Dynamic large financial networks via conditional expected shortfalls

In this article, we first generalize the Conditional Auto-Regressive Expected Shortfall (CARES) model by introducing loss exceedances of all (other) listed companies in related to each firm, thus proposing CARES-X (where ‘X’, as usual, stands for eXtended case large-dimensional problems). Second, construct a regularized network US financial Least Absolute Shrinkage and Selection Operator estimation step. Third, also propose calibration approach uncovering relevant edges between nodes, finding that estimated structure dynamically evolves through different market risk regimes. We ultimately show knowledge extreme links provides useful information, since intensity these has strong implications on portfolio risk. Indeed, it allows us design effective management mitigation allocation strategies.