Most long-term allocators use monthly or quarterly performance data to assess the risk of financial assets. A typical process involves taking monthly performance figures over a sufficiently long period of time (10 or 20 years), calculating monthly asset volatilities and correlations over these periods, annualizing volatilities by multiplying monthly figures by the square root of 12 (while leaving correlation estimates untouched) and then deploying results as “long-term risk estimates”.
This approach works well in the context of estimating asset risks over short investment horizons of say a month or a quarter. But in the context of much longer investment horizons, it would make more sense to lengthen the time horizons of our risk estimation as well. One way to go about this would be to decompose historical asset price patterns into persistent components and short-term “noise” and then focus on the former.
The two components of S&P 500 differ not only in terms of speed of change but also in terms of their long-term dynamic. The slow, “persistent” component grows over time, reflecting growth in real economy and corporate earnings. The fast, “transient” component is directionless and strongly mean reverting. Granted, these characteristics were effectively assured by our decomposition methodology, but they also reflect the dual nature of the price dynamic of most financial assets, including public equity.