The world is brimming with Betas. the U.S.-listed catalog of index funds now totals more than 800, available in three basic design wrappers: mutual funds, ETFs and the latest twist known as ETNs (exchange-traded notes), according to Morningstar Principia. More are arriving all the time, offering an ever broader list of betas in both conventional and, increasingly, alternative forms. But no matter how many betas the financial industry securitizes for general consumption, the old challenge of mixing and managing them in pursuit of strategic goals isn't getting any easier. If anything, portfolio design is becoming tougher in a world flooded with choice. More betas can be better, of course, but with rapidly expanding possibilities for customizing comes a higher risk of mediocrity, or worse.
The hazard convinces some to look to Mr. Market for buy and sell signals in running a multi-asset class portfolio. The strategy falls under the heading of rebalancing, although the variations on the theme are numerous. A popular system is setting strategic weights for asset classes and rebalancing back to those weights when they're breached. In the June issue of Wealth Manager, we spoke with one strategist who advocated rules for when and how to rebalance in the quest for superior results ("Weights and Bands," p. 59, June 2007).
This month we visit with Mebane Faber, who promotes a related strategy for capitalizing on market volatility--albeit one that takes a more aggressive tone on allocations. The portfolio manager of Los Angeles-based Cambria Investment Management, Faber says there are benefits to using market-momentum signals for switching between all cash and long positions for each asset class in a diversified portfolio. The main attribute is sharply lowering risk, while maintaining a comparable level of return relative to simply buying and holding the same assets.
The risk reduction flows from a rules-based strategy for an equal-weighted mix of five major asset classes, Faber explains. The tactical asset allocation model for a given asset class is one of moving from a 100 percent weighting to all cash when-- based on month-end closing prices--the relevant index closes below its trailing 10-month moving average. When the month-end index closes above its moving average, the allocation for the asset class swings back to 100 percent investment.
Simple but effective, argues Faber, who previously worked as a quantitative analyst at a futures broker/dealer and before that, as an equity analyst at the Genomics Fund. He adds that the strategy, when used for building a portfolio of multiple, long-only betas, shares more than a passing resemblance to a hedge fund of funds--less the high fees, liquidity issues, etc.
Faber detailed his findings in a paper published in the Spring 2007 issue of The Journal of Wealth Management ("A Quantitative Approach to Tactical Asset Allocation"). The paper's basic conclusion may or may not persuade, but it's still worth a read for its perspective on the relationships between risk management, market momentum and rebalancing/market timing in a multi-asset class context.
As for Faber's motivation for undertaking the study, his inquiry is only partly academic. He is managing director as well as portfolio manager of the recently launched asset management arm of the boutique investment bank Cambria Capital. As you might expect, the portfolio strategy for Cambria Investment Management's high-net-worth clients is informed by Faber's research.
But no matter how impressive the back-testing appears, there's always the threat of getting snookered by history. What looks triumphant on paper is all too often disappointing when deployed with real money going forward. Still, he is confident that his findings transcend the pitfalls of data mining.
"A key check against data mining is a system that works across a number of time periods as well as a number of different markets," Faber told Wealth Manager in a recent interview. "We tested it in more than 20 markets and over numerous decades, and the results were consistent."
For a discussion of the strategy's broader implications, read on.
What have you learned about tactical asset allocation from your research?
The first lesson is that diversification works, as per Markowitz, who showed that owning a number of risky assets can create a portfolio that's much less risky [than the components in isolation].
In the paper, I use the indices of five asset classes--U.S. stocks, foreign stocks, bonds, commodities and real estate. The last two--commodities and real estate--are asset classes that most people tend to exclude. But adding more risky asset classes brings down the risk of the portfolio as a whole. Worldwide exposure to market betas can form a highly diversified portfolio.
The second lesson is that using simple risk-management--in this case, a price-based, trend following system--can vastly improve risk-adjusted returns. It does that not by trying to beat the market, necessarily, but by reducing risk as measured by volatility and drawdowns [declines from a performance peak]. Overall, the model improves the risk-adjusted returns of the portfolio relative to a buy-and-hold approach.
Can you put a number on the improvement?
On average, my timing model--using five asset classes--reduces portfolio volatility as measured by standard deviation by about 35 percent compared to a buy-and-hold, equal-weighted strategy with the same asset classes (see table below). The model also reduces risk, measured by the maximum drawdown, by around 50 percent.
When you're dealing with retail investors, the main statistic they look at when they think of risk is drawdown--how much they're losing or how much their account is down. They don't notice the day-to-day fluctuations as much; they care much more about losing money. So drawdown, to me, is the most important statistic when you're thinking about a portfolio for individuals.
Do the results of your tactical asset allocation (TAA) study support or refute the Brinson study, which asserted asset allocation's influence on portfolios?
I think it agrees with it. A lot of people misquote the study, saying that the majority of returns come from asset allocation; in fact, it's the variability of returns that are driven by asset allocation. We agree with that. But I also like to say that most people don't include certain asset classes, especially commodities and real estate, which add a heck of a lot of value in a diversified portfolio. Excluding them can show how much you can improve your portfolio results by using those asset classes. So, yes, we consider [asset allocation] to be the biggest decision. Unfortunately, most people don't do it correctly.
Your study considers TAA with an equal weighting of five broadly defined asset classes. Does this strategy have any basis in what's practiced in the real world?
If you look at the portfolios of the Harvard and Yale endowments, and strip out the private equity and hedge fund exposures that private investors don't have easy access to, the portfolio weightings are roughly 20 percent for each of the five asset classes I studied. Bonds are a little underweight, and foreign equities are a little overweight, but overall it's close to an equal weighting.
Do you limit client portfolios to five broad asset classes?
For larger accounts, we identify as many as 40 asset classes, but I kept the study as simple as possible. I wanted to examine the five broadest asset classes, and readers can take it from there. In practice, you want as many uncorrelated sources of beta as possible. As the market evolves, as innovation delivers more products, a lot of what was considered alpha in the past is getting commoditized and reappearing as a low-cost source of beta.
How do you come up with 40 asset classes?
For example, you could break U.S. equity up by size--large cap, mid-cap. Or, you could break it out into the 10 sectors. Why? Because utilities shouldn't have a whole lot of correlation with say, consumer staples or basic materials.
With bonds, you find the least amount of improvement by timing the components, but that's because fixed income is the least volatile asset class overall. Of course, you could add high-yield bonds or corporate bonds, in which case you might be able to extract more alpha from timing.
Your study doesn't use emerging market equities. Do you exclude developing market stocks for clients?
No, we use emerging markets. Generally, we use 10 asset classes for smaller accounts, including foreign developed and foreign emerging markets. For larger accounts, we time the top-10 country components within the two foreign market buckets with ETFs and closed-end funds. There are still a few holes in ETF coverage. We'd like to see emerging market bonds, foreign developed market bonds and municipals, for example.
Is there a particular type of market that offers a better tailwind for the TAA strategy reviewed in your paper?
And the worst climate?
A choppy, sideways market.
Why? Because it could generate a lot of false signals that trigger trades?
Right. But it's not as bad as it used to be because commissions are low, and so the cost of getting out isn't so much.
How would taxes impact your TAA model in the real world?
One of the nice things about the model is that it's relatively inactive. As presented in the paper, it does less than one round trip per asset class per year. And when you look at the return distribution of the trades from the model, all of the losses were short-term losses and the majority of the gains were long-term gains. So it's fairly tax efficient from that perspective.
Ideally, you'd trade this in a tax-deferred account, as you would any active strategy. But I doubt that the impact of trading is going to be very significant for the TAA strategy compared to a buy and hold.
In another paper ("Comparing Returns: Market Timing Versus Hedge Fund Indices," March/April 2007, The Technical Analyst), you write that your five-asset-class TAA portfolio is similar to a hedge fund of funds strategy.
Comparing the timing model to hedge fund of fund and hedge fund indices shows that a worldwide asset allocation strategy is the same as hedge fund of fund indices. In fact, the returns of the TAA model are superior to hedge fund of fund indices. One of the biggest problems with the hedge fund space is fees. If you tack on 2 percent and 20 percent [a typical hedge fund fee structure] and another 1 and 10 [for the fund of funds fees], the underlying portfolio has to return something like 18 percent or 19 percent [a year] just to return what a buy-and-hold strategy for the five conventional asset classes would return.
Speaking of hedge fund indices, some say the investable kind--which are set to become available as mutual funds and perhaps ETFs at some point--are similar to owning a portfolio of the major asset classes.
That's basically our thesis, too. By doing something as simple as just buying the five asset classes in the TAA strategy, you'll do as well as a hedge fund of funds, and you'll avoid the headaches and problems of investing in hedge funds, such as liquidity requirements and the extra paperwork.
James Picerno (firstname.lastname@example.org) is senior writer at Wealth Manager.