This talk has been automatically captioned. You can remove these by pressing CC on the video toolbar.
Name: Corina Birghila
Talk Title: Distributionally robust bounds for tail indices
Abstract: In this work, we provide robust bounds on the index of tail heaviness in the context of model misspecification. They are defined as the optimal value when computing the worst-case tail behaviour over all models within some neighbourhood of the reference model. The choice of the discrepancy between the models used to build this neighbourhood plays a crucial role in assessing the heaviness of the asymptotic bounds. To this end, we evaluate the robust tail behaviour in ambiguity sets based on the Wasserstein distance and Csiszar f-divergence and obtain explicit expressions for the corresponding asymptotic bounds. Numerical simulations are provided to emphasize the difference between these bounds and the importance of the choice of discrepancy measure.
This talk is a contributed talk at EVA 2021.