New Developments in Heterogeneous Panel Data models with Cross-sectional Dependence

In panel data analysis, a key objective is estimating the mean slope parameter, which reflects the average effect of an explanatory variable across cross-sectional units. The Common Correlated Effects Mean Group (CCEMG) estimator, introduced by Pesaran (2006), was developed for this purpose under broad assumptions. This thesis aims to improve CCEMG methodology, particularly by developing alternative variance estimators to enhance inference in realistic settings. An empirical application using Australian wheat yield data illustrates the approach. The results demonstrate meaningful alignment with state-level weather sensitivity, highlighting the practical advantages of the proposed new methods.