Following in the footsteps of earlier research, the authors confirm that non-capitalization-weighted index strategies deliver higher returns than cap-weighted strategies. The excess returns persist regardless of whether deviations from capitalization weights are motivated by sensible investment beliefs or not. Creating portfolios using criteria other than market capitalization inherently slants them toward value and size, which explains the outperformance.
What’s Inside?
The authors compare performance metrics of portfolios constructed on the basis of common investment strategies with the performance of a capitalization-weighted index using the largest 1,000 US firms. These portfolios generate greater returns (11.75% on average versus 9.66%) and higher Sharpe ratios than the cap-weighted index. The authors invert the weighting of each of these constructed portfolios in two different ways and surprisingly find that the upside-down portfolios generate even greater excess returns (12.88% on average versus 9.66%). Even randomly generated portfolios (i.e., Malkiel’s monkey) outperform the cap-weighted index before transaction costs. Positive alphas on the upside-down portfolios challenge the assumption motivating the original strategies. When using the capital asset pricing model (CAPM), observed alphas are positive, and most are economically and statistically significant. But when the Fama–French four-factor model (FF4) is applied, the alphas are not significant, except for a few portfolios that invert fundamental indexation strategies.
How Is This Research Useful to Practitioners?
These findings should remind investors to be cautious about benchmarks and portfolios that deviate from a cap-weighted index because the excess returns arise primarily as a result of increased exposure to well-documented risk factors—namely, value and size. CAPM alpha and investment themes are not what drive the excess returns on the studied portfolios. Instead, the inherent and often unintentional slant toward value and small-cap stocks produces the returns. This value and size tilt exists even if a portfolio is constructed to capture strong earnings growth.
The authors do observe a few exceptions for the upside-down portfolios, which is unexpected because these portfolios are not based on superior market knowledge. They note that further studies may discover a currently unidentified explanatory factor.
Investors making the decision to follow a particular size or style tilt could forgo paying for stock selection skill and passively invest, aiming to incur as little cost as possible. Investors who seek active management (i.e., alpha) could benchmark their performance against equal- or capitalization-weighted portfolios that are transparently constructed to reflect their investment style. This action would help investors avoid falling into the trap of benchmarking against other active or quasi-active portfolios that are neither transparent nor specifiable in advance.
How Did the Authors Conduct This Research?
Using CRSP/Compustat merged data from 1964 to 2012 for the top 1,000 US-listed firms, the authors evaluate the returns of annually rebalanced portfolios that are constructed using such investment criteria as volatility, market beta, downside semideviation, minimum variance, book value, and earnings. They also simulate a dart-throwing monkey. For each portfolio, they evaluate two additional portfolios that are constructed by inverting the weights used in the original portfolio. The authors compare Sharpe ratios, value added, tracking error, information ratios, and the CAPM and FF4 alphas. They do not take costs or investment capacity into account.
Performance is measured relative to a capitalization-weighted portfolio of the same constituent firms as well as an equal-weighted portfolio, which also outperforms the cap-weighted portfolio. It becomes apparent that an equal-weighted portfolio is a reasonable benchmark for non-price-weighted portfolios.
In their study of global portfolios, the authors find similar results.
Abstractor’s Viewpoint
The authors are clearly concerned that capitalization weights may be a problematic index construction method, but the results of their rigorous analysis highlight the obfuscation opportunities available to promoters of active (i.e., non-capitalization-weighted) index products by confirming that the excess returns are simply a repackaging of the well-known risk premiums documented by the four-factor model (also see Carhart, Journal of Finance 1997). Non-capitalization-weighted index benchmarks deviate from the aggregate distribution of investors’ wealth in their respective market segments, and the resulting size and style biases are hazardous because of their opacity. Unskilled passive investors should stay with capitalization weights.
The positive and significant alphas observed using the four-factor method on the inverse fundamentally weighted portfolios are consistent with Perold’s caution against “fundamentally flawed indexing” (Financial Analysts Journal 2007).