A0551
Title: Shrinkage estimation of multiple structural breaks in spatial panel data models with multifactor error structure
Authors: Chaowen Zheng - University of Southampton (United Kingdom) [presenting]
Siqi Dai - Hunan University (China)
Abstract: The aim is to consider a spatial panel data model where multiple structural breaks occur in both the coefficients for the spatial lagged dependent variable and regressors. The model can accommodate cross-sectional dependence arising from spatial dependence and unknown common factors. To tackle the challenging issues of endogeneity and time heterogeneity, a novel penalized generalized method is proposed for moments estimation with common correlated effects (PGMM-CCEX). Specifically, the PGMM-CCEX method uses cross-sectional averages of regressors as factor proxies when constructing the instrumental variables and employs adaptive group fused lasso to detect multiple structural breaks. It is shown that the PGMM-CCEX method can consistently estimate the number of breaks and their dates, and the resulting regime-specific coefficients are also consistent and asymptotically normally distributed. Monte Carlo simulations show that the PGMM-CCEX method has the superior finite-sample performance of the proposed estimators, which is quite satisfactory. An empirical application to cross-country economic growth of 103 countries from 1970-2019 reveals a more complete and time-varying picture of the driving forces behind economic growth.
