Title: Testing structural breaks in large dynamic models based on extremal distribution
Authors: Zuzana Praskova - Charles University (Czech Republic) [presenting]
Abstract: A linear dynamic panel data model is considered. Its coefficients can change in a time that is unknown and same for all the panels. An Erdos - Darling type test procedure to detect a change, based on the quadratic form of cumulative sums of weighted LSE residuals, is proposed. The asymptotic distribution of the test statistic as the number of panels and number of observations converge to infinity is studied, both under the null hypothesis of no change and under alternatives of a change. The finite sample behaviour of the test is also studied.