The triumph of mediocrity in investing

Smart beta — a broad term for relatively passive investment strategies that use some alternative method of weighting stocks besides market-capitalization weights — are increasingly being considered, if not adopted, by institutional investors. They promise outperformance over traditional capitalization-weighted indexes on a risk-adjusted basis via the “factors” or other systematic “rules” they adopt in their stock selection approach.

There are some investors who still believe that following market-cap-weighted portfolios represents passive and diversified investments, unlike these systematic alternative beta approaches. These investors will be surprised by the results of our new research.

“Smart” beta is not quite appropriate as a term to cover all non-cap-weighted approaches that aim to outperform their market indexes, just as the term “hedge fund” covers very different investment strategies. Some smart beta approaches are factor-based. In other words, they use factors such as value, size and momentum to produce excess returns. But factor-based investing isn't new to quantitative equity managers who have long employed multifactor models in their portfolios. What is new is that the portfolio construction process of factor-based smart beta often reverts to simple weighting schemes without consideration of risk.

The second type of smart beta is diversification-based. It uses a systematic, rules-based approach to select equities based on some dimension of equality, such as portfolio weight, expected return, expected risk-adjusted return and risk contribution, for all stocks. We might call these approaches “naïve beta” because of their seemingly naïve assumption of equality — that all investments are the same in the dimension that is selected.

We researched four dimensions: portfolio weight; expected return; risk-adjusted return; and risk contribution. These, in turn, produced four types of naïve beta strategies: equally weighted, or EQ; minimum variance, MV; maximum diversification, MD; and risk parity, RP, respectively.

Here are the major findings.

All naïve beta strategies outperform the S&P 500 index, partly due to sector biases in the index. The index suffers from large allocation shifts — as over time, it adds weights to sectors with either the highest volatility or the lowest risk-adjusted returns.

Not all naïve beta strategies perform the same. MV and MD portfolios have high sector concentration (in defensive sectors) and high turnover.

The RP portfolio has diversified sector allocation and low portfolio turnover. This is the one we recommend for institutional investors seeking real diversification.

Of the four naïve beta portfolios, the EQ portfolio doesn't consider risk, just equal weighting of the investments. The other three do consider risk inputs: MV and MD optimize on risk inputs, which means they are highly sensitive to risk and return estimates. This research finds that for both these portfolios, the optimal allocations are highly concentrated in sectors and they can have high turnover — despite their systematic approach to allocation in one dimension with equality assumption.

The RP portfolio is the most robust in the risk budgeting process without optimization. Risk budgeting states the level of risk contribution from individual investments and derives the weights that satisfy the prescribed risk contributions. It is a portfolio construction process that uses risk inputs without optimization.

In addition, the RP portfolio incorporating a risk-budgeting approach is robust to the choice of risk model while consistently delivering a diversified portfolio with low portfolio turnover.

Our research offers a detailed analytical framework for the four naïve beta portfolios, followed by an empirical example of these portfolios based on sector returns of the S&P 500 index. To understand the results, we model the sectors into two groups of assets, where assets are more correlated within each group than across the groups. We look at analytic solutions of the portfolio weights within the groups and come up with explanations for the sector concentration or diversification.

In the empirical example, we look at sector allocation portfolios, based on the 10 S&P 500 sectors, using monthly returns from January 1990 to November 2014. Sector allocation is a practical concern for equity managers. It offers a manageable set of investment choices to infer results on portfolio weights, and it hasn't been an easy task for active managers to outperform the S&P 500 over this period, so it would be interesting to see how naïve betas did.

The results are quite distinct.

The MV portfolio is highly concentrated in the two sectors with lowest volatility: consumer staples and utilities.

The MD portfolio is less highly concentrated, with majority of exposures in utility, consumer staples, information technology and three other sectors.

The RP portfolio is much more diversified. All 10 sectors have weights between 7% and 14%, with the highest weights in low-volatility sectors (consumer staples and utilities) and lowest weights in high-volatility sectors (financials and technology)

Even though all three naïve beta portfolios are based on assumptions of equality, the MV and MD portfolios are concentrated across sectors, due to the optimization procedure. In contrast, the RP portfolio is more diversified across sectors, due to its risk budgeting process. In some sense, the RP portfolio improves on the EQ portfolio by incorporating risk inputs.

We also looked the sensitivity of the different naive beta approaches to asset correlations. Broadly, in terms of the sector correlations, the cyclical sectors tend to have greater correlations due to common exposure to economic growth, while the defensive sectors have lower correlations. The question is how are portfolio weights and turnover affected by this correlation structure within each type of naïve beta portfolio. Our analytic results show why the MV and MD portfolios are concentrated in defensive sectors while the RP portfolio is more diversified.

On the other hand, the index's sector weights change either due to price fluctuations of sectors or changes in the index constituents. Over the full 1990 to 2014 period, the technology and financials gained the most weights (13.7% and 7.7% respectively), health care gained a bit less and the other seven sectors declined in weight. Note the same two sectors — technology and financials — were severely affected by extreme market events: the tech bubble and the credit crisis. Given that the weights of these two sectors have been high and rising in the index, one can see how these sectors contributed to the Index's overall volatility. Not surprisingly, sector shifts caused by changes in the index's constituents have hurt overall index returns.

The significant shifts in (the S&P 500's) sector allocation are akin to making an active or “smart” bet on different sectors' relative future performance. Those bets have been detrimental to both the index's return and risk. In investing, being “smart” might not be a winning strategy.

Naïve beta portfolios have outperformed cap-weighted indexes in non-U.S., global and emerging-market equities, as well. The fundamental reason is the capitalization-weighted indexes have strong biases in their sector, country and stock allocations, due to two interconnected reasons. One is the short-term market momentum and the other is the change in index constituents, which is a highly active process on the part of index providers.

One cannot help but arrive at the conclusion that the capitalization-weighted indexes are only passive from a mechanical perspective and are quite active from an investment perspective. In contrast, naive beta based on an assumption of equality for some aspect of all investments is a truly passive investment. Its performance advantage, in a sense, is a triumph of mediocrity.

Edward Qian is chief investment officer in PanAgora Asset Management's Multiasset Investment Group in Boston. This article is derived from “The Triumph of Mediocrity: A Case Study of Naïve Beta,” research by Mr. Qian; Nicholas Alonso, a portfolio manager; and Mark Barnes, a director; at PanAgora and published in The Journal of Portfolio Management.