Investors are undoubtedly familiar with the concept of diversification, which is a fundamental idea underlying modern portfolio theory.
The phrase, “Don’t put all your eggs in one basket” is one that has been familiar to investors for thousands of years. At least intuitively, it has been so since ancient Roman entrepreneurs diversified against the loss of their cargo by shipping trade goods on multiple merchant ships, rather than on a single ship.
When one is properly diversified, one is not overly concentrated in one investment. The diversified investor spreads risk over many different areas, capturing the returns from multiple asset classes while reducing the volatility of individual return streams. This generates more consistent and stable overall returns for the portfolio. While economist Milton Friedman astutely observed that “there’s no such thing as a free lunch,” in the context of portfolio management, diversification is what AQR Capital’s chief investment officer Cliff Asness calls “the one free lunch of investing.” In speeches, Asness has observed, “When you see a free lunch, the only rational thing to do is eat.”
Which Asset Classes?
In today’s sophisticated global financial markets, there are countless possibilities available to the creative investor. An individual could invest in anything from Apple (AAPL) or Exxon Mobil (XOM) to esoteric securities traded on various international exchanges. Perhaps in an ideal world, an investor might strive for exposure to every asset class under the sun. From a practical standpoint, this is impossible. The question then becomes one relating to how to select asset classes that help achieve one’s personal risk and return goals after controlling for the FACTS: fees, access (liquidity), complexity, taxes, and search costs.
Consider the Yale endowment, the gold standard for institutional investing. Here are the fiscal-2016 target allocations Yale announced in September 2015:
- Absolute return: 21.5%
- Leveraged buyouts: 16.0%
- Foreign equity: 14.5%
- Venture capital: 14.0%
- Real estate: 13.0%
- Natural resources: 8.5%
- Bonds and cash: 8.5%
- Domestic equity: 4.0%
This is a wide-ranging list of asset classes, and several create implementation challenges for the average individual investor. For example, how does one establish exposure to private equity, an asset class characterized by high fees, illiquidity and variable manager performance?
Individual investors should accept their own limitations and not feel compelled to invest exactly how Yale invests. Instead, investors can reduce the number of asset classes for consideration to a manageable number, where investment can be practically achieved while receiving the lion’s share of the benefits that accompany a much greater degree of diversification.
Exchange-traded fundsoffer a compelling option for achieving the desired exposures in these asset classes. The primary advantages of using ETFs is that they provide the desired exposure, are low cost and are tax efficient.
Won’t a Complex Approach Provide a Good Solution?
Thus far, we have determined which baskets we’ll put our eggs in. The next stage of the discussion involves how many eggs to put in each basket. So where should an investor start? Perhaps looking to the winner of a Nobel Prize and his ultra-sophisticated methodology?
Harry Markowitz won a Nobel Prize for his work on mean-variance optimization, which offers a compelling asset allocation story. Mean-variance optimization involves calculating the historical returns of each asset and correlations for each pair of assets and solving for the efficient frontier at which returns are maximized and risk is minimized.
But a funny thing happened when his ideas were applied in the real world: Mean-variance performed poorly. The reason why is that the historical returns and correlations, which informed the input for the optimization algorithm, were different from what investors actually experienced in the real world in the future. That is, the estimates were “unstable,” in the sense that they turned out to be guesses that simply weren’t any good.
This is something to keep in mind when considering any optimization method for asset allocation: Complexity does not equal value.
Markowitz himself may have realized the value of simplicity intuitively. Jason Zweig’s book, “Your Money and Your Brain” (Simon & Schuster, 2008), highlights an interesting conversation with Harry Markowitz. In their conversation, Zweig asks Markowitz how he invests his own money. Markowitz responds by admitting,
“I should have computed the historical covariance of the asset classes and drawn an efficient frontier…I split my contributions 50/50 between bonds and equities.”
Consider the irony: The founder of modern portfolio theory uses an equal-weight allocation.
Today, now that there is enough data, we can show that Markowitz’s model doesn’t outperform a simple equal-weight allocation. The reason for this underperformance is a not a critique of the model, which is clearly an incredible intellectual achievement but has everything to do with the practical realities of accurately estimating a covariance matrix.
Complexity and fancy models don’t necessary translate into good practical ideas. This implies that a Ph.D. in finance is not needed to implement a reasonable asset allocation strategy.
Maybe allocating using complicated approaches such as mean-variance optimization aren’t ideal. Perhaps simpler is better.
Let’s go back to Yale, which could be a reasonable baseline. So how does Yale do it? If Yale’s target allocation percentages for fiscal 2016 are examined, it’s obvious that they use different weights for each asset class. Furthermore, Yale’s allocations for 2016 differ from those targeted for 2015 or even 2014.
David Swensen, chief investment officer for the Yale endowment, has dozens of analysts and economists working for him, as well as many specialized investment companies and third-party advisers. This team of experts all give him input and perspective on how to allocate. Do you have a similar network to help inform your decision-making? Probably not. Rather than rely on an army of analysts, perhaps there is a simpler, rules-based approach that can be relied upon.
Simple Versus Complex Asset Allocation Models
Based on a review of Yale’s portfolio, we see how a large endowment allocates its portfolio. We have also seen that a complex approach may not be appropriate to determine our allocations. So next, we review three simplified approaches: Yale’s David Swensen, author and financial theorist William Bernstein and a traditional 60%/40% equity/bond mix.
David Swensen’s Portfolio
Swensen’s book, “Unconventional Success” (Free Press, 2005), argues for a simple portfolio consisting of the following: 30% domestic stocks, 20% foreign stocks, 20% real estate, 15% inflation-protected bonds (e.g., Treasury Inflation-Protected Securities, or TIPS), and 15% in conventional bonds.
William Bernstein’s Portfolio
In “The Intelligent Asset Allocator” (McGraw-Hill Education, 2000), Bernstein argues for a portfolio consisting of: 25% domestic stocks, 25% foreign stocks, 25% small-cap stocks, and 25% bonds.
The 60/40 Portfolio
Back in the 1960s and 1970s, a lot of research was conducted to try to assess how pension funds allocated. It was observed that allocations seemed to coalesce around a simple strategy: 60% to domestic equity and 40% to bonds.
Does Asset Allocation Even Matter?
It turns out that the details of a chosen approach to asset allocation are largely irrelevant in the long run. The differences in returns are often noise.
Some academic research on asset allocation concludes that investors should hold the market portfolio or the portfolio of all risky assets weighted by their respective value in the market portfolio. Other research, such as Markowitz’s approach discussed above, clings to the tenets of mean-variance or other efforts to “optimize,” to reduce estimation error and identify superior portfolio rules. Yet as the data in Table 1 shows, the specific approach may not matter that much, and perhaps a simpler approach—such as equal weighting—can do just as well.
Victor DeMiguel, Lorenzo Garlappi and Raman Uppal have a surprising answer to this question in their paper “Optimal Versus Naive Diversification: How Inefficient Is the 1/N Portfolio Strategy?” (The Review of Financial Studies, 2009). The authors examined the traditional mean-variance asset allocation approach, but also included 12 other complex and sophisticated approaches related to Markowitz’s original approach to asset allocation. The authors compared these 13 different optimization models against good old-fashioned equal-weighted (“1/N”) portfolios.
The authors didn’t stop there. They analyzed the different asset allocation techniques on eight different datasets to ensure their findings were sound. They examined a variety of sector portfolios, country portfolios and factor portfolios and even generated simulated data for their horse race across the different asset allocation approaches.
What they found was remarkable. Equal-weight portfolios met or beat all of the fancy portfolios. In their own words:
“We evaluated the 14 models across seven empirical datasets. Our results found that none were consistently better than the 1/N rule in terms of Sharpe ratio (risk-adjusted return), certainty-equivalent return (the amount return required to be indifferent between an uncertain return and a certain return), or turnover. This indicates that the gain from optimal diversification is more than offset by estimation error.”
The DeMiguel, Garlappi and Uppal paper perplexed the academic establishment and drove home one of the key takeaways of successful asset allocation: Complexity does not necessarily add value. Simply stated, it is hard, and perhaps impossible, to do much better than 1/N in a competitive world with so much volatility and uncertainty.
[Victor DeMiguel, Yuliya Plyakha, Raman Uppal, and Grigory Vilkov have been working on finding methods to improve the 1/N strategy. Their recent paper, “Improving Portfolio Selection Using Option-Implied Volatility and Skewness,” (Journal of Financial and Quantitative Analysis, December 2013), suggests that it might be possible to beat the equal-weight construct, but the analytical and data capabilities required are high.]
A Good Starting Point
The research suggests that an equal-weight allocation approach will work just fine. But how does one decide which asset classes to include?
A good starting point is Mebane Faber and Eric Richardson’s Ivy Portfolio. In “The Ivy Portfolio: How to Invest Like the Top Endowments and Avoid Bear Markets” (John Wiley & Sons, 2011), they show that the world of endowment returns can boil down to five core asset classes, all of which are equal-weighted at 20%:
- Domestic equity: S&P 500 Total Return Index
- International equity: MSCI EAFE Total Return Index
- Real estate: FTSE NAREIT All Equity REITs Total Return Index
- Commodities: Goldman Sachs Commodity Index
- Bonds: Merrill Lynch 7–10 year Government Bond Index
Investors seeking to add complexity and additional asset classes need to be judicious and consider the FACTS along the way.
We’ll leave the last word to David Swensen, Yale’s successful CIO: Stick to a simple diversified portfolio, keep your costs down and rebalance periodically to keep your asset allocations in line with your long-term goals.
Investors wanting more information on the simple allocation approach and how it compares to more complex strategies can read Alpha Architect’s white paper, “The Robust Asset Allocation blog.alphaarchitect.com/2014/12/02/the-robust-asset-allocation-raa-solution/. In this research paper, investors will also find a link to a related paper discussing Alpha Architect’s FACTS approach to portfolio management in more depth.Solution” at
This post was adapted from Wesley Gray’s article in the November 2015 issue of the AAII Journal. Be sure to visit AAII.com for complete article, including analysis of the performance of three asset allocation strategies from 1979 through 2014.