| 研究実績の概要 |
Many advantages of the panel data analysis rest on the common breaks assumption, under which the breaks occurred at a common location for all individuals. Nevertheless, as the institutional and cultural background, preferences, and level of economic development vary substantially among different individuals, the break points may be distinct across individuals. Moreover, a major concern for the panel data model is the unobserved heterogeneity due to unobservable common factors. In this research project, I have developed new approaches for break points detection in general and realistic panel data models.
(1) I propose an approach to test whether the break in parameters in panels occurs at a common location across individuals. (2) I relax the common break assumption and allow that the break points are common in each group but vary across groups. When the break structure is misspecified, I show that the estimated common break point is still consistent for one of the true break points. (3) New tests for structural changes are introduced in a more general and realistic panel data model, which allows for cross-sectional dependence in error terms. (4) The empirical analysis strongly demonstrates the usefulness of the proposed methods in addressing real-world economic issues.
|
