What’s the difference between Residual Volatility factor and idiosyncratic risk?

[rdymke]

I know the Res Vol factor measures positions that inherently riskier than other positions. But some literature like this:

Are Factors a Thing of the Past?

Over the long term, the Value style had outperformed Growth investing in the US market.

www.msci.com

Says that Res Vol measures idiosyncratic risk. So why even strip it out at all? Why not just consider it idiosyncratic risk?

 

[Ken Abbott]

One reason is that regulatory agencies insist that banks measure both systematic and idiosyncratic risk.

 

[zeroquant]

Idiosyncratic risk is just that, the part of risk not captured by something systematic. Portfolio managers also likes to see it decomposed to show that they are trading alpha (idiosyncratic) and not beta (systematic—market, industry, well known factors like momentum, value, etc). Residual vol assumes that you have a risk model specified—in the case of the paper linked, the barra factor risk model—and you take the residual of the actual realized risk and the part explained by barra. You can see how residual vol is then a way to measure idiosyncratic risk since barra risk model captures exposure to market, industry and common factors. In general, the low vol factor is constructed by stripping out some simple exposures, and at that point it becomes a systematic factor that you can incorporate into a more elaborate risk model.

One other key point as well is your use of the term ”factor”. Factors are considered tradeable portfolios—it could be as simple as doing a sort and long the top 10% and shorting the bottom 10%—so they have some sort of transformation on top of the raw measure. Asset managers that subscribe to the asset pricing perspective of quantitative finance (DFA, AQR, etc) write a lot about them.