PT - JOURNAL ARTICLE
AU - Hight, Gregory N.
AU - Haley, Joseph D.
TI - Low-Risk Benchmarking Transcends Rebalancing Methods
AID - 10.3905/joi.2020.1.159
DP - 2021 Jan 31
TA - The Journal of Investing
PG - 7--30
VI - 30
IP - 2
4099 - http://joi.pm-research.com/content/30/2/7.short
4100 - http://joi.pm-research.com/content/30/2/7.full
AB - The dismal performance of managed investment and the success of equal allocation and minimum variance models prompt these questions: Can rebalancing driven by minimum variance and maximum diversification rebalancing outperform naïve models? Can a minimum variance model produce higher risk-adjusted returns than a maximum diversification model when security selection favors low-correlated assets? This study uses expected shortfall to measure risk, bootstrapping to transform fat-tailed distributions so they are suitable for t-tests, and factor analysis to help explain the relative performance of the models. The minimum variance and maximum diversification models outperformed naïve models, and the minimum variance models produced higher risk-adjusted returns than the maximum diversification model. Market factors adequately explained differences in returns between rebalancing models, but adding a factor for the effect of rebalancing procedures improved all models. All models constrained by the lower risk benchmark produced higher risk-adjusted returns than corresponding models constrained by the higher risk benchmark. This outcome suggests many rebalancing models—perhaps even return-based models—could produce superior risk-adjusted returns if lower risk benchmarks constrain risk.TOPICS: Portfolio theory, portfolio construction, risk management, factor-based models, exchange-traded funds and applicationsKey Findings▪ The two minimum variance models produced statistically significant superior risk-adjusted returns.▪ All models constrained by the lower risk benchmark produced higher risk-adjusted returns than corresponding models constrained by the higher risk benchmark.▪ Although market factors explained differences in model returns, adjusted R2 values increased when regression models included a rebalancing factor.