This is the first of several articles about diversification and risk reduction. The basic point of Part 1 is that investors got what they implied they wanted when they diversified equities globally: they obtained exposure to the global equity market factor.
Rather than blaming the "failure of diversification" for terrible performance, investors should properly allocate blame to their overly bullish equity forecasts, and consequently to their overly high allocations to equities.
Has diversification stopped working? Have the benefits of diversification disappeared? We have all read and heard concerns of this kind expressed time and again over the past six to twelve months, always with reference to the fact that since the end of October 2007, global equity markets (along with other asset classes such as commodities) have plummeted in lock step. The complaint seems to be that somehow global (equity) diversification was supposed to “lower risk” and protect investors from large downdrafts --- so clearly it has failed.
There are several intertwined issues here, including a very basic one: what is diversification, and what is it supposed to achieve? We suggest that the previous complaint is at least partly based on a fundamental misunderstanding of the point of diversification in general, and of global equity diversification in particular. This is not just a pedantic concern about defining terms or concepts. If investors misidentify important characteristics of the current investment environment, and misdiagnose problems in their portfolios and investment strategies, then corrective action going forward is less likely to be effective. We explore the concept of diversification in Part 1.
However, there is another legitimate question that may be underlying recent concerns about risk: have the risk properties of global investing somehow changed in the past year? In particular, have markets exhibited a level of risk that is unprecedented, and so could not have been expected by investors? We will examine some aspects of global risk and the potential for risk reduction using global asset classes in follow-up papers (Parts 2 and 3).
In its most basic sense, “diversifying” or the “process of diversification” means to vary, to give variety, or to spread out. In an investment context we vary or spread out our investments by holding multiple securities, and the reason for doing so is to reduce risk. Introductory finance textbooks generally have a section on this topic, and they show how, within a market such as the Canadian equity market, adding securities reduces overall portfolio risk. This happens dramatically at first, as you add a second, a third, and a fourth security to your portfolio, but as you reach perhaps 20 securities, the “marginal reduction in risk” gets to be quite small. Of course the ultimately “diversified” portfolio within a market holds all the securities in the market.
There are two related themes in this process of diversification. The first is the point that adding securities lowers risk due to the “imperfect correlation”(1) between individual stocks in a market. This is the risk reducing aspect of diversification. But the second point is that you can’t “diversify away” the market itself by adding more and more diversifying holdings --- in fact the more holdings you add, the closer you get to the market. In the jargon of finance theory, we say that you can diversify away non-systematic or non-market risk, but you cannot diversify away systematic or market risk.
So here is a good definition of diversification as found on the Bloomberg Financial Website in their Financial Glossary:
Diversification: Dividing investment funds among a variety of securities with different risk, reward, and correlation statistics so as to minimize unsystematic risk.(2)
Now, when we take the step of diversifying our equity portfolios globally, the same process is in play. The risk we are able to reduce is the “unsystematic” risk relative to the global market. By adding global securities, or (in our case) S&P500 and EAFE ETF’s, and perhaps Emerging Markets as well, we are hoping to lower portfolio risks by adding imperfectly correlated securities. Using the finance jargon, we are also lowering our non-systematic risk relative to the global equity market, and we are getting closer and closer to holding the global market itself --- the systematic global equity risk that we cannot diversify away.
When we lower our non-systematic risk, we are lowering the risk that our portfolio will behave very differently from the market itself. So when a globally diversified portfolio experiences a global selloff, such as the 2008 plummet generated by the real estate crash and broader financial system blow up, it just goes down! We had (and are still experiencing in March 2009) a global market phenomenon. Global equity diversification cannot protect an investor from a global market event --- by definition.
And of course when the global phenomenon is an upward swing, no one complains! For example, global equity markets were running up in lock step, from March 2003 to October 2007. In Figure 1 we show three investments that a Canadian investor could have held: the large cap TSX60 ETF(XIU), as well as the S&P500 (XSP) and EAFE (XIN) ETF’s both with foreign currencies hedged into Canadian dollars.(3) We use the currency hedged ETF’s here because they give investment returns that closely approximate the returns of “local” investors in global markets, removing most of the impact of Canadian dollar movements.(4) While the Canadian market is a tiny portion of the global “developed” equity market (probably less than 3%), together the S&P500 and EAFE compose more than 90% of the developed equity market. So a Canadian investor holding a large allocation of XIN and XSP, along with his Canadian stocks, has gone a long way towards diversifying globally.
There are three circles in figure 1. The left circle indicates the period from March 2003 to May 2005: the three ETF’s generated almost identical returns, even over shorter periods. The middle circle indicates the period from June 2005 to October 2007, with the three ETF’s still moving largely together, although the S&P 500 trails the overall return of the other two ETF’s, while still moving upward in close synchronization. The right hand circle shows the market selloff. All three ETF’s move downward in dramatic fashion, although the TSX 60 ETF bucked the overall trend until May 2008, as commodities continued to rally.
From a purely pictorial view, the 2008 downtown (right hand circle) had more divergence in return than the first leg of the bull market (left hand circle), and at least as much divergence as the middle circle, the second half of the bull market. In other words, global equity markets moving together is not a new story, and a globally diversified portfolio would have achieved its goal, both on the upside and on the downside, of capturing systematic global returns. Of course we are less likely to be concerned about the issue of synchronized investment returns when they are positive over a long stretch of time!
In sum, global diversification has worked --- investors got what they implied they wanted when they sought global diversification: strong exposure to the global market "come what may". They got it on the upside, and they got it on the downside!
So What are Investors Really Complaining About?
So what might be underlying the complaints that global diversification has failed? Since we are contending that investors have focused on the idea of diversification as "risk reducing", without distinquishing systematic from non-systematic risk, a legitimate complaint might be that the realized risks in global markets have been much higher than could have been expected by any reasonable investor. We will examine this possibility in Part 2, by looking at a number of direct measures of risk relative to history. In Part 3 we will extend this further by looking at recent correlations between global markekts. Here the point is that since imperfect correlations reduce (non-systematic) risk, if correlations are extremely high then the overall level of risk might be higher than expected.
If either case turns out to be true --- equity markets have undergone unprecedented high levels of risk as measured directly, or unprecedented high levels of correlation --- then while it might be technically wrong to say that diversification has failed, it might be legitimate to point out that we have been "blind-sided" by unpredictable risks. However, we find that with one exception ( risk as measured by short-term daily price volatility), on balance realized risks have not been higher than historical precedent would allow for.
There also may be a second "psychological" aspect of this complaint about the failure of diversification. We have observed that over the past 20 years, the concept of “diversification” has been used to justify forays into return enhancement, without necessarily justifying the “return” portion of the equation. If this observation is correct, then it is understandable that for those who followed this approach, disappointment in global equity returns implies a failure of diversification
What we mean is this. Diversification was the mantra that Canadian investors used to justify higher levels of investments in the US market in the late 1980’s, because that market had been outperforming Canada. Then in the 1990’s the push was to “diversify” into EAFE markets, which had been outperforming North America. Then in the early 2000’s the push for “diversification” was into emerging markets and commodities --- which in turn had been outperforming developed markets. Our point is that the cloak of diversification --- who could argue against reducing risk? --- was an easier justification for chasing return than actually sticking your neck out and forecasting returns in those markets to justify the investment.
As with any investment, these “diversifying” investments should have been made on the basis of clear return expectations combined with an analysis of their contribution to the risk of the investor’s portfolio. We are not sure what level of expected returns were being forecast by investors in September of 2007, and we are certainly not suggesting that investors should have forecasted a minus 40% return. But one would think that even a reduction in equity expectations, founded on low current dividend yields at that time and moderate expectations for ongoing economic growth, might have led to reductions in equity positions.
To some people these points will seem like "mere semantics". But in order to improve your chances of obtaining reasonable investment success going forward, don’t you have to make the correct analysis of what went wrong in the past? What you do in the future might be quite different if you think, on the one hand, that “diversification failed” (so it's not really your fault), or, on the other hand, that your return expectations were unrealistic (and you had better improve your approach to generating return expectations going forward).(5)
(1) If you search for a definition on the internet, or check with a finance or statistical textbook, you will find that the “correlation coefficient” is a precisely defined mathematical term. In colloquial English we can say several very general things such as: most equity securities within a market such as Canada tend to be positively (but not perfectly) correlated because they usually “move together” or “move in the same direction”. If two securities were perfectly positively correlated (with a correlation coefficient of 1.0), they would always have proportionately the same investment return over a given period --- and there would be no diversifying or risk-reducing benefit by holding them together. If two securities were uncorrelated (with a correlation coefficient of 0), they would move independently of each other. Holding uncorrelated securities generates very significant risk-reducing benefits. Pairs of securities can also be negatively correlated, meaning that they tend to move in opposite directions. Negative correlations generate powerful risk reduction, but in practice it is rarely found within an asset class such as equities.
In reality, most securities within a market have positive (but not perfect) correlation with each other – they all tend to generally move in the same direction. This means that we expect that almost any pair of securities within a market would have a correlation coefficient between 0 and 1.0. (Return to text)
(2) http://www.bloomberg.com/invest/glossary/bfglosd.htm (Return to text)
(3) For illustrative purposes we prefer to use actual investment vehicles that investors actually could have held, rather than more abstract indexes which for most investors are harder to understand, follow, and verify. (Return to text)
(4) Chart Essays #1 and #2 on this website show the impact of hedging or not hedging foreign currency exposure over this time frame. (Return to text)
(5) In a future publication on this website we will present our approach to long-term forecasting for equities, fixed income, and commodities. (Return to text)
Structured Capital presents a book by Tim Appelt:
Historical analysis of long-term global equity and bond returns is used to develop an analytical framework for a historical attribution of returns. In turn this attribution approach is used to develop expectations of future returns that acknowledge the past but take into account current market conditions.