However, in the last decade, we have also observed some undeniable mispricings—technology stocks in early years of the decade and homebuilders and mortgage bankers in mid-2007. Arguably, financial and consumer cyclical stocks of early 2009 were significantly undervalued. The S&P 500 capitalization-weighted index, by systematically overweighting the overpriced and underweighting the underpriced stocks, trailed the S&P Equal Weight Index by 3.8% per annum. Unpleasantly for investors, both active and passive approaches have delivered poor results.
While we believe strongly in markets being inefficient, we underperform the benchmark net of costs. Additionally, we believe that cap-weighting is an inappropriate passive investment vehicle where prices are inefficient as the index overallocates to expensive stocks and underallocates to cheap stocks. There is a third option for clients who wish to allocate to equities—non-price-weighted strategy indexes, which offer investors an alternative and complementary choice. Since the publication of “Fundamental Indexation” in the Financial Analysts Journal,1 many asset managers and indexers have created a dizzying array of “alternative betas” or “strategy indexes” designed to offer investors passive investment vehicles that are grounded in the hypothesis of market inefficiency.
We have studied the similarities and differences among these alternative beta strategies. Our comprehensive research, which was published in the Financial Analysts Journal, compares the performance of several of the well-known alternative betas using a common data set and investment parameters.2 We summarize the main findings of that research in this issue of Fundamentals.
The non-price-weighted strategies examined can be classified into two categories: heuristic-based-weighting methodologies and optimization-based-weighting methodologies.
The heuristic-based strategies include naïve Equal-Weighting and its extensions that seek to eliminate the undesirable characteristics of a simple equal-weighting strategy (e.g., Equal-Weighting’s sensitivity to the number of stocks in the portfolio). The strategies examined are Diversity-Weighting, which has limited turnover and tracking error relative to the cap-weighted benchmark as it is mathematically an interpolation of equal-weighting and cap-weighting; Risk-Clusters Equal-Weighting, which groups securities by country and risk factors, intuitively provides more robust diversification as it equal weights uncorrelated risk factors rather than individual securities; and the Fundamental Index Strategy, which completely severs the link with market prices, and instead uses variables tied to the economy to select and weight securities.
Optimization-based strategies are generally more complicated; they require complex mathematical and computational routines to arrive at a mean-variance optimal portfolio. While they are theoretically attractive, their models are difficult to apply in practice. Ad hoc assumptions for estimating the expected returns for all stocks and their covariance matrix are often required. The optimization-based strategies we look at are the Minimum-Variance strategy, a popular approach which assumes uniform expected returns for all stocks and targets the left end of the efficient frontier; the Maximum Diversification Index, which incorporates information on expected stock returns and seeks to reduce portfolio volatility; and Risk Efficient Indexation, which assumes risk and return are related to their downside risk and includes carefully designed portfolio constraints.
Our research involved simulations of the alternative beta strategies using a consistent database, risk factor construction, and portfolio parameters. Total returns were calculated for each strategy at a monthly frequency from 1964 through 2009 for the U.S. strategies, and from 1987 through 2009 for the global strategies. We compared these strategies to two leading cap-weighted indices—the S&P 500 for U.S. strategies and the MSCI World for global strategies. The choice of date ranges depended entirely on the breadth of historical data.
Portfolio parameters were synchronized to achieve a controlled environment for performance comparison. As Table 1 shows, all of the strategies produced meaningfully higher returns than their cap-weighted benchmarks over the full sample period. In general, the optimized strategies have higher tracking errors and lower volatilities, and the heuristic-weighting strategies tend to have relatively higher volatilities and lower tracking errors. As expected, the minimum-variance portfolios show the lowest volatilities of the strategies considered.