What Free Lunch? The Costs of Overdiversification

Shawn McKay, Robert Shapiro, and Ric Thomas, Financial Analysts Journal, CFA Institute, vol 74, no 1 (First Quarter 2018): 44–58

The authors examine the costs of excess diversification to highlight the trade-offs among diversification, active risk, and excess return in active equity allocations. They find that too much diversification frequently reduces risk levels to a point where achieving return targets may be unrealistic. Additionally, they find that as diversification increases, fees per unit of active risk and fees per unit of active share increase.

The authors analyze a data set of 663 large-cap active equity funds. From this data set, ten funds were chosen with the highest levels of active risk, active share, and percentage of active risk attributable to stock-specific risk. The study showed that as each fund is added to a basket, all three risk attributes declined significantly; when all ten were combined, the active share and stock-specific risks were close to those of the average of the 663-fund data set.

As investors are not likely to pick the ten funds with the highest risk attributes, the authors reviewed 10,000 random combinations of portfolios with two to ten funds. As the number of funds in the portfolio increased, achieving meaningful active risk became highly unlikely. This finding was then compared to the fundamental law of active management, which states that active return is a function of forecast accuracy, the number of forecasts made, and active risk. Holding the number of forecasts constant, the accuracy of manager forecasts needed to achieve a 1% excess return increased exponentially as the active risk was reduced below 2%.

The authors also discuss how active risk becomes more expensive as additional funds are added to a portfolio. They do this by analyzing the fees per unit of active risk and fees per unit of active share. As more funds are added to a portfolio, the authors find that both fees per unit of active risk and fees per unit of active share increase. They argue that hiring a high number of active managers essentially gives an investor an index-like equity allocation, but at a higher cost.

While the authors note that they do not suggest avoiding all active management, they do caution investors on the costs and risks of adding an excessive number of active managers. They advise investors to consider the fees they are willing to pay per unit of active risk when making decisions on how many and which active managers should be included in a portfolio.

When Diversification Fails

Sebastien Page and Robert A. Panariello, Financial Analysts Journal, CFA Institute, vol 74, no 3 (Third Quarter 2018): 1–14

This article explores the use of diversification as a tool for managing loss exposure within investment portfolios. The authors analyze the benefits of diversification by examining asset class correlations during tail events. They find that diversification often fails during market downturns—when it is needed most.

Investors frequently cite diversification as a prudent method of defending against outsized losses when markets tumble; however, the authors argue that diversification often fails in this respect due to the asymmetric correlation of most asset classes during market tail events. In other words, when equity markets sell off, correlations between assets tend to increase. Many prior studies on the topic failed to discover these asymmetries or incorrectly assumed the existence of symmetry.

The authors examine traditional asset class correlations between various asset class pairs, styles, sizes, and geographies. All were found to be highly asymmetric, thus proving the ineffectiveness of diversification in a market sell-off. The authors furthered their analysis by adding non-traditional assets. Neither hedge funds nor private investments (the latter of which was adjusted for its smoothing bias) proved ineffective as diversifiers.

The authors also reviewed whether risk factors were helpful as a source of diversification. They demonstrated that some risk factors delivered superior diversification benefits compared to traditional asset classes. The authors argue that risk factors may be more helpful because it’s different from how investors are typically positioned and it encompasses a broader universe of assets.

According to the authors, diversification tends to be driven by fundamentals in normal market conditions. But, in times of panic, investors often sell risk—regardless of the underlying fundamentals—creating the asymmetry in asset class correlations. The findings of this study suggest that investors should look beyond traditional diversification strategies. They should consider using hedging and dynamic strategies, such as risk factors with embedded short positions or defensive momentum strategies, as a means to manage portfolio risk.

Liquidity as the New Leverage: Will the Machines Amplify the Next Downturn?

Charles P. Himmelberg and James Weldon, Goldman Sachs, May 22, 2018

The authors find that the increased prevalence of high-frequency trading (HFT) strategies  may be contributing to the increased number of flash crashes and believe—similar to how the rise of leverage provoked the global financial crisis around a decade ago—the current state of trading liquidity could exaggerate the next downturn.

The authors evaluate the largest one-day move in the CBOE Volatility Index (VIX), which occurred on February 5, 2018. After finding no fundamental reason to justify a jump of its magnitude, the authors assert that poor trading liquidity was a contributing factor. The authors maintain that HFT algorithms behave similarly and can feed off each other in a correction; moreover, these algorithms may misinterpret negative market news.

The authors note that investor desire for risky yield alternatives increased following the global financial crisis (GFC). Like the rapid growth in financial engineering that occurred prior to the last crisis, the authors worry that new financial innovations, such as high-frequency trading, may similarly exacerbate the next market downturn.

Although HFT algorithms have an advantage over human traders in processing normal market information, the authors conclude that these algorithms sometimes fail to process complex market information correctly. The authors emphasize the importance of tail risk analysis in assessing overall market risk and suggest investors not be complacent about the quality of trading liquidity for even the most liquid markets.

Reconsidering Hedge Fund Contagion

Richard Sias, H.J. Turtle, and Blerina Zykaj, The Journal of Alternative Investments, vol 22, no 1 (Summer 2018): 27–38

In times of crisis, hedge fund contagion has been widely blamed for exacerbating losses—exactly what hedge funds are supposed to protect against. The authors examine previous studies and present new research that supports the assertion that hedge funds not only do not make these situations worse but also help protect investors during market shocks.

The hedge fund industry has grown dramatically over the past 20 years. Over that time, their perceived impact on the global marketplace has been controversial. Hedge funds have been widely associated with major negative market events; investors, regulators, and the popular press have often cited hedge fund contagion as causing, or at least largely contributing to, the 1998 financial crisis, the 2007 quant crisis, and the 2007–08 GFC. The authors argue that evidence of contagion (broadly defined as a left-tail event causing increased correlations between assets and thus steep losses) is actually quite scarce in the hedge fund industry.

The authors review two recently published academic studies providing evidence that hedge funds are no more likely to cause contagion than other asset classes. The first study disputes findings from an older study, showing what was previously thought to be evidence of hedge fund contagion caused by liquidity shocks was actually the result of model mis-specification. In the second study, the authors show hedge funds were less likely than other investment vehicles to crowd into the same stocks and that there is no evidence of demand shocks in hedge funds leading to broad contagion. The effects of hedge funds on two major negative market events, the 2007 quant crisis and the 2007–08 GFC, also have been misinterpreted according the authors.

Hedge funds exiting equity positions due to declining leverage and investor withdrawals is often cited as a core contributor to the 2007–08 GFC. The authors provide evidence that hedge funds were actually more stable and less susceptible than other assets classes to increasing correlations causing steep declines. Similarly, during the 2007 quant crisis, a large negative shock followed by a quick rebound was viewed as evidence of a liquidity shock causing contagion across hedge funds; however, the authors show this shock only affected market-neutral strategies, not all hedge funds. Correlations actually declined between sub-categories of hedge funds, increasing the diversification benefit when it was needed most.

Properties of Long/Short Commodity Indices in Stock and Bond Portfolios

Tom Erik Songsten Henriksen, The Journal of Alternative Investments, vol 20, no 4: 51–68

Institutional investor interest in commodities exploded in the 2000s. The author examines the impact of adding a long/short commodity index to a traditional portfolio. Their results indicate that a long/short commodity index tended to reduce a traditional portfolio’s total return but improve its risk-adjusted return.

The rationale for adding commodities to a stock/bond portfolio is well documented and empirically tested. Commodity portfolios have a low correlation to stocks and bonds and can serve as a hedge against unexpected inflation. The bulk of commodity-
related research is focused on passive, long-only positions in a basket of commodity futures, but, the author argues, given a long-term expected return of zero (considering that commodity prices are mean reverting over economic cycles), a long/short strategy may be more appropriate than passive, long-only positions.

To test long/short commodity indexes, the author employs four strategies: (1) “term structure” establishes positions based on the degree of backwardation; (2) “market neutral” has no spot price exposure; (3) “momentum” employs a 12-month moving average as a trading tool; and (4) “fundamental” relies on supply/demand forecasts to make investment decisions. The author uses eight different stock and bond indexes as well as various weighting schemes to analyze how an allocation to commodities would impact a traditional portfolio.

Over the full period reviewed (2002–15), the fundamental commodity index performed best, both on a stand-alone basis and as part of a stock/bond portfolio. The term structure strategy performed the worst. With very few exceptions, the addition of a long/short commodity index, regardless of strategy, lowered the total portfolio’s volatility and tail risk. But these positive characteristics were often accompanied by a reduction in performance.