Tuesday, June 18, 2019

Factor portfolio construction - All alternative risk premia are not the same

An investor could say, "I would like to have exposure to risk premium X." However, there is no one single definition or consensus for how to implement a risk premia portfolio. First, there has to be agreement on what it means to say you want a specific factor exposure. There may be agreement from an academic definition, but once you start getting into the investment details, there can be a lot of differences in how to access this premium. 

Here are some of the choice issues for investing in a risk premium:
  • Universe for application of the risk premia. This universe could be, in the case of equities, a large cap set of stocks or a much broader universe which includes mid and small cap stocks. There also has to be the decision whether to include global equities and emerging markets.
  • Monthly or daily reset or rebalancing. The timing for when to rebalance will matter both for performance, risk, and costs.
  • Long-only or some variation of short exposure with constraints. Performance will differ markedly when you include short positions.
  • A pure factor exposure or controlling exposure to other factors matter because over-exposure to one factor may be driven by other risks.
  • Equal weighted, risk weighted, or cap-weighted allocations will impact risk exposures. 
  • Constrained or unconstrained with respect to tracking error or exposures will impact concentration risks. 
The construction issues are relevant for all asset classes. A recent thought piece from the Axioma Group examined one risk factor, profitability, and compared the return results using different portfolio assumptions. The range of performance and correlations are startling. (See "What exactly is a factor? How factor portfolio construction impacts exposures, returns and attributions" from Axioma)

Table 1 shows the definitions for the choice set analyzed. Table 2 shows that the correlations between the factor and different implementation strategies can vary between .83 and .21 over a thirteen-year period. You may not get what you think you paid for with this risk premium.

The performance differences are significant with a spread in the cumulative gain of over 50 percent. The scaled performance differences are over 15 percent. Now, on an annual basis this may only a little more than 1 percent, but these differences add up.  The weighting scheme will also generate significant differences both in the short and long-term. if you made the "wrong" weighting choice, you would have likely exited this risk premium even though it would have generated strong gains.

Implementation matters, which means that knowing the details will also matter for any risk premia choice. This requirement for understanding the details is no different than knowing the differences in manager strategies. There is value with doing a deep due diligence.

No comments: