Today we bring you more of the most commonly asked questions about risk, fielded by our colleagues in Japan. Read part 1 of our blog for more insight.
3) Out of sight, out of mind? Or do statistical factors really add value?
In Japan there seems to be a clear preference for fundamental risk models. Perhaps because of their risk reporting needs, many clients are more concerned about being able to explain the numbers they are reporting to their clients than they necessarily are about how the numbers themselves are derived, or maybe even how good of an actual estimate those numbers may be. The discussion simply goes, “Our tracking error is XYZ%, and is driven 15% by risk factor A, 5% by risk factor B, etc.”
In this context, statistical factors can be viewed with great trepidation, regardless of whether they are viewed within a purely statistical or a hybrid framework. The fear is that the investor’s client will look at the risk report and say, “I see you also have 10% exposure to Statistical Factor 3. What’s this exactly?” The image alone of being caught out in front of the client with a long, drawn out “Uhhhhhh…” is often too much to bear.
Naturally, we find that this focus is often driven by the internal versus external nature of the risk function at the investment firm. Fundamental models seemingly have the advantage of being more intuitive to explain (e.g., we have a large exposure to size and momentum), which make them ideal for reporting purposes. But they can also suffer from being pre-specified in markets that are changing. On the one hand, it’s true that sometimes these changes are transient, and hence may be noise. Other times this is not the case. While fundamental models will dump these changes into the residual “Stock Specific” category, blind and statistical factors can alert you to systematic changes that are happening in the market right now. You may have to do some research of your own to grasp what those changes may represent, but if your goal is risk management over risk reporting, wouldn’t you rather know that these changes are occuring?
4) Risk isn’t for stock pickers
The final question I often get goes like this, “We are stock pickers. Hence, most of our risk is stock specific, and that’s what the model shows. So why do I need a risk model to show me what I already know?”
It may generally be true that investors who tend to have smaller, more concentrated portfolios relative to the benchmarks they invest against should have a higher percentage of their risk in the stock specific category. But while this “should” be the case, it isn’t always.
Take for example the three funds shown in the screenshot below. These are three Japan equity mutual funds that can be found in FactSet’s Ownership database, but which have been renamed Fund A, B and C respectively for anonymities’ sake. Fund A is rather concentrated with only 30 positions held, while Fund B is broader at 77 and Fund C even broader at 1,389. While Fund B has the highest tracking error, a greater percentage of that is coming from “Factor Active Risk” than both Fund A (with half the positions) and Fund C (with 18 times as many positions). In particular, we can see that there seems to be a clear “Industry” bet that’s helping to drive this funds tracking error, in addition to a significant bet on “Leverage”.
Assuming we have actually done a good job of picking stocks, this bias could still end up being a significant drag (or boost as it may be) on our performance. Was this bet intentional? Are we comfortable with this exposure? Or do we want to hedge this exposure out in some fashion? The risk model can help you identify and hedge these risks… but only if you have one.