentially disastrous outcomes (see LTCM and portfolio insurance). Value investors protect against loss by buying instruments that have limited downside in even the worst-case scenarios. Even though you may not know how one position will work out, when combining several of these positions, the portfolio should have limited downside as a whole. As for the upside, the sky is the limit.

I would also like to point out that it is very refreshing to see somebody of Bernstein's stature bringing up a very simple yet effective concept of risk mitigation. When you do not know or do not understand an investment, then using the "run like hell" approach may simply be the best way to avoid that particular risk. Good players realize when they do or do not have an edge, and pick their shots accordingly.