Title: Branching processes under model misspecification
Authors: Anand Vidyashankar - George Mason University (United States) [presenting]
Abstract: Inference for parameters of branching processes with and without immigration is well-understood. However, when the true probability model is misspecified, statistical estimators' behavior and the resulting inference concerning parameters are unknown. We address the estimator's asymptotic behavior when the offspring distribution is misspecified. Specifically, focusing on the offspring mean, we investigate the limiting distribution of the estimators when the true distribution belongs to a family of distributions belonging to the Kullback-Leibler class. Based on these results, we derive useful insights into conditional and marginal inference for these processes. Applications to robust inference are also provided.