Today, I’m tackling what the real problems are with robo-advisory services – who should use them, and how not to get crushed when they go haywire.
Depending on what theories and math robo-advisors are wired for, they construct a “personalized” portfolio based on forms you fill out online, and they automatically rebalance your portfolio when threshold weightings of positions in your portfolio get out of balance.
Ghosts in the Machine
The first problem with these services starts there, in their construction of portfolios.
They are personalized- but only to the degree that you fit into a model that fits thousands or hundreds of thousands of other hopeful investors. So, don’t count on your portfolio being different to the point where you believe it’s immune from what anyone (or everyone) else might suffer through if markets blow up.
Generally, you’re put into “passive” low-cost indexed ETFs. Investors plowing in larger sums at some services, like Wealthfront, can have a combination of individual securities and one or two “completion ETFs” to track an index.
If you’re investing 100% in U.S. equities and expect robots to diversify you, they will… but they won’t.
The bottom line is, no matter how robots break up your funds- whether they put you in several big index ETFs tracking the likes of the S&P 500, the Dow Jones Industrials, or the Nasdaq- you’re essentially correlated to the market.
Robots can put you into large-cap ETFs, small-cap ETFs, growth ETFs, value-oriented ETFs, into ETFs indexed to divided sectors or industries, or smaller subsets of equities based on fundamentals, dividends, almost any subset of stocks based on almost any theme. There are lots of indexed “products.”
But you’re still correlated to the market.
It may matter what indexes you’re in on the upside, in the short run. It won’t really matter in the long run if you’re well diversified across all these indexes and groups. You’re being indexed enough, diversified enough, to essentially just follow the general market.
On the downside it matters, because correlation is what it is: a phenomenon, whereby most equities breakdown when faced with widespread selling by individuals, hedge funds, and mutual funds. It matters especially in the short-run, regardless of how “uncorrelated” to each other these equities are supposed to be.
Then there are the ETFs. That’s what you’re mostly invested in with robo-advisory services.
If you don’t remember what happened to ETFs last August, you need to be reminded here and never forget.
Because ETFs are composed of actual stocks, or futures, other assets, or derivatives for that matter, they are priced based on the sum of their parts. Last August, before markets opened, futures prices were down sharply. Everyone knew stocks would likely selloff hard at the open. And… they did.
The problem with ETFs suddenly surfaced. How can you open trading in a security if the price of that security (an ETF), is based on other stocks that aren’t open, or opening, and have no prices? You can’t really. If you do, you’re just guessing.
So while lots of ETFs weren’t opened for trading, the constituent stocks that they’re made of were going down.
What happens to your portfolio if your robots can’t sell your ETFs while the stocks that make them up are going down?
You could be devastated.
The Faults in Correlation
Now, we’re back to correlation. If stocks are going down and some investors can’t sell what they want to sell, they’ll sell whatever they have to, including other asset classes.
That’s cross-asset correlation. Everything gets sold and sometimes there are no safe harbors. Usually when there’s panic selling, the Treasury bond market becomes a safe harbor and prices of bonds rise.
Maybe your robot will have you in a small T-bond position. You’ll at least have that.
Robo-advisor advocates would argue it would be unlikely an investor would be 100% in just U.S. equities based on their models. But their models are based on investor inputs. And if an investor wants all equities, and all-American equities, they’ll get that.
As far as cross-asset diversification, especially achieved by mean-variance analysis (a favorite of most robo-advisors), Mark Broadie, Professor of Business at Columbia University, has demonstrated through simulations that “the error maximization property of mean-variance analysis becomes more pronounced as the number of asset classes increases.”
In other words, you’re more prone to larger standard deviation moves on both the upside and downside the more diversified you are in different asset classes.
And, you know now that matters more on the downside, because that’s when the big moves – the “fat tails” wreck all normal distribution-based models… which all of the robo-advisors are based on.
You don’t have to worry about much on the upside. Your robo-service will make you money, but being as indexed as you’ll be, don’t expect to outperform the market.
On the downside… good luck. You’ll get creamed along with everyone else when markets tank, maybe worse if the ETF market implodes from structural issues.
That’s in the short-term. If you’re a long-term investor you’ll be told to sit back and let the markets come back and lift you up when the correction or panic passes.
While this is historically true, what if you need to take money out in the short-term? What if you don’t have a long investing horizon? What if you were just trying to maximize your investment portfolio before moving more into bonds?
Portfolio construction is a problem, diversification is a problem, ETFs are a problem, correlation is a problem, the normal distribution math used is a problem – there are lots of problems with robo-advisory services.
If you’re an investing beginner and having a portfolio automatically constructed and rebalanced for you gets you into the market, I’m all for it. Go with it.
But, understand the pitfalls in blindly expecting a robot to understand you, the markets, and protect you from serious harm in the short-term, or the long-term. I only advise you use a robo-advisor if you manage that account and plan your contingent moves into the future and be ready to execute them when you have to.