Diversification: Recent Legal and Academic Perspectives

• Publication year: 2003
• Author: By Patrick J. Collins, Ph.D., CLU, CFA

DIVERSIFICATION: RECENT LEGAL AND ACADEMIC PERSPECTIVES

By Patrick J. Collins, Ph.D., CLU, CFA^*

I. INTRODUCTION: A SEMINAL ARTICLE IN THE HISTORY OF MODERN PORTFOLIO THEORY

James Lorie (co-founder of the Center for Research in Security Prices at the University of Chicago) published, in 1975, an important article on portfolio diversification.¹ Investors, in Lorie's opinion, fail to appreciate certain critical implications of academic research with respect to construction of investment portfolios. Specifically, Lorie states that merely picking "good" or "safe" stocks cannot reduce risk. Portfolio construction is not optimal when it merely bundles together stocks of blue chip companies (firms exhibiting strong current financial statements and favorable accounting ratios). Rather, portfolio risk reduction depends on combining securities with differing economic characteristics as measured by their volatility and tendency to move either in tandem or separately from each other. Investors who seek only to maximize returns (i.e., those who do not care about risk) will put all of their money in the single most promising security (or sector). Investors who are concerned about risk, however, will employ a strategy of diversification.

Lorie points out that modern portfolio theory views risk as being composed of two elements. The first source of risk ("risk" is defined as uncertainty about the magnitude and direction of future price changes) is "market risk." Market risk includes changes in rates of industrial production, inflation, national income, unemployment, tax rates, regulatory policies and so forth. This risk cannot be eliminated through diversification simply because it exists as part of the marketplace itself. The only way to avoid market risk is to avoid the market - i.e., not investing. The second source of risk is "idiosyncratic" or "firm-related" risk. This is the risk that affects individual companies. It includes poor decision making by management, technological obsolescence, litigation risks, strikes and labor unrest, and so forth. Examples of firm-related risk include the placement of arsenic in Tylenol manufactured by Johnson & Johnson, the crash of the corporate jet carrying the top management of Texas Gulf Sulphur, the reactor failure of the General Public Utilities facility at Three Mile Island, and Enron's stealth accounting procedures. Scientific diversification aims at reducing idiosyncratic risk. One of the central tenets of modern portfolio theory is that idiosyncratic risk is uncompensated risk - because idiosyncratic risk can be eliminated, the market does not reward those who voluntarily elect to assume it.

In the modern sense of the term, diversification does not mean owning many investments. Rather, it refers to eliminating non-market risk. One statistic that captures the degree of closeness between the actual investment portfolio and the market in general is correlation (or, the square of correlation known as the R2 or the coefficient of determination statistic). Lorie states, however, that some investors fail to understand the deleterious consequences of failing to eliminate idiosyncratic risk. He points out that even small departures from perfect correlation with the benchmark or "bogey" portfolio can result in significant underperformance. Measuring systematic or market risk via the Beta statistic (with a Beta of 1.0 equaling the risk of the market), Lorie points out that "a portfolio with a beta of 1.0 and diversified 95% (coefficient of determination of 0.95) would fairly often have returns as much as 4.5 percentage points different from the market as a whole." This means that a portfolio that is 97.5% correlated to the target benchmark could underperform the bogey by substantial margins. Merely matching Betas (e.g. XYZ stock's Beta is close to that of the market and therefore the stock is about as risky as owning the market in general) is wholly inadequate as a portfolio design or risk management strategy.² Alternately, picking a few "safe" or low Beta stocks may, in fact, be more risky that owning a higher Beta but more well diversified portfolio. Lorie cautions investors that "departing from perfect diversification in an attempt to identify undervalued securities" may be a far riskier venture than conventional wisdom implies.

II. NEW APPROACHES TO ASSET MANAGEMENT: WHAT IS THE ROLE OF THE INVESTMENT FIDUCIARY?

For approximately a quarter of a century, Lorie's cautions remained largely unheeded by certain segments of the investment management industry. Even though ERISA, enacted in 1974, augments the duty to diversify, ERISA views diversification primarily as a technique for limiting large losses. Although this is certainly a valid reason for avoiding concentrated investment positions, Lorie's 1975 article focuses not on the risk of catastrophic loss but, rather, on "tracking error" risk. Failure to diversify completely (what Lorie called "the wisdom of extreme diversification") allows for the possibility of long-term underperformance relative to a comparable market-oriented benchmark. At the time of Lorie's essay, however, investors could not own such a benchmark, and this is perhaps one reason why the essay failed to command greater attention. Today, however, investable benchmarks that replicate capital market returns are available through low cost index investments.

The cataclysmic shattering of wealth brought about by the OPEC oil crisis of the 1970s turned the money management industry's attention towards theories of asset pricing based on models other than discounted projected earnings or relative valuation. With many folklore investment homilies destroyed, Wall Street sought more credible asset management guidelines.³ Money management "pros" soon trumpeted quantitative performance measures such as Alpha (a risk-adjusted measure of value added or subtracted by the manager) and Beta (a measure of systematic risk relative to the market). Nevertheless, instead of embracing broad-scope diversification, many managers continued to pick a few good stocks. Paradoxically, they could point to

[Page 35]

academic publications to justify their portfolio composition strategies. Independent evidence suggested that owning twenty to fifty stocks provides a well-diversified portfolio; and, if a money manager has superior stock selection ability, the investor can both control risk and enjoy market-beating performance. Without doubt, this asset management approach generates intriguing and compelling sales stories. What happened to Lorie's "wisdom of extreme diversification"?

Ironically, it was Lorie's earlier research, co-published in 1970 with Lawrence Fisher, which commanded the attention of the money management industry.⁴ The 1970 study reported that a sixteen-stock portfolio achieves 90% diversification relative to a portfolio comprised of all stocks listed on the New York Stock Exchange and, furthermore, a thirty-two-stock portfolio achieves 95% diversification. Other investigators writing in the 1970s and 1980s confirmed Lorie and Fischer's conclusions.⁵ Although independent studies reach slightly different conclusions based on analytical methodologies and the sample periods under investigation, most demonstrate that a portfolio having relatively few stocks achieves substantial risk reduction benefits. Which Lorie is correct — the 1970 Lorie or the 1975 Lorie? The answer has important implications for defense of surcharge cases in which the investment fiduciary is accused either of imprudent asset concentration risk or of long-term chronic underperformance relative to readily available, low-cost, passively-managed investment alternatives.⁶

III. A BRIEF DIGRESSION: SOME STATISTICAL CONCEPTS

The answer is that both articles are correct but, in the main, investors misinterpret the implications of the first. In 1970, Fisher and Lorie calculate the dispersion in returns achieved, on average, by forming portfolios of various sizes. Dispersion of returns is measured by the standard deviation statistic - the higher the dispersion, the riskier the investment. But standard deviation measures total risk — both market risk and idiosyncratic risk. The purpose of diversification is not, however, to eliminate market risk but to diminish idiosyncratic risk only (eliminating market risk also eliminates all return in excess of the risk-free rate — a result that most investors would find distressing!). A reduction in the total dispersion or variability of returns does not measure the benefits of diversification. A reduction in idiosyncratic risk (as measured by the R2 coefficient of determination) is, according to Lorie, the more correct measure of effective diversification. An R2 statistic of 1.00 indicates that the actual portfolio is perfectly diversified with the market benchmark. To the extent that the statistic has a value lower than 1.00 (with a lower-bound value of zero), the returns of the actual portfolio will fail to match those of the benchmark — hence, "tracking error" risk is heightened.⁷ A portfolio with an R2 value of 0.86, for example, has 86% of its returns explained by the market and 14% of its returns attributable to risk factors other than those represented by the market. That is to say, it has 14% idiosyncratic risk that, with further diversification, could be eliminated. Despite the fact that a portfolio might have a standard deviation equal to that of the benchmark, tracking error risk still exists if the R2 statistic is less than 1.00. The 1970 study measures the reduction in total return variability by adding additional stocks to a portfolio. The standard deviation statistic is a valuable guide to quantifying the perils of asset concentration (ERISA's concern with large losses). However, this is quite different from measuring the benefits of diversification as a strategy to eliminate the risk of persistent underperformance of the market-oriented benchmark index. Thus, the standard deviation...