Stable vs. Unstable Markets: A Tale of Two States

One of the bedrocks of modern capital market theory is that market risk and related statistics are stable over the long run. Nobel prizes have been won for this insight, and it is taught in the best business schools. Regulations have also been written based upon this assumption. Yet, experience does not support this idea. We know that often markets have periods of relative stability, but they can also be followed by years where it seems that all is chaos (and not in the physics sense of the word). In academia, there are theories that compete with the Capital Asset Pricing Model (CAPM), the main proponent of stable markets, but these competing theories are generally considered impractical since they don't lend themselves to easy solutions. The answers we often receive from the usable models, however, go horribly wrong when markets go south. Could it be that we're using those models simply because " the light is better here? " But how can we use the models that are " impractical? " In this paper, we show convincing evidence that there are actually two separate market states, each corresponding to these competing models. In essence, the CAPM and its critics are both right, but only part of the time. The implications from this for asset allocation and plan management are profound. There will be periods where using standard techniques for asset allocation or investment management will work well. When the environment changes, however, those processes may no longer work with reliability. This is particularly true of diversifi cation because assets that diversify one another in one state fail in the other state when they are truly needed. The rules change, and if investors hope to adapt successfully, they will need to know the new rules when the change occurs. This paper offers an explanation for that problem-two different market states causing assets to behave very differently. A subsequent paper will provide a roadmap to help asset owners anticipate and navigate deftly through the changes in market states. How do we distinguish these two market states? In one state, markets are statistically well behaved. They can be modeled using standard statistical analysis. Volatility is stable and low. Correlations are stable. Tail events (3 standard deviation or larger returns in either direction) are rare. The periods correspond to low volatility