Rethinking Modern Portfolio Theory

The world of institutional Chief Investment Officers and Risk Managers was a very orderly place until the credit crisis of 2008. As the correlation between asset classes went to 1.0, all asset values declined at the same time and risk metrics proved worthless…the defects in the application of “Modern Portfolio Theory” became highly visible and leading investment professionals began the search for better methodologies. In this paper I endeavor to explore some of those important issues.


Firstly, the basic principle of Modern Portfolio Theory appears sound; its frailties show up in its application, more particularly in the effort to apply quantitative methods. The basic concept is that an investor can optimize the risk adjusted returns of a multi-asset class portfolio by configuring the mix of asset performance characteristics. The application of this principle is far more complex than has previously been practiced. How often have I seen the seductive charts showing the “Efficient Frontier”, as if one could simply compute the best mix of investable assets. If only life were that simple.

The Fatal Flaw: Risk and Volatility are not the same

A fatal flaw in the application of the principle was the assumption that risk and volatility were the same thing. The formulation of Modern Portfolio Theory required data to be inserted in equations. Leading economists and academics struggled with this fact and, so, decided that volatility (for which they had ex-post data) was the measure of risk. Further they assumed that the historical volatility of each asset group would be predictive of the forward volatility/risk. From a macro point-of-view, much of the volatility data used in today’s models relates to historical periods that are not relevant looking forward. For the five year leading up to 2008, volatility was low and markets were stable. Then in late 2008 and early 2009 volatility went through the roof. Going forward which data do you use? Perhaps neither characterizes the forward risk situation..

A practical example of the frailty of the assumption (that volatility and risk are the same) came in 2007 with the notorious collapse of the Bear Stearns’ High Grade Structured Credit Strategies Fund which exhibited very low volatility statistics until it completely collapsed to near zero value (in a couple of months) with no warning. Even the title of the fund implied quality and low risk. Contributing to this insidious development was the fact that there was no underlying trading of the fund’s assets, therefor the manager (Bear Sterns) was marking the value on investors’ monthly statements to a model, not to an actively traded market (because there was none).

The Bear Sterns fiasco points to another flaw in the use of volatility statistics, namely that they can be misleading when applied to illiquid asset categories, such as real estate, private equity and some hedge fund strategies. Furthermore, if a security or portfolio trades with high volatility through a wide range, but experiences more “up days” than “down days”, moving higher over time, does that imply that it is high risk? To a pragmatic investor, risk is the likelihood of losing money…not the range of price variation over relatively short periods of time. Such variations do not reflect the investor’s holding period or the fact that he/she has the choice of exit timing. While there certainly is a wide range of risk characteristics of different asset classes and individual investments, risk cannot be quantified with any degree of precision. It therefore must lie in the domain of business judgment.

The fundamental issue here is that it is pure folly to think that one can mathematically determine risk to a couple of decimal points using statistical ex-post volatility data to project ex-ante outcomes. The assessment of risk must consider a range of issues and scenarios…but ultimately requires a high level of investment judgment. Within any asset class, investment strategy or fund, the range of risk characteristics must be evaluated by the investor on a prospective basis. Such risks might include the stability of a fund’s investment leadership, its team, the degree of liquidity of its assets, its net market exposure and the amount of leverage it uses or does not use.

Efficient Frontier

The efficient frontier has been described by a line plot on a graph that presents risk vs. reward (expected returns). It is used to create a collection of assets that optimize the overall risk-return characteristics of an investment portfolio. Since it is not possible to reasonably quantify risk or the returns that will be realized, one might question the application of this type of analysis. So, is there an “efficient frontier” and where is it? The principle is sound, but the challenge comes in its application. One cannot draw a line on a two-dimensional piece of paper that describes it.

For the efficient frontier methodology to work, it must rely on a few key assumptions; namely that the markets are rationale, that it prices securities in an efficient manner and that forward risk can be quantified with some degree of precision using historical volatility data. Increasingly, investors are coming to the realization that none of these assumptions are true. Another assumption intrinsic to this methodology is that the distribution of expected outcomes (returns) may be characterized by a Gaussian curve. In fact, many asymmetric or by-modal curves may describe expected returns on an ex-ante basis.

Diversification

In spite of the fact that diversification offered no safe “hiding place” during the 2008 credit crisis, it continues to be an important discipline in constructing portfolios that manage the two strange “bedfellows” of risk and opportunity. However, it is not a two dimensional phenomenon as has been historically displayed on risk-return charts. Missing from the picture is the consideration that many asset classes are cross correlated to the same risk drivers, so the benefits of diversification in such cases can be minimal. A weakening economy may put downward pressure on equity prices, but also would cause commodities and real estate to drop in value. Thus, in such a case, diversification across these asset classes offers little risk mitigation value. One must look at the correlations between asset classes to understand whether diversification can deliver risk abatement value or not.

Also, diversification should not lead an investor to buy some of everything. Portfolio construction should reflect a carefully constructed mix of asset classes and investment strategy types should consider correlations, lack of correlations and counter-correlations. The mix of these should reflect the scenarios that the investor believes are most likely to develop. As important as the risk considerations are, it is even more important that the portfolio mix reflect the investor’s judgment of the profile of opportunities that are expected during the time horizon.

Another important consideration is the risk parameters and capital market assumptions that one puts into the asset allocation process. The validity of outcomes can only be as good as the quality of the inputs. Here also it is easy to fall into the trap of characterizing a situation with single numbers. The risk profile of any asset class can be better described through a probability distribution rather than a single number. The shape of that risk profile curve varies depending on future scenarios of events not known at the time the asset allocation strategy is formed. The same can be said for the capital markets assumptions (ex-ante return projections). In an environment with many unknowns and unknowable’s, the investor’s judgment comes seriously into play. Historical data can only be a small guide to future expectations. To rely upon it exclusively, is foolhardy.

The Skill Factor

Economists all neglect the effects of skill on investment risk and outcomes. Helicopters may be unsafe when flown by an amateur pilot, but may be very safe when operated by a Marine copter pilot with thousands of hours of experience. It is the same with investment managers. This shows up, in particular, in those areas where managers have a high degree of discretion and ability to affect the course of events. When one looks at the investment categories of venture capital, private equity, hedge funds, real estate and some others, one observes a wide range of performance outcomes. The dispersion of return performances from the top quartile performers to the bottom quartile is substantial. On the other hand for managers of traditional long-only equity funds, mutual funds and bonds, the variations (from market benchmarks) are narrow.

Since economists deal with statistical methods, it is necessary to have a big sample which lumps good, bad and average managers together. In this the attribute of skill is lost. This principle also comes into play on a fund-specific level. Many fund managers seeking to avoid risk or variance from benchmarks build large diversified portfolios with sector allocations that track closely to their benchmarks. I call these “benchmark huggers”. In doing so, they dilute to positive effects of skill in the performance of their portfolios. In the hands of a skilled investment manager, a sensible level of concentration and directionality is a good thing.

Liquidity

A basic economic principle is that one must pay a higher price for liquid assets than illiquid assets. While, in general, this may be true, investors overlook the advantages of flexibility offered by liquid assets. Most investors look at the need for liquidity to be a use-of-funds or spending requirement. They overlook the value that it has in providing the ability to quickly adapt to changing conditions or unforeseen events. Harvard Management learned this lesson the hard way during the recent credit crisis when they found their investment portfolio to be substantially committed to illiquid holdings at a time when rapid change put a premium of flexibility. In addition to causing distress, this lack of capital flexibility presents an opportunity cost since such portfolios cannot seize the opportunities that emerge during such periods. In many situations, the flexibility to easily sell an asset and re-deploy one’s capital in order to re-align investment exposures can be very valuable…worth the extra price. In portfolios with a mix of liquid and illiquid assets, one must be aware of the tradeoffs and intelligently weigh the benefits of an illiquid investment against the cost of compromising flexibility. It is not always the case that the illiquid investment offers a better return opportunity.

Strategy

In an environment of uncertainty, with tools of limited value, what is an investor to do? It gets down to a matter of practical judgment. Often investment managers set asset allocations for the long term. Investment environments change. They are dynamic. It is not a good idea to fix asset allocations, to let them become “straight jackets” over a long time horizon. There is much opportunity in getting tactical allocations right and, conversely, a lot to be lost in getting them wrong. Therefore, it is important to bring the best thinking to the process of dynamically allocating assets from a tactical standpoint. ”Tactical” implies time horizons of 1-2 years whereas strategic implies a longer timeline from an asset allocation standpoint. In making such tactical allocations, one must recognize the uncertainty of the decision making environment and the imperfection of even the best investor’s ability to judge the future. For this reason it is important for every investor, in setting tactics and strategy, to ask themselves “What if I am wrong?”. Good investors deal with this by factoring into their decisions the prospect of less likely scenarios.

The look forward is perhaps the most important determination that an investor will make.

The process of asset allocation is best accomplished by carefully thinking through and researching the various high probability scenarios that are likely to develop. In doing so an investor must anticipate changes that are likely to develop and their impact on the various asset types and strategies in their portfolio. Only then can one consider the mix that presents the best composite return potential and the lowest risk profile.

The degree of concentration in cross-correlated assets depends on the investor’s level of conviction on his/her primary scenarios. If there is a high degree of uncertainty about the catalytic drivers of the scenarios, then a more uncorrelated asset mix (diversification) is warranted. Conversely, when uncertainty is low and conviction is high, more concentrated, directional strategies can be employed. At the end of the day, it is a matter of optimizing the probability of successful outcomes while assuming the level of risk that fits an investor’ tolerance level and objectives.

Investors live in a world characterized by uncertainty and change. As they continue in the challenge to manage the two strange “bedfellows” of risk and opportunity, it is important to constantly re-examine the validity of the tools and theories practiced in the investment industry and look for better methodologies.