Share this Post on Twitter

Tuesday, April 28, 2015

The Uncertain Future of Robo-Advice

I had a Twitter back and forth with the founder of Wealthfront regarding the tax efficiency of their platform vs. using the same asset allocation model within an ETF (see here). I thought it might be of interest to provide additional details, as well as some thoughts on robo-advice more broadly.



Investopedia defines a robo-advisor as:

A robo-advisor is an online wealth management service that provides automated, algorithm-based portfolio management advice without the use human financial planners. Robo-advisors (or robo-advisers) use the same software as traditional advisors, but usually only offer portfolio management and do not get involved in more personal aspects of wealth management, such as taxes and retirement or estate planning.  
Let me be clear... I am absolutely THRILLED that robo-advisors exist. If nothing else, they pressure fees traditional advisors and ETFs charge, while an allocation made to a robo-advisor (by a hands-off investor) is likely to result in materially better performance than could be expected from many brokers that charge exorbitant fees. In addition, I've seen first hand from a "techie" brother in-law that a small, (but real and growing) community in the Silicon Valley has been drawn to the automation robo-advisors provide, which hopefully leads to a better financially educated community for all involved. BUT... I really don't understand where the 'actual' service they provide sits on the advice / allocation spectrum that consists of a traditional advisors that act as fiduciary on one side (higher fee, higher interaction, higher education benefits to their clients) and the multi-asset / target date ETFs on the other (lower fee, lower interaction, lower education).


Where's the Advice?

While I think robo-advisors can create a decent portfolio based on an initial survey and may be able to provide decent adjustments over time (none of this is all that difficult... most robo-advisors and ETFs get to the same place), over the years I've realized that 90%+ of the value a traditional advisor provides is the “advice”. Advice does not only consist of the beginning "how much risk are you comfortable with" phase or the year-to-year "has anything changed" phase, but a real thorough understanding of changes that have taken place in their client's life. As important is the "advisor as therapist", talking their clients down from the ledge when investment behaviors ebb and flow with the market.

I question whether robo-advisors can manage investor emotion. Investors generally ignore advice when it goes against their view and/or when they have real fear (of missing out or of losing money). There are times when an investor benefits immensely from a simple "slap across the face" from their advisor (i.e. when markets are off 50% and they want to sell). What will happen to allocations on robo-platforms when markets (eventually) turn lower? Wealthfront has already outlined that their investor base has changed their risk tolerance as markets have moved and this took place in what has generally been an up-and-to-the-right market.


Tax Efficiency: Self-Contained ETFs Rule the Land

Ignoring investor behavior for a moment, robo-advisors platforms are simply inferior to an existing alternative option available to all retail investors in an area they claim superiority... tax efficiency. Without getting into how ETF sausages are made, target-date and multi-asset ETFs can avoid capital gains most of the time (high level details here) even when the ETF rebalances within the wrapper from an outperforming holding to an underperforming holding. This is HUGE for tax efficiency, as an investor may not have to pay capital gains taxes from the day they buy into the ETF, until the day they eventually sell, allowing returns to compound to a greater extent over time.

On the other hand, a robo-advisor can do everything right to optimize tax efficiency (tax loss harvesting / use dividend payments and contributions to limit the capital gains associated with rebalancing), but may still force an investor to realize capital gains as there is a limit to how much they can offset as the asset base grows relative to the contributions made by the investor.

For example... the charts below outline the level of rebalancing (as a percent) that can be supported by flows and dividend / interest payments for a portfolio over time assuming a $10,000 annual contribution, 8% annual returns, and a 3% dividend + interest income (slightly higher than the current "yield" of a 60/40 portfolio). The left hand chart shows the level as a percent over time, while the right hand chart shows the level as a percent in dollar terms using the same assumptions.

Using this simply for illustrative purposes, the key is that over time, the level of rebalancing that can be supported by contributions and distributions narrows materially towards the 3% distribution rate as contributions become smaller relative to the asset base. This also ignores the fact the contributions eventually turn to withdrawals, thus the level of rebalancing supported by flows will turn negative.


Compare the above 3-4% figure to the rebalancing required to maintain a simple 60/40 portfolio over time. The chart below outlines the shift from stocks to bonds (or bonds to stocks) required over rolling 12-month periods to maintain the 60/40 weights. Note the levels are materially higher when rebalancing is needed most (i.e. after sharp moves in stocks) and completely ignores the rebalancing associated with an asset allocation “glidepath” towards more bonds as an investor approaches retirement; rebalancing that is typically made when an investors has an asset base materially higher. This level only goes higher if an investor delays the rebalancing process (the benefit of doing so outlined here).


In my opinion, robo-advice is more of a feature than a solution. Robo-advisors lack the advice / hand-holding of a traditional advisor and the tax efficiency of ETFs.  If you desire the latter, why not just allocate to a multi-asset ETF and have some sort of "lock box" feature that requires a 30 minute conversation with a financial advisor, a full week to think, then another 30 minute conversation if you ever want to trade / sell the ETF? To me, that may result in all the behavior correction investors need at a reduced cost.