We can finally fulfill MPT's risk-management promise.
In most aspects of our lives, risk is something to
be avoided. As has often been observed, we live in a highly risk-averse
society surrounded with an ever-increasing array of laws and
regulations designed to protect us against the hazards of daily life.
It is, in many ways, as if we seek to make risk itself disappear.
There is one important area of our lives, however, where risk is something to be managed, not avoided; where the exposure to risk is unavoidable and, in fact, desirable. When we invest money in a portfolio of financial assets, we are seeking to expose ourselves to risk in a way that helps us achieve our financial goals. This, as we all know, is due to the fact that the relationship between risk and reward (expected return) is one of the only inviolate relationships in the financial markets. If one assumes more risk, the potential for reward should be proportionately higher. Thus, in the world of investing, risk is most definitely not a four-letter word.
But what is risk? It is widely accepted that the definition of risk is the variability of returns over time. This definition carries with it the assumption that, for each investor, there is a level of risk that is both tolerable from a psychological standpoint and appropriate given one's investment horizon and wealth level, as well as a host of other considerations.
Most investment strategies seek to manage that key variable-risk. This is due to the fact that return is considered to be the dependent variable. That is, control the level of portfolio risk and an expected level of return will be realized based upon the historical relationship between the two. This linkage ultimately gave rise to the concept of risk-adjusted returns. The basic idea is that the absolute return of any given asset carries little informational value. What is really important is the amount of return per unit of risk. Risk-adjusted return attempts to answer the question, "Did the investor assume too much risk for the amount of return that was realized?"
In fact, an entire industry has grown up around the idea that this relationship can be quantified with increasing levels of precision. Measures such as Standard Deviation, Sharpe Ratios, Information Ratios, Treynor Ratios, Value at Risk (VAR), Down Market Capture and Tracking Error are all designed to identify the amount of risk in a portfolio of assets either in absolute terms, relative to a specific benchmark or in relation to actual returns.
A risk-based approach to portfolio management makes sense for a couple of reasons. The first is that an investor's risk tolerance evolves over time. As investors approach the end of their investment horizon, the ability to withstand substantial declines in portfolio valuations (risk) decreases as the need to access portfolio assets approaches.
This hard reality is compounded by the fact that risk is asymmetric; a 50% decline in the value of an asset requires a 100% gain to return to the initial value. Thus, from a long-term investor's standpoint, the ability to avoid substantial portfolio losses outweighs the rewards associated with outsized gains. In other words, risk needs to be managed.
How? By not putting all of one's eggs in a single basket. It is not an overstatement to say that the blending of financial assets with differing risk characteristics is the cornerstone of modern financial theory. The central tenet of Modern Portfolio Theory (MPT), developed by Harry Markowitz in 1952, was that given a pool of risky assets, overall risk can be reduced by combining assets that exhibit low correlations of returns. Thus, from a risk management standpoint, the correlation between various assets, not their actual volatility, is the critical factor.
According to MPT, diversification is the key. But is this enough? The answer from our standpoint is both yes and no. Yes, in the sense that combining financial assets with low correlations is the key to controlling portfolio risk. No, in the sense of how diversification has been implemented in practice.
The standard industry definition of a diversified portfolio is one that includes allocations to U.S. stocks, international stocks, bonds and money market instruments. Within those broad allocations, equity assets can be further divided by market capitalization (large, mid-sized and small) and by investment style (growth, blend and value). Additional allocations are often made to international equities (developed and emerging markets), and fixed-income exposure is often diversified across maturity and credit quality.
How effective is this approach? Clearly, using fixed income and equities to diversify portfolio risk is an effective technique. Looking at historical correlations between the Standard and Poor's 500 Index and the Citigroup/Salomon Bros. U.S. Broad Investment Grade Bond Index, correlations over the last 25 years have been low and, at times, negative.
There are however, periods when correlations between stocks and bonds "spike" and the diversification benefit is reduced. That was clearly the case in 1985, when correlations reached +0.86, and again in 1995 when they reached +0.99. Moreover, bonds are not immune from correlation spikes during periods of market turmoil, as we saw in 1998 during the Long Term Capital Management crisis, when stock/bond correlations rose to +0.44.
The diversification benefit from adding an allocation to international stocks, however, is far from clear. In looking at correlations between the S&P 500 and the MSCI EAFE Index, we see that correlations have steadily increased over time. This makes intuitive sense, as the integration among the world's economies and capital markets has marched inexorably forward. We see in Table 1, that correlations between domestic and international stocks jumped to over +0.90 in 1980 and have remained high ever since.
But as we discussed earlier, risk management is all about mitigating the large losses that can be so destructive to an investment objective when markets go through their periodic crises. In this, international stocks appear to be ineffective. During the two-month period encompassing the 1987 stock market crash, daily correlations were +0.86.
If adding international stocks to a portfolio isn't an effective risk management technique, what about mixing up market capitalization and investment style? Many of us have been taught to think that if we have allocations to large-, mid- and small-capitalization stocks as well as allocations to growth, blend and value, then we have covered our bases. But is this true?
Unfortunately, the answer appears to be no. Looking at five-year snapshots going back to 1980, both of these techniques offer questionable diversification benefits. In looking at monthly correlations between the large-cap Russell 1000 Index and the small-cap Russell 2000 Index correlations were +0.90 or higher in five out of six observations, the exception being 2000 when the correlation between the two indexes dipped to +0.55.
The limits of style-based diversification are a little less clear-cut, ranging from +0.99 in 1995 to -0.39 in 2000. However, from a risk management perspective, style-based diversification suffers from the same correlation spikes seen in other asset classes. For example, during the bursting of the "tech bubble" in March of 2000, the correlation between Growth and Value was +0.92.
A related shortcoming associated with traditional risk reduction approaches is their reliance on long-only investment strategies. The long-only constraint makes effective risk management a challenging proposition.
The common perception of risk is that it is a relatively static target. Under the current approach to MPT, all one needs to do is 1) determine the investor's risk tolerance, 2) establish the appropriate stock/bond cash allocation, 3) diversify by market cap, style, maturity and credit quality and, 4) periodically rebalance. Layered on top of all this is often a tactical asset allocation overlay, which typically results in relatively modest changes to the portfolio's asset mix.
But, as Table 3 points out, portfolio risk is anything but static. The traditional portfolio construction approach results in dramatic swings in risk exposure over time. The classic 60/40 stock-bond portfolio-continuously rebalanced-had a standard deviation as low as 8.75% in the 1970s and as high as 14.5% in the 1980s. Thus, the traditional approach resulted in a near-doubling of portfolio risk from one decade to the next. This is far outside the definition of acceptable risk tolerance under most investment programs and raises serious potential suitability issues for the financial intermediary.
So in a real sense, traditional risk management techniques result in precious little risk management. It is as if the investor is being managed by risk, not managing it.
If traditional asset allocation is not an effective risk management strategy, are there other options? The good news is that there are. We are at a point in the evolution of the financial industry where what were once considered nontraditional investment strategies are now widely available and highly liquid-strategies that have been used and proven in the institutional marketplace for many years.
The key, as always, is correlation. If one is to get "true diversification" in a portfolio of financial assets, then it is imperative to include asset classes that do not respond to, or at least respond differently, to the same set of inputs. The essential shortcoming of traditional stock-bond asset allocation is that, under many circumstances, these assets respond to changes in interest rates, inflation, economic growth and changing risk premiums in the same manner.
However, the recent explosion of investable investment strategies and asset classes now makes it possible to achieve the kind of risk management originally promised by MPT. Once esoteric investing strategies, such as managed currency and commodity futures, real estate, short selling, arbitrage and event-driven strategies, allow portfolio risk management to be taken to the next level.
Lets take a look at how this might work in practice. If an investor held a portfolio of 50% S&P 500 stocks, 30% long Treasury bonds, 15% international stocks (MSCI EAFE) and 5% cash for the ten-year period of 1994-2004, that portfolio would have earned an annualized return of 9.85% and had a standard deviation of 9.26%, as shown in Table 4.
Now let's add some of the asset classes and investment strategies discussed above. By slightly paring back allocations to traditional asset classes and adding small allocations to real estate, commodities and hedge fund strategies, the characteristics of the portfolio are vastly improved. What we see in Table 5 is that portfolio risk has been substantially reduced-annualized standard deviation falls by nearly 14%-with a slight increase in annual returns.
A couple of important benefits arise from this exercise. The first is the clear reduction in portfolio risk by the introduction of investment strategies and asset classes that exhibit "true" diversification. This is now a more efficient portfolio with improved risk/reward characteristics.
Perhaps more importantly, this is now a more flexible portfolio. If the investor was comfortable with the original risk profile of the portfolio, she is now free to add leverage-in a measured way-to enhance expected return without increasing risk. Moreover, by moving away from the constraints imposed by long-only mandates through exposure to hedge fund strategies, the investor has introduced an absolute return component to the portfolio, which historically has had a positive impact on the risk profile of traditionally structured portfolios.
Thus, by broadening her investment horizon, the investor has taken advantage of the only free lunch in finance-diversification.
David Reilly is director of portfolio
strategies at Rydex Investments. He often is quoted by the media and
has published numerous articles on financial topics.