Many portfolios were constructed using mean-variance optimization (MVO), a strategy that seeks to minimize the volatility of an overall portfolio by diversifying across uncorrelated asset classes. At this top-down level, risk is defined as volatility, or how much an investment deviates from its long-term average. The very technique used in the creation of many portfolios- mean-variance optimization — reveals the objective of the strategy. MVO optimizes the return-versus-volatility trade-off.
Once the overall portfolio strategy was set, the job of actually investing was often implemented by a collection of active managers.
Active managers often have a different definition of risk. For many active managers risk is measured against a passive market benchmark, such as the S&P 500 or the Russell 2000. Positions are taken to over- or underweight an aspect of a benchmark. Success is measured in terms of benchmark metrics – for example, relative metrics like alpha or information ratio. Insights are gleaned via attribution analysis quantifying the sector or stock picks that helped or hurt relative performance.
What do all of these steps have in common? The framework for thinking of and measuring risk is all relative to a market benchmark.
So what is missing in this approach? How does this model sometimes fall woefully short? The elephant in the room is market risk, sometimes known as systematic risk. Neither the top-down asset allocation strategy or bottom-up stock selection, explicitly addresses systematic risk.
When the market racked up losses of over 50% in 2007-08, both the asset allocators and the stock selectors were able to claim they were doing what they were hired to do – build a portfolio in the asset allocator’s case and pick winning stocks in the stock picker’s case. Yet, we find that neither the asset allocators nor the stock pickers are addressing market risk.