Dynamic Correlation or Tail Dependence Hedging for Portfolio Selection

We solve for the optimal portfolio allocation in a setting where both conditional correlation and theclustering of extreme events are considered. We demonstrate that there is a substantial welfare loss indisregarding tail dependence, even when dynamic conditional correlation has been accounted for, andvice versa. Both effects have distinct portfolio implications and cannot substitute each other. We alsoisolate the hedging demands due to macroeconomic and market conditions that command importanteconomic gains. Our results are robust to the sample period, the choice of the dependence structure,and both varying levels of average correlation and tail dependence coefficients.