But there's a catch. Over this span, the U.S. stock market suffered some setbacks that lasted a few months, but never had a severe and prolonged decline, nothing like the dot-com collapse from 2000 to 2003, or the deep recession from 2007 to 2009. How might low-volatility funds perform, were they active during such periods?
Although no low-volatility fund or ETF was actively available at those times, back-dated results exist for the indexes that dictate stock-selections for USMV and other similar funds. The index values were generated by MSCI. They indicate what a fund currently pegged to the index might have done, had it been active since 1988 and had it been able to track the index closely. As I noted in an earlier post, retroactively generated indexes show drawdowns of about -40% for minimum-volatility stocks, both in the U.S. and internationally, between 1994 and 2019. Admittedly, broad markets were even more extreme, reaching cumulative losses of -55% or worse. Still, the index history argues that buying and holding a minimum volatility fund or ETF is no guarantee of shelter from stormy markets.
Rotating Minimum-Volatility Funds
A tsunami sinks all boats, even those more bouyant than most. To be secure when a tsunami threatens, you have to be out of the water, far inland, on high ground.
For investors exposed to financial storms, shelter is found in methods that remove most or all assets from the stock market (or in special instruments like put-options and inverse ETFs that offer downside insurance, typically with additional cost or risk). The method described here, called rotation, requires modest effort and some discipline, but is accessible to ordinary investors.
Shown below are the hypothetical results from using one specific version of a rotation strategy for the 25-year period from July 1994 through June 2019. It is not a buy-and-hold strategy. Nor is it buy-and-rebalance. Instead, it represents a portfolio that is invested, most of the time, in a specific mix of mostly U.S. and international minimum-volatility indexes, plus a small portion of long-term bonds. Sometimes, in a simple but well-defined manner, it rotates to being invested entirely in a broad selection of bonds.
Building a Rotation Portfolio
In terminology introduced in my post Time to Rotate Out of Stocks?, the strategy of the portfolio in the chart is Minimax Rotate 80-or-0, applied to minimum-volatility funds. The objecitve is a compromise between maximizing returns (by holding up to 80% in low-volatility stocks) and minimizing drawdowns (by sometimes rotating to 0% in stocks).
The data in the chart simulates this portfolio with historical data for mutual funds and index values. If implemented today with ETFs, it would be constructed as follows:
- At all times, 20% of the portfolio is invested in BLV, a Vanguard ETF that holds a mix of long-term corporate and government bonds. The simulation constructed an equivalent fund by combining actual historical returns of two Vanguard mutual funds, VWESX (long-term investment-grade bonds) and VUSTX (long-term U.S. Treasuries).
- Either 20% or 0% is invested in EFAV, an iShares ETF of minimum volatility stocks in developed countries outside the U.S. The simulation used the corresponding MSCI index.
- Either 60% or 0% is invested in USMV, an iShares ETF of minimum volatility stocks in the U.S. The simulation used the corresponding MSCI index.
- When assets are rotated out of EFAV and USMV, they are held in BND, a Vanguard ETF that tracks an index of the entire U.S. bond market. The simulation used actual historical data for VBMFX, Vanguard's equivalent mutual fund.
To manage its holdings, the portfolio follows a few simple rules:
- Once a month, compare the 12-month total return of EFAV (non-U.S. minimum volatility stocks) to that of BND (the bond alternative). If BND has the better 12-month return but funds are currently invested in EFAV, then move the EFAV amount to BND.
- Similarly, at the same time every month, compare the 12-month total returns of USMV (U.S. minimum volatility stocks) and BND (the bond alternative). If BND has the better 12-month return but funds are currently invested in USMV, move the USMV amount to BND.
- If, at the same time every month, BND holds funds but has a smaller 12-month total return than EFAV, USMV, or both, transfer assets from BND to the better-performing alternative(s). For example, suppose EFAV and USMV were both at $0; both had better 12-month returns than BND; and the amount held in BMD were $10,000. Then $7500 would go USMV and $2500 to EFAV because their targets, 60% and 20%, respectively, have a 3-to-1 ratio. As another example, suppose that both EFAV and USMV had better 12-month returns than BND; that funds were already invested in USMV from a previous month, but EFAV had $0; and that BND held $3,500. Then $3,500 would be moved to EFAV.
- At the same time every month, rebalance funds if the drift from current targets exceeds 5%. For example, suppose the targets this month were 60% BND, 20% EFAV, and 20% BLV, but the actuals were 63%, 22% , and 15% respectively. The two above-target ETFs sum to an excess of 5%. Therefore, move funds from BND and EFAV into BLV, to achieve the target percentages. In the 25-year simulation, rebalancing occurred rarely, and typically not when rotation was necessary. It can happen, however, that rotation and rebalancing are both required in the same month. For that reason, the monthly calculations should occur in a particular order. First, determine whether rebalancing is necessary; second, assess whether to rotate funds; and third, transfer amounts as needed.
Some Technical Details
- Why these target percentages? For low-volatility stocks, my analysis of historical data implied that 80% is the best ceiling, and 0% the best floor, if the investor's goal is to find an optimal compromise between maximizing returns and minimizing drawdowns. The ceiling value of 80% is higher for low-volatility stocks than for broad market indexes such as the CRSP total market index or the S&P 500 (50% to 60% was best as a minimax ceiling for broad indexes). Another finding in my historical simulation was that a modest portion of non-U.S. low-volatility stocks generated better results than investing only in the U.S. version or equally in the U.S. and non-U.S. versions. Dividing the 80% ceiling into 60% U.S. and 20% non-U.S. worked well historically, although there is no guarantee that it would be the optimal mix in the future.
- Why a 12-month window? Low-volatility stocks fall less rapidly than the average for all stocks. Consequently, a longer window is needed to detect a reliable change in outlook. The historical data implied that for low-volatility funds, 12 months worked better than shorter windows.
- Why BND as the bond alternative? A broad bond index fund like BND has, on average, less volatility than a fund that holds long-term bonds, particularly Treasuries. To evaluate whether a bond-fund or a stock-fund is currently the better choice, the historical data favored comparing funds with similar volatilities. Thus, less volatile stock-funds like USMV and EFAV are best compared with a low-volatility bond-fund like BND.
- Why once a month? If the decision points are too frequent, performance can decline. Although I did not have sufficient data to conduct a thorough search for the very best level of frequency across countries and times, I did conduct some analysis of recent data for the U.S. Taking action once a month tended to be better than weekly and quarterly, although sometimes every other month was virtually as good as every month. That said, the specific time of the month might not matter. My analysis used end-of-month values, but one might do equally well by checking every month on the 15th day, for example, or at some other reference point. The idea is to be consistent, every month, just once a month.
- Do flip-flops occur? Despite the 12-month window, the simulation had rare occasions when the portfolio would have repurchased a particular ETF one month after having sold that same ETF. In such cases, the simulation temporarily held funds in the equivalent of BSV, a Vanguard index of short-term bonds. Doing so could be useful to comply with a brokerage firm's limits on frequent trading or to avoid the tax-consequences of a "wash sale." Alternatively, one could invest temporarily in a replacement fund that is similar, but that tracks a different index (for example, SPLV instead of USMV).