FINRA Reports on Robo-Advisors

A year ago FINRA and the SEC issued a joint investor alert regarding robo-investment advisors and other automated tools that offer advice to investors. The alert clearly suggested that investors should be wary because an automated tool may rely on incorrect economic assumptions.

That warning has dissuaded just about no one. Robo-advisors are a fad that is growing exponentially.  Mainstream brokerage firms have acquired existing robo-advisor firms and launched their own robo-platforms.

I have written several articles where I explained why robo-advisors cannot work. In addition to the fact that they may rely on incorrect assumptions, they are also not connected to the real world in any way.  While most investors are concerned with slowing growth in China, the changing price of oil and the continuing rise in interest rates, robo-advisors concern themselves with none of these.  Robo-advisors do not look at what is happening in the broad economy or what is likely to happen.

FINRA has just published a new report on digital investment advice that demonstrates just how poorly robo-advisors perform.  The report asks a great many important questions about the use of robo-investment advisors but answers none.

You can view the report here. .

The report defines the investing process in six steps; customer profiling, asset allocation, portfolio selection, trade execution, portfolio re-balancing, tax-loss harvesting and portfolio analysis. I don’t quibble with these steps, however, it is how the robo-advisors execute them that is the problem.

The report notes that there is a wide disparity between the platforms regarding the amount of information that they acquire about a prospective customer.  I wasn’t actually troubled by this. The brokerage industry knows how to open an account and obtain enough information to ascertain a customer’s investment objectives and risk tolerance.  It is a task that the industry performs every day.

Where the report was most enlightening was in its side by side comparison of the portfolios that seven robo-advisors would have constructed for a hypothetical 27 year old who was investing for retirement. The disparities are enormous. The amount of equities purchased ranged from 51% to 90%; the amount of foreign issues from 22% to 48%. One portfolio contained 5% in gold and precious metals; another 14% in commodities. The rest had no investments in either category.

Presumably these allocations would not change over the next 40 years as the customer added to the account every year. The FINRA report defines re-balancing as maintaining the target allocation.  I would assume that risk tolerance declines with age, but the FINRA report makes no mention of how that money would be invested in the future.  Even if the allocations were static, the end result would vary greatly platform to platform.

So what, you say? The end result of a 40-year investment portfolio is going to vary greatly if you compare the portfolios prepared by live advisors as well.  True, but the difference is that live advisors are looking at the real world and absorbing real information.  The algorithm that is at the core of a robo-advisor, once set, may never be modified and never looks at real world events.

I have read a lot of articles about robo-advisors.  Few actually try to evaluate the algorithms of one versus another. The FINRA report suggests that member firms need to effectively govern and supervise the algorithms that they use. An algorithm is a mathematical formula. Exactly how does the investment committee at a brokerage firm supervise it?

The FINRA report gives passing acknowledgement that many of the robo-advisors are based in Modern Portfolio Theory (MPT) and gives a shout-out to Prof. Harry Markowitz who first proposed it and later won the Nobel Prize.  I went back and re-read Prof. Markowitz’ article that was published in 1952.  The math was over my head.

Markowitz was a theoretical mathematician who applied his theory to economics, specifically to investment portfolio construction. He theorized that investors would want to maximize their investment returns while at the same time minimizing their risks.

He accomplishes this by theorizing that a diverse portfolio of non-correlated assets will perform better than a portfolio of concentrated assets.  He creates a series of mathematical formulas (algorithms) to show that it is possible.

As a practical matter it is the idea that the various assets in the portfolio are not correlated to each other that is at the core. As oil prices come down for example, we know that it has a negative effect on the profit of oil companies but a positive effect on the profits of airlines and trucking companies. Higher interest rates are good for banks, but not so good for home builders.

Now go back and look at the categories of equities that make up the robo-advisor portfolios; large, mid sized and small cap domestic securities, developing and emerging foreign markets.  How are these non-correlated?  Haven’t builders, large and small in all countries been helped by low interest rates? Aren’t airlines of all sizes in all countries going to be negatively affected if oil prices shoot back up?

MPT necessarily looks at the effect that interest rates and commodity prices have on portfolio securities. I realize that those 5 categories of equities, large, medium and small cap, developed and emerging markets have become standard in some quarters but in a global economy they are less and less non-correlated.  That correlation is even higher if you use baskets of securities (ETFs) because some number of the securities in every basket will be affected by changes in macro market conditions.

Markowitz’ formulas are based upon what he believed investors should want; good returns with less risk. The “efficient frontier” that evolved from his equations is a point on a graph, not a place on a map.  It is a hypothetical goal.

The algorithms that support the robo-advisors are the same, hypothetical. They do not look at the real world, nor are they concerned with it.  They just tell you, if you do X, Y should happen.  That might be true is the world were stagnant, but it isn’t.

One difference between Markowitz and the new generation of algorithm writers is that Markowitz published his equations in a peer reviewed journal. The robo-advisors will never let anyone see what is behind the curtain.

The fact that FINRA can demonstrate the wide disparity between the portfolios that various algorithms would create makes shopping for a robo-advisor all the more necessary while at the same time demonstrating that it is an impossible task.

If you hire a human investment advisor you can sit down and talk to them first. You can ask them what they know and what they think. You can ask them about the markets and about the future. Chances are because they deal with a lot of people they have heard the same questions before and have had an opportunity to consider their answers.

Robo-advisors are a fad. They are popular because they are cheaper. They are sold by disparaging human investment advisors who some people think cannot do any better than average. That begs the question that I have asked before: why would anyone want to hire an average investment advisor?