Optimizing Performance in Lower Risk Betterment Portfolios
In this methodology, we provide insight into how we optimize the performance of the lower risk bonds in Betterment's portfolios.
TABLE OF CONTENTS
- The Role of Ultra Low-Risk Assets in a Bond Portfolio
- How we optimize ultra-low-risk bonds to target a higher yield
- Why Two Low-Volatility Funds Result in Our Ultra-Low-Risk Asset Allocation
- Using 30-day historical yields to inform future yields
- Continued bond portfolio research
In this methodology, we provide insight into how we optimize the performance of the lower risk bonds in our 100% bond portfolio. Primarily, Betterment’s optimization method involves the inclusion of short-term, investment-grade bonds in lower-risk allocations of the Betterment Portfolio Strategy.
Why are we exploring this part of how we manage the Betterment Portfolio Strategy?
First, over time, we’ve improved the mix of bonds in our portfolios to control risk without compromising expected performance—the main focus of this methodology.
Second, many investors may not yet know about the important role of ultra-low-risk bonds in the portfolio we recommend. When investors opt for a 100% bond, 0% stock allocation in their Betterment portfolio, the only assets in the portfolio are bonds with ultra-low-risk profiles.
The Role of Ultra Low-Risk Assets in a Bond Portfolio
One of our key investing principles is maintaining diversification. Effective diversification means taking as little risk as possible to achieve your growth target.
For portfolios with lower risk levels, adding in ultra-low-risk bonds can help reduce risk without adversely affecting returns. We consider U.S. short-term Treasuries and other US. short-term investment-grade bonds to be ultra-low-risk (although all investments carry some measure of risk).
At every risk level, Betterment invests in a portfolio that we expect to have the highest rate of return relative to its risk. These portfolios are considered to be on the “efficient frontier,” or the boundary of highest returning portfolios at any given level of risk. By further diversifying bond holdings with ultra-low-risk assets, the Betterment Portfolio Strategy pursues higher expected returns with less risk than portfolios that do not include these low-risk assets.
Graphically, you can see below that the highest returning portfolios for lower risk levels (i.e., levels of volatility) are portfolios that include ultra-low-risk bonds (the black line). Additionally, certain low risk portfolios could not be achieved at all without adding ultra-low-risk assets. As you can see below, portfolios constructed without ultra-low-risk bonds (the blue line) are unable to achieve a volatility lower than approximately 7%.
Figure 1. Betterment’s efficient frontier including ultra-low-risk bonds
Expected returns are computed by Betterment using the process outlined in our methodology optimizing the Betterment Portfolio Strategy. Volatilities are calculated by Betterment using monthly returns data provided by Xignite.
At a certain point, including ultra-low-risk bonds in the portfolio no longer improves returns for the amount of risk taken. This point is called the ‘tangent portfolio.’ For the Betterment portfolio strategy, the tangent portfolio is our 43% stock portfolio. Portfolios with a stock allocation of 43% or more do not include ultra-low-risk bonds.
When a portfolio includes no stocks—100% bonds—the allocation suggests an investor has no tolerance for market volatility, and thus, our recommendation is to put the investor’s money completely in ultra-low-risk bonds.
How we optimize ultra-low-risk bonds to target a higher yield
As you can see in the chart below, we include our U.S. Short-Term Investment-Grade Bond ETF and our U.S. Short-Term Treasury Bond ETF in the portfolio at stock allocations below 43% for both the IRA and taxable versions of the Betterment Portfolio Strategy.
At 100% bonds and 0% stocks, a Betterment portfolio consists of 80% U.S. short-term treasury bonds and 20% U.S. short-term investment-grade bonds. If an investor were to increase the stock allocation in their portfolio, the allocation to ultra-low-risk bonds decreases, though the relative proportion of short-term U.S. treasuries to short-term investment-grade bonds remains the same. Above the 43% stock allocation threshold, these two assets are no longer included in the recommended portfolio because they decrease expected returns given the desired risk of the overall portfolio.
In line with our fund selection process, we currently select JPST – JPMorgan Ultra-Short Income ETF to gain exposure to the U.S. short-term investment-grade bonds and we’ve selected SHV – iShares Short Treasury Bond ETF to gain exposure to U.S. short-term treasury bonds.
To summarize the fund selection process, we start with the universe of bond ETFs with average maturities of less than 3 years, given the relationship between maturity length and risk. We further reduce the set of candidates by ruling out ETFs with unfavorable risk characteristics, including those with excessive interest-rate risk or high overall volatility. We then filter for funds with sufficient liquidity, so that we can maintain low costs for investors. Finally, we select the fund with the lowest combination of expense ratio and expected trading costs: JPST. The same process led to our selection of SHV.
Why Two Low-Volatility Funds Result in Our Ultra-Low-Risk Asset Allocation
Short-term US treasuries and investment-grade bonds are both inherently low-risk assets. As can be seen from the chart below, short-term U.S. treasuries (SHV) have low volatility (any price swings are quite mild) and smaller drawdowns (the length and magnitude of periods of loss are muted). Though slightly more volatile than short-term treasuries, the same can be said for short-term investment grade bonds (JPST).
Figure 2. SHV and JPST
The above chart shows the historical growth of $1 invested in each investment option from 9/30/2013 (the first date both funds SHV and NEAR (which is used as a historical proxy for JPST) were in existence) to 5/23/2018. Performance is net of fund-level fees and does not include any management fees from Betterment. Dividends are assumed to be reinvested. The composite is assumed to be rebalanced daily at closing prices. JPST fund inception as in May, 2017. For returns data in May, 2017 and earlier, we use returns from a comparable fund, NEAR. JPST replaced NEAR as our U.S. short-term investment-grade bond fund in Dec, 2019. Data: xIgnite.
It’s also worth noting that these two asset classes do not always go down at exactly the same time. By combining these two asset classes, we are able to produce a two-fund portfolio with a higher potential yield and the same low volatility.
In fact, combining these asset classes resulted in smaller historical drawdowns in performance that lasted for fewer days than was the case for either asset class individually. As you can see from the chart below, the combination of U.S. short-term treasury bonds and U.S. short-term investment-grade bonds used for the Betterment 100% bond, 0% stock portfolio (blue) generally had shorter, less severe periods of down performance than either fund by itself.
Figure 3. Drawdowns in performance
The above chart shows the largest drawdowns in performance from Sept 30, 2013—the first date both funds SHV and NEAR, which is used as a historical proxy for JPST—were in existence) to 11/14/2019. Performance is net of fund-level fees and does not include any management fees from Betterment. Dividends are assumed to be reinvested. The composite portfolio (blue) is assumed to be rebalanced daily at closing prices to maintain a 80% SHV, 20% JPST weighting. JPST fund inception was in May, 2017. For returns data in May, 2017 and earlier, we use returns from a comparable fund, NEAR. JPST replaced NEAR as our U.S. short-term investment-grade bond fund in Dec, 2019. Data: xIgnite.
Using 30-day historical yields to inform future yields
A reasonable question about this methodology is how to interpret the potential returns of the composite looking forward. As with other assets, the returns for ultra-low-risk bonds include both the possibility of price returns and income yield. Generally, price returns are expected to be minimal, with the primary form of returns coming from the income yield.
Below you can see that the prices for the composite of 80% SHV and 20% JPST tends to stay fairly constant, while the price with dividends grows through time. This shows that the yield (paid by the funds through monthly dividends) is responsible for almost all of the growth in these funds.
Figure 4. Growth of $100 in the Betterment 0% stock portfolio
The above chart shows the historical growth of $100 invested in a portfolio that consists of 80% SHV and 20% JPST. Data is from 6/1/2017 (the first full month both funds SHV and JPST were in existence) to Nov. 14, 2019. Performance is net of fund-level fees and does not include any management fees from Betterment. Dividends are assumed to be reinvested. The portfolio is assumed to be rebalanced daily at closing prices. Data: xIgnite.
Looking at 30-day SEC yield—a standardized calculation of yield that includes fees charged by the fund—we can get a good sense of the expected performance for these low-volatility assets. Why can we believe this?
First, performance is determined by both yield and price change, and because there is low price volatility in these assets, yield is the primary component of performance. We use the SEC 30-day historical yield as an expectation of annualized future yield because it’s the most recent 30 days of yield performance, and we generally expect future yield to be similar to the last 30 days, although past performance does not guarantee future results. We expect this to be the case because the monthly turnover in these funds is relatively low. However, the yield can be expected to change—either up or down—as market conditions, including interest rates, change.
The yields you receive from the ETFs in Betterment’s 100% bond portfolio are the actual yields of the underlying assets after fees. Betterment does not adjust the yield you earn according to our discretion, as a bank savings account could. A bank may choose not to adjust its interest rate higher as prevailing rates rise, or may cut its interest rate. Because we are investing directly in funds that are paying prevailing market rates, you can feel confident that the yield you are receiving is fair and in line with prevailing rates.
Below we can see that over the 30 days ending Nov. 14, 2019, SHV had an annualized yield of 1.59%, net of fund fees, which is dispersed to shareholders on a monthly basis. Over the same period, JPST yielded 2.12%. The 100% bond portfolio, composed of 80% SHV and 20% JPST, has yielded 1.70%.
Table 1: Risk and Yield
|Ultra low-risk bond baseline (SHV)||Additional candidate fund (JPST)||SHV 80% + JPST 20%|
|SEC 30-day yield (includes fund expense ratios)||1.59%||2.12%||1.69%|
|Deepest drawdown return||-0.12%||-0.34%||-0.12%|
SEC 30-day yields are as of Nov. 14, 2019. Annual volatility and drawdown return are calculated from Sept. 30, 2013 —the first date both funds SHV and NEAR, which is used as a historical proxy for JPST—were in existence) to Nov 14, 2019. For returns data in May, 2017 and earlier, we use returns from a comparable fund, NEAR. JPST replaced NEAR as our U.S. short-term investment-grade bond fund in Dec, 2019.SEC 30-day yields, annual volatility and drawdown are net of fund fees and do not include Betterment’s management fee. Data from Xignite and Betterment calculations.
Bond portfolio research never stops.
By combining multiple low-risk assets, we seek to deliver higher expected returns, through higher yields, while keeping risk in check. The diversification benefits of U.S. short-term treasuries and investment-grade bonds allow us to construct low-risk portfolios with shorter and less severe downturns.
As always, we iterate on our portfolio optimization methods. We update our changes in this overview of our financial advice as we develop improved ways of helping you reach your financial goals.