Sunday, October 28, 2012

AQR on Making Low Vol Better

 AQR has been offering 'low vol' funds institutionally for a couple of years, but last July started offering them to retail customers too.  As they are titled 'Defensive Equity' funds, they escaped my radar but really these are 'low vol' funds (eg, U.S. Defensive Equity Fund AUEIX, the AQR International Defensive Equity Fund (ANDIX), and the AQR Emerging Defensive Equity Fund (AZEIX)).

I spoke with Adrea Frazzini (right), a portfolio manager at AQR, and one of the lead authors on a major bit of research in this area, Betting Against Beta.  He noted their approach is basically to minimize volatility while trying to retain industry and factor neutrality, where the factors are such things as value, size and momentum. Considering that many think low vol is redundant in the context of a value exposure, this approach would appeal to those who already have a value exposue.

The idea here is that this approach generates less  tracking error that would come from, say, the SPLV, the popular low vol ETF that just swipes the bottom 100 stocks from the SP500. Basically, SPLV incidentally takes on various amounts of factor risk at any one time as some industries or extreme-factor loading stocks become less volatile.  By minimizing factor exposures tracking error will be less correlated with factors, which I can empathize with because if you underperform by 5% one year it feels better if its from something you can't control (the serenity prayer).

There's great danger and opportunity on this path. That is, SPLV is incredibly simple but it gets you a long way towards capturing the low vol effect: much lower vol, perhaps a slight return bump. In contrast, Russell's failed LVOL ETF tried to take it to the next level by getting really complicated, although I can see how, with a bunch of PhDs in a room it all made perfect sense. SPLV nicely keeps US low vol funds focused.

Like the other strategies in the low volatility space AQR expects betas of around 0.7, and volatility of about 1/3 lower than their regional equity benchmarks. Such a claim is realistic because that has been the result of other funds that have plied this strategy in real time over the past 7 years. As long as the Security Market Line (Beta on x-axis, average return on y-axis) is empirically flat this approach seems a no-brainer to any MBA because you can get the same return for 70% of the volatility; a low beta/vol focus has a higher Sharpe. [I'd say things about AQR's performance, but our far-sighted regulators make that parlous in their efforts to protect widows and orphans].

Frazzini mentioned that in non-developed countries, the volatility reduction is even greater within those regions, and clearly for any investor taking advantage of international diversification seems a straightforward way to reduce volatility and covariance with one's income much more than staying local.


Anonymous said...

My guess is that the fairly high management fees for these AQR funds will eat up whatever benefit they purport to provide.

Ben said...

Of course, now this fact about volatility has been discovered and is widely known, it can be packaged in such a way as to make money reliably. (That's the pitch anyway).

But since it is now widely known (or becoming so), surely everyone and his dog will pile on, and bid away the excess profits?

Eric Falkenstein said...

Well, you could say the same thing about value, which persisted. Momentum perhaps is an example of something that went away.

Ben said...

Volatility is just statistical analysis of the historical price, and easy to automate, no?

The difference is that "value" requires judging the intrinsic value of the operation, which is time consuming, and skilled.

So Value investing cannot be automated.

(And skilled labour is in short supply, so the minimum return from value has a floor, which is the wage the analyst/investor could earn elsewhere.)

To take a liberty, and paraphrase a comment of mine from an earlier thread:

We can turn the information (in Shannon's sense - raw data) into knowledge by expending effort to discover patterns in it (E.g., be the first to discover momentum, or "vol of vol").

Since this requires effort, you may for a while be the only one who knows of the pattern, and you can trade on it.

But once three people have expended that effort, you can't make much money trading it any more because you are bidding aganst them. (The new floor is the running cost of the automatic trading system.)

Anonymous said...

Saying value cannot be automated is an ignorant statement and disqualifies you from this conversation.

We are all only talking about highyly rule-based simple systematic strategies, including value.

Go post on the Buffet-is-God boards.

Anonymous said...

Didn't notice that you dropped Claude Shannon's name whlie misunderstanding this whole discussion.

I take it back, you aren't just ignorant.

Anonymous said...

Misspelling "highly" is ignorant and disqualifies you from this conversation.

Steve said...

You said value "persisted" - not that it persists.. Do you think value no longer works as a risk premium? I have some doubts that accademics overstate it due to ignoring the illiquidity of the cheapest, smallest stocks when grouped into deciles. Eg.

You also failed to mention that Andrea has combined low price vol with low fundamental vol (ie. ecomic beta) which I felt was a novel element to their approach. Any thoughts on this?

Anonymous said...

Attacking a spelling error when the issue was a thinking/content error is pathetic. Spelling disqualifies one from nothing. The moronic invocation of Shannon while not understanding the basic argument, but feeling you had to post, and then attacking a spelling error, disqualifies you from honor and respect.

If you're two different people, cool, two fools.