Then there’s taxes—half of the firms I surveyed offer municipal bonds in the portfolios; the other half do not. Robo investors should take their tax rates and fixed allocation needs into account when considering the all-in cost. Although Future Advisor doesn’t use muni funds, it does offer asset-location services that can help you avoid taxes on your fixed-income funds.
Tax-loss harvesting can play a huge role in cost reduction, though it is not without risks. Wealthfront, Betterment and Future Advisor each offer tax-loss harvesting services. Each claims hefty returns from the practice.
Rebalancing can reduce a portfolio’s risk. Although all the robo advisors offer some type of rebalancing, FutureAdvisor and Invessence’s are triggered by the calendar, while Betterment and Wealthfront rebalance with cash flows, and sometimes in conjunction with tax-loss harvesting.
Jon Stein, Betterment’s chief executive officer, told me that Betterment’s focus on net after-tax returns—asset location, tax loss harvesting and rebalancing—has more impact on long-term investment outcomes than tactical asset allocation.
In other words, Betterment’s focus affects returns more than tweaking a portfolio’s balance between emerging markets and U.S. value stocks could. That’s quite a claim. I’ll take a hard look at that later on in this blog series.
The more similar the portfolios are to each other, the more the service costs, and the more cost-saving services will be a deciding factor. But, as we saw, there’s some surprising variation in these portfolios.
So, let’s set aside the fees-and-service comparisons, and dig in hard to the investment philosophies powering the robots.
Some robo portfolios are built with inflation control in mind; others seek factor exposure; a few try as best they can to simply mimic the market; and one includes the risk/return expectations of its investment committee. Most use optimization techniques, but one explicitly builds separate sleeves for each asset class.
Here’s a top-line breakdown of their approaches:
Wealthfront, Betterment, Covestor, Wise Banyan and Invessence all use some form of optimization. Invessence uses a quadratic optimizer; the rest use a more conventional mean-variance model. Each determines expected returns and volatilities for the asset classes they want to consider, along with correlations among the asset classes, and then uses a complex algorithm to determine the best weightings for each risk level.
Because mean-variance optimization can sometimes produce highly skewed portfolios, especially when using historical asset class returns as inputs, most of these firms modify their inputs or constrain the outputs.
That’s where things get interesting.