Robust portfolio selection with subjective risk aversion under dependence uncertainty

This paper solves the robust portfolio decision problem with subjective risk aversion under dependence uncertainty. We use spectral risk measures to capture investors’ subjective risk aversion while constructing mixture R-vine copula uncertainty with R-vine copula change-point detection to characterize dependence uncertainty. We introduce worst-case spectral risk measures (WSRM) and formulate portfolio decision problems with WSRM as risk measures. We conduct an empirical study using eight global stock indices. The results demonstrate that our robust portfolio outperforms two other portfolios without incorporating dependence uncertainty or considering dependence uncertainty constructed only through common economic crisis information in static and dynamic investments encompassing the full out-of-sample period and crisis periods across different risk aversion levels. The performance of our dynamic robust portfolio increases with investors’ risk aversion levels. Our robust portfolio strategy can help investors with diverse risk preferences achieve substantial profits and enable those with high-risk aversion to reap the rewards of their risk aversion.