At the heart of investing theory is the idea that investors can generate more return with less risk by combining assets in the portfolio that are not well correlated with one another. Weakly correlated assets tend to damp out total portfolio risk to some extent without decreasing return—and this is the value of diversification.
In my own analysis using Quantext Portfolio Planner [QPP] and in a review of a wide range of institutional research, I have previously suggested that really effective diversification is worth as much as 2.5% per year in annual return for an investor with a portfolio with the same risk level as a generic mix of 60% domestic stocks and 40% bonds.
As a rule-of-thumb, the well-diversified portfolio at this risk level is projected to have an annual standard deviation in return (a standard measure of risk) of about 10% and expected return of about 10%--a 1-to-1 ratio between risk and return. To build such a well-diversified portfolio, the investor or advisor needs to identify a set of core asset classes from which to determine the portfolio’s asset allocation.
What is an Asset Class?
What are the “building blocks”—the core asset classes-- of a well-diversified investment portfolio? Most investors will list stocks, bonds and cash. Most investors will break stocks out into domestic and foreign stocks, and emerging markets and developed markets are typically treated as distinct. Some investors think about stock funds in terms of how large the companies are, and many consider large company and small company stocks as distinct asset classes.
The Intelligent Portfolio, by Christopher Jones, provides a succinct definition: an asset class is “a category of investments such as stocks, bonds, or cash that share similar risk and return characteristics.” In the last ten years, the list of major asset classes that is discussed often includes commodities, REIT’s, and (among institutional investors) private equity.
I add an additional component to the definition of an asset class. An asset class is a category of investments that are similar to one another but also that behave distinctively relative to other asset classes.
The “Starter Set” of Core Asset Classes
For our purposes, let’s say that we want to start by defining a minimal set of asset classes that can yield a portfolio that is well diversified. We want to build a list of possible investments that does not include unnecessary redundancy. What does that list look like? Even a question that sounds as basic as this one will illicit a different answer from different experts. To get started, I propose the following “starter list” of asset classes, along with index funds (Exchange Traded Funds) that track them:
Why choose this set of basic asset classes? Why don’t we have “value” and “growth” breakdowns? Are utilities (NYSEARCA:IDU) and natural resources (NYSEARCA:IGE) distinct asset classes that are worthy of consideration? TIPS (NYSEARCA:TIP) are Treasury Inflation Protected Securities—government bonds that increase in value with CPI (a measure of inflation). TIPS are, in a nutshell, bonds with some embedded inflation protection. Are TIPS a distinct asset class from bonds? I treat utilities, energy, TIPS, commodities, and REIT’s as distinct asset classes because they behave differently from one another and differently from broad equity classes. In plain terms, they do not track the major equity or bond indices. As such, these “asset classes” are valuable in diversifying a portfolio.
There are a wide variety of ways to break out the important asset classes. For the purposes of this discussion, what we care about is how many distinct “types” of investments we need to consider to build a portfolio which captures most of the available diversification benefits. There are certainly more nuanced breakdowns that we might look at, but I believe that the case can be made that those listed above are sufficient for generating a portfolio that will out-perform the vast majority of mutual funds. For examples of the weights that end up assigned to each of these asset classes using our portfolio theory approach, see the linked article.
Building a list of “core” asset classes for a portfolio starts with asset classes that do not track well with one another—i.e. that exhibit low correlation.
The Perils of Diversifying by Names
It is common for investors to believe that they are diversified because they buy assets that sound as though they should act differently. Many mutual funds (and even broad asset classes) that “sound like” different investments actually track one another very closely—which means that having all of these different funds in a portfolio does not actually add value. This point is easy to demonstrate. The chart below compares investments in three Vanguard funds—large cap, mid cap, and small cap stocks funds—for two years. These three funds, despite representing different sizes of companies, have tracked one another remarkable closely for the last two years:
Vanguard Large Cap Stock Fund [VLACX] vs. Mid Cap Stock Fund [VIMSX] vs. Small Cap Stock Fund [NAESX] for two years through July 2008
Over the past two years, investors who thought they were diversified because they spread assets across these three funds have not generated much diversification benefit—these three funds have tracked one another very closely. Diversification is especially important as a way to manage risk—to protect the downside—and it should be obvious that these three funds of companies sorted by size have all gone down together. Given these high correlations to one another, why do we include small cap stocks and large cap stocks as distinct entities in our “Starter Set” of asset classes? First, small cap stocks provide higher expected returns than large cap stocks (on average). Second, small cap stocks and large cap stocks provide different diversification benefits with respect to other asset classes.
How about foreign stocks? The chart below compares an investment in the Vanguard Large Cap fund [VCACX] with an investment in the Vanguard Developed Markets fund [VDMIX]. Even these two have tracked remarkably closely over the past two years.
Vanguard Large Cap Stock Fund [VCACX] vs. Developed Markets Index [VDMIX] for two years through July 2008
Spreading money between this international fund and this domestic stocks fund provides very little protection during the recent market declines. By no means am I saying that any of these funds are bad—these are all fine funds. The issue that I am raising is that many investors would think that they had reasonably well-diversified equity portfolios if they invested in four Vanguard funds representing domestic large cap, mid cap, and small cap funds, along with an international stock fund.—but such a portfolio was not well diversified at all.
A reasonable definition of effective diversification is that a diversified portfolio is somewhat insulated from moves in any single asset class. This is surely not the case in if you simply mix an EAFE index with an S&P500 index. Still, the EAFE index does provide some diversification benefit—and this is why we have included it in our core set.
By contrast to these fairly well-correlated “asset classes,” consider some of my alternative suggestions from the Starter Set over the same period:
The central idea in diversifying a portfolio is to combine assets in a portfolio that do not all go the same direction at the same time. The chart above provides a nice example of why we will be looking at utilities (IDU), real estate (ICF), and energy stocks (IGE) as distinct asset classes. The portfolio value of combining various asset classes can only be determined via quantitative analysis.
Too many investors have been trained that combining broad index funds for large-, mid-, and small-cap funds is a good practice but that adding sector-specific funds (such as the utility index) is not terribly desirable. Rather than going by name, it makes far more sense to apply tools that analyze the behavior of different portfolio components—and that is the core theme of portfolio theory.
These charts are merely presented as examples—correlation between asset classes over some period of time must be calculated between asset returns—and these charts are showing prices. This belief has a solid foundation in financial theory, but I cannot expand upon it here.
It is all well and good to write about the portfolio benefits of energy and utilities or the limited value of international developed stocks after the fact of this market decline. The reader may be somewhat more engaged if he/she takes the time to go back and see that I have been writing about the portfolio benefits of utilities and energy and that the portfolio benefits of emerging and developed foreign markets had been over-stated for quite some time—considerably before this bear market.
In this discussion, I am not holding out this particular set of ETFs as the ideal “core asset classes” but rather as a decent place to start. Based on the statistical properties of these ETFs, they each add some incremental value to the portfolio. In other words, each of these asset classes works well with the other parts of the portfolio to assist in generating more return for a given level of risk. The choice of the set of core asset classes should be determined by the properties of these asset classes—they should not all track together—and the bear market of 2008 has provided a compelling real-life example.
Defining the set of core asset classes is not a trivial under-taking. You need a statistical model (like QPP) that generates forward-looking projections of portfolio risk and return, using projections for the individual asset classes. When you are considering whether an asset class adds value, you add it to a model portfolio and project forward. If this asset increases return at a given risk level, it is worth considering. If this process is done using historical data only (i.e. looking backwards), you will end up with a portfolio that would have done well historically but that often tends to under-perform as you go forward.
This result is nicely demonstrated in an analysis by William Bernstein in The Intelligent Asset Allocator that shows that investors who build portfolios relying simply on trailing performance do very poorly. Forward-looking statistical models are the standard of practice for this type of analysis.
I believed that a lot of the reason that retail investors generate such poor results is that they do not have any solid idea of what a set of core asset classes is. For investors who have professional advisors, sitting down and engaging in a discussion of the choice of core asset classes in the portfolio will be time well spent.
There is certainly room for debate as to the optimal set of core asset classes. PIMCO’s Mohamed El-Erian proposes the following.
The reader will note similarities between these asset class choices and the core asset classes proposed here. One interesting apparent difference is that Mr. El-Erian has “infrastructure” broken out into its own class as a real asset. This relates to my use of utilities as a core asset class. Utilities represent a centrally important “infrastructure” investment in both developed and emerging economies.
Further, if you look at the expanded set of core asset classes that I discuss, you will see specific allocations to other “infrastructure” equity classes, such as transportation and telecom. In the All-ETF portfolio that I proposed in June of 2007, for example, there are specific allocations to IYT (transport) and IYZ (telecom).
I believe that an expanded set of core asset classes adds value, but the basic set proposed here is a reasonable place to start—particularly for the investor who is not yet conversant with a portfolio-centered approach to asset allocation. It is crucial to understand why you are investing in specific assets/asset classes.
I will attempt to summarize in brief. A crucial step for investor and advisors is to define the universe of core asset classes that will be used in a portfolio. There should be good reasons for specifying each choice. The “invest a bit in everything” approach is a poor approach to portfolio construction.
I have proposed a minimal “starter set” of asset classes and some ETFs that capture these classes. There is certainly room for debate in this area. Portfolio analysis using QPP (a forward looking portfolio model) has long suggested that utilities, for example, deserve to be treated as a distinct asset class.
By contrast, QPP has suggested that breaking out “value” and “growth” adds less value. QPP suggests that TIPS provide unique value that is not captured by generic bond indices. QPP has also favored commodities (NYSEARCA:DJP) and natural resources (IGE) as distinct asset classes. The analysis from QPP is bolstered by a range of perspectives in the importance in assets that will tend to keep pace with inflation from PIMCO, David Swensen, and Ibbotson [pdf file] (among others).
What is most important is not that everyone agrees with this set of core asset classes, however. What is important is that investors and advisors have an objective plan and reasoning as to the universe of asset classes that will be considered. The choice of asset classes to be considered in the portfolio will, ultimately, have a substantial impact on the performance of the portfolio.