Rob Arnott: Index Investors ‘Buy High, Sell Low’

Rob Arnott: Index Investors ‘Buy High, Sell Low’

It's because they're too obsessed with tracking error, says Research Affiliates’ CEO.

Reviewed by: Lara Crigger
Edited by: Lara Crigger

Rob ArnottRob Arnott, founder and chairman of Research Affiliates, is the face of smart beta. In addition to his firm's own indexes, he has partnered with indexers such as FTSE Russell, S&P Dow Jones Indices, Citigroup and others to create benchmarks based on his Fundamental Index concept, which weights indexes according to accounting metrics rather than market capitalization.

Arnott knows indexes backward and forward, which is why the title of his Inside Smart Beta & Active ETFs Summit keynote address, "Buy High & Sell Low With Index Funds," caught our attention.

In advance of the upcoming conference, to be held June 6-7, 2018 in New York City, we spoke with Arnott to get his thoughts about what indexes are getting wrong, and how investors can beat the average index fund. Let's get right to it: What do you mean, "buy high and sell low with index funds"? What do index funds have to do with it?

Rob Arnott: It's actually really simple. Indexing is now seen as the go-to strategy, the way to "win" in the markets because active managers, in general, don't keep pace. So we looked at how index funds manage their portfolios to get some clues as to ways to win. When they buy, what do they buy? When they sell, what do they sell? And so on.

Index funds aren't passive. They trade, because the marketplace is constantly evolving. So when they sell, sometimes those sales are related to corporate actions: mergers and acquisitions, that kind of thing. The ones that aren't, however, are discretionary deletions. Discretionary deletions are companies of sufficiently low market capitalization, and sufficiently low interest to the marketplace that they're deemed unimportant. Generally, they're cheap.

But the stocks that are added are newly hot companies with large market capitalizations. They're, more or less, embarrassing to not have in the index. Generally, they're priced at lofty multiples.

So when you look at the spread in price between the ones that are added and the ones that are deleted, it's about a 3-to-1 ratio in valuation multiples; meaning, you're buying stocks three times as expensive as the ones you're selling. You're buying high and selling low. Does that valuation impact performance somehow?

Arnott: Yes. We tested index data from 1989 to 2017, and found that over the next 12 months, additions underperformed discretionary deletions to the index by 2,300 basis points. That's huge.

So if you're running an index fund, and you're willing to tolerate just a little bit of tracking error, you can do your customers an enormous favor by not trading when the S&P or the Russell changes, and instead putting a Post-It note on your calendar for 12 months from now that says "trade then."

The other thing is that before 1989, the S&P Index committee would wait to change an index until the market closed for the day, then announce the results the next day. There's no way to game that. But in 1989, it changed its protocol in 1989 to say, "We're dropping A and adding B, and we will do it on thus-and-such date."

This gives the index funds a grace period in which they can trade. Any trading costs they incur, and any movement in the stock price driven by their tremendous trading volume, also shows up in the index. So if they spend 500 basis points on trading costs, that impacts not just the index funds, but the index against which they're measured. So they seem to have zero tracking error.

We find that measuring before the rebalancing date and after, deletions beat additions by 570 basis points. If you include the rebalancing date, it widens out to 873 basis points. You're saying this is an opportunity for active managers to swoop in and capture some alpha.

Arnott: Well, some brave index fund manager should say, "Our primary goal should not be zero tracking error. It should be to pick up the crumbs along the path and give our clients some alpha." That means, when the index is changed, just wait a while for prices to settle down, for the additions to settle and the deletions to rebound, then trade a few months later. You can do that with startlingly little track error.

And there's one more thing. The No. 1 stock in any sector, in any country, or in the world index [MSCI ACWI] also usually trades at lofty multiples. It's usually a big, successful company, and it's trading expensively. We did a paper on this, "Too Big To Succeed," which showed the No. 1 stock in any sector or country typically underperforms by about 500 basis points, per annum, over the next 10 years. The No. 1 stock in the world tends to underperform the ACWI index by 1000 basis points.

So, what if an index fund just left out the No. 1 stock in the world? Today that would mean you have the ACWI index minus Apple; a few years ago, it would have been the ACWI minus Exxon Mobil. If you do that, you can add on the order of 20 basis points per annum, just by leaving out the No. 1 company by market cap.

So pester your ETF providers to launch an index fund that is lazy on trading, that trades three months late and leaves out the No. 1 company in each sector, each country and in the world, then reintroduces it when it's no longer No. 1.

This will have the effect of adding a little tracking error, a few tens of basis points. Your turnover would be roughly doubled. But if you accept a few tens of basis points of tracking error, you can add a few tens of basis points of incremental return. Why not? Because typically passive investors don't buy index funds because they want any tracking error. They like an index, and they want to replicate that index. What you're proposing has a high psychological hurdle to get over.

Arnott: Yes, but think about it: Why are we obsessed with tracking error? If you introduce 1% tracking error, that's not going to increase your volatility by 1%, because tracking error is uncorrelated with the market. So your risk will still be your risk, just on a slightly different trajectory.

I've talked with index fund providers about [this], and the general reaction is, "Yes, but we can't do that; it introduces tracking error." OK, why is tracking error more important than alpha? I think because, if you're selling an ETF that's based on an index, you want to get as close to replicating that index's performance as possible. If not, why not just go with an active manager instead?

Arnott: Right. And if you're willing to take a little bit of tracking error, once in a while it can go against you. So if you take 1% tracking error, with an expectation that you're going to boost the return by 50 basis points on average, that means that maybe, one year out of six you'll be 50 basis points under the index. You have to be willing to accept that to get your 50 basis point average gain.

But also, too many active managers are too greedy. They charge a fee that often matches their whole alpha. If they're wrong, and their alpha goes negative, you've got a high fee and a negative alpha. Well, that's not very great. But just because most active managers don't beat the market doesn’t mean it's hard to beat the market. How’s that?

Arnott: Most active managers anchor on the market. They're also worried about tracking error. So if Tesla and Netflix are in the index, they're happy to go underweight if they think they're expensive. But they're not happy to eliminate them entirely. Or, take Apple; they might be willing to underweight it, but, "My goodness, that's the largest stock in the index; I don't dare not hold it!" So you're left with anchoring on the market as an important source of performance drag for active managers.

Active managers are also human beings. As human beings, whatever has given us joy and profit, we want more of; that means we tend to buy high. Whatever's given us pain and losses, we tend to want less of; that's why we sell low. As investment managers, it's really hard to say, "This stock has been dreadful, I want more of it."

What we're suggesting is some extremely simple rules of thumb to help you avoid buying high and selling low. This isn't how we manage money; we're more active than this. But if you're running an index fund and you're willing to take a 1% tracking error, go for it.

Contact Lara Crigger at [email protected]

Lara Crigger is a former staff writer for and ETF Report.