Title: Sequential break detection in panel data
Authors: Zuzana Praskova - Charles University (Czech Republic) [presenting]
Abstract: A panel data model with lagged dependent variables and unobserved individual effects is considered and a sequential procedure to detect change in coefficients of lagged variables is proposed. The test statistic to detect change is based on quasi-likelihood scores and quasi-maximum likelihood estimators computed from a training data set. Asymptotic properties of the test statistic are studied in case that both the number of panels and the number of observations in the training set are sufficiently large.