The Short-Term Nature Of Robo Portfolios

Updated on

The Short-Term Nature Of Robo Portfolios by Stephen J. Huxley and John Y. Kim

Robo advisors have become what textbooks refer to as “disruptive” advancement: the traditional way of doing business in an industry changes dramatically, often due to the introduction of new technology. While technology has had an impact on the financial services industry, it is the quality of the robo advisors’ investment advice that will determine whether they become a lasting paradigm.

One of the many considerations involved in investment strategies is the appropriate time horizon. The asset allocation of robo advisors suggests they are more consistent with equity portfolios structured for short horizons. When evaluated in the context of time segmentation, the equity portfolios recommended by robo advisors for “moderate” investors were most strongly correlated to portfolios designed for one to three years. From a time segmentation perspective, short term does not imply that the portfolio is traded frequently, but rather that it is comprised of equity asset classes that tend to perform better in the short term at the expense of better long-term performance. Most financial planners emphasize that lifetime planning, by definition, should focus on the long term.

Fairly evaluating the investment performance of robo advisors is challenging. It is a dubious practice to judge any investment opportunity that has yet to produce a meaningful track record. Most of the robo portfolios are so new that it is difficult to make comparisons over a sufficiently long period.

However, to their credit, robo providers publish the allocations and funds they use in their portfolios. The allocations change but not precipitously. They are a little different in 2016 compared to 2015 but by less than a few percentage points for any one asset class. They will likely be a little different again in 2017 and in the future, but the changes are minor, at least so far, suggesting they intend to follow a strategic rather than a tactical asset allocation approach.

The open publication of their portfolios reveals the asset classes and allocations they use. Consistency in their allocations and a track record of 30 years or more for these asset classes permits estimates of how well their portfolios would have done for much longer spans of time than would otherwise be possible.

We compared several robo advisors’ equity portfolios to portfolios built to minimize worst-case scenario returns for time horizons from one to 40 years. Given that the average client of a robo advisor is often cited as under the age of 40 and assuming their only objective is saving for retirement, these younger investors should be investing for the long run. Unless robo risk tolerance questionnaires are sufficiently accurate to assess the true aspirations and needs of these younger clients, there is the risk of a mismatch between a robo advisors’ recommendations and the clients’ actual financial planning horizon. By attempting to lower near-term volatility, robo advisor portfolios sacrifice both long-term expected and downside performance for time horizons typically relevant to these clients.

Our definition of risk is not the traditional mean-variance focused volatility (typically quarterly or annual standard deviation). From a financial planning perspective, the risk of shortfall is far more relevant. We utilize the minimax principle and minimize the worst case downside performance time horizons driven by the timing of spending needs as our primary risk metric. For the average robo advisor client, the time horizon on their equity investments is well beyond 10 years, whereas the recommendations of robo advisors are more closely correlated with a one- to three-year horizon.

Background

The purpose of this paper is to investigate the asset allocation recommendations of robo advisors in the context of financial planning where investors typically require long time horizons on their equity investments. We briefly examine the performance over the past 31 years (1985-2015) of the recommended equity allocations for a “moderate” investor using a sample of four robos: Betterment, Motif, Schwab and Wealthfront. Performance was tabulated using www.portfoliovisualizer.com, a website that makes such comparisons relatively easy.

The evidence suggests that the portfolios they recommend for “moderate” investors are neither significantly better nor worse than the S&P 500 Index. However, there is a common theme that all the robos share in terms of performance – their short-term nature. In terms of allocations among asset classes, their portfolios correlate most strongly with equity portfolios designed for short-term, one- to three-year holding periods rather than portfolios designed for longer time horizons.

The average age of for robo clients has been reported to be late 30s to early 40s – a client base that generally requires long time horizons for the bulk of their savings that have been earmarked for retirement. Therefore, one would expect the portfolio to be designed for longer time horizons. If the robo risk tolerance questionnaires are sufficiently valid from a psychometric perspective to determine the true nature of each client’s risk tolerance, then there may not be a problem. But there is significant debate about the validity of risk tolerance questionnaire, and if the questionnaires fail to deliver, it brings into question the possibility of a mismatch between robo recommendations and the nature of their customer base.

The robos

To disguise their identities, the four robos examined are randomly ordered in the charts and tables below. That is, Betterment is not robo 1, Motif is not robo 2, etc.

For their moderate investor portfolios, one robo had an allocation close to 50/50 but three had 70% to 80% in equities (including minor allocations to commodities, REITS and, in one case, gold). Wealthfront provides the most extensive explanation of its investment methodology (Malkiel, 2016).

In this analysis, only allocations to equities (domestic, international and emerging) plus the alternatives listed above were considered (they were recalibrated to sum to 100%). There were several reasons for restricting the analysis to equities. First, the goal was to make an apples-to-apples comparison among the equity recommendations of the robos. Second, the equity portion of the portfolio traditionally serves as the long-term driver of growth for the overall portfolio. Third, most of the robos use index-based ETFs, so it is the allocations among these funds and the asset classes within them that will determine differences in equity performance (Brinson, Hood and Beebower, 1986, 1991). Finally, all but one of the robos had most of their allocations in equities, and the one who did not would likely be unable to keep up with the performance of the others in this sample over the long run.

The data

Data on annual returns and other portfolio computations came from Portfolio Visualizer, which tracks returns for all 36 asset classes shown in Exhibit 1 back to 1985 (with the lone exception of did-cap value, back to 1987), and as far back as 1972 for many.

Exhibit 1. Asset Classes Covered in Portfolio Visualizer

Domestic Stocks International Stocks Fixed Income Alternatives
US Stock Market Intl Stock Market Total Bond REIT
Large Cap Value Intl Value Stocks Global Bond Gold
Large Cap Blend Intl Small Cap Stocks TIPS Precious Metals
Large Cap Growth Intl Developed Markets Long Term Treasuries Commodities
Mid Cap Value Emerging Markets 10-year Treasuries
Mid Cap Blend Intl Pacific Region Intermediate Term Treasuries
Mid Cap Growth Intl Europe Short Term Treasuries
Small Cap Value Cash / Money Market
Small Cap Blend Corporate Bonds
Small Cap Growth High Yield Bond
Micro Cap Short Term Inv Grade
Long Term Tax Exempt
Intermediate Term Tax Exempt
Short Term Tax Exempt

The actual funds used by each robo can be found on their websites. Reports in The Wall Street Journal (Moyer, 2015 (1) and (2)) listed the allocations to specific funds in mid-2015. A check in mid-2016 revealed only minor changes from mid-2015.

The asset classes contained in each fund can be determined from Morningstar. For example, VTI (Vanguard Total Stock Market ETF) was used in three of the robo portfolios. Morningstar partitions funds and ETFs into their style box, listing the approximate percentage of each component equity classes (see Exhibit 2). On the day these allocations were tallied (5/31/2016), 24% of VTI is large value, 25% is large blend, etc. These allocations may drift slightly due to changes of a few percentage points as the markets shift, but most funds attempt to limit deviations from their target allocations

Exhibit 2. Asset Classes Comprising VTI Mutual Fund

Robo Portfolios

Robo Portfolios

Once the asset classes are known for each robo portfolio, Portfolio Visualizer’s evaluation tool will compute compound annual growth rate (CAGR), inflation-adjusted CAGR, standard deviation, best year, worst year, Sharpe ratio, Sortino ratio, U.S. market correlation, international market correlation and other characteristics. (See site for details on methodologies.) The calculations are based on data as far back as the shortest historical record. All data in this paper are based on 1985-2015.[1]

The free website allows the user to save up to 10 different portfolios for comparison purposes. It also includes a number of “lazy” portfolios, based on portfolios openly published by various commentators. The complete database and its sources are readily accessible. Clearly, a lot of work has gone into this site.

See the full article here.

Leave a Comment