What Makes Something An (Alternative) Asset Class, Anyway?

Posted by Michael Kitces on Wednesday, September 19th, 11:06 am, 2012 in Investments

The most typical definition of an asset class is a group of securities that have similar risk/return characteristics, and behave similarly in the marketplace. Thus, for instance, stocks, bonds, and cash represent the three most common asset classes, as each have different risk/return characteristics and they behave very differently in response to various economic and market events.

One of the most common ways to attempt to determine whether an investment represents a unique asset class is to examine its correlation with other investments. After all, two investments that have different risk/return characteristics and behave differently in response to market events would likely show little similarity in returns over time, thereby exhibiting a low correlation. In turn, given how Modern Portfolio Theory demonstrates that investments with a low correlation to the rest of the portfolio can lower the overall volatility of the portfolio - even if the underlying investment itself is volatile on its own - advisors have increasingly sought out low correlation "alternative" asset classes and investments to manage risk through diversification.

The caveat, however, is that in an increasingly complex financial world, having a low correlation alone can actually be a remarkably poor and misleading indicator of what constitutes an (alternative) asset class.

(Editor's Note: A version of this article originally appeared in the September 2012 issue of the Journal of Financial Planning.) 

Instability of Correlations

One of the primary problems with using correlation as a measure to determine what may be an alternative or different asset class is the fact that correlations themselves are often unstable. For instance, as many advisors and their clients noticed in late 2008, many asset classes that were previously low correlation actually had a very high correlation during the market decline. The end result, at least in retrospect, was a number of asset classes were actually low correlation to stocks during the bull markets but high correlation during bear markets - which means they were a drag on returns when markets were up, yet provided little diversification value when needed. The reason? Because in the end, the asset classes were still subject to the same underlying risk factors; and when those factors, including an economic recession and a contraction of credit, actually did occur, many previously-thought-to-be-separate asset classes moved in sync. Which means the bottom line is that a low correlation only matters if it's expected to be low when it's needed, not just on average. 

Notably, it's also important that the correlation be genuinely low in the first place; in a world where correlations have crept higher across many asset classes over the past several decades, so too has the threshold for what a "low" correlation really is to begin with. While once correlations had to be 0.2 - 0.4 to be considered "relatively low", many advisors now view 0.5 - 0.7 as being fairly low... despite the fact that in the end, such correlations wouldn't have been truly expected to produce much diversification even with traditional portfolio design. 

Low Correlations through Active Management

Another confounding factor to using correlation as a means of determining whether something is an asset class is the fact that some investments are actively managed - which leads to the possibility that low correlations may simply be a result of the varying decisions of the investment manager, not a difference in the underlying characteristics of the investment.

Historically, this hasn't necessarily been an issue; "actively managed equities" has never been viewed as a different asset class than equities, but simply represents the opportunity for the manager to add (or subtract) alpha, on top of the underlying returns of the asset class itself.

However, in today's environment with an increasing number of "go-anywhere" investment managers, the situation is less clear. An investment manager who was in cash in 2008 and equities in 2009 would show a very favorable return, but would exhibit a relatively low correlation to cash (because half the time was in equities) and a relatively low correlation to equities (because half the time was in cash). 

If the fund was simply a balanced fund that maintained a 50/50 exposure to stocks and bonds, it would be relatively straightforward to see that the fund represents not a new asset class, but simply a 50/50 exposure of two underlying asset classes. However, the reality is that when the manager chooses how much or whether to invest in each underlying asset class, and shifts over time, the correlations may decline, the asset class of the fund itself hasn't changed, just the allocations have.

As a result, it may arguably be more effective to evaluate such funds relative to the asset class of their benchmark - what they intended to mimic or track before adding or subtracting alpha - rather than what the manager (temporarily) holds. After all, if a fund's goal is to beat the equities asset class, which it does by holding equities except when they are risky and overvalued and cash represents a better investment opportunity, does the asset class of the fund change the day it shifts to cash, or just its short-term allocation?

Trying To Determine An (Alternative) Asset Class

So how should a planner determine whether an investment constitutes a unique asset class or not, especially given the rising popularity of "alternative" asset classes? 

The key is a return to the original definition of an asset class - a group of investments that shares common risk/return characteristics and behaves similarly in response to economic and market events. If an investment meets these criteria, then it should also have a low correlation to other asset classes. However, as shown earlier, a low correlation alone is not a sufficient criterion for determining an asset class, as there are many factors that can result in a low correlation that have little to do with what defines an asset class (and/or may not represent a distinct asset class at the time it's needed!).

When we look to the underlying investment characteristics, some asset classes meet the test far more effectively than others. For instance, real estate and commodities have unique characteristics, are driven by economic and market fundamentals that are different than other traditional asset classes, and while they happened to move in sync during the 2008 financial crisis, they clearly would not have a high correlation in all economic environments. For instance, while stocks, real estate, and commodities may move in lockstep during a deflationary environment (while government bonds go the other way), a rising inflationary environment can send commodities soaring while stocks, bonds, and real estate take a hit (at least, until their prices/rents can adjust and generate higher revenues). In turn, real estate may also be sensitive to interest rates, which affects the capitalization rate that in turn impacts real estate prices, in a manner that is similar to bonds but very different than how interest rates impact commodities or equities.

On the other hand, in some situations the lines are less clear. Large capitalization stocks are often viewed as a separate asset class from smaller capitalization stocks, even though both will typically rise in a growth environment and decline in a recession (at least, unless there are significant differences in valuation). Similarly, there is much debate about whether sectors of the economy - such as utilities versus technology stocks - constitute separate asset classes or not. While they are both equities, the reality is that their correlations may be lower and their returns may actually diverge more through an economic cycle than large versus small cap stocks, as technology stocks often rise and fall with the economy, while utilities typically move in a "counter-cyclical" fashion.

An especially problematic scenario for evaluating asset classes is the emergence of many "hedge fund strategies" funds and managed futures investment offerings, that often mix together multiple asset classes to begin with, along with shifting allocations over time, to generate low correlations to traditional asset classes. Yet if actively trading equities is not a separate asset class than equities in general, does that mean actively trading commodities is not a separate asset class from commodities, in the case of many managed futures funds? If the fund shifts between commodities, stock, and interest rate futures, does that constitute a new asset class, or simply a combination of the underlying, existing asset classes? If the goal of a merger arbitrage fund is to generate returns in excess of cash while neutralizing market risk, even though the underlying investments are equities, does that mean it's a stock asset class, a cash asset class, or an entirely separate asset class? What about other active hedge fund strategies?

What do you think? Do you consider large cap stocks a different asset class than small cap stocks? What about utilities versus technology stocks? Are managed futures and hedge fund strategies a new alternative asset class, or just the active trading of the underlying asset classes? What makes something an asset class? Is it just a low correlation, or does it require something more?

(This article was featured in the Carnival of Investing #5 on Investeem and the Nerdy Finance Carnival #9 on Nerdwallet.)


  • Avi Oren

    Mike, while reading the above on Alternative Investments, I am pondering if MultiSector Bond funds should be included there as separate asset classes. Take a look at this section of a post, indicating an approach which, imo, is presently popular amongst pre-retirees:

    "My pre-retirement portfolio is designed to capture the income yield, without touching principle, of selected bond funds. I try to select funds which pay higher than average yield (5% or more) in that effort. I only use the number of funds necessary to produce the amount of money needed to cover established ongoing expenses. Funds, not needed to produce that amount of income, have all income and
    capital gains reinvested for future use/needs."

    Bond funds mentioned are: PONPX; DBLTX; OSTIX; PAUIX; FSICX; LSIIX; IVHIX;etc.

    Any thoughts?

    • Michael Kitces

      Avi,
      I don't see this as any kind of asset class distinction. The investors are still buying bonds. The fact that they buy "some" bonds according to certain criteria, rather than "all" bonds, doesn't make the investment less of a bond investment.

      In a similar manner, buying higher dividend, or higher growth, or higher earning stocks isn't a different asset class. It's just a subset of the "stocks" asset class. For that matter, so is buying stocks that start with the letter "M" as an arbitrary example.

      Granted, some criteria for stock (or bond) selection may be more appealing from an investment perspective, but buying bonds with higher coupons doesn't define a separate asset class from bonds, just as buying companies with good earnings or companies that start with the letter "M" do not define a separate asset class from stocks.

      Simply put, there's a difference between investment selection WITHIN an asset class, and selection OF an asset class.
      - Michael

  • http://www.investeem.com Investeem

    Michael, thanks for contributing this post to the Carnival of Investing. We have included in the exclusive Editor's Picks section.

  • http://www.linkedin.com/in/bernardmack Bernie Mack

    Thanks for the discussion Michael. This is exactly an area that I have tried to get more definition on and also understand how it should be represented in portfolios and explained to clients. First should there be a distinction between alternative asset classes (real estate, commodities, etc.)and alternative investment strategies (absolute return, long short, etc.)

    Next, are we looking for low correlation asset class or non-correlated asset classes. Correlations as you pointed out shift over time. As an asset class becomes more liquid and easily tradable(GLD, DBC), it can also take on similar short term behavioral based trading characteristics as other asset classes. So non-correlation can be determined by economic and market fundamentals that are different than other traditional asset classes but also recognizes there will be times when correlations move to 100% due to behavioral forces. So do you discount the value of modern portfolio theory? How do we represent a "go anywhere" portfolio manager or an alternative strategy in a model portfolio allocation?

Blog Updates by Email

Nerd’s Eye View Praise

@MichaelKitces Twitter


Out and About

Thursday, November 6th, 2014

Cutting Edge Tax Planning Developments & Opportunities Future of Financial Planning in the Digital Age @ ACP 2014 Conference

Friday, November 7th, 2014

Setting a Proper Asset Allocation Glidepath in Retirement @ Charles Schwab IMPACT 2014

Monday, November 10th, 2014

Expanding the Framework of Safe Withdrawal Rates Future of Financial Planning in the Digital Age @ PICPA 2014 PFP Conference

VIEW FULL SCHEDULE