How to Construct a Low-Cost Conservative Portfolio

Loss potential of traditional asset allocations

One of the striking findings from PIMCO’s research was that many asset allocations that have traditionally been considered fairly safe actually had loss potentials much higher than plan consultants indicated were prudent. PIMCO’s researchers concluded that many target-date funds had potential loss levels considerably worse than the acceptable levels suggested by plan consultants. This problem was especially acute within 10 years of expected retirement, when investors expect to be taking the least risk.

My research has provided risk projections that have been fairly accurate. In 2007, for example, the risk projections generated by Quantext Portfolio Planner (QPP), the Monte Carlo-based portfolio management tool that I designed, correctly predicted that market volatility would increase substantially.

Let’s start by showing risk projections for a range of traditional asset allocation benchmarks and for mutual funds that are specifically classified as conservative.

All statistical models, including Monte Carlo simulations, are of limited value for estimating the probability of extreme events, not least because the historical record contains few occurrences of those events. When I compared the projected risk levels from Monte Carlo simulations in 2007 to what actually occurred in 2008, the magnitude of losses in equities was three standard-deviation events – with a probability of only 0.15% in a 12-month period. We have no way of knowing how probable 2008’s outcomes were, given conditions in 2007. As Moshe Milevsky pointed out in 2009, however, it is far more important to estimate the worst possible outcomes (and to make sure that these are survivable) than to worry about whether the estimated probability of this outcome is accurate. Extreme events will occur more frequently than models suggest and it is prudent to plan for them. I will use an estimated three-standard-deviation outcome as the basis for estimating maximum potential losses.

Below are the current projected risk levels for a number of stock/bond allocations. These values are generated by QPP, based on the last five years of return data

Monte Carlo estimates of 12-month maximum loss potential and 2008 returns

The estimated worst-case loss potential for the S&P500 is -37%, remarkably close to the actual loss in 2008. In the case of bond returns, 2008 was not an example of what a worst case looks like now. In 2008, bonds were a safe asset class, with AGG delivering a 7.6% return.

Given the increased risk in bonds today that QPP predicted, relative to 2008, I verified its results. I calculated a two- and three-standard-deviation loss, given the current yield as the expected return and using the option-implied volatility as the expected standard deviation of return. According to Morningstar, the current implied volatility for at-the-money put options on the longest-dated options on AGG is 6.1%. With a 12-month yield of 2.4%, a two-standard-deviation loss results in a return of -9.8%. Even though we have not experienced substantial losses in bonds in recent memory, we must still plan for the possibility.

Given that the 100% bond portfolio (100% AGG) has the lowest risk among these portfolios, why not simply treat a 100% bond portfolio as the default conservative portfolio? The answer is that yields are near historical lows, and investors cannot afford to have portfolios so vulnerable to interest rate risk.

To provide additional insight into low-risk portfolios, the table below shows the estimated risk levels, 2008 returns, trailing 12-month yields and correlations of a range of bond funds and conservative asset allocation mutual funds.

Aside from three bond ETFs (IEF, BSV, and LQD), these are all mutual funds that are classified as “conservative allocation” by Morningstar. The 80% AGG / 20% SPY portfolio is included as a reference.

Projected loss potential, yield, and interest rate exposure for conservative allocation funds

These results highlight the tradeoffs between portfolio loss potential and interest rate exposure. Funds with low loss potential tend to have high sensitivity to changes in interest rates. Risk-averse investors have traditionally held large allocations of Treasury bonds, but this type of portfolio tends to be adversely affected by rising rates. To illustrate interest-rate exposure, I show the correlation of fund returns to the 10-year Treasury bond yield using the last five years of data.

While it is a good bet that Treasury bond yields must eventually go up, the timing is unknown. We could end up with a decade-long period of extremely low interest rates and bond yields, similar to Japan. The most conservative approach is to create low-risk portfolios with fairly neutral exposure to interest rates. A portfolio for which returns have zero correlation to Treasury bond yields would be insensitive to changes in interest rates.

The Vanguard Life Strategy Income fund (VASIX) is attractive in this regard, with a modest -10.5% correlation between its returns and the 10-year Treasury bond yield, along with an estimated 12-month maximum loss of -10%. The Nationwide Investor Destinations Conservative fund (GIMCX) also looks relatively attractive. The one drawback of these funds is that their yields are low; both have 12-month yields of 2.2%.

The Vanguard Wellesley Income fund (VWINX) has a 12-month yield of 3%, but it has a potential worst-case 12-month return of -14%. Given that this estimated loss level is notably worse than the fund’s 2008 return and that my model has inherent uncertainties, VWINX is attractive for risk-averse investors despite its higher projected loss potential.

The 20% SPY/80% AGG portfolio has a projected worst-case return of -10% over a 12-month period and a -27% correlation to Treasury bond yields. The various conservative allocation funds do a better job of managing interest-rate risk than the 20/80 portfolio does, but their estimated potential losses are worse.

Building a better portfolio

In light of estimated risk levels and interest-rate exposures, what alternatives exist to traditional asset allocations and conservative allocation funds? I started with a series of index ETFs representing a range of asset classes and added the conservative allocation funds above. I used QPP and an optimizer to create a portfolio with zero correlation to the 10-year Treasury bond yield, an estimated 12-month three-standard-deviation return of no worse than -10% and maximum yield given those constraints. I allowed unlimited exposure to a short-term bond fund (BSV) but constrained the maximum allocation to any other fund to 15%.

This portfolio has the following features:

1. Projected 12-month loss potential of -9.8% (estimated three-standard-deviation loss)
2. Correlation to 10-year Treasury bond yield of 0%
3. Trailing 12-month distribution yield of 3.25%
4. 2008 return of -8.47% (model weights at the start of the year and does not rebalance)
5. Asset-weighted expense ratio of 0.43%

This portfolio has the key features that investors seeking safe portfolios require, although its 12-month potential return of -9.8% may not meet everyone’s needs.

What clearly differentiates this portfolio from the low-risk conservative allocation funds is its neutral exposure to interest rates and substantially higher yield. In historical terms, a 3.25% yield is nothing special. In the current environment, however, a 3.25% yield provides a 1.75% real return, based on the current Consumer Price Index, which has risen by only 1.5% in the past year.

Discussion

Investors must understand the spectrum of risk and return that the current market presents. Bond investors have reason to be cautious. Stocks don’t look particularly cheap, either, with the Shiller PE10 ratio at almost 23 and the S&P500 dividend yield near record lows. Stocks or bonds at current prices do not offer much of a margin of safety. Investors who have low risk tolerance must carefully diversify to achieve return while limiting risk – particularly exposure to interest-rate changes. A common choice for such investors is the conservative allocation funds. My analysis of these funds demonstrates, however, that they have widely varying risk levels and exposure to interest-rate risk.

It is possible to create a low-cost portfolio that is neutral with respect to 10-year Treasury bond yield and has higher yield and lower risk than the two most attractive conservative allocation funds, the Vanguard Wellesley (VWINX) and Berwyn Income (BERIX) funds. A 50/50 allocation between VWINX and BERIX has a 12-month distribution yield of 2.75% and estimated loss potential of -14.8% for a 12-month period, considerably worse than that of the model portfolio (3.25% and -9.8%, respectively). The estimated loss potential for the 50% VWINX/50% BERIX portfolio is consistent with the losses from those funds in 2008.

The model portfolio developed here is tailored to balance interest-rate exposure, loss potential and yield in the current market. In other market conditions, this portfolio might be less optimal, but its features are well suited for investors in today’s unprecedented bond-market conditions. Conservative investors will ultimately judge their advisors on whether they skillfully managed the range of risks in an environment with little margin for error. The model portfolio presented here achieves those goals based on sound methodology.

Geoff Considine is founder of Quantext and the developer of Quantext Portfolio Planner, a portfolio management tool. More information is available at www.quantext.com.

Geoff’s firm, Quantextis a strategic adviser to FOLIOfn,Inc. (www.foliofn.com), an innovative brokerage firm specializing in offering and trading portfolios for advisors and individual investors.

Read more articles by Geoff Considine