Is Modern Portfolio Theory Dead? Come On.

Editor’s note: Paul Pfleiderer is the C.O.G. Miller Distinguished Professor of Finance at the Stanford Graduate School of Business and co-founder of Quantal International.

A few weeks ago, TechCrunch published a piece arguing software is better at investing than 99% of human investment advisors. That post, titled Thankfully, Software Is Eating The Personal Investing World, pointed out the advantages of engineering-driven software solutions versus emotionally driven human judgment. Perhaps not surprisingly, some commenters (including some financial advisors) seized the moment to call into question one of the foundations of software-based investing, Modern Portfolio Theory.

Given the doubts raised by a small but vocal chorus, it’s worth spending some time to ask if we need a new investing paradigm and if so, what it should be. Answering that question helps show why MPT still is the best investment methodology out there; it enables the automated, low-cost investment management offered by a new wave of Internet startups including Wealthfront (which I advise), Personal Capital, Future Advisor and SigFig.

The basic questions being raised about MPT run something like this:

  • Hasn’t recent experience – i.e., the financial crisis — shown that diversification doesn’t work?
  • Shouldn’t we primarily worry about “Black Swan” events and unforeseen risk?
  • Don’t these unknown unknowns mean we must develop a new approach to investing?

Let’s begin by briefly laying out the key insights of MPT.

MPT is based in part on the assumption that most investors don’t like risk and need to be compensated for bearing it. That compensation comes in the form of higher average returns. Historical data strongly supports this assumption. For example, from 1926 to 2011 the average (geometric) return on U.S. Treasury Bills was 3.6%. Over the same period the average return on large company stocks was 9.8%; that on small company stocks was 11.2% ( See 2012 Ibbotson Stocks, Bonds, Bills and Inflation (SBBI) Valuation Yearbook, Morningstar, Inc., page 23. ).  Stocks, of course, are much riskier than Treasuries, so we expect them to have higher average returns — and they do.

One of MPT’s key insights is that while investors need to be compensated to bear risk, not all risks are rewarded. The market does not reward risks that can be “diversified away” by holding a bundle of investments, instead of a single investment. By recognizing that not all risks are rewarded, MPT helped establish the idea that a diversified portfolio can help investors earn a higher return for the same amount of risk.

To understand which risks can be diversified away, and why, consider Zynga. Zynga hit $14.69 in March and has since dropped to less than $2 per share. Based on what’s happened over the past few months, the major risks associated with Zynga’s stock are things such as delays in new game development, the fickle taste of consumers and changes on Facebook that affect users’ engagement with Zynga’s games.

For company insiders, who have much of their wealth tied up in the company, Zynga is clearly a risky investment. Although those insiders are exposed to huge risks, they aren’t the investors who determine the “risk premium” for Zynga. (A stock’s risk premium is the extra return the stock is expected to earn that compensates for the stock’s risk.)

Rather, institutional funds and other large investors establish the risk premium by deciding what price they’re willing to pay to hold Zynga in their diversified portfolios. If a Zynga game is delayed, and Zynga’s stock price drops, that decline has a miniscule effect on a diversified shareholder’s portfolio returns. Because of this, the market does not price in that particular risk. Even the overall turbulence in many Internet stocks won’t be problematic for investors who are well diversified in their portfolios.

Modern Portfolio Theory focuses on constructing portfolios that avoid exposing the investor to those kinds of unrewarded risks. The main lesson is that investors should choose portfolios that lie on the Efficient Frontier, the mathematically defined curve that describes the relationship between risk and reward. To be on the frontier, a portfolio must provide the highest expected return (largest reward) among all portfolios having the same level of risk. The Internet startups construct well-diversified portfolios designed to be efficient with the right combination of risk and return for their clients.

Now let’s ask if anything in the past five years casts doubt on these basic tenets of Modern Portfolio Theory. The answer is clearly, “No.” First and foremost, nothing has changed the fact that there are many unrewarded risks, and that investors should avoid these risks. The major risks of Zynga stock remain diversifiable risks, and unless you’re willing to trade illegally on inside information about, say, upcoming changes to Facebook’s gaming policies, you should avoid holding a concentrated position in Zynga.

The efficient frontier is still the desirable place to be, and it makes no sense to follow a policy that puts you in a position well below that frontier.

Most of the people who say that “diversification failed” in the financial crisis have in mind not the diversification gains associated with avoiding concentrated investments in companies like Zynga, but the diversification gains that come from investing across many different asset classes, such as domestic stocks, foreign stocks, real estate and bonds. Those critics aren’t challenging the idea of diversification in general – probably because such an effort would be nonsensical.

True, diversification across asset classes didn’t shelter investors from 2008’s turmoil. In that year, the S&P 500 index fell 37%, the MSCI EAFE index (the index of developed markets outside North America) fell by 43%, the MSCI Emerging Market index fell by 53%, the Dow Jones Commodities Index fell by 35%, and the Lehman High Yield Bond Index fell by 26%. The historical record shows that in times of economic distress, asset class returns tend to move in the same direction and be more highly correlated. These increased correlations are no doubt due to the increased importance of macro factors driving corporate cash flows. The increased correlations limit, but do not eliminate, diversification’s value. It would be foolish to conclude from this that you should be undiversified. If a seat belt doesn’t provide perfect protection, it still makes sense to wear one. Statistics show it’s better to wear a seatbelt than to not wear one.  Similarly, statistics show diversification reduces risk, and that you are better off diversifying than not.

Timing the market

The obvious question to ask anyone who insists diversification across asset classes is not effective is: What is the alternative? Some say “Time the market.” Make sure you hold an asset class when it is earning good returns, but sell as soon as things are about to go south. Even better, take short positions when the outlook is negative. With a trustworthy crystal ball, this is a winning strategy. The potential gains are huge. If you had perfect foresight and could time the S&P 500 on a daily basis, you could have turned $1,000 on Jan. 1, 2000, into $120,975,000 on Dec. 31, 2009, just by going in and out of the market. If you could also short the market when appropriate, the gains would have been even more spectacular!

Sometimes, it seems someone may have a fairly reliable crystal ball. Consider John Paulson, who in 2007 and 2008 seemed so prescient in profiting from the subprime market’s collapse. It appears, however, that Mr. Paulson’s crystal ball became less reliable after his stunning success in 2007. His Advantage Plus fund experienced more than a 50% loss in 2011. Separating luck from skill is often difficult.

Some people try to come up with a way to time the market based on historical data. In fact a large number of strategies will work well “in the back test.” The question is whether any system is reliable enough to use for future investing.

There are at least three reasons to be cautious about substituting a timing system for diversification.

  • First, a timing system that does not work can impose significant transaction costs (including avoidable adverse tax consequences) on the investor for no gain.
  • Second, an ill-founded timing strategy generally exposes the investor to risk that is unrewarded.  In other words, it puts the investor below the frontier, which is not a good place to be.
  • Third, a timing system’s success may create the seeds of its own destruction. If too many investors blindly follow the strategy, prices will be driven to erase any putative gains that might have been there, turning the strategy into a losing proposition. Also, a timing strategy designed to “beat the market” must involve trading into “good” positions and away from “bad” ones. That means there must be a sucker (or several suckers) available to take on the other (losing) sides. (No doubt in most cases each party to the trade thinks the sucker is on the other side.)

Black Swans

What about those Black Swans? Doesn’t MPT ignore the possibility that we can be surprised by the unexpected? Isn’t it impossible to measure risk when there are unknown unknowns?

Most people recognize that financial markets are not like simple games of chance where risk can be quantified precisely. As we’ve seen (e.g., the “Black Monday” stock market crash of 1987 and the “flash crash” of 2010), the markets can produce extreme events that hardly anyone contemplated as a possibility. As opposed to poker, where we always draw from the same 52-card deck, in financial markets, asset returns are drawn from changing distributions as the world economy and financial relationships change.

Some Black Swan events turned out to have limited effects on investors over the long term. Although the market dropped precipitously in October 1987, it was close to fully recovered in June 1988. The flash crash was confined to a single day.
This is not to say that all “surprise” events are transitory. The Great Depression followed the stock market crash of 1929, and the effects of the financial crisis in 2007 and 2008 linger on five years later.

The question is, how should we respond to uncertainties and Black Swans? One sensible way is to be more diligent in quantifying the risks we can see. For example, since extreme events don’t happen often, we’re likely to be misled if we base our risk assessment on what has occurred over short time periods. We shouldn’t conclude that just because housing prices haven’t gone down over 20 years that a housing decline is not a meaningful risk. In the case of natural disasters like earthquakes, tsunamis, asteroid strikes and solar storms, the long run could be very long indeed. While we can’t capture all risks by looking far back in time, taking into account long-term data means we’re less likely to be surprised.

Some people suggest you should respond to the risk of unknown unknowns by investing very conservatively. This means allocating most of the portfolio to “safe assets” and significantly reducing exposure to risky assets, which are likely to be affected by Black Swan surprises. This response is consistent with MPT. If you worry about Black Swans, you are, for all intents and purposes, a very risk-averse investor. The MPT portfolio position for very risk-averse investors is a position on the efficient frontier that has little risk.

The cost of investing in a low-risk position is a lower expected return (recall that historically the average return on stocks was about three times that on U.S. Treasuries), but maybe you think that’s a price worth paying. Can everyone take extremely conservative positions to avoid Black Swan risk? This clearly won’t work, because some investors must hold risky assets. If all investors try to avoid Black Swan events, the prices of those risky assets will fall to a point where the forecasted returns become too large to ignore.

A third and arguably pathological response to the Black Swan problem is to say that nothing is safe. An extreme event could significantly reduce the value of any asset (“We may not have seen it, but this doesn’t mean that it couldn’t happen”). I doubt anyone has gone to this nihilistic extreme, and I mention it to make it clear that being aware of the potential for unknown unknowns is useful, but not at the cost of decision-making paralysis.

Of course, if you are that privileged investor with a reliable enough crystal ball, by all means use it. The problem lies in knowing whether it is reliable enough.

Although unknown unknowns and Black Swan events make evaluating investment risks more challenging, they don’t change the value of diversification and controlling the risks we do know about.

It’s particularly important that young people at the beginning of their investing careers understand why the sloppy arguments against MPT are so dangerous. With its insights about diversification and controlling risk, MPT provides the best foundation for developing low-cost portfolios like the ones being used by the Internet startups to “eat the personal investing world.”